Information Asymmetry, Mutual Funds and Earnings Management: Evidence From China

China Journal of Accounting Research 6 (2013) 187–209
Contents lists available at ScienceDirect
China Journal of Accounting Research

journal homepage: www.elsevier.com/locate/cjar
Information asymmetry, mutual funds and

earnings management: Evidence from China
Yunhao Dai, Dongmin Kong ⇑, Li Wang
School of Economics, Huazhong University of Science and Technology, China
A R T I C L E I N F O A B S T R A C T
Article history: This paper investigates how information asymmetry and mutual fund owner-
Received 19 August 2012 ship affect listed companies’ earnings management. We show that (1) reducing
Accepted 25 March 2013 information asymmetry improves firms’ earnings management behavior; (2)
Available online 30 April 2013 relative to short-term mutual funds, long-term mutual funds promote earnings
quality by adopting a monitoring role; and (3) by dividing firms into high/low
information asymmetry groups, we find that the information environment sig-
JEL classification: nificantly increases the effect of long-term mutual funds on firms’ earnings
G12 management. In this paper, we provide new evidence for the role that institu-
G14 tional investors play in a typical emerging capital market. Our results have
G18 clear policy implications: to increase earnings quality, it is essential to improve
M41 information transparency and develop long-term institutional investors.
Ó 2013 Production and hosting by Elsevier B.V. on behalf of China Journal of
Keywords: Accounting Research. Founded by Sun Yat-sen University and City Univer-
Information asymmetry sity of Hong Kong.
Mutual funds
Earnings management
1. Introduction
Earnings information is an important basis for investors’ decision making. In recent years, due to a series of
significant accounting fraud and violation cases both at home and abroad, the issue of earnings management
has attracted considerable attention from academics and practitioners. However, the focus of most studies is
⇑ Corresponding author. Tel.: +86 15927068886.

E-mail addresses: [email protected] (Y. Dai), [email protected] (D. Kong), [email protected] (L. Wang).
Production and hosting by Elsevier
1755-3091/$ - see front matter Ó 2013 Production and hosting by Elsevier B.V. on behalf of China Journal of Accounting Research.
Founded by Sun Yat-sen University and City University of Hong Kong.
http://dx.doi.org/10.1016/j.cjar.2013.03.001
188 Y. Dai et al. / China Journal of Accounting Research 6 (2013) 187–209
limited to how to measure earnings management behavior, the factors that influence earnings management
and the economic consequences of earnings quality.
The majority of related studies overlook the effects of information asymmetry and institutional ownership
and their influence on company earnings management issues. Information asymmetry usually refers to the
inequality in the amount of information held by different market participants. It is well known that accounting
information provides relatively accurate financial information to market participants. This information helps
investors to understand the company’s operating activities and thus reduces the information asymmetry
between investors and management (Armstrong et al., 2010; Bhattacharya et al., forthcoming). The informa-
tion environment is likely to impose certain externalities on accounting information: a company’s degree of
information asymmetry can be decreased by an effective information disclosure system, which provides a
transparent information environment for the company’s financial reports and accounting earnings quality.
Managers under stronger supervision are more likely to provide high-quality accounting reports that further
promote improvements in corporate earnings quality (Hunton et al., 2006; Xia and Lu, 2005).
Institutional investors also play an increasingly important role in capital markets. Stock markets in the
United States, for example, are largely dominated by institutional investors. Similarly, institutional investors
have developed rapidly in China since the first securities investment funds were established in 1998.1 Today,
Chinese institutional investors are among the top 10 shareholders in more than half of the listed companies.
Can the increasing number of institutional investors improve the governance structure of companies? Institu-
tional investors are professionals who have advantages in terms of financial support and information discov-
ery – they are good at detecting companies’ actual operating conditions and supervising management.
However, there is considerable academic controversy over how institutional investors affect earnings manage-
ment. On the one hand, they may use their professional advantage to oversee the management of listed com-
panies and participate in the corporate governance process, thereby effectively inhibiting accruals-based
earnings management behavior (Prowse, 1990; Brous and Kini, 1994; Warfield et al., 1995; Cheng, 2006).
On the other hand, there are concerns about their negative influence, as their herd-like and short-sighted
behavior can exacerbate the extent of earnings management and thus reduce earnings quality (Graves,
1988; Porter, 1992; Deng and Tang, 2010).
Compared with capital markets in Western countries, Chinese listed companies have suffered from the
‘dominance’ of non-tradable shares, as a result of which institutional investors fail to play a full role. Zhao
and Zheng (2002) argue that the relatively small proportion of Chinese institutional investors, together with
the insufficiency of market information disclosure mechanisms and regulatory instruments, increases the infor-
mation search and interpretation costs for Chinese institutional investors. Therefore, the extent of the infor-
mation asymmetry between investors and listed companies is likely to further influence the relationship
between institutional investors and earnings management.
In summary, can the level of transparency in the information environment influence company earnings
management behavior in the Chinese capital market? As the enthusiasm for Chinese mutual funds continues
to heat up, do the diverse types and characteristics of mutual funds have different effects on corporate gover-
nance? What is the combined effect of these two important factors on earnings management? Our aim in this
paper is to explore these as yet unresolved issues.
This paper is innovative in the following two respects. First, although some scholars examine the relation-
ships between the degree of information asymmetry, institutional ownership and earnings management, most
studies are based on the mature markets of developed countries. As there is still a lack of comprehensive
research into these three factors in China, in this paper we investigate their interaction in the emerging Chinese
stock market. Second, previous studies generally analyze overall mutual fund behavior, but they either do not
classify funds according to their different investment characteristics or use biased classifications (Brown and
Goetzmann, 1997; Sun et al., 2012). To distinguish between the corporate governance behavior of different
1
There are various types of institutional investors. We choose securities investment funds (mutual funds) as a proxy variable for
institutional investors in this study because mutual funds are the largest institutional investors in China and are the main securities
investment funds affecting corporate governance, and because the data is relatively easy to access. Hence, unless otherwise stated, when
referring to institutional investors, we mean mutual fund investors.
Y. Dai et al. / China Journal of Accounting Research 6 (2013) 187–209 189
types of funds, we reclassify mutual funds as long-term and short-term funds using the method proposed by
Yan and Zhang (2009).
The paper is organized as follows. Section 2 presents the literature review and research hypotheses. Section 3
describes the data sources, variable definitions and empirical models. Section 4 presents the econometric mod-
els and results of the empirical analysis. Section 5 presents the results of additional tests. Section 6 summarizes
and concludes the paper.
2. Literature review and research hypotheses
Earnings quality is a measure of the profitability of listed companies. It indicates the degree of match
between a company’s profitability and its cash flow. If they are not well matched, there are potential operating
problems. Earnings management occurs when management tries to control or adjust reported accounting
earnings information to maximize their own interests. It is clear that earnings management will cause the com-
pany’s earnings quality to deviate from its true level. Information asymmetry usually refers to inequalities
between the information held by market participants, which can affect investors’ decision-making. Studies
conducted both in China and abroad show that a company’s earnings management is closely related to its
information environment. Both Dye (1988) and Trueman and Titman (1988) find that information asymmetry
between shareholders and management is a necessary condition for the existence of earnings management.
Schipper (1989) defines earnings management as management’s manipulation of the disclosure of financial
reporting by purposefully using their information advantage to seek private interests, which often go against
the interests of the business owners. When information asymmetry is high, stakeholders do not have sufficient
resources, incentives or access to relevant information to monitor managers’ actions, which gives rise to the
practice of earnings management. Richardson (2000) provides empirical evidence that information asymme-
try, as measured by the bid-ask spread and analysts’ forecast dispersion, is positively related to the level of
earnings management. Hunton et al. (2006) uses experimental methods to investigate whether greater trans-
parency reduces earnings management attempts. In the experiment, 62 financial executives and chief executive
officers decide which available-for-sale security to sell from a portfolio under different levels of transparency of
income reporting and projected earnings. The results suggest that more transparent reporting requirements
can reduce earnings management attempts or change the focus of earnings management attempts to less visible
methods. Jo and Kim (2007) examine the relationship between disclosure frequency and earnings management
and the effect of this relationship on post-issue performance, using a sample of seasoned equity offerings
(SEOs). They find that firms with extensive disclosure are less likely to face information problems, leading
to less earnings management and better post-issue performance. The above discussion suggests that foreign
scholars generally agree that earnings management increases as the degree of information asymmetry
increases.
The Chinese literature concerning earnings management and the information environment is still relatively
sparse. Yang (2005) investigates the relationship between information disclosure and earnings management
using a client–agency model, which suggests that information forecasts can decrease the information asymme-
try between managers and owners. Mangers should face an additional cost for their deliberate forecasting of
inaccurate information, so that managers’ information forecasts can reduce the possibility of earnings man-
agement. Xia and Lu (2005) use listed companies’ condemned announcements as a proxy for information dis-
closure and their results show that the degree of earnings management and the quality of information
disclosure are negatively related, implying that listed companies may lower information disclosure quality
to conceal their earnings management. Fang and Hong (2007) conclude that the quality of corporate disclo-
sure has an effect on analysts’ behavior. Specifically, it improves the accuracy of predictions and reduces the
dispersion between analysts, thus ensuring that earnings data are closer to reflecting a company’s actual profit.
From the results of studies conducted both in China and abroad, we conclude that the information asym-
metry between investors and listed companies leads to management speculation. The accounting earnings
information advantage of listed companies’ management drives them to adjust their accounting records
and information disclosure content to achieve their own personal interests or to benefit the minority. There-
fore, improving the transparency of information disclosure to reduce the degree of information asymmetry
can effectively regulate major shareholders’ violation of interests and improve corporate governance.
Furthermore, it can inhibit earnings management and promote companies’ accounting reporting quality to
reduce investment risk. Accordingly, we propose the following hypothesis:
Hypothesis 1. The information asymmetry of listed companies is positively related to the degree of earnings
management
Institutional investors are often characterized as “sophisticated investors” who have advantages in acquir-
ing and processing information compared to individual investors (Bushee, 1998; Bartov et al., 2000; Jiambalvo
et al., 2002). However, domestic and foreign scholars continue to disagree over their specific role in improving
reporting quality. As developing capital markets are not perfect, with serious information asymmetry and a
lack of funds and technical expertise, institutional investors have little motivation or ability to participate
in company governance and thus act as “traders” rather than “owners”. Foreign studies consider institutional
investors to be more concerned about current profits than corporate governance, as most institutional inves-
tors are still short-term speculators who pay little attention to long-term investment. When a listed company
performs poorly, institutional investors tend to “vote with their feet” by selling stock. To prevent the loss of
these important institutional investors, the company’s management may manipulate earnings to increase
reported earnings. As a result, a growing number of significant shareholders who are short-sighted and spec-
ulative induce companies to provide low-quality accounting information in an attempt to maintain their
attractiveness to institutional shareholders. For instance, Graves (1988) believes that fund managers look
mainly for short-term gains from their equity investments. In response to a desire for advancement and job
security, institutional investors encourage managers to forego an increase in risky and long-term investments
in favor of increasing their short-term financial profitability. Due to the information asymmetry between man-
agers and investors, Froot et al. (1992) show that it is more cost-effective for institutions to invest based on
short-term performance, rather than valuing the long-term prospects of the firms in their diversified portfolios.
In conclusion, institutional investors may lead to lower-quality accounting information because their frequent
trading and short-term focus may encourage managers to exercise short-sighted discretion in reporting a firm’s
financial performance.
As the development of foreign capital markets involves not only further improvements in laws and regu-
lations, but also the growing scale of institutional investors, the cost of institutional investors “voting with
their feet” is increasing. Compared with exiting the market negatively through stock-selling, institutional
investors increasingly take an active part in corporate governance by proposing shareholder bills or soliciting
proxy voting rights, because they are more capable of achieving higher returns by monitoring managerial
behavior than individual investors. Correspondingly, some foreign scholars support the monitoring effect
argument, which assumes that institutional ownership will bring high-quality accounting information. Prowse
(1990) finds that institutional investors that take part in corporate governance play a supervisory role in earn-
ings management. Brous and Kini (1994) suggest that higher levels of institutional ownership are associated
with more effective monitoring of the use of cash a firm obtains from equity issues, due to their higher own-
ership stake in the firm. Warfield et al. (1995) find that managerial ownership is positively associated with the
explanatory power of earnings for predicting returns and inversely related to the magnitude of accounting
accrual adjustments. Bushee (1998) holds that more sophisticated institutional investors remove incentives
for managers’ opportunistic behavior through closer monitoring of managerial behavior, either through expli-
cit governance activities2 or through the implicit collection and dissemination of information in the stock mar-
ket.3 According to this view, institutional investors play an active role in improving financial reporting quality
because they are willing to monitor and discipline managers, thus ensuring that managers maximize the long-
term value of the firm rather than their own interests. For example, there is evidence to suggest that firms with
higher AIMR disclosure rankings have greater institutional ownership (Bushee and Noe, 2000) and managers
are less likely to cut R&D to reverse an earnings decline when institutional ownership is high. Chung et al.
2
For example, in the early 1990s, some institutional investors (i.e., the California Public Employees Pension Fund and J.P. Morgan)
lobbied for the removal of CEOs at several large, poor performing firms, including Kodak, IBM, Westinghouse, Borden, American
Express and GM (Kahn and Winton, 1998).
3
Previous studies find that relatively more future earnings information is impounded in stock prices (Jiambalvo et al., 2002) and
inefficient pricing of earnings is reduced (Bartov et al., 2000) when institutional ownership is high.
(2002) also find evidence that is consistent with institutional investors monitoring and constraining the self-
serving behavior of corporate managers.
The recent foreign literature suggests a new method for studying this issue, by reclassifying institutional
investors based on their historical investing characteristics. Following Bushee (1998), Liu and Peng (2006)
group mutual funds into three categories4 and find that the accuracy of accruals is negatively related to the
shareholdings of short-term institutional investors, but positively related to the shareholdings of long-term
institutional investors. This suggests that short-term investors decrease earnings quality, whereas long-term
investors play a role in monitoring management. Koh (2007) indicates that long-term institutional investors
can mitigate aggressive earnings management, whereas transient institutional ownership is only associated
with aggressive earnings management among firms that need to meet their earnings benchmarks.
With the rapid development of institutional investors in China, several domestic researchers examine the
role of institutional investors with mixed results. Cheng (2006) finds that timeliness is positively associated
with the shareholdings of institutional investors, whereas the extent of earnings management is negatively
associated with the shareholdings of institutional investors. However, he does not consider the endogeneity
problem between earnings quality and institutional shareholdings. Gao and Zhang (2008) use data from
Chinese listed companies to show that institutional investors are involved, to an extent, in corporate
governance and thus restrain earnings management. Huang (2009) finds an inverse-U relationship between
the shareholdings of institutional investors and earnings management. Additionally, Deng and Tang (2010)
query whether institutional investors can restrain earnings management. They find that the shareholdings
of institutional investors are positively related to earnings management and furthermore, this relationship
is more prominent in state-owned enterprises. Their results suggest that the majority of institutional investors
in China are still myopic, and increase rather than reduce earnings management.
According to the above discussion, we find that the question of whether institutional investors are short-
sighted is a contentious issue, especially in the emerging capital market of China. Supposing the aim of insti-
tutional investors is to seek profit maximization, then we have reason to conjecture that institutional investors
are motivated to support management to manipulate earnings. Nevertheless, several studies consider how the
characteristics of institutional investors affect the extent of firms’ earnings management by classifying institu-
tional investors into different types. To address these issues, we propose Hypothesis 2:
Hypothesis 2. Relative to short-term mutual funds, the monitoring effect of long-term mutual funds on listed
companies’ earnings management is stronger
To date, most studies consider how the information environment or institutional investors affect earnings
management. However, they rarely study the issue by incorporating the relationship between information
asymmetry and institutional investors. As Ramalingegowda and Yu (2012) suggest, the higher the sharehold-
ings of institutional investors, the higher the earnings conservatism (an important component of earnings
quality). When the extent of information asymmetry is higher, the relationship between the shareholdings
of institutional investors and earnings conservatism is more prominent. Although the mutual fund industry
has recently experienced fast development in China’s capital market, gaps still exist compared with the same
industry in developed capital markets. The role of mutual funds in China is still limited to stabilizing market
efficiency and monitoring the behavior of managers. With respect to the cost when mutual funds play a role in
monitoring, the higher the information asymmetry, the higher the cost of supervision. Once the cost exceeds
the potential profit, mutual funds will no longer supervise the firm and will vote with their feet (Coffee, 1991).
However, when the extent of information asymmetry is low, the cost of monitoring is also low and mutual
funds can still gain high potential profits.5 Therefore, compared with developed capital markets, do mutual
funds in China behave differently when they face firms with different levels of information asymmetry? For
instance, do long-term mutual funds tend to have a greater monitoring role in comparison with short-term
mutual funds when the extent of information asymmetry is low? Considering that the extent of information
4
Bushee (1998) classifies institutional investors into three types. Due to data limitations, we choose the classification method used by
Yan and Zhang (2009).
5
We thank the referees for this suggestion.
asymmetry will affect the investment behavior of mutual funds, we group firms according to the extent of
information asymmetry and then study the effect of mutual fund ownership on earnings management. We pro-
pose Hypothesis 3:
Hypothesis 3. When there is low information asymmetry, the monitoring effect of long-term mutual funds on
listed companies’ earnings management is stronger
3. Data and research design
3.1. Sample selection and data sources
As the number and size of institutional investors has increased since 2004, we choose all companies listed on
the Chinese A-share market during the 2004–2010 period. Financial data and institutional investor sharehold-
ings data are obtained from the WIND Database, and stock return and corporate governance data are drawn
from the CCER Database. We initially group all listed companies into 13 industries following the CSRC
industry classification standard. However, since most companies belong to manufacturing industries, we
use subcategories for the manufacturing industry, which results in a total of 22 industry categories.
During the sample selection process, we exclude financial companies, companies listed on the GEM and
“Special Treatment” or “Particular Transfer” companies. We also exclude companies with missing data. To
minimize the influence of outliers, the top and bottom 1% of the variables are winsorized. In total, we have
7286 firm-year observations in our sample.
3.2. Variable definitions
We use empirical models to study the relationship between earnings management and information asym-
metry and long-term or short-term mutual funds. The variables we use are defined as follows.
3.2.1. Earnings management (EM)

