Governing The Firm

Governing the Firm
Workers’ Control in Theory and Practice
Most large firms are controlled by shareholders, who choose the board
of directors and can replace the firm’s management. In rare instances,
control over the firm rests with the workforce. Many explanations for
the rarity of workers’ control have been offered, but there have been
few attempts to assess these hypotheses in a systematic way.
In Governing the Firm, Gregory Dow draws on economic theory,
statistical evidence, and case studies to frame an explanation. His fun-
damental hypothesis is that labor is inalienable, while capital can be
freely transferred from one person to another. This implies that worker-
controlled firms typically face financing problems, encounter collective
choice dilemmas, and have difficulty creating markets for control po-
sitions within the firm. Together, these factors can account for much
of what is known about the incidence, behavior, and design of worker-
controlled firms.
A policy proposal to encourage employee buyouts is developed in
the concluding chapter.
Gregory K. Dow is Professor and Chair of the Department of Eco-

nomics at Simon Fraser University, British Columbia, Canada. He pre-
viously taught at the University of Alberta and Yale University, and he
has served as a visiting professor at the University of New South Wales,
the Erasmus Institute for Philosophy and Economics, and the Swedish
Collegium for Advanced Studies in the Social Sciences. Professor Dow
is associate editor of the Journal of Economic Behavior and Organiza-
tion, and he has published numerous articles on labor-managed firms
and other topics in economic theory in leading journals such as the
American Economic Review and the Journal of Political Economy. He
has also written many book chapters in edited volumes.
Governing the Firm
GREGORY K. DOW
Simon Fraser University, Canada
  
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo
Cambridge University Press

The Edinburgh Building, Cambridge  , United Kingdom
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521818537
© Gregory K. Dow 2003
This book is in copyright. Subject to statutory exception and to the provision of

relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.
First published in print format 2003
-
isbn-13 978-0-511-05506-5 eBook (Adobe Reader)
-
isbn-10 0-511-05506-4 eBook (Adobe Reader)
-
isbn-13 978-0-521-81853-7 hardback
-
isbn-10 0-521-81853-2 hardback
-
isbn-13 978-0-521-52221-2 paperback
-
isbn-10 0-521-52221-8 paperback
Cambridge University Press has no responsibility for the persistence or accuracy of

s for external or third-party internet websites referred to in this book, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
Contents
List of Tables and Figures page x

Preface xi
1 Introduction 1
1.1 Economic Systems 1
1.2 The Control Dimension 4
1.3 Looking for Clues 8
1.4 A Projected Synthesis 12
1.5 The Plan of the Book 18
2 Normative Perspectives 23
2.1 Why Care About Workers’ Control? 23
2.2 Equality 24
2.3 Democracy 27
2.4 Property 32
2.5 Dignity 34
2.6 Community 38
2.7 The Author Shows His Cards 41
3 Workers’ Control in Action (I) 45
3.1 Surveying the Terrain 45
3.2 The Plywood Cooperatives 50
3.3 The Mondragon Cooperatives 57
4 Workers’ Control in Action (II) 67
4.1 The Lega Cooperatives 67
4.2 Employee Stock Ownership Plans 76
4.3 Codetermination 83
5 Conceptual Foundations 92
5.1 The Theory of the Firm 92
vii
viii Contents
5.2 The Nature of Authority 98

5.3 The Locus of Control 101
5.4 Why Firms Cannot be Owned 107
5.5 Asset Ownership 110
5.6 Residual Claims 114
6 Explanatory Strategies 117
6.1 The Symmetry Principle 117
6.2 The Replication Principle 119
6.3 Transaction Costs 121
6.4 Optimal Contracting 126
6.5 Adverse Selection 129
6.6 Repeated Games 132
6.7 Historical Explanations 135
6.8 Cultural Explanations 138
6.9 The Strategy to be Pursued 140
7 A Question of Objectives 142
7.1 What do LMFs Maximize? 142
7.2 The Illyrian Firm 143
7.3 Membership Markets and Labor Markets 148
7.4 Membership Markets and Stock Markets 152
7.5 Imperfect Membership Markets 156
7.6 What does the Evidence Say? 161
7.7 Some Lessons 163
8 Views from Economic Theory (I) 165
8.1 Explaining the Rarity of Workers’ Control 165
8.2 Asset Ownership: Incentives and Information 166
8.3 Asset Ownership: Bargaining and Investment 168
8.4 Can Asset Specificity Explain the Rarity of Workers’
Control? 172
8.5 Work Incentives without Risk Aversion 173
8.6 Work Incentives with Risk Aversion 178
8.7 Can Work Incentives Explain the Rarity of Workers’
Control? 179
9 Views from Economic Theory (II) 185
9.1 Capital Constraints 185
9.2 Debt Financing 186
9.3 Equity Financing 188
9.4 Can Capital Constraints Explain the Rarity of Workers’
Control? 189
9.5 Portfolio Diversification 193
9.6 Can Portfolio Diversification Explain the Rarity of
Workers’ Control? 197
Contents ix
9.7 Collective Choice 200

9.8 Can Collective Choice Explain the Rarity of Workers’
Control? 204
10 Transitions and Clusters 207
10.1 Organizational Demography 207
10.2 Formation Rates 207
10.3 Worker Takeovers 212
10.4 Degeneration 221
10.5 Investor Takeovers 224
10.6 Survival Rates 226
10.7 Business Cycles 228
10.8 Clusters 230
11 Toward a Synthesis 234
11.1 The Causal Tapestry 234
11.2 Credible Commitment toward Labor 240
11.3 Credible Commitment toward Capital 245
11.4 The Composition of Control Groups 249
11.5 The Commodification of Control Rights 251
11.6 Intellectual History and Current Debates 253
11.7 Is Workers’ Control a Unitary Phenomenon? 256
12 Getting There from Here 260
12.1 Practical Considerations 260
12.2 A Modest Proposal 262
12.3 Reassuring Shareholders 273
12.4 Governing Firms 276
12.5 Trading Jobs 279
12.6 Sample Calculations 282
12.7 The Long and Winding Road 287
References 291
Index 313
List of Tables and Figures
Tables
1.1 Ownership and control as independent dimensions of
firm organization page 3
5.1 A classification of governance structures: Unbundled
input supply 111
5.2 A classification of governance structures: Bundled input
supply 113
12.1 Years needed to achieve workers’ control with dividends
reinvested in purchase of equity shares 284
12.2 Years needed to achieve workers’ control with dividends
paid out as cash 285
12.3 Steady-state employee share ownership with dividends
reinvested and equity claims sold to outsiders on departure 286
Figures
7.1 Income maximization in the Illyrian firm 145
x
Preface
Most large enterprises in the developed world are controlled by share-

holders, who choose the board of directors and can therefore replace the
firm’s top management. But in a few firms, control rests with the employ-
ees, who have an analogous right to choose the board of directors and
hire or fire top managers. Why is the first pattern so common and the
second so rare? Economists have had surprising difficulty in framing a
satisfactory answer to this seemingly straightforward question.
Because economists have no generally accepted explanation for the
prevalence of capitalist firms, they lack a good explanation for the persis-
tence of capitalism itself. They have also been obliged to remain largely
silent on various contemporary policy debates. Should employees be rep-
resented on a firm’s board of directors? Does it make sense to subsidize
stock purchases by employees? Should governments encourage worker
buyouts of closing plants or failing firms? Most economists could undoubt-
edly formulate opinions on these matters, but they would have trouble
locating a systematic body of theoretical or empirical research with which
to inform their opinions. With or without participation from economists,
however, policy issues of this sort have become increasingly prominent
across Europe and North America.
This book seeks to explain why investor-controlled firms are common
and worker-controlled firms are rare. The answer is not as simple as the
question, but one can develop a theoretical story that is logically co-
herent and accounts for a good deal of the empirical evidence. While
there is certainly room for debate, I believe that at least in broad out-
line a satisfactory explanation will need to look like the one suggested
here.
xi
xii Preface
Most of the book is devoted to the explanatory task, in part because it

is interesting and in part because it is hard to formulate policy recommen-
dations without first gaining an understanding of how the world works.
But the idea of workers’ control has stimulated a great deal of fervor and
hostility over the last two centuries, and it is impossible to disregard the
normative issues at stake. Accordingly, one chapter is devoted entirely
to normative ideas, and the concluding chapter proposes a strategy for
expanding the scope of workers’ control in modern economies. I am not
wedded to this particular strategy, but I believe that workers’ control is
currently undersupplied and that the theory developed in this book has
implications for what could be done about it.
The book is intended for a broad audience, including readers with
backgrounds in business, law, and public policy, as well as political science,
sociology, philosophy, and other academic disciplines. At the same time, I
have tried not to water down the theoretical discussion in a way that would
make it seem simplistic to economists. I do not back away from theoretical
concepts, and an economist will often be able to spot the models lurking in
the verbal underbrush. However, prior familiarity with economic theory
is not needed to grasp the key arguments, and there is almost no use of
mathematics. The book should thus be useful for graduate students and
advanced undergraduates in any of the social sciences.
Because the book offers a theoretical synthesis rather than just a lit-
erature review, I have not discussed every worthy piece of research on
workers’ control, which would have been impossible in any event. Some-
times I discuss a few groundbreaking articles on a topic but ignore later
extensions. In other cases, I cite surveys or summaries to avoid extensive
reviews of original research. Questions of emphasis have been resolved
by applying my own idiosyncratic preferences, and my knowledge of the
field remains highly incomplete. Apologies are due to all writers whose
valuable contributions have been overlooked.
I have accumulated many debts over the years, not all of which are
obvious from the citations in the text. At the risk of indulging in super-
fluous autobiography, the easiest way to acknowledge these debts is to
recount how I became interested in the topic. I first came across the idea
of workers’ control in 1974–75 as an undergraduate sociology major at
Amherst College, in a class on modern social theory taught by Norman
Birnbaum. At about the same time, I attended a lecture given by Sam
Bowles in which he emphasized the authoritarian nature of the capitalist
firm. Sam observed that when a boss wants a worker to go to the other
end of the factory, he doesn’t go there and hold up a five dollar bill; he
Preface xiii
just tells the worker to go. The relevance of this point will become clear
in Chapter 5.
While working on a master’s degree in public policy at the University
of Michigan, I began to explore the literature on workers’ control more
systematically. During this period, I discovered Katrina Berman’s 1967
book Worker-Owned Plywood Companies; Carole Pateman’s 1970 book
Participation and Democratic Theory; David Ellerman’s 1975 article in
Administration and Society; and Jaroslav Vanek’s 1975 anthology Self-
Management: Economic Liberation of Man. These authors persuaded me
that workers’ control was both morally attractive and viable in the real
world. But in the course of haranguing my fellow policy students about
the virtues of workplace democracy, I began to see some loose ends. For
example, if workers ran the firm, who would finance investment projects?
Capitalists?
For a while, I put these questions to one side and tried to learn enough
economics to get a doctorate from Michigan. Upon becoming an assistant
professor at Yale, my previous interests resurfaced, stimulated by some
reading in comparative economics that included Benjamin Ward’s famous
1958 article on the Illyrian firm, discussed in Chapter 7. In 1983, this
led to my first working paper on labor-managed firms and membership
markets.
The initial stirrings of the present book occurred in 1992 at the
Swedish Collegium for Advanced Study in the Social Sciences (SCASSS)
in Uppsala. Bo Gustafsson asked me to give a seminar directed mainly
to non-economists on the theory of the labor-managed firm. I decided to
survey the reasons economists had given for the rarity of such firms. A
few years later, Sam Bowles and Pranab Bardhan asked me to write up
a similar survey for their project on “The Costs of Inequality” funded by
the MacArthur Foundation. I dusted off the earlier SCASSS notes, called
in Louis Putterman for reinforcement, and together we wrote a long
working paper that became the foundation for Chapters 8 and 9. Other
versions of this material appeared in a 1999 volume, Employees and
Corporate Governance, edited by Margaret Blair and Mark Roe; in a 2000
article in the Journal of Economic Behavior and Organization; and in a
2001 article in the Review of Industrial Organization. Louis will not agree
with everything in Chapters 8 and 9 but he will undoubtedly recognize the
arguments.
The book as a whole was written during a year of study leave from
Simon Fraser University made possible by Nancy Olewiler and John
Pierce. Initial drafts of Chapters 1 and 5–9 were put together in October
xiv Preface
and November of 2000 at Erasmus University in Rotterdam. My visit to

Rotterdam was organized by Uskali Mäki of the Erasmus Institute for
Philosophy and Economics, John Groenewegen of the Faculty of Eco-
nomics, and Bart Nooteboom of the Faculty of Business. All three com-
mented on the material in these chapters while it was still in a highly
disorganized state. I am also grateful to the graduate students at EIPE
who suffered through my rough drafts and gave me an opportunity to re-
fine the presentation. Chapters 2, 10, and 11 were first assembled during
a visit to the School of Economics at the University of New South Wales
in April 2001. This enjoyable sojourn was arranged by Peter Robertson
with the support of John Piggott and Hazel Bateman.
Not everyone who deserves explicit thanks can be mentioned here,
but a few names stand out. Louis Putterman and Gil Skillman have often
enriched my understanding, and their friendship has been a pleasure. I
am also indebted to another co-author, Xiao-yuan Dong, for educating
me about China and emphasizing that workers’ control has a place on
the agenda of non-Western economies too. Sam Bowles and Herb Gintis
have frequently given me the benefit of their comments and criticisms,
and I have learned a good deal over the years from Avner Ben-Ner and
Bentley MacLeod.
Sam Bowles, Doug Kruse, John Pencavel, and Louis Putterman com-
mented on large parts of the manuscript, and Gil Skillman offered ex-
ceptional service as a sympathetic critic. Partly due to constraints of time
and space, I have not taken all of their advice, but I am grateful for it and
regret the flaws that went unrepaired. Three anonymous referees also
helped shape the book. Scott Parris of Cambridge University Press sup-
ported this project enthusiastically, and his e-mails were always a welcome
sight. At the production stage, Ronald Cohen carried out the copyediting
with a light and deft touch.
A group of hard-working journal editors deserves recognition: Mike
Montias, Joe Brada, and John Bonin at the Journal of Comparative Eco-
nomics and Richard Day at the Journal of Economic Behavior and Or-
ganization. All have waded through rough drafts of my journal articles
and tried their best to make sense out of them. My colleagues at Simon
Fraser have been generous and candid in commenting on my seminars and
working papers. Margaret Blair of the Brookings Institution provided fi-
nancial support and invited me to a series of conferences on employment
and the organization of the firm that helped focus my thinking. Much of
my research in this area has been supported by the Social Sciences and
Humanities Research Council of Canada. The colleagues and funders
Preface xv
listed here are absolved of all responsibility for the opinions expressed,
especially the foolish ones.
My greatest thanks go to Margaret Duncan for . . . well, everything.
But let’s just recall all those dinner conversations about dysfunctional
organizations. Theory is one thing and practice is another, but sometimes
the twain do meet.
The form of association, however, which if mankind continues to
improve, must be expected in the end to predominate, is not that which
can exist between a capitalist as chief, and work-people without a voice
in the management, but the association of the labourers themselves
on terms of equality, collectively owning the capital with which they
carry on their operations, and working under managers elected and
removable by themselves.
John Stuart Mill, Principles of Political Economy, Book IV,

Chapter VII, Section 6. First published in 1848; this passage was
added to the third edition in 1852. Reprinted in 1987 by Augustus
M. Kelley, Fairfield, NJ, 772–773.
For the women who ran the braiding machines,
the men who ran the tipping machines,
and especially the guy who worked in the wax room.
Governing the Firm
1
Introduction
1.1 Economic Systems

People care about the organization of firms. Their concerns include alien-
ation in the workplace and participation in decision-making, wage and
job security, the risks associated with employee stock ownership and
profit-sharing, and a host of other matters. Alienated workers are un-
likely to contribute their best efforts to the success of the firm. Em-
ployees may wonder whether the information they reveal now will be
used against them later, or whether authority delegated today will be
revoked tomorrow. Information technology may make worker knowl-
edge the most important input to the production process, but workers
who fear downsizing may be reluctant to invest in skills that would
only be valuable at their current job. Globalization expands export
markets but simultaneously poses risks to wages and jobs. Employee
stock ownership is often claimed to increase productivity, but employ-
ees may well hesitate to place their life savings in the hands of their own
employer.
Many of these dilemmas would be mitigated or eliminated if work-
ers had ultimate control over the firms to which they supply their labor.
Presumably if workers ran their own firms they would feel less alienated;
they would willingly disclose information that might improve efficiency;
they would be less worried about layoffs, downsizings, or plant closures;
and they would closely monitor how their savings were used by the firm.
Why not, then, adopt an economic system in which workers rather than
investors hold ultimate authority within firms? Why not implement a sys-
tem of workers’ control?
1
2 Introduction
A prudent worker, investor, or citizen would no doubt ask a few

questions before signing on to a program of this kind. Are any firms
controlled by workers today? What is their track record? How are they
organized? And perhaps the most important question: If workers’ con-
trol is so advantageous, why is it so rare? This book seeks to answer these
questions, especially the last one.
It may seem surprising that economists have not yet come to any con-
sensus on the reasons for the rarity of workers’ control. Despite much
attention within the profession to the organization of firms, the question
of why large firms are conventionally controlled by investors rather than
workers has not been high on the economic research agenda, perhaps
for the same reason that fish do not study water. It is also surprising that
economists have not been more interested in policy questions relating to
workers’ control: for instance, the subsidies widely granted for employee
stock ownership, or the support local governments sometimes provide to
employee buyouts of failing firms. By and large, the profession has been
content to view such measures as ad hoc political exercises that do not
require any novel theoretical or conceptual framework.
I will return to the economic literature on workers’ control later in
this chapter, but first a short historical digression may be useful in setting
the stage. The twentieth century witnessed a global struggle between two
alternative systems: capitalism and socialism. As normally defined, capi-
talism and socialism differ along the dimension of asset ownership. Under
capitalism, the physical assets used in production are owned by private
investors, while under socialism they are owned by the state. Soviet cen-
tral planning is, of course, discredited (Kornai, 2000), but socialism more
broadly remains identified with the public ownership of capital goods
(Stiglitz, 1994). Those in the West who favor this alternative have largely
embraced some form of market socialism: that is, a combination of public
asset ownership with reliance on market forces to handle most resource
allocation problems (Le Grand and Estrin, 1989; Bardhan and Roemer,
1993; Roemer, 1994).
Economic systems, however, vary along another dimension that is quite
distinct from the private or public ownership of physical assets: control
over production activities. Here the distinction is between firms controlled
by capital suppliers and those controlled by labor suppliers. The claim that
workers should control firms democratically differs sharply from the claim
that productive assets should be owned by the state. Nevertheless, writers
on the political left have sometimes advocated both ideas simultaneously,
perhaps believing that if one alternative to capitalism is good, then two
1.1 Economic Systems 3
Table 1.1 Ownership and control as independent dimensions of firm

organization
Asset ownership
Private Public

 Control by capital Capitalist firm Socialist firm
Control rights

Control by labor Laborist firm Self-managed firm
must be better. An unfortunate result is that the tensions between these

two modes of organization have often been glossed over.
Table 1.1 shows that four separate economic systems can be distin-
guished by treating the ownership of physical assets and control over
production as independent dimensions of firm organization. The terms
“capitalist firm” and “socialist firm” in Table 1.1 are standard. Although
differing in the dimension of asset ownership, both assign ultimate man-
agerial control over the firm to capital suppliers, located either in the
private sector or in the state. The “laborist” firm is a coinage meant to
suggest that assets are privately owned but firms are controlled by their
workforces rather than investors. The fourth cell can be identified with
the Yugoslav system of workers’ “self-management” that developed dur-
ing 1949–91, where the assets of firms were viewed as social property
(in practice, owned by the state) but workers within each enterprise had
substantial control over their use.
Like public ownership, the idea of workers’ control has a long pedigree,
motivating a diverse assortment of organizational experiments ever since
the Industrial Revolution. By contrast with public ownership, however,
there have been few attempts to impose workers’ control on entire soci-
eties by political fiat. The only major historical example along these lines
is that of the former Yugoslavia. It is difficult to draw lessons from this
experiment because the effects of workers’ control were confounded with
those of price regulation, poorly defined property rights, and political au-
thoritarianism. It is therefore impossible to use the Yugoslav experience
as a basis for predictions about the performance of workers’ control in a
setting of free markets, private ownership, and political democracy. Read-
ers interested in the Yugoslav case may want to examine Horvat (1971),
Estrin (1983), Sacks (1983), and Stephen (1984: ch. 5). A concise overview
of the entire 1949–91 period is provided by George (1993: ch. 3).
My working definition of capitalism is that it is an economic system
in which most large enterprises are under the ultimate control of
4 Introduction
capitalists – that is, individuals or groups who supply capital to firms.

The relevant alternative for this book is “laborism,” in which most large
firms are under the ultimate control of labor suppliers. In each case,
private asset ownership is taken for granted, so I will be concerned only
with the left column of Table 1.1. The term “socialism” will be discarded
because of its association with public ownership, which is irrelevant to
the question at hand.
It is vital to stress at the outset that private ownership of productive
assets does not necessarily imply capitalism in the sense that firms are
actively managed by the suppliers of capital (or their agents). In princi-
ple, worker-controlled firms can lease physical assets from their owners
on a competitive market without ceding control over the production pro-
cess, just as conventional firms routinely lease computers, offices, aircraft,
machinery, and land. There is nothing about private asset ownership that
implies routine control over the ways in which assets are used, any more
than owning an apartment building implies that a landlord can supervise
the daily lives of tenants. There are of course practical reasons why various
assets are seldom leased, and these will be explored later. But in theory,
workers could hire an asset from its owner, rather than being hired by the
asset owner. Therefore, ownership does not imply control.
If it is not feasible to lease physical assets, a worker-controlled firm
could finance collective asset ownership using debt contracts under which
workers retained control over the firm’s activities except in extreme cir-
cumstances such as default on contractual obligations. Workers could also
draw upon their personal savings to finance collective asset ownership by
the firm, in which case they would become suppliers of both capital and
labor. Such a firm qualifies as “worker-controlled” or “labor-managed” as
long as votes are proportional to labor supply rather than capital supply.
The fact that capitalists frequently do exercise control over the activ-
ities of workers is not a logical necessity but an empirical fact requiring
an explanation. A system of private asset ownership and decentralized
markets need not have led to capitalist firms. However, despite occasional
successes, worker-controlled firms occupy only a marginal position in the
economies of North America and Western Europe. This presents a puzzle:
Why does workers’ control remain rare? Or, in another formulation: Why
does capitalism persist?
1.2 The Control Dimension

Modern economic theory suggests that a firm should be seen as a set of in-
complete contracts among input suppliers (Williamson, 1985; Grossman
and Hart, 1986; Milgrom and Roberts, 1990a). In this view, the right to
make decisions not previously determined by contracts must be assigned
to some person or group. Such control rights could be used to deter-
mine a firm’s product line, investment strategy, wages and employment,
production methods, or working conditions, for example. In most large
enterprises, these decisions are made by managers who are ultimately
hired and fired by a board of directors. The directors in turn are chosen
by shareholders.
This is the archetypal capitalist firm. In its “laborist” counterpart, man-
agers may also be hired and fired by a board of directors, but the board
is chosen by, and is accountable to, the workforce. Firms of both types
must somehow induce managers to pursue the goals promulgated by the
board of directors, a problem that will become relevant on occasion in
later chapters. But the key question here is why ultimate control over the
firm normally rests with investors (or their representatives) rather than
with workers (or their representatives).
Because a variety of ambiguities can arise in practice, including the
possibility that workers may supply both labor and capital to some firms,
it is necessary to refine the rough descriptions given above. A capital-
managed firm (KMF) is defined as an enterprise in which ultimate con-
trol is allocated by virtue of, and in proportion to, capital supply, while a
labor-managed firm (LMF) assigns control by virtue of, and in proportion
to, labor supply. Other complications – for example, involving codetermi-
nation and collective bargaining – are addressed in Chapter 5, but these
preliminary definitions will suffice for the moment.
The reference to “ultimate” control is intended to bypass the prob-
lem of managerial incentives. Even if the shareholders can agree among
themselves on the objectives of the firm, their control over the board of
directors is limited, and directors likewise have limited control over top
managers. This is the famous “separation of ownership and control” first
highlighted by Berle and Means (1932), which is not to be confused with
the very different distinction between ownership and control drawn in
Table 1.1. Despite these incentive problems, however, it is nevertheless
significant that the board of directors can dismiss the top management of
the firm and that shareholders can replace the directors.
It could be objected that a focus on ultimate control is misplaced be-
cause workers can influence the decisions of their organizational superiors
even in capitalist firms. For example, employees might exercise influence
through committees dealing with innovation, product quality, working
conditions, or grievances. More covertly, they might exert social pressure
on managers, give them false information, or bribe them (Milgrom and
6 Introduction
Roberts, 1990a). It may therefore be misleading to say that such firms are
controlled by investors.
I ignore this objection for the following reasons. In an investor-
controlled firm, the authority delegated to employee teams or committees
can always be revoked. This is not to say that such participative processes
are a fraud. They play an important role in conveying information to man-
agers, motivating workers, accelerating responses to market conditions,
and taking some of the load off vertical communication channels. But to
grasp the deeper structure of the firm, it is illuminating to ask who has
ultimate authority, meaning authority that cannot be revoked by anyone
else. It may be quite costly or difficult in the short run to revoke delegated
authority – and to this extent workers do have some de facto control –
but in the long run, investors generally have the upper hand. In capitalist
firms, the shareholders directly or indirectly determine whether other in-
dividuals, such as managers and shop floor employees, will continue their
association with the firm. Employees often have a good deal of informal
influence, but at the end of the day they cannot fire their bosses.
Another potential objection is to deny that the large corporation is a
capitalist firm, or capital-managed, by arguing that most shareholders are
not really capital suppliers. The point is that when one person sells existing
shares to a second person, the firm itself does not gain access to any new
capital, and so (it may be argued) the new shareholder does not supply any.
This argument is mistaken because the current shareholders as a group
can liquidate the firm’s assets, though more than a simple majority vote is
generally required. If they do so, they will share in whatever funds are left
after various creditors (bondholders, suppliers, tax authorities) have been
paid. By refraining from liquidation, the shareholders keep part of their
wealth tied up in the firm and available for its use. As individual shares
are traded, each shareholder inherits the voting rights of the previous
shareholder, gains an identical claim on dividends and on the firm’s non-
human assets in the event of liquidation, and thus occupies a structurally
identical position within the firm. It therefore makes sense to say that one
capital supplier replaces another.
More generally, any firm whose complexity extends beyond sole pro-
prietorship has some governance structure that defines the organizational
roles assigned to suppliers of particular inputs. Whenever there is turnover
among input suppliers, the rights and duties associated with such roles
must be transferred from one person to another according to the pro-
cedures specified by the governance structure. In professional partner-
ships, all partners must usually agree before a partner can be replaced.
Some worker-controlled firms permit members to sell their positions to

replacement workers with the approval of other members. Shareholders
in a publicly traded firm, however, are uniquely free to sell their roles in
the firm without first obtaining permission from any other parties. A key
theoretical question is why it appears to be easy to create markets for the
position of “capital supplier” but more difficult to create parallel markets
for the position of “labor supplier.”
Where do control rights come from? They may arise in two general
ways: through contracts, laws, or constitutions on the one hand; or through
social customs, conventions, or norms on the other. The first case is the
most familiar. In modern societies, a group of input suppliers can typi-
cally draw up a partnership agreement or a corporate charter that says
who will have the authority to make managerial decisions. This agree-
ment or charter becomes the ultimate source of authority within the firm,
and the resulting authority may be delegated to other parties such as
managers. Sometimes a specific assignment of control rights to capital or
labor suppliers is fixed by law (as in German codetermination, discussed
in Section 4.3) or by a constitution (as in the former Yugoslavia), but
there is usually considerable scope for voluntary contracting. In this set-
ting, whether property rights over non-human assets lead to control rights
over the firm as a whole depends on the nature of the contract.
A less standard way of thinking about control rights, but one that should
not be dismissed a priori, involves social custom or non-cooperative equi-
librium in the absence of a supporting legal framework. As a thought
experiment, imagine that there are a number of locations at which pro-
duction can occur, and that the first person to reach a production site
can exclude others who arrive later, perhaps because of social convention
or through physical force. If others are admitted to the site, they cannot
subsequently be excluded against their will. In this environment, the first
arriver might charge a price for admission, with anyone paying the price
being allowed in. After a group of input suppliers has been assembled
in this way, it is necessary to make coordinated decisions about produc-
tion activities. At this stage, each person could threaten to leave, taking
their resources with them, unless the joint decisions are satisfactory. This
would give each some bargaining power over the group’s decisions, and
thus a measure of control. Other input suppliers who remained vulnera-
ble to expulsion and were easily replaced would of course lack parallel
influence over production activities. If the people with bargaining power
all happen to be capital suppliers, it would be reasonable to speak of a
capital-managed firm, and similarly if they are all labor suppliers.
8 Introduction
The point of this parable is conceptual. Control rights need not derive
solely from legally recognized property rights or contractual agreements.
I will not be concerned here with firm organization in an anthropolo-
gist’s tribe or an anarchist’s utopia, but clearly one could discuss control
rights over firms in such situations without lapsing into incoherence. More
significantly, one should not underestimate the role of social custom or
convention as a widely accepted source of authority, even in contempo-
rary societies. Contracts exist, but so do stable patterns of behavior when
the courts are not watching. It often makes sense to accept the boss’s
authority simply because it is correctly expected that everyone else will
do the same. Moreover, a boss often has ways of enforcing compliance
without resorting to a lawsuit: for instance, by threatening an employee
with dismissal.
The term “control rights” as used in this book does not necessarily refer
to legal or contractual rights, and never refers to moral rights. The meaning
is causal, not normative. References to “control rights” can always be
translated as “control abilities, or capacities, or powers.” To say that a
group of people has control rights in a given firm means (roughly) that
they can directly or indirectly allocate the resources of input suppliers
located inside the firm, distribute rewards or penalties among them, and
perhaps terminate their relationships with the firm. These powers often
have a formal contractual basis, but this is not essential. I will return to
these issues at greater length in Chapter 5.
1.3 Looking for Clues

A naive economics student might expect the discipline to have developed
a cogent and empirically supported explanation for the conventional as-
signment of control rights to investors rather than workers. After all, this
problem is one of the most basic that could be addressed by economists,
with implications for labor economics, comparative economics, industrial
organization, finance, and economic history, among other fields. Such a
student would be disappointed. Remarkably, economists still lack a com-
monly accepted rationale for the prevalence of capitalist firms in market
economies.
As will become painfully clear, hypotheses about the rarity of workers’
control are in excess supply. A small band of empirical researchers has
labored to redress the balance, but abstract modeling has outpaced the
evidence. Much theoretical discussion leans toward casual storytelling
rather than thoughtful analysis informed by factual knowledge. Writers
occasionally announce that they have discovered why LMFs are rare with-
out showing any awareness that other hypotheses exist, or that accepted
evidence calls their own story into question. These difficulties are exac-
erbated by the ideological passions the subject tends to provoke. Some
writers are convinced of the virtues of workers’ control, and view its rarity
as evidence for a massive market failure. Others, convinced of the mar-
ket’s wisdom, infer from the rarity of workers’ control that it lacks all
merit and barely warrants discussion.
One unfortunate consequence of this disarray is that economists are
left with little to say about a wide range of current policy issues. Numerous
academicians and practitioners have called for increased worker partic-
ipation in the management of firms, including board representation for
employees (Blair, 1995; Levine, 1995). Many firms have experimented
with profit-sharing plans, employee share ownership, and more employee
participation in decision making (Kruse, 1993; Blair, Kruse, and Blasi,
2000; Ben-Ner, Burns, Dow, and Putterman, 2000). A few large firms in
the United States have been completely taken over by their employees
(see Chapters 4 and 10), often with the aid of tax incentives or loan guar-
antees made available by various levels of government.
These developments raise provocative questions. Is there a process of
institutional evolution underway that is leading systematically to greater
workers’ control within the modern firm? If so, what forces are driving
this process? What are the key barriers to the further spread of workers’
control? Is there any rationale for policy intervention to reduce or elimi-
nate these barriers? Is it possible to combine the advantages of investors’
control in raising capital and diversifying risks with the advantages of
workers’ control in stimulating effort, reducing conflict, and tapping em-
ployee knowledge? Do the answers depend on the characteristics of the
industry, the nature of production technology, or the size of the firm?
I do not pretend to supply complete answers to these questions, but
I hope that after finishing this book, the reader will find some answers
more reasonable than others. Many pages will be devoted to historical,
case-study, and econometric evidence on the subject, as well as critiques
of received theoretical wisdom, because I believe there is a high marginal
benefit from sifting through existing data and arguments in a system-
atic way. Readers who seek a decisive mathematical proof or a definitive
regression equation will not find it here, but they will find a gradual accu-
mulation of evidence pointing toward specific conclusions.
Consider an example of what needs to be explained. A fact that
leaps out from historical and case study accounts is that traditional
10 Introduction
labor-managed firms are not distributed randomly across industries.

They occur mainly in professional services (law, accounting), craft manu-
facturing (printing, leather goods, glassware, furniture), low-skill service
tasks (reforestation, refuse collection, taxi service), and construction.
As a general rule, LMFs do not engage in large-scale, capital-intensive
production activities. There are exceptions, but this generalization is
quite robust for the subset of LMFs that have been formed directly
as producers’ cooperatives rather than arising through conversion of a
pre-existing KMF. Further facts about the distribution of LMFs across
industries, conversions of KMFs into LMFs (or conversely), and the
organizational designs of successful LMFs will surface later.
An ideal theory of control rights would accomplish the following things:
(1) explain why LMFs account for a small share of aggregate output, as-
sets, and employment relative to KMFs; (2) explain the observed distribu-
tion of LMFs across industries; (3) explain the organizational transitions
in which a firm changes from KMF to LMF through an employee takeover,
or from LMF to KMF through an investor takeover; and (4) explain the
design features that successful LMFs typically have. No theoretical frame-
work comes remotely close to satisfying all of these requirements. Instead,
there is a consistent tendency among theorists to focus on one fact – the
rarity of LMFs – and to emphasize one causal mechanism that allegedly
explains this fact.
To see how one might begin to construct a more complete theory of
the LMF from the ground up, consider momentarily a world in which
there are no qualitative distinctions between inputs of capital and labor,
in either physical or institutional respects. This is the level of abstraction
found in microeconomic theory textbooks, for example. For reasons to
be discussed in Chapters 6 and 7, in such a world there is no reason why
suppliers of one input or the other would more often acquire control
over firms. The distribution of KMFs and LMFs across industries would
be random, or a matter of historical accident, and thus KMFs would be
unlikely to dominate the economy as a whole.
Explanations for systematic differences between KMFs and LMFs
must therefore be grounded on qualitative asymmetries between capi-
tal and labor as inputs. Perhaps the most fundamental asymmetry of this
sort is that ownership of physical assets can be shifted from one person to
another, while the capacity to supply labor services cannot be. In short,
the capacity to supply labor is inalienable, both in a physical sense and
also institutionally (contracts involving slavery or voluntarily accepted
servitude are unenforceable in court).
This has many implications. A firm can obtain its capital inputs either
as an owned stock or as a leased flow, but it can only obtain labor as a
flow. A worker’s time and skill cannot exceed natural bounds, but there
is no upper limit on an investor’s wealth. A labor supplier must often
be in close proximity to other labor suppliers to join in production, and
cannot be in more than one place at a time, but one person can own many
physical assets in dispersed locations. Because labor inputs depend on
the characteristics of the person supplying them, they are often highly
heterogeneous, while financial capital is not.
Although these are suggestive observations, simply knowing that cap-
ital and labor differ, or even how they differ, is not enough. Another
portion of the theory must address what the controllers of firms do. What
sorts of decisions do they make? What can one say about the preferences,
beliefs, and constraints of individual controllers? Do controllers face im-
portant collective action problems? How does an input supplier become a
member of the controlling group or exit from this group? What does this
imply about the behavior of the firm or the likely performance of KMFs
and LMFs in specific environments? Juxtaposing information about the
asymmetries of capital and labor inputs with information about the role of
control groups within firms provides the basis for an explanatory strategy.
Later I will outline the nature of this strategy in greater detail.
Nothing thus far implies that the relative economic efficiency of KMFs
and LMFs must play a central explanatory role. Efficiency arguments
clearly become relevant if one’s theory says that surviving organizational
structures will necessarily satisfy some efficiency criterion. But someone
else’s theory might be based on subtle forms of market failure, or game
theoretic models in which multiple equilibria arise. Theories of this sort
may offer explanations for the observed distribution of LMFs while assert-
ing that everyone could be made better off through policies to encourage
the creation of more LMFs. Researchers will naturally pursue diverse ex-
planatory strategies (see Chapter 6), but perhaps all can agree that no
theory should receive extra points for assuming the efficiency of market
outcomes, or for assuming the contrary. A theory should be awarded a
great many points, however, if it successfully explains the incidence, be-
havior, and design of LMFs.
An argument occasionally advanced against the usefulness of studying
LMFs is that in a competitive economy, surviving firms must necessarily
maximize profit. Firms that fail to satisfy this criterion, it is said, will be re-
placed by those that satisfy it, and thus if LMFs are to be viable at all,
they must behave in a profit-maximizing way. However, for this to be
12 Introduction
true, certain conditions must be met. First, markets must be complete (in
the sense that all relevant features of a firm’s inputs and outputs can be
specified in binding contracts) and competitive (in the sense that perfect
substitutes for any input or output can be obtained at an identical price).
Second, and relatedly, economic profit must be the sole survival test.
Because LMFs exist, and have been shown empirically to deviate in var-
ious ways from the behavior of comparable KMFs (see Chapter 7), these
assumptions can safely be rejected.
Many economists see worker-controlled firms as a fringe phenomenon
having little or no significance for the economy as a whole, and deserving a
corresponding amount of research attention. This attitude is understand-
able but mistaken. Even if one lacks interest in any normative project to
advance the cause of workers’ control, surely it is important to explain
why firm governance in private ownership economies is predominantly
capitalist in form. This cannot be done without considering what it means
for a firm to have a capitalist structure and identifying the main alterna-
tive ways in which firms could be organized. The contrast with worker-
controlled firms is surely an obvious one. Moreover, such firms are not
hypothetical entities. They exist, they have often survived in competitive
environments for long periods of time, and a large amount of information
is available about them. The proof of the pudding is in the eating, but I
believe that a systematic attempt to understand the successes and failures
of workers’ control can shed light on capitalism itself, including the rea-
sons for its broad success and its continuing limitations. As a bonus, the
reader may gain an illuminating vantage point from which to ponder the
economic theory of the firm.
1.4 A Projected Synthesis

I will not undertake a full synthesis of theoretical ideas about LMFs until
Chapter 11, but it would be unfair, and unrealistic, to expect readers to
wait that long without a glimpse of the larger picture. This section provides
a preliminary sketch. Many causal linkages and qualifications are omitted
at this stage for expositional convenience.
Firms will be viewed as coalitions of input suppliers characterized by
incomplete or missing contracts, so control over production activities must
be assigned to some person or group. The only candidates for control
are capital and labor suppliers. I am not concerned in this book with
firms controlled by consumers or suppliers of raw materials, for example.
I assume that within limits, firm controllers can enforce their decisions
by means of binding contracts or some type of self-enforcement. The

latter could involve threats of retaliation in situations of non-compliance,
perhaps through wage reductions or dismissal.
Firms can be distinguished from markets by the absence of short-term
bargaining, among other things, so individual agents do not normally hag-
gle or attempt to reprice their entire relationship with the firm whenever
controllers issue new instructions. Obstacles to short-term bargaining in-
clude the time costs involved, private information, collective-action prob-
lems, and the inability of controllers to make credible promises about their
own future behavior. Whatever the reason, the lack of such bargaining
makes it difficult or impossible for the people affected by a control group’s
decisions to alter those decisions, although they can exit to the market
if sufficiently dissatisfied. Individuals or groups who enjoy control rights
can therefore impose uncompensated costs or benefits on other coali-
tion members. One implication is that the Coase Theorem (Coase, 1960),
which says that bargaining will allocate resources efficiently if transaction
costs are zero, does not apply within the firm.
Controllers take account of their own interests directly, but at best will
internalize the interests of other input suppliers in a partial and indirect
manner. The danger associated with centralized authority in firms, as else-
where, is that it can be abused in a self-interested way through decisions
that inflict costs on other agents (Dow, 1987). If the investor-controllers in
a KMF can discover a way to increase their payoffs while imposing costs
on employees, they will take advantage of this opportunity. Likewise, the
worker-controllers in an LMF will abuse outside investors if they find this
advantageous.
Some of the most significant forms of abuse have an intertemporal
dimension. It is frequently difficult for a control group to credibly reassure
non-controllers that they will be rewarded later for sacrifices made today.
This is not always impossible: If the parties place enough weight on future
payoffs, promises of this kind may become credible (see Section 6.6).
But agents do discount the future, and reputation is often an imperfect
safeguard. In some settings, the temptation to abuse non-controllers may
be too great to resist. Problems of intertemporal credibility are therefore
a prime hunting ground in looking for behavioral asymmetries between
KMFs and LMFs.
Consider a stylized fact about LMFs. When created de novo, rather
than through an employee buyout of a KMF, they rely almost exclusively
on financing from the personal savings of members and the retained earn-
ings of the firm itself. Loans from banks are rare unless there are easily
14 Introduction
marketed structures or inventories that can be used as collateral. Non-

voting equity shares are almost never used as a way of attracting capital or
transferring risk. A reasonable conjecture is that outside investors fear the
consequences of putting funds into an enterprise over which they have no
direct control (Williamson, 1985: ch. 12; Putterman, 1993). Other things
being equal, this predicts that LMFs will be rare when (1) workers are
poor; (2) the industry is capital intensive; (3) physical assets cannot easily
be leased; (4) the required assets are too specialized to serve as collateral
on loans; and (5) the firm is unlikely to engage in repeated transactions
with investors.
KMFs may have similar trouble making credible commitments to their
employees. Workers frequently make large investments whose value de-
pends on the future behavior of their employer. Such investments in-
clude decisions to locate in a specific neighborhood or to acquire certain
skills. Investments of this sort imply that workers will receive quasi-rents,
which are the difference between actual wages and the worker’s next best
alternative on the outside labor market. These quasi-rents are at risk in a
number of ways: The firm may withhold promised salary increases, it may
innovate in ways that make current skills obsolete, or it may shut down its
local office and shift operations elsewhere. In each case, there is a danger
that the investors who control the KMF will ignore costs to workers in
the form of lost wages or unemployment.
Another kind of commitment problem involves the incentives of
employees to supply information. It is often observed that workers are
reluctant to reveal their true abilities or the productivities of their jobs
to an employer. The reason is that employers can exploit this knowledge
to make incentive systems more stringent, perhaps by reducing piece
rates or raising output quotas, a phenomenon called the ratchet effect
(Weitzman, 1980; Dearden, Ickes and Samuelson, 1990; Allen and Lueck,
1999; Carmichael and MacLeod, 2000). Various distortions can occur as
a result, including low effort levels and opposition to innovation. Al-
ternatively, the employer may construct an incentive scheme that induces
efficient effort levels but only by giving rents to productive workers, which
is costly from the employer’s standpoint.
The LMF has advantages in solving both problems. The worker-
controllers of the firm internalize the quasi-rents they derive from spe-
cialized investments in skill or location, and take these quasi-rents into
account in making shutdown decisions. The ratchet effect is more com-
plicated, but there are reasons to believe that LMFs are less likely to suf-
fer from such difficulties. Workers are typically well informed about the
efforts and abilities of their co-workers, and the informational asymme-

tries involved are therefore muted. But perhaps more important, man-
agers accountable to their subordinates are unlikely to make incentive
systems more stringent if this is not what the firm’s controllers want them
to do. The LMF therefore solves the problem by providing a credible com-
mitment to workers that any information they reveal about productivity
will not be used against them later.
As in the case of incentives for LMFs to abuse outside investors, the
importance of these issues will vary across industries. Other things equal,
LMFs can be expected to have credibility advantages when (1) workers
make large investments in firm specific skills or location; (2) workers have
substantial private information that is valuable for managerial decisions;
and (3) there are no good alternative ways to reassure workers that their
interests will be protected, such as collective bargaining or the reputation
of the firm.
Thus far I have focused on the difficulties that controllers face in mak-
ing credible commitments to non-controllers (one form of the incentive
problem known to economists as “moral hazard”). A different problem
arises from the fact that individuals may have private information that can-
not be conveyed to others in a credible way (resulting in a somewhat differ-
ent problem, “adverse selection”). Although by itself this does not imply
any particular bias toward KMFs or LMFs, it is relevant that controllers
sometimes enjoy better access to information than non-controllers. Most
obviously, a controlling group can threaten to fire managers who are dis-
covered to have omitted or distorted crucial information.
Thus, in KMFs, investors are likely to know more than workers about
the firm’s technological options or product-market environment, while
in LMFs, workers are likely to have informational advantages over in-
vestors. The differing location of the informational problem can imply
that KMFs and LMFs are asymmetrical in their behavior or perfor-
mance. For example, a KMF might have difficulty convincing its work-
ers to accept pay cuts even when product demand is low, because em-
ployees distrust unsupported claims of financial distress. Conversely, an
LMF might have difficulty in convincing investors to supply new financ-
ing even when product demand is high, because investors distrust rosy
scenarios.
Another set of issues involves the composition or internal structure
of each potential control group. The number of investors in a firm is
not necessarily the same as the number of workers, and these groups
may also differ systematically in the extent to which members display
16 Introduction
diversity in preferences or beliefs. Group size is important because control

groups may encounter free riding, bargaining, or other collective action
problems when exercising their authority. Heterogeneity of preferences
or beliefs can make it hard for a control group to function as a cohesive unit
(Hansmann, 1996) and may leave the firm open to a takeover (Skillman
and Dow, 2002, unpublished manuscript).
The characteristics of capital and labor help to determine the size and
composition of each potential control group. There is considerable flexi-
bility about the size of the control group in a KMF. Such firms range from
closely held firms with a single founder-owner to publicly traded firms
with millions of shareholders. Tradeoffs clearly do arise: For instance,
widely held firms have better access to capital while suffering from more
severe free riding among shareholders with respect to the task of mon-
itoring managers. But in general, KMFs have some degree of freedom
with respect to the number of investor-controllers.
LMFs, on the other hand, must have a control group that is approxi-
mately the same size as the firm’s workforce. Since human capital is in-
alienable and cannot be accumulated without limit by single individuals,
larger firms necessarily have more controllers. This does not imply that
LMFs are automatically disadvantaged relative to KMFs because other
things may not be equal. For example, it is often suggested that workers
gain information about the performance of managers as a byproduct of
doing their jobs, which alleviates the free-rider problems associated with
monitoring.
Keeping size fixed, there is also the question of heterogeneity. As will
be seen in Chapter 9, one popular hypothesis about the rarity of workers’
control asserts that workers are usually more diverse in their preferences
toward firm policies than investors. The claim is that investors agree unan-
imously, or nearly so, on the maximization of profit or stock market value,
while workers frequently disagree over the relative importance of safety,
job security, wages, convenient scheduling, or a congenial social atmo-
sphere, for example.
Whether this argument makes sense depends to a large degree on
what one assumes about the structure of capital and labor markets. The
preferences of the individual investors or workers who supply inputs to
a given firm vary with the opportunities these agents have had to sort
themselves among firms or to diversify their input contributions across
multiple firms. Capital and labor clearly differ in relevant ways. Financial
capital is highly mobile, whereas labor is much less so; financial capital
is easily diversified across firms, while labor is not; and so on. One can
therefore construct a theory about differences in the behavior of KMFs

and LMFs that relies on the causal channel of preference heterogeneity.
A final causal mechanism also involves the nature of capital and labor
markets. I will show in Chapter 7 that the behavior of KMFs and LMFs
depends on whether there are markets for control positions in each type
of firm. For various reasons, it is more difficult to establish a market for
membership in an LMF than a market for shares in a KMF. The empirical
evidence indicates that LMF membership markets are imperfect when
they exist at all, and various consequences follow from this. I postpone
the details until Chapters 7 and 10, but in this case, LMFs may lack strong
investment incentives and may “degenerate” into capitalist firms if they
are financially successful.
The difficulty of commodifying control positions in LMFs implies that
firms of this kind are less likely to be created in the first place, because
entrepreneurs can more readily appropriate the rents from technological
or organizational innovation by setting up KMFs. Similarly, it is difficult
for anyone to privately appropriate the potential social benefits from
transforming a KMF into an LMF. To do so, buyout organizers would
need to know what value employees place on control rights and then
insist that they pay corresponding prices for continued membership in the
transformed firm. This is generally impossible, and thus employee buyouts
are likely to involve free-rider problems in which the benefits are widely
diffused among workers while the transition costs are concentrated on a
few individuals. I will argue later that for these reasons, workers’ control is
an undersupplied organizational form. Some policy proposals to address
this problem will be developed in Chapter 12.
The discussion to this point can be summarized as follows. The key
causal factors are (1) commitment problems facing the members of con-
trol groups; (2) the composition of control groups as defined by their size
and heterogeneity; and (3) the difficulties involved in commodifying con-
trol positions. I suggest in Chapter 11 that asymmetries between KMFs
and LMFs in all three respects can be traced to the fact that capital is alien-
able, while labor is not. Taken together, these causal factors can account
for the general rarity of LMFs, much of the evidence regarding their dis-
tribution across industries, some of what we know about organizational
transitions, and some of what we know about LMF design.
This theoretical framework is not simple, because the inalienability of
labor has multiple and diverse ramifications. The literature accordingly
features many monocausal stories emphasizing a single factor believed by
an author to be especially important – for example, financial constraints
18 Introduction
or collective-choice problems. Even granting that some of these stories

have a common foundation in labor’s inalienability, it remains of inter-
est to assess which of the causal channels are most significant. This is a
challenging research agenda because much of the evidence is anecdotal.
Moreover, the small set of econometric studies available suffer from mea-
surement problems, selection biases, and questions about the direction of
causality.
A deeper problem is that workers’ control may not be a unitary phe-
nomenon. It is not obvious, for example, that the obstacles to de novo
creation of an LMF are the same as those facing employees who want to
buy out a large conventional firm. These two kinds of LMFs differ in sys-
tematic ways: Traditional producers’ cooperatives are often small, labor
intensive, and have little division of labor, while capitalist firms bought out
by employees are often large, more capital intensive, and have a substan-
tial division of labor. At best, a theory accommodating both will explain
why the factors shaping LMFs operate in different ways in these two
settings. At worst, it may turn out that no theoretical unification can be
achieved because there is little or no overlap among the causal factors
responsible for the two phenomena (Mäki, 2001). I provisionally adopt
the more optimistic view.
In Chapter 6, I will explore the possibility that legal, historical, or cul-
tural factors might account for the rarity of LMFs. I do not entirely reject
hypotheses of this sort, but I criticize them for their vagueness and fail-
ure to grapple with empirical evidence on LMFs. Relatedly, some LMFs
were evidently the pet projects of unusually intelligent, charismatic, or al-
truistic people. Economists will probably not have much luck explaining
idiosyncratic personal elements in the formation of individual firms, but
may do better in explaining why certain organizational structures thrive
in a given market environment. To use a biological analogy, we may know
little about what causes a particular organizational mutation, but we may
know a great deal about the selection pressures operating on it and the
prospects for its long-term survival.
1.5 The Plan of the Book

Chapter 2 is devoted to normative perspectives on workers’ control. Al-
though this book focuses mainly on the explanatory task, much of the
literature on the subject has been driven by normative concerns. Ignor-
ing these concerns would be a disservice both to the reader and to the
literature. Apart from the argument of economic efficiency, advocates
of workers’ control have advanced many moral arguments on its behalf.

These include egalitarian distribution, democratic theory, just appropria-
tion, and concepts of dignity and community. I will argue that LMFs are
unlikely to serve as effective instruments for income or wealth redistri-
bution, although they will not aggravate existing inequalities, and may
have some modest equalizing effects. Arguments from democratic polit-
ical theory are plausible, but raise the question of whether democratic
rights should go to all workers or only those with high costs of leaving the
firm. There is also controversy over the question of whether democratic
rights are inalienable, or can be given up freely by those workers who are
prepared to cede control to an employer in order to achieve higher in-
comes. I am not persuaded by another argument, that firms must respect
a “labor theory of property,” partly because the conclusions drawn from
this theory are sensitive to seemingly innocuous aspects of contractual
relationships. More convincing is the idea that LMFs promote values of
dignity and community.
Chapters 3 and 4 provide the empirical foundation for much of the
rest of the book. I briefly sketch the historical emergence and present
status of workers’ cooperatives in North America and Western Europe,
and then present several case studies of workers’ control. These include
the plywood cooperatives in the U.S. Northwest, the Mondragon system
in Spain, and the Lega federation in Italy. In each case, I describe the
history of the cooperatives, their governance structures, and the ways in
which they have attracted capital and labor inputs. Chapter 4 goes on to
examine employee stock ownership plans (ESOPs) in the United States
and codetermination in Germany.
The analytical portion of the book begins with Chapter 5, which lays
the groundwork by defining what is meant by a firm and a governance
structure. I will define a firm to be a coalition of input suppliers whose
production activities are coordinated by means of a common authority
structure. Under this definition, it is impossible to own a firm, though one
can own non-human assets used in production. Chapter 5 also clarifies
the definitions of KMFs and LMFs, the nature of asset ownership, and
the meaning of residual claims.
The methodological discussion in Chapter 6 starts by proposing two
heuristic rules, the symmetry principle and the replication principle, which
prove helpful in sifting through hypotheses about LMFs. The first holds
that asymmetries between KMFs and LMFs must be traced back to asym-
metries in the characteristics of capital and labor inputs. The second
holds that an explanation is incomplete unless it says why any alleged
20 Introduction
advantages of KMFs cannot be replicated within LMFs, and conversely.

Thus a claim about asymmetries in firm behavior or performance should
invoke the identity of the firm’s ultimate controllers and their inherent at-
tributes, rather than stopping with more superficial and malleable aspects
of the firm. One should also take care not to build avoidable inefficiencies
or market failures into the definition of the LMF itself.
The second task in Chapter 6 is to survey the range of explanatory
strategies that can be used to account for empirical observations about
LMF design, incidence, behavior, and performance. These include ap-
proaches directly postulating efficient market outcomes (transaction cost
economics and optimal contracting), as well as approaches that assume
individual optimization and some form of market equilibrium but do not
assume efficiency (adverse selection and signaling, along with repeated
game models). Later sections of Chapter 6 deal with historical and cultural
explanations for the rarity of LMFs.
Chapter 7 reviews earlier work from comparative economics in which
the LMF was taken to maximize net income per worker. I reject this
“Illyrian” hypothesis for its logical shortcomings and its inconsistency
with the evidence on actual LMF behavior. The rest of Chapter 7 shows
that at least in principle, competitive markets for LMF membership are
perfect substitutes for labor markets and stock markets in the sense that
they induce LMFs to choose production and investment plans that max-
imize profit or present value. To the extent that real LMFs deviate from
these goals, the reasons must be related to missing or imperfect member-
ship markets. Sources of imperfection in membership markets are then
discussed, together with the resulting asymmetries between KMFs and
LMFs. Empirical evidence bearing on this topic is also reviewed.
Chapters 8 and 9 survey a range of hypotheses about workers’ control
taken from contemporary economic theory. I group these into several
categories – namely, hypotheses about asset ownership, work incentives,
capital constraints, portfolio diversification, and collective choice. Within
each group, I review the nature of the explanation offered for the rarity
of workers’ control or its distribution across industries, give a theoretical
evaluation, and undertake a brief comparison with the available evidence.
The main conclusions from this exercise are as follows: (1) Theories of the
LMF need to explain why non-human assets must be owned by the firm
rather than leased, but there are good reasons why this is often true; (2) it is
improbable that LMFs are rare because of deficient work incentives, as
sometimes claimed, and in fact LMFs often seem to have some produc-
tivity advantages; (3) the limited wealth of most workers, combined with
problems in attracting outside financing, does help explain a number of

empirical regularities; (4) portfolio-diversification arguments, although
popular explanations for why LMFs are rare, run into logical com-
plications, and at least for the present, lack strong empirical support;
(5) collective-choice problems do seem to account for various important
facts about the design and distribution of LMFs.
Questions of LMF rarity and distribution are approached from a very
different angle in Chapter 10, which focuses on firm births, transitions, and
deaths. I suggest that there is a bias against LMF formation as a result of
informational asymmetries, financing problems, and the advantage of the
KMF as a vehicle for appropriating entrepreneurial rent. The evidence
suggests that once LMFs have been established, their death rates are no
higher, and probably lower, than those of similar capitalist firms.
I also examine transitions resulting from employee buyouts of KMFs
and investor buyouts of LMFs. Most employee buyouts occur in finan-
cially distressed firms. Buyouts of healthier firms are obstructed by several
factors, including collective-action problems for large employee groups,
bargaining problems involving cost sharing and the willingness of em-
ployees to pay for a conversion to the LMF form, and perhaps also finan-
cial constraints. These factors tend to stabilize KMFs even though they
may have lower productivity than an LMF could achieve. On the other
hand, LMFs do appear vulnerable to investor takeover, at least more so
than KMFs are to employee takeover. The central reason probably in-
volves imperfections in membership markets, which can cause an LMF to
“degenerate” through the replacement of members by hired employees.
Chapter 11 develops my proposed theoretical synthesis, the broad out-
lines of which have already been discussed. Chapter 12 returns to the
normative concerns of Chapter 2 by proposing a strategy to increase the
extent of workers’ control in modern economies. The strategy is meant to
promote employee buyouts of conventional firms, and draws on the theo-
retical conclusions of Chapter 11 to construct a policy framework for this
process. Three key goals are reassuring outsider shareholders that their
interests will be protected during the transition phase, minimizing gov-
ernance problems among workers themselves, and managing transfers of
membership rights.
One regrettable omission is that I do not address the fascinating forms
of workers’ control that have arisen in developing countries, central and
eastern Europe, the former Soviet Union, and China. It was necessary
to limit the scope of the book because of limitations on space and my
own expertise. The political and legal institutions of these countries, their
22 Introduction
histories, and their level of economic development often diverge enor-

mously from those of advanced capitalist countries. There are good rea-
sons to wonder whether the causal forces stressed here also operate in
these economies, or whether they operate with similar intensity relative
to other influences. Nevertheless, I hope this book will prove useful to
readers who want to study or implement workers’ control outside the
narrow confines of North America and Western Europe.
2
Normative Perspectives
2.1 Why Care About Workers’ Control?

One reason why an economist should care about workers’ control is that
capitalist firms dominate large sectors of every modern economy and are
major features of everyday life. Given the apparent feasibility of firms
controlled by workers, this empirical regularity requires an explanation.
Most of this book addresses the theoretical question of why LMFs are
rare. This intellectual project should be of interest to anyone who wants
to understand the way the world works. At the same time, workers’ con-
trol taps into deeply held moral and philosophical beliefs about the pur-
poses and design of an economic system. Views expressed in the literature
run from sweeping denunciations of capitalism to equally sweeping pro-
nouncements that the market has revealed the futility of workers’ control.
This chapter surveys a series of moral justifications for workers’ con-
trol, including those based on egalitarianism, democratic theory, property
rights, and ideas of dignity and community. Most of the chapter will de-
scribe and critique arguments developed by other writers. Because each
writer makes a distinctive case against the capitalist firm, there is no
consistent version of workers’ control that all would embrace as an ideal.
Some writers advocate a radical transformation of the entire economic
system, while others endorse only minor tinkering with the status quo or
the use of workers’ control in particular contexts.
Accordingly, I do not specify any one policy proposal or organizational
alternative for sustained consideration. The focus of my assessment varies
from one author to another depending on the nature of the argument un-
der discussion. In some cases, I concentrate on logical difficulties, in other
23
24 Normative Perspectives
cases on empirical claims that seem dubious, and in still other cases on
the cogency of the moral principles involved. The chapter is best seen as
a buffet of normative views from which the reader can sample, drawing
whatever organizational inferences seem warranted. I do, however, out-
line my own stance toward the end of the chapter. Readers who want a
concrete set of policy ideas will find them in Chapter 12.
I do not address the various efficiency arguments for or against
workers’ control in this chapter. These are dealt with at length in
Chapters 6, 11, and 12. The moral arguments developed here are largely
independent of efficiency considerations. Nevertheless, it will be conve-
nient to introduce a few elementary efficiency concepts in Section 2.2
in order to help clarify the relationship between efficiency and equality.
Frequently I will highlight practical objections to a normative claim, in-
cluding the existence of potential conflicts with efficiency goals. The intent
of these remarks is not to suggest that efficiency trumps other normative
principles, but rather to flag situations in which efficiency issues might
become relevant as part of a more comprehensive policy assessment.
2.2 Equality
Some authors (Miller, 1989: 32, and Plant, 1989: 70–72) favor LMFs
because they believe firms of this kind promote greater equality of in-
come or wealth. In evaluating such claims, it is important to keep in mind
a few conceptual points about efficiency and equality. Economists define
a change in the allocation of resources to be an efficiency improvement
(in the Pareto sense) if it makes some people better off while harming
no one. A situation is defined to be efficient (in the Pareto sense) if it is
impossible to make anyone better off without harming others – that is, if
no efficiency improvement is possible.
Here the change in question is a transition from the KMF to the LMF,
either within a single firm or for all firms simultaneously. Suppose one
argues that such a transition would make the poor better off without
hurting anyone else, including the rich. Logically this claim implies that the
transition is an improvement in the Pareto sense and that the current
system of KMFs is not efficient. If one imposes the constraint that the
non-poor cannot be made worse off, it follows that any hope for LMFs to
make the poor better off is hostage to the likelihood of efficiency gains. In
the absence of such gains, one must either abandon the goal of a transition
to LMFs or remove the constraint that the non-poor cannot be harmed.
One may of course be perfectly willing to make the rich worse off in order
2.2 Equality 25
to make the poor better off, but it seems best to make this case directly
rather than trying to circumvent the distributional conflict by means of
unexamined efficiency claims that may not hold up under scrutiny.
Suppose, for simplicity, that there are only two relevant groups, workers
(viewed as poor) and investors (viewed as rich), and suppose that creation
of LMFs does not enhance economic efficiency in the Pareto sense. This
rules out full compensation for investors if the ownership of firm assets
is transferred to workers, or if returns from existing financial claims are
diminished. Outright wealth confiscation may not be needed because sub-
sidies to LMFs can be financed through progressive taxation, achieving
a similar result. Indeed, the history of Western economies suggests that
redistribution through the tax system has more often been politically
acceptable than extinction of property rights or confiscation of financial
claims. But if it is politically feasible to achieve greater equality by sub-
sidizing LMFs through the tax system, then presumably it is feasible to
use the tax system directly for redistributive purposes. In this case, the
more roundabout pursuit of equality through the creation of LMFs is
superfluous, or at least a less effective means to the given end.
Apart from shifts in the ownership of physical assets or financial wealth,
it can be argued that LMFs would reduce inequality because they would
pay more equal wages than their KMF counterparts. Here the issue cen-
ters around equality among workers rather than between workers and
investors. A response often voiced by skeptics is that an LMF could not
pay its workforce more equal wages than a similar KMF, given a com-
petitive external labor market where workers with specified skills receive
the going wage, because workers at the top end of the wage distribution
would flee to KMFs. Nevertheless, existing LMFs do adopt flatter wage
scales than similar KMFs (see Chapters 3 and 4).
This suggests that there are degrees of freedom in the real world that are
omitted in competitive supply and demand models of the labor market.
Search and mobility costs, investments in firm specific human capital,
and private information may give firms a degree of monopsony power
over some workers, and majority voting may prompt LMFs to utilize this
power differently from KMFs (Manning, 1988; Moene, 1993). Another
possibility is that KMF managers may be more insulated from investor
oversight than LMF managers are from worker oversight, resulting in
systematic differences in top managerial salaries.
LMFs surrounded by capitalist firms may attract applicants who
strongly prefer an egalitarian workplace. If there are relatively few
workers who have such preferences, there would be no reason to
expect similarly egalitarian results from an economy in which LMFs

were widespread. Assuming that the observed equalizing tendencies of
LMFs do not just reflect self-selection among workers with egalitarian
preferences, one could argue that an economy dominated by LMFs would
have a more compressed overall wage distribution as a result of majority
voting within each firm. Because such equilibria could quickly unravel if
capitalist firms were free to enter and bid up wages at the high end, those
seeking equality through LMFs have to contemplate a ban on KMFs as
part of their political program.
Even without KMF entry, an economy consisting entirely of LMFs
would be likely to retain a measure of income inequality across firms
because of variations in location, technical or organizational competence,
vintage of capital stock, and supply and demand shocks. As Nozick re-
marks: “Since a system of syndicalist factories would involve great in-
equalities of income among workers in different factories (with different
amounts of capital per worker and different profitability), it is difficult to
see why people who favor certain egalitarian end-state patterns think this
is a suitable realization of their vision.” (1974: 251). It is not obvious that
such inequalities would be “great” empirically, but the general point is
valid.
One solution is to tax away the rents derived from product-market im-
perfections. There are severe informational problems in this approach
because quantifying such rents requires knowledge of the cost and
demand conditions facing individual firms. A second possibility is to
pursue policies encouraging the free entry of new LMFs in the long run in
order to avoid persistent product-market rents in the first place. It is un-
clear how such pro-competitive policies could be made any more effective
than they are in capitalist economies. Finally, it might be possible for LMFs
to pool their incomes, a strategy that has attractive insurance features and
reduces the degree of income inequality across firms. But extensive in-
come pooling can lead to incentive problems, which, if significant, would
imply a tradeoff between egalitarian objectives and efficiency at the level
of individual firms.
My general conclusion is that the potential benefits of LMFs in pro-
moting greater equality of income or wealth have often been oversold or
at least subjected to insufficient critical analysis. The goal is laudable, but
it is far from clear that workers’ control would be an effective strategy
in pursuing it. Other strategies, such as progressive income taxes, inheri-
tance taxes, universal social programs, and centralized wage bargaining,
are almost surely more effective tools for this purpose. On the other
2.3 Democracy 27
hand, policies to encourage the conversion of KMFs into LMFs need

not aggravate current inequalities among workers or between workers
and investors, and might reduce wage inequality to some degree within
individual firms. LMF advocates need not fear that such efforts would
subvert the pursuit of greater equality in other policy arenas.
2.3 Democracy
Several political theorists, including Carole Pateman (1970), Michael
Walzer (1983), and Robert Dahl (1985), have argued that the firm resem-
bles the state and that justifications for political democracy carry over
to democracy at work. Dahl, for example, holds that an inalienable right
to democratic government arises whenever an association of competent
people must reach binding joint decisions (1985: 57–59), and “if democ-
racy is justified in governing the state, then it is also justified in govern-
ing economic enterprises” (1985: 135). Dahl acknowledges that firms lack
the coercive powers of a state, but stresses that firms do make collectively
binding decisions concerning the time, place, and methods of production,
and that these decisions are ultimately backed up by threats of firing.
Walzer agrees that control over employees within a firm is a kind of
power closely analogous to that exercised by the state over its subjects
(1983: 291). He recognizes that the analogy works better for cities than
nations because it is easier for individuals to move between cities. Cities,
like firms, are “created by entrepreneurial energy, enterprise, and risk
taking; and they, too, recruit and hold their citizens, who are free to come
and go, by offering them an attractive place to live.” (1983: 295). More-
over, cities resemble firms as sites of cooperative action and decision
making. Walzer maintains that since ownership is not an acceptable basis
for political power in cities and towns, it is likewise unacceptable in facto-
ries. “What democracy requires is that property should have no political
currency, that it shouldn’t convert into anything like sovereignty, author-
itative command, sustained control over men and women.” (1983: 298)
Displacement of collective decision-making by the power of property is
“the usurpation of a common enterprise” (1983: 300).
Walzer concludes that “at a certain point in the development of an
enterprise, then, it must pass out of entrepreneurial control; it must be
organized or reorganized in some political way, according to the pre-
vailing (democratic) conception of how power ought to be distributed”
(1983: 303). Although he accepts the idea that private entrepreneurship is
necessary for the formation and development of firms, the ultimate goal is
conversion to an LMF. He does not address the point at which this should
occur, the form that such LMFs should take, or how entrepreneurs should
be compensated when transitions occur.
Walzer admits that the problem of protecting citizens from abuse by
the state might differ in some ways from the protection of worker interests
on the job, but he downplays such distinctions. One apparent difference
is that municipal governments make decisions about a wide range of issues
affecting life at all hours of the day and night, while authority in the
firm is normally more limited in scope, duration, and location. A local
government also has access to coercive instruments (fines, taxation, arrest)
that go well beyond the harshest sanction an employer can impose on an
employee (in general, dismissal). Finally, it is only partially true that a
town is a voluntary association in the same sense as a firm. People are
born into towns, and exit costs can be extremely high due to location-
specific investments made over a lifetime.
Arneson (1993: 138–43) denies that the state and the firm are parallel
cases. Most importantly, states have powers of coercion from which citi-
zens need to be protected, and exit from the territory controlled by a state
is very costly even under democratic conditions. Employees, by contrast,
can exit as long as a competitive labor market exists. Arneson grants that
the case for workplace democracy is strengthened when exit costs are
high, but he emphasizes that workers have diverse preferences and that
some workers are willing to sacrifice democratic protections for higher
income or other benefits. Accordingly he rejects the claim that there is an
inalienable right to workers’ control. Miller (1989: 36) agrees that workers
are not all equally willing to give up income for the sake of workplace con-
trol, and believes along with Arneson that these individual preferences
should be respected.
The most explicit arguments among economists in support of work-
place democracy come from Bowles and Gintis (1990, 1993a, b, c, d).
They frame the analogy with political democracy in strong terms: “Any
compelling argument for democratic governance of the state entails
democratic governance of firms as well; and arguments that deny the legit-
imacy of democratic governance of firms equally oppose democratic gov-
ernance of the state.” (1993a: 87). Central to their approach is a definition
of power: “Agent A has power over agent B if, by imposing or threatening
to impose sanctions on B, A is capable of affecting B’s actions in ways
that further A’s interests, while B lacks this capacity with respect to A”
(1993b: 14). They then develop a theory of the labor market in which
firms create effort incentives by paying high wages and threatening to
2.3 Democracy 29
fire shirkers, arguing that this scenario satisfies the definition of power
provided earlier.
The normative punchline is that power should be accountable “to
the extent that this is not precluded by other valued social objectives.”
(1993b: 20). This caveat suggests that democratic rights could be alienable
at the discretion of the worker. A democratic society might also not insist
on accountability in firms if this clashed with other policy objectives such
as allocative efficiency or economic growth. In fact, Bowles and Gintis
(1993a, b, c) conclude that shared firm ownership (that is, control) be-
tween labor and capital suppliers is needed in order to ensure that firms
give adequate weight to risk-taking and innovation.
Robin Archer constructs a justification for workers’ control that
attempts to avoid reliance on the assumption of high exit costs. Archer
asserts that “all individuals who are subjected to the authority of an asso-
ciation should share direct control over the decisions of that association”
(1996: 18). Those affected by an association’s decisions but not subject to
its authority (for example, suppliers and customers) should have access
only to indirect methods of control such as exit. Archer explicitly rejects
the argument that workers should have control over firms because they
face high exit barriers. Instead, he says, “Direct voice control should be
exercised by labour because the employees who sell this labour are the
only human individuals subject to the authority of the firm” (1996: 21).
Further, “Even if workers can leave a firm, they are still subject to its
authority while they work for it. For the duration of their employment
they are bound to comply with certain decisions of the firm and it is this
which justifies their claim to direct control” (1996: 24).
Archer buttresses this view by pointing out that even if exit from the
territory of a state were easy, we would still distinguish between political
democracy and dictatorship. The argument for workers’ control is not
independent of its economic efficiency, however. Archer gives primacy
to opportunities for personal choice, and if worker-controlled firms were
inefficient, the gain from control over the workplace would have to be
weighed against the loss of other economic opportunities.
Archer’s effort to avoid reliance on exit costs is not entirely successful.
The key phrase in his argument is “subject to the authority of the firm,”
and it is not clear what this means in a context where the cost of exit
from the firm is literally zero. If workers are free to leave, then the source
of the firm’s authority is fuzzy at best, and employment becomes hard
to distinguish from ordinary market contracting. Since Archer specifi-
cally says that suppliers and customers do not need a direct role in firm
governance, one might think that the same would be true for workers
who function essentially as self-employed contractors. Thus, democratic
theorists have not yet provided a rationale for workers’ control that is
independent of exit conditions, although arguments based on dignity and
community may provide such a rationale (see Sections 2.5 and 2.6). More
generally, with the exception of Bowles and Gintis, the literature in this
area rarely spells out how exit costs arise or how the labor market differs
from other sorts of markets.
A further problem shared by virtually all arguments from democratic
theory is that the heavy normative reliance on exit costs can lead to
results that seem anomalous from a distributional standpoint. Suppose
worker A is a highly paid computer programmer who has substantial
firm-specific human capital and receives a large quasi-rent. Worker B is
a janitor who is paid very little but could obtain virtually the same wage
at a competing firm across the street. Democratic political theory would
likely favor giving worker A a role in firm governance (A’s quasi-rents
would be sacrificed in moving to another job, leaving A vulnerable to
abuse by her present employer), but not worker B (who can easily quit
if the boss makes unacceptable demands). If the right to participate in
firm governance depends upon whether one faces high exit barriers, it
may turn out that affluent workers are entitled to participate in managing
firms, while poor workers are not.
I have not yet distinguished between representative and participatory
democracy. Even assuming an inalienable right to self-government within
the firm, the analogy between state and firm suggests that this right could
be honored by having rank-and-file workers periodically elect represen-
tatives to a board of directors, and having this board supervise the firm’s
managers. However, such representative mechanisms have been criticized
by the advocates of participatory democracy, including Carole Pateman
and Ronald Mason.
Mason defines democracy as “a type of community rule in which the
process of decision making generally entails widespread and effective
participation of community members” (italics deleted; 1982: 30). The rel-
evant community need not be a nation-state or have territorial bound-
aries. Participatory theorists are generally reluctant to draw sharp lines
between government and other social venues in which collective deci-
sions are made, and often claim that there are spillovers from one venue
to another. Pateman, for example, sees workers’ control as a way of pro-
moting greater participation in the polity (1970: 43). More bluntly, Mason
asserts that “to increase participation in government, it is most efficient
2.3 Democracy 31
to increase participation in the workplace” (1982: 78). Some economists

likewise argue that workplace democracy would facilitate the democrati-
zation of other institutions, notably the educational system (Bowles and
Gintis, 1976; Marglin, 1991).
Pateman and Mason agree that participation is good because it creates
a better kind of human being. As Pateman says, “The major function of
participation in the theory of participatory democracy is . . . an educative
one, educative in the very widest sense, including both the psychological
aspect and the gaining of practice in democratic skills and procedures”
(1970: 42). She goes on to say that “the justification for a democratic
system in the participatory theory of democracy rests primarily on the hu-
man results that accrue from the participatory process . . . output includes
not just policies (decisions) but also development of the social and politi-
cal capacities of each individual, so that there is ‘feedback’ from output to
input” (1970: 43). Pateman and Mason believe these benefits go beyond
the acquisition of democratic skills, and extend to attitudinal changes. In
this they are joined by some economists interested in the endogeneity of
preferences, who suggest that an economy consisting of labor-managed
firms might inculcate attitudes of cooperation and concern for others
(Bowles and Gintis, 1990, 1993b, c, d).
Dahl (1985) is skeptical of such arguments, citing the research of
Greenberg (1981) who found that the U.S. plywood cooperatives did not
nurture any discernible feelings of cooperation or generosity extending
to the larger society. Greenberg (1986) found only a weak correlation
between membership in these firms and participation in politics. Per-
haps surprisingly, though, there was strong evidence that the members
of plywood cooperatives held more conservative political views than the
employees of conventional firms.
Other researchers have obtained different results. For example, Smith
(1985) finds a pronounced positive correlation between participation in
firm decisions and in community political affairs. However, it is impos-
sible to know whether there is any causal relationship between the two.
Such an association could arise because an underlying psychological trait
such as a sense of personal efficacy causes people to engage in both kinds
of participation, rendering the correlation between the two variables spu-
rious. The empirical literature from political science and sociology often
takes it for granted that the causal arrow runs from job experience to
personality, but an economist might be more tempted to argue that per-
sonality (that is, a preference ordering) influences the job characteristics
sought by workers.
The emphasis given by participatory theorists to educative functions,

acquisition of democratic skills, and the like suggests that mere annual
election of a board of directors by firm employees would be insufficient.
Pateman sometimes appears willing to accept such representative insti-
tutions (see 1970: 72 and also 1975). Gunn, drawing on Pateman’s views,
adopts a stronger position: “Most vital to the worker-controlled firm are
forms of direct democracy, not the more common ‘competition among
leaders (elites) for the votes of the people at periodic, free elections’ of
representative democracy” (1984a: 37). Mason takes a stronger view still,
seeing representative institutions as at best a necessary evil.
Much of the empirical work cited by participatory theorists is con-
cerned with the psychological effects of control over one’s job and imme-
diate work environment. From this standpoint, one can argue that having
self-governing teams at the shop floor level is the best way of educating
employees in the skills needed for collective decision-making. Nominal
control over a distant board of directors, especially in a large firm, might
add little to the benefit derived from daily experience with group problem-
solving. Thus, if a capitalist firm delegated enough authority to shop floor
teams, a participatory theorist might be untroubled by the absence of
workers’ control over the firm as a whole. Such conclusions are drawn ex-
plicitly both by Pateman (1970: 83) and Mason (1982: 189). Participatory
theory thus does not offer any real rationale for workers’ control over top
management, and may rule out managerial hierarchies that are useful on
productivity or efficiency grounds.
Pateman acknowledges that some inefficiency might result from the
introduction of democratic decision-making in firms, but argues that
whether the goal should be abandoned depends on the weight one gives
to other objectives such as the educative one (1970: 108). Interestingly,
Pateman does not claim that participatory democracy in the workplace
is an inalienable right, but instead opts for something resembling cost-
benefit analysis. Likewise, Mason occasionally suggests that it would be
possible to have too much democracy. He describes the competing values
only vaguely, but they likely include economic efficiency. Mason never-
theless believes that more participation is almost always better.
2.4 Property
A highly original argument for workplace democracy has been proposed
by David Ellerman (1992). Ellerman begins from the premise that as
a matter of physical fact, the capacity for intentional and responsible
2.4 Property 33
human action is inalienable. He then argues that the employment con-

tract pretends to the contrary that these capacities can be transferred to
the employer, who is depicted as renting the labor power of the worker.
Because responsible human agency cannot be transferred, the employ-
ment contract is fraudulent and should be ruled invalid by the courts.
The only way in which an authority relationship can be made consistent
with the de facto inalienability of human decision-making powers is by
having workers voluntarily delegate these powers to the boss. Therefore
it is necessary to replace employment contracts with democratic worker-
controlled firms.
This raises a number of intriguing questions. First, if all that is required
to avoid the taint of fraud is an act of delegation, why cannot the worker’s
voluntary acceptance of an employment contract be construed as just such
a legitimizing act of delegation? It is not at all obvious why delegation
must be a matter of collective rather than individual consent. Second,
there is a danger of proving too much. Authority relationships are often
accepted without any formal delegation in organizational settings out-
side the workplace, including universities, hospitals, and churches. Are all
of these authority structures unacceptable, or is there something unique
about employment? Third, it is hard to see in what sense there is a serious
issue of fraud. No employee can be unaware that she retains inherent
decision making powers that have not been handed over to the boss.
Employees know they are not robots. Likewise, every employer knows
that employees retain the capacity to shirk, argue, offer alternatives, and
sabotage projects. Because neither party can be in any doubt about the
true nature of the situation, the fact that lawyers or economists occa-
sionally talk metaphorically about “renting people” carries little weight.
Finally, there is the problem that Ellerman repeatedly tries to move from
a physical description of the world (inalienable labor) to a normative
conclusion (workers should control firms).
Ellerman’s other line of attack involves a labor theory of property,
as distinct from the Marxian labor theory of value. In Ellerman’s view,
people have an inalienable right to appropriate the positive and negative
fruits of their labors because they are human agents who bear responsi-
bility for their own actions. For production, this implies that labor should
appropriate all output and be liable for all inputs used up. An employ-
ment contract violates this principle because the employer appropriates
the output produced by workers.
At bottom, one either accepts this distributive principle, or not. But
some skepticism is needed because the conclusions are highly sensitive
to the manner in which contracts are described. To illustrate, suppose

a worker-controlled firm specializes in producing widgets to order. At
the beginning of each period, the buyer places an order for a particular
type of widget. Assume for simplicity that the workers are indifferent
about which kind of widget they produce. After getting an order, the
firm rents a widget-making machine, applies some labor, transfers the
widget to the buyer, and accepts payment. This is all fully consistent with
Ellerman’s requirement that workers pay for their inputs and appropriate
their outputs.
Now make the following modifications: (1) the buyer offers to pay
the price p in advance before saying which kind of widget is desired,
rather than after taking delivery; (2) the buyer acquires the widget-making
machine but continues to rent it to the firm in each period on unchanged
terms for the fee r; (3) the buyer suggests that it would be convenient
to replace the net payment to the firm, p − r , with a single payment,
w = p − r , and the workers agree to this. We have now reached a situation
observationally identical to a capitalist firm in which workers receive a
wage in advance, are supplied with necessary machinery, and are told
what to do by the capitalist, who then puts the finished widget in a truck
and drives away. Yet it is difficult to pin down the stage at which we went
from a situation respecting labor’s inalienable right of appropriation to
one where this right is violated.
Despite these problems, Ellerman is correct in stressing the factual
inalienability of labor. As discussed in Chapter 1, I will make use of this
idea in constructing a positive theory of workers’ control. Ellerman is also
correct in insisting that nothing about the ownership of a physical asset
logically entitles the owner to run the firm or appropriate the profit from
production activities (for an early statement of this point, see Ellerman,
1975).
2.5 Dignity
Perhaps the most common criticism of employment relationships is that
being subject to someone else’s authority is an affront to human dignity.
A closely related view is that such authority relations are incompatible
with membership in a community where persons are regarded as having
equal worth. These propositions have been unpacked in a variety of ways
by philosophers, political scientists, and sociologists.
Human dignity is sometimes taken to imply that labor should not be
treated as just another commodity to be bought and sold on a market.
2.5 Dignity 35
Ellerman (1992) sees employment as analogous to slavery for reasons of

this sort. Similar statements have occasionally been made by the Catholic
Church (Skalicky, 1975). However, some caveats are needed. First, labor
markets do not literally involve the purchase and sale of people. At most,
they involve the purchase and sale of labor services, which can be better
viewed as the rental of a human being rather than the ownership of one.
Due to the incompleteness of contracts, employers in practice only rent
the capacity to provide labor services, and incentives are still required
to elicit productive effort once the worker is on the job. Hence workers
do not really “sell” either themselves or their decision-making capacities
to an employer.
If one accepts a strong version of the “anti-commodification” thesis for
labor, one might insist that workers should never pay a membership fee to
join an LMF because this reinstates objectionable aspects of the capitalist
labor market. For example, in criticizing the use of membership markets
by the plywood cooperatives (see Section 3.2), Gunn says that “workers’
self-management is based on the principle that the right to control is
based on work, not on capital ownership. That right should remain with
those who work; it is not property to be bought and sold” (1984a: 122–23).
Similar ideas have been expressed by Ellerman (1984a: 260–62), who
holds that in a firm controlled by workers, voting rights and rights to share
in any economic profit should be regarded as personal rights attaching to
the functional role of working in the firm, and must therefore be non-
tradeable.
Such statements obscure a fundamental distinction between buying
and selling the right to membership in a democratic firm, and the subordi-
nation characteristic of a capitalist employment relationship. The former
involves the terms on which members are admitted, while the latter in-
volves the nature of control rights after a worker enters the firm. In later
chapters, I argue that a ban on admission fees carries a price tag: LMFs
that cannot use a membership market must compete against KMFs with
one hand tied behind their backs. One need not constrain LMFs in this
way even if one believes the labor market is morally problematic be-
cause it involves unaccountable authority relationships in the capitalist
firm. The authority question has already been addressed by ensuring that
members will possess control rights within the LMF after they join it.
Hiring processes under workers’ control have received some atten-
tion from Michael Walzer. He suggests that “qualification in the strict
sense – the ability to do the job or to learn to do the job – seems only
the first requirement for admission. The present members are free, if they
like, to set additional requirements, having to do with the sense they have
of their common life . . . if they cannot choose their co-workers, it is dif-
ficult to see in what sense they can be said to ‘control’ their workplace”
(1983: 161–62). This implies that LMFs should avoid a radical commodifi-
cation of membership, where individual members can sell their positions
to whomever they wish. At a minimum, insiders should be able to veto
new members (for further discussions of this issue see Sections 7.5 and
12.5).
I now want to put aside concerns about market transactions and focus
instead on the question of why workers might suffer a loss of dignity or
self-respect in a capitalist firm. The fundamental problem is that such firms
involve a unidirectional authority relationship between a boss and subor-
dinate, where one commands and the other obeys. Such master-servant
relationships are by their nature not conducive to dignity or self-respect on
the part of the subordinate (Bowles and Gintis, 1993d: 89). For example,
superiors sometimes feel free to express scorn, contempt, or indifference
toward a subordinate. These experiences are not only unpleasant, but
can lead subordinates to internalize the attitudes expressed by superiors.
There is a danger that workplace authority will be interpreted by both
parties as expressing some larger truth about the relative social worth of
the individuals involved.
In a worker-controlled firm, there are still typically bosses, or at least
managers, but the boss is ultimately accountable to subordinates as a
group or to their representatives. It is unwise to express attitudes of scorn
or contempt toward people who can fire you. When there is a direct
personal cost attached to such behavior, beyond possible costs associated
with compensating wage differentials in a capitalist firm, such attitudes are
less likely to be expressed. More fundamentally, accountability changes
or at least clarifies the nature of the authority relationship. In the LMF,
it is clear to everyone that authority is an instrumental, utilitarian social
arrangement serving the interests of the subordinates, chosen by them,
and continuing only with their consent. This makes it more difficult for
anyone to interpret an authority relationship as having roots in the relative
social worth of the parties.
The idea that subordination on the job leads to a devaluation of the
person is not a new one. Pateman says that “it is almost part of the
definition of a low status occupation that the individual has little scope for
the exercise of initiative or control over his job and working conditions,
plays no part in decision making in the enterprise and is told what to
2.5 Dignity 37
do by his organisational superiors” (1970: 50). Sennett and Cobb point

to survey data on the perceived desirability of occupations in reaching
similar conclusions, remarking that “occupations in which the individual
possesses some degree of autonomy – that is, some degree of freedom
from authority and from having to define his own function in terms of the
shifting demands of others – are more desirable than jobs where a person
has to deal with others and respond to them” (1972: 235–6). According
to Sennett and Cobb, taking orders is viewed by the subordinate as evi-
dence for a lack of ability, talent, or merit, since truly talented people are
believed able to get jobs in which they decide autonomously how to do
their work. This gives the subordinate a way to identify her own position
in the scale of social worth.
One can, of course, find occupations that combine low status with con-
siderable personal initiative (for example, scavenging for bottles to be
recycled), but such instances more often resemble self-employment than
employment within a capitalist firm. Likewise, as Nozick (1974: 246) ob-
serves, there are cases where authority relationships need not lead to a loss
of self-respect – for example, members of a symphony orchestra, draftees
in an army, or people on the way up a career ladder. But in the capitalist
firm, workers do not generally have any rare skill widely respected by
the rest of society (the orchestra); their situation is not clearly coercive
(the draftees); and their subordination may not be temporary (the rising
stars on a career track). Without special circumstances of this kind, it does
appear true that there is a chronic tendency in authority relationships for
respect to become unidirectional. An LMF avoids this because it is pre-
sumed in such a firm that workers are knowledgeable enough to assess
the performance of their own managers, at least on average and in the
long run.
David Miller (1993) maintains that the goal of market socialism is to
create equality of status rather than simply material equality, although the
latter is also important. Since Miller’s vision of market socialism involves
worker-controlled firms, his views are relevant here. He wants a society
“in which people regard and treat one another as equals and are not
separated from one another by vertical differences of class,” and goes on
to distinguish this from a hierarchical system in which interactions are
structured on the basis of the caste or class to which people belong. The
notion of class is not that of Marx, but rather one of a social group standing
in a hierarchical relation of superiority or inferiority to other groups.
Although it may be unrealistic to think that LMFs would remove all
such class distinctions, Miller’s emphasis on workers’ control as a partial

solution is not misplaced.
2.6 Community
There is a close connection between dignity and community. For example,
Walzer (1983: 272–80) says that self-respect originates in a knowledge of
one’s own competence relative to the norms and expectations of a given
peer or reference group, while community implies a sense of belonging
to some group of voluntarily associated and self-governing individuals.
Self-respect can therefore be viewed as knowledge that one measures up
to the standards of a particular community, which may be professional or
occupational in nature.
Mason approaches the idea of community from a slightly different
angle, stressing consciousness of shared interests: “Community refers to
a group of people bound into self-conscious units by common interests,
concerns, and problems” (1982: 153–4, emphasis deleted). He goes on
to say that democracy may be a relevant concept for the workplace if
this shared consciousness exists among workers, and that everyone at
the workplace would then be considered a member of the community in
question. He is silent on the prospects for workplace democracy if these
preconditions happen not to be satisfied, either because workers have
conflicting interests or are unaware of the interests they share.
Whatever definition one chooses, it seems clear that KMFs and LMFs
do not create an equivalent sense of community. KMFs do, of course, bring
together a set of individual employees with parallel interests through
a series of bilateral employment contracts. These employees may even
organize themselves in opposition to the owners or managers of the firm –
for example, by forming a union. But in a KMF, employees typically have
only weak norms about doing their work well, and the oppositional nature
of workplace relations can lead to norms against cooperating with bosses.
Any sense of community that does emerge tends to be limited to workers
located at roughly the same hierarchical stratum in the firm.
The shareholders of a capitalist firm may not constitute a community
under either of the definitions offered here. Thomas Weisskopf (1993: 126)
notes that capital suppliers typically have unequal membership rights,
little social contact with one another, few ties to the geographical area
in which the enterprise is based, and no enduring shared identity. The
LMF differs because it is a self-organizing association of workers who
jointly choose who will be admitted and how their shared activities will
2.6 Community 39
be structured. Because all workers are equal in the formal sense that each
has an equal vote when policy is made or representatives are elected, there
is an obvious community of members that can function as a substrate for
the emergence of social norms, including norms about the value of good
work. These norms then provide a basis for self-respect among those
workers who conform to them.
The members of an LMF often live in close proximity to their place of
employment. The LMF thus tends to link shared membership in a local
community (for example, a town) with membership in the economic com-
munity of the firm. This enables the LMF to take advantage of pre-existing
loyalties to create a common identity at work. It also implies that LMFs
tend to act spontaneously in the larger interests of the local community,
because LMF members share interests with nearby non-members. As
Weisskopf (1993) remarks, worker-controlled firms usually favor stable
employment and locally oriented investment, reinforcing the stability of
the residential communities in which members live. The greater willing-
ness of LMF members to supply local public goods, and to restrict local
public bads such as pollution, also serves the interests of the surrounding
community.
Thus far, the arguments developed in this section, although involving
psychological mechanisms not normally considered by economists, are
consistent with the commitment of most economists to the centrality of
individual welfare. But it seems clear that many people have moral intu-
itions that depend on the relational features of a society or community, in
ways that are at least partly detached from standard concepts of individual
welfare. Such intuitions are most commonly expressed in the language of
rights (Nozick, 1974), though some philosophers prefer to frame the issue
as one of “non-domination” (Walzer, 1983).
Perhaps the most obvious example of a relational value is freedom.
This is usually said to be a value in itself, independently of whether it
results in higher GDP per person, or whether moving from unfreedom
to freedom raises everyone’s utility level. This view is widely endorsed
by writers on the political left as well as the political right. For example,
Rawls (1971) assigns civil liberties lexical priority over the distribution
of primary goods. Bowles and Gintis also note that even if workers are
materially better off in undemocratic firms, to say this “is to reduce the
political and moral assessment of a relationship to its economic conse-
quences. It is thus analogous to defending slavery on the grounds that
slave living standards were higher than that of free agricultural labor”
(1996a: 74).
Suppose that given a choice between dictatorship with high individual

welfare (the trains run on time) and democracy with somewhat lower
individual welfare (the trains are often late), most people would choose
democracy. Is it just that democracy happens to be an argument in each
person’s utility function? This is surely one possible interpretation. But
another interpretation is that people care directly about the perceived
fairness or justice of social relationships and simply disapprove of the
power asymmetry between dictator and citizen on moral grounds. If one
can mount a non-instrumental, non-utilitarian defense of freedom over
dictatorship, then on similar grounds one should find it possible to criticize
caste systems or impermeable class boundaries.
Do such arguments extend to the community called the firm? Here the
argument is inevitably weaker because employment is only one dimension
of life, the sanctions that can be imposed on recalcitrant subordinates are
less severe, and association with the firm is in some sense voluntary. But
the underlying point that people care about social relationships apart
from their consequences for individual welfare could also carry over to
this context. Employment may be only one dimension of life, but it is
an important one; sanctions may not involve the firing squad, but being
fired is painful; and normally it is hard to turn down a job if one doesn’t
know when the next offer will arrive. Under these circumstances, the
relational aspects of unaccountable authority appear morally suspect, and
encouragement of more LMFs (or even an outright ban on KMFs) might
seem to be an appropriate response.
A skeptic could counterattack on several fronts. It might be argued that
the day-to-day experience of subjection to authority is likely to be rather
similar in KMFs and LMFs, especially if the LMF is large, the voice of any
given individual is small, substantial power is delegated to managers, and
there is little opportunity for direct participation in job-related decisions.
This resembles the argument by some participatory theorists that formal
control over the firm as a whole is less important than the content of daily
worklife.
The skeptic might also complain that too much weight is given in the
definition of community to the requirement that an association of in-
dividuals be self-organizing and self-governing. This clearly stacks the
deck in favor of the LMF. But in reality, some sense of community might
well emerge among employees of a KMF, while the members of an LMF
merely show up, do their jobs, and go home. In short, “community spirit”
might be rather loosely related to control over firms, weakening the case
for the LMF. Finally, the skeptic might object that if workers place a high
value on democratic governance, the market will supply it (Nozick, 1974:

ch. 8). Because this rarely occurs, the presumption is that most workers
don’t care very much about their lack of autonomy on the job.
The latter criticism can be blunted in several ways. First, workplace
democracy is a local public good, and may be undersupplied from an ef-
ficiency standpoint (see Chapters 11 and 12). Second, it strains credulity
to apply the concept of compensating wage differentials to self-respect.
Although workers’ control may yield heightened self-respect over time,
it is implausible that workers would accept a lower initial wage be-
cause they know that working in a certain kind of firm will eventu-
ally lead them to revise their beliefs about their own social worth.
To put it another way, it is hard to imagine people changing jobs in
order to prove that their own negative beliefs about themselves are
false.
Finally, even if one believes that the market selects efficient organiza-
tional forms, it is doubtful that efficiency ideas can do justice to the issues
raised in this section. A society might see the equal worth of its citizens
as a principle that requires institutional recognition not just in the vot-
ing booth but also in the workplace. Such a society might well refuse to
sacrifice dignity, autonomy, and self-respect for more consumer goods,
much as it would refuse to sacrifice political democracy for an efficient
dictatorship.
2.7 The Author Shows His Cards

