Вы находитесь на странице: 1из 11

ARTICLE IN PRESS

Journal of Accounting and Economics 44 (2007) 287297


www.elsevier.com/locate/jae

A discussion of corporate disclosure by family rms


Amy P. Hutton
Carroll School of Management, Boston College, USA
Available online 2 February 2007

Abstract

Using a unique empirical setting, family rms in the S&P 500, Ali et al. [Ali, A., Chen, T.-Y.,
Radhakrishnan, S., 2007. Corporate disclosures by family rms. Journal of Accounting and
Economics, doi:10.1016/j.jacceco.2007.01.006] contribute to a growing body of research on the
relation between corporate governance and corporate disclosure quality. Using an indicator variable
for sub-sample membership as an instrument for differing agency costs, the authors interpret their
ndings as consistent with family rms facing lower overall agency costs and providing higher quality
corporate disclosures. However, their empirical ndings are open to alternative interpretations and in
totality present relatively weak, indirect evidence of a relation between corporate governance and the
quality of corporate disclosure.
r 2007 Elsevier B.V. All rights reserved.

JEL classification: D82; G30; M41

Keywords: US family rms; Corporate disclosures; Earnings quality; Corporate governance; Management
forecasts; Analyst forecasts; Bid-ask spread

1. Introduction

Several years before the fall of Enron, regulators and other capital market participants
made known their growing concerns about questionable disclosure practices and the
quality of earnings reported by US Corporates (see Levitt, 1998; Brown, 1999; Parfet,
2000). Since Enron, corporate America has come under even greater scrutiny and increased
regulation. E.g., Sarbanes Oxley is, at least in part, intended to enhance the role of
corporate governance in safe guarding the quality of reported earnings and overall
corporate disclosure.

Tel.: +1 617 552 1951.


E-mail address: amy.hutton@bc.edu.

0165-4101/$ - see front matter r 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jacceco.2007.01.004
ARTICLE IN PRESS
288 A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297

With growing interest in the topic, recently researchers have attempted to examine
directly whether and how corporate governance affects earnings quality and overall
corporate disclosure quality. Specically, several research teams use the international
setting to compare internationally the quality of corporate governance and disclosure (see
e.g. Ball et al., 2003; Bushman et al., 2004; Khanna et al., 2004). While demonstrating that
higher quality governance is associated with higher quality disclosure and greater
transparency, the international research does not provide direct evidence on how specic
differences in corporate governance practices across US rms affects the quality of their
reported earnings and disclosure practices. Ali, Chen and Radhakrishnan (2007, hereafter
ACR) attempt to address this very interesting question: Does cross-sectional variation in
US corporate governance practices help to explain cross-sectional variation in managerial
choices of corporate disclosure policies?
Using Business Weeks split of the S&P 500 into family and non-family rms, the
ACR paper examines whether family rms corporate disclosure policies differ from
those of non-family rms. They argue that family rms face less severe agency problems
arising from the separation of ownership and control, but more severe agency
problems arising between controlling and non-controlling shareholders. The authors
use an indicator variable for sub-sample membership as an instrument for differing
agency costs and examine the relation between agency costs and corporate
disclosure choice. The authors interpret the evidence presented as largely consistent with
family rms facing lower overall agency costs and providing higher quality corporate
disclosures.
The paper is a reasonable rst attempt at demonstrating an association between agency
costs and corporate disclosure within the US setting, in as much as it uses an interesting
empirical settingfamily rms in the S&P 500. However, the decision to examine family
rms also raises concerns that key results are spurious. As I discuss below, if family rms
only retain ownership and control of successful rms and if, as documented by Miller
(2002), disclosure quality is positively related to rm performance, then the relation
between family rms and disclosure choices may be spurious. Other empirical ndings
presented in the paper are also open to alternative interpretations. Thus, in totality the
paper presents relatively weak and indirect evidence of a relation between corporate
governance and the quality of corporate disclosure.
Section 2 outlines the research questions examined by ACR. Section 3 discusses Business
Weeks denition of family rms, sample selection bias, and descriptive information.
Section 4 comments on the tests reported in the paper and the interpretation of results.
Section 5 summaries the contribution of the paper and provides suggestions for future
research.

