Farmer Book 2008

Expectations, Employment and Prices
Roger E. A. Farmer1 November 2005 This draft: May 16, 2008
1 To
Roxanne.
Contents
Preface 1 What this Book is About 1.1 The Nature of the Enquiry . . . . . . . . 1.2 The Theory Summarized . . . . . . . . . 1.3 The Theory of Aggregate Supply . . . . 1.4 The Theory of Aggregate Demand . . . . 1.5 Additional Features of my Interpretation 2 The 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 Labor Market Components of the General Theory . Households . . . . . . . . . . . . . . Firms . . . . . . . . . . . . . . . . . Search . . . . . . . . . . . . . . . . . The Social Planner . . . . . . . . . . Aggregate Demand and Supply . . . Eective Demand and the Multiplier A Denition of Equilibrium . . . . . Remarks on Ricardian Fiscal Policies Concluding Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix 1 3 4 5 7 8 11 12 13 14 16 17 19 22 24 26 27 29 30 32 33 34 36 40
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
3 Aggregate Demand and Supply 3.1 Households . . . . . . . . . . . 3.2 Firms . . . . . . . . . . . . . . 3.3 Search . . . . . . . . . . . . . . 3.4 The Social Planner . . . . . . . 3.5 Aggregate Supply and Demand 3.6 Keynesian Equilibrium . . . . . v
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
vi 3.7 3.8
CONTENTS Keynes and the Social Planner . . . . . . . . . . . . . . . . . . 42 Concluding Comments . . . . . . . . . . . . . . . . . . . . . . 44 47 47 48 49 49 51 51 52 53 54 55 57 59 59 60 60 61 61 63 65 66 69 70 71 73 76 77 78 80 81 83 84 85
4 Saving and Investment 4.1 The Model Structure . . . . . . . . . . . . . . . . . 4.2 Households . . . . . . . . . . . . . . . . . . . . . . 4.2.1 The Initial Old . . . . . . . . . . . . . . . . 4.2.2 The Initial Young . . . . . . . . . . . . . . . 4.2.3 The Third Generation . . . . . . . . . . . . 4.3 Firms . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Search . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 The Social Planner . . . . . . . . . . . . . . . . . . 4.6 Investment and the Keynesian Equilibrium . . . . . 4.7 The Denition of Equilibrium . . . . . . . . . . . . 4.8 Aggregate Demand and Supply . . . . . . . . . . . 4.9 Finding Values for the other Variables . . . . . . . 4.9.1 First Period Price and Output . . . . . . . . 4.9.2 Second Period Capital and Output . . . . . 4.9.3 Rental Rates and Consumption Allocations . 4.9.4 Second Period Prices . . . . . . . . . . . . . 4.10 Fiscal Policy in a Keynesian Model . . . . . . . . . 4.11 Concluding Comments . . . . . . . . . . . . . . . . 5 Presenting Business Cycle Facts 5.1 Whats Wrong with the HP Filter? 5.2 Measuring Data in Wage Units . . 5.3 Choosing a Per-Capita Measure . . 5.4 Interpreting Wage Units . . . . . . 5.5 The Components of GDP . . . . . 5.6 Concluding Comments . . . . . . . 6 The 6.1 6.2 6.3 6.4 6.5 6.6 Great Depression Preferences . . . . . . . . . . The Consumption Function . Technology . . . . . . . . . . Aggregate Supply . . . . . . . Search and the Labor Market Expectations Shocks . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
CONTENTS 6.7 6.8 6.9 6.10 6.11 6.12 6.13 6.14 7 The 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 The Social Planning Problem . . Demand Constrained Equilibrium Keynes and the Great Depression Eciency of Equilibrium . . . . . Household and Government . . . Crowding Out . . . . . . . . . . Concluding Comments . . . . . . Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
vii 85 87 90 92 94 96 100 102
War-Time Recovery Household Structure . . . . . . . . . . Dynamic Eciency . . . . . . . . . . . The Household Problem . . . . . . . . The Aggregate Consumption Function Aggregate Equations of Motion . . . . Steady State Equilibria . . . . . . . . . Concluding Comments . . . . . . . . . Appendix . . . . . . . . . . . . . . . . 7.8.1 The Household Problem . . . . 7.8.2 The Consumption Function . .
105 . 105 . 108 . 110 . 110 . 112 . 113 . 119 . 120 . 120 . 121 . . . . . . 123 123 124 125 132 132 135
8 Post War Experience 8.1 Introduction . . . . . . . . . . . . . . . . 8.2 The Impact of the Fed-Treasury Accord 8.3 Real Economic History . . . . . . . . . . 8.4 Monetary Economic History . . . . . . . 8.5 The Equity Premium . . . . . . . . . . . 8.6 The Theory Summarized . . . . . . . . .
9 Explaining Stagation 137 9.1 Adding Money to the Model . . . . . . . . . . . . . . . . . . . 138
Preface My preface will appear here.
ix
Chapter 1 What this Book is About

Those who cannot remember the past are condemned to repeat it. George Santayana, The Life of Reason (1905). This book is about the business cycle and how to control it. I will provide a theory of economic uctuations that explains why major recessions occur and I will explain how scal and monetary policy can be used to maintain high and stable employment. The reasoning I will provide is Keynesian and is based on ideas from the General Theory of Employment Interest and Money (1936). The remedies I will suggest are standard and are part of the current arsenal of policies employed by the governments of all market economies. What is new in this book is a theory of individual behavior that explains why these policies are appropriate and how they work. A by-product of my analysis is an explanation of how ination and unemployment can occur together and a theory of what it means to maintain full employment. Although economic uctuations in the U.S. have been relatively mild in recent decades, during the Great Depression of the 1930s the unemployment rate exceeded 20% for a protracted period of time. The Great Depression is not unique, and similar episodes have been a recurrent feature of capitalist economies since the beginning of the industrial revolution. Timothy Kehoe and Edward Prescott (2007) dene a great depression to be a period of diminished economic output with at least one year where output is 20% below the trend. By this denition Argentina, Brazil, Chile, and Mexico have all experienced great depressions since 1980. What causes big uctuations in economic activity? For several decades after the publication of the General Theory economists thought they had 1
CHAPTER 1 WHAT THIS BOOK IS ABOUT
an answer; the Great Depression was thought to be a failure of an unregulated capitalist economy to eciently utilize available resources. But with the resurgence of classical ideas in the 1970s, the key premise of the General Theory, that market economies are not inherently self-stabilizing, has been called into question. Although there has been a recent resurgence of Keynesian ideas under the rubric of new-Keynesian economics, the models studied by the new-Keynesians are hybrids that incorporate a classical core. New-Keynesian models allow for temporary deviations of unemployment from its natural rate as a consequence of sticky prices but they contain a stabilizing mechanism that causes a return to the natural rate over time. Since the return to the steady state is typically rapid in these models, the welfare costs of business cycles are also small and arise from second order consequences of deviating from the social planning optimum.1 In his 1966 book, Axel Leijonhufvud made the distinction between Keynesian economics and the economics of Keynes. By Keynesian economics, he meant the interpretations of Keynes that became incorporated into the IS/LM model and ultimately, into the new-Keynesian paradigm. Leijonhufvud pointed out that the assumption that the General Theory is about sticky prices is central to Keynesian economics but it is not a central argument of the text of the General Theory. This book provides an alternative microfoundation to Keynesian economics that does not rely on sticky-prices. In successive chapters I construct a series of models that build on a single idea. Each of them is constructed around a conventional dynamic general equilibrium model in which real resources must be used to move unemployed workers into jobs using a search technology. Although this technology is convex, I assume that the planning optimum cannot be decentralized as a competitive equilibrium because moral hazard prevents the creation of markets for the search inputs. As an alternative, I introduce an equilibrium concept called demand constrained equilibrium, in which the level of economic activity is determined by investor condence or animal spirits. I refer to the resulting model as old-Keynesian to dierentiate it from new-Keynesian economics that incorporates the natural rate hypothesis of Edmund Phelps (1970) and Milton Friedman (1968). In contrast to new-Keynesian models, those described in this book display
Gal, Gertler, and Salido (2007, page 56) nd that the average welfare costs of business uctuations in a New-keynesian model are less than .0.01% of steady state consumption, a number that is consistent with that cited by Lucas (1987, Chapter IV) in the context of the real business cycle model.
1
1.1 THE NATURE OF THE ENQUIRY
multiple stationary perfect foresight equilibria, and there is a dierent stationary unemployment rate, for each possible level of beliefs.
1.1
The Nature of the Enquiry
I began this chapter by posing a question: Why do capitalist economies sometimes go very wrong? One answer that has become fashionable is to argue that episodes like the Great Depression of the 1930s are an aberration. Economists who take this view, point to post-war U.S. experience in which unemployment has been relatively low and business cycles relatively mild for long periods of time. The Depression is seen as an unusual episode that requires explanation but it is thought that the place to look for such an explanation is in the actions of the regulatory authorities. An early example is the book by Milton Friedman and Anna Schwartz (1963) who argued that the Depression was caused by incompetent monetary policy in the 1920s. More recently Cole and Ohanian (2004) have argued that Herbert Hoovers regulatory policies deepened the depression in the early 1930s. This is a remarkable turn of intellectual thought from the prevailing mood in the early post-war period when President Richard Nixon is famously quoted as saying We are all Keynesians now. The view of business cycles that currently dominates the economics profession is that of real business cycles according to which most economic uctuations are caused by unforeseen shocks to total factor productivity. According to this view, the market system is ecient and the allocation of resources in a laissez-faire economy is, to a rst approximation, a reection of the allocation that would be made by a benevolent social planner whose goal is to maximize social welfare. This view is more extreme than that which Keynes ascribed to the classical economists of whom he considered Pigou a prime example. Pigou (1929), Lavington (1922) and Haberler (1937) all recognized a role for total factor productivity, but they also recognized alternative sources of aggregate uctuation including errors of optimism and pessimism, harvest variations and autonomous monetary movements (Pigou, 1929, Chapter 22). In Industrial Fluctuations Pigou concludes that although ...the popular opinion that industrial uctuations as such must be social evils is invalid, nevertheless ...industrial uctuations produced in the ways described [above] are certainly social evils. The General Theory makes a break from 1920s business cycle theory by
separating the theory of distribution from the theory of aggregate economic activity. Although Keynes believed that the market system is probably an ecient way of deciding which kinds of goods are produced for a given volume of employment, he argued that unregulated capitalist systems do not produce an ecient allocation of labor. His arguments had an important inuence on political economy and resulted in the current state-capitalist system in which the monetary and scal authorities in most Western democracies are charged with the objective of maintaining a high and stable level of employment. The scal and monetary policy environment since 1940 has been very dierent from that of the 1920s and 1930s as a direct consequence of ideas that arose from the publication of the General Theory and hence it is invalid to use the post-war period as an example of the success of the market system. Rather, it represents prima facie evidence for the success of Keynesian economics. A good example is provided by the U.S. economy. In 1929 federal and state taxes together accounted for 11% of gdp but by 1999 that gure had risen to 36%. When the Fed was created in 1913, policy makers had little conception of how to run a successful monetary policy and for most of the 1930s the short term interest rate was too low to be used as an eective instrument of control. Since 1945 however, the size of government has been large enough to create an automatic stabilizing mechanism that generates budget decits in recessions and potential surpluses in expansions. Further, the Fed actively pursues a countercyclical monetary policy by lowering the interest rate during recessions and raising it during expansions. These are exactly the policies that Keynes argued for in the 1930s and it is ironic that the success of these very policies should be seen by some as evidence against the theories that spawned them.
1.2
The Theory Summarized
Although the current volume is heavily inuenced by the General Theory, it is not a simple translation of that book into the language of modern economics. Seventy years of history since the publication of the General Theory have produced data that invalidates at least some of its key themes. Most notable amongst these is the experience of stagation in the 1970s that is inconsistent with the reverse L theory of aggregate supply outlined in Chapter 21 of the General Theory on the theory of prices. There are two key ideas in the General Theory that set it apart from pre-
1.3 THE THEORY OF AGGREGATE SUPPLY
Keynesian economics: The rst is that there is something distinctive about the labor market that makes the marginal disutility of labor dierent in general from the real wage. The second is that aggregate economic activity is determined by the animal spirits of investors. This book will preserve both of these ideas, modied in a way that respects recent developments in dynamic stochastic general equilibrium theory. The integration of stochastic dynamics into modern economic theory provides a set of mathematical tools that enable me to integrate dynamic ideas into macroeconomics in a way that was not possible in the 1930s. The ideas that I have identied as central to Keynesian economics can be separated into theories of aggregate supply and aggregate demand. Although the form with which I will state these ideas is dierent from existing interpretations of the General Theory, the intellectual predecessors are those formulated by Keynes in 1936, who expressed the following sentiment in a 1937 article in response to his critics: I am more attached to the comparatively simple fundamental ideas that underlie my theory than to the particular forms in which I have embodied them, and have no desire that the latter should be crystallized at the present stage of the debate. If the simplest basic ideas can become familiar and acceptable, time and experience and the collaboration of a number of minds will discover the best way of expressing them. Keynes (1937, Pages 211-212). The following two Sections outline briey my main arguments and explain how they are related to the fundamental ideas of the General Theory.
1.3
The Theory of Aggregate Supply
Chapter 2 presents a reformulation of the theory of aggregate supply. Keynes theory has been widely criticized for its lack of microfoundations and it is often asserted that if the marginal disutility of labor is not equal to the real wage, as Keynes assumed in Chapter 2 of the General Theory, then unemployed workers would be expected to oer to work for a lower wage. This argument is based on the implicit assumption that the labor market is an auction in which unemployed workers can eectively signal their willingness to work to prot maximizing rms.
Following arguments by Patinkin (1989) and Clower (1965), the disequilibrium literature of the 70s, exemplied by Barro and Grossman (1971), Benassy (1975), and Malinvaud (1977), tried to address this point by constructing an explicit theory of transactions at disequilibrium prices. This literature was ultimately judged to be unsuccessful by a generation of economists who followed the equilibrium approach of Lucas and Rapping (1969) and Lucas (1972). Although there were contemporary writers, (Axel Leijonhufvud (1966) is a leading example) who claimed that Keynesian economics was never about sticky prices, Leijonhufvud and his followers never managed to formulate an alternative theory that was capable of answering the new-Classical criticism that disequilibrium theory is empty. This argument rested on the fact that it contains what Lucas called free parameters and comes down to the claim that, as a consequence, the theory is untestable. In this book I pick up on recent literature due to Shimer (2005) and Hall (2005) that follows earlier work in search theory. Pissarides (2000) provides an excellent summary. In this tradition it is assumed that the process by which an unemployed worker nds a job requires the input of resources on the part of the rm and time on the part of the worker. When a worker and a rm meet, they determine the wage to be paid through a Nash bargain. Shimer pointed out that this assumption does not provide a good quantitative explanation of employment uctuations and Hall proposed to replace it with an alternative wage determination mechanism; he assumed that the real wage is determined one period in advance. Shimers criticism is commonly referred to as the Shimer puzzle and it has generated a considerable amount of recent work amongst economists and graduate students who are exploring alternative wage determination mechanisms in an attempt to reconcile the volatility of vacancies and unemployment with a model in which economic uctuations are driven by productivity shocks. In this book I take an alternative approach. I develop a series of models in which the labor market is cleared by search but instead of closing it with an explicit bargaining assumption, I assume only that all rms must oer the same wage. This leads to a theory in which there are many wages all of which are consistent with a zero prot equilibrium and it provides a microfounded analog of Keynes idea that there are many levels of economic activity at which the macroeconomy may be in equilibrium. To select an equilibrium and close the model I introduce the idea that households form beliefs about the future value of productive capital and I show that for any sequence of self-fullling beliefs, less than a given bound, there exists a Key-
1.4 THE THEORY OF AGGREGATE DEMAND
nesian equilibrium. This equilibrium will in general be inecient in the sense that a benevolent social planner would prefer a dierent employment level that may be higher or lower. Hence, I am able to articulate the Keynesian story of the Great Depression in a model with well dened microfoundations in which no individual agent has an incentive to deviate from his chosen action.
1.4
The Theory of Aggregate Demand
In addition to his theory of aggregate supply, Keynes contributed a theory of eective demand based on the multiplier. Problems with this theory were already apparent in the 1950s when it was realized that estimates of the marginal propensity to consume were typically much lower in cross-section than time-series data. This led to Friedmans (1957) book, A Theory of the Consumption Function, in which he proposed the concept of permanent income as a way of resolving the disparity between dierent estimates. But this was not the only important debate that characterized the macroeconomics of the immediate post-war period. The exercise of providing a dynamic foundation to Keynesian economics led to an attack on the intellectual foundations of the multiplier and a debate over the eectiveness of scal policy. Blinder and Solow (1973), on purely theoretical grounds, argued that one dollar of expenditure by government might, under some circumstances, perfectly crowd out one dollar of private consumption expenditure and aggregate demand would be unaltered. The debate was ultimately resolved by recognizing that crowding out would occur only if government bonds are not perceived as net-wealth by the community as a whole although this debate, and other preoccupations of the post-war Keynesian economists, became irrelevant when Keynesian economics was replaced by the real business cycle paradigm. In this book I will have cause to breathe new life into many of the old debates. I will construct a microfoundation to the theory of aggregate demand based on the assumption that agents are forward looking with rational expectations of future prices. Although this is a departure from Keynes theory of expectations, it is a departure worth making since it allows me directly to compare the implications of the theory with modern neoclassical alternatives. Keynes believed that some forms of uncertainty cannot be quantied and
that agents must act on the basis of partial information. Although the agents in my model will be able to form probability distributions over future events - not all of these events will be fundamental in the sense in which that word is now used in general equilibrium theory to describe uncertainty due to changes in preferences, endowments and technology. Although these sources of uncertainty will be present in the models I describe in this book, in addition, agents will be required to form expectations of the future actions of others. It is here that I capture the Keynesian idea of the importance of animal spirits. If investors today believe that all future investors will be pessimistic then this belief will be self-fullling. In contrast to the previous formulations of this idea that I described in my book on the Macroeconomics of Self-Fullling Prophecies (1993), agents in the models I will describe in the following chapters may form self-fullling beliefs that lead to an increase in the unemployment rate in the steady state.
1.5
Additional Features of my Interpretation
Keynes built a static theory of aggregate supply based on the assumption that the money wage and the existing stocks of capital equipment could be taken as given. Although he argued that his main ideas did not depend on these assumptions, the attempt that he made in Chapter 21, to provide a dynamic theory of wages and prices must ultimately be judged a failure. Keynes argued that there exists a critical level of employment, beyond which, increases in aggregate demand would lead to ination. According to this theory, a necessary requirement for the emergence of ination is the overemployment of existing resources. This theory is at odds with the observation of the 1970s in which we saw the simultaneous occurrence of high levels of unemployment and ination and hence the Keynesian theory must be revised. In this book I will amend the Keynesian theory by developing a microfounded theory of aggregate supply that avoids this problem. This theory, outlined in Chapter 2, does not lead to a reverse L theory of aggregate supply and in my version of the theory of aggregate supply, there may be overemployment just as there may be underemployment. Overemployment occurs if rms devote too many resources to the activity of searching for new workers and, as a consequence, fewer workers are available to produce commodities.
1.5 ADDITIONAL FEATURES OF MY INTERPRETATION
The General Theory, in contrast to much of the economics of Keynes followers, used only two units of measurement; money-value, measured in currency units, and ordinary-labor measured in hours. By restricting himself to the description of purely aggregate relationships between nominal gdp and employment Keynes was able to describe the forces that he believed determine the level of aggregate economic activity without making special assumptions about industrial structure. This level of generalization was possible because Keynes was concerned with determining the level of economic activity at a point in time and for this purpose it is legitimate to take the stocks of all existing capital goods as xed. I will show, in Chapter 3, that a theory of aggregate supply based on search theory can be expressed at this same level of abstraction and it is relatively simple to construct a static theory that allows for intermediate goods, multiple produced goods, and multiple capital goods. It is possible to put this theory together with Keynes theory of aggregate demand and to derive familiar Keynesian propositions. This exercise is instructive since it illustrates the power of the search-based version Keynes theory of aggregate supply. But although the static theory can be well expounded using the simple apparatus of the multiplier, the Keynesian theory of aggregate demand does not extend easily to dynamic models. Chapter 4 continues the exposition of the Keynesian model by embedding a search theory of the labor market into a two-period model with overlapping generations. This extends the models of the previous chapters by allowing for the distinction between saving and investment that Keynes thought to be central to the determination of the level of aggregate economic activity. The chapter provides an elaboration of the Keynesian idea that investment and savings are equated not by changes in the rate of interest; but by changes in gdp. But how does this argument extend to a fully dynamic economy? The labor market model of chapters 2-4 are set into simple one or two period economies in which income provides the scale variable to determine consumer behavior. But the experience of post-war macroeconomics taught us that Keynesian relationships like the consumption function or the demandfor-money function would need to include not income but wealth as an explanatory variable. This was the lesson of Friedmans work on the permanent income expounded in his 1957 book on the consumption function and it makes sense not only from an empirical point of view (it helps explain a discrepancy between cross-section and time series estimates of the consumption function), but also from a theoretical point of view. A long-lived agent
10
making plans for the innite future should be concerned about his wealth, not just his current income. Incorporating a search-theoretic model of the labor market into an innite horizon maximizing model in a consistent way provides a new set of challenges that go beyond the economics of the General Theory that I take up in chapters 6-8. Chapters 6, 7 and 8 deal with simple explanations of stylized facts, each associated with a dierent historical episode. But how should these facts be presented? Most recent work on business cycles, motivated by the Real Business Cycle, (RBC) agenda, has used the rst and second moments of detrended gdp, consumption, investment and labor hours as the salient facts of the business cycle. According to this agenda, a good model is one that can generate articial data with statistical properties that replicates those of the detrended data where detrending is achieved by removing a separate low frequency component from each time series in both the original and the articial data. Chapter 5 argues that this approach misses important business cycle facts and it suggests an alternative approach inspired by Chapter 4 of the General Theory - the measurement of data in wage units. Using data presented in the way described in Chapter 5, Chapter 6 takes up the task of articulating the Keynesian explanation of the Great Depression in a multi-commodity dynamic model with an innitely lived representative household and a search-theoretic model of the labor market. In the chapter I dene a notion of demand constrained equilibrium, (I also refer to this as Keynesian equilibrium) and I contrast its properties with the social planning solution. Since this is a representative agent economy the social welfare function can be unambiguously dened. In a Keynesian equilibrium the level of economic activity is indexed by the animal spirits of investors, dened as a bounded self-fullling sequence of stock-market prices. For every such sequence there is a Keynesian equilibrium, all but one of which is socially inecient. Unlike the ineciencies arising from new-Keynesian models the deviations from rst-best are rst order and can be large.
Chapter 2 The Labor Market

