Professional Documents
Culture Documents
J of Accounting Research - 2006 - WANG - Founding Family Ownership and Earnings Quality
J of Accounting Research - 2006 - WANG - Founding Family Ownership and Earnings Quality
J of Accounting Research - 2006 - WANG - Founding Family Ownership and Earnings Quality
x
Journal of Accounting Research
Vol. 44 No. 3 June 2006
Printed in U.S.A.
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
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]).
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
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-
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.”
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
624 D. WANG
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.
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
626 D. WANG
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:
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
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
FOUNDING FAMILY OWNERSHIP 629
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
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.
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
630 D. WANG
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
where:
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.
+ 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
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-
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.
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
FOUNDING FAMILY OWNERSHIP 633
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
However, firm histories usually identify important persons as sons, daughters, cousins, son-in-
laws, grandsons, etc.
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
634 D. WANG
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%
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.
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
636 D. WANG
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.
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
FOUNDING FAMILY OWNERSHIP 637
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
T A B L E 2 — Continued
Panel C: Persistence of transitory losses sample (N = 3,552)
Family Firms (N = 1,372) Nonfamily Firms (N = 2,180)
Std. Std.
Mean Dev. Median P25 P75 Mean Dev. Median P25 P75
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
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];
SIZE t = natural log of total assets at t;
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;
AGE t = firm age in years at t;
LOSS t = one if net income at t is negative, zero otherwise;
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;
MB t = market-to-book ratio at t;
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, zero otherwise;
P25 = 25th percentile;
P75 = 75th percentile.
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
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
FOUNDING FAMILY OWNERSHIP 639
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.
Variable definitions:
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;
SIZE t = natural log of total assets at t;
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;
AGE t = firm age in years 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.
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
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.
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).
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
FOUNDING FAMILY OWNERSHIP 641
TABLE 4
Multivariate Analysis of Earnings Informativeness and Founding Family Ownership
Dependent Variable: RET t
Model 1 Model 2 Model 3
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
Note that p-values are two-tailed values computed using White’s [1980] heteroskedasticity-corrected
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.
Variable definitions:
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;
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, 0 otherwise;
H CEO t = one if the CEO is hired from outside the founding family at t, 0 otherwise;
SIZE t = natural log of total assets at t;
LEV t = leverage at t, measured by total liabilities divided by total assets;
MB t = market-to-book ratio 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;
AGE t = firm age in years at t;
LOSS t = one if net income at t is negative, zero otherwise.
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
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
FOUNDING FAMILY OWNERSHIP 643
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
Model 1 Model 2 Model 3
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
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
644 D. WANG
T A B L E 5 — Continued
Dependent Variable: ΔNI t
Model 1 Model 2 Model 3
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
Note that p-values are two-tailed values computed using White’s [1980] heteroskedasticity-corrected
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.
Variable definitions:
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, zero otherwise;
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;
SIZE t = natural log of total assets at t;
LEV t = leverage at t, measured by total liabilities divided by total assets;
INST t = percentage of common stock owned by institutions at t;
INSIDER t = percentage of common stock owned by nonfamily insiders at t;
AGE t = firm age in years at t.
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
Variable definitions:
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;
FOUNDING FAMILY OWNERSHIP
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;
NI t = net income at t, scaled by market value of equity at the end of t−1;
NI t −1 = change in net income before extraordinary items at t−1, scaled by average total assets at t−1;
645
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
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
646 D. WANG
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
648
T A B L E 7 — Continued
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
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
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.
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 = change in net income before extraordinary items at t, scaled by average total assets at t−1;
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;
SIZE t = natural log of total assets at t;
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;
AGE t = firm age in years at t;
LOSS t = one if net income at t is negative, zero otherwise;
NI t = net income at t, scaled by market value of equity at the end of t−1;
MB t = market-to-book ratio at t;
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.
FAM OWN t = percentage of common stock owned by founding family members at t.
FOUNDING FAMILY OWNERSHIP
649
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
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
650 D. WANG
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.
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
FOUNDING FAMILY OWNERSHIP 651
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.
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
652 D. WANG
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
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.
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
FOUNDING FAMILY OWNERSHIP 655
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.”
Journal of Accounting & Economics 39 (2005): 295–327.
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.”
Journal of Accounting & Economics 7 (1985): 85–108.
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.”
Journal of Accounting & Economics 33 (2002): 375–400.
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.
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
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.