Accrual-Based and Real Activity Earnings Management at The Back Door Evidence From Chinese Reverse Mergers

Pacic-Basin Finance Journal 35 (2015) 317339
Contents lists available at ScienceDirect
Pacic-Basin Finance Journal

journal homepage: www.elsevier.com/locate/pacfin
Accrual-based and real activity earnings

management at the back door: Evidence from
Chinese reverse mergers
Tingting Zhu a, Meiting Lu b, Yaowen Shan c,, Yuanlong Zhang d
a
International Finance and Risk Management Research Center, School of Finance and Statistics, East China Normal University, China
b
Department of Accounting and Corporate Governance, Macquarie University, Australia
c
UTS Business School, University of Technology, Sydney, Australia
d
School of Finance and Statistics, East China Normal University, China
a r t i c l e i n f o a b s t r a c t
Article history: We examine how Chinese reverse merger (RM) rms trade off and con-
Received 30 May 2014 duct income-increasing earnings management through accrual-based
Accepted 26 January 2015 and real activities manipulation strategies. We nd that Chinese RM
Available online 31 January 2015
rms engage in both real activities and accrual-based manipulation at
higher levels than non-Chinese RM rms, regular US rms and other
JEL classication:
Chinese US-listed rms. Further analysis suggests that Chinese RM
G30
rms use real activities and accrual-based manipulation as substitutes
G34
M41
and tend to transition to real activities management in the years after
a reverse takeover. Big 4 auditors can effectively constrain both real ac-
Keywords: tivities and accrual-based earnings management in Chinese RM rms.
Earnings management We also nd that accruals manipulation is more costly relative to real ac-
Reverse merger tivities management in the short term because it predicts changes in
Accrual-based manipulation post-acquisition operating performance in Chinese RM rms. Overall,
Real activities manipulation the results provide practical implications to regulators, investors and au-
Firm performance
ditors on the channels through which Chinese RM rms manipulate
earnings and the economic consequence of those manipulations.
2015 Elsevier B.V. All rights reserved.
We are grateful for suggestions from Balasingham Balachandran (the editor), an anonymous reviewer, Martin Bugeja, Peter Easton,
Jere Francis, Wen He, Wei Jiang, Zoltan Matolcsy and Stephen Taylor, as well as workshop participants at the University of Technology
Sydney and East China Normal University. We also appreciate the research assistance of Dongjuan Zang, Yuxing Hou, Sheng Guo, Jiaqi
Ren, Zhuoli Wang and Jiaye Zhang. Tingting Zhu gratefully acknowledges the work sponsored by Shanghai Pujiang Program (No.
12PJC042), and Yaowen Shan acknowledges the support received from a UTS Business School research grant.
Corresponding author at: University of Technology, Sydney Business School, PO Box 123, Broadway, NSW 2007, Australia. Tel.: +61 2
9514 3627.
E-mail address: yaowen.shan@uts.edu.au (Y. Shan).
http://dx.doi.org/10.1016/j.pacn.2015.01.008
0927-538X/ 2015 Elsevier B.V. All rights reserved.
318 T. Zhu et al. / Pacic-Basin Finance Journal 35 (2015) 317339
1. Introduction
The recent accounting crisis and the market collapse of US-listed Chinese companies have attracted exten-
sive media coverage, shareholder concerns and regulatory attention. The concerns are particularly focused on
Chinese rms listed in the United States through reverse mergers (RMs). In these mergers a publicly traded US
rm is acquired by a Chinese private rm, which subsequently controls the combined business and goes pub-
lic in the US market. In June 2011, the US Securities and Exchange Commission (SEC) issued an investor warn-
ing related to the accounting and disclosure practices of Chinese RM rms. Further regulatory actions by the
SEC, the NASDAQ and the New York Stock Exchange (NYSE) resulted in more than 30 Chinese RM rms being
subject to suspended trading, halted trading or delisting for various reasons. New rules were also adopted to
toughen the standards for RM companies.1
In this study, we examine how Chinese RM rms trade off and conduct two distinct behaviors for income-
increasing earnings management, namely accrual-based earnings management (AEM) and real activities
earnings management (REM), at the back door.2 Chinese RMs present a unique setting for understanding
earnings management. While prior research extensively documents earnings management in initial public of-
ferings (IPO),3 recent studies argue that the evidence is driven by measurement errors in discretionary ac-
cruals arising from the investment of IPO proceeds in working capital (Armstrong et al., 2012; Ball and
Shivakumar, 2008; Cecchini et al., 2012). In contrast, reverse merger, as an alternative way of going public,
does not raise capital by issuing stocks. Consequently, tests of earnings management around RMs are not
undermined by the measurement errors identied in IPOs arising from the investment of raised funds (Ball
and Shivakumar, 2008). Thus, our study adds to the IPO literature and provides novel data on income-
increasing earnings management behaviors of rms around the time of going public.
In addition, examining Chinese RMs also enables conducting powerful tests on earnings management. Un-
like regular US rms listed through an IPO, an RM bypasses regulatory requirements set for IPO rms and does
not require the support of underwriters who monitor the IPO process. The lack of underwriter monitoring and
less stringent regulation requirements in the process of reverse acquisition increase earnings management
opportunity in Chinese RM rms (Feldman, 2009; Lee and Masulis, 2011). Moreover, a typical RM rm
tends to have relatively low fundamental quality and a high degree of information asymmetry (Adjei et al.,
2008; Gleason et al., 2005). Compared with regular US rms and other Chinese US-listed rms, Chinese RM
rms with inferior fundamentals have higher incentives and greater opportunity to manage earnings in the
years around reverse acquisition to boost stock prices, sustain overvaluation or increase the probability of rais-
ing capital in the future.
Finally, the background for Chinese RM rms is a transitional economy without a tradition of transparency
and effective legal enforcement. In contrast, other RM rms mostly originate in countries with strong legal en-
forcement and regulations such as the United States, United Kingdom and Canada (more than 90% of the sam-
ple). Ball et al. (2003) and Leuz et al. (2003) nd that the extent of earnings management tends to be greater
in rms from countries with weak legal enforcement. In addition, He et al. (2012) argue that the US stock ex-
changes and information intermediaries have become lax in screening Chinese rms for listing in recent years,
leading to the selection of a large group of Chinese RM rms with inferior quality in corporate governance.
Since good corporate governance serves as an effective mechanism to alleviate income-increasing earnings
management behaviors (Cornett et al., 2008; Hazarika et al., 2012; Laux and Laux, 2009), Chinese RM rms
with ineffective governance structure are more likely to engage in earnings manipulation.
We examine two distinct channels of income-increasing earnings manipulations. Accrual-based earnings
management involves managerial intervention in the nancial reporting process via accounting choices and
estimates under generally accepted accounting principles (GAAP). In contrast, real activities earnings man-
agement involves managers taking actions that have cash ow consequences, such as adjusting the timing
or scale of the rm's operating, investing and/or nancing transactions. Examples of real activities
1
On November 8, 2011, the NYSE and NASDAQ, with the approval of the SEC, adopted new rules that toughen the standards for RM
companies. Under these rules, a RM company that wants to apply for listing on a US stock exchange is required to (1) complete a one-
year seasoning period following the RM; and (2) le all required reports with the SEC, including an audited nancial statement.
2
Throughout this study, we use accrual-based earnings management, accrual-based manipulation, accruals earnings management, and
accruals manipulation interchangeably. We also use real activities earnings management, real activities manipulation and real activities
management interchangeably.
3
See for example, Teoh et al. (1998a, 1998b), Teoh and Wong (2002), Morseld and Tan (2006), and Fan (2007).
T. Zhu et al. / Pacic-Basin Finance Journal 35 (2015) 317339 319
management include reducing discretionary expenditures (e.g., R&D expenses), overproducing products to
lower reported cost of goods sold (COGS), and boosting sales through increased price discounts or more le-
nient credit terms (Roychowdhury, 2006).
While companies can use AEM and/or REM to achieve certain earnings targets, several aspects of the two
mechanisms differ. Managers may prefer one strategy over the other depending on the circumstances. First,
REM is less susceptible to the scrutiny of auditors and regulators; thus, managers tend to prefer taking real
economic actions to improve accounting appearances over engaging in AEM (Graham et al., 2005). Second,
AEM is more costly in the short term, whereas REM affects a rm's cash ows and is likely to result in greater
long-term negative impacts on rm performance (Graham et al., 2005). Third, AEM is largely constrained by
the underlying accounting relation presumed in nancial statements and can be reversed in future periods.
Firms that previously engaged in AEM may have to switch to manipulating real activities, if their motive of
earnings management remains in place (Cohen and Zarowin, 2010; Gunny, 2010).4
Using earnings management measures adjusted for rm performance,5 we nd that Chinese RM rms en-
gage in signicantly higher levels of REM and AEM than other RM rms, a matched sample of US domestic
companies6 and US-listed Chinese non-RM rms during years around reverse acquisition. Consistent with
the costs and balance-sheet constraints of AEM, we also nd that Chinese RM rms use REM and AEM as sub-
stitutes. Chinese RM rms conduct AEM predominantly in the year in which the reverse takeover occurs. Due
to limits in the amount of AEM in which a rm can engage, Chinese RM rms also resort to REM by substan-
tially reducing discretionary expenses in the year of reverse acquisition. However, in years after reverse merg-
ing, the rms tend to transition to REM.
We then investigate whether Big 4 auditors effectively monitor and constrain earnings management in
Chinese RM rms.7 Prior studies suggest that Big 4 auditors restrict the attempted use of AEM (Becker
et al., 1998; Francis et al., 1999), but evidence regarding REM indicates that the presence of Big 4 auditors is
associated with higher levels of real activities manipulation (Chi et al., 2011). In contrast, our results show
that Chinese RM rms with Big 4 auditors have lower levels of both AEM and REM through discretionary
expenses than rms with non-Big 4 auditors.
Finally, we examine the impacts of earnings management on future operating performance of Chinese RM
rms. We nd that the decline in their operating performance during the two years subsequent to a reverse
takeover is predominantly driven by AEM rather than REM. This nding supports the view that AEM tends to
be more costly in the short term, while REM may have a larger negative impact on future performance in the
long term (Graham et al., 2005).
Our study adds to the growing literature that examines various aspects of Chinese RM rms to understand
their fraudulent issues, including internal control weakness (Baker and Biddle, 2012), nancial health and
performance (Ang et al., 2012; Jindra et al., 2012; Lee et al., 2013), market reactions (Darrough et al., 2012)
and adverse selection problems (He et al., 2012). Our paper is closely related to several concurrent studies
(Chen et al., 2013; Chu et al., 2014; Givoly et al., 2014) that test nancial reporting quality in Chinese RMs,
yet it differs from these studies in several important ways.
First, while the concurrent studies examine whether and why Chinese RM rms have lower nancial
reporting quality, our study focuses on a unique perspective and addresses different research questions. In
particular, we investigate how Chinese RM rms trade off and conduct income-increasing AEM and REM be-
haviors, and how earnings management affects future performance. The concurrent studies all use accrual-
based measures to examine nancial reporting quality of Chinese RMs; however, earnings management
4
Managers also have to determine the extent of real activities management early in the nancial year because real activities, as op-
posed to accounting transactions, cannot be adjusted at the scal year-end. There is a risk involved with relying on accrual-based manip-
ulation alone since the amount being manipulated may fall short of the desired threshold (Roychowdhury, 2006; Zang, 2012).
5
Since Chinese RM rms are found to have signicantly higher rm performance (in terms of return-on-asset [ROA]), we employ
performance-adjusted measures of earnings management. In particular, we measure AEM by abnormal accruals estimated from the mod-
ied Jones model with ROA as an additional control (Dechow et al., 1995; Kothari et al., 2005). Similarly, we use performance-adjusted
abnormal discretionary expenditure and abnormal production costs to measure REM (Cohen et al., 2013; Roychowdhury, 2006).
6
The matched sample of regular US rms is generated by matching regular US rms with Chinese RMs based on year, industry, rm
size and sales growth. Our following analysis shows that Chinese RM rms differ signicantly in rm size and sales growth from regular
US rms. We require the matched US rms to have a rm size during the RM year within 20% and the closest growth rate in sales.
7
The Big 4 auditors are Deloitte, Ernst & Young, KPMG and PricewaterhouseCoopers. Throughout the paper, we use Big 4 generically to
designate Big 4, Big 5 and Big 6 auditors, depending on the sample period.
behaviors, which are the focus of our study, do not necessarily lead to poor nancial reporting quality identi-
ed by accrual-based measures. In fact, REM cannot be detected by accrual-based measures of nancial
reporting quality, and rms can use REM as a substitute for AEM. Consistent with this notion, we nd that
Chinese RMs use REM and AEM as substitutes and tend to transition to REM in the years after reverse takeover.
Second, the concurrent studies all use the absolute value of discretionary accruals (i.e., unsigned measures)
and document that Chinese RMs exhibit lower nancial reporting quality. However, the use of absolute discre-
tionary accruals is likely to lead to biased empirical conclusions (Hribar and Nichols, 2007) because Chinese
RM rms are more volatile in fundamentals.8 Our study follows the suggestion of Hribar and Nichols
(2007) and uses the signed measure of discretionary accruals to provide a more rened and robust
examination.
Equally as important, our research design of adopting performance-adjusted earnings management mea-
sures also differs from the design of concurrent studies, which use accrual-based measures of earnings man-
agement and do not control for rm performance. Kothari et al. (2005) show that failure to control for
differences in fundamentals (such as rm performance) between the treatment and control groups may
lead to biased or over-claimed empirical conclusions in testing earnings management. This could be a serious
concern because the nancial performance of Chinese RM rms is found to be signicantly better than either
their US RM counterparts or a group of industry-size-date matched rms from the same exchange (Lee et al.,
2013).
Third, our study differs from the concurrent studies by examining the impact of earnings management on
future operating performance of Chinese RM rms. By doing so, we contribute to the literature of earnings
management and reverse takeover. We nd that performance changes in the two years following a reverse
takeover are mainly associated with AEM. These results contradict those of Cohen and Zarowin (2010),
who nd that the underperformance of seasoned equity offering (SEO) rms is largely driven by REM. Our re-
sults also provide a possible explanation of earnings management behaviors to the fact that reverse mergers
exhibit underperformance in operations and protability after reverse acquisition (Gleason et al., 2005).
Finally, our study adds to previous literature on the interaction between audit quality and earnings man-
agement. We nd that Big 4 auditors restrict the attempted use of not only AEM but also REM through discre-
tionary expenses. Our results contrast sharply with those of Chi et al. (2011) and Zang (2012), who nd that
Big 4 auditors cannot restrict REM. Our results also add to those of Chu et al. (2014), who document that the
presence of a Big 4 auditor in RMs is associated with higher nancial reporting quality as measured by accrual-
based proxies.
The rest of the paper is organized as follows. Section 2 discusses the institutional background of Chinese
RMs, reviews relevant literature and develops testable hypotheses. Sample construction, the measurement
of earnings management and descriptive statistics are discussed in Section 3. Section 4 presents empirical
results, and Section 5 concludes the paper.
2. Literature review and hypothesis development
2.1. Background and related literature on Chinese RM rms
An RM (also called a back door listing) is the acquisition of a publicly traded company by a private com-
pany (e.g., a Chinese private operating company) to enable the private company to bypass the lengthy and
complex process of going public. The public company, typically with little business activity and assets, is re-
ferred to as a shell company. Once the merger is completed, the owners and management of the private
company obtain effective voting and operational control of the combined business, and the private held com-
pany subsequently becomes publicly listed. In recent years, reverse acquisition has become an increasingly
8
In particular, Hribar and Nichols (2007) show analytically that the mean of absolute discretionary accruals varies with the standard
deviation of signed discretionary accruals. As a result, tests based on absolute discretionary accruals are exposed to a class of correlated
omitted variables that are generally not a concern in research using signed discretionary accruals. Hribar and Nichols (2007) nd that
the absolute discretionary accruals have the highest correlations with volatility of sales and volatility of cash ows, but that prior work
virtually ignores these characteristics. Using simulation analysis, they show that even modest correlation between these volatility metrics
and the partitioning variable (e.g., the indicator of Chinese RM rms) severely increases the risk of incorrectly rejecting the null hypoth-
esis. In fact, RMs, particularly Chinese RM rms, are found to have higher debt and higher volatility in fundamentals (Adjei et al., 2008; Chu
et al., 2014; Gleason et al., 2005; Lee et al., 2013).
popular means for Chinese rms to go public in the US market. During 20072010, there were more than 150
Chinese private operating companies that accessed US markets by acquiring a shell company publicly traded
in the United States.
In June 2010, Muddy Waters published a research report that claimed that Orient Paper, a Chinese RM
company, had signicantly overstated its revenues and assets by around $30 million since 2009. The detailed
report, together with subsequent extensive negative media coverage and signicant market reaction, trig-
gered a wave of regulatory investigations and shareholder litigation against Chinese RM rms.
The accounting crisis and market collapse of Chinese RM rms in the United States has prompted academic
researchers to examine these companies from various perspectives. One stream of the literature examines
whether the integrity of the US market is impaired by Chinese RM rms with low fundamental quality.
Jindra et al. (2012) compare US-listed Chinese IPO rms with Chinese RM rms and nd that Chinese RM
rms have a smaller size, lower analyst following and higher leverage. Similarly, He et al. (2012) document
that Chinese RM rms have worse fundamental characteristics compared with their domestic peers, while
those of Chinese IPOs are superior. The researchers suggest that reverse acquisitions have induced an adverse
selection problem in the US market and the accounting scandals in Chinese RM rms have punished all US-
listed Chinese rms. In contrast, Lee et al. (2013) nd no evidence that US-listed Chinese RMs have poor fun-
damental quality. Specically, they show that Chinese RM rms are nancially healthier than US RM rms at
the initial 10-K reporting date and continue to exhibit better performance in terms of protability, nancial
leverage and market capitalization than their US counterparts.
Another stream of the literature focuses on accounting and disclosure practices among Chinese RM rms
and their impact on stock prices. Chen et al. (2013), Chu et al. (2014) and Givoly et al. (2014) nd that Chinese
RMs tend to have lower nancial reporting quality, as measured by the absolute value of accrual-based prox-
ies. In addition, Baker and Biddle (2012) document that Chinese RM rms are more likely to disclose that their
internal controls are ineffective under SOX 302, particularly in the areas of nancial statement preparation,
personnel and remediation. Ang et al. (2012) examine the determinants of committing nancial fraud for
companies listed overseas. They suggest that Chinese RM rms, compared with US-listed Chinese IPO rms,
have a higher probability of fraud, particularly those with a more autocratic governance structure and a
more concentrated ownership structure. Finally, Darrough et al. (2012) examine the spillover effect of the
news about nancial fraud allegedly committed by Chinese RM rms. They show that the stock market reac-
tion to fraud news appears to be China bashing rather than RM bashing.
2.2. Hypothesis development
2.2.1. Earnings management of Chinese RM rms

