Chen - 2013 - Underwriter Reputation Issuer Ownership and Pre-Ipo Earnings Management Evidence From China - FM

Underwriter Reputation, Issuer
Ownership, and Pre-IPO Earnings

Management: Evidence from China
Chao Chen, Haina Shi, and Haoping Xu
This paper investigates the correlation between pre-initial public offering (pre-IPO) earnings
management and underwriter reputation for issuers with different ownership structures in China.
We document a significantly inverse relationship between underwriter reputation and pre-IPO
earnings management for non-state-owned enterprises (NSOE) issuers only, while no significant
association is found for state-owned enterprises (SOE) issuers. We also find that for the NSOE new
issue market, underwriter reputation is positively correlated with issuer post-IPO performance
indicating that prestigious underwriters can incrementally improve issuer post-IPO performance.
A growing number of studies have documented that financial intermediaries improve informa-
tion quality, which, in turn, optimizes resource allocation. In particular, one important type of
financial intermediary, underwriters signal and certify issuer earnings quality during initial public
offerings (IPOs). Reputable underwriters mitigate information asymmetry between issuers and
public investors, reduce issuers costs of raising capital (Carter and Manaster, 1990; Dai, Jo, and
Schatzberg, 2010), and in general, enhance long-term returns to issuers and market efficiency
(Carter, Dark, and Singh, 1998; Bharat and Kini, 1999; Jo, Kim, and Park, 2007; Dong, Michel,
and Pandes, 2011).
However, the existing evidence is derived from developed markets and may not be equally
applicable to transitional markets, such as China, for the following reasons. First, one of a
transitional markets most important characteristics is government intervention in the capital
market, which may affect the role of financial intermediaries. In China, the government plays dual
roles in the capital market. It not only sets regulatory policies, but also participates in economic
activities. For example, from 2002 to 2008, nearly 49% of issuers were state-owned enterprises
(SOEs). When compared with developed markets, this feature of government ownership affects
underwriters incentives to supervise their clients, as well as their behavior in obtaining clients.
Additionally, transitional markets are often criticized for their weak enforcement of investor
protections (La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 1998). Without proper incentives
We are grateful to an anonymous referee, Charles Chen, Shimin Chen, Bill Christie (Editor), Dionysia Dionysiou, Yuan
Ding, Yuan-Cheng Hu, Jeong-Bon Kim, Jay Ritter, Shuang Xue, and Hongqi Yuan for their helpful comments and insights.
We would also like to thank the participants of the 2010 American Accounting Association Annual Meeting, 2010 Financial
Management Association Annual Meeting, 2011 MFS Annual Conference, 2010 National Taiwan University IEFA Annual
Conference, and the research workshops at Peking University, Tsinghua University, Jiangxi University of Finance and
Economics, and Fudan University. Special thanks go to an anonymous reviewer. The authors are grateful for the financial
support from the National Nature Science Foundation of China (Nos. 71072004, 71202056, and 71272073), and the
Financial Research Center of Fudan University. All remaining errors are our own.
Chao Chen is a Professor in the School of Management at Fudan University in Shanghai, China. Haina Shi is an
Assistant Professor in the School of Management at Fudan University in Shanghai, China. Haoping Xu is an Associate
Professor in the School of Management at Fudan University in Shanghai, China.
Financial Management Fall 2013 pages 647 - 677
648 Financial Management r Fall 2013
and/or strong enforcement mechanisms, underwriters may not act in the best interest of investors.
Consequently, it is important to note that the underwriters role in improving market efficiency
is constrained by the institutional background.
As one of the fastest growing markets in the world, China was ranked the second largest
economy in terms of gross domestic product (Bloomberg, 2010) in 2010 and its IPO market had
become the most active in terms of the number of issuers and total issuing amount. For instance,
in 2010, a total of 345 issuers were successfully listed on Chinas Shanghai and Shenzhen stock
markets, with a total issuing amount of about $81 billion USD (the China Securities Regulatory
Commission (CSRC), 2010). Therefore, it is important for researchers to provide more insight
into this rapidly growing IPO market.