2. Research questions

The family rm versus the non-family rm distinction is employed to identify rms


facing differing unresolved agency problems. Then, tests are developed to assess whether
differences in agency problems are associated with differing corporate disclosure practices.
The authors examine several aspects of corporate disclosure: Quality of earnings,
disclosure of bad news through management earnings forecasts, and voluntary disclosure
of corporate governance practices in regulatory lings.
ARTICLE IN PRESS
A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297 289

The authors focus on regressions aimed at measuring the relative level of the variables
of interest across family and non-family rms. The research questions addressed in this
paper are:

1. Are discretionary accruals smaller in absolute magnitude for family versus non-family
rms?
2. Are signed discretionary accruals smaller for family versus non-family rms?
3. Do the components of earnings do a better job predicting cash ows for family versus
non-family rms?
4. Do family rms have more persistent earnings than non-family rms?
5. Do family rms have larger earnings response coefcients than non-family rms?
6. Are managers of family rms more likely to provide an earnings warning for the same
level of negative earnings news?
7. Do family rms provide less voluntary disclosure about their corporate governance
than non-family rms?

Once the authors present results interpreted as indicating that family rms have better
disclosure practices than non-family rms, they go on to assess whether these rms
experience the usual benets of better quality disclosure (Botosan, 1997; Healy et al., 1999;
Welker, 1995). Once again the authors employ regression analyses to assess whether the
level of the variables of interest differ across family and non-family rms. The specic
research questions addressed to assess whether family rms enjoy the benets of better
quality disclosure are: Compared to non-family rms,

8. Do family rms enjoy greater analyst following?


9. Is the dispersion in analysts forecasts lower for family rms?
10. Are analysts forecast errors smaller for family rms?
11. Are analysts forecast revisions less volatile for family rms?
12. Do family rms face lower bid-ask spreads?

While the research questions addressed in the paper are of interest, there are several
empirical opportunities that are overlooked by the authors: (1) Since earnings management
via manipulation of real activities is often value destroying and since founding families
often hold large ownership stakes in their rms, one would expect family rms not to
change the real business activities of the rm to meet an earning benchmark.1 Neither
ACR nor Wang (2006), a contemporaneous paper that also documents earnings quality for
family rms in the S&P 500, examine whether family rms engage in less manipulation of
real activities to manage earnings than non-family rms. (2) Since founding families are
long-term investors with potentially under-diversied portfolios, one would expect family
rms voluntary disclosures in the quarterly earnings process to differ from non-family
rms. (3) Given differences in corporate governance and disclosure practices of family
versus non-family rms documented in both ACR and Wang (2006), it would be
interesting to know whether institutions invest more or less heavily in family rms.
Univariate analyses presented in both papers suggest that institutions invest less heavily in
family rms. Multivariate analysis is needed to draw useful inferences.
1
Roychowdhury (2006) provides a methodology for measuring manipulation of real business activities.
ARTICLE IN PRESS
290 A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297