My view includes the following elements: (1) As a result of market failures,
LMFs may yield efficiency benefits in some sectors of the economy where
they are currently absent; (2) while the analogy between state and firm
is inexact, there is merit in the opinion of political theorists that workers
should have governance rights within firms subject to the constraint that
other social objectives are not seriously impeded; (3) workers’ control
would likely reduce the psychological damage and social stratification
resulting from the existence of unaccountable authority within firms. At
this stage it is premature to address point 1. The remainder of this chapter
discusses points 2 and 3 in more detail.
I disagree with Dahl, Walzer, and Ellerman that workers have an in-
alienable right to democratic self-government within firms, and side in-
stead with Bowles, Gintis, Miller, and Arneson in believing that such a
right, if recognized by the rest of society, should be freely alienable. Work-
ers themselves might give more weight to high wages or a simple life than
to participation in firm governance, and it is possible that more of the

latter could imply less of the former. I am reluctant to override such judg-
ments by workers because participation in democratic governance should
not itself be imposed undemocratically.
I also agree with Bowles and Gintis that there may be good reasons
for withholding control rights even when workers do want them. The in-
terests of society as a whole may demand that other goals take priority,
perhaps including efficiency goals such as preserving entrepreneurial in-
centives. There is no contradiction in this stance. Democracy at the level
of the larger society, as expressed through the political system, may quite
reasonably trump the narrower interests of workers in particular firms.
However, I do favor the recognition of an alienable right to democratic
workers’ control in situations where there is no pressing social need that
would be sacrificed.
This may seem like thin gruel compared with the ringing manifestos for
workplace democracy sometimes found in the literature. But my view has
more normative force than may appear at first glance. The establishment
of democratic governance within a firm is a constitutional innovation:
It specifies procedures the firm will subsequently use in making deci-
sions about many local public goods of importance to workers, including
production methods, administrative rules, working conditions, and social
atmosphere. An employer cannot legitimately claim that employees have
chosen to forego governance rights merely because they have accepted a
series of bilateral employment contracts in which the subject never came
up. No individual worker values democratic governance as much as work-
ers taken as a group, and thus no individual worker will be prepared to
bribe the employer to institute such governance rules for the benefit of
all. It might be argued that an employer would supply democratic gover-
nance if workers as a group were willing to pay enough for it in reduced
wages, but without a direct expression of worker preferences it is hard
to see how an employer could estimate the total demand for democratic
safeguards. It is also hard to see how workers’ control could be insti-
tuted while the employer remained on the scene. Something resembling
an employee buyout is required.
My normative conclusion is that the right to workers’ control is indeed
alienable, but only collectively so, and only through an explicit democratic
process in which workers as a group are offered a choice among two or
more governance systems. Since I will propose a plan for this purpose in
Chapter 12, there is no need to elaborate the details here. However, a few
general points can be made.
Society as a whole, using standard political institutions, would set the

ground rules by determining which firms need to hold a control refer-
endum, and how such a referendum would be carried out. If workers’
control is adopted, I do not favor confining control rights to workers who
have high costs of exiting from the firm. There is no clear dividing line
separating “high” from “low” exit costs – these costs are often private
information known only to the worker involved, and they fluctuate with
market conditions. Furthermore, if one accepts the importance of dignity
and community, then it is unacceptable to stratify the firm by granting
voting rights to some workers but not others. Exceptions should be made
only in narrowly defined circumstances – for example, an independent
contractor, someone who is clearly a temporary member of the firm, or
someone serving a probationary period.
I do not agree with the tendency among participatory theorists to min-
imize the value of formal control rights or representative democracy. Au-
tonomous teams and other modes of participation at the shop-floor level
are fine as far as they go, but they do not substitute for control over the
authority structure of the firm as a whole because they do not address
the problem that delegated authority can be revoked at any time and
do not promote a sense of community across hierarchical ranks. More-
over, if specialized administrative structures are required in large firms,
then representative institutions appear indispensable as a way of making
managerial authority accountable to the workforce in such firms. At the
same time, it is vital to face squarely the difficulties that arise in devising
electoral processes for highly heterogeneous workforces (see Sections 9.7
and 12.4).
If workers vote to take over control of a firm, they will need to com-
pensate existing shareholders appropriately. Because I am only concerned
with broad normative principles, it is unnecessary to specify the mechan-
ics here. My reasons for requiring compensation are practical ones: This
is essential in order to encourage continued investment in capitalist firms,
and to ensure the political viability of a program to facilitate transforma-
tion of KMFs into LMFs. I will address these issues at greater length in
Section 12.3.
A fundamental objection to this idea is that if workers lack the wealth to
take control of a firm, rights to self-government will turn out to be empty.
The appropriate remedy is to persuade the larger society that workers’
control should be supported by financing a part of the compensation to
existing investors out of tax revenues. Lest this seem naively utopian,
it should be noted that such a process has been underway for several
decades in the United States through tax subsidies to employee stock

ownership plans (see Section 4.2). If workers’ control yields an efficiency
gain because some market failure is being rectified, subsidies may be
justified for that reason alone. But even if there is an efficiency loss, society
may still subsidize transitions to workers’ control in some or all firms
because the other values at stake are deemed sufficiently important.
As was mentioned previously, I do not believe that LMFs are useful
vehicles for the redistribution of income or wealth. It is highly improbable
that these firms will become common enough any time soon to have effects
on the aggregate distribution of income that would rival those of familiar
redistributive methods such as income taxation. Thus, efforts to bring
about redistribution are better directed toward these traditional policy
instruments. Moreover, any expansion in workers’ control involving the
confiscation of financial claims on firms, or a sharp reduction in their value,
would face vigorous political resistance. It would also trigger capital flight
from the political jurisdiction involved.
There are good reasons other than redistribution to promote more
widespread workers’ control. Given the difficulties outlined here, it is un-
wise to hobble such efforts by linking them with quixotic forms of income
or wealth redistribution. The key objective, in my view, is to empower
workers rather than to expropriate capitalists. From this point of view, a
better political and economic strategy is to devise transitions to workers’
control, either within individual firms or for the society more broadly,
which minimize costs borne by other segments of society.
Before one can seriously evaluate policies designed to promote the
spread of LMFs, it is necessary to gain an understanding of the reasons
for their formation, survival, and (in some cases) failure. I will therefore
put normative issues aside until Chapter 12, and attempt instead to explain
the present rarity of workers’ control. The first step, in Chapters 3 and 4,
will be to examine some success stories.
3
Workers’ Control in Action (I)
3.1 Surveying the Terrain

Before assessing candidate explanations for the rarity of workers’ control,
one must become acquainted with some of the rich descriptive material
on the topic. This literature is overwhelming in both scale and scope,
making a complete survey impossible. But it is vital to investigate a few
cases of workers’ control in order to anchor later theoretical analysis in
factual bedrock. I have chosen some examples for which there is good
information on the historical origins and organizational structures of the
firms involved. Priority has also been given to examples that have inspired
ongoing research by economists and others.
By these criteria, it is essential to pause for photo opportunities at
the U.S. plywood cooperatives, the Mondragon system in Spain, and the
Lega federation in Italy. All three cases are clearly success stories in the
annals of workers’ control, and have been studied extensively. While not
representative in any statistical sense, these cases raise intriguing ques-
tions. How did such firms thrive in competition with capitalist rivals? Why
has their success not been replicated elsewhere? Why plywood? Why the
Basques? Why northern Italy?
I also inspect two less exotic species: employee stock ownership plans
(ESOPs) in the United States and legislated codetermination in Germany.
The firms in these categories tend to occupy an intermediate position be-
tween the abstract extremes of management by capital suppliers or by
labor suppliers. Firms with ESOPs, for example, range from conventional
corporations that offer minority stock ownership to employees as part of a
pension plan to fully democratic enterprises governed on the principle of
45
46 Workers’ Control in Action (I)
one vote per worker. In most cases, these firms do not exhibit full-blown
workers’ control, but they do supply important evidence bearing on the-
oretical and policy issues that take center stage in later chapters.
Although the case studies in this chapter and the next are framed
predominantly in descriptive rather than theoretical terms, they are not
merely random collections of facts. I have focused on information that
seemed useful in assessing the hypotheses about LMFs to be presented
in Chapters 7–11. Each case sketches the size of the firms, the technology
they use, the product markets in which they operate, and their financing
needs. Features of special interest include the ways in which these LMFs
have attracted capital (from their own members or external sources) and
labor (including the terms on which members enter and exit from the firm,
as well as the hiring of non-member labor, if any is used). The narratives
also pay particular attention to the entrepreneurial activities that led to the
formation of the firms, the procedures used to govern them, the methods
by which they allocate net income, any distinctive dilemmas, tensions, or
conflicts they have faced, and the legal or political environment in cases
where it was especially supportive or constraining. More examples could
have been added (for example, law firms or the Israeli kibbutzim), but
space constraints prevented their inclusion.
To provide context for the case studies that follow, I briefly summa-
rize the recent state of workers’ control in North America and Western
Europe. The focus is on traditional producers’ cooperatives, although I
do append a few remarks on professional partnerships at the end of this
initial survey. The rest of the chapter provides detailed descriptions of the
plywood cooperatives and the Mondragon group. Chapter 4 offers similar
descriptions of the Lega federation in Italy, ESOPs in the United States,
and the German system of codetermination. The case studies in these two
chapters are arranged roughly in order from firms that evolved without
much government involvement to those for which political intervention
was crucial.
Experiments with workers’ control can be traced at least to the
Industrial Revolution. The starting point is inevitably hazy because
pre-industrial artisans often formed themselves into cooperating groups
that would probably satisfy most contemporary definitions of workers’
control. Indeed Estrin (1989: 169–70) suggests that early workers’ coop-
eratives can be regarded as efforts to restore conditions of pre-industrial
production, when artisans often owned their tools and worked indepen-
dently. Craft techniques became less viable in the aftermath of the Indus-
trial Revolution. This led to a search for institutional arrangements that
might deliver a measure of self-determination in the modern economic

environment. A great many blueprints for workers’ control, both practi-
cal and utopian, were developed over the following two centuries. That
history is a subject for another book, but interested readers can examine
the first ten chapters in the anthology edited by Vanek (1975), as well as
Vanek’s own introduction, to gain an overview of the diverse doctrines
that emerged.
Modern-day workers’ cooperatives exist, but have failed to transform
the economic system as their early advocates hoped. Instead, they are
a peripheral form of organization that prospers only in narrow niches.
Even in countries with thriving cooperative sectors, a very small fraction
of the total workforce is employed in such firms (about 2–3 percent in
Italy, whose sector is by far the largest in the West). The best data are
for the 1970s and 1980s, but it is unlikely that the situation has changed
dramatically since then.
Confining our attention solely to worker-controlled firms, Italy had
about 11,000 firms with 428,000 workers in 1981 and France had about
1,300 firms with 34,000 workers in 1986 (Bonin, Jones, and Putterman,
1993). Toward the end of the 1990s, the Mondragon group in Spain
employed over 40,000 workers (see Section 3.3). Between 1976 and 1981,
a period of quite rapid entry into this sector, over 14,000 cooperatives
employing 223,000 workers were created in the European Community
(Ben-Ner, 1988a). Further information on European workers’ cooper-
atives can be found in Defourney, Estrin, and Jones (1985); Jones and
Svejnar (1985); Estrin, Jones, and Svejnar (1987); Pérotin (1987); Ben-Ner
(1988a); Estrin and Jones (1992, 1995, 1998); and the references on the
Lega in Chapter 4.
Workers’ cooperatives range from a few large Italian construction firms
with about 3,000 workers, to individual Mondragon enterprises and U.S.
plywood cooperatives with around 200–300 workers, and French con-
struction firms averaging 60 workers. Most countries also have many tiny
cooperatives with 10 or fewer workers (Bonin, Jones, and Putterman,
1993). Except possibly for the Mondragon system as a whole, no workers’
cooperatives are comparable in size to large multinational corporations.
In part this stems from the fact that cooperatives are less diversified ge-
ographically and by product line. The exceptions, notably including the
Mondragon and Lega coops, all involve federations or other sheltering
institutions.
A similar degree of variation exists when cooperatives are ranked by
the size of the capital investment by a typical member. The shares of
U.S. plywood coops were sold for $90,000 in 1983 (Bonin, Jones, and
Putterman, 1993), and shares in sanitation coops in the San Francisco area
were sold for $50,000 in 1976 (Russell, 1985a: 75). The capital accounts
of individual workers at Mondragon ranged up to $50,000 in the 1980s.
Similar accounts in Italy were worth $500–1,125 for 1976–80, in France
$2,300–3,700 for 1979, and less than $200 in the U.K. for 1968. The mem-
bers of French cooperatives extended loans to their firms that on av-
erage were somewhat less than the balances in their individual capital
accounts, while members of Italian cooperatives made loans that were
larger than the balances in these accounts (Estrin, Jones, and Svejnar,
1987).
The financial arrangements of European cooperatives deserve further
comment. A typical French firm, for example, obtains capital from work-
ers in three ways (Estrin and Jones, 1992, 1995, 1998). First, individual
capital accounts are opened when workers pay a fee and become mem-
bers. Second, the firm has collective reserves accumulated out of retained
earnings. Third, the members can loan funds to the cooperative and be
repaid with interest. The first and second of these, which could both be
called equity capital, differ because balances in individual accounts can
be recovered on exit from the firm, while collective reserves cannot be.
In practice, individual capital accounts are more like loans because the
principal is repaid at its book value on departure, and rates of return
on these accounts are pre-determined. Fluctuations in the firm’s operat-
ing income are handled through changes in the contributions made to
collective reserves and the bonuses paid to workers. Current labor in-
come is divided between a flat wage and a variable bonus, presumably to
balance the benefits from income-smoothing against the incentive effects
of profit-sharing.
Ben-Ner (1988a: 10–11) reports that Western European workers’ co-
operatives are primarily found in construction and certain branches of
manufacturing. The latter include printing and publishing; mechanical
and metal products; clothing, textiles, and leather; and glass, ceramic,
wood, and furniture products. Cooperatives had essentially no presence
in chemicals and pharmaceuticals; food and beverages; iron, steel, and
other metals; or the transportation equipment and automobile manu-
facturing industry. Italian cooperatives are especially prominent in con-
struction, although they engage in many other activities as well (see
Section 4.1). French coops have traditionally been most concentrated in
construction, printing, and mechanical engineering, with recent growth
in electrical products, consulting, and other services. British coops have
focused on clothing, footwear, and printing (Estrin, Jones, and Svejnar,

1987).
Jones (1979) finds that U.S. workers’ cooperatives have historically
displayed a strong craft orientation, with past or present fields of activity
including metal foundries, barrel-making, shingle-making, and plywood.
Gunn (1984a: ch. 3) studies a set of U.S. reforestation cooperatives that
arose in the 1970s, and Russell (1985a,b) describes U.S. scavenging and
taxi coops. Conte and Jones (1985) report that many standard industrial
classifications (SIC codes) have few or no cooperatives, confirming the
notion that U.S. workers’ cooperatives have generally been confined to
a narrow range of industries. Such cooperatives account for a negligible
share of employment in the United States. The plywood coops as a group
have never had more than a few thousand members at a single point in
time, and after their early rapid growth, the reforestation coops had about
650 members in 1982.
The portrait emerging from this review is that workers’ coopera-
tives tend to favor labor-intensive industries that lack significant scale
economies, require few specialized physical assets, and have minimal bar-
riers to entry. This picture is reinforced by the large labor coefficients
found for samples of such firms when production functions are estimated
(for instance, Defourney, Estrin, and Jones, 1985). Such generalizations
must be regarded with caution, however. There are some Mondragon and
Italian cooperatives specializing in producer and consumer durables that
have considerable capital intensity. It is difficult to generalize about the
human capital of workers in cooperatives. In certain industries, such as
reforestation, skills are easily learned on the job, but craft occupations
sometimes demand considerable experience, and professional activities
can require many years of education.
The preceding comments focused on traditional workers’ cooperatives
and excluded what is arguably the largest category of worker-controlled
enterprise in modern economies: the professional partnership. The paral-
lels between such partnerships and other forms of workers’ control have
been widely recognized (for example, Jensen and Meckling, 1979, and
Russell, 1985a: ch. 5). Hansmann (1996) notes that “employee owner-
ship” was long the dominant mode of organization in the United States
for law, accounting, investment banking, management consulting, adver-
tising, architecture, engineering, and medicine.
Such firms can be organized either as partnerships or limited lia-
bility corporations. However, U.S. rules of professional conduct explic-
itly prohibit lawyers from serving as the employees of a profit-seeking
organization unless the organization is wholly owned by lawyers practic-

ing in it (Hansmann, 1996: 85). Firms of this type can be quite large by
the standards of workers’ cooperatives. The world’s largest law firm as
of March 2000 was assembled by merging partnerships in the United
States, the U.K., and Germany, and at the time of the merger employed
3,100 lawyers including 570 partners (New York Times, 2000).
Professional partnerships in a number of fields have been drifting to-
ward control by investors, at least in the United States. Hansmann remarks
that “in advertising, investment banking, and primary medical care, for
example, investor-owned firms have recently come to occupy large shares
of the market” (1996: 67). Advertising firms began converting to this mode
of organization in the early 1960s, and most large firms in that industry
are now investor-owned. Similarly, investment banking firms began aban-
doning the partnership form in the 1970s, and medical partnerships have
recently lost ground to investor-owned health maintenance organizations
(Hansmann, 1996: 96). In the fall of 2000, the 2,500 partners of Andersen
Consulting, the world’s largest consulting firm, approved a proposal to
sell a portion of the company to outside investors in order to raise capital.
At the same time, several other large consulting partnerships were either
being acquired by corporations or were planning to offer stock to the
public (International Herald Tribune, 2000).
3.2 The Plywood Cooperatives

The plywood cooperatives of Washington and Oregon have been a main-
stay of the literature on labor-managed firms since the 1960s and are
among the best researched LMFs in the world. In this survey, I mostly use
descriptive material from Berman (1967), Bellas (1972), and Greenberg
(1986). Research comparing the behavior and performance of the coops
with conventional firms in the same industry is reviewed in Chapter 7 and
elsewhere. Just three plywood coops remained by the end of the century
(Pencavel, 2001), but I will often speak of them in the present tense for
expositional purposes.
Olympia Veneer, the first plywood cooperative and the sixth plywood
mill built in the United States, was organized in 1921 by Edward Westman
and J. J. Lucas. Initially, 125 shares were sold at $500 each to experienced
loggers, carpenters, and mechanics in the Olympia, Washington, area to
finance construction. Westman was Swedish born and sold numerous
shares to Scandinavian friends. In July 1921, the owners were asked to
give another $500 each, and did so by running down savings accounts or
pledging future wage deductions. The organizers also secured a bank loan
of $25,000. Operations began in August 1921.
In return for his investment of $1000, each member gained the right
to work in the plant and share equally in any profit. If he wished to
sell, an owner was obliged first to offer his share to the company at a
market value determined by the shareholders. Newcomers had to be ap-
proved by a majority of the board of trustees who managed the firm,
although major decisions such as this were ratified at shareholder meet-
ings. By 1923, shares were selling to outsiders at a price of $2,550, and
the cooperative was recognized for high output per person, high qual-
ity, and innovative production methods. During the first half decade
of operation, 100 unskilled non-members were hired, while the num-
ber of working owners dropped from 118 to 92. Over the next thirty
years, Olympia came to resemble a closely held corporation. In 1946, the
original plant was sold to a conventional lumber company, and by 1952
there were fewer than fifty working shareholders left in the coop. Most
had acquired shares subsequent to the firm’s creation; only four to five
founders were still working, and 1,000 non-owners were employed. Fi-
nally, in 1954, Olympia was sold off to U.S. Plywood Corporation. Berman
(1967) estimates that twenty-three individuals averaged $652,000 each
as their return on an initial investment of $1,000 and a year of unpaid
labor.
No further plywood coops were formed until just before World War
II, when three were created. In 1939, Anacortes Veneer was organized
when 250 workers put up $1,000 each, and another $1,000 later. By 1951,
the shares were worth $28,000 and Anacortes was paying wages up to
double the prevailing union rate. Peninsula Plywood Corporation was
established in 1941 by 280 owners who each invested $1,000. Puget Sound
Plywood was created in the same year by 285 owners who also invested
$1,000 apiece. The first Oregon plywood cooperative, Multnomah, sold
300 shares for $2,500 each plus subsequent payroll deductions. Within
two years, Multnomah shares were worth $12,000–15,000. These striking
successes have been attributed to strong wartime demand for plywood
and rapid expansion in the use of plywood for housing construction after
the war ended.
Most plywood coops arose during the period 1949–56, when twenty-
one of the twenty-four operating in 1964 (the date of Berman’s research)
were organized. Of these twenty-four, nine took over existing plants,
thirteen built their own, and two took over plants not currently in opera-
tion. Reported motives for the sale of mills by conventional firms include
bankruptcy, depletion of nearby timber supplies, labor unrest associated

with union activity, and low labor productivity.
The funds raised by the sale of initial shares in a coop were limited
by two factors: the number of jobs to be created, and what potential
members could afford. In addition to initial share offerings and wage
deferrals, financing methods included mortgaging property such as land,
buildings, equipment, and inventories; borrowing from family and friends;
borrowing from banks; use of installment payments to buy machinery and
equipment; and loans from the Small Business Administration. Smaller
amounts were sometimes obtained from sales agents, former owners, sup-
pliers, customers, private individuals, and existing coops. Machinery was
occasionally leased, but this does not appear to have been common.
Observers agree that the coops had no difficulties in arranging short-
term loans to cover raw material, inventories, or operating expenses.
Loans could be collateralized using logs, sales contracts, or accounts re-
ceivable, and were occasionally unsecured. The coops would also some-
times ask their members for personal loans, give them promissory notes, or
sell them preferred stock. Some of the early coops sold non-voting shares
to the public, but this involved only local buyers and was not widespread
(Berman, 1967: 131–32).
Plywood coops were often organized by members or organizers of al-
ready existing coops. Entrepreneurial initiatives also came from workers
seeking self-employment, local businesspeople interested in the economic
vitality of the region, and individual promoters. The latter were attracted
by the rapid growth of share prices in the early 1950s, and hoped to achieve
quick capital gains by organizing more coops. These individuals often had
little knowledge of the industry and no intention of working in a mill. Two
such promotional efforts resulted in criminal trials for securities fraud
and mail fraud (Berman, 1967: 113–16), at which the principal promoter
was sentenced to twelve years in prison and another to seven years. This
brought the formation of new cooperatives to an abrupt halt.
The heyday of the plywood coops was the period 1942–57, when
Berman estimates they accounted for about 20–25 percent of total pro-
duction capacity in the industry. By 1964, this had fallen to 14 percent.
No new coops were created during 1956–1972 despite the construction of
ninety conventional plywood mills in the Pacific Northwest during these
years. At the end of this period, Bellas (1972: 15) found that twenty-one
coops were active. Berman and Berman (1989) reported that their data
included all ten remaining cooperatives still manufacturing softwood ply-
wood, along with seven mills formerly coops but no longer operating
as coops. In 1990, according to Craig and Pencavel (1995), seven of the

thirteen plywood mills in Washington state were coops, and by 2001 only
three coops were left (Pencavel, 2001). As Bellas (1972: 31) says, “The
plywood cooperatives are not a movement but a phenomenon. Exclud-
ing the four earliest mills, this phenomenon took place between 1949 and
1955.”
The plywood cooperatives display some variability in size, product mix,
financial results, and degree of integration backward into veneer produc-
tion and logging, or forward into distribution, but several generalizations
can be made. Each coop has a single plywood mill, although some have
offsite logging and selling activities. At least in the early phase of the in-
dustry’s growth, the coops sold much of their output through middlemen
of various kinds, so forward integration was unnecessary. Entry was also
facilitated by the existence of a trade association that developed uniform
product standards, advertised, gave technical support, and engaged in
research and development.
Most labor in the mills is semi-skilled, and Greenberg’s (1986) respon-
dents report that the majority of jobs can be learned in a few days. Jobs
are often rotated. The common view is that a worker-owner can do every
job in the plant except for specialized functions such as machine main-
tenance and accounting. All worker-owners normally receive equal pay,
and some coops attempt to equalize hours as well. This contrasts with
union contracts in conventional plywood mills, where the wages of the
highest paid workers differ from the lowest paid by a factor of about
2.5 (Craig and Pencavel, 1995: 133). The most common exception to the
rule of equal pay is the general manager, who is usually a hired out-
sider and receives substantially higher pay than members themselves.
Other exceptions include log buyers, sales managers, plant superinten-
dents and office managers. By all accounts, the coops rely heavily on
mutual monitoring among members to elicit work effort, and employ sub-
stantially fewer supervisors than similar conventional mills (Greenberg,
1984).
In 1931, a tax court ruled that Olympia Veneer was justified in pay-
ing higher wages than similar conventional mills because it had higher
labor productivity, even though high wages reduced the amount of in-
come subject to corporate taxation. Later coops have been able to avoid
corporate income taxes by paying wages, year-end bonuses, advances on
retained earnings, or “patronage distributions,” which exhaust annual net
income. The result is that no dividends are paid on shares. Net income is
distributed according to the hours worked by the members rather than
their individual capital investments. Historically, non-working share own-

ers have not participated in the earnings of the firm.
The major benefit to holding a share is a guarantee of being able to
work in the mill for whatever wage the firm is currently able to pay. Share
ownership thus functions like unemployment insurance. Members may
work elsewhere if the cooperative is paying low wages relative to their
outside alternatives, but they are sure to have a job if they want one, and
under normal conditions most owners do work in the coop. An owner
is entitled to a job in preference to non-owners, so if any non-owners
are currently employed, they can be fired to make room for an owner.
When low plywood prices lead conventional firms to lay off labor or shut
down a mill entirely, the coops first lay off non-members and then share
the available work among owners, cutting overtime and hourly wages if
necessary.
The governance structure of a typical coop is fairly simple. Ultimate
authority is vested in the general membership, which meets annually to
elect directors and more often to approve large capital expenditures, ad-
mission of a new member, or other specific proposals. Every member has
an equal vote regardless of invested capital. The directors must be owners
currently working in the mill. The board hires the general manager, who
has authority over day-to-day operations. There are also usually a num-
ber of committees dealing with innovation, grievances, and other matters.
Members are entitled to see whatever data or documents they wish, and
routinely exercise this right (Greenberg, 1986: 50). In general, the coops
exhibit a robust form of organizational democracy, where representative
institutions are reinforced by rather widespread participation on the part
of individual workers (Greenberg, 1984).
In all plywood coops, any resale of shares is limited to potential mill
workers. The firm usually has a first right to buy a departing owner’s share
at the current market price, defined as a bona fide offer to an owner, except
for transfers within the owner’s immediate family. If the firm chooses not
to exercise this right, the transfer to a new owner must be approved by a
majority either of the board of directors or a meeting of the membership.
A probationary period is normally required unless the new member is
already working in the mill. Despite these restrictions, shares are adver-
tised in local newspapers, and sell easily through private negotiation. Poor
firms that cannot buy back shares sometimes allow sales to outsiders in-
volving low down payments and long payoff periods. This practice, called
“renting” a share, is generally seen as compromising the incentive effects
of share ownership (Bellas, 1972: 25).
There is some disagreement in the literature over whether members

can be fired. Berman asserts that disobedience of a foreman’s orders is
cause for dismissal, or possibly a lesser penalty such as a fine or suspension,
and reports that “in most plants one or two shareholders have actually
been fired” (1967: 156). No managers interviewed by Berman indicated
that they felt unable to discharge low-productivity members. But writing
only five years later, Bellas (1972: 25) says that “the ability of the organi-
zation to discharge an owner is severely restricted,” and remarks that “an
incompetent or trouble-making owner, by using established procedures,
can often retain a job that a non-owner would have been discharged from,
for performing in a similar fashion” (p. 55). An expelled member cannot
work at the mill and must find a buyer for his share, which would otherwise
be worthless.
There is agreement that managers and foremen have a difficult time.
It is clear from Greenberg’s (1986) interviews that members feel free to
“thumb their noses” at foremen. At least in the 1960s and early 1970s,
turnover among managers was very high, and their job satisfaction was
lower than at conventional mills (Bellas, 1972). According to Berman,
managers in her interviews complained of an inability to plan for the
future: “One manager states that he is barred from considering any project
for improving the company with a payoff longer than two years” (1967:
164). Similarly, Bellas remarks that “a manager who establishes a two-
year investment plan on the day he is hired has only a 50 percent chance
of being around to see its results. Furthermore, the next manager may
disapprove of the plan and institute another” (1972: 79).
The cooperatives mainly manufacture sanded wood, which is more
labor-intensive and a higher quality product than sheathing or rough ply-
wood. Some produce their own veneer from logs, while others buy veneer
and process it into plywood. The technological aspects of plywood pro-
duction will not be summarized here, but Craig and Pencavel (1995: 137)
note that “there is no meaningful difference between the co-ops and the
conventional firms in the vintage of machines and technology of produc-
tion.” All observers appear to share Greenberg’s view that “the produc-
tion process in cooperative and conventional mills is indistinguishable; it
is universally noisy, dirty, dangerous, monotonous, and relentless” (1986:
81). Incentive pay is occasionally used to induce members to accept par-
ticularly bad jobs, although seniority seems to be a more common basis
for such decisions.
The typical size of a coop mill and a conventional mill as measured
by number of workers is nearly identical, and falls in the 200–300 range
(Craig and Pencavel, 1995). This figure includes hired non-owners in the
case of the cooperatives. In the coops studied by Greenberg, 16–36 percent
of the workforce were non-members. These hired workers consisted pri-
marily of (1) the plant manager and foremen; (2) office staff; (3) workers
in geographically segregated operations such as logging, veneer produc-
tion, and outlet stores; (4) newcomers who planned to buy a share, in
some cases the sons of current members; (5) a few highly skilled (and
highly paid) tradespeople such as electricians and specialists in machine
maintenance; and (6) relatively transient younger workers (1986: 60–61).
Further information on the role of non-member labor can be found in
Berman (1982: 81–85).
The plywood industry is characterized by input and output prices that
are highly volatile, both cyclically and from year to year. Each mill pro-
duces a small fraction of total output, and is a price-taker for its inputs
and outputs. The real price of logs (the crucial raw material) has gradually
risen in the Pacific Northwest over the last four decades as a result of the
depletion of old growth forest and increasingly stringent environmental
regulations (Craig and Pencavel, 1995). Entry by newer mills in the south-
ern United States has simultaneously put downward pressure on plywood
prices. In 1960, 99 percent of U.S. plywood production came from the
Northwest, but by 1992 this figure had fallen to 23 percent (see Table 1 in
Craig and Pencavel, 1995). The Southern mills, operated mostly by large
corporations and including no cooperatives, have focused on sheathing,
which is less labor-intensive. The coops have specialized in sanded ply-
wood because it “puts a premium on worker effort” (Bellas, 1972: 30). The
coops have also responded to this new competition with greater mecha-
nization, changes in work organization, and efforts to cut back on wasted
materials.
Despite the lack of new entry, the gradual attrition among existing co-
operatives, and the broader decline of the industry in the Pacific North-
west, the plywood coops deserve to be counted as a success story. The
closure of several coops in the 1990s does not reflect any deficiency in
this organizational form, but rather a long-term contraction of the re-
gional industry, which has also taken a toll among conventional mills. The
remaining coops have survived for several decades in competition with
unionized mills run by large corporations (Weyerhauser, Georgia-Pacific,
Boise Cascade), as well as numerous smaller mills owned by individual
proprietors.
Katrina Berman, who has written extensively on the formative years
of the coops, believes certain features of the plywood industry proved
favorable during the peak period of 1940–55. These included moderate

plant size, labor costs that were a significant fraction of total cost, a fairly
narrow range of required skills, an ability to rely on external marketing
agencies, and the presence of a supportive trade association. The lack
of scale economies and the labor-intensive nature of production meant
that financing was within reach of many potential applicants for mem-
bership, and the narrow range of required skills helped widen the mar-
ket for shares. Technology made both the quantity and quality of output
dependent upon the care and judgment exercised by workers, and the
significance of labor costs gave the cooperatives a competitive advantage
they could exploit.
Berman points out that these characteristics are not unique to the
plywood industry, and stresses that other factors played an important role.
The U.S. Northwest had a history of experimentation with producers’ and
consumers’ cooperatives; the Scandinavian origins of the pioneers gave
them a common cultural heritage; Olympia Veneer was organized by
unusually able individuals; and the successes of the first several coops
provided a template that later organizers were able to draw on. Although
Berman does not emphasize the point, the geographical clustering of the
coops evidently helped to create a critical mass. For example, lenders
gained familiarity with the coops over time, and the secondary market for
membership shares was relatively active.
3.3 The Mondragon Cooperatives

The renowned Mondragon group, taken as a whole, has developed into the
world’s first transnational labor-managed firm. My account is drawn from
books by Thomas and Logan (1982), Bradley and Gelb (1983), Wiener
(1987), Whyte and Whyte (1991), and Cheney (1999). I divide the nar-
rative into two phases, pre-1985 and post-1985, because major structural
changes were instituted in the late 1980s and early 1990s.
Don Jose Maria Arizmendi-Arrieta was a Basque priest who arrived
in the small industrial town of Mondragon in 1941, after supporting the
Republican side in the Spanish civil war and narrowly escaping execu-
tion by Franco’s forces. In 1943, Arizmendi opened a technical training
school whose graduates took local factory jobs. After failing to democra-
tize the major local employer, five of these graduates organized a firm of
their own, Ulgor, in 1954 as a limited liability company. Initial financing
came from relatives of the organizers and local residents who sympa-
thized with their ideals. Contributions were motivated by the prestige of
Arizmendi and the Technical Training School graduates, along with the
prospect that the firm would provide jobs in the community. The founders
bought a small workshop and with twenty-four employees began produc-
ing paraffin-fired cooking stoves and heaters. In 1956, this group built a
new facility in the town of Mondragon, located in a province of the Basque
region of northern Spain. Two years later, Ulgor was reorganized under
the Spanish legislation governing cooperatives.
Over the next few years, Ulgor began to diversify by licensing products
from other countries, including butane-fired cookers from Italy and elec-
trical products from Germany. A casting shop and foundry were created in
1958. At this time, several other coops were being created independently
through the transformation of conventional firms in the region. It became
clear that all of the coops shared problems involving inadequate access
to capital and expertise, as well as the fact that under Spanish law they
were excluded from the state-funded social security system. Moreover,
“outside sources of loans could not be offered adequate guarantees in the
form of collateral or equity participation, as both were legally proscribed”
under Spanish cooperative legislation (Thomas and Logan, 1982: 21).
After considerable research, Arizmendi discovered that the coops
could legally create a support organization that would attract local sav-
ings and lend these funds to the coops. Ulgor, two other manufacturing
coops, and a consumers’ coop affiliated in 1959 to create the Caja Laboral
Popular (CLP), the bank that later became the pivotal institution of the
Mondragon group. Other cooperatives could apply for membership by
signing a Contract of Association with the CLP, and would receive finan-
cial and technical assistance (a sample Contract of Association appears
in Appendix A of Wiener, 1987). The contract required a coop to submit
a balance sheet and annual budget to the CLP, along with monthly up-
dates in a standard format. The CLP engaged in constant monitoring of
financial performance, with each coop being comprehensively audited at
least once every four years. Affiliated coops conducted all banking and
financial operations through the CLP, and deposited their surplus funds
there. The CLP’s functions in the areas of pensions and insurance were
spun off into a separate cooperative in 1967.
Other rules in the Contract of Association imposed a uniform internal
structure on member cooperatives. Authority within a cooperative was
vested in the General Assembly of its members, which met at least once
per year. All decisions were on the basis of one vote per person, with no
one other than a working member having a vote. The Assembly elected
a Governing Council of nine members from within its ranks to four-year
terms. This Council met at least once per month and appointed managers
(often from outside the coop) as well as the immediate subordinates of
these managers. Managers carried out their administrative duties for a
term of four years, with renewal contingent on a performance review,
and could not be dismissed except for extraordinary reasons and with the
approval of the General Assembly. Only in exceptional circumstances was
a cooperative allowed to employ non-members, and these were limited to
10 percent of the workforce (usually people with special skills who were
needed temporarily).
The governance structure also included several bodies that were
mainly, or entirely, advisory in nature. A Social Council elected by the
membership was to deal with safety, social security, wages, grievances,
and related issues. The Management Council consisted of top managers
and executives, perhaps augmented by relevant outsiders, and provided
advice to the Governing Council. Finally, a “Watchdog Council” consist-
ing of three elected members of the general assembly was to monitor
everyone else, with the power to call on the CLP for support if necessary.
The structure of the CLP itself was similar, except that its General Assem-
bly included not only its own workers, but also representatives of other
coops in the group as well as individual members of other coops. This
system was set up in order to make it difficult for any one constituency to
dominate the policies of the bank. Most of the organizational structure de-
scribed here still exists within each cooperative in the Mondragon group,
although the administrative superstructure for the group as a whole and
the role of the bank have undergone extensive renovations, as discussed
later.
Applicants were to be screened carefully not only for skill and edu-
cation but also for their degree of social integration into the community,
with a six-month probationary period. There was a strong emphasis on
socialization into the ethics of cooperation. New members were required
to pay an up-front fee to finance the investment expense associated with
their workplace. According to Thomas and Logan (1982: 149), the entry
fee in 1958 was roughly twice the average annual earnings, but by 1977
it had declined to five to six months of earnings. They estimated in the
early 1980s that this fee covered only about 20 percent of the investment
cost per new worker. Bradley and Gelb (1983) estimated at the time of
their writing that the fee was about one year’s earnings for entrants at the
lower wage levels. Most interviewees in their control group of employees
at conventional firms cited the entry fee as the primary barrier to joining
Mondragon, along with the inability to withdraw their investment later.
Bradley and Gelb concluded that the fee screened out workers who did
not intend to stay, as well as those having a past history of unemployment
(1983: 76).
Workers who could not afford to pay the entire entry fee were allowed
to make a down payment (typically 25 percent) accompanied by wage
deductions during the first two years of membership. Most of the entry fee
was credited to an interest-bearing capital account set up in the worker’s
name that was adjusted annually for inflation. Workers were not permitted
to withdraw the principal from this account until retirement, but received
interest payments largely in the form of cash. On retirement, the total
value had to be distributed within two years under Spanish law. Voluntary
departure could result in a penalty of up to 30 percent, although this was
discretionary “and imposed only when capital withdrawal [was] seen as a
threat to the enterprise” (Bradley and Gelb, 1983: 18).
Wages formally were viewed as advances on anticipated profits. At
the outset, the maximum wage in any coop was limited to three times
the lowest wage, although special bonuses to managers could raise this
to a factor of 4.5. The general pattern was that the lowest paid workers
would receive slightly more than comparable workers in other local fac-
tories, and people in the middle range about the same, while highly skilled
workers and upper managers would receive half or less of what they could
earn elsewhere. Bradley and Gelb found a high level of forced managerial
turnover: In one year, managers in twelve out of eighty coops might be
dismissed (1983: 62–63). They suggested that managers refrained from
quitting the group altogether largely for social, moral, and ideological
reasons.
The position of any individual member on the wage scale depended
on a system of points, which were awarded for qualifications, responsibil-
ity, hard or dangerous work, and social integration. The precise weights
varied across coops depending on circumstances. There was also a goal of
preserving approximately equal average earnings across individual coop-
eratives regardless of their product line, although members of successful
coops would accumulate more in their individual capital accounts than
those in coops with less success.
After deducting wages, interest, and depreciation from net income,
10 percent of the remaining profit was given to a Social Fund sponsor-
ing community projects, a minimum of 20 percent was allocated to a
Collective Reserve Fund, and the remainder (up to a maximum of 70 per-
cent) was credited to the individual capital accounts of the members in
proportion to their wages (not their capital stakes). As the level of profit
increased, the fraction set aside for mandatory reserves grew. When losses
occurred, no more than 30 percent of these losses could be covered out
of reserves; any remaining losses were debited from individual accounts.
The 1960s and 1970s were a period of very rapid growth at Mondragon.
By 1979, there were 135 coops: 74 industrial (producing machine tools and
electronic equipment, consumer durables such as refrigerators, and vari-
ous other goods), 1 retail, 5 agricultural, 36 educational, 14 housing, and
5 coops devoted to diverse services including research and development.
The latter reflected a shift away from licensing foreign technology in favor
of in-house innovation. Mondragon avoided primary extractive activities
and raw material processing, as well as the chemical industry. The total
workforce was now over 15,000, with the average size of an industrial coop
at about 225, and coops accounted for 12.5 percent of industrial employ-
ment in Guipuzcoa province. Cooperatives in the group were scattered
throughout the Basque provinces, although the nucleus was still in the
town of Mondragon. At this time, the Caja Laboral Popular had ninety-
three branches and about 300,000 deposit accounts (Thomas and Logan,
1982).
New cooperatives could be (and were) created in any of five ways:
(1) a group of founders could organize a firm and ask to sign a Contract
of Association; (2) an existing coop could spin off a division to avoid ex-
cess size; (3) the management services division of the CLP, established in
1969, could identify a suitable entrepreneurial opportunity; (4) an exist-
ing coop could request affiliation; or (5) an existing capitalist firm could
be converted into a cooperative. In 1979, there were two new coops of
type 3, two of type 4, and four of type 5, with eight other possible new
entrants under study. Of the jobs existing at Mondragon, 10 percent had
resulted from the conversion of capitalist firms, while the remainder had
been generated internally. The objective of the CLP management services
division at that time was to add 10 new coops per year, 6 by converting
conventional firms and 4 to be created de novo. Of the 103 coops estab-
lished through 1986, only 3 had shut down by the early 1990s (Whyte and
Whyte, 1991: 3).
A turning point came in 1974, when the Mondragon group suffered
its first and still its only strike. In response to a job evaluation program
at Ulgor that led to a reduction in pay for 22 percent of all jobs, about
400 workers went on strike. The Governing Council, with the support of
the General Assembly, fired seventeen leaders (mostly women) and dis-
ciplined the others. In 1977, the strike organizers were finally reinstated.
Among other consequences, this experience convinced Mondragon’s
leadership that Ulgor was too large (3,250 workers at the time), and that
future cooperatives should not grow past 350–500 members in order to
preserve a suitably participatory environment. Instead, scale economies
would be gained by gathering individual coops into groups defined by
geography or product line. Despite this 1974 episode, serious disciplinary
problems are reportedly rare. Penalties can include suspension, loss of
income for up to sixty days, or in extreme cases expulsion.
Shop floor conditions at Mondragon have never differed dramatically
from those in conventional firms. Bradley and Gelb (1983) reported that
the supervisory structures they observed at Mondragon were virtually
identical to those at two nearby conventional firms. Thomas and Logan
remarked, “Mondragon cooperatives are [not] able to offer on a huge
scale less monotonous and more interesting work than elsewhere . . . exist-
ing technologies severely restrict the experiments that can be undertaken
in this respect. The necessity to compete in national and international
markets leaves insufficient space to implement alternative manners of
work organization on a large scale.” (1982: 70). Recent observers tend to
agree that in the name of international competitiveness, the Mondragon
cooperatives have imitated organizational trends fashionable elsewhere
(Cheney, 1999).
In the early 1980s, the Mondragon group faced its first real adversity
in the form of a world-wide recession. Although the coops had never
adopted a formal policy of lifetime employment, there had been no lay-
offs in any coop prior to 1983 (Bradley and Gelb, 1983: 34), and the lead-
ership went to great lengths to ensure that their extent was minimized.
The strategy for saving jobs was to maintain output while accepting low
margins, which helped in gaining market share internationally; take re-
ductions in personal income; and reallocate workers from weak coops to
stronger ones. Before a coop was allowed to transfer any of its members
elsewhere, it was required to reduce or eliminate the distribution of profit
to its members’ accounts and to cut wages. This policy resulted in 1,500
temporary transfers and 400 permanent transfers, with only 30 members
out of 18,000 becoming dependent on Mondragon’s internal system of
unemployment insurance (Wiener, 1987: 17).
A major change in Mondragon’s administrative superstructure be-
gan to take shape in 1984–1985. A new supervisory body was created,
the Cooperative Congress, along with a new top management body, the
Council of Groups. Head office functions were still to be performed by
the management services division (Division Empresarial), but this was
spun off from the CLP and made accountable to the Council of Groups.
Coops were organized in a matrix structure according to both geographic

region and product line. In particular, seventy industrial coops were
placed in twelve regional groups that dealt with shared personnel, legal,
and financial issues. At the same time, coops were assigned to sectoral
(product line) groups to avoid competition among them and gain scale
economies.
The Congress has 350 members elected by individual coops in rough
proportion to the size of their memberships, meets every two years, and
provides policy guidance to the Council of Groups on the entry of new
coops, social security issues, training, R&D, and external relations. A
standing subcommittee ensures that decisions of the Congress will be
implemented by coops. The Council of Groups wields executive author-
ity over all coops, subject to the policies formulated by the Congress, and
arbitrates among them if necessary. The Council includes the chair of the
Congress; the directors of the regional groups; the directors of the coops
responsible for retailing, banking, social security, and research; and the
director of management services. Council members have votes in propor-
tion to the size of the group they represent. Despite this centralization, the
individual coops remain separate legal entities that control management
through the Congress (Wiener, 1987: 21–23).
In December 1991, on the eve of tariff elimination throughout the
European Union, the Cooperative Congress pushed ahead with the re-
grouping of coops along product rather than geographical lines. The
umbrella organization was renamed “Mondragon Cooperative Corpora-
tion” (MCC), and members were renamed “associates.” MCC’s product
lines were narrowed to focus on core activities, and the old management-
services division of the Caja Laboral was spun off as a separate coop (LKS)
providing financial, organizational, and technical consulting services.
Salaries of top managers were allowed to rise to 70 percent of the out-
side market level (in 1987 the Congress permitted each cooperative to
pay a maximum wage equal to six times the lowest wage, but this still left
top managers earning about half what they could obtain elsewhere). Only
about thirty managers were actually eligible for the 70 percent provision,
but this symbolic linkage to the external market provoked much contro-
versy within the group. The coops were also given greater flexibility in
using contract employees, who were typically paid about 80 percent of
what associates received and enjoyed no share in profits (Cheney, 1999:
86). A new class of temporary associates, who were usually hired for three
years, was created. MCC had 300 temporary associates in 1998, many of
whom became full members as expansion proceeded.
MCC now has about 150 member cooperatives, 42,000 employees,

$7 billion U.S. in annual sales, accounts for about 30 percent of all eco-
nomic activity in the Basque region, and is the tenth largest private firm in
Spain (Cheney, 1999: 11). MCC includes dozens of manufacturing firms,
including the leading Spanish tool and die maker and the leading refrig-
erator manufacturer. It also has representation or has acquired firms in
about twenty other countries (Cheney, 1999: 42), including China, Mexico,
Argentina, Morocco, India, Brazil, and Colombia. MCC has shifted its
growth strategy away from starting new coops in favor of acquiring cap-
italist firms and converting them to cooperative status. However, this
policy has not yet been applied to firms acquired in other countries, a sit-
uation that the managers interviewed by Cheney (1999) generally viewed
as anomalous.
The sweeping organizational changes of the 1990s did not go unchal-
lenged. Most member coops voted to become part of MCC after the
decisions of the Congress in 1991, but a few seceded to form a new group
called Ulma with about 1,200 members. These coops, located in the orig-
inal Mondragon heartland, wanted to preserve local autonomy and the
principle of geographical clustering. Opposition also came from a quasi-
union, KT, which was organized in the early 1980s and gained much of its
support in the Fagor group, the descendent of the original Ulgor cooper-
ative. The members of KT favored greater local autonomy, orientation of
coops along regional rather than product lines, solidaristic wages rather
than tying wages to the outside market, and more emphasis on the social
aspects of the cooperative movement rather than on becoming competi-
tive internationally.
Throughout the 1990s, internal debates occurred over whether
Mondragon had “sold its soul” or whether democratic governance was
still in good working order. Cheney cites the declining importance of gen-
eral assemblies, the increasing influence of the management councils, and
a stronger emphasis on participation at the job level rather than at the
firm or system level as causes for concern (1999: 148). Mondragon has
clearly retained democratic internal structures at the level of the individ-
ual coops, and in a less direct way at the level of the MCC as a whole. But
formal structures can lose their operational relevance over time. It re-
mains to be seen whether the growing distance of top managers from the
rank and file can be reconciled with the democratic principles on which
the movement was founded.
What made Mondragon such a success? Almost all observers mention
the political and cultural environment of the Basque region in the 1940s
and 1950s. The region was never fully integrated into the Spanish state
and had been on the losing side of the Spanish civil war. Before the war,
regional labor unions and the Basque Nationalist Party supported the co-
operative movement. While trade unions and the Basque language were
repressed under the Franco dictatorship, cooperatives were not, leaving
a unique outlet for local economic, political, and social organization.
The region was evidently fertile ground: In 1972, the Basque region had
144 workers’ cooperatives not formally affiliated with the Mondragon
system (Whyte and Whyte, 1991: 11–12). In the 1980s, after Spain had
become a democracy, Bradley and Gelb (1983: 53) called the cooperatives
“intensely nationalistic.” Apart from ethnic homogeneity, Bradley and
Gelb point to some other factors, including physical isolation, a high
saving rate, and a history of small-scale credit unions (1983: 64–65). The
importance of these factors has been controversial. Wiener, for example,
argues that local circumstances were “far from conducive to success”
(1987: 67).
A second set of explanations centers around the organizational design
of the coops themselves. The constitutional rules of the group ensured a
high level of retained earnings by forcing members to allocate a large pro-
portion of profit to collective reserves, and made it difficult to withdraw
funds from individual accounts. Combined with a strategy that kept wages
from rising relative to the external labor market, this implied that any
productivity advantage at Mondragon would be translated into saving, in-
vestment, and expansion. The entry fee not only helped to finance capital
investment, but screened out applicants lacking a long-term commitment
to the enterprise. The absence of tradeable shares made the coops virtu-
ally invulnerable to outside takeover, although Whyte and Whyte (1991:
209, 223) do report that two small coops were bought out by private firms.
Even these design features might not have ensured success for an
isolated coop. Everyone agrees that Mondragon’s cooperative bank,
the CLP, was vital to its expansion. The CLP mobilized the savings of
thousands of depositors and funneled them into the growth of the coops.
The CLP also tightly monitored the performance of individual coops and
supplied technical expertise. When difficulties did emerge, it could put
a representative on a troubled coop’s Governing Council, replace its
management, or impose a recovery plan. Finally, the CLP adopted an
entrepreneurial role, identifying new opportunities and making strate-
gic decisions on behalf of the group. “If one factor can be singled out
as responsible for the success of the Mondragon co-operatives it is that
even the smallest of them is both assisted and disciplined by a battery of
management services on a scale appropriate to a company as big as the

whole Group” (Wiener, 1987: 32).
Part of the reason for Mondragon’s explosive growth in its first decades
appears to be that the CLP had too much capital rather than too little.
A quirk in Spanish banking laws of the 1950s identified by the founder
Arizmendi was that a cooperative bank was allowed to pay a slightly
higher interest rate to depositors than other banks, giving it an advan-
tage in attracting savings. Thomas and Logan observe that the CLP was
simultaneously subject to regulatory limits on the kinds of investments
it was permitted to make, so that apart from lending to the coops them-
selves, the only investment option was officially approved bonds carrying
a low rate of interest (1982: 79–80). The result was that “the Group’s
growth in the late sixties and early seventies was for all practical purposes
bank driven, and it is hard to imagine that the cooperatives would have
been able to finance their development from other more conventional
sources” (Wiener, 1987: 25). Eventually the funds available to the CLP
outstripped the investment opportunities available to the coops (Whyte
and Whyte, 1991: 199), leading the CLP to seek regulatory approval for
lending to conventional firms in the region. One reason for the reorgani-
zation of the Mondragon group beginning in the mid-1980s was the fear
by CLP personnel that affiliated coops would benefit from a soft bud-
get constraint, rather than facing the same degree of scrutiny as external
borrowers.
Under its present organizational structure, the Mondragon Cooper-
ative Corporation seems able to grow almost indefinitely through the
absorption of additional firms and their conversion to cooperative form,
much as conventional global corporations acquire and spin off divisions.
Even so, there is only one Mondragon. Given the widespread attention
this template has received in recent decades, particularly from proponents
of workers’ control, it is puzzling that there have not been successful repli-
cations. One may infer that although in its mature form the MCC structure
is eminently viable, the odds are stacked against the spontaneous emer-
gence of a strong alliance of producer cooperatives operating under the
discipline of a common financial institution.
4
Workers’ Control in Action (II)
4.1 The Lega Cooperatives

Italy has the largest number of workers’ cooperatives in the Western
world, and the largest fraction of the workforce employed by such firms.
Ammirato (1996: 319) reports that in 1989 there were 10,445 workers’
cooperatives affiliated with one of four national federations, along with
many others not so affiliated. This is likely to be an underestimate because
cooperatives in the agricultural, housing, transport, and fishing sectors
have been excluded along with so-called “mixed” coops. But data on the
Italian coops are notoriously bad (Oakeshott, 1978; Zevi, 1982; Earle,
1986: 63–66), with Earle remarking that nobody really knows “how many
co-ops are genuine and operative, and how many are dormant, embryonic,
phantasmal or bogus” (1986: 203). For example, entirely conventional
firms have sometimes registered as cooperatives in order to gain access
to public subsidies. All numerical estimates must therefore be taken with
a generous serving of salt.
By contrast with the plywood cooperatives and Mondragon, which
developed with little state involvement, the Italian cooperative movement
was heavily politicized from the outset, and has enjoyed tax advantages
as well as preferential access to public land and contracts, job creation
programs, loans and grants, and the financial expertise of public banking
and research institutions. The national federations provide member co-
operatives with more support than the plywood coops ever derived from
their industry association, but have less formal authority than the central
agencies of the Mondragon group. Here I will focus on the oldest and
best-known federation, La Lega Nazionale delle Cooperative e Mutue
67
68 Workers’ Control in Action (II)
(National League of Cooperatives and Mutual Aid Societies), or in brief

the Lega. Except where noted, the following account relies on Ammirato
(1996).
In 1989, the Lega had 11,398 member coops (including consumer coops,
which are not generally organized under the principle of workers’ con-
trol) and 3.4 million individuals as members. It directly employed 230,734
workers and accounted for 2.73 percent of national GDP. In the region of
Emilia Romagna (which includes Bologna and Ravenna), where the Lega
has historically been the strongest, it employed 62,500 and accounted
for 12.75 percent of regional GDP. Manufacturing activities include,
among other things, ceramics, machinery, rubber products, furniture, and
high tech equipment. Non-manufacturing activities include construction,
agriculture, fishing, retailing, housing, transportation, food catering, and
health services. Some manufacturing and construction cooperatives in
the Lega have 500–3,000 workers, but even in these fields, smaller firms
are more common. Holmstrom (1989: 29) reports that in 1980 the Lega’s
construction cooperatives averaged 269 employees and engineering and
metalworking cooperatives averaged 135, with most other manufactur-
ing activities in the 40–75 range. Among Lega service firms, cleaning and
maintenance coops averaged 267 workers, restaurant coops averaged 188,
and the rest were much smaller.
The first Italian workers’ cooperative was organized in 1856 by eighty-
four glass blowers of Altare, a town in Liguria. This example was soon
followed by other cooperatives across northern Italy involving tailors,
printers, masons, carpenters, and carders. In Emilia Romagna, work-
ers established cooperatives to perform public works such as construct-
ing roads and erecting buildings. During 1884–1885, some workers from
Ravenna, members of the General Association of Landworkers, won a
public contract to drain the Ostia marshes near Rome. Hundreds died of
malaria in the process (Earle, 1986: chs. 1–2).
In this environment, the Lega was established in 1886 as the Federation
of Italian Cooperatives by 248 cooperatives, which together had 74,000
members. The current name was adopted in 1893. The central goal from
the start was to lobby government for tax concessions and access to public
works contracts. The leader of this organizational effort, Luigi Luzzatti,
was a member of Parliament who had drafted legislation, passed in 1886,
which gave legal status to cooperatives and required that elections to
the board of directors be based on one person, one vote. In 1896, 98
of the 254 coops affiliated with the Lega were producers’ cooperatives;
however, 3,698 coops of all kinds existed in Italy, and the vast majority did
not join the Lega. Luzzatti had risen to Treasury Minister in the national
govenment by 1899 and helped to pass legislation that exempted coops
from competitive bidding on small contracts and from posting bonds. He
also helped to establish the Lega’s first national financial institution, the
Institute for Cooperative Credit, in 1904. At the local level, Lega coops
were favored with public contracts by socialist-controlled councils.
A high water mark came in 1920 when there were about 8,000 coops
affiliated with the Lega, including 2,700 workers’ cooperatives, located
mostly in northern Italy. There were also roughly 7,000 coops affiliated
with the Catholic National Confederation, including 694 workers’ coop-
eratives. At this time, the president of Fiat reportedly offered to sell the
firm’s Turin plant to a cooperative, but the Lega was reluctant to assume
the associated financial burden (Roelants, 2000). Some writers doubt
the sincerity of this offer because it came at a time when the Turin plant
was under occupation by employees in a labor dispute. The proposal
was forgotten after a wage increase brought an end to the occupation
(Earle, 1986: 23–24). This period of rapid growth ended abruptly under
the Fascists between 1921 and 1943, when many coops were suppressed
and others were placed under tight political control. The movement re-
vived after World War II, and by 1947 the Lega had about 4,700 member
coops (again, many of these were consumer coops).
The postwar constitution made the state responsible for promoting en-
terprises in the cooperative sector. This was followed by the Basevi Law
of 1947 under which coops had to operate on the principle of one person,
one vote; maintain an “open door” policy for new members; ensure that
members were at least 50 percent of the workforce; and obey limits on the
capital stakes of members and the rate of interest paid on members’ capi-
tal contributions. Members’ shares were made non-transferable. At least
20 percent of firm profit had to be placed in a collective reserve fund,
which could not be recouped by individuals leaving the firm; no more
than 20 percent could be used to supplement wages; and the remainder
had to be used either for reinvestment or social security purposes. Pro-
fessionals were prohibited from creating cooperatives, and no more than
4 percent of a coop’s workforce could be white collar workers. In the
event of dissolution, any net assets went to public or charitable purposes
rather than to the members. These constraints were offset by some ad-
vantages: Workers’ cooperatives were exempted from taxes if labor costs
exceeded 60 percent of total costs, and from competitive bidding on small
public works contracts. Subject to these legislative rules coops were to
be governed by a general assembly of the membership, whose powers
included electing the board of directors, approving the firm’s budget, and
deciding the distribution of profits.
Because of the Lega’s affiliation with the Communist Party, relations
with the national government turned hostile between 1948 and 1962,
but cooperatives in the group were still favored by local governments
in Emilia Romagna, Tuscany, and Umbria, especially with respect to con-
tracts issued to construction cooperatives. Economic performance was
unimpressive in this period, a fact that Ammirato attributes to financial
constraints and an emphasis on political rather than economic problems.
Indeed, Oakeshott (1978) expressed concern that the Lega might be lit-
tle more than a tool of the Italian Communist Party. On the other hand,
Earle asserted just eight years later that “the days of using the League
as a ‘transmission belt’ for party orders ended twenty years ago” (1986:
36). Ammirato’s opinion is that the Lega gradually took a less ideological
hue between 1963 and 1980. In this phase, it created sectoral associations
(product line groups) at the national level along with territorial structures
at the regional level, further diversified its activities, and cultivated inter-
nal expertise in planning, marketing, and accounting. In 1963, the group
purchased an insurance company (Unipol) and in 1969 it created a finan-
cial consortium (Fincooper), which finally provided a reliable mechanism
for internal capital allocation.
The Lega sought scale economies in this period through mergers among
member coops and acquisitions of other coops or private firms. Between
1974 and 1978, for example, the Lega transformed 100 failing private
firms with about 10,000 employees into cooperatives. As of 1978, most
construction coops, which tended to be older, had been organized di-
rectly by workers; manufacturing coops, which tended to be newer, had
arisen mostly as a result of efforts to preserve jobs in failing capitalist
firms (Zevi, 1982). Zevi also notes that the construction coops obtained
the vast majority of their financing through retained earnings, with some
additional financing from members’ loans and very little from members’
share capital (the latter kinds of financing were subject to legislative ceil-
ings on the allowable rate of return; these ceilings were raised a few
years later as will be explained). Manufacturing cooperatives made more
use of bank debt, perhaps reflecting their origins as transformed private
firms, but internally generated funds were also the dominant financial re-
source in these firms. There was steady growth in the number of Lega
cooperatives, aggregate coop membership, and total sales throughout the
1970s, although the construction cooperatives remained by far the most
important producers’ cooperatives measured by number of firms as well
as employment. By the mid-1980s, the Lega as a whole had become the

fourth-highest export earner in Italy after Fiat, Montedison, and Olivetti.
This growth was aided by a relaxation of earlier legislative constraints,
beginning in 1971 with an increase in the amount that members could in-
vest in firm shares, and in the allowable percentage of white-collar work-
ers. Members were allowed to lend funds to their cooperatives for the first
time in 1973, with favorable tax treatment of the interest received, and
by the mid-1970s, profits destined for collective reserves were no longer
taxed. Further relaxations came in 1983 when the ceiling on the inter-
est paid on member loans and shares was raised from a flat 5 percent
to 2.5 percent above the rate on government bonds. This immediately
increased the allowable interest rate to 18.5 percent because of the high
nominal interest rates at the time. Coops were also allowed to use profits
to add to members’ capital stakes without taxation if this involved no
more than 10 percent of the previously existing amount, and the ceiling
on the amount members could invest in their firms was raised substan-
tially. Although the tax rate on interest paid to members was also raised
slightly, this was still only half the tax rate paid by bank depositors, and
coops were now permitted to issue bonds. The limit on white-collar per-
sonnel was simultaneously increased to 20 percent. The ceilings on loans
and share investments by coop members were again increased in 1992.
Outside equity shares (voting or non-voting) were authorized, as long as
the votes on these shares did not exceed 30 percent of the voting rights in
the general assembly. Non-voting shares were offered a return that was
2 percent higher than on members’ shares. This was accompanied by a
requirement that medium-to-large coops provide certified annual reports.
The number of manufacturing cooperatives in the Lega grew from
403 in 1980 to 660 in 1989, with most of this increase coming through
the conversion of failing private firms. In 1986, the three major Italian
cooperative federations and the three national trade union federations
created CFI (Campagnia Finanziara Industriale), a risk-capital company
that aimed to save the jobs of laid-off workers by converting bankrupt
small and medium-sized industrial enterprises into workers’ cooperatives.
From 1986 to 1994, CFI converted 115 such enterprises into cooperatives,
a process that involved 4,000 workers (Roelants, 2000). These efforts were
encouraged by the Marcora Law, which came into effect in 1987 and cre-
ated a fund to finance worker takeovers of failing firms. Employees in such
firms were required to invest some personal wealth, up to half of which
could take the form of foregone unemployment benefits. The state would
also provide grants to the firm in lieu of unemployment compensation.
Cooperatives assisting in a reorganization were eligible for grants if they

matched the amount invested in the reorganized firm by the state. From
1987 to 1991, 86 coops involving 3,254 employees were created in this
way, mostly in furniture, clothing, textiles and footwear, and printing.
About 2,000 cooperatives were members of the financial consortium
Fincooper by 1990, which along with Unipol (the sixth largest insurance
company in Italy) is now at the center of the Lega federation. Each coop-
erative makes a minimum required investment in Fincooper shares, and
may also make deposits there that are subject to half the normal tax rate
on interest income. Fincooper in turn extends or guarantees loans to the
individual cooperatives, monitors their financial performance, assists with
international transactions, provides consulting services, and makes equity
investments in companies important to the cooperative sector as a whole.
Unipol, whose shares began trading on a stock exchange in 1989, is 70 per-
cent owned by the cooperative movement and engages in numerous joint
ventures with other European insurance companies including Lagun Aro,
the insurance arm of the Mondragon group. Among other contributions
to the Lega, it has financed office space and industrial plants that were
leased to individual cooperatives. A similar leasing function for machinery
and equipment has been assigned to Unintesa, another Lega enterprise.
Because many Italian cooperatives are not ultimately controlled by
their workforces, it is difficult to obtain good data on the importance
of worker-controlled firms by industry. Looking at the cooperative sec-
tor as a whole (not just the Lega), workers’ cooperatives are estimated
to account for 8 percent of national sales in construction, and a share
of about 10 percent for pottery, woodworking, glass-making, and cer-
tain fields of mechanical engineering (Jones and Svejnar, 1985). In 1991,
worker-controlled firms accounted for 28 percent of employment in clean-
ing services, 16 percent in road transport of goods, 15 percent in vigilance
services, and 13 percent in welfare services (Pittatore and Turati, 2000).
The latter authors observe that these service industries have some char-
acteristics in common: Tasks are standardized and homogeneous, firms
need little capital, and investments are not highly specialized.
Professional services are often organized as partnerships rather than
cooperatives. However, the Lega does have a number of professional
cooperatives, which according to Ammirato by the 1990s included ap-
proximately 120 cooperatives organized by architects, engineers, urban
planners, and economists. Other cooperatives have been established for
managerial training, organizational consulting, research and develop-
ment, advertising, and taking over private firms. Consortia link sets of
related firms in agriculture, manufacturing and consumer retailing, as

well as other sets in construction and manufacturing.
A typical organizational structure within an individual cooperative
runs as follows (Holmstrom, 1989: ch. 3). Workers can join after a proba-
tionary period, usually subject to a requirement that a minimum amount
of capital be contributed to the firm. Members meet annually in a general
assembly to approve the budget and other major proposals. They also
elect a council whose membership serves for two years. The entire coun-
cil may turn over simultaneously, or half the council may come up for
election every year. This council in turn chooses a president from among
its number and appoints a professional management board to implement
its decisions. The board often includes a director, top managers, and sev-
eral major department heads. The council meets weekly or every other
week to oversee the activities of the management board. All employees,
whether or not they are coop members, also vote for a separate works
council, which represents them in their capacity as union members. In the
larger cooperatives, base pay is set through collective bargaining involv-
ing an industrial union, and closely parallels wage agreements with other
employer federations (Holmstrom, 1989: ch. 4).
Member cooperatives are affiliated with the Central Association of the
Lega. Each cooperative elects representatives to regional congresses. The
latter elect delegates to the national congress, which is held every four
years, under a system in which regions with a higher cooperative den-
sity have more voting power. Consortia also elect representatives to the
national congress directly. The national congress elects a general council,
which then chooses the board of directors, executive council, president,
and two vice presidents. The general council meets at least four times a
year, and the board of directors meets at least once a month. Day-to-day
operations of the Lega are handled by the executive council, president,
and vice presidents. Parallel governance structures are used within the
sectoral and territorial associations.
The Lega supervises member coops to be sure that they comply with
legislation and normative principles, and checks their annual reports. It
also provides legal, accounting, and industrial relations services, monitors
the economic performance of individual coops, undertakes comparisons
with private enterprise, facilitates networking on technical issues, repre-
sents the interests of member coops in lobbying government, and builds
alliances with political parties, trade unions, other cooperative federa-
tions, and small businesses. Other activities include assistance with the
creation of new cooperatives, de novo or by takeovers of private firms.
This entrepreneurial role is significant: In the Lega, “The problem caused

by cooperatives’ low rate of formation has been alleviated by promoting
cooperatives from worker takeovers of private enterprises” (Ammirato,
1996: 136).
Individual cooperatives are autonomous and are free to join or leave
the Lega. Not only does the Lega lack strong authority over member
coops, it similarly lacks authority over its sectoral and territorial associ-
ations as well as the consortia formed by these bodies. Ammirato plays
down the contemporary role of financial resources or expertise as factors
giving influence to the central association, stressing instead “prestige, a
common history and a political subculture spread throughout the move-
ment” (1996: 291). He further states that coordination is not achieved
within the Lega through hierarchy, but rather through a network of inter-
locking directorships that afford opportunities for informal consultation.
There have been a number of econometric studies of Italian workers’
cooperatives, but most have used samples that included only coops (Jones
and Svejnar, 1985; Estrin, Jones, and Svejnar, 1987), making it difficult to
compare such firms directly with capitalist competitors. One important ex-
ception is Bartlett, Cable, Estrin, Jones, and Smith (1992), who construct a
sample of 49 coops and 35 private firms in Tuscany and Emilia Romagna
matched by size (about 100 workers each) and sector (light manufac-
turing, which includes metalworking, woodworking, and clothing), using
survey data collected in 1985–86. Most of the coops are affiliated with
the Lega. Bartlett et al. find that the coops had (1) sharply compressed
wages relative to private firms, with blue collar workers having similar
wages but managers receiving substantially less; (2) significantly fewer
managers as a fraction of the workforce; (3) more hours worked per per-
son; (4) significantly lower fixed assets per worker; (5) significantly higher
value added per person, per hour worked, and per unit of fixed capital;
and (6) many fewer strikes, fewer days per worker lost to absenteeism,
and lower quit rates. Interestingly, the higher productivity in 5 did not
imply higher wages. Bartlett et al. suggest that it was absorbed through
higher returns on capital contributions of coop members or more rapid
depreciation of fixed assets.
Another possibility is that the productivity advantage of the coops
is illusory. Upon estimating production functions using the same data
set, Estrin (1991) found no significant difference in productivity between
coops and private firms when labor input was proxied by the number
of employees, and significantly lower productivity for the coops when
labor input was proxied by total blue-collar hours worked. Estrin suggests
reasonably that the latter result occurs because blue-collar labor is used
more intensively by the coops than by the private firms, and thus provides
a poor measure of overall labor input. Even granting this, Estrin was still
unable to detect superior productivity among the coops.
To what degree can one generalize from the success of the Lega? One
lesson is that it is possible to create a large, vibrant cooperative sector if
the state is prepared to provide sustained legislative and financial support.
The Lega experience also refutes the idea that such firms suffer from deep
organizational flaws that predispose them to failure. On the other hand,
it is not surprising that granting sufficiently large subsidies for workers’
control would lead to the creation of many worker-controlled firms. Al-
though it is far from clear to what degree the Lega’s success is explainable
by such interventions (other firms in the private and public sectors also
enjoy substantial state support in Italy), skeptics remain free to argue that
the social benefits derived from a large cooperative sector are too modest
to justify the expenditures needed to create and sustain it.
I will not attempt any such cost-benefit comparison here. More im-
portant for my purposes is the question of whether the Lega provides
evidence that is useful in explaining why workers’ control is normally
rare. Hansmann takes a negative view, commenting that “Italian worker
cooperatives . . . operate in an environment of favoritism and constraints
that make it difficult to draw general conclusions from them about the
behavior and viability of fully worker-owned firms” (1996: 104). I am less
pessimistic about the opportunities to learn from the Lega experience. The
cross-industry incidence of workers’ cooperatives described by Pittatore
and Turati (2000), for example, provides useful insights. The nature of the
supporting institutions that have been created to assist individual coops
within the Lega implicitly sheds light on the obstacles that isolated work-
ers’ cooperatives encounter. Perhaps most importantly, the Mondragon
and Lega cases both indicate that a successful workers’ cooperative sec-
tor requires an institutional mechanism for the routine creation of new
coops, either de novo or through conversion of capitalist firms. No such
mechanism existed after the initial burst of cooperative formation in the
U.S. plywood industry, and the result has been a gradual decline in the
population of coops during the last half century.
Hansmann is correct, though, in observing that the state has often
favored coops in some ways, while simultaneously imposing a wide range
of constraints on their design and behavior. Given the broad success of the
cooperative movement in Italy, it would be hard to argue that cooperatives
have been disadvantaged on balance. But the constraints should be kept
in mind: Either by direct legislation or in order to qualify for subsidies,

Italian coops have had to limit the size of wage differentials, restrict the
use of white-collar labor, allocate net income in prescribed ways, place
ceilings on capital contributions from their members, pay limited interest
on the loans and shares of members, forego transferable membership
rights, and (for much of their history) avoid issues of bonds or shares to
the general public. It is difficult to generalize from this case about the
design features that a less-constrained worker-controlled firm might find
useful or appropriate. The Lega cooperatives may have succeeded despite
their organizational structures, rather than because of them.
4.2 Employee Stock Ownership Plans

