J of Accounting Research - 2006 - WANG - Founding Family Ownership and Earnings Quality

DOI: 10.1111/j.1475-679X.2006.00213.
x
Journal of Accounting Research
Vol. 44 No. 3 June 2006
Printed in U.S.A.
Founding Family Ownership

and Earnings Quality
D E C H U N WA N G ∗
Received 1 January 2005; accepted 18 November 2005
ABSTRACT
This study investigates the relation between founding family ownership and
earnings quality using data from the Standard & Poor’s 500 companies. Ex-
isting literature has documented that financial reporting is of higher quality
when firms have stronger corporate governance mechanisms and when there
is greater demand for quality financial reporting. I provide two competing
theories of the effect of founding family ownership on the demand and sup-
ply of earnings quality: the entrenchment effect and the alignment effect. The
empirical results show that, on average, founding family ownership is associ-
ated with higher earnings quality. In particular, I find consistent evidence that
founding family ownership is associated with lower abnormal accruals, greater
earnings informativeness, and less persistence of transitory loss components in
earnings. In addition, the results suggest a nonlinear relation between family
ownership and earnings quality.
1. Introduction
This study investigates whether founding family ownership affects the
quality of financial reporting. Founding family ownership is an important
∗ University of Nebraska–Lincoln. This paper is partly based on my dissertation completed at

the University of Missouri–Columbia. I thank my dissertation committee co-chairs, Jere Francis
and Inder Khurana, for their generous and consistent guidance and encouragement in com-
pleting this paper. I also thank other committee members—Raynolde Pereira, Kenneth Shaw,
and John Howe—for their helpful comments. This paper has benefited from the comments
of workshop participants at the University of Missouri–Columbia, the University of Nebraska–
Lincoln, Georgia State University, and the University of Florida. Special thanks to Arthur Allen,
Ken Reichelt, Sugata Roychowdhury (the discussant at the 2004 AAA annual meeting), and
Ray Ball (the editor) for their valuable suggestions. Finally, I greatly appreciate the comments
and suggestions from an anonymous reviewer. All errors are my own.
619
Copyright
C , University of Chicago on behalf of the Institute of Professional Accounting, 2006
1475679x, 2006, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1475-679X.2006.00213.x by <Shibboleth>-member@city.ac.uk, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
620 D. WANG
ownership structure. 1 Most firms around the world are family-owned busi-
nesses (Burkart, Panunzi, and Shleifer [2003]). Even among the Standard
and Poor (S&P) 500 and Fortune 500 companies, which are the least likely
to be family owned, one third have founding family members actively in-
volved in the businesses (Anderson and Reeb [2003a], Shleifer and Vishny
[1986], Weber et al. [2003]). It is also a unique ownership structure, in that
“founding-families represent a unique class of shareholders that hold poorly
diversified portfolios, are long-term investors (multiple generations), and
often control senior management positions” (Anderson and Reeb [2003a,
p.1304]).
One of the typical characteristics of publicly traded companies in the
United States is the separation of ownership from control, which gives rise to
agency conflicts between managers and outside shareholders (Demsetz and
Lehn [1985], Jensen and Meckling [1976], Shleifer and Vishny [1997]). On
one hand, professional managers of diffusely owned firms have incentives
to report financial accounting information that deviates from the substance
of underlying economic transactions to maximize private benefits at the
cost of shareholders or creditors (Christie and Zimmerman [1994], Healy
and Kaplan [1985], Leuz, Nanda, and Wysocki [2003], Warfield, Wild, and
Wild [1995]). Therefore, ownership structures affect the supply of quality
financial reporting (Fan and Wong [2002], Francis, Schipper, and Vincent
[2005], Warfield, Wild, and Wild [1995]). On the other hand, accounting
earnings are used to mitigate agency conflicts by aligning the interests of
managers with those of outside shareholders or creditors (Bushman and
Smith [2001], Christie and Zimmerman [1994], Healy and Kaplan [1985],
Watts and Zimmerman [1986]). Hence, there is demand for quality financial
reporting by shareholders, creditors, and other users of financial statements
for the purpose of efficient contracting and monitoring. In turn, the demand
for quality financial reporting creates incentives for firms to provide high-
quality financial statements in order to obtain better contracting terms (Ball,
Kothari, and Robin [2000], Ball, Robin, and Wu [2000, 2003], Ball and
Shivakumar [2005a]).
Founding family ownership could affect the demand and supply of quality
financial reporting in one of two competing ways: the entrenchment effect
and the alignment effect. The entrenchment effect motivates financial state-
ment suppliers (firms) to opportunistically manage earnings. It is consistent
1 Founding family ownership refers to firms with substantial common stock held by family
members or with founding family members actively involved in the management or the board
of directors. Over 80% of companies in North America and 35% of the Fortune 500 compa-
nies are businesses with founding family involvement or ownership (Anonymous [1998]). In
addition, “(f)amily businesses account for 78 percent of all new job creation, 60 percent of the
nation’s employment and 50 percent of the Gross National Product” (National Underwriter
[1998]). More than 40% of large firms in Western Europe are controlled by families (Faccio and
Lang [2002]) and almost all the biggest firms in Eastern Asia are managed by entrepreneurial
founding families (Anonymous [1996]).
FOUNDING FAMILY OWNERSHIP 621
with the traditional view that family firms are less efficient because concen-
trated ownership creates incentives for controlling shareholders to expro-
priate wealth from other shareholders (Fama and Jensen [1983], Morck,
Shleifer, and Vishny [1988], Shleifer and Vishny [1997]). Family members
usually hold important positions on both the management team and the
board of directors. Thus, these firms may have inferior corporate gover-
nance because of ineffective monitoring by the board. Another source of
entrenchment is potentially greater information asymmetry between found-
ing families and other shareholders. Fan and Wong [2002, p. 403] argue
that concentrated ownership limits accounting information flows to outside
investors, while Francis, Schipper, and Vincent [2005] suggest that infor-
mation asymmetry lowers the transparency of accounting disclosures. 2 As
a result, family members have both the incentive and the opportunity to
manipulate accounting earnings for private rents. Therefore, the entrench-
ment effect predicts that founding family ownership is associated with the
supply of lower earnings quality.
The supply of lower earnings quality resulting from the entrenchment ef-
fect of family ownership, however, can be attenuated by potentially greater
demand for higher earnings quality from family firms by users of financial
statements. 3 The entrenchment effect of family ownership motivates users
of financial statements to demand high-quality earnings from family firms to
better safeguard their assets and interests. In other words, family firms will
face stricter contracting terms that are more sensitive to financial reporting
quality if contracting parties perceive that family ownership creates an en-
trenchment effect on financial reporting. If so, family firms have incentives
to meet the greater demand by users of financial statements and to provide
higher-quality accounting information in exchange for better contracting
terms, such as a lower cost of capital.
A competing view is the alignment effect, which is based on the argument
that family firms have incentives to report earnings in good faith and thus
earnings are of higher quality. The alignment effect implies that concen-
trated ownership creates greater monitoring by controlling owners (Dem-
setz and Lehn [1985], Shleifer and Vishny [1997]), suggesting that con-
trolling families might monitor firms more effectively. For example, family
firms are able to make decisions more quickly and have incentives to cre-
ate long-term employee loyalty (Weber et al. [2003]). Moreover, because of
family members’ long-term and sustainable presence in the firm and their
intention to preserve the family name, founding families have a greater
2 Fan and Wong [2002] consider the entrenchment effect and the information effect as two
complementary effects. Because both effects increase agency costs, I refer to the entrenchment
effect more generally as including both entrenchment resulting from controlling sharehold-
ers’ effective control and information asymmetry between controlling shareholders and other
shareholders.
3 I thank an anonymous reviewer for the argument that family ownership affects both de-
mand and supply of earnings quality.

622 D. WANG
stake in the firm than nonfamily professional executives. Conversely, man-

agers with short-run objectives may report earnings that maximize their per-
sonal wealth at the cost of shareholders (Christie and Zimmerman [1994]).
Hence, founding families are more likely to forgo short-term benefits from
managing earnings because of the incentives to pass on their business to
future generations and to protect the family’s reputation. Accordingly, the
alignment effect implies that founding family firms are less likely to engage
in opportunistic behavior in reporting accounting earnings because it po-
tentially could damage the family’s reputation, wealth, and long-term firm
performance. Thus, family firms are motivated to report earnings of higher
quality than nonfamily firms.
While the alignment effect motivates family firms to report higher-quality
earnings, it may reduce the demand for high-quality financial information by
contracting parties if they believe that family ownership enhances corporate
governance. For example, outside shareholders of family firms may rely less
on the quality of financial statements to monitor family members (insiders)
because outside shareholders’ interests are better aligned with those of the
founding families. In other words, the contracting terms for family firms are
less sensitive to the quality of financial information than those for nonfamily
firms. This, in turn, gives family firms less incentive to report high-quality
financial information.
Overall, the two competing theories of the effect of family ownership
on the demand and supply of earnings quality indicate that the relation
between family ownership and earnings quality is an empirical issue. Us-
ing data from the S&P 500 companies for the period 1994–2002, I find
that, on average, founding family ownership is significantly associated with
higher earnings quality. Specifically, founding family ownership is associ-
ated with lower abnormal accruals, greater earnings informativeness, and
less persistence of transitory loss components in earnings, after controlling
for institutional and nonfamily insider ownership, size, risk, and other firm-
specific variables. The results are also robust to controlling for potential
bias from time-series correlation, executive compensation, and nonfamily
blockholder ownership. In addition, I find evidence that the relation be-
tween family ownership and earnings quality is nonlinear.
My findings indicate that founding family ownership enhances the com-
munication between insiders and users of financial statements through
higher-quality accounting earnings. The higher earnings quality of family
firms may result from better alignment of interests between family mem-
bers and other shareholders, or it may be attributed to greater demand
for earnings quality by users of financial statements. The results in my pa-
per are consistent with and extend the study of Warfield, Wild, and Wild
[1995] that documents a positive relation between managerial ownership
and earnings quality. This paper also corroborates the findings of recent
studies that family firms perform better and incur a lower cost of debt than
nonfamily firms (Anderson, Mansi, and Reeb [2003], Anderson and Reeb
[2003a]).
This study adds to a growing body of research on earnings quality

and ownership structure (Fan and Wong [2002], Francis, Schipper, and
Vincent [2005]), and more broadly, corporate governance (Agrawal and
Chadha [2005], Bowen, Rajgopal, and Venkatachalam [2003], Klein [2002],
Srinivasan [2005]). Amid the growing concern about earnings quality in cor-
porate America (Lev [2003]), this study provides insights into earnings qual-
ity within a unique ownership context. Understanding how earnings quality
varies with ownership structure provides potential benefits to investors. Fi-
nally, this study documents evidence challenging the traditional view that
family firms have entrenched ownership and thus have greater incentives to
opportunistically manage reported earnings than nonfamily firms.
The rest of this paper is organized as follows. Section 2 develops the
study’s hypothesis and section 3 presents the empirical models. Section 4
reports the sample and descriptive statistics. Section 5 reports the results of
the abnormal accruals analysis, the earnings informativeness analysis, and
the analysis of persistence of transitory losses. Section 6 presents additional
tests, and section 7 concludes the paper.
2. Motivation and Theoretical Development

Shleifer and Vishny [1997] argue that large shareholders have incentives
to maximize their own benefits at the cost of other shareholders. Consis-
tent with this argument, Fama and Jensen [1983] and Morck, Shleifer, and
Vishny [1988] show theoretically and empirically that combined ownership
and control create greater agency conflicts. Concentrated ownership carries
potential benefits, however, in that large shareholders may exert greater
monitoring on management (Shleifer and Vishny [1997]). Demsetz and
Lehn [1985] find evidence that controlling shareholders have strong incen-
tives to diminish agency problems and maximize firm value. In other words,
concentrated ownership aligns the interests of controlling shareholders with
those of noncontrolling shareholders. Grossman and Hart [1986] present
theoretical evidence that removing firm ownership from managers leads
to greater transaction costs, indicating that managerial ownership has the
potential to reduce agency conflicts.
The relation between founding family ownership and earnings quality
potentially fits in the realm of agency theory, in which family members
expropriate wealth from other shareholders by managing accounting earn-
ings ( Jensen and Meckling [1976]). 4 Prior studies investigate the relation
between earnings quality and various features of ownership structure. Specif-
ically, earnings are more informative for firms with less concentrated own-
ership in East Asian countries (Fan and Wong [2002]). In the U.S. context,
4 Burkart, Panunzi, and Shleifer [2003, p. 2168] document a nonpecuniary benefit of found-
ing family control that does not compromise the wealth of other shareholders. It is a significant
amenity potential of family members to have “. . . his child run the company that bears the family
name.”
624 D. WANG
Francis, Schipper, and Vincent [2005] document lower earnings response

coefficients (ERCs) for firms with ownership structures that have unequal
voting rights. Warfield, Wild, and Wild [1995] examine the relation between
managerial ownership and earnings quality and find that higher managerial
ownership reduces managers’ incentives to report accounting earnings that
deviate from the substance of underlying economic transactions. While it is
possible that family members manage earnings for private gains at the ex-
pense of other shareholders, existing theories predict that founding family
ownership affects the demand and supply of earnings quality in two com-
peting ways: the entrenchment effect and the alignment effect.
2.1. FOUNDING FAMILY OWNERSHIP AND THE ENTRENCHMENT EFFECT