3. Sample development, selection bias, and descriptive statistics

3.1. Sample development and selection bias

Inferences drawn from the ndings documented in this paper depend critically on the
classication scheme for family and non-family rms. ACR splits the S&P 500 into family
and non-family rms based on Business Weeks classication provided in a special report
on family rms in its November 10, 2003 issue.2 The purpose of the Business Week special
report is to compare the performance of family and non-family rms over the time period
19922002. Business Weeks denition of a family rm, any company where founders or
descendants continue to hold positions in top management, on the board, or among the
companys largest stockholders, results in 177 of the rms in the July 2003 S&P 500 stock
index being classied as family rms.
Based on Business Weeks denition, the sub-sample of family rms in the S&P 500
includes a wide range of rms from Microsoft, known for its conservative accounting
choices, to Campbell Soup Co., Cendant, and Computer Associates, all known for their
accounting scandals. A supplement to the Business Week special report, Dening
Family, indicates that where founders owned stock but were not involved in day-to-day
operations, Business Week used its judgment, taking into account the size of the founders
ownership stake. The supplement also indicates that Business Week included rms without
founders or descendents present, if either had been active for the bulk of its sample period.
The decision to rely on Business Week for the classication of rms leaves the ACR
paper open to several criticisms. First, it is unclear that the family rm distinction dened
by Business Week is the most powerful classication for addressing ACRs research
questions.3 Second, even if one agrees with the criteria Business Week used to dene a
family rm, a quick review of the Business Week list of family rms and the accompanying
details indicating why each rm is classied as a family rm, raises concerns about Business
Week as an objective, scientic research entity. For instance, Business Week classied
Chubb Corporation (ticker symbol CB) as a family rm because the founders heir, Percy
Chubb III, is a non-voting director emeritus on the insurance companys board of
directors. The Chubb family retains no signicant stock ownership in the company. As
another example, Business Week classied Cummins Inc. (ticker symbol CMI) as a family
rm because director William Miller is a distant relative of W.G. Irwin, the banker who
supplied the start-up capital in 1919. The Irwin family retains no signicant stock
ownership in the company. Neither of these rms appear to have founding families
retaining signicant control over the corporation either from a managerial decision
making or from a stock ownership perspective. There are other questionable classications
readers may review on the Business Week website noted in footnote two of this discussion.
A third criticism arises from the decision to focus on the family rm distinction per se as
a way to identify differing agency costs. If families tend to exit or lose control of poor
performing businesses, then a sample selection bias existsfamily rms will systematically

2
See the following website for details on each of the 177 rms from the July 2003 S&P 500 stock index that are
dened by Business Week (2003) as a family rm: http://bwnt.businessweek.com/family_companies/2003/
index.asp?sortCol=rank&sortOrder=ASC&pageNum=4&resultNum=50&sector=
3
See Wang (2006) for an alternative classication of the S&P 500 into founding family rms and non-founding
family rms. Wang also employs a continuous variable, founding family ownership.
ARTICLE IN PRESS
A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297 291

outperform non-family rms. The authors decision to focus on the S&P 500 has the
potential to mitigate this performance bias, since all rms in the S&P 500 perform well
historically or they would not be included in the S&P 500. However, prior research
demonstrates that within the S&P 500 family rms are more protable than non-family
rms (Anderson and Reeb, 2003). Thus, the performance bias prevails in the sample of
S&P 500 rms and it is important for interpreting ACRs results, since Miller (2002)
documents that rm performance is positively associated with voluntary corporate
disclosure. As the authors note (on page 18 of the text), the key ndings in the paper may
reect spurious correlation. Specically, better rm performance is associated with the
family rm distinction; better rm performance is also associated with higher quality
disclosure; thus, while the family rm indicator variable is positively associated with higher
quality disclosure, it is potentially spurious correlation.
Finally, the focus on the S&P 500 limits generalization of the ndings. In particular,
ndings presented in this paper do not rule out managers of family rms being more likely
to resort to manipulating reported results when faced with several years of poor
performance. In fact, when faced with such extreme circumstances, CEOs that are
founders are documented to be more likely to manipulate reported results than non-
founder CEOs (see Dechow et al., 1996). Anecdotal evidence also suggests that family
rms resort to earnings management when facing downturns: Adelphia, Campbell Soup,
Cendent, Computer Associates, and Rite Aid Corp., just to name a few famous cases.