Employee stock ownership plans (ESOPs) developed in the United States
as an outgrowth of employer-sponsored retirement systems. A statu-
tory framework for ESOPs was created by the Employee Retirement
Income Security Act (ERISA) of 1974. In 1999, ESOPs held about $500
billion in assets, and other forms of employee ownership such as stock
options were worth about $300 billion more (National Center for Em-
ployee Ownership, 1999: 5). ESOPs thus represent just one element in a
broadly based trend toward increased employee ownership of U.S. cor-
porations. To keep the discussion focused specifically on ESOPs, I ig-
nore employee stock ownership through stock options, profit sharing,
and 401(k) plans.
Legally speaking, an ESOP is a defined-contribution employee benefit
plan that invests primarily in the stock of the company sponsoring
the plan. The firm may promise to make regular contributions on the
employee’s behalf, but the ultimate payout is uncertain and depends on
the performance of the firm’s shares. The value of an employee’s account
depends on the history of contributions made by the firm, changes in the
market value of the assets held by the ESOP, and any dividends on those
assets that have been retained in the account rather than being paid out
to the employee as cash.
In any ESOP, the firm creates a trust that acquires shares in the firm on
behalf of its employees. There are two general types of ESOP, depending
on the way in which these shares are financed. In a non-leveraged ESOP,
the firm either donates stock to the trust, or donates cash to buy stock from
current shareholders. A leveraged ESOP borrows from a bank or other
lender and uses this loan to buy stock from current shareholders. In a non-
leveraged ESOP, the firm’s contributions can be completely discretionary,
but in a leveraged ESOP they are fixed by the schedule of debt repayment,
usually occurring over five to ten years.
Federal law prohibits defined-benefit pension plans and most defined-
contribution plans from investing more than 10 percent of their assets
in company stock, but this ceiling does not apply to ESOPs, which can
own up to 100 percent of the firm’s shares. Normally, any full time em-
ployee who is twenty-one or older and has had at least one year of service
with the firm is eligible to participate. When other eligibility criteria are
used, there are legal rules to ensure that firms do not discriminate in
favor of highly paid employees such as top management. ESOPS may
exclude non-resident aliens, employees in a separate line of business, or
union members as long as retirement benefits have been the subject of
good-faith bargaining.
Similar non-discrimination rules govern the ways in which firms can
allocate ESOP shares to the accounts of individual workers. The usual
procedure is for such allocations to be proportional to salary or some
other measure of compensation. In principle, however, a range of other
formulas can be used, including equal per capita amounts or credits based
on hours of work or duration of service. In a non-leveraged ESOP, shares
are allocated to each eligible employee when the firm contributes to the
plan, but in a leveraged ESOP, shares are released to individual accounts
as the underlying loan is repaid. When an employee leaves the firm, the
ESOP pays out the value of that person’s account in the form of cash
or stock. Vesting rules specify the minimum time an employee must stay
with the firm to avoid forfeiture of this payout. Common formulas are
100 percent vesting after five years, or 20 percent vesting a year starting
in the third year.
Contributions to an ESOP can be deducted from corporate taxes up to
a maximum of 25 percent of annual payroll for eligible employees. In the
leveraged case, this includes a tax deduction for repayment of the principal
of the loan used to finance ESOP stock purchases, on the grounds that
such loans merely finance deferred employee benefits, which would be
treated as a deductible expense in any case. More important is the fact
that firms can deduct dividend payments on shares held by the ESOP
(whether or not these shares have already been allocated to individual
accounts, and whether or not the dividends are passed through as cash
payments to employees). The firm can also deduct any dividends used
to repay the loans through which shares were purchased by a leveraged
ESOP. Employees are thus a preferred source of equity finance since
they are the only potential shareholders not subject to double taxation
of dividends (in the normal case, dividends are first subject to corporate
income tax and then again to personal income taxes paid by the individual
shareholder).
The main role of ESOPs in closely held firms is to create a market
for the shares of small business owners who lack a family successor and
do not want to sell out to a larger firm, or who prefer to phase out their
participation in the firm gradually. Indeed, about half of ESOPs in closely
held firms are used to buy out an owner (NCEO 1999: 4). Since 1984, small
business owners who sell more than 30 percent of their firm to an ESOP
have been exempt from capital gain taxes if the proceeds of the sale are
reinvested in the securities of other U.S. corporations. The value of such
assets must be assessed by a qualified independent appraiser, and will
generally be adjusted upward if the ESOP will achieve majority control
or the shares are readily marketable. Publicly traded firms can also use
ESOPs to cash out the shares of large stockholders who prefer to avoid
transactions on the open market.
Under federal law, in a publicly traded company voting rights on the
shares held by a non-leveraged ESOP must be passed through to employ-
ees. This extends to all issues on which shareholders would normally vote,
including elections to the board of directors. The same is true for lever-
aged ESOPs on those shares already allocated to individual accounts.
Voting rights on any shares not yet allocated are held by a trustee who
is usually appointed by management, but has a fiduciary responsibility to
act in the interest of employees. Since in a leveraged ESOP shares are
transferred to employee accounts as the loan used to finance the ESOP
is repaid, the rate at which employees acquire votes depends on the rate
at which the firm retires this debt. Many publicly traded firms have im-
plemented a policy of “mirror voting,” where employees cast confidential
votes on allocated shares and the trustee votes unallocated shares in the
same proportions.
Employee voting rights are weaker in closely held firms. Here, employ-
ees are only entitled by law to vote on major proposals for which approval
by a supermajority is usually needed, such as merger, sale, or liquidation
of the firm. The firm is under no obligation to permit employee voting
with regard to the board of directors, and most privately held firms appear
not to do this. As is the case for a publicly traded firm, voting rights on
unallocated shares in a leveraged ESOP are retained by a trustee.
In 1999, there were about 11,000 ESOPs, with 90 percent of them
in privately held firms and 10 percent in firms that are publicly traded
(NCEO, 1999). This is an increase from 8,000 ESOPs in 1987 (Blasi, 1988).
According to the National Center for Employee Ownership (NCEO),

employee ownership in 1999 accounted for 8 percent of corporate stock,
up from 3.87 percent in 1983 (Blasi, 1988). These figures include not just
ESOPs but also 401(k) plans, stock options, and other forms of employee
ownership, although, as previously mentioned, ESOPs account for at least
half of the shareholdings of employees.
Blasi (1988) estimated that in 1987, about 1,000–1,500 ESOPs had ma-
jority employee ownership. These mainly involved privately held firms,
although not necessarily firms that were small in size: 12 firms among the
400 largest privately held companies in the United States were regarded
by the NCEO as having majority employee ownership (Blasi, 1988: 75).
Minority employee ownership was more common. In the early 1990s,
about 9,000 closely held firms had an employee ownership stake of at
least 4 percent through one or more employee benefit plans, including
ESOPs (Blasi and Kruse, 1991: 11).
Reliable information on employee ownership is difficult to obtain
even for publicly traded firms, but the number of such firms with sig-
nificant employee ownership has clearly been growing over the last two
decades. An exhaustive search by Blair, Kruse, and Blasi (2000) identified
twenty-seven publicly traded U.S. corporations that in 1983 had 20 per-
cent or more of their shares owned by employees. These vanguard firms
were distributed across a range of sectors, including manufacturing, trade,
finance, and services; none were in the mining, construction, or chemical
industries. Five were major defense contractors. By 1990, there were 148
publicly traded companies in the United States with similar levels of em-
ployee ownership (Blasi and Kruse, 1991, Appendix A). The distribution
of the latter firms across industries closely resembled the corresponding
distribution for the universe of publicly traded firms as a whole (Blair,
Kruse, and Blasi, 2000: 283).
Although not all of the vanguard firms were large, Blasi (1988: 90)
discovered that eleven firms on the Fortune 500 list were approaching
20 percent ownership by employees in 1981. About a decade later, there
were nineteen firms on the Fortune 500 list with employee stakes of 20 per-
cent or more, and another twenty firms were approaching this level (Blasi
and Kruse, 1991, Table 5). A number of large publicly traded firms in
the financial, transportation, and retail sectors had comparable employee
ownership by 1990. Even when employee claims do not reach such a high
threshold, they can still be significant. Blasi and Kruse (1991) found that
of the 7,000 or so U.S. corporations whose stock was traded on a pub-
lic exchange in 1989, about 1,000 had employee ownership of 4 percent
or more. Often this is the largest single block of shares, although it is

questionable whether employees are a cohesive voting group.
What are the principal motivations for firms to establish ESOPs? Apart
from the tax advantages discussed earlier, and the usefulness of ESOPs
in closely held firms as a way of buying out retiring owners, several other
motivations have been discussed. These include the use of ESOPs as a de-
fense against takeovers and as a way of saving a failing firm by transferring
ownership to employees (usually in a context where the employees are
taking substantial pay cuts). While many examples of both kinds have re-
ceived extensive media coverage, Blasi and Kruse (1991) emphasize that
a large majority of ESOPs in publicly traded firms were not created in
response to an identifiable takeover threat, and that a very small fraction
have involved a full employee buyout of a failing firm or resuscitation of a
plant on the verge of shutdown. Blair, Kruse, and Blasi (2000) have, how-
ever, found that ESOPs promote stability in the sense that they lower the
rate of firm disappearance through takeover or bankruptcy. It is also clear
that in specific industries facing import competition or deregulation, such
as steel, shipbuilding, trucking, airlines, and health care, ESOPs have of-
ten played a prominent role in employee buyouts or concessionary wage
bargaining by unions (see Blasi and Kruse, 1991, Appendix F).
Most of the ESOPs in publicly traded firms have arisen as part of a
larger corporate financial strategy to reorganize pension programs such
as defined-benefit and profit-sharing plans in tax-favored ways. This mo-
tive is consistent with the broad distribution of ESOPs across industries
reported by Blair, Kruse, and Blasi (2000). In particular, ESOPs do not
conform to the stereotype that employee ownership can only thrive in
industries with small minimum efficient scale or very labor-intensive pro-
duction methods. Of course, the creation of an ESOP may simultaneously
have side effects that are attractive to management, such as deterring
takeover threats or enhancing productivity within the firm.
Blasi and Kruse (1991: 42) note that tax subsidies can usually offset
only about 30–40 percent of a firm’s ESOP contributions. This raises
the interesting question of who ultimately bears the after-subsidy cost
of the shares held on behalf of employees in an ESOP. There are three
possibilities. The first is that claims of existing shareholders are reduced
in value by the issue of new shares without any corresponding increase in
earnings; by reductions in the cash balances of the firm; or by a reduction in
net earnings due to the debt incurred to finance employee share purchases.
It is unlikely that dilution in the claims of shareholders is a major factor,
because other than in the context of a takeover bid there is little evidence
that the market value of a firm falls when an ESOP is announced. Indeed,
the firm’s stock market value often appears to rise (Blasi and Kruse,
1991: 181–83). The second possibility is that an ESOP may be partly self-
financing if it raises productivity – for example, by means of stronger work
incentives, more information-sharing, reduced conflict within the firm, or
better monitoring of managers by employees. The empirical evidence for
this view is also weak, at least for firms that do not substantially alter their
organizational cultures at the same time (see Blasi, 1990; Conte and
Svejnar, 1990; and Kruse and Blasi, 1997).
The remaining potential source of ESOP financing is employees them-
selves. Given competitive labor markets, employees would be expected
to pay most of the net cost to the firm through explicit payroll deductions,
lower wages, or conversion of assets that would have been used to fund
alternative forms of deferred compensation. In this case, the major ben-
eficiaries of ESOP subsidies would be the firm’s existing shareholders. A
cynic might thus see ESOP tax breaks as a way of bribing the firm’s own-
ers to compensate employees through company stock rather than cash
wages, where the bribe is large enough to cover any risk premium that
must be paid to employees, but the result is to leave employees no bet-
ter off or worse off than they otherwise would have been. Less cynically,
employees and shareholders might divide up the value of the tax breaks
in a more balanced way.
Should a firm with majority employee ownership through an ESOP
be regarded as an LMF? In general, no, because votes are still attached
to shares of common stock, and these shares must be acquired through
capital investment. As indicated, it is unclear who precisely is making
these investments, but suppose for the sake of argument that employees
are paying the after-subsidy cost of the firm’s ESOP contributions. Even
in this case, votes are proportional to invested capital, not labor contri-
butions: All that employee status provides is an opportunity to purchase
shares on better terms than are available to the typical outside investor. At
most, one may argue that some of the shares held by the ESOP were paid
for by taxpayers and that employees vote these shares on the taxpayers’
behalf.
At the same time, ESOP legislation is flexible enough that it can be used
to establish a labor-managed firm if this is desired. Recall that shares can
be allocated to the accounts of individual employees using any formula
that does not discriminate in favor of highly compensated employees or
officers of the firm. In particular, it is possible to award shares on an equal
per capita basis so all full time employees with more than one year of
service gain equal voting rights. In a closely held firm with substantially
full employee ownership (in practice, about 85 percent or more of the
firm), departing workers can be required to accept a cash payout or to
sell their accumulated stock back to the firm at its current fair market
value, which is determined by an outside appraiser. This ensures that the
firm’s ownership structure remains stable and that voting rights do not
diffuse away to outside investors over time. Such a firm would deserve to
be called labor-managed.
Few ESOP firms have adopted a structure of this kind; see, however,
the example of Seymour Specialty Steel cited by Blasi (1988: 31), where
a privately held firm elects the board of directors on a one worker, one
vote basis. More often, shares, and hence voting rights, are allocated in
proportion to employee compensation rather than hours or on a per capita
basis. Still, a believer in marginal productivity theory might argue that
compensation reflects labor inputs adjusted for past investments in human
capital, and that a firm where votes are proportional to wage rates or
earnings could still qualify as labor-managed. If one also believes that
employees implicitly bear the after-subsidy costs of ESOP shares in rough
proportion to their earnings, one can argue that workers are supplying the
firm with bundled contributions of labor and capital in a fixed ratio, and
that votes are awarded in proportion to the size of these bundles. Clearly,
though, this deviation from the rule of one vote per person would be
unacceptable to most advocates of workers’ control.
Even where employees hold large share positions, this does not auto-
matically mean that employee representatives will obtain seats on the
board of directors. Blasi and Kruse (1991: 216) point out that non-
managerial workers serve on the board in fewer than 10 of the 1,000
publicly traded firms with the most employee share ownership. The deci-
sion to create an ESOP rests with the firm’s board or top management,
as does any decision to end the plan. For example, the management of a
closely held firm might terminate an ESOP in order to prevent employ-
ees from voting on a major reorganization favored by other share owners.
Such maneuvers can only be blocked if there are worker representatives
on the board of directors, but, as noted, this is highly unusual.
In sum, ESOPs are inherently messy from a control perspective, pri-
marily because the allocation of shares and therefore voting rights among
employees is highly flexible, but also because the true incidence of the
firm’s ESOP contributions is hard to pin down. At the extremes, it
is not difficult to identify ESOP firms that are clearly KMFs or, in a
few cases, LMFs. There is also a heavily populated intermediate range
where employees form a substantial or even a majority voting block, but

where this role has been achieved through financial investments made on
privileged terms rather than through the supply of labor.
The factors that induce firms to adopt ESOPs need not overlap strongly
with factors promoting employee participation in firm decision-making.
This point deserves emphasis because many studies have reported the ex-
istence of substantial complementarities between ESOPs or other forms
of employee financial ownership and a participative organizational cul-
ture (Doucouliagos, 1995; Levine, 1995; Ben-Ner, Han, and Jones, 1996).
There may well be such complementarities from a productivity stand-
point, but empirical work by Ben-Ner, Burns, Dow, and Putterman (2000)
establishes that the two kinds of participation are driven by rather dif-
ferent firm and industry characteristics. The involvement of employees
in decision-making is positively related to the complexity of production
tasks, while ESOP adoption and coverage are positively related to the
average wage and length of tenure of the firm’s employees, suggesting
that the personal wealth of workers plays a significant role.
It is far from clear whether the U.S. trend toward more employee share
ownership will continue unabated, or whether a natural upper limit will
eventually be reached. One upper bound might result from the limited
willingness of employees to pay for retirement benefits out of current
wages. Blasi and Kruse (1991: 128) hazard a guess that replacing benefit
costs with a system of employee ownership would not generate more
than a 20 percent employee ownership stake in most companies. Another
potential limit on ESOPs is the degree to which employees are willing to
forego the diversification of personal financial wealth. Yet another might
be the reluctance of outside investors to put themselves at the mercy of a
voting majority of workers. I will say more about these issues in Chapter
12. But if highly concentrated employee share ownership can deliver large
productivity gains, perhaps in conjunction with other innovations such as
a participative organizational culture, then employee share accumulation
might proceed further. ESOPs are a moving target, and their continuing
evolution warrants scrutiny from students of workers’ control.
4.3 Codetermination
Employee stock ownership plans in the United States frequently give
workers a major role as suppliers of equity capital without commensu-
rate control rights. Conversely, the system of codetermination in Germany
gives workers control rights by placing their representatives on company
boards of directors, but does not require them to make any equity
investment in the firm. I conclude this chapter with an overview of codeter-
mination policies and a few of the controversies they have spawned. The
term “codetermination” is often used broadly to encompass works coun-
cils at the plant level or even collective bargaining, but here I will limit its
meaning to mandatory employee representation on supervisory boards.
Employees have enjoyed minority board representation since the early
1970s in a number of European countries, including Sweden, Denmark,
Norway, Austria, and Luxembourg (see Svejnar, 1982a), but Germany
has moved the furthest in this direction.
It is necessary to begin with some background on German corporate
governance. All large corporations have a two-tier board structure, with
both a “supervisory board” and a “management board.” The supervisory
board monitors the financial condition of the firm, ratifies important in-
vestments and acquisitions, approves the annual financial statement and
balance sheet, and approves any dividend payout. The management board
directly operates the firm. In public firms, the supervisory board appoints
the management board. Members of the management board usually have
five-year terms but can be fired by the supervisory board for cause. Man-
agers do, however, have considerable discretion. For example, their deci-
sions about product strategy and plans for financing investments need not
be ratified by the supervisory board. In 86 percent of large public firms,
the supervisory board meets only the legal minimum of twice per year
(Gelauff and den Broeder, 1996: 50).
Codetermination was established in German coal and steel compa-
nies (the so-called “Montan” sector) by the 1951 Codetermination Act.
For firms in the smallest affected size class, this Act requires a supervisory
board with eleven members, of whom five are chosen by shareholders and
five by employees (three of the latter are selected directly by the labor
unions involved). Four members within each group must be shareholders
or employees. The fifth is an outsider who cannot belong to an employ-
ers’ federation or a labor union and has not been either a shareholder
or an employee of the firm in the past year. Employee board members
must include at least one blue-collar and one white-collar worker who are
elected in separate procedures and approved by the relevant labor unions.
The eleventh member is a neutral chairman elected by all other members
of the supervisory board. To be nominated, a candidate must receive at
least three votes from each of the shareholder and employee groups, so
either caucus can block an unacceptable nominee. In larger firms, the su-
pervisory board has more members, up to a maximum of twenty-one, and
the preceding arrangements are scaled up accordingly. The management

board includes another worker representative, the labor director, who
must receive not only a majority of votes from the supervisory board but
also a majority among its employee members. The Montan system is nor-
mally referred to as one of “full parity” (the description in this paragraph
is taken from Pistor, 1999: 168–69, and Baums and Frick, 1999: 208–10).
The Works Constitution Act of 1952 extended similar rules to other in-
dustries, but required only that labor representatives constitute one third
of the supervisory board. There is also a Works Constitution Act of 1972,
which left the central principles of the 1952 Act in place. Given the pre-
ponderance of shareholder representatives, it seems safe to assume that
the role of worker representatives in such firms is mainly informational.
The Codetermination Act of 1976 extended “near parity” on the super-
visory board to all limited liability firms outside the Montan sector with
more than 2,000 employees. The size of the supervisory board varies with
the number of employees, but in all cases there is equal representation
of shareholders and employees. A majority of the employee members
are chosen by having the blue-collar and white-collar workers separately
elect delegates and then having these delegates elect the board members
for their constituencies. These groups have board seats in proportion to
their numbers. A minority of employee board members represent labor
unions, and are elected by the same delegates after having been nomi-
nated by the relevant union. These members need not be employees of
the firm.
The supervisory board elects a chairman from among the current board
members. In principle, this person could belong either to the shareholder
or employee group but must receive a two-thirds majority from the board
as a whole. In case of deadlock, which given the composition of the board
is likely, the chairman is elected by the shareholder group and the deputy
chairman by the employee group. The expression “near parity” refers to
the fact that the chairman has a tie-breaking vote on any decisions before
the supervisory board. Members of the management board must also
be approved by a two-thirds majority of the supervisory board, but the
chairman is empowered to break any impasse, and will normally represent
the interests of shareholders (this description draws from Svejnar, 1982b;
Baums and Frick, 1999; and Pistor, 1999).
This differs slightly from the system in the Montan industries, where
it is the neutral chairman, perhaps an academic or local elected official,
who has the authority to break a tie. Outside the Montan sector, the labor
director on the management board is selected by the supervisory board
in the same way as other managers, rather than being appointed solely
by worker representatives. Also, unions in the Montan sector have a role
in choosing all worker representatives on the supervisory board, which
is not true in the industries covered by the 1976 Act. By the end of the
1970s, about 30 percent of all private-sector workers were employed in
companies that had some labor members on the supervisory board. Most
of these were under the near parity system defined by the 1976 Act rather
than the 1952 “one third-rule” or the 1951 “full parity” Montan system
(Gelauff and den Broeder, 1996).
Here is the current situation. No firm with fewer than 500 employees
is required to have employee representation on its supervisory board, if
it has such a board at all. For any limited liability firm with 501–1,000 em-
ployees, one-third of this board must be worker representatives. The same
requirement applies to limited liability firms with 1,001–2,000 employees
except those in a Montan industry, where employee representation is one-
half and the other Montan rules apply. Finally, in all limited liability firms
with more than 2,000 employees, one-half of the supervisory board must
be worker representatives. If such a firm is in a Montan industry, the 1951
rules apply; otherwise the 1976 rules are relevant (for more details on the
legal framework for codetermination, see Gurdon and Rai, 1990).
The degree to which workers have gained real power as a result of
this legislation is unclear, and the consequences of codetermination have
been a subject for debate. Jensen and Meckling (1979: 474–75) report that
Montan firms altered their organizational structures to evade the 1951
legislation, leading to several additional laws to prevent such avoidance.
They report similar responses to the 1952 legislation establishing the one-
third rule. In the years prior to passage of the 1976 law, it ran into strong
opposition from corporations and employers’ associations. After its pas-
sage, nine companies, twenty-nine employers’ associations, and a feder-
ation representing shareholders challenged the law in court, but it was
upheld by the Federal Constitutional Court on March 1, 1979.
Employers took numerous evasive actions in the aftermath of the 1976
legislation. One strategy was to sell divisions or break a firm into pieces
to reduce employment below the minimum threshold for codetermina-
tion. Another was to reorganize a firm using a legal form not covered by
the law. Other steps included delegating more power to management by
modifying corporate charters to reduce the set of transactions requiring
approval from the supervisory board, or shifting approval from the super-
visory board to the shareholder meeting. These tactics led to a reduction
in the number of firms covered by the 1976 law from 650 to 480 over a
three-year period, with 50 firms changing their charters and 120 reducing
their workforces below 2,000 employees (Benelli, Loderer, and Lys, 1987).
A number of ploys led to court battles, including electing a second (non-
employee) deputy chairman; adopting quorum rules treating shareholder
and employee representatives differently; giving exclusive authority over
executive employment contracts to shareholder meetings; and delegating
decisions to committees on which employees lacked parity. Of these ma-
neuvers, the first three were struck down in court, but the last was upheld
(Pistor, 1999). Some observers report that boards have become essentially
superfluous: “Meetings of the supervisory board in Germany are often
characterized by the absence of debate and by consensus; the subjects
brought before the board are hardly ever controversial” (Gelauff and
den Broeder 1996: 59). But others stress that supervisory boards have al-
ways been relatively weak in Germany and that their largely ceremonial
role cannot be blamed entirely on codetermination (Pistor, 1999; Roe,
1999).
If firms successfully evaded the intent of the legislation after its pas-
sage, one would anticipate little or no effect on their behavior and per-
formance. If avoidance strategies were not fully successful and workers’
control shifted rents away from shareholders, one should find evidence
that shareholders became worse off. A third scenario is that greater in-
fluence for workers could have generated productivity gains that more
than compensated for any redistribution away from shareholders, so prof-
itability or stock market value might even have increased as a result of
codetermination. Of course, in the last case it is unclear why firms would
not introduce codetermination spontaneously.
Virtually all empirical studies of codetermination face methodological
problems of the following kind. Suppose, for example, one would like to
know whether the 1976 law increased the capital intensity of firms. No
control group is available to help answer this question because all firms
with more than 2,000 employees are subject to the 1976 rules. A researcher
is therefore forced to proceed in one of two ways. First, direct before-and-
after comparisons can be made for the affected firms. This approach is
open to the objection that other things may have changed simultaneously
(oil price shocks, the business cycle), so that the effects of codetermination
are confounded with other phenomena. Second, affected firms can be
compared against smaller firms that are not subject to codetermination
rules. The latter firms may differ from the former in ways related to size
but not codetermination. With these caveats in mind, I undertake a brief
review of empirical research on the subject.
In a pair of studies using slightly different data and methods, Svejnar

(1981, 1982a) estimated that the 1951 legislation imposing codetermina-
tion on the Montan industries had a statistically insignificant effect on
wages in the coal industry but led to a significant 5–6 percent wage in-
crease for iron and steel. There is some possibility that the wage effect
for iron and steel may have resulted from an increase in industry profit,
rather than greater bargaining power for labor due to codetermination
itself (Svejnar, 1982a). In related research, Svejnar (1982c) found that the
1951 law had no effect on productivity in any Montan industry.
The effects of the 1976 “near parity” rules on shareholders have been
explored in several studies with mixed and somewhat puzzling results.
From questionnaire data on 63 firms, Gurdon and Rai (1990) conclude that
codetermination decreased capital-labor ratios and revenue per worker,
while increasing the rate of profit on assets. However, using a produc-
tion function approach for a sample of 112 metalworking firms, FitzRoy
and Kraft (1993) found that codetermination reduced productivity (sig-
nificant at the 10 percent level) and return on shareholders’ equity (sig-
nificant at the 5 percent level), although it did not affect labor cost per
worker. FitzRoy and Kraft speculate that codetermination may have so-
cial benefits in the form of employment stabilization, more training, and
better industrial relations but acknowledge that these effects are hard to
quantify.
Schmid and Seger examined the effect of the 1976 requirements on the
stock market value of the affected firms and concluded that the market
value of the equities in these firms could have been up to 24 percent higher
by 1991 if the near-parity codetermination rules had not been introduced
(cited in Jürgens, Naumann, and Rupp, 2000: 63). But other researchers
have obtained quite different results. Baums and Frick (1999) used daily
stock-return data to investigate the impact of twenty-three court decisions
extending or restricting codetermination on the stock-market value of
the firms involved, and found no statistically significant effects. They also
searched for any effects of the 1976 legislation and the 1979 Constitutional
Court ruling on sectoral stock-market indexes for industries that have
relatively high coverage rates under codetermination, and again could
not find a statistically significant relationship.
Benelli, Loderer, and Lys report that an event study of the Constitu-
tional Court’s ruling on the 1976 legislation “failed to detect any abnormal
stock price changes associated with that event . . . our results fail to sup-
port the contention that codetermination makes stockholders better off.
However, there is also little evidence for the opposite notion” (1987: 573).
Jürgens, Naumann, and Rupp similarly conclude that “the existence of co-
determination per se has no definite implications for shareholders, mainly
because the system can be implemented and operated in so many different
ways” (2000: 65).
Apart from the question of whether shareholders gained or lost wealth
as a result of codetermination, other issues have been explored. For ex-
ample, Benelli, Loderer and Lys (1987) deduce from the theory of finance
that (1) workers will want firms to use equity financing rather than debt
because debt claims compete with those of workers in cases of bankruptcy;
(2) workers will try to limit the payout of dividends even if this leads the
firm to pursue projects with negative net present value; (3) workers will
seek to increase wages, limit the growth of the labor force, and favor
capital-intensive investment projects; and (4) workers may favor projects
with negative net present value if they diminish the variance of firm value
sufficiently. These predictions have been endorsed by Pejovich (1994).
The empirical results of Benelli, Loderer, and Lys (1987) do not lend
much support to these ideas. As mentioned earlier, the 1976 legislation
applies to all companies with more than 2,000 employees, so there is no
way to devise a control group of firms that are similar in size but exempt
from codetermination. Instead, these authors controlled for size by using
matched pairs of firms, and searched for possible effects on firm behavior
and performance variables by examining the appropriate ratios between
codetermined firms and other firms before and after the legislation. They
find that “there is no evidence that employees use codetermination to
affect firm policies” (1987: 566). In particular, they find no statistically
significant effects on dividend payout, ratio of debt to total assets, ra-
tio of net investments to total assets, capital intensity, employees’ total
pay, variability of salaries, or variance of book return on assets. There
is some evidence that the variance of book return on equity fell. In a
non-parametric test, they find that all variables moved in the directions
predicted by theory, but never at the 5 percent level of significance. The
authors suggest that these negative results occur because employees have
heterogeneous objectives and thus do not constitute a unified block in
opposition to the shareholder goal of wealth maximization.
A related interpretation of codetermination is proposed by Henry
Hansmann. He argues that for those firms affected by the 1976 Act, “a
variety of elements . . . seem well designed to prevent politics at the board
level, both among workers and between workers and shareholders, from
yielding instability and inefficient decisions” (1996: 111). These include
the following: (1) labor lacks full parity because ties are broken by a
shareholder representative; (2) the dual board structure means labor is

only represented on a board that makes broad policy decisions; (3) man-
agerial representatives may side with shareholders on a particular issue;
(4) many issues are decided through industry-wide collective bargaining
(for instance, wage levels) or in the works councils of individual plants.
Point 4 implies that decisions that could trigger serious conflicts between
workers and shareholders are often removed from the jurisdiction of su-
pervisory boards. It should be noted, however, that an individual firm or
plant may pay wages well above the minimum level set through industry-
wide agreements; see Hassel (1999). Hansmann summarizes by remarking
that codetermination does not appear to have had much effect on firm
decisions (1996: 112).
Several economists have developed an efficiency case in favor of
codetermination. Freeman and Lazear (1995) show that if this policy
simultaneously raises productivity and enhances worker bargaining
power, shareholders may resist it because they become worse off even
though the total size of the pie expands. Thus the policy may need to
be imposed by law. Freeman and Lazear argue that codetermination
can increase productivity through better communication between man-
agement and workers. Another argument is advanced by Gelauff and
den Broeder, who believe that codetermination helps to enforce implicit
agreements between shareholders and employees and thereby enhances
workers’ incentives to invest in specialized human capital (1996: 42).
Smith (1991) focuses explicitly on the problem of managerial discretion
and argues that codetermination can generate efficiency benefits by curb-
ing managerial incompetence, shirking, information hoarding, and oppor-
tunism toward the firm-specific investments of employees. In Smith’s view,
codetermination does not emerge spontaneously because the benefits to
workers and possibly also shareholders, though large, are widely diffused
over these groups, while the costs are concentrated on top management.
Managers are therefore both motivated and strategically positioned to
block codetermination proposals, while those favoring such proposals
confront a free-rider problem. Smith also argues that firms find it costly
or impossible to make credible commitments to workers that their quasi-
rents will be respected, unless there is a legally mandated framework
such as codetermination to facilitate enforcement of these commitments
by employee representatives.
One mystery about codetermination is why employers fought against
the 1976 law with such vehemence before and after passage when it has
subsequently been so difficult to find evidence that codetermination had
any real effects on firm behavior or performance. To be sure, empirical

work has been hampered by the lack of control groups, small sample
sizes, and various measurement and causality problems. But given such
strong opposition, the inability of most researchers to discover any effect
of the Constitutional Court’s 1979 decision on the stock-market values of
affected firms is striking.
A plausible answer is that the opposition to codetermination came
primarily from managers who did not want to share power with their em-
ployees or be supervised by them. This is consistent with Pistor’s view
of the constitutional challenge: “The main challengers of the law were
not shareholders, but companies and employers’ associations made up of
company executives . . . the main issue in the entire proceeding was not
whether codetermination would dilute shareholder value, but whether it
interfered with the interests of the corporation and the way it was man-
aged” (1999: 176). Whatever the truth may be, there is now a broad con-
sensus in favor of codetermination in Germany, partly because it reduces
conflict over downsizing and plant closures. Jürgens, Naumann, and Rupp
say that criticism of codetermination is now “almost taboo” (2000: 72).
Ironically, the main negative aspect of codetermination reported by
recent interview respondents is that the weakness and fractionalization
of supervisory boards have increased the difficulty of controlling top man-
agers (Pistor, 1999: 188–91). Moreover: “The system of codetermination
is seldom attacked openly by company management. On the contrary,
quite a few top managers praise the system warmly when confronted
by institutional investors. A cynic might observe that German managers
have ulterior motives for praising a system that serves as a protection for
incumbent management against hostile takeovers” (Jürgens, Naumann,
and Rupp, 2000: 72). Thus, although top managers initially opposed code-
termination, they appear to have made their peace with the system either
by evading its requirements or by turning it to their own advantage.
5
Conceptual Foundations
5.1 The Theory of the Firm

It is impossible to study workers’ control of firms without first saying
what a firm is. I define a firm as a set of agents supplying inputs to a
common production process, where the productive activities of the agents
are coordinated through an authority structure and the resulting outputs
are sold on a market. Inputs may include labor, physical assets, financial
wealth, raw materials, land, or any other resource that can be owned by
an individual or group. More will be said about the nature of authority
structures later.
In principle, a firm might consist of one person or a household, but
I am concerned only with enterprises in which groups of agents come
together for the specific purpose of production. The condition that output
be sold on a market rules out production activities directed solely toward
household consumption. It can be argued that households produce labor
itself, and that labor services are often sold on a market. However, I do
not address “firms” of this kind.
I do not want to become committed to a purely technological view
of the firm, so a common production process is deemed to exist when-
ever input suppliers are coordinated directly or indirectly by a com-
mon authority. In this context, indirect coordination means through a
chain of authority relationships rather than by market contracting (this
distinction will be discussed shortly). The boundaries of the firm are
defined by the boundaries on the exercise of authority, but explaining
the size and scope of the firm is not my present task. Ultimate author-
ity is often vested in a group rather than an individual, and in these
92
cases I speak of a control group. Section 5.2 will develop this idea more
formally.
Authority structures are highly variable. Some firms have a dictatorial
structure in which subordinates merely follow orders without participat-
ing in joint decision-making at all. Other firms formulate plans through
majority voting among all affected agents. Most of the cooperatives dis-
cussed in Chapters 3 and 4 have a blended system in which workers serve
under the authority of managers at the operational level but elect rep-
resentatives who can hire and fire the managers. I will generally use the
expression “governance structure” when discussing the political mecha-
nisms for choosing top managers.
All authority structures within firms have certain key elements. A co-
herent plan for deploying inputs is formulated and then modified over
time in response to new information about the firm’s environment; this
plan is not normally blocked by individual haggling after it has been
adopted by the control group or its designees; and the plan is enforceable
in the sense that agents have suitable incentives to implement the tasks
assigned to them. I do not assume that authority must have a contractual
basis under which sanctions are imposed by a court when deviant behavior
occurs. This is one possibility, but compliance might also be self-enforced
by having firm members themselves punish those who deviate.
The reference to coordination in the definition of the firm implies that
no authority structure exists if there is no core person or group who
imposes order on the actions taken by other members of the firm. Of
course, in any large firm some decisions are delegated to managers or
production workers, and no person has access to all the information on
which the firm’s actions are based (Hayek, 1945). All that is required
is a set of communication and decision-making procedures that impose
a measure of coherence on the actions taken by input suppliers. This
coherence is ultimately ensured by some central control group, but such
a group cannot usually make every decision within the enterprise.
The agents who ultimately control the firm, and those wielding del-
egated authority, need not renegotiate the compensation packages of
the other agents each time new tasks are assigned or resources are
reallocated. An employee’s wage rate does not vary from minute to
minute, increasing for harder tasks and decreasing when tasks become
easier. Tasks can be altered, within acceptable limits, even though the
compensation provided by the firm is left unchanged. These adjust-
ments may shift the worker’s welfare up or down but not by so much
that the worker prefers to quit the firm. Authority can thus be defined
94 Conceptual Foundations
as control over behavior that is exercised in the absence of continual

repricing.
The key difference between firm and market is the absence of efficient
short-term bargaining within the firm. If such bargaining is feasible, then it
can be shown under fairly general conditions that the allocation of inputs
under a series of short-term bargains is the same as the allocation under
long-term contracting (Fudenberg, Holmstrom, and Milgrom, 1990). Be-
cause the latter is the best one could expect from an efficiency standpoint,
there is nothing to be gained from an authority structure. But if efficient
short-term bargaining is infeasible for reasons of informational asymme-
try, this conclusion no longer holds, and an efficiency case can be made
for the use of authority relationships.
The obstacles to efficient bargaining are legion. Well-known examples
include the opportunity costs of the bargainers’ time, which are usually
high if multilateral negotiations are needed in a large production team;
the possibility of multiple equilibria in bargaining games (Milgrom and
Roberts, 1990a); temptations for the parties to dissipate resources by
strategically prepositioning themselves prior to the bargaining process,
either by acquiring better information (Barzel, 1982, 1989) or by manip-
ulating the characteristics of physical assets (Dow, 1985, 1988, 1989); and
delays or impasses arising when the bargainers have private information.
Some ex post inefficiency is inevitable for any bargaining mechanism
when there is uncertainty about whether gains from trade exist (Myerson
and Satterthwaite, 1983). In a model developed by Garvey (1993), for
example, there is a tradeoff between deadweight losses from failure to
agree in some states of the world under bargaining, and deadweight
losses from failure to internalize costs borne by a subordinate under
authority.
Agents who are part of a firm and subject to an authority relationship
remain free to renegotiate their relationship with the firm at any time.
Employees are not slaves and they can walk away from a firm if they wish,
or refuse unpleasant tasks unless given additional compensation, although
they may be fired as a result. The point is simply that employees do not
exercise this right under normal conditions. Whatever the reasons for this
choice, the fact that agents treat decisions as authoritative means that the
firm can reliably terminate negotiations over its resource-allocation plan
and move on to the implementation stage.
Authority in the sense of control over behavior does not exist unless
firm decisions are enforceable. Even if agents do not demand wage ad-
justments or issue counterproposals when asked to do something, they
could deviate from their assigned duties or do nothing. Incentives must

therefore exist for agents to carry out their particular portions of the firm’s
current plan. These incentives are shaped by the compensation package
an agent accepts on joining the firm. One possibility is a legally bind-
ing contract under which workers accept the firm’s authority within a
given domain with the threat that financial penalties will be imposed by a
court if they later fail to comply. But such contracts are seldom practical
because the actions of subordinates are hard to document and a clever
employee can always find ambiguities in the instructions of superiors.
Furthermore, access to the courts can be expensive and their decisions
difficult to predict.
For these reasons among others, firms typically rely on self-
enforcement, as when a firm punishes shirking or insubordination with
dismissal. Since authority can be abused, one must also consider the pun-
ishment strategies available to subordinates, which typically include shirk-
ing and quitting. Dismissal and quitting are, to be sure, extreme responses.
More moderate carrots and sticks, also self-enforced, are generally ade-
quate to maintain compliance under routine conditions.
Disputes between employers and workers do sometimes wind up in
court (Epstein, 1984; Rosen, 1984; Finkin, 1986), but the outcomes often
serve only to reinforce practices that would normally be self-enforced.
Holmstrom and Tirole observe in this connection that “legal systems
generally define authority by usage . . . [A]uthority is thought to be le-
gitimate if it corresponds to good practice or prevailing customs” (1989:
124). Finkin (1986) makes the same point: The courts look to customary
informal practice in the same firm or similar firms when deciding whether
a subordinate had a legal obligation to obey a specific command. For this
reason, it is frequently a useful theoretical simplification to limit attention
to strategies through which authority relationships can be self-enforced,
without bringing in the enforcement role of the courts as a separate con-
sideration.
Agents are members of different firms if they are subject to decisions
reached under different authority structures. When a worker is subject to
the direct or indirect control of a particular control group, I will say that
she is located “inside” the firm defined by that group. Otherwise she is
“outside” the firm and engaged in ordinary market exchange. This raises
the question of whether someone can be a member of several firms si-
multaneously. There is no conceptual reason to rule this out. A worker
could have multiple jobs, or an investor could have voting rights in mul-
tiple firms. As long as the plans of the firms do not come into direct
conflict – for instance, by telling the same worker to be in two different

places at once – no problem arises. For the most part, however, it does
no harm to picture firms as coalitions with the feature that an individual
worker belongs to one such coalition at most.
The literature of the new institutional economics describes the firm
in a bewildering variety of ways. My approach descends from that of
Ronald Coase (1937). In Coase’s formulation, a firm is an arena in which
resources are allocated by authority or command, while a market uses
the price system to allocate resources. He remarks that “if a workman
moves from department Y to department X, he does not go because of a
change in relative prices, but because he is ordered to do so.” In a firm, a
worker “for a certain remuneration (which may be fixed or fluctuating),
agrees to obey the directions of an entrepreneur within certain limits”
(emphasis deleted). Therefore, “the distinguishing mark of the firm is the
supersession of the price mechanism.” There is much textual evidence that
Coase thought a key advantage of authority was its ability to overcome
short-term bargaining inefficiencies.
One drawback in Coase’s treatment is that he relies too heavily on
legal institutions as the foundation for authority (Mäki, 1998a). This is
easily remedied: What matters to the theory of the firm is not the law’s
attitude toward the relationship between employer and employee, but the
existence of authoritative coordination itself. If compliance is enforced
through the courts using legally binding contracts, that is fine, but if it
is enforced through threats of dismissal, that is fine too. Another way in
which I depart from Coase is that I do not see the structure of a firm as
the solution to a problem of transaction cost minimization. This point will
be addressed in Section 6.3 and need not be pursued here.
Looking at more recent authors, my view is most consistent with that of
Milgrom and Roberts, who say that “the crucial distinguishing character-
istic of a firm is not the pattern of asset ownership but the substitution of
centralized authority for the relatively unfettered negotiations that char-
acterize market transactions” (1990a: 72). Another point of agreement is
that authority will be used in self-interested ways by whoever has it. As
Milgrom and Roberts comment, “The authority to intervene inevitably
implies the authority to intervene inefficiently. Yet such interventions,
even if they are inefficient overall, can be highly beneficial for particu-
lar individuals and groups” (1990a: 72). I argue in Chapter 11 that this
is a key factor determining the allocation of control rights among input
suppliers.
Many economists and other social scientists have described the firm as
an authority structure and distinguished it from market exchange on this
basis. Influential examples include Marx (1867), Knight (1921), Weber
(1946), Simon (1951), Arrow (1974), and Radner (1992). Williamson
(1975, 1985) largely adopts this viewpoint, as do Bowles and Gintis
(1993b). The presence or absence of authority is thus a standard line
of demarcation between firm and market.
Other views are less useful for my purposes. In the “nexus of contracts”
approach, the firm is a specialized set of market contracts linking suppliers
of inputs with consumers of output (Alchian and Demsetz, 1972; Jensen
and Meckling, 1976; Cheung, 1983). This does not remove the need to
characterize “firm-like” contracts. One might say, for example, that the
firm is a set of incomplete contracts involving a central agent who has the
right to fill in any contractual gaps when the need arises, to alter the mem-
bership of the production coalition, and to sell these rights to someone
else. This is roughly the notion adopted by Alchian and Demsetz (1972).
But if one continues down this road, it becomes difficult to separate the
“nexus” framework from a definition of the firm as an authority structure.
Another idea is that property rights over physical assets are central to
the firm, and that firm boundaries thus coincide with the set of productive
assets brought under unified ownership (Grossman and Hart, 1986; Hart
and Moore, 1990; Moore, 1992; Hart, 1995). In this view, asset owners
retain all decision-making powers not contractually assigned to someone
else. Because contracts between the owners and users of assets are in-
complete, patterns of asset ownership have efficiency implications and
are arranged to maximize total surplus for the agents involved.
The main problem is that this approach does not consider the full
range of contracts that can arise between owners and users of assets.
It is possible to write a contract under which the asset owner retains
all but a short list of rights specifically granted to a user. But it is also
possible to write a contract under which the asset owner retains only a
short list of enumerated rights for the duration of the contract, while
all unenumerated decision-making powers are assigned to the user. As
mentioned in Chapter 1, a familiar example is the usual landlord-tenant
contract under which the tenant can use an apartment in a great many
ways, most of which are not spelled out in advance. The apartment
owner, however, cannot enter unannounced, cannot evict the tenant
without notice, and so on. A firm controlled by its workforce could lease
physical assets from external owners in a similar way.
Other ways of defining a firm or drawing boundaries around it stress

differences in the flows of information occurring inside and outside the
firm (Riordan, 1990); the role of legal distinctions between employees and
contractors (Masten, 1988; Williamson, 1991); or the organizational and
technological competence of the enterprise (Demsetz, 1991; Winter, 1991;
Chandler, 1992; Teece, Rumelt, Dosi, and Winter, 1994). Each approach
has merit, but none highlights the control issues that are of interest here.
5.2 The Nature of Authority

Authority is exercised within a limited domain. Employees are not usu-
ally expected to rob banks or to donate their kidneys to the boss. But
the boundaries on acceptable uses of authority are vague and prone to
conflict. Although social conventions and norms play a major role, an
economic strategy for locating the “zone of acceptance” (to use Simon’s
1951 term) is to compare the cost to a subordinate of complying with
some command against the maximum penalty the superior can impose
for non-compliance.
The cost of compliance is defined as the difference between the
short-run payoff the subordinate gets by following orders, and the largest
short-run payoff the subordinate can achieve from an optimal deviation.
Usually the maximum penalty a superior can impose is dismissal. If this
penalty is feasible, the zone of acceptance is the set of all actions whose
cost of compliance does not exceed the discounted present value of the
loss from dismissal. The greater the surplus a subordinate gets by staying
in the relationship, net of the expected costs of future actions that will be
demanded in equilibrium, the greater will be the cost of dismissal, and
the larger will be the zone of acceptance. Other things being equal, the
domain of authority is wider when unemployment spells are long, search
costs in the labor market are high, unemployment insurance is limited,
and match-specific investments are needed before beginning a new job
(Shapiro and Stiglitz, 1984; Bowles, 1985; MacLeod and Malcomson, 1998;
Baker, Gibbons, and Murphy, 1999; Dow, 2000).
A deeper problem is determining when one person has authority over
another. At what point along the continuum from suggestions to com-
mands does authority first make its appearance? Or, if authority is seen
as an alternative to bargaining, how much haggling is allowed before a
relationship involves market pricing rather than authority? The answer
is that these are necessarily matters of degree. An economic relationship
5.2 The Nature of Authority 99
is more firm-like as the range of activities that might be demanded by an

employer expands, as the cost of exit increases, and as renegotiation of
compensation packages becomes more infrequent.
One way to cut through these difficulties is to say that person A has
authority over person B if A can expel B from the firm but not vice
versa. Expulsion threats are only a means of enforcing authority and are
not fundamental to the idea of authority itself. But this criterion is often
a valuable clue about who has authority over whom. Large capitalist
firms have authority structures that are clearly asymmetrical: A manager
can fire a front line worker, leaving the rest of the firm intact, while the
converse is not true. The worker can quit, but this is not the same thing
as staying in the firm and expelling the manager.
How does this square with the definition of a firm as a set of human
agents? It can be objected that no one can really be expelled from “the
firm” because any change in the set of agents constituting the firm auto-
matically creates a different firm. In particular, when a superior expels
a subordinate two new firms are created, one that is the same as before
but without the subordinate, and another that consists of the subordinate
alone (who may eventually join an alternative group after a period of
search).
One way to handle this at a conceptual level is to say that since the
subordinate was not a member of the previous firm’s control group,
her departure does not alter this group, so the “daughter” firm con-
taining the same control group is (by convention) considered to be the
same firm. The continuity of the firm is then interrupted only if there
is turnover in the membership of the control group itself. This is not
entirely satisfactory either, because if shareholders constitute a control
group in a corporation, one would not want to say that a new firm
has been created because one person sold shares to another. A fur-
ther convention for such cases might be that of the two daughter firms,
the one with a majority of members (perhaps weighted by voting rights
or bargaining power) from the original control group is the successor
firm.
Whatever convention one may adopt about the identity of the firm,
the asymmetrical nature of expulsion powers remains relevant. When
a superior expels a subordinate, this yields two daughter firms, one of
which is identical to the previous firm except that it does not include the
erstwhile subordinate. Because the subordinate cannot create a daughter
firm through her own unilateral action that includes every member of
the earlier firm except her superior, the superior has authority over the
subordinate but not conversely.
This idea can be formalized by defining a relation of the form “S → i”,
where S is a set of agents, i is an individual agent who is not a member of
S, and S → i means that the coalition S can expel i from the enterprise.
When compliance with authority is backed up by dismissal threats, one can
map the authority structure of a firm by determining all of the coalitions S
that can potentially expel each given person i. Thus if “power to dismiss”
is observable, then authority structures themselves will be observable.
Alternatively, when compliance is secured through rewards or penalties
other than dismissal, one can map out the system of authority relationships
within a firm by determining which groups have the power to impose these
rewards or penalties on each individual agent.
Under a few reasonable axioms, it can be shown that an authority re-
lation between coalitions and individuals defines a unique minimal sub-
set of the firm’s membership called the ultimate control group (Dow,
2002). Roughly speaking, this is the smallest subset of agents holding
formal authority that cannot be revoked by some alternative set of agents.
This control group directly or indirectly allocates the resources of the
firm.
The ultimate control group could consist of one agent (as when the
firm is run by a single entrepreneur), the entire firm membership (as in
a democratic firm that owns all of its non-labor inputs and gives every
worker a vote), or some intermediate subset of agents (as in a corpora-
tion where the shareholders have ultimate authority, and hire managers
and workers). I assume that a control group exists within each firm, so
firms can be classified according to the inputs supplied by their control-
ling members. The members of an ultimate control group do not exercise
authority over each other, and therefore some collective choice process
such as bargaining or voting must be used by this group. In most cases of
interest, a voting process is used, so I will often refer to “votes” and “con-
trol rights” interchangeably. A democratic partnership does not violate
the definition of a firm offered in Section 5.1 as long as the collective de-
cisions of the membership are authoritative from the standpoint of each
individual member.
In large firms, subordinates routinely make many economic decisions
in which their superiors choose not to intervene (Aghion and Tirole, 1997;
Baker, Gibbons, and Murphy, 1999). They may also achieve some informal
influence over their superiors through biased information supply, gift ex-
change, or personal relationships (Milgrom, 1988; Milgrom and Roberts,
1988, 1990a). Without denying the reality of such influences, I am con-

cerned here with the locus of ultimate control. It is vital to identify this
locus because authority delegated by the ultimate controllers can always
be revoked.
My view in this connection is consistent with that of Baker, Gibbons,
and Murphy (1999) but differs from that of Aghion and Tirole (1997),
who maintain that subordinates can be granted formal authority through
irreversible commitments supported by contracts, corporate charters,
customs, or the law. A reasonable test of this claim is whether there are
any circumstances under which these “subordinates” can be expelled from
the firm by their nominal “superiors.” If the answer is no, it is not appro-
priate to label them as subordinates. Instead they are members of the
ultimate control group. If the answer is yes, their authority is revocable
after all. Matters may not always be clear cut, because a hierarchical su-
perior may be able to fire a subordinate only at a high cost, or only in
certain states of the world. But subject to these caveats, it is generally
clear who has ultimate control in a large firm.
The firms discussed in this book usually have a governance structure
that includes a board of directors that hires and fires top management. An
easy test in searching for the ultimate control group is to ask who chooses
the board of directors. The relevant procedure is often specified in legis-
lation, corporate charters, bylaws, or other formal documents. However,
corporate charters and the like are only diagnostic devices, and one can
frequently identify an ultimate control group even when these devices are
absent.
5.3 The Locus of Control

In principle, anyone could have control rights over a firm – steel suppliers,
electricity suppliers, consumers, or perhaps entrepreneurs who supply no
inputs other than the effort involved in negotiating contracts and making
decisions when prevailing contracts are silent. For analytical simplicity, I
consider only suppliers of capital and labor inputs as candidates for mem-
bership in the ultimate control group. This presumes that other inputs
can be acquired, and outputs can be sold, in ways that do not affect firm
governance. In reality, suppliers of other inputs sometimes do exercise
control – for example, farmers who supply their harvest to agricultural co-
operatives (Hansmann, 1996). Consumer cooperatives are a familiar type
of organization, and production can be managed by non-profit organiza-
tions or government as well. These cases fall outside the scope of the book.
I define a capital-managed firm (KMF) as a firm in which ultimate

control is held by suppliers of physical or financial capital, and a
labor-managed firm (LMF) as a firm in which ultimate control is held
by suppliers of labor. This terminology requires comment because the
governance structure standing in contrast to the capitalist firm has re-
ceived an enormous variety of names, including producers’ cooperative,
workers’ cooperative, self-management, economic or industrial democ-
racy, labor-managed firm, and workers’ control. I will use labor-managed
firm and workers’ control interchangeably.
The term “producers’ cooperative” is sometimes used in a similar way,
but carries excess baggage because of its historical association with con-
crete organizational practices such as egalitarian wage policies, job ro-
tation, and suspicion of administrative hierarchies. Not only are these
practices not essential to the concept of ultimate control by workers, in
some cases they may well undermine firm performance. Furthermore,
the firms normally called producers’ cooperatives are typically small and
labor-intensive, although there is no reason in principle why large capital-
intensive firms could not also be controlled by workers. And finally, the
term producers’ cooperative is sometimes used broadly to include firms
that are controlled by the suppliers of agricultural produce or other inputs.
For clarity, I will speak of workers’ cooperatives where such terminology
is appropriate, as in Chapters 3 and 4.
The concept of self-management is linked with the economic system
that emerged in Yugoslavia between 1949 and 1991 and is unsuitable as an
umbrella term. Economic and industrial democracy are vague labels that
have often been applied to an excessively broad range of organizational
structures, including firms controlled by their workforces, socialist firms
with public ownership, codetermination, collective bargaining, and even
firms controlled by shareholders (who exercise their democratic voting
rights at annual meetings).
Calling a firm “labor-managed” directs attention to control issues
with only minimal danger that non-essential firm characteristics will
surface in the mind of the reader. The most significant danger of this kind
involves an association that has developed in comparative economics
between the general concept of an LMF and the specific behavioral
hypothesis that such firms maximize income per worker. I reject this
assumption in Chapter 7, and the reader should bear in mind that the
term “LMF” as used here emphatically does not entail maximization of
income per worker. A related problem is that some writers have sought
to distinguish an LMF, which finances all of its investment projects

through external funding, from a “worker-managed firm,” which finances
investment out of retained earnings. This distinction plays no role in my
terminology.
I next try to clear away some possible sources of confusion. First, capi-
tal and labor may be supplied by the same agents, or there may be overlap
between these two sets of agents. In principle, however, the distinction
between KMF and LMF remains clear. If control rights are awarded by
virtue of and in proportion to capital supply, then the firm is a KMF; if
control rights are awarded by virtue of and in proportion to labor supply,
then the firm is an LMF. Some Microsoft employees own stock in the
company, but since voting rights are proportional to the supply of equity
capital, Microsoft is a KMF. Conversely, the cooperatives of Chapter 3
often require new members to contribute to the capital of the firm. But
votes are equal across workers rather than varying with the amount of
capital supplied, so these firms are LMFs. The distinction between KMF
and LMF may not be very useful for small firms that lack a formal gover-
nance structure and obtain capital and labor from the same individuals,
such as the Mom and Pop grocery store or the family farm. Even in large
corporations, this distinction is occasionally fuzzy (recall the discussion
of ESOPs in Chapter 4). But usually it is not hard to discover whether
control rights are attached to capital supply or labor supply.
A second difficulty is that control is sometimes shared by capital and
labor suppliers even when these inputs are provided by distinct agents,
as in the firms subject to German codetermination law. A U.S. example
of this sort, United Airlines, will be discussed in Chapter 10. There is no
theoretical reason why the concept of a control group could not be applied
to such cases. But it is analytically useful to draw a sharp contrast between
control by capital and labor suppliers, while attaching appropriate caveats
whenever necessary.
Another complicating factor is collective bargaining, which some
writers view as a close cousin of codetermination. At least in North
America, unions have traditionally been more concerned with wages,
benefits, and job security than with negotiating a direct role in firm man-
agement. For instance, unions rarely seek seats on the board of directors.
It thus seems reasonable to regard unions as imposing contractual limits
on managerial behavior rather than attempting to control the firm itself by
altering core elements of its governance structure (Freeman and Medoff,
1984; Rock and Wachter, 1999).
A fourth conceptual (and complicating) issue is that control rights are

often held by a subset of capital suppliers (shareholders but not bondhold-
ers), or a subset of labor suppliers (lawyers but not janitors). Members
of the ultimate control group may have unequal decision-making weight
(dual classes of shareholders with unequal voting rights) and some parties
may have control only in narrowly defined states of the world (lenders
in situations of bankruptcy). These facts do not lead to any definitional
problem but do complicate the explanatory task.
The ideas developed here can now be tested for their consistency with
intuitions about firm organization. First, consider partnerships. These can
be either KMFs or LMFs depending on how the control rights are as-
signed. A real-estate partnership where votes are allocated according to
invested capital is a KMF. On the other hand, a law firm in which every
partner has an equal vote regardless of prior capital contributions is an
LMF, albeit one in which labor suppliers who are not partners lack control
rights.
The plywood cooperatives from Chapter 3 give all members a right to
participate in management, and membership is defined by labor supply.
There is a membership market on which workers can purchase shares
from departing members or the firm itself. When a share is purchased
from the firm, which normally occurs only at the formation stage, the
transaction helps to finance plant and equipment. One might therefore
argue that control is derived from the supply of capital to the firm through
the purchase of membership shares. But members who pay more for their
shares do not have extra votes, and members with more than one share
(a rare occurrence) still have just one vote. To this extent, the plywood
coops count as labor-managed firms. However, this description could be
questioned on two other grounds: The firms often hire non-member labor,
and retirees sometimes retain their voting rights.
Finally, consider the typical large corporation. From a formal stand-
point, this case is straightforward. Employees receive instructions from
managers who are organized in a hierarchical chain of authority, the peak
managers are accountable to the board of directors, and the directors
in turn are elected by a majority of shareholders at an annual meeting.
By the criterion of ultimate control, the shareholders run the firm, which
is thus a KMF. This would still be true even if some shareholders also sup-
plied labor to the firm, because voting rights are proportional to equity
investments.
There are two main objections to this reasoning. The first centers
around the weak nature of the control exercised by shareholders. Berle
and Means (1932) argued, and many subsequent writers have agreed,
that the nominal control rights assigned to shareholders are a sham. In
very large corporations with widely dispersed shareholdings, no single
person or small group can replace the incumbent directors unilaterally,
and the cost of organizing a majority of shareholders to do so is prohibitive.
Moreover, individual shareholders face strong incentives to free-ride on
monitoring and enforcement activities carried out by other sharehold-
ers, so there is little information on which to base a move against the
incumbent directors or managers. At worst, the incumbent board mem-
bers may be mere cronies of the CEO and other top managers, who
therefore run the firm in their own interests with little fear of outside
intervention. It could even be argued that because the incumbent man-
agers supply a form of labor, this “managerial” firm is in fact an LMF,
albeit one in which a very narrow subset of the firm’s labor suppliers
exercise control. More typically, managers are portrayed as running the
firm primarily in their own interests (Jensen, 1986; Jensen and Warner,
1988), or as arbitrating among various stakeholder groups such as share-
holders, employees, and the community (Aoki, 1990; Blair and Stout,
1999).
I nevertheless classify the Berle-Means firm as a KMF. The criterion of
ultimate control compels this conclusion, because every non-shareholder
can potentially be replaced involuntarily while the voting rights held by
shareholders cannot be taken away by anyone else. The problem is that
the ultimate controllers face obstacles in exercising their formal rights
effectively. But this is only an extreme example of a common predica-
ment. Ultimate controllers often face an incentive problem vis-à-vis their
subordinates, and collective-action problems vis-à-vis each other. Parallel
issues arise, although perhaps in differing degree, when workers elect the
board of directors. For an example, consider the difficulties that frontline
workers in the Mondragon group confront in exercising effective control
over distant top managers (Section 3.3).
Such factors may well have a bearing on the question of whether con-
trol rights will be assigned to capital or labor. But there is no qualita-
tive line between control groups with large incentive problems and high
collective-action costs versus those with small incentive problems and
low collective-action costs. The question is whether one thinks that con-
trol rights matter at all in governance structures of the former type. The
history of corporate takeovers during the last several decades suggests
that the latent control rights of shareholders are not a negligible element
in corporate governance.
Another possible objection was anticipated in Chapter 1. Perhaps,

it could be said, shareholders do not really supply capital. True, equity
capital is supplied at an initial public offering and perhaps again if the
firm decides to issue new shares, but the vast majority of stock-market
transactions simply pass existing financial claims from one person to an-
other. How then can one say that suppliers of capital control large publicly
traded corporations?
The answer depends on how one thinks about governance structures.
A transaction in the stock market transfers a bundle of rights from one
person to another – a claim on any dividends that may be paid, a claim on
the underlying assets of the firm if it is liquidated or sold, and voting rights.
Suppose this package is transferred from an initial equity investor, who
directly supplied financial capital to the firm, to someone else. The second
agent now takes on the role of the first. In particular, the new shareholder
takes over her predecessor’s claim on the proceeds from any asset sale
by the firm. As long as the shareholders as a group choose not to liquidate
the firm, it is sensible to say that all of them, old and new, are supplying
capital to it. Thus the role of “capital supplier” in the firm’s governance
structure is private property that can be alienated at a mutually agreeable
price. The stock market is not used solely to raise capital for the firm,
although this is one of its functions, but also to transfer the role of capital
supplier from one party to another.
A similar conceptual point will arise in Chapter 7 in connection with
markets for membership in LMFs. There I argue that no conflict need
arise between the assignment of control rights to current labor suppliers
and a reliance on markets to transfer the role of “labor supplier” from one
worker to another. If the firm wants to expand, it can sell newly created
membership rights and thereby increase its labor inputs, but individual
workers could use the same market to transfer existing membership po-
sitions to someone else. This is in fact the system used by the plywood
cooperatives, which treat the bundle of rights called “membership” as the
private property of an individual worker.
Finally, it is an interesting exercise to compare the present definitions of
KMF and LMF with the enterprise structures proposed by some contem-
porary market socialists. For example, Bardhan and Roemer (1993: 7)
remark that such proposals usually involve firms “operating indepen-
dently of state control, with boards of directors representing either work-
ers or various institutions (banks, mutual funds, pension funds) that
hold stock in the firms or are responsible for their financing.” Although
Bardhan and Roemer view all of the options listed as being compatible
with market socialism, on my criterion it should be clear that if the

board of directors represents workers, then the firm is an LMF, while
if the board represents financial interests, the firm is a KMF. This is true
whether the assets of financial institutions happen to be owned publicly or
privately.
5.4 Why Firms Cannot Be Owned

In a society where slavery and indentured servitude are not legally en-
forceable, no one can own a firm because a firm is a set of human agents.
It is of course possible for someone to own an asset used by a firm, to
own a claim on the net income from the firm’s activities, to control the
allocation of resources through the firm’s governance structure, or even
to own a controlling position within this governance structure. But the
human beings constituting a firm cannot be bought and sold.
Why then do economists and the financial press frequently say that
someone bought or sold a firm? This is a shorthand way of saying that a
role in the governance structure of the firm has been transferred from one
agent to another. In capitalist firms, it is often true that the same person
or group (1) owns the physical assets needed for production; (2) bears
risk; and (3) is the ultimate control group. Nothing in the definition of
a firm prevents these roles from being bundled together, or from being
sold as a unit to another person or group. However, the employees of
the firm are not sold, and are free to quit if they do not like the new
“owners,” just as they were free to quit under the old “owners.” The new
owners may expect that most current employees will choose to stay in
place and the value of this expectation may be built into the price paid for
“the firm,” but expectations of this kind do not reflect any legally binding
commitment.
The bundling of asset ownership, claims to net income, and control
rights lies at the heart of the concerns in this book. It would be a funda-
mental mistake to assume that these roles are inevitably linked, because
one way to create workers’ control is to separate them. For example, a
worker-controlled firm might lease assets from outside owners who do not
participate in the firm’s governance structure, or finance assets by selling
non-voting equity shares. A worker team could occupy all three roles by
having its members jointly purchase physical assets out of personal sav-
ings, but it would only qualify as an LMF if control were awarded on the
basis of labor supply rather than capital supply. To the extent that the sup-
ply of capital, residual claims, and control rights are bundled empirically,
this is not a matter of logical necessity but rather an empirical fact that
must be explained. For the sake of clarity it is vital to avoid casual refer-
ences to “the ownership of the firm.”
The importance of this is reinforced by the observation that capital
supply, residual claims, and control rights are routinely decoupled in prac-
tice. For example, employees in capitalist firms may receive profit-sharing
bonuses and function as residual claimants even though they do not share
in the ownership of any asset and have no formal control rights. Mem-
bers of ESOPs have residual claims and contribute capital, but often lack
the voting rights enjoyed by other shareholders (Section 4.2). Conversely,
employees can enjoy some formal control rights through collective bar-
gaining, consultative committees, or codetermination (Section 4.3) while
lacking an ownership stake in the firm’s physical assets or a contractually
recognized residual claim.
It may seem unnecessary to belabor these points, but the idea of “firm
ownership” is so deeply entrenched in the economic literature that fur-
ther discussion is warranted. I thus take a moment to review some com-
ments on this subject by Putterman (1988a). Putterman correctly notes
that there are (at least) three conceptions of the firm at large in the eco-
nomics profession, which he labels “the firm as commodity,” “the firm as
coalition,” and “the firm as association (or polity).” The first treats the
firm as something that can be bought and sold, while the others assert
that the firm is an organization consisting of human beings who cannot
be owned. According to Putterman, the coalitional view stresses the per-
vasiveness of free-riding problems, while the associational view stresses
problems of power, authority, and governance. Because I see no conflict
between these two conceptualizations, I consider only the commodity and
coalitional views.
As Putterman says, again correctly, the new institutional economics
has frequently adopted the following approach. Since human beings can-
not be owned but we want to talk about ownership of the firm, the firm
must be something other than a set of human beings. Thus, assume it
is a contractual role or position – namely, the right to make decisions
or wield control over production activities. The employees who actu-
ally engage in these production activities are, on this view, not part of
the firm. It may be hard for the legal entity known as the firm to lease
all required physical assets because these assets are specialized or their
use is difficult to monitor (see Sections 8.2 and 8.3), so the firm must
own these assets. Efficiency now dictates that control rights over the
firm be treated as a commodity in order to guarantee optimal investment
decisions. Otherwise, the controllers of the firm would be unable to ap-

propriate the present value of investment returns accruing after their
departure (see Section 7.4). Since the firm becomes a commodity, it
naturally follows that if workers want to control the firm, they must
buy it.
Throughout the remainder of this book, I adopt the following alterna-
tive view. As everyday common sense suggests, firms are organizations
consisting of people. The firm called General Motors includes the em-
ployees of GM. There is an ultimate control group for GM – namely,
its shareholders. But when someone buys a share (or even a majority
of shares) in GM on the stock market, they are not “buying the firm.”
Rather they are buying a role or position in the firm, the role of being a
GM shareholder. This role is defined by a bundle of rights including the
right to receive dividends, the right to a share of GM’s assets if they are
liquidated, and the right to vote at shareholder meetings. Such roles can
most certainly be treated as commodities and passed from one person to
another. But the firm is not bought or sold in the process, because it is
larger than its shareholders. It includes all of the agents who are members
of the ultimate control group or subject to its authority.
The firm need not be treated as a commodity in order to ensure efficient
investment decisions, nor could it be. What does achieve this purpose is
making the role of controller into a commodity. The incumbents in con-
trolling positions can then transfer these roles to their successors at a
price that properly capitalizes the future investment returns involved. It
makes no difference in principle whether such control positions are oc-
cupied by suppliers of capital or labor, as long as the role of controller
has an appropriate price attached to it. It may of course be true that
markets for control have different features depending on whether the
control group consists of capital or labor suppliers, and such differences
may have a bearing on who the control group turns out to be (see
Section 7.5).
It is necessary to exercise equal care with regard to the idea of profit.
In economic theory, the profit of the firm in a given time period is total
revenue minus total opportunity costs (explicit and implicit) for all mem-
bers of the coalition constituting the firm. If capital suppliers enjoy an
income stream from the firm that is identical to the best income stream
they could have gained by investing elsewhere, their opportunity costs
are covered and they get a normal rate of return. If capital suppliers do
not participate in the firm’s governance structure, this return is normally
all they will receive, even though there may be imperfect competition in
the product market. Any economic profit will likely go to someone else,
specifically the labor suppliers. If capital suppliers do participate in the
firm’s governance structure, and economic profit is positive, they may get
more than a normal rate of return by using their control rights to ensure
that labor suppliers are paid normal wages that only cover their oppor-
tunity costs. To reduce confusion, I will consistently use “profit” to mean
pure economic profit (also called “surplus”), not payments specifically to
capital suppliers. The latter are referred to as dividends, interest, rental
payments, or returns on capital.
A final remark on the relationship between control rights and residual
claims. It has usually been assumed in the new institutional economics that
control rights follow residual claims, because whoever gets the residual
will have proper incentives to make managerial decisions. But an alter-
native view is that residual payments follow control: Whatever group has
control will have the authority to determine how the coalition’s net income
is allocated and can appropriate it for themselves, subject to participation
and incentive constraints for the other agents. Different theories of the
firm may highlight one causal mechanism or the other, so it should not
be taken for granted that the locus of residual claims is fundamental
while the locus of control merely follows after. Indeed, one could argue
that both features of firm organization are determined simultaneously by
some set of exogenous factors.
5.5 Asset Ownership

Although workers cannot be owned by anyone else, the assets that provide
capital services can be owned in many ways – for example, by outsiders
whose only role is to lease their assets to the firm; by insiders who supply no
labor but have control rights in the firm; or by workers who supply labor,
have control rights, and own the non-human assets used in production. In
each case, assets may be owned by individuals or by groups.
Table 5.1 classifies firms on two dimensions. The rows indicate the
locus of control, while the columns indicate whether asset ownership
is by individuals or groups. I assume temporarily that labor and capital
are supplied by separate agents, so there is no need to consider the
complications associated with bundled input supply. All assets are
taken to be privately owned. These assets may include land, buildings,
computers, client lists, and so on, but for convenience I refer to them
simply as “machines.” If ownership is shared, the owners must adopt some
collective choice procedure to decide how their assets will be used (Barzel,
Table 5.1 A classification of governance structures: Unbundled input supply
Asset ownership
Individual Shared ownership
ownership of of non-human
non-human assets assets


 Control by Machine-owners’ Joint-stock



 capital cooperative using company using



 hired workers hired workers



 (KMF) (KMF)

Control rights

 Control by Workers’ Workers’



 labor cooperative leasing cooperative leasing



 machines from machines from a



 individuals (LMF) joint-stock

company (LMF)
1989: 99–100). This does not imply that such groups acquire direct control
over the production activities of the firm, because they may decide to
lease their assets to others who actually manage the production process.
Thus a group of agents sharing ownership of an asset may or may not
correspond to the ultimate control group defined in Section 5.2.
In a world where physical assets are owned by individuals (the left
column of Table 5.1), two cases could arise. The firm could be governed
by a collection of individual machine owners who contribute the service
flows from their assets to the firm and hire employees at a market wage.
The resulting machine-owners’ cooperative appears in the upper left cell.
Because ownership is individual, each asset owner is free to withdraw her
personal machine from the firm and put it to some other use. The firm
could also be governed by a group of workers who supply labor services
and lease machines from individual owners at market prices. This is the
simple workers’ cooperative shown in the bottom left cell.
Now move to the right column of Table 5.1. In a world where one or
more machines are owned by a group (think of them as shareholders),
there are again two cases. First, the machine owners might hire workers at
a market wage and exercise control over both their assets and the firm it-
self, as in the upper right cell. The conventional corporation has this form,
as do partnerships based on the supply of capital. It is of no importance
whether such a firm is established by having machine owners transfer
their personal assets to the firm in exchange for financial compensation,
or by having investors contribute funds to be used for the purchase of
jointly owned machines. In either situation, no individual investor owns

an identifiable chunk of the firm’s physical assets, and therefore cannot
unilaterally withdraw a specific machine from the firm. Instead, decisions
about asset use are made collectively (for instance, a decision to liquidate
may require a two-thirds vote). The same voting procedure is generally
used both to decide how to manage jointly owned machines and how to
direct the labor services supplied by employees.
In the lower right cell, the machine owners still decide as a group
how their assets will be used (for instance, they might vote on whether
to shift their machines to another industry, or on how best to maintain
them). But it is labor suppliers who decide how to manage the production
process as long as the machines are available, just as in the lower left
cell, where each machine had an individual owner. Suppose the machine
owners are shareholders in a corporation, and this corporation is the legal
owner of the machines in question. The lower right cell then describes a
worker-controlled firm that leases assets from a corporation. Here there
are two separate collective-choice processes – one through which the
shareholders manage their assets, and another through which workers
govern their firm.
To complete the discussion, it is necessary to consider the case in which
inputs are bundled: that is, workers supply both labor services and capital.
Such organizational forms can emerge in various ways. When assets are
individually owned, as in the left column of Table 5.2, workers might use
personal tools (hammers, computers) to engage in production activities,
or perhaps occasionally borrow tools belonging to other workers. When
assets are jointly owned by workers as a group, other arrangements can
be used. The founding workers might pool their savings and authorize the
firm as a legal entity to buy machines on behalf of the group; they might
borrow funds from a bank and repay the loan over time; or they might
donate personally owned machines to the firm.
In all cells of Table 5.2, labor services and machine services are supplied
by the same set of agents. It might therefore be tempting to say that
all represent instances of workers’ control. This is misleading, however,
since the criterion used to assign control rights also matters. The firm
is an LMF if votes are proportional to labor supply, but it is a KMF if
votes are proportional to equity investments, debt obligations, or machine
donations.
Suppose that every machine has an individual owner (the left col-
umn). If voting rights are assigned according to the value of the machines
contributed, or the value of their service flows, the firm is a KMF, as in
Table 5.2 A classification of governance structures: Bundled input supply
Asset ownership
Individual Shared ownership
ownership of of non-human
non-human assets assets

 Control by Machine-owners’ Joint-stock



 capital cooperative to company to



 which machine which



 owners supply labor machine-owners



 services (KMF) supply labor; votes



 on the basis of



 capital supplied

 (KMF)
Control rights 




 Control by Workers’ Workers’



 labor cooperative to cooperative with



 which workers jointly-owned



 supply personally machines; votes



 owned machines on the basis of



 (LMF) labor supplied

(LMF)
the upper left cell. In deciding how to vote on any given issue, machine
owners may of course take into account that they also have interests as
workers, but control is still proportional to capital supply, not labor supply.
However, if workers contribute roughly equal amounts of labor and have
equal voting rights, or varying amounts of labor and votes are adjusted
to reflect this, then the firm is an LMF, as shown in the lower left cell. An
example is a law partnership in which all partners supply equal labor and
each has one vote, although they also own personal computers that differ
in value.
When the members of a production team supply labor and also
jointly own a large, lumpy asset such as a factory (the right column of
Table 5.2), again it is necessary to look at the basis on which control rights
are allocated. If votes are proportional to capital supply, then the firm is
a KMF, as in the upper right cell, although again shareholders may take
into account that they also have interests as employees. If workers collec-
tively own the firm’s productive assets and exercise control on the basis of
labor supply, the firm is an LMF, as in the lower right cell. Some examples
include the plywood cooperatives and the Mondragon group discussed in
Chapter 3.
There is no conflict between collective asset ownership (by sharehold-

ers in a KMF, or by workers in an LMF) and the existence of individually
owned roles or positions in the firm that can be traded on a market (the
role of being a shareholder in a KMF, or the role of being a member
in an LMF). These are independent ideas. This is important because the
literature on LMFs often collapses the two notions. For example Ben-
Ner (1988b) uses “collective ownership” for a system where individual
workers lack tradeable claims on the LMF, and “individual ownership”
for a system where a membership market exists. In my framework, all
four combinations are logically possible – collective asset ownership with
or without commodification of membership, and likewise for individual
asset ownership.
Collective asset ownership is common, and indeed standard for en-
terprises that are organized as corporations. It nonetheless presents a
major puzzle for economics. As has been pointed out by Holmstrom
(1999), modern property-rights theory can at most explain patterns of
asset ownership among individuals. Within the framework of the latter
theory, a collective ownership system in which each shareholder owns
10 percent of all the firm’s assets is inferior on incentive grounds to a
carefully crafted system of individual ownership where assets are parceled
out to stimulate investments by workers in asset-specific human capital
that maximize joint surplus. Large firms nonetheless vest ownership of al-
most all assets in a legal entity controlled by shareholders. One rationale
is that centralization of decision-making in an authority structure may
be difficult unless asset ownership is also centralized, because individual
worker-owners would be more inclined to haggle over directives issued
by the central authority (Hermalin, 1999). As yet there is no satisfactory
formal analysis along these lines.
5.6 Residual Claims

The conventional concept of “firm ownership” embraces several rights:
ownership of at least some physical assets used by the firm, receipt of
income from the firm, control over the firm, and the right to sell the
preceding three rights to someone else (Putterman, 1988a, 1993). Since I
have already considered asset ownership and control rights in some detail
and have indicated how roles in a firm might be bought and sold, the only
remaining item on the agenda is the right to receive income from a firm.
No one doubts that in a private-ownership economy, asset owners are
free to make whatever contracts they find advantageous, and to generate
5.6 Residual Claims 115
income by doing so. But it is not obvious how, or whether, this idea extends
to the firm itself. If one thinks that firms can be owned, the question is
trivial: The firm is an asset like others, and whoever owns it can keep
the income derived from it. But, of course, firms as coalitions of input
suppliers cannot be owned.
The new institutional economics tries to get around this problem by
talking about “residual claims,” understood to mean the claim on what-
ever firm income is left over after other parties have received their con-
tractual payments. But even if one regards the idea of a residual claimant
as unproblematic, it does not follow that the owners of non-human assets
must be residual claimants. Such agents might rent their assets to one
or more users who keep the firm’s net income after meeting contractual
obligations, including rental payments to suppliers of non-human assets.
Thus, an asset owner’s right to receive income generated from the
asset does not imply anything about who has residual claims in the
firm.
There is also a deeper problem. A standard idea is that residual
claimants bear risk. Everyone else has the security of a contractual guar-
antee, so risks associated with the firm must be borne by those agents
who get the leftovers. This is potentially quite misleading. Theoretically
speaking, an agent in the firm bears risk if her income or wealth is contin-
gent on an unknown state of the world. This could be true in principle for
virtually everyone connected with the firm, including agents not normally
considered to be residual claimants. Creditors, suppliers, customers, and
employees can all suffer losses in wealth if the firm becomes bankrupt,
for instance, which is usually possible in some states of the world.
Even without bankruptcy, employees bear risk in numerous ways. They
may be paid on the basis of piece rates, with their productivity being
subject to random events; they may compete in promotion tournaments
whose outcome is partly dependent on the luck of the draw; they may
participate in profit-sharing plans, or be paid a year-end bonus that de-
pends on firm performance; they may be able to bargain for higher wages
when the firm prospers, but not otherwise; or they may be at risk of los-
ing rents or quasi-rents if the firm closes the factory or office where they
work. A central point in the literature of principal-agent theory is that
workers must usually bear risks for incentive reasons (see Holmstrom
and Milgrom, 1991; Sappington, 1991).
What seems to be meant by “residual claimant” is that under normal
circumstances, barring situations generally believed to have low probabil-
ity, everyone else will get what they have been promised, perhaps except
for minor random variations that are weakly correlated across individ-
ual members of the firm, while any major fluctuation in revenue or cost
affecting the firm as a whole is borne by the specified residual claimant(s)
alone. It is commonly taken for granted in the theory of the firm that resid-
ual claims in this sense must go together with control rights for incentive
reasons.
This proposition has roots in the neoclassical theory of the competi-
tive firm where the “owner of the firm” chooses a production plan, pays
the prevailing market prices for the inputs, sells the output at a similar
market price, and keeps the resulting profit. When this residual claimant
or “owner” maximizes profit, the aggregate social surplus from the firm’s
production activities is also maximized, because competitive prices reflect
the social costs of inputs and the social benefits of outputs. Since the so-
cial surplus is fully appropriated by the residual claimant, this is the right
person to make decisions on behalf of the firm.
In a world of incomplete contracts where input suppliers have rents
or quasi-rents at stake, it is much less obvious that a particular agent
should be singled out as “the residual claimant” and awarded control
rights on efficiency grounds. Such an agent may well find it attractive to
implement production plans that reduce or eliminate the rents or quasi-
rents of others, because from the standpoint of the decision maker, these
are costs rather than part of the social surplus to be maximized. The
problem is that the decision-maker no longer faces the true social cost
of the inputs involved, and bargaining may not function effectively as a
way of making this agent internalize all of the costs borne by other firm
members. Similar problems generally arise no matter who is designated
as the ultimate controller of the firm, so second-best comparisons are
inevitable if one wants to make efficiency judgments about how control
rights should be awarded.
In later chapters, I avoid loose references to residual claims where pos-
sible, because it is best not to encourage unwanted associations with the
competitive firm where the net income going to the “owner” is identical
with the social surplus generated by the firm. This terminology is deeply
embedded in the new institutional economics, and it is not possible to
eliminate it completely. The reader should therefore be on guard and re-
call from time to time that (1) every member of the firm generally bears
some risk; (2) control groups do not usually internalize all costs and ben-
efits to other members of the firm; and (3) this failure to appropriate costs
and benefits borne by other input suppliers is quite relevant in explaining
the empirical features of KMFs and LMFs.
6
Explanatory Strategies
6.1 The Symmetry Principle

This chapter has two broad objectives. First, I develop a pair of heuristic
tools, the symmetry and replication principles, which are helpful in prob-
ing hypotheses about LMFs. The rest of the chapter surveys strategies that
could be used to explain the facts about LMFs. These include theoretical
frameworks that rely primarily on efficiency concepts (transaction-cost
economics, contract theory) and others that assume optimizing behavior
but leave more room for inefficiency (adverse selection, repeated games).
I also consider two approaches not derived from microeconomic theory,
those invoking history and culture. My goal is not to evaluate the empir-
ical merits of any theory, but to present a theoretical menu from which
choices can later be made.
A useful source of intellectual discipline in generating explanations for
the rarity of workers’ control runs as follows. If there were competitive
markets for all relevant goods and services, firms would not exist in a
meaningful sense because all production activities could be coordinated
through market contracting rather than authority. In this environment,
each agent would maximize the profit obtained by transactions in in-
put and output markets because each would want to have the maximum
possible income available for consumption purposes. There would be no
debate about control rights or the objectives of firms.
The question of control over firms only becomes interesting after
one has identified certain departures from the world of complete and
competitive markets. These may involve informational asymmetries, an
inability to make binding commitments about future actions, or other
117
118 Explanatory Strategies
considerations. But even in an imperfect world, where authority within

firms can have economic consequences, it is hardly obvious why the sup-
pliers of capital rather than labor would control the firm. To see this,
suppose capital and labor were symmetrical inputs in both physical and
institutional respects. “Symmetry” here does not imply that capital and
labor are mathematically symmetrical inputs in a production function.
Rather, it is meant to rule out qualitative asymmetries and conjure up the
parallel treatment given to capital, labor, land, and other inputs in text-
book microeconomic theory. In this situation, KMFs and LMFs would
be symmetrical in their behavior and performance, and their distribution
across industries would be fundamentally random.
KMFs and LMFs are clearly asymmetrical in the real world. Aside
from the obvious point that LMFs are rare relative to KMFs, Chapter 3
made clear that LMFs are not scattered across industries in a random
way. Traditional workers’ cooperatives often carry out semi-skilled or
craft manufacturing activities, low-skilled service tasks, and construction.
LMFs adopting the partnership mode are generally confined to profes-
sional services. Except for conventional firms that have been taken over
by employees, LMFs are seldom found in industries with large economies
of scale or scope, high capital intensity, or very specialized physical as-
sets. There is also econometric evidence indicating that these two kinds
of firm behave differently even when they operate in the same industry
(see Section 7.6).
The symmetry principle suggests that these systematic differences be-
tween KMFs and LMFs should be explained by identifying qualitative and
causally relevant asymmetries in the features of capital and labor them-
selves. Such asymmetries may be either physical or institutional in nature.
Probably the most fundamental is that the ownership of non-human assets
can be shifted from one person to another, while endowments of time and
skill cannot. In short, the capacity to supply labor services is inalienable.
I have previously alluded to a few implications of this fact in Chapter 1,
and will return the same theme in Chapter 11.
In the absence of a qualitative asymmetry of this sort, it is usually pos-
sible to prove “equivalence” or “isomorphism” theorems showing that
the resource allocations supportable in an economy consisting entirely of
KMFs coincide with the allocations supportable in an economy consist-
ing entirely of LMFs. Accordingly, one has no basis on which to explain
why these two organizational forms behave asymmetrically in the real
world. Examples of equivalence theorems include Drèze (1976, 1989)
and Roemer (1988: ch. 7). Such results can be obtained even in some
6.2 The Replication Principle 119
environments where the use of market contracting is severely limited – for

example, by ruling out binding commitments about future actions (Dow,
2000). A specific theorem that will play a key role in Chapter 7 shows
that competitive markets for membership in LMFs are a perfect substi-
tute for labor markets and stock markets, and that LMFs having access
to such markets will maximize profit or present value just as a capitalist
firm would do (Sertel, 1982; Dow, 1986, 1993a, 1996; Fehr, 1993).
Of course, it is tautological to say that if capital and labor are sym-
metrical inputs, then KMFs and LMFs will behave symmetrically. But
as Chapter 7 will show, this tautology has often been ignored by writers
who attempt to explain why LMFs are rare. Intuitively it should be evi-
dent that one cannot hope to account for differences between KMFs and
LMFs without talking about differences between capital and labor. The
nub of the symmetry principle is that any explanation that fails to identify
an exogenous qualitative distinction between these inputs is wide open to
a critique involving equivalence theorems, and must therefore be judged
incomplete. Equivalence theorems are warning flags suggesting that the
explanatory task may not be quite as easy as it appears.
6.2 The Replication Principle

