Journal of accounting, auditing finance tập 26, số 02, 2011 4

Accounting Properties of
Chinese Family Firms

Journal of Accounting,
Auditing & Finance
26(4) 623–640
Ó The Author(s) 2011
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0148558X11409147


Shujun Ding1, Baozhi Qu2, and Zili Zhuang3

Abstract
We posit that family firms in China exhibit accounting properties consistent with the prevalence of Type II agency problems. In contrast to the owners of non-family firms, the
owners of family firms have more incentives to seek private benefits of control at the
expense of minority shareholders and provide lower-quality earnings for self-interested purposes. The empirical evidence presented in this study suggests that the accounting earnings
of listed Chinese family firms are less informative, and family firms employ less conservative
accounting practices than their non-family counterparts. We also find that Chinese family
firms have higher discretionary accruals compared to non-family firms, which is consistent
with the view that family firms engage in more opportunistic reporting behavior. Overall,
our study suggests that family ownership in China is associated with lower earnings quality,
which is in sharp contrast to the findings of prior studies that examine such ownership in
the U.S.
Keyword
agency problems, accounting properties, family firms, China
This study examines the accounting properties of family firms in China. The majority of
the world’s firms can be classified as family firms to some extent (Claessens, Djankov, &
Lang, 2000; La Porta, Lopez-de-Silanes, & Shleifer, 1999), and such firms thus play a critical role in modern economies. Recent studies examining the accounting properties of

family firms primarily focus on the United States and offer interesting results. Wang
(2006), for example, finds that founding family ownership is associated with more informative earnings, more conservative reporting, and lower discretionary accruals. Ali, Chen, and
Radhakrishnan (2007) analyze the typical agency problems faced by family firms and find

1

University of Ottawa, Ontario, Canada
Skolkovo Institute for Emerging Market Studies, China
3
The Chinese University of Hong Kong
2

Corresponding Author:
Baozhi Qu, Skolkovo Institute for Emerging Market Studies, Unit 1608, North Star Times Tower, No. 8
Beichendong Road, Chaoyang District, Beijing, China, 100101
Email:

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


624

Journal of Accounting, Auditing & Finance

that these firms report higher quality earnings. These important articles shed light on how
family ownership affects accounting properties and disclosure when Type I agency problems, which arise from the separation of ownership and management and thus can be mitigated by family ownership, dominate Type II agency problems, which stem from conflicts
between controlling and noncontrolling shareholders.
Fan and Wong (2002) find listed firms in East Asia to be characterized by less informative accounting earnings. This lack of high-quality information disclosure has been said to
be responsible, at least in part, for the 1997 Asian financial crisis (Ho & Wong, 2001). Fan
and Wong (2002) examine seven East Asian jurisdictions but exclude China. However, we

believe that examining the interactions among institutional arrangements, family ownership,
and accounting properties in China would offer important incremental insights. The country’s weak legal system makes it easier for controlling shareholders to expropriate minority
shareholders (the Type II agency problem), thus providing us with a good opportunity to
investigate whether family firms tend to have different disclosure incentives and hence,
exhibit different accounting properties, in an environment in which the Type II agency
problem is more pervasive than it is in developed markets such as the United States.
It is well recognized in the literature that Type II agency problems can lead to the
manipulation of accounting earnings by family firms. For example, Ali et al. (2007) suggest
that a variety of incentives arising from these agency problems may lead family firms to
manipulate accounting earnings to facilitate private benefit-seeking behavior. For instance,
these firms may be motivated to manipulate earnings to ‘‘hide the adverse effect of a
related party transaction’’ (p. 243). Ali et al. further point out that family owners usually
have a high level of influence over the firm’s board and top management, which is certainly the case in China. Hence, they are able to manipulate earnings more easily should
they choose to do so. Although legal institutions that are designed to protect the rights of
minority shareholders may help to mitigate the differences in Type II agency problems
between family and nonfamily firms, given that such institutions are either nonexistent or
ineffective in China (Allen, Qian, & Qian, 2005), family firms are expected to be subject
to more severe Type II agency problems and, accordingly, to have poorer earnings quality.
Using a sample of all listed nonstate firms in China from 2003 to 2006, we first examine
the informativeness of accounting earnings and find the reported earnings of family firms
to be less informative than those of nonfamily firms.1 We then use a piecewise serial
dependence model to test the relationship between family firms and the persistence of the
transitory loss components in earnings, which measures accounting conservatism. The
family firms in our sample are found to use less conservative accounting practices than
their nonfamily counterparts. Finally, we examine the relationship between discretionary
accruals and family firms and find these firms to have a higher level of such accruals. Our
results remain robust across different model specifications and to the inclusion of different
control variables.
The findings of this study make several contributions to the literature. First, we document the importance of Type II agency problems to financial reporting. The effect of
agency conflicts on disclosure has been extensively investigated (see Healy & Palepu,

2001, for a review), and recent studies have examined the impact of family ownership on
corporate disclosure (e.g., Ali et al., 2007; Chen, Chen, & Cheng, 2008; Wang, 2006).
However, the aforementioned studies focus on the U.S. market, which is characterized by
Type I agency problems (e.g., Ali et al., 2007). Our focus on Chinese family firms offers
us an opportunity to isolate a setting in which Type II agency problems dominate. Our
investigation of the accounting properties of these firms thus provides insights that are

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


Ding et al.

625

complementary to those of previous studies (e.g., Ali et al., 2007; Wang, 2006) examining
the way in which agency conflicts affect accounting properties.
Second, an examination of family firms in China, where privatization took place only
recently and the founders of family firms still run their firms directly, enables us to sharpen
our tests of the impact of family ownership on financial reporting and disclosure. In contrast to Chinese family firms, the majority of those in the United States are entrepreneurial
firms. The founders of U.S. family firms often hire professional managers. When these
founders retire, their families usually hold only ‘‘marginal ownership’’ (Burkart, Panunzi,
& Shleifer, 2003, p. 2168). Prior studies that use samples of U.S. family firms use either
the S&P 500 (e.g., Ali et al., 2007; Wang, 2006) or the S&P 1500 (e.g., Chen et al., 2008),
which may raise concerns about the sample (Hutton, 2007). Recent studies have shown that
findings involving family firms are ‘‘indeed sensitive’’ to the sample used (Miller, BretonMiller, Lester, & Cannella, 2007, p. 831); these authors have discovered that findings
based on Fortune 1000 firm data simply cannot be replicated in randomly drawn samples
of smaller public companies. By no means do we suggest that our study uses a noise-free
setting, but the early stages of both Chinese family firms and the Chinese stock market
may offer a more powerful context for our tests.
This study also differs from prior studies that examine ownership concentration in

East Asia (e.g., Fan & Wong, 2002) and earnings quality in China (e.g., Firth, Fung, &
Rui, 2007), none of which investigates the difference between family and nonfamily
firms. Furthermore, Fan and Wong (2002) did not include China in their sample, and Firth
et al. (2007) compare earnings quality between the country’s state- and nonstate-owned
firms. Our study focuses on the impact of family ownership on corporate disclosure and
thus adds to this literature.
The rest of the article proceeds as follows. The next section discusses China’s institutional background. Section titled ‘‘Relevant Literature and Hypothesis Development’’
reviews the relevant literature and develops our hypothesis. Section titled ‘‘Sample and
Empirical Analysis’’ discusses our sample and empirical tests, and the last section concludes the article.

Institutional Background of China
China had a centrally planned economy for the three decades following the birth of the
People’s Republic of China in 1949. The country’s economic reforms and opening-up
policy began in 1978 and initially focused on rural areas. In the 1980s, these reforms,
which blended central-planning elements with market-based practices, were extended
beyond the agricultural sector to state-owned enterprises (SOEs). It was not until 1992 that
the Chinese Communist Party formally announced, at its 14th National Congress, that
China was adopting a socialist market-based system. A significant chapter in the country’s
transition to this economic system was the establishment of the Shanghai and Shenzhen
Stock Exchanges in the early 1990s, and its capital markets have experienced unprecedented growth since then.
The majority of listed firms on the country’s stock market are the result of the corporatization of SOEs. Typically, an operational unit of a large SOE was carved out, with its net
assets converted to nontradable shares at a certain rate. The remaining shares were then
issued to the public and can be traded. As in other countries in which the government
keeps a controlling stake in listed (and partially privatized) SOEs, the Chinese central and
local governments remain, either directly or indirectly, the controlling shareholders of these