The extant literature provides ample evidence that rms have incentives to overstate earnings around sig-
nicant corporate events such as takeovers and equity offerings to boost stock prices, lower their nancing
costs or both (e.g., Cohen and Zarowin, 2010; Louis, 2004; Teoh et al., 1998a, 1998b).9 In addition, Jensen's
(2005) agency theory of overvalued equity suggests that managers of overvalued companies turn to further
earnings management and even fraudulent practices to maintain the appearance of growth and value crea-
tion. Chi and Gupta (2009) and Badertscher (2011) provide supportive evidence that overvaluation motivates
managers to engage in income-increasing earnings management to sustain an overvalued stock price.
A reverse merger, compared to regular US rms listed through an IPO, bypasses regulatory requirements
set for IPO rms by the stock exchange and effectively allows access to the nancial market (Feldman,
2009).10 It does not require raising capital or obtaining the support of underwriters who oversee the IPO pro-
cess. Prior literature suggests that RM rms are generally younger, smaller in size, less protable and more
likely to be delisted within a few years of going public (e.g., Adjei et al., 2008; Gleason et al., 2005). As a result,
RM rms are typically characterized as having lower fundamental quality and a higher degree of information
asymmetry. Therefore, Chinese RM rms with inferior fundamental quality have higher incentives to engage
9
Louis (2004) nds that in share-for-share takeovers, acquiring rms tend to overstate their earnings preceding a stock swap
announcement.
10
For example, the standards set by NYSE require IPO rms to have a minimum of $2 million of earnings in each of the two most recent
years or $75 million revenues in the most recent nancial year. In contrast, the requirements for maintaining an existing listing (such as
the acquisition of a public rm via RM) are not as stringent.
in income-increasing earnings management to sustain overvaluation or reduce nancing costs, compared to

regular US rms and Chinese US-listed non-RM rms.
In addition, the presence of a reputable underwriter effectively restrains earnings management by IPO
rms (Lee and Masulis, 2011). The lack of underwriter monitoring and less stringent disclosure and compli-
ance requirements in the process of a reverse acquisition also increase the opportunity for earnings manage-
ment in Chinese RM rms. Thus, we expect that Chinese RM rms, due to the lack of market and regulatory
scrutiny in the RM process, have higher incentives and opportunity to manipulate earnings, resulting in a
higher level of income-increasing earnings management than regular US rms and Chinese non-RM rms.
We also predict that Chinese RM rms engage in more earnings manipulation than other RM rms. Unlike
other RM rms originating in countries with strong legal enforcement and regulation (such as the United
States, United Kingdom and Canada), Chinese RM rms arise from a transitional economy without a tradition
of transparency and effective law enforcement. Ball et al. (2003) and Leuz et al. (2003) suggest that country-
specic institutional structure and legal environment signicantly determine managers' incentives and be-
haviors in nancial reporting; they also nd that the level of earnings management tends to be higher in coun-
tries with weak legal enforcement.
In addition, extensive literature suggests that good corporate governance serves as an effective mechanism
to alleviate income-increasing earnings management behaviors (Cornett et al., 2008; Hazarika et al., 2012;
Laux and Laux, 2009). He et al. (2012) provide evidence that US listings have selected a large group of Chinese
RM rms with inferior fundamental quality and corporate governance. Since corporate governance shapes the
actions of the CEO and top managers, Chinese RM rms with ineffective governance structure are more likely
to engage in earnings manipulation. The reasons for adverse selection are two-fold. First, the unique approval
system of IPO listing in China prevents less protable companies and those without political connections from
going public in China; therefore, small private rms with inferior fundamental and accounting quality tend to
have high incentives to seek access to and raise capital in the US markets (Chen and Yuan, 2004; Piotroski and
Zhang, 2012).11 Second, the US stock exchanges and information intermediaries have become lax in screening
Chinese rms for listing due to the China growth story, particularly after the promulgation of SOX (Hung et al.,
2012). Inferior Chinese rms have been selected by information intermediaries and tend to use reverse take-
overs to bypass the stringent IPO listing process (He et al., 2012).12
In summary, we predict that Chinese RM rms undertake higher levels of income-increasing earnings
management around the year of a reverse takeover compared with other RM rms, regular US rms and Chi-
nese non-RM rms, because of the lack of scrutiny associated with the RM process and the less effective law
enforcement and governance environment in China. Prior research suggests that both AEM and REM are likely
to be used by rms with strong incentives to inate reported earnings. In events of SEO, Cohen and Zarowin
(2010) and Kothari et al. (2012) nd that SEO rms conduct both real activities and accrual-based manipula-
tion during the offering year. Furthermore, Wongsunwai (2013) report that IPO rms in the US undertake
AEM and REM during the IPO year. In contrast to several concurrent studies of Chinese RMs testing accrual-
based measures of nancial reporting quality, our study examines both AEM and REM. Our rst hypothesis
is as follows:
H1. Chinese RM rms exhibit higher levels of income-increasing accrual-based and real activities earnings
management around the year of reverse acquisition compared with non-Chinese RM rms, regular US rms
and Chinese non-RM rms.
2.2.2. The trade-off between AEM and REM