One corollary of the symmetry principle is the replication principle. This
asserts that a hypothesis about the rarity of LMFs that cites a purported
advantage of the KMF must explain why this advantage cannot be repli-
cated under workers’ control. Likewise, any claim that LMFs are in
some way superior to KMFs must say why the same advantage cannot
be achieved in investor-controlled firms. If an alleged advantage can be
replicated under the alternative governance structure, the hypothesis in
question fails to accomplish what it set out to do, which is to account
for differences between the two structures. The replication principle thus
requires that explanations focus concretely on why the physical or insti-
tutional differences between capital and labor induce suppliers of these
inputs to utilize control rights in different ways.
The replication principle has considerable bite because I have adopted
a narrow focus on control as the demarcation line between KMFs and
LMFs. If a hypothesis claims that LMFs are rare because they have prob-
lem x, while KMFs do not, and problem x is not related to the structure
of control rights, a critic can object by asking how KMFs are able to avoid
problem x, and then pointing out that LMFs could use the same solu-
tion. The only way to escape this trap is by linking problem x directly
with workers’ control or providing a series of logical steps that eventually

trace back to the assignment of control rights.
A few examples may be helpful. It is sometimes said that LMFs are
rare because they involve dysfunctional organizational practices such as
job rotation, direct democracy, or a rejection of administrative hierarchy,
but KMFs do not. For argument’s sake, suppose these practices really
are harmful to firm performance. Even so, nothing in the definition of
LMFs rules out specialization, representative democracy, or an adminis-
trative hierarchy. The only requirement is ultimate control by workers. If
workers choose, they can certainly assign specialized tasks, elect a board
of directors, or establish complicated administrative structures. Unless it
can be shown that it would be impossible for a worker-controlled firm to
adopt measures of this sort, such explanations for LMF rarity should be
rejected.
Proponents of workers’ control are often guilty of similar sins. It has
been argued, for example, that because LMFs give residual claims to
workers, they can elicit more effort than comparable KMFs. Leave aside
the question of why such productivity advantages do not enable LMFs
to drive KMFs out of business. The relevant fact is that profit-sharing in
KMFs also raises labor productivity among the firms adopting it, and the
same may be true when employees have some share ownership in a KMF
(Section 8.7). Profit-sharing gives workers a residual claim, but is quite
compatible with ultimate control by investors. Hence LMF proponents
are vulnerable to the retort that the alleged incentive benefits of the LMF
are achieveable by the KMF, so workers’ control is unnecessary. This
retort is not decisive (see Section 11.6), but the point is that concrete
organizational practices need not be linked inextricably to a particular
control structure.
Differences in the ways capital and labor suppliers would use con-
trol rights must in the last analysis stem from differences in the prefer-
ences or opportunities of the individuals who supply these inputs. For
instance, workers may be less able than investors to finance certain in-
vestments because of their lower personal wealth; they may have better
information on shop-floor conditions; they may be less able to diver-
sify their inputs across firms; they may be better able to make cred-
ible commitments of type x, but less able to make commitments of
type y; and so on. Explanations that rely on factors of this sort are
not vulnerable to a replication critique in any obvious way. I will ar-
gue in Chapter 11 that there is a small set of causal channels through
which characteristics of capital and labor lead to differences in the
preferences or opportunities of controllers, and thus to differences in

firm behavior.
6.3 Transaction Costs

In this section, I examine the explanatory strategies pursued by two
members of the transaction-cost school: Oliver Williamson and Henry
Hansmann. While they adopt similar conceptual starting points, these
authors pose the problem of LMF rarity in different ways and they high-
light different causal factors. My purpose in this section is not to rehash
old debates but to use the work of these authors to make some broader
methodological points.
Williamson (1980) provides an illuminating example of his method in
an article on pin-making at the time of the Industrial Revolution. This
example was chosen in part because it is well known from the writings of
Adam Smith, and in part because it was used by Marglin (1974) as the
basis for criticisms of capitalist organization. The context is that Marglin
argued against the efficiency of capitalist firms, while Williamson sought
to refute what he saw as the neglect of transaction-cost issues by Marglin.
Williamson’s own story was later criticized from various directions (see
Putterman, 1981; Willman, 1982; S. R. H. Jones, 1982, 1983; Williamson,
1981, 1983a, b; Szostak, 1989).
Williamson begins by specifying a discrete set of governance structures
that could conceivably be used to organize pin-making activities. These
vary on two dimensions: the ownership structure for non-human assets
(by individual workers, by workers collectively, or by an outsider), and the
nature of the contractual relationships (essentially, a distinction between
repeated bargaining and administrative ways of coordinating production).
Though this is not done in Williamson’s 1980 article, the resulting struc-
tures can be arranged in a 3 × 2 table with six cells. The three bargaining
modes are Federated (individual asset ownership), Communal (collective
asset ownership), and Inside Contracting (assets owned by a capitalist).
The three administrative modes are Putting Out (individual asset owner-
ship), Peer Group (collective asset ownership), and Authority Relation
(assets owned by a capitalist). Some of these organizational forms have
occurred historically, while others are hypothetical.
To identify which of these six governance structures is most efficient,
Williamson sets out eleven criteria for efficiency, including three for
product-flow attributes, three for assignment attributes, and five for in-
centive attributes. Each of the six structures is given a 0 or 1 on each
dimension. The overall ranking puts the Authority Relation in the lead
with 9 points out of a possible 11, followed closely by the Peer Group with
8 points, and the remainder trailing behind. This victory for the Authority
Relation is then used to argue against Marglin’s assertion that the capital-
ist mode of organization emerged historically for reasons of exploitation
rather than efficiency.
This explanatory strategy is open to various criticisms. First, the re-
sults are highly sensitive to the ways in which governance structures are
described. Putterman (1984: 177–78) noted in a subsequent review, for ex-
ample, that the Peer Group is weighed down by two arbitrary constraints:
It rotates leadership rather than using specialized managers, and it pays
workers on the basis of average product. The first constraint is costly be-
cause there is no reason to think that well-qualified people will be placed
in managerial roles, and the second implies that free-riding in the supply
of effort goes unchecked. Putterman appropriately objects that the game
has been rigged: Real LMFs need not labor under such self-imposed re-
strictions. This is a violation of the replication principle from Section 6.2,
because LMFs could certainly hire specialized managers and pay workers
based on their estimated productivity. To be fair, however, it should be
noted that Williamson’s interest in the Peer Group was motivated by his
desire to address critics of capitalism who favored something resembling
this structure.
Williamson’s conclusions are also highly sensitive to the list of included
efficiency criteria. Since the Authority Relation won by a single point, it is
obvious that the results could have been reversed by adding or subtracting
criteria. Williamson does not evaluate governance structures according
to whether they provide incentives for workers to disclose private infor-
mation to management, or guarantee workers that investments in firm-
specific human capital will be protected. Both criteria would tilt the scales
toward the Peer Group. Conversely, he omits a criterion for investment
in non-human assets. Judging by remarks he makes about collective as-
set ownership, Williamson clearly believes that including such a criterion
would favor the Authority Relation.
A number of other difficulties are apparent. The binary assessment of
a governance structure on each efficiency dimension is problematic (in an
earlier working paper, a four-point scale was used), as is the unweighted
summation over all eleven dimensions. In the absence of historical re-
search, one cannot know, but perhaps in the late eighteenth-century pin-
making industry, efficient product flow was vastly more important to a
firm’s success than strong effort incentives, or vice versa.
Finally, Williamson remarks: “Note that each of the eleven perfor-

mance statements that follow is of a ceteris paribus kind” (1980: 22). This
means that interaction effects are ignored; the benefit obtained by solving
a particular efficiency problem does not depend on whether some, all, or
none of the other problems have also been solved. This assumption is
necessary for an additive aggregation scheme to make sense, but one can
surely question whether it is a reasonable depiction of the economic situa-
tion. In reality, there may be substantial complementarities among certain
organizational practices (Milgrom and Roberts, 1990b; Holmstrom and
Milgrom, 1994).
Once again I want to stress that I am not concerned here with a two-
decades-old debate over the pin-making industry. Rather, I want to scru-
tinize the general method. In addressing the question of why LMFs are
rare, or why they are found in some industries but not others, transaction-
cost writers necessarily use methods similar to those described earlier.
The reason is that efficiency is interpreted in transaction-cost economics
as a scalar index that reflects the aggregate transaction cost of a particu-
lar governance structure in a particular environment, and the theory says
that actual modes of organization do a better job on this efficiency index
than any feasible alternatives.
Thus if one wants to explain why LMFs are rare in the petrochemi-
cal industry, it is necessary to argue that such firms would fail to econ-
omize on transaction costs as well as existing KMFs in that industry. It
is hard to see how one could do this in practice without making a list
of the dimensions on which transaction costs arise, assessing the perfor-
mance of LMFs and KMFs along each dimension, and then summing
over dimensions to obtain a comprehensive efficiency comparison. Such
a procedure is inevitably subject to biases and uncertainties that enter
into the description of each governance structure, the list of efficiency
dimensions adopted, the weight given to each dimension, the potential
neglect of interactions among dimensions, and the scoring system. If the
KMF gets 9 points while the LMF gets 8, how confident can one be that
transaction-cost theory has explained the prevalence of capitalist petro-
chemical companies?
Williamson has subsequently given much more weight to financing is-
sues, arguing that labor rarely obtains representation on the board of
directors because shareholders are at greater risk of opportunistic expro-
priation and therefore require control rights as safeguards (1985: ch. 12). I
accept the significance of this factor in Chapters 9–11, but it is interesting
that it does not appear anywhere on the 1980 list of efficiency criteria.
This again indicates the arbitrariness that can potentially enter into the
process of compiling such lists.
More importantly, Williamson’s 1985 explanatory strategy is somewhat
different. Instead of undertaking a detailed study of a specific industry,
the later Williamson cites one efficiency dimension common to many in-
dustries as the central reason for LMF rarity. A comprehensive balancing
of factors specific to an industry is no longer used. Presumably, however,
cross-industry variation in the prevalence of LMFs could still be dealt
with by identifying variations in the importance of this dominant causal
factor relative to others.
Henry Hansmann, another writer in the transaction-cost school, pur-
sues a similar strategy, although his dominant causal factor is the cost
of collective decision-making rather than the danger of opportunism to-
ward investors. The substantive merits of the collective-choice hypothesis
are considered in Sections 9.7 and 9.8, and are not of present concern.
Hansmann starts by assuming there are several stakeholder groups: in-
vestors, employees, customers, suppliers, and so on. Only one of these
groups will get ownership rights, which combine control rights and resid-
ual claims. The other groups will have a contractual relationship to the
firm. The cost of having stakeholder group j be the “owners” is equal
to the ownership cost for group j itself, plus the sum of the contracting
costs for all other stakeholders, where i = j. Predictions about the domi-
nant governance structure in an industry are generated by identifying the
group j that minimizes total cost (Hansmann, 1996).
In practice, Hansmann characterizes industries by the presence or ab-
sence of LMFs, considers several cost and benefit dimensions sequentially,
and in each case asks whether the importance of the factor in question is
correlated with the distribution of LMFs across industries. If not, he re-
jects it as a minor factor. Thus he tests impressionistically for the amount
of cross-industry variance in organizational form explained by each fac-
tor viewed separately. In the absence of good quantitative data, this is
not a bad strategy for inferring the relative importance of causal factors.
The limitations are twofold: Interactions among causal mechanisms are
not easy to spot because individual factors are compared against the data
one at a time, and judgments about relative importance depend heav-
ily on subjective assessments concerning the degree of fit with observed
cross-industry patterns.
What about efficiency explanations more generally? It is common
in the literature (and even more in casual conversation) to encounter
statements of the form “labor-managed firms are rare because they are
inefficient.” This does not qualify as an explanation unless more is said,

because it is necessary to cite some asymmetry between KMFs and LMFs
that would account for the alleged difference in their efficiency. At best it is
a heuristic device for seeking an explanation: If one thinks that markets fa-
vor efficient governance structures and discard others, then it may be help-
ful to look for differences in the efficiency features of KMFs and LMFs.
Williamson and Hansmann have taken serious steps in this direction.
There are, however, a number of hurdles that any such explanation
must surmount. Most obviously, it is necessary to define efficiency. The
next task is to justify the belief that efficient modes of organization are
more likely to be observed. The primary mechanisms that have been
proposed in the transaction-cost literature are foresighted organizational
design, learning, and evolutionary selection. For several reasons, it can
be debated whether these mechanisms accomplish the required theoreti-
cal job (Dow, 1987; Williamson, 1987; Hodgson, 1996). First, it is unclear
whether economic selection processes operate on the payoffs to individu-
als (thus promoting strategic behavior) or on the payoffs to groups (thus
promoting efficiency). Second, it is unclear whether ex ante side payments
are sufficiently unconstrained to permit a clean separation between effi-
ciency and distributional issues. If not, governance structures can prevail
as a result of the gains they deliver to particular stakeholder groups, rather
than to participants in general (Dow, 1993b, c). Third, the transaction-cost
school usually plays down ex ante informational asymmetries, which can
yield failures in the markets for capital and labor that may be highly rele-
vant for the survival prospects of KMFs and LMFs. Even when this issue
is discussed, as it is by Hansmann (1996), the implications for efficiency
often go unexamined (for more on this, see Section 6.5).
Transaction-cost writers usually assume that the comparative efficiency
of various governance structures can be assessed for a given firm in isola-
tion, without examining the equilibrium in surrounding labor and capital
markets. This rules out potential externalities across firms (see Sections 6.5
and 6.6). And finally, survival tests do not support the view that LMFs are
inefficient once they have been created. As I will discuss in Section 10.6,
LMFs often have lower failure rates than KMFs in direct survival contests.
The main problems for LMFs are that they are rarely created, and that
KMFs are rarely converted into LMFs. It may be possible to account for
this finding within the transaction-cost framework, but a straightforward
application of natural-selection ideas will not suffice.
For these reasons, the case for transaction-cost economics as an ex-
planatory strategy is less than overwhelming. Fortunately, one can accept
many of the substantive arguments developed by users of this framework

without accepting the efficiency baggage that comes along for the ride.
For example, in my view, evidence about the cross-industry distribution
of LMFs and their design characteristics favors the idea that collective-
choice problems are relevant in explaining the general rarity of LMFs.
This does not mean that the transaction-cost framework is the only way
to think about these problems, or that actual organizational forms are
necessarily efficient.
6.4 Optimal Contracting

Another important component of organizational economics is contract
theory. I will focus primarily on principal-agent models. Broader versions
of contract theory, including mechanism design, have played little or no
role in discussions of LMFs. The new property-rights framework of Hart,
Grossman, and Moore is sometimes classified under the rubric of contract
theory, but since this has been discussed in Section 5.1, it is unnecessary to
add much here. Principal-agent models have been used to study incentives
in partnerships and cooperatives as well as asset ownership by workers
(Sections 8.5 and 8.6). Related issues arise in the context of debt financing
(Section 9.2).
The model has the following structure. A principal (P) cares about the
choice made by an agent (A). For example, P may be an employer and A
may be a worker who chooses an effort level, where A’s effort affects firm
output and thus the revenue stream enjoyed by P. The problem is that P
does not observe A’s effort, or it is impossible to specify a required effort
level in a legally binding contract. However, the wage paid to A can be
conditioned on other things, such as the output produced. Unfortunately,
output depends not just on A’s effort, but also on random events not
observed by P. If output is low, this may be due to shirking by A, but
may also reflect events over which A had no control (the photocopier fell
apart at a crucial time). It is usually assumed that P is risk neutral and A
is risk averse.
Under these conditions, there is a tradeoff between having P provide
insurance to A through a fixed wage (which is attractive because A is risk
averse) and having P provide strong effort incentives to A by making the
wage depend on output (so A is less inclined to shirk). A standard result
is that an optimal contract involves a compromise between these two
objectives: A must bear some risk in order to have effort incentives, but
cannot bear all the risk because this is inefficient for insurance reasons
6.4 Optimal Contracting 127
(see Sappington, 1991; Mas-Colell, Whinston, and Green, 1995: ch. 14;
Salanie, 1997).
With a suitable reinterpretation of the model, the principal could be
a lender and the agent could be a borrower. Effort might then refer to
the resources used to identify suitable investment projects. The principal
could also be an insurance company, in which case the agent would devote
effort to reducing the risk of loss. In all of these contexts, the problem is
one of “moral hazard,” which occurs when informational asymmetries
arise only after a contract is already in place. The alternative is “adverse
selection” (see Section 6.5), where informational asymmetries arise even
before a contract has been accepted.
Formally, solving such a problem involves writing down P’s objective
(maximizing expected profit, for example), a participation constraint for
A (there is some minimum level of expected utility that A must be offered,
or A will reject the contract), and an incentive constraint for A (after
the contract is in place, A must find it optimal to choose the level of
effort demanded by P). In the employment context, this yields some wage
contract that is optimal for the principal, where a contract is a function
linking the wage with output. More complex models can involve many
principals, many agents, or many time periods, but it is sufficient here to
consider the simple case just outlined.
Some qualitative features of an optimal contract are noteworthy. First,
the agent’s participation constraint is generally binding, which means the
agent is indifferent between accepting the contract or rejecting it. Hence
all of the ex ante surplus from the relationship goes to the principal.
However, one could reverse the model by letting the agent propose a
contract to the principal. With a properly specified minimum acceptable
utility level for the principal, the same solution would be derived.
Another result is that the incentive constraint is generally binding.
In applications, this often implies some form of quantity rationing. For
example, in an employment setting, the employee would like to have more
insurance (less dependence of the wage on output), and may offer to work
at a lower average wage in return, but the employer will reject offers of
this kind because the employee would shirk under such a contract. A
promise made by the employee not to shirk lacks credibility in the eyes of
the employer. In applications to the credit market, the optimal contract
generally specifies that the borrower can have access to a fixed total loan
or line of credit, because larger loans create unacceptable incentives to
shirk or choose risky projects. In the case of insurance contracts, the agent
will be rationed through deductibles or co-payment clauses that force the
agent to bear risk. An agent who insists upon full coverage will be rejected
because this eliminates any incentive to take precautionary measures. It
can be shown that in each of these cases, the optimal contract is efficient
in a second-best sense: A social planner operating under information
constraints as severe as those facing the principal would be unable to
make both parties better off.
Thus, principal-agent theory, although differing from the transaction-
cost approach in many important ways, comes out in roughly the same
normative place. Assume that the utility level available to the agent by
exit to the outside market is fixed. If one believes that the rarity of LMFs
is explained by their failure to provide strong work incentives, and that
this implies a lower expected surplus in the principal-agent framework
than a KMF could achieve, then one is committed to the normative con-
clusion that LMFs are not efficient in a second-best sense. Hence it is
impossible to shift from a KMF to an LMF without making at least
one participant worse off. Alternatively, one might devise a principal-
agent model having the opposite implication, that LMFs are efficient,
while KMFs are not. The point is that, as in transaction-cost economics,
there is little or no separation between the normative and explanatory
agendas.
Some limitations on principal-agent theory need to be kept in mind
(Salanie, 1997). First, the minimum acceptable utility level for the agent
is exogenously given. The model thus involves partial equilibrium with
a vengeance. Not only is no attempt made to examine the economy as
a whole, there is usually no attempt to model equilibrium in the labor,
capital, and product markets where the outside options of the parties are
determined. Often when market interactions are taken into account, and
almost always in general equilibrium, the desirable efficiency properties
of principal-agent models break down (see Section 6.5 for similar remarks
in the context of adverse selection).
Second, all bargaining power in the model is exogenously assigned to
the principal. The situation involves an ultimatum game where P proposes
a contract to A, who either says yes or no. If A says yes, the optimal
contract goes into effect, and if A says no, both players get their outside
alternatives. In this game, the principal extracts all the surplus from the
relationship and leaves A indifferent between saying yes and no (given
indifference, A is assumed to say yes). One can also turn the model around
and have A propose a contract to which P says yes or no, which gives all
the bargaining power to the agent. But there is usually no attempt to
study cases where both parties have some ex ante bargaining power, a
point that could become relevant if the relative viability of organizational

forms depends on the upfront surplus appropriated by various classes of
input supplier (Dow, 1993b, c).
Third, it is assumed that there are no ex ante informational asymme-
tries. Before the principal and agent meet, each knows everything worth
knowing about the other. Thus it is only ex post, after the contract is in
place, that an informational problem arises (in which P cannot observe
A’s effort or the random state of the world relevant for A’s productivity).
The problem is therefore one of moral hazard, not adverse selection (see
Section 6.5 on the latter). This matters because the second-best efficiency
of optimal contracts need not carry over to adverse selection models in
which there is private information ex ante.
Finally, a vital assumption is that wages can be made dependent on
output in a legally binding way. Because of this, the agent need never
fear that the principal will cheat by refusing to pay the promised wage.
The moral hazard problem is therefore one-sided; it is only the agent
who might cheat. But it may be difficult to assess the quality of output,
or the courts may be too costly or unreliable as enforcers of employment
contracts. The moral hazard problem then becomes two-sided because
the principal is also tempted to renege on the wage. This leads to the
analysis of repeated games (Section 6.6).
The “new property rights” school of Hart and his co-authors (Sections
5.1 and 8.3) resembles contract theory in a number of respects. Neither
attempts to model equilibrium on the wider markets in which the firm
participates, although the property-rights approach takes a modest step
in the direction of endogenizing outside options by suggesting that they
depend on the assets an agent owns when she leaves the firm. Neither ad-
dresses the point that side payments among the players may be restricted
by their wealth, or that bargaining inefficiencies can arise due to private
information. Finally, both assume that organizational structures are cho-
sen ex ante to maximize the joint surplus of the parties, and in this sense
both are committed to second-best efficiency explanations.
6.5 Adverse Selection

“Adverse selection” can be distinguished from “moral hazard” through
the timing of the information problem. Under adverse selection, some par-
ties have private information about their own characteristics even before
any contract is proposed or accepted, while in a moral hazard model, there
are no informational asymmetries prior to contract acceptance. Used-car
markets provide the classic example of adverse selection: Sellers typically

know a lot about the true quality of the car they are offering for sale, while
buyers cannot easily distinguish a good car from a bad one (Akerlof, 1970).
Similar phenomena arise in labor markets, where a job applicant may
know more about her true skills than the employer does (Weiss, 1990); in
capital markets, where a borrower may know more about the true riski-
ness of a proposed investment project than the lender does (Stiglitz and
Weiss, 1981); or in insurance markets, where the buyer of insurance may
know more about the risk of a loss than the insurer does (Rothschild and
Stiglitz, 1976).
In such models, the price paid by buyers (assumed here for expositional
purposes to be the uninformed market participants) equals the expected
value of the goods or services offered by sellers (assumed to be the in-
formed participants). But when the market price is below the value to a
seller of keeping the good, the seller stays out of the market. Thus the ex-
pected value to the buyer must reflect voluntary decisions by sellers about
whether or not to participate in the market, given the prevailing price.
Under reasonable assumptions, this can lead to an equilibrium where the
highest quality goods, most productive workers, least risky projects, and
so on, drop out of the market, leaving only the less desirable goods or
services. In extreme cases, the entire market can collapse, so that none
of the potential gains from trade are realized. The same holds if it is the
buyers who are informed and the sellers who are not, as is more likely to
be true in an insurance market.
Adverse selection can lead to inefficiency in partial-equilibrium labor-
market models, where firms take prices and wages as given, at least if mul-
tiple equilibria exist (Mas-Colell, Whinston, and Green, 1995: ch. 13). In
such models, the equilibrium with the lowest wage is not efficient because
an uninformed social planner (one confronting the same information con-
straints as the uninformed employers in the model) can make every agent
at least as well off, and some better off, by imposing a minimum-wage
law. This occurs because low-skill workers receive a higher wage, high-
skill workers move into the labor market and capture some surplus not
previously available to them, and employers have zero profit both before
and after the minimum-wage law due to free entry by firms.
Such inefficiencies do not inevitably arise for adverse-selection models
in a partial equilibrium setting. But efficiency becomes quite problematic
from a general equilibrium perspective because adverse selection virtu-
ally guarantees the existence of tax or subsidy policies that can make
everyone better off. Similar remarks apply to moral hazard, as in the
principal-agent models from Section 6.4. The reason is that incentive and
information problems are like externalities for the economy as a whole,
and implementation of suitable taxes or subsidies can correct these ex-
ternalities (Greenwald and Stiglitz, 1986, 1988).
Because under adverse selection it is the sellers with the most desir-
able goods or services who drop out of the market, there are incentives
for these sellers to find ways of separating themselves from their lower-
quality rivals so they can capture gains from trade. Simply claiming to be
a high-quality seller will not work, because anyone can make such a claim.
But there may be signals that are rather cheap for high-quality sellers to
send and very costly for lower-quality sellers. This offers a credible way
for the former to separate themselves from the latter. Examples include
warranties on used cars (cheap if the car is good because it is unlikely
to need repairs, but expensive if the car is bad and very likely to need
repairs), and investing some of one’s own wealth in a project in order
to convince a bank to extend a loan (cheap if the project is good, but
expensive if failure is likely).
Signaling need not be efficient. It is easy to construct partial-
equilibrium examples in which an uninformed social planner can make
every agent at least as well off, and some better off, by banning the signal-
ing activity (Mas-Colell, Whinston, and Green, 1995: ch. 13). This makes
low-quality sellers better off because they are no longer distinguishable
from high-quality sellers. It can sometimes make high-quality sellers bet-
ter off if signaling is costly and there are few low-quality agents in the
population to depress the equilibrium price. Finally, if uninformed buy-
ers break even in both situations, they are indifferent.
These points are potentially relevant for attempts to explain the rarity
of LMFs. It is commonly argued, for example, that LMFs are rare because
workers lack wealth and must attract outside financing to establish firms
(Sections 9.1–9.4). But, it is said, if they try to borrow from banks, such
workers will face credit rationing as a result of moral hazard or adverse
selection. Capitalist investors with enough personal wealth will not con-
front this problem. Using the adverse-selection version of the story, an
inefficient equilibrium could arise in which some good projects available
to LMFs are not funded because adequate credit is not available. Sig-
naling is not necessarily a solution since even if workers did invest some
of their own limited wealth in an LMF to convey their confidence in the
project to lenders, the signaling equilibrium could easily be inefficient.
One can also devise other stories about failures in the markets for labor
or insurance that might hamper the creation of LMFs.
Such inefficiencies are not trivial in the sense of being inevitable given
the relevant informational constraints. Rather, policy interventions de-
signed by governments with no special access to information, but with
powers to tax, subsidize, and regulate, could make everyone better off.
Hence one should not assume hastily that the rarity of LMFs reflects some
intrinsic and irremediable design flaw. Instead, one should acknowledge
that LMFs could be rare because of market failures for which remedies
can be devised. This possibility does not surface in transaction-cost eco-
nomics or principal-agent theory because both focus on individual firms
taken one at a time, with outside options exogeneously given. From an
explanatory standpoint, it may be more illuminating to examine equilib-
ria in the markets for labor, capital, and insurance, and to endogenize the
organizational structure of firms as one component in this wider analytical
framework.
6.6 Repeated Games

Contractual agreements enforced at low cost through the courts are not
the only way to support cooperative behavior. In addition to judicial en-
forcement, two other mechanisms have been explored by economists –
market enforcement and internal enforcement (Hart and Holmstrom,
1987: 141–42).
Market enforcement arises when third parties external to a relation-
ship can observe events occurring within that relationship and take actions
to punish behavior deviating from existing agreements, conventions, or
norms. Firms that cheat employees by reneging on promised wages, for ex-
ample, may have difficulty in hiring future workers. If consumers see that
a firm has cheated on promises about product quality, each may decide
not to buy from the firm in the future, even if they were not personally
victimized (Klein and Leffler, 1981). When adverse selection is an im-
portant factor, it is common to discuss equilibria in reputational terms.
Firms may want to maintain a reputation for good behavior because if
they do not, other parties will draw negative inferences about their true
characteristics and refuse to do business on favorable terms in the future.
Sometimes, however, outsiders lack reliable information about the
behavior of the parties within a relationship. It is often impossible for
workers in other firms to discover whether a firm cheated an employee
by withholding wages for no good reason, or whether (as the employer
claims) this was done in retaliation for gross misconduct. Likewise it may
be impossible for other employers to know whether a worker was fired
6.6 Repeated Games 133
for shirking, or (as the worker claims) this was an unjust act by a spiteful
boss. When information flows in a labor or capital market are thin and
unreliable, market enforcement breaks down because of the inability of
third parties to administer sanctions when and only when they are needed.
This leaves internal enforcement, meaning enforcement strategies that
are internal to the relationship itself. A familiar example is the version of
efficiency-wage theory in which firms offer employees a premium above
the market-clearing wage, and threaten to fire those who are caught
shirking (Shapiro and Stiglitz, 1984). Similarly, employees might retaliate
against a firm that fails to honor its promises, even if there is no legally
binding contract involved (Bull, 1987). Related models can be used to
address relationships between a firm and its main bank where the firm’s
line of credit is cut off if it is caught taking undue risks. Baker, Gibbons,
and Murphy (1999) model authority within the firm in a parallel way.
As Hart and Holmstrom (1987: 74–75) note, the partition of economic
relationships by the nature of the enforcement mechanism (judicial, mar-
ket, internal) corresponds to a division in terms of modeling strategies.
Transactions for which judicial enforcers exist are modeled as constrained
optimization problems, where the problem is to discover the terms of an
optimal contract. When judicial enforcers are absent, however, it is neces-
sary to rely on models of non-cooperative equilibrium. For pure internal
enforcement, this involves a repeated game framework, while market
enforcement requires that repeated game concepts be supplemented by
some notion of market equilibrium. This is also necessary for models with
internal enforcement if the outside options of the parties are treated as
endogenous.
A need for repetition of the basic game arises because deviations from
equilibrium behavior today must be deterred by a credible threat of re-
taliation tomorrow. This requires either infinitely many repetitions or, in
practice, some uncertainty about when the last round of the game will
occur, because with a known finite end point, there is nothing to pre-
vent a player from deviating in the last round. In games of the prisoners’
dilemma variety, where there is a tension between group cooperation and
individual self-interest, cooperation may then unravel all the way back to
the first round of the game.
Economists commonly refer to patterns of behavior supported by mar-
ket or internal enforcement as implicit contracts (Bull, 1987; MacLeod
and Malcomson, 1989). In my view, this terminology is misleading and I
will dispense with it. In such situations, there is no contract in the legal
sense, and what is really involved is a non-cooperative equilibrium in a
repeated game. There may be metaphorical value in the concept of an

implicit contract if a non-cooperative equilibrium mimics the solution of
a constrained optimization problem. But there is no reason to attach these
connotations to all repeated game equilibria, especially those where an
agreement among the parties is lacking or there are serious inefficiencies.
Infinitely repeated games can have many equilibria. A well-known
technical result, the Folk Theorem, shows that if agents put enough weight
on future payoffs, then almost any feasible array of payoffs can occur in
equilibrium (Fudenberg and Tirole, 1991: ch. 5). The only qualification is
that no player can be driven below the payoff she can guarantee herself
by choosing her best reply to the actions of the other agents. The Folk
Theorem shows that efficient payoffs can potentially be achieved in a
repeated game. However, in the real world, people do discount future
payoffs. If this discounting effect is large enough, then efficient outcomes
may be excluded. Even if agents put enough weight on the future to make
efficient outcomes possible, there is no guarantee that such equilibria
will actually occur. Since there are usually many equilibria, coordination
failures can arise in which the prevailing equilibrium is worse for everyone
than some alternative equilibrium.
Why is all this relevant to issues of organizational form? An illustration
may help to show what is at stake. In a highly influential article, Holmstrom
(1982) investigated effort incentives in partnerships using a principal-
agent framework where the principal was able to make legally binding
commitments about the way in which wages of individual workers would
vary with the total output of the production team. He concluded that
partnerships, and by implication LMFs, would be disadvantaged relative
to capitalist firms for incentive reasons (this argument will be discussed
in more detail in Section 8.5).
This conclusion depends on the assumption that wage agreements can
be enforced judicially. If instead they must be self-enforced, in a one-
shot game the principal has no reason to pay the promised wages, and
the incentive system of the capitalist firm collapses. This can be avoided
in a repeated game setting, but then moral hazard becomes bilateral:
Workers can shirk and quit, while employers can cheat on wages and
fire workers. Upon solving for the market equilibrium as a whole, one
finds that there is no incentive advantage for capitalist firms. Any allo-
cation of resources that can be supported by firms having an outside as-
set owner or “principal” can also be supported by firms where worker
teams own assets jointly and no principal exists (Dow, 2000). Within
this framework, the equilibrium organizational form is a matter of social
convention or historical accident. Unless groups of agents can coordi-

nate their strategies, a departure from the usual assumption for a non-
cooperative game, inefficient market equilibria can persist.
6.7 Historical Explanations

The preceding review of explanatory strategies was carried out roughly
in order of the importance assigned to efficiency as a theoretical premise.
The time has now come to consider strategies that leave conventional
economic theory behind altogether.
An important intellectual strategy of this sort involves the idea of path
dependence. The shortest statement of this idea is offered by Lindblom,
who remarks: “Not by logic but by history, owners of capital have be-
come the owners of the enterprise” (1977: 105). A less compact version
runs like this. The organizational structures of firms have co-evolved with
other factors: the law, regulatory institutions, technology, and the wealth
distribution, for example. Thus the current environment is biased in fa-
vor of KMFs. Their dominance is an accident of history, which could have
unfolded in a way that favored LMFs instead.
Such claims involve two causal links. First, there is a short-run link
running from the exogenous institutions, technology, or wealth distribu-
tion to endogenous organizational structures (KMF or LMF) that must
be adapted to, or compatible with, the background environment. The
more distinctive feature of the story is that the background environment
is endogenized via a long-run dynamic link running from the prevalence
of KMFs back to the institutions, technology, or wealth distribution of
the society. Perhaps over quite long historical periods, these background
conditions adapt to suit the needs of capitalist firms.
There is wide agreement among economic historians that legal insti-
tutions played an important role in the development of modern capital-
ism (Rosenberg and Birdzell, 1986; North, 1990; Field, 1991), but there
is disagreement over the claim that modern law is tilted against LMFs.
Ellerman (1984a) argues that legal codes are biased in assuming that cap-
ital supply and control rights must be bundled together. Accordingly, he
believes that it is hard to design a firm based on the principle of one
vote per worker without also forcing workers to supply equal amounts
of capital. On the other hand, Hansmann (1996: 86) grants that minor
legal problems may exist in setting up an LMF under U.S. law, but be-
lieves that this factor is too small to explain the general rarity of such
firms.
An argument linked more directly to the early legal framework for

firm governance has been advanced by Everett and Minkler (1993). They
point out that LMFs need to issue financial claims with variable returns
in order to shift risks, but these claims must be non-voting if workers’
control is to be maintained. Such financial instruments would probably
not be viable under a system of unlimited liability, because then outside
investors would be forced to risk their entire personal wealth in supplying
funds to a firm over which they had no control. LMFs therefore have a
special need for limited liability, but this did not become widely available
in Britain until the Companies Act of 1856. By then, capitalist firms had
a long head start and other economic institutions had already evolved to
accommodate these firms, placing LMFs at a disadvantage from which
they never recovered.
Other path-dependence stories involve technology. Given an arbitrary
starting point involving capitalist firms, perhaps technological innovation
has been pushed in a direction complementary to the organizational struc-
ture of such firms. The result in the long run, it could be argued, has been
to lock in production techniques (perhaps involving large scale, high capi-
tal intensity, or generic skills for workers) for which investor control is the
most suitable mode of organization. This feedback loop thus steered the
economy away from workers’ control. Some writers have studied tech-
nological innovations they believe were motivated by efforts to lower
employee bargaining power (Braverman, 1974; Noble, 1977), a process
that has been formalized by Dow (1985) and Skillman and Ryder (1993).
The most ambitious theoretical move in this direction is that of Pagano
(1993), who modifies transaction cost economics to include feedback
from the property-rights structure of firms to technology, in addition to
the conventional causal link running from technology to property-rights
arrangements. Pagano and Rowthorn (1996) develop a formal model
of “organizational equilibrium,” defined as a situation in which two
optimality conditions hold simultaneously: (1) firms choose production
techniques to maximize profits given existing property rights, and (2)
property-rights arrangements within firms maximize profits given existing
production techniques. Pagano and Rowthorn suggest that this can be
understood as something like a Nash equilibrium in a non-cooperative
game, where the players are firm managers in case 1 and financiers in case
2. They go on to show that multiple equilibria can arise, where in one equi-
librium, investor-controlled firms use specialized equipment and generic
labor, while in the other, worker-controlled firms use specialized skills and
generic equipment. In this model, an equilibrium need not be efficient.
The most straightforward path-dependence story involves the interac-

tions between governance structures and wealth distribution. One such
model is supplied by Bowles and Gintis (1996b). They argue that LMFs
are limited by worker wealth, combined with the fact that workers are
risk averse and do not want to put all of their wealth into a single firm
(see Chapter 9). Conditions of wealth inequality thus favor the creation
of capitalist firms. But Bowles and Gintis also argue that there is a feed-
back effect through which a capitalist economy perpetuates inequality. If
more workers were in democratic firms, they would bear more risk. As-
suming risk aversion is a decreasing function of wealth, such workers
would save more, and over time this would promote greater wealth
equality in the society. At an arbitrary moment in time, perhaps at the
beginning of the Industrial Revolution, an unequal distribution of wealth
was historically given. Capitalist firms thus prevailed, and the ongoing
inequality of the wealth distribution reproduced capitalism. But if the
starting point had been egalitarian, worker-controlled firms would have
been more common and would have stabilized the initially egalitarian
wealth distribution.
A final path-dependence idea involves the concept of macroeconomic
externalities, and has been proposed by David Levine along with several
co-authors (Levine and Tyson, 1990; Levine, 1993; Levine and Parkin,
1994; Levine, 1995). Suppose LMFs do not lay off workers when demand
falls, because the members prefer to absorb demand fluctuations through
earnings rather than layoffs. The practice of avoiding layoffs is costly to
the firm if recessions are long and deep, and may threaten the survival
of the enterprise. Suppose, on the other hand, that KMFs hold nominal
wages fixed over the business cycle and absorb reductions in demand
through layoffs. If KMFs are dominant, their practices will aggravate the
business cycle for macroeconomic reasons, leading to longer and deeper
recessions than in an economy where LMFs were the dominant mode of
organization. Thus if KMFs are already dominant, an LMF may not be
viable, but if LMFs were already widespread, the resulting dampening of
the business cycle would enable these firms to flourish. A few pieces of
empirical evidence bearing on this idea will be reviewed in Section 10.7.
Without attempting any overall evaluation of the various theories
sketched here, it can be noted that each needs to identify some factors that
currently stabilize the economic system in favor of KMFs. Thus it should
be possible to screen path-dependence theories in two stages. First, the
proposed short-run mechanism explaining why LMFs are now at a disad-
vantage relative to KMFs can be investigated. If evidence on this part of
the theory is supportive, it may be worth undertaking historical research

to assess the proposed long-run feedback loop running from firms to the
institutional or technological environment. But if there is no evidence for
the short-run causal channel from environment to firms, the theory as a
whole is in trouble. Unfortunately, many path-dependence theories fail to
make contact with detailed empirical evidence about the internal design
of LMFs, their distribution across industries, their rates of formation and
failure, and so on. This makes it difficult to conduct the first stage in the
evaluation process described.
6.8 Cultural Explanations

LMFs have often been created by groups at odds with the surrounding
mainstream culture. Many utopian communities of the 1800s and “alter-
native” enterprises of the 1970s were motivated by a desire to secede
from capitalist society (on the latter, see Jackall and Crain, 1984). An-
other regularity apparent from the case-study literature is that LMFs
have often been established by people sharing a distinctive ethnicity or
religious identity, even when no overarching ideological motivation was
evident.
Examples are readily cited. Scandinavian immigrants had a large role
in the creation of the first plywood cooperative, and the distinctive fea-
tures of Basque history and culture are routinely mentioned in dis-
cussions of the Mondragon system (Sections 3.2 and 3.3). Members of
San Francisco-area scavenger cooperatives were almost universally of
Italian background. Kinship connections among them were common, and
membership rights were often passed down intergenerationally to male
relatives (Russell, 1985a: ch. 3; Benham and Keefer, 1991). Russell also
notes the role of Jewish immigrants in the taxi cooperatives of Boston
and Los Angeles. Benham and Keefer discuss the Hutterite religious
communes, whose members have close kinship ties and little contact with
the outside world.
One argument is that some specific cultures are predisposed to favor
the formation of LMFs or to support their long-run survival, while others
are indifferent or hostile toward LMFs. The concrete elements that might
place a culture in one category or the other are not often spelled out, but
one can speculate that groups favoring egalitarian values, personal reci-
procity rather than anonymous market exchange, or loyalty under condi-
tions of insecurity or external threat might support LMFs. A danger with
this sort of argument is that it is easy to spin out cultural rationalizations
6.8 Cultural Explanations 139
for observed cases of LMF success, while ignoring other societies with
parallel cultural features but few LMFs. However, there are cases where
the story fits well. For example, Pencavel explains the lack of plywood
coops in the southeastern U.S. in terms of independently measured dif-
ferences in “social capital,” especially trust levels, for the northwestern
and southeastern regions of the country (2001).
A different variety of explanation stresses the value of homogene-
ity or uniformity among LMF members, whatever the underlying social,
ethnic, or religious basis for this uniformity may be. This is clearly what
Benham and Keefer have in mind in discussing Mondragon, the U.S.
scavenger cooperatives, and the Hutterite communes. These authors ar-
gue that homogeneity among the members of LMFs tends to reduce the
costs of collective choice (see Sections 9.7–9.8). Gamson and Levin (1984)
have pointed out that many small U.S. cooperatives growing out of the
countercultural movement of the 1960s were based almost entirely on
shared opposition to mainstream institutions, and often failed because
the members had no shared culture defining their obligations to the firm
itself.
There have been some efforts to link the existence of multiple equi-
libria in repeated games to the idea of culture. Kreps (1990) and Miller
(1992: chs. 9–10) suggest that firms differ in the social conventions they
adopt to guide the behavior of employees in repeated game environments,
and speculate that these conventions might be altered in productivity-
enhancing directions. Ideas of this sort are often invoked in an informal
way by authors who suggest that profit-sharing and employee share own-
ership will not lead to productivity gains without simultaneous changes in
corporate culture to promote employee participation in decision-making
and to enhance job security. Repeated game ideas have also been used to
model cultures in a wider sense. Weitzman and Xu (1994), for example,
suggest that Chinese township and village enterprises succeeded because
the surrounding society had a culture of high trust that facilitated coop-
eration despite poorly defined property rights.
Cultural stories often have obvious impressionistic merit. My view is
that cultural factors are helpful in understanding why organizational mu-
tations have occurred in specific times and places, but do not generally
explain why LMFs persist in particular industries or why LMFs are trans-
formed into KMFs and vice versa. In short, culture has little apparent
connection to the design features and selection pressures that account for
economic success or failure. To understand the latter, the hypotheses to
be discussed in Chapters 7–11 seem more promising.
6.9 The Strategy to be Pursued

Given the enormous range of explanatory strategies that could be em-
ployed, what strategy will be adopted here? Chapter 7 will start by using
the symmetry principle to clear away some confusion about the goals
of LMFs and to critique the intellectual history of the LMF in the field
of comparative economics. Chapters 8 and 9 will undertake a systematic
review of hypotheses about LMFs culled from the new institutional eco-
nomics and modern microeconomic theory. Because most of the hypothe-
ses in these chapters only address the static distribution of LMFs across
industries, Chapter 10 will separately address organizational transitions
from KMF to LMF or vice versa. Finally, Chapter 11 will attempt to syn-
thesize those ideas that have survived theoretical and empirical scrutiny.
Mäki (1998b) suggests that science generally has two major objectives:
unification, and understanding the causal structure of the world. Roughly
speaking, the first involves maximizing the number of things that can be
explained using a given set of causal factors, while the second involves
minimizing the number of causal factors used to explain a fixed set of
things. The present project seeks greater unification in the sense that I
would like to account for several things simultaneously: the aggregate
rarity of LMFs, their distribution across industries at a point in time, their
life-cycle features (especially their probabilities of formation, survival,
and transition to KMF), and the design features that successful LMFs
often have. It is by no means clear that it is possible to unify all of these
within a single theory, because the underlying causal processes may simply
be different. One test of this will be the degree of theoretical coherence
achieved in Chapter 11.
Where I think it is clearly possible to make some progress is in min-
imizing the set of causal forces needed to explain a given set of facts.
A great many hypotheses have been proposed to explain the rarity of
LMFs. Many of these hypotheses have been exposed to little or no sus-
tained criticism, either with respect to their internal logic or the weight of
the evidence. Attempts to compare rival hypotheses directly have been
few (some recent exceptions include Bonin, Jones, and Putterman, 1993;
Putterman, 1993; Hansmann, 1996; Dow and Putterman, 1999, 2000; and
Ben-Ner, Burns, Dow, and Putterman, 2000). It may be fairly easy to dis-
pose of some unpromising hypotheses. During this process of elimination,
it may also become apparent that certain other hypotheses are relatively
robust in the sense that they are consistent with much of the evidence and
are rarely in serious empirical trouble.
6.9 The Strategy to be Pursued 141
To close with an appropriate quote from Mäki: “A good theory is real-

istic in the sense that it contains significant truth, that it truly represents
the way the relevant domain of the world works – if it captures the ma-
jor elements of the causal processes and mechanisms responsible for its
workings.” (1998b: 314) The next five chapters attempt to deliver a good
theory of workers’ control.
7
A Question of Objectives
7.1 What do LMFs Maximize?

Economists have long debated what the LMF maximizes. From the
standpoint of the modern theory of the firm, it is not obvious that
LMFs maximize anything. Firms are complicated coalitions in which each
agent has her own preferences, information, and feasible actions. These
coalitions face difficult coordination, bargaining, and incentive problems.
Whether the strategic actions of individual labor and capital suppliers
translate into the pursuit of some objective by the coalition as a whole
is very much open to debate.
Early writers made strong claims about the behavior of LMFs: They
were assumed to maximize net income per worker. This objective has
implications for input demand and output supply that differ drastically
from those of profit maximization, and not usually in attractive ways.
Many writers skeptical of LMFs seized on these perversities as a way of
explaining why LMFs are rare. More surprisingly, many proponents of the
LMF took the hypothesis of per capita income maximization to heart, and
thereby dug themselves into a deep analytical hole from which it is hard
to climb out. I will argue here that it is time to stop digging. The model
in question is not plausible in theory, it bears little resemblance to the
organizational practices of real LMFs, and it has only modest empirical
support.
This chapter is organized as follows. First, I will retrace some intellec-
tual history by summarizing the properties of the “Illyrian” or per-capita-
income-maximizing LMF. Next, I criticize this model by applying the
symmetry principle discussed in Chapter 6: Why would KMFs maximize
142
absolute profit, while LMFs maximize net income per unit of an input?
Reflection on this question leads to a theory of LMFs with markets for
membership rights. Such firms have precisely the same behavioral prop-
erties as the profit-maximizing firms of textbook theory. In particular, it
can be shown at the level of theory that a membership market is a perfect
substitute for the conventional labor market.
After developing these ideas in a static context, I consider a number
of criticisms of the LMF that center on investment decisions. It was often
claimed in the early literature that LMFs tend to underinvest because
workers cannot capitalize any investment returns occurring after their
planned departure from the firm. This proposition has become a piece
of folk wisdom about LMFs, but a market for membership solves the
problem in precisely the same way that a stock market solves the parallel
problem for a capitalist firm.
After disposing of the alleged static and dynamic perversities of the
LMF at the level of theory, I turn to a different problem: the question
of why explicit membership markets are relatively unusual despite their
attractive theoretical properties. There are a number of possible expla-
nations, prominently including the inalienability of labor. This suggests
that in practice, membership markets are unlikely to achieve everything
theory would predict.
Direct econometric tests of alternative behavioral hypotheses about
the LMF are few, but some have been carried out. For the most part, these
indicate that LMFs do differ from similar KMFs in their responses to ex-
ogenous demand or cost shocks, with LMFs having somewhat less elastic
output supply and highly inelastic labor demand (at least with respect
to membership; many LMFs hire non-member labor in ways that do not
differ much from KMFs). This rigidity of the LMF would be a bad thing
if input and output markets were perfectly competitive, because theory
shows that profit-maximizing firms already meet the usual conditions for
allocative efficiency in a competitive economy. It does not take a huge
departure from competitive assumptions to alter this assessment, how-
ever, and I will argue that the LMF’s reluctance to shed labor may actually
be a good thing in the real world.
7.2 The Illyrian Firm

The earliest formal model of the labor-managed firm was provided by
Ward (1958) in his classic article “The Firm in Illyria: Market Syndical-
ism,” where the title alluded to the system of workers’ self-management
144 A Question of Objectives
prevailing at the time in Yugoslavia. Later authors extended Ward’s

model and used it to study the Soviet collective farm (Domar, 1966),
as well as a hypothetical economy consisting entirely of LMFs (Vanek,
1970). Although the firm in this theoretical framework could be termed
the “Ward-Domar-Vanek” firm after these pioneers, it will be convenient
to refer to it simply as the Illyrian firm.
The key assumption of the model is that an LMF chooses levels for all
of its input and outputs, notably including labor itself, to maximize the
net income of the firm per unit of labor. Comparative static analysis of
the Illyrian model predicts a number of differences between LMFs and
profit-maximizing firms. LMFs have a less-elastic supply response to an
increase in output price because they limit membership when firm profit is
positive, in order to restrict the size of the group entitled to share in these
profits. In extreme cases, the output supply curve can bend backward, so
that increases in output price reduce both the quantity supplied and the
number of firm members. Other comparative static perversities include
a tendency for output to increase when fixed costs rise, because then it
is desirable to spread the firm’s higher fixed cost over more members. A
profit-maximizing firm has no short-run response to changes in fixed cost.
Reviews of such comparative static results are available in Ireland and
Law (1982), Stephen (1984), and Bonin and Putterman (1987).
A second problem with the Illyrian firm is that it leads to an inefficient
allocation of labor across firms in the short run. There is no guarantee
that when various different firms maximize net income per worker, the
resulting income per worker will be the same in each. But if not, then
the value of labor’s marginal product will also differ across firms. In this
situation, it is possible to obtain efficiency improvements by shifting labor
from firms where the value of its marginal product is low to firms where
it is high. No such problem exists in a competitive labor market with
profit-maximizing firms because all firms will equate the value of labor’s
marginal product to the same equilibrium wage rate. Only in a long-
run competitive equilibrium with free entry of new firms, where pure
economic profit is driven to zero, does the problem of inefficient labor
allocation disappear among LMFs (see Vanek, 1970; Meade, 1972; Dreze,
1976). But for many industries, free entry is not a reasonable assumption,
and in this case the problem persists even in the long run.
To see the logical difficulties of the Illyrian model it is helpful to con-
sider Figure 7.1, which shows the value of labor’s average and marginal
product as a function of labor input for a competitive firm. The input of
labor is denoted by L, the firm’s output is Q, the price per unit of output
Figure 7.1. Income maximization in the Illyrian firm
is p, and net income per worker is y. To carry out the analysis with the
fewest distractions, I assume that labor is the only input and there is no
fixed cost. I also assume that each worker supplies one unit of labor, so
L can be viewed as the number of worker-members in a labor-managed
firm. All of these assumptions can be relaxed without affecting the cen-
tral points of the analysis. By standard definitions, average product at
input L is AP(L) = Q(L)/L and marginal product is the rate of change
in total output evaluated at L – that is, MP(L) = Q/L (this becomes
a derivative in the continuous case).
An Illyrian firm maximizes the value of labor’s average product (which
is identical in this context to net income per worker) by choosing the labor
input L0 . The resulting income per worker is y0 = p·AP(L0 ). I assume this
exceeds the wage w each member can get on the external labor market –
for instance, by working for a capitalist firm – so y0 > w, as in the graph. If
y0 < w occurred instead, the LMF would shut down because its members
would leave to take jobs in capitalist firms. The assumption y0 > w implies
that pure economic profit is positive because y = p·AP(L) = pQ/L > w
implies pQ − wL > 0. Thus revenue is greater than labor cost, which
is the only cost in the model. Textbook analysis shows that a profit-
maximizing firm would operate at the larger input level L∗ where the
value of the marginal product of labor equals the market wage – that is,
p·MP(L∗ ) = w.
The problem with the Illyrian analysis can now be spotted. Suppose
there are many people working in capitalist firms at the market wage
w. Because all of them are receiving less than the Illyrian income yo,
they would all like to join the LMF. Moreover, there are bargains that
can be reached between these outsiders and LMF insiders that would
make everyone better off. Because the Illyrian firm maximizes labor’s
average product, it must operate at the level of labor input where the
average and marginal product curves of labor intersect in Figure 7.1 (this
is a mathematical necessity). But this implies that the value of labor’s
marginal product at L0 must exceed the outside wage w. It follows that an
outside worker can offer to join the firm in exchange for some payment
between p·MP(L0 ) and w – for instance, w1 . This is clearly better for
the outsider than working in a capitalist firm. The insiders in the LMF
are also better off because the value of the output produced by a new
member is p·MP(L0 ), but they only have to pay the newcomer w1 to
obtain this output. The surplus p·MP(L0 ) − w1 > 0 can be divided among
the existing LMF membership.
Unless such arrangements are explicitly prohibited, the original pair
(L0 , y0 ) cannot be an equilibrium, and there is no justification for thinking
that the LMF will maximize the value of labor’s average product. The
logic of the preceding argument suggests that further deals will be made
with new applicants until the value of labor’s marginal product equals
the outside wage, because prior to this there are still mutually beneficial
bargains that can be struck between insiders and outsiders. But if the
equilibrium result is the labor input L∗ at which p·MP(L∗ ) = w, there is
no difference between the labor input of the LMF and an identical profit-
maximizing firm. The only circumstance in which the preceding argument
would not apply (barring complete shutdown of the LMF because y0 < w)
is when y0 = w, so that pure economic profit is zero. In this situation,
the firm must already have p·AP = p·MP = w, and again there is no
departure from profit maximization.
The flaw in the Illyrian model is that it suppresses the labor market
by assuming that the LMF ignores the gap between what insiders and
outsiders are paid. Ward (1958) was quite aware of this, and explicitly
discussed the rigidities resulting from the absence of a conventional labor
market. It can be debated whether such rigidities actually arose in the

economic system prevailing in Yugoslavia during the 1950s, which was
the system Ward set out to model. But nothing about the principle of
workers’ control requires the absence of a labor market, so it is misleading
to treat this “missing market” assumption as a defining characteristic of
the LMF rather than as one possible hypothesis about LMF behavior. As
an empirical matter, most real LMFs charge entry fees (see Chapters 3–4),
so bargains are in fact struck between insiders and outsiders, and a labor
market therefore exists.
The problems with the Illyrian model go deeper still, because it as-
sumes that when there is an exogenous increase in output price, the firm
will expel some members without compensation in order to raise income
per person for those who remain behind. Only those who stay could ac-
tually benefit from such decisions. It is conceivable that this is a minor-
ity of the initial membership, raising the question of how a minority of
workers would manage to expel a majority. But even when it is a ma-
jority that can gain by expelling co-workers, it is unclear how a specific
group of victims would be chosen. A minority about to be thrown out has
strong incentives to suggest a different list of victims and demand another
vote.
Scenarios of this sort do not even remotely resemble the governance
procedures or property rights found in real LMFs, a point made repeat-
edly by early critics of the Illyrian model (for one example, see Robinson,
1967). LMFs are renowned for maintaining membership levels under ad-
versity while sharing economic pain through lower wages or hours for
their existing members, and rarely expel members except for poor work
performance. Moreover, LMFs usually compensate departing members
for their past financial contributions to the firm (see Chapters 3–4 and
the sources cited there). This conflicts with the Illyrian assumption that
LMFs expel some members without compensation to maximize per capita
income for those who remain.
Elaborations of the Illyrian model in the 1970s and 1980s showed that
some glaring comparative static perversities (for instance, the backward
bending output-supply curve) did not extend to multiple inputs or out-
puts, to variable hours or effort, to environments with uncertainty, or to
situations in which the LMF was indirectly exposed to the wage available
on the external labor market. There is no need to discuss these epicycles
at length, because the Illyrian model is simply the wrong starting point
for understanding LMF behavior.
7.3 Membership Markets and Labor Markets

The long detour caused by the Illyrian model is puzzling in light of the
symmetry principle from Section 6.1. Without an underlying asymmetry in
the nature of capital and labor, either it should make no difference which
set of input suppliers chooses the firm’s production plan, or the KMF and
LMF should suffer from symmetrical distortions. The Illyrian framework,
though, identifies no qualitative asymmetry between capital and labor. If
one sensibly assumes that capitalist firms maximize the absolute level of
profit rather than net income per unit of capital, why doesn’t parallel rea-
soning (whatever it may be) apply to a “laborist” firm? Or, if LMFs are
believed to maximize net income per unit of labor, why don’t KMFs maxi-
mize net income per unit of capital, rendering the KMF equally perverse?
An important step in thinking about the problem is to recognize that
there is nothing distinctively “capitalist” about the profit-maximizing firm
that inhabits the microeconomic textbooks. As Putterman (1988a: 247)
remarks, the neoclassical or Walrasian firm merely consists of a set of
feasible production plans with an attached “owner” who chooses a plan
and gets the resulting net income. The “owner” maximizes total profit for
the usual reason that consumers seek to relax their budget constraints
whenever possible. But there is no reason why this owner would person-
ally supply capital, labor, raw materials, or any other input. In Walrasian
general-equilibrium theory, every input is a hired input (assets can be
leased if capital services are needed), and in this sense every input plays a
symmetrical role in the theory. Casual identification of the Walrasian firm
with “capitalist” firms, as opposed to “laborist” firms, which are somehow
different, is unwarranted.
One good reason for modeling capitalist firms using Walrasian
assumptions is that there is an institutional device, the stock market, that
generally induces capital suppliers to support the maximization of total
profit or present value. I return to the role of the stock market later in
this chapter, but the point here is that the same reasoning can be applied
to “laborist” firms. Given a suitable institutional framework, it should
be possible to induce labor suppliers to support profit maximization by
an LMF. Furthermore, Section 2 offered a clue about the nature of the
required mechanism. Because the perverse behavior of the Illyrian firm
can be traced to a neglect of the labor market, what is needed is a way to
model labor-market transactions for the LMF.
By the mid 1980s, it had become evident that the solution was to
introduce a market for membership in LMFs, which under competitive
conditions is a perfect substitute for the usual labor market (Sertel, 1982;
Dow, 1986, 1996). Just as the stock market induces the members (share-
holders) of a KMF to support profit maximization rather than maximiza-
tion of net income per unit of capital, a membership market induces
LMFs to maximize profit rather than net income per worker. The reason
is that the membership market forces LMF insiders to value labor at its
external opportunity cost, and therefore leads them to operate at a point
where the value of labor’s marginal product equals the outside wage. The
result is comparative static behavior identical to that of an ordinary profit-
maximizing firm, as well as an efficient allocation of labor across firms in
both the short and long run.
To see how this works, consider Figure 7.1 and start at the Illyrian
solution (L0 , y0 ). What is the price of a membership deed in this situation?
An outsider currently receives the capitalist wage w, but by switching
places with an insider would receive y0 . The outsider is therefore willing
to pay anything up to y0 − w for the membership rights already held by
an insider. Competition among outsiders drives the price of membership
up to this level.
Now suppose the insiders are debating whether or not the firm should
expand. If they decide to do so, the insiders create a new membership
deed and sell it to an outsider. In return, the new member will supply
labor and get an equal per capita share of the firm’s net income. Suppose
a new member adds some small amount L to the firm’s labor input, so
L1 = L0 + L is total input after expansion, and let Q1 = Q0 + L·MP0
be the associated output, where MP0 is the marginal product of labor
evaluated at L0 .
Every outsider knows that if she buys the new membership deed, total
labor will be L1 and she will receive the income y1 per unit of labor after
joining. Therefore an outsider is willing to pay up to (y1 − w)L for
membership: y1 − w is the net gain per unit of labor from joining the
LMF rather than working in a capitalist firm, and L is the number of
units the new member will supply (equal to one under the assumptions of
Section 7.2). Again, competition among outsiders drives the price of the
membership deed up to this ceiling.
The total revenue collected by the insiders from whichever new-
comer happens to be selected is (y1 − w)L. This revenue is divided
equally among the insiders. Expansion thus gives a typical insider y1 +
(y1 − w)L/L0 , where the first term is the usual income per worker af-
ter the new worker joins, and the second term is the insider’s share of
the newly obtained membership revenue. Expansion is desirable if this
sum exceeds y0 . Substituting y = pQ/L for dividends where appropri-

ate, some algebra shows that the required inequality holds if and only if
p·MP0 > w – that is, if the value of labor’s marginal product exceeds the
outside wage.
But this is precisely the same as the condition under which a profit-
maximizing firm would expand. It is easy to reverse the argument by as-
suming that an LMF that wants to reduce its labor input must buy back the
membership rights of one or more insiders. This guarantees that contrac-
tion occurs if and only if p·MP0 < w, which is the condition under which
a profit maximizing firm would contract. The LMF is thus in equilibrium
only when p·MP = w, which is the necessary condition for profit max-
imization from textbook theory. Because this condition forms the basis
for any comparative static analysis, the LMF with a competitive member-
ship market has exactly the same responses to changes in output price as
a profit-maximizing firm. The tendency of the Illyrian firm to restrict its
membership in response to an increase in output price disappears.
This line of argument is perfectly general and extends to a firm with any
number of inputs and outputs, a firm with fixed costs, and so on. The same
intuition applies in every case: Because current insiders internalize what
outsiders are willing to pay for membership, they treat the external wage
as the true cost of labor. This removes any tendency toward maximization
of net income per unit of labor. Instead, the insiders are concerned with
net income per unit of labor plus membership revenue, and the second
item ensures that their actual objective function is total profit. For other
expositions of this point, see Sertel (1982, 1987, 1991), Stephen (1984:
38–40), Fehr (1993), and Dow (1996).
An alternative way of thinking about the role of the membership mar-
ket is to notice that it brings about expansion or contraction whenever
there are gains from trade between insiders and outsiders. Mutual gains
can be derived from expansion if the value of labor’s marginal product
is above the outside wage, and conversely mutual gains can be derived
from contraction (having an existing member exit) if the value of labor’s
marginal product is below the outside wage. The membership market
plays exactly the same role as any other competitive market: It exhausts
these gains from trade. When applicants compete to join the firm, a new
member derives no surplus from the membership transaction because
(net of the membership price) the new worker gets the outside wage. All
surplus is therefore appropriated by the insiders. Since the latter inter-
nalize all of the social benefits associated with expansion or contraction,
they make these decisions efficiently.
This arrangement fills the gap created by the missing labor market in
Illyrian theory. In the same way, the membership market removes the
Illyrian problem of inefficient labor allocation across firms. Since LMFs
all treat the external wage as the cost of labor, they all equate the value
of labor’s marginal product to a common wage just as in a capitalist labor
market. If all firms are LMFs, then the external wage becomes a shadow
wage, rather than an explicit wage paid by capitalist firms, but the idea is
the same (Dow, 1996).
This is perhaps a good time to dispel some frequent misconceptions
about the operation of a membership market. First, it does not necessarily
impose harsh financial burdens on new applicants. For a perfectly com-
petitive firm that leases all of its physical assets and has zero economic
profit, the equilibrium price of membership is zero, because outsiders are
indifferent between joining a KMF and an LMF. The equilibrium price
will be positive but small for a labor intensive firm in a competitive in-
dustry that is obliged to use collective asset ownership in place of leasing.
The conditions under which a market for membership would yield a high
equilibrium price include a high degree of capital intensity, inability to
lease assets, and a low debt to equity ratio. But these are circumstances
under which any LMF would have difficulty financing its operations (see
Chapter 9); the problem is not attributable to membership markets per se.
In such cases, an LMF might allow a new member to pay off the required
amount over time through payroll deductions.
Second, it is important to recognize that the model differs fundamen-
tally from the original Illyrian model in a key respect. Even if a firm wants
to reduce its size, no worker is ever expelled against her will (Dow, 1996).
Incumbent members have property rights that cannot be violated arbi-
trarily. Under adverse conditions, the membership rights of some insiders
may be bought back by the firm at a mutually acceptable price, but this
is not the same as expulsion. Indeed, whenever economic profit can be
increased by reducing the firm’s labor input, the equilibrium membership
price induces insiders to leave voluntarily in exchange for the repurchase
of their positions.
A common normative objection is that membership in a worker-
controlled firm is a “personal right” rather than a “property right”, and
therefore should not be for sale (see, for example, Ellerman, 1984a). Taken
literally, this objection rules out any fee for admission to an LMF, so
most of the LMFs in Chapters 3–4 are violating these alleged personal
rights. But those who advance the argument are often willing to let LMFs
charge entry fees as long as these fees are construed as contributions to
the working capital of the firm, or as a way of ensuring commitment to

the enterprise. What is unacceptable is setting the entry fee at a market-
clearing level. There is nothing, however, in the concept of a personal
right that would explain why this particular price should be ruled out. It
is also hard to understand why someone would be entitled to share in the
profits of a workers’ cooperative without first paying some compensation
to the existing members of that cooperative.
The normative objection to membership markets based on the idea of
personal rights involves a misconception about workers’ control. Surely if
this principle means anything, it means that the current labor suppliers are
the people entitled to determine whether the firm will expand, whether
an applicant for membership is acceptable, and the terms on which an
applicant can join. But if an LMF can regulate admission in such ways, it
is reasonable to think that it can set a minimum price to be paid by new
members. If this price happens to be set at a level where the demand for
membership positions equals the supply, so be it.
7.4 Membership Markets and Stock Markets

A major claim about LMFs in the early literature was that such firms
would tend to underinvest. The idea was that members who planned to
leave the LMF in the near future would have no stake in the returns on
long-term investments, preferring instead to be paid a high wage today. If
most of the membership planned to leave in five years, for example, there
would be little support for an investment project whose returns would
start to arrive six years from now, especially if it had to be financed out
of retained earnings that could have been used to pay wages instead. As
a result, LMFs would not make some investments that a capitalist firm
would regard as attractive. The same reasoning applies to maintenance
activities whose returns extend beyond the expected departure date of
most members. Thus there is a danger not only that the LMF will fail to
grow at the same rate as a KMF, but that it will allow its existing capital
stock to depreciate over time without engaging in repairs, replacement,
or modernization.
This has become known as the horizon problem. A closely related phe-
nomenon, the common-property problem, arises because future members
will share in the investment returns made possible by the sacrifices of
current members. The latter group will therefore hesitate to make invest-
ments that require admission of new members. These criticisms of LMFs
were advanced by Pejovich (1969), Furubotn and Pejovich (1970), and
Furubotn (1971, 1976), and were later taken up by Jensen and Meckling
(1979). As with the Illyrian hypothesis, concerns about the horizon prob-
lem were initially voiced in the context of the Yugoslav self-management
system, but then were quickly generalized to Western LMFs.
Vanek (1971a, b) noticed the horizon problem at roughly the same
time as Furubotn and Pejovich. He concluded that external financing was
essential to the success of LMFs, and that the past failures of such firms
could be attributed to a reliance on retained earnings. One attractive
aspect of debt financing, in particular, was that the repayment schedule
could be synchronized with the arrival of investment returns. Vanek later
observed that individual capital accounts could substitute for external
financing if these accounts were recoverable with interest when work-
ers left the firm. Ellerman (1984a, b, 1986) also advocated the use of
individual capital accounts. The ghost of the horizon problem lingers in
recent work by Jossa and Cuomo (1997), who recommend that LMFs
use debt financing rather than retained earnings to finance investment
projects.
The claim that horizon and common-property problems afflict LMFs
but not KMFs raises a warning flag under the symmetry principle. One’s
immediate response should be to look for a market failure that has been
smuggled into the definition of the LMF but not the KMF. In this case,
it is not necessary to look very far. The KMF does not suffer from an
underinvestment problem because those who hold shares in the firm can
sell their shares when they wish to exit, and the market value of the
shares will reflect whatever the present value of future investment re-
turns is believed to be. Therefore, a shareholder in Toyota or Siemens
who plans to sell out tomorrow may favor an investment project that
will not pay off for twenty years, as long as someone else will buy the
shares at a price reflecting the present value of future returns from the
project.
But Jensen and Meckling assert explicitly that an LMF has no trade-
able claims on investment returns: “What the term ‘labor-managed’ really
means is that the models being used presume there are legal prohibitions
against the existence of tradeable residual claims on the entire sequence of
future cash flows generated by the firm” (1979: 475, emphasis in original).
The same assumption is implicit in the criticisms of Pejovich and Furubotn,
and immediately leads to the horizon problem. In discussing Western
cooperatives, which they distinguish from Yugoslav LMFs, Jensen and
Meckling further assume that coops “will admit new members without
charge” (1979: 499), ensuring a common-property problem.
This is puzzling on several fronts. First, workers’ cooperatives usually

do charge new members an admission fee (see Chapters 3–4), so the asser-
tion that LMFs have no such fees is incorrect. Second, nothing about the
concept of workers’ control requires a ban on tradeable residual claims,
and the plywood cooperatives actually have such claims. As long as the
firm is managed by its current labor suppliers at each point in time, it
qualifies as an LMF. This does not conflict with the practice of having
new members pay a price to join, or having departing members receive a
price for their membership rights when they leave. A precisely analogous
procedure is used among shareholders in publicly traded capitalist firms,
which qualify as KMFs because current capital suppliers have votes at
each point in time. Finally, in most Western economies there is no legal
prohibition on membership transactions. If some LMFs do not permit
them, that is a matter of choice, not law.
The solution is again a market for membership rights. Such a mar-
ket enables each departing member to appropriate the present value of
future investment returns by selling her membership position, either to
a replacement worker or back to the firm itself. When new members
enter, they must either buy a membership position from an incumbent
or a newly created position from the firm itself. This eliminates both
the horizon and common-property problems by allowing current work-
ers to capitalize future returns and requiring newcomers to compensate
incumbents for sharing in the income streams generated by past invest-
ments. A membership market therefore induces LMF members to make
investment decisions that maximize present value and are identical to
the decisions made by a KMF with access to a stock market (Dow, 1986,
1993a).
The idea that tradeable membership rights induce present-value max-
imization by an LMF, and that such markets are analogous to stock mar-
kets, was expressed as least as far back as Schlicht and von Weizsäcker
(1977). Remarkably, however, this point continues to generate confu-
sion. Nuti (1996), for example, criticizes Schlicht and von Weizsäcker for
failing to recognize that an LMF with a membership market would maxi-
mize present value per job rather than total present value. This complaint
attests to the continuing influence of the Illyrian model, but it is incor-
rect, apparently because Nuti has overlooked the effects of side payments
between newcomers and incumbent members. Whatever the source of the
misunderstanding, it should be clear from Dow (1986, 1993a) that the rel-
evant maximand is total present value, not present value per person, per
job, or per hour of labor.
This conclusion applies whether the LMF owns or leases physical assets,
and is not affected by the method of financing (debt, equity, or retained
earnings). The incentive role of membership markets in ensuring effi-
cient investment even under a system of collective asset ownership is
particularly important. Because the underinvestment concept is deeply
ingrained, those with some sympathy for LMFs often feel obliged to rec-
ommend rental of all non-human assets or full debt financing of invest-
ment. But neither alternative is really feasible (see Chapters 8–9), so
LMFs inevitably appear defective relative to KMFs, which combine col-
lective asset ownership with a stock market.
Other authors with less sympathy for the LMF assume by definitional
fiat that such firms are “forbidden to hold claims or rights in durable pro-
ductive resources like those held by individuals” (Jensen and Meckling,
1979: 477). More recently, Pejovich has repeated this claim: “Employ-
ees . . . have non-tradeable claims on the firm’s cash flows; . . . neither
employees nor anyone else have transferable rights in the firm’s capi-
tal assets” (1994: 226). While such assumptions may apply to the case of
Yugoslav self-management, which is now only of historical interest, they
are analytically crippling if one wants to understand how LMFs function
in contemporary Western economies. Of course, one can accept this con-
ceptual point while believing that membership markets are impractical
in the real world. Such arguments need to be developed explicitly (see
Section 7.5).
Other devices for circumventing the horizon and common-property
problems do not yield the same equivalence with the KMF stock market.
For example, Mondragon’s use of individual capital accounts (Section 3.3)
enables the system to secure loan financing from its members. There is
nothing wrong with this practice, but members still cannot capitalize the
value of their membership positions on departure, and therefore do not
get any signals about the present value of future investment returns.
I am not arguing that if only LMFs would establish membership mar-
kets, all would be well. Explicit markets of this kind are relatively unusual,
and many successful LMFs, including Mondragon, get by without them.
But the unusual nature of membership markets is puzzling, given their
strong theoretical advantages and the fact that KMFs routinely use stock
markets. Such asymmetries provide valuable clues in deciphering the rea-
sons for the rarity of LMFs themselves. However, these clues cannot be
interpreted properly until it is recognized that in a world of complete
and competitive markets, KMFs and LMFs would be symmetrical in their
static and dynamic behavior. This proposition provides a vital point of
departure for more realistic comparisons between the two governance

structures.
7.5 Imperfect Membership Markets

Examples of LMF membership markets do exist. U.S. cases include the
plywood cooperatives in the Pacific Northwest and the scavenger coop-
eratives in the San Francisco Bay area. In both cases, new applicants can
buy membership rights from departing workers at a mutually agreeable
price, subject to approval from other members (on the plywood case, see
Craig and Pencavel, 1992, 1993, 1995, and Pencavel and Craig, 1994; on
scavenger coops, see Russell, 1985a: ch. 3). A membership market also
evolved informally in two Los Angeles taxi cooperatives during the 1970s,
but it is unclear whether other cooperative members had to approve these
transactions in advance (Russell, 1985a: 122–23). There is historical evi-
dence for tradeable memberships in two more sets of U.S. firms, cooperage
and shingle-weaving cooperatives, although details are sparse (Stephen,
1984: 159–66).
Many European cooperatives ask new members to pay an upfront
fee (see Chapters 3 and 4). These fees may not always be set at a market-
clearing level (for a remark on the existence of queueing among applicants
at Mondragon, see Bonin, Jones, and Putterman, 1993: 1295). However,
in the French cooperatives, a significant fraction of the workforce often
consists of employees who could become members in exchange for a
nominal fee but choose not to do so (Estrin and Jones, 1995), and this also
seems to be the case in many Italian cooperatives. It may therefore be true
that most European coops have market-clearing fees, and Mondragon is
an exception. In any event, where the European coops clearly diverge
from the U.S. examples mentioned here is that membership rights are not
assigned any market value when a worker leaves. Instead, workers are
given refunds on their individual capital accounts that are unrelated to
the present value of future profits.
One common story is that membership markets are unusual because
workers cannot afford to buy their way into a firm. Wealth constraints will
be discussed more generally in Chapter 9, but a few comments are ap-
propriate here. The equilibrium price of membership reflects the present
value of the difference between the income stream a worker would get
from an LMF and whatever stream of wage payments the worker would
get on the external labor market, assuming that non-pecuniary factors are
equivalent. The size of this gap is determined by the value of jointly owned
assets per worker in the LMF as well as rents in the product market, if any.
As noted in Section 7.3, this gap is small for a labor-intensive firm in a com-
petitive industry, and zero if all physical assets can be leased or financed
by debt. Thus wealth constraints cannot explain the lack of membership
markets in industries approximating this description. Furthermore, some
workers do have personal savings, and it is unclear why LMFs where jobs
are in excess demand would not ration those jobs using a price mecha-
nism that is profitable to insiders. In any case, LMFs often impose upfront
fees on new members. What they usually do not do is allow workers to
trade membership rights on an open market, and this prohibition is what
requires explanation.
It might seem that risk aversion would be the main obstacle to mem-
bership markets, since workers would need to put a substantial chunk
of their financial portfolio into a single asset whose returns are highly
correlated with the returns on their human capital. There is some evi-
dence that membership shares in the plywood cooperatives have been
consistently underpriced (Craig and Pencavel, 1992), in the sense that the
present value of membership turned out to be substantially larger ex post
than the price paid ex ante. Craig and Pencavel suggest that this apparent
underpricing may reflect implicit compensation for risk.
A complete discussion of the diversification hypothesis will also be
postponed until Chapter 9, but again some comments are in order. If
workers are concerned about risk, this should of course affect membership
prices, just as risk affects the prices of other assets. A risk discount on a
membership claim does not imply a market failure, any more than the risk
discount on a share of common stock implies a market failure in the stock
market. It also needs to be remembered that the relevant comparison for
present purposes is between LMFs with membership markets and LMFs
without them. Most of the force of the risk-aversion argument is already
spent once LMFs exist, because assets have somehow been financed and
workers have chosen to bear the risks of running a firm. Market prices
for membership merely repackage these risks, so it seems implausible
that risk aversion can explain the decision by many LMFs not to use
membership markets.
A deeper explanation stems from the inalienability of labor. When
control rights are tied to labor supply, it is impossible to transfer control
from person A to person B without replacing A’s labor services with B’s
labor services. But in a corporation, the voting rights tied to a share of
common stock can be passed from A to B without any modification in
the physical assets of the firm. A’s role as a partial owner of these assets is
simply transferred to B. The alienability of non-human assets thus makes

it possible to separate the market for KMF control from decisions about
the use of physical inputs within the firm. The market for LMF control,
however, is subject to all of the same frictions as the labor market itself.
A catalogue of labor-market frictions is easily compiled. First, it is dif-
ficult to be in two places at once, and travel costs are often significant,
so most workers hold only one job at a time. Because of costs associ-
ated with search, turnover, and relocation, workers also tend to change
jobs infrequently. Hence, the membership market for an LMF will only
become active intermittently as jobs open up. The heterogeneity of la-
bor services implies that each job is likely to involve distinct skills, and
thus a separate membership market, as contrasted with the KMF, where a
single stock market suffices for the entire firm. Because some jobs may at-
tract relatively few qualified candidates, a membership market can easily
degenerate into bilateral bargaining between an applicant and the cur-
rent insiders, or an applicant and a departing member. Bargaining of this
sort will usually involve private information because outsiders may know
little about the firm’s true financial prospects, while insiders may know
little about the true abilities or preferences of applicants.
A membership right is also likely to be an illiquid asset. Because trans-
actions are infrequent and require advance approval by other members
(perhaps for adverse-selection reasons as discussed later), workers can-
not necessarily convert their membership rights into cash quickly when
cash happens to be needed. Thus they may be reluctant to hold the mem-
bership asset in the first place. This lack of liquidity could help to explain
the apparent underpricing of membership rights in the plywood coops ex-
amined by Craig and Pencavel, although it surely does not explain all of
it. It has also been suggested that plywood coops have seldom expanded
through the creation of new membership shares because there is no well-
established central market on which these shares could be sold (Berman,
1982).
In contrast, investors can hold partial claims on the assets of many
different firms simultaneously. These claims can be traded almost instan-
taneously, with a high volume of trade for the shares of most large firms.
Informational asymmetries concerning firm value are mitigated by the ex-
istence of some informed traders, so that significant information is built
into a firm’s share price. Moreover, there are scale economies in doing re-
search on the underlying values of firms. Putterman (1993: 251) points out
that the equity and debt claims of smaller firms are generally not traded
openly on financial markets because given the thinness of the market,
it would not be profitable to gather and disseminate information on the

financial condition of the firm. This problem is more severe in an LMF
membership market because no one other than a potential supplier of
labor to the firm will care about the value of its membership rights.
There are also asymmetries with respect to takeovers. A small group
can acquire control of a conventional firm by buying out a majority of
its shareholders. But a takeover in an LMF poses major collective-action
problems because outsiders must assemble a team capable of supplying
51 percent of the firm’s labor services and buy out the membership rights
of 51 percent of the current workforce. This effectively disables takeover
threats as a strategy for disciplining managers. However, the balance in
this respect is not tilted entirely against the LMF: Takeovers in stock
markets are also subject to free-rider problems (see Grossman and Hart,
1980; Stiglitz, 1985). LMFs may also have less need for takeovers because
workers have easier access to information on managerial performance.
One proposed reason for the rarity of LMFs relies on asymmetries be-
tween stock markets and membership markets in a relatively explicit way.
Milgrom and Roberts (1992: 331–32) argue that investors should man-
age firms for reputational reasons. Their point is that a firm’s reputation
for selling quality goods or treating suppliers fairly is a valuable com-
modity, and could be severely damaged by managerial decisions aimed
at reaping gains from short-term opportunism. Such inefficient decisions
are less likely to occur if the controllers of the firm can easily sell their
positions to replacement controllers when they leave the firm. The incum-
bent decision-makers will then take into account the full present value of
the firm’s reputation, because they will be concerned about the price at
which their positions in the firm can be sold in the future. Milgrom and
Roberts argue that this is more likely if control rights are assigned to eq-
uity investors than workers, since the latter have to change jobs in order to
transfer their claims on the firm, while the former only need to sell their
shares. Whatever the merits of this story may be, it does pass method-
ological muster because it cites an underlying asymmetry between capital
and labor to justify the idea that membership markets are an imperfect
substitute for stock markets.
Another source of asymmetries between the two kinds of market in-
volves adverse selection, and the fact that labor is highly heterogeneous
while financial capital is not. In their roles as decision-makers, residual
claimants, and asset owners, LMF insiders take a keen interest in the
characteristics of any new member. A severe externality problem arises if
departing members are allowed to sell their membership positions directly
to prospective replacements because the worker leaving the firm will not
be concerned with the quality of the replacement worker and will sell out
to the highest bidder, whatever the cost to insiders who stay behind. To
appreciate the problem, an academic economist need only imagine a sys-
tem in which departmental colleagues can transfer their professorships to
whoever will pay them the most money. A workable membership market
requires that a member sell her position back to the LMF so that the
insiders can internalize quality effects when seeking a replacement, or at
least that insiders have a veto over proposed membership transactions.
This is incompatible with anonymous trading of the sort found in stock
markets.
The adverse-selection problem has further implications for member-
ship markets. In general, labor services and physical assets may both have
quality variations that cannot be observed ex ante. A KMF locates these
problems in separate markets, each with one-sided adverse selection:
the labor market and the equity market. An LMF instead combines them
by creating a membership market with two-sided adverse selection, where
insiders do not know the quality of new applicants and outsiders do not
know the value of joining the firm. This can lead to equilibria in which
membership transactions reduce total welfare, so from a social standpoint
it would be best to ban them (Dow, 2002, unpublished manuscript).
A separate problem with membership markets is that workers usually
have diverse preferences about the policies to be pursued by a firm, while
outside investors agree on the maximization of profit or present value. For
reasons to be explained in Section 9.7, this can induce a majority of LMF
insiders to sell out to investors, who then convert the firm into a KMF.
Real LMFs typically circumvent this problem by making it impossible for
individual members to sell their positions, and their control rights, to out-
siders of their own choosing. This solution, although perhaps necessary to
the continuing viability of the LMF, in effect shuts down the membership
market.
An imperfect membership market can have serious consequences for
LMF viability. Suppose LMFs enjoy static labor-productivity advantages
over KMFs but suffer a dynamic disadvantage because a stock market
fully capitalizes investment returns while membership markets do not.
Under these conditions, LMFs invest less in collectively owned assets and
grow more slowly than otherwise identical KMFs (Dow, 1993a). For cer-
tain parameter values, there are perverse incentives for current members
to transform LMFs into KMFs by selling out to capitalist investors, al-
though in present-value terms, the total surplus from an LMF exceeds
7.6 What does the Evidence Say? 161
the total surplus from a KMF. The reason is that current LMF members
do not completely internalize the benefits to other workers from join-
ing the LMF in the future. Transformation into a KMF enables current
members to appropriate a larger present value at the expense of these
prospective future members.
Having said this, it is important to bear in mind that LMFs do ex-
ist, despite having either imperfect membership markets or none at all.
The frictions surrounding membership markets are not inevitably fatal to
LMFs, which may have some offsetting advantages (see the discussion of
labor productivity in Section 8.7). There are other ways to price future
income streams, including a reliance on informal compensation packages
when a member departs. And finally, if the rate of labor turnover is low,
the horizon problem will not be severe.
7.6 What does the Evidence Say?