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


626

Journal of Accounting, Auditing & Finance

firms (Chen, Firth, Gao, & Rui, 2006). According to Chen et al. (2006), 30% of the shares
in a typical listed SOE are owned by the government or government-related agencies, and
another 30% are held by legal entities that are usually controlled by the state. The remaining 40% are owned by individuals (including management and employees), private institutions, and foreign investors. In non-SOE listed firms, the controlling shareholder may be a
family (discussed below) or some other type of nonstate entity (such as a foreign investor,
the firm’s employees, etc.).
Hence, the Chinese stock market presents two unique ownership features. First, although
all shareholders have the same rights, there are six types of shares: state, legal entity, foreign, management, employees, and other individuals (Firth, Fung, & Rui, 2007). The shares
held by state and legal entities cannot be traded on the market, whereas those owned by
individuals are actively traded. Second, ownership is highly concentrated. The state, and/or
a legal entity shareholder, often controls the listed company, and, typically, there are no
other block holders (Chen et al., 2006). Of the aforementioned six types of shares, those
held by management, employees, and foreign investors usually account for less than 3%
(Firth et al., 2007).
Relative to SOEs, private companies, including family firms, are a recent product of the
country’s economic reforms and opening-up policy. The first group of entrepreneurs generally comprised farmers and workers who had been laid off as a result of the SOE reforms.
It is estimated that around 140,000 such entrepreneurs started up family businesses in the
early 1980s. The expansion of economic reform has led to the rapid growth of these family
firms, which have had a presence in China’s capital market since its inception. In the first
10 years of this market’s establishment, the number of listed family firms increased annually by 83.8% (Zhang & Zhang, 2004). In all, 36% of these firms went public through
Initial Public Offerings (IPOs), 3% were listed through management buyouts, and the
remainder obtained listing status by acquiring existing listed companies (Zhang & Zhang,
2004). Although Chinese family firms have clearly become increasingly significant, surprisingly, to the best of our knowledge, no one in the literature has studied the impact of
family ownership on the accounting properties among the nonstate firms in China.

Relevant Literature and Hypothesis Development
Agency Problems and Family Firms
Firms face two types of agency problems, both of which have significant implications for

the accounting properties of family firms (e.g., Ali et al., 2007; Wang, 2006). The first
type, known as Type I agency problems, results from the separation of ownership and control, and may lead managers to act in their own best interests rather than those of the shareholders (Jensen & Meckling, 1976). Type I agency problems are typical in countries in
which ownership is diffuse, such as the United States. The second type of agency problem,
known as Type II, stems from the conflict between controlling and noncontrolling shareholders (Ali et al., 2007), and is common in regions in which the ownership of listed firms
is usually concentrated in the hands of a single shareholder, as is generally the case in East
Asia (Fan & Wong, 2002). Both types of agency problems result in incentives and disincentives for accounting transparency and corporate disclosure.
Compared with their nonfamily counterparts, family firms usually exhibit different patterns in both types of agency problems. Several factors influence the Type I agency problems in these firms. First, family members usually hold positions among top management

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


Ding et al.

627

or serve on the board of directors and are sometimes directly involved in the firm’s operations. As a result, they usually have better knowledge of the daily operations of the firm,
which enables them to monitor managers more effectively and reduce opportunistic behavior on the part of the latter (Anderson & Reeb, 2003). Second, family firms are long-term
oriented, and thus, the managers of these firms are less likely to seek short-term benefits by
manipulating accounting earnings (e.g., Chen et al., 2008). Third, family firms are more
sensitive to negative market events, such as litigation (Chen et al., 2008). For all of
these reasons, family firms may be less subject to the severe agency problems that often
arise from the separation of ownership and control and more likely to disclose higher quality earnings.
However, Type II agency problems may lead the controlling owners of these firms to
engage in opportunistic activities. Family owners may use their controlling positions in the
firm to expropriate outside shareholders through various channels, such as related-party
transactions (Anderson & Reeb, 2003) and freezing out minority shareholders (Gilson &
Gordon, 2003), and they may pursue their own interests at the expense of those of noncontrolling shareholders (Ali et al., 2007). Correspondingly, the controlling shareholders of
family firms have more incentives to hide relevant information by disclosing lower quality
earnings, as such opacity helps them to expropriate outside shareholders.
The severity of one type of agency problem over the other determines the quality of the

information that firms disclose. For example, the U.S. market is characterized by diffuse
ownership, and Type II agency problems are significantly alleviated due to the wellestablished investor protection mechanisms in that country. As a result, Type I agency
problems tend to dominate Type II in the United States, which means that the family firms
there are more likely to disclose higher quality earnings, as family ownership mitigates
Type I problems. Ali et al. (2007) and Wang (2006) provide evidence to support this argument. Our study, in contrast, examines a setting in which Type II agency problems are
likely to be dominant, thus enriching our understanding of the impact of such agency problems on accounting properties.

Hypothesis Development
Both types of agency problems exist in China, although, as noted, Type II problems are
predominant for several reasons. First, the existence of dominant shareholders is a typical
feature of listed firms in China. Second, the country’s investor protection mechanisms are
weak, despite the rapid development of its macro-legal environment. The Chinese legal
system has been heavily influenced by the civil law tradition. La Porta, Lopez-de-Silanes,
Shleifer, and Vishny (1998) argue that legal protection for shareholders is weakest in countries with a civil law legal origin. Furthermore, Allen et al. (2005) provide evidence to
show that creditor and shareholder protection in China is even worse than that in other
major emerging markets. The existence of dominant shareholders, in conjunction with
weak protection for investors, renders Type II agency problems more salient, which helps
to explain why the China Securities Regulatory Commission, the country’s stock market
regulator, has repeatedly asserted that its top priority is strengthening minority shareholder
protection.
The governance mechanisms and incentive structure of family firms differ from those of
nonfamily firms in several important aspects, all of which have significant implications for
Type II agency problems. The key difference lies in their different ultimate controllers.
Unlike family firms, which are controlled by an individual person and his or her family,

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


628

Journal of Accounting, Auditing & Finance

nonfamily firms are controlled by a group of legal entities, employees, or private investors,
including institutional investors. The benefits of expropriation for these controlling shareholders, which are typically institutions, organizations, or a group of private investors that
are independent of one another, are thus diluted. As a result, the large shareholders of nonfamily firms have fewer incentives to expropriate minority shareholders (Villalonga &
Amit, 2006). Family owners in China, in contrast, usually have absolute control over their
firm’s board and management and are less constrained by its corporate governance system.
Such a control structure makes it less costly for them to expropriate minority shareholders.
Furthermore, the private benefits of control are not diluted, as they all go to the family
owners. Family owners thus have stronger incentives to seek private benefits at the expense
of minority shareholders and may have more significant Type II agency problems compared with nonfamily firms. Consequently, their disclosures tend to be more opaque for
self-interested purposes. For example, their accounting may be less informative, and their
earnings may be managed to bury the wealth effects (transfers) of their expropriation
activities.
Type I agency problems, in contrast, which arise from the separation of ownership and
management, are similar for family and nonfamily firms in China because, on average,
both have a concentrated ownership structure. Large shareholders are likely to monitor
management effectively and are sometimes directly involved in management. For example,
chairman of the board is an executive position in China. This chairman, who is presumed
to represent the interests of the controlling shareholder(s), is often directly involved in
operations (Chen et al., 2006). In addition, chief executive officers (CEOs) and other top
managers are usually appointed by the controlling shareholders (or, in some cases, are actually the founders or their family members).
In summary, compared with their counterparts in the United States, Chinese family firms
constitute a unique sample that is characterized by a higher degree of Type II agency problems. Consequently, we posit that Chinese family firms have different disclosure incentives
than such firms in the United States, and we thus put forward the following hypothesis.
Hypothesis: Family firms in China report lower quality accounting earnings than
their nonfamily counterparts.

Sample and Empirical Analysis
This section presents our empirical analyses. Our sample covers all non-state listed companies in China from 2003 to 2006. We exclude finance firms, although their inclusion has

no quantitative effect on our results. General accounting data and data on family firms are
obtained from the Guotaian (GTA) databases, which are widely used in accounting and
finance research using Chinese data (e.g., Haw, Qi, Wu, & Wu, 2005; Sun & Tong, 2003;
Wei, Xie, & Zhang, 2005). In this study, a family firm is defined as a firm that is controlled by a private person and his or her family through direct stock ownership or through
a pyramid structure. In addition, for a firm to be considered a family firm, the ownership
stake of the controlling family owner (the largest shareholder) must be greater than or
equal to 10%. Given the short history of the Chinese stock market and the rarity of mergers
and acquisitions among the country’s listed companies, most, if not all, listed Chinese
family firms are still controlled by their founders and their families. The nonfamily firms
in our data set primarily include the following types of companies: dispersedly held companies with no controlling owner (or family), companies that are controlled by a group of

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


Ding et al.