3.2. Descriptive statistics

The authors attempt to provide external validation of the Business Week classication
by providing descriptive statistics on salient rm characteristics in Table 2. Panel B is the
most revealing with its comparison of the family and non-family rms corporate
governance variables. Family rms have much larger share ownership by ofcers and
directors, consistent with the authors contention that these rms face less severe agency
problems arising from the separation of ownership and control. Consistent with family
rms facing higher agency costs arising between controlling and non-controlling
stockholders, the percentage of independent directors is lower for family rms and the
percentage of rms with an outside blockholder is also lower for family rms.
Finally, Table 2 Panel B indicates that the frequency with which the CEO serves as the
Chairman of the Boards is lower for family rms. This statistic is difcult to interpret,
however. One interpretation is that family rms have better corporate governance with
reduced opportunities for managerial entrenchment as power is shared between two
individuals who provide a check and balance on one another. However, in this current
context, what really matters is the identities of these two individuals. If both are members
of the founding family, then perhaps this statistic suggests greater opportunity for
entrenchment of the founding family and higher agency costs arising between controlling
and non-controlling shareholders. The authors do not examine the identities of the CEO
and Chairman as a way to identify entrenchment of the founding family.

4. Empirical tests

The authors compare several aspects of corporate disclosure across family and non-
family rms: Measures of earnings quality, voluntary disclosure of bad news through
ARTICLE IN PRESS
292 A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297

management earnings forecasts, and voluntary disclosure of corporate governance


practices in regulatory lings. Once results are interpreted by the authors as indicating
that family rms have better disclosure practices, the authors go on to examine whether
family rms enjoy the usual benets of better quality disclosure. Finally, to gain added
condence that the differences in agency problems across family and non-family rms
drive their main results, the authors present analyses of sub-samples of family rms. In this
section of the discussion, I comment on several of the empirical tests pointing out
important caveats to the authors interpretations.
As a general comment, it is worth noting that for all of their analyses, the authors focus
on pooled time-series, cross-sectional regressions aimed at measuring the relative level of
the variables of interest across family and non-family rms. The authors pool observation
across 5 years, 19982002, to increase the power of their tests. However, pooling
observation across time raises concerns about the independence of the observations, since
many of the variables of interest do not change much from year to year for the individual
rms. While these concerns are addressed in footnote 14 of the text, the statistical
signicance of the papers ndings is potentially overstated in the presented tables.

4.1. Earnings quality

The authors examination of earnings quality is vefold. They examine signed and
unsigned discretionary accruals, predictability of cash ows, earnings persistence, and
earnings response coefcients. The tests are generally well done, often following closely the
methods developed in prior research. Evidence is presented that is consistent with family
rms having more negative discretionary accruals, earnings components that are better
able to predict cash ows, and higher earnings response coefcients. My discussion focuses
on the inferences that can be drawn from tests with signicant ndings.

4.1.1. Signed discretionary accruals


The authors rely on one model of accruals, the performance-adjusted modied Jones
model, when analyzing discretionary accruals (Dechow et al., 1995; Kothari et al., 2005).
Although this is a state-of-the-art model, providing alternative measures of discretionary
accruals is useful, since inferences drawn from earnings management tests depend critically
on proper specication of the accruals model (see Ball and Shivakumar, 2005a; Cohen
et al., 2006). One alternative accruals model to consider is the non-linear accrual model
presented in Ball and Shivakumar (2005a). Ball and Shivakumar (2005b) demonstrate that
this non-linear model via its incorporation of the role of accruals in timely gain and loss
recognition is a substantial specication improvement, explaining up to three times the
amount of variation in accruals as equivalent linear specications.4
Researchers examining signed discretionary accruals at the rm-level (rather than the
rm-year level) either explicitly or implicitly assume that without specic incentives to
decrease reported earnings, all managers face incentives to accelerate income recognition,
and interpret more positive discretionary accruals as indicative of more earnings
management. Yet, accrual manipulation for the purpose of manipulating reported
earnings is all about the timing of income recognition, e.g., either increasing net assets to
4
Wang (2006) demonstrates that family rms in the S&P 500 have smaller absolute value of abnormal accruals
using the Ball and Shivakumar model.
ARTICLE IN PRESS
A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297 293

accelerate the recognition of income or decreasing net assets to delay the recognition of
income. Research designs that fail to specify income-increasing or income-decreasing
incentives in each reporting period often lead to invalid inferences. For instance, in the
ACR paper, the more negative discretionary accruals for the family rms is just as likely to
be indicative of more rather than less earnings management, when one considers family
rms incentives to report lower earnings, such as their incentives to minimize taxes or
reduce political costs (e.g., Microsoft).