The empirical verdict on the Illyrian model is mixed. The behavior of the
plywood cooperatives deviates from the predictions of the Illyrian model
in various ways, although some tests fail to reject the model and others
show that the behavior of the coops is clearly different from profit max-
imization. Smith (1984) finds that plywood coops give positive weight
to both employment and income per worker, a result later confirmed
by Craig and Pencavel (1993). This conflicts with the Illyrian hypothesis,
which says that income per worker is the sole objective. The work of Craig
and Pencavel is particularly interesting because they use a general objec-
tive function that includes the Illyrian model, profit maximization, and
employment maximization as special cases, with profit maximization as
an intermediate case between the extremes of income and employment
maximization. Their results indicate that the plywood coops fall between
profit and employment maximization. Thus the coops assign more weight
to the employment objective and less to income per worker than a con-
ventional profit-maximizing firm, the opposite of the Illyrian prediction.
But in other statistical work, Pencavel and Craig (1994) were unable to re-
ject various theoretical aspects of the Illyrian model, and Pencavel (2001,
personal communication) regards this model as fitting the facts about
plywood coop behavior remarkably well.
All researchers who have studied the question find that plywood
cooperatives have virtually no short-run response of membership size
to demand or cost shocks. There is no evidence that members are ever
expelled in response to favorable (or unfavorable) demand conditions.
Most authors who have done econometric work on the plywood cooper-
atives have also failed to reject the hypothesis of a zero short-run output
supply elasticity. Thus a change in the price of plywood neither increases
nor decreases output. This result has been reported both by Berman and
Berman (1989) and by Craig and Pencavel (1995). Indeed in the 1950s,
when their aggregate market share was considerably larger, the coop-
eratives were notorious within the industry for maintaining output re-
gardless of demand, thereby driving prices down even further in bad
economic times (Berman, 1967: ch. 12). In earlier work, however, Craig
and Pencavel (1992) found a small positive supply response. There is lit-
tle evidence for a backward-bending supply curve, although one weak
exception is reported by Berman and Berman (1989).
Because the cooperatives have much less elastic input and output re-
sponses than the conventional plywood mills, pure profit maximization
cannot provide an adequate theory of coop behavior. To put the con-
trast concisely, the coops tend to respond to external shocks by adjusting
earnings, while conventional mills adjust the quantities of inputs and out-
puts. Since models based on perfect membership markets predict profit
maximization, one can infer some degree of failure in the membership
market. The source of the imperfection is unclear, although whatever fac-
tors are involved may also account for the underpricing of membership
shares found by Craig and Pencavel (1992). At the same time, Berman
and Berman (1989) cannot reject the hypothesis that labor inputs are
allocated efficiently across plywood firms in the short run. This is not a
strong test of any particular behavioral model, and even with imperfect
membership markets the result is unsurprising given routine hiring of
non-member labor by the plywood cooperatives (see Section 3.2).
The apparent difference in objectives between cooperatives and con-
ventional firms leads to econometric conundrums. As Craig and Pencavel
(1995) point out, one can either assume that both firm types maximize
profit and infer from the data that they use different technologies, or as-
sume that they use identical technologies and infer that their behavior is
different. Impressionistic reports stress the nearly identical technology in
the two kinds of firm (see Section 3.2), lending support to the idea that
real behavioral differences exist.
There is less evidence on the underinvestment issue. After reviewing
studies prior to 1993, Bonin, Jones, and Putterman conclude that “no
strong empirical support for the underinvestment hypothesis is found
either in France or the U.K.,” and their overall view is that “the em-
pirical literature contains no econometric support for this hypothesis”
7.7 Some Lessons 163
(1993: 1311, 1316). While there have been no attempts to estimate

investment demand functions for KMFs and LMFs operating in the
same industry, Estrin and Jones (1998) do estimate such a function for a
sample of French cooperatives. They find no evidence that collective asset
ownership impairs investment, although they do find that the volume of
investment is positively related to the debt-equity ratio. This is consistent
with the hypothesis that coops are rationed in the credit market, and that
when debt financing becomes more available, they invest more. Also
relevant is the fact that many European cooperatives have constitutional
rules mandating a minimum level of investment out of net income. This
places investment decisions beyond the reach of majority voting, at least
to some degree (Bonin, Jones, and Putterman, 1993: 1311). While such
rules may lead to undesirable rigidities, they are one means through
which LMFs can overcome the horizon problem in practice.
7.7 Some Lessons

The two central lessons for LMF design coming out of this chapter are
the need to expose insiders to the true opportunity cost of labor so they
will make efficient decisions about the expansion or contraction of the
firm, and the need to compensate members for the investment returns
they forego on departure. If explicit markets for membership are not
feasible, then alternative institutional procedures should be developed
to handle these tasks. Otherwise the LMF may be subject to “degener-
ation,” where members are gradually replaced by hired labor over time
(a process discussed in Section 10.4) and may hesitate to invest out of
retained earnings. Of course, institutional details will depend on the cir-
cumstances, but a variety of organizational solutions are suggested in
Chapter 12.
Whether the difficulties and compromises that necessarily follow from
imperfect or missing membership markets provide any leverage in ex-
plaining the rarity of LMFs is yet to be determined. One would anticipate
that successful LMFs have found ways around the problems described
here, or have emerged in niches where these problems are not very se-
vere. If assets are easily leased or financed through debt, for example,
there is less need to finance investment out of retained earnings, and im-
perfect membership markets will have little effect on investment. The
same is true when labor turnover is low.
When competitive markets exist for all goods and services, profit-
maximizing firms satisfy standard theoretical conditions for efficient
resource allocation. The hypothesis of profit maximization is rejected

empirically for the plywood cooperatives, so it can be argued that they
violate these efficiency conditions. But one can also argue that under
competitive conditions, non-profit-maximizing firms should be driven out
by profit maximizers. This has not proven true in the U.S. plywood indus-
try, where most cooperatives exited very slowly in response to a secular
decline of the entire industry in the Northwest region, and a handful were
able to survive for a half century.
What then should one make of the finding that plywood cooperatives
have inelastic output-supply curves and are not very responsive to de-
mand or cost shocks? If layoffs are costly to workers, and KMFs rely
excessively on layoffs due to bargaining inefficiencies or an inability to
make credible commitments, the fact that LMFs assign some positive
weight to employment in their objective functions and retain labor in
times of economic adversity may well yield a social benefit. However,
this argument cannot be pushed too far because LMFs do sometimes lay
off non-member employees when demand falls, and in this respect they
resemble conventional firms. Job security for members is combined in
practice with large earnings fluctuations for the members and job insecu-
rity for non-members. Whether this combination is better or worse from
a social standpoint than the employment practices of conventional firms
is difficult to judge.
8
Views from Economic Theory (I)
8.1 Explaining the Rarity of Workers’ Control

The time has come to survey some hypotheses concerning the rarity of
LMFs and their observed distribution across industries. This chapter and
the next address hypotheses offered by microeconomic theorists and writ-
ers in the new institutional economics, drawing on earlier work by Dow
and Putterman (1999, 2000). Here I focus on asset ownership and work
incentives. Chapter 9 continues this survey and critique by addressing
capital constraints, portfolio diversification, and collective choice. Using
the case studies from Chapters 3–4 and other empirical evidence, I will
identify strengths and weaknesses in each approach, along with puzzles
deserving further attention. While no hypothesis escapes unscathed,
theory and evidence together suggest that some are more promising
than others. Chapter 11 will weave the more successful ideas, and those
of Chapter 10 on organizational demography, into a broad theoretical
synthesis.
It may be that all of the hypotheses in the next few chapters con-
tain some kernel of truth. The constraints on workers’ control that prove
binding could vary by industry, society, or historical epoch. Moreover,
important interactions are likely to be overlooked when hypotheses are
artificially grouped into a few simple categories. But one must start some-
where, and even if the reader rejects the synthesis proposed in Chapter 11,
the effort made here to separate wheat from chaff should help clarify the
issues at stake.
165
166 Views from Economic Theory (I)
8.2 Asset Ownership: Incentives and Information

Theoretical analyses of the LMF sometimes fail to state clearly whether
non-human assets are leased from outsiders or owned by the firm itself.
Since this issue is central for several of the hypotheses in this chapter
and the next, I begin by considering several factors that can make leas-
ing unattractive. These factors by themselves do not explain the rarity
of LMFs, because collective asset ownership by LMFs will need to be
addressed in Chapter 9. But if all required asset services could be ob-
tained on competitive markets through leasing contracts, it would be hard
to explain the dominance of KMFs because there would be little reason
for suppliers of non-human assets to possess control rights.
Leasing is clearly problematic for a wide range of intangible assets such
as technical know-how (Jensen and Meckling, 1979). But technological
and organizational capabilities are embodied in workers themselves, in-
dividually or in their roles as part of a production team. Thus, such assets
are already owned individually or jointly by workers and pose no spe-
cial difficulties for the creation of LMFs. Managers in KMFs usually have
specialized knowledge that is relevant for firm operations, but specialized
managers are also needed in most LMFs and can be regarded as addi-
tional members of the production team. Similarly, reputations are not
easily leased from outsiders, but again an LMF could invest in building
a reputation with customers or suppliers (see, however, the remarks on
this point in Section 7.5). For these reasons, I will focus on physical assets
in what follows.
If the individuals using an asset do not own it, they may have weak
incentives to maintain it properly or avoid its misuse, especially when
they are being rewarded primarily for other activities such as producing
output (Holmstrom and Milgrom, 1991). As Alchian and Demsetz (1972)
point out, depreciation on a physical asset is likely to be greater when it
is used in the absence of the owner than when its use is observed and
the owner charges a price that accurately reflects losses in its value. They
cite several activities, including watch repair, engineering, carpentry, and
truck driving, in which workers normally own their own tools, arguably
because monitoring by an outside owner would be impractical.
This concept is easily generalized: Why introduce an irrelevant outsider
as an asset owner, which creates unnecessary monitoring and incentive
problems, when instead asset users could be the owners and internalize
the tradeoff between output and asset value? As will be seen in Chapter 9,
some authors answer this question by introducing considerations such as
8.2 Asset Ownership: Incentives and Information 167
capital constraints or portfolio diversification. For the moment, however,

I want to pursue a different line of reasoning suggested by Alchian and
Demsetz.
By a logical extension of their argument, it might appear that large
indivisible assets such as oil refineries and steel mills should be owned
jointly by the production teams using them, because worker-owners would
be in the best position to monitor how physical assets are used. This is not
the conclusion Alchian and Demsetz are trying to reach. Instead, they
want to explain why the worker teams using such assets are generally
organized as KMFs, with the physical assets being owned by a party such
as an outside investor who is external to the team itself. To account for this,
they point out that joint ownership by a large team can lead to free-rider
problems. Assets owned by everyone may in practice be owned by no one,
and thus may be abused or poorly maintained. Concentrated ownership
and active monitoring of asset use by the owner is one way to overcome
this problem.
The potential for abuse of costly and delicate equipment is often quite
real (Alston and Gillespie, 1989). However, small- to medium-sized pro-
duction teams might find ways to handle these dilemmas. The problem is
analogous to that of preserving a common pool resource jointly exploited
by a local community. As Ostrom (1990) has shown, small-scale commu-
nities can often resolve such collective-action problems through suitable
institutions.
Free-riding with respect to asset maintenance in a production team is
closely related to the repeated prisoners’ dilemma game, and the issues
arising in this context are identical to those discussed in Section 6.6. An
efficient equilibrium can arise as long as the number of players is not
too large, each player can monitor the actions of others, and each places
enough weight on her own future payoffs. But such benign outcomes
are not inevitable. Efficiency may become problematic once team size
exceeds quite modest levels, and thus joint ownership may not yield a
satisfactory level of maintenance activity within the LMF. If other organi-
zational solutions such as the designation of specialized monitors also fail,
and leasing is unworkable, a KMF with intensive monitoring may be the
least of the evils from an efficiency standpoint (see also Sections 8.5–8.7).
Apart from moral-hazard issues connected with asset maintenance,
adverse selection may also tend to forestall leasing. Suppose that an asset
constructed by someone else can be of variable quality, but quality is not
obvious until some time has elapsed. For example, the durability of a
photocopier may be unknown until after it has been used for a while and
experience has accumulated about how often a repair technician must be

called. The same principle extends to a wide range of other physical assets:
machinery, factories, and so on. If the terms of a leasing contract must be
finalized in advance, but reliability, durability, or other quality dimensions
will not become clear for months or years, a market can emerge in which
the price of a leased asset reflects the average quality of the assets supplied
and only low-quality assets are available.
This is the same kind of adverse-selection process as was discussed in
Section 6.5, except that in the standard model, goods are purchased rather
than leased. Similar problems arise in leasing markets if warranties or
other contractual remedies are infeasible (consider the rich scope for op-
portunism on both sides if a warranty were offered against defects in the
construction of a factory). Now suppose a firm has both the competence
and resources to carry out the design and construction of a physical asset
internally. If the firm is willing to pay for high quality, it may be preferable
to avoid leasing or second-hand markets where low quality is common-
place, and construct the asset in-house, perhaps under the direction of the
same managers who will ultimately be responsible for its use.
8.3 Asset Ownership: Bargaining and Investment

Productivity benefits can often be obtained from investments in special-
ized assets that have a large sunk-cost component. Such investments gen-
erate quasi-rents after the asset exists, because there is a gap between the
value of the asset in its existing use and the next best return that could be
obtained by shifting the asset to some other use. This creates a temptation
for opportunistic behavior by other parties who would like to appropri-
ate such quasi-rents for themselves (Klein, Crawford, and Alchian, 1978).
For this to occur, there must be some reason why it is impossible to write
and enforce complete contracts about the services derived from use of
the assets, because if ironclad leasing contracts were feasible, there would
be no scope for ex post opportunism once the asset was in place. More-
over, Fudenberg, Holmstrom, and Milgrom (1990) show that a series of
short-term contracts can perfectly replicate a single long-term contract
as long as there is no information asymmetry when renegotiation occurs,
so opportunistic seizure of quasi-rent is not likely to be a major concern
unless private information is important (see Section 5.1).
In his transaction-cost framework, Williamson (1975, 1979, 1985) ar-
gues that because of bounded rationality, agents cannot write contracts
in which they spell out the actions to be taken in all possible states of
the world. A dilemma therefore arises: Ironclad contracts lack flexibil-

ity, while incomplete contracts enable agents to bargain over quasi-rents
after a specialized asset exists. Assuming that bargaining occurs under in-
complete information, delays or impasses are likely (Kennan and Wilson,
1993), and agents who expect to play such games will generally take costly
actions in advance to safeguard their interests.
Williamson and other transaction-cost theorists conclude that it may
be efficient to bring related specialized assets under unified ownership,
thus avoiding any need for costly ex post bargaining over the stream of
quasi-rent. The implications for workers’ control are similar. Suppose that
control rights of the sort described in Chapter 5 enable the controlling
party to appropriate quasi-rents (Dow, 1993b). If workers have special-
ized skills that yield quasi-rents but physical assets are generic, then un-
necessary bargaining can be avoided by giving control over the firm to
workers and having them lease or own the physical assets. Conversely, if
physical assets yield quasi-rents but labor is generic, bargaining can be
avoided by giving control to asset owners who hire workers on a com-
petitive labor market. When physical assets and workers’ skills are both
specialized, the organizational solution suggested by transaction-cost rea-
soning is once again unified ownership. Because slavery is illegal, this can
only be implemented by having workers jointly own the physical assets.
A somewhat different version of the asset-specificity hypothesis is due
to Grossman and Hart (1986) and Hart and Moore (1990) – henceforth
GHM. Informal expositions can be found in Hart (1991, 1995) and
Moore (1992). Unlike Williamson or Klein, Crawford, and Alchian, GHM
assume bargaining is costless and efficient, so there is no problem of adapt-
ing ex post to unforeseen states of the world. The issue is whether there
are adequate incentives for ex ante investments in skills that are comple-
mentary to particular physical assets. Ownership of the assets serves as
a bargaining chip that determines the ex post return each party receives
on these investments, and by affecting investment levels, asset ownership
affects the total surplus derived from the firm. GHM argue that the struc-
ture of ownership rights will be arranged ex ante to maximize this surplus.
They abstract from the capital constraints and diversification issues to be
discussed in Chapter 9.
The underlying reason why ownership affects bargaining power is that
contracts are incomplete (for reasons of bounded rationality, if nothing
else), and property rights bestow residual rights of control on asset own-
ers (that is, the right to make decisions about asset use that have not
already been contracted away). The assignment of property rights thus
has allocative consequences. For example, an asset owner may use resid-
ual control rights to siphon off an unauthorized share in the returns from
her asset, even if this undermines efficiency by reducing the incentives of
other parties to invest in complementary skills.
But the pattern of asset ownership does not all by itself determine the
distribution of bargaining power among agents. The substantive content
of the contracts among them also matters. If prevailing contracts have
resolved the problem of contractual incompleteness in favor of agents
other than the nominal owner by authorizing the users to do what they
like with the asset (except perhaps for a set of explicitly forbidden uses),
the users will possess effective control, and may enjoy a great deal of
bargaining power for the remaining period of the contract. The owner
does retain the option of withdrawing access to the asset at the contract
renewal stage, but in the meantime control can be exercised by other
parties. As I mentioned in Section 5.1 in connection with the property-
rights framework, the key issue is precisely what sort of contract will be
adopted in equilibrium: one in which the asset owner retains all but a few
explicitly enumerated rights granted to the users, or one in which the users
can do as they like provided that they do not engage in certain prohibited
activities.
GHM do not give much attention to the allocation of control rights
between capital and labor, being concerned instead mainly with vertical
and lateral integration among firms. But if one abstracts from joint asset
ownership by workers and confines attention to leasing arrangements,
the GHM model can be used to compare KMFs and LMFs. Suppose it is
possible to assign control rights to input suppliers without altering asset
ownership, just by modifying the governance structure of the firm. The
KMF allows asset owners to make all production decisions not previ-
ously specified by contract, subject to explicit constraints on managerial
authority over employees, and the LMF allows workers to make all pro-
duction decisions subject to explicit constraints on the ways in which as-
sets can be used. Assume that either side can terminate the relationship
at will.
After the choice between KMF and LMF has been made, each input
supplier makes an independent decision, which is non-contractible, about
a variable related to productivity. In the case of the asset owner (K), this
could be an investment to improve machinery; in the case of workers (L),
it could be an investment in skills. If there were no more negotiation,
the party with control rights would then make any remaining managerial
decisions in a way that maximized her own payoff. But suppose an efficient
bargain is reached concerning these other decisions, where the payoffs

from disagreement at the bargaining stage are those that result from a
unilateral choice by the party with control rights, subject to the constraint
that the other party does not prefer to quit.
Under this interpretation, the analysis of GHM shows that if the ben-
efits or costs of non-contractible investments accrue primarily to capital
suppliers, while the payoff to labor suppliers is not sensitive to these
investments, the efficient organizational form is the KMF (see Proposi-
tion 1(B) in GH 1986). The converse implies that the LMF is efficient.
With some further technical assumptions, GH show that control by K
is efficient when K’s non-contractible investment in machinery is more
important than L’s investment in skills, and overinvestment by K is less
harmful than underinvestment (GH, 1986: 708). For example, we should
expect to see KMFs when investments in specialized machinery matter
more for productivity than investments in specialized skills, because giving
control to capitalists will enhance their bargaining position and increase
the return on machinery investments.
All of this assumes that capital and labor are supplied by separate
agents, and has no relevance for LMFs with collective asset ownership by
workers themselves. However, Hart and Moore (1990) suggest reasons
why collective ownership may not be observed. Suppose a specialized
physical asset is required by a team of workers, and the individual
members of the team may make non-contractible investments in human
capital. HM derive the following propositions: (1) If only one agent
needs to make an investment in human capital, that agent should own
the physical asset; (2) if the asset only affects the marginal benefit from
human capital investment for one agent, that agent should own the
asset; and (3) if other parties cannot derive any benefit from the asset
without having a given agent in their coalition, the essential agent should
own it.
In any of these cases, there is an argument for individual rather than
collective asset ownership, assuming patterns of asset ownership maxi-
mize overall surplus. Now assume instead that the returns from human
capital investments by the individual team members are only affected by
the availability of the physical asset if at least half the members of the team
are also present. For this case, HM show (their Proposition 7) that con-
trol over the asset should reflect majority voting among team members,
a form of collective asset ownership.
HM thus provide a theory that can distinguish between individual and
collective asset ownership on efficiency grounds. Moreover, their story
does not appeal to work incentives, capital constraints, or risk aversion.

The main difficulty is with the partial nature of their results. Asset owner-
ship by an individual is unambiguously predicted only under a highly re-
strictive set of circumstances. Likewise, collective ownership is predicted
only under restrictive conditions. Most real-world cases do not satisfy any
of the sufficient conditions developed by HM, and in these cases their
theory is silent on the nature of the optimal ownership arrangement.
8.4 Can Asset Specificity Explain the Rarity of Workers’ Control?

The transaction-cost ideas developed in Section 8.3 are not sufficient
to show why workers’ control is rare. The reason is that this theoretical
framework says nothing about the feasibility or desirability of collective
asset ownership. At best, asset specificity might explain why LMFs do not
generally lease specialized physical assets from outside owners. Klein,
Crawford, and Alchian (1978) are explicit about this. They observe that
in principle, workers could run firms, but say that it would be too costly
for them to rent plant and equipment “because such a specific investment
could be rather easily appropriated from its owners after it is constructed.
Therefore it is unlikely to be built.” But they acknowledge that workers
could “own the firm” as long as they also own these firm-specific assets.
The same criticism does not apply to the property-rights framework
of Hart and Moore, which as noted does yield some predictions about
collective asset ownership.
The predictions outlined in Section 8.3 are broadly consistent with
the distribution of LMFs and KMFs across industries. As discussed
in Section 3.1, traditional workers’ cooperatives generally emerge in
labor-intensive industries with little need for specialized physical assets.
Activities that require large, lumpy, specialized assets (oil refineries, auto
assembly plants) are rarely carried out by LMFs. There is also casual
evidence that generic assets need not be a stumbling block for LMFs. For
example, United Airlines, the largest publicly traded U.S. firm with ma-
jority employee share ownership (see Section 10.3), uses general purpose
assets – aircraft – for which second-hand and leasing markets exist. There
is, however, no econometric evidence establishing a link between the rel-
ative specificity of physical assets across industries and the cross-sectional
distribution of LMFs.
The evidence concerning human capital specificity is even murkier.
The analyses of Grossman and Hart (1986), Hart and Moore (1990), and
Dow (1993b) suggest that LMFs are more likely when human capital
is more specialized to the firm (GHM) or the industry (Dow), holding

physical asset specificity constant. But this hypothesis is in trouble even
in terms of casual empiricism. Hansmann (1996: 72) points out that LMF
members “appear unusually mobile,” citing cooperatives found in the
taxicab, refuse collection, and plywood industries. LMFs are common
in the professions, but doctors, lawyers, and accountants typically have
generic skills. There is evidence that conventional firms are more likely
to create formal channels for employee participation in decision-making
when the production technology is complex, perhaps because front-line
workers have private information that is valuable to managers. However,
evidence that such participation is linked to investments in firm-specific
skills is much weaker (Ben-Ner, Burns, Dow, and Putterman, 2000).
8.5 Work Incentives without Risk Aversion

The remainder of this chapter examines hypotheses about the incentives
for workers to supply effort. All assume that workers’ effort contributions
are either difficult to verify or else non-contractible for some other reason,
such as the cost of writing detailed contracts or using the court system for
enforcement purposes. In their simplest forms, these stories only provide
reasons why someone other than workers should run the firm; without
some non-contractibility problem affecting the supply of capital, they do
not indicate why it is specifically capitalists who will acquire control rights.
One can nonetheless complete at least some of these stories in such a way
that the controlling party is identified as a capital supplier. Hypotheses
that do not involve the assumption of risk aversion are examined here,
and those that do involve this assumption are considered in Section 8.6.
A prominent early effort to explain why capital suppliers normally
manage the firm is that of Alchian and Demsetz (1972). They begin from
the premise that teamwork is often productive, but makes it difficult to ob-
serve individual effort. Unless pay is closely tied to effort, team members
will shirk. The solution is to appoint a monitor who pays out wages based
on estimated effort. Shirking by the monitor is prevented by giving this
agent sole claim on the firm’s residual income. The monitor manages the
firm because the information derived from monitoring is valuable in coor-
dinating production tasks. Finally, the monitor will own any physical assets
required by the production team because these assets would otherwise
be overused or poorly maintained, a central owner has strong incentives
to curb such abuse, and the same person should logically monitor both
effort and asset utilization.
The outcome is a capitalist firm, although Alchian and Demsetz

acknowledge that hard-to-monitor professional or artistic activities may
be carried out by small partnerships. Strictly speaking, their hypothe-
sis only provides a rationale for sole proprietorships because in larger
firms, most employees are supervised by other employees, not a resid-
ual claimant. To explain the corporate form, Alchian and Demsetz also
need to show why it is desirable to have multiple shareholders, because
each individual shareholder is tempted to free-ride on the monitoring
efforts of others. This free-rider problem could be outweighed by the
need to overcome wealth constraints or spread risk, for example (see
Chapter 9).
In the terminology of Table 5.2 in Section 5.5, Alchian and Demsetz
are asserting the efficiency of a degenerate type of machine owners’
cooperative in which all physical assets are owned by a single individual
who hires labor on a competitive market. They reject the notion that an
LMF could lease assets from outsiders by stressing the monitoring prob-
lems surrounding asset maintenance. They also reject the idea that an
LMF could own physical assets collectively for reasons already discussed
in Section 8.2: With shared ownership, machines will be abused and
maintenance will be neglected.
There are numerous theoretical objections to this story. First, vertical
monitoring by a specialist owner may be less reliable or more costly than
horizontal monitoring among a group of co-workers (Putterman, 1984).
Second, as discussed later it may be possible to motivate team members to
supply effort efficiently without monitoring individuals at all, through the
use of group bonuses or penalties (Holmstrom, 1982). Third, the Alchian
and Demsetz story implicitly relies on an assumption of limited liability
because it is possible to approximate an efficient equilibrium in a produc-
tion team without specialized monitoring as long as unlimited penalties
can be imposed on individuals (Legros and Matthews, 1993).
A number of authors have pointed out that shirking on either effort
supply or asset maintenance can be deterred in a repeated game frame-
work if workers put sufficient weight on future payoffs (see MacLeod,
1984, 1988; Weitzman and Kruse, 1990; Putterman and Skillman, 1992;
Dong and Dow, 1993). This does not require that team members be able
to observe one another’s effort levels (it suffices that team output can
be observed by all workers) and does not require that output be verifi-
able in court (unlike the principal-agent models in Section 8.6, where
wages are contractually linked with verifiable output measures). The
analysis also goes through when output depends on individual efforts
in a stochastic way, although this does make it more difficult to achieve

an efficient equilibrium. For Folk Theorem reasons (see Section 6.6),
there can be many equilibria in such a model, including highly inef-
ficient equilibria. Consequently, while it may be possible in theory to
obtain efficient behavior, this is by no means assured, especially in large
teams.
Another theoretical difficulty with the Alchian and Demsetz approach
is that it one-sidely emphasizes shirking by production workers, without
appreciating the potential for opportunistic behavior on the part of the
monitor. A residual claim may well motivate the monitor not to shirk in
measuring effort levels accurately, but a dishonest monitor could lie about
what these effort levels actually were. There are strong incentives to do
so, because as the sole residual claimant, the monitor gains a dollar in
income for every dollar not paid out in wages. Employees have no legal
recourse in this scenario since by assumption effort is not verifiable in
court. Alchian and Demsetz may have in mind a reputational mechanism
that deters cheating by employers in the absence of legal safeguards. But
such mechanisms need to be studied using a repeated game framework
(see, for example, Dow, 2000).
A number of authors subsequent to Alchian and Demsetz have taken
up the problem of effort supply under conditions of team production.
Undoubtedly the most influential is Holmstrom (1982), who considers a
team in which each worker independently chooses an effort level, there
is a deterministic relationship between effort inputs and team output,
and team output is verifiable in court although individual effort is not.
Holmstrom shows under these assumptions that as long as the entire
output of the team must always be distributed among its members (the
budget-balancing constraint), it is impossible to devise a contract that
elicits efficient effort levels. A free-rider problem is inevitable because in-
creasing incentives for one team member necessarily decreases incentives
for someone else.
Holmstrom also shows that if the budget-balancing constraint is re-
moved, group bonuses or penalties can be used to obtain efficient effort
levels. Imagine that a “principal” (someone outside the team) offers the
following contract to a group of “agents” (workers). If team output equals
or exceeds the level that would occur when everyone contributes her effi-
cient effort level, each agent gets an equal per capita share in the efficient
output level, and the principal gets whatever is left over, if anything. But if
output ever falls below this target, each agent gets nothing, and the prin-
cipal keeps the entire output. Under reasonable conditions, this induces
each agent to supply an efficient level of effort, and in equilibrium the

principal gets nothing. The Alchian and Demsetz dilemma is resolved by
punishing all team members simultaneously when any individual shirks.
The function of the principal is not to monitor individual effort, but to
eliminate the budget-balancing constraint.
One complication is that there is often more than one equilibrium
under a contract of the Holmstrom type. While efficient effort choices
are one equilibrium, it may also be an equilibrium for everyone to supply
zero effort. Thus efficiency cannot be ensured merely by adopting the
right contract. Some coordination or mutual trust among team members
is also needed.
It is debatable whether the principal should be interpreted as a capi-
talist, or whether the model should be viewed as a theory of the capitalist
firm. Holmstrom has encouraged this view by remarking: “The fact that
capitalistic firms feature separation of ownership and labor implies that
the free-rider problem is less pronounced in such firms than in closed
organizations like partnerships” (1982: 325). But nothing in the formal
model implies that the principal must be a capital supplier. All the prin-
cipal does is sop up any output when shirking occurs, so anyone external
to the team could play this role. Indeed, it is vital that the principal not
be a supplier of any non-contractible input because otherwise it would
be necessary to consider the principal’s incentive to shirk and the team
would merely have one additional member. Thus, if the principal is re-
garded as a capitalist, capital must be a fully contractible input. But then
it is unclear why workers could not rent capital while having someone
else serve as the budget breaker for incentive purposes.
As in the Alchian and Demsetz framework, one must consider the
possibility of an opportunistic principal. One possible mode of cheating
is for the principal to collude with a team member to induce a minor
amount of shirking, and then divide the team’s output with this turn-
coat (Eswaran and Kotwal, 1984; Andolfatto and Nosal, 1997). A further
problem arises if output itself is non-verifiable, perhaps because quality
cannot be readily measured. This makes it impossible for the principal to
pre-commit to a binding contract linking wages with output. When there
is a danger that the principal might cheat on promised wages, ownership
in a repeated game context is neutral: Asset ownership by a principal
is no better or worse than collective asset ownership by the team itself
(Dow, 2000).
Subsequent work has elaborated on the many issues raised by
Holmstrom (1982). MacLeod (1987) shows that if a single production
team can be divided into subteams with independently measurable out-

puts, penalties for shirking within one subteam can be paid to members
of the other. This enables an LMF to maintain efficient levels of individ-
ual effort by administering group penalties internally, without using an
external principal to relax the balanced budget constraint. McAfee and
McMillan (1991) show that monitoring individual efforts may in fact be
useful if there are simultaneous problems of moral hazard (shirking) and
adverse selection (workers who have unknown abilities). In their model,
the function of monitoring is to decrease the size of the payments the
principal must make to individual workers, and thereby to reduce the
principal’s temptation to cheat.
Work by Legros and Matthews (1993) casts doubt on the idea that a
budget-breaker is necessary for economic efficiency. They derive neces-
sary and sufficient conditions for a partnership organization to achieve
efficient effort levels while dividing all its output among the team mem-
bers, so that the budget-balancing requirement remains in force. Their
results show that partnerships can approximate the efficient outcome as
closely as desired if there is no limit on the personal liability of workers. In
this case, one can always devise an array of equilibrium strategies where
no worker can gain by deviating to some lower effort level, because doing
so would reduce team output and trigger a severe punishment for the
shirker in the form of mandatory income transfers to other members of
the team.
Of course, in reality there are restrictions on the penalties that can be
imposed in a production team: Workers have limited wealth, and shirkers
cannot usually be shot at dawn. Nevertheless, the point is an important
one. If workers cannot be deprived of much wealth because they don’t
have much to begin with, or if for institutional reasons it is impossible to
impose penalties of the required size, second-best efforts may fall well
short of the first-best levels. It may therefore be attractive to use costly
monitoring or to bring in an outsider who can impose group penalties as
in the Holmstrom (1982) scheme.
More generally, such theoretical models help to emphasize that work
incentives and thus firm organization may depend on the distribution of
wealth. These implications are developed further by Legros and Newman
(1996), who show that an unequal wealth distribution in combination with
imperfect financial markets can lead to an equilibrium in which firms use
costly monitoring to maintain work incentives, although more aggregate
surplus could be generated under conditions of wealth equality. I return
to issues involving wealth and capital markets in Chapter 9.
8.6 Work Incentives with Risk Aversion

Knight (1921) maintained that firms would be controlled by the venture-
some, who would offer insurance to the timid. The risk-bearing func-
tion of entrepreneurs implies that they must control the firm because
hired employees will lack the proper incentives to make managerial deci-
sions. Related ideas have been developed by later writers (Meade, 1972;
Drèze, 1976, 1989; Kihlstrom and Laffont, 1979). The implications of risk
aversion for the financing of firms will be explored in Sections 9.5–9.6, but
here I continue to focus on work incentives. The existence of a production
team is no longer central to the analysis, although the general argument
applies to that case as well.
Suppose the revenue generated for a firm by a worker is a random
variable whose distribution depends on the worker’s effort. The stochastic
nature of the worker’s marginal revenue product may result from techno-
logical uncertainties about mechanical breakdowns and power failures,
or variability in market prices. Under these conditions, compensating the
worker out of her own revenue stream will expose her to risk. A risk-
averse worker would prefer to trade this uncertain earnings stream for
a certain one with the same mean. Knight’s proposition is that such a
worker can be given a fixed payment only if she cedes control over pro-
duction to the insurer, since an income guarantee eliminates any incentive
to enhance the value of the firm. The insurer will also need to monitor
employees and pay wages based on effort.
As in the Alchian and Demsetz framework, there is a question about
the incentive of the insurer to cheat employees by failing to pay promised
wages. If effort is not verifiable, then an entrepreneur-insurer would be
free to make false claims of shirking to justify wage reductions. If the
entrepreneur is contractually committed to pay a fixed wage but ex-
tracts effort by threatening to dismiss shirkers (Shapiro and Stiglitz, 1984;
Bowles, 1985), the moral-hazard problem on the employer side takes a
different form: Now there is a danger that non-shirkers will be dismissed
without justification (Dow, 2000).
Modern principal-agent theory adopts a starting point similar to Knight
by assuming that agents (workers) are risk averse, while principals
(entrepreneurs) are risk neutral. It diverges from Knight in stressing
that workers inevitably control their own effort, which is hard for the
principal to observe. This leads to a tradeoff between incentives and in-
surance (Holmstrom and Milgrom, 1991, 1994; Sappington, 1991). An
optimal contract balances these considerations by having workers bear
8.7 Work Incentives and Workers’ Control 179
some risk to provide incentives, but also get some insurance to limit the
costs of risk-bearing. By monitoring labor inputs, it is usually possible
for the principal to reduce the part of employee income that depends on
revenue, and thus the compensation paid to workers for bearing risk. But
monitoring is costly and imperfect, so workers are only partially insured,
and retain some residual claim. There is no concern in this approach that
the principal might cheat by holding back wages because it is assumed
that wages can be linked with revenue in a legally enforceable way.
Nothing in the original Knightian theory of entrepreneurship, or for
that matter in most modern principal-agent models, implies that the
principal who provides insurance to workers must also supply capital
to the firm. These theoretical perspectives leave open the possibility that
physical assets are fully contractible and readily leased. If so, the principal
could be anyone who is less risk averse than labor suppliers, and the anal-
ysis of incentive problems does not lead to a theory of the capitalist firm.
Knight’s own discussion suggests that the roles of insurer and financier
can be connected by extending moral-hazard ideas to the sphere of equity
supply, an issue to be taken up in Section 9.3.
Holmstrom and Milgrom (1991, 1994) bridge the gap between capital
supply and insurance by reintroducing problems of asset maintenance.
In their framework, a principal provides workers with partial insurance,
and workers may or may not own physical assets depending on avail-
able monitoring techniques and worker attitudes toward risk. Workers
tend to own assets when they are not too risk averse and there is reli-
able information about their activities. An interesting wrinkle is that if
workers do own the assets they work with, they can be given stronger
incentives to generate output than if the assets are owned by an em-
ployer, because in the former case the workers will internalize tradeoffs
between output and asset value. This contradicts the Alchian and Demsetz
prediction that effort incentives will be stronger in firms with an outside
asset owner. Indeed, in the Holmstrom and Milgrom framework, the only
thing standing in the way of LMFs with collective capital ownership is the
need for a risk-neutral principal who can supply insurance to risk-averse
workers.
8.7 Can Work Incentives Explain the Rarity of Workers’ Control?

The array of models concerned with work incentives does not provide
any decisive theoretical argument for or against workers’ control. With
the partial exception of Alchian and Demsetz (1972) and Holmstrom and
Milgrom (1991, 1994), most research in this area fails to address the supply
of capital and labor inputs within a unified analytical framework. Thus the
literature on work incentives generally fails to show why it is specifically
capital suppliers, rather than non-capital-supplying principals or insurers,
who have control over the firm. Incentives to supply labor may be part of
the explanation, but non-contractibility problems affecting the supply of
capital must be considered simultaneously.
Given the inconclusive nature of the theoretical literature, one might
hope for a clear empirical result regarding the effects of workers’ control
on productivity. I survey some of the relevant evidence later, but first it is
necessary to define with greater precision what is meant by productivity.
Some authors define (labor) productivity as a firm’s output divided by
the size of its workforce or total hours of work. Crude ratios of this sort
are inadequate because they depend on the levels of other inputs. If two
firms have the same output and are otherwise identical, but one uses
more capital while the other uses more labor, the capital-intensive firm
will have a higher output per unit of labor. This does not imply that it
is any more productive; it has merely substituted one input (capital) for
another (labor). Indeed, the firm may be incurring unnecessary costs if
capital is expensive and labor is cheap.
For these reasons, economists normally compare firms using a measure
called “total factor productivity,” which divides output by a weighted sum
of all inputs, taking account of each input’s share in total cost. Unfortu-
nately, matters become more complicated when one is concerned with the
effect of firm governance on productivity, because it could be argued that
workers’ control tends to enhance the effectiveness of labor inputs even
when all other inputs are held constant. For example, workers’ control
might stimulate increased effort at a given wage. This would show up sta-
tistically as a rise in output per worker or per hour even though capital and
other inputs are unchanged. On the other hand, workers’ control might
lead to more information-sharing and thus technological innovations that
enhance total factor productivity without any bias toward labor in par-
ticular. The studies to be cited here use a variety of strategies to control
for non-labor inputs, and differ in the quality of their data collection and
econometric techniques. I can only skim over the surface of the relevant
literature, so interested readers are urged to consult the original publica-
tions and evaluate the reliability of reported results for themselves.
The Alchian and Demsetz hypothesis, which argues that claims on
firm net income must be concentrated in the hands of a single individ-
ual to ensure high labor productivity, conflicts with recent evidence on
profit-sharing. Using U.S. data, Kruse (1993) estimates that adoption of

profit-sharing raises labor productivity by about 4–5 percent. Similar re-
sults are obtained for the U.K. (Wadhwani and Wall, 1990) and for Japan
(Jones and Kato, 1995). One can question these results by raising issues
of causality or selection bias, but they are not surprising. Any firm that
offers employees a claim on its profit stream must pay for that claim in
one of three ways: (1) cutting base wages so that overall labor compen-
sation does not rise; (2) reducing the dividend stream to investors; or
(3) increasing productivity (either labor or total factor) by enough to
avoid 1 and 2. In a competitive labor market, it will not be possible to re-
duce total compensation outright, and even if total compensation is kept
the same, risk-averse employees are made worse off when their earn-
ings have a larger variable component but the same mean. Hence 1 will
not generally be feasible. Investors are unlikely to favor 2 so the usual
case must be 3. The fact that so many firms find it attractive to imple-
ment profit-sharing thus contradicts the idea that residual claims must be
assigned to a central monitor for productivity reasons.
Even so, it would be unwise to dismiss concerns about work incentives
altogether. LMFs do seem to avoid highly capital-intensive activities, and
one potential explanation is the Alchian-Demsetz notion that large, com-
plex physical assets are vulnerable to misuse or poor maintenance when
collectively owned by the workforce. An observation consistent with this
idea is the fact that workers are paid more, other things being equal, in
capital-intensive industries. This could reflect efficiency wage payments
designed to deter potential misuse of valuable equipment (Holmstrom
and Milgrom, 1991). But there are other explanations for this wage pat-
tern, including a reduction in labor turnover to keep specialized equip-
ment fully utilized, and the capture of quasi-rents by employees through
bargaining. Ideally, one would like evidence that KMFs are more common
in industries where abuse of equipment by workers is highly costly and
easily forestalled by vertical supervision, and that workers are monitored
intensively in such industries, but such data do not currently exist.
There is some disagreement over whether the cross-industry distribu-
tion of LMFs is consistent with expectations based on monitoring costs.
Hansmann (1996: 70–71) says that “the existing pattern of employee
ownership is just the reverse of what one would expect if it were pri-
marily a response to the difficulty of monitoring employees,” arguing that
LMFs are a common mode of organization for law firms in which the
productivity of individual workers can be monitored relatively easily, but
are quite rare in manufacturing activities involving large teams where
vertical monitoring is difficult. But Russell (1985b) argues that profes-

sional partnerships, scavenger cooperatives, and taxi cooperatives are all
in part a response to difficulties in monitoring sales and service efforts,
under conditions in which scale economies render sole proprietorships
infeasible.
In an impressionistic but comprehensive review, Ben-Ner, Han, and
Jones (1996) found evidence for complementarities between residual
claims and participation in decision-making – that is, the productivity
benefits from one variable appeared to be magnified by an increase in
the other. Kruse (1999, unpublished manuscript) found that while the
majority of studies of employee share ownership were unable to reject
the null hypothesis that there is no significant relationship with firm per-
formance, a meta-analysis of ESOP studies did reject this null hypoth-
esis. The average estimated productivity difference between ESOP and
non-ESOP firms was 6.2 percent, and the average productivity increase
after adoption was 4.4 percent (see also Blasi, Conte, and Kruse, 1996,
and Kruse and Blasi, 1997). Using firm-level panel data, Kumbhakar and
Dunbar (1993) estimated that the growth rate of output increased by
about 1.8–2.7 percent for each year elapsing after creation of an ESOP.
Studies of profit-sharing alone cannot establish much about the rel-
ative merits of KMFs and LMFs. Capitalist firms routinely use profit-
sharing as a motivational device, and thus by the replication principle
from Section 6.2 it is hard to claim that LMFs have a strong advantage
in this regard (see the further discussion of this point in Section 11.6). A
more directly relevant literature uses samples of workers’ cooperatives
that vary in their degree of participation in decision-making (measured
by the share of workers who are voting members), the level of profit-
sharing (measured by the relative size of bonuses paid), and financial
participation (measured by individually or collectively owned assets per
worker). Of greatest interest for a theory of control rights is participation
in voting. Econometric studies of French and Italian cooperatives have
repeatedly found that productivity depends positively and significantly on
the fraction of the workforce who are voting members, even after reverse
causality is taken into account (Defourney, Estrin, and Jones, 1985; Jones
and Svejnar, 1985; Estrin, Jones, and Svejnar, 1987; Estrin and Jones, 1992,
1995). There is also some evidence that Italian cooperatives that rely more
heavily on the knowledge of production workers have better profitability
and higher income per member (Smith, 1994).
On reviewing eleven econometric studies concerned with the ef-
fects of worker decision-making participation on LMF productivity,
Doucouliagos (1995) found that this relationship was positive in seven

cases, negative in two, and zero in two. Excluding four studies that pooled
data for KMFs and LMFs or examined firms whose LMF status was
doubtful, meta-analysis showed that the population correlation was pos-
itive at conventional significance levels. For participatory capitalist firms,
meta-analysis conducted using twelve studies indicated that the same
correlation was positive but not quite significant after three studies of
codetermined firms that found significant negative effects were excluded.
The results do not directly compare productivity in LMFs and KMFs.
The cleanest test would be a direct comparison of productivity in KMFs
and LMFs within the same industry. A few researchers have studied
matched sets of firms for Italy, Sweden, and France, with what Bonin,
Jones, and Putterman (1993: 1305) have termed “inconclusive” results.
The best example of data collection and econometric work in this area
is probably Craig and Pencavel (1995), who estimated a 6–15 percent
productivity advantage for the plywood cooperatives relative to conven-
tional mills. However, there are potential selection biases that make
it hard to interpret the results. For example, workers can and do move
between coops and conventional firms, and the coops may attract mem-
bers who have intrinsically higher productivity. If so, Craig and Pencavel
overestimate the advantages of the cooperative governance structure.
But conversely, Bonin, Jones, and Putterman (1993: 1305) point out that
successful plywood coops have disproportionately left the sample over
time by selling out to conventional rivals, so Craig and Pencavel could be
underestimating the productivity gains from cooperative governance.
According to case-study evidence, shirking is not an important prob-
lem in LMFs. Almost all reports emphasize the prevalence of mutual
monitoring and relatively low levels of formal supervision (Weitzman
and Kruse, 1990; Bonin, Jones, and Putterman, 1993). The plywood coop-
eratives, for example, not only have a high degree of mutual monitoring
but also use fewer supervisors than conventional firms (Greenberg, 1986:
43–44), and by some accounts have fewer non-production employees in
general (Berman and Berman, 1989: 289, note 9). Coops in the Lega fed-
eration also seem to use fewer managerial and supervisory employees
than comparable capitalist firms (Bartlett et al., 1992).
A related question is whether workers in LMFs will free-ride in their
responsibilities to monitor managers of the firm. Milgrom and Roberts
(1992: 563), for example, express some concern on this point. But as Craig
and Pencavel (1993: 291, note 4) point out, hired managers in plywood
coops tend to complain that they are monitored too closely by coop
members. While one might not be inclined to give these complaints much
weight, Chapter 3 indicated that managers in the plywood coops and at
Mondragon were often dismissed by worker-members. If there is a ba-
sis for any concern, it might be that LMF members cannot resist the
temptation to interfere in managerial decision-making.
Despite all the measurement, causality, and selection issues that bedevil
econometric work in this area, it is fair to conclude that there is little
evidence for inferior productivity in LMFs, and it appears reasonable to
believe that LMFs would provide modest productivity benefits in some
industries. This is consistent with independent evidence that once LMFs
exist, they generally have lower failure rates than comparable KMFs
(see Section 10.6).
One could try to salvage the hypothesis that LMFs are rare because
of weak productivity performance by claiming that LMFs will arise only
when they have no disadvantage on this dimension, and that the lack of
LMFs elsewhere in the economy reflects the defective work incentives
from which LMFs would suffer if they entered other sectors. But asser-
tions of this kind are untestable. In fact, the evidence cited here justi-
fies reversing the question. Given the productivity advantages that LMFs
would likely enjoy in some industries where they do not currently exist,
especially industries with limited economies of scale and labor-intensive
production techniques, what are the non-productivity barriers restraining
the further spread of such firms? The next chapter takes up the search for
these barriers.
9
Views from Economic Theory (II)
9.1 Capital Constraints

Workers who want to create a labor-managed firm will usually find that
inputs must be acquired today, while output and hence revenue will not
arrive until tomorrow. There are several potential solutions. First, work-
ers might attempt to match revenue flows against expenditure flows by
leasing assets. But leasing is not always practical, as was discussed in
Sections 8.2–8.4, and it is hard to avoid the need for working capital.
Second, workers might use their personal savings and those of family
members or friends to finance jointly owned assets or other upfront costs.
Poor workers, however, may not have the resources needed for this. Third,
they might finance a firm by borrowing from banks, suppliers, or cus-
tomers, or by selling equity claims to outside investors.
Many writers have argued that limited worker wealth combined with
credit rationing is the key barrier to LMF creation. For example, Drèze
asserts that “funding difficulties are the main reason why labor-managed
firms are not spreading within capitalist economies,” and adds, “labor-
managed firms will spread more easily in sectors where capital invest-
ment per worker is not high” (1993: 261–62). A number of authors blend
an emphasis on capital constraints with arguments about the cost to work-
ers of making undiversified investments in their own firms (Bowles and
Gintis, 1990, 1993b, d, 1994, 1996b; Putterman, 1993). I consider portfolio
diversification separately in Sections 9.5–9.6, but for the moment it will
be simpler to assume that all parties are risk neutral.
For convenience, I split the discussion into two sections that deal with
debt and equity. Comprehensive asset-leasing is ruled out for one or more
185
186 Views from Economic Theory (II)
of the reasons given in Sections 8.2–8.3. Unless otherwise, noted, “capital”

in this chapter refers to liquid funds rather than physical assets.
9.2 Debt Financing

For concreteness, I consider loans from banks, but similar issues arise
for workers seeking credit from suppliers or customers. Microeco-
nomic theory identifies two principal reasons why workers might face
problems in the credit market: moral hazard and adverse selection. Both
were mentioned in Chapter 6, but here I develop their specific implica-
tions for the problem of financing collective asset ownership by an LMF.
Moral hazard arises when worker-borrowers have incentives that run
contrary to the interests of lenders after a loan contract is in place. Schlict
and von Weizsäcker (1977), in one of the earliest applications of moral-
hazard ideas to LMF financing, argued that such firms would have partic-
ular problems in obtaining credit because workers could leave a failing
firm. Gui (1985) showed formally that if LMFs can liquidate projects to
avoid debt repayment, there is sometimes an upper limit on the level of
debt financing the market will supply. Moore (1992) and Hart and Moore
(1994) obtain similar results by assuming that human capital is inalien-
able and that entrepreneurs who finance investment projects through
debt remain free to walk away ex post. These incentive conflicts between
borrower and lender intensify as the ratio of debt to equity grows. The
consequences can include interest rates that rise with leverage, or outright
credit rationing (Stiglitz and Weiss, 1981).
Another form of moral hazard arises when, after gaining access to
liquid funds, workers pursue risky projects because high payoffs go mainly
to the borrower (subject to a loan repayment obligation), while low pay-
offs are inflicted mainly on the lender (because the borrower can declare
bankruptcy). Parallel issues arise if workers can shirk after their projects
have been financed. Eswaran and Kotwal (1989) argue that investors run
firms because entrepreneur-borrowers are tempted to substitute inputs fi-
nanced through loans for effort contributions. To get around this problem,
lenders take over the management of the production process. One way
to interpret this proposition is that it generalizes the notion of equipment
abuse developed in Section 8.2: Just as workers may substitute machine
services for their own effort, they may also substitute borrowed funds for
their own effort.
Short of taking direct control over production activities, lenders may
prefer to deal with undemocratic firms having a single peak owner because
9.2 Debt Financing 187
it is easier to influence the policies of such firms (Gintis, 1989). Lenders

may want such influence in order to prevent the pursuit of excessively
risky projects, or excessively conservative ones. Levin takes a similar
view, remarking: “[T]here is no guarantee of a single owner or officer
who always represents the workers, given the democratic management
[of workers’ cooperatives]. As a result, the financial community cannot
obtain the leverage over cooperatives seeking to borrow that they can
over capitalist firms” (1984: 246).
Adverse selection, in contrast, involves an informational asymmetry
that exists even before the loan contract is signed: Some borrowers
have better skills or projects than others, but lenders cannot distinguish
between these borrower types ahead of time. Thus a safe borrower can find
it difficult to convince a lender that the probability of loan repayment is
high, and again may face credit rationing (Stiglitz and Weiss, 1981; Stiglitz,
1987).
To be sure, high-quality borrowers will look for ways of signal-
ing their true type. One solution is for borrowers to invest some of
their own funds in a project to convey their confidence in its merits
to lenders or other outsiders. Unfortunately, this solution is not avail-
able when workers have desirable projects but very limited personal
wealth, and in any case signaling is costly (without information asym-
metry, it might be efficient for the lender to finance the entire project). A
different remedy is for lenders to screen projects carefully before grant-
ing loans, but this is also costly. For reasons of this sort, equilibria in-
volving adverse selection and signaling can prove to be inefficient (see
Section 6.5).
Another solution involves the use of collateral that is forfeited to the
bank in case of default. Leaving aside the possibility that workers might
have sufficient personal wealth to serve as collateral, physical assets
purchased with loan proceeds could play this role, just as a house serves
as collateral for a mortgage. But not all projects generate equally good
collateral. Work in progress for which there is a ready market may be
used to collateralize loans. The same is true for fixed assets with a strong
second-hand or leasing market. But as Williamson (1988) observes,
very specialized assets with a large sunk-cost component, or intangible
assets such as customer goodwill, are unattractive as collateral and must
be financed in some other way. Workers cannot generally offer their own
future labor income as collateral because human capital is inalienable
and cannot be forfeited to the bank in case of default (Hart and Moore,
1994).
9.3 Equity Financing

An alternative to debt for workers with limited personal wealth is selling
non-voting equity shares in order to attract capital and reallocate risk
without sacrificing control rights. Such proposals have been advanced
by some supporters of workers’ control (Vanek, 1970; McCain, 1977),
while encountering opposition from others (Jossa and Cuomo, 1997). An
intriguing feature of this discussion identified by Putterman (1988a) is
that many prominent arguments about the role of shareholders in large
corporations make little or no reference to their voting rights, suggesting
that these rights could easily be shifted from shareholders to employees
without any serious disruption in the firm’s access to equity finance.
But as Putterman (1982, 1984, 1988a, 1993) also stresses, moral hazard
is likely to intrude. Investors who lack control may worry that workers will
use their decision-making powers in an opportunistic fashion. Non-voting
equity capital is likely to be made available only on unfavorable terms,
if at all, unless workers can credibly show that their interests converge
with those of outside investors. Anything less than full convergence of
interests between investors and workers raises the cost of capital, placing
LMFs at a disadvantage. Bonin, Jones, and Putterman (1993: 1316) give
this mechanism a central place in their explanation for the rarity of LMFs.
Williamson (1988) offers a variation on this story by observing that
physical assets with many other uses can serve as collateral on loans,
while assets with a large sunk-cost element are not easily redeployable
and cannot be used to collateralize a loan. Such assets must therefore be
financed by equity capital – that is, from the personal wealth of one or
more investors. Because workers are likely to have problems making
equity investments, Williamson says that “one possible adaptation is for
worker-managed firms to specialize in industries and technologies where
the exposure of specific assets is limited” (1985: 267, note 21). On the
other hand, if equity capital is to be obtained from outside investors, then
the equity suppliers will require representation on the board of directors
because otherwise their investments would be vulnerable to expropriation
(1985: ch. 12). For example, if the directors of the firm were workers, they
might embezzle equity funds, refrain from paying dividends in order to
raise wages, or dissipate resources on projects of dubious value. In short,
control rights are assigned to capital suppliers in order to protect these
agents from opportunistic abuse by other parties, including workers.
This argument cuts two ways, however. One could equally well argue
that workers are vulnerable to opportunistic abuse by investors, who
9.4 Capital Constraints and Workers’ Control 189
might deprive them of the returns on firm-specific investments in skills or

residential location, perhaps by closing production facilities prematurely.
At a more mundane level, employers might assign tasks that are more
dangerous or less enjoyable than promised, relying on search costs to
guarantee that workers will not quit. Perhaps KMFs must therefore pay
premium wages to workers who lack control rights in order to compen-
sate them for accepting such hazards, placing KMFs at a disadvantage
in the labor market compared with LMFs. Williamson acknowledges this
possibility, noting that workers might be willing to work for lower wages
in an LMF in order to avoid oppressive treatment (1980, 1985: 267).
The balance turns on the relative size of the temptation facing each
party if it gains control, the severity of abuse when it occurs, the extent
to which control rights are effective safeguards against abuse, and the
extent to which there are other reliable safeguards such as reputation.
If a worker-controlled firm need never return to the equity market
after making an initial investment in plant and equipment, shareholders
without voting rights should be worried that dividends may be few and
far between. Conversely, if an investor-controlled firm has no plans for
future hiring and is contemplating a plant closure, it is the employees who
should be worried about a lack of voting rights. Williamson’s view is that
dangers of the former kind are more common and severe than those of
the latter kind (1985: 302–6).
One potential asymmetry between KMFs and LMFs arises from the
fact that it can be costly to create a reputation for cooperative behavior,
but a reputation of this sort could be a vital prerequisite for attracting
inputs without granting control rights to their suppliers. If workers lack
the initial funds to make such “trust” investments, they may be caught in
a vicious circle from which there is no escape, because there is no way
of reassuring buyers of non-voting equity shares that they will not be
victimized in the future. If capitalists need to invest in a reputation for
cooperation in order to reassure non-voting workers, however, they are
likely to have access to the funds needed for this purpose.
9.4 Can Capital Constraints Explain the Rarity of Workers’ Control?

In many cases, employees cannot finance a firm’s capital stock exclusively
through their own resources. In a back-of-the-envelope calculation
involving U.S. data for 1988, Bowles and Gintis (1996b: 95) estimate that
the average net worth of a worker (computed using the least wealthy
80 percent of families) was roughly equal to half the value of the capital
stock per employee in the U.S. economy. Furthermore, about half of the
net worth of the families in question was tied up in cars and homes.
Although the figures are approximate, the orders of magnitude reveal that
many employees would have to undertake major capital market transac-
tions if they wanted to buy out the firms in which they work.
The theoretical arguments surveyed here concerning failures in the
capital market are consistent with the fact that traditional workers’
cooperatives have been financed almost entirely by the personal savings
of their members and retained earnings (Bonin, Jones, and Putterman,
1993: 1296). Short-term credit is sometimes available from banks, sup-
pliers, or customers, but generally requires tangible and unspecialized
inventories, structures, or equipment as collateral. Inventories of logs
and veneer have been used in this way by the plywood cooperatives,
which have apparently had little trouble borrowing from banks and sup-
pliers. Nevertheless, Pencavel (2001) suggests that bankers have found
it awkward to deal with the democratic governance system used by the
coops, and that the coops have likewise been reluctant to accept contrac-
tual restrictions imposed by bankers. Non-voting equity is almost never
used by LMFs. Where this strategy has been tried, as it occasionally was
in the early days of the plywood coops, firms apparently relied on close
social connections to the local community as a substitute for contractual
safeguards to investors.
There is also evidence that when LMFs and KMFs compete in the
same industry, the former choose lower capital-labor ratios than the latter.
George (1993: 78–81) finds that such differences arise in Denmark for
the baking industry (significant at the 5 percent level) and construction
(at the 10 percent level). Batstone (1982: 107) shows that mean capital-
labor ratios in a subset of French printing and construction cooperatives
are well below corresponding industry averages, although he does not
conduct any statistical tests. Berman and Berman (1989) find evidence
that the plywood cooperatives use less capital-intensive methods than
their capitalist rivals. Although the empirical methods can be questioned
in each case, such observations are consistent with the notion that LMFs
are more often rationed in the capital market or face systematically higher
financing costs than KMFs.
The exceptions to these generalizations frequently prove the rule. The
Mondragon group has entered some industries having capital intensities
above the norm for workers’ cooperatives, but all observers stress the
unique role played by Mondragon’s cooperative bank in financing such
activities (Section 3.3). Furthermore, the leadership at Mondragon made
9.4 Capital Constraints and Workers’ Control 191
a conscious decision early in the group’s development to avoid certain

highly capital-intensive industries such as mining and chemicals (Thomas
and Logan, 1982: ch. 4), and still adheres to these self-imposed restrictions.
Where the wealth hypothesis becomes more questionable is in rela-
tion to LMFs that have arisen abruptly through an employee buyout
(see Section 10.3) or through a gradual accumulation of shares in em-
ployee pension plans that eventually led to majority worker ownership.
The growing number of large U.S. firms in which employees have sig-
nificant share ownership provides the best evidence against the wealth
hypothesis, because clearly workers are sometimes willing and able to
supply large amounts of capital. The discussion of ESOPs in Section 4.2
raised the question of who actually bears the main burden of this wealth
accumulation: other shareholders, taxpayers, or employees. I argued that
ESOPs have probably not been financed through dilution of the claims
of existing shareholders and that subsidies can only account for a minor-
ity of the sums involved. Barring unreasonably high productivity benefits
from ESOPs, it seems likely that employees themselves have borne most
of the burden. But the ability of workers to forego consumption varies
widely. ESOPs are more likely to be created, and cover a larger frac-
tion of the workforce when they do exist, in firms where employees have
higher average wages and longer average tenure (Ben-Ner, Burns, Dow,
and Putterman, 2000). This suggests that limited personal wealth often
does constrain employees as potential suppliers of capital, although the
constraint is not equally tight for all segments of the workforce.
There is also evidence bearing on the claim that LMFs have limited
access to capital because of the hazards confronting non-voting equity
suppliers. While it is no longer unknown for employees to hold a majority
of voting shares in publicly traded firms through ESOPs, such employees
are both labor suppliers and capital suppliers, and voting rights normally
remain tied to capital supply. Moreover, employees rarely elect labor rep-
resentatives to the board of directors in U.S. firms, even when as a group
they constitute a voting majority (Blair, Kruse, and Blasi, 2000). These
factors apparently ensure sufficient convergence of interests between
workers and outside investors, despite the likelihood that workers’ goals
will depart from maximization of the firm’s stock-market value. As will
become clear from the United Airlines case in Section 10.3, when em-
ployees do obtain board representation it has sometimes been regarded
as necessary to craft complex safeguards to protect minority sharehold-
ers. Thus, while workers may be better able to supply capital than has
sometimes been assumed, true workers’ control within a firm may indeed
create hazards for outside shareholders who lack votes or find themselves
in a minority voting position.
The premium for voting shares compared with non-voting shares pay-
ing similar dividends is often found to be small, especially in the United
States. Zingales (1995), for example, obtains a median voting premium of
3 percent and a mean of 10.5 percent using U.S. data. However, he cites
evidence for a number of other countries (Sweden, the United Kingdom,
Canada, Switzerland, Israel, and Italy) indicating that the premium may
be in the 20–80 percent range. Zingales conjectures that such differences
may arise because minority shareholders in the United States enjoy more
protection through regulation and lawsuits than those in other countries.
If one takes the low U.S. estimates literally, they seem to suggest that
voting rights are of little value to shareholders, at least when no takeover
battle is imminent, and thus that LMFs could attract investment using
non-voting equity shares without suffering any major disadvantage in
the cost of capital. This would be a hasty conclusion. For most purposes,
the interests of voting and non-voting shareholders are closely aligned.
Corporate charters in the United States, for example, forbid payment of
greater dividends to shares having superior voting rights (Zingales, 1995).
The existence of a class of equity investors who hold voting rights may be
essential for the effective supervision of management, and the protection
of equity suppliers as a group (Putterman, 1988b). Suppliers of non-voting
equity might not enjoy comparable protection in an LMF, and such firms
might have to offer a substantially greater premium if they wanted to
keep control in the hands of workers.
Despite this generally supportive assessment of the wealth hypothesis,
cautionary remarks are in order. There are no data comparing the cost
of external capital for KMFs and LMFs, so it is impossible to determine
directly whether LMFs are disadvantaged in the credit market relative to
similar KMFs. Employees in KMFs occasionally appear to have enough
wealth to buy the firms employing them, but do not (professional sports
franchises and film production companies come to mind). Union pension
plans can also have assets exceeding the net worth of the firms that employ
the union’s members. These examples suggest that lack of funds is not
always the binding constraint on workers’ control, and that other factors,
such as portfolio diversification, are probably important as well.
Workers not only occasionally own substantial shares in conventional
firms, they also sometimes pay large upfront fees to join LMFs. In some
plywood cooperatives, for example, the asking price for the shares of re-
tirees reached $90,000 in 1980 (Pencavel, 2001), and by one account a
share sold for $100,000 (Gunn, 1984a). By 1990, the share prices for prof-
itable coops had fallen back to $55,000 (Pencavel, 2001). The Mondragon
system also imposed steep entry fees during its formative years, with new
members asked to pay amounts ranging from six months’ to two years’
income (see Section 3.3). While again this suggests that capital constraints
are not always binding, the workers who paid these prices were not in any
sense a representative sample of the labor force as a whole.
Firms themselves could help workers overcome capital-market imper-
fections. If workers’ control would improve productivity or otherwise
enhance the value of the firm, shareholders and employees could pre-
sumably split these gains by arranging for employees to buy out the firm.
Capital-market constraints should not be a serious obstacle because the
shareholders could provide financing through a payroll deduction plan
backed by the firm’s own access to credit. I will discuss such employee
buyouts in Chapters 10 and 12.
The evidence adduced in favor of the wealth hypothesis frequently has
a number of alternative interpretations. For example, the observation that
LMFs avoid capital-intensive industries might be explained by (1) limits
on worker wealth and access to financial capital; (2) free-rider problems
associated with maintenance of jointly owned assets as in Chapter 8;
(3) a correlation between capital intensity and scale economies, along
with the limited gains from mutual monitoring in large firms; (4) a corre-
lation between capital intensity and asset specificity, along with problems
connected with idiosyncratic physical assets as in Chapter 8; (5) worker
risk aversion and a reluctance to accept undiversified financial portfolios,
as will be discussed later; or (6) horizon problems associated with collec-
tive asset ownership, as in Chapter 7. Thus the mere fact that LMFs rarely
engage in capital-intensive activities is not strong support for the wealth
hypothesis. More generally, this illustrates the pitfalls in trying to draw
sharp causal inferences from highly limited data on LMFs.
9.5 Portfolio Diversification

Another popular hypothesis about the rarity of LMFs is that workers do
not manage firms because this would require them to put much or all of
their wealth into a single firm, foregoing the benefits of diversification
in coping with risk. Often this idea is accompanied by the further point
that investing funds in their own firms would compel workers not only
to forego diversification of their financial investments, but also to invest
their financial and human capital in the same venture.
Other things being equal, the members of an LMF would like to sell
claims on the firm’s risky profits to outsiders and use the proceeds to
diversify their own financial portfolios (Drèze, 1976). But if workers’
control is to be maintained, these outside equity shares must be non-
voting, which may be infeasible or expensive for the reasons discussed
in Section 9.3. The familiar alternative choice is for workers to give up
control to investors as the price of diversification. Of course, if LMFs
had sufficiently large productivity advantages, then the higher average
income achieved under workers’ control might compensate even risk-
averse workers for random variation in their income streams, but here I
ignore this possibility.
Under the diversification hypothesis, the prevalence of LMFs depends
on a mix of factors: the cost per worker of financing a firm, how wealthy
workers are, how risk-averse they are, how highly correlated the firm’s
profit is with that of other firms in the economy, and how much workers
are willing to pay to obtain control rights in the firms that employ them.
Versions of this hypothesis have been advanced by many authors including
Montias (1970), Meade (1972), Neuberger and James (1973), Schlicht and
von Weiszäcker (1977), Jensen and Meckling (1979), Gui (1985), Drèze
(1976, 1989), Putterman (1993), and Bowles and Gintis (1993b, 1994).
Bonin, Jones, and Putterman (1993: 1316) conclude that the inability of
LMF members to diversify their portfolios and moral hazard in capital
markets are the main reasons for the rarity of workers’ control.
As with the other hypotheses surveyed here and in Chapter 8, there
are numerous variants of the diversification story. In a simple version,
one assumes directly that investors are more willing to tolerate risk than
workers; that the market or technological risks facing a firm cannot be
shifted to other parties through contracts with banks, suppliers, customers,
or insurers; that equity capital must be supplied by workers if they want
control of firms; and that the size and allocation of risks do not vary
with organizational form. Given all of this, the rarity of LMFs follows
easily. But it is instructive to investigate the consequences that follow
from relaxing each of these assumptions in turn.
One minor variation is to eliminate the assumption that investors are
inherently more tolerant of risk, replacing it by an assumption that the
agents in the economy differ in their risk attitudes and sort themselves
into equity investors and employees accordingly (Knight, 1921; Kihlstrom
and Laffont, 1979). But even this degree of heterogeneity in risk attitudes
is unnecessary. The central idea is just that the typical worker’s opti-
mal financial portfolio, abstracting from control issues, contains a smaller
equity claim on the worker’s own firm than would be necessary if the work-
ers in that firm wanted to control it. This could be true even if everyone
in the economy had the same risk attitudes and wealth.
A more important assumption is that the risks associated with running
a firm cannot be transferred to other parties, such as banks or insurance
companies, without at the same time handing over control rights. The
usual reason given is moral hazard: Shifting risks to another party means
partially or fully insuring the worker-managers, who then do not have
proper incentives to manage the firm efficiently (Putterman, 1993). Simi-
larly, Bowles and Gintis (1993b, 1994) emphasize that workers must have
an equity position in LMFs because of the incentive conflicts that would
otherwise develop between borrowers and lenders.
The entire diversification story is hostage to this incentive argument,
and it is possible to imagine scenarios where the argument fails. For in-
stance, suppose that moral-hazard problems can be eliminated by having
insurers monitor LMF performance and threaten to cut off the supply of
further insurance if shirking is ever detected. Alternatively, suppose that
suppliers of non-voting equity capital behave in this way. If these threats
are effective, the risk-bearing parties need not gain direct managerial
control over the firm.
Another crucial premise for the diversification hypothesis, rarely com-
mented on explicitly, is that the risks facing the firm are exogenously given
and their allocation across agents is independent of organizational form
(but see Drèze, 1993, and Miceli and Minkler, 1995). No doubt in their
origins many risks are invariant to organizational form, including tech-
nical breakdowns and price movements on a competitive market. But
the way in which risks are divided among the input suppliers is strongly
dependent on the firm’s governance structure. A firm’s response to a ran-
dom shock depends on decisions made by the firm’s control group, and
these decisions depend on who the controllers are.
One implication is that workers may have the greatest demand for con-
trol rights in precisely those industries where the market is most volatile.
When they lack control rights, workers may face a risk of layoffs, but with
control they can respond to adversity in other ways such as reduced wages
or shorter hours. Thus workers’ control is in itself a form of insurance
against the risk that investors will make decisions contrary to workers’
interests. Even when control comes at the cost of concentrated financial
investments for an employee, and thus greater portfolio risk, this may look
like a bargain compared with the alternative of allowing shareholders to
decide whether employees will lose their jobs.
Bowles and Gintis (1993b, 1994) develop a quite different argument

involving the dependence of risk on governance structure. In their view,
different types of firms choose projects with distinct risk characteristics.
Specifically they argue that because workers are risk averse and lack diver-
sification, LMFs will select excessively conservative investment projects
by giving undue weight to risks that are socially diversifiable. KMFs, on
the other hand, tend to be unconcerned with idiosyncratic risks. A key
incentive issue confronting designers of LMFs is thus to find ways of
encouraging more risk-taking and innovation.
As it is normally formulated, the diversification hypothesis involves a
comparison between an LMF and a publicly traded KMF in which shares
are widely held and investors are approximately neutral toward idiosyn-
cratic risks. A comparison that is seldom made, but may prove more
enlightening, involves a closely held KMF whose founder has a large
fraction of her personal wealth tied up in the firm. This involves a lack of
diversification comparable to that of LMFs, but presumably the founders
of closely held KMFs are willing to accept this added risk in exchange for
the control rights that go with a position as the sole or dominant share-
holder. Moreover, it is frequently said that share concentration provides
dominant shareholders with a powerful incentive to monitor managers.
But perhaps LMFs have advantages in this respect as well, since workers
often have information that would be useful in monitoring managers.
Many writers argue that in addition to the difficulties previously iden-
tified, workers face a further problem because they cannot diversify their
human capital across firms. This lack of diversification in the labor market,
it is claimed, makes workers even less willing to invest the financial re-
sources that would be needed to acquire control over a firm. A typical
statement of this idea runs as follows: “The fact that workers are not or-
dinarily owners of their firms may be explained by workers’ high demand
for portfolio diversification given their low wealth, their relatively undi-
versifiable labor, and the fact that their incomes are already linked to the
fortunes of their firms by way of the returns on that labor” (Putterman
1993: 253). Similar arguments have been advanced by Ben-Ner (1988b).
The claim that the non-diversifiability of human capital exacerbates the
underlying problem of financial non-diversification is not as obvious as it
may appear. For instance, suppose workers will be paid the same wage
with certainty in all KMFs and LMFs, so that risks are borne entirely
by equity suppliers in each firm. Here, the inability of workers to diver-
sify their labor supply is obviously irrelevant because they gain nothing
by doing so. Of course, they might still prefer not to tie up most of their
9.6 Portfolio Diversification and Workers’ Control 197
financial wealth as equity capital in one firm for reasons already given,
but that is a separate issue. The same logic applies if the equilibrium wage
is a random variable but identical across firms, as might be true for a
competitive labor market subject to unpredictable exogenous shocks.
An inability to diversify in the labor market does matter if firms pay
wages that are less than perfectly correlated. This could occur for various
reasons, including search costs or firm-specific investments in location and
human capital. Here, the inability of workers to diversify their labor across
firms can have real consequences for their willingness to supply equity to
the particular firm in which they work. But even if one grants that this
may have some effect on portfolio decisions, the relevant comparison is
not between a world where diversification by labor is feasible and one
where it is not. The real comparison is between economies populated by
KMFs and LMFs for a given set of diversification possibilities.
Finally, the fact that risks are allocated in part through a firm’s gover-
nance structure has implications for the treatment of firm-specific human
capital. Far from being a reason why workers would not want control,
as in the usual diversification story, the existence of large firm-specific
investments by labor could well increase the demand for control rights,
because control offers protection against the danger that a firm might
respond to an external shock by destroying quasi-rents currently enjoyed
by the workforce (Blair, 1995). In fact, the prospect that workers would
lose such quasi-rents has apparently precipitated a number of employee
buyouts of capitalist firms (see Section 10.3).
9.6 Can Portfolio Diversification Explain the Rarity

of Workers’ Control?
The risk-aversion framework accounts for the fact that LMFs do not often
appear in capital-intensive sectors of the economy, because workers would
have to make large equity investments that would oblige them to forego
diversification. It is also consistent with the evident fact that diversification
plays a key role in the investment strategies adopted by most employee
pension plans, and by workers who manage personal savings of their own.
At the same time, there are some intriguing anomalies. As discussed in
Section 4.2, ESOPs are exempt from conventional legal restrictions on the
fraction of a pension plan’s assets that can be invested in a company’s own
stock. ESOPs typically use most or all of their assets to buy shares in the
employee’s firm, and an ESOP account is likely to be a large proportion
of total wealth for at least some employees. This concentration of assets
is hard to square with the diversification hypothesis as a theory about the

rarity of LMFs. The same is true for the increasing use of profit-sharing,
stock options, and other forms of risky compensation in conventional
firms (Kruse, 1993). It can still be argued, however, that such organiza-
tional devices impose less risk on workers than they would need to bear
if they wanted to exercise full control over the firm.
Supporters of the diversification hypothesis do not always take the
same view about its predictions regarding the cross-industry distribution
of LMFs. At an early stage, Meade predicted that for LMFs to arise, “the
risk of fluctuations in the demand for the product must not be too great”
(1972: 426). However, plywood cooperatives face large variations in input
and output prices (for data see Craig and Pencavel, 1992; Pencavel and
Craig, 1994), but have nevertheless survived and attracted new members
for over half a century. Similar remarks apply to construction cooperatives
like those of Italy (Section 4.1), which are exposed to major demand
fluctuations over the business cycle, and the airline industry, which in the
United States has seen several major employee stock purchases (Gordon,
1999).
This evidence is clearly in conflict with the Meade formulation. Bonin,
Jones, and Putterman, who are aware of the evidence concerning the ply-
wood and construction cases, provide a different prediction: “We should
find a higher concentration of PCs [producer cooperatives] in industries
that exhibit more pronounced cycles so that members can partially in-
sure themselves against employment risk” (1993: 1314). While perhaps
correct, this turns the diversification hypothesis on its head by suggest-
ing that risk-averse workers will establish LMFs in industries that are
especially risky.
Empirical research on ESOPs and profit-sharing has not revealed any
relationship between the adoption of such plans and the risk properties of
the industries in which the adopting firms reside (Ben-Ner, Burns, Dow,
and Putterman, 2000). One interpretation of these negative results is that
ESOPs, profit-sharing plans, and LMFs may have productivity advantages
that enable firms to compensate employees for bearing additional risk by
paying higher wages. But it is also possible that the right measures of
industry risk have not yet been found, or that workers regard the risks
involved as small.
There is good evidence that governance structures condition the way in
which risks are allocated within firms. One of the most robust empirical
generalizations about LMFs is that they respond to negative demand
shocks by maintaining employment while restricting hours, wage rates,
9.6 Portfolio Diversification and Workers’ Control 199
or both. This is true of the plywood cooperatives (Pencavel and Craig,

1994) as well as Mondragon (Bradley and Gelb, 1987, and references in
Section 3.3) and most other workers’ coops (Bonin, Jones, and Putterman,
1993). Such behavior differs from the typical KMF response of absorbing
shocks through layoffs, and as was discussed in Section 7.6, it is consistent
with the view that LMFs deviate from profit maximization by giving extra
weight to employment in their objective functions (Craig and Pencavel,
1993).
If risk aversion were the central factor behind the prevalence of cap-
italist firms, one might expect it to be equally important in shaping the
employment contracts within capitalist firms. In particular, if employees
were concerned about risks associated with undiversified human capital,
one would expect that in industries where layoffs are costly to workers
and occur with a high probability, firms would have to pay employees
a wage that included a large risk premium. There is evidence that the
present value of lost income due to layoffs is quite large for many work-
ers (see Schultze, 2000 and references cited there). But Murphy and Topel
(1987: 130) have found that a one standard-deviation increase in the vari-
ance of annual weeks worked (4.75 weeks) only requires compensation
in average annual earnings of about $45. This modest effect suggests that
risk aversion is unlikely to be a key factor in shaping organizational form,
although compensating differentials of this sort may simply be hard to
detect.
Finally, there is an extensive literature on the role of risk aversion as a
determinant of agricultural contracting arrangements. As in the principal-
agent framework from Section 6.4, it has been claimed that sharecropping
represents a compromise between the need to maintain effort incentives
and the benefit from having a risk-neutral landlord insure a risk-averse
farm worker. But empirical work has cast doubt on this story, suggesting
that the risks associated with particular crops or farm locations have little
or no power to explain the contractual arrangements observed in North
American agriculture (Allen and Lueck, 1998). The alternative hypoth-
esis for which Allen and Lueck do find support is one of pure moral
hazard: The terms of agricultural contracts are strongly influenced by the
amount of damage that can be done to the land by a non-owning farmer
(a problem analogous to the equipment abuse dilemma in Section 8.2).
As Hansmann (1996: 78) notes, agriculture is quite volatile, but family
farmers have long been willing to bear these risks. It is unclear why work-
ers in the manufacturing and service sectors would not be prepared to do
the same.
9.7 Collective Choice

A small but expanding literature claims that LMFs are at a disadvan-
tage relative to KMFs because worker preferences tend to be more het-
erogeneous than those of investors. The idea is that capital suppliers
unanimously support the maximization of profit or present value, while
workers have widely diverse attitudes toward effort, scheduling, job secu-
rity, safety, social atmosphere, and other aspects of the workplace. Writers
taking this approach have in general come to one of two conclusions:
Either that it is more costly for LMFs to make decisions than it is for
KMFs, or that LMFs are an unstable mode of organization.
The most prominent exponent of the collective choice hypothesis has
been Henry Hansmann (1988, 1990a, b, 1996; but see also Jensen and
Meckling, 1979; Ben-Ner, 1988b; Benham and Keefer, 1991; and Milgrom
and Roberts 1992: 562–63). Hansmann adopts a transaction cost approach
(Section 6.3) and suggests that heterogeneity of worker preferences tends
to increase the cost of reaching decisions in LMFs. The precise nature of
these costs depends on the decision-making mechanism.
If decisions are reached through bargaining among members of the
control group, one must recognize that bargaining can be costly under
conditions of private information (Kennan and Wilson, 1993). This prob-
lem is aggravated by a high-dimensional policy space and the presence
of bargainers who have highly diverse preferences over this space, com-
pared with bargaining among investors for whom the central concern is
simply how to divide the wealth from profit maximization. Even without
private information, complex multilateral bargaining is likely to be costly
in time and other resources.
If decisions are instead reached through voting, a problem of voting
cycles arises. Whenever the policy space has more than one dimension,
for each possible decision there is typically another decision preferred by
a majority of voters (Plott, 1967; Kramer, 1973; McKelvey, 1979). There-
fore if workers elected a board of directors, they might be unable to
provide their representatives with coherent guidance about the objec-
tives to be pursued. Hansmann (1996: ch. 5–6) argues that each possible
solution to the cycling problem creates some disadvantage for the LMF
in relation to the KMF. For instance, the firm could screen new mem-
bers, avoid an extensive division of labor, or restrain skill and wage in-
equalities in order to limit the heterogeneity of preferences. It could also
stay at a suboptimal scale to limit the number of decision-makers. Each
of these responses imposes a cost on the firm, either directly or through
lower productivity. Hansmann does not deny that conflicts also arise from
time to time among investors, but he maintains that the costs of collective
choice are usually greater in LMFs, placing these firms at a competitive
disadvantage. In fact, “if costs associated with collective self-governance
were not a problem, employee ownership would be far more widespread
than it is” (1996: 92).
According to Hansmann, such costs can only be avoided by limiting
the scope of enterprise democracy itself. For example, the governance
structure of the firm might be designed in a way that restricts the right of
workers to place new proposals on the agenda; replaces direct participa-
tion with representative democracy; disenfranchises certain groups such
as clerical workers whose interests diverge from those of the dominant
coalition; or delegates many substantive decisions to managers. Thus, it is
argued, in order to survive in a competitive world, the LMF must settle
for a pale imitation of democratic procedures.
Zusman (1992) specifies the transaction costs arising from collective
choice in more detail, suggesting that firm governance structures will
minimize the sum of bargaining costs and the members’ risk premiums.
Zusman handles the cycling problem by treating it as one of uncertainty:
If the firm adopts a voting process under which cycling is possible, then
every efficient managerial choice is assigned equal subjective probability.
Similarly, all bargaining coalitions have equal probability of forming, and
if a constitutional rule creates a privileged position of authority, then all
members of the firm are assumed to have the same subjective probability
of occupying that position. Zusman illustrates these concepts using an
example with three workers and two dimensions of choice.
A further variation on the collective choice theme is the idea that firms
adopting the principle of one vote per worker will suffer from excessive
egalitarianism with respect to income distribution (Kremer, 1999, unpub-
lished manuscript). Suppose LMF members differ in their productivities,
and the median ability is below the mean. Then a majority of voters can
gain by making differences in pay smaller than differences in productivity,
thus transferring resources from the most productive to the least produc-
tive workers. Ability and effort are not easily distinguished, so this policy
resembles a tax that tends to dull effort incentives.
In the absence of mobility barriers, high-ability workers could respond
by fleeing to jobs in firms owned by shareholders where they would be
paid their marginal product. But if workers must invest some capital in
the LMF and are uncertain about their own abilities ex ante, there will
be scope for redistributive activities. LMFs could benefit by committing
themselves in advance not to behave in this way, but Kremer argues that
it is too difficult to entrench the required constitutional safeguards. As he
recognizes, the argument would not work if there were a perfect ex post
market for membership rights, but he points out that such markets may
fail in practice because of adverse selection on worker ability.
Unlike most writers in this category Kremer confines attention to a one-
dimensional space where a median voter equilibrium exists. The core of
his argument is the proposition that the controlling majority in an LMF
cannot credibly promise to respect the quasi-rents of a minority. Thus
the argument has more in common with the literature on exploitation of
minority shareholders in KMFs than with theories based on preference-
aggregation issues.
Other writers have argued that preference heterogeneity among work-
ers makes the LMF an unstable mode of organization. For example,
Ognedal (1993) suggests that mixed patterns of share ownership, where
both workers and external investors have voting rights proportional to
their shares, are often unstable if unrestricted trading in shares is allowed.
Her analysis relies on cooperative game theory and rests on the idea that
majority voting can be exploited to redistribute profit away from a mi-
nority of shareholders. The instability result is thus closely related to the
problem of voting cycles.
Skillman and Dow (2002, unpublished manuscript) also study the rel-
ative stability of the KMF and LMF when individuals are free to sell their
ownership shares and hence their votes. Assuming that financial capital
is costlessly mobile and that investors therefore operate in a perfectly
competitive environment, theorems drawn from the finance literature
(Hart, 1979) imply that within each firm, investors unanimously favor
profit maximization. However, because of larger mobility costs, workers
are confined to monopolistically competitive regional labor markets. This
leads workers to have heterogeneous preferences toward firm policies.
If investors can commit themselves in advance to implement particular
policies, the LMF is generically unstable because there is almost always a
policy change that induces 51 percent of the members to sell their gover-
nance rights. Even if investors are unable to make binding commitments
of this kind, LMFs are vulnerable to takeover by investors unless a ma-
jority of the current membership opposes any move in a profit-enhancing
direction.
The fundamental problem behind the collective-choice hypothesis is
the familiar one of preference aggregation (Arrow, 1963), and the idea
behind all versions of the hypothesis is that this aggregation problem is
in some sense more difficult for workers as a group than for capitalists as
a group. The key asymmetry between capital and labor in this hypothe-
sis is that workers must generally be physically present at a production
site in order to supply their inputs, while owners of capital need not be.
This has a wide variety of implications – for instance, financial capital is
more mobile than human capital and it is easily diversified across firms.
Furthermore, the physical presence of workers opens up the possibility
that they will have heterogeneous preferences toward local public goods
in the workplace.
As noted earlier, the mobility and diversifiability of capital may lead
to stock market equilibria in which shareholders unanimously agree on
firm policies. It can clearly be debated whether the conditions required
for theoretical demonstrations of this sort apply to real stock markets.
Casual empiricism indicates that shareholders sometimes do disagree
over the strategies pursued by firms. Disagreements can result from dif-
ferences in beliefs, risk attitudes, time horizons, or tax status (Bainbridge,
1996) as well as differences in the cost of trading stock, nearness to
voting control, or auxiliary relationships to the firm as employee, sup-
plier, or purchaser (Benham and Keefer, 1991). However, despite these
caveats, it seems reasonable to postulate shareholder unanimity as a first
approximation.
On the labor-market side, workers would be unanimous in their pol-
icy preferences within each firm if there were more firms than types of
workers so that workers could sort themselves appropriately across firms
(Jensen and Meckling, 1979: 491). Indeed, without scale economies, every
firm could consist of a single worker and preference heterogeneity would
never arise. Alternatively, no collective-choice problem would arise if the
number of firms was large compared with the number of employment di-
mensions workers cared about, because this would generate a complete
array of implicit prices for job characteristics.
When there are too few firms compared with the size of the charac-
teristics space, in equilibrium there will be preference diversity among
the workers within each firm, and the policies of firms will be semi-public
goods. Highly restrictive conditions on preferences are needed to ensure
that profit maximization by firms is consistent with the efficient supply
of these public goods (Drèze, 1976; Drèze and Hagen, 1978). As Drèze
points out, it may be possible to make everyone better off simultaneously
by altering working conditions and reallocating workers across firms.
If these public good problems can be solved by an LMF, this yields
an efficiency argument in favor of LMFs. Related ideas have been used
to argue for the view that worker voice expressed through collective
bargaining can provide efficiency benefits (Freeman, 1976; Freeman and
Medoff, 1984). Hansmann recognizes that workers may have trouble cred-
ibly communicating their preferences to managers of KMFs, and hence
“investor-owned firms may be handicapped in fashioning the most effi-
cient package of financial compensation and working conditions for their
employees.” But he goes on to make the empirical claim that “employee
ownership tends to appear in precisely those settings in which manage-
ment is likely to have relatively little difficulty understanding employees’
preferences” (1996: 73). This suggests that LMFs arise not as a solution
to public goods problems, but rather only in environments where such
problems are minor or trivial.
9.8 Can Collective Choice Explain the Rarity of Workers’ Control?