Recent research suggests that managers use REM and AEM as substitutes (Badertscher, 2011; Cohen and
Zarowin, 2010; Zang, 2012). Cohen and Zarowin (2010) show that SEO rms' choices between two alternative
strategies vary predictably as a function of their ability to use accruals manipulation and the costs of doing so.
11
Chinese rms that apply for domestic IPOs are required to go through careful screening by the regulators against protability, rm
size and earnings management (Chen and Yuan, 2004). In addition, the determination process of IPO listing candidacy is found to be high-
ly political, and rms with strong political connections are more likely to obtain listing (Piotroski and Zhang, 2012; Yang, 2012).
12
See Wall Street Journal, May 10, 2012, China's Square Peg vs. the West Round Hole. After the promulgation of SOX (Section 302 in
2002 and Section 404 in 2006), many Chinese state-owned rms turned away from the US market owing to litigation and reputation risks
(Hung et al., 2012). As a result, US information intermediaries had to lower their standards and began to take in substandard clients such
as Chinese privately controlled rms.
Badertscher (2011) examines overvalued rms and nds that during the sustained period of overvaluation
managers use AEM in early years but REM in later years. Zang (2012) argues that rms face different levels
of constraints for the two earnings management methods and documents evidence that a rm's trade-off
decision depends on the relative costliness of the two strategies.
Consistent with existing studies, we predict that Chinese RM rms trade off AEM and REM methods and
use these two strategies as substitutes, and the relative degree of AEM vis--vis REM depends on the relative
costliness of these two methods. First, REM is less likely to be constrained and discovered by auditors and reg-
ulators than AEM because real operational decisions are more rmly within the domain of managers' exper-
tise and less likely to be discovered by auditors (Graham et al., 2005). Thus, the detection costs are relatively
low for REM. Second, AEM is largely constrained by the presumed relation in accounting statements. Given
the balance-sheet constraints of AEM, rms that utilized higher levels of AEM in previous years are likely to
switch to manipulating real activities in the current and future periods (Gunny, 2010). As a result, the exibil-
ity of a rm's accounting system can lead to differential costs and preferences for using REM over AEM. Third,
for rms with poor nancial performance and health, REM is more likely to be perceived as more costly than
AEM (Zang, 2012). This is because the primary goal for rms with poor performance and health is to improve
operations, and REM tends to impose higher economic costs in the long run (Graham et al., 2005). Therefore,
our second hypothesis is stated as follow:
H2. Chinese RM rms use real activities manipulation and accrual-based manipulation as substitutes.
2.2.3. Big 4 auditors and earnings management

Previous studies suggest that high-quality auditing potentially constrains the attempted use of AEM
because it is more likely to discover the behavior. Becker et al. (1998) and Francis et al. (1999) show that
rms with Big 4 auditors tend to have less AEM, implying that Big 4 auditors effectively restrict clients' oppor-
tunistic reporting behavior.
However, evidence relating audit quality to the use of REM is mixed. An auditor's responsibility is to ensure
that the accounting reports faithfully represent a rm's real activities, and they are not required to evaluate
managers' motivations for real activities decisions. As a result, REM draws less scrutiny from auditors
(Graham et al., 2005), and rms with high-quality auditors may resort to higher levels of REM to avoid the
monitoring of AEM. Consistent with this view, Chi et al. (2011) document that Big 4audited rms switch
to higher levels of REM. In contrast, Zang (2012) fails to nd a signicant and positive relationship between
REM and Big 4 auditors.
We hypothesize that Chinese RM rms with non-Big 4 auditors, given their strong incentives of earnings
management, engage in higher levels of income-increasing earnings management than those that do not em-
ploy Big 4 auditors. Auditing theory suggests that Big 4 auditors are of high quality because of the higher risk
to their reputation (DeAngelo, 1981) and the higher risk of litigation (Dye, 1993). In contrast, a small auditor
with fewer clients and less wealth at risk from litigation tends to have higher incentives to cheat in order to
retain an existing client. Therefore, our third hypothesis is stated as follows:
H3. Chinese RM rms with Big 4 auditors undertake less income-increasing earnings management than
those with non-Big 4 auditors.
2.2.4. Earnings management and future operating performance

Both AEM and REM impose economic costs on rms, but their impacts differ based on the length of time
that passes before a negative consequence is apparent in the rm's operating performance (Graham et al.,
2005). In the short term, REM is less costly than AEM since it is subject to less scrutiny by regulators and in-
vestors (Graham et al., 2005). On the other hand, REM could potentially create higher economic costs than
AEM in the long term.13 Consistent with prior studies (e.g., Cohen and Zarowin, 2010; Teoh et al., 1998a),
we expect that the economic costs arising from earnings manipulation would be reected in a rm's future
operating performance. To the extent that Chinese RM rms manage earnings upward around RMs using ei-
ther AEM or REM, we predict a negative association between the extent of earnings management and future
13
Firms engaging in real activities management are found to overproduce unnecessary inventory and aggressively reduce discretionary
expenses such as R&D investments, which can hurt long-term future cash ows (Roychowdhury, 2006).
operating performance in Chinese RM rms. Since AEM is more costly than REM in the short term, the nega-
tive association is likely to be stronger for AEM (Graham et al., 2005).
The negative effects of both AEM and REM behaviors on rm performance are documented in prior re-
search in the context of IPOs and SEOs. Teoh et al. (1998a) nd that IPO rms with high levels of AEM during
the IPO year experience a decline in future period stock returns and operating performance. Rangan (1998)
documents similar results for SEO rms in that the extent of AEM predicts post-SEO return and operating
underperformance. With respect to REM, Cohen and Zarowin (2010) show that operating performance sub-
sequent to SEOs is negatively related to the extent of earnings management. Consistent with the long-term
costs of REM, they nd that the decline in operating performance following SEOs is driven largely by REM rel-
ative to AEM. The above argument and supportive evidence lead to the following hypotheses:
H4a. Future operating performance of Chinese RM rms is negatively related to the extent of earnings
management in the year of reverse acquisition.
H4b. The negative association between accrual-based earnings management and future operating perfor-
mance of Chinese RM rms is stronger than the association between real activities management and future
performance in the short run.
3. Data and methodology
3.1. The sample
Our initial sample consists of all rms (excluding nancial and utility rms) that have required accounting
and stock price data from the merged CRSP and Compustat databases for the period of 19902011. To identify
US-listed Chinese RM rms, we rst compile a sample that consists of all US-listed Chinese rms and remove
rms having IPO dates and/or underwriter information. To ensure the accuracy of the sample of Chinese RMs,
we crosscheck the list with the Bloomberg list of Chinese RMs.14 We also manually search each rm's SEC
lings (10-K reports and 8-K forms) through the SEC website and identify RMs through the Corporate histo-
ry section in 10-K reports and the Change in shell company status section in 8-K forms. Our nal sample
includes 132 Chinese RMs with available accounting information.
To test our hypotheses, we also identify three control samples: non-Chinese RM rms, a matched sample
of regular US rms and US-listed Chinese non-RM rms.15 The sample of non-Chinese RM rms is obtained
from the Securities Data Corporation (SDC) database.16 The matched sample of regular US rms is generated
by matching regular US rms with Chinese RMs based on year, industry, rm size and sales growth. Our fol-
lowing analysis shows that Chinese RM rms differ signicantly in rm size and sales growth from regular US
rms. We require that the matched US rms have a rm size during the RM year within 20% and the closest
growth rate in sales.17
Table 1 presents the industry distribution of Chinese RM rms as well as the three control samples.18 We
nd that RMs tend to cluster in specic industries; Chinese RM rms are more prevalent in manufacturing
(29%), high technology (13%), consumer nondurables (13%) and health (12%) industries. Similarly, non-
Chinese RMs are more likely to be companies in high technology (23%) and health (18%) industries.
3.2. Measures of earnings management
We use abnormal accruals as the proxy for accrual-based earnings management (AEM). Kothari et al.
(2005) show that accruals of rms that have experienced unusual performance are expected to be systemat-
ically non-zero, and as a result, rm performance should be controlled when estimating abnormal accruals.
14
As of June 22, 2011, there were 83 Chinese RMs included in Bloomberg's table.
15
Our results remain quantitatively and qualitatively similar if we use the full sample or alternative control sample such as US-listed
Chinese IPO rms, US RM rms, US-listed RM rms originating from foreign countries other than China, US-listed non-RM foreign rms
and regular Canadian rms. See Section 4.4 for more details.
16
In SDC, the reverse takeover ag indicates whether the corresponding transaction is a reverse acquisition.
17
The results remain similar if we use a matched sample of regular US rms based on year, industry and rm size.
18
The industry classication is based on the Fama-French ten industries classication.
Table 1
Industry distribution of sample rms by different list types.
This table shows the distribution of the sample across ten industries according to Fama-French industry classication.
Industry Chinese RM rms Non-Chinese RM rms Regular US rms Chinese non-RM rms
Consumer nondurables 17 18 784 7