4.1.2. Predictability of cash flows


Next, the authors analyze the ability of earnings components to predict cash ows
(Dechow et al., 1998; Barth et al., 2001). They classify earnings as high quality if the
absolute value of the residuals from the cash-ow prediction model falls in the lower half
of the sample (Cohen, 2006). Univariate analysis indicates that this discrete measure of
earnings quality does not vary systematically across family and non-family rms.
However, logistic regression analyses indicate that, after including many control variables,
the absolute value of the residuals from the cash-ow prediction model fall below the
median more frequently for the sub-sample of family rms. The authors interpret this as
evidence that family rms earnings components predict cash ows better than the earnings
components of non-family rms (Table 4).
It is not clear to me that the residuals from the industry-specic cash-ow prediction
model are an indicator of earnings quality. The absolute value of the residual for a
specic rm could be large simply because the rm operates differently from other rms in
its industry (2-digit SIC grouping). For instance, if a rm is signicantly better at
managing its working capital accounts than other rms in its industry (e.g., Wal-Mart
Stores), then its residual from the industry-specic cash-ow prediction model will be
large, even if in an absolute sense its earnings components are good at predicting its cash
ows.5 One alternative approach would be to estimate a rm-specic time-series model of
the relation between earnings components and future cash ows and then use the standard
deviation of the residuals from this model as a measure of nancial reporting quality
(Cohen, 2006).
Additionally, as in the discretionary accruals tests discussed above, inferences drawn
from these tests depend critically on proper specication of the cash-ow prediction model.
If the model omits variables correlated with the partitioning variable, Type I errors
are likely.6 This is a concern for the ACR paper because, for instance, I would expect
both the unsigned residuals from the cash-ow prediction model and the family rm
indicator variable to be correlated with the volatility of earnings and/or the volatility of
sales, neither of which are control variables in ACRs analyses of the predictability of cash
ows.

5
To see this, note the discussion in Barth et al. (2001) on page 34 that e.g., the coefcient on ,INV reects the
payment deferred to next period of the current period change in inventory. If a particular rm is able to defer
more of its payments to future periods for the same level of increases in inventory (e.g., Wal-Mart with its power
over its suppliers), then the estimated industry-specic coefcient will be biased when applied to the particular
rm and the estimated residual will be large for this rm, regardless of how well this rms earnings components
predict its future cash ows.
6
See Hribar and Nichols (2006) for a useful discussion of this issue in the context of earnings management tests
that use unsigned discretionary accruals.
ARTICLE IN PRESS
294 A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297

4.1.3. Earnings response coefficients


These tests are fairly well done, as extensive control variables are included to address the
signicant differences in rm characteristics across the family and non-family rms.
However, at least one variable is missing from the list of control variables, earnings
persistence, which is known to be associated with earnings response coefcients (see Collins
and Kothari, 1989) and demonstrated to be related to the family rm indicator variable in
Table 5 Panel A. Thus, the coefcients reported in Table 6 Panel B are potentially biased.