629

Table 1. Descriptive Statistics
Family firm

RET
EARNINGS
DNI
ABS_ACC
SIZE
BETA
MB
LEVERAGE
LOSS

SEO
CFO
ROA
RETVOL
MVE

Nonfamily firm

n

M

Median

n

M

Median

917
829
870
782
922
841
922
922
922
922

869
923
912
922

0.006
0.007
0.009
0.112
20.764
1.117
2.924
0.585
0.180
0.018
0.056
20.012
0.438
1,553.690

20.062
0.001
0.001
0.067
20.729
1.110
2.239
0.535
0.000
0.000

0.049
0.022
0.396
1,016.750

613
606
566
579
619
592
619
619
619
619
589
619
619
619

20.048***
0.007
0.010
0.101
20.814
1.091
2.854
0.647***
0.216*
0.019

0.046**
20.028**
0.391***
1,498.132

20.091**
0.001
0.001
0.068
20.850
1.100
2.090
0.554**
0.000*
0.000
0.044*
0.018**
0.355***
1,084.356

Note: RET = cumulative market-adjusted returns over the 12-month period from 8 months before the fiscal year-end
to 4 months after it (that is, from May 1 of year t to April 30 of year t 1 1); EARNINGS = the annual change in net
income, deflated by the market value of equity at the beginning of the year; DNI the change in net income, calculated
as the net income of year t minus that of year t 2 1, scaled by the book value of equity at the beginning of year t;
ABS_ACC = the absolute value of discretionary accruals (performance-matched discretionary accruals calculated following Ali, Chen, & Radhakrishnan, 2007); SIZE = the natural logarithm of the year-end book value of total assets; BETA =
the stock beta at year t; MB = the market-to-book ratio, calculated as the year-end share price divided by the book
value of equity per share; LEVERAGE = the leverage ratio of the firm at the end of the year, calculated as the year-end
book value of total liability divided by total assets; LOSS = a dummy variable that equals one if net income \ 0 and
zero otherwise; SEO = a dummy variable that equals one if the company has seasoned equity offerings and zero otherwise; CFO = cash flow from operations scaled by beginning-of-year total assets; ROA = return on assets, measured by
net income divided by average total assets; RETVOL = annual stock volatility calculated using monthly stock returns

over the year; and MVE = the market value of equity, in millions of Chinese yuan.
*** indicates that the mean (or median) value of the variable for family firms is significantly different from that for
nonfamily firms at the 1% level; ** indicates a significance level of 5%; and * indicates a significance level of 10%.

investors (who are not from the same family), companies that are controlled by a group of
legal entities (such as not-for-profit organizations, township and village organizations, etc.),
companies whose shares are held by employees or their unions, foreign-invested companies, and other nonstate firms that are not family controlled.
Table 1 presents the descriptive statistics of our sample. On average, the family firms
are smaller and have a higher level of abnormal returns, lower leverage ratios, greater cash
flows from operations, greater profitability, and a higher level of return volatility than the
nonfamily firms. These characteristics suggest that family firms’ Type I agency problems
are at least not as severe as that of nonfamily firms even though both groups of firms have
similar level of ownership concentration, a result that is consistent with prior studies (e.g.,
Ali et al., 2007).2 Differences in accounting properties between family and nonfamily firms
are thus likely to be driven by Type II agency problems. Finally, at the bottom of Table 1,
we can see that both groups of firms have a similar degree of market capitalization.
Table 2 presents the sample distribution for family and nonfamily firms by year and
industry. The total number of both types of firms increased steadily from 2003 to 2006. Of
the 1,542 observations, approximately 59.8% are family firms, about two thirds of which

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


630

Journal of Accounting, Auditing & Finance

Table 2. Sample Distribution by Year and Industry
Number of firms
Year

2003
2004
2005
2006
Total observations

Family firm

Nonfamily firm

Total

99
243
271
309
922

198
134
126
162
620

297
377
397
471
1,542

86
57
100
625
54
922

63
28
96
379
54
620

149
85
196
1,004
108
1,542

Industry code
2: Utilities
3: Properties
4: Conglomerates
5: Industrials
6: Commerce
Total observations

are industrial firms. In the overall sample, the firms are from five different industries, with

the majority industrial firms (67.8% for family firms and 61.1% for nonfamily firms).
We examine three attributes of accounting earnings in Chinese family firms: earnings
informativeness; the persistence of transitory loss components in earnings, which is a measure of conservatism; and discretionary accruals. These three attributes have also been
examined in studies of the earnings quality of U.S. family firms (e.g., Ali et al., 2007;
Wang, 2006) and that of East Asian firms in general (Fan & Wong, 2002).3

Informativeness of Accounting Earnings
We follow the common practice in the literature to measure the informativeness of accounting earnings (e.g., Ali et al., 2007; Collins & Kothari, 1989) and examine whether those of
the family firms in our sample are less informative. The primary estimation model is given
by the following:
RET 5a1b1 EARNINGS1b2 EARNINGS3FAMILY 1b3 EARNINGS3SIZE1
b4 EARNINGS3BETA1b5 EARNINGS3MB1b6 EARNINGS3LEVERAGE1
b7 EARNINGS3RETVOL1INDUSTRY EFFECTS1YEAR EFFECTS1error;
ð1Þ
where RET represents cumulative market-adjusted returns over the 12-month period from
8 months before the fiscal year-end to 4 months after it (from May 1 of year t to April 30 of
year t 1 1), which includes the earnings announcement period; EARNINGS is the annual
change in net income,4 deflated by the market value of equity at the beginning of the year;
and FAMILY is a dummy variable that equals one if the firm is a family firm and zero otherwise. Following the prior literature (e.g., Ali et al., 2007; Collins & Kothari, 1989), we
include the following control variables in our regression models: firm size (SIZE, which is

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


Ding et al.

631

Table 3. Correlations Among Variables
RET

Family firms
RET
1.000
EARNINGS 0.194*
SIZE
0.040
BETA
0.027
MB
0.082*
LEVERAGE 20.113*
LOSS
20.218*
SEO
0.093*
RETVOL
0.157*
Nonfamily firms
RET
1.000
EARNINGS 0.294*
SIZE
0.115*
BETA
0.043
MB
20.001
LEVERAGE 20.148*
LOSS
20.275*

SEO
0.003
RETVOL
0.016

EARNINGS

SIZE

BETA

MB

LEVERAGE

LOSS

SEO

RETVOL

1.000
20.041
0.094*
0.026
0.026
20.382*
0.083*
0.155*

1.000
20.156* 1.000
20.100* 0.048
1.000
20.202* 0.161* 20.116*
1.000
20.230* 0.126* 20.003
0.417*
1.000
0.153* 20.024
0.010 20.037 20.064
1.000
20.263* 0.182* 0.083*
0.123*
0.070* 20.068* 1.000

1.000
20.050
0.143*
0.001
0.080*
20.328*
0.027
0.062

1.000
20.147* 1.000
20.142* 20.073
1.000
20.288* 0.162* 20.199*

1.000
20.297* 0.122* 20.002
0.471*
1.000
0.098* 20.031 20.015 20.074 20.074
20.250* 0.026
0.026
0.293*
0.308*

1.000
0.006

1.000

Note: RET = cumulative market-adjusted returns over the 12-month period from 8 months before the fiscal yearend to 4 months after it (that is, from May 1 of year t to April 30 of year t 1 1); EARNINGS = the annual change in
net income, deflated by the market value of equity at the beginning of the year; SIZE = the natural logarithm of the
year-end book value of total assets; BETA = the stock beta at year t; MB = the market to book ratio, calculated as
the year-end share price divided by the book value of equity per share; LEVERAGE = the leverage ratio of the firm
at the end of the year, calculated as the year-end book value of total liability divided by total assets; LOSS = a
dummy variable that equals one if net income \ 0 and zero otherwise; SEO = a dummy variable that equals one if
the company has seasoned equity offerings and zero otherwise; RETVOL = annual stock volatility calculated using
monthly stock returns over the year.
* indicates that the correlation is significant at the 5% level or better.

the natural logarithm of the year-end book value of total assets); stock beta (BETA); growth
potential (MB, which is the market-to-book ratio, calculated as the year-end share price
divided by the book value of equity per share); risk of bankruptcy (LEVERAGE, which is the
leverage ratio of the firm at the end of the year, calculated as the year-end book value of
total liability divided by total assets); and return volatility (RETVOL, which represents annual

stock volatility and is calculated using monthly stock returns over the year). Finally,
INDUSTRY EFFECTS and YEAR EFFECTS are dummy variables that are included to control
for industry and time-fixed effects, respectively. To mitigate the undue influence of outliers,
the continuous variables used in our analysis are winsorized at the 1st and 99th percentiles.5
Simple correlation analysis (see Table 3) reveals that cumulative market-adjusted returns
and earnings are more positively correlated for nonfamily firms than they are for family
firms (both correlation coefficients are significant at the 5% level), which is consistent with
the supposition that family firms are characterized by less informative earnings.
Table 4 presents our estimation results. Robust standard errors adjusted for clustering
and heteroscedasticity are reported in parentheses for all of the coefficient estimates
(Petersen, 2009). Regression (1) in Table 4 is conducted to determine whether firm