Consumer durables 9 9 359 1
Manufacturing 38 36 1768 18
Energy 5 28 636 4
High technology 17 98 3055 65
Telephones and television 2 33 619 5
Shops 13 39 1646 9
Health 16 75 1525 9
Utilities 0 7 445 2
Others 15 85 2148 30
Total 132 428 12,985 150
Our analysis conrms that the operating performance of Chinese RM rms differs signicantly from that of
both non-Chinese RM rms and matched regular US rms. We thus follow Kothari et al. (2005) and measure
AEM by the residuals from the modied Jones model with an additional control of return-on-asset (ROA).19
The model is estimated cross-sectionally each year for every two-digit SIC industry with at least 10 rm-
year observations.
ACCt 1 1=Assetst1 2 SALESt ARt 3 ROAt t 1
where ACC represents current accruals, which are dened as the increase in account receivables
(RECCH, Compustat item) plus the increase in inventory (INVCH) minus the increase in account payable
(APALCH) minus the increase in tax payable (TXACH) minus the net change in other current assets
(AOLOCH); Assetst 1 is lagged total assets; SALESt is the change in sales from year t 1 to t scaled by lagged
total assets; ARt is the change in accounts receivables from year t 1 to t scaled by lagged total assets; and
ROA is return on assets measured as income before extraordinary items scaled by lagged total assets.20
We employ two measures of real activities management following prior studies (Cohen and Zarowin,
2010; Roychowdhury, 2006). Abnormal discretionary expenses (REM1) capture the reduction of discretion-
ary expenditures, including cuts in R&D expenditures, advertising expenses, and selling, general and admin-
istrative expenses (SG&A). Abnormal production costs (REM2) represent manipulating production costs by
overproducing in order to lower cost of goods sold (COGS). The lower (higher) the amount of abnormal
discretionary expenses (abnormal production costs), the more likely the rm engages in REM. To estimate
these two measures, we run the following models for each industry and year with at least 10 rm-year
observations:
DISEXPt =Assetst1 0 1 1=Assetst1 2 SALESt1 =Assetst1 3 ROAt t 2
PRODt =Assetst1 0 1 1=Assetst1 2 SALESt =Assetst1
3 SALESt =Assetst1 4 SALESt1 =Assetst1 5 ROAt t 3
where DISEXP is the sum of R&D, advertising and SG&A expense, and PROD is the sum of COGS and change in
inventory. Other variables are dened in the same way as in Eq. (1). Consistent with AEM, we control for rm
performance in the spirit of Kothari et al. (2005).21
19
Our results are similar if we use the performance-matched approach as in Kothari et al. (2005). Those results are not presented here
but are available upon request.
20
Our results remain similar if we measure accruals using total accruals (the difference between income before extraordinary items and
operating cash ows scaled by lagged total assets) or measure abnormal accruals using alternative accrual models as in Dechow et al.
(1995), Dechow and Dichev (2002) and McNichols (2002). Detailed discussions are provided in Section 4.6.
21
The results are even more signicant if we do not control for rm performance. Those results are not presented here but are available
upon request.
3.3. Summary statistics
Table 2 reports descriptive statistics of the variables employed in the analysis for Chinese RM rms and the
control samples. Consistent with Hypothesis 1, we nd that the extent of both AEM and REM for Chinese RM
rms is signicantly higher than non-Chinese RM rms, regular US rms and Chinese non-RM rms. The
mean (median) value for performance-adjusted AEM, REM1 and REM2 for Chinese RM rms is 4.404,
17.724 and 7.715 (2.625, 10.902 and 6.579) respectively, while the values for non-Chinese RM rms
and regular US rms are much smaller in absolute values and close to zero.
We also nd a signicant difference in rm characteristics between Chinese RM rms and the three con-
trol samples. Chinese RM rms tend to outperform both non-Chinese RMs and US rms in terms of rm per-
formance (ROA) and sales growth (Growth). Our main analysis uses the performance-adjusted measures for
both AEM and REM because of the importance of controlling for rm performance in estimating earnings
management measures (Kothari et al., 2005). As in the study by Adjei et al. (2008), Chinese RM rms are
found to be younger and smaller in size and to have lower market valuation compared with non-Chinese
RM rms, regular US rms and Chinese non-RM rms. Finally, Chinese RM rms are less likely to hire Big 4
auditors than the control samples (11% compared with 66%, 74% and 85%), suggesting that Big 4 auditors
might restrain their earnings manipulation behaviors.
Table 2
Summary statistics.
This table presents summary statistics for earnings management and control variables. AEM is abnormal accruals estimated from Kothari
et al. (2005)'s performance adjusted modied Jones model, and reported as percentages of lagged total assets; REM1 is performance ad-
justed abnormal discretionary expenses, and REM2 is performance adjusted abnormal production cost estimated from the
Roychowdhury (2006)'s models, and both are reported as percentages of lagged total assets; Age is the log of the number of months
the rm rstly appears in CRSP database; BM is the book value of equity divided by market value of equity; Growth is sales growth;
Lev is total debt divided by total assets; ROA is net income before extraordinary items and reported as percentages of lagged total assets;
Financing is the sum of net debt and equity nancing cash ow and reported as percentages of lagged total assets; Sizemv is the log of
market value of equity; Sizeta is the log of total assets; Big4 is an indicator variable set to 1 if the rm hires a Big 4 auditor and 0 otherwise;
Loss is an indicator variable set to 1 if the rm's net income is negative and 0 otherwise; Suspect is an indicator variable set to 1 if net
income over lagged total asset is higher than 0 but lower than 0.005; Hasdebt is an indicator variable equal to 1 if the rm has any
long-term or short-term debt outstanding at the beginning or the end of the year; Mfg is an indicator variable equal to 1 if the rm belongs
to a manufacturing industry and 0 otherwise. Abnormal changes in ROA over one year (MatchROA1) is changes in Chinese RM rm's
ROA minus changes in ROA of the control rm matching based on size and performance over the rst year after the reverse merger,
and reported as percentages. Abnormal changes in ROA over two years (MatchROA2) is changes in Chinese RM rm's ROA minus chang-
es in ROA of the control rm matching based on size and performance over the rst two years after the reverse merger, and reported as
percentages.
Chinese reverse merger rms Non-Chinese reverse Regular US rms Chinese non-RM rms
merger rms
Mean Std dev Median Mean Std dev Median Mean Std dev Median Mean Std dev Median
AEM 4.404 13.966 2.625 0.236 10.668 0.077 0.034 9.104 0.353 1.407 11.157 0.184
REM1 17.724 40.217 10.902 0.873 43.110 0.175 1.206 40.323 0.573 7.512 36.889 4.395
REM2 7.715 34.252 6.579 1.209 35.675 0.195 0.971 32.028 0.378 6.975 27.798 4.240
Age 3.207 1.305 3.135 4.469 1.108 4.635 4.571 1.232 4.727 3.492 1.116 3.664
BM 1.101 1.368 0.722 0.522 0.980 0.439 0.546 0.969 0.467 0.978 0.954 0.700
Growth 0.450 0.869 0.295 0.377 1.050 0.095 0.246 0.762 0.083 0.552 0.955 0.306
Lev 0.176 0.195 0.100 0.273 0.256 0.235 0.270 0.267 0.222 0.157 0.191 0.072
ROA 8.254 26.822 11.157 12.997 34.768 0.557 5.743 29.456 2.661 8.409 21.732 8.248
Financing 17.398 40.308 4.049 19.031 53.137 0.817 12.769 45.939 0.000 26.945 65.949 0.659
Sizemv 3.853 1.929 4.212 4.361 2.442 4.112 4.805 2.373 4.767 6.282 1.972 6.181
Sizeta 4.152 1.482 4.418 4.447 2.420 4.034 5.023 2.276 4.935 6.142 1.960 5.780
Big4 0.113 0.317 0.000 0.663 0.473 1.000 0.744 0.436 1.000 0.850 0.358 1.000
Loss 0.294 0.456 0.000 0.512 0.500 1.000 0.377 0.485 0.000 0.218 0.413 0.000
Suspect 0.013 0.115 0.000 0.023 0.149 0.000 0.020 0.140 0.000 0.023 0.149 0.000
Hasdebt 0.826 0.379 1.000 0.879 0.326 1.000 0.864 0.342 1.000 0.698 0.459 1.000
Mfg 0.755 0.430 1.000 0.387 0.487 0.000 0.467 0.499 0.000 0.437 0.496 0.000
MatchROA1 0.497 28.045 0.283
MatchROA2 0.396 27.695 1.649
4. Empirical results
4.1. Univariate analysis of earnings management in Chinese RM rms
We rst offer preliminary insights by conducting univariate tests on earnings management behaviors in
Chinese RM rms. In particular, we compare the mean and median values of earnings management measures
for Chinese RM rms with those of non-Chinese RM rms, Chinese non-RM rms and regular US rms
matched by industry, size and sales growth. The results presented in Panel A of Table 3 are consistent with
Hypothesis 1; that is, Chinese RM rms engage in signicantly higher levels of income-increasing AEM and
REM than non-Chinese RM rms, regular US rms and Chinese non-RM rms. The difference is both statisti-
cally and economically signicant. For example, the difference in the mean value of abnormal accruals (AEM)
is 4.17% of total assets between Chinese RMs and other RM rms. In contrast, operating performance (ROA) is
only 8.25% for an average Chinese RM rm. To the extent that Chinese RM rms manage earnings at higher
levels than other companies, the nding implies that about half of operating performance in Chinese RM
rms may be attributable to aggressive earnings manipulation.
We also examine the time-series proles of mean and median earnings management measures for Chi-
nese RM rms for years 0 to + 3 relative to the year of reverse merger (year 0). Given the consistency of
the results presented in Panel B of Table 2, we interpret our results on the basis of the median value. The me-
dian abnormal accruals in the year of reverse acquisition are positive and signicant (9.03% of total assets).
The level of AEM declines continuously in the years after RM but remains statistically signicant, which is
Table 3
Earnings management for Chinese RM rms around reverse merger.
Panel A: This table compares the mean and median of earnings management variables between Chinese RM rms and the two control
samples. AEM is abnormal accruals estimated from Kothari et al. (2005)'s performance adjusted modied Jones model, and reported
as percentages of lagged total assets; REM1 is performance adjusted abnormal discretionary expenses, and REM2 is performance
adjusted abnormal production cost estimated from the Roychowdhury (2006)'s models, and both are reported as percentages
of lagged total assets. Two tailed p-values are reported in parentheses.
Chinese RM rms vs. Chinese RM rms vs. matched Chinese RM rms vs. Chinese
non-Chinese RM rms regular US rms non-RM rms
Difference Difference Difference Difference Difference Difference

in mean in median in mean in median in mean in median
Abnormal accruals (AEM) 4.168 2.702 5.138 3.615 2.997 2.810