There is controversy in the existing literature regarding the definition and measurement of earnings man-
agement. Previous studies measure the extent of earnings management in different ways, such as timeliness,
smoothness and persistence (Hunt et al., 1996; Lev and Thiagarajan, 1993). However, these methods generally
measure the earnings management of a particular group of companies, but not individual companies. The
existing literature usually uses the relationships between earnings, operating cash flows and accruals to mea-
sure firm-level earnings management. We also follow this method by adopting the model of Dechow and
Dichev (2002; hereafter referred to as the DD model) model and extending Ball and Shivakumar’s (2005)
piecewise nonlinear regression model (hereafter referred to as the BS model).6
The DD model considers the relationship between accruals and cash flows, and adopts the mapping rela-
tionships between current accruals at t and cash flows in periods t 1, t and t + 1, to measure earnings man-
agement, as follows:
DWC t ¼ a0 þ a1 CFOt1 þ a2 CFOt þ a3 CFOtþ1 þ et ð1Þ
The principle of the DD model is to measure the degree of match between accruals and cash flows. When
the part of accruals (et) that cannot be explained by operating cash flows is relatively stable, the accruals qual-
ity is high and investors’ uncertainty over predictions of future operating cash flows is low. This helps to
ensure the high quality of information concerning future cash flows. Therefore, the extent of earnings man-
agement is low, and vice versa.
The DD model requires 7 continuous years of information from annual reports. In Eq. (1), DWCt is total
accruals at t, measured as operating profit minus operating cash flow; and CFOt1, CFOt and CFOt+1 represent
6
We do not report the results of the BS model as they are consistent with those of the DD model. We provide a detailed description of
the measurement method for the BS model in Appendix A. Furthermore, the Pearson correlation between |BS_eq| obtained from the BS
model and |DD_eq| obtained from the DD model is 0.94, so using them as the same proxy measure is reliable.
the operating cash flows in periods t 1, t and t + 1, respectively. All of the above variables are scaled by aver-
age total assets to eliminate the size effect. Using Eq. (1), we regress by each industry and each year separately to
obtain the residual et, which represents abnormal accruals. In the DD model, there are generally two ways to
calculate earnings management. The first uses the absolute value of the residual as the proxy for earnings man-
agement. The second uses the standard deviation of the residual from t 4 to t for each company to obtain the
measure of earnings management. If the volatility of abnormal accruals over 5 years is weak, then the risk of
earnings is low, and so is the extent of earnings management.
Koh (2007) points out that there is a difference between negative and positive earnings management.
Negative earnings management occurs when management is concerned that firm performance will deteriorate
in the future, so they use accounting conservatism to hide the firm’s current performance and thus artificially
inflate future profits. Positive earnings management is due to company management attempting to hide the
firm’s current poor performance by artificially inflating profits. Although both types of earnings management
deceive investors and reduce earnings quality, their motivation and channels of implementation are different.
In this paper, we attempt to distinguish the behavior of mutual funds by differentiating between positive and
negative earnings management, so we choose the original value of et as the proxy for earnings management,
which is defined as DD_eq. If DD_eq > 0, then the accruals profit based on working capital exceeds the actual
accruals profit calculated by 3 continuous years of operating cash flows, suggesting that accrual profit may be
increased artificially, i.e. positive earnings management DD_eq(+). DD_eq() reflects negative earnings man-
agement. A higher value of |DD_eq| indicates higher earnings management.
3.2.2. Information asymmetry (InfoAsy)—based on the standard deviation of idiosyncratic risk (SD)
Following Dierkens (1991) and Kong and Fu (2005)7, for each company in each year, we regress the market
model. Then, we calculate the standard deviation of the difference between the actual return and the normal
return estimated by the market model and use it as the proxy for information asymmetry. The formula is as
follows:
e0t ¼ rit ^rit ð2Þ
where rit is the actual daily return of companies in period t, and ^rit is the normal daily return estimated by the
following market model:
^rit ¼ ^ ^ mt
ai þ br ð3Þ
where ^ ^ are estimated using the stock return data for companies in period t 1.
ai and b
3.2.3. The classification of institutional investors (Insti) as long- and short-term mutual funds (long/short_ratio)
Previous studies rarely group institutional investors according to their characteristics, but some studies
show that there is a bias error in the investment style of mutual funds (Brown and Goetzmann, 1997; Sun
et al., 2012). Therefore, we do not adopt the classification directly from the database, but follow Yan and
Zhang (2009) in classifying mutual funds into long- and short-term investors based on their portfolio turnover
over the past year.
First, we calculate aggregate purchases and sales for each mutual fund k:
7
We thank the referee for this suggestion. Although previous studies propose different measures of information asymmetry, we choose
to follow Dierkens (1991), who suggests that the standard deviation based on idiosyncratic risk actually reflects the uncertainty of firm
value, and this uncertainty stems from the information asymmetry between the company management and investors. When the extent of
information asymmetry is higher, the disagreement between investors and management regarding firm value is higher, and the
management have more chance to hide idiosyncratic information. As the extent of information asymmetry reduces, investors and
management have consistent opinions about firm value and uncertainty is reduced. Although the bid-ask spread can also proxy for
information asymmetry, it tends to reflect the information asymmetry in trading, which is more likely to be due to the different private
information owned by investors. As this paper is interested in the information asymmetry between the company management and
investors, we believe it is feasible to measure information asymmetry following Dierkens (1991) and Kong and Fu (2005).
X
Nk
CR buy k;t ¼ jS k;i;t P i;t S k;i;t1 P i;t1 S k;i;t DP i;t j ð4Þ
i¼1
S k;i;t >S k;i;t1
X
Nk
CR sellk;t ¼ jS k;i;t P i;t S k;i;t1 P i;t1 S k;i;t DP i;t j ð5Þ
i¼1
S k;i;t 6S k;i;t1
where CR_buyk,t and CR_sellk,t are mutual fund k’s aggregate purchases and sales for period t, respectively;
Pi,t and Pi,t1 are the share prices for stock i at the end of period t and t 1; Sk,i,t and Sk,i,t1 are the number
of shares of stock i held by investor k at the end of periods t and t 1, respectively; and DPi,t is the price
change from t 1 to t. If the number of shares of stock i held by mutual fund k is less in period t than it
is in period t 1, then mutual fund k sold stock i, and the capital change in stock i will be accounted for
in aggregate sales, otherwise the capital change will be accounted for in aggregate purchases.
Then, we calculate mutual fund k’s churn rate (CR) for period t:
minðCR buy k;t ; CR sellk;t Þ
CRk;t ¼ PN k S k;i;t P i;t þS k;i;t1 P i;t1 ð6Þ
i¼1 2
and obtain mutual fund k’s average churn rate based on the turnover over the past year:
1
AVG CRk;t ¼ ðCRk;t þ CRk;t1 Þ ð7Þ
2
Given the average churn rate measure, for each period t, we sort all mutual funds into three tertile portfo-
lios based on AVG_CR. Those ranked in the bottom tertile are classified as long-term mutual funds and those
ranked in the top tertile are classified as short-term mutual funds. Finally, for each stock, we define the long-
term (short-term) institutional ownership (hereafter long_ratio and short_ratio) as the ratio of the number of
shares held by long-term (short-term) mutual fund investors and the total number of shares outstanding.
3.2.4. Other control variables