The entrenchment effect is based on the argument that concentrated
ownership creates incentives for controlling shareholders to expropriate
wealth from other shareholders (Fama and Jensen [1983], Morck, Shleifer,
and Vishny [1988], Shleifer and Vishny [1997]). In turn, the entrenchment
effect implies that family members, as controlling shareholders, may extract
private benefits from the firm at the cost of minority shareholders. For
example, DeAngelo and DeAngelo [2000] document that the management
team of Time Mirror Company in 1994 dramatically cut dividends to other
shareholders, but maintained a special dividend for the Chandler family.
Thus, the entrenchment effect predicts that family firms report earnings
of lower quality because family members may have greater incentives to
manage earnings for their private benefits.
In addition to academic studies, there is ample anecdotal evidence that
immediate family relationships can generate agency conflicts. As a result,
some firms consider that having immediate family members on the board
and the management team could have a negative impact on the company.
For example, the 2002 proxy statement of Chiron Corporation states “There
is no family relationship between any of the nominees or between any of
the nominees and any of Chiron’s executive officers” (Chiron Corporation
[2002, p. 6]). Examples abound of the entrenchment effect in family-owned
firms. For example, in June 2002, the Securities and Exchange Commission
filed charges against the former CEO of Rite Aid Corp., the son of its founder,
and against other top management members for overstating pretax income
by $2.3 billion from 1997 to 1999 for annual bonuses (SEC [2002]). Camp-
bell Soup Co., a firm with founding family ownership, provides another
example of accounting fraud. The company was charged with artificially
boosting profits in the 1990s by using fraudulent shipments.
2.1.1. Supply of Earnings Quality. Prior literature largely focuses on the

way in which ownership structure affects the supply of earnings quality. The
study by Fan and Wong [2002], which documents lower ERCs for firms with
greater ownership concentration, is consistent with the entrenchment effect
of founding family ownership on earnings quality in East Asian countries.
In East Asia, almost all of the largest firms are managed by entrepreneurial
founding families (Anonymous [1996]). The findings of Fan and Wong

[2002] imply that family ownership is associated with the supply of lower
earnings quality. Although their evidence may indicate that greater family
ownership creates greater agency conflicts and information asymmetry, the
legal protection of minority shareholders is weaker, and financial reporting
is less transparent in East Asia than in the United States. In addition, the
demand for earnings quality is lower in East Asia than in the United States
because users of financial statements can rely on information from private
sources in East Asian countries (Ball, Robin, and Wu [2003]). Therefore,
the conclusion of Fan and Wong [2002] does not necessarily generalize
to a country such as the United States with stronger investor protection,
more transparent disclosure, and greater demand for public accounting
information.
In the U.S. context, Francis, Schipper, and Vincent [2005] examine the
relation between earnings and dividend informativeness and stock classes.
They find that earnings of firms with dual class equity structures are less
informative than earnings of firms with single class equity structures. Firms
with dual class equity structures have inferior corporate governance and thus
report earnings of lower quality. Their findings may indicate an entrench-
ment effect of founding family ownership on earnings quality because dual
class firms tend to have higher ownership concentration (Francis, Schipper,
and Vincent [2005]).
2.1.2. Demand for Earnings Quality. Ball, Robin, and Wu [2003] document
evidence that the incentives of financial statement preparers play an essen-
tial role in reporting high-quality financial information. Extending this line
of research, Ball and Shivakumar [2005a] argue that market demand in-
creases the quality of financial reporting. They document evidence that
private firms in the U.K. market recognize economic losses in a less timely
manner than public firms, although private and public firms comply with
substantially equivalent accounting standards. They conclude that earnings
of public firms are of higher quality because the market demands higher
earnings quality from public firms than it does from private firms.
Financial reporting can be of different quality if the incentives for high-
quality financial reporting are different for family and nonfamily firms.
While the entrenchment effect predicts that family ownership is associated
with the supply of lower earnings quality, users of financial statements may
demand greater earnings quality from family firms if they perceive that fam-
ily ownership is associated with inferior corporate governance. To protect
their assets and interests, users of financial statements may set contract-
ing terms that are more sensitive to the quality of accounting earnings
with family firms. As a result, the perceived entrenchment effect of fam-
ily ownership may incur greater demand for earnings quality by users of
financial statements. This, in turn, motivates family firms to report earn-
ings of higher quality for better contracting terms, such as a lower cost of
capital.
626 D. WANG
2.2. FOUNDING FAMILY OWNERSHIP AND THE ALIGNMENT EFFECT

The alignment effect is based on the notion that the interests of found-
ing families and other shareholders are better aligned because of the large
blocks of stock owned by family members and their long-term presence.
Therefore, according to the alignment effect, founding families are less
likely to expropriate wealth from other shareholders through managing
earnings. Because the wealth of founding families is closely tied to firm
value, families have strong incentives to monitor employees (Anderson and
Reeb [2003a]) and to create long-term loyalty in employees (Weber et al.
[2003]). In addition, long-term orientation and reputation protection dis-
courage family firms from opportunistically managing earnings, because
earnings management activities are more likely to be short-term oriented
and perhaps even detrimental to long-term firm performance. Stronger
monitoring mechanisms, such as “No Absentee Landlords” (Weber et al.
[2003, p.110]), are observed in the boards of directors of founding fam-
ily firms (Anderson and Reeb [2003b], Weber et al. [2003]). In turn, the
strong monitoring mechanisms motivate family members to communicate
more effectively through higher-quality accounting information with other
shareholders and creditors, thereby reducing the cost of debt (Anderson,
Mansi, and Reeb [2003]).
Consistent with the alignment effect, founding family firms seem to per-
form better and have stronger corporate governance. Anderson and Reeb
[2003a] find evidence that founding family firms are better performers than
nonfamily firms, as measured by accounting performance (return on assets)
and market performance (Tobin’s Q ). In addition, Anderson, Mansi, and
Reeb [2003] document evidence that founding family firms are associated
with a lower cost of debt. Although their findings are not directly related
to earnings quality, they imply a positive relation between founding family
ownership and corporate governance.
The business success of founding family firms is not uncommon. For ex-
ample, the Walton family founded Wal-Mart Stores Co., presently the largest
retailer in the world, reporting annual sales of $286 billion worldwide (2004
fiscal year). The Walton family is one of the richest families in the United
States and the family continues to be the largest shareholder group. There-
fore, there exists both academic and anecdotal evidence consistent with the
alignment effect that founding family ownership creates incentives for fam-
ily members to maximize the wealth of all shareholders. Overall, the long-
term business horizon, a higher stake in the firm and incentives to preserve
the family’s reputation constrain founding families from opportunistically
managing accounting earnings for private gains.
2.2.1. Supply of Earnings Quality. As argued above, the alignment ef-
fect predicts that family ownership is associated with stronger corporate
governance. The stronger corporate governance in turn mitigates man-
agers’ opportunism in managing reported earnings (Klein [2002]). Thus,
family firms may report earnings of higher quality relative to nonfamily

firms. Founding families have incentives to produce higher-quality earnings
through less aggressive earnings management behavior in order to avoid
potential damage to the family’s reputation and improve long-term firm
performance. The finding of Warfield, Wild, and Wild [1995] that greater
managerial ownership is associated with higher earnings quality suggests a
potential alignment effect of founding family ownership on earnings qual-
ity. Because founding families tend to own large blocks of stock, family firms
may have a higher level of managerial or insider ownership. In turn, found-
ing family firms are expected to report earnings of higher quality.
2.2.2. Demand for Earnings Quality. Shareholders may rely less on public
financial information to monitor managers if the interests of insiders and
outsiders are better aligned. As such, shareholders of family firms may have
less incentive to demand high-quality financial information because family
member managers, who may also be influential shareholders, have access to
family firms’ private information. Similar arguments can be made for cred-
itors. If creditors believe that family ownership is associated with stronger
corporate governance, the debt contracting terms for family firms will be
less sensitive to earnings quality because creditors may assume that financial
statements of family firms are prepared in good faith. Overall, the alignment
effect implies that family ownership reduces the demand for quality financial
reporting.
Earnings are of lower quality if the incentives to report high-quality fi-
nancial information are low (Ball, Kothari, and Robin [2000], Ball Robin,
and Wu [2000, 2003], Ball and Shivakumar [2005a]). The lower demand
for earnings quality reduces the incentives of family firms to report higher-
quality financial statements. Ball, Robin, and Wu [2003] document evidence
that financial reporting quality in four Asian countries with common law tra-
ditions (Hong Kong, Malaysia, Singapore, and Thailand) is no higher than
that in code law countries. The demand for high-quality earnings is low in
the four countries even though investor protection is considered stronger in
common law countries than in code law countries (La Porta et al. [1998]).
The findings of Ball, Robin, and Wu [2003] may imply that family firms re-
port earnings of lower quality because their incentives to report high-quality
earnings are lower than those of nonfamily firms.
Overall, because existing theories provide competing and alternative pre-
dictions about the effects of founding family ownership on earnings quality,
the study’s hypothesis is nondirectional and states:
H1: Earnings quality is systematically related to founding family

ownership.
The directional relation between founding family ownership and earnings

quality becomes an empirical question.
628 D. WANG
3. Model Description
3.1. MEASURES OF EARNINGS QUALITY
While managerial opportunism reduces earnings quality, there is no
single measure of earnings quality in the existing literature. As such, I
adopt three commonly used proxies for earnings quality: abnormal accru-
als, earnings informativeness, and persistence of transitory loss components
in earnings. The first measure, the absolute value of abnormal accruals, is
based on the Dechow and Dichev [2002] model, as modified by Ball and
Shivakumar [2005b]. Ball and Shivakumar [2005b] incorporate the asym-
metrically timely recognition of unrealized gains and losses into the Dechow
and Dichev [2002] abnormal accruals estimation model. Earnings quality
is considered to be lower when actual accruals deviate from expected ac-
cruals based on underlying economic transactions. The second measure is
earnings informativeness, that is, ERCs. Earnings informativeness has been
widely used as a market measure of earnings quality (Fan and Wong [2002],
Francis, Schipper, and Vincent [2005], Warfield, Wild, and Wild [1995]).
Greater earnings informativeness implies that earnings are of higher qual-
ity. The third measure is persistence of transitory loss components in earn-
ings. Earnings are conservative when transitory loss components in earnings
are less persistent than transitory gain components (Ball and Shivakumar
[2005a], Basu [1997]). Existing literature has provided ample evidence that
earnings are of higher quality when reported earnings are more conserva-
tive (Ball, Kothari, and Robin [2000], Ball, Robin, and Wu [2000, 2003],
Ball and Shivakumar [2005a], Basu [1997]). I adopt the Basu [1997] model
as modified by Ball and Shivakumar [2005a] to investigate the relation be-
tween family ownership and persistence of transitory loss components in
earnings.
3.2. EXPERIMENTAL VARIABLES
Following Anderson and Reeb [2003a], founding family ownership is de-
fined and tested in two ways. First, a binary variable (F FAM ) is coded one if
founding family members are either on the board of directors or in the top
management of the company, irrespective of the level of family common
stock ownership, and coded zero otherwise. The influences of founding
families exerted on the firm, represented by voting power, may go beyond
the common stock percentages owned by family members. Therefore, the
binary variable of family ownership is used as the primary analysis. For com-
pleteness, a secondary variable (FAM OWN ) is used to measure founding
family ownership based on the percentage of common stock owned by fam-
ily members, with a larger value indicating greater family interests in the
firm.
As a supplement to the binary and continuous variables that measure
family ownership, founding family firms are classified further into three
groups with different CEO attributes: founder CEO (F CEO), which equals
one if a family firm has the founder as CEO and zero otherwise, descendant
CEO (D CEO), which equals one if the family firm is headed by a family
descendant and zero otherwise, or hired CEO (H CEO), which equals one
if the CEO is hired from outside the family and zero otherwise, with non-
family firms as the default comparison group. Family CEOs (founders and
descendants) might be drawn from a suboptimal labor pool that limits more
talented professional executives from running the firm (Anderson and Reeb
[2003a]). Thus, family firms with family members as CEOs might perform
poorly. Conversely, family CEOs can enhance firms’ wealth because they pos-
sess special expertise (Morck, Shleifer, and Vishny [1988]) and intentions
of long-term presence (Anderson, Mansi, and Reeb [2003]). Therefore, I
do not make differential predictions across the three types of founding fam-
ily firms. Rather, results of family ownership with different CEO attributes
are presented only for the purpose of shedding more light on the relation
between family ownership and earnings quality.
3.3. EMPIRICAL MODELS
3.3.1. Abnormal Accruals Analysis. The conventional linear discretionary
accruals models introduced by Jones [1991] and Dechow and Dichev [2002]
have been widely used in accounting literature to estimate abnormal accru-
als (Dechow, Sloan, and Sweeney [1995], DeFond and Jiambalvo [1994],
Francis et al. [2005]). In a recent study, however, Ball and Shivakumar
[2005b] argue that the conventional linear accruals models fail to recognize
the nonlinear nature of the accounting accruals process. By incorporating
asymmetrically timely loss recognition, they provide piecewise nonlinear ab-
normal accruals estimation models that substantially increase the models’
explanatory powers relative to traditional linear models.
Thus, I adopt the Dechow and Dichev [2002] model, as modified by Ball
and Shivakumar [2005b], to estimate abnormal accruals. 5 Specifically, the
following piecewise nonlinear abnormal accruals model is used to estimate
abnormal accruals: 6
ACC t = α0 + α1 CF t + α2 CF t−1 + α3 CF t+1 + α4 DCF t + α5 DCF ∗t CF t + e t (1)
where:
ACC t = total accruals at t, scaled by average total assets at t; total accruals
are earnings before extraordinary items minus operating cash
flows;
CF t = operating cash flows at t, scaled by average total assets at t;
5 The results are robust to using the modified Jones [1981] model (Dechow, Sloan, and
Sweeney [1995]) for estimation of abnormal accruals.