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Abnormal Dis Exp (REM1) 16.851 10.727 25.482 14.761 10.213 6.507
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Abnormal Prod. Cost (REM2) 6.506 6.384 12.582 10.608 0.740 2.339
(0.00) (0.00) (0.00) (0.00) (0.72) (0.58)
Panel B: This table displays the mean and median values of earnings management variables for Chinese reverse merger rms for
years 0 to +3 relative to the year of reverse merger (year 0). AEM is abnormal accruals estimated from Kothari et al. (2005)'s
performance adjusted modied Jones model, and reported as percentages of lagged total assets; REM1 is performance adjusted
abnormal discretionary expenses, and REM2 is performance adjusted abnormal production cost estimated from the Roychowdhury
(2006)'s models, and both are reported as percentages of lagged total assets. Two tailed p-values are reported in parentheses.
Year 0 Year +1 Year +2 Year +3
Mean Median Mean Median Mean Median Mean Median
Abnormal accruals 12.365 9.031 7.254 4.525 6.080 2.995 3.617 2.205
(AEM) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.01) (0.03)
Abnormal Dis Exp 24.292 20.877 12.489 6.352 18.911 14.972 11.184 7.106
(REM1) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.00) (0.01)
Abnormal Prod. 10.156 10.514 19.574 17.769 7.193 9.586 4.536 1.730
Cost (REM2) (0.61) (0.24) (0.00) (0.00) (0.10) (0.08) (0.20) (0n47)
Indicates signicance at the 1% level.
consistent with accrual-based manipulation being largely constrained by the underlying accounting relation
in the balance sheet (Gunny, 2010).
The results for REM present a slightly different pattern. During the RM year, we nd evidence of signicant
negative abnormal discretionary expenses, indicating that Chinese RM rms substantially reduce R&D, adver-
tising or SG&A expenses to manage earnings upwards. However, the results also show that Chinese RM rms
do not manage earnings upwards via overproducing to achieve a lower COGS. In the years after a reverse take-
over, abnormal discretionary expenses remain negative and statistically signicant. The extent of REM via dis-
cretionary expenses in the years after a reverse acquisition is consistently below the level in the RM year, with
an increased level of earnings management occurring in year +2 relative to year +1 and year +3. However,
the level of abnormal production costs increases dramatically above the RM year level in the rst post-RM
year (year +1) but reduces signicantly in year +2 and becomes close to zero in year +3.
Collectively, the analysis of the time-series patterns of earnings management measures shows that Chinese
RM rms undertake both REM and AEM in the year of reverse acquisition. This is consistent with the view
that, given the balance sheet constraint of AEM, the amount being manipulated can fall short of the desired
threshold and managers consequently have to rely on both types of manipulation (Roychowdhury, 2006;
Zang, 2012). The results in years after RM also provide preliminary evidence on the trade-off between AEM
and REM. Real activities manipulation via discretionary expenses and production costs is used during post-RM
years when AEM is largely constrained by accounting statements.
4.2. Multivariate regression analysis of earnings management in Chinese RM rms
We then conduct multivariate analysis to examine earnings management behaviors in Chinese RM rms
by including a set of controls that are correlated with earnings management in previous studies. Specically,
we estimate the following regression:
Xn
Earnings management CHNRM i1
i Controli Year Dummies Industry Dummies
4
where earnings management is equal to AEM, REM1 or REM2. CHNRM is an indicator variable equal to one for
Chinese RMs, and zero otherwise. In addition to the CHNRM indicator variable, we also include a set of control
variables that have been shown in prior research to be important in explaining one or both types of earnings
management (Badertscher, 2011; Cohen and Zarowin, 2010; Roychowdhury, 2006). We also control for the
year and industry effects, and report rm-clustering adjusted t-statistics.
For AEM, control variables include rm performance (ROA), rm size (Sizeta), the level of external nanc-
ing (Financing), rm age (Age), sales growth (Growth), nancial leverage (Lev), Big 4 auditor indicator (Big4)
and loss indicator (Loss). We expect that rm age and size have a negative association with earnings manage-
ment measures because smaller and younger rms are more vulnerable to business risk and therefore are
more likely to manage earnings in order to avoid bankruptcy. We also expect growth and performance to
be positively associated with the level of earnings management (Kothari et al., 2005; McNichols, 2002). We
control for nancial structure because a highly leveraged rm faces high bankruptcy risk and therefore has in-
centives to manage earnings to avoid default or bankruptcy. Meanwhile, highly leveraged rms face greater
monitoring from their creditors, which will constrain earnings management. Finally, high-quality auditors
are more likely to restrict earnings management and we expect a negative coefcient.
With respect to REM, we include rm performance (ROA), rm size (Sizeta), the level of external nancing
(Financing), sales growth (Growth), nancial leverage (Lev), and indicator variables for debt (Hasdebt), sus-
picious rms with strong incentives of earnings management (Suspect), and manufacturing rms (Mfg). ROA,
rm size and sales growth are included to control for systematic variations in discretionary expenses and pro-
duction costs due to growth opportunities, size and measurement error correlated with performance. Finan-
cial structure variables (Lev, Financing and Hasdebt) are employed to control for the incentives for real
activities manipulation to avoid violating debt covenants or the pressure from capital provider on managers
to avoid reporting losses. Industry membership (Mfg) is also controlled because overproduction as an earn-
ings management strategy is only available to manufacturing rms. We expect that return on assets, nancial
structure and industry membership have a positive relation with the extent of real activities management,
whereas rm size has a negative relation.
The results presented in Table 4 support the view that Chinese RM rms engage in a higher degree of AEM.
For example, when the matched sample of regular US rms (the sample of non-Chinese RM rms) is used as
the control sample, the coefcient on CHNRM is 5.138 (3.912), with a signicant t-statistics of 8.50 (3.08). The
coefcients are attenuated but remain signicant when control variables are included. For control variables,
the coefcients on ROA, Financing, Lev and Loss are (marginally) signicant with expected signs. Overall,
the results suggest that Chinese RM rms have higher levels of abnormal accruals than non-Chinese RM
rms, regular US rms and Chinese non-RM rms.
The results for REM are presented in Table 5. For the sample including matched regular US rms, the co-
efcient on CHNRM is signicantly negative (coefcient = 25.482; t = 7.97) for abnormal discretional
expenses (REM1) and positive (coefcient = 12.582; t = 6.05) for abnormal production costs (RME2).
This indicates that, relative to regular US rms, Chinese RM rms engage in more REM and they tend to do
so by reducing discretionary expenses and reporting lower costs of goods sold through overproduction. Im-
portantly, the coefcient on CHNRM remains signicant with similar magnitude even after including rm-
level controls. The results for the sample including non-Chinese RM rms are consistent with those discussed
above. The only exception is that, when compared to Chinese non-RM rms, Chinese RM rms engage in more
REM via reducing discretionary expenses, but they do not report lower costs of goods sold through
overproduction.
Table 4
Regression results for accrual-based earnings management.
This table presents regression results of accrual-based earnings management in Chinese RM rms. CHNRM is an indicator variable set to 1
if the rm is a Chinese reverse merger rm and 0 otherwise. AEM is abnormal accruals estimated from Kothari et al. (2005)'s performance
adjusted modied Jones model, and reported as percentages of lagged total assets; Age is the log of the number of months the rm rstly
appears in CRSP database; BM is the book value of equity divided by market value of equity; Growth is sales growth; Lev is total debt di-
vided by total assets; ROA is net income before extraordinary items and reported as percentages of lagged total assets; nancing is the
sum of net debt and equity nancing cash ow and reported as percentages of lagged total assets; Sizeta is the log of total assets; Big4
is an indicator variable set to 1 if the rm hires a Big 4 auditor and 0 otherwise; loss is an indicator variable set to 1 if the rm's net income
is negative and 0 otherwise. Figures in parentheses are robust t-statistics using standards errors corrected for clustering by rm.
Chinese RM rms vs. Chinese RM rms vs. Chinese RM rms vs. Chinese
matched regular US rms non-Chinese RM rms non-RM rms
Abnormal accruals (AEM) Abnormal accruals (AEM) Abnormal accruals (AEM)
Constant 0.371 3.534 0.288 2.821 2.890 10.152

(0.75) (0.97) (0.46) (2.08) (2.54) (3.52)
CHNRM 4.939 2.761 3.912 2.936 3.062 2.896
(6.11) (2.16) (3.08) (2.29) (3.69) (2.18)
ROA 0.034 0.024 0.042
(1.25) (1.75) (1.44)
Sizeta 0.006 0.072 0.356
(0.01) (0.67) (1.22)
Financing 0.126 0.053 0.018
(8.56) (5.57) (1.29)
Age 0.287 0.734 0.974
(0.68) (2.93) (2.38)
Growth 0.097 0.260 1.016
(0.09) (1.15) (1.09)
Lev 3.854 0.966 0.754
(2.29) (0.99) (0.30)
Big4 1.473 0.409 1.678
(1.62) (0.88) (1.39)
Loss 3.050 0.847 4.465
(2.38) (1.90) (4.34)
Industry dummies No No Yes Yes Yes Yes
Year dummies No Yes Yes Yes Yes Yes
N 909 559 4654 3273 1271 877
Adj. R2 0.041 0.177 0.017 0.084 0.051 0.146
Indicates signicance at the 1% level for two tailed test.
330
Table 5
Regression results for real activities earnings management.
This table reports regression results of real activities earnings management in Chinese RM rms. REM1 is performance adjusted abnormal discretionary expenses, and REM2 is performance adjusted ab-
normal production cost estimated from the Roychowdhury (2006)'s models, and both are reported as percentages of lagged total assets; Growth is sales growth; Lev is total debt divided by total assets;
ROA is net income before extraordinary items and reported as percentages of lagged total assets; Financing is the sum of net debt and equity nancing cash ow and reported as percentages of lagged total
assets; Sizeta is the log of total assets; Suspect is an indicator variable set to 1 if net income over lagged total asset is higher than 0 but lower than 0.005; Hasdebt is an indicator variable equal to 1 if the rm
has any long-term or short-term debt outstanding at the beginning or the end of the year; Mfg is an indicator variable equal to 1 if the rm belongs to a manufacturing industry and 0 otherwise. Figures in
parentheses are robust t-statistics using standards errors corrected for clustering by rm.
Chinese RM rms vs. matched regular US rms Chinese RM rms vs. non-Chinese RM rms Chinese RM rms vs. Chinese non-RM rms
T. Zhu et al. / Pacic-Basin Finance Journal 35 (2015) 317339

Abnormal Dis Exp (REM1) Abnormal Prod Cost Abnormal Dis Exp (REM1) Abnormal Prod Abnormal Dis Exp (REM1) Abnormal Prod Cost
(REM2) Cost (REM2) (REM2)
Constant 7.696 29.753 5.028 29.657 0.913 12.138 1.165 4.735 5.030 3.639 3.729 2.438
2.12 (3.79) (3.85) (6.05) (0.27) (2.36) 0.48 (1.54) (1.41) (0.81) (1.19) (0.64)
CHNRM 25.482 24.067 12.582 14.093 19.230 18.694 5.160 4.812 12.124 12.161 0.864 0.612
(7.97) (7.04) (6.05) (6.87) (4.53) (4.57) (1.70) (1.84) (4.97) (4.86) (0.39) (0.27)
ROA 0.272 0.056 0.043 0.028 0.002 0.034
(2.53) (0.83) (1.00) (0.70) (0.02) (0.46)
Suspect 3.688 6.035 7.067 0.608 12.798 11.745
(0.43) (0.92) (2.12) (0.17) (3.27) (2.87)
Mfg 1.661 0.409 2.452 2.352 1.819 2.699
(0.31) (0.12) (0.49) (0.63) (0.55) (0.92)
Hasdebt 8.711 7.139 1.752 0.716 0.389 2.455
(1.80) (2.03) (0.52) (0.33) (0.12) (0.95)
Sizeta 9.802 3.616 2.437 0.738 4.789 1.018
(7.30) (4.75) (3.76) (1.88) (6.85) (1.55)
Financing 0.102 0.062 0.088 0.013 0.092 0.009
(1.24) (1.59) (3.41) (0.44) (3.52) (0.33)
Lev 15.61 8.13 8.402 0.749 16.782 17.511
(1.46) (1.21) (1.66) (0.18) (2.34) (2.78)
Growth 9.285 0.02 6.319 2.333 3.721 3.113
(4.06) (0.01) (4.42) (2.45) (2.21) (1.63)
Industry dummies No No No No Yes Yes Yes Yes Yes Yes Yes Yes
Year dummies No Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes
N 853 771 774 684 4178 3602 4219 3590 4178 3602 4219 3590
Adj. R2 0.076 0.194 0.031 0.081 0.025 0.079 0.009 0.02 0.025 0.079 0.009 0.02
Combining the results of Tables 4 and 5, we nd evidence consistent with Hypothesis 1 that Chinese RM
rms exhibit higher levels of income-increasing AEM and REM compared with matched regular US rms,
non-Chinese RM rms and Chinese non-RM rms. The results also indicate that, while the recent accusation
by both the public and the regulatory authorities against Chinese RM rms focuses extensively on earnings
management via accounting transactions and choices (such as accruals manipulation), REM also serves as
an important mechanism through which Chinese RM rms employ to manage reported earnings upwards.
4.3. Accrual-based vs. real activities earnings management in Chinese RM rms
To examine the trade-off between AEM and REM, we use a two-stage process based on the Heckman (1979)
method following Badertscher (2011) and Cohen and Zarowin (2010). In the rst stage, we model a rm's de-
cision to manage reported earnings as a function of earnings management incentives and rm characteristics,
regardless of which method is being used. In particular, we estimate the following cross-sectional model:
X
n
Prob EM i Determinant i 5a
i1
where Prob_EM measures whether a rm-year is classied as being engaged in earnings management or not;
it is an indicator variable set to one if either a rm's AEM or REM is above the industry-year median. The ex-
planatory variables included in Eq. (5a) are rm performance (ROA), market capitalization (Sizemv), the
book-to-market ratio (BM), nancial leverage (Lev), the level of external nancing (Financing), and suspi-
cious rms with strong incentives of earnings management (Suspect). We also include CEO duality (Duality)
and CEO ownership (CEOown) to control for managerial incentives.
In the second stage, we follow Badertscher (2011) and include the alternative earnings management
choice as an additional explanatory variable in Eq. (4) as follows:
Xn
AEM REM i
i Controli Year Dummies 5b
where REM is REM1 or REM2. We include the inverse Mills ratio estimated from the rst stage to control for
the self-selection bias. To capture the relative costliness of the two earnings management methods, we in-
clude variables that represent auditor scrutiny (Big4), regulatory scrutiny (Litigation), accounting exibility
(NOA), and nancial performance (ROA and Loss) in Eq. (5b). Given the trade-off between REM and AEM,
we expect a positive association between abnormal accruals (AEM) and abnormal discretionary expenses
(REM1), and a negative relation between abnormal accruals (AEM) and abnormal production costs (REM2).
The results are reported in Table 6. In the rst stage, we nd that market capitalization, the level of external
nancing and CEO ownership are important determinants of earnings management decisions in Chinese RM
rms, consistent with the notion that capital market incentives and managerial incentives are most important
in affecting earnings management activities (Fields et al., 2001). The results for the second stage reveal a sub-
stitution effect between AEM and REM via discretionary expenses. In particular, the coefcient on REM1 is pos-
itive and signicant, which is consistent with prior studies (Badertscher, 2011; Cohen and Zarowin, 2010; Zang,
2012) and supports Hypothesis 2 (i.e., Chinese RM rms trade off the two types of earnings management).
However, no substitution effect exists between AEM and REM through production costs (REM2). The coef-
cient on the inverse Mills Ratio is signicant, suggesting the importance of controlling for self-selection bias.22
Taken together, the results from our univariate analysis (Fig. 1 and Panel B of Table 3) and regression anal-
ysis provide corroborating evidence that Chinese RM rms use REM and AEM as substitutes. The results
22
For robustness, we examine an alternative version in predicting the likelihood of earnings management in the rst stage, where the
two equity incentive variables, namely CEO duality (Duality) and CEO ownership (CEOown), are not included. This is because the require-
ment of the availability of corporate governance information will reduce the sample size and may lead to sample selection bias. The results
reported in Exhibit 3 remain similar to those in Table 6. We also employ a Probit model rather than OLS regression in the second stage,
where the dependent variable is an indicator variable set to one if a rm's AEM (REM) is higher than the industry-year median, and zero
otherwise. The key independent variable is an indicator variable set to one if a rm's alternative manipulation mechanism is higher than
the industry-year median and zero otherwise. Alternative manipulation mechanism is REM1 or REM2. The results are reported in Exhibit
4. Consistent with the trade-off between REM and AEM, we nd a negative association between the likelihood of AEM and the likelihood
of REM via discretionary expenses (REM1), but there is no signicant relation between the likelihood of AEM and REM through produc-
tion costs (REM2).
Table 6
Regression results on the trade-off between real activities and accrual-based earnings management in Chinese reverse merger rms.
This table reports regression results on the trade-off between real activities and accrual-based earnings management in Chinese RM rms.
In the rst stage, we estimate a probit model that predicts a rm's decision to manage reported earnings as a function of earnings man-
agement incentives and rm characteristics. The dependent variable is Prob_EM, an indicator variable set to one if either a rm's AEM or
REM is above the industry-year median. ROA is net income before extraordinary items and reported as percentages of lagged total assets;
Sizemv is the log of market value of equity; BM is the book value of equity divided by market value of equity; suspect is an indicator var-
iable set to 1 if net income over lagged total asset is higher than 0 but lower than 0.005; duality is an indicator variable set to one if the CEO
is also the chairperson of the board, and zero otherwise; CEOOwn is the percentage of ownership held by the CEO of the rm; Financing is
the sum of net debt and equity nancing cash ow and reported as percentages of lagged total assets; Lev is total debt divided by total
assets. In the second stage regression, the dependent variable is AEM, abnormal accruals estimated from Kothari et al. (2005)'s perfor-
mance adjusted modied Jones model, and reported as percentages of lagged total assets. REM1 is performance adjusted abnormal dis-
cretionary expenses, and REM2 is performance adjusted abnormal production cost estimated from the Roychowdhury (2006)'s models,
and both are reported as percentages of lagged total assets; Inverse Mills ratio is estimated from the rst stage; Growth is sales growth;
Sizeta is the log of total assets; Big4 is an indicator variable set to 1 if the rm hires a Big 4 auditor and 0 otherwise; Loss is an indicator
variable set to 1 if the rm's net income is negative and 0 otherwise; Litigation is an indicator variable equal to 1 if the rm is classied in
the industry with any of the following four-digit SIC codes: 2833-2836, 8731-8734, 7371-7379, 3570-3577, 3600-3674; NOA is net oper-
ating assets, calculated as the sum of shareholders' equity less cash and marketable securities and plus total debt at the beginning of the
year, deated by total sales. Figures in parentheses are robust t-statistics using standards errors corrected for clustering by rm.
First stage Second stage
Prob_EM AEM AEM
(AEM N median) (AEM N median)
Constant 0.681 15.402 6.671