Each chapter of this book is based on a model of the labor market where the search inputs of workers and rms are combined to produce matches. A match is an employed worker in place at a rm. Search requires two inputs, 1) the time spent searching by workers and 2) the resources needed to post vacancies by rms. Most existing search models do not assume that these inputs are traded in competitive markets. Instead, they assume that vacancies and unemployed workers are matched randomly. If searching workers and vacancy posting rms take the real wage as given, the resulting general equilibrium model has fewer equations than unknowns. In search models it is typical to assume that the match technology satises standard neoclassical properties of monotonicity, dierentiability and constant returns-to-scale. These assumptions allow one to prove versions of the rst and second welfare theorems in a general equilibrium model with search; every competitive equilibrium is Pareto optimal and every Pareto optimal allocation can be decentralized as a competitive equilibrium. But what might this decentralization look like? The natural decentralization would posit the existence of a large number of competitive employment agencies. Each agency would operate a match technology and would purchase search inputs from rms and workers. The agency would purchase, from an unemployed worker, the exclusive right to match that worker with a vacancy. From a rm with a vacant job, the agency would purchase the right to match that vacancy with an unemployed worker. The agency would operate a matching process and resell the joint product, a worker-rm match, back to the worker-rm pair. Why is this decentralization implausible? First, it involves transactions 11
12
CHAPTER 2 THE LABOR MARKET
that we do not observe in the real world. There are no private institutions that pay money to unemployed workers for the right to nd them jobs. Nor do we nd private employment agencies that pay rms for the privilege of acting as their recruiting agents. A moments reection suggests that these markets do not exist because of the moral hazard associated with monitoring the motives of the participants. Ecient operation of these markets requires exclusivity of contracts. If such markets existed it would be dicult or impossible to prevent an unemployed worker from selling the exclusive right to be matched to multiple agencies and to turn down job oers when presented on spurious but hard-to-monitor grounds. Since there may be legitimate reasons to refuse a job, the requirement that all potential matches must be accepted is not a feasible solution to this problem. Casual observation of state run employment agencies suggests that this problem is present in practice and is a signicant impediment to the ecient operation of a matching market.
2.1
Components of the General Theory
What are the key ideas of the General Theory and how might one construct a microfounded model that embodies these ideas? First, there is the assertion in Chapter 2, that the real wage is not equated to the marginal disutility of labor. Second, in chapter 3, there is the principle of eective demand based on the multiplier and taught to several generations of undergraduate students in the form of the Keynesian cross. This chapter is an attempt to make sense of these ideas using a theory of labor market search. In the General Theory, eective demand is driven primarily by the animal spirits of investors. Animal spirits are a key component of autonomous investment expenditure that, in turn, is the prime cause of uctuations in eective demand. To capture this idea one requires a dynamic model since investment involves plans that span at least two periods. However, autonomous expenditure is also determined by government spending and by recognizing this I will be able to explain how a search model of the labor market can be embedded into general equilibrium in a relatively simple environment; a one period model that abstracts from capital in which output and employment are driven by scal policy. The one period model I will describe is simpler in some directions, but more complicated in others, than the rst dynamic model that I introduce in chapter 4. It is simpler since I abstract from capital and assume that
2.2 HOUSEHOLDS
13
all output is produced from labor. It is more complicated since I introduce a government sector and discuss an issue that was important in the 1970s but has become less actively debated in recent decades; that of Ricardian equivalence. Sections 2.2 and 2.3 are about the microeconomic behaviors of households and rms. The questions I will study are not ones that occupied Keynes who was concerned solely with relationships between aggregates. But they are questions that I will need to address in an enquiry that seeks to provide microfoundations to Keynes concepts of aggregate demand and supply. Initially, I will simplify the environment by studying a model in which there is a single period, a single commodity, and a large number of identical households and rms. These are not simplications found in the General Theory and, as I will show in Chapter 3, they are unnecessary simplications if one is interested in a comparative static view of macroeconomic activity. Keynes took the existing stocks of capital and existing nominal wages as historically determined and he showed how eective demand would determine economic activity. My goals are more comprehensive. I want rst, to show how the principle of eective demand can be consistent with individual behavior at a point in time. But beyond that, I want to link the periods in a dynamic general equilibrium model where the people in my model form rational expectations of future events. Although the single agent, single good ction is unnecessary to explain eective demand in a comparative static model, it considerably simplies the nature of aggregate dynamics.
2.2
Households
The model consists of a unit mass of families each of which has a unit mass of members. These assumptions allow me to abstract from the fact that unemployed workers are typically worse o than employed workers. In this model, the family self-insures its unfortunate members. The utility of the family is represented by an increasing concave function j J = j (C) (2.1)
where J represents the utility of the familys consumption, C. Since all families are identical I will refer to the consumption of an individual family and to aggregate consumption with the same symbol.
14
Each family has a measure 1 of workers all of whom begin the period unemployed. Leisure has no utility and each household solves the problem
{C,H}
max j (C)
(2.2)
such that pC wL (1 ) + T, H 1, L = qH, U = H L. (2.3) (2.4) (2.5) (2.6)
Equation (2.3) is a budget constraint. Each familys consumption is constrained by its after tax employment income. w is the money wage, p is the money price, L is the measure of employed workers, is the tax rate and T is a lump-sum transfer, measured in dollars. H represents the measure of household members that search for employment and Equation (2.4) constrains this to be no greater than 1, the household size. The measure of household members that successfully nd jobs is represented by q H; where q is taken as given by the household in a search market equilibrium; Equation (2.5) is the relationship between employment and search. Finally, Equation (2.6) denes the measure of unemployed, U, to be those searching workers who do not nd jobs. This problem has the trivial solution H = 1, pC = wL (1 ) + T, L = q. (2.7) (2.8) (2.9)
Since there is no utility to leisure all workers look for a job and since there is no motive to save; all income is consumed.
2.3
Firms
Firms produce output using a constant-returns-to-scale technology in which labor is the sole input. There is free entry and each rm solves the problem
{V,Y,L,X}
max pY wL
(2.10)
2.3 FIRMS such that X + V = L, L = qV. Y AX,
15
(2.11) (2.12) (2.13)
Equation (2.11) is the production function, Y is output, A > 0 is a productivity parameter and X is the measure of workers employed by the rm in direct production. A rm that employs L workers may allocate them to produce commodities (this is the measure X) or to the recruiting department (this is the measure V ). A rm that devotes V workers to recruiting will hire qV workers where q is taken as given by the rm. I have assumed that labor, rather than output, is used to post vacancies in contrast to most search models. This innovation is not important and is made for expositional simplicity and to allow me, later in the book, to write down models that can easily be compared with more familiar real business cycle economies. The timing of the employment decision deserves some discussion since it allows the rm to use workers to recruit themselves. If a rm begins the period with no workers, and if workers are an essential input to recruiting, it might be argued that the rm can never successfully hire a worker. Since I will be thinking of the time period of the model as a quarter or a year, this assumption should be seen as a convenient way of representing the equilibrium of a dynamic process. The rm puts forward a plan that consists of a feasible 4tuple {V, Y, L, X}. Given the exogenous hiring elasticity, q, a plan to use V workers in recruiting results in qV workers employed of whom X are used to produce commodities. Solving the rms problem leads to the correspondence, if pA 1 1 w > 0, q (2.14) V = [0, ] if pA 1 1 w = 0, q 0 if pA 1 1 w < 0. q This expression is closely related to the condition that would arise in a model where labor is hired in a spot market. In a model of that kind there is no need for a recruiting department and one would require the real wage, w/p, to equal the marginal product of labor, w A= (2.15) p
16
in order for the rm to produce positive output. If productivity, A, were greater than w/p, the rm would be willing to expand without limit. If A were less than the real wage, the rm would shut down. The correspondence represented in (2.14) is similar to the spot-market case but productivity is weighted by the hiring eectiveness parameter q, which represents the number of workers that can be hired by a single recruiter.1 As this parameter gets large, the relative size of the recruiting department shrinks, and in the limit the production function of the search model converges to that of the spot-market model. The technology of the spot-market model delivers a restriction on the real wage in equilibrium in the form of Equation (2.15). The search-market equivalent is the equation 1 w A 1 = . (2.16) q p Equation (2.16) is consistent with a range of equilibrium real wages since the hiring eectiveness parameter, q, is an endogenous variable. Existence of a solution with non-zero output requires q > 1. If q is large, a small recruiting department can support a large workforce and productivity and the real wage will be high in a zero prot equilibrium. If q is small, the reverse is true and in the limit, as q approaches 1, the entire workforce is engaged in recruiting and there is no-one left to produce commodities.
2.4
Search
I have described how individual households and rms respond to the aggregate variables w, p, q and q. This section describes how aggregate matches and aggregate employment are related to the aggregate measure of vacancies. This requires a description of the match technology which takes the form, m = H 1/2 V 1/2 , (2.17) where m is the measure of workers that nd jobs when H unemployed workers search and V vacancies are posted by rms. I have used bars over variables to distinguish aggregate from individual values. Since leisure does not yield disutility and hence H = 1, this equation simplies as follows, (2.18) m = V 1/2.
q is closely related to labor market tightness, which Mortensen-Pissarides dene to be the ratio of vacancies to unemployment.
1
2.5 THE SOCIAL PLANNER
17
A further simplication follows from the fact that, since all workers are initially unemployed, employment and matches are the same thing and hence,2 L = V 1/2 . (2.19)
2.5
The Social Planner
In Section 2.6 I will dene an equilibrium concept that captures the idea of eective demand. Before taking this step, it is helpful to have a benchmark against which to measure the properties of equilibrium. Consider a benevolent social planner who maximizes the welfare of the representative family. The planner solves the problem
{C,V,L,H}
max J = j (C) C AX, L = X + V,
(2.20) (2.21) (2.22) (2.23) (2.24) (2.25)
L = H 1/2 V 1/2, H 1, H = L + U.
Since the objective function is increasing in C, the inequalities (2.21) and (2.24) will hold with equality. Using this fact and combining equations (2.22)(2.24) leads to the expression C = AL (1 L) , which is maximized at 1 1 A V = , L = , C = . 4 2 4
2
(2.26)
(2.27)
In a dynamic model, employment will appear as a state variable in a programming problem since it takes time to recruit new workers. In this chapter, I abstract from this aspect of labor market dynamics.
18
C = AL (1 L )
A 4
L*
Figure 2.1: The Social Planning Solution Figure 2.1 illustrates the nature of this solution on a graph.3 Since there is a representative family in this economy the only eective decision of the social planner is how many workers to allocate to recruiting. Given the search technology, the optimum is achieved at V = 1/4. Any additional allocation of workers to recruiting would be counter-productive. Although the social planner could increase employment, the additional employed workers would not produce additional output - they would simply be recruiting additional recruiters and the resulting allocation would leave less, not more, output available for consumption. The social planning solution provides a clear candidate denition of full employment - it is the level of employment L that maximizes per-capita output. In the General Theory, Keynes argued that a laissez-faire economic system would not necessarily achieve full employment and he claimed the
I have drawn this graph for the case B = 1 and in this case full employment occurs when 50% of the labor force is employed. 100% emplyment would result in zero output. Dierent values of B will modify these values but the message remains the same. For example, if B = 3/4, full employment occurs at 33% unemployme
3
2.6 AGGREGATE DEMAND AND SUPPLY
19
possibility of equilibria at less than full employment as a consequence of what he called a failure of eective demand. Section 2.6 makes this notion precise in a micro-founded model based on labor market search.
2.6
Aggregate Demand and Supply
Before giving a formal denition of equilibrium I will outline Keynes principle of eective demand in the context of a one-good general equilibrium model where the spot-market model of the labor market is replaced by an appropriate denition of search market equilibrium. Keynes dened the aggregate supply price of a given volume of employment to be the expectation of proceeds which will just make it worth the while of the entrepreneurs to give that employment (Keynes, 1936, Page 24). I will return to this denition in Chapter 3 where I provide a multi-sector version of the model. In the one-sector representative agent version the following simplications are possible. First, if one assumes rational expectations and no uncertainty then expectations of proceeds may be replaced by proceeds. Second, proceeds are dened as factor cost plus prots and, in the representative agent environment, this is equivalent to the value of nominal gdp. Third, since I will be concerned with a real model, nominal and real gdp can be set equal to each other by choosing an appropriate numeraire. It is tempting to notice that, in the one good model, it is possible to choose a price normalization rule by setting p = 1. I will resist this normalization since it does not generalize to the multiple good world and instead I will choose the normalization w = 1. This implies that p is the inverse of the real wage. This fact is important in interpreting the aggregate supply and demand diagram of the General Theory. The Keynesian aggregate supply function may be represented by a diagram that measures the aggregate supply price in units of money on the vertical axis, Keynes called this Z, and ordinary units of labor on the horizontal axis, Keynes called this N. I have replaced the N of the General Theory with the symbol L to be consistent with the notation introduced earlier in the chapter. Using this notational change one can write Keynes
20 Aggregate Supply Function as,
Z = (L) .
(2.28)
Bear in mind that in the General Theory, Z is the value of a set of heterogenous commodities and it is only in the one commodity model that it can be reduced to the expression Z pY, (2.29) where Y is the number of physical units of the produced good. More generally, Z would be dened by the expression Z
n X i=1
pi Yi .
(2.30)
To complete the description of his equilibrium concept Keynes dened D to be the proceeds which entrepreneurs expect to receive from the employment of L men, the relationship between D and L being written D = f (L) which can be called the Aggregate Demand Function. (Keynes, 1936, Page 25, L is substituted for N from the original). Like Z, D is measured in monetary units. Keynes principle of eective demand amounts to the propositions that 1) employment is determined by the intersection of the aggregate demand and supply schedules and 2) equilibrium may occur at a point less than L , the full employment level that I have dened as the solution to the social planning problem. To elucidate the properties of aggregate supply Keynes asked us to consider what would happen if, for a given value of employment, aggregate demand D is greater than aggregate supply Z. In that case... there will be an incentive to entrepreneurs to increase employment beyond L and, if necessary, to raise costs by competing with one another for the factors of production, up to the value of L for which Z has become equal to D. (Keynes, 1936, Page 25, N replaced by L and italics added).
21
In the one-good representative agent model the principle of eective demand implies that competition between prot maximizing rms will cause the real wage to adjust to the point where prot is equal to zero. The following algebra establishes that for values of D in a given interval, there will exist a real wage that has this property. Consider the implications of assuming that the economy is in a symmetric equilibrium in which rms take the hiring eectiveness parameter q as given. Symmetry implies that the variables V and V , L and L are equal and the assumption that rms take q parametrically implies; L = qV. From the properties of the aggregate technology L = V 1/2 . (2.32) (2.31)
It follows, in a symmetric equilibrium, that the following expressions characterize the relationships between q,V and L. q= 1 , V = L2 . L (2.33)
Combining the rst of these expressions with the zero prot condition, Equation (2.16), it follows that there exists a zero prot equilibrium for any value of L [0, 1] with a real wage given by the expression, w = A (1 L) , p (2.34)
which, given the normalization w = 1, implies p= 1 . A (1 L) (2.35)
The aggregate supply function is found by combining (2.11), (2.12), (2.13) and (2.33) to yield the expression, Z = pY = L. (2.36)
One can also combine Equations (2.35) and (2.36) to yield the following expression for the physical quantity of output produced in this economy Y = AL (1 L) (L) . (2.37)
22
Z
A 4
Y=
(L )
p (L )
(L )
(L ) = L
45
L*
Figure 2.2: The Aggregate Supply Fucntion The aggregate supply function, (L) and the output function Y = (L) are graphed on Figure 2.2 as the solid line and the dashed curve. Moving along the aggregate supply function from zero to L , output is increasing. Moving beyond L , aggregate supply as dened by Keynes continues to increase as the price level rises but the physical quantity of the produced good falls. Although it is tempting to refer to (L) as the aggregate supply function this would be a mistake since, as Keynes made clear in the General Theory, this measure cannot easily be generalized beyond the one-good case.
2.7
Eective Demand and the Multiplier
In modern DSGE models the government is assumed to choose expenditure and taxes subject to a constraint. Models that incorporate a constraint of this kind were dubbed Ricardian by Robert Barro (1974). But in models with multiple equilibria there is no reason to impose a government budget
2.7 EFFECTIVE DEMAND AND THE MULTIPLIER
23
constraint and Eric Leeper (1991), discussing models of monetary and scal policy, has argued that one should allow government to choose both taxes and expenditure and that this choice selects an equilibrium. He calls a policy in which the government choose both taxes and expenditure, an active scal regime. The modied-search model of the labor market is one with multiple equilibria and hence, one can close the model in the way advocated by Leeper. To derive the aggregate demand function for the one-good representative agent economy one need only recognize that materials balance requires D = pC. (2.38)
This is the GDP accounting identity in a model with no government expenditure and no investment. Aggregate demand is related to employment by the expression D = (1 ) wL + T, (2.39) where recall that is the tax rate and T represents lump-sum transfers measured in dollars. Since we have chosen w as the numeraire, set equal to 1, it follows that aggregate demand is equal to D = (1 ) L + T, (2.40)
where all terms of this equation are in monetary units. Recall, from Equation (2.36) that aggregate supply is given by the expression; Z = L. Substituting (2.41) into (2.40) it follows that D = (1 ) Z + T, (2.42) (2.41)
and that in equilibrium when D = Z, the equilibrium value of income, Z K is given by T ZK = , (2.43) where the superscript K is for Keynesian. The aggregate supply function, Equation (2.36), and the aggregate demand function, Equation (2.40), are depicted in Figure 2.3 together with the physical quantity of output, Y . The equilibrium depicted in Figure(2.3) is one where there is positive unemployment since at LK , the Keynesian equilibrium, is less than L , the
24
D, Z
A 4 1
Y = (L )
Aggregate Supply
Z =L
45
D = (1 ) L + T
Aggregate Demand
LK
L*
Figure 2.3: Aggregate Demand and Supply social planning optimum. If government were to increase transfers, T , it would be possible to increase the Keynesian equilibrium to a point to the right of L . A policy of this kind would result in a higher price level (a lower real wage), less output than at L , and over employment since the Keynesian equilibrium would be associated with too many workers employed.
2.8
A Denition of Equilibrium
This section provides a formal denition of equilibrium based on the ideas sketched out above. I will appropriate a term, demand constrained equilibrium, that has was used by Jean Pascal Benassy (1975), Jacques Dreze (1975) and Edmond Malinvaud (1977) in a literature on x-price economics that was developed in the 1970s. Although xed-price models with rationing of the kind studied by these authors are sometimes called demand constrained equilibria; that is not what I mean here. Instead I will use the term to refer to a
2.8 A DEFINITION OF EQUILIBRIUM
25
competitive search model that is closed with a materials balance condition. The common heritage of both usages of demand constrained equilibrium is the idea of eective demand from Keynes General Theory. Denition 2.1 (Demand Constrained Equilibrium) For any given and T a symmetric demand constrained equilibrium (DCE) is a real wage w/p, an allocation {Y, C, V, L, X} and a pair of numbers q and q, with the following properties. 1) Feasibility: Y AX, (2.44) LV C Y, (2.45) (2.46) (2.47) (2.48)
1/2
X + V = L, p T (1 ) AX. w 2) Consistency with optimal choices by rms and households: if A 1 1 w > 0, q p V = [0, ] if A 1 1 w = 0, q p 0 if A 1 1 w < 0, q p p T C = L (1 ) + . w w q=L L q= . V 3) Search market equilibrium:
(2.49)
(2.50)
(2.51) (2.52)
To summarize, the modied-search model of the labor market provides a micro-foundation to the Keynesian cross that characterized textbook descriptions of Keynesian economics in the 1960s. Income, equal to output, is demand determined and is equal to a multiple of exogenous expenditure. Since I have abstracted from saving and investment, aggregate expenditure is determined as a multiple of transfer payments where the multiplier is the inverse of the tax rate.
26
2.9
Remarks on Ricardian Fiscal Policies
A large literature has developed recently on Ricardian and non-Ricardian scal policies, (see for example, the papers by Cochrane (1999), Sims (1994) and Woodford (1995) in which it has been argued that, contrary to the monetarist view, the price level can, under some circumstances, be determined by scal policy. In an incisive survey of this literature, Kocherlakota and Phelan (1999) have argued, I think correctly, that the scal theory is a selection device in a model in which there would otherwise be multiple equilibria. In the context of this literature Leeper (1991) characterizes monetary and scal regimes as active and passive. A passive monetary regime is one in which the nominal interest rate is set by the Fed to respond weakly to a nominal anchor. If the Fed perceives an increase in ination it raises the interest rate less than one-for-one. An active regime is one where the response to ination is more aggressive and a one per cent increase in ination is met with a greater than one percent increase in the interest rate. In simple representative agent economies, an active policy can be shown to lead to a unique equilibrium whereas a passive policy is associated with the existence of multiple equilibria that take the form of many paths leading back to a unique steady state. These results lead naturally to the question: What determines the price level if the monetary regime is passive? The scal theory of the price level answers this question by arguing that the government is a large agent that does not act as a price taker. Whereas a typical agent in a general equilibrium model chooses an allocation taking prices as given; in the scal theory the government chooses a plan that would violate its budget constraint for all but a single price system. A policy of this kind is called non-Ricardian or active. An active scal regime is characterized by the fact that the government budget equation holds as an equilibrium condition for only one price system as opposed to a passive scal policy in which it holds for all feasible price systems. It is precisely the possibility that scal policy may be active that determines the equilibrium values of output and employment in the model that I have constructed in this chapter. But the model I have formulated is not a simple restatement of the new-Keynesian models studied by Cochrane, Sims and Woodford. Old Keynesian economics diers from these models because in an old Keynesian model there is a multiplicity of steady states.
2.10 CONCLUDING COMMENTS
27
2.10
Concluding Comments
The model I have described has many features in common with the Keynesian cross that was taught to several generations of undergraduates. That model was criticized by Patinkin (1989) amongst others since it lacked a coherent theory of the labor market. Patinkin put together Keynesian economics with general equilibrium theory by including the real value of money balances in utility and production functions. Although this route was intellectually coherent, it castrated the main message of the General Theory: that the level of economic activity is demand determined in equilibrium.4 A major criticism of Keynesian theory is that, when augmented by a classical model of the labor market, it gives unemployed workers an incentive to oer to work for a lower wage. The combination of a complete set of Walrasian markets and a demand-determined level of economic activity is inconsistent: it results in a system with one more equation than unknown. The search-based model of the labor market described in this chapter provides a microfoundation to Keynesian economics that is not Walrasian since by assumption, agents do not trade the inputs to the search technology in competitive markets. The search technology has two inputs, unemployed workers and vacancies and a single price, the real wage. The resulting model lacks one equilibrium condition which makes it a perfect partner for the Keynesian theory of demand determination. The resulting synthesis leads to a coherent theory of output and relative prices that does not suer from the classical criticism that unemployed workers have an incentive to oer to work for a lower wage.
It for precisely this reason that Keynes called his book the General Theory. He viewed the Walrasian equilibrium as on of many possible rest points of the system.
Chapter 3 Aggregate Demand and Supply

The concepts of aggregate demand and supply are widely used by contemporary economists. My purpose in this chapter is to explain the meaning that Keynes gave to them. Aggregate demand and supply are typically explained in the context of a one commodity model in which real gdp is unambiguously measured in units of commodities per unit of time. In the General Theory there is no assumption that the world can be described by a single commodity model. Chapter 4 of the General Theory is devoted to the choice of units. Here Keynes is clear that he will use only two units of measurement, a monetary unit (I will call this a dollar) and a unit of ordinary labor. The theory of index numbers as we understand it today was not available at the time and Keynes use of these units to describe relationships between the components of aggregate economic activity was clever and new. Keynes chose an hours unit of ordinary labor to represent the level of economic activity because it is a relatively homogeneous unit. To get around the fact that dierent workers have dierent skills he proposed to measure labor of dierent eciencies by relative wages. Thus ...the quantity of employment can be suciently dened for our purpose by taking an hours employment of ordinary labour as our unit and weighting an hours employment of special labor in proportion to its remuneration; i.e. an hour of special labour remunerated at double ordinary rates will count as two units... [General Theory, Page 41] The other unit that Keynes uses in the general theory is that of monetary 29
30
CHAPTER 3 AGGREGATE DEMAND AND SUPPLY
value. His aggregate demand and supply curves are relationships between the value of aggregate gdp measured in dollars and the volume of aggregate employment measured in units of ordinary labor. This is not the same as the relationship between a price index and a quantity index that is used to explain aggregate demand and supply in most modern textbooks.
3.1
Households
Sections 3.1 and 3.2 extend the model of Chapter 2 by adding multiple goods. I will begin by describing the problem of the households. Since I am going to concentrate on the theory of aggregate supply, I will continue to assume the existence of identical households, each of which solves the problem
{C,H}
max J = j (C) ,
(3.1) (3.2) (3.3) (3.4) (3.5)
p C (1 ) Lw + r K + T, H 1, L = qH, U = H L.
Each household has a measure 1 of members. C is a vector of n commodities, p is a vector of n money prices, w is the money wage, r is a vector of money rental rates and K is a vector of m factor endowments. I use the symbol rj to 0 refer to the j th rental rate. The factors may be thought of as dierent types of land although in later chapters, when I introduce investment, they will have the interpretation of dierent types of capital goods. I will maintain the convention throughout the book that boldface letters are vectors and xy is a vector product. The household decides on the measure H of members that will search for jobs, and on the amounts of its income to allocate to each of the n commodities. T is the lump-sum household transfer (measured in dollars) and the income-tax rate. A household that allocates H members to search will receive a measure qH of jobs where the employment rate q is taken parametrically by households. I will assume that utility takes the form j (C) =
n X i=1
gi log (Ci ) ,
(3.6)
3.1 HOUSEHOLDS where the utility weights sum to 1,

n X i=1
31
gi = 1.
(3.7)
Later, I will also assume that each good is produced by a Cobb-Douglas production function and I will refer to the resulting model as a logarithmicCobb-Douglas, or LCD, economy. Although the analysis could be generalized to allow utility to be homothetic, and technologies to be CES, this extension would considerably complicate the algebra. My intent is to nd a compromise model that allows for multiple commodities but is still tractable and for this purpose, the log utility model is familiar and suitable. The solution to the problem has the form H = 1, pi Ci = gi Z D , (3.8) (3.9)
where gi is the budget share allocated to the i0 th good. For more general homothetic preferences these shares would be functions of the price vector p. Household income, Z is dened as Z Lw + r K, (3.10)
and is measured in dollars. The term Z D in Equation (3.9) represents disposable income and is dened by the equation Z D = (1 ) Z + T. (3.11)
Since all income is derived from the production of commodities it follows from the aggregate budget constraints of households, rms and government that Z is also equal to the value of the produced commodities in the economy, Z
n X i=1
pi Yi .
(3.12)
The equivalence of income and the value of output is a restatement of the familiar Keynesian accounting identity, immortalized in the textbook concept of the circular ow of income.
32
3.2
Firms
There are n m commodities. Output of the i0 th commodity is denoted Yi , and is produced by a constant returns-to-scale production function Yi = i (Ki , Xi ) , (3.13)
where Ki is a vector of m capital goods used in the i0 th industry and Xi is labor used in production in industry i. The j 0 th element of Ki , denoted Ki,j , is the measure of the j 0 th capital good used as an input to the i0 th industry and Ki is dened as, Ki (Ki,1 , Ki,2 . . . , Ki,m ) . The function i is assumed to be Cobb-Douglas,
i,1 i,2 i,m i (Ki , Xi ) Ai Ki,1 Ki,2 . . . Ki,m Xibi ,
(3.14)
(3.15)
where the constant returns-to-scale assumption implies that the weights ai,j and bi sum to 1, m X ai,j + bi = 1. (3.16)
j=1
Since the assumption of constant returns-to-scale implies that the number of rms in each industry is indeterminate I will refer interchangeably to Yi as the output of a rm or of an industry. Each rm recruits workers in a search market by allocating a measure Vi of workers to recruiting. The total measure of workers Li , employed in industry i, is (3.17) Li = Xi + Vi . Each rm takes parametrically the measure of workers that can be hired, denoted q, and employment at rm i is related to Vi by the equation, Li = qVi . The rm solves the problem
{Ki ,Vi ,Xi ,Li } a
(3.18)
max
pi Yi wLi r Ki
a a
(3.19) (3.20)
i,1 i,2 i,m Yi Ai Ki,1 Ki,2 . . . Ki,m Xibi ,
3.3 SEARCH Li = Xi + Vi , Li = qVi .
33 (3.21) (3.22)
Using equations (3.21) and (3.22) we can write labor used in production, Xi , as a multiple, Q, of employment at the rm, Li Xi = Li Q, where Q is dened as 1 Q= 1 . q max pi Yi wLi r Ki ,
a a a
(3.23)
(3.24)
We may then write the problem in reduced form,