4.2. Earnings warnings

Next, the authors examine whether managers of family rms, in comparison to managers
of non-family rms, are more likely to issue earnings forecasts for a given magnitude of bad
news (Table 7). At rst glance, the tests appear to follow closely Kasznik and Lev (1995).
However, as I discuss below, a close look reveals that the tests differ in important ways.
First, in contrast to prior work, the development of ACRs hypotheses regarding the
association between the likelihood of management earnings forecasts and the magnitude of
the negative earnings news (Hypotheses 2a and 2b) implicitly assumes that management
can withhold bad news indenitely. Otherwise, controlling shareholders would merely
delay rather than escape scrutiny of their private rent seeking and entrenchment activities.
Prior empirical work assumes that managers who do not warn are able to withhold bad
news only temporarily, that is until the annual earnings announcement (e.g., Skinner, 1994;
Kasznik and Lev, 1995). Under this assumption, the primary motivation for managers to
provide earnings warnings is to lower expected litigation costs (Skinner, 1994). Interpreting
ACRs ndings when it is assumed that bad news is revealed at the annual earnings
announcement, family rms are more likely to warn for a given magnitude of bad news
simply because the families themselves bear more of the litigation costs of not warning
given their signicant ownership stakes in the rms. In this light, family rms decisions to
warn are not necessarily indicative of less opportunistic behavior or lower agency costs.
Also in contrast to Kasznik and Lev (1995), ACR does not constrain the horizon of the
management forecasts.7 The reason this constraint is important, examining forecasts made
well in advance of the earnings announcement increases the chances that the forecast was
not intended to warn investors about the upcoming earnings surprise. In fact, if through its
forecast management moved investors expectations away from actual earnings, then the
forecast assists in creating the earnings surprise. This is a concern for the ACR paper, since
the authors did not document nor control for the sign of the news in the management
earnings forecast. Thus, it is unclear whether the management forecasts included in their
analyses were actually intended to warn investors of the upcoming earnings surprise.

4.3. Benefits of better financial disclosure by family firms

To examine whether family rms benet from the higher quality earnings and greater
likelihood of an earnings warnings, next the authors examine whether family rms have
7
Kasznik and Lev (1995) examine only management forecasts made within 60 days of the annual earnings
announcement. Since Kasznik and Lev hand collected their data, I presume that they checked to ensure that the
sign of the news in the management earnings forecast matched the sign of the news at the earnings announcement
date, even though I could not nd a mention of this constraint in the text of the paper.
ARTICLE IN PRESS
A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297 295

greater analyst coverage, lower dispersion in analysts forecasts, greater forecast


accuracy, smaller volatility in forecast revisions as well as lower bid-ask spreads
(Table 9). The tests follow closely those presented in Lang and Lundholm (1996) and
Healy et al. (1999).
A confounding factor in these tests is the fact that family rms provide less transparent
disclosures of their corporate governance practices. In footnote eight the authors note this
to be a potential problem for tests examining differences in bid-ask spreads, but argue that
it is not likely to be an issue for tests examining analyst following and the properties of
analyst forecasts. However, if institutions shy away from investing in family rms because
of the lack of transparency in their corporate governance practices (as suggested by
univariate tests presented in Panel A of Table 3) and if analyst coverage (number and
quality of analyst covering the rm) is affected by the level of institutional investment, then
analyst following and the properties of analyst forecasts may well be affected by the less
transparent disclosure of their corporate governance practices. To understand the relations
more fully, a detailed analysis of institutional investment is needed.
On a related point, the documented differences in the characteristics of analysts outputs
are consistent with family and non-family rms having different earnings guidance
policies. Specically, the documented differences in analyst following and analysts outputs
suggest that family rms provided more effective earnings guidance to analysts prior to
Regulation Fair Disclosure (Reg FD). To investigate this alternative interpretation one
could examine whether analysts quarterly forecasts are systematically pessimistic in the
pre-Reg FD time period for family rms, and not so for non-family rms (see Matsumoto,
2002; Hutton, 2005). Additionally, greater analyst following for family rms could results
from their superior performance, as analysts have been documented to prefer following
strong performers (McNichols and OBrien, 1997).