6 The average adjusted R 2 value from this piecewise nonlinear accruals model, estimated
using all Compustat firms from 1994 to 2002, is 42.80%. In contrast, the average adjusted
R 2 based on the traditional linear Dechow and Dichev [2002] model, ACC t = α 0 + α 1 CF t
+ α 2 CF t −1 + α 3 CF t +1 + e t , is 37.36%. It is apparent that adding the nonlinear nature of the
accounting accruals process to the traditional accruals model substantially improves model
specification.
630 D. WANG
CF t −1 = operating cash flows at t−1, scaled by average total assets at t;

CF t +1 = operating cash flows at t+1, scaled by average total assets at t;
DCF t = one if the change in cash flows at t is less than zero (CF t – CF t −1
< 0), and zero otherwise;
e t = error term.
The interaction term, DCF t ∗ CF t , is the proxy for economic losses. Consis-
tent with Ball and Shivakumar [2005b], equation (1) is estimated in each
industry measured by a three-digit Standard Industrial Classification (SIC)
code and requires at least 30 observations in each industry regression. The
error term, e t , is expected to capture the unexpected portion of total accru-
als that deviate from economic transactions.
The absolute value of the residuals (ABS ACC t ) from equation (1) is the
proxy for earnings management. The absolute value is used because earn-
ings management can involve either income-increasing accruals or income-
decreasing accruals to meet earnings targets (Bowen, Rajgopal, and Venkat-
achalam [2003], Klein [2002], Reynolds and Francis [2000], Warfield, Wild,
and Wild [1995]). A higher value means a greater level of earnings man-
agement or lower earnings quality. To test the relation between family own-
ership and earnings management, ABS ACC t is employed as the dependent
variable in the following equation:
ABS ACC t = δ0 + δ1 FAM PROXYt + δ2 SIZE t + δ3 ROAt
+ δ4 LEV t + δ5 GROWTH t + δ6 INST t + δ7 INSIDER t
+ δ8 AGE t + δ9 LOSS t + e t (2)
where:
ABS ACC t = absolute value of abnormal accruals at t;

FAM PROXY t = proxies of founding family ownership at t (defined in
section 3.2);
SIZE t = natural log of total assets at t;
ROA t = net income at t divided by average total assets at t;
LEV t = firm leverage at t, measured by total liabilities divided by
total assets;
GROWTH t = growth rate in sales at t;
INST t = institutional ownership at t, measured by the percentage
of total common equity owned by institutions;
INSIDER t = nonfamily insider ownership, measured by the percent-
age of total common equity owned by managers and
directors, excluding the percentage of total common
equity owned by family members;
AGE t = firm age in years at t;
LOSS t = one if net income < 0, and zero otherwise;
e t = error term.
Following prior literature (Beker et al. [1998], Cheng and Warfield [2005],
Reynolds and Francis [2000]), variables are included for firm size (SIZE t ),
profitability (ROA t ), risk for bankruptcy (LEV t and LOSS t ), and growth op-
portunities (GROWTH t ). In addition, other ownership features, such as in-
stitutional ownership and insider ownership, can affect both the monitoring
mechanisms and the incentives to manage earnings. Thus, I control for in-
stitutional ownership (INST t ) and nonfamily insider ownership (INSIDER t )
in order to tease out the effect of founding family ownership. Lastly, firm
age (AGE t ) is controlled for because, intuitively, older firms are less likely
to be founding family owned (Anderson and Reeb [2003a]).
The coefficient on FAM PROXY t (δ 1 ) tests the relation between family
ownership and the absolute value of abnormal accruals. A positive estimate
will indicate that family firms have earnings that are of lower quality; how-
ever, if the estimate on δ 1 is negative, it implies that family firms report
earnings of higher quality.
3.3.2. Earnings Informativeness. The second proxy of earnings quality is

earnings informativeness, measured by the ERCs (Fan and Wong [2002],
Francis, Schipper, and Vincent [2005], Warfield, Wild, and Wild [1995]).
The following model is adopted to investigate the relation between earnings
informativeness and family ownership. 7, 8
RET t = β0 + β1 NI t + β2 NI t ∗ FAM PROXY t + β3 NI t ∗ SIZE t
+ β4 NI t ∗ LEV t + β5 NI t ∗ MB t + β6 NI t ∗ INST t
+ β7 NI t ∗ INSIDER t + β8 NI t ∗ AGE t + β9 NI t ∗ LOSS t
+ Industry Fixed Effects + e t (3)

where:
RET t = 12-month cumulative raw return ending three
months after the fiscal year-end at t;
NI t = net income for year t, scaled by the market value
of equity at the end of t−1;
MB t = market-to-book ratio at t;
Industry fixed effects = industry dummies based on three-digit SIC codes.
All other variables are as previously defined. Consistent with H1, the coef-
ficient on FAM PROXY t (β 2 ) is expected to be significantly different from
7 As a sensitivity analysis, I include FAM PROXY in the model to allow the intercept to vary
across family and nonfamily firms. The results are robust to the addition of FAM PROXY as an
additional control.
8 The results are robust to using the Center for Research in Security Prices (CRSP) equal-
weighted or value-weighted market-adjusted 12-month cumulative returns and to using net

income before extraordinary items.
632 D. WANG
zero. A positive (negative) estimate on β 2 will be evidence that family firms

are associated with more (less) informative earnings than nonfamily firms.
3.3.3. Persistence of Transitory Loss Components in Earnings. Basu [1997]

finds evidence that negative earnings changes (transitory loss components
in earnings) are less persistent than positive earnings changes. He argues
that the asymmetric persistence of negative earnings changes is the result
of the conservative nature of accounting earnings. Ball and Shivakumar
[2005a] adopt Basu’s [1997] serial dependence model and find that the
transitory loss components in earnings of public firms are less persistent (or
are recognized in a more timely fashion) than those of private firms in the
U.K. market. Thus, I adopt the Basu [1997] piecewise serial dependence
model as modified by Ball and Shivakumar [2005a] to test the relation be-
tween family ownership and persistence of transitory loss components in
earnings. 9, 10, 11
NI t = λ0 + λ1 DNI t−1 + λ2 NI t−1 + λ3 NI t−1 ∗ DNI t−1
+ λ4 FAM PROXY t + λ5 DNI t−1 ∗ FAM PROXY t
+ λ6 NI t−1 ∗ FAM PROXY t + λ7 NI t−1 ∗ DNI t−1 ∗ FAM PROXY t
+ λ8 SIZE t + λ9 DNI t−1 ∗ SIZE t + λ10 NI t−1 ∗ SIZE t
+ λ11 NI t−1 ∗ DNI t−1 ∗ SIZE t + λ12 INST t + λ13 DNI t−1 ∗ INST t
+ λ14 NI t−1 ∗ INST t + λ15 NI t−1 ∗ DNI t−1 ∗ INST t
+ λ16 INSIDER t + λ17 DNI t−1 ∗ INSIDER t + λ18 NI t−1 ∗ INSIDER t
+ λ19 NI t−1 ∗ DNI t−1 ∗ INSIDER t + λ20 LEV t + λ21 DNI t−1 ∗ LEV t
+ λ22 NI t−1 ∗ LEV t + λ23 NI t−1 ∗ DNI t−1 ∗ LEV t + 24 AGE t
+ λ25 DNI t−1 ∗ AGE t + λ26 NI t−1 ∗ AGE t
+ λ27 NI t−1 ∗ DNI t−1 ∗ AGE t + Industry Fixed Effects + e t (4)
9 Basu’s [1997, table 1] reverse regression model is more widely used in accounting literature
to examine earnings conservatism or timely loss recognition. However, I do not find a significant
relation between family ownership and timely loss recognition using Basu’s [1997] reverse
regression approach. Gigler and Hemmer [2001] argue that Basu’s [1997] reverse regression
approach may generate biased results because it does not control for the potential effect of
voluntary disclosures on stock prices.
10 Basu [1997, p. 19] argues “. . . timeliness and persistence are different ways of viewing
the same phenomenon.” Therefore, throughout this paper, I adopt persistence of transitory loss
components in earnings instead of timely loss recognition to distinguish between Basu’s [1997] serial
dependence model and reverse regression model.
11 Basu’s [1997] serial dependence model has two potential limitations: it cannot distinguish
the transitory components in earnings from random accruals errors, and it cannot identify
whether the transitory loss components in earnings are recognized in a timely manner (Ball
and Shivakumar [2005a]). Therefore, the results in the analysis of persistence of transitory
losses should be interpreted with caution.
where:
NI t = change in net income before extraordinary items at t, scaled
by average total assets at t−1;
NI t −1 = change in net income before extraordinary items at t−1,
scaled by average total assets at t−1;
DNI t −1 = one if NI t −1 < 0, and zero otherwise.
All other variables are the same as previously defined. Consistent with
Ball and Shivakumar [2005a], the coefficient on NI t −1 (λ 2 ) is ex-
pected to be insignificant, and the coefficient on NI t −1 ∗ DNI t −1
(λ 3 ) is expected to be negative and significant. The coefficient on
NI t −1 ∗ DNI t −1 ∗ FAM PROXY t (λ 7 ) captures the incremental persis-
tence of transitory losses for family firms relative to nonfamily firms. A pos-
itive estimate on λ 7 will indicate that transitory losses for family firms are
more persistent (lower earnings quality). In contrast, a negative estimate
on λ 7 will indicate that family ownership is associated with less persistent
transitory losses (higher earnings quality).
4. Sample Description
4.1. DATA COLLECTION PROCESS
The empirical analysis is performed on firms that were listed on the S&P
500 index either on December 31, 1994 or on December 31, 2002. 12 Consis-
tent with Anderson and Reeb [2003a], firms in regulated industries (utilities
and the financial industries) are excluded. Table 1, panel A describes the
number of firms identified from the 1994 and 2002 S&P 500 indices. Af-
ter deleting firms that are in the financial and utilities industries and firms
with unavailable proxy statements, 397 firms (132 family firms) are iden-
tified from the 1994 S&P 500 index and 379 firms (151 family firms) are
identified from the 2002 S&P 500 index. Among these firms, 234 firms (76
founding family firms) appear in both years as S&P 500 index constituents.
Therefore, 542 (397 + 379 − 234 = 542) unique firms are identified for
further data collection. Of the 542 unique firms, 207 (132 + 151 − 76 =
207) are unique founding family firms.
I obtain 4,195 firm-year proxy statements from Lexis-Nexis for the nine-
year period from 1994 through 2002 for the 542 unique firms identified in
panel A of table 1. I examine these proxy statements and corporate histo-
ries to identify family firms, family members (founders and descendants),
family holdings, and CEO attributes. 13 Corporate histories are drawn from
12 In Compustat (Research Insight), December 31, 1994 is the earliest date that the S&P 500
index constituents can be identified.

13 Sometimes, the descendants’ family names are not the same as those of the founders.
However, firm histories usually identify important persons as sons, daughters, cousins, son-in-
laws, grandsons, etc.
634 D. WANG
the Hoovers database or from the individual firm’s website. Institutional

ownership and insider ownership are obtained from Compact Disclosure.
Firm age is determined from the firm’s history. If a firm’s starting year is
not stated in its history, firm age is taken from Moody’s Manual. Stock
prices (returns) are obtained from CRSP. All other variables are available on
Compustat.
The data selection process for the final analysis is reported in table 1,
panel B. I start with 4,195 firm-year observations that have proxy statements
available in Lexis-Nexis from 1994 through 2002 for the 542 unique firms. To
obtain the data for the abnormal accruals analysis, I delete 162 observations
without data for abnormal accruals, 94 observations with acquisitions, and
365 observations with missing values in institutional ownership, nonfamily
TABLE 1
Sample Description
Panel A: Family and nonfamily firms in the 1994 and 2002 S&P 500 indices
December 31, December 31,
1994 S&P 500 2002 S&P 500
Constituents Constituents
Number of firms 500 500
Less:
Financial institutions (SIC Code: 6000–6999) and (98) (121)
utilities companies (SIC Code: 4900–4999)
Proxy statement not available (5) (0)
Number of firms identified for further data collection 397 379
Number of founding family firms 132 151
Percentage of founding family firms 33.25% 39.84%
Number of unique firms identified for further data 542
collection
Number of unique founding family firms 207
Percentage of unique founding family firms 38.19%
Panel B: Sample selection