{Ki ,Vi ,Xi ,Li }
(3.25) (3.26)
i,1 i,2 i,m Yi Ai Lbi Qbi Ki,1 Ki,2 . . . Ki,m . i
The solution to this problem is characterized by the rst-order conditions ai,j pi Yi = Ki,j rj , j = 1, . . . , m, (3.27) (3.28)
bi pi Yi = wLi .
Using these rst-order conditions to write Li and Ki,j as functions of w, r, and pi and substituting these expressions into the production function leads to an expression for pi in terms of factor prices, w ,r . (3.29) pi = pi Q The function pi : Rm+1 R+ is known as the factor price frontier and is homogenous of degree 1 in the vector of m money rental rates r and in the productivity adjusted money wage, w.
3.3
Search
I have described how individual households and rms respond to the aggregate variables w, p, r, q and q. This section describes the process by which
34
searching workers are allocated to jobs. I assume that there is an aggregate match technology of the form, m = H 1/2 V 1/2 , (3.30)
where m is the measure of workers that nd jobs when H unemployed work workers are allocated to recruiting in aggregate by all rms. ers search and V I have used bars over variables to distinguish aggregate from individual val ues. Since leisure does not yield disutility and hence H = 1, this equation simplies as follows, m = V 1/2. (3.31) Further, since all workers are initially unemployed, employment and matches are equal, L = V 1/2 . (3.32) Jobs are allocated to the i0 th rm in proportion to the fraction of aggregate recruiters attached to rm i; that is, Vi Li V 1/2 , V where Vi is the number of recruiters at rm i. (3.33)
3.4
The Social Planner

n X i=1
In the multi-good economy, the planner solves the problem max j (C) =
a a
{C,V,L,H } a
gi log (Ci )
(3.34) (3.35) (3.36)
b i,1 i,2 i,m Ci Ai Ki,1 Ki,2 . . . Ki,m (Li Vi ) i , i = 1, . . . n, n X i=1
Ki,j Kj , j = 1, . . . m Li = H V 1/2 Vi ,
(3.37) (3.38)
H 1.
35
Since the optimal value of H, denoted H , will equal 1, Equations (3.37) and (3.38) can be rearranged to give the following expression for aggregate employment as a function of aggregate labor devoted to recruiting, L
n X i=1
Li = V 1/2 .
(3.39)
Combining this expression with Equation (3.37) leads to the following relationship between labor used in recruiting at rm i, employment at rm i, and aggregate employment, Vi = Li L. (3.40) Equation (3.40) implies that it takes more eort on the part of the recruiting department of rm i to hire a new worker when aggregate employment is high; this is because of congestion eects in the matching process. Using Equation (3.40) to eliminate Vi from the production function we can rewrite (3.35) in terms of Ki and Li ,
i,1 i,2 i,m Ci = Ai Ki,1 Ki,2 . . . Ki,m Lbi (1 L)bi . i
(3.41)
Equation (3.41) makes clear that the match technology leads to a production externality across rms. When all other rms have high levels of employment it becomes harder for the individual rm to recruit workers and this shows up as an external productivity eect, this is the term (1 L), in rm i0 s production function. The externality is internalized by the social planner but may cause diculties that private markets cannot eectively overcome. I will show below that this externality is the source of Keynesian unemployment. To nd a solution to the social planning solution, we may substitute Equation (3.41) into the objective function (3.34) and exploit the logarithmic structure to write utility as a weighted sum of the logs of capital and labor used in each industry, and of the externality terms that depend on the log of (1 L). The rst-order conditions for the problem can then be written as Pn gi bi gi bi , i = 1, . . . , n, (3.42) = i=1 Li (1 L) gi ai,j = j , i = 1, . . . , n, j = 1, . . . , m, Ki,j n X Ki,j = Kj , j = 1, . . . , m.
i=1
(3.43) (3.44)
36
the The variable j is a Lagrange multiplier on P j 0 th resource constraint. Using Equation (3.42) and the fact that L = n Li , it follows from some i=1 simple algebra, that the social planner will make the same allocation of labor in the LCD economy as in the simple one-good model studied in Chapter 2; L = 1/2. (3.45)
The rst-order conditions can also be used to derive the following expression for the labor L used in industry i; i gi bi L . L = Pn i gi bi i=1 (3.46)
To derive the capital allocation across rms for capital good j, the social planner solves Equations (3.43) and (3.44) to yield the optimal allocation of capital good j to industry i; gi ai,j Ki,j = Pn Kj. (3.47) i=1 gi ai,j For the LCD economy, the social planner sets employment at 1/2 and allocates factors across industries using weights that depend on a combination of factor shares and preference weights.1
3.5
Aggregate Supply and Demand
This section derives the properties of the Keynesian aggregate supply curve for the LCD economy. When I began this project I thought of aggregate supply as a relationship between employment and output. Intuition that was carried over from my own undergraduate training led me to think of this function as analogous to a movement along a production function in a onegood economy. This intuition is incorrect: A more appropriate analogy would be to compare the aggregate supply function to the rst order condition for labor in a one good model. Consider a one-good economy in which output, Y, is produced from labor L and capital K using the function Y = A L K L1 , (3.48)
The fact that L is equal to 1/2 follows from the assumption that the elasticity of the matching function is 0.5. I will maintain this assumption in the current chapter but I will need to relax it later in the book when I calibrate a model of this kind to match U.S. data.
1
3.5 AGGREGATE SUPPLY AND DEMAND
37
where A may be a function of aggregate employment because of the search externalities discussed above. It would be a mistake to call the function, A (L) K L1 , (3.49)
where L is replaced by L, the Keynesian aggregate supply function. At the risk of boring the reader through repetition, I will restate some passages that I cited in Chapter 2 relating to Keynes denitions of aggregate supply and demand. The points that I want to make are worth repeating because the original intent of the General Theory has been obfuscated by decades of misinterpretation. Keynes dened the aggregate supply price Z to be the expectation of proceeds which will just make it worth the while of the entrepreneurs to give that employment (Keynes, 1936, Page 24). By aggregate demand he meant, the proceeds which entrepreneurs expect to receive from the employment of L men, the relationship between D and L being written D = f (L) which can be called the Aggregate Demand Function. (Keynes, 1936, Page 25, L is substituted for N from the original). Keynes then asked us to consider what would happen if, for a given value of employment, aggregate demand D is greater than aggregate supply Z. In that case... there will be an incentive to entrepreneurs to increase employment beyond L and, if necessary, to raise costs by competing with one another for the factors of production, up to the value of L for which Z has become equal to D. (Keynes, 1936, Page 25, N replaced by L and italics added). It is not possible to understand this denition without allowing relative prices to change since the notion of competing for factors requires an adjustment of factor prices. In a general equilibrium environment Walras law implies that one price can be chosen as numeraire; the price chosen by Keynes
38
was the money wage. Given a value of w, competition for factors requires adjustment of the money price p and the rental rate r to equate aggregate demand and supply. In the one-good economy, the equation that triggers competition for workers is the rst-order condition w (1 ) Y = . L p (3.50)
Aggregate demand Z, is price times quantity. Using this denition, Equation (3.50) can be rearranged to yield the expression, Z pY = wL , (1 ) (3.51)
which is the Keynesian aggregate supply function. By xing the money wage Keynes was not assuming disequilibrium in factor markets; he was choosing a numeraire. Once this is recognized, the Keynesian aggregate supply curve takes on a dierent interpretation from that which is given in introductory textbooks. A movement along the aggregate supply curve is associated with an increase in the price level that reduces the real wage and brings it into equality with a falling marginal product of labor. In an economy with many goods the aggregate supply price, Z, is dened by the expression n X Z pi Yi . (3.52)
i=1
For the LCD economy the aggregate supply function has a particularly simple form since the logarithmic and Cobb-Douglas functional forms allow individual demands and supplies to be aggregated. The rst order condition for the use of labor at rm i has the form Li = bi Yi pi . w (3.53)
To aggregate labor across industries we need to know how relative prices adjust as the economy expands. To determine relative prices we must turn to preferences and here the assumption of logarithmic utility allows a simplication since the representative agent allocates xed budget shares to each commodity Yi pi = gi Z D , (3.54)
3.5 AGGREGATE SUPPLY AND DEMAND
39
where Z D is disposable income. Combining Equation (3.53) with (3.54) yields the expression, bi gi Z D . (3.55) Li = w Summing Equation (3.55) over all i industries and choosing w = 1 as the numeraire leads to the expression ZD = where 1 L (L) ,
n X i=1
(3.56)
gi bi .
(3.57)
Since the government budget must also balance T = Z, it follows that income and disposable income must be equal, that is; Z D = Z (1 ) + T = Z, and Eq. (3.56) can be written as, Z= 1 L (L) . (3.60) (3.59) (3.58)
Equation (3.60) is the Keynesian aggregate supply curve for the multi-good logarithmic-Cobb-Douglas economy. To reiterate; the aggregate supply curve in a one-good economy is not a production function; it is the rst-order condition for labor. In the LCD economy it is an aggregate of the rst order conditions across industries with a coecient that is a weighted sum of preference and technology parameters for the dierent industries. Can this expression be generalized beyond the LCD case? The answer is yes, but the resulting expression for aggregate supply depends, in general, on factor supplies, that is, Z will be a function not only of L but also of K1 , . . . Km . The following paragraph demonstrates that, given our special assumptions about preferences and technologies, these stocks serve only to inuence rental rates.
40
CHAPTER 3 AGGREGATE DEMAND AND SUPPLY The rst order condition for the j 0 th factor used in rm i can be written
as Ki,j =
ai,j Yi pi rj
(3.61)
Combining the rst order conditions for factor j and summing over all i industries leads to the expression Kj =
n X i=1
Ki,j =
Pn
i=1
ai,j Yi pi . rj
(3.62)
Exploiting the allocation of budget shares by consumers, Equation (3.54), one can derive the following expression, j Z rj = , Kj where j
n X i=1
(3.63)
ai,j gi .
(3.64)
Equation (3.63) determines the nominal rental rate for factor j as a function of the aggregate supply price Z and the factor supply Kj .
3.6
Keynesian Equilibrium
What determines relative outputs in the Keynesian model and how are aggregate employment, L, and the aggregate supply price Z, determined? As in the one-good model aggregate demand follows from the gdp accounting identity, n X D= pi Ci . (3.65)
i=1
In an economy with government purchases and investment expenditure this equation would have two extra terms as in the textbook Keynesian accounting identity that generation of students have written as Y = C + I + G. (3.66)
3.6 KEYNESIAN EQUILIBRIUM
41
P In our notation C is replaced by n pi Ci , Y is replaced by D, and G and I i=1 are absent from the model. The Keynesian consumption function is simply the budget equation n X pi Ci = (1 ) Z + T, (3.67) and since D = i=1 pi Ci and Z = L, from Equation (3.56), the aggregate demand function for the LCD economy is given by the equation, D = (1 ) L + T. Pn
i=1
(3.68)
In a Keynesian equilibrium, when D = Z, the value of income, Z K is given

The Aggregate Supply Function The Aggregate Demand Z = 1 L Function
Z, D
D = (1 ) ZK
+T
T 1
LK
1
Figure 3.1: Aggregate Demand and Supply by the equality of aggregate demand and supply; that is, ZK = T , (3.69)
42
and equilibrium employment is given by the expression. LK = Z K . The Keynesian aggregate demand and supply functions for the LCD economy are graphed in Figure 3.1.
3.7
Keynes and the Social Planner
How well do markets work and do we require government micro-management of individual industries to correct inecient allocations of resources that are inherent in capitalist economies? Keynes gave a two part answer to this question. He argued that the level of aggregate economic activity may be too low as a consequence of the failure of eective demand and here he was a strong proponent of government intervention. But he was not a proponent of socialist planning. In this section I will show that the model outlined in this chapter provides a formalization of Keynes arguments. If eective demand is too low, the model displays an inecient level of employment and in this sense there is an argument for a well designed scal policy. But the allocation of factors across industries, for a given volume of employment, is the same allocation that would be achieved by a social planner. To make the argument for scal intervention one need only compare aggregate employment in the social planning solution with aggregate employment in a Keynesian equilibrium. The social planner would choose 1 L = . 2 The Keynesian equilibrium at LK = T , (3.71) (3.70)
may result in any level of employment in the interval [0, 1]. For any value of LK < L we may say that the economy is experiencing Keynesian unemployment and in this case there is a possible Pareto improvement that would make everyone better o by increasing the number of people employed. The formalization of Keynesian economics based on search contains the additional implication that there also may be overemployment since LK may
3.7 KEYNES AND THE SOCIAL PLANNER
43
be greater than L . Overemployment is also Pareto inecient and welfare would, in this case, be increased by employing fewer workers across the board. Although a value of LK greater than L is associated with a higher value of nominal gdp (Z K > Z ), there is too much production on average and by lowering L back towards L the social planner will be able to increase the quantity of consumption goods available in every industry. In an overemployment equilibrium the additional workers spend more time recruiting their fellows than in productive activity. In the limit, as employment tends to 1, nominal gdp tends to its upper bound, 1/. But although gdp measured in wage units always increases as employment increases, for very high values of employment the physical quantity of output produced in each industry is very low and in the limit at L = 1, Yi is equal to zero in each industry and pi is innite. Every employed worker is so busy recruiting additional workers that he has no time to produce commodities. What about the allocation of factors across industries. Here the capitalist system fares much better. Equations (3.46) and (3.47), that determine factor allocations in the social planning solution, are reproduced below gi bi L = Pn L , i gi bi i=1 (3.72)
Given the resources Kj for j = 1, ..., m, Equation (3.73) implies that these resources will be allocated across industries in proportion to weights that depend on the preference parameters gi and the production elasticities ai,j . Equation (3.72) implies that the volume of resources employed, L , will be allocated across industries in a similar manner. Contrast these equations with their counterparts for the competitive equilibrium. The factor demand equations (3.61), and the resource constraints (3.62) are reproduced below, Ki,j =
n X i=1
gi ai,j Ki,j = Pn Kj. gi ai,j i=1
(3.73)
ai,j Yi pi , rj Pn
i=1
(3.74)
Kj =
Ki,j =
ai,j Yi pi . rj
(3.75)
44
Consumers with logarithmic preferences will set budget shares to utility weights (3.76) pi Yi = gi Z. Combining this expression with Equations (3.61) and (3.75) leads to the following equation that determines the allocation of factor j to industry i in a Keynesian equilibrium, gi ai,j Ki,j = Pn Kj . gi ai,j i=1 (3.77)
This expression is identical to the social planning solution, Equation (3.73). What about the allocation of labor across industries? The rst order conditions for rms imply bi pi Yi = wLi . (3.78) Combining this expression with Equation (3.76) and using the fact that Z K = LK gives, gi bi Li = Pn LK . (3.79) gi bi i=1 Pn where I have used the fact that i=1 gi bi . Equation (3.79) that determines the allocation of labor across industries is identical to the social planning solution with one exception; the ecient level of aggregate employment L is replaced by the Keynesian equilibrium level LK . It is in this sense that Keynes provided a General theory of employment; the classical value L is just one possible rest point of the capitalist system, as envisaged by Keynes, and in general it is not one that he thought would be found by unassisted competitive markets.
3.8
Concluding Comments
It is dicult to read the General Theory without experiencing a disconnect between what is in the book and what one has learned about Keynesian economics as a student. The most egregious misrepresentation is the notion of aggregate demand and supply that we teach to undergraduates and that bears little or no relationship to what Keynes meant by these terms. The representative textbook author has adopted the Humpty Dumpty approach
45
that, ...when I use a word, it means just what I choose it to mean - neither more nor less.2 The textbook aggregate demand curve slopes down; the Keynesian aggregate demand curve slopes up. The textbook aggregate demand curve plots a price against a quantity; so does the Keynesian aggregate demand curve, at least in name, but the aggregate demand price and the aggregate supply price of the general theory are very dierent animals from the price indices of modern theory. Beginning with Patinkin (1989), textbook Keynesians have tried to t the round peg of the General Theory into the square hole of Walrasian general equilibrium theory. The fact that the t is less than perfect has caused several generations of students to abandon the ideas of the General Theory and to follow the theoretically more coherent approach of real business cycle theory. The time has come to reconsider this decision.
The quote is from Alices Adventures in Wonderland, by Lewis Carroll.
Chapter 4 Saving and Investment

The models I have discussed so far are missing a central component of the General Theory; the idea that investment is the driving force of business cycles. Chapter 4 introduces this idea by developing a model with saving and investment. In the General Theory, Keynes argued that the distinction between these concepts was central to his theory of eective demand. I will explain this distinction with a two-period model populated by three generations of households. One of these generations, the young in the rst period, save for the future and invest in capital to produce commodities in the second period. In Walrasian general equilibrium models saving and investment are brought into equality by changes in intertemporal prices. In the Keynesian model they are equated by changes in employment. Explaining the dierence between these two mechanisms is the main purpose of this chapter. In earlier chapters I showed that there may be many equilibria, indexed by the stance of scal policy. That is not an entirely satisfactory account of Keynes message since Keynes saw scal policy as the remedy to mass unemployment, not its cause. To explain this idea, Section 4.10 introduces scal policy into the two period model and shows how government may design a tax-transfer system to restore full-employment.
4.1
The Model Structure
The chapter builds a model economy that has all the same features as the one commodity environment that I introduced in Chapter 2. In addition it 47
48
CHAPTER 4 SAVING AND INVESTMENT
has an extra period and a produced factor of production, capital. This richer structure allows me to discuss the idea that unemployment is produced by a lack of investment spending. Whereas eective demand in Chapter 2 was a function of scal policy; in this chapter it will also depend on the beliefs of investors. Two competing dynamic general equilibrium models are widely used in macroeconomics. The rst assumes the existence of a representative family that makes decisions for the innite future. The second is the overlapping generations model of Allais (1947) and Samuelson (1958). This latter model is the natural general equilibrium environment in which to discuss Keynesian economics since the representative agent environment places strong restrictions on the equilibrium interest rate that limit the possibility to discuss meaningful scal policies. I will assume that there are two periods, labeled 1 and 2 and three generations labeled 0, 1, and 2. In period 2 there are two generations alive. Generation 0 is old and owns the capital stock. Generation 1 is young and owns an endowment of time. At the end of period 1, generation 0 dies. At the beginning of period 2 generation 2 is born and is endowed with a production technology. Throughout the chapter a superscript will index the period in which a generation was born and a subscript will index calendar time, thus xs is the date t value of the variable x associated with the generation born t in period s.
4.2
Households
This section describes, in turn, the economic choices made by agents of each generation. I will begin by describing the choices made by the old and the young in the rst period; I refer to them as generations 0 and 1. I will then move on to the second period of the model and introduce the choices of a third generation that I refer to as generation 2. The decisions of generations 0 and 2 are limited and most of the action in this model takes place with the choices made by generation 1. In later chapters I will adapt the same structure by adding more periods and more generations each of which behaves like generation 1.
4.2 HOUSEHOLDS
49
4.2.1
The Initial Old
subject to the constraint
There are two coexistent generations in period 1. Generation 0, solves the problem 0 max j 0 C1 , (4.1) 0 {C1 }
0 p1C1 [(1 ) p1 + r1] K1.
(4.2)
There is unique commodity in each period that may be consumed or accumulated to be used as capital in production in the subsequent period. This commodity has money price p1 in period 1. K1 is an initial stock of capital owned by generation 0, r1 is the money rental rate for capital in period 1, 0 is the depreciation rate and C1 is consumption of generation 0 in period 0 0 1. Since I assume that the utility function j 0 (C1 ) is increasing in C1 , the households decision problem has the trivial solution
0 p1 C1 = [(1 ) p1 + r1 ] K1 ,
(4.3)
which directs the household to consume all of its wealth.1
4.2.2
The Initial Young
As in previous chapters I assume a unit measure of households with preferences over current consumption of household members. The representative generation 1 household receives utility from consumption in periods 1 and 2 and solves the problem
1 1 {C1 ,C2 ,K2 ,H1 }
max
1 1 1 1 j 1 C1 , C2 = g1 log C1 + g2 log C2 ,
(4.4)
where the preference weights g1 and g2 sum to 1, g1 + g2 = 1.