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


632

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012

0.038 (0.036)
0.704*** (0.097)
—
—
—
—
—
—
Yes
Yes
.058

1,424

(1)
0.036 (0.038)
0.876*** (0.111)
20.201** (0.085)
—
—
—
—
—
Yes
Yes
.061
1,424

(2)
0.037 (0.038)
20.082 (2.469)
20.166** (0.085)
0.063 (0.110)
0.083 (0.293)
0.043** (0.022)
20.530* (0.301)
20.114 (0.791)
Yes
Yes
.081
1,349

(3)

Regression Models

0.038 (0.038)
0.022 (2.471)
20.180** (0.086)
0.056 (0.110)
0.082 (0.294)
0.046** (0.022)
20.534* (0.301)
20.060 (0.792)
Yes
Yes
.081
1,342

(4)
0.031 (0.039)
20.696 (2.517)
20.157* (0.086)
0.083 (0.112)
0.203 (0.301)
0.041* (0.023)
20.527* (0.305)
20.001 (0.812)
Yes
Yes
.080
1,268

(5)

Note: The dependent variable is RET, which is cumulative market-adjusted returns over the 12-month period from 8 months before the fiscal year-end to 4 months after it
(from May 1 of year t to April 30 of year t 1 1); EARNINGS = the annual change in net income, deflated by the market value of equity at the beginning of the year; FAMILY = a
family dummy that equals one if the firm is a family firm and zero otherwise; SIZE = the natural logarithm of the year-end book value of total assets; BETA = the stock beta at
year t; MB = the market-to-book ratio, calculated as the year-end share price divided by the book value of equity per share; LEVERAGE = the leverage ratio of the firm at the
end of the year, calculated as the year-end book value of total liability divided by total assets; RETVOL = annual stock volatility calculated using monthly stock returns over the
year; INDUSTRY EFFECTS = dummy variables that control for industry fixed effects; and YEAR EFFECTS = dummy variables that control for calendar year fixed effects. In
Regression (4), firms that issue both A shares (in the mainland Chinese stock markets) and H shares (in the Hong Kong stock market) are excluded from the sample; in
Regression (5), real estate firms are excluded. Continuous variables are winsorized for exceptionally high/low values. Standard errors (in parentheses) are robust standard
errors adjusted for clustering and heteroscedasticity. Values are bold to highlight rows of our focal interest.
*** indicates a significance level of 1%, ** indicates a significance level of 5%, and * indicates a significance level of 10%, all two-tailed.

Constant
EARNINGS
EARNINGS3FAMILY
EARNINGS3SIZE
EARNINGS3BETA
EARNINGS3MB
EARNINGS3LEVERAGE
EARNINGS3RETVOL
INDUSTRY EFFECTS
YEAR EFFECTS
Adjusted R2
No. of observations

Dependant variable: RET

Table 4. Earnings Informativeness and Family Firms

Ding et al.

633

earnings are generally informative, with a positive sign expected for b1. The estimation
result generally confirms this expectation, with the coefficient being positive and highly significant. The sign of this coefficient estimate also remains positive in Regression Model (2).
In Regressions (2) and (3) in Table 4, the main coefficient of interest is that of
EARNINGS 3 FAMILY. If it is positive, then family firms’ earnings are more informative
than those of their nonfamily counterparts. A negative coefficient would indicate the opposite. Family firms are found to disclose less informative earnings, as the coefficient estimates
have a negative sign in both of these regression models and are significant at the 5% level.
Our results remain robust across different model specifications and after controlling for
variables that are commonly used in the literature (e.g., Ali et al., 2007; Collins & Kothari,
1989; Wang, 2006), including firm size, stock beta, leverage, the effects of growth (MB),
stock price characteristics, and the like. The coefficients on the control variables, when significant, have the expected signs (except for that on EARNINGS 3 BETA). For example,
firms with better growth potential tend to report more informative earnings, and those with
a greater degree of leverage less informative earnings. In addition, because firms that issue
both A shares in the mainland Chinese stock markets and H shares in the Hong Kong stock
market are likely to be subject to more stringent monitoring and accounting quality regulations, we run a sensitivity test by excluding firms with dual A and H shares (column 4 of
Table 4). Our main results hold in this test. Finally, column 5 presents the estimation
results with real estate firms excluded from the sample. Our main results are unchanged,
but the coefficient estimate becomes only marginally significant. Our estimates may thus
be sensitive to the exclusion of certain industries such as property and real estate in the
sense that the significance level of our estimation results is somehow reduced by their
exclusions.

Persistence of Transitory Loss Components in Earnings
Researchers have long argued that the transitory loss components in earnings are less persistent than positive earnings changes, possibly as a result of the conservative nature of
accounting earnings (Basu, 1997). In this section, we use a piecewise serial dependence

model (Ball & Shivakumar, 2005; Basu, 1997; Wang, 2006) to test the relationship
between family firms and the persistence of transitory loss components in earnings.6 The
primary estimation model is given by
DNIt 5a1b1 DDNItÀ1 1b2 DNItÀ1 1b3 DNItÀ1 3DDNItÀ1 1
b4 FAMILYt 1b5 DDNItÀ1 3FAMILYt 1b6 DNItÀ1 3FAMILYt 1
b7 DNItÀ1 3DDNItÀ1 3FAMILYt 1b8 SIZEt 1b9 DDNItÀ1 3SIZEt 1
b10 DNItÀ1 3SIZEt 1b11 DNItÀ1 3DDNItÀ1 3SIZEt 1

ð2Þ

b12 LEVERAGEt 1b13 DDNItÀ1 3LEVERAGEt 1b14 DNItÀ1 3
LEVERAGEt 1b15 DNItÀ1 3DDNItÀ1 3LEVERAGEt 1
INDUSTRY EFFECTS1YEAR EFFECTS1error;
where DNIt is the change in net income, calculated as the net income of year t minus that
of year t 2 1 and scaled by the book value of equity at the beginning of year t; DNItÀ1 is
the change in net income in year t 2 1; and DDNItÀ1 is a dummy variable that equals one
if DNItÀ1 is negative and zero otherwise. All of the other variables are as previously

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


634

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012

0.045 (0.033)
20.040* (0.024)
20.035 (0.068)
20.260*** (0.048)
0.031 (0.021)

20.010 (0.031)
20.008*** (0.002)
0.218*** (0.048)
—
—
—
—
—
—
—
—
Yes
Yes
.110
1,350

(1)

(3)
20.095 (0.280)
0.322 (0.383)
0.219 (0.526)
1.772*** (0.665)
0.034* (0.020)
0.013 (0.028)
20.006* (0.003)
0.314*** (0.062)
0.011 (0.015)
20.013 (0.019)
20.017 (0.025)

20.108*** (0.034)
20.122 (0.099)
20.182 (0.127)
0.039 (0.153)
20.063 (0.050)
Yes
Yes
.182
1,343

(2)
20.094 (0.274)
0.377 (0.380)
0.210 (0.519)
1.799*** (0.666)
0.034* (0.019)
0.012 (0.028)
20.006* (0.003)
0.313*** (0.061)
0.011 (0.014)
20.016 (0.019)
20.017 (0.025)
20.109*** (0.034)
20.119 (0.099)
20.196 (0.126)
0.028 (0.153)
20.066 (0.050)
Yes
Yes
.197

1,350

Regression Models

20.079 (0.282)
0.437 (0.393)
0.268 (0.530)
1.765*** (0.665)
0.045** (0.020)
20.007 (0.028)
20.006* (0.003)
0.324*** (0.062)
0.008 (0.015)
20.017 (0.020)
20.018 (0.025)
20.108*** (0.033)
20.062 (0.100)
20.243 (0.129)
0.011 (0.158)
20.066 (0.049)
Yes
Yes
.185
1,270

(4)

Note: The dependent variable is DNIt , which is the change in net income, calculated as the net income of year t minus that of year t 2 1, scaled by the book value of equity at
the beginning of year t; DDNItÀ1 = a dummy variable that equals one if DNItÀ1 \ 0 and zero otherwise; FAMILY = a family dummy that equals one if the firm is a family firm and
zero otherwise; SIZE = the natural logarithm of the year-end book value of total assets; LEVERAGE = the leverage ratio of the firm at the end of the year, calculated as the yearend book value of total liability divided by total assets; INDUSTRY EFFECTS = dummy variables that control for industry fixed effects; and YEAR EFFECTS = dummy variables that

control for calendar year fixed effects. In Regression (3), firms that issue both A shares (in the mainland Chinese stock markets) and H shares (in the Hong Kong stock market)
are excluded from the sample; in Regression (4), real estate firms are excluded. Continuous variables are winsorized (1% in each tail). Standard errors (in parentheses) are
robust standard errors adjusted for clustering and heteroscedasticity. Values are bold to highlight rows of our focal interest.
*** indicates a significance level of 1%, ** indicates a significance level of 5%, and * indicates a significance level of 10%, all two-tailed.