Persistence of
Abnormal Earnings Transitory
Accruals Informativeness Losses
Initial sample with proxy statements available in 4,195 4,195 4,195
Lexis-Nexis from 1994 through 2002 for the
542 unique firms identified in panel A
Less: observations without abnormal accruals (162)
Less: observations without stock returns from (135)
CRSP
Less: observations with acquisitions (94) (95) (95)
Less: observations with missing values on (365) (317) (396)
institutional ownership, nonfamily insider
ownership and other control variables
Less: top and bottom 1% of ABS ACC t (70)
Less: top and bottom 1% of RET t and NI t (146)
Less: top and bottom 1% of NI t and NI t −1 (127)
Less: observations with |studentized residuals| > 4 (48) (19) (25)
Number of observations in the final analysis 3,456 3,483 3,552
T A B L E 1 — Continued
Panel C: Founding family ownership by year (based on abnormal accruals sample)
Percentage
Median% of
Avg. % of of Family
Common Common Percentage Firms
Stock Stock of Family % of Family with
Total No. of Percentage Owned by Owned by Firms with Firms with Hired
Fiscal No. of Family of Family Family Family Founders Descendants Outsiders
Year Firms Firms Firms Members Members as CEOs as CEOs as CEOs
1994 393 162 41.22% 11.54% 6.04% 32.72% 24.69% 42.59%
1995 400 165 41.25% 11.55% 5.80% 32.73% 24.24% 43.03%
1996 393 156 39.69% 11.45% 5.30% 32.05% 23.08% 44.87%
1997 402 156 38.81% 11.10% 5.27% 35.26% 19.87% 44.87%
1998 399 161 40.35% 10.53% 4.50% 33.54% 19.88% 46.58%
1999 395 157 39.75% 10.17% 4.40% 33.76% 18.47% 47.77%
2000 393 154 39.19% 9.63% 4.42% 31.17% 17.53% 51.30%
2001 340 132 38.82% 8.81% 4.48% 23.48% 16.67% 59.85%
2002 341 124 36.36% 7.40% 4.05% 20.97% 16.94% 62.10%
Total 3,456 1,367 39.55% 10.35% 5.01% 31.02% 20.34% 48.64%
Variable definitions:
ABS ACC t = absolute value of abnormal accruals at t, estimated based on the Dechow and Dichev [2002]
model as modified by Ball and Shivakumar [2005b];
RET t = 12-month buy-and-hold stock return ending three months after the fiscal year-end at t;
NI t = net income at t, scaled by market value of equity at the end of t−1;
NI t = change in net income before extraordinary items at t, scaled by average total assets at t−1;
NI t −1 = change in net income before extraordinary items at t−1, scaled by average total assets at t−1.
Abbreviations: CRSP, Center for Research in Security Prices; S&P, Standard and Poor’s; SIC, Standard
Industrial Classification.
insider ownership, and other control variables. 14 In addition, to avoid outlier

effects, observations of the top and bottom 1% of the dependent variable (70
observations) and observations with |studentized residuals| >4 (48 observa-
tions) are excluded. 15 After these screenings, 3,456 observations remain for
the abnormal accruals analysis. As shown in table 1, panel B, similar screen-
ings leave 3,483 observations for the earnings informativeness analysis and
3,552 observations for the analysis of persistence of transitory losses.
Table 1, panel C presents a description of family ownership over time us-
ing the sample for the abnormal accruals analysis. There are 3,456 firm-year
observations in the abnormal accruals analysis, of which, 1,367 (39.55%) are
14 In all analyses throughout this paper, the results are robust to including observations with
acquisitions.
15 In the abnormal accruals and earnings informativeness analyses, the results are robust to
inclusion of these outliers. The outliers do affect the results in the analysis of persistence of
transitory losses, however, because the coefficients on test variables become insignificant if the
extreme variables are included in the regressions. Further screenings show that the inference—
family ownership is associated with less persistent transitory losses—still holds when the top
and bottom 2% or 3% of NI t and NI t −1 , and/or |studentized residuals| > 3 are excluded.
This indicates that inclusion of outliers in the analysis of persistence of transitory losses works
against finding results.
636 D. WANG
family firm-year observations. On average, family members own 10.35% of

common stock. The average (median) family ownership measured by per-
centage of common equity is 11.54% (6.04%) in 1994, but this decreases to
7.40% (4.05%) in 2002. 16 Of the family firms, 32.72% are run by founder
CEOs in 1994 and 20.97% are run by founder CEOs in 2002. The corre-
sponding percentage of family firms run by family descendants is 24.69% in
1994, but this decreases to 16.94% in 2002. In comparison, the percentage
of family firms run by hired nonfamily CEOs is 42.59% in 1994, and this
increases to 62.10% in 2002. 17
The percentage of family firms run by family members (founders and
their descendants) is declining over time, while the percentage run by hired
nonfamily member CEOs is increasing. This phenomenon might mitigate
the criticism that family ownership violates the axiom of having the best
persons running the business because of the increasing percentage of family
firms run by outsider CEOs.
4.2. DESCRIPTIVE STATISTICS
The descriptive statistics for the variables other than family ownership in
the abnormal accruals analysis are reported in table 2, panel A for family
firms (N = 1,367) and nonfamily firms (N = 2,089). The average abnormal
accruals (ABS ACC t ) are 0.056 for family firms and 0.053 for nonfamily firms.
This is not surprising, because family firms tend to be smaller. The natural
log of total assets (SIZE t ) is 7.947 for family firms and 8.444 for nonfamily
firms. Relative to nonfamily firms, family firms have a greater return on
assets (ROA t ), a lower leverage ratio (LEV t ), a higher growth rate in sales
(GROWTH t ), and a lower level of institutional ownership (INST t ). Although
family firms have a lower mean percentage of nonfamily insider ownership
(INSIDER t ), the median of nonfamily insider ownership is higher for family
firms than for nonfamily firms. In addition, family firms tend to be younger
firms (AGE t ), and they are less likely to report a loss (LOSS t ).
Table 2, panel B reports the descriptive statistics of the sample in the earn-
ings informativeness analysis. There are 1,363 family firm-year observations
and 2,120 nonfamily firm-year observations. On average, family firms have a
higher 12-month cumulative return (RET t , 0.182 for family firms and 0.119
for nonfamily firms), consistent with the findings in Anderson and Reeb
[2003a], who report better performance of family firms. The means of net
income (NI t ) are 0.041 and 0.040 for family firms and nonfamily firms, re-
spectively. In addition, family firms have a higher market-to-book ratio (MB t ,
16 The average family ownership is lower than that reported by Anderson and Reeb [2003a]
for 1992 (17.9%). Given that the S & P 500 index constituents are different in 1992 and that
there is a tendency for founding family ownership to decline over time, it is not surprising
that founding family ownership is lower in this study than in the study of Anderson and Reeb
[2003a].
17 The descriptive statistics of the family ownership variables (F FAM , FAM OWN , F CEO ,
t t t
D CEO t , and H CEO t ) in the earnings informativeness analysis and the analysis of persistence
of transitory losses are very similar and are therefore not reported.
4.553 for family firms and 4.162 for nonfamily firms). The descriptive statis-
tics of firm size (SIZE t ), leverage (LEV t ), institutional ownership (INST t ),
nonfamily insider ownership (INSIDER t ), firm age (AGE t ), and the dummy
variable for losses (LOSS t ) are comparable with those reported in panel A.
Panel C presents the descriptive statistics of the sample in the analysis
of persistence of transitory losses. The means of change in income before
extraordinary items (NI t ) are 0.015 and 0.008 in year t, for family and
nonfamily firms, respectively. The means of DNI t −1 are 0.307 for family
firms and 0.369 for nonfamily firms, indicating that nonfamily firms are
more likely to report negative earnings changes (NI t −1 < 0). The descrip-
tive statistics of firm size (SIZE t ), leverage (LEV t ), institutional ownership
(INST t ), nonfamily insider ownership (INSIDER t ), and firm age (AGE t ) are
comparable with those reported in panel A.
4.3. CORRELATION
Consistent with the descriptive statistics in table 2, panels A through C, un-
tabulated Spearman correlation results indicate that family ownership is pos-
itively correlated with abnormal accruals (ABS ACC t ), 12-month cumulative
stock return (RET t ), change in net income (NI t and NI t −1 ), return on
assets (ROA t ), growth rate (GROWTH t ), market-to-book ratio (MB t ) and
TABLE 2
Descriptive Statistics
Panel A: Abnormal accruals sample (N = 3,456)
Family Firms (N = 1,367) Nonfamily Firms (N = 2,089)
Std. Std.
Mean Dev. Median P25 P75 Mean Dev. Median P25 P75
ABS ACC t 0.056 0.068 0.033 0.015 0.067 0.053 0.067 0.031 0.014 0.063
SIZE t 7.947 1.312 7.923 7.133 8.770 8.444 1.302 8.452 7.568 9.360
ROA t 0.075 0.082 0.074 0.034 0.116 0.057 0.084 0.057 0.023 0.095
LEV t 0.504 0.199 0.516 0.366 0.628 0.613 0.193 0.626 0.493 0.737
GROWTH t 0.182 0.308 0.110 0.026 0.245 0.095 0.232 0.061 −0.007 0.149
INST t 0.579 0.205 0.593 0.467 0.728 0.622 0.215 0.667 0.531 0.766
INSIDER t 0.038 0.060 0.019 0.010 0.040 0.046 0.114 0.007 0.003 0.024
AGE t 56.03 39.09 48 21 87 74.58 40.65 76 42 103
LOSS t 0.105 0.306 0 0 0 0.147 0.355 0 0 0
Panel B: Earnings informativeness sample (N = 3,483)

Std. Std.
RET t 0.182 0.456 0.130 −0.118 0.391 0.119 0.376 0.096 −0.126 0.322
NI t 0.041 0.058 0.047 0.027 0.066 0.040 0.067 0.050 0.025 0.071
SIZE t 7.952 1.252 7.902 7.196 8.722 8.482 1.328 8.474 7.571 9.419
LEV t 0.498 0.191 0.510 0.364 0.622 0.611 0.183 0.626 0.498 0.730
MB t 4.553 4.544 3.335 2.020 5.521 4.162 4.864 2.861 1.816 4.746
INST t 0.588 0.199 0.602 0.476 0.732 0.625 0.214 0.670 0.537 0.767
INSIDER t 0.036 0.053 0.019 0.010 0.040 0.044 0.110 0.007 0.003 0.023
AGE t 57.85 38.88 51 23 89 74.88 40.45 76 42 103
LOSS t 0.098 0.298 0 0 0 0.133 0.340 0 0 0
638 D. WANG
Panel C: Persistence of transitory losses sample (N = 3,552)
Std. Std.
NI t 0.015 0.056 0.013 −0.007 0.037 0.008 0.056 0.008 −0.013 0.029
NI t −1 0.011 0.049 0.012 −0.006 0.031 0.005 0.050 0.007 −0.012 0.024
DNI t −1 0.307 0.461 0 0 1 0.369 0.483 0 0 1
SIZE t 7.999 1.271 7.942 7.205 8.800 8.498 1.308 8.487 7.589 9.426
LEV t 0.509 0.198 0.520 0.371 0.630 0.620 0.188 0.630 0.502 0.741
INST t 0.583 0.202 0.598 0.473 0.729 0.622 0.214 0.667 0.531 0.766
INSIDER t 0.038 0.062 0.019 0.010 0.040 0.044 0.110 0.007 0.003 0.023
AGE t 58.03 38.56 52 23 89 75.21 40.31 77 43 103
ABS ACC t = absolute value of abnormal accruals at t, estimated based on the Dechow and Dichev [2002] model
as modified by Ball and Shivakumar [2005b];
ROA t = return on assets at t, measured by net income divided by average total assets;
LEV t = leverage at t, measured by total liabilities divided by total assets;
GROWTH t = sales growth rate at t;
INST t = percentage of common stock owned by institutions at t;
INSIDER t = percentage of common stock owned by nonfamily insiders at t;
LOSS t = one if net income at t is negative, zero otherwise;
NI t −1 = change in net income before extraordinary items at t−1, scaled by average total assets at t−1;
DNI t −1 = one if NI t −1 < 0, zero otherwise;
P25 = 25th percentile;
P75 = 75th percentile.
nonfamily insider ownership (INSIDER t ). In comparison, family ownership

is negatively correlated with firm size (SIZE t ), leverage (LEV t ), institutional
ownership (INST t ), firm age (AGE t ), losses (LOSS t ), and the dummy variable
for transitory losses (DNI t −1 ).
5. Multivariate Results
5.1. FOUNDING FAMILY OWNERSHIP AND ABNORMAL ACCRUALS
The results in table 3 are from the ordinary least squares regressions using
the absolute value of abnormal accruals as the dependent variable. Three
models are presented, each with an alternative proxy of founding family
ownership: a binary variable (F FAM t ), a continuous variable (FAM OWN t ),
and three binary variables with different CEO attributes (F CEO t , D CEO t ,
and H CEO t ). All models are significant at p < 0.001. The adjusted R 2
values are 0.13 or higher. Throughout this paper, the p-values on the
independent variables are two-tailed values calculated based on White’s
[1980] heteroskedasticity-corrected standard errors. In model 1, the coeffi-
cient on F FAM t is −0.009 with a p-value < 0.001, suggesting that family firms
report a lower level of abnormal accruals. The magnitude of the coefficient
TABLE 3
Multivariate Analysis of Abnormal Accruals and Founding Family Ownership
Dependent Variable: ABS ACC t
Model 1 Model 2 Model 3
Independent Expected
Variable Sign Estimate p-value Estimate p-value Estimate p-value
Intercept ? 0.134 0.000 0.138 0.000 0.135 0.000
F FAM t ? −0.009 0.000
FAM OWN t ? −0.068 0.000
F CEO t ? −0.001 0.864
D CEO t ? −0.018 0.000
H CEO t ? −0.009 0.000
SIZE t − −0.006 0.000 −0.006 0.000 −0.006 0.000
ROA t ? 0.050 0.176 0.056 0.137 0.050 0.183
LEV t + −0.030 0.000 −0.029 0.000 −0.031 0.000
GROWTH t + 0.029 0.000 0.028 0.000 0.027 0.000
INST t − −0.014 0.014 −0.019 0.001 −0.015 0.008
INSIDER t − 0.003 0.862 0.003 0.822 0.003 0.822
AGE t ? −0.0002 0.000 −0.0002 0.000 −0.0002 0.000
LOSS t + 0.047 0.000 0.046 0.000 0.047 0.000
F -value 57.57 0.000 60.41 0.000 49.09 0.000
Adj. R 2 0.130 0.134 0.133
N 3,456 3,456 3,456
Note that p-values are two-tailed values computed using White’s [1980] heteroskedasticity-corrected
standard errors. Test variables and their statistics are indicated in bold typeface.
ABS ACC t = absolute value of abnormal accruals at t, estimated based on the Dechow and Dichev [2002]
as modified by Ball and Shivakumar [2005b];
F FAM t = one if a firm is a founding family firm at t, zero otherwise;
FAM OWN t = percentage of common stock owned by founding family members at t;
F CEO t = one if the CEO is the founder of the family firm at t, zero otherwise;
D CEO t = one if the CEO is a descendant of the founding family at t, zero otherwise;
H CEO t = one if the CEO is hired from outside the founding family at t, zero otherwise;
GROWTH t = sales growth rate at t; .
LOSS t = one if net income at t is negative, zero otherwise.
(−0.009) indicates that, on average, family firms report lower abnormal ac-
cruals equivalent to 11.13% of their pretax income. 18 Apparently, the lower
abnormal accruals reported by family firms have an economically material
effect on earnings using the standard 5% rule of thumb for materiality. As
18 In model 1, the coefficient on F FAM is −0.009. This suggests that, on average, family
t
firms report lower abnormal accruals (in dollar amount) by a magnitude of −0.009 ∗ (average
assets at t). The average total assets are $9,130 million and the average pretax income is $738
million. Therefore, family firms report lower abnormal accruals equivalent to 11.13% (−0.009 ∗
9,130/738 = −0.1113) of their pretax income.
640 D. WANG
such, family firms report both statistically and economically significant lower
abnormal accruals than nonfamily firms.
Consistent with the results in model 1, the coefficient on family owner-
ship (FAM OWN t ) in model 2 is −0.068 with a p-value < 0.001. The result
indicates that greater family ownership is associated with lower abnormal
accruals. Specifically, when family-owned common stock increases by 1%,
family firms report lower abnormal accruals by an amount equivalent to
0.84% of their pretax income. 19 In model 3, the coefficient on D CEO t is
−0.018 and is significant at p < 0.001. The coefficient on H CEO t is −0.009
with a p-value < 0.001. An F -test shows that the magnitude of the coefficient
on D CEO t is greater than that on H CEO t at p < 0.05. The coefficient on
F CEO t is insignificant at the conventional level. In sum, the results in table 3
show consistently that earnings of family firms are of higher quality.
The signs of coefficients on the control variables are generally consistent
with predictions and prior literature. Specifically, the coefficients on SIZE t
and INST t are negative and significant at p < 0.01 or lower, indicating that
large firms and firms with greater institutional monitoring have lower ab-
normal accruals. Firms with a greater growth rate (GROWTH t ) and firms
with negative income (LOSS t ) are associated with greater abnormal accru-
als. The signs of these coefficients are consistent with prior studies (Becker
et al. [1998], Chung and Kallapur [2003], Reynolds and Francis [2000]).
However, the coefficient on LEV t is negative and significant at p < 0.001 even
though it is predicted to be positive. 20 Frankel, Johnson, and Nelson [2002]
and Cheng and Warfield [2005] also report a negative relation between
leverage and abnormal accruals.
5.2. FOUNDING FAMILY OWNERSHIP AND EARNINGS INFORMATIVENESS