1
(4.5)
Throughout the book I will abstract from the bequest motive. Adding bequests will not change the main message of the book provided bequests are given because the giver obtains direct utility from the size of the gift.
50
Each household member is endowed with a single unit of time in period 1 and a fraction H1 of all members search for a job where H1 1. (4.6)
Since leisure does not yield utility, H1 will be chosen to equal 1. Of the workers that search, a fraction L1 nd a job and the remaining U1 are unemployed, hence, L1 + U1 = H1 = 1. (4.7) The relationship between H1 and L1 is given by the expression L1 = qH1 , (4.8)
where q is taken parametrically by the household. Generation 1s allocation problem is subject to the sequence of budget constraints 1 (4.9) p1C1 + p1 K2 w1 L1 ,
1 p2 C2 (r2 + p2 (1 )) K2 ,
(4.10)
where w1 is the money wage in period 1, K2 is capital carried into period 2, r2 is the money rental rate in period 2, and p2 is the money price in period 2. Households may borrow and lend with each other at money interest rate i and hence the intertemporal budget constraint is,
1 p1C1 +
p2 1 C w1 L1 . 1+i 2
(4.11)
The solution to this problem is characterized by the consumption allocation decisions 1 p1 C1 = g1 w1L1, (4.12)
1 p2 C2 = g2 w1L1, 1+i and the no-arbitrage condition, r2 p2 1+i= +1 , p2 p1
(4.13)
(4.14)
that denes the money interest rate i at which households have no desire to borrow or lend with each other.
4.3 FIRMS
51
4.2.3
The Third Generation

(4.15) (4.16)
In period 2 there is a third generation that solves the problem 2 max j 2 C2 , a 2 {K2 ,C2 }
2 p2 C2 p2 Y2 r2K2 ,
where output Y2 is produced with the technology

Y2 K2 .
(4.17)
The solution to this problem is given by the expression,

2 C2 = Y2
r2 K2 . p2
(4.18)
In later chapters, when I introduce an innite horizon model, each generation will be modeled like that of generation 1. To keep this two-period example as simple as possible I assume, in this chapter, that generation 2 owns the technology described by Equation (4.17) and that members of this generation rent capital from generation 1 and produce output Y2 . There is no labor market in period 2.
4.3
Firms
I have described production in period 2. This section describes the choices made by rms in period 1. Since the structure of this problem is a special case of the problem described in Chapter 3, I will be relatively brief in my description. There is a large number of competitive rms each of which solves the problem p1 Y1 w1 L1 r1 K1 , (4.19) max
{Y1 ,K1 ,V1 ,L1 ,X 1 }
subject to the constraints,

1 Y1 AK1 X1 ,
(4.20) (4.21) (4.22)
L1 = X1 + V1 , L1 = qV1.
52
As in Chapters 2 and 3, the rm must choose a feasible plan {Y1 , K1 , V1 , L1 , X1 } to maximize prot taking the wage w1 , the rental rate r1, the price p1 and the recruiting eciency q as given. A rm that allocates V1 workers to recruiting will hire qV1 = L1 workers of which X1 will be allocated to productive activity. The solution to this problem is characterized by the rst-order conditions (1 ) and the factor price frontier p1 = w1 [1 ] Q 1 r1 , (4.25) w1 Y1 = , L1 p1 (4.23) (4.24)
Y1 r1 = , K1 p1
where the aggregate productivity variable 1 Q= 1 , q is taken as given by the individual rm.
(4.26)
4.4
Search
The search technology is identical to that described in Chapter 2. There is a match technology of the form, 1/2 L1 = H1 V11/2 , (4.27)
where L is employment, equal to the measure of workers that nd jobs when H1 unemployed workers search and V1 workers are allocated to recruiting by = 1 and hence rms. Households choose H L1 = V11/2. In a symmetric equilibrium, (4.21), (4.28) and (4.26) imply Q = 1 L1 . (4.28)
(4.29)
53
4.5
The Social Planner
How would a benevolent social planner arrange production and consumption in this economy? This section addresses that question by studying the solution to the following constrained optimization problem. 0 1 1 2 max (4.30) 0 j 0 C1 + 1j 1 C1 , C2 + 2j 2 C2 , 0 1 1 {K2 ,L1 ,C1 ,C1 ,C2 }
0 1 C1 + C1 + K2 K1 L1 (1 L1 )1 + K1 (1 ) , 1 1 2 C2 + C2 K2 + (1 ) K2 ,
(4.31) (4.32)
where the numbers i are welfare weights that sum to 1, 1 + 2 + 3 = 1. (4.33)
The social planner chooses K2 , the amount of capital to carry into period 2, L1 , employment in period 1, and a way of allocating commodities to indi0 1 1 2 viduals, {C1 , C1 , C2 , C2 }. His problem is characterized as the maximization of (4.30) subject to the two feasibility constraints (4.31) and (4.32). The solution to this problem requires that the two inequalities (4.31) and (4.32) should hold with equality and, in addition, the following six rst order conditions should be satised,
1 1 + 2 K2 = 0, 1 1 1 1 = 0, (1 ) K1 L1 (1 L1 ) L1 1 L1 0 0 0 j1 C1 = 1 , 1 1 1 1 j1 C1 , C2 = 1 , 1 1 1 1 j2 C1 , C2 = 2 , 2 2 2 j1 C2 = 2 .
(4.34) (4.35) (4.36) (4.37) (4.38) (4.39)
The six equations (4.34) (4.39), and the two constraints (4.31) and (4.32) 0 1 1 2 determine the six variables, K2 ,L1 , C1 , C1 , C2 and C2 and the two Lagrange multipliers 1 and 2 associated with the inequality constraints (4.31) and (4.32).
54
The most important of these conditions is Equation (4.35) which implies that optimal employment in period 1, call this L , occurs when 1 1 L = . 1 2 (4.40)
This problem diers from a conventional social planning problem since there is an externality in the technology that is internalized by the social planner - this is the occurrence of the term 1 L1 in the production function in period 1. In all other ways, the problem is conventional. Given L1 , the planner chooses how to allocate commodities across individuals and across time. I will show below that the existence of this externality may make it dicult or impossible for a market economy to make the right employment decision and I will formalize this idea in the concept of a demand constrained equilibrium. But if this problem can be corrected, the existence of commodity and asset markets implies that a decentralized economy can produce a Pareto ecient allocation.
4.6
Investment and the Keynesian Equilibrium
We are used to thinking of general equilibrium in Walrasian terms. Agents take prices and endowments as given and form demands an equilibrium is a set of prices and an allocation of commodities such all markets clear and no individual has an incentive to alter his allocation through trade at equilibrium prices. For the Keynesian model we will require a dierent equilibrium concept since, by construction, there are not enough markets to determine equilibrium allocations. This section extends the DCE equilibrium concept of Chapter 2 to the two-period model. I will use this extended concept to introduce the idea that investment determines economic activity and I will show that there is an interval such that, for any value of investment expenditure in that interval, there exists a demand constrained equilibrium. Since the Keynesian model is missing an equation, there are many equivalent candidates for an equation with which to close it. Following the General Theory, this chapter closes the model with the assumption that investors form a set of beliefs about the future. Keynes called this animal spirits. But this assumption has many representations all of which will be consistent
4.7 THE DEFINITION OF EQUILIBRIUM
55
there is an equilibrium, characterized by values for prices {p1 , p2 , w1 , i} , con0 1 1 2 sumption allocations {C1 , C1 , C2 , C2 } employment L1, unemployment U1 , productions Y1 and Y2 and capital K2 such that no individual has an incentive to change his behavior given the prices and the quantities demanded and supplied for commodities in each period and for borrowing, lending and capital in the asset markets. In the labor market, employment is determined by matching the equilibrium numbers of searchers on each side of the market.
with a self-fullling equilibrium. In this chapter I have chosen to represent the assumption by assuming that the value of capital is determined by the beliefs of investors. Specically, let (4.41) I1 p1 K2 . I will refer to I1 as investment although this is a misnomer since it is in fact the money value of the next periods capital stock. I have chosen this denition because it simplies the equilibrium concept. By assuming that entrepreneurs have xed beliefs about the appropriate value of I1, I will be able to separate the equation that determines aggregate demand and employment from the equations that determine relative prices. The resulting dichotomy allows me to provide an interpretation of textbook Keynesian models that has a rm microfoundation. If I1 is large I will say that investors are optimistic and if I1 is small they 1 such that for any value of are pessimistic. I will show that there is value I I1 0, I1 , (4.42)
4.7
The Denition of Equilibrium
This section extends the denition of a demand constrained equilibrium from Chapter 2 to the two-period model with capital. Since this concept is based on ideas from the General Theory I will also refer to it as a Keynesian equilibrium and I will refer to equilibrium values of variables in the model with the superscript K, for Keynes. These values are to be contrasted with the superscript that denotes the social planning optimum when the planner uses the welfare weights i . Denition 4.1 (Demand Constrained Equilibrium) Let I1 be given by the equation, I1 = (1 ) (1 b) , (4.43)
56 where
For any given I1 0, I1 a symmetric demand constrained equilibrium (DCE) is (i) a six-tuple of prices {p1 , p2 , w1 , r1 , r2 , i}, (ii) a production plan {Y1 , Y2, K2, V1, L1 , X1}, 0 1 1 2 (iii) a consumption allocation {C1 , C1 , C2 , C2 } and (iv) a pair of numbers q and q: with the following properties. 1) Feasibility: 1 (4.45) Y1 AK1 X1 ,
0 1 C1 + C1 + K2 K1 (1 ) Y1 , 1 2 C2 + C2 Y2 + K2 (1 ) Y2 K2 ,
b = + g1 (1 ) .
(4.44)
(4.46) (4.47) (4.48) (4.49) (4.50) (4.51)
L1 V11/2 ,
X1 + V1 = L1 , I1 K2 = . p1 2) Consistency with optimal choices by rms: r1 Y1 = , p1 K1 Y1 w1 = (1 ) , p1 L1 r2 Y2 = , p2 K2 1 w r

1 1
(4.52) (4.53) (4.54) . (4.55)
p1 =
[1 ] Q
3) Consistency with optimal choices by households:

0 p1C1 = [(1 ) p1 + r1 ] K1 , 1 p1 C1 = g1 w1L1,
(4.56) (4.57)

1 p2 C2 = g2 w1 L1 , 1+i 2 p2C2 = p2Y2 r2 K2.
57 (4.58) (4.59)
4) Search market equilibrium: q = L1 , q= L1 , V1 (4.60) (4.61) (4.62)
L1 = V11/2 .
In Section 4.8 I will show that a DCE exists and in Section 4.9 I show how to compute the prices and allocations associated with this equilibrium.
4.8
Aggregate Demand and Supply
To show existence of a Keynesian equilibrium, this section develops aggregate demand and supply equations and shows that the equality of aggregate demand and supply results in an equilibrium employment level LK that is 1 feasible and that satises the optimality conditions of households. In Section 4.9 I show that there exist prices that support this allocation as a demand constrained equilibrium. An important feature of a Keynesian equilibrium is that there is a dierent demand constrained equilibrium for every value of I1 in the interval 0, I1 : All of these equilibria have the property that no investor has an incentive to deviate from his plan. As in previous chapters I will choose the money wage w1 as the numeraire and I dene aggregate supply to be the money value of gdp at which employers are indierent to hiring L1 workers. The function (L1 ) that has this property is found from the rst order condition for labor (4.23) and is given by the expression 1 Z1 = L1 (L1) . (4.63) 1 As in the General Theory I refer to Z1 p1Y1 as the aggregate supply price of employment, L1 . Period 1 aggregate demand, D1 is equal to i 0 h 1 1 (1 ) p1 K1 , D1 = p1 C1 + C1 + I (4.64)
58
where the rst term in square brackets is the money value of aggregate consumption and the second is the money value of investment. Using Equations (4.3) and (4.12), leads to the expression D1 = [(1 ) p1 + r1 ] K1 + g1w1 L1 + I1 (1 ) p1 K1, which, by using the rst-order conditions, (4.23) and (4.24) can be simplied as follows (4.65) D1 = bZ1 + I1 , where, This is the point where the denition of investment as I1 , the money value of period 2 capital, rather than p1I1 , the money value of additions to capital, leads to a considerable simplication of the equations that determine equilibrium. If I had chosen p1I1 as the object of investors beliefs, the equation that determines equality of aggregate demand and aggregate supply would have contained the additional term (1 ) K1 p1 . There is no conceptual diculty in following this alternative denition but it would break the separation of the equations that determine equilibrium prices from those that determine aggregate demand and supply. K K The Keynesian equilibrium occurs when D1 = Z1 . Imposing this condition and solving Equations (4.63) and (4.65) leads to the following expression for the equilibrium value of the aggregate supply price,
K Z1 =
b = + g1 (1 ) .
(4.66)
Equilibrium employment, LK is equal to 1
1 I1 . 1b
(4.67)
K LK = (1 ) Z1 . 1
(4.68)
The Keynesian equilibrium is illustrated in Figure 4.1. Since employment must lie in the interval [0, 1] and aggregate supply is dened by Equation (4.63) it follows that the maximum value of aggregate supply is equal to 1/ (1 ) . It follows from the linearity of the aggregate demand and supply equations that there exists a Keynesian equilibrium for any value of I1 where 0, I I = (1 ) (1 b) . (4.69) The following section establishes this claim formally by showing how the other variables of the model are determined.
4.9 FINDING VALUES FOR THE OTHER VARIABLES
59
D1 , Z1
Aggregate Supply
Z1 =
1 1
1 L1 1
Aggregate Demand
I1
1 1
D1 =
b L1 + I1 1 L
LK
Figure 4.1: Aggregate Demand and Supply
4.9
Finding Values for the other Variables
I have shown how aggregate supply and employment are determined in period 1 in a Keynesian equilibrium. It remains to be shown how the prices 0K 1K 1K 2K pK , pK , 1 + iK , the consumption allocations C1 , C1 , C2 and C2 , the 1 2 K capital stock K2 and the outputs Y1K and Y2K are determined in equilibrium. This section applies some simple algebra by rearranging rst-order conditions and budget identities and may be skipped without loss of content if the reader is inclined.
4.9.1
First Period Price and Output
I turn rst to the determination of prices and of the physical value of output, Y1K in the Keynesian equilibrium. The equilibrium price in period 1, pK can 1
60
be found by solving the equation pK = 1 where

K Z1 , Y1K
(4.70)
is the physical value of output. Using (4.68) this gives the following expression for the money price pK 1 K L1 1 1 K p1 = (4.72) 1 , A (1 ) K1 (1 LK ) 1
1 1 Y1K = AK1 LK 1 LK , 1 1
(4.71)
as a function of the endowment of capital, K1 and the value of employment at the Keynesian equilibrium, LK . pK is an increasing monotonic function 1 1 of LK , reecting the fact that the real wage falls as investors become more 1 optimistic and the economy moves up the aggregate supply curve. This is the same mechanism that was discussed in Chapters 2 and 3.
4.9.2
Second Period Capital and Output
K I now turn to the variables K2 and Y2K . Given I1 and pK it follows from the 1 1 that denition of I I1 K K2 = K , (4.73) p1 and hence K Y2 = K2 . (4.74)
4.9.3
Rental Rates and Consumption Allocations
Next consider the determination of real rental rates and consumption allo0K cation to each generation. Generation 0 consumes the amount C1 which is found from the budget equation,
0K C1 = (1 ) K1 + K r1 K1 . pK 1
(4.75)
The real rental rate r1 /p1 is found from the rst order condition for rental capital in period 1 K r1 YK = 1 . (4.76) pK K1 1
4.10 FISCAL POLICY IN A KEYNESIAN MODEL Equations (4.75) and (4.76) imply,
0K C1 = (1 ) K1 + Y1K .
61
K The second period real rental rate r2 /pK is found from the rst-order con2 ditions, K r2 YK = 2K , (4.77) pK K2 2
and generation 2s consumption is

2K C2 = Y2K K r2 K K = (1 ) Y2K . pK 2 2
(4.78)
Generation 1s consumption in period 2 is found from market clearing

1K 2K K C2 + C2 = (1 ) K2 + Y2K ,
(4.79) (4.80)
as
1K K C2 = (1 ) K2 + Y2K .
4.9.4
Second Period Prices
Finally we can solve for pK / 1 + iK , the present value of pK , from Equation 2 2 (4.13) as 1K pK C2 2 = g2 w1 LK . (4.81) 1 1 + iK which can be rearranged to give
K g2 (1 ) Z1 pK 2 = . 1K 1 + iK C2
(4.82)
It is worth pointing out that pK and iK are not separately dened in this 2 model since there is no role for a separate unit of account in period 2.
4.10
Fiscal Policy in a Keynesian Model
A Keynesian equilibrium can result in any value of employment in the interval [0, 1], but the social planner will choose L = 1/2. It follows that unless investors happen fortuitously to choose the correct value of I1 , the Keynesian
62
equilibrium may be one of over or under employment. This section describes the Keynesian remedy for this problem by putting scal policy into the model. Since there are three generations, a scal policy could conceptually consist of a level of government expenditure and a set of taxes and transfers indexed by the age of the household. I will exclude government expenditure since that raises the issue of public goods and instead I will consider policies that consist of an income tax rate , levied in period 1, and a transfer payment T to generation 1. I will show that for any value of I1 , there exists a tax-transfer policy { , T } that implements the full employment level of employment, L .2 1 Conceptually, it is possible to tax wage income and capital income at dierent rates and the generational burden of these taxes will dier. I will be concerned with the question; can scal policy maintain full employment? The answer to this question is yes and further, there are many policies that can implement full employment. To implement a policy that maintains full employment let 1 be the tax 0 1 rate on income and let T1 and T1 be lump-sum transfers to generations 0 and 1 in period 1. There are no taxes or transfers in period 2 and the tax rate on rental income is the same as the tax rate on wage income. These assumptions imply that aggregate demand is given by the expression 0 1 D1 = (1 1 ) Z1 + T1 + g1 (1 1 ) (1 ) Z1 + T1 + I1 . (4.83) The rst term on the right side is the consumption from after-tax rental in0 come of the old generation in period 1 and T1 is the transfer to the initial old. The term in square brackets is after-tax income of the young generation. Their labor income (1 ) Z1 is taxed at rate 1 and they receive a transfer T1. The parameter g1 is the marginal propensity to consume for these individuals. It is clear from this expression that there will be many choices 0 1 of T1 , T1 and 1 that force the equilibrium value, at which D1 equals Z1 , to occur at the planning optimum Z1 ; but most of these solutions will cause the government to accumulate debt that will need to be repaid in the second period. Instead, let us conne ourselves to tax transfer policies for which
0 1 T1 + T1 = 1 Z1 ,
(4.84) (4.85)
where
Z1 =
2
1 , 2 (1 )
I have called this full employment in line with the language of the General Theory although the model introduced here admits of over employment as well as under employment.
63
is the value of aggregate supply in the social planning optimum. These are balanced budget policies. Substituting (4.84) into (4.83) and solving for the Keynesian equilibrium leads to the expression,
K Z1 = 0 1 I1 + T1 (1 ) (1 g1) (1 g1) T1 . (1 ) (1 g1)
(4.86)
Notice that if the tax rate on wage and rental income is the same, as I have assumed, then a transfer to the old has a positive eect on equilibrium 0 income and a transfer to the young has a negative eect; T1 enters with a 1 K positive and T1 with a negative sign. It follows that if Z1 < Z (the case of underemployment in the Keynesian equilibrium) then the optimal policy is K to transfer income to the old and if Z1 > Z (the case of overemployment) it is to transfer income to the young. The reason is that the young save a fraction g1 of their income while the old consume all of it and so a transfer to the old increases aggregate demand while a transfer to the young reduces it.
4.11
Concluding Comments
The main early criticisms of Keynes work were theoretical, not empirical. It was pointed out that the General Theory does not have a satisfactory theory of the labor market. In Chapters 2 and 3 I constructed one-period demanddriven models to address these criticisms. Both of these chapters were based on a search-theoretic model of the labor market. Their purpose was to provide a microfoundation to the Keynesian theory of aggregate supply. Recall that the aggregate supply function (L1 ) is a relationship between Z1 , the supply price measured in dollars, and L1 , employment. Z1 is that expectations of proceeds [...i.e. nominal gdp...] that will just make it worth the while of the entrepreneurs to give that employment. Chapter 4 has developed the rst, and simplest, of several models that embody Keynes theory of eective demand. Aggregate demand D1 is the proceeds that entrepreneurs expect to receive from the employment of L1 men and it can be broken into two components, consumption and investment, each measured in dollars. I have provided a microfounded model in which consumption expenditure is a linear function of income and investment expenditure is determined by beliefs of investors about future productivity,
64
so called animal spirits. A Keynesian equilibrium, formally dened in this chapter as a demand constrained equilibrium, is a value of employment at which aggregate demand and aggregate supply are equal. Keynes emphasized that saving and investment are equated not by the interest rate, but by the level of economic activity. This chapter has provided an interpretation of that idea. We are used to teaching macroeconomics in the language of Walrasian general equilibrium theory. In Walrasian theory it is prices that clear markets and it is a change in the rate of interest that equates saving and investment. By adding a search externality and removing the spot market for labor I have provided a framework where there are not enough Walrasian prices to equate demands and supplies for all of the quantities. This framework goes beyond the General Theory by providing an explicit microfoundation to the Keynesian idea of aggregate supply.
Chapter 5 Presenting Business Cycle Facts

The rst four chapters of this book have presented relatively simple models of the aggregate economy designed to illustrate the mechanisms that distinguish a Keynesian from a classical model of the business cycle. In subsequent chapters I will present progressively more elaborate versions of articial economies that incorporate these same mechanisms with the ultimate goal of providing systems of equations that can be confronted with the data. As I describe how these economies work I will show how they explain business cycle facts. This chapter presents a way of presenting these facts that diers from approaches taken in most recent work in macroeconomics. Business cycles are recurrent events in which many dierent time series move together. To present their common features in a way that is apparent to visual inspection it helps to separate the trend in each time series from the cycle. Almost all recent papers on business cycles have removed these trends by passing each aggregate time series through the Hodrick-Prescott lter. Instead, I will present a method suggested by Keynes discussion of measurement in Chapter 4 of The General Theory; the measurement of aggregate variables using wage units. My purpose for presenting this alternative way of representing data is to illustrate some connections between medium frequency movements in economic time series that are not apparent in data that has been HP ltered. 65
66
CHAPTER 5 PRESENTING BUSINESS CYCLE FACTS
5.1
Whats Wrong with the HP Filter?
Before the advent of real business cycle theory in the 1970s macroeconomists confronted models with data using econometric methods. In the 1960s and early 1970s the state of the art in macroeconometrics consisted of the estimation of large simultaneous equation models identied by exclusion restrictions suggested by Keynesian theory. These models were estimated with systems methods and used as guides to policy makers. A good example is the Wharton model developed by Lawrence Klein at the University of Pennsylvania. The Keynesian econometric methodology developed by Klein and associates was criticized by Lucas in his 1976 Critique of Econometric Policy Evaluation on the grounds that microfounded structural equations should contain expectations of future variables. Since the parameters of these expectations should depend on the parameters of the rules followed by the policy authorities Lucas argued that rational expectations assumption would invalidate the practice of using xed parameter models as policy guides. The profession responded to the Lucas Critique in two dierent ways. The rst, introduced to economists in the Lucas and Sargent (1981) book Rational Expectations and Econometric Practice, was to develop appropriate econometric methods to estimate parameters in rational expectations environments. The second, explained most clearly in Ed Prescotts 1996 article, The Computational Experiment was to develop a new methodology, calibration, that lowered the standards of what it means for a model to be successful by requiring that a good model should explain only a limited set of empirical moments. In the introduction to their 1981 book, Lucas and Sargent argued for the development of econometric techniques that would recognize the crossequation restrictions imposed on a model by the rational expectations assumption. However, when simple RBC models were confronted with data they often performed poorly. As a response to this road block in the development of a new paradigm, Ed Prescott argued that one could not expect that a science in its infancy would initially outperform established science. In Prescotts view this did not mean that economists should abandon the new approach; he argued instead that a new paradigm brings its own anomalies and puzzles and that as real business cycle theory is developed it will eventually solve these puzzles and supplant its predecessor. In a related paper Hodrick and Prescott introduced the eponymous HodrickPrescott (HP) lter that soon became the dominant method for confronting
5.1 WHATS WRONG WITH THE HP FILTER?
67
business cycle models with business cycle facts. I will make two separate but related arguments in this chapter. First, I will argue that the correct approach to confronting models with data is the one advocated by Lucas and Sargent and that the new paradigm is now ready to be confronted with a full set of empirical moments. Second, I will argue that the ubiquitous use of the HP lter masks a serious shortcoming of the RBC model and that the use of ltered data to evaluate it overstates the models conformity with the facts. The rst issue of confronting models with data is beyond the scope of the current work and will be the focus of a subsequent companion volume. The second; the issue of presenting limited sets of stylized facts by detrending data is the main focus of this chapter. I will develop an alternative methodology to the HP lter that corrects what I will argue is a serious shortcoming of the HP approach. There are advantages to the presentation of HP ltered data and, properly used, the confrontation of theory with data can be simplied with its careful use. But the HP lter also has disadvantages that are often overlooked. Most business cycle models have implications for the comovements of variables at all frequencies. By ltering out the low frequency movements, one ignores implications of these movements for the relationships between economic variables that could potentially be used to discriminate between theories. Models in which the trend in data is assumed to be generated by a common non-stationary productivity shock should be ltered by removing a common trend. Instead, the HP-lter removes a dierent non-linear trend from every series thereby purging the data of potentially useful information. These low frequency movements are critical in assessing explanations of unemployment that deny the natural rate hypothesis.1 Figures 5.1 and 5.2 compare data detrended with the HP lter with my proposed alternative; measuring GDP in wage units. Notice from Figure 5.2 that the unemployment rate has low to medium frequency movements that move closely with real gdp. Compare this with Figure 5.1 which plots unemployment against real GDP measured as percentage deviations from the HP trend. Gdp measured in wage units moves more closely with unemployment
A second objection to the use of ltered data is one that is often raised by those who favor a traditional econometric approach to the analysis of time series data. Since the lter is two sided, it is not consistent to assume that ltered data can be modelled as if it were chosen by a forward looking rational agent operating in a world with no trend in productivity. Despite this objection, a number of inuential papers do use HP-ltered data in this way. See for example, the work of Thomas Lubik and Frank Schorfheide 2003.
1
68
2 3 Percent of labor force (Inverted scale) 4 5 6 7 8 9 10 50 55 60 65 70 75 80 85 90 95 00 Shaded areas are NBER recessions
.08 .06 Percent deviation from trend .04 .02 .00 -.02 -.04 -.06 -.08
Unemployment Rate
Real GDP
Figure 5.1: Unemployment and HP ltered GDP
1.80 1.75 1.70 1.65 1.60 1.55 1.50 1.45 1.40 Wage units
Unemployment Rate (Left scale) Gdp per member of labor force (Right scale)
Figure 5.2: Unempoyment and GDP per member of the labor force
5.2 MEASURING DATA IN WAGE UNITS
69
at medium to low frequencies whereas these movements are ltered out of the gdp series when it is detrended with the HP lter. It is for this reason that I am advocating the use of wage units as a way of measuring real magnitudes. An example of the kind of movement I have in mind is the slow increase in unemployment that runs from 1969 through 1985 which is mirrored in Figure 5.2 by a decrease in real gdp. This movement is absent from the GDP series in Figure 5.1 in which real GDP has been detrended with the HP lter. One would like to be able to ask the question: did the low frequency fall in gdp cause the increase in the unemployment rate or did causation run in the opposite direction? If one adopts a classical approach to the labor market then the low frequency movements in unemployment must be explained by movements in the natural rate of unemployment. In a Keynesian model, in contrast, low frequency movements in the unemployment rate are potentially explained by low frequency movements in aggregate demand. Since these low frequency movements are not present in the HP ltered data it is not possible to address the question of the possible dependence of the natural rate on scal or monetary policy.
5.2
Measuring Data in Wage Units
What are wage units? The inationary component of any two nominal series can be removed by taking their ratio. A natural series to use for this purpose is the nominal wage which grows for two reasons. It grows because of changes in the value of the monetary unit but it also grows because the real wage increases as the economy becomes more productive. Deating all nominal series by a measure of the money wage should lead to time series that are stationary if data are well described by series that uctuate around trends that grow at the same rate. The following gures depict US annual data from 1950 through 2004, described as per-capita ratios of nominal series, deated by the money wage. In all case, the nominal wage is constructed by dividing compensation to employees from the national income and product accounts by full and parttime equivalent employees. The wage series used is depicted in Figure 5.3 and is measured in units of thousands of dollars per year. My proposal, explained in this chapter, is to report all real magnitudes by dividing the corresponding nominal series by the nominal wage, measured as compensation to employees divided by full and part time equivalent em-
70