(2.14) (2.66) (1.06)
Abnormal Dis Exp (REM1) 0.034
(1.74)
Abnormal Prod Cost (REM2) 0.009
(0.37)
Inverse Mills ratio 18.501 20.667
(1.81) (1.83)
ROA 0.009 0.074 0.001
(1.42) (1.65) (0.02)
Sizemv 0.095
(1.94)
BM 0.011
(0.16)
Suspect 0.240
(0.36)
Duality 0.122
(0.43)
CEOOwn 0.965
(1.87)
Sizeta 0.901 0.426
(1.07) (0.49)
Financing 0.005 0.117 0.116
(1.25) (5.19) (4.34)
Growth 0.323 0.178
(0.32) (0.15)
Lev 0.249 6.267 10.789
(0.36) (1.24) (1.89)
Big4 0.636 0.147
(0.25) (0.06)
Litigation 2.310 1.072
(0.88) (0.42)
NOA 1.476 0.034
(1.49) (0.03)
Loss 2.621 3.494
(1.01) (1.35)
Industry dummies No Yes Yes
N 332 218 175
Pseudo R2 0.056
Adj. R2 0.256 0.238
A Accrual-based earnings management

14
Percentage of total assets

12
10
8
6
4
2
0
0 1 2 3
Years relative to the RM year, year 0
Median Mean
B Real activities earnings management

25
20
15
Percentage of total assets
10
5
0
-5
-10
-15
-20
-25
-30
0 1 2 3
Years relative to the RM year, year 0
REM1-Mean REM2-Mean REM1-Median REM2-Median
Fig. 1. Earnings management for Chinese RM rms around reverse merger. Panel A: This gure presents the time series of accrual-based
earnings management measured by abnormal accruals in and after the year of reverse merger. Abnormal accruals are estimated from
Kothari et al. (2005)'s performance adjusted modied Jones model, and reported as percentages of lagged total assets. Panel B: This gure
presents the time series of real activities earnings management measured by abnormal discretionary expenses (REM1) and abnormal pro-
duction costs (REM2). REM1 is performance adjusted abnormal discretionary, and REM2 is performance adjusted abnormal production
cost estimated from Roychowdhury (2006)'s model. Both are reported as percentages of lagged total assets.
highlight the importance of understanding the channels of earnings manipulation in Chinese RMs and show that
different manipulation strategies are employed in different years. For instance, AEM is conducted mainly in the
year of reverse acquisition. Due to limits in the amount of AEM in which a rm can engage, Chinese RM rms
also conduct income-increasing REM via discretionary expenses in the RM year. In years after RM, however,
Chinese RM rms tend to transition to REM due to the reversing nature and balance-sheet limits of accruals.
4.4. The role of auditors in constraining earnings management
To investigate the role of Big 4 auditors in restricting earnings management, we extend Eq. (4) by includ-
ing an interaction term between Big 4 and CHNRM as below:
Earnings management CHNRM Big4 CHNRMBig4

Xn 6
i1 i Controli Year Dummies Industry Dummies
Notes to Table 6:
where Big4 is an indicator variable equal to one if the rm is a client of Big 4 auditors, and zero otherwise. We
expect , the coefcient of the interaction term CHNRM* Big4, to have the sign opposite that of .
Table 7 presents the results for the sample of Chinese RM rms and non-Chinese RM rms.23 Consistent
with the role of Big 4 auditors in restricting earnings management, the estimated coefcient on the
CHNRM*Big4 interaction term is signicantly negative (coefcient = 2.950; t = 2.23) for abnormal ac-
cruals (AEM) and positive (coefcient = 13.690; t = 2.59) for abnormal discretionary expenses (REM1).
This supports the monitoring role of Big 4 auditors in constraining income-increasing earnings management
done by inating abnormal accruals or the reduction of discretionary expenses. However, we do not nd any
evidence that Big 4 auditors restrict REM through lowering production costs, which may be due to the com-
plexity of the auditing procedure in relation to the production costs. Overall, our results provide evidence that
Chinese RM rms with Big 4 auditors tend to have lower levels of AEM and REM through discretionary ex-
penses than those with non-Big 4 auditors.
4.5. Earnings management and post-RM operating performance
Finally, to examine the impact of earnings management on operating performance of Chinese RM rms in
the years after reverse takeover, we compare the difference in ROA changes of Chinese RM rms to that of the
matched US rms based on size and rm performance over the subsequent two years. We employ the follow-
ing regression model to test whether the abnormal changes in ROA can be predicted by earnings management
measures:
Xn
Abnormal changes in ROA Earnings management i1
i Controls Year Dummies
Industry Dummies 7
where the dependent variable, abnormal changes in ROA, includes one-year and two-year ahead return on as-
sets. Earnings management is equal to AEM, REM1 or REM2. Following previous studies (Gong et al., 2008;
Cohen and Zarowin, 2010; Rangan, 1998), we control for the normal amount of mean reversion in ROA by
subtracting the change in ROA of a size-matched and performance-matched regular US rm over the same
period. Specically, we select a regular US rm whose assets are within 80% to 120% of the corresponding
Chinese RM rm and has the closest ROA in the year of reverse takeover. According to H4a, we expect that ab-
normal changes in ROA are negatively associated with abnormal accruals (AEM) and abnormal production costs
(REM2) and positively related to abnormal discretionary expenses (REM1). Following previous literature
(Cohen and Zarowin, 2010; Gong et al., 2008), control variables include market capitalization (Sizemv), the
ratio of book value to market value of equity (BM), nancial leverage (Lev), rm age (Age) and sales growth
(Growth).
The results in Table 8 show that the coefcient on AEM is negative and signicant, suggesting that Chinese
RM rms with high levels of AEM during the RM year experience a decline in operating performance in the
subsequent two years. The negative impact of AEM is particularly signicant in the rst year after RM. How-
ever, inconsistent with Gunny (2010) and Cohen and Zarowin (2010), we do not nd a signicant relation be-
tween REM and the rm's future operating performance, which accordingly supports H4b that the effect of
AEM on future performance is stronger than the REM effect in the short run (i.e. one or two years).
Collectively, the results suggest that the decline in operating performance during the subsequent two
years after reverse takeover is predominantly driven by AEM. It also conrms the view that AEM tends to
be more costly than REM in the short-term (Graham et al., 2005; Roychowdhury, 2006)
4.6. Additional tests24
4.6.1. Alternative samples for tests

The above analysis utilizes two control samples: non-Chinese RM rms and a matched sample of regular
US rms. We also performance the above analysis by using the full sample, including all available US rms as
23
The results for the Chinese non-RM rms and the matched regular US rms are consistent with those presented in Table 7. They are
not tabulated here but are available upon request.
24
The results are not tabulated here due to space constraints but are available upon request.
Table 7
Regression results on the impact of Big 4 auditors on earnings management in Chinese RM rms.
This table presents regression results on the impact of Big 4 auditors on earnings management in Chinese RM rms. The sample includes
Chinese RM rms and non-Chinese RM rms. CHNRM is an indicator variable set to 1 if the rm is a Chinese reverse merger rm and 0
otherwise. AEM is abnormal accruals estimated from Kothari et al. (2005)'s performance adjusted modied Jones model, and reported as
percentages of lagged total assets; REM1 is performance adjusted abnormal discretionary expenses, and REM2 is performance adjusted
abnormal production cost estimated from the Roychowdhury (2006)'s models, and both are reported as percentages of lagged total as-
sets; Age is the log of the number of months the rm rstly appears in CRSP database; BM is the book value of equity divided by market
value of equity; Growth is sales growth; Lev is total debt divided by total assets; ROA is net income before extraordinary items and report-
ed as percentages of lagged total assets; nancing is the sum of net debt and equity nancing cash ow and reported as percentages of
lagged total assets; Sizemv is the log of market value of equity; Sizeta is the log of total assets; Big4 is an indicator variable set to 1 if
the rm hires a Big 4 auditor and 0 otherwise; loss is an indicator variable set to 1 if the rm's net income is negative and 0 otherwise;
suspect is an indicator variable set to 1 if net income over lagged total asset is higher than 0 but lower than 0.005; Hasdebt is an indicator
variable equal to 1 if the rm has any long-term or short-term debt outstanding at the beginning or the end of the year; Mfg is an indicator
variable equal to 1 if the rm belongs to a manufacturing industry and 0 otherwise. Figures in parentheses are robust t-statistics using
standards errors corrected for clustering by rm.
Dependent variable: Dependent variable: Dependent variable:

Abnormal accruals Abnormal Dis Exp Abnormal Prod Cost
(AEM) (REM1) (REM2)
Constant 0.177 2.573 5.988 12.497 1.889 5.368