Equation (3) examines the differential earnings informativeness of family
firms and nonfamily firms from regressions of annual return on net income.
The results are presented in table 4. All models are significant at p < 0.001.
The adjusted R 2 values are about 0.14. In model 1, the coefficient on net
income (NI t ) is 5.382 and significant at p < 0.001. 21 The coefficient on the
test variable (NI t ∗ F FAM t ) is 0.804 with a p-value of 0.002. This suggests
19 The average total assets are $9,130 million, and the average pretax income is $738 mil-
lion. Therefore, family firms report lower abnormal accruals equivalent to 0.86% (−0.068% ∗
9,130/738 = −0.84%) of their pretax income if family ownership increases by 1%.
20 Dropping LEV from the regressions does not affect the signs and significance levels of
t
the test variables.
21 The magnitude is comparable with that in Fan and Wong [2002, table 6], who report an
earnings response coefficient (on NI t ) of 5.50 for Malaysia and 7.12 for South Korea. In the
U.S. context, the coefficient on NI t seems to be high compared with that reported by Francis,
Schipper, and Vincent [2005, table 3], who report a coefficient of 1.0146 on net income. In
this paper, however, the coefficient on NI t captures the earnings response coefficient for profit
firms of large S&P 500 companies. In a simple regression model, RET t = a +bNI t + cNI t ∗
F FAM t + error term, the coefficient is 1.192 on NI t and is 1.055 on NI t ∗ F FAM t (both signif-
icant at p < 0.001).
TABLE 4
Multivariate Analysis of Earnings Informativeness and Founding Family Ownership
Dependent Variable: RET t
Independent Expected
Variable Sign Estimate p-value Estimate p-value Estimate p-value
Intercept ? −0.001 0.986 −0.009 0.907 0.000 0.999
NI t + 5.382 0.000 5.510 0.000 4.968 0.000
NI t ∗ F FAM t ? 0.804 0.002
NI t ∗ FAM OWN t ? 4.891 0.004
NI t ∗ F CEO t ? 2.088 0.000
NI t ∗ D CEO t ? 0.006 0.987
NI t ∗ H CEO t ? 0.546 0.085
NI t ∗ SIZE t + −0.085 0.378 −0.068 0.487 −0.093 0.335
NI t ∗ LEV t − −0.855 0.221 −0.906 0.191 −0.603 0.381
NI t ∗ MB t + 0.204 0.000 0.200 0.000 0.196 0.000
NI t ∗ INST t + −0.601 0.285 −0.579 0.304 −0.503 0.373
NI t ∗ INSIDER t + 0.956 0.390 0.802 0.473 0.970 0.374
NI t ∗ AGE t ? −0.014 0.000 −0.015 0.000 −0.011 0.000
NI t ∗ LOSS t − −3.394 0.000 −3.387 0.000 −3.317 0.000
F -value 4.830 0.000 4.810 0.000 4.940 0.000
Adj. R 2 0.141 0.140 0.146
N 3,483 3,483 3,483
standard errors. Coefficients on 140 industry dummies based on three-digit Standard Industrial Classifica-
tion codes are not reported for brevity. Test variables and their statistics are indicated in bold typeface.
D CEO t = one if the CEO is a descendant of the founding family at t, 0 otherwise;
H CEO t = one if the CEO is hired from outside the founding family at t, 0 otherwise;
LOSS t = one if net income at t is negative, zero otherwise.
that, on average, the ERC of family firms is 14.94% (0.804/5.382 = 0.1494)

higher than that of nonfamily firms, and this is economically significant
using the 5% rule of thumb for materiality.
In model 2, the coefficient on the continuous family ownership variable
(NI t ∗ FAM OWN t ) is 4.891 with a p-value of 0.004, suggesting that greater
family ownership is associated with more informative earnings. If family
ownership increases by 1%, the ERC of family firms will be 0.89% (4.891%/
5.510 = 0.89%) higher than that of nonfamily firms. In model 3, the coef-
ficient on founder CEO (NI t ∗ F CEO t ) is 2.088 with a p-value < 0.001, and
the coefficient on outsider CEO (NI t ∗ H CEO t ) is 0.546 with a p-value of
0.085. The coefficient on descendant CEO (NI t ∗ D CEO t ) is not significant
642 D. WANG
at the conventional level. Overall, the results in models 1 through 3 provide

evidence that family firms report higher-quality earnings.
The signs of the coefficients on the control variables are, in general, con-
sistent with prior literature (Fan and Wong [2002], Francis, Schipper, and
Vincent [2005]). Firms with a higher market-to-book ratio (NI t ∗ MB t ) have
higher ERCs, and firms with losses (NI t ∗ LOSS t ) have lower ERCs. In addi-
tion, earnings of older firms (NI t ∗ AGE t ) tend to be less informative. The
coefficients on firm size (NI t ∗ SIZE t ), leverage (NI t ∗ LEV t ), institutional
ownership (NI t ∗ INST t ), and nonfamily insider ownership (NI t ∗ INSIDER t )
are insignificant at the conventional level.
5.3.
FAMILY OWNERSHIP AND PERSISTENCE OF TRANSITORY LOSS
COMPONENTS IN EARNINGS
A firm is considered to have greater financial reporting quality if transi-
tory loss components in earnings are less persistent into future periods than
transitory gain components (Ball and Shivakumar [2005a], Basu [1997]).
The results of the analysis of persistence of transitory losses are presented
in table 5. All three models are significant at p < 0.001. The adjusted R 2 val-
ues are 0.128 for model 1 and 0.134 for models 2 and 3. The coefficient on
NI t −1 ∗ DNI t −1 is predicted to be negative because transitory loss compo-
nents are less persistent than transitory gains (Ball and Shivakumar [2005a],
Basu [1997]). The test variables are F FAM t , FAM OWN t , F CEO t , D CEO t ,
and H CEO t , interacted with NI t −1 and DNI t −1 . In model 1, the coeffi-
cient on NI t −1 ∗ DNI t −1 ∗ F FAM t is −0.377 with a p-value of 0.018, which
indicates that transitory losses of family firms are less persistent into future
periods than those of nonfamily firms. The coefficient on NI t −1 ∗ DNI t −1
is −1.372 with a p-value of 0.006. This suggests that family firms’ transitory
losses are 27.48% (−0.377/−1.372 = 27.48%) less persistent than those of
nonfamily firms. As a result, the coefficient on NI t −1 ∗ DNI t −1 ∗ F FAM t
is also economically significant using the standard 5% rule of thumb for
materiality.
Consistent with the results in model 1, the coefficient on
NI t −1 ∗ DNI t −1 ∗ FAM OWN t in model 2 is −4.067 with a p-value <
0.001. The negative coefficient indicates that higher family ownership
is associated with less persistent transitory losses. Specifically, a 1% in-
crease in family ownership reduces the persistence of transitory losses of
family firms by 2.85% (−4.067%/−1.425 = 2.85%) relative to nonfamily
firms. In model 3, the coefficients on NI t −1 ∗ DNI t −1 ∗ F CEO t and
NI t −1 ∗ DNI t −1 ∗ D CEO t are insignificant at the conventional level. The
coefficient on NI t −1 ∗ DNI t −1 ∗ H CEO t is −0.513 with a p-value of 0.003.
Overall, these results indicate that family firms are more conservative in
reporting transitory losses than nonfamily firms. Therefore, earnings of
family firms are of higher quality.
The coefficient on NI t −1 ∗ DNI t −1 ∗ SIZE t is positive and significant
at p < 0.05, indicating that larger firms are less conservative in report-
ing transitory losses. The coefficients on NI t −1 ∗ DNI t −1 ∗ INST t and
NI t −1 ∗ DNI t −1 ∗ LEV t are negative and significant at p < 0.10, indicating
that firms with greater institutional monitoring and a higher leverage ratio
are associated with less persistent transitory losses.
In summary, the results in the abnormal accruals analysis, the earn-
ings informativeness analysis, and the analysis of persistence of transitory
losses show consistently that founding family ownership is associated with
greater earnings quality. The results support the alignment effect of family
TABLE 5
Multivariate Analysis of Persistence of Transitory Loss Components in Earnings and Founding
Family Ownership
Dependent Variable: ΔNI t
Exp.
Independent Variables Sign Estimate p-value Estimate p-value Estimate p-value
Intercept ? 0.036 0.041 0.030 0.090 0.029 0.114
DNI t −1 ? −0.019 0.367 −0.022 0.301 −0.023 0.297
NI t −1 0 0.898 0.001 0.918 0.000 0.920 0.001
NI t −1 ∗ DNI t −1 − −1.372 0.006 −1.425 0.003 −1.542 0.002
F FAM t ? −0.002 0.545
DNI t −1 ∗ F FAM t ? −0.005 0.420
NI t −1 ∗ F FAM t ? 0.223 0.022
ΔNI t −1 ∗ DΔNI t−1 ∗ F FAM t ? −0.377 0.018
FAM OWN t ? −0.012 0.472
DNI t −1 ∗ FAM OWN t ? −0.008 0.765
NI t −1 ∗ FAM OWN t ? 1.999 0.001
ΔNI t −1 ∗ DΔNI t −1 ∗ FAM OWN t ? −4.067 0.000
F CEO t ? 0.010 0.221
DNI t −1 ∗ F CEO t ? −0.000 0.983
NI t −1 ∗ F CEO t ? 0.208 0.238
ΔNI t −1 ∗ DΔNI t −1 ∗ F CEO t ? −0.041 0.885
D CEO t ? −0.001 0.846
DNI t −1 ∗ D CEO t ? −0.000 0.986
NI t −1 ∗ D CEO t ? 0.210 0.317
ΔNI t −1 ∗ DΔNI t −1 ∗ D CEO t ? −0.151 0.616
H CEO t ? −0.007 0.083
DNI t −1 ∗ H CEO t ? −0.008 0.217
NI t −1 ∗ H CEO t ? 0.164 0.106
ΔNI t −1 ∗ DΔNI t −1 ∗ H CEO t ? −0.513 0.003
SIZE t ? 0.002 0.301 0.002 0.186 0.002 0.167
DNI t −1 ∗ SIZE t ? 0.003 0.147 0.003 0.167 0.004 0.119
NI t −1 ∗ SIZE t ? −0.110 0.001 −0.116 0.000 −0.115 0.000
NI t −1 ∗ DNI t −1 ∗ SIZE t ? 0.194 0.000 0.200 0.000 0.210 0.000
INST t ? 0.002 0.774 0.003 0.671 0.002 0.728
DNI t −1 ∗ INST t ? −0.004 0.734 −0.003 0.825 −0.004 0.721
NI t −1 ∗ INST t ? 0.249 0.142 0.351 0.045 0.254 0.144
NI t −1 ∗ DNI t −1 ∗ INST t ? −0.560 0.065 −0.663 0.028 −0.564 0.063
INSIDER t ? −0.029 0.086 −0.024 0.148 −0.027 0.117
DNI t −1 ∗ INSIDER t ? 0.050 0.070 0.048 0.086 0.054 0.049
NI t −1 ∗ INSIDER t ? 0.822 0.126 0.682 0.201 0.850 0.115
NI t −1 ∗ DNI t −1 ∗ INSIDER t ? 0.490 0.572 0.564 0.522 0.607 0.473
LEV t ? −0.046 0.000 −0.046 0.000 −0.047 0.000
DNI t −1 ∗ LEV t ? −0.026 0.129 −0.023 0.181 −0.024 0.171
NI t −1 ∗ LEV t ? −0.261 0.219 −0.260 0.222 −0.230 0.284
NI t −1 ∗ DNI t −1 ∗ LEV t ? −0.665 0.060 −0.664 0.057 −0.670 0.058
AGE t ? −0.000 0.536 −0.000 0.701 −0.000 0.979
DNI t −1 ∗ AGE t ? 0.000 0.762 0.000 0.810 0.000 0.805
644 D. WANG
Dependent Variable: ΔNI t
Exp.
Independent Variables Sign Estimate p-value Estimate p-value Estimate p-value
NI t −1 ∗ AGE t ? −0.001 0.360 −0.001 0.221 −0.001 0.406
NI t −1 ∗ DNI t −1 ∗ AGE t ? −0.001 0.740 −0.000 0.978 −0.001 0.895
F -value 4.080 0.000 4.250 0.000 4.090 0.000
Adj. R 2 0.128 0.134 0.134
N 3,552 3,552 3,552
standard errors. Coefficients on 142 industry dummies based on three-digit Standard Industrial Classification
codes are not reported for brevity. Test variables and their statistics are indicated in bold typeface.
DNI t −1 = one if NI t −1 < 0, zero otherwise;
AGE t = firm age in years at t.
ownership on the supply of earnings quality, or alternatively, the entrench-