50 Thousands of dollars per year Shaded areas are NBER recessions
40
30
20
10
0 1930
1940
1950 1960
1970
1980
1990
2000
Average annual wage
Figure 5.3: The average annual real wage 1929-2004 ployees. I have chosen this measure of employment because it is continuously available from 1929 through the present day. Because my goal is to end up with a stationary measure I also need to remove the growth in GDP that occurs from increases in the number of working people. The following section discusses the best way to accomplish this.
5.3
Choosing a Per-Capita Measure
The number of employed people may uctuate for two reasons. First, more or less people may choose to enter the labor force in response to changes in incentives or changes in tastes. Second, the number of people employed may uctuate as members of the labor force enter and leave paid employment. These two reasons for uctuating employment correspond to dierent economic mechanisms. Simple Keynesian models account for changes in employment by assuming a xed or exogenous labor force and modeling changes in employment by assuming that the unemployment rate varies with aggregate demand. Real business cycle models abstract from unemployment and assume that all variations in employment are caused by variations in labor
5.4 INTERPRETING WAGE UNITS
71
force participation. In practice, both margins are likely to be important. Consider the following two alternative ways of removing the population trend. First, one might divide gdp in wage units by the adult population; a series constructed this way is reported in Figure 5.4. Second, one might divide by the labor force (employed plus unemployed adults); this series is reported in Figure 5.2 and reproduced in Figure 5.5 for comparison with Figure 5.4. Notice that the series in Figure 5.5 is approximately stationary and tracks unemployment very closely while that in Figure 5.4 has a marked upward trend. The dierence between them is accounted for by a substantial change in labor force participation in the post-war period that was accounted for by a big increase in the number of women who seek employment outside the home. It is likely that the secular increase in labor force participation was caused by exogenous changes in household preferences although there were surely also economic factors at work. An example of a non-economic cause is the experience of working women during WWII that caused many of these women to remain active in the labor force once the war ended and to inuence their own daughters to prepare for a working role outside of the home. Because of the secular trend in post-war participation I have deated all of the real measures presented in this book by the civilian labor force.
5.4
Interpreting Wage Units
To see what the use of wage units means in a concrete case, suppose that the data were generated by an economy in which a unique good was produced by competitive rms from capital and labor with a Cobb-Douglas production function using a constant returns-to-scale technology. In this case the rst order condition for labor would imply bY p = wL (5.1)
where b is a the elasticity of labor, Y is the quantity of output produced in physical units, p is the dollar price per unit, L is the number of worker-hours used to produce Y and w is the wage per worker hour. We have data on compensation to employees, wL, the number of full and part time equivalent employees, L, and nominal gdp, pY . In 1929 pY was equal to 103.7 billion dollars, compensation to employees was equal to 51.1 billion dollars, 37.7 million full or part time equivalent people were employed and the labor force
72
.95 .90 .85 .80 .75 .70 .65 .60 .55 Wage units
Unemployment Rate
Gdp per adult
Figure 5.4: Unemployment and GDP per adult
1.80 1.75 1.70 1.65 1.60 1.55 1.50 1.45 1.40

Wage units
Unemployment Rate (Left scale) Gdp per member of labor force (Right scale)
Figure 5.5: Unempoyment and GDP per member of the labor force
5.5 THE COMPONENTS OF GDP
73
(call this N) consisted of 47.6 million people. The yearly wage is computed as 51, 100, 000, 000 (wL) = = 1, 355 w= L 37, 700, 000 dollars per year. Gdp is computed as Z= 103, 700, 000, 000 1 pY 1 = N w 47, 600, 000 1, 355 (5.2)
which for 1929 gives a value of gdp per member of the labor force of 1.6 in units of pure number. If we let u= N L N
be the unemployment rate then combining this denition with Eqns (5.2) and (5.1) implies that Z should be equal to the inverse of the elasticity parameter b multiplied by one minus the unemployment rate. Z= (pY ) L 1L 1 = = (1 u) . N (wL) bN b
In other words, if the economy produced a single good then gdp in wage units would be equal to the inverse of the elasticity of labor in production (also equal to labors share of national income) multiplied by the employment ratio. Since there are multiple goods produced in the real economy this statement oversimplies the denition but it is still true that all measurements in wage units have units of pure numbers and that to the extent that production technologies are close to Cobb-Douglas, gdp measured in this way is a weighted multiple of the employment rate.
5.5
The Components of GDP
This section illustrates some comovements between the components of gdp and the unemployment rate that would be ltered out by the HP lter but that are clear in the data represented in wage units. Figure 5.6 illustrates the comovements of private investment expenditure, government purchases, private consumption expenditure and net exports with unemployment. In each of the panels of this gure one of the components of gdp is measured
74
on the right axis in wage units and the unemployment rate is measured on the left axis in percentage points with an inverted scale. The scale of the left axis, measuring the unemployment rate, is the same in all four panels but the right scale is dierent for each graph. Investment uctuates between 0.22 and 0.31, consumption uctuates between 0.96 and 1.16 and government purchases uctuate between 0.24 and 0.4. By contrast net exports uctuate between plus .02 and minus .08. These four series together add up to gdp which uctuates between 1.5 and 1.74 over the same period (see Figure 5.5) The top left panel illustrates that investment per member of the labor force moves quite closely with unemployment at high frequency, but there are lower frequency movements that it does not pick up. For example, between 1969 and 1980 unemployment trends up (recall that the axis is inverted) where as investment uctuates around a constant level. In contrast to the stationary uctuations in investment over this period, consumption and government purchases trend down mirroring the increase in the unemployment rate (see the bottom left and top right panels of Figure 5.6). This period is often referred to as the productivity slowdown since conventional measures of gdp per person grow at a slower rate in the 1970s. The Keynesian models I will develop in this book suggest that this slowdown was associated by an increase in the unemployment rate that itself was caused by the reduction in government purchases accompanying the end of the Vietnam War. The Keynesian model suggests that consumption depends on government purchases through a multiplier eect and in Chapter 7 I will explain this mechanism and develop a model that can account for the 1970s slowdown. The downward trend in the consumption series that runs from the late 1960s to 1980 is sharply reversed in 1981 and the subsequent drop in trend unemployment accompanies an increase in consumption from 0.96 in 1981 to 1.16 in 2004. Unlike the slowdown in the 1970s, this reversal of trend unemployment is not accompanied by a parallel increase in government purchases which instead continue to decline. This suggests that the same mechanism that accounts for the 1970s slowdown cannot be what is happening in the 1980s. If one is looking for policy induced explanations for movements in unemployment, an obvious candidate is the dramatic change in monetary policy that occurred in 1980 when Paul Volcker took over as Chairman of the Fed. Figure 5.7 plots the ex-post real interest rate (measured as the t-bill rate minus the annualized rate of change of the gdp deator) against consumption in wage units. The dramatic turnaround in consumption is accompanied by an increase in the real interest rate at the time of the Volcker
5.5 THE COMPONENTS OF GDP

2 3 Percent of labor force (Inverted scale) 4 5 6 7 8 9 10 50 55 60 65 70 75 80 85 90 95 00 Shaded areas are NBER recessions .34 .32 Percent of labor force (Inverted scale) .30 .28 .26 .24 .22 .20 .18 Wage units 2 3 4 5 6 7 8 9 10 50 55 60 65 70 75 80 85 90 95 00 Shaded areas are NBER recessions .52 .48 .44 .40 .36 .32 .28 .24 .20 Wage units
75
Unemployment Rate
Pri vate investment
Unemployment Rate
Government purchases
2 3 Percent of labor forc e (Inverted scale) 4 5 6 7 8 9 10 50 55 60 65 70 75 80 85 90 95 00 Shaded areas are NBER recessions
1.20 1.16 Percent of labor forc e (Inverted scale) 1.12 1.08 1.04 1.00 0.96 0.92 0.88 Wage units
2 3 4 5 6 7 8 9 10 50 55 60 65 70 75 80 85 90 95 00 Shaded areas are NBER recessions
.04 .02 .00 -.02 -.04 -.06 -.08 -.10 -.12 Wage units
Unemployment Rate
Private c onsumption
Unemployment Rate
Net exports
Figure 5.6: The Components of GDP measured in wage units

10 8 Percent per year 6 4 2 0 -2 1965 Shaded areas are NBER recessions 1.16 1.12 1.08 Wage units 1.04 1.00 0.96 0.92 1995
1970
1975
1980
1985
1990
Real Interest Rate
Consumption
Figure 5.7: The real interest rate and consumption
76
policy initiative. In Chapter 8 I will provide a development of the Keynesian model that can account for these events.
5.6
Concluding Comments
The data presented in this chapter is suggestive and it does not constitute formal statistical evidence for or against any given theoretical model. But the development of theory is guided by stylized facts and for the past twenty years, business cycle stylized facts have been those that were dened by real business cycle theorists. The RBC agenda is to construct models that can match the volatility and cross correlations of time series data that has been ltered by removing all but a narrow band of frequencies. By using this limited set of moments a generation of theorists has been starved of a set of the medium frequency facts. By detrending data in the way described in this chapter a new set of stylized facts become apparent and their explanation provides a new set of puzzles that I will attempt to make sense of in the following chapters.
Chapter 6 The Great Depression

In this chapter I develop an innite horizon model in which employment is determined by aggregate demand and I use it to tell the Keynesian story of the Great Depression. According to this story, the stock market crash of 1929 was due to a loss of condence in the economy that caused a calamitous drop in aggregate demand. This in turn caused an increase in unemployment that was socially inecient in the sense that the unemployed persons could and should have been protably employed in productive activity. This story was taught to generations of undergraduates as the leading explanation of the Great Depression by Keynesian economists of the post war period. The competition to this explanation was not the RBC model of Kydland and Prescott (1982) but an alternative story of market failure promoted by Milton Friedman (1948). Friedman disputed the impetus to the depression, for him it was the failure of the Fed to maintain sucient liquidity, but he did not dispute the fact that unemployment in the depression was socially inecient. Why do I need to revisit a story that was accepted by several generations of economists? The answer is that the theoretical foundations of this story have been discredited because Keynes did not construct a credible microfoundation to the theory of aggregate supply. In this chapter I will use the search model developed earlier to provide such a foundation. In so doing I will have cause to revisit an important debate that arose in the post-war literature. Does scal policy matter? Before the rational expectations revolution of the 1970s macroeconomists attempted to extend Keynesian economics to dynamic environments by building microfoundations to each of the components of the Keynesian model. Lucas (1967) and Treadway (1971) estimated models of investment, 77
78
CHAPTER 6 THE GREAT DEPRESSION
Milton Friedman (1957) provided a permanent income theory of the theory of consumption function and Friedman (1956) breathed new life into the Quantity Theory of Money by making the case for a stable demand-for-money function. A central goal of this research was to provide a quantitative explanation of the eects of government policy on employment and prices. This goal ran into theoretical diculties when Alan Blinder and Robert Solow (1973) pointed out that dynamic Keynesian models had no role for scal policy since a one dollar increase in government expenditure was predicted to crowd out an equal amount of private consumption expenditure if consumption and investment functions were derived from optimizing behavior by a representative family. In section 6.12 I will show that crowding out is a logical consequence of the representative agent model in which government cannot inuence the real rate of interest. Hence, the model developed in this chapter does not provide a good vehicle for explaining the wartime recovery.
6.1
Preferences
I will study a multi-commodity intertemporal representative family model in which there is a single capital good in xed supply. The simplication of non-reproducible capital enables me to draw out a relationship between the value of the stock market (represented by the value of capital) and the level of economic activity. There is a unit measure of identical representative innitely lived families. There are n consumption goods and Kt = 1 units of capital. Preferences are described by the following logarithmic utility function " # n X X Jt = s1 gi log (Ci,s ) , (6.1)
s=t i=1
where
n X i=1
gi = 1,
(6.2)
and the gi are preference weights. Each family sends a measure 1 of members to look for a job every period. All jobs last for one period and there is 100% labor market turnover. These assumptions are very strong and are made to facilitate the exposition of the model.
6.1 PREFERENCES The household faces the sequence of budget constraints pk,t Kt+1 = (pk,t + rrt ) Kt + wtLt
n X i=1
79
pi,t Ci,t ,
t = 1, ...,
(6.3) (6.4) (6.5) (6.6)
Ht 1, Lt = qt Ht , Ut = Ht Lt , and the no-Ponzi scheme constraint,

T
lim QT Kt+1 0. t
(6.7)
The notation is dened as follows. Ht is the measure of family members that search, Lt is the measure that nd a job and Ut is the measure that remain unemployed. wt is the money wage, pi,t is the money price of good i, qt is the probability that a searching worker will nd a job, Ci,t is consumption of good i, Kt is the familys ownership of capital, rrt is the rental price of capital and pk,t is the money price of a unit of capital. All of these terms are dened for each date t. All date t prices are in dierent date-t units of account that I refer to as date-t money. The variable QT represents the t relative price of date-T money in terms of date-t money and is given by the expression QT = t Qt t
T 1 Y k=t
1 , (1 + ik )
T > t,
(6.8) (6.9)
= 1.
I assume riskless borrowing and lending at money rate of interest it and a no arbitrage condition then implies that 1 + it = pk,t+1 + rrt+1 , pk,t (6.10)
where it is the money rate of interest between dates t and t + 1. Since all families are identical there will be no borrowing or lending in equilibrium.
80
6.2
The Consumption Function
Since the household derives no disutility from work, it will choose to send all of its members into the labor force to look for a job. At the end of each period all workers are red and, in the next period, the entire labor force is rehired. I make this assumption to facilitate exposition. Dropping it is an important extension that I will leave for a later chapter. The rst order conditions for the problem are represented by an Euler equation and a set of intertemporal rst-order conditions that together imply 1 = (1 + it ) , Ct Ct+1 where consumption expenditure, Ct is dened as Ct
n X i=1
(6.11)
pi,t Ci,t .
(6.12)
Equation (6.11) describes how aggregate consumption expenditure, measured in dollars, evolves over time. This equation will be central later in this chapter when I discuss crowding out. For completeness, it may be helpful also to write down the solution to the households problem by deriving an equation in which consumption expenditure is described as a function of prices, its time endowment and the hiring probability; all variables that it takes as given in equilibrium. This solution requires rst that we dene some alternative concepts of wealth. Let ht+1 ht = wt Lt + , (6.13) 1 + it be the human wealth of the family.1 By iterating this equation forwards and using the no-Ponzi scheme constraint (6.7) human wealth can be written in terms of prices and hiring probabilities, ht =
1
X s=t
Qs ws Ls = t
X s=t
Qs ws qs . t
(6.14)
This is a slightly non-standard denition since wealth would normally be dened as the net present value of the labor endowment. Here it is the net present value of labor income.
6.3 TECHNOLOGY The household also has nancial wealth in the forms of claims to capital (pk,t + rrt ) Kt . The sum of nancial and human wealth is total wealth Wt , Wt = (pk,t + rrt ) Kt + ht .
81
(6.15)
The solution to the household problem is to spend a xed fraction of total wealth on consumption goods and consumption expenditure is given by Ct = (1 ) Wt . (6.16)
I will need Equation (6.16) in Chapter 7 when I discuss a more complicated family structure although for the purpose of solving the representative agent model in this chapter it does not play a central role.
6.3
Technology
This section describes the production side of the economy. There is a unit measure of non-reproducible capital that must be allocated across industries in every period. There is also a unit measure of workers, all of whom will be allocated, in equilibrium, to the activity of labor market search. I assume that each industry is described by a Cobb-Douglas production function and that capital is rented in a competitive rental market. Labor is hired in a search market. I assume further that labor in each rm is divided between a recruiting department and a production department as in previous chapters. These assumptions lead to the observation that average and marginal products are equal and are equated to factor prices. Production is competitive and has the same structure as that described in chapter 3. Output of the i0 th commodity is denoted Yi,t , and is produced by a Cobb-Douglas function
ai bi Yi,t Ki,t Xi,t ,
(6.17) (6.18) (6.19)
where ai + bi = 1, and
n X i=1
Ki,t = Kt.
82
Ki,t is the rental demand for capital by rm i and Xi,t is the rms allocation of labor to production. Market clearing in each industry implies that Ci,t = Yi,t . (6.20)
Firms maximize prots taking pi,t , wt , rrt and qt as given. Each rm solves the problem
{Ki,t ,Vi,t ,Xi,t ,Li,t }
max
ai bi pi,t Ki,t Xi,t wtLi,t rrt Ki,t
(6.21) (6.22) (6.23)
Li,t = Xi,t + Vi,t , Li,t = qtVi,t ,
where Li,t is total labor hired by rm i and Vi,t is the labor that it allocates to recruiting. Substituting Eqns (6.22) and (6.23) into (6.21) and dening Qt = (1 1/qt ) , leads to the reduced form expression for prots;
ai pi,t Qbi Ki,t Lbi wt Li,t rrt Ki,t t i,t
(6.24)
(6.25)
which is maximized when ai, pi,t Yi,t = rrt Ki,t , and bi pi,t Yi,t = wt Li,t . (6.27) (6.26)
These expressions are identical to those that hold in an economy with a competitive labor market. This economy diers from the competitive model since the recruiting eciency parameter Qt is endogenously determined by aggregate economic activity but is taken parametrically by the rm; hence there is an externality in the labor market that is not priced. The following section combines Eqns (6.26) and (6.27) with consumer rst-order conditions to obtain some simple aggregate equilibrium relationships.
6.4 AGGREGATE SUPPLY
83
6.4
Aggregate Supply
In this Section I will introduce the variable Zt to denote nominal gdp. Recall that Ct is the nominal value of aggregate consumption and since there is no investment or government expenditure these two variables will be identical as a consequence of accounting identities. My goal here is to nd a relationship between Zt and Lt that I refer to as aggregate supply. From the solution to the households problem it follows that the consumer allocates a fraction gi of total consumption expenditure to good i; that is, pi,t Ci,t = gi Ct . Since all production of good i is consumed, this also implies that pi,t Yi,t = gi Ct , and, dening
n X i=1
(6.28)
(6.29)
pi,t Yi,t Zt ,
(6.30)
it follows that,
pi,t Yi,t = gi Zt .
(6.31)
Substituting (6.31) into the rst order condition for the choice of labor in industry i leads to the expression bi gi Zt = wt Li,t , (6.32)
which can be summed over all industries to give the following expression, Zt = wt Lt , where
n X i=1
(6.33)
gi bi .
(6.34)
Since money in each period is simply an accounting device there is a degree of freedom in choosing a price normalization in each period. I will choose the date t numeraire to be labor by setting wt = 1, t = 1, .... (6.35)
84
This normalization implies that pi,t is the inverse of the real product wage for each commodity and it allows me to write Eq (6.33) as Zt = 1 Lt , (6.36)
an equation that I referred to earlier (in Chapter 3) as the Keynesian aggregate supply curve. A similar exercise using the rst order condition for rental capital yields the expression Zt = rrt Kt , (6.37) where
n X i=1
gi ai .
(6.38)
In Section 6.8 I will use equations (6.36) and (6.37) to describe how the properties of aggregates behave in a demand constrained equilibrium.
6.5
Search and the Labor Market
As in Chapter 3 I assume there is an aggregate match technology that results in the following expression for aggregate employment,
1/2 Lt = Vt ,
(6.39)
where Lt is the measure of workers that nd jobs when a measure 1 of workers search and Vt workers are allocated to recruiting in aggregate by all rms. Each rm faces an individual hiring equation Li,t = qtVi,t , which, when aggregated over all rms, yields the expression L t = qt V t . (6.41) (6.40)
These equations can be rearranged to nd an expression relating the measure of workers that can be hired by a single recruiter, qt , to aggregate employment, Lt , 1 qt = . (6.42) Lt
6.6 EXPECTATIONS SHOCKS
85
6.6
Expectations and Technology Shocks
As in earlier chapters of this book, the absence of markets for the search time of workers and recruiters leads to an equilibrium model with one less equation than unknown. If rms and workers take all wages and prices as given then there is an equilibrium for every value of the sequence of hiring eectiveness parameters {Qs } . There are many possible ways of resolving s=t this indeterminacy each of which corresponds to a dierent possible belief about the future. In the General Theory, Keynes argued that the level of economic activity is pinned down by the state of long term expectations. In our model, this concept is represented by a self-fullling sequence of values for the capital good, that is, a sequence {pk,t }. I will call the economy dened in this chapter, a Keynesian economy. Parallel to it, there is another economy, identical in all respects, with three exceptions. First, the labor market is cleared by sequence of spot market auctions. Second, the household labor endowment is dierent in every period and third, the economy is hit by a sequence of technology shocks common to each industry. I call this economy - the Walrasian analog economy. For every sequence of non-negative numbers {pk,t }, there exists a demand constrained equilibrium (a concept that I dene carefully below) to the Keynesian economy. Associated with every sequence {pk,t } in the Keynesian economy, there corresponds a sequence of productivity shocks {Qt } and a sequence of labor endowments {Ht } such that every demand constrained equilibrium of the Keynesian economy is associated with a Walrasian equilibrium of the Walrasian analog economy.
6.7
The Social Planning Problem
Before discussing the properties of a Keynesian equilibrium I will provide a benchmark for what full employment means in this economy, by solving a social planning problem. As in earlier chapters in this book, one can show that the Keynesian equilibrium mimics the decisions of a Social Planner by allocating resources across industries in an ecient fashion; but the Keynesian equilibrium may fail to maintain full employment in a well dened sense. The purpose of this section is to dene what this means. Consider the problem,
86
CHAPTER 6 THE GREAT DEPRESSION " #
{C i,s ,Xi,s ,Vi,s ,Li,s ,Hs ,Ls ,Vs }
max
Jt =
such that
X s=t
s1
n X i=1
gi log (Ci,s )
(6.43)
ai b Ci,s Ki,s Xisi
i = 1, ...n
(6.44) (6.45)
Xi,s + Vi,s = Li,s ,