Constant
DDNItÀ1
DNItÀ1
DNItÀ1 3DDNItÀ1
FAMILYt
DDNItÀ1 3FAMILYt
DNItÀ1 3FAMILYt
DNItÀ1 3DDNItÀ1 3FAMILYt
SIZEt
DDNItÀ1 3SIZEt
DNItÀ1 3SIZEt
DNItÀ1 3DDNItÀ1 3SIZEt
LEVERAGEt
DDNItÀ1 3LEVERAGEt
DNItÀ1 3LEVERAGEt
DNItÀ1 3DDNItÀ1 3LEVERAGEt
INDUSTRY EFFECTS
YEAR EFFECTS
Adjusted R2
No. of observations

Dependent variable: DNIt

Table 5. Persistence of Transitory Loss Components in Earnings and Family Firms

Ding et al.

635

Table 6. Discretionary Accruals and Family Firms
Regression Models
Dependent variable: ABS_ACC
CONSTANT
FAMILY
SIZE
LEVERAGE
MB
LOSS
CFO
ROA
SEO
INDUSTRY EFFECTS
YEAR EFFECTS
Adjusted R2
No. of observations

(1)

(2)

(3)

0.135*** (0.016)
0.014* (0.009)

—
—
—
—
—
—
—
Yes
Yes
.019
1,283

0.394** (0.160)
0.016* (0.008)
20.015* (0.009)
0.046 (0.029)
0.005 (0.005)
20.042*** (0.012)
20.050 (0.076)
20.112 (0.111)
20.004 (0.017)
Yes
Yes
.053
1,283

0.394** (0.164)
0.016* (0.008)
20.015* (0.009)
0.046 (0.029)

0.005 (0.005)
20.043*** (0.012)
20.050 (0.077)
20.116 (0.114)
20.005 (0.017)
Yes
Yes
.052
1,276

Note: The dependent variable is ABS_ACC, which is the absolute value of discretionary accruals (performancematched discretionary accruals calculated following Ali et al., 2007); FAMILY = a family dummy that equals one if
the firm is a family firm and zero otherwise; SIZE = the natural logarithm of the year-end book value of total
assets; MB = the market to book ratio, calculated as the year-end share price divided by the book value of equity
per share; LEVERAGE = the leverage ratio of the firm at the end of the year, calculated as the year-end book value
of total liability divided by total assets; LOSS = a dummy variable that equals one if net income \ 0 and zero otherwise; SEO = a dummy variable that equals one if the company has seasoned equity offerings and zero otherwise;
CFO = cash flow from operations scaled by beginning-of-year total assets; ROA = return on assets, measured by net
income divided by average total assets; INDUSTRY EFFECTS = dummy variables that control for industry fixed
effects; and YEAR EFFECTS = dummy variables that control for calendar year fixed effects. Real estate firms are
excluded from the regressions. In Regression (3), firms that issue both A shares (in the mainland Chinese stock
markets) and H shares (in the Hong Kong stock market) are excluded from the sample. Continuous variables are
winsorized (1% in each tail). Standard errors (in parentheses) are robust standard errors adjusted for clustering
and heteroscedasticity. Values are bold to highlight rows of our focal interest.
*** indicates a significance level of 1%, ** indicates a significance level of 5%, and * indicates a significance level of
10%, all two-tailed.

defined. The coefficient of interest on DNItÀ1 3DDNItÀ1 3FAMILYt captures the incremental difference in accounting conservatism between family firms and nonfamily firms
(Ball & Shivakumar, 2005). A positive b7 would suggest that the former are less conservative in accounting (lower earnings quality) than the latter and vice versa.
Table 5 reports the estimation results. Similar to those reported in the previous section,
the robust standard errors adjusted for clustering and heteroscedasticity are reported in parentheses. The coefficient estimates on DNItÀ1 3DDNItÀ1 3FAMILYt are positive in both
Regression Models (1) and (2), and are significant at the 1% level, which means that

family firms use less conservative accounting than do nonfamily firms. Following the arguments in the prior literature (Ball & Shivakumar, 2005; Wang, 2006), this finding provides
evidence to indicate that Chinese family firms issue lower quality financial reports than do
nonfamily firms.
Another interesting result from Model 2 of Table 5 is that the coefficient estimate on
DNItÀ1 3DDNItÀ1 3SIZE t is significantly negative (at the 1% level), thus indicating
that larger firms are more conservative, which is consistent with the notion that
such firms may provide higher quality financial reports. In general, the overall regression

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


636

Journal of Accounting, Auditing & Finance

models are highly significant. Industry and year dummies are again included to
control for fixed industry and time effects, and, similar to the previous section, we run
sensitivity tests by excluding A and H dual share firms (Regression 3 in Table 5) and real
estate firms (Regression 4 in Table 5) from the sample, but the main results remain
unchanged.
It must be stated here that the usual caveat applies when interpreting these findings.
As Ball and Shivakumar (2005) and Wang (2006) point out, the serial dependence model
(Basu, 1997) is limited by its potential inability to distinguish the transitory components
in earnings from random accruals errors and from some types of earnings management.
In addition, this model may be unable to determine whether the recognition of the transitory loss components in earnings is timely or untimely (Ball & Shivakumar, 2005).

Discretionary Accruals
Following the literature (e.g., Ali et al., 2007; Ashbaugh, LaFond, & Mayhew, 2003), we
estimate the following model to examine the relationship between discretionary accruals
and family firms:

ABS ACC5a1b1 FAMILY 1b2 SIZE1b3 LEVERAGE1b4 LOSS1b5 CFO
1b6 ROA1b7 SEO1b8 MB1INDUSTRY EFFECTS

ð3Þ

1YEAR EFFECTS1error:
The dependent variable ABS_ACC is the absolute value of discretionary accruals, which
are performance-matched discretionary accruals calculated as in Ali et al. (2007). More
specifically, we match firms by return on assets (ROA) within their industry, that is, utilities, conglomerates, industrials, and commerce, with the property and real estate industry
excluded from this analysis.7 We include control variables following the literature (e.g., Ali
et al., 2007; Wang, 2006). Specifically, we control for the risk of bankruptcy
(LEVERAGE), firm size (SIZE), and growth potential (MB). Firms may manage their earnings to meet the regulatory standards for stock rights offerings, and we thus control for this
effect by including a SEO dummy that equals one if the firm had a seasoned equity offering
and zero otherwise. CFO is defined as cash flows from operations scaled by total assets at
the beginning of the year, and ROA is the current year’s return on assets. LOSS is a dummy
variable that equals one if net income \ 0 and zero otherwise. All of the other variables
are as previously defined. Finally, we include industry and year dummies to control for
time and industry effects. A positive coefficient on the family dummy would indicate that
family firms are associated with a higher level of discretionary accruals.
The descriptive statistics reported in Table 1 suggest that the mean value of ABS_ACC
for family firms is higher than that for nonfamily firms, although the difference is statistically insignificant. Table 6 reports the multiple regression results with the robust standard
errors adjusted for clustering and heteroscedasticity in parentheses. The coefficient estimates of the family dummy in Models 1 and 2 are significantly positive at the 10% level,
thus indicating that the level of discretionary accruals is higher for family firms than for
their nonfamily counterparts and that greater opportunistic reporting behavior exists among
the former. If a higher level of discretionary accruals proxies for lower quality earnings,
then this result is consistent with those on earnings informativeness and conservatism.

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012

Ding et al.

637

These findings are robust to the inclusion of the various control variables commonly
used in the literature, and the signs of the coefficients on these variables are also generally
consistent with the prior literature. For example, the coefficient on SIZE is negative and
significant at the 10% level, which suggests that large firms have a lower level of discretionary accruals. The coefficient on LOSS has a negative sign, which may reflect the
greater degree of monitoring by the government and the market that loss firms receive in
China, thus making it more difficult for these firms to manipulate earnings.8 In addition, as
previously stated, if a firm issues both A shares on the Shanghai or Shenzhen stock
exchange and H shares on the Hong Kong exchange, then it may be subject to greater regulatory scrutiny and therefore less likely to manage earnings. As a sensitivity test, we
exclude dual A and H share firms from our sample (column 3 of Table 6), but our main
results hold. However, caution must be exercised in interpreting our discretionary accruals
results, as the FAMILY coefficient is only marginally significant across all of the regressions in Table 6.
In summary, using a portfolio of earnings quality measures, including earnings informativeness, conservatism, and discretionary accruals, we find evidence that is consistent with
the notion that family firms in China exhibit certain accounting properties that stem from
Type II agency problems.