ment effect on the demand for earnings quality.
6. Additional Analysis
6.1. TIME-SERIES CORRELATION
The sample used in this study includes nine years of observations. If the
residuals in annual regressions are correlated, the coefficients on the test
and control variables are biased. To mitigate this concern, I adopt Fama-
MacBeth regressions and alternative methods of computing significance
levels (Barth [1994]; Healy, Kang, and Palepu [1987]).
For simplicity, table 6 only reports the coefficients on the test variables.
The coefficients on the control variables are comparable with those reported
in tables 3 through 5 and are therefore not reported. In Fama-MacBeth re-
gressions, the mean estimate is the average of the coefficients in annual
regressions, and the t-value is the t-statistic based on nine estimated coeffi-
cients. The results are reported in table 6. In the abnormal accruals analy-
sis, the Fama-MacBeth estimates of the coefficients on F FAM t , FAM OWN t ,
D CEO t , and H CEO t are negative and significant at p = 0.003 or lower,
comparable with those in table 3. The Z1 and Z2 statistics based on Barth
[1994] and Healy, Kang, and Palepu [1987] show comparable significance
levels. The results indicate that the time-series correlation is not a concern.
In the earnings informativeness analysis, the mean coefficient estimates
on the test variables are comparable with the pooled regression results
TABLE 6
Summary of the Estimates on Test Variables after Controlling for Potential Time-series Correlation Bias
Fama-MacBeth Regressions Z1 Z2
Mean Estimate t-value p-value Z1-value p-value Z2-value p-value
Test variables in abnormal accruals analysis
F FAM t −0.010 −6.257 0.000 −4.333 0.000 −6.774 0.000
FAM OWN t −0.081 −6.467 0.000 −6.196 0.000 −6.776 0.000
F CEO t −0.002 −0.795 0.449 −0.595 0.552 −0.671 0.502
D CEO t −0.018 −5.541 0.001 −4.453 0.000 −5.337 0.000
H CEO t −0.010 −4.252 0.003 −3.656 0.000 −4.109 0.000
Test variables in earnings informativeness analysis
NI t ∗ F FAM t 0.863 2.887 0.020 3.478 0.001 3.323 0.001
NI t ∗ FAM OWN t 4.027 2.268 0.053 2.345 0.019 2.255 0.024
NI t ∗ F CEO t 2.046 3.003 0.017 5.131 0.000 3.563 0.000
NI t ∗ D CEO t 0.030 0.102 0.921 −0.235 0.814 −0.255 0.799
NI t ∗ H CEO t 0.651 2.588 0.032 2.337 0.019 2.200 0.028
Test variables in analysis of persistence of transitory losses
NI t −1 ∗ DNI t −1 ∗ F FAM t −0.421 −3.504 0.008 −2.641 0.008 −3.398 0.001
NI t −1 ∗ DNI t −1 ∗ FAM OWN t −3.149 −2.452 0.039 −2.951 0.003 −2.282 0.023
NI t −1 ∗ DNI t −1 ∗ F CEO t −0.098 −0.337 0.745 −0.125 0.900 −0.111 0.912
NI t −1 ∗ DNI t −1 ∗ D CEO t −0.333 −0.542 0.603 −0.721 0.471 −0.574 0.566
NI t −1 ∗ DNI t −1 ∗ H CEO t −0.537 −2.937 0.019 −2.458 0.014 −3.272 0.001
Note that Z1 and Z2 statistics test whether the time-series mean t-statistics from yearly regressions is statistically different from zero (Barth [2004]; Healy, Kang, and Palepu
t t¯ √
[1987]). Z1 = √1N Nj=1 √k /(kj −2) , Z2 = s tde v(t)/ N−1
, where t is t-statistic and k is the degrees of freedom for year j, and N is the number of years.
j j
FOUNDING FAMILY OWNERSHIP
645
DNI t −1 = one if NI t −1 < 0, zero otherwise.
646 D. WANG
reported in table 4. The Fama-MacBeth t-statistics are slightly smaller than

those reported in table 4 but still significant at p < 0.10 or lower. The infer-
ence that family ownership is associated with greater earnings informative-
ness is consistent with the pooled regression results. The corresponding Z1
and Z2 statistics show significance levels comparable with those in table 4.
In the analysis of persistence of transitory losses, the Fama-MacBeth co-
efficients on NI t −1 ∗ DNI t −1 interacted with F FAM t , FAM OWN t , and
H CEO t are negative and significant at p < 0.05. The Z1 and Z2 statistics
show similar results. These results indicate that family firms are associated
with less persistent transitory losses than nonfamily firms, consistent with
the results from the pooled regressions in table 5.
6.2. NONLINEARITY AND FOUNDING FAMILY OWNERSHIP

Anderson and Reeb [2003a] document an inverted U–shaped relation
between founding family ownership and firm performance. I also examine
whether the relation between earnings quality and founding family own-
ership is nonlinear. I investigate this issue in two ways. First, I create two
dummy variables: strong family ownership (S FAM t ) and weak family own-
ership (W FAM t ). S FAM t equals one if the percentage of stock owned by
family members is greater than or equal to the median of founding family
ownership (about 5%), and zero otherwise. Similarly, W FAM t equals one if
the percentage of stock owned by family members is lower than the median
of founding family ownership, and zero otherwise. The default comparison
group is nonfamily firms. Second, I add the square of founding family own-
ership (FAM OWN t )2 to model 2 in tables 3 through 5 to test whether there
is an inverted U–shaped relation between family ownership and earnings
quality.
The results are presented in table 7. Panel A compares the strong and weak
family ownership (S FAM t vs. W FAM t ). In the abnormal accruals analysis,
the coefficient on S FAM t is −0.017 with a p-value < 0.001. In compari-
son, the coefficient on weak family ownership (W FAM t ) is −0.001, but it is
not statistically different from zero. This indicates that the relation between
family ownership and abnormal accruals may not be linear. In the earn-
ings informativeness analysis, the coefficient on NI t ∗ S FAM t is 1.059 with a
p-value of 0.002. The coefficient on NI t ∗ W FAM t is 0.585 with a p-value of
0.051. An F -test shows that the coefficient on NI t ∗ S FAM t is not significantly
higher than that on NI t ∗ W FAM t at p < 0.10. In the analysis of persistence
of transitory losses, the coefficient on NI t −1 ∗ DNI t −1 ∗ S FAM t is −0.663
and significant at p < 0.001. The coefficient on NI t −1 ∗ DNI t −1 ∗ W FAM t
is −0.074 but insignificant at the conventional level. Overall, the results in
table 7, panel A provide evidence that the relation between founding family
ownership and earnings quality may be nonlinear.
Panel B of table 7 presents the results of testing whether there is an in-
verted U–shaped relation between founding family ownership and earnings
quality. In the abnormal accruals analysis, the coefficient on FAM OWN t is
−0.145 and on (FAM OWN t )2 is 0.215 (both are significant at p < 0.001).
TABLE 7
Nonlinear Relation between Earnings Quality and Founding Family Ownership
Panel A: Strong versus weak founding family ownership and earnings quality
Abnormal Accruals Earnings Informativeness Persistence of Transitory Losses
Dependent Variable: ABS ACC t Dependent Variable: RET t Dependent Variable: NI t
Independent Variable Estimate p-value Independent Variable Estimate p-value Independent Variable Estimate p-value
Intercept 0.140 0.000 Intercept −0.003 0.968 Intercept 0.035 0.048
S FAM t −0.017 0.000 NI t 5.335 0.000 DNI t −1 −0.021 0.326
W FAM t −0.001 0.622 NI t ∗ S FAM t 1.059 0.002 NI t −1 0.861 0.001
SIZE t −0.006 0.000 NI t ∗ W FAM t 0.585 0.051 NI t −1 ∗ DNI t −1 −1.315 0.009
ROA t 0.055 0.142 NI t ∗ SIZE t −0.075 0.437 S FAM t −0.005 0.302
LEV t −0.030 0.000 NI t ∗ LEV t −0.852 0.219 DNI t −1 ∗ S FAM t −0.001 0.853
GROWTH t 0.028 0.000 NI t ∗ MB t 0.205 0.000 NI t −1 ∗ S FAM t 0.398 0.000
INST t −0.018 0.001 NI t ∗ INST t −0.610 0.274 ΔNI t −1 ∗ DΔNI t −1 ∗ S FAM t −0.663 0.000
INSIDER t 0.003 0.863 NI t ∗ INSIDER t 0.973 0.383 W FAM t 0.002 0.678
AGE t −0.0002 0.000 NI t ∗ AGE t −0.015 0.000 DNI t −1 ∗ W FAM t −0.010 0.149
LOSS t 0.047 0.000 NI t ∗ LOSS t −3.415 0.000 NI t −1 ∗ W FAM t −0.023 0.859
ΔNI t−1 ∗ DΔNI t −1 ∗ W FAM t −0.074 0.707
SIZE t 0.002 0.246
DNI t −1 ∗ SIZE t 0.003 0.156
NI t −1 ∗ SIZE t −0.115 0.000
NI t −1 ∗ DNI t −1 ∗ SIZE t 0.196 0.000
INST t 0.001 0.905
DNI t −1 ∗ INST t −0.002 0.899
NI t −1 ∗ INST t 0.348 0.047
NI t −1 ∗ DNI t −1 ∗ INST t −0.659 0.030
INSIDER t −0.028 0.094
DNI t −1 ∗ INSIDER t 0.049 0.078
NI t −1 ∗ INSIDER t 0.786 0.141
NI t −1 ∗ DNI t −1 ∗ INSIDER t 0.472 0.590
LEV t −0.046 0.000
DNI t −1 ∗ LEV t −0.025 0.150
NI t −1 ∗ LEV t −0.239 0.266
NI t −1 ∗ DNI t −1 ∗ LEV t −0.695 0.051
AGE t −0.000 0.443
DNI t −1 ∗ AGE t 0.000 0.735
NI t −1 ∗ AGE t −0.001 0.363