n X i=1
Li,s = Ls , s = t, . . . .
n X i=1
(6.46)
n X i=1
Ki,s = 1,
Vi,s = Vs , s = t. . . .
(6.47) (6.48) (6.49)
1/2 Ls = Vs1/2 Hs , s = t, . . .
Hs 1, Li,s =
s = t, . . .
Vi,s Ls s = t, . . . . (6.50) Vs Equation (6.43) is the objective function of the social planner (identical to that of the representative agent) and Eqns (6.44)-(6.50) dene the constraint set. As in earlier chapters, this problem can be simplied by using the match technology to eliminate Vi,s , Xis and Vs from the problem. Using Eqns (6.44), (6.45) and (6.50) one can write the production function for good i as
ai Ci,s Ki,s Lbi (1 Ls )bi . i,s
(6.51)
Using this simplication the social planning problem can be restated as " # n X X max Jt = s1 gi log (Ci,s ) (6.52)
{Ci,s ,Li,s ,Ls } s=t i=1
subject to Eqn (6.51) for each commodity at each date and the set of labor constraints represented by Eqn (6.46).
6.8 DEMAND CONSTRAINED EQUILIBRIUM
87
Proposition 6.1 (SP) The solution to the social planning problem has the following properties. Aggregate employment each period is given by the expression, 1 (6.53) Ls = , s = t . . . 2 and labor and capital are allocated across industries according to the equations, gi bi gi ai Li,s = , Ki,s = , s = t, . . . . (6.54) P P where n gi bi = and n gi ai = . i=1 i=1 As in the earlier models of this book, with a similar labor market structure, there is an optimal unemployment rate of 50%. I will return to this result below when I discuss whether scal policy can improve on the equilibrium.
6.8
Demand Constrained Equilibrium
I begin this section by dening a demand constrained equilibrium, which I refer to interchangeably as a Keynesian equilibrium, for the dynamic economy with multiple commodities. I will proceed to show that a Keynesian equilibrium exists and that it is represented by a set of aggregate equations that determine employment and gdp and a separate set of equations that describe how labor and capital are allocated across industries. Denition 6.1 A (bounded) state of (long-term) expectations is a non-negative sequence {pk,s } with a bound B such that s=t pk,s < B for all s. I dene the state of expectations to be a sequence of beliefs about the value of capital in all future periods. In a more general model, there will be a dierent sequence of beliefs for every type of reproducible capital and discrepancies between expectations and the interest rate will cause changes in investment expenditures. In this model I am abstracting from investment spending by assuming that there is a unique non-reproducible capital good.
88
Even in this simple environment changes in beliefs about the value of capital will have an eect on expenditure since long-term expectations inuence wealth which, in turn, inuences consumption expenditure. The following denition is of a demand constrained equilibrium in the innite horizon economy. Following this denition, I show that aggregate variables in a DCE follow a relatively simple equation. Denition 6.2 (Demand Constrained Equilibrium) For any state of expectations a demand constrained equilibrium (DCE) is an ntuple of price sequences {pi,s } i = 1, . . . n, a sequence of rental rates {rrs } a set of s=t s=t quantity sequences {Yi,s , Xi,s , Vi,s , Li,s , Ci,s , Hs } and a pair of sequences of s=t numbers {s , qs } , such that the following equations hold for all s = t, . . . : q s=t 1) Feasibility and Market Clearing.
ai bi Yi,s = Ki,s Xi,s ,
(6.55) (6.56) (6.57) (6.58) (6.59) (6.60) pi,s Yi,s , Ki,s hs , 1 + is
Ci,s = Yi,s ,
n X i=1
Vs =
Vi,s ,
Xi,s + Vi,s = Li,s , n n X X Ls = Li,s , Ks = Ki,s ,

i=1 i=1
1 1
Ls = Hs2 Vs2 , Hs = 1, pi,s Yi,s , Li,s Ks = 1. 2) Consistency with optimal choices by rms and households. 1 = bi,s rrs = ai,s
(6.61) (6.62) (6.63) (6.64)
Ci,s = gi (1 ) Ws , Ws = (pk,s + rrs ) Ks + hs , 1 + is = 3) Search market equilibrium: qs = Ls , qs = Ls . Vs hs = Ls + pk,s+1 + rrs+1 . pk,s
(6.65) (6.66)
6.8 DEMAND CONSTRAINED EQUILIBRIUM
89
Equations (6.55)-(6.60)) dene technologies, adding up constraints and market clearing conditions. Eqns (6.61)-(6.64) are rst order conditions that dene solutions to individual optimizing problems and (6.65) and (6.66) dene search market equilibrium. Proposition 6.2 (DCE) There exists a unique Demand Constrained Equilibrium for every state of expectations with bound B . (1 )
In a DCE, for s = t, . . ., aggregate expenditure, aggregate employment and the rental rate are described by Equations (6.67)-(6.69), Zs = 1 pk,s (1 ) , (6.67) (6.68) (6.69)
Ls = Zs , rrs = pk,s (1 ) .
Equations (6.70)-(6.71), which hold for all i = 1, ...n and all s = t, ..., determine the allocation of factors across industries, Ki,s = ai, gi , (6.70) (6.71)
Li,s = bi,gi Zs . The price in wage units of each commodity is given by the expression pi,s = Zs ai ai bi bi 1 1 , bi 1 Zs
(6.72)
and the physical quantity of each good produced is given by the equation Yi,s a i ai = (bi Zs )bi gi (1 Zs )bi . (6.73)
90
6.9
Keynes and the Great Depression
According Keynes, the Great Depression was caused by a failure of aggregate demand and the model developed in this chapter provides a simplied framework for understanding his explanation. In 1929 the stock market fell 13% in one day. The drop in stock market value was followed by drop in expenditure on new capital goods from 16% of gdp in 1929 to 6% in 1932 and a corresponding dramatic increase in unemployment from 8% to 25% of the labor force. The economy did not recover until 1942 when the United States entered World War II.
Z Z= 1 pk ,1929 Z= 1
Z=
pk ,1933
L*1933
L*1929
Figure 6.1: The Keynesian Explanation for the Great Depression Figure (6.1) illustrates the Keynesian explanation for these events. In 1929 investors lost condence in the economy causing a self-fullling drop in stock market prices and a subsequent fall in investment purchases. This in turn triggered a drop in consumption expenditure through a multiplier eect. In the model of this chapter pk is an exogenous driving variable and a fall in pk causes an increase in unemployment. On the gure, Z falls from 1 p to 1 pk,1933 and as the economy moves down the aggregate supply k,1929 curve employment falls from L 1929 to L1933 . Is this explanation consistent with the data?
6.9 KEYNES AND THE GREAT DEPRESSION
91
4 3 Normalized units 2 1 0 -1 -2 1930 1932 1934 1936 1938 1940 1942 1944 Investment in wage units (left sca le) Real stock market index (left scale) Unemployment Rate (right scale)
0 5 10 15 20 25 30 % of labor force (Inverse scale)
Figure 6.2: Investment, Stocks and Unemployment in the Depression
.8 .7 .6 Wage units .5 .4 .3 .2 .1 .0 1930 1932 1934 1936 1938 1940 1942 1944
0 5 10 % of labor force (Inverse Scale) 15 20 25 30 35 40
Unemployment Rate (right scale) Investment (left scale) Government Expenditure (left scale)
Figure 6.3: Unemployment Investment and Government Purchases in the Recovery
92
Figure (6.2) plots the value of the Standard and Poors Stock Market index in constant dollars against an investment series and the unemployment rate plotted on an inverted scale. This gure shows that although the model does not explain the investment data, since capital is xed, it does capture the increase in unemployment that accompanies the crash. Notice, however, that the recovery in the unemployment rate that occurred in the 1940s is not accompanied by an increase in the value of the S&P index - neither is it accompanied by an increase in private investment expenditure. Figure (6.3) plots the investment and unemployment series from Figure (6.2) and, in addition it plots the data for government purchases. The investment and government purchases data on this gure are measured in wage units and are comparable to each other. Notice that although investment falls, government purchases shoot up as the United States enters World War II in 1942. The Keynesian model explains the recovery with this fact since, in the textbook static version of the model, equilibrium gdp and employment are functions of autonomous expenditure which consists of the sum of investment and government purchases. I now turn to the question: Can the Dynamic Keynesian model developed in this chapter explain the wartime recovery?
6.10
Eciency of Equilibrium
Before discussing scal policy one would like to understand why it might be necessary. Keynes argued that unemployment was a waste of resources and that full employment could be restored by government expenditure nanced either by taxes or by borrowing. In this section I will give the rst of these assertions a theoretical foundation in the model developed in this chapter by comparing the eciency of a Keynesian equilibrium with that of the social planning optimum. Although the model I have described is inspired by the General Theory, it is not identical to it and, as in previous chapters, overemployment as well as underemployment is a possibility. Recall that the model-economy has a stationary ecient employment level of 50% in every period. Although a unit mass of workers searches for employment it is not ecient for all of them to be employed. Any employment level greater than 50% would require that each rm allocate too many of its workers to the recruiting department and would result in a fall in the physical output produced in each industry and
6.10 EFFICIENCY OF EQUILIBRIUM
93
a corresponding drop in social welfare. In a Keynesian equilibrium the value of GDP is proportional to the value of physical capital and is given by the expression Zt = pk,t . Employment, proportional to GDP, is equal to Lt = Zt, (6.75) (6.74)
where = (1 ) / (). In an economy with a stock market, pk,t represents the value of equity. The value of pk,t is determined by the state of long term expectations, which was also famously described in the General Theory as the animal spirits of investors. In the model of this chapter there is no underlying uncertainty in the physical environment but in reality technology, preferences and endowments as well as political and social variables are themselves changing in unknown ways. Keynes argued that the future cannot be quantied in a way that has become common in modern macroeconomics as agents are assumed to know the probability distributions of all uncertain future events. As a consequence, the belief of agents in the form of the animal spirits of investors becomes an independent driving force of the business cycle. I have argued elsewhere (1999) that animal spirits should be modeled by building general equilibrium models in which there is an indeterminate continuum of equilibria, indexed by beliefs. In my earlier work these equilibria represented dierent non-stationary paths each of which converged to the same steady state. The model in this chapter is an extension of this idea to allow beliefs to inuence the steady state itself. How does the Keynesian equilibrium compare with the social planning solution? The answer is that there is a continuum of Keynesian equilibria indexed by {pk,t } . In any given period there will be a unique value p = k,t 1 , 2 (6.76)
such that when the stock market price pk,t is equal to p , the Keynesian k,t Equilibrium implements the social planning solution and Lt = 1/2. If pk,t < p , there will be ineciently high unemployment and if pk,t > p gdp and k,t k,t employment will be too high. In this case gdp is high because prices are
94
high and welfare and physical output could be increased by an increase in the unemployment rate: in common parlance, the economy is overheating. In either case, if pk,t is too high or too low, the Keynesian equilibrium will be inecient in the sense that a dierent belief by investors would result in an unambiguous increase in social welfare. In a calibrated model one might expect the socially ecient unemployment rate to be considerably less than 50%; perhaps 5% would be a good guess. If unemployment rose to 20% (or higher) as it did in the early 1930s the additional 15% represents workers that could have been gainfully employed in producing consumption and investment goods. Even if only half of this additional unemployment was due an ineciency of the kind modeled in this chapter, the implied welfare loss would be substantial. During the 1930s government spending was widely discussed as a possible remedy to the Great Depression but the remedy was not eectively put into practice until 1942 when the United States entered World War II. In the textbook Keynesian model consumption is a function of income that is itself the sum of investment, consumption and government spending. In this simple model an increase in government spending causes an increase in equilibrium income that in eect, pays for itself. The following section explores the possibility of telling a similar story in the context of the intertemporal representative agent model of this chapter.
6.11
Household and Government
Consider the following variation on the model developed so far. Let there be a government that purchases commodities Gi,t in each period. To keep the model simple I will assume that Gi,t = gi Gt (6.77)
where the weights gi for i = 1, . . . n are the same as the preference weights of the consumer. This assumption allows me to abstract from distribution eects associated with changes in the composition of aggregate demand between consumption and government purchases. To pay for its purchases, the government levies an income tax at rate t on labor income, or it may borrow money from households by issuing debt Bt . The assumption that there is no capital tax is not inconsequential since one might wish to use capital taxes or subsidies to inuence intertemporal prices. Since tax-subsidy schemes of this
6.11 HOUSEHOLD AND GOVERNMENT
95
kind are not the focus of the expansionary scal policies that I am interested in I will abstract from capital taxation in this section and I will return to the issue later in the book. The government faces the following sequence of constraints, Bs+1 = Bs (1 + is1 ) + Gs s Ls , s = t, ... with the no-Ponzi scheme condition
T
(6.78)
lim QT BT 0. t
(6.79)
Here, tLt is the tax revenue from the labor income tax and I have used the normalization ws = 1 to remove ws from the ow budget constraint. The sequence of constraints (6.78) together with Eqn (6.79) , is equivalent to the single innite-horizon constraint
X s=t
Qs G s t
+ Bt (1 + it1)
X s=t
Qs s Zs , t
(6.80)
where I have replaced Lt by Zt from the aggregate supply curve. How does the introduction of a government that spends, taxes and issues debt, inuence the solution to the consumers problem? Recall that the household solves the problem, " # n X X s1 max Jt = gi log (Ci,s ) . (6.81)
{Ci,s } s=t i=1
When we introduce taxes and government debt into the model the budget constraint faced by the household becomes n ! X X X Qs pi,s Ci,s Qs s L s + A t . (6.82) t t
s=t i=1 s=t
where At = pk,t + rrt + Bt (1 + it1 ) , (6.83) represents its initial wealth. Aggregating rst-order conditions for this problem leads to the consumption Euler equation, 1 = (1 + it ) , Ct Ct+1 (6.84)
96
and riskless arbitrage implies, 1 + it = pk,t+1 + rrt+1 . pk,t (6.85)
It follows from this analysis that the introduction of government purchases does not alter the households consumption-Euler equation. This fact has an important implication for the usefulness of the representative agent model in telling the story of the 1940s recovery.
6.12
Crowding Out
In this section I dene a government expenditure plan and a scal policy and using these denitions I describe the characteristics of an equilibrium in an economy with government. I dene a class of scal policies that restricts spending by government to have the same distributional pattern as spending by households. Although this restriction is not strictly necessary in the sense that one could dene an equilibrium without it, the assumption simplies the characterization of an equilibrium. It would not be surprising if expenditure by government on a particular sector of the economy has distributional consequences by changing relative prices but that is not what one normally means by the eectiveness of scal policy. A scal policy as I dene it has two components. First, it is a decision by government to purchase a given quantity of goods and services in every period. Second; it is a decision on whether those purchases should be nanced by raising taxes or by borrowing. Denition 6.3 (Expenditure Policy) A (distributionally neutral) expenditure policy is a set of non-negative sequences {Gi,s } and an initial debt s=t level Bt (1 + it1 ) such that Gi,s = gi Gs (6.86) for all s. An expenditure policy together with a pair of sequences { s , Bs } s=t is called a scal policy. If there exists a pair of sequences { s , Bs } such s=t that the budget equation
X s=t
Qs G s t
+ Bt (1 + it1 )
X s=t
Qs s Zs , t
(6.87)
6.12 CROWDING OUT
97
is satised then the expenditure policy {Gi,s } is said to be feasible for price s=t sequence {Qs }. t Given this denition, a particular class of stationary policies is of particular interest. Denition 6.4 (Stationary Fiscal Policy) A feasible (distributionally neutral) scal policy is stationary if the sequences {Gi,s , s , Bs } do not depend s=t on s. Given these denitions I now provide a relatively straightforward extension of Denition 6.2 to show how a Keynesian equilibrium is modied in the presence of government expenditure. Denition 6.5 (DCEG) For any state of expectations {pk,s } and any s=t distributionally neutral expenditure policy {Gi,s } a demand constrained s=t equilibrium with Government (DCEG) is an ntuple of price sequences {pi,s } s=t i = 1, . . . n, a sequence of rental rates {rrs } and implied present value s=t prices {Qs } , a set of quantity sequences {Yi,s , Xi,s , Vi,s , Li,s , Ci,s , Hs } , a t s=t s=t set of tax and debt sequences { t , Bs } such the the policy is feasible for s= the present value prices {Qs } , and a pair of sequences of numbers {s , qs } , q t s=t such that the following equations hold for all s = t, . . . : 1) Feasibility and Market Clearing.
ai bi Yi,s = Ki,s Xi,s ,
(6.88) (6.89) (6.90) Ks =

n X i=1
Ci,s + Gi,s = Yi,s , Xi,s + Vi,s = Li,s , Vs =

n X i=1 n X i=1
Vi,s ,
Ls =
Li,s ,
1
Ki,s ,
(6.91) (6.92) (6.93)
Ls = Hs2 Vs2 , Hs = 1, Ks = 1.
2) Consistency with optimal choices by rms and households. 1 = bi,s pi,s Yi,s , Li,s rrs = ai,s pi,s Yi,s , Ki,s (6.94)
98
CHAPTER 6 THE GREAT DEPRESSION Ci,s = gi (1 ) Ws , hs (1 s ) hs = Ls + , 1 + is As = (pk,s + rrs ) Ks + Bs (1 + is1) Ws = As + hs , pk,s+1 + rrs+1 1 + is = . pk,s (6.95) (6.96) (6.97) (6.98) (6.99)
3) Search market equilibrium: qs = Ls , qs = Ls . Vs (6.100) (6.101)
This denition diers in three ways from (6.2). First, Eqn (6.89) is modied to recognize the allocation of resources between household and government sectors. Second, the denition of human wealth in Eqn (6.96) is modied to include only the after tax value of labor income. Third, nancial wealth of the household sector, dened in Eqn (6.97) includes government debt. Denition 6.6 (Stationary DCEG) A DCEG is stationary if all variables are independent of calendar time. I now have enough machinery to dene the main idea of this section. I will deal with the case in which households have stationary pessimistic expectations in the sense that pk is constant and permanently less than p . k Stationarity is a strong assumption but a useful one since it is the case that Keynes believed was characteristic of the Great Depression. In the General Theory he argued that unemployment may be an equilibrium phenomenon (in the sense of a stationary state). To capture this feature I assume that agents expectations are unchanging and that the economy is in a stationary Keynesian equilibrium with an unemployment rate that is ineciently high. I would like to be able to model Keynes prescription of increased government expenditure as a way out of the Great Depression. The following proposition demonstrates that the representative agent environment is not a good vehicle with which to make this case because one dollar of government expenditure is predicted to crowd out an equal amount of private consumption expenditure.
6.12 CROWDING OUT
99
Proposition 6.3 (Crowding Out) Let {pk,s } be a bounded stationary state of expectations such that pk,s = pk < , s = t . . . .. (1 )
1 pk,s (1 ) , s = t. . . . . There exists a unique stationary Demand Constrained Equilibrium with government. This equilibrium has the following characteristics. Aggregate expenditure, aggregate employment and the rental rate are described by Equations (6.102)-(6.104), 1 pk (1 ) , (6.102) Zs = Z = Ls = L = Z, (6.103) pk (1 ) rrs = rr = . (6.104) Equations (6.105)-(6.106), which hold for all i = 1, ...n and all s = t, ..., determine the allocation of factors across industries, ai, gi , (6.105) Ki,s = Ki = Gs = G Li,s = Li = bi, gi Z. (6.106) The price in wage units of each commodity is given by the expression bi Z ai 1 bi 1 pi,s = pi = , (6.107) ai bi 1 Z
Let {Gs } be a stationary sequence of expenditures such that
and the physical quantity of each good produced is given by the equation a i ai Yi,s = Yi = (bi Z)bi gi (1 Z)bi . (6.108) Consumption and government purchases of each commodity are allocated as follows Z G Yi,s , Ci,s = (6.109) Z G Gi,s = Yi,s . (6.110) Z
100
The proof of this proposition mirrors that of Proposition 6.2 and it hinges on the fact that the household Euler equation is unchanged by the introduction of government. In aggregate, it can be written as pk,s+1 + rrs+1 1 , (6.111) = Cs Cs+1 pk,s or, using the rst order conditions from production, 1 pk,s+1 + Zs = . Cs Cs+1 pk,s In a stationary equilibrium this implies 1 1 . Zs = pk
(6.112)
(6.113)
Equation (6.113) implies that gdp in a stationary equilibrium is independent of government expenditure and is a function only of the state of expectations. Since Z = C + G, (6.114) it follows that a one dollar increase in government expenditure must crowd out one dollar of private consumption expenditure. This is exactly the point made by Blinder and Solow in 1973 and since government spending and private spending is allocated in the same proportion across industries the allocation of each commodity to households and government is in proportion to aggregate spending as in Eqns (6.109) and (6.110).
6.13
Concluding Comments
What have we learned from this exercise? When economists of the postwar period began to provide microfoundations to Keynesian economics they turned to the Ramsey model of a representative agent as the simplest formal framework within which to model the evolution of dynamic equilibrium models. The static consumption function modeled consumption as a function of income but it was soon realized that a forward looking agent should be concerned not just with current income but with wealth or in Friedmans terms, permanent income.
101
The Keynesian remedy to unemployment was to replace investment by government expenditure and the Hicks-Hansen framework illustrated in a relatively simple model why this ought to work. The rational expectations revolution of the 1970s threw out the Hicks-Hansen apparatus because it did not cope well with the simultaneous appearance of high ination and high unemployment in the 1970s. But the model could not have been rejected so quickly on empirical grounds if it was not already on weak theoretical foundations. The rst part of this book has attempted to shore up the theoretical foundations of aggregate supply by providing a sound microfoundation to the idea that there may be a continuum of stationary equilibrium unemployment rates indexed by beliefs. This has led us down a road and towards a conclusion that was trod in 1973 by Blinder and Solow. Chapter 7 moves forward by breaking away from the representative agent framework and providing a model in which scal policy matters.
102
6.14
Appendix to Chapter 6
Proof of Proposition 6.1. Let i,s be the multiplier on the i0 th constraint (6.51) at date s, and let be the multiplier on (6.46). The following three rst order conditions follow from the choice of Ci,s , Li,s and Ls s1 gi = i,s , Ci,s i,s bi Ci,s = , Li,s n X i,s bi Ci,s = . 1 Ls =1 s1 = Ls . Combining (6.115) with (6.117) yields, s1 = (1 Ls ) . Together, these equations imply Ls = 1/2. (6.120) (6.119) (6.115) (6.116) (6.117)
Combining (6.115) with (6.116) and summing over i gives, (6.118)
To obtain the allocations of labor across industries combine Eqns (6.115), (6.116) and (6.118). The allocation of capital follows from a similar analysis using the rst-order condition for capital. Proof of Proposition 6.2. The proof of existence is constructive. Since labor supply is bounded above by 1, and since, in a DCE, Ls = Zs , from Eq (6.36), Zs is bounded above by 1 . By aggregating the consumer Euler equations one obtains Eq (6.11) which can be combined with (6.10) and the market clearing conditions to give 1 pk,s+1 + rrk,s+1 = . (6.121) Zs Zs+1 pk,s Using Eq (6.37), (6.60) and rearranging terms, 1 pk,s+1 = + , Zs Zs+1 pk,s pk,s
(6.122)
6.14 APPENDIX which can be iterated forward to obtain the expression 1 = 1 + + 2 ... . Zs pk,s
103
(6.123)
Since (0, 1) the innite sum on the RHS of (6.123) converges to (1 ) and rearranging this expression then leads to Eq (6.67). Since Zs is bounded above by 1 , it follows that a valid equilibrium requires pk,s . (1 ) (6.124)
Eq (6.68) follows from (6.36) and (6.69) follows from combining (6.67) with (6.121). Equations (6.70) and (6.71) follow directly from rearranging the rst order conditions for the rm. To obtain Equation (6.72), note that the production function for good i can be written, in reduced form, as
ai Yi,s = Qbi Ki,s Lbi , s i,s
(6.125)
where it follows from (6.24) (6.36), and (6.42) that Qs = (1 Zs ) . (6.126)
Using the fact that the consumer spends a fraction gi on good i leads to the equation gi Zs . (6.127) pi,s = Yi,s Combing Equations (6.125) and (6.127), substituting for Ki,s and Li,s from (6.70) and (6.71) leads to Eq (6.72). Equation (6.73) is derived similarly.
Chapter 7 The War-Time Recovery