(0.29) (2.07) (2.36) (3.05) (1.25) (3.13)
CHNRM 4.649 3.522 17.251 19.338 7.504 5.790
(5.21) (3.31) (5.88) (6.15) (3.35) (3.48)
Big4 0.617 0.559 7.724 5.265 2.514 2.628
(1.37) (1.13) (4.02) (2.29) (1.43) (1.49)
CHNRMBig4 2.950 2.384 13.690 17.797 0.782 0.257
(2.23) (1.76) (2.59) (3.70) (0.14) (0.05)
Firm-level controls No Yes No Yes No Yes
Industry dummies Yes Yes Yes Yes Yes Yes
Year dummies Yes Yes Yes Yes Yes Yes
N 4652 3273 4176 3601 4216 3588
Adj.R2 0.018 0.084 0.034 0.084 0.012 0.021
well as alternative control samples such as US RM rms, US-listed RM rms originating from foreign countries
other than China, US-listed non-RM foreign rms and regular Canadian rms. The results are qualitatively
similar to those reported above.
4.6.2. Controlling for corporate governance

Inferior corporate governance is one of the reasons that Chinese RM rms engage in a higher level of earn-
ings management. In the previous analyses, we do not control for corporate governance variables mainly due
to the availability of data.25 We have hand-collected the corporate governance information for the period of
20012011 from 10-Ks, and proxy statements on the SEC website. In particular, we have collected block own-
ership (holdings by large shareholders with 5% of ownership or higher), insider ownership (holdings by the
directors and ofcers), CEO ownership, and board characteristics such as board size, board independence
and CEO duality (whether the CEO is also the chairperson of the board). We then include block ownership
(Block), board independence (BoardInd) and CEO duality (Duality) as additional controls in Eq. (4).26 The
sample size reduces substantially after including the three corporate governance variables.27 The results
25
Commonly-used corporate governance databases such as ExecuComp, GMI rating and RiskMetrics do not cover most of the sample
rms (Chinese RMs, non-Chinese RMs, Chinese non-RM rms, and matched regular US rms) in our study.
26
Based on prior studies (Badertscher, 2011; Cohen and Zarowin, 2010; Zang, 2012), we expect that earnings management is positively
associated with CEO duality, and negatively related to board independence. The association between block ownership and earnings man-
agement is mixed (Hope, 2013). If large shareholders' interests are aligned with those of minority shareholders and they monitor the -
nancial reporting process, a greater block ownership lowers earnings management. On the other hand, if they are not aligned, large
shareholders' attempts to exploit minority shareholders will increase earnings management.
27
In most cases, the sample size for the analyses with controls for corporate governance is about one-third of the sample size for those
without the controls. This may lead to sample selection bias that relates to a rm's information environment and disclosure requirements.
Table 8
Regression results on the association between post-acquisition operating performance and the extent of earnings management in the RM
year.
This table reports regression results on the association between post-acquisition operating performance and the extent of earnings man-
agement in the year of reverse merger. Abnormal changes in ROA over one year (MatchROA1) is changes in Chinese RM rm's ROA
minus changes in ROA of the control rm matching based on size and performance over the rst year after the reverse merger, and re-
ported as percentages. Abnormal changes in ROA over two years (MatchROA2) is changes in Chinese RM rm's ROA minus changes
in ROA of the control rm matching based on size and performance over the rst two years after the reverse merger, and reported as
percentages. AEM is abnormal accruals estimated from Kothari et al. (2005)'s performance adjusted modied Jones model, and reported
as percentages of lagged total assets; REM1 is performance adjusted abnormal discretionary expenses, and REM2 is performance adjusted
abnormal production cost estimated from the Roychowdhury (2006)'s models, and both are reported as percentages of lagged total as-
sets; Sizemv is the log of market value of equity; BM is the book value of equity divided by market value of equity; Lev is total debt divided
by total assets; Age is the log of the number of months the rm rstly appears in CRSP database; Growth is sales growth. Figures in
parentheses are robust t-statistics using standards errors corrected for clustering by rm.
Dependent Dependent Dependent Dependent Dependent Dependent

variable: variable: variable: variable: variable: variable:
MatchROA1 MatchROA2 MatchROA1 MatchROA2 MatchROA1 MatchROA2
Abnormal accruals (AEM) Abnormal Dis Exp (REM1) Abnormal Prod Cost (REM2)
Constant 2.815 17.369 0.176 14.401 7.023 24.827

(0.23) (0.97) (0.02) (1.16) (0.57) (1.40)
Abnormal accruals 0.212 0.205
(AEM) (1.70) (1.03)
Abnormal Dis Exp 0.031 0.062
(REM1) (0.91) (1.12)
Abnormal Prod Cost 0.054 0.032
(REM2) (0.76) (0.35)
Sizemv 3.381 0.266 2.053 0.635 2.466 3.131
(1.84) (0.12) (1.62) (0.33) (1.53) (1.44)
BM 0.378 6.996 0.358 6.507 0.262 6.456
(0.19) (2.39) (0.23) (2.49) (0.12) (2.06)
Lev 18.853 26.937 15.675 18.464 8.638 13.219
(2.01) (1.92) (2.09) (1.74) (0.71) (0.91)
Age 2.204 2.239 1.560 0.814 1.355 1.774
(2.44) (1.39) (1.90) (0.76) (1.48) (1.13)
Growth 7.332 10.075 5.826 7.572 1.395 3.078
(3.28) (1.54) (2.46) (1.32) (0.53) (1.22)
Year dummies Yes Yes Yes Yes Yes Yes
N 150 134 148 131 134 118
Adj.R2 0.111 0.099 0.129 0.123 0.059 0.073
(untabulated) are consistent with those without controlling for corporate governance. The estimated coef-
cients on board independence and CEO duality are not statistically signicant, but we nd a signicantly neg-
ative association between block ownership and AEM.
4.6.3. Alternative measures of earnings management