There is extensive evidence in the literature that effective corporate governance contributes to the improve-
ment of listed companies’ accounting information. There are two views on this issue. The first is agency theory,
which examines the relationship between management ownership, corporate governance and information con-
tent. Klassen (1997) and Warfield et al. (1995) both find that the pressure from capital markets induces com-
panies with low management ownership to choose accounting measures to increase earnings, which
consequently reduces the information content of earnings. The second concerns outside blockholders and
focuses on the role of corporate governance and its effect on earnings quality. Kaplan and Minton (1994)
and Kang and Shivdasani (1995) find that because outside blockholders have strong incentives to obtain
information about companies and monitor management, they play a positive role in corporate governance.
Therefore, to control for the influence of corporate governance on earnings management, we choose five cor-
porate governance variables as follows:
Duality: A dummy variable that equals 1 if the chairman and the CEO are the same person, and 0
otherwise;
Out_ratio: The proportion of independent directors on the board;
Top1: The proportion of shareholdings of the largest shareholder;
Top2_10: The aggregate proportion of shareholdings by the second to the 10th largest shareholders; and
CR_5: The aggregate proportion of shareholdings of the top five shareholders.
In addition, financing needs, ultimate controllers, auditors and other firm factors may affect earnings man-
agement and information asymmetry. Thus, we also use the following control variables:
Offering: Following Ljungqvist et al. (2009), we use a corporate financing needs dummy variable, which
equals 1 if the company has an SEO, rights offering or debt issue in period t, and 0 otherwise;
Table 1
Descriptive statistics for |DD_eq|.
Year Obs Mean SD Min P25 P50 P75 Max
2004 934 0.034 0.042 0.000 0.008 0.019 0.041 0.259
2005 1012 0.032 0.040 0.000 0.007 0.017 0.039 0.259
2006 1025 0.035 0.044 0.000 0.009 0.021 0.041 0.259
2007 1057 0.041 0.049 0.000 0.011 0.025 0.049 0.259
2008 1095 0.038 0.047 0.000 0.008 0.021 0.045 0.259
2009 1072 0.039 0.049 0.000 0.010 0.023 0.046 0.259
2010 1091 0.037 0.046 0.000 0.011 0.024 0.041 0.259
Notes: This table reports descriptive statistics for |DD_eq| from 2004 to 2010. Obs: sample size for each year. Mean: sample average for
each year. SD: sample standard deviation for each year. Min: sample minimum value for each year. P25: sample first quartile for each year.
P50: sample median for each year. P75: sample third quartile for each year. Max: sample maximum value for each year.
SOE: A dummy variable that equals 1 if the company is a state-owned enterprise, and 0 otherwise;
Audit: A dummy variable that equals 1 if the audit opinion is unqualified, and 0 otherwise;
TopAudit: A dummy variable that equals 1 if the company’s audit firm is a Big 4 firm, and 0 otherwise.8
LnSize: Natural logarithm of total assets at the end of the year;
LnMB: Natural logarithm of the market-to-book ratio; and
To: The turnover of listed companies in each year, because Hakim et al. (2008) find that high earnings qual-
ity is positively associated with high liquidity.
4. Empirical analysis
4.1. Descriptive statistics
We calculate abnormal accruals for each firm following the DD model, namely DD_eq. Table 1 presents the
summary statistics for each year. It shows that since 2004, the earnings quality of listed companies in China
first declines and then increases, and peaks in 2007.
Table 2 presents the summary statistics for the key variables. The table shows that the average |DD_eq| is
0.037, and the range is between 0 and 0.259. The average |BS_eq| is 0.036, and the correlation coefficient of
|DD_eq| and |BS_eq| is 0.94, thus it is reliable to use them as the same proxy measure. The mean of SD, which
is a proxy for information asymmetry, is 0.024, and the standard deviation is 0.006, suggesting that differences
in information asymmetry exist among companies. Furthermore, the average proportion of shareholdings of
long-term mutual funds is 0.021 and for short-term mutual funds is 0.036. The summary statistics for the cor-
porate governance variables show that ownership concentration in Chinese listed companies is relatively high.
The mean of Top1 is 37.9% and the maximum value is 75.1%, while the average TOP2_10 is 18.5%. Consid-
ering the corporate governance structure, on average, the proportion of independent directors is relatively
high, with a mean of 55.2%, which can play a positive role in monitoring controlling shareholders and man-
agement and protecting minority investors. Meanwhile, the average Duality is 9.6%, which may reduce earn-
ings quality.
Panel B of Table 2 shows that the extent of earnings management is significantly positively related to infor-
mation asymmetry and the proportion of shareholdings of long- and short-term mutual funds. This suggests
that the extent of earnings management increases as the extent of information asymmetry increases, and both
long-term and short-term mutual funds are likely to increase earnings management behavior. Both company
size and market-to-book ratios are positively related to abnormal accruals, which shows the dominant role
of speculation. The negative correlation coefficient between turnover and abnormal accruals is consistent with
the conclusion in Hakim et al. (2008). Considering the corporate governance variables, Top1 and CR_5 are both
positively related to abnormal accruals, which suggests that large shareholders have an incentive to implement
8
We thank the referees for this suggestion.
Table 2
196
Descriptive statistics and correlation coefficient matrix.
|DD_eq| |BS_eq| SD Long_ratio Short_ratio TO Duality Out_ratio Top1 Top2_10 CR_5 LnSize LnMB Offering SOE Audit TopAudit
Panel A: Summary statistics
N 7286 7286 7286 7286 7286 7286 7286 7286 7286 7286 7286 7286 7286 7286 7286 7286 7286
Mean 0.037 0.036 0.024 0.021 0.036 6.428 0.096 0.552 0.379 0.185 0.530 21.831 1.110 0.131 0.669 0.989 0.073
Std.Dev 0.046 0.044 0.006 0.041 0.066 3.977 0.295 0.107 0.157 0.128 0.150 1.121 0.698 0.338 0.470 0.102 0.260
Min 0.000 0.000 0.012 0.000 0.000 0.787 0.000 0.250 0.090 0.010 0.198 19.838 0.177 0 0 0 0
P25 0.009 0.009 0.020 0.000 0.000 3.298 0.000 0.500 0.252 0.075 0.424 21.001 0.583 0 0 1 0
P50 0.021 0.021 0.024 0.000 0.001 5.527 0.000 0.500 0.360 0.162 0.537 21.709 1.057 0 1 1 0
P75 0.043 0.043 0.028 0.023 0.043 8.735 0.000 0.600 0.502 0.276 0.638 22.503 1.565 0 1 1 0
Y. Dai et al. / China Journal of Accounting Research 6 (2013) 187–209

Max 0.259 0.253 0.039 0.210 0.314 18.980 1.000 0.900 0.751 0.551 0.910 25.356 3.434 1 1 1 1
Panel B: Correlation matrix