Conclusion
Family firms have become an increasingly important area of research (e.g., Bennedsen,
Nielsen, Perez-Gonzalez, & Wolfenzon, 2007). Previous accounting studies using U.S.
family firm data provide evidence on the degree of transparency and disclosure in these
firms when Type I agency problems dominate. However, compared with U.S. family firms,
which are usually managed and controlled by professional managers, those in Asian economies, including China, are more pervasive and diverge more from their nonfamily counterparts. Because of the weak investor protection mechanisms and less-advanced market
development in these economies, Type II agency problems are likely to play a larger role.
The use of Chinese data thus helps to alleviate some of the data problems seen in U.S.
studies and to isolate Type II agency problems, which has sharpened our tests of the interactions among institutional development, incentives, ownership structure, and accounting
properties. Different from the findings reported by Ali et al. (2007) and Wang (2006) on
U.S. family firms, we find that such firms in China disclose less informative earnings, are

less conservative in their financial reporting, and are more likely to engage in the manipulation of discretionary accruals. These findings are consistent with our argument that
Chinese family firms have greater Type II agency problems than their nonfamily
counterparts.
Several limitations must be acknowledged. This study does not examine the effect of
China’s convergence to International Financial Reporting Standards (IFRS), a process that
began in 1993. Following the country’s new generally accepted accounting principles
(GAAP), which became effective in 2007, it is clear that the Chinese authorities believe the
convergence between Chinese GAAP and IFRS to be nearly complete (Ding & Su, 2008,
p. 478). The findings of the current study suggest that incentives and institutions, perhaps
in combination with accounting standards and education, may have an impact on corporate
transparency. Our sample period ended before the new standards were implemented in
2007. A possible direction for future research would be to examine whether China’s staged
approach to IFRS convergence has any impact on the results reported here. The

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


638

Journal of Accounting, Auditing & Finance

profitability requirement and valuation of IPOs have been dramatically changed over the
past 15 years, and corporate laws are increasingly well defined.9 Future research could also
examine the effect of these changes on our results. In addition, some of our results (such as
those of our discretionary accruals analysis) are marginally significant, and caution should
thus be exercised in their interpretation.
Authors’ Note
The authors are grateful for the helpful comments from the anonymous referee, Charles Chen, Yuan
Ding, Gordon Richardson, and seminar participants at the Peking University, Shanghai Jiaotong
University, and the 2009 Shanghai Winter Finance Conference of China. They also thank Qingbo

Yuan for his excellent research assistance. The usual disclaimer applies.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/
or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this
article.

Notes
1. The authors investigate nonstate firms in this article and thus ‘‘nonfamily firms’’ refers to nonstate, nonfamily firms hereafter.

2. In untabulated analysis, the largest shareholders of family firms own about 35.5% of shares, and
3.
4.
5.

6.

7.
8.
9.

the largest shareholders of nonfamily firms own about 36.2% of shares; the difference in ownership concentration between family and nonfamily firms is insignificant.
Fan and Wong (2002) use only earnings informativeness, not the other two attributes.
The authors also run regressions using operating income rather than net income, and the results
are similar.
Speculative trading is an important factor that may affect the value of RET for firms in China. It
is well known that speculative trading, which tends to create outliers with exceptionally high- or

low-RET values, is rife in the Chinese stock markets. We tackle this problem by winsorizing
RET at the 10th and 90th percentiles.
The authors also tried Basu’s (1997) reverse regression approach. However, the relationship
between family ownership and timely loss recognition is nonsignificant even though the coefficient estimate has the expected sign. As Gigler and Hemmer (2001) argue, Basu’s approach may
generate biased results because it does not control for the potential effect of voluntary disclosures
on stock prices.
Including these firms in the analysis does not change our main results.
In China, a listed firm with losses may be designated as Special Treatment (ST) or Particular
Treatment (PT) firms by the regulatory body. ST/PT firms are usually subject to greater regulatory and market scrutiny.
The authors thank the anonymous referee for raising this point. In China, firms were required to
achieve an annual profitability level of at least 10% for three consecutive years to be qualified
for an IPO, but the criteria were changed later to 10%, on average, for three consecutive years.
Furthermore, the Chinese Corporate Law was enacted in 1993 and was amended 3 times in
1999, 2004, and 2005. It is believed that these amendments have led to significant improvements. The latest Corporate Law, for instance, has codified independent directorship and

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


Ding et al.

639

addressed serious constraints facing supervisory boards, which are part of the dual-board structure in Chinese listed companies.

References
Ali, A., Chen, T. Y., & Radhakrishnan, S. (2007). Corporate disclosures by family firms. Journal of
Accounting and Economics, 44, 238-286.
Allen, F., Qian, J., & Qian, M. (2005). Law, finance, and economic growth in China. Journal of
Financial Economics, 77, 57-116.
Anderson, R., & Reeb, D. (2003). Founding-family ownership and firm performance: Evidence from

the S&P 500. Journal of Finance, 58, 1301-1328.
Ashbaugh, H., LaFond, R., & Mayhew, B. W. (2003). Do non-audit services compromise auditor
independence? Further evidence. Accounting Review, 78, 611-639.
Ball, R., & Shivakumar, L. (2005). Earnings quality in U.K. private firms: Comparative loss recognition. Journal of Accounting and Economics, 39, 83-128.
Basu, S. (1997). The conservatism principle and asymmetric timeliness of earnings. Journal of
Accounting and Economics, 24, 3-37.
Bennedsen, M., Nielsen, K., Perez-Gonzalez, F., & Wolfenzon, D. (2007). Inside the family firm: The
role of families in succession decisions and performance. Quarterly Journal of Economics, 122,
647-691.
Burkart, M., Panunzi, F., & Shleifer, A. (2003). Family firms. Journal of Finance, 58, 2167-2201.
Chen, G., Firth, M., Gao, D., & Rui, O. (2006). Ownership structure, corporate governance, and
fraud: Evidence from China. Journal of Corporate Finance, 12, 424-448.
Chen, S. P., Chen, X., & Cheng, Q. (2008). Do family firms provide more or less voluntary disclosure? Journal of Accounting Research, 46, 499-536.
Claessens, S., Djankov, S., & Lang, L. H. (2000). The separation of ownership and control in East
Asian Corporations. Journal of Financial Economics, 58, 81-112.
Collins, D., & Kothari, S. P. (1989). An analysis of inter-temporal and cross-sectional determinants of
earnings response coefficients. Journal of Accounting and Economics, 11, 143-181.
Ding, Y., & Su, X. (2008). Implementation of IFRS in a regulated market. Journal of Accounting and
Public Policy, 27, 474-479.
Fan, J., & Wong, T. J. (2002). Corporate ownership structure and the informativeness of accounting
earnings in East Asia. Journal of Accounting and Economics, 33, 401-425.
Firth, M., Fung, P., & Rui, O. (2007). Ownership, two-tier board structure, and the informativeness of
earnings—Evidence from China. Journal of Accounting and Public Policy, 26, 463-496.
Gigler, F. B., & Hemmer, T. (2001). Conservatism, optimal disclosure policy and the timeliness of
financial reports. Accounting Review, 76, 473-493.
Gilson, R. J., & Gordon, J. (2003). Controlling controlling shareholders (Working Paper No. 228).
New York, NY: Columbia Law School, The Center for Law and Economic Studies.
Haw, I., Qi, D., Wu, D., & Wu, W. (2005). Market consequences of earnings management in response
to security regulations in China. Contemporary Accounting Research, 22, 95-140.
Healy, P., & Palepu, K. (2001). Information asymmetry, corporate disclosure, and the capital markets:

A review of the empirical disclosure literature. Journal of Accounting and Economics, 31, 405440.
Ho, S., & Wong, K. (2001). A study of the relationship between corporate governance structures and
the extent of voluntary disclosure. Journal of International Accounting, Auditing and Taxation, 10,
139-156.
Hutton, A. (2007). A discussion of ‘‘corporate disclosure by family firms.’’ Journal of Accounting
and Economics, 44, 287-297.
Jensen, M., & Meckling, W. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3, 305-360.
La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (1999). Corporate ownership around the world.
Journal of Finance, 54, 471-517.

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


640

Journal of Accounting, Auditing & Finance

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. (1998). Law and finance. Journal of
Political Economy, 106, 1113-1155.
Miller, D., Breton-Miller, I., Lester, R. H., & Cannella, Jr., A. (2007). Are family firms really superior performers? Journal of Corporate Finance, 13, 829-858.
Petersen, M. (2009). Estimating standard errors in finance panel data sets: Comparing approaches.
Review of Financial Studies, 22, 435-480.
Sun, Q., & Tong, W. (2003). China share issue privatization: The extent of its success. Journal of
Financial Economics, 70, 183-222.
Villalonga, B., & Amit, R. (2006). How do family ownership, control and management affect firm
value? Journal of Financial Economics, 80, 385-417.
Wang, D. (2006). Founding family ownership and earnings quality. Journal of Accounting Research,
44, 619-656.
Wei, Z., Xie, F., & Zhang, S. (2005). Ownership structure and firm value in China’s privatized firms:
1991-2001. Journal of Financial and Quantitative Analysis, 40, 87-108.