Hansmann (1996) and Benham and Keefer (1991) argue that almost all
LMFs take steps to limit preference diversity, the scope of democratic
governance, or both. Common strategies to control diversity include stay-
ing small and carefully screening new members for social compatibility.
Hansmann emphasizes that in most worker-controlled firms, the members
do extremely similar work and there is rarely much hierarchical authority
among them (1996: 91). Pittatore and Turati (2000) endorse this portrait
based on evidence from Italian workers’ cooperatives in the service sector
(see Section 4.1).
When workers do not have uniform tasks or skills, voting rights in
LMFs are often confined to a homogeneous subset of the workforce,
with others being hired as employees or contractors. Homogeneity is also
sometimes ensured in other ways – for instance, by giving preference to
applicants with kinship ties to current members, or by ensuring ethnic
or religious uniformity (see Section 6.8). Strategies to limit democratic
governance include appointment of persons or committees with agenda-
setting power; delegation of extensive authority to hired managers; and
establishment of rigid constitutional rules about wages, investment, job
security, or other matters (Benham and Keefer, 1991).
Another piece of evidence favorable to the collective-choice hypothe-
sis is that with the exception of Mondragon and some KMFs bought out
by their employees, there are few examples of multi-establishment LMFs.
This corroborates the hypothesis because spatial dispersion provides a
natural basis for the emergence of conflicting interests (for example, over
the allocation of investment funds among establishments). It is also worth
9.8 Collective Choice and Workers’ Control 205
recalling from Chapter 3 that in the plywood cooperatives and Mondragon

there have been periods in which hired managers were routinely dis-
missed, a form of instability that might indicate the presence of voting
cycles among members.
There is much evidence to support Kremer’s view that LMFs will adopt
internally redistributive policies. The Mondragon cooperatives have
rules specifying the maximum acceptable pay differential (Section 3.3),
and wage compression has been found in firms belonging to the Lega
(Section 4.1). But investor-owned firms are not immune to concerns with
pay equity (Baker, Jensen, and Murphy, 1988), and earnings compres-
sion may have beneficial effects on morale that outweigh the negative
incentive effects emphasized by Kremer. Most importantly, there is little
evidence that worker-controlled firms are plagued by weak work incen-
tives as Kremer’s theory predicts (see Section 8.7).
A key counterexample to the collective-choice hypothesis is the
Mondragon system. This group of cooperatives involves considerable
worker heterogeneity along several axes including skills, technology,
product line, and geographic location. Hansmann (1996: 98–103) argues
that the Mondragon group succeeded by relying on representative rather
than direct democracy, delegating substantial authority to managers,
and restricting decisions by individual cooperatives. Hansmann thus re-
gards the Mondragon case as corroborating his hypothesis. A different
lesson that could be drawn, however, is that collective-choice prob-
lems need not be fatal as long as LMFs adopt appropriate gover-
nance structures. On this interpretation, such problems might explain
certain aspects of organizational design in successful LMFs but not
the more general rarity of LMFs per se. A related point raised by
Pencavel (2001) is that causality may run from workers’ control to work-
force homogeneity, the reverse of what the collective-choice hypothesis
suggests.
Another possible set of counterexamples involves employee buyouts
of financially troubled capitalist firms. This subject will be discussed in
Section 10.3, so I will not go into detail here. However, such firms are not
usually found in the same industries as traditional workers’ cooperatives,
and the usual generalizations about size, labor intensity, and internal ho-
mogeneity of the workforce do not always apply. Hansmann argues that
these buyouts may be transitional phenomena rather than equilibrium
modes of organization (1996: 83–84) and that they often do not result in
direct voting rights for employees. But some employee buyouts have per-
sisted for at least one or two decades without reverting to a conventional
capitalist structure, and such buyouts do occasionally result in genuine

workers’ control.
Perhaps the main drawback of the collective-choice approach is that it
does not offer any leverage in explaining a range of other generalizations
about LMFs, including the fact that LMFs seldom arise de novo in indus-
tries that are capital intensive, have strong scale economies, or involve
substantial physical-asset specificity. For this reason, it is best seen as one
part of the story about the rarity of workers’ control, not the entire story.
10
Transitions and Clusters
10.1 Organizational Demography

The population of KMFs at any point in time depends on past rates of
KMF creation and destruction. It also depends on past rates at which
KMFs have been converted into LMFs, and vice versa. Given the small
population of LMFs, the latter effects are negligible relative to the stock
of KMFs. The number of LMFs at any point in time also depends on past
rates of creation and destruction as well as past rates at which KMFs have
become LMFs, and vice versa. However, transitions between the KMF
and LMF structures play a non-trivial role in determining how many
LMFs there are at a point in time, because such flows are significant
relative to the overall population of LMFs. Understanding these processes
is therefore an important objective in explaining why LMFs remain rare.
Section 10.2 begins with birth rates of new firms and considers why
there might be a systematic bias favoring the KMF structure at this stage.
Section 10.3 addresses worker buyouts of KMFs, and Sections 10.4 and
10.5 take up the opposite process, the conversion of LMFs into KMFs. I
discuss the evidence concerning survival rates of LMFs and KMFs in Sec-
tion 10.6. Section 10.7 takes a brief look at the effects of the business cycle
on the LMF population, and Section 10.8 examines the tendency of LMFs
to cluster in particular industries, geographical regions, and historical eras.
10.2 Formation Rates

The 1970s and early 1980s were a time of especially rapid LMF formation.
Avner Ben-Ner (1988a: 14) has quantified LMF birth rates for several
European countries between 1976 and 1984. The annual percentage
207
208 Transitions and Clusters
increase in the number of LMFs, with some variation by country in the

particular years covered, was 11.4 percent in France, 24.5 percent in the
Netherlands, and 35.3 percent in the U.K. These figures include only firm
creation, not transformations of KMFs into LMFs through buyouts by
employees, and compare favorably with birth rates for KMFs in France
and the U.K. at about the same time (5.3 percent for manufacturing in
France, 9.4 percent for all industries in the U.K.). Pérotin (1997: ch. 7)
found that in France, the population of workers’ cooperatives expanded
by 70 percent between 1971 and 1980 and 60 percent between 1980 and
1992. The corresponding figures for firms in general were 50 percent and
20 percent.
When interpreting such figures, it must be borne in mind that birth rates
for LMFs involve a low population base. Since there are few LMFs to be-
gin with, a sudden wave of firm formation automatically implies a high
rate of growth. For example, the Industrial Common Ownership Move-
ment in the U.K. estimated that in 1975, roughly 30 worker cooperatives
were using its model rules, while in 1982 this had risen to 400 (Cornforth,
1983; such estimates vary widely depending on the source). Staber (1989)
discovered an equally dramatic rise in formations of workers’ coopera-
tives in Atlantic Canada during the early 1970s. Despite this upsurge, it
is clear that in absolute numbers, the creation of new KMFs far outpaces
the creation of new LMFs in all years and in all Western economies. As
Aldrich and Stern remark in a historical study of 595 cooperatives in the
United States, the number created from 1835 to 1935 “represents only a
tiny fraction of the total number of business organizations created over
this time span” (1983: 376). The same is true today.
One reason for a systematic bias toward capitalist firms at the forma-
tion stage is that workers have limited personal wealth and cannot easily
attract external financing. To see the economic mechanisms at work, as-
sume there are disagreements about the profit to be derived from various
entrepreneurial projects. Each person has some favorite projects and be-
lieves that other projects are a waste of time and money. All potential
projects require assets that must be owned by the firm itself (they cannot
be leased for reasons discussed in Sections 8.2–8.3). Each project requires
the labor services of several people because of the advantages of team
production or a division of labor, and the cost of the necessary assets
exceeds the joint wealth of most such worker teams.
What must happen in order for an LMF to be created? Most obviously,
a number of labor suppliers who agree on the merits of a common project
must be assembled. At the outset, someone must incur the cost of finding
like-minded workers, and there is a free-rider problem because the people

bearing these search costs will not appropriate all the benefits that flow to
other LMF members after the firm is created. Assuming such a group has
been assembled, the workers must now find a lender who believes that
their chosen project will be profitable. Searching for a sympathetic lender
is once again costly. Moreover, adverse selection problems may arise if it
is common knowledge that some projects are completely worthless, but
lenders cannot tell in advance which projects these are. Workers who have
a good project may then face prohibitive interest rates or find it impossible
to get a loan.
A wealthy investor can pursue a similar project by purchasing the
necessary assets, hiring a group of employees, and telling them what to
do. Given sufficient personal wealth, the investor will not have to incur any
search cost to locate other like-minded investors. At worst, a small group
of partners can pool their financial resources in starting up a firm. The
free-rider problem in assembling a set of capital suppliers who agree on
the same project is less severe than the parallel problem for labor suppliers
if the required number of investors is smaller than the minimum size of an
efficient worker team. KMF employees need not have any opinion about
the merits of a project as long as their wages are paid on time.
Why does the KMF founder, who by assumption believes in a given
project, not extend a loan to poor workers wishing to create an LMF based
on that project, or offer to finance a transfer of the KMF’s assets to its
workforce through a debt contract? Once again, it is not enough to have
assembled a random sample of workers drawn from the local labor force.
The current employees may not agree among themselves on the value of
the project, so it would be necessary to recruit a group of workers with
relatively uniform beliefs. This is vital because normally it is impossible
to finance a firm entirely through debt (see Section 9.2), and thus the
workers would have to put some of their own wealth at risk. Moreover,
just as lenders are vulnerable to adverse selection when teams of workers
apply for loans, workers are vulnerable to the danger that the capitalist
actually has a bad project that is not easily distinguished in advance from
a good project.
Such problems are not just hypothetical. Berman (1967) has provided
a fascinating account of the speculative atmosphere that surrounded the
creation of plywood cooperatives in the early 1950s. Encouraged by the
rapid rise in the price of membership shares that had occurred in pre-
vious coops, promoters descended on Washington state and Oregon and
persuaded workers to join newly established cooperatives. The promoters
themselves usually had no intention of working in the plywood mills and

often had little knowledge of the industry, although they claimed to offer
financial and organizational expertise or access to novel technology. It
is clear that in many cases, promoters were attracted by the prospect of
quickly reselling their own shares to capture a fast capital gain. In two
cooperatives, this led to criminal trials for securities fraud and mail fraud.
Most observers link the abrupt end of coop formations around 1956 with
the publicity given to these events in the local media.
Thus far I have stressed obstacles to LMF formation resulting from
capital-market constraints. But heterogeneity of beliefs and adverse-
selection problems would also tend to prevent the formation of worker-
controlled firms even if large upfront investments were unnecessary. The
reason is that KMFs are an attractive vehicle for the appropriation of
entrepreneurial rents, while LMFs are not. This is a common theme in
the LMF literature, and I will review the ideas of several other writers
before offering my own interpretation.
In his study of LMF life cycles, Ben-Ner notes that “a self-interested
entrepreneur will not choose to establish a worker-owned firm and share
entrepreneurial profits with others, if the establishment of a capitalist firm
is a viable alternative” (1988b: 290). LMF formation thus often requires
support from “the state, political movements, trade unions, worker-owned
firm umbrella organizations, and philanthropists” (1988b: 291) willing
to create such firms even if they gain little personal profit. Jensen and
Meckling (1979: 489) and Estrin (1989) make similar remarks. Putterman
(1982) argues that the concentration of returns to innovation gives KMFs
an advantage under bounded rationality, which impedes negotiation over
the division of returns within the LMF and can imply that the per-person
benefit from LMF formation is too small to attract the attention of indi-
vidual workers.
Marglin (1974, 1984) likewise argues that capitalist firms are better or-
ganizational vehicles than LMFs for the appropriation of entrepreneurial
rent. In addition to the factors already discussed, an innovating en-
trepreneur has a strong incentive not to reveal crucial proprietary infor-
mation to anyone else because this only increases the number of poten-
tial competitors. In particular, an entrepreneur prefers not to share novel
ideas with a large number of co-workers, some of whom might leave to
start a competing firm.
In an LMF where all members share control rights, it would be dif-
ficult for any one person to withhold vital information from the other
members even if that person happened to be the founder of the firm.
In a KMF, this problem is less severe because employees can be hired

at a competitive wage and told what to do without being told why. Fur-
thermore, a division of labor among employees is helpful in preserving
the entrepreneur’s monopoly on key technical or market data. This is the
idea behind Marglin’s much quoted remark about the dominance of the
capitalist firm during the Industrial Revolution: “The capitalist division
of labor . . . was the result of a search not for a technologically superior
organization of work, but for an organization which guaranteed to the
entrepreneur an essential role in the production process, as integrator of
the separate efforts of his workers into a marketable product” (Marglin,
1974: 62; see also 71, footnote 11). But there is nothing unique about the
birth of industrial capitalism in this regard. The same process is at work
whenever any entrepreneur uses a capitalist firm to capture rents from
innovation.
The appropriation story does not logically necessitate a KMF because
in principle the entrepreneur could rent equipment or borrow the re-
quired capital. Thus the founder and controller of the firm need not be
a capital supplier. Marglin recognizes this, commenting at one point that
“the essence of the capitalist’s contribution is not capital, but organizing
ability” (1984: 150). But there are a number of reasons apart from credit
rationing why the entrepreneur would also own some or all of the firm’s
physical assets (Sections 8.2–8.3), and in any case the reasons for avoiding
workers’ control are clear enough.
The simple version of the appropriation problem, where LMFs force
entrepreneurs to share profits with many other workers, is actually a re-
cycled description of the common property problem from Section 7.4. It
therefore has a parallel theoretical solution: Set up a membership mar-
ket for the LMF. This creates a mechanism through which newcomers to
the firm can compensate incumbent members for sharing product-market
rents or returns on past investments. The difference here is that there are
no past investments and there is only one incumbent member of the firm,
the founding entrepreneur. But the entrepreneur might still capture the
present value of future profits by selling membership positions in the
LMF, rather than establishing a KMF and hiring employees at a market
wage.
What’s wrong with this scenario? Many things. First, prospective LMF
members would have to be persuaded that the entrepreneur’s idea was
as valuable as claimed. The resulting adverse-selection problem could
be severe because workers would need to form a precise estimate of
the firm’s present value despite its lack of any track record. To sell off
membership rights at a high price, the entrepreneur would have to disclose

some credible information about the technological or economic basis for
the rent involved, but dispersal of information would allow some member-
ship candidates to create their own firms instead. Finally, if potential re-
turns were large, prospective members might lack sufficient wealth to pay
the full present value of membership, if this were known or guessed. If the
workers asked a bank rather than the entrepreneur for a loan to finance
their membership rights, this would add another party who would have to
be convinced of the merits of the project, and yet another channel through
which information might leak out. Thus there is little prospect that a mem-
bership market can remove appropriation biases in favor of the KMF at
the firm-formation stage.
The rent-appropriation hypothesis has empirical content, although
there have been few, if any, direct tests. Most obviously it predicts that
KMFs will be common in industries having a high rate of technological
innovation, while LMFs will appear in industries with stable, well-known
production methods. This prediction is consistent with the emergence
of workers’ cooperatives in the plywood, clothing, construction, and re-
forestation fields, and with the prevalence of professional partnerships.
One might also speculate that other things being equal, LMFs would
arise more often when there are good substitutes for secrecy in protect-
ing proprietary information, such as cheaply and reliably enforceable
patents.
LMFs are sometimes created despite their deficiencies as rent-
appropriation devices. They might be formed by workers who regard
democratic governance as desirable in itself, or who place a particularly
high value on job security (Ben-Ner, 1988b; Bonin, Jones, and Putterman,
1993). At times, they have also been formed for ideological or religious
reasons (Benham and Keefer, 1991). But in a world where most firms are
created in order to bring material benefits to their founders, LMFs are
clearly the underdogs.
10.3 Worker Takeovers

Once a capitalist firm has grown and information asymmetries from the
formation stage have receded, an employee buyout might become a prac-
tical proposition, especially if LMFs have productivity advantages. How-
ever, most capitalist firms are not bought out by their employees. This
could indicate that LMF productivity gains are a mirage. But the evidence
from Section 8.7, imperfect though it is, suggests that LMFs often perform
at least as well as KMFs on standard measures of labor productivity. So

why don’t employees buy out shareholders and exploit these advantages?
They occasionally do, but infrequently and only under special condi-
tions. Abrupt transitions to workers’ control usually involve a firm that
is in financial distress. Ben-Ner’s 1976–84 data for several European
countries shows that in each year, at least twice as many conversions
of KMFs into LMFs occurred through employee takeovers of financially
troubled firms compared with takeovers of financially healthy firms. For
most countries studied, these numbers differed by a ratio of three or four
to one (1988a: 14). Quarter and Brown (1992) report that of the thirty-nine
worker buyouts operating in Canada in June 1990, half had arisen from
plants at risk of closure. Keeping in mind that most KMFs at any given
time are not in financial difficulty, the LMFs resulting from conversion
of KMFs thus emerge quite disproportionately from the ashes of failed
capitalist firms. In what follows, I review two North American cases of
employee takeover, Algoma Steel and United Airlines, and then venture
some analytical comments.
My account of the Algoma Steel case is based primarily on Gunderson
et al. (1995), with additional information taken from Nishman (1995).
Algoma is Canada’s third largest integrated steel producer, with 13 per-
cent of Canadian raw steel output. It is located in Sault Ste. Marie, Ontario,
near the confluence of Lakes Michigan, Superior, and Huron. From 1988
to 1992, it was owned by Dofasco, the largest Canadian steel producer.
Algoma lost money in seven out of nine years from 1982 to 1990. Among
its troubles at various times in this period were recessions, high interest
rates, a high Canadian dollar, competition from low-cost imports, and a
four-month strike in 1990. In January 1991, Dofasco announced that it
would end financial support to the plant. It was estimated at the time that
if Algoma closed, one family in four would leave Sault Ste. Marie, impos-
ing hundreds of millions of dollars in costs on the provincial and federal
governments. The provincial premier immediately convened a meeting of
Algoma and Dofasco executives, together with cabinet ministers, to dis-
cuss short-term financing arrangements that might keep the firm afloat.
In February 1991, Algoma applied for bankruptcy protection, and
it remained in this state until June 1992. During this time, it was kept
open through voluntary wage reductions and matching contributions from
the provincial government, banks, and Dofasco. In May 1991, Algoma
management proposed a restructuring plan under which the firm would
close half of its operations, lay off 3,000 out of roughly 5,000 employees,
impose a 20 percent pay cut, and eliminate pension indexing as well as
cost-of-living allowances. There would also be a seven-year moratorium

on strikes. In return, the employees would receive 14 percent ownership
of the firm’s common shares, a wage increase of 1.5 percent in 1992, and
wage increases of 3 percent annually between 1993 and 1996.
This was rejected by the United Steelworkers of America, which coun-
tered with a proposal in August 1991 involving majority employee share
ownership but only minority representation for workers on the board
of directors. The final agreement in January 1992 included a plan for
60 percent employee equity ownership by 1997, with major lenders tak-
ing the remaining 40 percent. According to Gunderson et al. (1995), this
agreement also involved a 16 percent wage reduction for hourly work-
ers, elimination of a week of vacation time for two years, and a bonus
freeze; salaried workers accepted a 14.5 percent pay cut. (Nishman, 1995,
reports that the overall wage rollback was 20 percent along with a bene-
fits freeze). The workforce was to be cut by 1,600 through attrition. The
Ontario provincial government provided loan guarantees of $110 million
(Canadian), while Dofasco gave $30 million in cash and $10 million in
guarantees. The federal government offered bridging pensions to older
workers. (Nishman’s version of the story again differs a bit; he cites
$95 million in loan guarantees and $50 million for training and retire-
ment benefits as a combined contribution of the Ontario and federal
governments.) As a result of Algoma’s bankruptcy, the equity acquired
by workers was nearly worthless.
All full-time employees received the same number of shares regardless
of their labor earnings. Employees had to hold their shares for five years
and could sell them only after permanent departure from the company.
Voting by employees was on a one person, one vote basis. Workers had
veto powers over the sale, merger, or transfer of the company; proposals
that would reduce employee ownership below 50 percent; other major
changes in the business plan of the company; and purchases of businesses
or assets outside the steel industry. There was a thirteen-member board
of directors, including four directors chosen by the steelworkers’ union,
one by salaried employees, the CEO, and seven “independent” directors.
The latter included shareholder representatives along with former politi-
cians and others. Thus, although employees did not have a clear majority
on the board to reflect their majority equity stake, outside shareholders
also lacked a majority. The independent board chairman, Earl Joudrie,
had been instrumental in saving the company via his appointment by the
Ontario government as Chief Restructuring Officer in 1991, and was still
chairman as of 2001. The buyout package also involved the establishment
of a joint steering committee to develop a workplace-participation pro-

cess aimed at increasing productivity.
Although it is unclear to what degree this can be attributed to the re-
structuring of the company, Algoma had a rapid recovery in 1993 and
1994, with its price per share rising to $16.50 from a low of 5 cents. Af-
ter several good years, Algoma encountered slowing U.S. demand, high
debt burdens, and intense foreign competition. Union agreements have
recently imposed wage freezes and workforce reductions through attri-
tion. In April 2001, the firm again sought bankruptcy protection from its
creditors (Algoma Steel, 2001).
My discussion of United Airlines (UAL) relies on a detailed case study
by Jeffrey Gordon (1999), supplemented with some background informa-
tion from Henry Hansmann (1996). At this writing, it remains to be seen
whether the buyout of UAL by three employee groups will continue as
a successful venture. Conflicts over wages and other issues led to work
slowdowns in 2000 and 2001, but United is not alone among U.S. airlines
in having experienced persistent labor conflict.
A first effort at an employee buyout of UAL occurred when 7,000
pilots negotiated a preliminary agreement to purchase the firm in 1989.
This agreement failed in part because of opposition from the 23,000 mem-
ber machinists’ union (Hansmann, 1996: 117–18). In July 1994, the pilots’
and machinists’ unions together acquired 45 percent of UAL’s stock and
salaried employees acquired another 10 percent, with 45 percent remain-
ing in the hands of public shareholders. Employees agreed to wage cuts
of 15 percent for at least three years and accepted changes in work rules
for twelve years. The wage cuts were valued at $3.3 billion and work-rule
changes at $1.6 billion, so the employee contribution totaled $4.9 billion.
The old UAL shareholders received $2.1 billion in cash plus 45 percent of
the common equity in the recapitalized airline, the latter being valued at
$1.1 billion. Because the 55 percent claim going to the employees was in
convertible preferred shares, these were assigned a greater than propor-
tional value of $1.9 billion. The employees were also given a job-security
promise. It can be inferred that this job-security pledge and the new labor-
control rights to be described next were assigned an implicit value of at
least $1.4 billion (the wage loss of $3.3 billion minus the share valuation
of $1.9 billion, ignoring the cost of the work-rule changes to employees;
see Gordon, 1999: 340–41).
The UAL buyout did not result from a threat of imminent financial dis-
aster. UAL did, however, suffer large losses (about $1 billion) between
1990 and 1993 and the buyout was an element in a long-term shift in
corporate strategy. The immediate impetus to the buyout came from the
pilots, who were attempting to protect high earnings in a changing indus-
try environment. Gordon estimates that the average pilot enjoyed rents
and quasi-rents with a present value of $900,000 (1999: 339, footnote 19).
For shareholders and management, the main attraction of the deal was an
ability to reduce costs on a new short-haul subsidiary (Shuttle by United).
These parties also believed that the new ownership structure would en-
hance productivity. Gordon notes that the majority equity stake for UAL’s
employees was “an unprecedented outcome in a large public corpora-
tion not at insolvency’s door” (1999: 339). He also notes that despite the
modest equity capital from employees (most of the labor contribution
took the form of wage cuts and promised productivity gains), “UAL has
maintained its customary relationships with external capital suppliers to
finance the transaction, airline purchases, working capital, and long-term
capital needs” (1999: 339).
A twelve-member board of directors was created, with complex rules
governing the representation of interested parties. The board consisted
of (1) three “employee directors,” including two directors chosen by the
leadership of the pilots’ and machinists’ unions, and a third by a com-
mittee of salaried and managerial employees; (2) five “public directors”
elected by non-employee shareholders, three of whom have never been
employees and the other two being the CEO and the COO; and (3) four
“independent directors,” at least two of whom are senior executives or
board members of major public companies. The four independent direc-
tors are nominated by a committee consisting of the incumbent indepen-
dent directors plus two employee directors, with approval required from
a majority of independent directors and at least one union director. The
three outside public directors are nominated by the incumbent outside
public directors, and the two management public directors are chosen by
a majority of the full board (Hansmann, 1996, 326–27, footnote 79).
Several features of this structure are noteworthy. First, the directors
who represent the employees are not elected by the employees them-
selves, but instead are chosen by the leaders of the respective labor unions.
Second, the three employee directors are balanced precisely by three di-
rectors representing outside shareholders. To have a clear voting majority
on an issue, either group would have to gain the support of four other di-
rectors. Merely gaining the support of top management would not suffice.
However, the independent directors do constitute such a pivotal group,
and because they are partially self-perpetuating, they are not entirely
beholden either to the employees or the outside shareholders. Employee
directors have a strong say in the appointment of this group, while out-
side shareholders have no say, as befits the greater share ownership of
employees, but on any given issue the independent directors can coalesce
with outside shareholders and veto proposals favored by employees.
In addition to the rather byzantine board structure, there is a sys-
tem of committees to deal with more specialized issues. Perhaps most
important is the Labor Committee, which handles amendments to
collective-bargaining agreements. This committee consists of three or
more directors, including at least one outside public director, one in-
dependent director, and at least one other, but no union director. This
ensures that the unions will not be in the position of negotiating with
themselves. There is also a Compensation Committee that consists of the
CEO, a public outside director, two independent directors, and all three
labor directors, and has a voice in determining managerial salaries. The
CEO’s own salary is delegated to a Compensation Administration Com-
mittee with two independent directors and one outside public director but
no employee representatives. Finally, there is a committee to formulate
strategy for the new shuttle subsidiary (Gordon, 1999: 343).
Extraordinary matters such as mergers and acquisitions, entry into a
new business, asset sales, and sales of securities that would dilute labor’s
ownership share require a 75 percent majority on the board, including
one union director, or a 75 percent vote at a shareholder meeting where
employees would hold 55 percent of the votes. Thus, in practice, the three
labor directors have a veto over such matters. Employee stock is held by
an ESOP, and the trustee is instructed to vote that stock as directed by the
employees, subject to the complication that Department of Labor rules
require the trustee to exercise independent judgment. Although employ-
ees do not directly elect directors of the firm, their voting rights could be
useful in cases where the board divides equally, or for amendments to the
firm’s charter (Gordon, 1999: 344–45).
Although the employee buyout is clearly intended as a long-term ar-
rangement rather than a transitional scheme, some problems are foresee-
able for the future. First, as workers retire, their preferred stock will be
converted into common stock, which they can sell. This will alter the rel-
ative voting power of various groups over time. Second, employees who
are hired after the initial six-year period of stock accrual will own little if
any stock in the firm. The ESOP has been constructed so that labor as a
group will retain 55 percent of the voting rights in the firm, and influence
over nine of the twelve directors, until employee ownership falls below
20 percent. This is not expected to happen until about 2016 at the earliest.
But well before that date, voting rights will be exercised by union lead-
ers despite the fact that most union members will own no stock. Gordon
comments: “This could be a prescription for disaster, since the employ-
ees’ economic interest as shareholders is an important check on the use of
governance power for distributive purposes” (1999: 346). One potential
solution that may be adopted in the future is to devise some way for new
employees to purchase shares. On the positive side, United outperformed
its major competitors between 1994 and 1997. There is no evidence that
outside shareholders were hurt by the buyout, and there is some evidence
that labor productivity improved: Sick time, worker compensation claims,
disability claims, and the filing of grievances all declined (1999: 348).
What can be said at a theoretical level about the reasons for such
employee takeovers? There is substantial unanimity on a number of
points. First, information asymmetries between employees and manage-
ment matter – for example, management usually knows more about the
firm’s future profitability than do the employees. Gordon argues that such
asymmetries interfere with the efficient renegotiation of fixed contractual
claims when a firm encounters financial distress, or more generally a tur-
bulent economic environment: “Equity-for-concessions transactions may
conserve value in firms facing economic transitions by providing a mecha-
nism that promotes agreement rather than a deadlock in the renegotiation
of the employee share of cash flow” (1999: 325).
A formal model along these lines has been developed by Ben-Ner and
Jun (1996), who consider a non-cooperative bargaining game between
management and a labor union in which management is better informed
about the financial state of the firm. They show that in equilibrium, the
union presents management with a choice: Pay a high wage, or sell the firm
to the employees at a low price. In firms that are highly profitable, man-
agers are more likely to accept the high wage demand (selling a profitable
firm for a low price is even less attractive), while in firms that are less
profitable, the employee buyout option is more likely to be chosen (such a
firm cannot afford to pay a high wage). Ben-Ner and Jun argue that their
framework is consistent with the fact that employee buyouts are more
common among financially troubled firms. Kovenock and Sparks (1990)
similarly show that when managers can offer workers both a straight wage
and an equity claim through an ESOP, larger equity claims are generally
offered when profits are low.
Another theme involves the inability of shareholders to make credi-
ble commitments to their employees about future managerial decisions.
Employees are often concerned that demands for concessions, if accepted,
will be followed by more demands tomorrow. One way to avoid this prob-
lem is to give workers control rights in the firm. For example, the promise
of job security extended to UAL employees was enforced by substantial
employee influence on the board of directors and at shareholders’ meet-
ings. A related aspect of the credibility problem is that employees need
to be sure they will capture a share of the future returns made possible
by their present sacrifices. An equity claim has attractive features in this
way because workers obtain payments that are strictly proportional to the
payments made to outside shareholders, and workers can take advantage
of public information about stock prices. In theory, it might be possible to
write other kinds of profit-sharing contracts to replicate this result, but
such contracts could be open to manipulation and would require special-
ized monitoring and enforcement efforts.
Given their advantages in coping with informational asymmetries and
guaranteeing that promises will be kept, why are employee takeovers not
more common? A full answer will have to await Chapter 11, but some
remarks can be made here. First, collective bargaining is a substitute for
employee share ownership. The Ben-Ner and Jun model shows that wage
bargaining will be used when the firm’s financial condition is strong, so it
is only weaker capitalist firms that are susceptible to employee buyout.
Browning (1987) similarly shows that ailing firms may be bought out by
risk-averse workers if current wages are rigid and the firm cannot con-
dition future wages on the state of the world. Viable firms, however, are
never bought out if profits and reservation wages of workers are posi-
tively correlated over the business cycle. Moene and Wallerstein (1993)
argue that collective bargaining will usually create monopoly rents, while
formation of an LMF will not, so again workers may prefer union wage
negotiations in relatively prosperous industries.
Second, the problem of credible commitment cuts two ways. If workers
are to run the firm, they will have to find some way to reassure sharehold-
ers that promised dividends will actually be paid, or else do without access
to outside equity capital. Similar problems arise in less-severe form for
debt capital, where there is a danger that workers may find it attractive
to default on their obligations to lenders. In either case, the inability of
workers to reassure investors may create serious problems in financing
a takeover. It is instructive in this respect to consider the safeguards de-
signed in the UAL case to prevent workers from raising their wages at
the expense of dividends. Likewise, workers in the Algoma case did not
get a majority of board seats despite owning a majority of shares, perhaps
because the outside shareholders feared that wages would be raised at
the expense of dividends. Any difficulties in financing a takeover will re-

inforce the tendency for buyouts to involve firms in poor financial health
because the market price of such firms is lower and employees need to
contribute less of their own wealth. If employees have large quasi-rents
at stake, the benefits from forestalling complete failure of the firm can be
substantial relative to the cost of acquiring shares that have a low market
value to outsiders.
Third, major collective-action problems arise in the course of an em-
ployee buyout. The benefits of a takeover are diffused widely throughout
the workforce, while the cost to a small group of employees of organiz-
ing a takeover effort is high. There is no easy way for such a group of
employee-entrepreneurs to privately appropriate the social benefits from
the transition. It is noteworthy that both the Algoma and UAL cases
involved unionized firms. In each case, union leaders represented work-
ers as a group and advanced coherent demands about the terms of a
buyout package. It is hard to see how such transitions can occur at all
unless workers first have some collective organization to speak on their
behalf.
Fourth, management is not the only party with private information.
Management does not know precisely what employees are willing to
pay for control rights over the firm. Moreover, an individual employee
may not know what other employees are willing to pay. A worker’s per-
sonal valuation of a takeover depends on private information about her
own alternatives in the labor market, her preferences about firm poli-
cies, and her beliefs about the preferences of other workers (these beliefs
are important in forecasting the policies to be adopted after a takeover).
This can lead to difficult bargaining problems among employees over
the manner in which buyout costs will be shared, especially in a firm
where there are several distinct groups of employees having divergent
interests.
To further complicate matters, Skillman and Dow (2002, unpublished
manuscript) show that even if there is no information asymmetry, an em-
ployee buyout will not succeed unless the takeover group has sufficiently
uniform preferences about firm policies (Section 9.7). Direct evidence that
employee preferences are sometimes not uniform, even when a plant is
about to close, has been provided by Browning and Lewchuk (1990). They
show for a failing firm in Canada that slightly more than half of the produc-
tion workers in their sample favored establishment of a workers’ coopera-
tive rather than a takeover by outside investors. Among those supporting
a cooperative, job security was the dominant motive. These respondents
were more experienced, and thus had more firm-specific human capital
at stake, than those who supported the alternative of investors’ control.
In recent empirical work, Chaplinsky, Niehaus, and Van de Gucht
(1998) examine the determinants of employee buyouts in the United
States. These authors compare buyouts having broad employee partici-
pation (EBOs) against management buyouts without such employee par-
ticipation (MBOs), where in each case a publicly traded firm or one of its
subsidiaries is taken private in the transaction. Their main conclusions are
that employee involvement in a buyout is more likely, other things being
equal, if the firm has (1) worse pre-buyout stock performance; (2) more
excess pension assets; (3) less long-term debt; and (4) a lower asset value
per employee. Result 1 is consistent with the fact that employee buyouts
are more common in distressed firms. Results 2, 3, and 4 are linked to
the ability of employees to finance a takeover. Excess pension assets can
be converted into employee equity, added debt is generally needed to
finance an ESOP loan, and a lower value of assets per employee reduces
financing requirements per worker.
It is also possible to view these empirical findings in a more strategic
light. Excess assets in a pension plan leave workers vulnerable to oppor-
tunistic behavior by managers or shareholders. Furthermore, poor stock
performance is correlated with a shorter expected lifespan for the firm,
which tends to weaken reputational constraints. An employee buyout can
therefore be interpreted as a safeguard to defend the interests of workers
in an endgame scenario by removing the opportunity for other players to
cheat on promises to employees. Under this hypothesis, the probability
of an employee buyout increases with the amount of employee wealth
that could be appropriated by others, and the likelihood that appropria-
tion will occur (for further remarks along the same lines, see Chaplinsky,
Niehaus, and Van de Gucht, 1998: 294).
10.4 Degeneration
An old theme in the literature on LMFs is that these firms will tend to “de-
generate” into capitalist firms, even if (perhaps especially if) they achieve
some measure of financial success. There are two strands in this story,
the first focusing on the tendency of successful LMFs to hire increasing
numbers of wage laborers and the second stressing direct takeovers by
investors. I take up the first idea here and the second in Section 10.5.
It is not hard to see why an LMF considering expansion might decide
to hire a non-member worker at a competitive wage, rather than bringing
this worker in as a full-fledged member. If all members share equally in

the firm’s net income, a new member would have to receive a payment
sufficient to cover the wage available on the external labor market (the
worker’s opportunity cost of joining), plus a share of firm profit. A hired
hand, however, only receives the external wage, leaving a larger net in-
come for the current incumbents to share among themselves if the firm
is profitable. The argument is identical when one of the current members
leaves and the remaining incumbents must decide how best to replace
this worker: If the firm is profitable, a hired hand is preferred. Since all
firms experience labor turnover through resignations and retirements, in
the long run the initial membership group will dwindle and hired hands
will increase. Ultimately, only a few founders will be left, and the firm will
become a KMF in all but name (Ben-Ner, 1984, 1988b).
Miyazaki (1984) formalizes the degeneration idea by comparing LMFs
and KMFs within a model where workers are risk-averse and the compet-
itive spot wage varies with the state of the world. Assuming the LMF can
temporarily hire non-member workers, pool the earnings of members who
work inside and outside the firm in order to compensate laid-off mem-
bers, and reallocate income across states of the world through a perfect
capital market, Miyazaki shows that the LMF degenerates into a KMF
when the KMF would have positive profit. But if the KMF has negative
profit, there is an incentive to transform it into an LMF because workers
can use the LMF to smooth their incomes relative to the spot labor mar-
ket, even though their average incomes are lower than what they would
get as hired employees.
Miyazaki also investigates the consequences of relaxing the preceding
assumptions, especially the payment of compensation to laid-off mem-
bers by the LMF and the existence of a perfect capital market. His results
tie together a key idea from Section 10.3 (low profitability leads to the
conversion of KMFs into LMFs) with the idea of degeneration in this
section (high profitability leads to the conversion of LMFs into KMFs).
It is unclear in the Miyazaki analysis why KMFs cannot supply equally
good income-smoothing services to risk-averse workers through implicit
contracts, although one suspects that a KMF would have difficulty com-
mitting itself to honor such promises.
The most comprehensive theoretical discussion of LMF life cycles is
that of Ben-Ner (1988b). Assuming an absence of tradeable membership
rights, Ben-Ner notes that an increase in profitability leads an LMF to
expand through hired labor up to the same size as a similar KMF. Members
are always better off hiring wage labor than adding new members, and
if they do this, the maximization of income per incumbent member is

equivalent to the maximization of total profit. Although a majority of
LMF incumbents might be tempted to replace some existing members
with wage laborers, this is unlikely since “most reasonable lay-off rules
that members would adopt (such as: members have equal probability of
being laid off, or laid-off members are compensated so as to be indifferent
between staying and leaving) will freeze membership at its initial level”
(1988b: 299, footnote 12).
Ben-Ner’s prediction is that a gradual reduction in membership
through retirements and quits will occur until the LMF is fully trans-
formed into a KMF. One countervailing factor is that productivity might
fall within the LMF if hired labor is used. This may lead to lower effort
incentives, conflicts between members and non-members, and the like.
If so, the LMF may prefer to limit the ratio of hired workers relative to
members in order to preserve its productivity advantages. Ben-Ner (1984)
shows that this mechanism could enable LMFs without hired hands to sur-
vive in marginal industries where capitalist firms are unprofitable. Some
corroborating evidence comes from Russell’s conclusion that scavenging
and taxi cooperatives arise as a result of the exceptional difficulty of mon-
itoring labor effort in these activities, and that this same factor tends to
prevent the coops from using hired labor (1985a: 179; 1985b).
A membership market of the sort described in Section 7.3 can prevent
degeneration, at least in principle. If new members must pay a fee to the
current members in exchange for admission into the firm, and this fee
reflects the value of the profit share awarded to the newcomer, as it will
in a competitive equilibrium, then the current members are completely
compensated and have no reason to prefer hired labor over the admission
of new members. Thus theories that assert the inevitability of degener-
ation must assume that membership markets are imperfect, perhaps for
one or more of the reasons given in Section 7.5.
The evidence on degeneration is mixed. One striking example is
Olympia Veneer, the first of the plywood cooperatives (Section 3.2).
Derek Jones (1982) cites various other U.S. cooperatives from the nine-
teenth and early twentieth centuries that either transferred control
rights to outside capital suppliers over time or made increasing use
of non-member labor. Craig and Pencavel (1992) find a statistically
significant trend toward greater use of non-member labor in the ply-
wood coops between 1958 and 1982, although some individual firms
moved in the opposite direction. These coops have rarely, if ever, ex-
panded by selling new membership shares rather than using hired labor
(Berman, 1982: 84–85). Other U.S. cases of degeneration include scav-

enging and taxi cooperatives (Russell, 1985a, b).
Hired workers have historically made up a large fraction of the
workforce in many European LMFs (sometimes more than 70 percent;
Ben-Ner, 1988a: 18). Interestingly, however, legislative changes and the
policies of a national federation raised the average membership propor-
tion in French cooperatives from 36 percent in 1960 to 60 percent in 1984
(Pérotin, 1987). In a study of all 283 French producer cooperatives that ex-
isted continuously between 1970 and 1979, as well as a cross-section of the
541 coops existing in 1979, Estrin and Jones (1992) did not find steady de-
generation as measured by the fraction of non-member workers. Instead
they found a U-shaped pattern in which the membership share initially
decreased with age for forty to forty-five years, after which it began to
rise. The evidence for degeneration was strongest for unprofitable firms,
contradicting theoretical predictions that degeneration should occur only
when an LMF is profitable. Estrin and Jones also found that the share of
members in the workforce was positively and significantly correlated with
labor productivity, which could account for the fact that more profitable
LMFs seemed to degenerate the least. They concluded strongly that “the
contention of many theorists that employee-owned firms will necessarily
degenerate in a capitalist environment is false” (1992: 336).
An explanation for this result offered by Estrin and Jones (1995) is that
the standard institutional assumptions behind the degeneration hypoth-
esis do not apply to French coops. In particular, insiders cannot exclude
hired workers from becoming full members if they so desire. To join, a
hired worker only needs to make a minimal capital contribution and must
then be accepted. This prevents members of more profitable coops from
expanding solely through hired labor. Moreover, any profit-sharing bonus
distributed to members must also be paid to non-members, reducing the
incentive to use non-member labor in the first place.
10.5 Investor Takeovers

LMFs can be converted into KMFs if the members decide to sell out
to investors. This process is often placed under the broad rubric of “de-
generation,” but the two kinds of transition need to be studied sepa-
rately. There is no doubt that LMFs have often sold out to investors
who then converted the firms into KMFs. For example, in the period
1970–1980, five plywood cooperatives were sold off to private investors
(Gunn, 1984a: 101). Russell (1985a: ch. 3, 1985b) likewise remarks that
10.5 Investor Takeovers 225
several scavenging cooperatives were sold to outsiders in the early 1970s.

Many professional partnerships in the United States have also shifted to
investor control in recent decades (Section 3.1).
Casual discussions of investor takeover, particularly in the context of
the plywood coops, often attribute the phenomenon to high prices for
membership shares coupled with the inability of workers to afford such
high prices. This general story is uncompelling. If it is hard to find buy-
ers, why is the price high? If workers could afford membership rights
at an earlier date, why has it become impossible for other workers to
buy in now? The first question might be answered by asserting that the
price reflects what outside investors are willing to pay, not what workers
are willing to pay, but the prices cited in many accounts refer to trans-
actions between incoming and outgoing workers. The second question
could be answered by arguing that collectively owned assets have grown
in value, perhaps as a result of technological change. In this case, the
firm might buy back the share of a retiring member, replace it with debt
supported by the additional firm assets, bring equity capital per worker
back to its original level, and reduce the membership price accordingly
(Hansmann, 1996).
Several other factors might lead to an investor takeover. Two given by
Hansmann (1996: 83–84) are a reversal of a mistaken decision to enter an
industry for which the LMF was poorly suited, and a return to KMF status
in cases where the LMF was a transitional device to salvage a troubled
KMF. Miyazaki (1984) argues that successful LMFs may be converted
into KMFs if their insurance features are no longer needed by workers
because firm profitability has improved. Another explanation is that of
Skillman and Dow (2002, unpublished manuscript), who show that for
collective choice reasons it is often possible for wealth-maximizing in-
vestors to design a takeover offer that 51 percent of the LMF members
will accept (see Section 9.7). In this analysis, the firm’s vulnerability to
takeover grows as most members approach retirement age and become
more interested in wealth maximization than working conditions or other
consumption benefits derived from firm membership.
Investor takeovers of LMFs may be nothing to worry about. After
all, the members sell out voluntarily and obtain a price they find accept-
able. If workers’ control is no longer attractive, why should workers be
obliged to keep a form of organization they do not want? One qualifi-
cation to this reasoning arises from the Skillman-Dow model just cited,
where investor takeovers usually inflict negative externalities on the mi-
nority of members who are opposed to selling out. Another caveat is that
when membership markets are imperfect, it can be in the unanimous in-

terest of current members to sell out even though the firm would generate
a larger social surplus in present value terms as an LMF (Dow, 1993a).
Thus the normative assessment of such transitions is less straightforward
than it may appear.
10.6 Survival Rates

Another demographic factor of interest concerns the survival proper-
ties of LMFs and KMFs. Ben-Ner (1988a: 14–15) has calculated annual
death rates for European LMFs during the 1970s and early 1980s. In-
cluding deaths from all sources, both conversion into KMFs and out-
right dissolution, the average rates are 6.9 percent (France), 28.6 percent
(Netherlands), 9.3 percent (Italy), 6.3 percent (UK), and 29.5 percent
(Sweden). Ignoring the Netherlands and Sweden, whose LMF sectors
were much smaller and newer, these rates compare favorably with the
10 percent death rate for KMFs in French manufacturing over a similar
period, and the 10.5 percent rate for all industries in the U.K. The KMF
figures include ownership transformations, which do not imply financial
failure. However, the LMF figures include investor takeovers, which like-
wise do not imply failure in the sense of bankruptcy or dissolution.
Controlling for the ages of firms and comparing death rates for KMFs
and LMFs in the U.K. for 1974–86, Ben-Ner (1988a: 18–19) finds that
the hazard rate for each type of firm peaks at about three years after
formation. The KMF hazard rate is at least three times higher at all ages.
This result does not control for size differences, but given the smaller
average size of LMFs and the fact that small firms are more likely to fail,
adjusting for size effects would probably tilt the balance further toward
an LMF survival advantage.
Pérotin (1997: ch. 7) finds that French workers’ cooperatives had exit
rates that were sometimes higher and sometimes lower than the rate
for French firms in general. But averaging over the period 1979–91, the
annual exit rate for cooperatives was 11 percent and that for all other
firms was 12 percent. Cornforth and Thomas (1994) report a death rate
for U.K. cooperatives of 16 percent per year in the recession of 1989–
91, and a rate between 10 and 16 percent for the period 1982–86.
They assert that these rates are similar to the rates for all small firms.
However, the cooperatives in their sample were especially small, with a
majority having fewer than five full-time workers. This may tend to bias
the LMF failure rate upward.
10.6 Survival Rates 227
Staber (1989) has examined failure rates for the universe of workers’
cooperatives in Atlantic Canada (specifically, New Brunswick, Nova
Scotia, and Prince Edward Island) between 1940 and 1987. In this group
of 205 coops, the highest hazard rate for death occurs in the age in-
terval of three to six years, and this is true for all economic sectors:
resources, manufacturing, construction, transportation and communica-
tion, and services. Comparing his results with data from Illinois and Ohio
on conventional retail firms, business consultancies, machine shops, and
restaurants, Staber finds much higher failure rates for these KMFs, es-
pecially in the early years of a firm’s existence. Within the LMF group,
resource cooperatives had the highest failure rates and service coopera-
tives the lowest.
The general conclusion one can draw from these results is that LMFs
are not rare because they fail disproportionately often. Once created,
they appear robust. Rather, they are rare because in absolute numbers
they are created much less often than KMFs, and to a lesser degree be-
cause successful LMFs may be transformed into KMFs (a similar point is
made by Hodgson, 1996). The flow from LMF to KMF is at least partly
offset by a flow of unsuccessful KMFs into the LMF column through em-
ployee buyouts. However, these buyouts are not common enough to alter
the relative frequency of the two firm types in a dramatic way.
A rather different kind of evidence on firm survival also deserves men-
tion. Blair, Kruse and Blasi (2000) exhaustively searched available data
sources to compile information on all publicly traded U.S. corporations
that in 1983 had 20 percent or more of their common stock in ESOPs,
profit-sharing, or other employee benefit plans (for the most part these
are not LMFs under the definition given in Section 5.3). They identified
twenty-seven such firms. They then compared this group with a control
sample matched by size and industry in an attempt to detect differences
in firm survival rates between 1983 and 1997.
Firms in either group that failed to survive could exit in various ways:
liquidation or bankruptcy; acquisition by another firm; or being taken
private in some form of buyout, not necessarily a buyout by employees.
Blair, Kruse, and Blasi found that the firms with employee ownership
(EO) had the stronger survival record. “Having substantial employee
share ownership in 1983 resulted in a statistically significant reduction in
the annual probability that a firm would disappear in each year between
1983 and 1997, given that the firm had survived up to that point. The
statistical significance of this finding is weakened – but is nonetheless still
significant at the p = .10 level – in the models that control for firm size
and capital intensity at the beginning of the period” (2000: 268). Firms in
the two groups differed in their mode of exit from the sample. A higher
percentage of EO firms left through merger and acquisition, but far fewer
were taken private.
Blair, Kruse, and Blasi summarize their conclusions as follows: “Our
study does not give us any reason to believe that organizational forms in
which a substantial block of (otherwise publicly traded) common stock
is held by or for employees is an inefficient, unstable, contentious, or
otherwise problematic arrangement, or that it is an arrangement used only
to facilitate major transitions. In fact, such arrangements seem at least as
stable as any other ownership arrangement and may well help to ‘stabilize’
a firm, making it more resistant to bankruptcy and unwanted takeovers
and somewhat less prone to labor strife and wrenching downsizings. We
also find no evidence that these benefits come at a cost in operating or
financial performance” (2000: 287–88).
In some related research, Chaplinsky, Niehaus, and Van de Gucht
(1998) examined a sample of employee buyout (EBO) and managerial
buyout (MBO) firms from 1980 to 1990 to discover whether there were
differences in outcomes depending on the level of employee participa-
tion in the transaction. By 1993, 14.8 percent of the MBOs had gone
into bankruptcy, 16.2 percent had been acquired by another firm, and
23.2 percent had reverted to their earlier publicly traded status, yielding a
“survival” rate of 45.8 percent. The figures for the EBOs were 12.5 percent
in bankruptcy, 18.8 percent acquired, and 18.8 percent public, for a sur-
vival rate of 50 percent. The similarity of these numbers led the authors
to conclude, “The evidence on survivorship . . . suggests that EBOs are a
viable organizational structure for certain firms” (1998: 317).
10.7 Business Cycles

There has been considerable debate about patterns of LMF creation and
destruction over the business cycle. I briefly review the various claims
that have been made and what evidence is available. This evidence is
relevant in assessing the story about macroeconomic externalities from
Section 6.7, but it is also of independent interest.
Ben-Ner stresses the strong growth of the LMF sector beginning in
the mid-1970s, “when economic stagnation accompanied by rising unem-
ployment rates set in, while capitalist sectors experienced slow growth
or even decline” (1988b: 310). Estrin (1989) argues that LMFs resist
closure in bad times, but that in an upswing they are inclined to hire
10.7 Business Cycles 229
wage labor rather than share the increased income from prosperity with
new members. According to Estrin, this accelerates the transformation
of LMFs into KMFs as predicted by the degeneration hypothesis from
Section 10.4.
Bonin, Jones, and Putterman identify effects running in two directions.
Economic growth is likely to be beneficial for LMFs because it gives
workers more income and thus a greater ability to accumulate personal
wealth that could be used to finance a firm. For the same reason, it is
likely to reduce worker risk aversion, and it will increase the demand
for participation in firm governance if this is a normal good. However,
recession increases the cost of job search and relocation, which they assert
will make LMFs more attractive (1993: 1312–13). The net effect is unclear,
but it seems reasonable to think that the positive effects of growth on
LMF formation are likely to be imperceptible over the shorter intervals
of the business cycle. Therefore, in the medium run, one might expect
both de novo LMF creation and employee buyouts to increase during
recessions.
Efforts to detect a cyclical pattern in LMF formation have been largely
unsuccessful. After excluding the Great Depression, when governmental
programs to create cooperatives were established, Conte and Jones (1985)
found no significant correlation between LMF formation in the U.S. and
the unemployment rate between 1820 and 1960. The results were also
unencouraging when the unemployment rate was replaced by cyclical
movements in an index of industrial production, a measure of strike ac-
tivity, or growth of craft-union membership. The picture is instead one
of slow but steady LMF formation, punctuated by exceptional years in
which many LMFs were created in short bursts.
In a study encompassing all of the 221 producers’ cooperatives existing
in Atlantic Canada during the period 1900–87, Staber (1993) found that
neither a dummy variable for recession nor the unemployment rate had
a significant effect on formations or dissolutions. However, most of these
were very small firms, with an average size of six to seven members, where
the members were often part-time or volunteer workers who routinely
supplemented their coop earnings with other jobs.
The only major study to find a significant connection between LMF
birth and death rates and the business cycle is Pérotin (1997: chs. 8–9).
In her data set, which consisted of French cooperatives for 1971–92,
the birth of cooperatives was countercyclical, increasing as a function of
the previous year’s unemployment rate. Growth in GDP had a negative
effect on formations. On the other hand, Pérotin found that GDP growth
reduced the death rate of cooperatives, while unemployment increased

their rate of closure.
10.8 Clusters
A pattern that emerges from the historical and case-study literature
on traditional workers’ cooperatives is their tendency to congregate in
certain industries, regions, and time periods. The reasons for cluster
formation are varied but include at least the following: (1) imitation of
successful firms; (2) tangible support from related firms, such as exper-
tise and financial assistance; (3) responses to parallel problems afflicting
an industry or region; and (4) entrepreneurial activities by cooperative
federations, labor unions, or governments.
Jones (1979, 1984) finds that past or present activities for U.S. cooper-
atives have included foundries, barrel-making, shingle-making, plywood,
reforestation, and refuse collection. Foundries were important between
1860 and 1890 in the industrial Northeast and Midwest; barrel-making
was prominent between 1870 and 1900 in Minnesota; the shingle-making
coops were formed during the 1910s on the Pacific coast; the plywood
and reforestation coops were also formed on the West coast in the 1950s
and 1970s, respectively; and the sanitation coops are a California phe-
nomenon. Jones identifies three further clusters not linked by product
similarities: a group sponsored by the Knights of Labor in the 1880s, and
two waves from 1860–1890 and 1920–1940 (the latter included “self-help”
cooperatives established during the Great Depression with direct govern-
mental assistance; see Jones and Schneider, 1984). Conte and Jones (1985)
observe that a great many standard industrial classifications (SIC codes)
include few or no coops, confirming that LMFs have arisen in a narrow
range of industries (for a contrary view, however, see Aldrich and Stern,
1983).
Because the U.S. reforestation cooperatives have not been discussed
in detail, it may be useful to sketch their history here (see Gunn, 1984a:
ch. 3, and 1984b). At the time of Gunn’s writing, this group included
twenty-three firms in Oregon, Washington, Montana, Idaho, and north-
ern California, with a total workforce of about 800. These firms worked
on a contract basis, mostly on public forest lands, and their key activity
was to replant trees after timber harvesting. They were labor-intensive,
owned few physical assets, were often small, and raised much of their
working capital through membership fees. Of the thirty-five coops cre-
ated between 1970 and 1984, twenty-three were operating in 1984 and
10.8 Clusters 231
twelve had dissolved. Changes in government forest policy and competi-

tion from immigrant labor eventually eliminated most of the coops, and
only one or two survive today (Pencavel, 2001).
Non-U.S. clusters include traditional British cooperatives in printing
and footwear. The first British footwear cooperative, Northamptonshire
Productive, was formed in 1881, and the last, Staffordshire Shoe, in 1929.
This group, geographically concentrated in the English Midlands, peaked
in number (fifteen to twenty) and membership (5,000) in the first quarter
of the last century, and slowly declined through 1970 (Backus and Jones,
1977).
Two common elements can be identified for many of the clusters iden-
tified here. First, there was typically a template firm whose organizational
structure was imitated when other coops were created in the same indus-
try. The Cooperative Barrel Company played this role for the Minnesota
cooperages, and Olympia Veneer did likewise for the plywood coops
(Jones, 1979, 1984). Among reforestation coops, the template firm was
Hoedads (Gunn, 1984a, b). Scavenger cooperatives in several California
cities were modeled after the early San Francisco coops (Russell, 1985a:
177).
A second recurring theme is that clusters of related firms often had
access to a pre-existing industry association that provided marketing or
other services, or rapidly created such an association themselves. Most
of the shingle-making cooperatives belonged to the Mutual Lumber
Mills, which functioned as a central marketing agency (Jones, 1979). The
plywood cooperatives belonged to a trade association that developed
product standards, advertised, provided technical support, and sponsored
research (Section 3.2). The forest cooperatives created their own federa-
tion, the Northwest Forest Workers’ Association.
Some clusters have emerged through the entrepreneurial activities of
a trade union, government, or other body. For example, Quarter and
Brown (1992) observe a clustering pattern for worker buyouts in Canada,
where there were thirty-nine such firms in operation in June 1990. Fifteen
were located in Quebec, including five ambulance services, and fourteen
were in British Columbia, including eight in highway services and three
in forestry. There were no worker buyout firms at that time in southern
Ontario, the heartland of Canadian industry where most plant closures
take place. In Quebec, nine of the fifteen buyouts were organized during
a two-year interval by a consulting group affiliated with a large labor con-
federation. In British Columbia, the highway cooperatives arose through
a privatization program under which the provincial government assisted
employees in buying firms. For North America as a whole, Quarter and

Brown stress the role of the United Steelworkers of America, which as-
sisted with employee buyouts of nine plants between 1985 and 1992.
The most obvious European cluster is the Mondragon system, which
evolved in a narrowly bounded geographical region of Spain (Section 3.3).
Mondragon is also in a sense its own federation because it has organized
the supply of banking, education, and technical support services within the
group. In Italy, many of the roughly 1,000 worker buyouts that occurred
in the 1980s were in Emilia Romagna where the Lega federation is strong
(Quarter and Brown, 1992: 105; on the Lega, see Section 4.1). In France,
there were fifty to a hundred worker buyouts per year in the 1980s, many
of these supported by CGSCOP, a federation of workers’ cooperatives
(Quarter and Brown, 1992: 106).
The role of template firms is readily explained by learning. Once it be-
comes known that an LMF has been successful in some industry, others
are more inclined to replicate this organizational form. This leads to spa-
tial and temporal clustering partly because industries themselves are geo-
graphically concentrated around raw materials or other resources, partly
because a region may face common economic problems to which LMFs
offer a solution (such as plant shutdowns), and partly because dense in-
formation networks develop locally through word of mouth, community
organizations, and media outlets.
These factors may be especially important for an organizational form
that is largely unfamiliar. Ben-Ner (1988b) asserts that the impediments
to LMF formation are increased by the rarity of LMFs themselves. Specif-
ically, he cites a scarcity of legal, organizational, and financial expertise,
and the tendency for capital suppliers to view LMFs as risky because
of their unfamiliarity. Bonin, Jones, and Putterman (1993: 1313) simi-
larly note that positive informational externalities may help overcome
the wealth obstacles to LMF creation, and may therefore generate waves
of LMF formation. They also suggest that the LMF sectors in Italy and
France gain from the presence of supporting organizations that dissemi-
nate information and reduce the risks perceived by lenders. On a smaller
scale, a reader of the literature on the plywood cooperatives is likely to
be impressed by their gradual acceptance among lenders as they became
a normal feature of the local landscape.
Related LMFs are often linked together by tangible forms of mutual
support that include personnel transfers, financial assistance, and tech-
nical advice. Gunn, for example, reports that other reforestation coop-
eratives were often started by workers who participated in the template
10.8 Clusters 233
firm Hoedads. He also observes that the Denver Yellow Cab Cooperative
Association provided advice, copies of its bylaws, and the time of one of
its founders to assist Capital City Co-op Cab in Sacramento, and that it
provided similar advice to a group of cab drivers in Minneapolis (1984a:
166–68; Russell 1985a: 177). The systematic creation of new LMFs de novo
and through reorganization of conventional firms by the Mondragon and
Lega groups is an analogous process, albeit on a vastly larger scale.
11
Toward a Synthesis
11.1 The Causal Tapestry

The time has now come to gather up the threads from the last few chapters
and weave an answer to the question, “Why is workers’ control so rare?”
As explained in Section 6.1, the challenge is to identify a physical or
institutional asymmetry between labor and capital that accounts for actual
asymmetries in the control rights held by the suppliers of these inputs. I
will argue that the symmetry between capital and labor is broken by the
fact that capital is alienable, while labor is not.
Ownership rights over productive non-human assets, ranging from
agricultural land and office space to machines and computer software,
are easily transferred from one person or group to another. The same
is true for financial wealth including cash, bonds, or claims on the net
incomes of firms. The bundle of ownership rights in this context is the
standard one: the right to decide how an asset will be used, the right to
derive income from its use, and, crucially, the right to transfer the first
two rights to other people. Nothing rules out the collective ownership
of non-human assets provided that there is a well-defined collective
choice procedure through which the owners exercise the rights just
described.
The capacity of a person to supply labor services, together with that
person’s skills, talents, experiences, and other aspects of human capital,
cannot be transferred to another person because these attributes are in-
tegral to personhood itself. It may be said that people own themselves
because they decide how to use their human capital and can derive in-
come from it, but they do not have a legally recognized right to transfer
234
these rights to someone else. Nor does it make physical sense to suggest
this.
Contemporary Western democracies reinforce the physical inalienabil-
ity of labor by limiting the contractual devices that might give one person
the power to command someone else’s labor services. Workers normally
retain the right to quit a job despite promises they may have given to the
contrary, and transparent attempts by employers to raise the costs of exit
from a job are routinely voided by the courts. However, an employer can
legally use a system in which wages rise with seniority, which discourages
experienced workers from quitting. In general schemes that tie workers
to firms through carrots pass legal muster, while schemes that achieve
similar results through sticks do not.
To say that labor is inalienable is deceptively simple. Many impli-
cations of this fact could conceivably prove relevant for the allocation
of control rights in firms. It is therefore necessary to sort through the
various consequences of labor’s inalienability and focus on a small sub-
set that have maximum explanatory power. Moreover, it is likely that
some of these consequences will favor the KMF while others favor the
LMF. The task is to discern whether certain effects of inalienability vary
in ways that match up with the governance patterns observed in the
world.
Another complication is that inalienability of labor, while funda-
mental, cannot do the explanatory job all by itself. Numerous auxiliary
assumptions about the nature of the world are also needed. Using the
standard economic classification of exogenous variables in terms of pref-
erences, technology, endowments, and institutions, the major assumptions
run as follows. People have diverse preferences toward income, effort,
risk, the time path of consumption, and working conditions. Many in-
dustries display at least an initial range of increasing returns to scale,
perhaps also with economies of scope; many have at least a moderate
degree of capital intensity; and some require specialized physical as-
sets. There are often productivity gains from a division of labor within
the firm, and from employment of workers with varied skills. Endow-
ments of both physical assets and financial wealth are distributed un-
equally. Labor endowments vary both in the nature of the skills possessed
by workers and in their productivity. Legally enforceable contracts are
generally costly and incomplete; most of the exchanges between capi-
tal and labor suppliers are enforced by the parties themselves through
repeated interactions or market reputation; and people frequently have
private information about the quality of assets, the quality of skills, or
236 Toward a Synthesis
the profitability of projects. These background conditions are of course

variable, and their intensity will modulate the strength of the main causal
forces.
The broad outlines of the story can now be presented. I will ar-
gue that differences in alienability affect control rights in firms through
three fundamental channels. First, there are commitment asymmetries,
involving the ability of firm controllers to extend credible commitments
to the suppliers of non-controlling inputs. Second, there are compo-
sition asymmetries, involving disparities in the characteristics of can-
didate control groups such as their size, the information available to
their members, and the degree of heterogeneity in the preferences of
members. Third, there are commodification asymmetries involving the
ease or difficulty with which control positions can be made tradeable
on markets. I do not claim that these are the only relevant causal fac-
tors, but each accounts for a set of important facts about workers’ con-
trol that would be difficult to explain in other ways. The rest of this
introduction offers a short overview of the arguments developed in later
sections.
Commitment asymmetries
The inalienability of labor implies that investors cannot buy human capi-
tal; they can only rent it. To frame the same idea a bit differently, physical
and human capital cannot be brought under common ownership by hav-
ing a capitalist firm purchase a stock of labor services and use it up over
time as the investors see fit. Instead, investors gain authority over a flow
of labor services and in return pay a flow of wages. In principle, capital
could be owned by the firm or rented, but often there are strong reasons
to prefer asset ownership by the firm (see Sections 8.2–8.4). Generally,
then, labor is rented, while at least some capital is owned.
The alienability of physical assets matters because it implies that access
to financing is crucial to firm formation. In capitalist firms, wealthy in-
vestors simply pool their funds and acquire physical assets. To compete
effectively, LMFs must find a way of financing a similar assortment of
owned assets, but they often founder on the shoals of limited wealth and
liquidity constraints. If physical assets were, like labor, inalienable, the
KMF would lose much of its competitive edge because both firms would
need to rent assets, and LMF financing problems would not be such an
important source of comparative disadvantage.
The problem for the KMF is that investors and workers have conflict-
ing interests in the workplace, so there is a chronic temptation for each
to engage in short-run opportunism at the expense of the other. Employ-

ment relationships remain viable largely because many small decisions
are continually made on each side and it is difficult for one party to seize
a significant amount of wealth from the other without quick detection and
retaliation, perhaps through firing or quitting. Although it is hobbled by
persistent low-level conflict, a KMF typically develops social conventions
defining what range of behavior by each side will be tolerated in routine
interactions. These conventions can be maintained and enforced by the
parties themselves even though contracts are largely incomplete in a legal
sense.
In an LMF, the conflict between capital and labor suppliers is located
in a different place and takes a different form. Here, an upfront invest-
ment of capital must be provided in exchange for interest or dividend
payments in the future. Once investors have committed a stock of funds
or physical assets to the firm, the LMF may be able to rely on retained
earnings for working capital and maintenance, which eliminates any need
to ask investors for additional infusions of cash in the near term. If the
members of the LMF decide to take advantage of this situation by paying
themselves high wages, depreciating equipment rather than supplying ef-
fort, or pursuing risky projects in which investors would be left holding
the bag after a failure, the investors have limited recourse.
Giving control rights to labor facilitates credible commitments to indi-
vidual workers within the firm, often with genuine productivity benefits,
but also makes it hard to reassure outside investors that their interests
will be protected. In a parallel manner, giving control rights to capital re-
assures investors that their interests will be protected but leaves workers
exposed. This superficial symmetry breaks down because the intertempo-
ral structure of the relationship between controllers and non-controllers
differs in the two types of firm. My contention is that despite persistent
conflict over the terms and conditions of employment in the KMF, the
greater problem from the standpoint of organizational viability is usu-
ally the inability of workers to offer credible guarantees of repayment to
investors in an LMF.
This is one important reason why LMFs are rare, but it cannot be
the only reason. Some sectors of the economy are labor-intensive, lack
large economies of scale or scope, and require only easily leased generic
assets, but nevertheless are dominated by capitalist firms. Examples in-
clude restaurants, real estate agencies, small retail operations, and other
service activities. Therefore it becomes necessary to invoke other factors.
Here I consider two: the internal composition of control groups, especially
size and degree of heterogeneity, and the commodification biases that

undermine LMFs.
Composition asymmetries
Even a small firm usually obtains labor services from several workers
simultaneously, although it may have only one capital supplier. The same
point applies more dramatically when economies of scale or scope are sig-
nificant, because then many workers are needed but a sufficiently wealthy
person might be able to finance the entire firm. The connection to the
inalienability of labor is straightforward. It is not usually feasible for a
single person to satisfy a firm’s entire labor requirement because the rate
at which one individual can supply labor services is subject to natural
limits. A large labor input per unit of time can only be obtained from a
group of cooperating individuals.
Thus it is frequently true, although not invariably so, that a KMF will
have a small control group while a comparable LMF would require a
considerably larger control group. This is the standard case for KMFs
organized as closely held corporations and not listed on stock exchanges.
The relative sizes of the two prospective control groups can imply greater
collective-action problems in the LMF. These problems can take many
forms, but especially involve free-riding and the costs of multilateral
bargaining.
The implications are not uniformly unfavorable for the LMF because
large publicly traded corporations may have many more individual share-
holders than employees because of a need for large amounts of capital,
the finite wealth of individual investors, and a preference for diversified
portfolios. Furthermore, inalienability of labor implies that workers must
be physically present at the work site, and hence that they may possess in-
formation about the immediate production process that outside investors
rarely have. This could conceivably tilt the balance in favor of the LMF
by making workers better informed about the behavior of managers and
less subject to free-rider problems in monitoring this behavior.
The internal heterogeneity of a control group also matters. Investors
can be seen as having approximately parallel interests in the maximization
of profit or stock-market value. The workers within a given firm, on the
other hand, can have diverse preferences because they are unable to diver-
sify their labor across many firms simultaneously, and labor markets do not
provide sufficient sorting opportunities. Such heterogeneity in the com-
position of the LMF control group predictably leads to collective-choice
problems (see Section 9.7).
Commodification asymmetries
The final causal channel, the ease or difficulty with which control positions
can be treated as a commodity, again traces back to differences in the
alienability of capital and labor. In a publicly traded corporation, shares
of stock with voting rights attached can be transferred from one person
to another without changing the firm’s physical assets, just by passing the
first shareholder’s role in the firm’s governance structure to the second
shareholder. Implicitly this involves a transfer of the shareholder’s claim
on collectively owned assets in the event of liquidation.
The same is not true if one tries to create a market for membership
rights in an LMF. Because of the definition of an LMF and the fact of
labor’s inalienability, it is impossible to transfer control rights without si-
multaneously replacing one person’s labor inputs with those of another
person. Such markets cannot be expected to function as well as capital-
ist share markets for the reasons discussed in Section 7.5. Among the
difficulties that could result from an imperfect LMF membership market
are underinvestment because of horizon problems, a failure to expand the
workforce because of common-property problems, degeneration through
the recruitment of wage labor, and investor takeovers in cases where the
LMF would have generated a larger social surplus. These were prominent
themes in Chapters 7 and 10.
A closely related aspect of the commodification problem, one that may
be even more important than those first listed, involves systematic biases
in the appropriation of costs and benefits. Section 10.2 indicated that
KMFs have strong advantages in capturing the rents from innovation
at the firm formation stage. Section 10.3 suggested that parallel appro-
priation problems arise later in the firm life cycle and obstruct employee
buyouts of capitalist firms. The governance and productivity benefits from
a transition to workers’ control are widely diffused among the workforce,
and a small group of workers can only capture a fraction of these bene-
fits by organizing an LMF. The organizers of a buyout do, however, bear
much of the cost. No corresponding problem arises in the other direc-
tion because a few investors can capture much of the social benefit from
transforming an LMF into a KMF, although they may not internalize all
of the costs (for example, losses to future workers or to workers who
were in the minority when the takeover was approved). These issues are
a subset of the more general commodification problem because if each
candidate for LMF membership could be charged a price to become a
member, an entrepreneur could internalize the social costs and benefits
of this governance structure and more LMFs would be organized.
I believe that these are the primary factors explaining the rarity of
workers’ control. Sections 2 and 3 of this chapter consider credible com-
mitment when the ultimate control group consists of investors and work-
ers, respectively. Section 4 takes up the internal composition of control
groups, and Section 5 addresses commodification issues. Citations to the
literature are kept sparse in these sections, but I have borrowed from
other authors without hesitation when I thought they were on the right
track. Some of my major intellectual debts are highlighted in Section 6.
Section 7 closes with a few methodological reflections.
11.2 Credible Commitment toward Labor

LMFs have no evident problem with respect to labor productivity. There
are few direct comparisons with KMFs, but the majority are favorable
to the LMF. More generally, there is considerable evidence showing
that residual claims and control rights for workers are associated with
higher productivity (see the references in Section 8.7). The burden of
proof clearly rests with those who believe that LMFs are inferior on this
dimension.
Even if one is unpersuaded by econometric investigations of productiv-
ity, there is an independent line of evidence suggesting that the economic
performance of LMFs, once they have been created, is not inferior to that
of KMFs. The information on firm survival in Section 10.6 revealed that
LMFs often have lower failure rates than KMFs. A reasonable inference
is that the small size of the LMF population is more closely related to
the rarity of LMF births and the low rate at which KMFs are bought
out by their employees than to any supposed inefficiencies arising after a
worker-controlled firm already exists.
Why might LMFs have labor-productivity advantages? It is useful to
approach this question the other way around, by asking why KMFs fre-
quently have problems with labor productivity. This makes it possible to
focus more directly on the reasons why investors have trouble making
credible commitments to workers in a KMF, and on the consequences
that follow. At a general level, capitalist firms suffer from various famil-
iar pathologies. Employees are vulnerable to higher production quotas,
deteriorating working conditions, failures to deliver promised wages or
benefits, raids on pension funds, layoffs, and plant closures, for example. A
capitalist firm that could somehow commit itself in advance not to under-
cut worker interests in these ways would undoubtedly save on real costs as-
sociated with the concealment or distortion of information, absenteeism,
shirking, strikes, sabotage and other retaliatory or self-protective actions

by workers. But because of its control structure, a capitalist firm has great
trouble committing itself not to behave in these ways. Problems of this
sort are rare, if not entirely absent, in LMFs.
This section will elaborate on the nature of the productivity problems
arising in the KMF. I organize the discussion around a series of “runs”
from “very short run” (everyday life in offices and factories), to “short
run” (minor changes in technical or organizational arrangements), “long
run” (wage and employment changes in response to market conditions),
and “very long run” (endgame scenarios involving mass layoffs or plant
closures). This does not correspond with the Marshallian concept of short
and long runs commonly found in economics textbooks, but it matches
up nicely with some concepts from game theory.
The very short run has the feature that the stakes are typically small
and abuses, if they are detected, can be punished quickly. Because inter-
actions are routine, standard ideas about repeated games can be applied.
The short run involves relatively minor technological or organizational
changes in production activities against a backdrop of fixed employment
levels and compensation packages. These changes occur less often, re-
quire agents to predict how future costs and benefits will be allocated,
and involve greater informational asymmetries. Wage and employment
adjustments occur still less often, entail higher stakes, have obvious redis-
tributive consequences, and frequently involve asymmetrical information
in an essential way. In the very long run, the relationship between investors
and workers is reaching an end, or is likely to do so, and repeated game
effects or concerns with reputation offer little protection to either side.
Very short run

At the level of daily operations, innumerable small adjustments have to
be made to adapt to a changing environment. Information must be con-
veyed, tasks must be assigned, effort must be supplied, and output must be
monitored. As a great many observers have stressed, this environment is
one of latent or open conflict between bosses and employees. Because no
employment contract is detailed enough to say what everyone must do in
all possible states of the world, and bargaining costs rule out continual re-
visions in compensation schemes, authority is used to coordinate worker
activities. In a capitalist firm, this authority structure serves investors. At
existing rates of employee compensation, investors want greater effort,
higher quality, fewer delays, and lower costs, while workers want greater
leisure, decreased supervision, more collegiality, and safer conditions.
What usually happens is that social conventions develop that implic-

itly specify acceptable effort levels, working conditions, and so forth, and
embody expectations about the responses that are likely to be triggered
if one side deviates. These conventions can be modeled as equilibria in
a repeated game, where either party can punish a deviation from equi-
librium by withdrawing future cooperation. If a worker shirks for one
day, that day’s output is lost and the employer has paid wages in ex-
change for nothing, but the worker can be fired or otherwise penalized
the next day. Similarly, if the employer cheats by not paying the promised
wage, this is usually not a year’s wage but perhaps a week’s or a day’s
wage, and the worker can quit immediately in response. Because of the
shortness of the time intervals between successive actions and the high
likelihood that serious cheating will be detected, an efficient equilibrium
may emerge (short time intervals have the same effect as a large weight
on future payoffs; see the discussion of the Folk Theorem in Section 6.6).
But, in general, there are many possible equilibria, ranging from efficient
to highly dysfunctional.
Short run
When managers in a capitalist firm want to reorganize production activ-
ities in some way that departs from normal routines, but do not want to
renegotiate the compensation packages of the affected employees, new
issues arise. First, if the proposal is just a disguised attempt to get more
effort for the same pay, or to do away with costly but valued working con-
ditions, this may not be obvious immediately. The stakes are therefore
higher on both sides: Bosses gain more by cheating on the firm’s social
conventions, if that is what they are trying to do, and workers risk suffering
larger losses. Relatedly, decisions along these lines are made less often,
and the conditions of the Folk Theorem are less likely to be satisfied, so
an efficient equilibrium may not exist. There are also collective-action
issues. Workers who are harmed may need help in mounting an effective
campaign against managerial decisions, and each worker may wait for
others to seize the initiative.
Conversely, even mutually beneficial plans can be derailed by lack of
trust among employees. This is especially likely when informational asym-
metries are important, as they usually are when it becomes necessary to
forecast the allocation of costs and benefits from some novel technolog-
ical or organizational arrangement. Managers in capitalist firms tend to
have better information than employees about a number of relevant vari-
ables, especially demand and cost data, and they cannot easily be forced
to reveal such information. This is an environment in which bargaining

costs can be high and impasses frequent.
A common problem in designing organizational innovations is eliciting
information from frontline employees. This can be difficult in a capitalist
firm where workers know it may be against their interests to disclose
such information, because it could be used against them later to raise
production quotas or lower piece rates. Such ratchet effects, which were
discussed in Section 1.4, follow directly from the inability of employers to
make credible commitments to their employees. Similar problems arise
when employers cannot credibly promise to reward workers later for
investments in firm-specific skills today, or provide credible assurances
that layoffs will not occur at some future date.
While participatory management supported by appropriate incentives
may take the KMF some distance in the direction of the LMF, it is doubt-
ful that KMFs can fully replicate the LMF’s advantage in eliciting private
information from frontline workers. The contrary has sometimes been
asserted; for example, Jensen and Meckling claim that if they chose to
do so, capitalist firms could “establish worker councils and agree not to
change production methods without worker approval. Moreover, em-
ployers would establish such practices if the benefits exceeded the costs”
(1979: 473). But the key question is whether employees would have any
confidence in worker councils established by their own bosses. At the end
of the day, the KMF can override decisions made by such bodies if some-
thing important is at stake. It may choose not to do so if a great deal of its
reputational capital is on the line and employees will have opportunities
for future retaliation, but LMFs need not rely on reputational mechanisms
in order to convince workers that their interests will be protected.
Long run
The period over which wages and employment vary is here termed “long
run” because a definitional feature of the firm is the inflexibility of com-
pensation packages in the short run (see Section 5.1). But compensation
schemes can be renegotiated if the need becomes pressing. The main
problem with wage adjustments from an employer standpoint is that they
are highly visible and clearly redistributive. Employees will want to know
whether there is a good reason for any cut in wages or benefits, and higher
dividends for investors do not usually count as a good reason.
Informational asymmetries play a crucial role in the context of conces-
sionary wage bargaining (see Section 10.3). Workers will not usually know
whether the firm’s managers and investors are telling the truth when they
cite adverse market conditions. This may lead to a signaling equilibrium

where managers must take costly actions to convince workers that the
alleged adversity is real. One common signal is to maintain wages while
laying off workers. Employees who see these layoffs will correctly infer
that a profitable firm would not resort to such actions, and will be less
likely to rise up in protest than if they were faced with wage cuts, which
would be attractive to investors even under profitable conditions.
The precarious job security of employees in a conventional firm can
have a number of implications for productivity. Employees who know they
may not be around in the long run will be less inclined to invest heavily
in firm-specific skills. More generally, anything that leads workers and
managers to place less weight on future rounds of their relationship makes
it harder to sustain cooperative modes of interaction. The willingness
of workers to supply effort and information is therefore systematically
weakened when employers cannot commit themselves to preserve the
relationship into the distant future.
Very long run

When the shutdown of a plant or firm is at issue, repeated game mecha-
nisms break down as a way of ensuring that promises will be kept because
there is a high probability that the relationship between employers and
employees will end soon. As noted in Section 10.3, this can explain the
marked tendency of employee buyouts to occur in financially troubled
firms. These are precisely the firms in which the capacity of investors for
credible commitment is nearest to collapse, and if there are large quasi-
rents or pension assets on the table, employees may conclude that a buyout
is their best or only safeguard.
None of this would matter if shutdown decisions were made efficiently,
taking into account any costs that employees bear in the process. Smooth
bargaining over a shutdown is unlikely for several reasons, however. First,
workers cannot borrow against the stream of future quasi-rents they de-
rive from preservation of the firm because their own labor income cannot
be used as collateral. Second, the lack of commitment devices means that
demands for concessions today, especially if they take the form of lump-
sum transfers to investors, may well be followed by additional demands
tomorrow. Third, informational asymmetries hinder wage bargaining. Fi-
nally, there are collective-action problems in having workers as a group
bribe an employer to keep a plant open. Someone must bear the cost of
gathering information, organizing workers, resolving disputes within the
workforce, and negotiating with investors. It is unclear how any individual,
or even a labor union, would appropriate the entire social benefit from
these efforts.
Once again, LMFs do not have such problems. Putterman points out
the “different set of considerations affecting decisions on layoffs and plant
closings in the cases of financier and worker-controlled firms. Ownership
rights being for sale, the [capitalist] firm should pick up and move, or shut
down and sell off its assets, whenever the discounted profit stream can be
made larger by doing so. Workers with ties to their communities will tend
to apply a different calculus” (1993: 259). One concrete example may be
illuminating. Berman (1967: ch. 9) reports that most conventional firms
in the plywood industry had, as of her writing, responded to the deple-
tion of local timber resources by physically shifting their facilities to less
depleted areas, while cooperatives with ties to their local communities
did not. More generally, in response to economic adversity, conventional
firms closed old plants, built new plants with an improved layout, and
consolidated their operations in more efficient plants. The cooperatives
responded with changes in work organization, tighter discipline, invest-
ments in labor-saving machinery to raise output per person, shifts in their
product mixes, and less wastage of raw materials (1967: 186–87).
11.3 Credible Commitment toward Capital

If one grants that LMFs might have productivity advantages in numerous
industries where they do not presently exist, the question is why LMFs
have not already conquered the world. For one answer, I turn to the
other half of the credibility equation, the fact that LMFs have trouble
making commitments to capital suppliers. This has prima facie merit as an
explanation because LMFs seldom appear in capital-intensive industries.
From a commitment standpoint, the major problem for the LMF is
that capital is a stock, and must be obtained now in exchange for a flow of
payments later. Workers could appropriate a large portion of this wealth
through rapid asset depreciation or they could find reasons not to make
promised payments on time. It is useful to distinguish among ways of
attracting outside capital according to the level of protection an investor
would enjoy in each case. In order of decreasing security, I will consider
leasing, debt, and equity.
Leasing
Patterns of depreciation for physical assets are affected by patterns of
use, and proper maintenance frequently calls for active monitoring and
intervention by the user. When an asset is leased from an outside owner

and the user does not bear reductions in the capital value of the asset, the
incentive is to depreciate the asset quickly to increase current earnings or
to reduce maintenance effort (see Section 8.2).
The most reliable sanction an absentee asset owner can impose on an
LMF is to end the leasing arrangement and retrieve the asset once trouble
is detected, but by then it may be too late. Apart from this, the owner’s
protection resides primarily in the LMF’s concern for its reputation, since
it may want to lease other assets in the future. This safeguard is most
effective when the assets involved are easy to inspect, their value is not
sensitive to patterns of past use, the firm renews its leasing contracts at
frequent intervals, and it is straightforward to quantify any damage. LMFs
do lease assets such as buildings, machinery, and vehicles, so the severity of
the problem should not be exaggerated. But the LMF’s inability to provide
a credible guarantee against uncontrolled depreciation will sometimes
prevent it from leasing productive assets at a price the members are willing
and able to pay.
Debt
An LMF that borrows from a bank to finance the purchase of a pro-
ductive asset is in a different position from an LMF that leases the asset
from an outside owner. First, the lender can lose the principal as well
as the interest payments due, while the lessor retains ownership of the
asset and faces only the danger of not being paid for its use. This distinc-
tion may not be very significant in practice if the LMF can depreciate
the asset in an opportunistic way. Another difference is that liquid funds
can be diverted to consumption purposes more easily than physical as-
sets can. Normally, however, lenders can ensure that a loan is used to
acquire an asset meeting certain broad specifications. This can be done,
for example, by doling out the loan in small steps as the asset is assem-
bled. A further way in which debt differs from a leasing arrangement
is that with leasing, it is the outside owner who chooses the characteris-
tics of the asset to be leased, while with debt, it is the borrower. This
leads to moral-hazard issues because after a repayment schedule has
been established, the borrower can pursue risky projects with high upside
potential.
There is also a more straightforward danger. After acquiring a pro-
ductive asset and generating a stream of income from its use, the firm
may stop making timely payments to the lender. Aside from instituting
legal proceedings, the lender can resort to two kinds of threat. The first
is to remove the asset if it was pledged as collateral. The effectiveness

of this depends on how much the LMF cares about losing the asset. If
its services were extracted with undue haste, this may not matter much.
Furthermore, if deterrence should fail and the asset must be repossessed,
a highly specialized asset will have little value in other uses.
The second threat is to withhold future loans. This is a routine aspect
of credit lines and other short-term borrowing, and in principle works in
the same way as efficiency wage incentives in the labor market. Preferred
borrowers are offered a premium through lower interest rates or other
favorable treatment, with the implication that behavior unacceptable to
the lender will result in termination. But the symmetry with the labor
market breaks down for durable assets, especially if retained earnings
provide sufficient internal resources after an initial lumpy investment
is made, because termination threats must be accompanied by a very
large premium for good behavior by the borrower in order to have any
effect.
I have framed the issue as one of non-commitment or failure to live up
to promises, but informational asymmetries also play a role. If borrow-
ers have unknown characteristics that make default more or less likely,
a bank will ration credit or charge safe borrowers a higher interest rate
than they would pay under complete information. A bank will also be
more concerned about the collateral value of an asset because the in-
formational asymmetry means that default will have a higher probability
in equilibrium. This problem ultimately derives from a safe borrower’s
inability to make a credible claim about her true type.
Empirical evidence from earlier chapters generally fits the picture
just outlined. It is not unknown for LMFs to rely partially on debt fi-
nance. Often this involves loans from members as in Mondragon and
other European cooperatives, where commitment problems are min-
imal because workers themselves run the firm. Some LMFs such as
the plywood cooperatives have built up regular relationships with local
banks. These normally involve short-term loans with generic assets
such as structures or log inventories used as collateral. On a larger
scale, the transfer of control to employees at United Airlines did not
interfere with the relationships between UAL and its creditors (see
Section 10.3). This smooth transition probably resulted in part from the
availability of generic collateral (passenger jets), together with reputa-
tional considerations: no one expected UAL to leave the airline indus-
try, and it would clearly need regular access to the credit market in the
future.
Equity
LMF advocates have sometimes suggested that such firms could attract
capital through non-voting equity shares. These would resemble shares in
corporations but without the usual voting rights. In this situation, investors
would be committing their funds not to any particular project or asset,
but to the firm as a whole, for whatever use the LMF management or
membership found expedient. In return, the equity claimants would get a
stream of dividends that would fluctuate with the fortunes of the firm, in
order to shift some income variability from risk-averse workers to more
risk-tolerant investors.
From the standpoint of credible commitment, the objection to this
proposal is clear. Once investors have turned their funds over to the firm,
there is no reason why dividends would ever be paid, apart from the
firm’s concern for its reputation or its need for further capital later. By
contrast with leasing or debt, there is no designated asset that investors
could extract from the firm in the event of non-payment, and they would
have minimal legal recourse if the LMF paid high wages or inflated costs
in other ways. And as with debt, the threat of non-renewal has little bite
when assets are durable, retained earnings are healthy, or the firm is about
to go out of business. Unsurprisingly, there are few documented cases in
which workers’ cooperatives have used non-voting equity.
Something approximating this situation occurs when employees buy a
majority of equity shares and relegate outside shareholders to a minority
position (see the Algoma and United cases in Section 10.3). If one believes
that employees and investors will each vote as a bloc on every issue,
then the outside investors in such firms are effectively non-voting. But
an employee buyout can be structured so that employees lack a voting
majority on the board despite having majority share ownership. There is
often a pivotal group of directors who are not fully accountable to either
of the two core shareholding groups, and who play a role like that of
the chairman under the “full parity” version of German codetermination
(see Section 4.3). Furthermore, the conflicting interests of workers and
investors are partly aligned by the fact that dividends are paid out in
fixed proportions to the inside and outside shareholders. Workers can
therefore gain significant control rights while continuing to pay dividends
to minority capital suppliers.
All the financing problems outlined here are exacerbated to the extent
that workers are risk-averse and do not want to put their limited wealth
into a single firm. Other things being equal, this increases the need for
external capital and tightens the financial noose around the LMF. I return
11.4 The Composition of Control Groups 249
to this topic in Section 11.6. For now, it suffices to point out that the argu-
ments to this point could all be formalized in a model with risk neutrality,
because they depend on a lack of intertemporal commitment devices or
information asymmetries but not risk-aversion. Whether risk-aversion
adds a lot or a little explanatory power is an empirical matter. Indeed,
most of the story would go through even without information asymme-
tries because it requires only a repeated game framework in which each
player’s behavior can be observed by the others. The key element in the
story is not the absence of information, but rather the absence of legally
binding commitments about future actions.
To the extent that employment relationships in the capitalist firm be-
come more like the financing problems of the labor-managed firm, the
distinction between the two begins to blur. An example arises when em-
ployees are asked to make large sunk investments in a community or
in firm-specific skills, in return for promises made by the employer that
such investments will be recouped through high wages later in life. The
intertemporal structure of this problem is not qualitatively different from
that of an LMF, which needs to attract upfront capital by promising a
future stream of interest or dividend payments. In each firm, there is a
danger that controllers will seize quasi-rents from non-controllers, and
there is a rationale for assigning control rights to provide some protec-
tion to the parties at risk. If the situations differ, it is perhaps by virtue of
the fact that a capitalist firm must frequently return to the labor market
to replace workers lost through turnover, and therefore may have more
reason to protect its reputation than an LMF, which rarely returns to the
capital market.
11.4 The Composition of Control Groups

The inalienability of labor in combination with economies of scale or
scope implies that a team of workers is needed in most firms. The same
is not true for capital, which can be supplied by one person or a small
group in a way that is less closely coupled with scale. The relative size of
candidate control groups, as well as their informational position relative
to the production process, has implications for the monitoring of man-
agerial performance. Here the comparison sometimes favors the LMF.
In a closely held corporation with one or a few shareholders and many
workers, an engaged owner can watch over management, and free-rider
problems in monitoring are negligible. But in large publicly traded cor-
porations, there are often many more shareholders than employees, and
employees as a group might more easily overcome free-rider problems

than the shareholders as a group. Through their involvement in produc-
tion, employees also have access to information about managers. In fact,
LMF members sometimes seem unable to commit themselves not to fire
the manager (see the plywood and Mondragon examples in Chapter 3).
Shareholders do have access to takeover mechanisms to discipline
managers. Such mechanisms do not function as smoothly, if at all, for
LMFs. However, takeovers do not eliminate free-riding among share-
holders (Grossman and Hart, 1980). The difficulties with LMF takeovers
can also be overstated. LMF groups that enjoy strong management and
organization have sometimes pursued an explicit strategy of taking over
failing KMFs (see the Mondragon system in Section 3.3 and the Lega
federation in Section 4.1).
If the entrenchment of top managers at the expense of shareholder
value is a serious problem, employees and shareholders are potential
allies because both could benefit at the expense of managers by having
employees gain control within the firm. This may help to explain why
employee buyouts organized at the initiative of top management rarely
lead to board representation for employees (see Section 4.2), and why
management initially fought against German codetermination with such
vigor (see Section 4.3).
Suspicions of this sort are entertained by Hansmann, who notes that
“the creation of ESOPs in which votes are not passed through to employ-
ees can probably be explained at least in part as an effort by corporate
managers to preserve or increase their own autonomy – that is, as a means
of protecting the managers both from hostile takeovers and from direct
accountability to the employees” (1996: 107). If managerial entrench-
ment is the problem and employee buyouts are the solution, why cannot
employees and shareholders reach a mutually advantageous agreement
under which employees take over the job of monitoring management?
One possible answer is that shareholders fear workers’ control more than
they fear managers’ control, but another is that these two coalitions are
large and confront major collective-action problems, while top manage-
ment does not.
Next, consider the relative heterogeneity of interests among work-
ers and investors when each is considered as a candidate control group.
There are some theoretical reasons to expect that shareholders will ap-
proach unanimity in their preferences about managerial actions. There is
much less reason to expect this of workers. The differences again derive
from labor’s inalienability, which has at least two effects: (1) Workers are
11.5 The Commodification of Control Rights 251
sorted discretely into firms rather than dividing their labor endowments
across numerous firms, and (2) there are many more potential investors
for any given firm than there are potential employees. As explained in
Section 9.7, these factors tend to imply that the members of a worker
team will have heterogeneous preferences toward the policies of the
firm.
One consequence is that workers in a labor-managed firm may face
more difficult voting or bargaining problems than the investors in a com-
peting capitalist firm, placing the LMF at a disadvantage. Another is that
LMFs may be susceptible to takeover by wealth-maximizing investors.
A third consequence is that employees in a KMF who are contemplat-
ing a buyout may be unable to agree on the policies they would pursue
after gaining control. The latter problem interacts with the financing con-
straints discussed in Section 11.3: A group small enough to agree on firm
policies might be unable to finance a buyout, while a group large enough
to buy out the firm might fail to agree on the policies it should pursue.
The heterogeneity of worker preferences can account for the tendency
of LMFs to congregate in industries where workers have similar skills
and firms have a limited division of labor. This is a feature of most
traditional workers’ cooperatives and many professional partnerships.
Such problems also help to explain the reluctance of LMFs to diversify
into new product lines or geographical locations, investor takeovers of
plywood cooperatives and professional partnerships, and the complex
governance structures frequently created during an employee buyout
(see Section 10.3). Asymmetries in the internal composition of control
groups can thus explain various observations about the design, incidence,
and life cycles of LMFs that would be difficult to rationalize in terms of
financial constraints alone.
11.5 The Commodification of Control Rights

The economic literature has often predicted that LMFs will display a
large number of behavioral pathologies, including perverse demand and
supply responses, a tendency to underinvest, a bias toward expansion
through hired labor rather than new members, and a tendency to sell out to
external investors (see Chapters 7 and 10). The symmetry principle from
Section 6.1 requires that we ask why KMFs do not suffer from parallel
pathologies. As I argued in Chapter 7, the central reason is that KMFs
treat control as one element in a larger bundle of ownership rights that
can be bought and sold quite easily.
Why then do LMFs not establish markets for membership rights, in

which the right to participate in governing the firm is passed from one
worker to another whenever there is turnover among the workforce of
the firm? The obstacle is again the inalienability of labor. Such a mar-
ket would require that the labor inputs of the firm be altered in every
transaction, while a share transaction in a capitalist firm does not re-
quire that physical assets be altered. Legal ownership of the firm’s phys-
ical assets, or the right to be one owner among others in cases of col-
lective asset ownership, can be transferred automatically when control
changes hands, so for most purposes, the stock market can be regarded
as a market in pure financial claims rather than a market on which inputs
are supplied to firms. Several implications of this fact were developed in
Section 7.5.
I do not want to suggest that imperfections in membership markets are
necessarily fatal for LMFs. Many workers’ cooperatives have been able
to thrive without membership markets, perhaps as a result of constitu-
tional rules preventing the use of non-member labor or mandating min-
imum levels of earnings retention. Professional partnerships and closely
held corporations have developed procedures for dealing with turnover
among the members of controlling groups without open, anonymous mar-
kets for control. In these contexts, a central part of the solution is for
independent professionals to appraise the fair market value of the part-
nership or share position to be sold. This process substitutes for a price
quoted on a stock exchange and reduces the frequency of bargaining
impasses.
The imperfections of membership markets are important in explain-
ing why KMFs are better rent-appropriation devices than LMFs at the
formation stage (see Section 10.2). Barriers to commodification can also
account for the gradual conversion of some workers’ cooperatives into
capitalist firms through expanded use of non-member labor (see Section
10.4). The empirical evidence for certain other problems allegedly arising
in the absence of membership markets, including Illyrian comparative
static effects and underinvestment, is generally weak, although rigidi-
ties in output supply and input demand may be symptomatic of market
imperfections (see Sections 7.6–7.7).
A point closely related to the rent-appropriation bias at the firm-
formation stage is that no one can appropriate the full social benefit of
transforming an existing KMF into an LMF. First, there is the free-rider
problem noted in Section 11.1. Anyone who tries to organize an employee
buyout must incur significant personal costs, but will not directly enjoy
more than a small fraction of the benefit. At least in theory, this could
be overcome in the same way that appropriation biases at the formation
stage could be overcome: have the organizer of the buyout sell the right
to participate in the transformed firm.
At the formation stage, this was infeasible largely for informational
reasons. For an established firm, informational problems still exist but
they take a different form. The issue now is not so much the value of
the firm being acquired, which for a public corporation can be roughly
ascertained from its share price, but what various employee constituencies
are prepared to pay to acquire it. The need to distribute the takeover
cost among employees can lead to complicated multilateral bargaining
problems with highly incomplete information.
There is also a separate problem involving property rights. The only
way one could charge a price for membership in the newly created LMF is
by threatening to exclude any worker who does not pay, in effect expelling
that person from the firm. But after the LMF exists, it will not be bound
by such commitments and will be free to readmit the recalcitrant worker.
It is also rather unclear who has the right to determine the membership
of the new LMF. One can imagine shareholders, managers, and various
rival employee coalitions all claiming a right to appropriate the value of
membership. This lack of clarity about property rights reflects a reality
emphasized in Section 5.4: No one owns the firm.
Appropriation biases, which are a subset of the commodification phe-
nomenon, can explain two crucial demographic observations about LMFs:
their low formation rate and the infrequent nature of transitions from
KMF to LMF. This point helps to reconcile evidence for productivity
benefits under workers’ control with the small number of such firms.
11.6 Intellectual History and Current Debates

For those favoring LMFs, the theoretical conclusions in Sections 11.2
and 11.3 can be summarized as “productivity optimism” combined
with “financing pessimism.” Other writers taking a similar view include
Ben-Ner (1988b), Putterman (1993), and Bowles and Gintis (1993b, 1994).
All see LMFs as an attractive solution to the workplace conflicts of cap-
italism, but emphasize that capital constraints usually limit the ability of
LMFs to exploit their productivity advantages vis-à-vis capitalist firms.
My focus on commitment and repeated interaction differs from a
more standard way of thinking about the productivity advantage of an
LMF, which stresses that LMF members have residual income claims
(Bowles and Gintis, 1993b; Putterman 1993). Those adopting this view
point out that these residual claims stimulate worker effort directly,
especially in smaller firms, and encourage mutual monitoring. But the
replication principle from Section 6.2 raises the question of whether
KMFs can also give workers residual claims. Capitalist firms often use
team bonuses, profit-sharing, ESOPs, and other incentive devices under
which workers become partial residual claimants.
What are the limits to this process, if any? A capitalist firm whose
workers became 100 percent residual claimants would only be able to
extract returns for its investors by charging workers a fixed rental fee for
the use of its assets, or by lending funds to teams of workers. A suspicion
arises that this is incompatible with the assignment of control rights
to investors, because the financial arrangements involved sound more
like something one would expect from an LMF. For incentive reasons,
control rights are normally correlated at least roughly with residual claims
(see Chapters 5 and 6).
One can circumvent objections based on the replication principle by
observing that LMFs easily assign complete residual claims to workers
by virtue of their control structure, while KMFs cannot do this without
placing control rights in one set of hands and residual claims in another.
Assuming control rights and residual claims are complementary, one can
construct a causal chain running from the LMF governance structure to
residual claims to productivity. Whether the direct effects from workers’
control are more or less important than the indirect effects operating
through residual claims is an empirical issue. My stress on commitment
rather than residual claims is accordingly more a matter of emphasis than
of theoretical disagreement with other writers.
My own guess is that the key obstacle preventing capitalist firms from
going further in the direction of residual claims for workers is not the
fact that control would eventually be decoupled from residual claims.
The deeper problem is that making pay contingent on individual or firm
performance requires considerable trust between those who manage and
those who are paid fluctuating incomes but lack control. Employees are
justifiably worried that profit-sharing plans can be manipulated, bonuses
can be withheld, and the true reasons for unexpectedly low earnings can
be hidden from view. This is consistent with Ben-Ner’s (1988b) empha-
sis on commitment problems in KMFs, and in particular his observation
that KMFs have trouble linking wages with productivity because of the
temptations that regularly arise ex post to distort information and renege
on contracts.
A number of authors have stressed the importance of credible

safeguards to outside investors along lines similar to Section 11.3. Jensen
and Meckling (1979: 487) say explicitly that outside investors would
not supply funds to an LMF in which workers retained complete con-
trol. Williamson may have been the first to notice that the distinction
between capital as a stock and labor as a flow has consequences for
control rights within firms. He emphasizes that shareholders must risk
their entire investment, while workers always retain their stock of hu-
man capital, and that the relationship between shareholders and the
firm does not come up for periodic renewal. He concludes from this
that the board of directors in the corporation functions as a safeguard
against expropriation of shareholder wealth (1985: 304–6). Putterman
(1988a, 1993) concurs with this line of reasoning. In his case study
of the United Airlines buyout, Gordon (1999: 323) likewise endorses
the idea that agreements with employees are more likely to be self-
enforcing than agreements with shareholders, and that the latter need
dominance on the corporate board to gain credible protection from
expropriation.
For the ideas about heterogeneity within control groups in Section 11.4,
I am indebted to Hansmann (1996), who more than anyone else has
shown that such preference diversity can explain empirical regularities
not easily explicable through capital constraints. Skillman has emphasized
that heterogeneity theories require rigorous microfoundations, and that
they can account for an assortment of facts about employee and investor
takeovers without the need to invoke transaction cost concepts (Skillman
and Dow, 2002, unpublished manuscript).
The commodification of control rights as a way of thinking about LMFs
has two sources: Sertel’s (1982) dissatisfaction with the perverse compar-
ative static predictions of the Illyrian model, and Dow’s (1986) dissat-
isfaction with assertions about underinvestment based on the horizon
problem. An early and influential discussion of rent appropriation at the
firm-formation stage was provided by Marglin (1974, 1984), but the gen-
eral notion that LMFs are unattractive to profit-seeking entrepreneurs is
widespread (see Section 10.2).
One difference between the present synthesis and the work of earlier
authors is that Putterman (1993) and Bowles and Gintis (1993b, 1994)
strongly emphasize risk aversion and the demand for portfolio diversi-
fication, while I do not. All agree that diversification is valuable, other
things being equal, and that workers cannot diversify their labor in-
puts across firms. But such arguments about the rarity of LMFs often
appear theoretically shaky, and empirical support is surprisingly sparse

(see Sections 9.5–9.6).
While taking on board both Williamson’s view that control rights give
shareholders protection against expropriation and Hansmann’s view that
preference heterogeneity can disadvantage the LMF, I do not follow their
lead by concluding that the market has already rendered an efficiency
verdict. My reasons were given in Section 6.3. Here it suffices to say that
the transaction-cost edifice fails to accommodate adverse selection and
signaling, multiple equilibria in repeated games, appropriation biases, or
the supply of local public goods. Any of these may lead to efficiency losses
that are remediable through careful policy intervention. I will return to
these themes in Chapter 12.
11.7 Is Workers’ Control a Unitary Phenomenon?