|BS_eq| 0.940
SD 0.028 0.029
Long_ratio 0.148 0.145 0.041
Short_ratio 0.245 0.235 0.061 0.571
TO 0.068 0.068 0.565 0.101 0.151
Duality 0.022 0.019 0.028 0.009 0.002 0.025
Out_ratio 0.027 0.028 0.081 0.021 0.002 0.106 0.038
Top1 0.043 0.047 0.134 0.055 0.079 0.144 0.058 0.042
Top2_10 0.016 0.021 0.035 0.178 0.166 0.091 0.042 0.011 0.464
CR_5 0.026 0.027 0.136 0.121 0.134 0.218 0.032 0.042 0.717 0.264
LnSize 0.370 0.352 0.082 0.395 0.502 0.087 0.034 0.076 0.178 0.013 0.147
LnMB 0.049 0.045 0.477 0.242 0.226 0.390 0.043 0.082 0.106 0.123 0.055 0.368
Offering 0.056 0.056 0.177 0.126 0.089 0.111 0.009 0.053 0.007 0.019 0.015 0.163 0.162
SOE 0.045 0.037 0.100 0.032 0.050 0.105 0.121 0.094 0.256 0.203 0.129 0.136 0.125 0.072
Audit 0.046 0.041 0.004 0.049 0.054 0.032 0.003 0.018 0.021 0.011 0.006 0.083 0.004 0.028 0.050 1.000
TopAudit 0.215 0.196 0.104 0.117 0.167 0.119 0.041 0.003 0.113 0.091 0.203 0.322 0.055 0.006 0.093 0.024 1.000
Notes: Panel A presents the descriptive statistics for all variables (dependent, independent and control). |DD_eq| and |BS_eq| are the absolute value of abnormal accruals for each firm,
following the DD model and BS model respectively, both of which have a positive relationship with the extent of earnings management. Long_ratio (Short_ratio) is the ratio between
the number of shares held by long-term (short-term) mutual fund investors and the total number of shares outstanding. The rest are control variables, as follows: Duality: a dummy
variable that equals 1 if the chairman and CEO are the same person, and 0 otherwise. Out_ratio: the proportion of independent directors on the board. Top1: the proportion of
shareholdings of the largest shareholder. Top2_10: the aggregate proportion of shareholdings of the second to the 10th largest shareholders. CR_5: the aggregate proportion of
shareholdings of the top 5 shareholders. LnSize: natural logarithm of total assets at the end of the year. LnMB: natural logarithm of the market-to-book ratio. TO: turnover of listed
companies each year. Offering: a corporate financing needs dummy variable, which equals 1 if the company has an SEO, rights offering or debt issue in period t, and 0 otherwise. SOE:
a dummy variable that equals 1 if the company is a state-owned enterprise, and 0 otherwise. Audit: a dummy variable that equals 1 if the audit opinion is unqualified, and 0 otherwise.
TopAudit: a dummy variable that equals 1 if the company’s audit firm is a Big 4 firm, and 0 otherwise. Panel B presents the Pearson correlation coefficients for each variable.
tunneling and propping behavior that benefits themselves but damages other shareholders’ interests (Liu and
He, 2004). The direct consequence of tunneling and propping is the reduction of earnings quality, so tunneling
behavior will reduce transparency and distort earnings. This finding is consistent with Bertrand et al. (2002).
In general, the relationships between the variables are consistent with our predictions. Next, we introduce
the control variables and use the regression models to investigate the specific effects of information asymmetry
and long- and short-term mutual fund ownership on earnings management.
4.2. Empirical results and analysis
We investigate how the extent of information asymmetry and long/short-term mutual fund ownership, and
a combination of the two, affect listed companies’ earnings management. To test Hypothesis 1, we examine the
relationship between information asymmetry and earnings management, and propose the following regression
model:
DD eqi;t ¼ ai þ b1 SDi;t þ b2 Con Vari;t þ ei;t ð8Þ
There is an endogeneity problem between information asymmetry and earnings management, because a
reduction in transparency may exacerbate earnings management behavior. High quality earnings information
may, in turn, reduce the extent of information asymmetry between shareholders and company management.
To mitigate this problem, we adopt a difference-in-difference regression method to control for endogeneity,
and propose the following regression model:
DDD eqi;t ¼ ai þ b1 DSDi;t þ b2 Con Vari;t þ ei;t ð9Þ
In Eqs. (8) and (9), DD_eqi,t is the abnormal accruals of firm i in period t, DDD_eqi,t is the change in abnormal
accruals from t 1 to t. To distinguish between positive and negative earnings management, we not only use
|DD_eqi,t|, but also divide DD_eqi,t into two groups based on whether the value of DD_eqi,t is larger than 0,
then obtain DD_eqi,t(+) and DD_eqi,t(). SDi,t is a proxy for the extent of information asymmetry for firm i in
period t, and DSDi,t is the change in information asymmetry from t 1 to t. Con_Vari,t indicates the other
control variables for firm i in period t, including LnSize, LnMB, TO, the corporate governance variables
(Top1, Duality, Out_ratio, CR_5, Top2_10) and the dummy variables (Offering, SOE, Audit, TopAudit) that
may affect the firm’s earnings management. The regression results are presented in Tables 3 and 4.
From the tables, we find that the absolute value of abnormal accruals |DD_eq| (|D_DD_eq|) is significantly
positively related to the information asymmetry measure SD (D_SD) in both Eqs. (8) and (9). These results
indicate that as the extent of information asymmetry increases, earnings management behavior becomes more
serious, thus confirming Hypothesis 1.
We examine positive and negative abnormal accruals separately and find that positive abnormal accruals,
DD_eq(+), are significantly positively related to the extent of information asymmetry, and this is also con-
firmed in the difference-in-difference regression. Negative abnormal accruals, DD_eq(), are only significantly
negatively related to the extent of information asymmetry in Eq. (8), suggesting that when a company’s infor-
mation transparency is low, both positive and negative earnings management increase, resulting in worse earn-
ings quality. To explain why the coefficient of D_DD_eq() is not significant, we consider a more common
situation in reality: if a firm’s management have an information advantage and have more private informa-
tion, they will usually seek to maximize profits to attract more investors. Therefore, they have an incentive
to cover up any potential operational problems and are more likely to manipulate profits, leading to positive
earnings management. Consequently, their earnings quality is worse, as in the well-known Enron scandal or
“Yin Guang Xia” incident. Such companies can successfully report deceptive earnings information due to the
serious information asymmetry between shareholders and management. In particular, because of the high cost
of supervision for minority shareholders, investors lack information and cannot determine the company’s
actual earnings, thus directly providing insiders with the opportunity to manipulate earnings and capture
short-term gains. In reality, even if a company has the ability to cover up good performance when they face
serious information asymmetry, they do not have the incentive to do so because the negative management will
have an adverse effect on the company’s short-term performance. Moreover, taking into consideration the
Table 3
Regression results for information asymmetry and earnings management.
|DD_abs| DD_eq(+) DD_eq()
SD 0.570*** 0.558*** 0.354***
(5.903) (2.753) (3.229)
TO 0.000*** 0.001*** 0.000
(2.636) (3.501) (0.388)
Duality 0.002 0.006 0.002
(0.839) (1.317) (0.873)
Out_ratio 0.008 0.001 0.011**
(1.608) (0.121) (2.076)
Top1 0.081*** 0.037 0.032
(3.247) (0.730) (1.077)
Top2_10 0.063*** 0.044 0.023
(3.000) (1.028) (0.928)
CR_5 0.064*** 0.052 0.008
(2.682) (1.094) (0.297)
LnSize 0.007*** 0.022*** 0.002*
(6.803) (10.601) (1.859)
LnMB 0.000 0.012*** 0.004***
(0.321) (4.102) (2.737)
Offering 0.000 0.003 0.002*
(0.276) (0.990) (1.661)
SOE 0.004 0.010 0.009***
(1.436) (1.467) (2.736)
Audit 0.025*** 0.015 0.016***
(5.615) (0.969) (3.804)
TopAudit 0.007** 0.002 0.003
(1.974) (0.273) (0.729)
Intercept 0.089*** 0.428*** 0.098***
(4.211) (8.891) (3.694)
Obs 7286 2873 4413
R2_adj 0.036 0.098 0.024
Notes: This table reports the regression results for Eq. (8). The data in brackets are the estimated t-values of the regression coefficients.
|DD_abs| is the absolute value of abnormal accruals for each firm, following the DD model, which has a positive relationship with the
extent of earnings management. We divide DD_eq into two groups based on whether the value of DD_eq is larger than 0, then we obtain
DD_eq(+) and DD_eq(). SD: the standard deviation of idiosyncratic risk, used as the measure of information asymmetry. Duality: a
dummy variable that equals 1 if the chairman and CEO are the same person, and 0 otherwise. Out_ratio: the proportion of independent
directors on the board. Top1: the proportion of shareholdings of the largest shareholder. Top2_10: the aggregate proportion of share-
holdings of the second to the 10th shareholder. CR_5: the aggregate proportion of shareholdings of the top 5 shareholders. LnSize: natural
logarithm of total assets at the end of the year. LnMB: natural logarithm of the market-to-book ratio. TO: turnover of listed companies for
each year. Offering: a corporate financing needs dummy variable, which equals 1 if the company has an SEO, rights offering or debt issue in
period t, and 0 otherwise. SOE: dummy variable that equals 1 if company is a state-owned enterprise, and 0 otherwise. Audit: a dummy
variable that equals 1 if the audit opinion is unqualified, and 0 otherwise. TopAudit: a dummy variable that equals 1 if the company’s audit
firm is a Big 4 firm, and 0 otherwise. Intercept: intercept term. Obs: number of observations. R2_adj: adjusted R-square.
*
Significant at the 10% level.
**
***
company’s corporate reputation, long-term development and investor confidence, the company will use less
negative earnings management.
According to the above, the higher the extent of information asymmetry, the more serious the extent of
earnings management, whether positive or negative. Therefore, to reduce earnings management, regulators
should strengthen the supervision of listed companies, improve the quality of information disclosure and
reduce the information asymmetry between investors and listed companies.
Table 4
Difference-in-difference regression results for information asymmetry and earnings management.
D_|DD_abs| D_DD_eq(+) D_DD_eq()
D_SD 0.328*** 0.845*** 0.117
(2.906) (2.863) (0.759)
D_TO 0.000 0.000 0.001***
(1.203) (0.403) (5.087)
Duality 0.001 0.007 0.000
(0.180) (0.806) (0.072)
D_Out_ratio 0.000 0.021 0.016**
(0.014) (1.494) (2.045)
D_Top1 0.066* 0.046 0.016
(1.870) (0.511) (0.322)
D_Top2_10 0.052* 0.078 0.023
(1.850) (1.085) (0.586)
D_CR_5 0.080** 0.126 0.022
(2.307) (1.458) (0.433)
D_LnSize 0.001 0.053*** 0.025***
(0.241) (8.928) (7.766)
D_LnMB 0.006*** 0.048*** 0.021***
(2.637) (7.965) (6.562)
Offering 0.001 0.003 0.001
(0.467) (0.710) (0.445)
SOE 0.002 0.012 0.004
(0.305) (0.741) (0.497)
Audit 0.003 0.064** 0.008
(0.321) (1.974) (0.829)
TopAudit 0.005 0.017 0.030***
(0.907) (1.369) (3.318)
Intercept 0.000 0.081** 0.011
(0.022) (2.379) (0.934)
Obs 5915 2270 3645
R2_adj 0.012 0.091 0.037
Notes: This table reports the difference-in-difference regression results for Eq. (9). The data in brackets are the estimated t-values of the
regression coefficients. D_|DD_abs| is the change in the absolute value of abnormal accruals for each firm, following the DD model, from
period t 1 to t. We divide D_DD_eq into two groups based on whether the value of D_DD_eq is larger than 0, then we obtain
D_DD_eq(+) and D_DD_eq(). D_SD: the change in the standard deviation of idiosyncratic risk, used as the information asymmetry
measure, from period t 1 to t. Duality: a dummy variable that equals 1 if the chairman and CEO are the same person, and 0 otherwise.
D_Out_ratio: the change in the proportion of independent directors on the board from period t 1 to t. D_Top1: the change in the
proportion of shareholdings of the largest shareholder from period t 1 to t. D_Top2_10: the change in the aggregate proportion of
shareholdings of the second to the 10th shareholders from period t 1 to t. D_CR_5: the change in the aggregate proportion of
shareholdings of the top 5 shareholders from period t 1 to t. D_LnSize: the change in the natural logarithm of total assets at the end of
the year from period t 1 to t. LnMB: the change in the natural logarithm of the market-to-book ratio from period t 1 to t. D_TO: the
change in the turnover of listed companies for each year from period t 1 to t. Offering: a corporate financing needs dummy variable,
which equals 1 if the company has an SEO, rights offering or debt issue in period t, and 0 otherwise. SOE: a dummy variable that equals 1
if the company is a state-owned enterprise, and 0 otherwise. Audit: a dummy variable that equals 1 if the audit opinion is unqualified, and 0
otherwise. TopAudit: a dummy variable that equals 1 if the company’s audit firms is a Big 4 firm, and 0 otherwise. Intercept: intercept term.
Obs: number of observations. R2_adj: adjusted R-square.
*
**
***
We also use a fixed-effect model to test Hypothesis 2. Because institutional investors may reduce their
investment in companies with low earnings quality, we also use a difference-in-difference regression to control
for this endogeneity problem, and propose the following equations:
DD eqi;t ¼ ai þ b1 Instii;t þ b2 Con Vari;t þ ei;t ð10Þ
DDD eqi;t ¼ ai þ b1 DInstii;t þ b2 Con Vari;t þ ei;t ð11Þ
Table 5
Regression results for long/short-term mutual funds and earnings management.