NI t −1 ∗ DNI t −1 ∗ AGE t −0.001 0.760
F -value 54.93 0.000 F -value 4.810 0.000 F -value 4.140 0.000
Adj. R 2 0.135 Adj. R 2 0.141 Adj. R 2 0.133
647
N 3,456 N 3,483 N 3,552
648
Panel B: Inverted U–shaped relation between earnings quality and founding family ownership
Abnormal Accruals Earnings Informativeness Persistence of Transitory Losses
Dependent Variable: ABS ACC t Dependent Variable: RET t Dependent Variable: NI t
D. WANG
Independent Variable Estimate p-value Independent Variable Estimate p-value Independent Variable Estimate p-value
Intercept 0.141 0.000 Intercept −0.007 0.927 Intercept 0.031 0.088
FAM OWN t −0.145 0.000 NI t 5.395 0.000 DNI t −1 −0.021 0.325
(FAM OWN t )2 0.215 0.000 NI t ∗ FAM OWN t 11.50 0.001 NI t −1 0.897 0.000
SIZE t −0.006 0.000 NI t ∗ (FAM OWN t )2 −19.87 0.009 NI t −1 ∗ DNI t −1 −1.372 0.005
ROA t 0.056 0.133 NI t ∗ SIZE t −0.063 0.513 FAM OWN t −0.025 0.498
LEV t −0.030 0.000 NI t ∗ LEV t −0.908 0.184 DNI t −1 ∗ FAM OWN t −0.03 0.634
GROWTH t 0.029 0.000 NI t ∗ MB t 0.202 0.000 NI t −1 ∗ FAM OWN t 3.917 0.001
INST t −0.019 0.001 NI t ∗ INST t −0.608 0.276 ΔNI t −1 ∗ DΔNI t −1 ∗ FAM OWN t −7.317 0.001
INSIDER t 0.003 0.839 NI t ∗ INSIDER t 0.850 0.445 (FAM OWN t )2 0.059 0.371
AGE t −0.0002 0.000 NI t ∗ AGE t −0.015 0.000 DNI t −1 ∗ (FAM OWN t )2 0.056 0.653
LOSS t 0.046 0.000 NI t ∗ LOSS t −3.401 0.000 NI t −1 ∗ (FAM OWN t )2 −7.097 0.038
ΔNI t −1 ∗ DΔNI t −1 ∗ (FAM OWN t )2 12.46 0.041
SIZE t 0.002 0.176
DNI t −1 ∗ SIZE t 0.003 0.167
NI t −1 ∗ SIZE t −0.118 0.000
NI t −1 ∗ DNI t −1 ∗ SIZE t 0.202 0.000
INST t 0.003 0.712
DNI t −1 ∗ INST t −0.003 0.813
NI t −1 ∗ INST t 0.354 0.039
NI t −1 ∗ DNI t −1 ∗ INST t −0.675 0.026
INSIDER t −0.025 0.137
DNI t −1 ∗ INSIDER t 0.048 0.083
NI t −1 ∗ INSIDER t 0.727 0.178
NI t −1 ∗ DNI t −1 ∗ INSIDER t 0.506 0.567
LEV t −0.046 0.000
DNI t −1 ∗ LEV t −0.024 0.159
NI t −1 ∗ LEV t −0.262 0.214
NI t −1 ∗ DNI t −1 ∗ LEV t −0.674 0.053
AGE t −0.000 0.628
DNI t −1 ∗ AGE t 0.000 0.780
NI t −1 ∗ AGE t −0.001 0.310
NI t −1 ∗ DNI t −1 ∗ AGE t −0.000 0.821
F -value 55.42 0.000 F -value 4.820 0.000 F -value 4.190 0.000
Adj. R2 0.136 Adj. R 2 0.141 Adj. R 2 0.135
N 3,456 N 3,483 N 3,552
Note that p-values are two-tailed values computed using White’s [1980] heteroskedasticity-corrected standard errors. In the analyses of earnings informativeness and persistence
of transitory losses, coefficients on industry dummies based on three-digit Standard Industrial Classification codes are not reported for brevity. Test variables and their statistics are
indicated in bold typeface.
ABS ACC t = absolute value of abnormal accruals at t, estimated based on the Dechow and Dichev [2002] model as modified by Ball and Shivakumar [2005b];
S FAM t = one if family ownership at t is greater than or equal to the median, zero otherwise;
W FAM t = one if family ownership at t is less than the median, zero otherwise;
GROWTH t = sales growth rate at t;
LOSS t = one if net income at t is negative, zero otherwise;
DNI t −1 = one if NI t −1 < 0, zero otherwise.
FAM OWN t = percentage of common stock owned by founding family members at t.
649
650 D. WANG
This means that there is a U-shaped relation between family ownership

and abnormal accruals. When founding family ownership is 33.72%, the
abnormal accruals are the lowest. Earnings are of lower quality with higher
abnormal accruals. Therefore, the relation between family ownership and
earnings quality is an inverted U shape with an inflection point of 33.72%.
Firms with 33.72% family ownership report lower abnormal accruals, equiv-
alent to 30.24% of their pretax income, relative to nonfamily firms. 22 In
addition, the results also suggest that family firms report lower abnormal
accruals than nonfamily firms up to 67.44% (0.145/0.215 = 67.44%) family
ownership. In other words, firms with family ownership >67.44% will report
higher abnormal accruals than nonfamily firms.
In the earnings informativeness analysis, the coefficient on
NI t ∗ FAM OWN t is 11.50 and the coefficient on NI t ∗(FAM OWN t )2 is
−19.87, and both are significant at p < 0.01 or lower. The results
indicate that there is an inverted U–shaped relation between family
ownership and earnings informativeness with an inflection point at 28.94%
family ownership. When firms have family ownership at the inflection
point (28.94%), the ERC of family firms is 30.84% ((11.50 ∗ 28.94% −
19.87 ∗ (28.94%)2 )/5.395 = 30.84%) higher than that of nonfamily firms.
Moreover, family firms have a higher ERC than nonfamily firms up to
57.88% (11.50/19.87 = 57.88%) family ownership, implying that firms with
family ownership greater than 57.88% will begin to have lower earnings
informativeness than nonfamily firms.
In the analysis of persistence of transitory losses, the coefficient on
NI t −1 ∗ DNI t −1 ∗ FAM OWN t is −7.317 with a p-value of 0.001 and the
coefficient on NI t −1 ∗ DNI t −1 ∗(FAM OWN t )2 is 12.46 with a p-value of
0.041. The results can be interpreted as a U-shaped relation between family
ownership and persistence of transitory loss components in earnings. Be-
cause lower persistence of transitory losses indicates better earnings quality
(i.e., more conservative earnings), the results suggest an inverted U–shaped
relation between family ownership and earnings quality with an inflection
point at 29.36%. Transitory losses of firms with 29.36% family ownership will
be 78.29% ((−7.317 ∗ 29.36% + 12.46 ∗ (29.36%)2 )/−1.372 = 78.29%) less
persistent into future periods than those of nonfamily firms. The results also
suggest that transitory losses of family firms are less persistent than those of
nonfamily firms up to 58.72% (7.317/12.46 = 58.72%) family ownership,
indicating that firms with family ownership greater than 58.72% will begin
to report transitory losses less conservatively than nonfamily firms.
Overall, based on the results in table 7, I conclude that the relation be-
tween family ownership and earnings quality is nonlinear. Specifically, there
is an inverted U–shaped relation between family ownership and earnings
quality. Family firms report earnings of higher quality than nonfamily firms
22 The average total assets are $9,130 million, and the average pretax income is $738 million.
Therefore, firms with family ownership of 33.72% report lower abnormal accruals equivalent to
30.24% ((−0.145∗33.72% + 0.215∗(33.72%)2 )∗9,130/738 = −0.3024) of their pretax income.
up to certain levels of family ownership (67.44% in the abnormal accruals

analysis, 57.88% in the earnings informativeness analysis, and 58.72% in
the analysis of persistence of transitory losses). Almost all family firms used
in this study have family ownership less than 58%, and this is consistent
with the empirical findings that, on average, family firms report earnings of
higher quality than nonfamily firms. 23 When family ownership exceeds cer-
tain levels (about 58–67%), however, family firms start to report earnings of
lower quality than nonfamily firms. This indicates that extremely high fam-
ily ownership will have an entrenchment effect on the supply of earnings
quality, or alternatively, an alignment effect on the demand for earnings
quality.
6.3. ROBUSTNESS TESTS

In the analysis of persistence of transitory losses, I adopt Basu’s [1997]
serial dependence model as modified by Ball and Shivakumar [2005a] to
investigate the relation between family ownership and persistence of transi-
tory losses. However, Basu’s [1997] serial dependence model has two poten-
tial limitations: it cannot distinguish the transitory components in earnings
from random accruals errors, and it cannot identify whether the transi-
tory loss components in earnings are recognized in a timely manner (Ball
and Shivakumar [2005a]). As an attempt to mitigate the concern, I investi-
gate the association between contemporaneous short-window stock returns
surrounding the quarterly earnings announcement date and the transitory
components in quarterly earnings. The key argument of the test is that stock
returns should be less sensitive to transitory losses that have a lower level
of persistence (Basu [1997]). Empirically, I replace the change in earnings
at t (NI t ) in equation (4) with the three-day abnormal returns surround-
ing the quarterly earnings announcement date. In addition, the change in
earnings at t−1 (NI t −1 ) is replaced by the change in quarterly earnings
(earnings in quarter q minus earnings in quarter q−4, scaled by total assets
at q−4). 24 Accordingly, DNI t −1 is substituted by a dummy variable that
equals one if the change in quarterly earnings is negative, and zero other-
wise. All other variables are defined the same as in equation (4), except that
corresponding quarterly data are used instead of annual data. Untabulated
results indicate that stock returns are less sensitive to transitory losses for
family firms relative to nonfamily firms, indicating that family firms report
transitory loss components in quarterly earnings more conservatively than
nonfamily firms. 25
23 Only three firm-year observations have family ownership greater than 58% in this study.
24 The results are robust to using the change in quarterly earnings per share (EPS in quarter
q minus EPS in quarter q−4, scaled by stock price at q−4) as a proxy for transitory components
in quarterly earnings.
25 This test still suffers from possible noise in measuring transitory loss components. As such,
the results may be interpreted as evidence that negative earnings changes for family firms are
noisier than negative earnings changes for nonfamily firms.
652 D. WANG
Executive compensation, such as bonuses and stock options, can affect

the quality of financial reporting (Aboody and Kasznik [2000], Bowen,
Rajgopal, and Venkatachalam [2003]). Therefore, I control for top execu-
tive bonuses and stock options because they may affect the incentives of top
executives to exercise accounting discretion when accounting earnings are
an integral part of executive compensation contracts (Bowen, Rajgopal, and
Venkatachalam [2003]). Top executive bonuses are defined as the bonuses
paid to top executives divided by the sum of salaries, bonuses, and other
compensation. Stock options are defined as the in-the-money exercisable
options divided by the sum of the in-the-money exercisable options and
unvested options. The bonuses and stock options are obtained from the Ex-
ecutive Compensation Database provided by Compustat. The coefficients
on the test variables are comparable with those reported in tables 3 through
5, and the significance levels are not affected by the inclusion of bonuses
and stock options as additional controls.
Blockholder ownership, measured by equity owned by shareholders who
hold 5% or more of the outstanding stock, can serve as a gauge of share-
holder monitoring. Therefore, as a sensitivity analysis, I also control for
blockholder ownership by nonfamily members and institutions. However,
the blockholder ownership provided by Compact Disclosure may contain
measurement errors. 26 Thus, I also use 5% (or more) institutional own-
ership to proxy for blockholder ownership. The coefficients and statistical
significance of the test variables are comparable with those reported in tables
3 through 5 after controlling for the proxies of blockholder ownership.
Audit quality affects the quality of financial reporting. Because the sample
used in this study is from the S&P 500, more than 99% of the firm-year
observations are audited by Big 5 auditors. Specifically, about 98.5% of family
firm-year observations and 99.5% of nonfamily firm-year observations are
audited by Big 5 auditors. To test whether Big 5 auditors affect the results,
I create a dummy variable (BIG5 t ) that equals one if a firm is audited by a
Big 5 auditor, and zero otherwise. 27 Incorporating BIG5 t into the regression
models does not change the signs and significance levels of the test variables
in tables 3 through 5. The coefficients on BIG5 t in the abnormal accruals
analysis, on NI t ∗ BIG5 t in the earnings informativeness analysis, and on
NI t −1 ∗ DNI t −1 ∗ BIG5 t in the analysis of persistence of transitory losses
are not statistically significant at the conventional level.
26 For example, Heflin and Shaw [2000] report an average of 12.3% blockholder owner-
ship for a sample of 173 firms (of 260 firms) that have blockholders, which is much smaller
than 26.9% of blockholder ownership for the sample used in the abnormal accruals analysis
based on the blockholder ownership data from Compact Disclosure.
27 Only 23 firm-year observations in the abnormal accruals analysis, 33 observations in the
earnings informativeness analysis, and 33 observations in the analysis of persistence of transitory

losses are audited by non–Big 5 auditors. The mean of BIG5 t for family (nonfamily) firms is
0.990 (0.991) in the abnormal accruals analysis, is 0.984 (0.995) in the earnings informativeness
analysis, and is 0.984 (0.995) in the analysis of persistence of transitory losses.
7. Summary and Conclusions

This study examines the potential impact of founding family ownership
on earnings quality. The contracting-based theory of accounting states that
managers either efficiently report accounting earnings to maximize a firm’s
value for all shareholders or they opportunistically manage earnings for
private gains at the cost of other shareholders (Christie and Zimmerman
[1994]). Existing theories predict that founding family ownership can af-
fect the demand and supply of earnings quality in two competing ways:
the entrenchment effect and the alignment effect. The entrenchment ef-
fect predicts that founding families expropriate wealth from other investors
through the supply of lower earnings quality. However, the supply of lower
earnings quality may be attenuated by the demand for higher earnings qual-
ity if users of financial statements perceive that family firms have inferior
corporate governance. In contrast, the alignment effect predicts that found-
ing family ownership aligns the interests of founding families with those
of other investors and is thus associated with higher-quality earnings. The
supply of higher earnings quality, however, may be mitigated by the lower
demand for higher earnings quality if users of financial statements take it
for granted that family firms have stronger corporate governance.
Using data from the S&P 500 companies during the period 1994–2002, I
document evidence that, on average, founding family ownership is associ-
ated with higher earnings quality, a finding consistent with the alignment
effect of family ownership on the supply of earnings quality, or alternatively,
the entrenchment effect on the demand for earnings quality. The results
are robust to alternative definitions of founding family ownership (a bi-
nary variable, a continuous variable, and the CEO attributes of family firms)
and three measures of earnings quality (abnormal accruals, earnings in-
formativeness, and persistence of transitory loss components in earnings).
In addition, the empirical results suggest that the relation between family
ownership and earnings quality is nonlinear.
There are three major limitations in this study. First, it is unclear whether
higher earnings quality is a result of the demand for greater earnings qual-
ity from family firms or a result of the supply of greater earnings quality
by family firms. Unfortunately, there is no way to identify and separate the
two effects in my study. Second, like other studies examining the conse-
quences of ownership structure (Fan and Wong [2002], Francis, Schipper,
and Vincent [2005], Warfield, Wild, and Wild [1995]), potential endogene-
ity is a concern. Founding families may choose to leave by selling their stock
if family firms have potential problems. Therefore, this study is not intended
to draw a causal relation between founding family ownership and earnings
quality. Rather, it documents an empirical association between a common
ownership structure and characteristics of accounting earnings. Third, the
inferences apply only to large publicly traded firms in the United States. Out-
side the United States, where minority investors’ legal protection is weaker,
the findings based on this study may not hold. Given that there may be a
654 D. WANG
greater percentage of founding family firms in smaller publicly-traded firms