Zhang, J., & Zhang, H. (2004). Family firms’ performance, market value, and corporate governance.
World Economy, 11, 1-13.

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


Does Enterprise Risk
Management Increase Firm
Value?

Journal of Accounting,
Auditing & Finance
26(4) 641–658
Ó The Author(s) 2011
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0148558X11409160


Michael K. McShane1, Anil Nair1, and Elzotbek Rustambekov1

Abstract
Enterprise risk management (ERM) has emerged as a construct that ostensibly overcomes
limitations of silo-based traditional risk management (TRM), yet little is known about its
effectiveness. The scant research on the relationship between ERM and firm performance
has offered mixed findings and has been limited by the lack of a suitable proxy for the
degree of ERM implementation. Using Standard and Poor’s newly available risk management
rating, the authors find evidence of a positive relationship between increasing levels of TRM
capability and firm value but no additional increase in value for firms achieving a higher ERM
rating. Considering these results, the authors suggest directions for future research.

Keywords
enterprise risk management (ERM), traditional risk management (TRM), S&P ERM rating,
insurance firms, firm value
The crisis that started in 2007 with U.S. financial institutions caused a panic that rippled
across global markets and practically froze credit markets in 2008. Some have blamed the
crisis on a ‘‘failure of conventional risk management in financial institutions’’ (Fraser &
Simkins, 2010, p. 27). Others have extended the blame to include enterprise risk management (ERM), a new paradigm that had started to supplant conventional risk management,
especially within the large financial institutions at the heart of the crisis (Hampton, 2009,
p. 66).
The crisis has once again brought risk management to the forefront, not just among top
executives within firms but also among members of Congress and government regulators.
However, this concern about risk management had been gaining steam for several years.
For instance, Section 404 of the Sarbanes-Oxley Act of 2002 requires a top-down risk
assessment, which includes the identification of material risks on financial statements. In
2004, the New York Stock Exchange (NYSE) implemented new corporate governance
rules requiring audit committees of listed firms to be more involved in risk oversight. The

1

Old Dominion University, Norfolk, VA, USA

Corresponding Author:
Michael K. McShane, College of Business and Public Admin, Old Dominion University, 2124 Constant Hall,
Norfolk, VA 23529, USA
Email:

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


642

Journal of Accounting, Auditing & Finance

new rules have motivated many boards to require the review and approval of risk
management processes and top risk exposures by their audit committee.
In response to the financial crisis, in October 2008, Congress enacted the Emergency
Economic Stabilization Act, which created the Troubled Asset Relief Program (TARP), to
help troubled financial institutions. TARP stipulates that participating firms must certify
that executive compensation programs do not encourage excessive risk taking. In May
2009, Senators Schumer and Cantwell proposed legislation, the Shareholder Bill of Rights,
which requires public companies to create stand-alone risk committees comprised entirely
of independent directors who are responsible for the establishment and evaluation of risk
management practices. In October 2009, the Federal Reserve proposed guidance that places
responsibility on the board of directors for establishing appropriate incentive compensation
arrangements and effectively monitoring risk exposures created by incentive compensation
arrangements. New rules from the Securities and Exchange Commission effective February
28, 2010, require enhanced risk-related disclosures in proxy and annual statements.
Disclosure is required indicating the relationship of a company’s compensation policies and
practices to risk management and the board of directors’ leadership structure and role in
risk oversight.
Driven by this intense flurry of government and stock exchange activities related to risk
management within corporations, trade and business publications directed at top management are full of articles related to ERM, yet academic research in the area is still rare. We
believe that one main roadblock to this research is the difficulty in developing a valid and
reliable measure for the ERM construct. Beasley, Pagach, and Warr (2008) and Hoyt and
Liebenberg (2011) use the appointment of a chief risk officer (CRO) as a proxy for ERM
implementation, whereas Gordon, Loeb, and Tseng (2009) develop their own ERM index.
Results on the relationship between ERM and various measures of firm value have been
mixed. Beasley et al. (2008) investigate equity market reactions to senior management
appointments to oversee a firm’s ERM processes. Their results suggest firm-specific benefits of ERM. For nonfinancial firms, they find that market reactions to appointment
announcements are positively related to firm size and volatility of previous earnings but

negatively related to leverage and the ratio of cash to liabilities. They cannot make the
same claim for financial firms and argue that these firms may be more driven by other
demands for risk management, such as from regulators. Hoyt and Liebenberg (2011) found
a positive relationship between firm value and the appointment of a CRO. Gordon et al.
(2009) found that the relationship between ERM and firm performance depended on how
well ERM implementation was matched with firm-specific factors.
We use a newly available measure to investigate the relationship between the extent of
risk management implementation and firm performance. Since 2007, Standard and Poor’s
(S&P) has included a risk management rating as a component in its overall rating of insurance companies. The rating is a sophisticated and comprehensive index that assesses the
risk management culture, systems, processes, and practice within the insurer.
S&P assigns risk management ‘‘ERM ratings’’ over five categories, which we interpret
as indicating increasing levels of risk management sophistication ranging over three traditional risk management (TRM) levels and two ERM levels. Our study offers a unique setting to investigate the relationship between risk management and firm value for two
reasons. First, insurance firms are arguably leaders in implementing sophisticated risk
management programs; second, the year 2008 was characterized by extreme uncertainty in
which a superior risk management program should provide an advantage. Overall, our
results indicate a positive relationship between ‘‘ERM rating’’ and firm value as the rating

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


McShane et al.

643

increases over the first three categories—the first three categories are indicative of increasing levels of TRM—but no additional increase in firm value as the rating moves beyond
TRM into what we consider the ERM realm.
The article is organized as follows. First, in our literature review, we cover the evolution
of risk management research from ‘‘irrelevance’’ to ERM, focusing on the distinction
between TRM and ERM. Next, we motivate the variables we use in the study, including a
description of S&P’s new risk management rating for insurance companies. In our research

design section, we describe the data and model. After detailing the results, we conclude
with suggestions for future research.

Literature Review
Risk management has been a widely debated topic from the early days of finance research,
where it was considered irrelevant (Modigliani & Miller, 1958) under perfect market conditions. The debate continues today as firms adopt ERM programs and accounting1 and
finance academics begin to investigate their effectiveness. The following discussion covers
the evolution of this topic and distinguishes between what we call ‘‘TRM’’ and ‘‘ERM.’’

TRM
Some finance scholars responded to Modigliani and Miller’s (1958) ‘‘risk management
irrelevance principle’’ by citing capital market imperfections and proposing theories that
explain why risk management can increase firm value. In TRM research, scholars propose
that the existence of these imperfections allows risks to impose real costs on firms and that
risk management can increase firm value by reducing total risk, typically measured as
some type of volatility. Researchers have identified various value-increasing benefits of
risk management that can generally be classified as reduction in expected costs related to
the following: tax payments, financial distress, underinvestment, asymmetric information,
and undiversifiable stakeholders.2
Such studies help in understanding the reasons that firms decide to hedge risk and provide a theoretical justification for the link between risk management and firm value.
Allayannis and Weston (2001) directly investigate the relationship between risk management and firm value. Among their sample of large nonfinancial firms with foreign currency
exposures, Allayannis and Weston find that firms using foreign currency derivatives
had, on average, almost a 5% higher firm value than nonusers. More studies (see, for example, Bartram, Brown, & Conrad, 2009; Carter, Rogers, & Simkins, 2006; Graham &
Rogers, 2002; Nelson, Moffitt, & Affleck-Graves, 2005) followed showing a positive relationship between risk management, specifically hedging using derivatives,3 and firm value.
However, Guay and Kothari (2003) question the results of these studies after finding
that derivatives positions of most nonfinancial companies are too small to significantly
affect firm value. They surmise that derivatives usage is likely a fine-tuning mechanism for
a firm’s much larger overall risk management program, which includes other activities,
such as operational hedges. In support of this view, Jin and Jorion (2006) investigate oil
and gas firms and find no evidence that firms using derivatives to hedge their oil and gas

risk increase firm value relative to firms that do not hedge.
The studies mentioned up to now investigate risk management using derivatives to
hedge risk as a proxy for risk management activities. Other studies investigate the relationship between financial and operational hedging and, typically, proxy financial hedging by