|DD_abs| DD_eq(+) DD_eq() |DD_abs| DD_eq(+) DD_eq()
Long_ratio 0.007 0.002 0.056***
(0.518) (0.080) (3.052)
Short_ratio 0.036*** 0.027* 0.002
(3.692) (1.762) (0.130)
TO 0.000 0.001*** 0.000** 0.000 0.001** 0.000**
(0.059) (2.617) (1.982) (0.255) (2.394) (2.184)
Duality 0.002 0.006 0.002 0.002 0.005 0.002
(0.973) (1.294) (0.824) (0.908) (1.179) (0.766)
Out_ratio 0.007 0.002 0.011** 0.007 0.002 0.011**
(1.529) (0.156) (2.119) (1.499) (0.155) (2.012)
Top1 0.066** 0.026 0.003 0.095*** 0.053 0.025
(2.549) (0.499) (0.100) (3.689) (1.003) (0.806)
Top2_10 0.046** 0.031 0.007 0.075*** 0.058 0.014
(2.091) (0.693) (0.262) (3.407) (1.273) (0.542)
CR_5 0.047* 0.036 0.020 0.074*** 0.062 0.001
(1.898) (0.741) (0.684) (2.986) (1.236) (0.018)
LnSize 0.007*** 0.022*** 0.003** 0.006*** 0.022*** 0.003**
(6.547) (10.581) (2.071) (6.060) (10.415) (2.140)
LnMB 0.001 0.011*** 0.005*** 0.000 0.012*** 0.005***
(0.494) (3.634) (3.518) (0.170) (3.867) (3.129)
Offering 0.000 0.003 0.002 0.000 0.003 0.002
(0.365) (1.230) (1.279) (0.374) (1.304) (1.297)
SOE 0.004 0.011 0.009*** 0.004 0.011 0.009***
(1.270) (1.573) (2.667) (1.282) (1.600) (2.701)
Audit 0.024*** 0.013 0.015*** 0.024*** 0.013 0.015***
(5.371) (0.838) (3.597) (5.453) (0.848) (3.644)
TopAudit 0.007* 0.002 0.003 0.006* 0.002 0.003
(1.953) (0.335) (0.592) (1.910) (0.265) (0.686)
Intercept 0.075*** 0.419*** 0.108*** 0.064*** 0.411*** 0.111***
(3.535) (8.705) (4.122) (2.985) (8.503) (4.221)
Obs 7286 2873 4413 7286 2873 4413
R2_adj 0.031 0.094 0.023 0.033 0.096 0.020
DD_eq(+) and DD_eq(). Long_ratio (Short_ratio) is the ratio between the number of shares held by long-term (short-term) mutual fund
investors and the total number of shares outstanding. Duality: a dummy variable that equals 1 if the chairman and CEO are the same
person, and 0 otherwise. Out_ratio: the proportion of independent directors on the board. Top1: the proportion of shareholdings of the
largest shareholder. Top2_10: the aggregate proportion of shareholdings of the second to the 10th largest shareholders. CR_5: the
aggregate proportion of shareholdings of the top 5 shareholders. LnSize: natural logarithm of total assets at the end of the year. LnMB:
natural logarithm of the market-to-book ratio. TO: turnover of listed companies for each year. Offering: a corporate financing needs
dummy variable, which equals 1 if the company has an SEO, rights offering or debt issue in period t, and 0 otherwise. SOE: a dummy
variable that equals 1 if company is a state-owned enterprise, and 0 otherwise. Audit: a dummy variable that equals 1 if the audit opinion is
unqualified, and 0 otherwise. TopAudit: a dummy variable that equals 1 if the company’s audit firm is a Big 4 audit firm, and 0 otherwise.
Intercept: intercept term. Obs: number of observations. R2_adj: adjusted R-square.
*
**
***
where Instii,t is the proportion of shareholdings held by long- or short-term mutual funds i in period t (Long/
Short_ratio). DInstii,t is the change in the proportion of shareholdings held by long- or short-term mutual
funds from period t 1 to t (D_Long/Short_ratio). Other variables are the same as those used above, and
the results are presented in Tables 5 and 6.
Table 6
Difference-in-difference regression results for long/short-term mutual funds and earnings management.
D_|DD_abs| D_DD_eq(+) D_DD_eq() D_|DD_abs| D_DD_eq(+) D_DD_eq()
D_Long_ratio 0.001 0.004 0.079***
(0.063) (0.140) (3.323)
D_Short_ratio 0.010 0.037* 0.016
(0.920) (1.737) (0.824)
D_TO 0.000 0.000 0.001*** 0.000 0.000 0.001***
(0.318) (0.303) (4.611) (0.227) (0.500) (5.178)
Duality 0.001 0.007 0.000 0.001 0.008 0.000
(0.239) (0.865) (0.062) (0.244) (0.880) (0.067)
D_Out_ratio 0.000 0.021 0.016** 0.000 0.020 0.016**
(0.053) (1.430) (2.104) (0.059) (1.402) (2.062)
D_Top1 0.061* 0.048 0.051 0.069* 0.085 0.006
(1.699) (0.531) (0.982) (1.905) (0.926) (0.115)
D_Top2_10 0.046 0.079 0.056 0.054* 0.116 0.013
(1.602) (1.062) (1.376) (1.859) (1.539) (0.317)
D_CR_5 0.072** 0.124 0.055 0.079** 0.157* 0.011
(2.055) (1.394) (1.086) (2.259) (1.760) (0.224)
D_LnSize 0.001 0.054*** 0.024*** 0.001 0.053*** 0.025***
(0.235) (9.028) (7.483) (0.309) (8.955) (7.797)
D_LnMB 0.006*** 0.047*** 0.021*** 0.006*** 0.048*** 0.021***
(2.740) (7.859) (6.723) (2.725) (7.892) (6.527)
Offering 0.001 0.004 0.001 0.001 0.004 0.001
(0.584) (0.835) (0.431) (0.580) (0.898) (0.406)
SOE 0.001 0.010 0.004 0.001 0.011 0.004
(0.151) (0.614) (0.503) (0.162) (0.655) (0.553)
Audit 0.002 0.063* 0.009 0.002 0.063* 0.009
(0.288) (1.930) (0.872) (0.281) (1.948) (0.836)
TopAudit 0.005 0.015 0.029*** 0.005 0.015 0.030***
(0.871) (1.250) (3.174) (0.877) (1.213) (3.318)
Intercept 0.001 0.079** 0.010 0.000 0.080** 0.011
(0.057) (2.291) (0.876) (0.043) (2.322) (0.955)
Obs 5915 2270 3645 5915 2270 3645
R2_adj 0.011 0.086 0.041 0.011 0.088 0.037
Notes: This table reports the difference-in-difference regression results for Eq. (11). The data in brackets are the estimated t-values of the
regression coefficients. D_|DD_abs| is the change in the absolute value of abnormal accruals for f each firm, following the DD model, from
period t 1 to t. We divide D_DD_eq into two groups based on whether the value of D_DD_eq is larger than 0, then we obtain
D_DD_eq(+) and D_DD_eq(). D_Long_ratio (D_Short_ratio) is the change in the ratio between the number of shares held by long-term
(short-term) mutual fund investors and the total number of shares outstanding from period t 1 to t. Duality: a dummy variable that
equals 1 if the chairman and CEO are the same person, and 0 otherwise. D_Out_ratio: the change in the proportion of independent
directors on the board from period t 1 to t. D_Top1: the change in the proportion of shareholdings of the largest shareholder from
period t 1 to t. D_Top2_10: the change in the aggregate proportion of shareholdings of the second to the 10th largest shareholders from
period t 1 to t. D_CR_5: the change in the aggregate proportion of shareholdings of the top 5 shareholders from period t 1 to t.
D_LnSize: the change in the natural logarithm of total assets at the end of the year from period t 1 to t. LnMB: the change in the natural
logarithm of the market-to-book ratio from period t 1 to t. D_TO: the change in the turnover of listed companies in each year from
period t 1 to t. Offering: a corporate financing needs dummy variable, which equals 1 if the company has an SEO, rights offering or debt
issue in period t, and 0 otherwise. SOE: a dummy variable that equals 1 if the company is a state-owned enterprise, and 0 otherwise. Audit:
a dummy variable that equals 1 if the audit opinion is unqualified, and 0 otherwise. TopAudit: a dummy variable that equals 1 if the
company’s audit firm is a Big 4 audit firm, and 0 otherwise. Intercept: intercept term. Obs: number of observations. R2_adj: adjusted R-
square.
*
**
***
The results in Tables 5 and 6 indicate a prominent difference between long- and short-term mutual funds.
When the dependent variable is negative earnings management, the coefficient of the proportion of sharehold-
ings held by long-term investors is significantly positive. This implies that long-term mutual funds can reduce
Table 7
Regression results for long/short-term mutual funds and earnings management after grouping based on the extent of information
asymmetry.
Low High
|DD_abs| DD_eq(+) DD_eq() DD_abs DD_eq(+) DD_eq() |DD_abs| DD_eq(+) DD_eq() DD_abs DD_eq(+) DD_eq()
Long_ratio 0.006 0.002 0.071** 0.023 0.011 0.018
(0.260) (0.056) (2.500) (1.047) (0.295) (0.572)
Short_ratio 0.028** 0.001 0.006 0.046*** 0.026 0.038
(1.992) (0.050) (0.287) (2.747) (0.921) (1.571)
TO 0.000 0.001** 0.000 0.000 0.001** 0.000 0.000 0.001* 0.000 0.000 0.001 0.000
(1.485) (2.293) (0.760) (1.262) (2.261) (0.928) (0.328) (1.732) (1.205) (0.009) (1.590) (1.363)
Duality 0.002 0.003 0.001 0.002 0.003 0.000 0.002 0.008 0.003 0.002 0.008 0.003
(0.569) (0.475) (0.146) (0.487) (0.476) (0.106) (0.544) (1.058) (0.870) (0.527) (1.046) (0.905)
Out_ratio 0.009 0.003 0.009 0.009 0.003 0.009 0.023*** 0.011 0.034*** 0.022*** 0.011 0.033***
(1.137) (0.211) (0.976) (1.174) (0.213) (0.984) (2.976) (0.629) (3.926) (2.930) (0.591) (3.897)
Top1 0.110*** 0.077 0.090* 0.131*** 0.077 0.116** 0.037 0.052 0.046 0.082** 0.015 0.017
(2.614) (0.947) (1.646) (3.121) (0.933) (2.138) (0.941) (0.588) (0.977) (2.074) (0.161) (0.365)
Top2_10 0.064* 0.069 0.049 0.086** 0.069 0.075* 0.036 0.002 0.048 0.079** 0.038 0.021
(1.804) (0.994) (1.110) (2.416) (0.970) (1.689) (1.058) (0.023) (1.202) (2.330) (0.483) (0.513)
CR_5 0.089** 0.090 0.068 0.109*** 0.090 0.091* 0.025 0.016 0.070 0.067* 0.020 0.043
(2.213) (1.172) (1.310) (2.692) (1.145) (1.765) (0.676) (0.187) (1.569) (1.781) (0.229) (0.968)
LnSize 0.011*** 0.024*** 0.001 0.011*** 0.024*** 0.002 0.006*** 0.024*** 0.007*** 0.005*** 0.023*** 0.007***
(6.126) (6.665) (0.545) (6.073) (6.749) (0.707) (3.899) (7.086) (3.547) (3.238) (6.802) (3.891)
LnMB 0.003 0.007 0.002 0.003 0.007 0.000 0.002 0.014*** 0.007*** 0.002 0.014*** 0.007***
(1.054) (1.252) (0.596) (1.382) (1.319) (0.138) (1.055) (2.956) (3.725) (0.912) (3.015) (3.688)
Offering 0.001 0.003 0.002 0.001 0.003 0.002 0.001 0.005 0.001 0.001 0.004 0.001
(0.583) (0.758) (0.787) (0.662) (0.753) (0.831) (0.545) (1.062) (0.512) (0.448) (1.042) (0.563)
SOE 0.001 0.005 0.001 0.001 0.005 0.002 0.007* 0.022** 0.011** 0.007* 0.022** 0.011**
(0.179) (0.411) (0.191) (0.156) (0.413) (0.287) (1.661) (1.989) (2.529) (1.687) (1.991) (2.576)
Audit 0.023** 0.044 0.015* 0.023** 0.044 0.015* 0.020*** 0.008 0.008 0.020*** 0.008 0.008
(2.488) (0.881) (1.822) (2.499) (0.881) (1.844) (3.572) (0.425) (1.468) (3.619) (0.429) (1.495)
TopAudit 0.013*** 0.012 0.001 0.013*** 0.012 0.001 0.002 0.011 0.001 0.002 0.010 0.002
(2.833) (1.479) (0.110) (2.751) (1.479) (0.133) (0.312) (0.991) (0.099) (0.260) (0.897) (0.173)
Intercept 0.184*** 0.434*** 0.016 0.179*** 0.435*** 0.009 0.054* 0.445*** 0.193*** 0.034 0.433*** 0.209***
(4.479) (4.836) (0.297) (4.385) (4.852) (0.165) (1.804) (6.010) (5.144) (1.110) (5.752) (5.504)
Obs 3645 1527 2118 3645 1527 2118 3641 1346 2295 3641 1346 2295
R2_adj 0.043 0.104 0.021 0.044 0.104 0.016 0.032 0.123 0.038 0.034 0.124 0.039
Notes: This table reports the regression results for Eq. (10) after dividing the sample into Low and High groups based on the extent of
listed companies’ information asymmetry. The data in brackets are the estimated t-values of the regression coefficients. |DD_abs| is the
absolute value of abnormal accruals of each firm, following the DD model, which has a positive relationship with the extent of earnings
management. We divide DD_eq into two groups based on whether the value of DD_eq is larger than 0, then we obtain DD_eq(+) and
DD_eq(). Long_ratio (Short_ratio) is the ratio between the number of shares held by long-term (short-term) mutual fund investors and
the total number of shares outstanding. Duality: a dummy variable that equals 1 if the chairman and CEO are the same person, and 0
otherwise. Out_ratio: the proportion of independent directors on the board. Top1: the proportion of shareholdings of the largest
shareholder. Top2_10: the aggregate proportion of shareholdings of the second to the 10th largest shareholders. CR_5: the aggregate
proportion of shareholdings of the top 5 shareholders. LnSize: natural logarithm of total assets at the end of the year. LnMB: natural
logarithm of the market-to-book ratio. TO: turnover of listed companies for each year. Offering: a corporate financing needs dummy
variable, which equals 1 if the company has an SEO, rights offering or debt issue in period t, and 0 otherwise. SOE: a dummy variable that
equals 1 if the company is a state-owned enterprise, and 0 otherwise. Audit: a dummy variable that equals 1 if the audit opinion is
unqualified, and 0 otherwise. TopAudit: a dummy variable that equals 1 if the company’s audit firm is a Big 4 audit firm, and 0 otherwise.
Intercept: intercept term. Obs: number of observations. R2_adj: adjusted R-square.
*
**
***
companies’ negative earnings management behavior and play a supervisory role of management, because they
have an information advantage and can communicate with listed companies effectively. Furthermore, their
large investments and long-sighted view of performance means they are more likely to actively participate
in corporate governance and monitor management, thereby reducing the company’s profit manipulation
behavior. This “supervision effect” increases the company’s earnings quality and increases mutual fund inves-
tors’ returns to compensate them for their monitoring costs and uncertainty risk.
Table 8
Difference-in-difference regression results for long/short-term mutual funds and earnings management after grouping based on the extent of information asymmetry.
Low High
D_|DD_abs| D_DD_eq(+) D_DD_eq() D_|DD_abs| D_DD_eq(+) D_DD_eq() D_|DD_abs| D_DD_eq(+) D_DD_eq() D_|DD_abs| D_DD_eq(+) D_DD_eq()
D_Long_ratio 0.046* 0.060 0.138*** 0.031 0.056 0.058
(1.945) (1.196) (3.499) (1.367) (1.097) (1.560)
D_Short_ratio 0.002 0.024 0.022 0.011 0.022 0.032
(0.095) (0.723) (0.694) (0.630) (0.596) (1.054)
*
D_TO 0.000 0.001 0.001 0.000 0.001 0.001** 0.000 0.000 0.001 ***
0.000 0.000 0.001***
(1.092) (1.016) (1.962) (0.755) (0.655) (2.558) (0.888) (0.485) (3.109) (0.764) (0.411) (3.379)
Duality 0.002 0.010 0.002 0.002 0.011 0.002 0.001 0.000 0.002 0.000 0.001 0.002
(0.469) (0.743) (0.204) (0.453) (0.795) (0.311) (0.129) (0.011) (0.324) (0.081) (0.082) (0.315)
Y. Dai et al. / China Journal of Accounting Research 6 (2013) 187–209