China keeps its IPO process under strict regulatory control. Appendix A illustrates how a po-
tential issuer proceeds through an IPO and becomes publicly listed. As demonstrated in Appendix
A, the CSRC, a government-controlled regulatory institution, has the right of final approval in
the qualification of new issues. However, reforms (see Appendix B for a detailed discussion
on these reforms) have been undertaken since 2001 aimed at strengthening the role of market
agents, especially underwriters, in enhancing the efficiency of the IPO process. Underwriters now
participate in the IPO process and lead a consultancy team that includes auditors and lawyers.
They advise issuers on corporate governance, financial systems, and regulatory compliance, as
well as corporate restructuring. The CSRC has also specified that the lead underwriter of an
underwriting project should be responsible for truthfulness of disclosure. If a false statement
is found, a penalty of up to five years in prison is imposed upon the signing project leader.
In addition, the underwriter is responsible to the issuer for up to two years after the IPO if
post-IPO financial performance deteriorates significantly. Moreover, since 2004, the CSRC has
adopted a sponsor system (i.e., Step 8 in Appendix A). Investment banks employ a sponsor
whose career is directly related to the issuers disclosure quality and post-IPO financial per-
formance. Given the critical role of underwriters in the IPO process in China, it is important
to investigate the effects of the underwriter role on pre-IPO accounting quality and post-IPO
peformance.
Additionally although the equity financing market in China functioned primarily for SOEs prior
to 2000, an increasing number of non-SOEs (NSOEs) have recently become important players in
the capital market, resulting in a market that consists of firms with different ownership structures.
Public investors can view the government as an implicit guarantee of an SOE issuers underlying
value for the following reasons. First, when compared with NSOEs, SOEs enjoy organizational
legitimacy and favorable access to valuable information about the policy process. SOEs are often
provided with preferential financial treatment and less policy discrimination (Huang, 2003). With
the governments support, SOEs usually enjoy favorable access to bank loans (Brandt and Li,
2003), lower costs of capital (Borisova and Megginson, 2011), monopoly advantages (Li, 2009),
and higher growth performance relative to their industry counterparts (Lin and Germain, 2003).
Moreover, in the case of financial distress, SOEs are more likely to receive government bailouts.
For example, during the 2008 financial crisis, PetroChina, a huge oil producer and distributor
controlled by the central government of China, was granted a government subsidy of RMB 16.9
billion, 12.34% of its net income that year. Thus, when compared with NSOE issuers, SOE issuers
are less likely to be adversely viewed by investors. In this case, underwriters of different issuers
are associated with different incentives in the underwriting task. Therefore, despite common
regulations, underwriters roles may be different for firms with different ownership backgrounds.
Chen and Xu (2011) examine the determinants of underwriters market shares in China and
find that underwriters with different characteristics tend to have different target clients and
behave differently in SOE and NSOE markets. In short, the unique feature of the Chinese capital
Chen, Shi, & Xu r Underwriter Reputation, Issuer Ownership, and Pre-IPO Earnings Management 649
market, consisting of firms with different ownership backgrounds under the same regulations and
identical macroeconomic conditions, allows us to investigate the different impact of underwriters
on accounting information quality and, in turn, post-IPO performance within different institutional
contexts.
Specifically, this paper investigates the role of underwriters in restricting earnings management
during the IPO process for issuers of different ownership backgrounds. We focus on earnings
management as IPO pricing depends heavily upon pre-IPO accounting earnings (Dechow, Kothari,
and Watts, 1998; Liu, Nissim, and Thomas, 2002). Cooked earnings in China could easily
misallocate resources since IPO pricing was determined by a regulated price/earnings (PE) ratio
before year 2009.1 Previous literature (Aharony, Lee, and Wong, 2000, 2010; Kao, Wu, and Yang,
2009) finds evidence of earnings manipulation by issuers in China who used accruals to manage
earnings prior to IPOs to obtain favorable IPO pricing. Based on a sample of 503 IPO issuers
in China from 2002 to 2008, we find that underwriter reputations have significantly negative
correlations with IPO earnings management if the issuer is an NSOE. However, no significant
relationship is documented if the issuer is government affiliated (i.e., an SOE). Further tests also
find that a prestigious underwriter can improve post-IPO performance, but only in the NSOE
market.