Having proposed a theoretical structure to explain the rarity of workers’
control, it is time to step back and consider the generality of this struc-
ture. Can any theory provide a satisfactory account of workers’ control
in all of its manifestations, from small nineteenth-century cooperatives to
the large-scale experiments at Mondragon and United Airlines? Can the
same causal mechanisms explain why entrepreneurs rarely create LMFs,
why employees rarely buy out large conventional firms, and why LMFs
sometimes become KMFs through investor takeovers? Do the economic
principles that govern labor-intensive cooperatives in which workers have
undisputed control also apply to ESOPs in large U.S. corporations, or to
firms operating under German codetermination law?
The reader could surely be forgiven for harboring some skepticism.
As an exhibit for the prosecution, I reprint a lengthy but undoubtedly
incomplete list of the conditions that have been said to favor LMF viability
(for details, see Dow, 2001).
1. Production technology and the market environment are common

knowledge.
2. Production depends strongly on teamwork and information
sharing.
3. Production does not require an extensive division of labor.
4. Production places workers in close geographic proximity.
5. Mutual monitoring is cheap and accurate compared with monitor-
ing by supervisors.
6. Non-human assets are easily maintained and not highly vulnerable

to abuse.
7. Non-human assets are not highly specialized.
8. High rates of investment in shared non-human assets are not
required.
9. Minimum efficient scale is small, and the cost penalty for operating
below minimum efficient scale is minor.
10. Workers have substantial wealth compared with the per capita
financing needs of their firms, or external financing is readily
available.
11. Workers are not strongly risk-averse, either because of their per-
sonal preferences or the availability of personal wealth that can be
diversified through financial markets.
12. The industry is characterized by relatively stable input and output
prices.
13. Rigidity in the face of short-run cost or demand shocks is not
strongly penalized.
14. Workers prefer job security even at the expense of fluctuations in
income or hours.
15. Egalitarian treatment of members does not conflict strongly with
a need for internal incentives, or a need to recruit members on
outside markets at unequal wages.
16. Workers are relatively homogeneous with respect to age, sex, and
ethnic, cultural, or religious background.
17. Adverse-selection problems are minor because worker skills and
the quality of non-human physical assets are easily ascertained in
advance.
18. Exogenous worker turnover is infrequent.
The purpose here is not to assess the empirical merits of individual items
on the list, which can clearly be debated, but to stress the sheer number of
potentially relevant conditions and the difficulty in framing a manageable
theory that accommodates most or all of them.
A pessimist might reasonably doubt the coherence of the subject, and
believe that a comprehensive theory of workers’ control is chimerical. Per-
haps research efforts should be devoted instead to constructing separate
theories for workers’ cooperatives, professional partnerships, employee
buyouts, ESOPs, codetermination, and so on. I will nonetheless attempt
to justify my opinion that workers’ control is a coherent topic for inquiry
and that intellectual payoffs can be realized by thinking about it in a

unified way.
The modern idea of the firm as a set of incomplete contracts among
input suppliers inevitably raises the issue of who makes decisions when
existing contracts are silent. Such contractual gaps must be filled one
way or another. Assuming it makes sense to construct a theory of the
firm – that is, if “the firm” is a sensible object of study at all – then it must
make sense to construct a theory of control rights within firms, because as
argued in Chapter 5, authoritative allocation is the defining feature of a
firm. Suppliers of capital and labor are the two most obvious constituen-
cies to wield control rights within the firm, so the need for a theory of
control leads directly to exploration of investors’ control, workers’ con-
trol, and the relative viability of these two governance structures. Thus,
the first unifying element is conceptual: Explanatory stories should de-
scribe the nature and behavior of control groups, and say why the persons
constituting a control group are capital or labor suppliers.
The distinction between alienable capital and inalienable labor pro-
vides the causal glue. To the extent that individual causal stories rest on
this common foundation, they are components in a single overarching the-
ory with multiple causal channels, not merely a grab bag of monocausal
hypotheses. Explanatory stories should therefore be asked to trace their
assertions about the viability of KMFs and LMFs back to the alienabil-
ity distinction that ultimately breaks the symmetry between capital and
labor.
I have sought to adhere to these methodological principles in devis-
ing the theoretical synthesis of this chapter. Clearly this program does
not result in a single neat mechanism that accounts for most or all of the
facts. The inalienability of labor has many implications for the way con-
trol groups function: the intertemporal structure of the transactions they
engage in, their size and internal heterogeneity, the options available for
handling turnover among controllers, and maybe others. The theory of
workers’ control should therefore be pictured as a bush: It is connected
to its root system at a single point, but sprouts several branches just above
the ground that then intertwine in complex ways.
The skeptic’s concerns about lack of unity in the subject matter of work-
ers’ control can perhaps be assuaged as follows. Problems of credible in-
tertemporal commitment affect the viability of LMFs through their access
to finance at all scales and life-cycle stages. A small team of workers ap-
proaching a bank for a loan faces exactly the same problem as a coalition
of thousands of employees seeking to buy out an established firm – how
to convince the lender that the loan will be repaid. An entrepreneur

thinking about creating an LMF must confront the problem that rents
from innovation will be shared with colleagues, just as a large group of
workers thinking about converting a KMF into an LMF will likely find
that appropriation and free-rider problems obstruct their efforts. Much
as a traditional cooperative may be tempted to hire wage labor if there
is no market for membership rights, large-scale employee buyouts will
be only transitory organizational experiments unless the equity shares of
older employees can be transferred on their retirement to new employees
who arrive after the buyout has occurred.
One last methodological objection remains to be addressed. Perhaps
the theoretical synthesis in this chapter is too flexible. Perhaps the theoret-
ical bush already has too many branches, and grows new ones too easily
whenever someone suggests a new causal story. Perhaps the idea that
asymmetries in commitment, composition, and commodification have
common roots in the inalienability of labor puts one on a slippery slope
toward rationalizing still more causal stories, each having some tenuous
connection to inalienability.
This is a danger, but it is not a fatal objection. Along the way, I have dis-
carded some theoretical stories either on logical grounds or for want of ev-
idence. Examples include the Illyrian model in which LMFs maximize in-
come per worker (Section 7.2) and assertions that LMFs are rare because
KMFs have better work incentives (Section 8.7). Other stories were put
on hold pending clarification and stronger evidence for their empirical im-
portance, including the portfolio-diversification idea (Section 9.6). Some
stories dropped away as the book progressed, such as those involving
path dependence, macroeconomic externalities, and culture (Sections 6.7
and 6.8), either because they remain stuck at an early theoretical stage,
lack strong empirical implications, or are not related in an obvious way to
alienability. The causal branches that have survived this pruning process
each account for a set of empirical regularities not readily explained by
the others.
The economics profession is in a much better position today than it
was thirty years ago to understand why workers’ control is rare. Perhaps,
therefore, we can now explain to our students why the economy around
them is primarily populated by capitalist firms. One last task remains: to
determine whether this new understanding has any policy implications.
In the final chapter, I return to the normative concerns of Chapter 2 and
develop some policy ideas consistent with the theoretical framework I
have outlined.
12
Getting There from Here
12.1 Practical Considerations

A classic tale from the state of Maine goes something like this. The res-
idents of a small Maine community notice a tourist from New York. He
comes to an intersection at the east end of the town, gazes at a com-
plicated set of road signs, and then drives back toward the west end of
town. Baffled by an equally complicated set of road signs, he turns around
and drives east again. This continues for a while. Finally he pulls up to
the general store and asks, “How do you get to Nesowadnehunk?” After
pondering this for a moment, one of the locals replies, “You can’t get
there from here.”
The conclusions of Chapter 11 may have left the reader somewhat
discouraged. The inalienability of labor is a fundamental physical fact,
and few of us would want to live in a society that pretended otherwise.
Workers’ control may thus appear infeasible, at least beyond the narrow
niches where it is currently found, despite the normative arguments in
its favor. But all is not gloomy, because some of the consequences of in-
alienability can be remedied through policy measures. Capital market im-
perfections can be corrected to some degree. Collective-choice problems
can be mitigated through sensible institutional design. The appropriation
and free-rider problems hindering employee buyouts of capitalist firms
can also be overcome to some extent. There is no plausible strategy for
converting every capitalist firm to workers’ control by tomorrow, but the
scope of workers’ control can be broadened in incremental ways. This
chapter offers some road signs that may prove useful in getting there
from here.
260
12.1 Practical Considerations 261
A few guidelines are adopted in the discussion that follows, not all
of them likely to be popular with every LMF advocate. But in my view,
these practical considerations are essential to any program of workers’
control that is meant to move beyond rhetoric. First, I steer away from
strategies involving large redistributions of income or wealth, not because
I oppose redistribution but because workers’ control is more likely to ex-
pand if it does not make politically influential segments of the population
significantly worse off.
Second, I propose institutional arrangements that resemble existing
models of the sort discussed in Chapters 3 and 4. Many elements in my pro-
posal are minor extensions or adaptations of known practices. Any nov-
elty derives principally from the recombination of the component parts.
The spirit of the approach is to acknowledge the limits of our present un-
derstanding, proceed incrementally, and leave room for future tinkering.
Third, I try to avoid unrealistic assumptions about monitoring and
enforcement. A workable governance structure needs to respect human
abilities to manipulate information, free-ride on colleagues, covertly ap-
propriate wealth, and engage in other strategic activities. In particular, it
is important not to create large temptations for shareholders or managers
to evade workers’ control by keeping firms small, incorporating in other
jurisdictions, failing to incorporate, delisting firms on stock exchanges, or
playing subtle accounting games.
Finally, I have designed the organizational proposals of this chapter
in light of the theoretical synthesis from Chapter 11. We are far from
completely understanding workers’ control and the reasons for its rarity,
but available theory and evidence should be exploited. There is no need
to wait until we have achieved complete consensus on the positive theory
of workers’ control before formulating policy ideas. The wait could be
quite long, and in the meantime many experiments with workers’ control
are already underway. Economists and other social scientists should learn
from these experiments, debate their meaning, and offer suggestions of
their own.
One major unknown is whether efficiency gains can be achieved
through the spread of workers’ control. If LMFs are held back by a re-
mediable market failure, then the case for policy intervention is stronger.
However, one can accept the normative arguments of Chapter 2 even
if LMFs lack efficiency advantages in a conventional economic sense.
In the latter situation, one would presumably be attracted to policies
that achieve workers’ control at the lowest-possible efficiency cost. Our
knowledge is too primitive to attach a confident positive or negative sign
262 Getting There from Here
to the efficiency properties of workers’ control, which undoubtedly vary

with industry characteristics and the specific organizational design one has
in mind. My proposals are designed to secure net benefits where these
exist, while limiting the extent to which workers’ control is imposed in
grossly inappropriate settings.
Efforts to encourage LMFs might involve interventions at various
stages in a firm’s life cycle: for example, (1) when the firm is organized as
a proprietorship or partnership; (2) after it has become a privately held
limited liability corporation; or (3) after it has sold shares to the general
public. Interventions at stage 1 face serious obstacles. Specifically, it is
difficult to design policies to counter informational asymmetries at the
firm-formation stage. I do not deny the value of capital-market interven-
tions to make credit more available to LMFs, but the KMF has a strong
advantage as a device through which entrepreneurs can appropriate the
rents from innovation. Accordingly, I am not optimistic that capital mar-
ket policies alone would substantially increase the relative birthrate of
LMFs and thereby raise their steady-state share of the firm population.
Turning to stages 2 and 3, it is easier to formulate strategies for a
transition from KMF to LMF when there is an open market for equity
shares. The conversion of privately held firms involves special consider-
ations, especially valuation problems, to be discussed later. Thus I con-
sider stage 3 for most of this chapter. Although much of the economy as
measured by sales, employment, or investment lies outside the realm of
the publicly traded corporation, this sector is clearly important. Indeed,
one can think of workers’ control as a developmental stage that may
become relevant after a firm has reached a level of size and maturity
sufficient to warrant policy attention.
Section 12.2 outlines a proposal for converting KMFs into LMFs. The
next three sections evaluate the strengths and weaknesses of the pro-
posal in light of three topics from Chapter 11: the difficulty of making
commitments to non-controlling groups, the internal composition of con-
trol groups, and commodification of control rights. Many qualifications
and caveats will be postponed until these sections. Section 12.6 provides
a few illustrative calculations, and Section 12.7 concludes with a larger
historical perspective.
12.2 A Modest Proposal

The general strategy is to expand workers’ control by facilitating em-
ployee buyouts of corporations listed on public stock exchanges. There
are three central elements: a labor trust, labor directors, and labor shares.
After describing these components of the strategy, I fill in some institu-
tional details about how the conversion process would be set in motion,
how financing might be arranged, and how firms might be linked through
a federation.
The labor trust

This is a legal entity through which employees buy a firm’s equity shares
on the stock market. Its function is to phase out the equity market by
taking the firm private, while phasing in a market for labor shares. The role
of the trust is similar to that of an employee stock ownership trust (ESOT;
see Chapter 4), but it differs in some ways, and a distinct label is warranted.
Equity shares are purchased at the prevailing market price using funds
generated from a system of payroll deductions. I will assume that the
trust is funded by taking an identical percentage from the paycheck of
each employee.
The dividends paid on equity shares held by the labor trust are con-
strained to equal those paid on the equity shares of outside shareholders.
Rather than being paid out directly to employees in the form of cash,
dividends on shares in the labor trust are used during the transitional
phase to buy additional shares from outside investors. For the transition
phase only, each employee has an individual capital account. The balance
in this account records the cumulative capital contribution made by the
employee to the labor trust through payroll deductions and reinvestment
of dividends. Assuming the share-purchase program is not terminated
by procedures to be discussed later, the labor trust will eventually hold
enough of the firm’s equity shares that it can acquire the rest in one trans-
action, perhaps by converting it into a moderate amount of debt. This is
analogous to the procedures through which publicly listed firms are taken
private in leveraged buyouts.
Just before the firm is taken private by employees, dividends must
be aligned with wages. Because of their differing wage histories, some
employees may have large balances in their capital accounts relative to
their current wage, while others are in the reverse position. The for-
mer need to sell equity claims to the latter to make dividend claims
proportional to wage claims. The price for these transactions may be
the current market price for equity shares, or an average computed
over several recent years to avoid excessive sensitivity to short-run
fluctuations. If necessary, the labor trust should help to finance these
transactions through loans or temporary wage transfers via the firm’s
payroll system. No net liability arises for the firm as a whole during this
process.
Once dividend and wage claims are aligned across employees, and
shareholders are bought out, the distinction between labor and capital
income is abolished by replacing the dividend stream with wages. Indi-
vidual capital accounts are simultaneously closed. From this point on, the
return to equity capital is implicitly distributed using the same procedures
as are used to determine wages. In practice, it will normally be desirable
to distribute a part of the firm’s net income as bonuses that reflect current
financial performance, hold back some for expansion, and add some to
the labor trust’s collective reserves.
Labor directors
The labor trust is run by its own board of directors elected by firm em-
ployees on the basis of one vote per person. The sizes of the capital
contributions made by employees are irrelevant in awarding voting rights.
The election of these labor directors should involve proportional repre-
sentation to ensure that all major interests have a voice in managing the
trust. Thus, blue-collar and white-collar employees might choose direc-
tors through separate voting procedures, as in German codetermination.
Electoral systems could also provide representation according to product
line, geographical location, or other criteria.
The firm’s overall board of directors comprises two groups: the la-
bor directors, who manage the labor trust, and the capital directors, who
represent outside shareholders. The number of seats on the board for
each group reflects the fraction of equity capital shares held by the labor
trust compared with the fraction held by outside shareholders. Because
the labor trust continues to purchase equity shares at the current market
price throughout the transition phase, the fraction of labor directors for
the firm as a whole will grow over time.
Assuming that the labor trust buys common stock until it exhausts
the outstanding shares, the firm will eventually be fully controlled by its
workforce. The supervisory role of the board is retained, so the employ-
ees’ elected representatives choose top management. After the transition
is complete, the firm’s capital stock is collectively owned and managed
by its workforce, rather than being collectively owned and managed by
outside investors.
Labor shares
When the labor trust is created, each employee receives a labor share free
of charge. An employee gets just one of these shares, regardless of wages
or balances in individual capital accounts, and each share carries a single

vote. A labor share entitles its owner to work in the firm, receive income,
and vote for directors of the labor trust.
Labor shares must be transferred when a worker leaves and is replaced
by someone else, both during the transition phase and afterward. This is
essential to avoid a situation where non-employees vote for labor direc-
tors. One way to do this is to let the departing worker sell her labor share
at a mutually agreeable price to anyone with suitable skills who wants
to replace her. In practice, though, the sale of a labor share to a new
worker would generally require approval from the labor directors or a
hiring committee.
Alternatively, the labor trust might prohibit unilateral sales of labor
shares but offer to buy back these shares at a publicly announced price (see
Section 12.5). The labor directors could then replace an outgoing worker
by reselling the labor share to an incoming worker at the same price. The
incoming worker would be asked for a down payment and would finance
the rest through payroll deductions or a loan. A similar procedure could
be used to create and sell new labor shares when the firm expands, or to
buy back and retire labor shares when the firm contracts.
Referendum procedures
I am not advocating that the system just described be imposed on all firms.
The principle of workers’ control demands that workers themselves have
a voice in this decision. Thus it is necessary to hold a referendum among
employees on the question of whether a labor trust should be organized,
where each employee has one vote. Basic parameters, including the rate
of payroll deduction, must be specified in a legally binding way prior
to this referendum. The creation of a labor trust is a public good from
an employee standpoint, and the use of payroll deductions to finance it
should be regarded as qualitatively similar to the collection of mandatory
union dues or public payroll taxes.
Very possibly, a majority of employees in many eligible firms will
choose not to set up a labor trust, either because they want to consume
now rather than later or because they want to diversify their savings. For
some workers, these considerations will outweigh any benefits derived in
the future from increased job security or protections against managerial
abuse. Such preferences should be respected if they are freely expressed.
A reasonable requirement for conducting a referendum is a petition
endorsed by a substantial minority of employees, perhaps 10 percent,
20 percent, or 30 percent. A higher threshold helps avoid costly and
frivolous referenda with no prospect for success, while a lower threshold

limits the severity of the free-rider problem confronting the proponents
of workers’ control. Petitioners should be required to incorporate stan-
dard features of a labor trust, labor shares, and so on into their proposed
referendum language, but within these constraints would be able to tailor
the proposal to the circumstances of their own firm.
Assuming a neutral regulatory body has certified that the petition
threshold has been exceeded, a referendum campaign can begin. It should
go without saying that rights to free speech and free assembly must
be vigorously enforced during any such campaign and that employees
must be free from intimidation by management (for example, dismissal
threats). The issues here are the same as those arising in the context of
union organizing drives in North America, and the regulatory experience
gained from such campaigns should be transferable to referenda on labor
trusts. Employees must have access to audited financial statements so that
any decision to start a buyout process is based on credible information.
A referendum is likely to stimulate a torrent of propaganda from top
management, either in the interests of investors or those of managers
themselves. Employees can give such pronouncements whatever weight
they deserve. More substantively, shareholders or managers might offer
employees higher wages or other benefits if they forego a labor trust.
These transfers can serve a useful purpose by inducing employees not to
pursue workers’ control in industries for which it is poorly suited. Again,
employees will have to assess the credibility of such promises. But the ref-
erendum remains an exercise in democracy among employees, not share-
holders. Each manager has only one vote, and shareholders have no vote
at all. The fate of shareholders under this scheme will be addressed in
Section 12.3.
It is also the workforce that must decide whether to end a labor trust,
or terminate further equity investment before reaching 100 percent share
ownership and complete control. If the labor trust is to be dissolved en-
tirely – for instance, by selling its equity claims to outside investors and
converting the firm back into a KMF – a referendum among employees
seems appropriate. As with major reorganizations in most corporations,
a supermajority of two-thirds or three-fourths of the workforce would be
a reasonable constitutional requirement.
Less dramatic cases may also arise in which the labor directors want to
suspend or slow the accumulation of equity shares temporarily without
dissolving the trust, perhaps in order to pay out some cash dividends.
For this, a simple majority of labor directors should suffice, and there is
little need for a referendum before resuming equity accumulation. As

a safeguard against unresponsive labor directors, employees might be
empowered at annual general meetings to halt the process of payroll
deduction if they so desire.
Eligible firms
As mentioned earlier, the referendum legislation should apply to all lim-
ited liability companies traded on a public stock exchange. On average,
these will tend to be fairly large firms, but an arbitrary size cutoff
(for example, 2,000 employees) should be avoided. Such thresholds can
encourage management to keep firms just below the cutoff, or to spin
off divisions as legally independent firms. They also disenfranchise em-
ployees in smaller firms that may be especially good candidates for
workers’ control (recall that successful workers’ cooperatives often have
200–500 members).
Public listing has many benefits as a criterion, including its usefulness as
a discrete and easily verified test of eligibility for a referendum. A market
for the firm’s shares also offers an uncontroversial valuation method in
a transition. For many firms, the advantages of going public, including
the opportunities to attract additional capital, diversify portfolios, and
gain enhanced liquidity, eventually become too large to forego. Assuming
reasonable safeguards for shareholders are in place (Section 12.3), these
advantages will usually offset incentives to remain private in order to
avoid a possible employee buyout in the future. But a few firms may decide
to stay private because their shareholders fear severe inefficiencies or
massive redistribution under workers’ control, or just prefer not to dilute
their existing control rights. Opting out procedures of this kind provide
useful screening devices that forestall the extension of workers’ control
to firms for which it is poorly suited.
Privately held limited liability firms would probably need separate rules
if they were covered at all. These firms differ from publicly traded firms in
that the problem of share valuation becomes harder to solve. Another dif-
ference is that the procedure for triggering a referendum may need modifi-
cation. For example, an entrepreneur approaching retirement might seize
the initiative and organize a labor trust in order to create a market for
her shares. A referendum among employees could then be used to ratify
the trust. ESOP legislation in the United States has proven attractive to
many founders of privately held firms who want to cash out at retirement
or achieve liquidity in a more gradual way, and labor trusts could play an
analogous role. The use of employee petitions as a triggering mechanism
could be replaced in privately held firms by a requirement that a vote on

a labor trust be held if certain well-defined events occur, such as a pro-
posed acquisition by another company, the listing of the firm on a public
exchange, or an imminent plant closure.
Even if one were to extend the proposal for labor trusts to privately held
companies, it would be important to retain the limited-liability criterion,
because this screens out many firms for which a labor trust would probably
be a bad fit. Proprietorships and partnerships with unlimited liability are
generally small, have a legal personality that is often tenuous at best, and
frequently dissolve when key individuals leave. Valuation of such firms is
also problematic because they may lack the lengthy track record of larger
firms.
Some pitfalls
The use of individual capital accounts raises serious problems about what
happens when someone leaves. If a worker resigns during the transition,
withdraws “her” shares, and becomes an outside investor or sells shares
on the market, this slows the process of accumulation by the labor trust
and postpones the acquisition of control rights by those workers who stay
behind. Assuming the remaining workers put a positive value on gaining
control in the future, this creates a negative externality. Similarly, if the
firm has already reached 100 percent worker control but workers are
free to sell “their” equity shares on exit, collective asset ownership and
perhaps workers’ control itself will soon unravel.
This instability is further aggravated if workers can sell “their” cap-
ital shares while still employed by the firm. For diversification reasons,
an individual worker will generally want to cash out her own capital ac-
count. In a large firm, the consequences for productivity are negligible as
long as everyone else retains her shares. But if all workers sell off their
shares, any productivity gains from collective asset ownership and work-
ers’ control will be dissipated. This is essentially a prisoners’ dilemma
game in which cooperative outcomes are undermined by the pursuit of
individual self-interest. It is also a major reason why the establishment of a
labor trust must involve a collective-choice process such as a referendum.
Capitalist firms that create ESOPs generally require that employees hold
their shares for a period of time, presumably because employers recognize
both the temptation for workers to diversify and the negative implications
this would have for productivity.
These pitfalls can be avoided through a constitutional clause prohibit-
ing the sale of a capital stake to outsiders while a worker remains in the
firm and requiring that workers sell “their” equity shares to incoming
employees or to the labor trust itself on departure, at the same time as
they sell their labor shares. A byproduct of this institutional design is
to give the labor trust a permanent legal identity on a par with that of
conventional corporations.
Normative objections
One might argue on egalitarian grounds against the proposal to make eq-
uity investments proportional to wages during a transition. Why not have
more highly paid employees provide a larger fraction of their paychecks
to the labor trust? The central problem with this idea is that when the tran-
sition phase ends, dividends and wages must be aligned in a proportional
way across workers. Otherwise, it would make no sense to drop the sep-
aration between the two in moving to a system of collective asset owner-
ship. If more highly paid workers contributed proportionately more to the
labor trust, they would accumulate proportionately larger capital stakes.
At the end of the transition phase, it would then be necessary for less
highly paid workers to buy back part of these capital balances. In the ab-
sence of compensation payments among employees, those with the larger
capital balances would implicitly have some of their wealth transferred to
other workers when the firm abandoned the distinction between capital
and labor income.
Perhaps one should not rule out possibilities of this kind, and merely
leave it to the workers petitioning for a referendum to decide whether
they want to use a negative income tax as part of their transition strat-
egy. In any event, a number of considerations should be borne in mind.
First, it may be desirable to avoid a sharp divergence of interests between
employees who think like workers (those who pay relatively little into
the labor trust) and employees who think like investors (those who pay
relatively more). Second, although the costs of establishing workers’ con-
trol are assigned through proportional wage deductions in my proposal,
the benefits after the transition might be distributed in a more egalitar-
ian manner, for example through wage compression. Finally, a constant-
percentage deduction approach is easily understood and administered.
It also derives automatic legitimacy from the fact that many other wage-
deduction systems operate in the same way.
One might object that workers should not have to purchase control
rights at all. If, on philosophical grounds, workers are entitled to govern
firms, it appears perverse to say that they must become equity investors,
and therefore capitalists, to achieve this goal. This argument has some
appeal, but it should be remembered that control rights are only attached
to capital during the transition phase. Afterward, everyone on the firm’s
board of directors is elected by the labor suppliers in a democratic fashion.
An alternative is to finance a labor trust through some form of corpo-
rate profits tax. This has practical difficulties. For one thing, such a tax
might be borne largely by workers through lower wages. For another,
returns to capital are a much smaller flow than returns to labor in many
firms, so a high tax rate on capital would be needed even to implement
majority workers’ control (let alone 100 percent control) over any reason-
able time horizon. In general, policies that finance labor trusts by taxing
dividends, or diluting existing shares by issuing new shares and giving
them to the trust free of charge, drive down the value of the firm’s stock.
This has unattractive consequences to be discussed in Section 12.3.
Other financial strategies

Readers familiar with U.S. ESOPs might contemplate another financial
strategy to ease the burden on workers: have the labor trust borrow money
from a bank and use the proceeds to buy up equity shares, which are cred-
ited to employee accounts. Unfortunately, this strategy involves double-
counting, because a stream of dividends previously flowing to outside
shareholders must now do two things at once. First, it must be paid out as
a dividend stream to employees in their capacity as inside shareholders, or
be reinvested until the transition phase is complete, after which it is again
paid to employees. At the same time, however, this dividend stream is
needed to pay the principal and interest on the bank loan. In the absence
of large external subsidies or productivity gains, this approach is a shell
game and does not enable workers to escape the cost of acquiring equity
stakes in the firm. Of course, if workers only want to borrow from a bank
in order to replace external equity with debt, that is a different matter.
If it is regarded as unfair for workers to bear the full cost of acquir-
ing equity shares in the transition, a suitable response is for governments
to subsidize purchases of common stock by labor trusts. Many Western
nations already subsidize employee stock ownership to varying degrees,
including the United States, the U.K., France, and Germany, and equiva-
lent policies could be applied in the present context. The size of any tax
concession to promote workers’ control is naturally a subject for political
debate.
Two efficiency justifications for such subsidies can be derived from
the theoretical synthesis of Chapter 11. The first stresses that workers
cannot ordinarily finance the equity requirements of a firm out of their
own resources, and must therefore borrow the necessary funds. But in-
formation asymmetries in the capital market make it difficult for workers
to finance a buyout using debt. Also, employees cannot borrow against
future labor income, and the returns on collectively owned assets are bun-
dled with labor income. If workers’ control would nonetheless increase
productivity sufficiently, a subsidy may be warranted.
Another efficiency argument is that individuals or small groups cannot
appropriate the full social benefit of converting a firm to workers’ control,
even if they face no financial barriers. The potential benefits of this gov-
ernance structure are widely diffused among the workforce. In order to
capture the benefit to each individual worker, the buyout organizer would
need to threaten each with expulsion from the firm after the transition
phase unless the worker paid a suitable price to the organizer. But no
one has the property rights needed to charge such a price. There is also
an informational problem because no one knows what each employee is
willing to pay for democratic governance. Furthermore, employees and
shareholders are large constituencies, and there are serious free-rider
problems in organizing a transaction to benefit both groups. Those in the
best position to do so – the incumbent managers – would likely oppose
workers’ control for their own reasons.
Because the market for employee buyouts is very imperfect or absent
altogether, at least in firms that are not on the brink of financial disas-
ter, workers’ control is currently undersupplied. Total economic welfare
would therefore rise if some capitalist firms located close to the present
margin of workers’ control were to be acquired by their employees, but
these efficiency-enhancing buyouts do not occur spontaneously. This mar-
ket failure can be corrected by subsidizing stock purchases by employees,
provided that this is one element in a larger program to achieve demo-
cratic governance. The rationale for subsidies given here does not run
afoul of the usual objection that the productivity benefits from ESOPs
involve no externality problem (Kaufman and Russell, 1995; Mitchell,
1995).
In addition to setting the subsidy level, a crucial problem is to identify
those firms for which a subsidy is appropriate. My proposal addresses
this issue in several ways: by treating workers’ control as a public good
supplied through voting; by imposing reasonable hurdles on the decision
to hold a referendum; by requiring workers to pay most of the cost of
acquiring control; by permitting investors to bribe workers not to take
control; and by limiting the set of affected firms to exclude those for
which workers’ control seems to be a poor fit on a priori grounds. These
provisions make it unlikely that a transition will ever get started in a firm
that is not a reasonable candidate for an employee buyout, and thus tend
to target any subsidies on the most promising firms.
A federation
The proposals just described owe much to U.S. ESOPs (for the concept
of the labor trust), German codetermination (for the role of labor direc-
tors), and the plywood cooperatives (for the use of labor shares). In my
view, the lessons derived from Mondragon and the Lega do not involve
the financial structures of individual enterprises, but rather the critical im-
portance of a federation that serves many LMFs simultaneously. Legisla-
tion authorizing labor trusts should thus create a federation consisting of
those firms choosing to establish a trust. The federation could be funded
by a modest extra wage deduction, as in the Lega. At a minimum, the
federation should include a financial institution to serve member firms,
and this institution should receive an initial endowment from public funds.
I will not go into details here, but possible organizational structures are
readily suggested by the Mondragon and Lega cases.
An attractive way for employees to diversify away risks is through par-
tial pooling of incomes across member firms within a federation. There
is some danger that excessive income-pooling could lead to free-riding
problems among individual firms. This is not a decisive objection because
many KMFs have profit-sharing plans that pool net incomes across divi-
sions. But risk-pooling can be expected to dampen incentives to a degree,
and is likely to imply more central monitoring by the federation. This
could tilt the balance away from a looser system of the Lega kind toward
a tighter one like that of Mondragon.
Aside from the evident static advantages of centralizing services for
which there are scale economies, a federation helps workers’ control
become self-reinforcing in a dynamic sense. As the federation expands,
financial reserves will be accumulated, opportunities for risk pooling
will grow, more specialized services can be offered, and lessons can be
drawn from earlier experience. Over time, individual firms will likely de-
rive greater advantages by joining, and workers facing the decision of
whether to create a labor trust will become more confident of receiving
professional support if they do so.
The Mondragon and Lega experiences also make it clear that a federa-
tion is essential if LMFs are to become self-replicating. A federation or its
financial institutions can buy out KMFs and transform them into LMFs,
spin off new LMFs as internal growth occurs within existing firms, and
assist workers who want to start LMFs de novo. It can also evolve into a
large or even transnational organization while retaining democratic gov-
ernance internally. This requires nested representational structures such
as those pioneered by Mondragon and the Lega, which are analogous to
the relationships among city, state or provincial, and federal governments.
In this approach, authority can be delegated to the center for certain pur-
poses, while retaining accountability of the central authority to individual
workers or their firms.
12.3 Reassuring Shareholders

Some readers may be impatient with the idea that workers need to buy
control rights from shareholders. Why not just pass a law transferring
voting rights from the owners of common stock to the employees of firms?
One probable effect is that the market price of a firm’s shares would drop
sharply as soon as the legislation was thought to have a serious chance
of passage. Shareholders would fear, perhaps with justification, that the
employees would use their newly found control rights to increase wages
at the expense of dividends and capital gains. On normative grounds, one
might view this with equanimity, believing that the shareholders have no
right to control the firm in the first place. There is of course the practical
problem that such redistributive effects kindle political resistance and
may sink the entire project of workers’ control, but this is not the point
on which I want to focus here.
It is an inescapable reality that any largely capitalist economy must rely
on capitalist firms for growth, job creation, and innovation. If investors
in capitalist firms are routinely exposed to threats of expropriation be-
cause their employees can opt for a referendum on workers’ control, and
the outcome of a successful referendum is likely to be the destruction of
shareholder wealth, investors will have little interest in establishing or
expanding firms. Shareholders can also shield themselves against expro-
priation. Potential strategies include avoiding legal structures covered by
referendum legislation, incorporating in jurisdictions where such legisla-
tion is absent, and pre-empting any referendum by taking firms private.
These responses undermine efforts on behalf of workers’ control. A better
strategy is to keep shareholder wealth intact when a firm makes the tran-
sition to workers’ control. This ensures the continuing creation of KMFs,
which offer the best available raw material for the creation of LMFs.
How can shareholders be convinced that the market value of their

shares will not collapse when a labor trust is established? One factor
postponing the day of reckoning is that shareholders continue to hold a
majority of seats on the board of directors throughout the early stages of
the transition process. Employees cannot immediately raise wages at the
expense of dividends. As discussed in Section 4.3, there is little evidence
that the passage of the 1976 German codetermination law or the later
court rulings upholding it diminished shareholder wealth, even though
this law required near parity on the board of directors. As long as em-
ployees are in the minority, even if that minority is large, their role on the
board is likely to remain mostly informational in nature.
The territory in which employee representation exceeds a majority
but falls short of 100 percent is less well charted. One safeguard may be a
need for additional equity financing in the future, but I do not put much
weight on this for reasons given in Section 11.3. A more promising point
is that workers receive dividends per share that equal those of outside
investors. A higher wage thus comes partly out of the returns on the
equity capital already accumulated by workers. Workers will not be in a
position to impose wage increases until they have a majority on the board
and hence own a majority of the firm’s equity shares, so this is not a trivial
consideration. It is still true that a portion of any wage increase can be
shifted to outsiders. But this temptation declines as the capital accounts
of workers grow, and falls to zero when employees gain complete control.
If national or industry-wide collective-bargaining exists, then to some
degree wages will be exogenous for an individual firm. This may curb ex-
cessive wage demands during the transition. But many countries lack such
institutions, and even those that have them experience some wage drift
at the firm level. In either case, it is probably best to remove bargaining
over wages and benefits from the purview of the board of directors. This
was done at United Airlines, for example, where changes in collective
bargaining agreements were left to a separate committee having share-
holder representatives but no union members (see Section 10.3). More
generally, one could mandate that wages and benefits be decided in a com-
mittee having an equal number of shareholder and employee members
regardless of the composition of the board of directors. A neutral referee
could be selected in advance to resolve deadlocks, perhaps through final
offer arbitration.
Employees might also divert resources to other less tangible forms of
consumption, such as roomy offices, travel opportunities, or on-the-job
leisure. Because the costs come partly at the expense of shareholders,
this provides another avenue for expropriation. This problem can also
be expected to decline in severity as employees accumulate more equity.
In particular, employees nearing retirement will need to sell their indi-
vidual capital accounts back to the firm or to a replacement worker at a
price determined on the stock market, and will not be favorably disposed
toward colleagues whose behavior diminishes the value of their accounts.
As workers gain control, productivity might also increase through mu-
tual monitoring among employees and better information-sharing. This
could fully or partially compensate the firm’s outside shareholders for any
opportunism on the part of employees.
It is also unclear whether this agency problem is any worse than the
parallel agency problem arising between shareholders and top managers
in a capitalist firm. Shareholders do have nominal control rights in the
latter case, but their control is highly attenuated when equity capital is
diffusely held. Managers in capitalist firms hold a much smaller fraction
of the firm’s equity than the majority position workers must acquire in
order to gain effective control under the proposal of Section 12.2. This
larger equity role obliges employees as a group to bear more of the cost
of resource diversion than is currently true for top managers.
Shareholders might even gain by replacing de facto managerial control
with de jure workers’ control because workers will monitor top managers
as well as one another. The case of German codetermination in Section 4.3
suggested that the political resistance to this institutional innovation came
more from managers than from shareholders. The scarcity of employee
representation on U.S. boards of directors, even in the firms where ESOPs
have large equity holdings, may stem more from a desire by top managers
to avoid supervision and accountability than from any threat perceived by
shareholders. In fact, the shareholders may prefer a predictable program
of share repurchase to the sporadic dividends paid out by a group of
self-perpetuating managers.
One motivation for the referendum proposal is to reduce the collective-
action costs facing the employee coalition, and the rationale for installing
the protections of this section is to assure shareholders that they will not
be made worse off should employees vote yes. If these procedures work
well, any net gain from workers’ control will be internalized by employees,
and managers will be unable to block the transition. The referendum
procedure is thus a crude way of testing for the existence of potential
gains: If workers take over and share prices do not fall, one can infer
that workers’ control was previously stalled by the resistance or inertia
of management.
Firm governance involves a wide range of issues other than distri-

butional conflict and agency problems. Examples include choices about
product line, investment strategies, production methods, incentive sys-
tems, job security, and working conditions. Employees often have diverse
attitudes toward such matters. In a transitional context, this diversity can
be useful in stabilizing the role of outside shareholders after they are
in a minority position. Some labor directors, such as those representing
managerial or professional interests, can be expected to vote with cap-
ital directors on certain issues. Shareholders will thus continue to have
a strong voice even when they have only 30–40 percent of board seats.
A more formal strategy to protect minority interests would be to impose a
supermajority rule for a subset of particularly important decisions, giving
shareholders a veto in sensitive domains until their voting strength falls
below a threshold such as one-third or one-quarter.
This precaution could be accompanied by a further requirement that
the labor trust complete the transition phase by acquiring the last 10, 15,
or 20 percent of total equity in a single discrete transaction. Under such
a rule, the last shareholders out the door would not have to fear being
in a position of negligible influence relative to workers. If the labor trust
lacked the resources for such a transaction, it might seek to arrange fi-
nancing through bank loans, bond issues, a government agency, or the
federation from Section 12.2. Alternatively, the trust might suspend its
equity purchases temporarily while accumulating sufficiently large inter-
nal reserves to take the firm private through an offer to all remaining
shareholders.
12.4 Governing Firms

What about the danger that workers would prove unable to manage a
large firm in a coherent way as a result of their diverse preferences and in-
terests? One potential objection can be handled immediately. The cycling
problems associated with majority voting (Section 9.7) are most likely to
occur under a system of direct democracy where issues are routinely put
to a vote and the agenda always remains open to further proposals. This is
not the setting under consideration. I take it as given that any reasonably
large worker-controlled firm will use representative methods and have an
administrative system with professional managers. Cycling is not a plausi-
ble worry for a board of directors of moderate size, at least no more than
it is for committees or legislatures of other sorts where diverse groups are
represented.
12.4 Governing Firms 277
Interactions within the boardroom are instead likely to involve

bargaining and quid pro quos among a small set of well-organized in-
terests. Assuming all directors have access to common information, and
side payments can be arranged if necessary, there is no reason why
such a committee could not bargain its way to efficient decisions. Spe-
cific outcomes, especially from a distributional standpoint, would depend
on the power of each coalition to block the policies favored by other
coalitions.
Some problems can nonetheless be anticipated. First, voting proce-
dures are poorly adapted to handle distributional conflicts. Absent other
institutional mechanisms, the group with the greatest voting strength can
seize the entire pie. Conflicts among employee groups are likely to revolve
around relative wages, especially because after a transition to workers’
control, wage payments include implicit returns from jointly owned cap-
ital assets, and there is no obvious benchmark analogous to wages on
an external labor market that can serve as a focal point in partitioning
this income stream. One solution is to hand off the task of developing
wage determination criteria to a separate committee, independent of the
board of directors, with any proposal requiring ratification by the work-
force. Systems of this sort have proven workable in large cooperatives,
and labor unions face similar internal conflicts in formulating a bargaining
strategy. The difficulties do not seem insurmountable.
A predictable outcome of workers’ control is that wages will be com-
pressed relative to similar capitalist firms. Whatever the rules for wage
determination may be, the skewed nature of the initial wage distribution
will imply that the wage of the median voter is below the mean. Experi-
ence with cooperatives and labor unions shows that the majority located
below the mean will use their political strength to limit the salaries paid at
the top. Similar forces will operate in a referendum on whether to pursue
workers’ control in the first place. Organizers can be expected to seek
support from the lower end of the wage distribution by emphasizing the
opportunities that will arise to reduce salaries at the upper end. More
than anything else, this may lead management to oppose a transition
to workers’ control.
Another problem is that some employee constituencies with limited
voting strength may find themselves constantly on the losing side of every
conflict. This danger is obvious for an at-large voting system where ev-
ery labor director is elected by the entire workforce, because a cohesive
majority could exclude every other group from the board. Proportional
representation by occupation, establishment, seniority, or other criteria
is indispensable to ensure that smaller groups will have a voice at the

board level. But the same problem arises within the board because rep-
resentatives of larger groups will outnumber representatives of smaller
groups. There is a continuing danger that democratic processes will degen-
erate into a form of dictatorship run in the interests of a single dominant
constituency.
One solution is to impose supermajority requirements on issues of spe-
cial concern to minority interests, and another is to give these interests
a veto over proposals in certain domains. The difficulty with these ar-
rangements is that they can paralyze the bargaining process, especially
if there is no easy way to split differences or compensate losers through
side payments. No standardized governance system of this kind is likely
to be satisfactory across a wide range of firms. Such procedures must be
tailored to the needs of particular firms in order to provide various coali-
tions with the right mix of influence and veto power. Procedural details
are therefore best left to the advocates of workers’ control within a firm,
subject to broad statutory constraints.
A specific basis for conflict among employees involves differences in
the amount of equity capital each member has invested in the enterprise.
This problem only arises in the transition phase because afterward, indi-
vidual capital accounts no longer exist. But in the transition, older workers
with more accumulated equity will be more inclined to support dividend
payments, while younger workers favor higher wages. Section 12.3 sug-
gested that these disagreements may help to reassure outside investors
that some employees share their interest in preserving the market value
of the firm. But such conflicts are nonetheless real, and cannot easily be
eliminated.
Many of the dilemmas outlined here can be circumvented by dele-
gating substantial authority to managers and giving them job security
for a defined period of time, subject to the usual caveat that managers
can be removed for corruption or gross negligence. Going to an ex-
treme in this direction, decisions by the board of directors might be
confined mainly to hiring managers and renewing or terminating their
contracts. This would put the burden of determining worker prefer-
ences squarely on the shoulders of managers, but presumably if they are
well paid they will accept this challenge. It also raises questions about
whether board representatives can credibly promise not to intervene in
managerial decisions. If they can, then one need not be quite as con-
cerned with collective-choice problems. The crucial tradeoff will instead
become the extent to which democracy must be sacrificed on the altar of

managerial authority in order to make workers’ control a viable gover-
nance structure.
12.5 Trading Jobs

Concern has repeatedly been expressed in the literature on LMFs that
any form of collective asset ownership makes it impossible for workers
to cash out their equity stakes when they leave the firm, and thus makes
firm members reluctant to finance investments out of retained earnings.
This is incorrect. A transferable labor share enables a worker to recover
the present value of her claim on future net income on departure and pre-
serves proper investment incentives for the membership. This system also
compensates insiders for admission of new members. Even so, it is desir-
able to adopt a constitutional rule forbidding non-member labor except
in narrow and well-defined situations, because even small imperfections
in the labor-share market can unravel workers’ control.
The theory behind these conclusions was discussed at length in
Chapter 7. Here I want to focus on the heterogeneity of occupations
and the role of career paths in an LMF. Such factors have been ignored in
most theoretical analyses of membership markets, but a practical scheme
must address them. I will also consider the valuation issue arising after the
transition phase when there is no longer an open market for equity shares.
To begin, imagine an LMF in which members receive an explicit wage
equal to their next best alternative in the labor market. Any returns on
collectively owned assets, gains from superior productivity, or product-
market rents are labeled as dividends, and awarded to individual workers
in some democratic way. Now consider a new member who expects to re-
main in a given job with a fixed wage and dividend for some time period,
and then depart. Ignoring credit rationing, risk aversion, and the like, the
equilibrium price of the labor share for this job is equal to the discounted
present value of future dividends associated with the job, including divi-
dends to be paid after the worker leaves, since at exit the worker’s labor
share can be sold to a replacement who will pay a similar present value.
A real LMF will not generally partition its net income in a way that
corresponds to the categories of economic theory. Instead, returns to
collective capital, productivity gains from workers’ control, and product-
market rents may all be lumped together as “wages,” or paid out in the
form of fixed wages supplemented by annual bonuses that reflect current
financial results. Some firm income will also be added to collective reserves
or used for expansion. Nevertheless, the principles in valuing a worker’s
labor share are fairly clear.
Let each labor share be tied to a job description and appropriate skill re-
quirements. Estimate a base wage that is the best available approximation
to the wage that a worker with these characteristics could obtain on
the outside labor market. Summing up these base wages for the labor
force and deducting them from sales revenue, along with expenditures
for materials and debt service, leaves a flow of dividends attributable to
implicit returns on assets owned by the membership, the firm’s idiosyn-
cratic productivity characteristics, and possibly rents due to imperfect
competition. The discounted present value of this dividend flow is the
shadow value of the firm. The difference between the actual income paid
to a worker in a given job (including bonuses, profit-sharing, and other
current income) and the base wage for the job is the portion of the divi-
dend flow paid out to that worker.
Assuming that base wages and the shadow value of the firm’s total
dividend stream can be estimated, the correct way to value the labor
share attached to a given job is to multiply the shadow value of the firm
by the fraction of current dividends paid to the incumbent in that job.
It is enough to compute the fraction of dividends actually paid out in
a given year because dividends that are reinvested in the firm will lead
to higher dividends later, and to a first approximation future dividends
will be awarded in much the same proportions as the dividends currently
being paid out.
Next, suppose that a worker who is already a member moves to a new
job within the same firm, perhaps after acquiring new skills. This job will be
associated with some other base wage on the external labor market. The
gap between the actual wage and the base wage may also differ compared
with the old job. This raises no problem in principle. The old and new jobs
both have shadow values for their corresponding labor shares, so the
worker can sell the old labor share back to the firm and buy the new labor
share instead. This will involve a net payment from the worker to the
labor trust if a promotion results in a larger claim on total dividends. To
avoid lumpy transfers, payments could be implemented through suitable
payroll adjustments in the first year or two of the new job.
The hard part is to estimate base wages and the shadow value of the
firm. It is not always obvious what someone’s opportunity cost is, and the
shadow value of the firm is dependent on expectations about future events.
There are, however, some grounds for optimism. Personnel departments
in large firms routinely survey labor-market conditions or hire consultants

to gather data on the wages paid by firms in the same industry or region.
These are precisely the data needed to estimate base wages. The task
of estimating a firm’s shadow value, given the base wages, differs little
from the task of valuing the equity claims of minority owners in closely
held capitalist firms, or valuing the positions of members in professional
partnerships. For this purpose, the LMF is not an exotic new species,
but just another corporation that has been taken private. Thus, existing
valuation expertise should transfer without great difficulty.
In practice, a combination of actual and imputed market values could
be used. There is little reason to avoid market-driven membership prices
on entry into the firm, whatever the particular job description may be, as
long as there is an ample supply of candidates. If the labor share price
posted by the firm is too high, few workers will apply. If the price is too
low, applicants will queue for admission. Either way, insiders have in-
centives to revise the price in the right direction. Market mechanisms
cope less well with career paths within the firm because of the pervasive-
ness of idiosyncratic skills and the corresponding absence of competition.
Capitalist firms deal with this issue in part by creating internal labor mar-
kets where wages are attached to jobs rather than persons, and bilateral
bargaining is minimized. A democratic firm could likewise attach posted
labor-share prices to jobs.
Some additional institutional features are needed. First, the firm must
stand ready to buy back anyone’s labor share at its posted price in order
to give workers confidence that their investments will have some liquidity.
Second, it is vital that valuations be carried out by independent profes-
sionals on the basis of generally accepted principles. Disagreements are
most likely to erupt over alleged errors with respect to outside labor-
market conditions, but these disputes would probably not differ qualita-
tively from disputes over wages and salaries in conventional firms. Such
conflicts might be resolved by a neutral arbitrator acceptable to both sides.
Of course, firms remain free to set actual wages in any way they like and
to revise their decisions from time to time. The point is that labor-share
valuations need to reflect these distributional decisions in a transparent
and reliable way.
It might also be desirable to set up a central place at which labor-share
transactions can be carried out among firms operating in the same industry
or requiring a similar set of skills. The federation outlined in Section 12.2
could offer this service. Finally, it would be useful for a regulatory body
to establish consistent disclosure requirements, or at least attest to the
solvency of firms in the labor-share market, to restrain adverse selection

and outright fraud. This would parallel the role of securities regulators in
the stock market.
12.6 Sample Calculations

A key issue in evaluating the proposal from Section 12.2 is the likely
duration of the transition phase leading to full workers’ control. One
would also like to know how long it would take for employees to achieve
a simple majority on the firm’s board of directors, or a two-thirds majority
if this is needed for major corporate reorganizations. Another point of
interest is the equilibrium price of a labor share after a transition has
been completed. Some illustrative calculations can be used to highlight
the major variables and derive very rough answers. The analysis in this
section involves a number of heroic simplifications, and is not meant as a
serious formal model.
Let us define the following notation:
Y = value-added per year (sales revenue minus expenditures on

materials)
W = wage costs per year
K = value of collectively owned assets
D = firm debt (bank loans, outstanding bonds)
E = firm equity (where K = D + E)
r = rate of return on debt and equity capital
α = fraction of wages contributed to the labor trust annually
β = W/Y = share of labor in value-added
γ = E/K = share of equity in total assets
Several simplifying assumptions are used. Value-added, wage costs, and

assets all remain constant over time. Debt and equity have the same rate
of return, so total capital costs are independent of the debt-equity ratio.
The parameter γ determining this ratio is exogenously fixed and risk
aversion is ignored. The firm operates in a competitive product market,
so Y = W + rK. Depreciation and taxes are also ignored.
Suppose initially that all of the equity capital E is supplied by outside

shareholders. The firm’s employees then create a labor trust. In the first
year, payroll deductions generate an amount αW, which is used to pur-
chase shares on the stock market. Let E1L = αW be the total amount of
equity capital in workers’ accounts at the end of the first year.
At the beginning of the second year, the equity shares in workers’
accounts receive a dividend payment equal to r E1L. This dividend is used
to acquire more equity shares from outside investors. In addition, workers
still have the principal E1L = αW in their accounts. Payroll deductions
generate another contribution of αW, so the balance in workers’ capital
accounts at the end of the second year is
E2L = αW + (1 + r )E1L = αW[1 + (1 + r )]
After T years, the total balance is
ETL = αW[1 + (1 + r ) + (1 + r )2 + (1 + r)T−1 ]

= αW[(1 + r)T − 1]/r
where the last equality results from a standard piece of algebraic

manipulation.
We want to know the first year T in which the employees will have
transferred an arbitrary fraction λ (perhaps half) of the firm’s equity cap-
ital from the outside shareholders to their individual capital accounts in
the labor trust. The task is therefore to find the value of T such that
ETL/E = λ. Using the zero profit assumption Y = W + rK and the defi-
nition β = W/Y, we get K = (1 − β)Y/r . Using the definition γ = E/K,
we get E = γK = γ (1 − β)Y/r . Substituting this result and the solution
for ETL into the equation ETL/E = λ yields
(1 + r )T = 1 + [λγ (1 − β)/αβ]
In the year T at which this equation holds with equality, the labor trust
reaches the fraction λ of equity capital. This occurs sooner if the payroll
deduction is large, if the firm is labor-intensive, if the equity share in assets
is low, and if the rate of return on capital is high.
Given some assumptions about the parameter values, a few numerical
calculations can be performed. As a baseline, I consider α = .075,
β = .70, and γ = .50. A 7.5 percent rate of payroll deduction is high but
not unrealistic, and the other values would not be unusual for North
American manufacturing. I use a baseline return on capital of 10 percent.
Table 12.1 Years needed to achieve workers’ control with dividends

reinvested in purchase of equity shares
λ (fraction of equity capital held by employees)
0.50 0.67 1.00

α: (.05), .075, (.10) (13), 10, (8) (15), 12, (10) (18), 15, (13)
β: (.50), .70, (.90) (16), 10, (4) (18), 12, (5) (>20), 15, (6)
γ : (.30), .50, (.70) (7), 10, (12) (9), 12, (14) (11), 15, (17)
r : (.05), .10, (.15) (19), 10, (7) (>20), 12, (8) (>20), 15, (10)
In practice, the after-tax return on equity is often higher and the return
on debt is often lower.
Table 12.1 shows how the number of years required to reach a given
employee capital share (λ) depends on the parameters. Figures not
in parentheses correspond to the baseline values stated earlier. Figures
in parentheses indicate how the dependent variable λ changes as one pa-
rameter varies, holding other parameters constant at their baseline levels.
In every case, a majority share is reached in less than two decades. In
the baseline case, only one decade is needed. The latter figure is greatly
reduced if the firm is especially labor intensive (4 years if β = .9) or
the return on capital is high (7 years if r = .15). Only two more years
are needed to achieve two-thirds share ownership in the baseline case,
with full workers’ control in 15 years. The time until full control is quite
sensitive to the degree of labor intensity (6 years for a labor inten-
sive firm, but more than 20 in the capital-intensive case) and to the return
on equity (with a high return, full control is reached in 10 years).
The effect of a government subsidy can be seen from Table 12.1.
Suppose tax breaks cover 33 percent of the cost of share purchases, a
subsidy comparable to that for ESOP programs in the United States (see
Section 4.2). If workers contribute 5% of their wages, then net of subsidy
the relevant figures are those for α = .075 in Table 12.1. The effects of a
50 percent subsidy, again with a 5 percent payroll deduction for workers,
are given by the figures for α = .10.
These results assume that dividends are reinvested in further stock
purchases during the transition phase, so that employee capital accounts
grow exponentially. An alternative is to pay cash dividends, which can
be used for consumption purposes. The total employee equity stake after
T years is ETL = αWT in this situation, with equity E computed as before.
The length of the transition phase under this linear accumulation plan is
Table 12.2 Years needed to achieve workers’ control with dividends paid
out as cash
λ (fraction of equity capital held by employees)
0.50 0.67 1.00

α: (.05), .075, (.10) (22), 15, (11) (29), 20, (15) (43), 29, (22)
β: (.50), .70, (.90) (34), 15, (4) (45), 20, (5) (67), 29, (8)
γ : (.30), .50, (.70) (9), 15, (20) (12), 20, (27) (18), 29, (40)
r : (.05), .10, (.15) (29), 15, (10) (39), 20, (13) (58), 29, (20)
shown in Table 12.2. A quick comparison with Table 12.1 reveals that the
failure to take advantage of compounding in rates of return leads to long
delays in achieving high levels of share ownership. Only if the firm is very
labor intensive (β = .9) can full control be achieved within a decade.
A final scenario is similar to that used in present-day ESOPs. Suppose
dividends are always reinvested to purchase additional shares, but when
an employee leaves the firm her accumulated shares are sold back to
the open market. Under this system, full workers’ control may never be
attained. Instead, the firm is likely to reach a steady state where new equity
purchases by incumbent employees are exactly offset by the equity sales
of outgoing employees, yielding no net increase in the total equity stake
held by the labor trust.
The steady-state employee capital share can be calculated using as-
sumptions about the length of a typical worker’s stay in the firm. Suppose
every worker stays 5 years and then leaves. Assuming that wages are flat
throughout a worker’s tenure in the firm, a fifth of the workforce at the
end of any given year will be newcomers who have saved (.2)αW, a fifth
will have saved (.2)αW[1 + (1 + r )], and so on, with the last fifth having
accumulated (.2)αW[1 + (1 + r ) + · · · (1 + r )4 ]. More generally, if every
worker stays T years and then leaves, each term is multiplied by 1/T,
and the last exponent is T − 1. Summing terms and using a few algebraic
tricks, the steady-state employee equity stake turns out to be:
ETL = αW[(1 + 1/r )[(1 + r )T − 1]/T − 1]/r
Total equity E is the same as before.

Table 12.3 shows that if accumulated capital shares are sold on the
market by retiring workers, it is hard for workers to acquire a substantial
equity stake even with reinvestment of dividends during employment.
When employees stay in the firm for 5 years, they never reach a majority
Table 12.3 Steady-state employee share ownership with dividends reinvested

and equity claims sold to outsiders on departure
T (duration of stay in years by an employee)

5 10 20
α: (.05), .075, (.10) (.08), .12, (.16) (.18), .26, (.35) (.50), .75, (1.0)
β: (.5), .7, (.9) (.05), .12, (.46) (.11), .26, (>1) (.32), .75, (>1)
γ : (.3), .5, (.7) (.20), .12, (.09) (.44), .26, (.19) (>1), .75, (.54)
r : (.05), .10, (.15) (.06), .12, (.19) (.11), .26, (.47) (.26), .75, (>1)
position. If they stay 10 years, a majority is reached only when the firm
is highly labor-intensive. For a 20-year stay, full control is possible but
only in exceptional circumstances: a high payroll deduction (or subsidy),
a high labor intensity, a high debt-to-equity ratio, or a high rate of return
on capital. There are still some parameter values under which workers
fail to achieve majority control, although normally this does occur.
The lessons from Tables 12.1–12.3 are unmistakable. The best hope
of reaching majority workers’ control, let alone 100 percent control, in a
reasonable time span rests with the strategy of reinvesting dividends in
further equity accumulation and having workers sell the shares in their
individual accounts to the labor trust on departure. To maintain the value
of workers’ equity, new and continuing workers would need to take over
the capital accounts of departing workers by making cash payments to
the trust.
Under present assumptions, the price of a labor share after the tran-
sition phase is equal to the implicit equity stake of a typical worker.
Let L be the number of workers, with W = zL, where z is the annual
income of a worker. Total equity capital is E, so the price of a labor
share is p = E/L = γ (1 − β)z/rβ. A more meaningful way to express
this is to compute the ratio of the labor share price to annual income, or
p/z = γ (1 − β)/rβ. The price increases relative to wages when the firm is
financed primarily by equity rather than debt, when it is capital intensive,
and when the rate of return on capital is low.
Baseline parameter values yield p/z = 2.14, so the price of a labor
share is slightly more than twice a worker’s annual income. This is some-
what lower than corresponding figures for the shares of plywood coopera-
tives during the early 1980s, which were often priced at a level 2.6–3.0
times higher than annual earnings (Craig and Pencavel, 1992), and it is
about the same as the entry fee of two times annual earnings reported
for Mondragon during its early growth phase (see Section 3.3). In a very
labor-intensive firm (β = .9), the ratio declines to p/z = 0.55, so the price
is only about half of annual earnings. On the other hand, in a very capital-
intensive firm (β = .5), the price rises to five times annual income, a level
that would be widely viewed as prohibitive.
Prices are likewise high at low interest rates because the implicit stream
of future dividends associated with a labor share is not strongly dis-
counted. However, r = .15 with other parameters set at baseline levels
reduces the price to 1.42 times annual income. When the financial param-
eter γ is close to zero and the firm is largely debt-financed, the price of
a labor share is also near zero. As equity becomes more important – for
example, because the firm’s physical assets are highly specialized or it
needs to make intangible investments with a low collateral value – the
price of a labor share rises proportionately.
12.7 The Long and Winding Road

The literature on LMFs contains many other proposed organizational
blueprints. Interested readers may want to consult Le Grand and Estrin
(1989), Bardhan and Roemer (1993), and Jossa and Cuomo (1997). All of
these sources frame the question from a broadly socialist point of view and
linger over problems of public asset ownership, which are not germane
here. But some market socialists are quite flexible about property rights,
and there is considerable overlap with the concerns of this book.
Before the nineteenth century, corporations in the U.K. and United
States were generally established by royal charter. Their shares were
sometimes sold when companies engaged in long-distance trade and
investments were not expected to pay off until the distant future (on
the case of the Dutch East India Company, see Steensgaard, 1982). In
the nineteenth century, corporations built turnpikes, canals, and railways,
but these activities also required specific approval from a governmental
authority. General statutes that enabled groups of investors to register cor-
porations for a wide variety of purposes, gave these corporations a distinct
legal identity, and routinely granted them limited liability did not exist in
the United States, the U.K., or France until the mid-1800s. Public listing
of industrial corporations on stock exchanges did not occur in the United
States until 1890–1914, and later in Europe (Rosenberg and Birdzell, 1986:
ch. 6–7). The large publicly traded corporation is thus a recent invention.
LMFs today are in a similar position to that of corporations before
general incorporation statutes. In the absence of widespread experience
with LMFs and a standard organizational template that can be used in
large firms, it is necessary to cobble together governance structures on a

case-by-case basis using whatever materials happen to be close at hand:
partnership or corporate law, ESOPs, codetermination ideas, and so on.
Further, it is necessary to do this in an environment where crucial exper-
tise is undeveloped and a nucleus of large LMFs has not yet emerged to
demand a suitable policy framework. I am not suggesting that legal institu-
tions, regulatory agencies, or professional services are decisive in explain-
ing the rarity of LMFs. The underlying factors are those in Chapter 11.
But sensible design and a supportive policy framework are important at
the margin.
The supply of policy measures is not independent of the demand. Thus
it is worth considering whether there are any causal forces that might
make LMFs viable in a wider range of industries over time. One candi-
date is the increasing average wealth accumulated by employees through
life-cycle saving, which has made employees more willing and able to hold
shares in their own companies. Simultaneously, profit-sharing and other
incentive plans under which workers bear significant risk have come into
broader use, not just in the United States but in many European coun-
tries (Perry and Kegley, 1990). If this process continues, employees may
eventually demand voting rights commensurate with their risk-bearing
role.
There is no reason to anticipate any systematic trend in capital intensity
or minimum efficient scale. Some innovations increase capital intensity,
while others decrease it; some can only be used at very large scale, while
others facilitate entry at small scales. If there is a secular trend toward
increasing average wealth but no strong trend in capital requirements,
the proportion of industries in which workers’ control is financially viable
should grow.
Other technological developments are evidently encouraging greater
involvement by employees in the decision-making processes of conven-
tional firms. The increasing use of self-directed teams, for example, is well
known (Appelbaum and Berg, 2000). It appears reasonable to attribute
such organizational innovations at least in part to the introduction of
flexible manufacturing systems (Milgrom and Roberts, 1990b) as well
as improvements in information technology (Bresnahan, Brynjolfsson,
and Hitt, 2000). More generally, Ben-Ner, Burns, Dow, and Putterman
(2000) find that employee participation in a broad range of decision-
making activities is strongly linked to the complexity of the production
process. These findings suggest that the productivity gains from work-
ers’ control may be increasing in certain sectors. The demand for worker
representation may also grow as investments in firm-specific human cap-