D_Out_ratio 0.016* 0.033 0.006 0.016* 0.033 0.005 0.021** 0.022 0.037*** 0.021** 0.021 0.036***
(1.800) (1.595) (0.442) (1.809) (1.596) (0.388) (2.372) (0.787) (3.246) (2.396) (0.741) (3.157)
D_Top1 0.112* 0.244 0.029 0.135** 0.303* 0.032 0.030 0.165 0.065 0.025 0.165 0.013
(1.751) (1.555) (0.285) (2.113) (1.940) (0.320) (0.577) (1.101) (0.868) (0.476) (1.068) (0.171)
D_Top2_10 0.060 0.216* 0.085 0.083* 0.275** 0.020 0.044 0.037 0.040 0.038 0.039 0.006
(1.197) (1.723) (1.118) (1.652) (2.193) (0.260) (1.053) (0.300) (0.675) (0.892) (0.305) (0.106)
D_CR_5 0.141** 0.339** 0.013 0.164*** 0.396*** 0.049 0.054 0.089 0.054 0.048 0.096 0.007
(2.240) (2.233) (0.131) (2.633) (2.642) (0.499) (1.067) (0.600) (0.741) (0.933) (0.630) (0.093)
D_LnSize 0.005 0.056*** 0.019** 0.005 0.055*** 0.018** 0.004 0.055*** 0.027*** 0.004 0.056*** 0.028***
(1.111) (5.061) (2.575) (1.118) (4.953) (2.457) (1.406) (6.269) (6.673) (1.276) (6.430) (6.992)
D_LnMB 0.002 0.043*** 0.019** 0.000 0.045*** 0.015** 0.008*** 0.050*** 0.023*** 0.008*** 0.050*** 0.023***
(0.301) (3.833) (2.497) (0.005) (4.035) (1.971) (2.939) (5.740) (5.956) (2.943) (5.729) (5.906)
Offering 0.001 0.006 0.001 0.001 0.006 0.000 0.002 0.010 0.002 0.002 0.010 0.002
(0.289) (0.842) (0.192) (0.211) (0.812) (0.070) (0.887) (1.304) (0.481) (0.864) (1.333) (0.474)
SOE 0.005 0.009 0.026 0.005 0.009 0.025 0.007 0.025 0.014 0.007 0.025 0.014
(0.468) (0.352) (1.402) (0.473) (0.365) (1.333) (0.962) (0.842) (1.579) (0.941) (0.832) (1.564)
Audit 0.021 0.191** 0.019 0.020 0.194** 0.019 0.000 0.042 0.005 0.001 0.043 0.005
(1.073) (2.320) (0.900) (1.057) (2.357) (0.874) (0.047) (1.049) (0.386) (0.062) (1.064) (0.371)
TopAudit 0.014* 0.006 0.010 0.014* 0.006 0.008 0.027** 0.007 0.061*** 0.026** 0.006 0.062***
(1.783) (0.369) (0.759) (1.713) (0.358) (0.643) (2.383) (0.284) (3.424) (2.344) (0.261) (3.482)
Intercept 0.025 0.188** 0.021 0.025 0.192** 0.019 0.008 0.067 0.019 0.008 0.067 0.019
(1.208) (2.230) (0.830) (1.182) (2.269) (0.745) (0.689) (1.523) (1.361) (0.691) (1.527) (1.372)
Obs 2954 1225 1729 2954 1225 1729 2961 1045 1916 2961 1045 1916
R2_adj 0.017 0.083 0.034 0.015 0.082 0.022 0.024 0.135 0.073 0.023 0.133 0.071
Notes: This table reports the difference-in-difference regression results for Eq. (11) after dividing the sample into Low and High groups based on the extent of listed companies’
information asymmetry. The data in brackets are the estimated t-values of the regression coefficients. D_|DD_abs| is the change in the absolute value of abnormal accruals for each
firm, following the DD model, from period t 1 to t. We divide D_DD_eq into two groups based on whether the value of D_DD_eq is larger than 0, then we obtain D_DD_eq(+) and
D_DD_eq(). D_Long_ratio (D_Short_ratio) is the change in the ratio between the number of shares held by long-term (short-term) mutual fund investors and the total number of
shares outstanding from period t 1 to t. Duality: a dummy variable that equals 1 if the chairman and CEO are the same person, and 0 otherwise. D_Out_ratio: the change in the
proportion of independent directors on the board from period t 1 to t. D_Top1: the change in the proportion of shareholdings of the largest shareholder from period t 1 to t.
D_Top2_10: the change in the aggregate proportion of shareholdings of the second to the 10th largest shareholders from period t 1 to t. D_CR_5: the change in the aggregate
proportion of shareholdings of the top 5 shareholders from period t 1 to t. D_LnSize: the change in the natural logarithm of total assets at the end of the year from period t 1 to t.
LnMB: the change in the natural logarithm of the market-to-book ratio from period t 1 to t. D_TO: the change in turnover of listed companies for each year from period t 1 to t.
Offering: a corporate financing needs dummy variable, which equals 1 if the company has an SEO, rights offering or debt issue in period t, and 0 otherwise. SOE: a dummy variable that
equals 1 if the company is a state-owned enterprise, and 0 otherwise. Audit: a dummy variable that equals 1 if the audit opinion is unqualified, and 0 otherwise. TopAudit: a dummy
variable that equals 1 if the company’s audit firm is a Big 4 audit firm, and 0 otherwise. Intercept: intercept term. Obs: number of observations. R2_adj: adjusted R-square.
*
**
203
***
However, we find a difference in short-term mutual funds’ investment behavior. Table 5 shows a significant
positive relationship between the proportion of shareholdings held by short-term investors and positive abnor-
mal accruals. The sign is consistent with our prediction in the difference-in-difference regression when the
dependent variable is D_|DD_abs|, although it is only marginally significant. This result suggests that an
increase in the proportion of shares held by short-term mutual funds reduces listed companies’ earnings qual-
ity and, in particular, increases positive earnings management. We also find that the coefficient for the propor-
tion of shareholdings held by short-term investors is not significant when the dependent variable is negative
abnormal accruals. As negative management will not increase short-term returns, there is less incentive to
implement negative earnings management to reduce financial reporting quality.
From the investment behavior of long- and short-term mutual funds, we conclude that compared with
short-term mutual funds, the monitoring effect of long-term mutual funds on listed companies’ earnings man-
agement is stronger, consistent with Hypothesis 2.
For Hypothesis 3, we divide the sample into Low and High groups based on the extent of listed companies’
information asymmetry, and investigate the effect of the proportion of shareholding held by long- and short-
term mutual funds on earnings management in different information environments. The results are presented
in Tables 7 and 8.
The results indicate that the speculation of short-term mutual funds is observable, regardless of the extent
of information asymmetry. In the low information asymmetry group, for each unit increase in the proportion
of short-term mutual funds, the company’s earnings management increases by 2.8%. In the high information
asymmetry group, the increase is 4.6%. This suggests that when short-term mutual funds invest in high infor-
mation asymmetry companies, they are more likely to focus on short-term returns. For long-term mutual
funds, when the extent of information asymmetry is low, the proportion of shareholdings is significantly pos-
itively related to negative abnormal accruals. This suggests that in a relatively transparent information envi-
ronment, long-term investors can mitigate negative earnings management and increase earnings quality.
However, when the extent of information asymmetry is high, all of the coefficients for long-term mutual funds
are non-significant. Therefore, when the information environment is relatively transparent, the monitoring
effect of long-term mutual funds on listed companies’ earnings management is stronger, consistent with
Hypothesis 3. Based on these results, we suggest that it is important to promote the development of long-term
institutional investors by expanding information channels and increasing the information transparency
between listed companies and investors.
In summary, we find that although short-term mutual funds may exploit their information advantage to
trade frequently, when the information environment is more transparent, long-term investors can mitigate
earnings management and the earnings information published by listed companies will be more credible. These
results reveal the importance of improving transparency for institutional investors. We propose that regulators
could prevent institutional investors from manipulating profits and reducing companies’ earnings quality by
strengthening information disclosure and transparency.
5. Additional tests
Because Chinese accounting standards changed significantly in 2006, we create a dummy variable to exam-
ine the effect of this change. The dummy variable is defined as follows: all years before and including 2006 take
the value of 0, and all subsequent years take the value of 1. The interaction terms of this year dummy variable
and the other main variables reflect the differences before and after the change in accounting standards. We
run the regression using Eqs. (12) and (13), and the results are shown in Tables 9 and 10.
DD eqi;t ¼ ai þ b1 SDi;t þ b2 after2006 þ b3 SD after2006 þ b4 Con Vari;t þ ei;t ð12Þ
DD eqi;t ¼ ai þ b1 Instii;t þ b2 after2006 þ b3 Instii;t after2006 þ b4 Con Vari;t þ ei;t ð13Þ
From Table 9, we find that the interaction term coefficients for information asymmetry and the year
dummy variable are both significantly positively related to positive and negative earnings quality. This indi-
cates that after 2006, the association between information asymmetry and positive earnings management is
Table 9
Regression results for information asymmetry and earnings management using year dummy variables.
|DD_abs| DD_eq(+) DD_eq()
SD 0.716*** 0.557* 0.601***
(4.839) (1.872) (3.478)
after2006 0.001 0.030*** 0.017***
(0.342) (3.452) (3.396)
SD after2006 0.148 0.659* 0.523***
(0.879) (1.906) (2.708)
TO 0.000*** 0.001*** 0.000
(2.648) (3.630) (0.662)
Duality 0.002 0.005 0.002
(0.808) (1.217) (0.866)
Out_ratio 0.007 0.004 0.013**
(1.487) (0.369) (2.421)
Top1 0.084*** 0.061 0.023
(3.357) (1.216) (0.782)
Top2_10 0.067*** 0.069 0.016
(3.170) (1.603) (0.651)
CR_5 0.065*** 0.060 0.011
(2.732) (1.256) (0.385)
LnSize 0.008*** 0.031*** 0.005***
(6.229) (11.813) (3.230)
LnMB 0.001 0.019*** 0.005***
(0.511) (5.741) (3.269)
Offering 0.000 0.002 0.002
(0.331) (0.929) (1.532)
SOE 0.004 0.013* 0.009***
(1.345) (1.770) (2.772)
Audit 0.024*** 0.015 0.016***
(5.566) (0.940) (3.873)
TopAudit 0.006* 0.003 0.003
(1.925) (0.571) (0.672)
Intercept 0.110*** 0.607*** 0.137***
(4.310) (10.491) (4.509)
Obs 7286 2873 4413
R2_adj 0.037 0.112 0.028
DD_eq(+) and DD_eq(). SD: the standard deviation of idiosyncratic risk as the information asymmetry measure. after2006: a year
dummy variable that equals 1 if the year is later than 2006 (excluding 2006), otherwise 0. SD after2006: interaction term of SD and
after2006. Duality: a dummy variable that equals 1 if the chairman and CEO are the same person, and 0 otherwise. Out_ratio: the
proportion of independent directors on the board. Top1: the proportion of shareholdings of the largest shareholder. Top2_10: the
aggregate proportion of shareholdings of the second to the 10th largest shareholders. CR_5: the aggregate proportion of shareholdings of
top 5 shareholders. LnSize: natural logarithm of total assets at the end of the year. LnMB: natural logarithm of the market-to-book ratio.
TO: turnover of listed companies in each year. Offering: a corporate financing needs dummy variable, which equals 1 if the company has an
SEO, rights offering or debt issue in period t, and 0 otherwise. SOE: a dummy variable that equals 1 if the company is a state-owned
enterprise, and 0 otherwise. Audit: a dummy variable that equals 1 if the audit opinion is unqualified, and 0 otherwise. TopAudit: a dummy
variable that equals 1 if the company’s audit firm is a Big 4 audit firm, and 0 otherwise. Intercept: intercept term. Obs: number of
observations. R2_adj: adjusted R-square.
*
**
***
more significantly positive, while the association with negative earnings management is more significantly neg-
ative. Overall, however, companies are more likely to artificially increase their earnings by using information
asymmetry after 2006.
Table 10
Regression results for long/short-term mutual funds and earnings management using year dummy variables.
Long Investor Short Investor
|DD_abs| DD_eq(+) DD_eq() |DD_abs| DD_eq(+) DD_eq()
***
Long_ratio 0.014 0.008 0.088
(0.844) (0.296) (3.869)
Long_ratio after2006 0.018 0.028 0.086***
(0.891) (0.814) (3.106)
Short_ratio 0.024* 0.043** 0.073***
(1.789) (2.116) (3.162)
Short_ratio after2006 0.016 0.017 0.083***
(1.180) (0.764) (3.649)
after2006 0.000 0.010*** 0.003** 0.000 0.010*** 0.004**
(0.133) (3.371) (2.202) (0.159) (3.380) (2.400)
TO 0.000 0.001** 0.000** 0.000 0.001* 0.000**
(0.064) (2.271) (2.119) (0.373) (1.947) (2.079)
Duality 0.002 0.005 0.002 0.002 0.005 0.002
(0.967) (1.246) (0.851) (0.920) (1.069) (0.775)
Out_ratio 0.008 0.001 0.013** 0.007 0.002 0.013**
(1.542) (0.093) (2.367) (1.515) (0.166) (2.349)
Top1 0.065** 0.024 0.003 0.093*** 0.069 0.012
(2.537) (0.468) (0.092) (3.590) (1.318) (0.388)
Top2_10 0.046** 0.030 0.006 0.073*** 0.076* 0.002
(2.091) (0.682) (0.241) (3.312) (1.659) (0.085)
CR_5 0.047* 0.025 0.013 0.072*** 0.066 0.005
(1.909) (0.498) (0.458) (2.914) (1.328) (0.170)
LnSize 0.006*** 0.028*** 0.005*** 0.006*** 0.028*** 0.005***
(5.069) (11.028) (3.668) (4.790) (10.939) (3.807)
LnMB 0.001 0.013*** 0.005*** 0.000 0.015*** 0.005***
(0.536) (4.188) (3.769) (0.308) (4.755) (3.638)
Offering 0.000 0.003 0.002 0.000 0.003 0.002
(0.388) (1.241) (1.264) (0.383) (1.291) (1.325)
SOE 0.004 0.013* 0.009*** 0.004 0.013* 0.009***
(1.282) (1.794) (2.789) (1.287) (1.884) (2.789)
Audit 0.024*** 0.012 0.015*** 0.024*** 0.012 0.015***
(5.358) (0.763) (3.653) (5.417) (0.776) (3.688)
TopAudit 0.007* 0.003 0.003 0.006* 0.003 0.003
(1.935) (0.544) (0.622) (1.880) (0.452) (0.643)
Intercept 0.066*** 0.538*** 0.161*** 0.057** 0.531*** 0.164***
(2.650) (9.560) (5.446) (2.286) (9.437) (5.625)
Obs 7286 2873 4413 7286 2873 4413
R2_adj 0.031 0.102 0.029 0.033 0.103 0.028
DD_eq(+) and DD_eq(). Long_ratio (Short_ratio) is the ratio between the number of shares held by long-term (short-term) mutual fund
investors and the total number of shares outstanding. after2006: year dummy variable that equals 1 if the year is later than 2006 (excluding
2006), otherwise 0. Long_ratio after2006 (Short_ratio after2006): interaction terms of Long_ratio(Short_ratio after2006) and
after2006. Duality: a dummy variable that equals 1 if the chairman and CEO are the same person, and 0 otherwise. Out_ratio: the
proportion of independent directors on the board. Top1: the proportion of shareholdings of the largest shareholder. Top2_10: the
aggregate proportion of shareholdings of the second to the 10th largest shareholders. CR_5: the aggregate proportion of shareholdings of
the top 5 shareholders. LnSize: natural logarithm of total assets at the end of the year. LnMB: natural logarithm of the market-to-book
ratio. TO: turnover of listed companies in each year. Offering: a corporate financing needs dummy variable, which equals 1 if the company
has an SEO, rights offering or debt issue in period t, and 0 otherwise. SOE: a dummy variable that equals 1 if the company is a state-owned
enterprise, and 0 otherwise. Audit: a dummy variable that equals 1 if the audit opinion is unqualified, and 0 otherwise. TopAudit: a dummy
variable that equals 1 if the company’s audit firm is a Big 4 audit firm, and 0 otherwise. Intercept: intercept term. Obs: number of
observations. R2_adj: adjusted R-square.
*
**
***
In Table 10, the coefficients for the interactions between long- and short-term investors and negative earn-
ings management are significantly negative, which suggests that both types of investors have a reduced mon-
itoring effect after 2006. Combining these results with those in Table 9, we conjecture, on the one hand, that
more companies might exploit information asymmetry to manipulate earnings after 2006, resulting in a reduc-
tion in institutional investors’ monitoring behavior. On the other hand, the rapid development of institutional
investors in China has created increasingly strong industry competition. Consequently, more and more insti-
tutional investors may shun long-term investments and monitoring of corporate governance in favor of short-
term returns. This would also explain the reduced monitoring effect of institutional investors in recent years.
6. Conclusion
In this paper, our sample includes all listed companies in the Chinese A-share market for the 2004–2010 per-
iod. We calculate earnings management based on the DD model proposed by Dechow and Dichev (2002),
define the standard deviation of idiosyncratic risk as the information asymmetry measure, divide mutual funds
in China into long- and short-term mutual funds following Yan and Zhang (2009), and use the proportion of
long- and short-term mutual funds as our proxy for institutional investors’ shareholdings. In addition, we con-
trol for companies’ fundamental characteristics, corporate governance, accounting standards, financing needs,
auditors and ultimate controllers and so forth, to investigate how the extent of information asymmetry and the
behavior of long- and short-term mutual funds affect listed companies’ earning management. We also divide
our sample into two groups based on the extent of information asymmetry and examine the effect of the inter-
action between information asymmetry and mutual funds on earnings management.
We find that low information asymmetry can increase companies’ earnings quality. This promotes effective
monitoring and encourages company management to publish reliable earnings information by expanding
information channels and increasing information transparency. Furthermore, the Chinese capital market is
speculative, and the good and the bad are intermingled in the mutual fund industry. The empirical results indi-
cate that compared with short-term mutual funds, long-term mutual funds play a supervisory role of company
management and effectively reduce negative management behavior, resulting in increased earnings quality.
To further examine the findings, we divide the sample into different groups based on the extent of informa-
tion asymmetry. Considering the difference in the investment style and objectives of long- and short-term
mutual funds, when information asymmetry is low, long-term mutual funds can monitor earnings manage-
ment more effectively and thus increase earnings quality. However, as the extent of information asymmetry
increases, the supervision effect of long-term mutual funds is seriously weakened, while short-term mutual
funds are associated with higher earnings management and thus reduce earnings quality. Therefore, it is
important for corporate governance to improve information disclosure systems in China and to regulate
the behavior of Chinese institutional investors.
Acknowledgements
We thank anonymous referees, Donghua Chen, Yuan Ding, Rui Ge, Feng Liu, Xijia Su, Jinsong Tan and
other participants at the CJAR Summer Research Workshop 2012 for their valuable suggestions. All errors
are our own. We also acknowledge financial support from the Nature Science Foundation of China (NSFC:
71173078; 70803013).
Appendix A
A.1. Ball and Shivakumar (2005) piecewise nonlinear regression model
Based on the DD model, Ball and Shivakumar (2005) propose a piecewise nonlinear regression model to
improve traditional models. They argue that traditional linear models cannot reflect the nonlinear properties
of accruals, so they introduce DCF and the interaction of DCF and CF to adjust the DD model. Wang (2006)
and Yang et al. (2007) both use the BS model to measure earnings management. The BS model increases the
explanatory power of traditional models, and it only requires financial information for 1 year before and after:
ACC t ¼ b0 þ b1 CF t þ b2 CF t1 þ b3 CF tþ1 þ b4 DCF t þ b5 DCF t CF t þ et ð13Þ