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


644

Journal of Accounting, Auditing & Finance

derivative usage and operational hedging by geographic and segment diversification.
Chowdhry and Howe (1999) argue that derivatives are used to mitigate short-term currency
exposures, whereas operational hedges are better suited for handling long-run currency
exposures. Later studies examine whether financial and operational hedging are substitutes
or complements, and most find evidence of a complementary relationship (see, for example, Allayannis, Ihrig, & Weston, 2001; Kim, Mathur, & Nam, 2006; Pantzalis, Simkins, &
Laux, 2001).
Another strand of the finance literature argues that firms should not engage in any effort
to manage idiosyncratic risk. In the 1960s, building on Markowitz’s (1952) work on diversification and portfolio theory, various researchers (Lintner, 1965; Mossin, 1966; Sharpe,
1964; Treynor, 1961, 1962) developed the capital asset pricing model (CAPM). In this
model, investors are compensated only for bearing systematic (nondiversifiable) risk but
not for bearing idiosyncratic (diversifiable) risk. In other words, a firm’s cost of capital
(required rate of return) should depend only on the firm’s systemic risk, not the total risk
of the firm, because investors can eliminate the diversifiable risks of individual firms by
holding a well-diversified portfolio. The systemic risk of a firm is also called ‘‘market
risk’’ because this risk (and the firm’s cost of capital) depends on the covariance of the
firm’s security returns with the returns of the broad market, not on the firm’s overall volatility (variance). The systemic risk of the firm is represented by the familiar b in the CAPM.
An implication of CAPM is that firms should not use risk management to reduce firmspecific risks because investors can eliminate firm idiosyncratic risks through
diversification.
However, several researchers countered with asset pricing models in which idiosyncratic

risk does matter, for example, because investors may hold undiversified portfolios (see, for
example, Goyal & Santa-Clara 2003; Green & Rydquist, 1997; Levy, 1978; Merton, 1987).
Froot and Stein (1998) develop a capital allocation/structure model for financial institutions4 in which information-intensive assets cannot be frictionlessly hedged. Froot (2007)
builds on this model to include customer aversion to insolvency risk, which is an important
consideration for financial institutions because their customers typically have a greater concern about solvency risk than do investors. Overall, an implication is that in deciding
whether to allocate capital for an investment, the decision should reflect the covariation of
the investment’s risk with the firm’s existing portfolio of risks.

ERM
Traditionally, risk management has been compartmentalized and uncoordinated within a
firm. Risk had been managed in silos with corporate risk managers focusing on pure risks,5
whereas the treasury department used derivatives to reduce financial risks, such as interest
rate, credit, market, and foreign exchange risk. ERM attempts to deal with additional risks
such as operational or strategic risks. The goal of ERM is the coordinated management of
all risks faced by a firm, whether it is risk related to corporate governance, auditing, supply
chains, distribution systems, IT, or human resources. Unlike TRM’s silo-based risk management, the purpose of ERM is to gain a systematic understanding of the interdependencies and correlations among risks. A fundamental concept of ERM is the aggregating of
risks into portfolios, then hedging the residual risk, which is more efficient and value maximizing than dealing with each risk independently. Applying concepts of portfolio theory,
ERM can increase firm value because the risk of an aggregate portfolio should be less than

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


McShane et al.

645

the sum of the individual risks if the risks are not 100% correlated, especially if natural
hedges exist.6
In a call for risk management research that focuses on the coordination and strategic
allocation of risk, Stulz (1996) proposes that academic theory expand beyond considering

that the goal of risk management is ‘‘variance minimization.’’ In other words, the goal of
risk management should not be to reduce total risk but to allocate risks to play on a firm’s
strengths. A basic concept of ERM is that a firm should reduce exposure to risk in areas
where it has no comparative information advantage and exploit risks in areas where it has
an advantage, meaning that total risk can possibly increase under ERM risk allocation.
Schrand and Unal (1998) posit that corporate managers should coordinate riskmanagement activities by hedging exposure to activities in which they are likely to earn
zero economic rents (homogeneous risks), such as investments in efficient markets, while
increasing exposure to core-business activities (Barney, 1991) in which they enjoy comparative information advantages. Such a coordinated approach can generate a decreasing,
neutral, or increasing effect on total firm risk. Since Schrand and Unal (1998), there has
been very little work related to coordinated risk management in the finance literature.
Recently, Zhang, Cox, and McShane (2011) use insurance industry data to investigate the
coordination of risks across completely different functions of the enterprise while controlling for other factors that affect hedging decisions. They consider investments to be a
homogeneous risk for insurers and underwriting to be a core-business risk, and find evidence that insurers are coordinating risk management by hedging investment risk to take
on more underwriting risk.
A few articles have indirectly investigated the determinants of ERM implementation
among firms. Liebenberg and Hoyt (2003) investigate the determinants of ERM adoption,
using the appointment of a CRO as a proxy for ERM implementation. Their main finding
is that more leveraged firms are more likely to appoint a CRO. In a similar study, Pagach
and Warr (2011) find that firms with more leverage, higher earnings volatility, poorer stock
performance, and a CEO whose compensation increases with stock volatility are more
likely to have a CRO. Using survey data, Beasley, Clune, and Hermanson (2005) find
ERM implementation in their sample of firms to be positively related to factors such as the
presence of a CRO, firm size, and whether the firm is in the insurance or banking industry.
Two studies indirectly investigate the relationship between ERM implementation and
firm value. Hoyt and Liebenberg (2011) find a positive relationship between firm value and
the appointment of a CRO. In an event study of the market reaction to the appointment of
senior executives to oversee a firm’s ERM process, Beasley et al. (2008) find firm-specific
benefits of ERM for nonfinancial firms, but not for financial firms. Gordon et al. (2009)
develop their own ERM index and find that the relationship between ERM and firm performance is conditional on the match between ERM implementation and firm-specific factors.
Beasley et al. (2008) indicate that a limitation of using the CRO variable is that it does

not capture the extent of ERM program implementation. In the next section, we describe
the measure used in this study, which we believe comprehensively captures the complexity
of ERM and reflects the extent of its implementation.

Variable Motivation
Our risk management variable is novel, but the other variables are motivated by the previous risk management literature.

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012


646

Journal of Accounting, Auditing & Finance

Dependent Variable (Firm Value)
Our dependent variable is firm value, which we proxy for using Tobin’s Q, the most commonly used measure of firm value in empirical risk management studies (Smithson &
Simkins, 2005). We calculate Tobin’s Q as the market value of equity plus the book value
of liabilities divided by the book value of assets. This version of Tobin’s Q is suitable for
insurance companies because the book value of an insurer’s assets is a good approximation
of replacement costs (Cummins, Lewis, & Wei, 2006; Hoyt & Liebenberg, 2011).

Independent Variable of Interest (ERM Rating)7
Financial rating firms, such as S&P, rate the ability of a firm to pay back creditors. A firm
with a higher rating will have lower borrowing costs, which should translate to higher firm
value, all else equal. This effect will be intensified for insurers because the policyholder is
a contingent debtholder. In essence, policyholders are both customers and main creditors of
insurance companies. As described in McShane, Cox, and Butler (2010), insurers with
higher ratings command higher premiums because they are perceived as safer by policyholders. Premiums are the main revenue source for an insurance firm, implying that a
higher credit rating leads to higher returns, and supporting empirical evidence has been
found (see, for example, Cummins & Nini, 2002).

We use the new risk management rating from S&P as a proxy for degree to which an
insurer has implemented a risk management program.8 S&P rates the financial strength of
insurers based on eight components and gradually started to add the newest component,
ERM, for insurers in 2006.9 S&P investigates the following to determine the ‘‘ERM
rating’’ for each insurer: risk management culture, risk control processes, emerging risks
management, risk and economic capital models, and strategic risk management. At the base
of the ERM program is the firm’s risk management culture. A major S&P consideration in
this area is the importance of the risk management process to C-suite executives because
ERM only works if the ‘‘tone is set at the top.’’ The governance structure should reflect
the influence of risk and risk management considerations on corporate-wide decision
making, including the transparency with which the risk management philosophy is communicated across the organization and the extent to which risk management influences management compensation and budgeting.
Next are the three pillars of the ERM program: (a) the ability of the insurer’s risk control processes in identifying, analyzing, and keeping losses within defined risk tolerances;
(b) the capability of the insurer to scan the environment to anticipate and prepare for emerging risks; and (c) the effectiveness of the insurer’s risk and economic capital models to
realistically provide insight into possible risks facing the insurer and support to other ERM
processes.
A strong risk management culture at the base and the three well-designed pillars are
essential to support the firm in achieving effective strategic risk management for which a
key consideration is the extent to which the insurer has integrated risk management with
core strategic planning processes. Firms with a higher ERM rating should have an advantage in anticipating and dealing with the next big risk, lower volatility of earnings, and
greater ability to allocate capital to attain higher risk-adjusted returns.
S&P places each insurer into one of five ‘‘ERM rating’’ categories. A weak ERM program lacks reliable loss control systems for one or more major risks. An adequate ERM
program has reliable loss control systems but may still be managing risks in silos instead of

Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012