4.4. Sub-sample tests

The authors conduct tests on sub-samples of the family rms to gain additional
condence that differences in the severity of agency problems across family and non-family
rms drives [their] results. Specically, they classify family rms along two lines: Whether
the rm has dual class shares and whether the CEO is the founder, a descendent of the
founder, or hired. Firms with founder CEOs and no dual class shares are expected to have
less severe agency problems and thus are more likely to exhibit better disclosure practices.
The authors interpret their ndings (in Table 10) as supportive.
Confounding these tests is the fact that the sub-samples of family rms are very unequal
in size. Specically there are only 30 rms with descendent CEOs and only 19 rms with
dual class shares. Thus, the lack of signicant coefcients for these dummy variables may
well reect a lack of power, leaving the added condence in the papers primary ndings
gained from these tests marginal at best.
On the other hand, it would have been interesting to know whether family rms with
family members in executive positions have lower quality disclosures. Sixty-three percent
of the family rms have family members in executive positions, indicating that a split of the
sample on this variable of interest would have provided more even sub-samples and
perhaps more powerful tests. Also, it would have been interesting to know whether family
rms disclosing information about signicant related party transactions (27% of the family
rm sample, reported in Table 8 Panel A) have lower quality disclosures.
ARTICLE IN PRESS
296 A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297

5. Conclusions and future research

The two agency issues examined in concert in this paper have been examined
individually in prior research. Wareld et al. (1995) examine directly the association
between managerial ownership and earnings informativeness and earnings management.
They nd that managerial ownership is positively associated with earnings informativeness
and negatively associated with the magnitude of discretionary accruals. Francis et al.
(2005) examine directly the informativeness of earnings when cash-ow rights are
separated from voting rights via dual class shares. They nd that earnings are generally less
informative for dual class rms.
This raises the question: How does an examination of family rms contribute to the
existing literature? By the authors own admission, family rms face less severe agency
problems arising from the separation of ownership and control and greater agency
problems between controlling and non-controlling shareholders. Thus, their sample is a
mixed bag of unresolved agency problems, making it more difcult to draw useful
inferences from their ndings.
Nevertheless, family rms offer an opportunity to examine a unique class of
shareholders that hold poorly diversied portfolios, are long-term investors (multiple
generations) and control senior management positions (Anderson and Reeb, 2003,
p. 1304). Beyond existing work on founding-family rms, which examines difference in
performance, capital structure, real investment decisions (rm diversication and capital
spending) and corporate disclosure, additional attributes worthy of examination in the
context of founding-family rms include differences in investor base (e.g., level of
institutional investment) and earnings management through real activities manipulation
(Roychowdhury, 2006).
Additionally, understanding the role of corporate governance mechanisms in limiting
the expropriation of rm wealth by founding families deserves further research. For
example, Anderson and Reeb (2004) demonstrate that founding-family rm performance
depends on board composition. It is likely that risk-aversion behavior in family rms as
well as the level of institutional investment and earnings management via real activities
manipulation also vary with the existence of corporate governance mechanisms intended
to limit the expropriation of rm wealth by founding families.

References

Ali, A., Chen, T.-Y., Radhakrishnan, S., 2007. Corporate disclosures by family rms. Journal of Accounting and
Economics, doi:10.1016/j.jacceco.2007.01.006.
Anderson, R.C., Reeb, D.M., 2003. Founding-family ownership and rm performance. Journal of Finance 58,
13011328.
Anderson, R.C., Reeb, D.M., 2004. Board composition: balancing family inuence in S&P 500 rms.
Administrative Sciences Quarterly 49, 209237.
Ball, R., Shivakumar, L., 2005a. Earnings quality in UK private rms. Journal of Accounting and Economics 39,
83128.
Ball, R., Shivakumar, L., 2005b. The role of accruals in asymmetrically timely gain and loss recognition. Journal
of Accounting Research 44, 207242.
Ball, R., Robin, A., Wu, J.S., 2003. Incentives versus standards: Properties of accounting income in four East Asia
countries. Journal of Accounting and Economics 36, 235270.
Barth, M., Cram, D., Nelson, K., 2001. Accruals and the prediction of future cash ows. Accounting Review 76,
2758.
ARTICLE IN PRESS
A.P. Hutton / Journal of Accounting and Economics 44 (2007) 287297 297