We also conduct sensitivity tests on different measures of earnings management. For AEM, we adopt abnor-
mal accrual measures based on the modied Jones model (Dechow et al., 1995), the Dechow and Dichev (2002)
model that considers the mapping between cash ows and accruals, and the McNichols (2002) modication of
the Dechow and Dichev (2002) model.28 The results for alternative AEM measures remain unchanged.
28
Allen et al. (2013) argue that accruals can be decomposed into (1) accruals that correctly anticipate future benets (i.e., good ac-
cruals) and (2) accrual estimation error (i.e., bad accruals). They suggest that tests of earnings management should rule out the pos-
sibility that the reversals are attributable to good accruals. However, the ndings in Allen et al. (2013) are highly contradictory to
Dechow et al. (2012), which develop a reversal framework to detect AEM. In addition, the decomposition model in Allen et al. (2013)
is criticized by Wysocki (2009) and Dechow et al. (2012). We use the McNichols (2002) model to control for good accruals (i.e., rm
growth and working capital uctuations) and nd that the results remain similar even after controlling for accrual reversal.
With respect to REM, we also follow Roychowdhury (2006) and construct abnormal operating cash ows
(CFO) that capture sales manipulation by accelerating the timing of sales through increased price discounts or
more lenient credit terms. Moreover, similar to Cohen and Zarowin (2010) and Zang (2012), we construct two
combined measures to capture the total effects of real activities management. One is to multiply abnormal dis-
cretionary expenses (REM1) by negative one and add it to abnormal production costs (REM2). The second
measure is to take the sum of abnormal operating cash ows and abnormal discretionary expenses (REM1)
and then multiply it by negative one. The larger the two combined measures, the larger the extent the rm
manipulates its real activities to boost earnings. We nd similar results when using the two combined
measures of REM. However, we do not nd any evidence that Chinese RM rms inate earnings upwards
by manipulating real activities via abnormal operating cash ows.
5. Conclusion
This study examines the channels through which Chinese RMs manage earnings upwards and the perfor-
mance consequences of earnings management. We document that Chinese RM rms use both AEM and REM
to inate earnings both in the year of reverse acquisition and afterward. We also nd a trade-off between REM
and AEM. Chinese RM rms tend to resort to REM in the years after RM due to the costs and constraints of
AEM. Chinese RM rms with non-Big 4 auditors tend to have higher levels of earnings management. Finally,
we show that income-increasing AEM has a negative impact on subsequent operating performance in Chinese
RMs, but there is no impact from REM.
Our study provides practical guidance and implications to a wide variety of interested groups, such as in-
vestors, regulators and external auditors. For investors it potentially enables them to identify bad apples and
to avoid experiencing signicant investment loss. For regulators, our results suggest that RMs should be
scrutinized more extensively at the back door to ensure transparency and credibility of nancial informa-
tion. By examining how Chinese RMs manipulate earnings upwards, our results also give regulators a lead
on where to look in conducting their oversight function.
For an external auditor who certies a rm's nancial statements, the results suggest that auditors should
consider additional procedures (e.g., more risk assessment procedures and substantive testing) in assessing
and auditing nancial statements of RMs, which could reduce their risk of costly shareholder litigation and
reputation damage. Recently, the SEC issued the initial decision sanctioning the Chinese afliates of the Big 4
auditors by suspending them from issuing the audit reports required from their China-based SEC registrant
audit clients for at least six months. This suggests that the audit quality is believed to be an important determi-
nant of earnings management behaviors of RMs. It is not clear whether auditors are aware of the associated
risks in RMs and how they can control the risk in the auditing process. We leave these questions for future
research.
Appendix. Variable denitions
Variable Denition
CHNRM An indicator variable set to 1 if the rm is a Chinese reverse merger rm and zero otherwise
AEM Abnormal current accruals estimated from Kothari et al. (2005)'s performance adjusted modied Jones
model, and reported as percentages of lagged total assets. Current accruals is dened as the increase in
account receivables (RECCH) plus the increase in inventory (INVCH) minus the increase in account payable
(APALCH) minus the increase in tax payable (TXACH) minus the net change in other current assets
(AOLOCH).
REM1 Performance adjusted abnormal discretionary expense estimated from the Roychowdhury (2006)'s model,
and reported as percentages of lagged total assets
REM2 Performance adjusted abnormal production cost estimated from the Roychowdhury (2006)'s model, and
reported as percentages of lagged total assets
MatchROA1 Changes in Chinese RM rm's ROA minus changes in ROA of the control rm matching based on size and
performance over the rst year after the reverse merger, and reported as percentages
(continued on next page)
(continued)
Appendix (continued)
Variable Denition
MatchROA2 Changes in Chinese RM rm's ROA minus changes in ROA of the control rm matching based on size and
performance over the rst two years after the reverse merger, and reported as percentages
Age The log of the number of months the rm rstly appears in the CRSP database
BM The book value of equity (CEQ) divided by market value of equity (CSHO * PRCC_F)
Growth Sales (SALE) growth
Lev Total debt (DLTT + DLC) divided by total assets (AT)
ROA Net income before extraordinary items (IB) divided by lagged total assets (AT), and reported as percentages
Financing The sum of net debt (DLTIS DLTR DLCCH) and equity nancing (SSTK PRSTKC DV) cash ow divided
by lagged total assets (AT), and reported as percentages
Sizemv The log of market value of equity (CSHO * PRCC_F)
Sizeta The log of total assets (AT)
Big4 An indicator variable set to 1 if the rm hires a Big 4 auditor and zero otherwise
Loss An indicator variable set to 1 if the rm's net income is negative and zero otherwise
Suspect An indicator variable set to 1 if net income/lagged total asset is higher than zero but lower than 0.005, and
zero otherwise
Hasdebt An indicator variable set equal to 1 if there is long-term or short-term debt outstanding at the beginning of
the year or at the end of the year, and zero otherwise
Mfg An indicator variable set equal to 1 if the rm belongs to a manufacturing industry, and zero otherwise
References
Adjei, F., Cyree, K., Walker, M., 2008. The determinants and survival of reverse mergers vs IPOs. J. Econ. Financ. 32, 176194.
Allen, E.J., Larson, C.R., Sloan, R.G., 2013. Accrual reversals, earnings and stock returns. J. Account. Econ. 56 (1), 113129.
Ang, J.S., Jiang, Z., Wu, C., 2012. Good apples, bad apples: sorting among Chinese companies traded in the U.S. Unpublished Working
Paper. Florida State University.
Armstrong, C.S., Foster, G., Taylor, D.J., 2012. Financial reporting quality in newly public companies: Are accruals opportunistic? Unpub-
lished Working Paper. The Wharton School, University of Pennsylvania.
Badertscher, B., 2011. Overvaluation and the choice of alternative earnings management mechanisms. Account. Rev. 86, 14911518.
Baker, R.R., Biddle, G.C., O'Connor, N., 2012. SOX internal control deciencies and auditors of U.S. -listed Chinese versus U.S. rms. Unpub-
lished working paper, The University of Hong Kong.
Ball, R., Shivakumar, L., 2008. Earnings quality at initial public offerings. J. Account. Econ. 45, 324349.
Ball, R., Robin, A., Wu, J.S., 2003. Incentives versus standards: properties of accounting income in four East Asian countries. J. Account.
Econ. 36, 235270.
Becker, C.L., DeFond, M.L., Jiambalvo, J., Subramanyam, K.R., 1998. The effect of audit quality on earnings management. Contemp.
Account. Res. 15, 124.
Cecchini, M., Jackson, S.B., Liu, X., 2012. Do initial public offering rms manage accruals? Evidence from individual accounts. Rev. Acc.
Stud. 17, 2240.
Chen, C.W., Yuan, H.Q., 2004. Earnings management and capital resource allocation: evidence from China's accounting-based regulation
of rights issues. Account. Rev. 79, 645665.
Chen, K.C., Cheng, Q., Lin, Y.C., Lin, Y.C., Xiao, X., 2013. Financial reporting quality of Chinese reverse merger rms: the reverse merger
effect or the China effect? Working Paper. Singapore Management University
Chi, J., Gupta, M., 2009. Overvaluation and earnings management. J. Bank. Financ. 33, 16521663.
Chi, W., Lisic, L.L., Pevzner, M., 2011. Is enhanced audit quality associated with greater real earnings management? Account. Horiz. 25,
315335.
Chu, C., Gotti, G., Schumann, K., 2014. Reverse mergers and earnings quality. Working Paper. University of Texas at El Paso.
Cohen, D., Zarowin, P., 2010. Accrual-based and real earnings management activities around seasoned equity offerings. J. Account. Econ.
50, 219.
Cohen, D., Pandit, S., Wasley, C., and Zach, T., 2013, Measuring real activity management, Unpublished working paper, University of
Rochester.
Cornett, M.M., Marcus, A.J., Tehranian, H., 2008. Corporate governance and pay-for-performance: The impact of earnings management.
J. Financ. Econ. 87, 357373.
Darrough, M., Huang, R., Zhao, S., 2012. The spillover effect of Chinese reverse merger frauds: Chinese or reverse merger? Unpublished
working paper. CUNY Baruch College.
DeAngelo, L.E., 1981. Auditor size and audit quality. J. Account. Econ. 3, 183199.
Dechow, P., Dichev, I., 2002. The quality of accruals and earnings: The role of accrual estimation errors. Account. Rev. 77 (Supplement),
3559.
Dechow, P., Sloan, R., Sweeney, A., 1995. Detecting earnings management. Account. Rev. 70, 193225.
Dechow, P.M., Hutton, A.P., Kim, J.H., Sloan, R.G., 2012. Detecting earnings management: a new approach. J. Account. Res. 50 (2),
275334.
Dye, R., 1993. Auditing standards, legal liability and auditor wealth. J. Polit. Econ. 101, 887914.
Fan, Q., 2007. Earnings management and ownership retention for initial public offering rms: theory and evidence. Account. Rev. 82,
2764.
Feldman, D., 2009. Reverse mergers: And other alternatives to traditional IPOs. John Wiley & Sons.
Fields, T.D., Lys, T.Z., Vincent, L., 2001. Empirical research on accounting choice. J. Account. Econ. 31 (1), 255307.
Francis, J., Maydew, E., Sparks, H., 1999. The role of Big 6 auditors in the credible reporting of accruals. Audit. J. Pract. Theory 18, 1734.
Givoly, D., Hayn, C., Lourie, B., 2014. Importing accounting quality: the case of foreign reverse mergers. Working Paper. Pennsylvania
State University.
Gleason, K.C., Rosenthal, L., Wiggins Iii, R.A., 2005. Backing into being public: an exploratory analysis of reverse takeovers. J. Bank. Financ.
12, 5479.
Gong, G., Louis, H., Sun, A., 2008. Earnings management and rm performance following open-market repurchases. J. Financ. 227,
947986.
Graham, J.R., Harvey, C.R., Rajgopal, S., 2005. The economic implications of corporate nancial reporting. J. Account. Econ. 40, 337.
Gunny, K.A., 2010. The relation between earnings management using real activities manipulation and future performance: evidence
from meeting earnings benchmarks. Contemp. Account. Res. 27, 855888.
Hazarika, S., Karpoff, J.M., Nahata, R., 2012. Internal corporate governance, CEO turnover, and earnings management. J. Financ. Econ. 104,
4469.
He, X., Wong, T., Young, D., 2012. US listing of Chinese rms: bonding vs. adverse selection. Unpublished working paper. The Chinese
University of Hong Kong.
Heckman, J.J., 1979. Sample selection bias as a specication error. Econometrica 47, 153161.
Hope, O.K., 2013. Large shareholders and accounting research. China J. Account. Res. 6 (1), 320.
Hribar, P., Nichols, C.D., 2007. The use of unsigned earnings quality measures in tests of earnings management. J. Account. Res. 45 (5),
10171053.
Hung, M., Wong, T., Zhang, T., 2012. Political considerations in the decision of Chinese SOEs to list in Hong Kong. J. Account. Econ. 53,
435449.
Jensen, M., 2005. Agency costs of overvalued equity. Financ. Manag. 34, 519.
Jindra, J., Voetmann, T., Walking, R., 2012. Reverse merger: the Chinese experience. Working Paper. Ohio State University.
Kothari, S., Leone, A., Wasley, C., 2005. Performance matched discretionary accrual measures. J. Account. Econ. 39, 163197.
Kothari, S.P., Mizik, N., Roychowdhury, S., 2012. Managing for the moment: the role of real activity versus accruals earnings management
in SEO valuation. Working Paper. Massachusetts Institute of Technology.
Laux, C., Laux, V., 2009. Board committees, CEO compensation, and earnings management. Account. Rev. 84, 869891.
Lee, G., Masulis, R.W., 2011. Do more reputable nancial institutions reduce earnings management by IPO issuers? J. Corp. Financ. 17,
9821000.
Lee, C., Li, K., Zhang, R., 2013. Shell games: are Chinese reverse merger rms inherently toxic? Unpublished working paper. Stanford
University.
Leuz, C., Nanda, D., Wysocki, P.D., 2003. Earnings management and investor protection: an international comparison. J. Financ. Econ. 69,
505527.
Louis, H., 2004. Earnings management and the market performance of acquiring rms. J. Financ. Econ. 74, 121148.
McNichols, M., 2002. Discussion of the quality of accruals and earnings: the role of accrual estimation errors. Account. Rev. 77 (Supple-
ment), 6169.
Morseld, S.G., Tan, C., 2006. Do venture capitalists inuence the decision to manage earnings in initial public offerings? Account. Rev. 81,
11191150.
Piotroski, J.D., and Zhang, T.Y., 2012. Politicians and the IPO mechanism: The impact of impending political promotions on IPO activity in
China. Unpublished working paper, Stanford University.
Rangan, S., 1998. Earnings management and the performance of seasoned equity offerings. J. Financ. Econ. 50, 101122.
Roychowdhury, S., 2006. Earnings management through real activities manipulation. J. Account. Econ. 42, 335370.
Teoh, S., Wong, T., 2002. Why new issues and high-accrual rms underperform: the role of analysts' credulity. Rev. Financ. Stud. 15,
869900.
Teoh, S., Welch, I., Wong, T., 1998a. Earnings management and the long-run underperformance of initial public offerings. J. Financ. 53,
19351974.
Teoh, S., Wong, T., Rao, G., 1998b. Are accruals during initial public offerings opportunistic? Rev. Acc. Stud. 3, 175208.
Wongsunwai, W., 2013. The effect of external monitoring on accrualbased and real earnings management: evidence from venture
backed initial public offerings. Contemp. Account. Res. 30 (1), 296324.
Wysocki, P., 2009. Assessing earnings and accruals quality: US and international evidence. Working Paper. Massachusetts Institute of
Technology.
Yang, Z.F., 2012. Do political connections add value to audit rms? Evidence from IPO audits in China. Contemp. Account. Res. 30,
891921.
Zang, A.Y., 2012. Evidence on the trade-off between real activities manipulation and accrual-based earnings management. Account. Rev.
87, 675703.

Accrual-Based and Real Activity Earnings Management at The Back Door Evidence From Chinese Reverse Mergers

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Accrual-Based and Real Activity Earnings Management at The Back Door Evidence From Chinese Reverse Mergers

Загружено:

Авторское право:

Доступные форматы

Pacic-Basin Finance Journal 35 (2015) 317339

Contents lists available at ScienceDirect

Pacic-Basin Finance Journal

Accrual-based and real activity earnings

2. Literature review and hypothesis development

2.1. Background and related literature on Chinese RM rms

2.2. Hypothesis development

2.2.1. Earnings management of Chinese RM rms

in income-increasing earnings management to sustain overvaluation or reduce nancing costs, compared to

2.2.2. The trade-off between AEM and REM

2.2.3. Big 4 auditors and earnings management

2.2.4. Earnings management and future operating performance

3. Data and methodology

3.1. The sample

3.2. Measures of earnings management

Consumer nondurables 17 18 784 7

ACCt 1 1=Assetst1 2 SALESt ARt 3 ROAt t 1

DISEXPt =Assetst1 0 1 1=Assetst1 2 SALESt1 =Assetst1 3 ROAt t 2

PRODt =Assetst1 0 1 1=Assetst1 2 SALESt =Assetst1

3 SALESt =Assetst1 4 SALESt1 =Assetst1 5 ROAt t 3

3.3. Summary statistics

4.1. Univariate analysis of earnings management in Chinese RM rms

Difference Difference Difference Difference Difference Difference

Abnormal accruals (AEM) 4.168 2.702 5.138 3.615 2.997 2.810

Year 0 Year +1 Year +2 Year +3

Mean Median Mean Median Mean Median Mean Median

4.2. Multivariate regression analysis of earnings management in Chinese RM rms

Abnormal accruals (AEM) Abnormal accruals (AEM) Abnormal accruals (AEM)

Constant 0.371 3.534 0.288 2.821 2.890 10.152

T. Zhu et al. / Pacic-Basin Finance Journal 35 (2015) 317339

4.3. Accrual-based vs. real activities earnings management in Chinese RM rms

First stage Second stage

Prob_EM AEM AEM

(AEM N median) (AEM N median)

Constant 0.681 15.402 6.671

A Accrual-based earnings management

Percentage of total assets

B Real activities earnings management

4.4. The role of auditors in constraining earnings management

Earnings management CHNRM Big4 CHNRMBig4

4.5. Earnings management and post-RM operating performance

4.6. Additional tests24

4.6.1. Alternative samples for tests

Dependent variable: Dependent variable: Dependent variable:

Constant 0.177 2.573 5.988 12.497 1.889 5.368

4.6.2. Controlling for corporate governance

Dependent Dependent Dependent Dependent Dependent Dependent

MatchROA1 MatchROA2 MatchROA1 MatchROA2 MatchROA1 MatchROA2

Constant 2.815 17.369 0.176 14.401 7.023 24.827

4.6.3. Alternative measures of earnings management

Appendix. Variable denitions

Вам также может понравиться