than those included in the S&P 500, investigation of smaller publicly-traded
firms offers a potential avenue for future research.
Overall, this study contributes to the literature on ownership structure,
corporate governance, and earnings quality. Warfield, Wild, and Wild [1995]
document a positive relation between insider ownership and earnings qual-
ity. More recent studies (e.g. Agrawal and Chadha [2005], Bowen, Rajgopal,
and Venkatachalam [2003], Fan and Wong [2002], Francis, Schipper, and
Vincent [2005], Klein [2002], Srinivasan [2005]) have documented the re-
lations between various corporate governance mechanisms and financial
reporting. This study adds to this stream of research by focusing on found-
ing family ownership, an ownership structure that is common in the United
States and dominant around the world.
REFERENCES
ABOODY, D., AND R. KASZNIK. “CEO Stock Option Awards and the Timing of Corporate Volun-
tary Disclosures.” Journal of Accounting Economics 29 (2000): 73–100.
AGRAWAL, A., AND S. CHADHA. “Corporate Governance and Accounting Scandals.” Journal of
Law and Economics 48 (2005): 371–406.
ANDERSON, R. C.; S. A. MANSI; AND D. M. REEB. “Founding Family Ownership and the Agency
Cost of Debt.” Journal of Financial Economics 68 (2003): 263–85.
ANDERSON, R. C., AND D. M. REEB. “Founding-Family Ownership and Firm Performance: Evi-
dence from the S&P 500.” The Journal of Finance 58 (2003a): 1301–28.
ANDERSON, R. C., AND D. M. REEB. “Who Monitors the Family?” Working paper, American
University and University of Alabama, 2003b.
ANONYMOUS. “In Praise of the Family Firm.” The Economist 338/7956 (1996): 16.
ANONYMOUS. “Some Facts on Family-owned Firms” National Underwriter 102/40 (1998): 10.
BALL, R.; S. P. KOTHARI; AND A. ROBIN. “The Effect of International Institutional Factors on
Properties of Accounting Earnings.” Journal of Accounting & Economics 29 (2000): 1–51.
BALL, R.; A. ROBIN; AND J. S. WU. “Accounting Standards, the Institutional Environment and Is-
suer Incentives: Effect on Timely Loss Recognition in China.” Asia Pacific Journal of Accounting
and Economics 7 (2000): 71–96.
BALL, R.; A ROBIN; AND J. S. WU. “Incentives Versus Standards: Properties of Accounting Income
in Four East Asian Countries.” Journal of Accounting & Economics 36 (2003): 235–70.
BALL, R., AND L. SHIVAKUMAR. “Earnings Quality in U.K. Private Firms: Comparative Loss Recog-
nition.” Journal of Accounting & Economics 38 (2005a): 83–128.
BALL, R., AND L. SHIVAKUMAR. “The Role of Accruals in Asymmetrically Timely Gain and Loss
Recognition.” Working paper, University of Chicago and London Business School, 2005b.
BARTH, M. “Fair Value Accounting: Evidence from Investment Securities and Market Valuation
of Banks.” The Accounting Review 69 (1994): 1–25.
BASU, S. 1997. “The Conservatism Principle and Asymmetric Timeliness of Earnings.” Journal
of Accounting & Economics 24 (1997): 3–37.
BECKER, C. L.; M. L. DEFOND; J. JIAMBALVO; AND K. R. SUBRAMANYAM. “The Effect of Audit
Quality on Earnings Management.” Contemporary Accounting Research 15 (1998): 1–24.
BOWEN, R. M.; S. RAJGOPAL; AND M. VENKATACHALAM. “Accounting Discretion, Corporate Gov-
ernance and Firm Performance.” Working paper, University of Washington and Duke Uni-
versity, 2003.
BURKART, M.; F. PANUNZI; AND A. SHLEIFER. “Family Firms.” The Journal of Finance 58 (2003):
2167–202.
BUSHMAN, R. M., AND A. SMITH. “Financial Accounting Information and Corporate Gover-
nance.” Journal of Accounting & Economics 32 (2001): 237–351.
CHENG, Q. AND T. D. WARFIELD. “Equity Incentives and Earnings Management.” The Accounting
Review 80 (2005): 441–76.
CHIRON CORP. “Proxy Statement Pursuant to Section 14(a) of the Securities Exchange Act of
1934.” May 16, 2002.
CHRISTIE, A., AND J. ZIMMERMAN. “Efficient and Opportunistic Choices of Accounting Proce-
dures: Corporate Control Contests.” The Accounting Review 69 (1994): 539–66.
CHUNG, H., AND S. KALLAPUR. “Client Importance, Nonaudit Services, and Abnormal Accruals.”
The Accounting Review 78 (2003): 931–55.
DEANGELO, H., AND L. DEANGELO. “Controlling Stockholders and the Disciplinary Role of
Corporate Payout Policy: A Study of the Times Mirror Company.” Journal of Financial Economics
56 (2000): 153–207.
DECHOW, P., AND I. DICHEV. “The Quality of Accruals and Earnings: The Role of Accrual Esti-
mation Errors.” The Accounting Review 77 (2002): 35–59.
DECHOW, P. M.; R. G. SLOAN; AND A. P. SWEENEY. “Detecting Earnings Management.” The Ac-
counting Review 70 (1995): 193–225.
DEFOND, M. L., AND J. JIAMBALVO. “Debt Covenant Violation and Manipulation of Accruals.”
Journal of Accounting & Economics 17 (1994): 145–76.
DEMSETZ, H., AND K. LEHN. “The Structure of Corporate Ownership: Causes and Conse-
quences.” Journal of Political Economy 93 (1985): 1155–77.
FACCIO, M., AND L. LANG. “The Ultimate Ownership of Western European Corporations.”
Journal of Financial Economics 66 (2002): 365–95.
FAMA, E., AND M. JENSEN. “Separation of Ownership and Control.” Journal of Law and Economics
26 (1983): 301–25.
FAN, J., AND T. J. WONG. “Corporate Ownership Structure and the Informativeness of Accounting
Earnings in East Asia.” Journal of Accounting & Economics 33 (2002): 401–25.
FRANCIS, J.; R. LAFOND; P. OLSSON; AND K. SCHIPPER. “The Market Pricing of Accrual Quality.”
FRANCIS, J.; K. SCHIPPER; AND L. VINCENT. “Earnings and Dividends Informativeness When Cash
Flow Rights Are Separated from Voting Rights.” Journal of Accounting & Economics 39 (2005):
329–60.
FRANKEL, F.; M. F. JOHNSON; AND K. K. NELSON. “The Relation Between Auditors’ Fees
for Nonaudit Services and Earnings Management.” The Accounting Review 77 (2002): 71–
105.
GIGLER, F. B., AND T. HEMMER. “Conservatism, Optimal Disclosure Policy and the Timeliness
of Financial Reports.” The Accounting Review 76 (2001): 473–93.
GROSSMAN, S. J., AND O. D. HART. 1986. “The Costs and Benefits of Ownership: A Theory of
Vertical and Lateral Integration.” The Journal of Political Economy 94 (1986): 691–719.
HEALY, P.; S. KANG; AND K. PALEPU. “The Effect of Accounting Procedure Changes on CEOs’
Cash Salary and Bonus Compensation.” Journal of Accounting & Economics 9 (1987): 7–34.
HEALY, P. M., AND R. S. KAPLAN. “The Effect of Bonus Schemes on Accounting Decisions.”
HEFLIN, F., AND K. W. SHAW. “Blockholder Ownership and Market Liquidity.” Journal of Financial
and Quantitative Analysis 35 (2000): 621–33.
JENSEN, M. C., AND W. H. MECKLING. “Theory of the Firm: Managerial Behavior, Agency Costs
and Ownership Structure.” Journal of Financial Economics 3 (1976): 305–60.
JONES, J. J. “Earnings Management During Import Relief Investigations.” Journal of Accounting
Research 29 (1991): 193–228.
KLEIN, A. “Audit Committee, Board of Director Characteristics, and Earnings Management.”
LA PORTA, R.; F. LOPEZ-DE-SILANES; A SHELIFER; AND R. VISHNY. “Law and Finance.” Journal of
Political Economy 106 (1998): 1113–55.
LEUZ, C.; D. NANDA; AND P. D. WYSOCKI. “Earnings Management and Investor Protection: An
International Comparison.” Journal of Financial Economics 69 (2003): 505–27.
LEV, B. “Corporate Earnings: Facts and Fiction.” Journal of Economic Perspectives 17 (2003): 27–50.
MORCK, R.; A. SHLEIFER; AND R. VISHNY. “Management Ownership and Market Valuation: An
Empirical Analysis.” Journal of Financial Economics 20 (1988): 293–315.
REYNOLDS, J. K., AND J. R. FRANCIS. “Does Size Matter? The Influence of Large Clients on Office-
level Auditor Reporting Decisions.” Journal of Accounting & Economics 30 (2000): 375–400.
656 D. WANG
SECURITIES AND EXCHANGE COMMISSION (SEC). “SEC Announces Fraud Charges Against For-
mer Rite Aid Senior Management.” Web site, http://www.sec.gov/news/press/2002-92.htm,
2002.
SHLEIFER, A., AND R. VISHNY. “Large Shareholders and Corporate Control.” Journal of Political
Economy 94 (1986): 461–89.
SHLEIFER, A., AND R. VISHNY. “A Survey of Corporate Governance.” Journal of Finance 52 (1997):
737–83.
SRINIVASAN, S. “Consequences of Financial Reporting Failure for Outside Directors: Evidence
from Accounting Restatements and Audit Committee Members.” Journal of Accounting Re-
search 43 (2005): 291–334.
WARFIELD, T. D.; J. J. WILD; AND K. L. WILD. “Managerial Ownership, Accounting Choices, and
Informativeness of Earnings.” Journal of Accounting & Economics 20 (1995): 61–92.
WATTS, R., AND J. ZIMMERMAN. Positive Accounting Theory. Englewood Cliffs, NJ: Prentice-Hall,
1986.
WEBER, J.; L. LAVELLE; T. LOWRY; W. ZELLNER; AND A. BARRENT. “Family, Inc.” BusinessWeek
(November 10, 2003): 100–14.
WHITE, H. “A Heteroskedasticity-consistent Covariance Matrix Estimator and a Direct Test for
Heteroskedasticity.” Econometrica 48 (1980): 817–38.

J of Accounting Research - 2006 - WANG - Founding Family Ownership and Earnings Quality

Uploaded by

Copyright:

Available Formats

You might also like

J of Accounting Research - 2006 - WANG - Founding Family Ownership and Earnings Quality

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

J of Accounting Research - 2006 - WANG - Founding Family Ownership and Earnings Quality

Uploaded by

Copyright:

Available Formats

DOI: 10.1111/j.1475-679X.2006.00213.

Founding Family Ownership

Received 1 January 2005; accepted 18 November 2005

∗ University of Nebraska–Lincoln. This paper is partly based on my dissertation completed at

mand and supply of earnings quality.

stake in the firm than nonfamily professional executives. Conversely, man-

This study adds to a growing body of research on earnings quality

2. Motivation and Theoretical Development

Francis, Schipper, and Vincent [2005] document lower earnings response

2.1. FOUNDING FAMILY OWNERSHIP AND THE ENTRENCHMENT EFFECT

2.1.1. Supply of Earnings Quality. Prior literature largely focuses on the

founding families (Anonymous [1996]). The findings of Fan and Wong

2.2. FOUNDING FAMILY OWNERSHIP AND THE ALIGNMENT EFFECT

family firms may report earnings of higher quality relative to nonfamily

H1: Earnings quality is systematically related to founding family

The directional relation between founding family ownership and earnings

Sweeney [1995]) for estimation of abnormal accruals.

CF t −1 = operating cash flows at t−1, scaled by average total assets at t;

+ δ4 LEV t + δ5 GROWTH t + δ6 INST t + δ7 INSIDER t

+ δ8 AGE t + δ9 LOSS t + e t (2)

ABS ACC t = absolute value of abnormal accruals at t;

3.3.2. Earnings Informativeness. The second proxy of earnings quality is

+ Industry Fixed Effects + e t (3)

weighted or value-weighted market-adjusted 12-month cumulative returns and to using net

zero. A positive (negative) estimate on β 2 will be evidence that family firms

3.3.3. Persistence of Transitory Loss Components in Earnings. Basu [1997]

index constituents can be identified.

the Hoovers database or from the individual firm’s website. Institutional

Panel B: Sample selection

insider ownership, and other control variables. 14 In addition, to avoid outlier

family firm-year observations. On average, family members own 10.35% of

Panel B: Earnings informativeness sample (N = 3,483)

nonfamily insider ownership (INSIDER t ). In comparison, family ownership

5.2. FOUNDING FAMILY OWNERSHIP AND EARNINGS INFORMATIVENESS

that, on average, the ERC of family firms is 14.94% (0.804/5.382 = 0.1494)

at the conventional level. Overall, the results in models 1 through 3 provide

ownership on the supply of earnings quality, or alternatively, the entrench-

DNI t −1 = one if NI t −1 < 0, zero otherwise.

reported in table 4. The Fama-MacBeth t-statistics are slightly smaller than

6.2. NONLINEARITY AND FOUNDING FAMILY OWNERSHIP

NI t −1 ∗ AGE t −0.001 0.363

N 3,456 N 3,483 N 3,552

This means that there is a U-shaped relation between family ownership

up to certain levels of family ownership (67.44% in the abnormal accruals

6.3. ROBUSTNESS TESTS

Executive compensation, such as bonuses and stock options, can affect

earnings informativeness analysis, and 33 observations in the analysis of persistence of transitory

7. Summary and Conclusions

greater percentage of founding family firms in smaller publicly-traded firms

You might also like

DNI t −1 = one if NI t −1 < 0, zero otherwise.

NI t −1 ∗ AGE t −0.001 0.363