ital increase (Blair, 1995; Levine, 1995).
A few other secular trends may push in the same direction. Workers’
control might be a normal good, and if so the demand for it will increase
as real incomes rise (Ben-Ner 1988b; Putterman, 1993). Higher average
education levels may also spread entrepreneurial skills more widely in
the population (Ben-Ner, 1988b). Greater capital mobility may mean
that KMFs will appropriate or destroy rents and quasi-rents traditionally
captured by labor, stimulating greater interest in workers’ control as a
countervailing strategy.
Contrary to the hopes and fears voiced in earlier eras, the road of
workers’ control leads neither to heaven nor hell. Further progress will
require creative institutional design, careful attention to transition prob-
lems, and a willingness to learn from experience. Most of all it requires
that dogmatism be replaced by pragmatism. LMFs will not dominate
Western economies any time soon, but it should prove possible to cre-
ate a robust worker-controlled sector through modest policy innovations
and small public subsidies. If in the process we learn how to reconcile
democratic governance with economic prosperity, that will be knowledge
worth having.
References
Aghion, Philippe and Jean Tirole, 1997, Formal and real authority in organizations,
Journal of Political Economy 105(1), February, 1–29.
Akerlof, George, 1970, The market for “lemons”: Quality uncertainty and
the market mechanism, Quarterly Journal of Economics 84(3), August,
488–500.
Alchian, Armen A. and Harold Demsetz, 1972, Production, information costs,
and economic organization, American Economic Review 62(5), December,
777–795.
Aldrich, Howard and Robert Stern, 1983, Resource mobilization and the creation
of U.S. producer’s cooperatives, 1835–1935, Economic and Industrial Democ-
racy 4, August, 371–406.
Algoma Steel, 2001, press release, http://www.algoma.com/, April.
Allen, Douglas W. and Dean Lueck, 1998, The nature of the farm, Journal of Law
and Economics 41(2), Part 1, October, 343–386.
Allen, Douglas W. and Dean Lueck, 1999, Searching for ratchet effects in
agricultural contracts, Journal of Agricultural and Resource Economics 24(2),
December, 536–552.
Alston, Lee J. and William Gillespie, 1989, Resource coordination and transac-
tion costs: A framework for analyzing the firm/market boundary, Journal of
Economic Behavior and Organization 11(2), March, 191–212.
Ammirato, Piero, 1996, La Lega: The Making of a Successful Cooperative Net-
work, Brookfield, VT, Dartmouth Publishing Company.
Andolfatto, David and Ed Nosal, 1997, Optimal team contracts, Canadian Journal
of Economics 30(2), May, 385–396.
Aoki, Masahiko, 1990, Toward an economic model of the Japanese firm, Journal
of Economic Literature 28(1), March, 1–27.
Appelbaum, Eileen and Peter Berg, 2000, High-performance work systems: Giv-
ing workers a stake, ch. 4 in Margaret M. Blair and Thomas A. Kochan, eds., The
New Relationship: Human Capital in the American Corporation, Washington,
DC, Brookings Institution Press, 102–144.
291
292 References
Archer, Robin, 1996, The philosophical case for economic democracy, ch. 2 in Ugo
Pagano and Robert Rowthorn, eds., Democracy and Efficiency in the Economic
Enterprise, New York, Routledge, 13–35.
Arneson, Richard J., 1993, Democratic rights at national and workplace levels,
ch. 3 in David Copp, Jean Hampton, and John E. Roemer, eds., The Idea of
Democracy, New York, Cambridge University Press, 118–148.
Arrow, Kenneth J., 1963, Social Choice and Individual Values, 2nd ed., New York,
John Wiley.
Arrow, Kenneth J., 1974, The Limits of Organization, New York, Norton.
Backus, David K. and Derek C. Jones, 1977, British producer cooperatives in the
footwear industry: An empirical evaluation of the theory of financing, Economic
Journal 87(347), September, 488–510.
Bainbridge, Stephen M., 1996, Participatory management within a theory of the
firm, Journal of Corporation Law 21(4), Summer, 657–730.
Baker, George, Robert Gibbons, and Kevin J. Murphy, 1999, Informal authority
in organizations, Journal of Law, Economics, and Organization 15(1), April,
56–73.
Baker, George, Michael Jensen, and Kevin Murphy, 1988, Compensation and
incentives: Practice vs. theory, Journal of Finance 43(3), July, 593–616.
Bardhan, Pranab K. and John E. Roemer, 1993, eds., Market Socialism: The Cur-
rent Debate, New York, Oxford University Press.
Bartlett, Will, John Cable, Saul Estrin, Derek C. Jones, and Stephen C. Smith,
1992, Labor-managed cooperatives and private firms in North Central Italy: An
empirical comparison, Industrial and Labor Relations Review 46(1), October,
103–118.
Barzel, Yoram, 1982, Measurement costs and the organization of markets, Journal
of Law and Economics 25(1), April, 27–48.
Barzel, Yoram, 1989, Economic Analysis of Property Rights, New York,
Cambridge University Press.
Batstone, Eric, 1982, France, ch. 5 in Frank H. Stephen, ed., The Performance of
Labour-Managed Firms, London, Macmillan Press, 99–121.
Baums, Theodor and Bernd Frick, 1999, The market value of the codetermined
firm, ch. 7 in Margaret M. Blair and Mark J. Roe, eds., Employees and Corporate
Governance, Washington, DC, Brookings Institution Press, 206–235.
Bellas, Carl J., 1972, Industrial Democracy and the Worker-Owned Firm: A Study
of Twenty-One Plywood Companies in the Pacific Northwest, New York, Praeger
Publishers.
Benelli, Giuseppe, Claudio Loderer, and Thomas Lys, 1987, Labor participation
in corporate policy-making decisions: West Germany’s experience with code-
termination, Journal of Business 60(4), October, 553–575.
Benham, Lee and Philip Keefer, 1991, Voting in firms: The role of agenda control,
size and voter homogeneity, Economic Inquiry 29(4), October, 706–719.
Ben-Ner, Avner, 1984, On the stability of the cooperative form of organization,
Journal of Comparative Economics 8(3), September, 247–260.
Ben-Ner, Avner, 1988a, Comparative empirical observations on worker-owned
and capitalist firms, International Journal of Industrial Organization 6(1),
March, 7–31.
References 293
Ben-Ner, Avner, 1988b, The life cycle of worker-owned firms in market economies:
A theoretical analysis, Journal of Economic Behavior and Organization 10(3),
October, 287–313.
Ben-Ner, Avner, W. Allen Burns, Gregory Dow, and Louis Putterman, 2000, Em-
ployee ownership: An empirical exploration, ch. 6 in Margaret M. Blair and
Thomas A. Kochan, eds., The New Relationship: Human Capital in the American
Corporation, Washington, DC, Brookings Institution Press, 194–240.
Ben-Ner, Avner, Tzu-Shian Han, and Derek C. Jones, 1996, The productivity ef-
fects of employee participation in control and in economic returns: A review of
empirical evidence, ch. 11 in Ugo Pagano and Robert Rowthorn, eds., Democ-
racy and Efficiency in the Economic Enterprise, New York, Routledge, 209–244.
Ben-Ner, Avner and Byoung Jun, 1996, Employee buyout in a bargaining
game with asymmetric information, American Economic Review 86(3), June,
502–523.
Berle, Adolf A. and Gardiner C. Means, 1932, The Modern Corporation and
Private Property, New York, Macmillan.
Berman, Katrina V., 1967, Worker-Owned Plywood Companies: An Economic
Analysis, Pullman, WA, Bureau of Economic and Business Research,
Washington State University Press.
Berman, Katrina V., 1982, The United States of America: A co-operative model
for worker management, ch. 4 in Frank H. Stephen, ed., The Performance of
Labour-Managed Firms, London, Macmillan Press, 74–98.
Berman, Katrina V. and Matthew D. Berman, 1989, An empirical test of the
theory of the labor-managed firm, Journal of Comparative Economics 13(2),
June, 281–300.
Blair, Margaret M., 1995, Ownership and Control: Rethinking Corporate
Governance for the Twenty-First Century, Washington, DC, Brookings Insti-
tution Press.
Blair, Margaret M., Douglas L. Kruse, and Joseph R. Blasi, 2000, Employee own-
ership: An unstable form or a stabilizing force?, ch. 7 in Margaret M. Blair and
Blair, Margaret M. and Mark J. Roe, 1999, eds., Employees and Corporate
Governance, Washington, DC, Brookings Institution Press.
Blair, Margaret M. and Lynn A. Stout, 1999, A team production theory of corpo-
rate law, Virginia Law Review 85, March, 247–328.
Blasi, Joseph R., 1988, Employee Ownership: Revolution or Ripoff ?, New York,
Harper and Row.
Blasi, Joseph R., 1990, Comment on Conte and Svejnar, in Alan S. Blinder, ed.,
Paying for Productivity: A Look at the Evidence, Washington, DC, Brookings
Institution Press, 172–181.
Blasi, Joseph R., Michael Conte, and Douglas Kruse, 1996, Employee stock owner-
ship and corporate performance among public companies, Industrial and Labor
Relations Review 50(1), October, 60–79.
Blasi, Joseph R. and Douglas L. Kruse, 1991, The New Owners: The Mass Emer-
gence of Employee Ownership in Public Companies and What It Means to
American Business, New York, HarperCollins.
294 References
Bonin, John P., Derek C. Jones, and Louis Putterman, 1993, Theoretical and em-
pirical studies of producer cooperatives: Will ever the twain meet?, Journal of
Economic Literature 31(3), September, 1290–1320.
Bonin, John P. and Louis Putterman, 1987, Economics of Cooperation and the
Labor-Managed Economy, Fundamentals of Pure and Applied Economics
No. 14, New York, Harwood Academic Publishers.
Bowles, Samuel, 1985, The production process in a competitive economy:
Walrasian, neo-Hobbesian, and Marxian models, American Economic Review
75(1), March, 16–36.
Bowles, Samuel and Herbert Gintis, 1976, Schooling in Capitalist America: Ed-
ucational Reform and the Contradictions of Economic Life, New York, Basic
Books.
Bowles, Samuel and Herbert Gintis, 1990, Contested exchange: New microfoun-
dations of the political economy of capitalism, Politics and Society 18(2), June,
165–222.
Bowles, Samuel and Herbert Gintis, 1993a, A political and economic case
for the democratic enterprise, Economics and Philosophy 9(1), April,
75–100.
Bowles, Samuel and Herbert Gintis, 1993b, The democratic firm: An agency-
theoretic evaluation, ch. 2 in Samuel Bowles, Herbert Gintis, and Bo
Gustafsson, eds., Markets and Democracy: Participation, Accountability and
Efficiency, New York, Cambridge University Press, 13–39.
Bowles, Samuel and Herbert Gintis, 1993c, A political and economic case for
the democratic enterprise, ch. 9 in David Copp, Jean Hampton, and John E.
Roemer, eds., The Idea of Democracy, New York, Cambridge University Press,
375–399.
Bowles, Samuel and Herbert Gintis, 1993d, The Revenge of Homo Economi-
cus: Post-Walrasian economics and the revival of political economy, Journal of
Economic Perspectives 7(1), Winter, 83–102.
Bowles, Samuel and Herbert Gintis, 1994, Credit market imperfections and
the incidence of worker-owned firms, Metroeconomica 45(3), October,
209–223.
Bowles, Samuel and Herbert Gintis, 1996a, Is the demand for workplace democ-
racy redundant in a liberal economy?, ch. 4 in Ugo Pagano and Robert
Rowthorn, eds., Democracy and Efficiency in the Economic Enterprise, New
York, Routledge, 64–81.
Bowles, Samuel and Herbert Gintis, 1996b, The distribution of wealth and the
viability of the democratic firm, ch. 5 in Ugo Pagano and Robert Rowthorn, eds.,
Democracy and Efficiency in the Economic Enterprise, New York, Routledge,
82–97.
Bradley, Keith, and Alan Gelb, 1983, Cooperation at Work: The Mondragon Ex-
perience, London, Heinemann Educational Books.
Bradley, Keith and Alan Gelb, 1987, Cooperative labour relations: Mondragon’s
response to recession, British Journal of Industrial Relations 25(1), March,
77–99.
Braverman, Harry, 1974, Labor and Monopoly Capital: The Degradation of Work
in the Twentieth Century, New York, Monthly Review Press.
References 295
Bresnahan, Timothy F., Erik Brynjolfsson, and Lorin M. Hitt, 2000, Technology,
organization, and the demand for skilled labor, ch. 5 in Margaret M. Blair and
Browning, Martin, 1987, Co-operatives, closures, or wage cuts: The choices facing
workers in an ailing firm, Canadian Journal of Economics 20(1), February,
114–122.
Browning, Martin and Wayne Lewchuk, 1990, Workers’ preferences for co-
operatives versus private buy-outs, Journal of Economic Behavior and Organi-
zation 14(2), October, 261–283.
Bull, Clive, 1987, The existence of self-enforcing implicit contracts, Quarterly
Journal of Economics 102(1), February, 147–159.
Carmichael, H. Lorne and W. Bentley MacLeod, 2000, Worker cooperation and
the ratchet effect, Journal of Labor Economics 18(1), January, 1–19.
Chandler, Alfred Jr., 1992, Organizational capabilities and the economic history
of the industrial enterprise, Journal of Economic Perspectives 6(3), Summer,
79–100.
Chaplinsky, Susan, Greg Niehaus, and Linda Van de Gucht, 1998, Employee
buyouts: Causes, structure, and consequences, Journal of Financial Economics
48(3), June, 283–332.
Cheney, George, 1999, Values at Work: Employee Participation Meets Market Pres-
sure at Mondragon, Ithaca, NY, ILR Press, Cornell University Press.
Cheung, Steven N. S., 1983, The contractual nature of the firm, Journal of Law
and Economics 26(1), April, 1–21.
Coase, Ronald H., 1937, The nature of the firm, Economica 4, November, 386–405,
also reprinted as ch. 2 in Oliver E. Williamson and Sidney G. Winter, eds.,
1991, The Nature of the Firm: Origins, Evolution, and Development, New York,
Oxford University Press, 18–33.
Coase, Ronald H., 1960, The problem of social cost, Journal of Law and Eco-
nomics 3, October, 1–44.
Conte, Michael A. and Derek C. Jones, 1985, In search of a theory of formation
for U.S. producer cooperatives: Tests of alternative hypotheses, in Barbara D.
Dennis, ed., Proceedings of the Thirty-Seventh Annual Meeting, Industrial Re-
lations Research Association Series, Dec. 28–30, 1984, Dallas; IRRA, Madison,
WI, 377–384.
Conte, Michael A. and Jan Svejnar, 1990, The performance effects of employee
ownership plans, in Alan S. Blinder, ed., Paying for Productivity: A Look at the
Evidence, Washington, DC, Brookings Institution Press, 143–172.
Cornforth, Chris, 1983, Some factors affecting the success or failure of worker
cooperatives: A review of empirical research in the United Kingdom, Economic
and Industrial Democracy 4(2), May, 163–190.
Cornforth, Chris and Alan Thomas, 1994, The changing structure of the worker
cooperative sector in the UK: Interpreting recent trends, Annals of Public and
Cooperative Economics 65(4), 641–656.
Craig, Ben and John Pencavel, 1992, The behavior of worker cooperatives: The
plywood companies of the Pacific Northwest, American Economic Review
82(5), December, 1083–1105.
296 References
Craig, Ben and John Pencavel, 1993, The objectives of worker cooperatives, Jour-
nal of Comparative Economics 17(2), June, 288–308.
Craig, Ben and John Pencavel, 1995, Participation and productivity: A compari-
son of worker cooperatives and conventional firms in the plywood industry, in
Martin Neil Baily, Peter C. Reiss, and Clifford Winston, eds., Brookings Papers
on Economic Activity: Microeconomics, Washington, DC, Brookings Institu-
tion Press, 121–174.
Dahl, Robert A., 1985, A Preface to Economic Democracy, Berkeley, University
of California Press.
Dearden, James, Barry W. Ickes, and Larry Samuelson, 1990, To innovate or not to
innovate: Incentives and innovation in hierarchies, American Economic Review
80(5), December, 1105–1124.
Defourney, Jacques, Saul Estrin, and Derek C. Jones, 1985, The effects of workers’
participation on enterprise performance: Empirical evidence from French co-
operatives, International Journal of Industrial Organization 3(2), June, 197–217.
Demsetz, Harold, 1991, The theory of the firm revisited, ch. 10 in Oliver E.
Williamson and Sidney G. Winter, eds., The Nature of the Firm: Origins, Evo-
lution, and Development, New York, Oxford University Press, 159–178.
Domar, Evsey, 1966, The Soviet collective farm as a producer cooperative,
American Economic Review 56(4), Part I, September, 743–757.
Dong, Xiao-yuan and Gregory Dow, 1993, Does free exit reduce shirking in pro-
duction teams?, Journal of Comparative Economics 17(2), June, 472–484.
Doucouliagos, Chris, 1995, Worker participation and productivity in labor-
managed firms and participatory capitalist firms: A meta-analysis, Industrial
and Labor Relations Review 49(1), October, 58–77.
Dow, Gregory K., 1985, Internal bargaining and strategic innovation in the theory
of the firm, Journal of Economic Behavior and Organization 6(3), September,
301–320.
Dow, Gregory K., 1986, Control rights, competitive markets, and the labor man-
agement debate, Journal of Comparative Economics 10(1), March, 48–61.
Dow, Gregory K., 1987, The function of authority in transaction cost economics,
Journal of Economic Behavior and Organization 8(1), March, 13–38.
Dow, Gregory K., 1988, Information, production decisions, and intra-firm bar-
gaining, International Economic Review 29(1), February, 57–79.
Dow, Gregory K., 1989, Knowledge is power: Informational precommitment in
the capitalist firm, European Journal of Political Economy 5(2/3), 161–176.
Dow, Gregory K., 1993a, Democracy versus appropriability: Can labor-managed
firms flourish in a capitalist world?, ch. 11 in Samuel Bowles, Herbert Gintis,
and Bo Gustafsson, eds., Markets and Democracy: Participation, Accountability
and Efficiency, New York, Cambridge University Press, 176–195.
Dow, Gregory K., 1993b, Why capital hires labor: A bargaining perspective,
American Economic Review 83(1), March, 118–134.
Dow, Gregory K., 1993c, The appropriability critique of transaction cost eco-
nomics, ch. 6 in Christos Pitelis, ed., Transaction Costs, Markets and Hierarchies,
Oxford, Basil Blackwell, 101–132.
Dow, Gregory K., 1996, Replicating Walrasian equilibria using markets for mem-
bership in labor-managed firms, Economic Design 2(2), November, 147–162.
References 297
Dow, Gregory K., 2000, On the neutrality of asset ownership for work incentives,
Journal of Comparative Economics 28(3), September, 581–605.
Dow, Gregory K., 2001, Allocating control over firms: Stock markets ver-
sus membership markets, Review of Industrial Organization 18(2), March,
201–218.
Dow, Gregory K., 2002, The ultimate control group, Journal of Economic Behavior
and Organization 49(1), September, 39–49.
Dow, Gregory K. and Louis Putterman, 1999, Why capital (usually) hires labor:
An assessment of proposed explanations, ch. 1 in Margaret M. Blair and Mark J.
Roe, eds., Employees and Corporate Governance, Washington, DC, Brookings
Dow, Gregory K. and Louis Putterman, 2000, Why capital suppliers (usually)
hire workers: What we know and what we need to know, Journal of Economic
Behavior and Organization 43(3), November, 319–336.
Drèze, Jacques H., 1976, Some theory of labor management and participation,
Econometrica 44(6), November, 1125–1139.
Drèze, Jacques H., 1989, Labour Management, Contracts and Capital Markets: A
General Equilibrium Approach, Oxford, Blackwell.
Drèze, Jacques H., 1993, Self-management and economic theory: Efficiency, fund-
ing, and employment, ch. 15 in Pranab K. Bardhan and John E. Roemer, eds.,
Market Socialism: The Current Debate, New York, Oxford University Press,
253–265.
Drèze, Jacques H. and Kåre P. Hagen, 1978, Choice of product quality: Equilib-
rium and efficiency, Econometrica 46(3), May, 493–513.
Earle, John, 1986, The Italian Cooperative Movement: A Portrait of the Lega
Nazionale delle Cooperative e Mutue, London, Allen and Unwin.
Ellerman, David P., 1975, The “ownership of the firm” is a myth, Administration
and Society 7(1), May, 27–42.
Ellerman, David P., 1984a, Workers’ cooperatives: The question of legal structure,
ch. 11 in Robert Jackall and Henry M. Levin, eds., Worker Cooperatives in
America, Berkeley, University of California Press, 257–274.
Ellerman, David P., 1984b, Theory of legal structure: Worker cooperatives, Journal
of Economic Issues 18(3), September, 861–891.
Ellerman, David P., 1986, Horizon problems and property rights in labor-managed
firms, Journal of Comparative Economics 10(1), March, 62–78.
Ellerman, David P., 1992, Property and Contract in Economics: The Case for
Economic Democracy, Oxford, Blackwell.
Epstein, Richard A., 1984, In defense of the contract at will, University of Chicago
Law Review 51(4), Fall, 947–982.
Estrin, Saul, 1983, Self-management: Economic Theory and Yugoslav Practice,
Cambridge, Cambridge University Press.
Estrin, Saul, 1989, Workers’ co-operatives: Their merits and their limitations, ch. 7
in Julian Le Grand and Saul Estrin, eds., Market Socialism, Oxford, Clarendon
Press, 165–192.
Estrin, Saul, 1991, Some reflections on self-management, social choice and re-
form in Eastern Europe, Journal of Comparative Economics 15(2), June,
349–366.
298 References
Estrin, Saul and Derek C. Jones, 1992, The viability of employee-owned firms:
Evidence from France, Industrial and Labor Relations Review 45(2), January,
323–338.
Estrin, Saul and Derek C. Jones, 1995, Workers’ participation, employee owner-
ship, and productivity: Results from French producer cooperatives, in Derek C.
Jones and Jan Svejnar, eds., Advances in the Economic Analysis of Participatory
and Labor-Managed Firms, vol. 5, Greenwich, CT, JAI Press, 3–24.
Estrin, Saul and Derek C. Jones, 1998, The determinants of investment in
employee-owned firms: Evidence from France, Economic Analysis 1(1),
February, 17–28.
Estrin, Saul, Derek C. Jones, and Jan Svejnar, 1987, The productivity effects of
worker participation: Producer cooperatives in Western economies, Journal of
Comparative Economics 11(1), March, 40–61.
Eswaran, Mukesh, and Ashok Kotwal, 1984, The moral hazard of budget-
breaking, Rand Journal of Economics 15(4), Winter, 578–581.
Eswaran, Mukesh, and Ashok Kotwal, 1989, Why are capitalists the bosses?
Economic Journal 99(394), March, 162–176.
Everett, Michael J. and Alanson P. Minkler, 1993, Evolution and organisational
choice in nineteenth-century Britain, Cambridge Journal of Economics 17(1),
March, 51–62.
Fehr, Ernst, 1993, The simple analytics of a membership market in a labor-
managed economy, ch. 16 in Samuel Bowles, Herbert Gintis, and Bo Gustafsson,
eds., Markets and Democracy: Participation, Accountability and Efficiency, New
York, Cambridge University Press, 260–276.
Field, Alexander James, 1991, Do legal systems matter?, Explorations in
Economic History 28(1), February, 1–35.
Finkin, Matthew W., 1986, The bureaucratization of work: Employer policies and
contract law, Wisconsin Law Review, 733–753.
FitzRoy, Felix R. and Kornelius Kraft, 1993, Economic effects of codetermination,
Scandinavian Journal of Economics 95(3), September, 365–375.
Freeman, Richard B., 1976, Individual mobility and union voice in the labor mar-
ket, American Economic Review 66(2), May, 361–368.
Freeman, Richard B. and Edward P. Lazear, 1995, An economic analysis of works
councils, ch. 2 in Joel Rogers and Wolfgang Streeck, eds., Works Councils:
Consultation, Representation, and Cooperation in Industrial Relations, Chicago,
University of Chicago Press, 27–50.
Freeman, Richard B. and James L. Medoff, 1984, What Do Unions Do?, New
York, Basic Books.
Fudenberg, Drew, Bengt Holmstrom, and Paul Milgrom, 1990, Short-term con-
tracts and long-term agency relationships, Journal of Economic Theory 5(1),
June, 1–31.
Fudenberg, Drew and Jean Tirole, 1991, Game Theory, Cambridge, MA, MIT
Press.
Furubotn, Eirek G., 1971, Toward a dynamic model of the Yugoslav firm, Canadian
Journal of Economics 4(2), May, 182–197.
Furubotn, Eirek G., 1976, The long-run analysis of the labor-managed firm: An
alternative interpretation, American Economic Review 66(1), March, 104–124.
References 299
Furubotn, Eirek G. and Svetozar Pejovich, 1970, Property rights and the behav-
ior of the firm in a socialist state: The example of Yugoslavia, Zeitschrift für
Nationalökonomie 30(3–4), 431–454.
Gamson, Zelda F. and Henry M. Levin, 1984, Obstacles to the survival of
democratic workplaces, ch. 9 in Robert Jackall and Henry M. Levin, eds.,
Worker Cooperatives in America, Berkeley, University of California Press,
219–244.
Garvey, Gerald, 1993, Does hierarchical governance facilitate adaptation to
changed circumstances? Some formal examples, Journal of Economic Behavior
and Organization 20(2), February, 187–211.
Gelauff, George M. M. and C. den Broeder, 1996, Governance of Stakeholder
Relationships: The German and Dutch Experience, Research Memorandum
No. 127, April, The Hague, Central Planning Board, Netherlands Bureau for
Economic Policy Analysis.
George, Donald A. R., 1993, Economic Democracy: The Political Economy of
Self-Management and Participation, London, Macmillan Press.
Gintis, Herbert, 1989, Financial markets and the political structure of the
enterprise, Journal of Economic Behavior and Organization 11(3), May,
311–322.
Gordon, Jeffrey N., 1999, Employee stock ownership in economic transitions: The
case of United and the airline industry, ch. 10 in Margaret M. Blair and Mark J.
Roe, eds., Employees and Corporate Governance, Washington, DC, Brookings
Greenberg, Edward S., 1981, Industrial self-management and political attitudes,
American Political Science Review 75(1), March, 29–42.
Greenberg, Edward S., 1984, The plywood cooperatives and democratic theory:
The case of the plywood firms, ch. 8 in Robert Jackall and Henry M. Levin,
eds., Worker Cooperatives in America, Berkeley, University of California Press,
171–214.
Greenberg, Edward S., 1986, Workplace Democracy: The Political Effects of Par-
ticipation, Ithaca, NY, Cornell University Press.
Greenwald, Bruce C. and Joseph E. Stiglitz, 1986, Externalities in economies with
imperfect information and incomplete markets, Quarterly Journal of Economics
101(2), May, 229–264.
Greenwald, Bruce C. and Joseph E. Stiglitz, 1988, Pareto inefficiency of market
economies: Search and efficiency wage models, American Economic Review
88(2), May, 351–355.
Grossman, Sanford and Oliver Hart, 1980, Takeover bids, the free-rider problem
and the theory of the corporation, Bell Journal of Economics 11(1), Spring,
42–64.
Grossman, Sanford and Oliver Hart, 1986, The costs and benefits of ownership:
A theory of vertical and lateral integration, Journal of Political Economy 94(4),
August, 691–719.
Gui, Benedetto, 1985, Limits to external financing: A model and an application
to labor-managed firms, in Derek C. Jones and Jan Svejnar, eds., Advances in
the Economic Analysis of Participatory and Labor-Managed Firms: A Research
Annual 1, Greenwich, CT, JAI Press, 107–120.
300 References
Gunderson, Morley, Geoffrey Sack, James McCartney, David Wakeley, and

Jonathan Eaton, 1995, Employee buyouts in Canada, British Journal of In-
dustrial Relations 33(3), September, 417–442.
Gunn, Christopher Eaton, 1984a, Workers’ Self-Management in the United States,
Ithaca, NY, Cornell University Press.
Gunn, Christopher Eaton, 1984b, Hoedads co-op: Democracy and cooperation at
work, ch. 7 in Robert Jackall and Henry M. Levin, eds., Worker Cooperatives
in America, Berkeley, University of California Press, 141–170.
Gurdon, Michael A. and Anoop Rai, 1990, Codetermination and enterprise per-
formance: Empirical evidence from West Germany, Journal of Economics and
Business 42(4), November, 289–302.
Hansmann, Henry, 1988, Ownership of the firm, Journal of Law, Economics, and
Organization 4(2), Fall, 267–305.
Hansmann, Henry, 1990a, When does worker ownership work? ESOPs, law
firms, codetermination, and economic democracy, Yale Law Journal 99, June,
1749–1816.
Hansmann, Henry, 1990b, The viability of worker ownership: An economic per-
spective on the political structure of the firm, ch. 8 in Masahiko Aoki, Bo
Gustafsson, and Oliver Williamson, eds., The Firm as a Nexus of Treaties,
London, Sage Publications, 162–184.
Hansmann, Henry, 1996, The Ownership of Enterprise, Cambridge, MA, The
Belknap Press of Harvard University Press.
Hart, Oliver, 1979, On shareholder unanimity in large stock market economies,
Econometrica 47(5), September, 1057–1083.
Hart, Oliver, 1991, Incomplete contracts and the theory of the firm, ch. 9 in
Oliver E. Williamson and Sidney G. Winter, eds., The Nature of the Firm:
Origins, Evolution, and Development, New York, Oxford University Press,
138–158.
Hart, Oliver, 1995, Firms, Contracts, and Financial Structure, New York, Oxford
University Press.
Hart, Oliver and Bengt Holmstrom, 1987, The theory of contracts, ch. 3 in Truman
Bewley, ed., Advances in Economic Theory – Fifth World Congress, Cambridge,
Cambridge University Press, 71–155.
Hart, Oliver and John Moore, 1990, Property rights and the nature of the firm,
Journal of Political Economy 98(6), December, 1119–1158.
Hart, Oliver and John Moore, 1994, A theory of debt based on the inalienability
of human capital, Quarterly Journal of Economics 109(4), November, 841–879.
Hassel, Anke, 1999, The erosion of the German system of industrial relations,
British Journal of Industrial Relations 37(3), September, 483–505.
Hayek, Friedrich A. von, 1945, The use of knowledge in society, American
Economic Review 35(4), September, 519–530.
Hermalin, Benjamin E., 1999, The firm as a noneconomy: Some comments
on Holmstrom, Journal of Law, Economics, and Organization 15(1), April,
103–106.
Hodgson, Geoffrey M., 1996, Organizational form and economic evolution, ch. 6
in Ugo Pagano and Robert Rowthorn, eds., Democracy and Efficiency in the
Economic Enterprise, New York, Routledge, 98–115.
References 301
Holmstrom, Bengt, 1982, Moral hazard in teams, Bell Journal of Economics 13(2),
Autumn, 324–340.
Holmstrom, Bengt, 1999, The firm as a subeconomy, Journal of Law, Economics,
and Organization 15(1), April, 74–102.
Holmstrom, Bengt and Paul Milgrom, 1991, Multitask principal-agent analyses:
Incentive contracts, asset ownership, and job design, Journal of Law, Economics,
and Organization 7(0), special issue, 24–52.
Holmstrom, Bengt and Paul Milgrom, 1994, The firm as an incentive system,
American Economic Review 84(4), September, 972–991.
Holmstrom, Bengt and Jean Tirole, 1989, The theory of the firm, ch. 2 in Richard
Schmalensee and Robert Willig, eds., Handbook of Industrial Organization,
vol. 1, Amsterdam, North-Holland, 61–133.
Holmstrom, Mark, 1989, Industrial Democracy in Italy: Workers’ Co-ops and the
Self-Management Debate, Aldershot, UK, Avebury.
Horvat, Branko, 1971, Yugoslav economic policy in the post-war period: Problems,
ideas, institutional developments, American Economic Review 61(3), Part 2,
June, 71–169, reprinted in David L. Prychitko and Jaroslav Vanek, eds., 1996,
Producer Cooperatives and Labor-Managed Systems, volume 2, Cheltenham,
UK, Edward Elgar, 3–101.
International Herald Tribune, October 14–15, 2000, “Andersen Consulting
Partners Study IPO,” Saturday/Sunday, 14.
Ireland, Norman J. and Peter J. Law, 1982, The Economics of Labour-Managed
Enterprises, London, Croom Helm.
Jackall, Robert and Joyce Crain, 1984, The shape of the small worker
cooperative movement, ch. 5 in Robert Jackall and Henry M. Levin, eds.,
Worker Cooperatives in America, Berkeley, University of California Press,
88–108.
Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance, and
takeovers, American Economic Review 76(2), May, 323–329.
Jensen, Michael C. and William Meckling, 1976, Theory of the firm: Managerial
behavior, agency costs and ownership structure, Journal of Financial Economics
3(4), October, 305–360.
Jensen, Michael C. and William Meckling, 1979, Rights and production functions:
An application to labor-managed firms and codetermination, Journal of
Business 52(4), October, 469–506.
Jensen, Michael C. and Jerold B. Warner, 1988, The distribution of power
among corporate managers, shareholders, and directors, Journal of Financial
Economics 20(1/2), January/March, 3–24.
Jones, Derek C., 1979, U.S. producer cooperatives: The record to date, Industrial
Relations 18(3), Fall, 342–357.
Jones, Derek C., 1982, The United States of America: A survey of producer
co-operative performance, ch. 3 in Frank H. Stephen, ed., The Performance
of Labour-Managed Firms, London, Macmillan Press, 53–73.
Jones, Derek C., 1984, American producer cooperatives and employee-owned
firms: A historical perspective, ch. 3 in Robert Jackall and Henry M. Levin,
eds., Worker Cooperatives in America, Berkeley, University of California Press,
37–56.
302 References
Jones, Derek C. and Takao Kato, 1995, The productivity effects of employee
stock-ownership plans and bonuses: Evidence from Japanese panel data,
American Economic Review 85(3), June, 391–414.
Jones, Derek C. and Donald J. Schneider, 1984, Self-help production cooperatives:
Government-administered cooperatives during the depression, ch. 4 in Robert
Jackall and Henry M. Levin, eds., Worker Cooperatives in America, Berkeley,
University of California Press, 57–84.
Jones, Derek C. and Jan Svejnar, 1985, Participation, profit sharing, worker
ownership and efficiency in Italian producer cooperatives, Economica 52(208),
November, 449–465.
Jones, S. R. H., 1982, The organization of work: A historical dimension, Journal
of Economic Behavior and Organization 3(2–3), June–September, 117–137.
Jones, S. R. H., 1983, Technology and the organization of work: A reply, Journal
of Economic Behavior and Organization 4(1), March, 63–66.
Jossa, Bruno and Gaetano Cuomo, 1997, The Economic Theory of Socialism
and the Labour-Managed Firm: Markets, Socialism, and Labour Management,
Cheltenham, UK, Edward Elgar.
Jürgens, Ulrich, Katrin Naumann, and Joachim Rupp, 2000, Shareholder value
in an adverse environment: The German case, Economy and Society 29(1),
February, 54–79.
Kaufman, Roger T. and Raymond Russell, 1995, Government support for profit
sharing, gainsharing, ESOPs, and TQM, Contemporary Economic Policy 13(2),
April, 38–48.
Kennan, John and Robert Wilson, 1993, Bargaining with private information,
Journal of Economic Literature 31(1), March, 45–104.
Kihlstrom, Richard E. and Jean-Jacques Laffont, 1979, A general equilibrium
entrepreneurial theory of firm formation based on risk aversion, Journal of
Political Economy 87(4), August, 719–748.
Klein, Benjamin, Robert Crawford, and Armen Alchian, 1978, Vertical integra-
tion, appropriable rents, and the competitive contracting process, Journal of
Law and Economics 21(2), October, 297–326.
Klein, Benjamin and Keith Leffler, 1981, The role of market forces in assuring
contractual performance, Journal of Political Economy 89(4), August, 615–641.
Knight, Frank H., 1921 [1964], Risk, Uncertainty and Profit, New York, Augustus
M. Kelley.
Kornai, Janos, 2000, What the change of system from socialism to capitalism does
and does not mean, Journal of Economic Perspectives 14(1), Winter, 27–42.
Kovenock, Dan and Roger Sparks, 1990, An implicit contract approach to
employee stock ownership plans, Journal of Comparative Economics 14(3),
September, 424–451.
Kramer, Gerald H., 1973, On a class of equilibrium conditions for majority rule,
Econometrica 41(2), March, 285–297.
Kreps, David, 1990, Corporate culture and economic theory, ch. 4 in James E. Alt
and Kenneth A. Shepsle, eds., Perspectives on Positive Political Economy, New
Kruse, Douglas L., 1993, Profit Sharing: Does It Make A Difference?, Kalamazoo,
MI, W. E. Upjohn Institute for Employment Research.
References 303
Kruse, Douglas L. and Joseph R. Blasi, 1997, Employee ownership, employee

attitudes, and firm performance: A review of the evidence, in David Lewin,
Daniel J. B. Mitchell, and Mahmood A. Zaidi, eds., The Human Resources
Management Handbook, Part I, Greenwood, CT, JAI Press, 113–151.
Kumbhakar, Subal C. and Amy E. Dunbar, 1993, The elusive ESOP-productivity
link: Evidence from U.S. firm-level data, Journal of Public Economics 52(2),
September, 273–283.
Le Grand, Julian and Saul Estrin, eds., 1989, Market Socialism, Oxford, Clarendon
Press.
Legros, Patrick and Steven Matthews, 1993, Efficient and nearly-efficient partner-
ships, Review of Economic Studies 60(3), July, 599–611.
Legros, Patrick and Andrew F. Newman, 1996, Wealth effects, distribution,
and the theory of organization, Journal of Economic Theory 70(2), August,
312–341.
Levin, Henry M., 1984, ESOPs and the financing of worker cooperatives, ch. 10
in Robert Jackall and Henry M. Levin, eds., Worker Cooperatives in America,
Berkeley, University of California Press, 245–256.
Levine, David I., 1993, Demand variability and work organization, ch. 10
in Samuel Bowles, Herbert Gintis, and Bo Gustafsson, eds., Markets and
Democracy: Participation, Accountability and Efficiency, New York, Cambridge
University Press, 159–175.
Levine, David I., 1995, Reinventing the Workplace: How Business and Employees
Can Both Win, Washington, DC, Brookings Institution Press.
Levine, David I. and Richard J. Parkin, 1994, Work organization, employment
security, and macroeconomic stability, Journal of Economic Behavior and
Organization 24(3), August, 251–271.
Levine, David I. and Laura Tyson, 1990, Participation, productivity, and the firm’s
environment, in Alan Blinder, ed., Paying for Productivity: A Look at the
Evidence, Washington, DC, Brookings Institution Press, 183–243.
Lindblom, Charles E., 1977, Politics and Markets: The World’s Political-Economic
Systems, New York, Basic Books.
MacLeod, W. Bentley, 1984, A theory of cooperative teams, CORE Discussion
Paper No. 8441, Université Catholique de Louvain, Belgium.
MacLeod, W. Bentley, 1987, Behavior and the organization of the firm, Journal
of Comparative Economics 11(2), June, 207–220.
MacLeod, W. Bentley, 1988, Equity, efficiency, and incentives in cooperative
teams, in Derek C. Jones and Jan Svejnar, eds., Advances in the Analysis of
Participatory and Labor-Managed Firms 3, 5–23.
MacLeod, W. Bentley and James M. Malcomson, 1989, Implicit contracts, incen-
tive compatibility, and involuntary unemployment, Econometrica 57(2), March,
447–480.
MacLeod, W. Bentley and James M. Malcomson, 1998, Motivation and markets,
American Economic Review 88(3), June, 388–411.
Mäki, Uskali, 1998a, Is Coase a realist?, Philosophy of the Social Sciences 28(1),
March, 5–31.
Mäki, Uskali, 1998b, Aspects of realism about economics, Theoria – Segunda
Epoca 13(2), 301–319.
304 References
Mäki, Uskali, 2001, Explanatory unification: Double and doubtful, Philosophy of

the Social Sciences 31(4), December, 488–506.
Manning, Alan, 1988, A model of the labour market with some Marxian and
Keynesian features, European Economic Review 32(9), November, 1797–1816.
Marglin, Stephen A., 1974, What do bosses do? The origins and functions of
hierarchy in capitalist production, Review of Radical Political Economics 6(2),
Summer, 60–112.
Marglin, Stephen A., 1984, Knowledge and power, ch. 9 in Frank H. Stephen,
ed., Firms, Organization and Labour: Approaches to the Economics of Work
Organization, London, Macmillan, 146–164.
Marglin, Stephen A., 1991, Understanding capitalism: Control versus efficiency,
ch. 6 in Bo Gustafsson, ed., Power and Economic Institutions: Reinterpretations
in Economic History, Brookfield, VT, Edward Elgar, 225–252.
Marx, Karl, 1867 [1977], Co-operation, ch. 13 in Ben Fowkes, trans., Capital: A
Critique of Political Economy, volume 1, New York, Vintage Books, a division
of Random House, 439–454.
Mas-Colell, Andreu, Michael D. Whinston, and Jerry R. Green, 1995, Microeco-
nomic Theory, New York, Oxford University Press.
Mason, Ronald M., 1982, Participatory and Workplace Democracy: A Theoretical
Development in the Critique of Liberalism, Carbondale, IL, Southern Illinois
University Press.
Masten, Scott, 1988, A legal basis for the firm, Journal of Law, Economics, and
Organization 4(1), Spring, 181–198.
McAfee, R. Preston and John McMillan, 1991, Optimal contracts in teams, Inter-
national Economic Review 32(3), August, 561–577.
McCain, Roger A., 1977, On the optimal financial environment for worker coop-
eratives, Zeitschrift für Nationalökonomie, 37(3–4), 355–384.
McKelvey, Richard D., 1979, General conditions for global intransitivities in
formal voting models, Econometrica 47(5), September, 1085–1125.
Meade, James, 1972, The theory of labour-managed firms and profit-sharing,
Economic Journal 82(325), supplement, March, 402–428.
Miceli, Thomas and Alanson J. Minkler, 1995, Transfer uncertainty and organiza-
tional choice, Advances in the Economic Analysis of Participatory and Labor-
Managed Firms 5, 121–137.
Milgrom, Paul R., 1988, Employment contracts, influence activities, and effi-
cient organizational design, Journal of Political Economy 96(1), February,
42–60.
Milgrom, Paul R. and John Roberts, 1988, An economic approach to influence
activities and organizational responses, American Journal of Sociology 94, July,
supplement, S154–179.
Milgrom, Paul R. and John Roberts, 1990a, Bargaining costs, influence costs, and
the organization of economic activity, ch. 3 in James E. Alt and Kenneth A.
Shepsle, eds., Perspectives on Positive Political Economy, New York, Cambridge
Milgrom, Paul R. and John Roberts, 1990b, The economics of modern manu-
facturing: Technology, strategy, and organization, American Economic Review
80(3), June, 511–528.
References 305
Milgrom, Paul R. and John Roberts, 1992, Economics, Organization, and Man-
agement, Englewood Cliffs, NJ, Prentice-Hall.
Mill, John Stuart, 1848 [1987], Principles of Political Economy, Fairfield, NJ,
Augustus M. Kelley.
Miller, David, 1989, Why markets?, ch. 2 in Julian Le Grand and Saul Estrin, eds.,
Market Socialism, Oxford, Clarendon Press, 25–49.
Miller, David, 1993, Equality and market socialism, ch. 18 in Pranab K. Bardhan
and John E. Roemer, eds., Market Socialism: The Current Debate, New York,
Oxford University Press.
Miller, Gary J., 1992, Managerial Dilemmas: The Political Economy of Hierarchy,
New York, Cambridge University Press.
Mitchell, Daniel J. B., 1995, Profit sharing and employee ownership: Policy impli-
cations, Contemporary Economic Policy 13(2), April, 16–25.
Miyazaki, Hajime, 1984, On success and dissolution of the labor-managed firm in
the capitalist economy, Journal of Political Economy, 92(5), October, 909–931.
Moene, Karl Ove, 1993, Contested power, in David Copp, Jean Hampton, and
John E. Roemer, eds., The Idea of Democracy, New York, Cambridge Univer-
sity Press, 375–399.
Moene, Karl Ove and Michael Wallerstein, 1993, Unions versus cooperatives,
ch. 9 in Samuel Bowles, Herbert Gintis, and Bo Gustafsson, eds., Markets and
Montias, John M., 1970, Market and price mechanism in socialist countries:
Discussion, American Economic Review 60(2), May, 321–323.
Moore, John, 1992, The firm as a collection of assets, European Economic Review
36(2–3), April, 493–507.
Murphy, Kevin and Robert Topel, 1987, Unemployment, risk, and earnings:
Testing for equalizing wage differences in the labor market, ch. 5 in Kevin Lang
and J. S. Leonard, eds., Unemployment and the Structure of Labor Markets, New
York, Blackwell.
Myerson, Roger and Mark Satterthwaite, 1983, Efficient mechanisms for bilateral
trading, Journal of Economic Theory 29(2), April, 265–281.
National Center for Employee Ownership, 1999, The ESOP Reader: An Intro-
duction to Employee Stock Ownership Plans, Oakland, CA.
Neuberger, Egon and Estelle James, 1973, The Yugoslav self-managed enter-
prise: A systemic approach, ch. 7 in Morris Bornstein, ed., Plan and Market:
Economic Reform in Eastern Europe, New Haven, Yale University Press,
245–284.
New York Times, March 26, 2000, “A Lawyer’s Lawyer, Bridging Borders,”
Sunday, section 3, 2.
Nishman, Robert F., 1995, Worker Ownership and the Restructuring of Algoma
Steel in the 1990s, Current Issues Series, Industrial Relations Centre, Queen’s
University, Kingston, Ontario, IRC Press.
Noble, David F., 1977, America by Design: Science, Technology, and the Rise of
Corporate Capitalism, New York, Oxford University Press.
North, Douglass, 1990, Institutions, Institutional Change and Economic Perfor-
mance, New York, Cambridge University Press.
306 References
Nozick, Robert, 1974, Anarchy, State, and Utopia, New York, Basic Books.
Nuti, Domenico Mario, 1996, Efficiency, equality and enterprise democracy, ch. 10
in Ugo Pagano and Robert Rowthorn, eds., Democracy and Efficiency in the
Economic Enterprise, New York, Routledge, 184–206.
Oakeshott, Robert, 1978, The Case for Workers’ Co-ops, London, Routledge and
Kegan Paul.
Ognedal, Tone, 1993, Unstable ownership, ch. 15 in Samuel Bowles, Herbert
Gintis, and Bo Gustafsson, eds., Markets and Democracy: Participation,
Accountability and Efficiency, New York, Cambridge University Press, 248–259.
Ostrom, Elinor, 1990, Governing the Commons: The Evolution of Institutions for
Collective Action, New York, Cambridge University Press.
Pagano, Ugo, 1993, Organizational equilibria and institutional stability, ch. 6
in Samuel Bowles, Herbert Gintis, and Bo Gustafsson, eds., Markets and
Pagano, Ugo and Robert Rowthorn, 1996, The competitive selection of demo-
cratic firms in a world of self-sustaining institutions, ch. 7 in Ugo Pagano and
Robert Rowthorn, eds., Democracy and Efficiency in the Economic Enterprise,
New York, Routledge, 116–145.
Pateman, Carole, 1970, Participation and Democratic Theory, New York,
Cambridge University Press.
Pateman, Carole, 1975, A contribution to the political theory of organizational
democracy, Administration and Society 7(1), May, 5–26.
Pejovich, Svetozar, 1969, The firm, monetary policy, and property rights in a
planned economy, Western Economic Journal 7(3), September, 193–200.
Pejovich, Svetozar, 1994, A property rights analysis of alternative methods of
organising production, Communist Economies and Economic Transformation
6(2), 219–230.
Pencavel, John, 2001, Worker Participation: Lessons from the Worker Co-ops of
the Pacific Northwest, New York, Russell Sage Foundation.
Pencavel, John and Ben Craig, 1994, The empirical performance of orthodox
models of the firm: Conventional firms and worker cooperatives, Journal of
Political Economy 102(4), August, 718–744.
Pérotin, Virginie A., 1987, Conditions of survival and closure of French worker
cooperatives: Some preliminary findings, in Derek C. Jones and Jan Svejnar,
eds., Advances in the Economic Analysis of Participatory and Labor-Managed
Firms 2, Greenwich, CT, JAI Press, 201–224.
Pérotin, Virginie A., 1997, What makes coops work? Institutional viability, firm
creation, survival and closure among workers’ cooperatives in France, unpub-
lished Ph.D. thesis, Ithaca, NY, Cornell University.
Perry, Charles R. and Delwyn H. Kegley, 1990, Employee Financial Participa-
tion: An International Survey, Multinational Industrial Relations Series No. 12,
Philadelphia, PA, Industrial Research Unit, The Wharton School, Vance Hall,
University of Pennsylvania.
Pistor, Katharina, 1999, Codetermination: A sociopolitical model with governance
externalities, ch. 5 in Margaret M. Blair and Mark J. Roe, eds., Employees and
Corporate Governance, Washington, DC, Brookings Institution Press, 163–193.
References 307
Pittatore, Silvia and Gilberto Turati, 2000, A map of property rights in Italy and the
case of co-operatives: An empirical analysis of Hansmann’s theory, Economic
Analysis 3(1), February, 23–48.
Plant, Raymond, 1989, Socialism, markets, and end states, ch. 3 in Julian Le Grand
and Saul Estrin, eds., 1989, Market Socialism, Oxford, Clarendon Press, 50–77.
Plott, Charles, 1967, A notion of equilibrium and its possibility under majority
rule, American Economic Review 57, September, 787–806.
Putterman, Louis, 1981, The organization of work: Comment, Journal of Eco-
nomic Behavior and Organization 2(3), September, 273–279.
Putterman, Louis, 1982, Some behavioral perspectives on the dominance of
hierarchical over democratic forms of enterprise, Journal of Economic Behavior
and Organization 3(2–3), June–September, 139–160.
Putterman, Louis, 1984, On some recent explanations of why capital hires labor,
Economic Inquiry 22(2), April, 171–187.
Putterman, Louis, 1988a, The firm as association versus the firm as commodity:
Efficiency, rights, and ownership, Economics and Philosophy 4(2), October,
243–266.
Putterman, Louis, 1988b, Asset specificity, governance, and the employment
relation, in G. Dlugos, W. Dorow, and K. Weiermair, eds., Management under
Differing Labour Market and Employment Systems, Berlin, Walter de Gruyter.
Putterman, Louis, 1993, Ownership and the nature of the firm, Journal of Com-
parative Economics 17(2), June, 243–263.
Putterman, Louis and Gilbert Skillman, 1992, The role of exit costs in the theory
of cooperative teams, Journal of Comparative Economics 16(4), December,
596–618.
Quarter, Jack and Judith Brown, 1992, Worker buyouts in Canada: A social net-
working analysis, Economic and Industrial Democracy 13(1), February, 95–117.
Radner, Roy, 1992, Hierarchy: The economics of managing, Journal of Economic
Literature 30(3), September, 1382–1415.
Rawls, John, 1971, A Theory of Justice, Cambridge, MA, The Belknap Press of
Harvard University Press.
Riordan, Michael, 1990, What is vertical integration?, ch. 5 in Masahiko Aoki, Bo
Gustafsson, and Oliver E. Williamson, eds., The Firm as a Nexus of Treaties,
London, Sage Publications, 94–111.
Robinson, Joan, 1967, The Soviet collective farm as a producer cooperative: Com-
ment, American Economic Review 57(1), March, 222–223.
Rock, Edward B. and Michael L. Wachter, 1999, Tailored claims and governance:
The fit between employees and shareholders, ch. 4 in Margaret M. Blair and
Mark J. Roe, eds., Employees and Corporate Governance, Washington, DC,
Brookings Institution Press, 121–159.
Roe, Mark J., 1999, Codetermination and German securities markets, ch. 6 in
Margaret M. Blair and Mark J. Roe, eds., Employees and Corporate Governance,
Washington, DC, Brookings Institution Press, 194–205.
Roelants, Bruno, 2000, Worker co-operatives and socio-economic development:
The role of meso-level institutions, Economic Analysis 3(1), February, 67–83.
Roemer, John E., 1988, Free to Lose: An Introduction to Marxist Economic
Philosophy, Cambridge, MA, Harvard University Press.
308 References
Roemer, John E., 1994, A Future for Socialism, Cambridge, MA, Harvard Uni-
versity Press.
Rosen, Sherwin, 1984, Commentary: In defense of the contract at will, University
of Chicago Law Review 51(4), Fall, 983–987.
Rosenberg, Nathan and L. E. Birdzell Jr., 1986, How the West Grew Rich: The
Economic Transformation of the Industrial World, New York, Basic Books.
Rothschild, Michael and Joseph E. Stiglitz, 1976, Equilibrium in competitive insur-
ance markets: An essay on the economics of imperfect information, Quarterly
Journal of Economics 90(4), November, 630–649.
Russell, Raymond, 1985a, Sharing Ownership in the Workplace, Albany, NY, State
University of New York Press.
Russell, Raymond, 1985b, Employee ownership and internal governance: An
“organizational failures” analysis of three populations of employee-owned
firms, Journal of Economic Behavior and Organization 6(3), September,
217–241.
Sacks, Stephen R., 1983, Self-Management and Efficiency: Large Corporations in
Yugoslavia, London, Allen and Unwin.
Salanie, Bernard, 1997, The Economics of Contracting: A Primer, Cambridge,
MA, MIT Press.
Sappington, David E. M., 1991, Incentives in principal-agent relationships, Journal
of Economic Perspectives 5(2), Spring, 45–66.
Schlicht, Ekkehart and Carl Christian von Weizsäcker, 1977, Risk financing
in labour managed economies: The commitment problem, Zeitschrift für die
gesamte Staatswissenschaft 133, special issue, 53–66.
Schultze, Charles, 2000, Has job security eroded for American workers?, ch. 2 in
Margaret M. Blair and Thomas A. Kochan, eds., The New Relationship: Human
Capital in the American Corporation, Washington, DC, Brookings Institution
Press, 28–65.
Sennett, Richard and Jonathan Cobb, 1972, The Hidden Injuries of Class, New
York, Knopf.
Sertel, Murat R., 1982, A rehabilitation of the labor-managed firm, ch. 2 in Workers
and Incentives, Amsterdam, North-Holland, 11–26.
Sertel, Murat R., 1987, Workers’ enterprises are not perverse, European Economic
Review 31(8), December, 1619–1625.
Sertel, Murat R., 1991, Workers’ enterprises in imperfect competition, Journal of
Comparative Economics 15(4), December, 698–710.
Shapiro, Carl and Joseph E. Stiglitz, 1984, Equilibrium unemployment as a worker
discipline device, American Economic Review 74(4), September, 433–444.
Simon, Herbert, 1951, A formal theory of the employment relationship, Econo-
metrica 19(1), January, 293–305.
Skalicky, K., 1975, The Catholic Church and workers’ participation, ch. 8 in
Jaroslav Vanek, ed., Self-Management: Economic Liberation of Man, Baltimore,
MD, Penguin Books, 110–126.
Skillman, Gilbert L. and Harl E. Ryder, 1993, Wage bargaining and the choice
of production technique in capitalist firms, ch. 13 in Samuel Bowles, Herbert
Gintis, and Bo Gustafsson, eds., Markets and Democracy: Participation, Ac-
countability and Efficiency, New York, Cambridge University Press, 217–227.
References 309
Smith, Stephen C., 1984, Does employment matter to the labour-managed firm?
Some theory and an empirical illustration, Economic Analysis and Workers’
Management 18(4), 303–318.
Smith, Stephen C., 1985, Political behavior as an economic externality: Econo-
metric evidence on the relationship between ownership and decision making
participation in U.S. firms and participation in community affairs, in Derek C.
Jones and Jan Svejnar, eds., Advances in the Economic Analysis of Participatory
and Labor-Managed Firms: A Research Annual 1, Greenwich, CT, JAI Press,
123–136.
Smith, Stephen C., 1991, On the economic rationale for codetermination law,
Journal of Economic Behavior and Organization 16(3), December, 261–281.
Smith, Stephen C., 1994, Innovation and market strategy in Italian industrial
cooperatives: Econometric evidence on organizational comparative advantage,
Journal of Economic Behavior and Organization 23(3), May, 303–320.
Staber, Udo, 1989, Age dependence and historical effects on the failure rates
of worker cooperatives: An event-history analysis, Economic and Industrial
Democracy 10(1), February, 59–80.
Staber, Udo, 1993, Worker cooperatives and the business cycle: Are cooperatives
the answer to unemployment?, American Journal of Economics and Sociology
52(2), April, 129–143.
Steensgaard, Niels, 1982, The Dutch East India Co. as an institutional innovation,
ch. 9 in Maurice Aymard, ed., Dutch Capitalism and World Capitalism, New
Stephen, Frank H., 1984, The Economic Analysis of Producers’ Cooperatives,
London, Macmillan Press.
Stiglitz, Joseph E., 1985, Credit markets and the control of capital, Journal of
Money, Credit, and Banking 17(2), May, 133–152.
Stiglitz, Joseph E., 1987, The causes and consequences of the dependence of
quality on price, Journal of Economic Literature 25(1), March, 1–48.
Stiglitz, Joseph E., 1994, Whither Socialism, Cambridge, MA, MIT Press.
Stiglitz, Joseph E. and Andrew Weiss, 1981, Credit rationing in markets with
imperfect information, American Economic Review 71(3), June, 393–410.
Svejnar, Jan, 1981, Relative wage effects of unions, dictatorship and codetermina-
tion: Econometric evidence from Germany, Review of Economics and Statistics
63(2), May, 188–197.
Svejnar, Jan, 1982a, Employee participation in management: Bargaining power
and wages, European Economic Review 18(3), July, 291–303.
Svejnar, Jan, 1982b, West German co-determination, ch. 10 in Frank H. Stephen,
ed., The Performance of Labour-Managed Firms, London, Macmillan Press,
214–225.
Svejnar, Jan, 1982c, Codetermination and productivity: Empirical evidence from
the Federal Republic of Germany, ch. 10 in Derek C. Jones and Jan Svejnar,
eds., Participatory and Self-Managed Firms: Evaluating Economic Performance,
Lexington, MA, D. C. Heath, 199–212.
Szostak, Richard, 1989, The organization of work: The emergence of the fac-
tory revisited, Journal of Economic Behavior and Organization 11(3), May,
343–358.
310 References
Teece, David, Richard Rumelt, Giovanni Dosi, and Sidney Winter, 1994,
Understanding corporate coherence: Theory and evidence, Journal of Eco-
nomic Behavior and Organization 23(1), January, 1–30.
Thomas, Henk, and Chris Logan, 1982, Mondragon: An Economic Analysis, pub-
lished in co-operation with the Institute of Social Studies at The Hague, London,
George Allen & Unwin.
Vanek, Jaroslav, 1970, The General Theory of Labor-Managed Market Economies,
Ithaca, NY, Cornell University Press.
Vanek, Jaroslav, 1971a, Some fundamental considerations on financing and the
form of ownership under labor management, Cornell Department of Eco-
nomics Working Paper no. 16, June, reprinted as ch. 8 in 1977, The Labor-
Managed Economy: Essays by Jaroslav Vanek, Ithaca, NY, Cornell University
Press, 171–185.
Vanek, Jaroslav, 1971b, The basic theory of financing of participatory firms, Cornell
Department of Economics Working Paper no. 27, July, reprinted as ch. 9 in 1977,
The Labor-Managed Economy: Essays by Jaroslav Vanek, Ithaca, NY, Cornell
Vanek, Jaroslav, ed., 1975, Self-Management: Economic Liberation of Man,
Baltimore, MD, Penguin Books.
Wadhwani, Sushil B. and Martin Wall, 1990, The effects of profit-sharing on em-
ployment, wages, stock returns and productivity: Evidence from UK micro data,
Economic Journal 100(399), March, 1–17.
Walzer, Michael, 1983, Spheres of Justice: A Defense of Pluralism and Equality,
New York, Basic Books.
Ward, Benjamin, 1958, The firm in Illyria: Market syndicalism, American Eco-
nomic Review 48(4), September, 566–589.
Weber, Max, 1946, Bureaucracy, in Hans H. Gerth and C. Wright Mills, trans.,
From Max Weber: Essays in Sociology, New York, Oxford University Press,
196–244.
Weiss, Andrew, 1990, Efficiency Wages, Princeton, NJ, Princeton University Press.
Weisskopf, Thomas E., 1993, A democratic enterprise-based market socialism,
ch. 6 in Pranab K. Bardhan and John E. Roemer, eds., Market Socialism: The
Current Debate, New York, Oxford University Press, 120–141.
Weitzman, Martin L., 1980, The “ratchet principle” and performance incentives,
Bell Journal of Economics 11(1), Spring, 302–308.
Weitzman, Martin L. and Douglas L. Kruse, 1990, Profit sharing and productivity,
in Alan Blinder, ed., Paying for Productivity, Washington, DC, Brookings
Weitzman, Martin L. and Chenggang Xu, 1994, Chinese township-village en-
terprises as vaguely defined cooperatives, Journal of Comparative Economics
18(2), April, 121–145.
Whyte, William F. and Kathleen K. Whyte, 1991, Making Mondragon: The Growth
and Dynamics of the Worker Cooperative Complex, 2d edition, Ithaca, NY, ILR
Press, an imprint of Cornell University Press.
Wiener, Hans, with Robert Oakeshott, 1987, Worker-Owners: Mondragon Revis-
ited, A project of the Anglo-German Foundation for the Study of Industrial
Society, London, Anglo-German Foundation.
References 311
Williamson, Oliver E., 1975, Markets and Hierarchies, New York, Free Press.
Williamson, Oliver E., 1979, Transaction-cost economics: The governance of
contractual relations, Journal of Law and Economics 22(2), October, 233–261.
Williamson, Oliver E., 1980, The organization of work, Journal of Economic
Behavior and Organization 1(1), March, 5–38.
Williamson, Oliver E., 1981, The organization of work: Reply, Journal of Eco-
nomic Behavior and Organization 2(3), September, 281–283.
Williamson, Oliver E., 1983a, Technology and the organization of work: A reply
to Jones, Journal of Economic Behavior and Organization 4(1), March, 57–62.
Williamson, Oliver E., 1983b, Technology and the organization of work: A rejoin-
der, Journal of Economic Behavior and Organization 4(1), March, 67–68.
Williamson, Oliver E., 1985, The Economic Institutions of Capitalism, New York,
Free Press.
Williamson, Oliver E., 1987, Transaction cost economics: The comparative con-
tracting perspective, Journal of Economic Behavior and Organization 8(4),
December, 617–625.
Williamson, Oliver E., 1988, Corporate finance and corporate governance, Journal
of Finance 43(3), July, 567–591.
Williamson, Oliver E., 1991, Comparative economic organization: The analysis
of discrete structural alternatives, Administrative Science Quarterly 36(2), June,
269–296.
Willman, Paul, 1982, Opportunism in labour contracting: An application of the
“organisational failures” framework, Journal of Economic Behavior and Orga-
nization 3(1), March, 83–98.
Winter, Sidney G., 1991, On Coase, competence, and the corporation, ch. 11 in
Oliver E. Williamson and Sidney G. Winter, eds., The Nature of the Firm: Ori-
gins, Evolution, and Development, New York, Oxford University Press, 179–195.
Zevi, Alberto, 1982, The performance of Italian producer cooperatives, ch. 12 in
Derek C. Jones and Jan Svejnar, eds., Participatory and Self-Managed Firms:
Evaluating Economic Performance, Lexington, MA, D. C. Heath and Company,
239–251.
Zingales, Luigi, 1995, What determines the value of corporate votes?, Quarterly
Journal of Economics 110(4), November, 1047–1073.
Zusman, Pinhas, 1992, Constitutional selection of collective-choice rules in a
cooperative enterprise, Journal of Economic Behavior and Organization 17(3),
May, 353–362.
Index
accountability, and authority assets and asset ownership:

relationships in LMFs, 36 alienability of, 158, 236; bargaining
adverse selection: and asset and investment, 168–72; incentives
ownership, 168; and debt financing, and information, 166–8; leasing of,
187; and firm formation, 209, 211; 166–8, 245–6; public and private,
and information, 15; and 2–4; and theory of the firm, 110–14
membership markets, 160; and asset-specificity hypothesis, 169–73
optimal contracting, 127; and association, firm as, 108–109
structure of LMFs, 129–32 asymmetrical authority structures,
advertising firms, 50 99–100
Aghion, Philippe, 101 authority: and dignity or self-respect
agriculture, and risk aversion, 199 in capitalist firm, 36–7; and
Alchian, Armen A., 97, 166, 167, 172, investor-controlled firm, 6; and
173–5, 179, 180–1 social custom, 8; and theory of the
Aldrich, Howard, 208 firm, 92–101; and workplace
Algoma Steel, 213–15, 219–20 democracy, 33, 278–9
alienability: of assets, 158, 236; of
capital, 17, 258; of right to workers’ Baker, George, 101, 133
control, 42. See also inalienability banks and banking: and credible
Ammirato, Piero, 67, 68, 70, 72, 74 commitment toward capital, 247;
Anacortes Veneer, 51 and democratic governance, 190;
Andersen Consulting, 50 and professional partnerships, 50.
anti-commodification, and labor, 35 See also capital markets
appropriation: and control rights, 253; Bardhan, Pranab K., 106–107, 287
and firm formation, 211–12 bargaining: and asset ownership,
Archer, Robin, 29–30 168–72; and collective choice, 200;
Arizmendi-Arrieta, Don Jose Maria, LMFs and short-term, 13; market
57–8, 66 and short-term, 94; and optimal
Arneson, Richard J., 28, 41 contracting, 128–9. See also
Arrow, Kenneth J., 97 collective bargaining
313
314 Index
Bartlett, Will, 74 capitalism, and asset ownership, 2–4

base wages, 280–1 capital-managed firms (KMFs): and
Basque Nationalist Party, 65 community, 38; and degeneration,
Baums, Theodor, 88 221–4; and equality, 24–5; and
Bellas, Carl J., 50, 52, 53, 55, 56 investor takeovers of LMFs, 224–6;
Benelli, Giuseppe, 88, 89 and locus of control, 102–107; and
Benham, Lee, 138, 139, 204 organizational demography, 207;
Ben-Ner, Avner, 47, 48, 83, 114, 182, and symmetry principle, 118;
196, 207–208, 210, 213, 218, 219, 222, systematic differences between
226, 228, 232, 253, 254, 288 LMFs and, 10–12; and theory of
Berle, Adolf A., 5, 104–105 firm, 5, 6; transformation of into
Berman, Katrina V., 50, 51, 52–3, 55, LMFs, 43
56–7, 162, 190, 209–10, 245 capital markets, and obstacles to
Berman, Matthew D., 162, 190 formation of LMFs, 209–10.
Blair, Margaret M., 79–80, 227–8 See also banks and banking;
Blasi, Joseph R., 79, 80, 82, 83, stock market
227–8 case studies. See Algoma Steel;
boards of directors: choice of, 101; and codetermination; Lega federation;
employee stock ownership plans, 82; Mondragon system; plywood
and labor trust, 264, 265, 266–7, 274, cooperatives; United Airlines
276; and theory of firm, 5 Catholic Church, 35
Bonin, John P., 144, 162–3, 183, 188, Catholic National Confederation, 69
194, 198, 229, 232 CFI (Campagnia Finanziara
Bowles, Samuel, 28–9, 39, 41, 42, 97, Industriale), 71
137, 189–90, 194, 196, 253, 255 Chaplinsky, Susan, 221, 228
Bradley, Keith, 57, 59–60, 62, 65 Cheney, George, 57, 64
Brown, Judith, 213, 231–2 China, and cultural explanations for
Browning, Martin, 219, 220 cooperation, 139
Burns, W. Allen, 83, 288 class, and market socialism,
business cycles, 219, 228–30. See also 37–8
life cycles coalitions: authority and employees,
96; firms as, 108–109
Cable, John, 74 Coase, Ronald H., 96
Caja Laboral Popular (CLP), 58, 61, Coase Theorem, 13, 96
65, 66 Cobb, Jonathan, 37
Canada: business cycles and codetermination (Germany), and case
producers’ cooperatives, 229; and studies of LMFs, 83–91, 275
clusters of cooperatives, 231–2; and collective-action problems, and
failure rates for workers’ employee buyouts, 220
cooperatives, 227; and formation collective asset ownership, 114
rates for workers’ cooperatives, 208; collective bargaining, 103, 200.
and worker takeovers, 213, 214, 220 See also bargaining
capital: alienability of, 17, 258; and collective choice: and decision-making
constraints, 185–6, 189–93; intensity in LMFs, 200–204; and explanations
of, 87, 288; mobility and expansion for rarity of workers’ control,
of workers’ control, 289. See also 204–206. See also preference
capital markets heterogeneity
Index 315
commitment problem: asymmetries of, Craig, Ben, 53, 55, 157, 158, 161, 162,
236–8; labor and credible, 240–5; 183–4, 223
and LMFs compared with KMFs, Crawford, Robert, 172
14–15, 17 credible commitment: toward capital,
commodification: asymmetries of, 245–9; toward labor, 240–5
239–40; and control rights, 17, 251–3 culture: and explanations for LMFs,
commodity, and firm as association or 138–9; and success of Mondragon in
coalition, 108–109 1940s and 1950s, 64–5
common property problems, and Cuomo, Gaetano, 153, 287
membership markets, 153–4, 155
Communist Party (Italy), 70 Dahl, Robert A., 27, 31, 41
community: and democracy, 30–1; and debt, and credible commitment
theory of workers’ control, 38–41 toward capital, 246–7
compensation, and transformation of debt financing, 186–7
KMFs into LMFs, 43 decision making, and collective choice,
compliance, and authority, 98 200–206
composition: asymmetries of, 238; of degeneration, and LMFs, 221–4
control groups, 16, 17 depreciation, and leasing of assets, 166
consulting partnerships, 50 Defourney, Jacques, 47
consumer cooperatives, 101 democracy: and efficiency goals, 42;
Conte, Michael, 49, 229, 230 and managerial authority, 278–9;
control, and theory of firm, 4–8. See and minority interests, 278; as public
also control groups; control rights good, 41; and theory of workers’
control groups: and authority, 100; control, 27–32. See also voting
composition of, 238, 249–51; and Demsetz, Harold, 97, 166, 167, 173–5,
LMFs compared to KMFs, 16, 17; 179, 180–1
and locus of control, 104; and den Broeder, C., 87, 90
theory of the firm, 93 Denmark, and capital-labor ratios in
control rights: and commitment LMFs, 190
asymmetries, 237; and dignity, and workers’ control, 34–8
commodification, 251–3; and LMFs diversification hypothesis, and
compared to KMFs, 10; and theory membership markets, 157
of firm, 5–8, 110. See also ownership dividends, and labor trust, 263–4, 280
rights; property rights Dofasco, 213, 214
cooperatives: clusters of, 230–3; and Domar, Evsey, 144
financial arrangements, 48–9; and Doucouliagos, Chris, 183
locus of control, 102; and modern Dow, Gregory K., 83, 136, 150, 154,
economic systems, 47–8; and rate of 165, 172–3, 202, 220, 225, 255, 288
firm formation, 208. See also Lega Drèze, Jacques H., 118, 185, 194
federation; Mondragon system;
plywood cooperatives; producers’ Earle, John, 67, 70
cooperatives econometrics: and alternative
coordination, and definition of the behavioral hypotheses for LMFs,
firm, 93 143; and differences in objectives of
Cornforth, Chris, 226 cooperatives and conventional
corporate charter, and control rights, 7 firms, 162; and studies of French and
corporations, history of, 287 Italian cooperatives, 182
316 Index
economics: literature and research on equity shares, and labor trust, 263
workers’ control, 8–12; workers’ equivalence theorems, 118
control and systems of, 1–4. See also ESOPs. See employee stock
econometrics ownership plans
education, and expansion of workers’ Estrin, Saul, 3, 46, 47, 74–5, 163, 210,
control, 289 224, 228–9, 287
efficiency: and adverse selection, 130; Eswaran, Mukesh, 186
and codetermination, 90; and Europe: cooperatives and
comparison of KMFs and LMFs, 11; membership markets, 156; and
and labor trust, 270–1; moral and financial arrangements of
philosophical concerns, 41–4; of cooperatives, 48–9; and history of
optimal contracts, 129; and corporations, 287; and rates of firm
transaction costs, 123, 124–5 formation, 207–208; and rates of
effort: and optimal contracting, 127; firm survival, 226; and worker
and work incentives, 173, 175 takeovers, 213. See also Denmark;
Ellerman, David P., 32–4, 35, 41, France; Italy; Netherlands; Spain;
135, 153 Sweden; United Kingdom
employee buyouts: and collective Everett, Michael J., 136
choice, 205–206; and free-rider exit costs, 29–30, 43
problems, 17; and labor trust, 271.
See also worker takeovers Fagor group, 64
Employee Retirement Income federation, and labor trust, 272–3, 281.
Security Act (ERISA), 76 See also Lega Federation
employees: authority and theory of Federation of Italian Cooperatives,
the firm, 94; and control of firm, 5–6; 68
employee stock ownership plans Fehr, Ernst, 150
and voting rights of, 78; and funding Fincooper financial consortium, 72
of employee stock ownership plans, firm: and asset ownership, 110–14; and
81; LMFs and equality of, 25; and authority, 98–101; formation rates
risk, 115. See also employee for LMFs, 207–12; and locus of
buyouts; employee stock ownership control, 101–107; and ownership,
plans; labor; wages; workers’ 107–10; and residual claims, 114–16;
control as set of incomplete contracts, 258;
employee stock ownership plans survival rates for LMFs, 226–8;
(ESOPs): and capital constraints, theory of, 4–8, 92–8. See also
191; and case studies of LMFs, capital-managed firms;
76–83; and characteristics of firms, labor-managed firms
45–6; and labor productivity, 182; FitzRoy, Felix R., 88
and labor trust, 267, 285; and Folk Theorem, 134, 175, 242
portfolio diversification, 197; and Fortune 500, 79
stock market, 81; and worker France: and cooperatives, 48, 156, 163,
takeovers, 217. See also pension 182, 190, 232; and history of
programs corporations, 287; number of
entrepreneurship, Knightian theory of, worker-controlled firms in, 47; and
179 rates of firm formation, 208; and
equality, and LMFs, 24–7 rates of firm survival, 226
equity financing, 188–9, 248–9 freedom, and relational values, 39
Index 317
Freeman, Richard B., 90 Han, Tzu-Shian, 182

free-rider problems: and asset Hansmann, Henry, 49, 50, 75, 89–90,
maintenance, 167; and control 121, 124–5, 135, 173, 181, 199,
groups, 16; and employee buyouts, 200–201, 204, 205, 215, 225, 250, 255,
17; and firm formation, 209; and 256
investor takeovers, 159 Hart, Oliver, 126, 129, 133, 169–72,
free speech and assembly, rights to, 172–3, 186
266 Hassel, Anke, 90
Frick, Bernd, 88 hierarchy. See class
Fudenberg, Drew, 168 history: of corporations, 287; of
Furubotn, Eirek G., 152 experiments with workers’ control,
46–7, 57. See also intellectual history
Gamson, Zelda F., 139 Holmstrom, Bengt, 95, 114, 133, 134,
Garvey, Gerald, 94 168, 175–6, 177, 179–80
Gelauff, George M. M., 87, 90 Holmstrom, Mark, 68
Gelb, Alan, 57, 59–60, 62, 65 horizon problem, and membership
General Motors, 109 markets, 152–3, 154, 155
George, Donald A. R., 3, 190 Horvat, Branko, 3
Germany. See codetermination Hutterite religious communes, 138
Gibbons, Robert, 101, 133
Gintis, Herbert, 28–9, 39, 41, 42, 97, Illyrian model, 143–7, 151, 161–3,
137, 189–90, 194, 196, 253, 255 252
Gordon, Jeffrey N., 215, 216, inalienability, of labor, 17, 18, 157–8,
218, 255 234–5, 236, 238, 249, 251–3, 258.
governance structure: classification of, See also alienability
111t, 113t; and codetermination, incentives: and asset ownership,
84–6; and control of firm, 6–7; of 166–8; authority and theory of the
Lega federation, 73; of Mondragon firm, 95; and optimal contracting,
system, 58–9, 62–3; and path 127–8; and risk aversion, 173–9; and
dependence, 137; of plywood workers’ control, 179–84
cooperatives, 54; and transaction income: LMFs and maximization of,
costs, 121–4; use of term, 93. 142–3; LMFs and redistribution of,
See also management; 44. See also wages
organization incomplete contracts, 258
government. See legal systems; policy; Industrial Common Ownership
taxation Movement, 208
Great Depression, and cooperatives, Industrial Revolution: dominance of
230 capitalist firm during, 211; and
Greenberg, Edward S., 31, 50, 53, experiments with workers’ control,
55, 56 46–7; pin-making and transaction
Grossman, Sanford, 126, 169–72, costs, 121–4; and unequal
172–3 distribution of wealth, 137
Gui, Benedetto, 186, 194 information: and asset ownership,
Gunderson, Morley, 214 166–8; and LMFs compared to
Gunn, Christopher Eaton, 32, 35, 230, KMFs, 14–15
232–3 Institute for Cooperative Credit, 69
Gurdon, Michael A., 88 institutional economics, 115
318 Index
insurance: and Lega federation, 72; Kreps, David, 139

Mondragon and internal system of Kruse, Douglas L., 79–80, 82, 83, 181,
unemployment, 62; and share 182, 227–8
ownership in cooperatives, 54 KT union, 64
intellectual history, and LMFs, 253–6
internal enforcement, and repeated labor: and credible commitment,
games, 133 240–5; inalienability of, 17, 18,
investments, and asset ownership, 157–8, 234–5, 236, 238, 249, 251–3,
168–72 258; intensity of, 284. See also
investors: and control of firm, 6; and employees; labor markets; labor
professional partnerships, 50. See unions; layoffs; plant closures;
also shareholders productivity; wages
investor takeovers, of LMFs, 159, labor directors, 264, 265, 266–7, 276.
224–6 See also boards of directors
Ireland, Norman J., 144 laborism, and economic systems, 3, 4
isomorphism theorems, 118 labor-managed firms (LMFs): and
Italy: and cooperatives, 48, 75–6, 156, adverse selection, 129–32;
182, 232; and funding of worker alternative hypotheses for, 140–1;
takeovers of failing firms, 71; and asset ownership, 166–72; and
numbers of worker-controlled firms, business cycles, 228–30; conditions
47, 67; postwar constitution and for viability of, 256–7; construction
cooperatives, 69–70; and rates of of theory of, 10–12; and credible
firm survival, 226. See also Lega commitment toward capital, 245–9;
federation cultural explanations for, 138–9; and
degeneration, 221–4; distribution of
James, Estelle, 194 across industries, 10; and employee
Japan, profit-sharing and labor stock ownership plans, 81, 82; and
productivity, 181 equality, 24–7; and firm formation
Jensen, Michael C., 86, 153, 155, 194, rates, 207–12; and firm survival
210, 243, 255 rates, 226–8; and Illyrian model,
Jones, Derek C., 47, 49, 74, 162–3, 182, 143–7, 161–3; and intellectual
183, 188, 194, 198, 223, 224, 229, 230, history, 253–6; and investor
232 takeovers, 224–6; and labor market,
Jossa, Bruno, 153, 287 25, 148–52; labor-productivity
Joudrie, Earl, 214 advantages of, 240–5; and locus of
Jun, Byoung, 218, 219 control, 102–107; and maximization,
Jürgens, Ulrich, 89, 91 142–3; and membership markets,
148–56; moral and philosophical
Keefer, Philip, 138, 139, 204 benefits of, 41–4; and opportunity
Klein, Benjamin, 172 cost of labor, 163–4; and optimal
KMFs. See capital-managed firms contracting, 126–9; and
Knight, Frank H., 97, 178–9 organizational demography, 207;
Knights of Labor, 230 and path dependence, 135–8;
Kotwal, Ashok, 186 plywood cooperatives as case study
Kovenock, Dan, 218 of, 50–7; and repeated games, 132–5;
Kraft, Kornelius, 88 and replication principle, 119–21;
Kramer, Gerald H., 202, 205 and stock market, 152–6; and
Index 319
symmetry principle, 117–19; and Luzzatti, Luigi, 68–9

synthesis of theoretical ideas, 12–18; Lys, Thomas, 88, 89
and theory of firm, 5; and
transaction costs, 121–6; and worker MacLeod, W. Bentley, 176–7
takeovers, 212–21. See also firm; macroeconomic externalities, and
workers’ control path dependence, 137
labor markets: and adverse selection, Mäki, Uskali, 140, 141
130; and dignity, 35; and Illyrian management: and buyouts, 221, 228;
model, 146–7; and membership codetermination of boards of, 84–6,
markets of cooperatives, 158, 163–4; 91; and control of firm, 5; of
and structure of LMFs, 25, Mondragon system, 60; of plywood
148–52 cooperatives, 55; workers’ control
labor shares. See membership markets and problems in, 276–9. See also
labor trust, as strategy to expand governance structure; organization
workers’ control, 263–73, 274, 276 Marglin, Stephen A., 121, 210, 211, 255
labor unions: and collective marketing, and clusters of
bargaining, 103; and worker cooperatives, 231
takeovers, 220 markets: failures of, 44, 271; repeated
lawyers, and professional partnerships, games and enforcement, 132; and
49–50 short-term bargaining, 94. See also
layoffs: and credible commitment to labor market; membership markets;
labor, 244; and Mondragon group, stock market
62. See also plant closures market socialism, 37, 287
Law, Peter J., 144 Marx, Karl, 97
Lazear, Edward P., 90 Mason, Ronald M., 30–1, 32, 38
leasing, of assets, 166–8, 245–6 Matthews, Steven, 177
Lega federation, 67–76, 232, 272–3 McAfee, R. Preston, 177
legal systems: authority and theory of McMillan, John, 177
the firm, 95; cooperatives and Meade, James, 194, 198
Spanish, 58, 60; LMFs and Means, Gardiner, 5, 105
economic history, 135–6 Meckling, William, 86, 153, 155, 194,
Le Grand, Julian, 287 210, 243, 255
Legros, Patrick, 177 medical partnerships, 50
leveraging, and employee stock membership markets: and
ownership plans, 76–7, 78 commodification of control rights,
Levin, Henry M., 139, 187 239, 252; and degeneration, 223;
Levine, David I., 137 imperfect, 156–61; and labor
Lewchuk, Wayne, 220 market, 148–52; and opportunity
life cycles, of firms, 222–3. See also cost of labor, 163–4; and stock
business cycles market, 152–6; and structure of
Lindblom, Charles E., 135 LMFs, 17, 35–6; and trading of jobs,
LMFs. See labor-managed firms 279–82
locus of control, and theory of the Microsoft, 103
firm, 101–107 Milgrom, Paul R., 96, 159, 168, 179,
Loderer, Claudio, 88, 89 180, 183
Logan, Chris, 57, 59, 66 Miller, David, 28, 37–8, 41
Lucas, J. J., 50 Miller, Gary J., 139
320 Index
Minkler, Alanson P., 136 organization, and repeated games,

minority employee ownership, 79 134–5. See also governance
minority interests, and democratic structure; management
processes, 278 organizational demography, 207
Miyazaki, Hajime, 222, 225 Ostrom, Elinor, 167
Moene, Karl Ove, 219 ownership, and theory of the firm,
Mondragon system: and capital 107–10. See also asset ownership
accounts, 48, 155; and capital ownership rights, 234. See also
constraints, 190–1, 193; as case study property rights
of LMF, 57–66, 272–3; and clusters
of cooperatives, 232; and collective Pagano, Ugo, 136
choice, 205; cultural explanations participatory democracy, theory of,
for, 138; and locus of control, 105; 30–2
number of employees, 47 partnerships: and control of firm, 6–7;
monitoring, and work incentives, and Lega federation, 72; as
174 worker-controlled enterprises,
Montan industries, 85–6, 88 49–50
Montias, John M., 194 Pateman, Carole, 27, 30, 31, 32,
Moore, John, 126, 169–72, 172–3, 36–7
186 path dependence, 135–8
moral hazard: and adverse selection, Pejovich, Svetozar, 89, 152, 155
129–30; and asset maintenance, 167; Pencavel, John, 53, 55, 139, 157, 158,
and control groups, 15; and debt 161, 162, 183–4, 190, 205, 223
financing, 186, 188; and optimal Peninsula Plywood Corporation, 51
contracting, 129 pension programs: and employee
Multnomah plywood cooperative, 51 stock ownership plans, 80; and
Murphy, Kevin J., 101, 133, 199 worker buyouts, 221. See also
employee stock ownership
Nash equilibrium, 136 plans
National Center for Employee Pérotin, Virginie A., 47, 226,
Ownership (NCEO), 79 229–30
Naumann, Katrin, 89, 91 personal right, and membership in
near parity, and codetermination, worker-controlled firm, 151–2
85, 88 Pistor, Katharina, 91
Netherlands, and rates of firm Pittatore, Silvia, 75, 204
formation and survival, 208, 226 plant closures: and credible
Neuberger, Egon, 194 commitment to labor, 244–5; and
“nexus of contracts” approach, to worker takeovers, 213. See also
authority, 97 layoffs
Niehaus, Greg, 221, 228 plywood cooperatives: as case studies
Nozick, Robert, 26, 37 of LMFs, 50–7; and clusters of
Nuti, Domenico Mario, 154 cooperatives, 230; and collective
choice, 205; cultural explanations
Oakeshott, Robert, 70 for, 138, 190; and degeneration,
Ognedal, Tone, 202 223–4; and democracy, 31; and firm
Olympia Veneer, 50–1, 53, 223, 231 formation, 209–10; and Illyrian
optimal contracting, 126–9 model, 161–2; and locus of control,
Index 321
104; and membership markets, 106, Putterman, Louis, 83, 108–109, 122,
156, 157, 158; number of members, 144, 148, 158–9, 162–3, 165, 183, 188,
49; and profit maximization, 164; 194, 196, 198, 210, 229, 232, 245, 253,
sale of shares in, 48 255, 288
policy, and conversion of KMFs into
LMFs, 26–7. See also legal systems; Quarter, Jack, 213, 231–2
taxation quasi-rents, 14, 30, 169. See also rents
portfolio diversification, 185,
193–9 Radner, Roy, 97
preference heterogeneity, and Rai, Anoop, 88
collective choice, 202–3 ratchet effect, 14
principal-agent theory: and adverse Rawls, John, 39
selection, 132; and optimal referendum procedures, and labor
contracting, 126, 128; and residual trust, 265–7, 275
claims, 115; and work incentives, reforestation cooperatives, 49, 230–1,
178–9 232–3
prisoners’ dilemma game, 167 rent-appropriation hypothesis, 212,
private asset ownership, 2, 3, 4 252–3
producers’ cooperatives: and business rents, LMFs and product-market, 26.
cycles, 229; as examples of LMFs, See also quasi-rents
10, 18; and locus of control, 102. repeated games, 132–5
See also cooperatives replication principle, 119–21
productivity: and credible residual claims: and capitalist firms,
commitment toward labor, 240–5; 254; and control rights, 110; and
and LMFs compared to KMFs, theory of the firm, 114–16
14. See also total factor risk and risk aversion: and equity
productivity financing, 249; and membership
profit: LMFs and maximization of, markets, 157; and path dependence,
11–12; and ownership of firm, 137; and portfolio diversification,
109–10 194–5, 199; and residual claims, 115;
profit-sharing: and labor productivity and work incentives, 173–9
of KMFs, 120, 181; and portfolio Roberts, John, 96, 159, 183
diversification, 198 Roemer, John E., 106–107, 118, 287
propaganda, and referendums on Rowthorn, Robert, 136
labor trusts, 266 Rupp, Joachim, 89, 91
property, and workplace democracy, Russell, Raymond, 49, 138, 182, 223
32–4 Ryder, Harl E., 136
property rights: and asset ownership,
114; and authority, 97; and control Sacks, Stephen R., 3
rights, 7, 253; and optimal sanitation cooperatives, 48
contracting, 129. See also control scavenger cooperatives, 138, 156, 231
rights Schlicht, Ekkehart, 154, 186, 194
public asset ownership, 2, 3, 4 search cost, and firm formation, 209
public goods: and collective choice, self-directed teams, 288
202–203; and workplace democracy, self-management, Yugoslav system of
41 workers’, 3, 102, 144, 147
Puget Sound Plywood, 51 Sennett, Richard, 37
322 Index
Sertel, Murat R., 150, 255 taxation: and efficiency goals, 44; and
Seymour Specialty Steel, 82 employee stock ownership plans,
shadow value, of firm, 280–1 77–8, 80; financing of LMFs by
shareholders: and control of firm, 6; progressive, 25; government
and locus of control, 104–105; subsidies and workers’ control, 284;
voting and capital constraints, 192; and plywood cooperatives, 53.
workers’ control and reassurance of, See also policy
273–6. See also investors taxi cooperatives, 138
shutdowns. See plant closures technology: and expansion of workers’
signaling: and adverse selection, 131; control, 288; and path dependence,
and credible commitment to labor, 136
244; and debt financing, 187 Thomas, Henk, 57, 59, 226
Simon, Herbert, 97 Tirole, Jean, 95, 101
Skillman, Gilbert, 136, 202, 220, Topel, Robert, 199
225, 255 total factor productivity, 180
Smith, Adam, 121 trade associations, and clusters of
Smith, Stephen C., 31, 74, 90 cooperatives, 231
social custom, and control rights, 7, 8 transaction costs: and adverse
socialism: and asset ownership, 2–4; selection, 132; and asset ownership,
and property rights, 287 169; and structure of LMFs,
social surplus, and residual 121–6
claims, 116 Turati, Gilberto, 75, 204
Spain: and banking laws, 66;
cooperatives and legal system, 58, Ulgor, 57–8, 61–2
60; and cultural environment of Ulma group, 64
Basque region, 64–5. See also ultimate control, 5–6
Mondragon system Unintesa, 72
Sparks, Roger, 218 Unipol, 72
Staber, Udo, 227, 229 United Airlines, 103, 172, 191, 215–18,
standard industrial classification 219, 220, 247, 274
(SIC codes), 230 United Kingdom: and clusters of
Stephen, Frank H., 3, 144, 150 cooperatives, 231; and financial
Stern, Robert, 208 arrangements of cooperatives,
stock market, and governance 48–9; and history of corporations,
structures, 106; and public listing, 287; LMFs and Companies
267. See also employee stock Act of 1856, 136; profit-sharing
ownership plans; investors; and labor productivity, 181;
shareholders and rate of firm formation,
supermajority, and democratic 208; and rate of firm survival,
processes, 278 226
supervisory boards, and United States. See plywood
codetermination, 84–6, 91. See also cooperatives; United Airlines
boards of directors United Steelworkers of America,
Svejnar, Jan, 47, 88 214, 232
Sweden, and rate of firm survival,
226 Van de Gucht, Linda, 221, 228
symmetry principle, 117–19 Vanek, Jaroslav, 47, 144, 153
Index 323
vesting rules, and employee stock workers’ control: and asset specificity,
ownership plans, 77 172–3; and capital constraints,
von Weizsäcker, Carl Christian, 154, 189–93; and collective choice,
186, 194 204–206; and community, 38–41; and
voting: and capital constraints, 192; democracy, 27–32; and dignity, 34–8;
and employee stock ownership duration of transition phase leading
plans, 78; and shareholder control of to, 282–7; economic literature,
firm, 6; workers’ control and systems research and case studies of, 8–12,
of, 277–8. See also democracy 45–50; and economic systems, 1–4;
and equality, 24–7; explanations for
wages: and credible commitment to rarity of, 165; future of, 287–9;
labor, 243–4; LMFs and equality, moral and philosophical benefits of,
25–6; and labor trust, 274; and 41–4; and portfolio diversification,
Mondragon system, 60; and value of 197–9; and problems in
labor shares, 280; workers’ control management, 276–9; and property,
and distribution of, 277. See also 32–4; and shareholders, 273–6;
income strategy for expansion of, 262–73; as
Wallerstein, Michael, 219 unitary phenomenon, 256–9; and
Walzer, Michael, 27–8, 35–6, 38, 41 work incentives, 179–84. See also
Ward, Benjamin, 143–4, 146–7 labor-managed firms
wealth: LMFs and redistribution of, workers’ cooperatives. See
44; and membership markets, 156–7; cooperatives
path dependence and distribution worker takeovers, 212–21. See also
of, 137 employee buyouts
Weber, Max, 97 Works Constitution Acts of 1952 and
Weisskopf, Thomas E., 38, 39 1972 (Germany), 85
Weitzman, Martin L., 139
Westman, Edward, 50 Xu, Chenggang, 139
Whyte, William F. and Kathleen K.,
57, 65–6 Yugoslavia, and self-management, 3,
Wiener, Hans, 57, 66 102, 144, 147
Williamson, Oliver E., 97, 121–4, 125,
168–9, 187, 188, 255, 256 Zevi, Alberto, 70
worker councils, 243 Zingales, Luigi, 192
workers. See employees; labor; “zone of acceptance,” and authority,
workers’ control; worker 98
takeovers Zusman, Pinhas, 201

Governing The Firm - Worker's Control in Theory and Practice - Gregory K. Dow (CUP, 2003)

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Governing The Firm - Worker's Control in Theory and Practice - Gregory K. Dow (CUP, 2003)

Загружено:

Авторское право:

Доступные форматы

Workers’ Control in Theory and Practice

Gregory K. Dow is Professor and Chair of the Department of Eco-

Cambridge University Press

© Gregory K. Dow 2003

This book is in copyright. Subject to statutory exception and to the provision of

First published in print format 2003

Cambridge University Press has no responsibility for the persistence or accuracy of

List of Tables and Figures page x

5.2 The Nature of Authority 98

9.7 Collective Choice 200

Most large enterprises in the developed world are controlled by share-

Most of the book is devoted to the explanatory task, in part because it

and November of 2000 at Erasmus University in Rotterdam. My visit to

John Stuart Mill, Principles of Political Economy, Book IV,

1.1 Economic Systems

A prudent worker, investor, or citizen would no doubt ask a few

Table 1.1 Ownership and control as independent dimensions of ﬁrm

must be better. An unfortunate result is that the tensions between these

capitalists – that is, individuals or groups who supply capital to ﬁrms.

1.2 The Control Dimension

Some worker-controlled ﬁrms permit members to sell their positions to

1.3 Looking for Clues

labor-managed ﬁrms are not distributed randomly across industries.

1.4 A Projected Synthesis

by means of binding contracts or some type of self-enforcement. The

marketed structures or inventories that can be used as collateral. Non-

efforts and abilities of their co-workers, and the informational asymme-

diversity in preferences or beliefs. Group size is important because control

therefore construct a theory about differences in the behavior of KMFs

or collective-choice problems. Even granting that some of these stories

1.5 The Plan of the Book

of workers’ control have advanced many moral arguments on its behalf.

advantages of KMFs cannot be replicated within LMFs, and conversely.

problems in attracting outside ﬁnancing, does help explain a number of

histories, and their level of economic development often diverge enor-

2.1 Why Care About Workers’ Control?

expect similarly egalitarian results from an economy in which LMFs

hand, policies to encourage the conversion of KMFs into LMFs need

to increase participation in the workplace” (1982: 78). Some economists

The emphasis given by participatory theorists to educative functions,

human action is inalienable. He then argues that the employment con-

to the manner in which contracts are described. To illustrate, suppose

Ellerman (1992) sees employment as analogous to slavery for reasons of

do by his organisational superiors” (1970: 50). Sennett and Cobb point

such class distinctions, Miller’s emphasis on workers’ control as a partial

Suppose that given a choice between dictatorship with high individual

value on democratic governance, the market will supply it (Nozick, 1974:

2.7 The Author Shows His Cards

to participation in ﬁrm governance, and it is possible that more of the

Society as a whole, using standard political institutions, would set the

decades in the United States through tax subsidies to employee stock

Workers’ Control in Action (I)

3.1 Surveying the Terrain

might deliver a measure of self-determination in the modern economic

focused on clothing, footwear, and printing (Estrin, Jones, and Svejnar,

organization unless the organization is wholly owned by lawyers practic-

3.2 The Plywood Cooperatives

bankruptcy, depletion of nearby timber supplies, labor unrest associated

as coops. In 1990, according to Craig and Pencavel (1995), seven of the

their individual capital investments. Historically, non-working share own-

There is some disagreement in the literature over whether members

favorable during the peak period of 1940–55. These included moderate