This chapter builds a dynamic model in which scal policy matters. The taking o point is a paper by Olivier Blanchard (1985) who showed how to combine the properties of an overlapping generations model with those of the innite horizon model in a tractable way. The resulting model is similar to that of Chapter 6 with a few key dierences. There are many generations alive at the same time; each generation has a dierent consumption allocation based on its date of birth. Because each generation has logarithmic preferences and the same human wealth it is possible to derive a simple set of equations in aggregate variables that describes the properties of an equilibrium. These equations are very similar to those of the representative agent economy; but the representative agents Euler equation is replaced by an aggregate consumption equation in which income and wealth can aect consumption in the steady state. The main purpose of this chapter is to demonstrate that this model can account quantitatively not only for the Great Depression, but also for the war-time recovery
7.1
Household Structure
In the model studied in this chapter there is a continuum of agents, indexed by date of birth. I refer to them interchangeably as households or dynasties. Each household survives into the subsequent period with a xed probability . Every period a measure (1 ) of households dies and, to keep the population xed, a measure (1 ) of new ones is created. There are no bequests in this economy and no population growth although these are features that 105
106
CHAPTER 7 THE WAR-TIME RECOVERY
can be added with little diculty. As in Chapter 6 there are n consumption goods and Kt = 1 units of capital. Preferences of a typical dynasty created at date h are described by the logarithmic utility function " # n X X h Jth = ()st gi log Ci,s , t h (7.1)
s=t i=1
where
the gi are preference weights and is the probability that the dynasty surh vives into the subsequent period. The superscript on Ci,s denotes the birth date of the dynasty and as in the previous chapter s and t index calendar time and i indexes commodity. Each dynasty sends a measure 1 of members to look for a job every period. All jobs last for one period and there is 100% labor market turnover. The dynasty faces the sequence of constraints Ah t+1 = (1 + it1) Ah t + wtLt + T Rt Tt At = 0, t Ht 1, L t = qt H t , Ut = Ht Lt , and the no-Ponzi scheme condition,
T n X i=1 h pi,tCi,t ,
n X i=1
gi = 1,
(7.2)
t = h, ..., (7.3) (7.4) (7.5) (7.6) (7.7)
lim QT Ah 0. t t+1
(7.8)
As in Chapter 6, Ht is the measure of family members that search, Lt is the measure that nd a job and Ut is the measure that remain unemployed. wt is the money wage, pi,t is the money price of good i, qt is the probability that a h searching worker will nd a job and Ci,t is consumption of good i by dynasty h. I have omitted the index h from Ht and Lt because the assumptions that families are large and the labor market turns over every period implies that
7.1 HOUSEHOLD STRUCTURE
107
every family will have the same employment level. The terms T Rt and Tt represent lump-sum transfers and taxes that are independent of household. Ah is the familys assets, and it is the interest rate received by households. t For the reasons discussed below, this is greater than the interest rate paid by government or by rms. All terms are dened for each date t. Although all date t prices are in date t units of account this is a real model and I choose the normalization wt = 1 for all t. Hence, eectively, all variables are measured in wage units and pi,t is the inverse of the real product wage. QT t represents the relative price of date T labor in terms of date t labor and is given by the expression QT = t Qt t
T 1 Y k=t
1 , (1 + ik )
T > t,
(7.9) (7.10)
= 1.
Assets Ah are claims against a competitive annuity sector that owns the t capital stock and outstanding claims to government debt, denoted Bt . In aggregate, I assume X Ah = At (7.11) t
h
where the sum is over all agents alive at date t. At is the total liabilities of the annuities sector. Assets of this sector are equal to the capital stock Kt , valued at price pk,t , and a stock of outstanding government liabilities, Bt. Hence, At = Bt 1 + ig (7.12) t1 + (pk,t + rrt ) Kt . Since the annuities sector is competitive, the stream of government interest payments and corporate rental payments received is paid out in aggregate to families. When a family dies its assets are returned to the annuity company and in return it receives a return greater than the interest received on the companys assets. I assume riskless borrowing and lending at money rate of interest it and a no arbitrage condition then implies that pk,t+1 + rrt+1 (1 + it ) = = (1 + ig ) , (7.13) t pk,t
where it is the money rate of interest between dates t and t + 1 received by families, pk,t is the value of a capital good, rrt is the rental rate and ig is the t
108
interest rate paid by the government on a one period security. These facts plus the assumption that Kt = 1 allow me to rewrite Eqn (7.12) as At = Bt (1 + it1 ) + (pk,t + rrt ) .
7.2
Government, Dynamic Eciency, and the Interest Rate
As in Chapter 6, the government purchases commodities Gi,t in each period and it allocates its expenditure across commodities in the same way as consumers, Gi,t = gi Gt . (7.14) The government faces the sequence of constraints, Bs+1 = Bs 1 + ig s1 + Gs Ts + T Rs , s = t, ... with the no-Ponzi scheme condition
T
(7.15)
lim QT BT 0, t
(7.16)
As in the previous chapter I assume no tax on capital and I have used the normalization ws = 1. The switch to lump-sum taxes, as opposed to the proportional labor income tax in Chapter 6, has no important consequences because labor is inelastically supplied. It has the benet of simplifying the algebra. The sequence of constraints (7.15) together with Eqn (7.16) , is equivalent to the single innite-horizon constraint
X s=t
Qs t
(Gs + T Rs ) + Bt (1 + it1 )
X s=t
Qs Ts . t
(7.17)
Recall that I assume that there a single unit of capital that does not depreciate and thus there is no investment expenditure in this model. This assumption is not inconsequential. In competitive models with a single innite horizon representative agent the equilibrium is dynamically ecient and competitive equilibria are Pareto optimal. Dynamic eciency implies that the interest rate will exceed the growth rate. In overlapping generations economies, of which the Blanchard structure used in this chapter is an example, this result no longer holds and equilibria may be both dynamically
7.2 DYNAMIC EFFICIENCY
109
inecient and Pareto ranked. Pareto inecient equilibria that arise as a consequence of dynamic ineciency have the property that the interest rate is less than the growth rate. Equilibria of this kind can never arise in models in which there is a single representative agent since the net present value of the aggregate endowment becomes innite. Since this endowment is owned by a single impatient individual an equilibrium price path that valued the aggregate endowment at innity would be inconsistent with market clearing. One would not expect Pareto eciency to hold in the economy described in this chapter since there is a missing market: this leads to the inecient search unemployment that is the focus of this book. One might nevertheless expect that, in a representative agent economy, the equilibrium interest rate would always be greater than the growth rate for the same reason that this property holds in Walrasian economies. Further, one would expect that in an overlapping generations model with Keynesian unemployment, one might nd equilibria in which the interest rate is less than the growth rate. The argument is the same as that put forward by Shell (1971) for the case of the standard overlapping generations model. Innite aggregate wealth is feasible in equilibrium as long every single individual has nite wealth: this requires that the model have an innite number of agents as they do in this model. Although Keynesian equilibria with a low interest rate are possible in some models with an innite number of agents, they are ruled out in this one by the assumption that capital does not depreciate. This places a lower bound of zero on the rate of interest for equilibria in which capital is held. In any such equilibrium, the following condition will hold, 1 pk,t+1 + rrt+1 Rg = t. Rt (1 + it ) = pk,t Rt is the gross real interest rate in wage units at which households can trade with annuity companies, (pk,t+1 + rrt+1 ) /pk,t is the gross real return on capital held in these companys portfolios and Rg is the gross real return at which t the government can borrow and lend. In a stationary equilibrium Rg = 1 + rr p
and since rr 0, the interest rate paid by government, Rg 1, must be greater than or equal to zero. This must be true, a fortiori, of the interest rate earned by households since R is greater than Rg by the multiple 1/.
110
7.3
The Household Problem
Sections 7.3 and 7.4 draw on the ideas developed in Blanchards (1985) paper to develop an analogue of the consumption Euler equation that holds for aggregate consumption in the economy with births and deaths. First, I restate the problem of each dynasty which chooses a consumption sequence h Ci,s to solve the problem " # n X X h max Jth = ()tt gi log Ci,s , (7.18)
{C} s=t i=1
subject to the constraints (7.3) and (7.8). Dene human wealth by the expression ht = Lt + T Rt Tt +
ht+1 , (7.19) (1 + it ) and notice that ht is independent of dynasty because all agents have the same expected future. Future income is discounted by the survival probability . Dening Cth as n X h h Ct = Ci,t (7.20)
i=1
which states that the dynasty devotes a constant fraction (1 ) of its wealth to consumption in each period where Cth , Ah and ht are all measured t in wage units.
one can derive the following expression for the value of dynastic consumption expenditure at date t Cth = (1 ) Ah + ht , (7.21) t
7.4
The Aggregate Consumption Function
Blanchards idea was to give every household the same death probability, independent of age. This assumption, together with that of logarithmic preferences, makes aggregation possible. Dene aggregate consumption and aggregate nancial wealth as X X Ct = Cth , At = Ah . (7.22) t
h h
7.4 THE AGGREGATE CONSUMPTION FUNCTION
111
Household wealth Ah and household consumption expenditure Cth will be t dierent across dynasties with dierent histories but, since the preferences lead to a consumption function that is linear in wealth, the consumption functions of individual households can be aggregated. The clever assumption that makes this work is that X ht = ht = ht, (7.23)
h
where ht is independent of superscipt-h because a dynasty that has existed for a thousand years has the same survival probability as one newly created. The last equality in (7.23) follows from the fact that I maintain a unit measure of dynasties at all time by assuming that there are as many births as deaths.1 If one thinks of dynasties as people, the literal interpretation would be that a ninety year old man has the same life expectancy as a new born baby. A more reasonable interpretation is that of Philippe Weil (1989) who suggested that we think of these agents as dynasties in which occasionally children are unloved and left no bequest. Denitions (7.22) and (7.23), combined with the individual consumption function (7.21), lead to the expression Ct = (1 ) [At + ht ] . (7.24)
Combining this with the denition of human wealth, Eq (7.19), leads to the following representation of aggregate consumption Ct = Ct+1 + (pk,t + Rt1 Bt + Zt + T Rt Tt ) Rt (7.25)
where the compound parameters, and are dened as follows, 1 = 1 + , Zt = =

n X i=1
(1 ) (1 ) ,
(7.26)
pi,tYi,t ,
(7.27)
is the value of gdp measured in dollars and Rt (1 + it ) ,

1
(7.28)
Formally, the summation sign represents the integral over a unit measure.
112
is a compact notation for the rate of return between periods t and t + 1. The derivation of Equations (7.24) and (7.25) is contained in Appendix 7.8. In the case = 1, it follows = 0, = and the model collapses to a representative agent economy. For values of less than 1, the model behaves very dierently from the representative agent model even when is close to 1. For the case in which 0 < < 1, aggregate consumption depends not only expected future consumption but also on income and wealth with a coecient . Although one might expect to be small if and are close to 1, the model with positive may still behave very dierently from the representative agent version.
7.5
Aggregate Equations of Motion
This section brings together four equations that together determine the properties of an equilibrium in this economy. My goal is to derive a graphical apparatus than can be used to analyze scal policy in a steady state equilibrium. The equations are as follows; Ct = Ct+1 + (pk,t + Rt1 Bt + Zt + T Rt Tt ) , Rt Rt1 1 = pk,t + Zt pk,t1 , (7.29)
(7.30) (7.31) (7.32)
Bt+1 = Bt Rt1 + Gt + T Rt Tt , Zt = Ct + Gt .
Equation (7.29) is an aggregate consumption function that describes how current consumption expenditure depends on expected future consumption expenditure, income and wealth. Since this is a perfect foresight model (at least for aggregate variables) expected future consumption is equal to realized future consumption. Wealth includes the value of the stock market, represented by pk,t and the value of government debt, Bt . Equation (7.30) follows from the zero prot assumption in the nancial services industry, Equation (7.31) is the government budget constraint and (7.32) is the gdp accounting identity. The denition of equilibrium given in Chapter 6 can be modied in a natural way to give a denition of equilibrium for this economy with many
7.6 STEADY STATE EQUILIBRIA
113
agents. The main complication involves recognizing that the allocations of each dynasty will dier according to its date of birth. Once this modication is made, the equations that describes aggregate variables, for any given bounded price sequence, are those represented by Equations (7.29) (7.32). Notice that when = 1, there is a single representative agent and, in this case, the model collapses to that of the representative agent economy studied in Chapter 6. One might be tempted to think that for close to 1, the model behaves in essentially the same way. This assumption is not correct as the following analysis shows.
7.6
Steady State Equilibria
Imposing the assumption that all variables are time independent leads to the following representation of a steady state equilibrium 1 = (pk + RB + Z + T R T ) , C 1 (7.33) R pk Z= (R 1) , (7.34) T G T R = B (R 1) , (7.35) Z = C + G. (7.36) A given state of long term expectations is captured by the self-fullling belief that the stock market price will equal pk . Taking pk as given and given a feasible scal policy {B, T, T R}, this system describes four equations in the four unknowns R, Z, C and G. The following analysis reduces this system to two equations in R and Z by eliminating G and C using Eqns (7.35) and (7.36) and replacing their values as functions of Z and R in Eqns (7.33) and (7.34). To facilitate the description and analysis of the steady state of the model 1 consider the following denition of the function g : , R+ , dened as follows, 1 . g (R) = (7.37) 1 1 R 1 This function is decreasing on , and has the property that g 1 as R .
114
Using the denition of g one can rearrange Equations (7.33) (7.36) to give the following two expressions IS Curve: and, IR Curve: Z= p (R 1) . (7.39) Z= g (R) (pk + B) + G, 1 g (R) (7.38)
Equation (7.38) is a steady state variant of the equation that, following Alvin Hansen and John Hicks, has been referred to in generations of Keynesian textbooks as the IS curve. It diers in two ways from the usual representation of the IS curve. First, the equation recognizes that the government budget constraint must be satised and it replaces government expenditure by a function of debt and the interest rate. Second, textbook treatments of this equation assume that the savings rate is constant and the IS curve slopes down because investment expenditure is interest sensitive. In this model there is no investment and instead, the IS curve slopes down because the Blanchard model requires that saving is a function of the interest rate. Equation (7.39) represents combinations of the interest rate and aggregate expenditure that are consistent with a zero prot equilibrium. On Figure 7.1 I have labelled this curve IR for interest rate. This curve replaces the vertical aggregate supply curve (gdp independent of R) of textbook classical models. Note also that I have plotted expenditure, Z on the vertical axis in contrast to the usual textbook convention of plotting R on this axis. The gure illustrates, in this model, the predicted eect of a drop in the value of the stock market. The point {Z , R } represents the models description of the state of the economy in 1929 before the market crash. This point is chosen in a way such that pk implements the social planning optimum.2 The point {Z1, R1 } is the position in 1933; after the crash. The eect of a fall in pk is to shift the IS curve to the left and the AS curve to the right. Since the positions of both AS and IS curves depend on p, Z is predicted unambiguously to fall, but the eect on R is ambiguous and depends on B. The picture draws the situation for B = 0 in which R is predicted to remain unchanged.
Of course thee is no presumption that the economy was at the social planning optimum in 1929. Indeed, there is considerable circumstantial evidence to suggest that pk in 1929 was well above p . I have nevertheless chosen this starting point for illustrative purposes. k
2
7.6 STEADY STATE EQUILIBRIA
115
* IR ( pk )
IR ( p1 )
Z1
* IS ( pk )
IS ( p1 )
R* , R1
Figure 7.1: The Keynesian Equilibrium in 1929 contrasted with that in 1933
Z
First best
IR ( p1 )
Z* Z1
Effect of fiscal, expansion
IS ( p1 , B2 )
IS ( p1 , B1 )
R * , R1
Figure 7.2: The Expansionary Eect of Fiscal Policy
116
Figure 7.2 illustrates the predicted eect of a scal expansion that raises government debt in the steady state from B1 its 1933 value, to B2, its 1945 value and Figure 7.3 shows the historical values of the debt to gdp ratio for the period 1929 through 1950 which went from 40% of gdp in 1933 to 120% in 1945.
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 30 32 34 36 38 40 42 44 46 48 50
Government debt as a fraction of gdp
Figure 7.3: The magnitude of changes in debt in the US data These pictures and equations are suggestive but they raise an important empirical question: How well can a calibrated model account for the magnitudes of the changes observed in the data? A rst guess at this question might give a pessimistic answer precisely because of the crowding out issue raised in Chapter 6. Although calibration of the survival probability is ultimately an empirical question, a model that required very low values of would not be plausible since it would suggest that the horizon of the average family is short. But if is close to 1; intuition suggests that crowding out will hold approximately and that wealth eects in consumption in the steady state will be small. This intuition turns out to be incorrect because the magnitude of aggregate wealth eects in consumption depends on the ratio of two factors. In the numerator of this ratio is the compound parameter
7.6 STEADY STATE EQUILIBRIA which is close to zero if is close to 1. In the denominator is the term 1 1 R
117
which is also small if the interest rate is close to . In calibrated models these two terms approximately balance each other and there are thus sizeable wealth eects in the steady state. Table 7 1 documents this assertion by studying the eects of a government scal expansion in a calibrated model. Table 7.1: 1929 1933 1945 1929 1933 Data 0.26 100 0.14 100 0.21 100 1.60 100 0.47 181 0.18 129 0.10 50 1.18 74 0.73 281 0.63 450 0.06 29 1.64 102 1945
Simulation Gov. (wage units) debt (% of 1929) 0.26 100 0.14 100 11.7 100 0.67 100 0.47 181 0.18 129 5.8 50 0.40 60 0.73 281 0.63 450 3.3 29 0.67 100
Gov. (wage units) exp. (% of 1929)
S & P index number price Gdp (% of 1929) (wage units) (% of 1929)
Fiscal Policy as a Response to a Fall in the Value of the Stock Market = 0.97, = 0.98, = 0.66 The table compares the values of gdp, government debt, government purchases and the value of the Standard and Poors stock market index for the years 1929, 1933 and 1945. In 1929, gdp measured in wage units was 1.6.; it fell to 1.18 in 1933 (its lowest value of the decade) and in 1945 had rebounded to its 1929 value. The stock market fell 50% from 1929 to 1933 and had fallen even further by 1945 when it was just 29% of its 1929 value. I argued in Chapter 6, that the Keynesian model can explain the fall between 1929 and 1933 but that the representative agent version would have diculty explaining the war-time recovery. By adding the Blanchard generational structure the theoretical possibility arises that the recovery might also
118
be explained since government debt, in that structure, is net wealth. But can the model explain the magnitude of the depression and of the recovery? To address this question I simulated the steady state of the model for three dierent values of government purchases, government debt and the stock market. The simulated data is reported in the left panel of Table 7.1. I chose plausible values of the three parameters , and . is the discount factor and I chose an annual discount factor of 0.97. The parameter represents the survival probability of a dynasty and here I chose 0.98 which implies that the expected duration of a dynasty is 50 years. The parameter is labors share of gdp which I set to 0.66. The results are not highly sensitive to plausible variations in these numbers although I plan to estimate these parameters formally in a separate exercise. Here I am merely interested in the answer to back of the envelope calculation to nd out if the model has a hope of explaining the data quantitatively. Since the government purchase and debt data are measured in wage units they are comparable and I fed these number directly into the calculation. I deated the stock market index by wage units to account for changes in the value of the labor standard during the depression; but, since the Standard and Poors value is an index number and not a dollar value it must still be scaled by a constant to bring it into units that are directly comparable with the dollar value of government debt. To pick this constant I chose a value that matched as closely as possible the return of gdp to its 1929 value in my computational experiment. The result is reported in the table.3 The table shows that the model can reproduce a large drop in gdp although it overstates the magnitude; simulated gdp in 1933 is 60% of its 1929 value in the model but 74% in the data. If the stock market price is calibrated to a higher value, the model can reproduce the percentage drop in gdp but it understate the magnitude of the recovery. Since my computations assume that 1929, 1933 and 1945 are all steady states it is perhaps not surprising that the numbers cannot come closer than this to the data. Notice also that the model produces GDP of 0.67 in 1929 whereas the true value is 1.6. Since the model excludes investment - this is also perhaps not surprising. The main feature that the reader should take away from this table is the fact that a model with long lived dynastys, each of which has an expected duration of 50 years, generated large wealth eects from government debt
I used a factor that set the stock market price equal to 80% of the ecient price in 1929 in a world with no government.
3
7.7 CONCLUDING COMMENTS and government purchases.
119
7.7
Concluding Comments
I ended Chapter 6 on a pessimistic note - although the search model of unemployment developed in that chapter could potentially explain the magnitude of the Depression; it seemed that the model could not account for the wartime recovery. In this chapter I amended that model by adding a richer generational structure and showed that scal policy of the same magnitude as that observed in the U.S. data can inuence the equilibrium of the economy in a quantitatively signicant way. This is perhaps surprising since agents in the model have realistically long horizons and one might think that the model behaves much like the representative agent economy of Chapter 6. We saw that in practice, this is not the case for realistic values of the model parameters.
120
7.8
7.8.1
Appendix to Chapter 7
The Household Problem
n X i=1
The budget constraints of the household consist of the sequence of equations, Ah t+1 = Rt1Ah t + Lt + T Rt Tt
h pi,t Ci,t , t h,
(7.40)
where I dene (1 + it ) Rt , together with the no-Ponzi scheme condition

T
lim QT Ah 0. t t+1
(7.41)
Since we assume the existence of perfect life insurance markets the agent may trade future claims to consumption that occur if and only if the agent is alive at that date. A promise to pay one unit of account at date s if and only if the agent is alive at that date will trade for price Qs st at date t < s. t Iterating Eqn (7.40) forwards and making use of (7.41) allows one to write a single lifetime budget constraint
X s=t
Qs st t
n X i=1
h pi,s Ci,s
X s=t
Qs st (Ls + T Rs Ts ) + Ah Rt1 , t t
(7.42)
or more compactly
X s=t h Qs st Cs ht + Ah Rt1 , t t
(7.43)
where human wealth ht is dened as ht = and

h Cs = X s=t
Qs st (Ls + T Rs Ts ) , t X
i h pi,s Ci,s ,
(7.44)
(7.45)
is consumption expenditure at date s. The objective function of the family is " # n X X h st Jth = () gi log Ci,s , t h (7.46)
s=t i=1
7.8 APPENDIX
121
and maximizing (7.46) subject to (7.42) leads to to the rst order condition gi ()st = Qs stpi,s , t h Ci,s (7.47)
P where is the Lagrange multiplier on (7.42). Since i gi = 1, Eqn (7.47) can be summed over all goods at each date to give the following expression relating consumption expenditure at date s to its present value price and the multiplier , h (7.48) st = Qs Cs . t Substituting (7.48) into (7.43) gives the following solution for 1 1 = (1 ) ht + Rt1 Ah . t
(7.49)
It follows from (7.48) that consumption expenditure at date t is equal to Cth = (1 ) ht + Rt1Ah . (7.50) t
7.8.2
The Consumption Function
This subsection derives an aggregate expression for the consumption function. Recall that human wealth at dates t and t + 1 are related by the recursion ht = Lt + T Rt Tt + ht+1 . Rt (7.51)
Rearranging Eqn (7.50), substituting it into (7.51) and making use of (7.40) gives the following expression, Cth Rt1 Ah = Lt + T Rt Tt t (1 ) h Ct+1 h h + Rt Rt1 At + Lt + T Rt Tt Ct , (7.52) Rt 1
which can be rearranged to give h Ct+1 1 (1 ) h h = Lt + T Rt Tt + At (1 ) + . Ct 1 Rt (1 ) (7.53)
122
h Since this equation is linear in Cth , Ct+1 Lt and Ah it can be summed over t the unit measure of households to yield the aggregate expression 1 (1 ) Ct+1 Ct = (Lt + T Rt Tt + At ) (1 ) + . (7.54) 1 Rt (1 )
Dene the following constants: 1 = 1 + , and notice that At = Rt1 Bt + (pk,t + rrt ) Kt. Using the facts that Lt + rrt Kt = Zt from the national income accounting identity, Kt = 1 from the non-reproducability of capital and Lt = Zt from the aggregate supply equation (6.36) developed in Chapter 6, we have the following intermediate expression, Lt + At = pk,t + Rt1 Bt + Zt . (7.56) = (1 ) (1 ) , (7.55)
Substituting this into Eqn (7.54) and making use of denition (7.55) yields the result, Ct = Ct+1 + (pk,t + Rt1 Bt + Zt + T Rt Tt ) Rt (7.57)
which is the equation we seek.
Chapter 8 The Post-War Experience: After the Accord

8.1 Introduction
This chapter is about the economic history of the United States in the period from 1951 through to the present. The chapter begins in 1951 when the accord between the Fed and the treasury allowed the Fed to conduct active monetary policy. Before that date the Fed had agreed to a policy of buying treasury bills at a xed low interest rate, a measure that was introduced to facilitate war time nancing. In the period after 1951 the Fed began actively to manipulate the short-term interest rate in an eort to manage the economy. With the end of war-time price controls, the fear was a recurrence of the Great Depression1 but ination soon took over as a more immediate concern. A series of recessions in the 1950s was followed by an economic expansion in the 1960s and during the 1970s growth slowed down. This slowdown is often attributed to increases to the price of oil in 1973 and again in 1979. At the same time that growth slowed, ination increased. In 1952 the annual rate of wage ination was 3%; by 1981 it had reached a peak of 11% . The increase in ination was accompanied by a simultaneous increase in the unemployment rate. The coincidence of low growth, high unemployment and high ination was dubbed stagation in the popular press. In 1980 the Fed took aggressive action to end ination by raising interest rates to unprecedented levels and
1
Robert Hetzel and Ralph Leach 2001, Page 34.
123
124
CHAPTER 8 POST WAR EXPERIENCE
the subsequent decades through the early part of the 21st century have been a period of high growth, low unemployment and low ination. This chapter documents the facts beginning in 1951 and ending in 2006.
8.2
The Impact of the Fed-Treasury Accord
Figure 8.1 illustrates the behavior of a short term interest rate, wage ination and the implied ex post real interest rate for the period from 1929 through 2006. The interest rate is the three month treasury bill rate for the period from 1934 to 2006 and for the years from 1929 to 1934 it is a six month commercial paper rate spliced and scaled to equal the t-bill rate in 1934. The rate of wage ination is constructed from the same annual wage series that was used in Chapters 6 and 7 to generate data in wage units. The ex-post real rate is constructed as Rt = (1 + it1 ) wt1 wt (8.1)
where it1 is the annualized t-bill rate between years t 1 and t, and wt is the money wage in year t constructed as compensation to employees divided by full and part-time equivalent employees. Notice that before 1951 the t-bill rate is smooth and wage ination is volatile. The period after 1951 is one of less volatile uctuations in wages but more volatile interest rate movements.2 Active monetary intervention in the form of counter-cyclical interest rate movements begins in 1951 with the Accord and for this reason I have chosen to focus on this period in the current chapter. A close inspection of Figure 8.1 reveals that for much of the period from 1951 through 1980, the real interest factor is less than one; that is, the real interest rate is negative. In contrast, the expansion of the 1980s and 1990s are accompanied by positive real rates. The real rate appears to have a trend break in 1980. From 1951 through 1980 it uctuates around a constant mean that is a little less then one. In 1983 it reaches a peak of a little over 5% and slowly returns to its pre 1980 trend. This shift in the real rate will be an important part of the picture I will paint and I will return to it in Section 8.4.
It is likely that some, but not all, of the reduction in wage volatility is due to improved methods of data collection as pointed out by Christina Romer 1986.
2
8.3 REAL ECONOMIC HISTORY