where ACCt is total accruals in period t, CFt, CFt1 and CFt + 1 are the operational cash flows in periods t,
t 1 and t + 1, respectively. All variables are scaled by average total assets to eliminate the size effect. DCFt
equals 1 if CFt–CFt1 < 0, and 0 otherwise. Following the CSRC industry classification standard, we regress
by each industry and each year, and the residual et is the proxy for earnings management, BS_eq. Similar to
the DD model, we distinguish between positive earnings management BS_eq(+) and negative earnings man-
agement BS_eq(). The higher the absolute value of |BS_eq|, the higher the extent of earnings management.
References
Armstrong, C.S., Guay, W.R., Weber, J.P., 2010. The role of information and financial reporting in corporate governance and debt
contracting. Journal of Accounting and Economics 50 (2–3), 179–234.
Ball, R., Shivakumar, L., 2005. Earnings quality in UK private firms: comparative loss recognition timeliness. Journal of Accounting and
Economics 39, 83–128.
Bartov, E., Radhakrishnan, S., Krinsky, I., 2000. Investor sophistication and patterns in stock returns after earnings announcements. The
Accounting Review 75 (1), 43–63.
Bertrand, M., Mehta, P., Mullainathan, S., 2002. Ferreting out tunneling: an application to Indian business groups. Quarterly Journal of
Economics 117, 121–148.
Bhattacharya, N., Desai, H., Venkataraman, K., forthcoming. Does earnings quality affect information asymmetry? Evidence from
trading costs. Contemporary Accounting Research.
Brous, P.A., Kini, O., 1994. The valuation effects of equity issues and the level of institutional ownership. Financial Management 23, 33–
46.
Brown, S.J., Goetzmann, W.N., 1997. Mutual fund styles. Journal of Financial Economics 43, 373–399.
Bushee, B.J., 1998. The influence of institutional investors on myopic R&D investment behavior. The Accounting Review 73 (3), 305–333.
Bushee, B.J., Noe, C., 2000. Corporate disclosure practices, institutional investors, and stock return volatility. Journal of Accounting
Research 38, 171–202.
Cheng, S.Q., 2006. An empirical study on the relationship between institutional ownership and the listed firms’ accounting earnings
information. Management World 9, 129–136 (in Chinese).
Chung, R., Firth, M., Kim, J.B., 2002. Institutional monitoring and opportunistic earnings management. Journal of Corporate Finance 8,
29–48.
Coffee, J.C., 1991. Liquidity versus control: the international investor as corporate monitor. Columbia Law Review 91 (6), 1277–1368.
Dechow, P.M., Dichev, I.D., 2002. The quality of accruals and earnings: the role of accrual estimation errors. Accounting Review 77, 35–
59.
Deng, K.B., Tang, X.Y., 2010. Can institutional investors really decrease earnings management? Evidence from China’s listed firms.
Industrial Economics Research 5, 71–86 (in Chinese).
Dierkens, N., 1991. Information asymmetry and equity issues. Journal of Finance and Quantitative Analysis 26, 181–199.
Dye, R.A., 1988. Earnings management in an overlapping generations model. Journal of Accounting Research 26, 195–235.
Fang, J.X., Hong, J.Q., 2007. Listed firms’ information disclosure quality and securities analysts’ earnings forecasts. Securities Market
Herald 3, 25–30 (in Chinese).
Froot, K.A., Scharfstein, D.S., Stein, J.C., 1992. Herd on the street: information inefficiencies in a market with short-term speculation.
Journal of Finance 47, 1461–1484.
Gao, L., Zhang, J., 2008. Corporate governance, institutional investors and earnings management. Accounting Research 9, 64–72 (in
Chinese).
Graves, S.B., 1988. Institutional ownership and corporate R&D in the computer industry. Academy of Management Journal 31 (2), 417–
428.
Hakim, F., Triki, F., Omri, A., 2008. Earnings quality and equity liquidity: evidence from Tunisia. International Journal of Managerial
and Financial Accounting 2, 147–165.
Huang, Q., 2009. An empirical study on the relationship between institutional investors and earnings management of China listed
companies. Modern Economic Science 4, 108–115 (in Chinese).
Hunt, A., Moyer, S., Shevlin, T., 1996. Managing interacting accounting measures to meet multiple objectives: a study of LIFO firms.
Journal of Accounting and Economics 21 (3), 339–374.
Hunton, J.E., Libby, R., Mazza, C.L., 2006. Financial reporting transparency and earnings management. The Accounting Review 81 (1),
135–157.
Jiambalvo, J., Rajgopal, S., Venkatachalam, M., 2002. Institutional ownership and the extent to which stock prices reflect future earnings.
Contemporary Accounting Research 19 (1), 117–145.
Jo, H., Kim, Y., 2007. Disclosure frequency and earnings management. Journal of Financial Economics 84 (5), 561–590.
Kahn, C., Winton, A., 1998. Ownership structure, speculation, and shareholder intervention. Journal of Finance 53 (1), 99–129.
Kang, J.K., Shivdasani, A., 1995. Firm performance, corporate governance and top executive turnover in Japan. Journal of Financial
Economics 38, 29–58.
Kaplan, S.N., Minton, B.A., 1994. Appointments of outsiders to Japanese boards: determinants and implications for managers. Journal of
Financial Economics 36, 225–257.
Klassen, K.J., 1997. The impact of inside ownership concentration on the trade-off between financial and tax reporting. The Accounting
Review 72 (3), 455–474.
Koh, P.S., 2007. Institutional investor type, earnings management and benchmark beaters. Journal of Accounting and Public Policy 26,
267–299.
Kong, D.M., Fu, K.H., 2005. A study on market reaction and impact factors of the Chinese stock market’s seasoned equity offerings.
World Economy 10, 51–59 (in Chinese).
Lev, B., Thiagarajan, S.R., 1993. Fundamental information analysis. Journal of Accounting Research 31 (2), 190–215.
Liu, F., He, J.G., 2004. Ownership structure and substantial shareholders’ choice in interest realizing methods: tentative study on
tunneling in Chinese capital market. China Accounting Review 2, 141–158 (in Chinese).
Liu, L.Y., Peng, E.Y., 2006. Institutional Ownership Composition and Accruals Quality. Working Paper.
Ljungqvist, A., Mally, C.J., Marston, F.C., 2009. Rewriting history. Journal of Finance 64, 1935–1960.
Porter, G., 1992. Accounting earnings announcements, institutional investor concentration and common stock returns. Journal of
Accounting Research 30 (1), 46–155.
Prowse, S.D., 1990. Institutional investment patterns and corporate financial behavior in the United States and Japan. Journal of
Financial Economics 27, 43–66.
Ramalingegowda, S., Yu, Y., 2012. Institutional ownership and conservatism. Journal of Accounting and Economics 53, 98–114.
Richardson, V.J., 2000. Information asymmetry and earnings management: some evidence. Review of Quantitative Finance and
Accounting 15 (4), 325–347.
Schipper, K., 1989. Commentary on earnings management. Accounting Horizons 3 (4), 91–102.
Sun, Z., Wang, A., Zheng, L., 2012. The road less traveled: strategy distinctiveness and hedge fund performance. Review of Financial
Studies 25 (1), 96–120.
Trueman, B., Titman, S., 1988. An explanation for accounting income smoothing. Journal of Accounting Research 26, 127–139.
Wang, D.C., 2006. Founding family ownership and earnings quality. Journal of Accounting Research 44, 619–656.
Warfield, T.D., Wild, J.J., Wild, K.L., 1995. Managerial ownership, accounting choices, and informativeness of earnings. Journal of
Accounting and Economics 20, 61–91.
Xia, L.J., Lu, X.N., 2005. On the relationship between earnings management and information disclosure quality of listed firms.
Contemporary Economy & Management 10, 145–150 (in Chinese).
Yang, D.M., 2005. The disclosure of information forecasts and earnings management. Chinese Journal of Management Science 7, 108–112
(in Chinese).
Yang, D.M., Lin, B., Xin, Q.Q., 2007. Earnings quality, investors’ reasonless behavior and earnings momentum. Journal of Financial
Research 2, 122–132 (in Chinese).
Yan, X.M., Zhang, Z., 2009. Institutional investors and equity returns: are short-term institutions better informed? Review of Financial
Studies 22, 893–924.
Zhao, T., Zheng, Z.X., 2002. Asymmetric information and manipulate stock in institutions behavior. Economic Research Journal 7, 41–48
(in Chinese).