Botosan, C., 1997. The impact of annual report disclosure level on investor base and the cost of capital.
Accounting Review 72, 323350.
Brown, P.R., 1999. Earnings management: a subtle (and troublesome) twist to earnings quality. The Journal of
Financial Statement Analysis Winter.
Bushman, R.M., Piotroski, J.D., Smith, A.J., 2004. What determines corporate transparency. Journal of
Accounting Research 42, 207252.
Business Week, 2003. Family Inc. November 10: http://www.businessweek.com/magazine/content/03_45/
b3857002.htm.
Cohen, D.A., 2006. Does information risk really matter? An analysis of the determinants and economic
consequences of nancial reporting quality. Working paper, Stern School of Business, NYU.
Cohen, D.A., Dey, A., Lys, T.Z., 2006. Trends in earnings management in the pre- and post-Sarbanes Oxley
periods. Working paper, Northwestern University.
Collins, D., Kothari, S.P., 1989. An analysis of inter-temporal and cross-sectional determinants of earnings
response coefcients. Journal of Accounting and Economics 11, 143181.
Dechow, P.M., Sloan, R.G., Sweeney, A.P., 1995. Detecting earnings management. Accounting Review 70,
193225.
Dechow, P.M., Sloan, R.G., Sweeney, A.P., 1996. Causes and consequences of earnings manipulation: an analysis
of rms subject to enforcement actions by the SEC. Contemporary Accounting Research 13, 136.
Dechow, P.M., Kothair, S.P., Watts, R.L., 1998. The relation between earnings and cash ows. Journal of
Accounting and Economics 25, 133168.
Francis, J.K., Schipper, Vincent, L., 2005. Earnings and dividend informativeness when cash ow rights are
separated from voting rights. Journal of Accounting and Economic 39, 329360.
Healy, P., Hutton, A., Palepu, K., 1999. Stock performance and intermediation changes surrounding sustained
increases in disclosure. Contemporary Accounting Research 16, 485520.
Hribar, P., Nichols, D.C., 2006. The use of unsigned earnings quality measures in tests of earnings management.
Working paper, Cornell University.
Hutton, A.P., 2005. Determinants of managerial earnings guidance prior to regulation fair disclosure and bias in
analysts earnings forecasts. Contemporary Accounting Research 22, 867914.
Kasznik, R., Lev, B., 1995. To warn or not to warn: management disclosure in the face of an earnings surprise.
Accounting Review 70, 113134.
Khanna, T., Palepu, K.P., Srinivasan, S., 2004. Disclosure practices of foreign companies interacting with US
markets. Journal of Accounting Research 42, 475508.
Kothari, S.P., Leone, A., Wasley, C.E., 2005. Performance matched discretionary accruals measures. Journal of
Accounting and Economics 39, 163197.
Lang, M., Lundholm, R., 1996. Corporate disclosure policy and analysts. Accounting Review 71, 467492.
Levitt, A., 1998. The Numbers Game. Speech by Chairman of the US Security and Exchange Commission.
Available on: http://www.sec.gov/news/speech/speecharchive/1998/spch220.txt.
Matsumoto, D.A., 2002. Managements incentives to avoid negative earnings surprises. Accounting Review 77,
483514.
McNichols, M., OBrien, P., 1997. Self-selection and analyst coverage. Journal of Accounting Research 35,
167199.
Miller, G.S., 2002. Earnings performance and discretionary disclosure. Journal of Accounting Research 40,
173204.
Parfet, W.U., 2000. Accounting subjectivity and earnings management: A preparer perspective. Accounting
Horizon 14, 481488.
Roychowdhury, S., 2006. Earnings management through real activities manipulation. Journal of Accounting and
Economics 42, 335370.
Skinner, D., 1994. Why rms voluntarily disclose bad news. Journal of Accounting Research 32, 3860.
Wang, D., 2006. Founding family ownership and earnings quality. Journal of Accounting Research 44, 619656.
Wareld, T.D., Wild, J.J., Wild, K.L., 1995. Managerial ownership, accounting choices, and informativeness of
earnings. Journal of Accounting and Economics 20, 6191.
Welker, M., 1995. Disclosure policy, information asymmetry and liquidity in equity markets. Contemporary
Accounting Research 11, 801828.

Вам также может понравиться