1.20 1.15 1.10 1.05 1.00 0.95 0.90 0.85 0.80 1930 1940 1950 1960 1970 1980 1990 Gross Treasury Bill Rate Wage inflation Gross Real Rate in Wage Units Fed Treasury Accord
125
2000
Figure 8.1: Interest Rates and the Accord
8.3
The Economic History of the Post-War Period: Real Variables
This section describes the behavior of real variables in the period from 1951 through 2006 and interprets these movements using the theory developed in Chapter 7. The behavior of gdp and unemployment is depicted in Figure 8.2. Gdp is measured on the left axis in wage units and unemployment is plotted on the right axis and measured as percent of the labor force on an inverted scale. NBER recessions are shaded in grey. The period can broadly be divided into four sub-periods for the analysis of the medium frequency movements that concern me. During the 1950s there is a downward trend in gdp and a slow but steady increase in the unemployment rate. Unemployment begins the decade at 4.5% and ends at over 7%. The 1960s see a reversal of this movement and there are seven years of uninterrupted expansion ending with a recession in 1969. From 1970
126
1.76 1.72 1.68 1.64 1.60 1.56 1.52 1.48

3 4 5 6 7 8 9 10 55 60 65 70 75 80 85 90 95 00 05
Gdp (Wage units)
Unemployment Rate
Figure 8.2: Unemployment and Gdp through 1980 growth slows down and unemployment moves from a low of 5% in the late 1960s to over 9% in 1980. Finally, this medium term trend is reversed in the 1980s and 1990s with a long expansion punctuated by a single brief recession in 1990 and ending with a second recession in 2001. In the Keynesian models developed in this book unemployment is caused by decient aggregate demand which is divided into three components; investment, consumption and government purchases of goods and services. Whereas a fall in investment spending is often blamed by Keynesians for the Great Depression, I will argue in this chapter that consumption and government purchases are more important causes of post-war movements in output and employment.3 Investment is highly correlated with business cycles at typical business cycle frequencies but it does not seem to be correlated with the more important medium term movements in employment that are the focus of this enquiry. Consider Figure 8.3 which plots investment (measured on the right scale) and gdp on the left scale. Both series are in wage units. Casual inspection
The drop in investment spending in the 1930s was a secondary consequence of the drop in stock market wealth that rendered the existing capital stock overvalued.
3
127
of this gure reveals that local peaks and troughs in the series often occur together, but the broad decade long swings in gdp are not associated with similar movements in investment. For example, the long contraction of gdp that begins in 1967 and ends in 1982 is not associated with any similar discernible trend in investment. If the two are series are passed through the Hodrick-Prescott lter with a smoothing parameter of 100 (typical for annual data) the ltered series reveal a much closer correlation of investment and gdp at typical business cycle frequencies. It is these movements that mod1.76 1.72 1.68 1.64 1.60 1.56 1.52 1.48 55 60 65 70 75 80 85 90 95 00 05 .34 .32 .30 .28 .26 .24 .22 .20
Gdp (Wage units) Private Investment (Wage units)
Figure 8.3: Gdp and Investment ern business cycle theorists focus upon and the RBC explanation for business cycles is that they occur as a consequence of technology shocks that cause random uctuations in aggregate productivity. These productivity shocks are autocorrelated and cause consequent movements in investment as businesses expand to take account of future opportunities. In this explanation, investment uctuations and gdp uctuations are both caused by the same driving force: productivity shocks. This explanation may well be correct and there is certainly an element of truth to the idea that some movements in aggregate economic activity are caused by supply side shocks of this nature. Figure ?? illustrates the HP-ltered movements in investment and gdp - these series are measured in
128
.08 .06 .04 .02 .00 -.02 -.04 -.06 -.08 55 60
65
70
75
80
85
90
95
00
05
Gdp in wage units (HP filtered) Investment in wage units (HP filtered)
Figure 8.4: Gdp and Investment (HP Filtered) wage units and a typical movement in investment and gdp at business cycle frequencies is plus or minus 0.3 units. The movements from 1970 through 1980 on Figure ?? are the kind that have in mind. These movements should be compared with the mean value of gdp over the period of 1.6 wage units. In contrast, the medium frequency movement in GDP depicted in 8.3 that begins with the 1967 peak and ends with the 1980 trough is a movement of 2 wage units from 1.72 to 1.52. I am excluding movements of this kind from those associated with supply driven business cycles since although they are associated with a spike in HP ltered investment in 1967, and a trough, in 1981, these extrema are not associated with the medium term trend in gdp. Figure 8.3 reveals that investment was acyclic over the period from 1967 to 1980 and, because a productivity shock should cause gdp and investment to move together, it seems dicult to attribute the medium term downward trend in gdp to a uctuation in productivity. In addition to investment spending, aggregate demand movements may be due to consumption, driven by movements in wealth and income, or movements in government purchases. Consider rst, movements in wealth and consumption. Figure 8.5 illustrates the behavior of two forms of wealth;
129
government debt and the value of the stock market. Government debt is measured on the right axis and the value of a stock market index (the Standard and Poors 500) is on the left axis. The units of the Standard and Poors index (henceforth the S&P) are not comparable with those of government debt since the S&P is an index number divided by the money wage while debt is a dollar value divided by the money wage. The units of debt are pure numbers whereas the units of the S&P series are $1. In the model developed in Chapter 7, steady state consumption is the following function of wealth and income, C = g (R) (pk + B + Z) where Eq (8.2) combines (7.33) and (7.34) and g (R) 1 1 . 1 R (8.3) (8.2)
Income, in the steady state, is determined by the equation Z = H (R) (pk + B) + G, where H (R) (8.4)
g (R) . (8.5) 1 g (R) Recall that Z is gdp, B is government debt, pk is the value of the stock market and R is the gross rate of interest. The following explanation is based on the two steady-state equations, (8.2) and (8.4), and although the relative importance of the dierent determinants of consumption and income will depend on the timing of events and on equilibrium dynamics, the steady state equations will give a rough picture of the plausibility of a demand driven explanation. How well does a theory based on Equations (8.2) and (8.4) explain the facts? Figure 8.6 plots consumption on the right axis and gdp on the left, both measured in wage units. Compare this with Figure 8.5 which plots wealth and income over the same period. From 1951 through 1980 government debt and the stock market move in dierent directions. Over this period consumption trends down in the rst part of the sample and begins a sharp upward trend in 1980. The post 1980s movements in consumption can potentially be explained by the simultaneous increase in stock market wealth and government debt; but what of the period from 1951 to 1980?
130
1.4 1.2 .4 .3 0.6 .2 .1 .0 55 60 65 70 75 80 85 90 95 00 05 0.4 1.0 0.8
S and P 500 (Index weighted by Wage) Government debt (Wage Units)
Figure 8.5: Two Wealth Measures
1.76 1.72 1.68 1.64 1.60 1.56 1.52 1.48 55 60 65 70 75 80 85 90 95 00 05
1.20 1.16 1.12 1.08 1.04 1.00 0.96 0.92
Gdp (Wage units) Private Consumption (Wage units)
Figure 8.6: Gdp and Consumption

1.76 1.72 1.68 1.64 1.60 1.56 1.52 1.48 55 60 65 70 75 80 85 90 95 00 05 .42 .40 .38 .36 .34 .32 .30 .28
131
Gdp (Wage units) Government Purchases (Wage Units)
Figure 8.7: Government purchases and GDP
Figure 8.7 plots gdp and government purchases of goods and services, a major component of gdp. Government purchases are measured on the right axis and gdp on the left, both measured in wage units. Notice from this gure that the medium frequency movements in government purchases from 1951 through 1980 are mirrored by the movements in gdp. The history of this period is one of wartime nance. The Korean war ended in 1953 and was followed by a decline in government purchases and a corresponding drop in gdp. This drop in government purchases was reversed in 1959 when America entered the Vietnam conict and the period from 1959 through 1967 was one of increased government expenditures driven largely by defense. The changes in government purchases over this period are responsible for the decline and subsequent increase in gdp. To summarize; economic activity is determined by aggregate demand which has three main determinants; government purchases, government debt and the value of the stock market. In the period since 1951 the major movements in government purchases have been driven by the requirements of wartime nancing and it is those movements that were mainly responsible
132
for a slowdown in the 1950s and a subsequent expansion in the 1960s. Superimposed on the movements in government purchases there was a decline in government debt from 1951 through 1979 that was then reversed and followed by a period of increasing debt. These changes in debt were associated with a wealth transfer from future to current generations that caused a consumption led boom in economic activity after 1980. A third major component of the movements in demand has been a slow moving but largely unpredictable movement in the value of the stock market caused by changes in market psychology.
8.4
The Economic History of the Post-War Period: Monetary Variables
The explanation I given to this point is missing a critical element - the role of monetary policy. A complete integration of monetary policy with a theory of real behavior based on wealth requires a discussion of how the interest rate on treasury securities is related to the stock market return. It is to this that I now turn.
8.5
The Equity Premium
The top panel of Figure 8.8 plots the ex-post real return on three month treasury bills against the ex post real stock market return on the S&P 500. Both real returns are computed by multiplying the gross nominal return by the ration wt/wt+1 where wt is the money wage series computed in the way explained in Chapter 5. This panel illustrates the equity premium puzzle, the fact that stock market returns are, on average much higher than the return to government securities. Over this period the real return on the S&P averaged 6.8% with a standard deviation of 13%. The average return on t-bills was negative 0.3% with a standard deviation of 2.5%. Most economic models of business cycles ignore this discrepancy and it is typical to calibrate a model to the stock market return. I cannot aord to take that approach in this chapter since the behavior of the t-bill rate is crucial for understanding the evolution of government debt and in the model I will build, government debt is net wealth to the community and is an important component of aggregate demand.
8.5 THE EQUITY PREMIUM
133
1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.7 55 60 65 70 75 80 85 90 95 00 05
Gross Real Rate in % (Wage is numeraire) Real Stock Return in % (Wage is numeaire) 1.4 1.2 1.0 0.8 1.08 1.04 1.00 0.96 0.92 55 60 65 70 75 80 85 90 95 00 05 0.6 0.4
Gross Real Rate in % (Wage is numeraire) Government debt (Wage Units)
Figure 8.8: Debt and the Interest Rate
134
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 55 60
65
70
75
80
85
90
95
00
05
Government debt (Wage Units) Primary budget deficit (Wage units)
Figure 8.9: Debt and the Decit
The lower panel of Figure 9.1 illustrates the behavior of government debt, measured in wage units, against the real t-bill rate from the top panel. Debt is measured on the right scale and the interest rate is on the left. The main point of this graph is to illustrate that the behavior of debt, in wage units, is explained principally by the interest rate. The value of debt fell from a little over 1.2 wage units (about 75% of gdp) to 0.5 wage units in 1980. During this period the real return to t-bills was negative in almost every year and never exceeded 1/3 of 1%. Debt fell because the economy grew faster than the interest rate that was paid by government on its debt. Figure 8.9 provides additional support for this statement. It illustrates the behavior of debt along with the real value of the primary decit, dened as government purchases plus transfers minus all government receipts (including payments into the social security trust fund). This graph shows that debt did not fall because it was paid o; in fact the government borrowed approximately 0.06 wage units more than it received on average over the period from 1951 to 2006.
8.6 THE THEORY SUMMARIZED
135
8.6
The Theory Summarized
What does the theory developed in this book have to say about the events described in Section 8.3? The story I will tell is one of movements in aggregate demand driven by two broad trends. A secular downward movement in government debt began at the end of World War II and continued until 1980. Debt that was accumulated during WWII was paid o in the immediate post-war period as a natural consequence of the growth of the economy and from seigniorage revenues on government debt that private agents are willing to hold for a roughly 7% discount over corporate equity. During the period from 1951 through This reduction in debt came to an abrupt end in 1980 as a direct consequence of the actions of the Fed which temporarily raised the real interest rate and caused the treasury to shift the debt burden to future generations, thereby generating a wealth eect that prompted current generations to increase consumption. expand raised has liquidity value to the nancial system that movements are explained by actions of the Fed. Superimposed on this are unpredictable uctuations in the real value of the stock market and temporary but important movements in government purchases driven by the Korean war (19501953), the Vietnam war (1959-1975) and end of the cold war in 1989. None of these individual features can account for the movements in unemployment, but together they present a fairly complete explanation of business cycle feature of the period. features can account for the Reagan prices that are unpredictable features. The beginning of the period coincides with the accord in 1951 and the end of the Korean war in 1953. Seven years of peace are associated with a drop in government purchases, a fall in aggregate demand and a steady increase in unemployment over the decade which reaches a peak of 7% in 1960. In 1959 America enters the Vietnam war and the period from 1959 through 1968 is associated with a big increase in defense expenditure and a corresponding increase in aggregate demand. Unemployment falls during this period to a low of 4.7% in 1969, the lowest gure since 1957.
Chapter 9 Explaining Stagation

To be completed
The starting point of the model developed in the Chapter is the real model of Chapter 7 that I repeat below, Ct = Ct+1 + (pk,t + Rt1 Bt + Zt + T Rt Tt ) Rt 1 pk,t + Zt Rt1 = , pk,t1 Bt+1 = Bt Rt1 + Gt + T Rt Tt , Zt = Ct + Gt . (9.1)
(9.2) (9.3) (9.4)
In that chapter I used labor as the numeraire and set the money wage in each period to one. Here I assume instead that the money wage may change each period. Since all variables are measured in wage units this assumption leads to the following modication of Eq (9.1), Ct = Ct+1 wt+1 + (pk,t + Rt1 Bt + Zt + T Rt Tt ) (1 + it ) wt
where wt is the money wage and it is the interest rate on treasury bonds that I take to be set by the Fed through open market operations. 137
138
CHAPTER 9 EXPLAINING STAGFLATION
9.1
Adding Money to the Model
Explain money in production function. Explain why log assumption is not quite right. Explain why it doesnt matter much.Concluding Comments
Bibliography
Allais, M. (1947): Economie et intrt. Imprimerie Nationale, Paris. Barro, R. J. (1974): Are Government Bonds Net Wealth?, Journal of Political Economy, 82(6), 10951117. Barro, R. J., and H. Grossman (1971): A General Disequilibrium Model of Income and Employment, American Economic Review, 61, 82 93. Benassy, J. P. (1975): Neo-Keynesian disequilibrium theory in a monetary economy, The Review of Economic Studies, 42, 503523. Blanchard, O. J. (1985): Debts, Decits, and Finite Horizons, Journal of Political Economy, 93(April), 223247. Blinder, A. S., and R. M. Solow (1973): Does Fiscal Policy Matter?, Journal of Public Economics, 2(November), 319337. Clower, R. W. (1965): The Keynesian Counterrevolution: A Theoretical Appraisal, in The Theory of Interest Rates, ed. by F. Hahn, and F. Brechling. McMillan, London. Cochrane, J. H. (1999): A Frictionless View of U.S. Ination, in NBER Macroeconomics Annual 1998, ed. by B. S. Bernanke, and J. J. Rotemberg, pp. 323384. MIT Press, Cambridge, MA. Cole, H. L., and L. E. Ohanian (2004): New Deal Policies and the Persistence of the Great Depression: A Geberal Equilibrium Analysis, Journal of Political Economy, 112(4), 779816. Dreze, J. H. (1975): Existence of an exchange economy with price rigidities, Interntaional Economic Review, 16, 310320. 139
140
BIBLIOGRAPHY
Farmer, R. E. A. (1993): The Macroeconomics of Self-Fullling Prophecies. MIT Press, Cambridge, MA. (1999): The Macroeconomics of Self-Fullling Prophecies. MIT Press, Cambridge, MA, second edn. Friedman, M. (1948): A Monetary and Fiscal Framework for Economic Stability, American Economic Review, 38(June), 245264. (1956): The Quantity Theory of Money A Restatement, in Studies in the Quantity Theory of Money, ed. by M. Friedman. University of Chicago Press, Chicago. (1957): A Theory of the Consumption Function. Princeton University Press. (1968): The Role of Monetary Policy, American Economic Review, 58(March), 117. Friedman, M., and A. J. Schwartz (1963): A Monetary History of the United States, 1867-1960. Princeton University Press, Princeton. Gal, J., M. Gertler, and D. L. Salido (2007): Markups, Gaps and the Welfare Costs of Business Cycle Fluctuations, The Review of Economics and Statistics, 89(1), 4459. Haberler, G. (1937): Prosperity and Depression. George Allen and Unwin Ltd. Hall, R. E. (2005): Employment Fluctuations with Equilibrium Wage Stickiness, American Economic Review, 95(1), 5065. Hetzel, R. L., and R. F. Leach (2001): The Treaury-Fed Accord: A New Narrative Account, Federal Reserve Bank of Richmond Economic Quarterly, 87(1), 3355. Kehoe, T. J., and edward C. Prescott (2007): Great Depressions of the Twentieth Century. Federal Reserve Bank of Minneapolis. Keynes, J. M. (1936): The General Theory of Employment, Interest and Money. MacMillan and Co.
BIBLIOGRAPHY
141
(1937): The General Theory of Employment, Quarterly Journal of Economics, pp. 209223. Kocherlakota, N., and C. Phelan (1999): Explaining the Fiscal Theory of the Price Level, Federal Reserve Bank of Minneapolis Quarterly Review, 23, 1423. Kydland, F., and E. C. Prescott (1982): Time to Build and Aggregate Fluctuations, Econometrica, 50, 13451370. Kydland, F. E., and E. C. Prescott (1996): The Computational Experiment: An Econometric Tool, Journal of Economic Perspectives, 10(Winter), 6985. Lavington, F. (1922): The Trade Cycle. P.S. King and Son. Leeper, E. M. (1991): Equilibria Under Active and Passive Monetary and Fiscal Policies, Journal of Monetary Economics, 27(1), 129147. Leijonhufvud, A. (1966): On Keynesian Economics and the Economics of Keynes. Oxford University Press. Lubik, T. A., and F. Schorfheide (2003): Testing for Indeterminacy: An Application to U.S. Monetary Policy, American Economic Review, forthcoming, Mimeo, Johns Hopkins University. Lucas, Robert E., J., and T. J. Sargent (1981): Rational Expectations and Econometric Practice. University of Minnesota Press, Minneapolis, MN. Lucas, Jr., R. E. (1972): Expectations and the Neutrality of Money, Journal of Economic Theory, 4, 103124. (1987): Models of Business Cycles. Basil Blackwell, Oxford, UK. Lucas, R. E. J. (1967): Optimal Investment Policy and the Flexible Accelerator, International Economic Review, 8, 7885. Lucas, R. E. J., and L. A. Rapping (1969): Real Wages Employment and Ination, Journal of Political Economy, 77, 72154.
142
BIBLIOGRAPHY
Malinvaud, E. (1977): The theory of unemployment reconsidered. Basil Blackwell, Oxford. Patinkin, D. (1989): Money Interest and Prices. The MIT Press, Cambridge, Massachusetts, second abridged edn. Phelps, E. S. (1970): The New Microeconomics in Ination and Employment Theory, in Microeconomic Foundations of Employment and Ination Theory, ed. by E. S. Phelps. Norton, New York. Pigou, A. C. (1929): Industrial Fluctuations. McMillan. Pissarides, C. (2000): Equilibrium Unemployment Theory. MIT Press, Cambridge, 2nd edition. Romer, C. D. (1986): Is the Stabilization of the Post War Economy a Figment of the Data, American Economic Review, 76(46), 341352. Samuelson, P. A. (1958): An Exact Consumption-Loan Model of Interest with or without the Social Contrivance of Money, Journal of Political Economy, 66, 467482. Santayana, G. (1905): The Life of Reason. C Scribners Sons, New York. Shell, K. (1971): Notes on the Economics of Innity, Journal of Political Economy, 79, 10021011. Shimer, R. (2005): The Cyclical Behavior of Equilibrium Unemployment and Vacancies, American Economic Review, 95(1), 2549. Sims, C. A. (1994): A Simple Model for the Determination of the Price Level and the Interaction of Monetary and Fiscal Policy, Economic Theory, 4, 381399. Treadway, A. B. (1971): The Rational Multivariate Flexible Accelerator, Econometrica, 39(5), 845855. Weil, P. (1989): Overlapping Generations of Innitely Lived Agents, Journal of Public Economics, 38, 183198. Woodford, M. (1995): Price-Level Determinacy without Control of a Monetary Aggegrate, Carnegie-Rochester Conference Series on Public Policy, 43, 146.

Farmer Book 2008

Загружено:

Сведения о документе

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

Farmer Book 2008

Загружено:

Авторское право:

Доступные форматы

Expectations, Employment and Prices

Roger E. A. Farmer1 November 2005 This draft: May 16, 2008

CONTENTS Keynes and the Social Planner . . . . . . . . . . . . . . . . . . 42 Concluding Comments . . . . . . . . . . . . . . . . . . . . . . 44 47 47 48 49 49 51 51 52 53 54 55 57 59 59 60 60 61 61 63 65 66 69 70 71 73 76 77 78 80 81 83 84 85

vii 85 87 90 92 94 96 100 102

9 Explaining Stagation 137 9.1 Adding Money to the Model . . . . . . . . . . . . . . . . . . . 138

Preface My preface will appear here.

Chapter 1 What this Book is About

CHAPTER 1 WHAT THIS BOOK IS ABOUT

1.1 THE NATURE OF THE ENQUIRY

The Nature of the Enquiry

CHAPTER 1 WHAT THIS BOOK IS ABOUT

The Theory Summarized

1.3 THE THEORY OF AGGREGATE SUPPLY

The Theory of Aggregate Supply

CHAPTER 1 WHAT THIS BOOK IS ABOUT

1.4 THE THEORY OF AGGREGATE DEMAND

The Theory of Aggregate Demand

CHAPTER 1 WHAT THIS BOOK IS ABOUT

Additional Features of my Interpretation

1.5 ADDITIONAL FEATURES OF MY INTERPRETATION

CHAPTER 1 WHAT THIS BOOK IS ABOUT

Chapter 2 The Labor Market

CHAPTER 2 THE LABOR MARKET

Components of the General Theory

CHAPTER 2 THE LABOR MARKET

such that pC wL (1 ) + T, H 1, L = qH, U = H L. (2.3) (2.4) (2.5) (2.6)

2.3 FIRMS such that X + V = L, L = qV. Y AX,

(2.11) (2.12) (2.13)

CHAPTER 2 THE LABOR MARKET

2.5 THE SOCIAL PLANNER

The Social Planner

max J = j (C) C AX, L = X + V,

(2.20) (2.21) (2.22) (2.23) (2.24) (2.25)

CHAPTER 2 THE LABOR MARKET

2.6 AGGREGATE DEMAND AND SUPPLY

Aggregate Demand and Supply

20 Aggregate Supply Function as,

CHAPTER 2 THE LABOR MARKET

2.6 AGGREGATE DEMAND AND SUPPLY

which, given the normalization w = 1, implies p= 1 . A (1 L) (2.35)

CHAPTER 2 THE LABOR MARKET

Eective Demand and the Multiplier

2.7 EFFECTIVE DEMAND AND THE MULTIPLIER

CHAPTER 2 THE LABOR MARKET

2.8 A DEFINITION OF EQUILIBRIUM

CHAPTER 2 THE LABOR MARKET

Remarks on Ricardian Fiscal Policies

2.10 CONCLUDING COMMENTS

Chapter 3 Aggregate Demand and Supply

CHAPTER 3 AGGREGATE DEMAND AND SUPPLY

(3.1) (3.2) (3.3) (3.4) (3.5)

3.1 HOUSEHOLDS where the utility weights sum to 1,

CHAPTER 3 AGGREGATE DEMAND AND SUPPLY

i,1 i,2 i,m Yi Ai Ki,1 Ki,2 . . . Ki,m Xibi ,

3.3 SEARCH Li = Xi + Vi , Li = qVi .

We may then write the problem in reduced form,

i,1 i,2 i,m Yi Ai Lbi Qbi Ki,1 Ki,2 . . . Ki,m . i

CHAPTER 3 AGGREGATE DEMAND AND SUPPLY

The Social Planner

(3.34) (3.35) (3.36)

b i,1 i,2 i,m Ci Ai Ki,1 Ki,2 . . . Ki,m (Li Vi ) i , i = 1, . . . n, n X i=1

3.4 THE SOCIAL PLANNER

CHAPTER 3 AGGREGATE DEMAND AND SUPPLY

Aggregate Supply and Demand