Information Asymmetry, Mutual Funds and Earnings Management: Evidence From China

Uploaded by

Copyright:

Available Formats

Information Asymmetry, Mutual Funds and Earnings Management: Evidence From China

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Information Asymmetry, Mutual Funds and Earnings Management: Evidence From China

Uploaded by

Copyright:

Available Formats

China Journal of Accounting Research 6 (2013) 187–209

Contents lists available at ScienceDirect

China Journal of Accounting Research

Information asymmetry, mutual funds and

⇑ Corresponding author. Tel.: +86 15927068886.

Production and hosting by Elsevier

2. Literature review and research hypotheses

3. Data and research design

3.1. Sample selection and data sources

3.2. Variable deﬁnitions

3.2.1. Earnings management (EM)

3.2.4. Other control variables

4.1. Descriptive statistics

Y. Dai et al. / China Journal of Accounting Research 6 (2013) 187–209

Panel B: Correlation matrix

4.2. Empirical results and analysis

Y. Dai et al. / China Journal of Accounting Research 6 (2013) 187–209

DD eqi;t ¼ ai þ b1 Instii;t þ b2 after2006 þ b3 Instii;t after2006 þ b4 Con Vari;t þ ei;t ð13Þ

A.1. Ball and Shivakumar (2005) piecewise nonlinear regression model

ACC t ¼ b0 þ b1 CF t þ b2 CF t1 þ b3 CF tþ1 þ b4 DCF t þ b5 DCF t CF t þ et ð13Þ

You might also like