Our study contributes to the existing literature along the following dimensions. First, the major-
ity of existing research focuses on underwriter reputation regardless of institutional background.
These studies consider the market as a whole, with the assumption that interaction between issuers
and underwriters are cross-sectionally homogeneous. Our study addresses this issue by studying
a single market while analyzing differences in market segments based on ownership background.
We find that underwriter reputation works well in alleviating issuers pre-IPO earnings manage-
ment and improves postissue performance in NSOE market segments only, rather than in the
whole market. Additionally, our evidence supplements the literature regarding mechanisms for
improving information quality in the transitional market. We study the correlation between un-
derwriter reputation, pre-IPO earnings management, and post-IPO returns in China. Our results
indicate that the underwriters role in reducing earnings management and improving postissue
returns is conditional upon the issuers ownership background.
The rest of this paper is organized as follows. Section I reviews the related literature and develops
the hypothesis. Section II presents our data sources and measurements of key variables and
descriptive statistics. Our research design and main regression results are discussed in Section III,
while our robustness checks are conducted in Section IV. The role of the underwriters reputation
on post-IPO performance is explored in Section V. Section VI provides our conclusions.
I. Related Literature and Hypothesis
A. Earnings Management and Underwriter Reputation during IPOs

Public information about the issuer is limited during an IPO, which constrains the investors
abilities to evaluate the issuers prior performance. Therefore, the accounting information in the
prospectus is of great significance. Underwriters set IPO offer prices based on the issuers current
earnings and industrial PE ratios. Alternatively, investors rely on reported earnings to make their
1
The CSRC set an upper bound for the issuing PE ratio during our sample period. For example, the bound was set at
30 from 2004 to 2008. In practice, over 50% of issuers set their IPO pricing exactly at the upper bound. To differentiate
from real market-oriented IPO pricing, we call this the regulated PE ratio. In May 2009, the CSRC removed the regulated
upper bound of the PE ratio and required that IPO pricing be based on book building.
investment decisions. Therefore, issuers should bear considerable discretion in releasing good
news to provide a vision of firms future (Ho, Huang, Lin, and Lin, 2010) and manipulating
earnings prior to IPOs. However, the accounting accrual may be reversed later leading to poor
post-IPO performance. Teoh, Welch, and Wong (1998a) and Teoh, Wong, and Rao (1998b) find
evidence supporting this conjecture. Specifically, they document that issuers use positive accru-
als to report favorable pre-IPO earnings and that issuers with aggressive earnings management
experience decline in long-term stock returns and earnings performance. DuCharme, Malatesta,
and Sefeik (2004) report similar results, where some firms opportunistically manipulate earnings
upward prior to stock issues. Darrough and Rangan (2005) note a positive correlation between
discretionary accruals in IPO years and managerial selling, providing evidence of managers op-
portunistic motivation in earnings manipulation during the IPO process. Several studies (Aharony,
Lee, and Wong, 2000, 2010; Kao, Wu, and Zhang, 2009) also document overall evidence of IPO
earnings management by issuers in China.
To alleviate information asymmetry in the IPO process, third party certifiers, such as auditors,
underwriters, and attorneys, are required to assist in the IPO process. Michaely and Shaw (1994)
determine that IPOs underwritten by reputable investment banks are less likely to be underpriced
and tend to perform better in the long run. Brau and Fawcetts (2006) survey of chief financial
officers indicates that hiring a top investment bank to underwrite an IPO is a positive signal. Kim,
Palia, and Saunders (2010) find evidence that underwriters are able to differentiate the quality
of issuers and reflect the information in the underwriting spread and initial returns. Titman and
Trueman (1986) provide theoretical support for the relationship between underwriter quality and
IPO pricing. In particular, they find that an issuer with favorable information about its firm value
is more likely to choose a high-quality underwriter than an issuer with less favorable information.
In other words, issuers that are confident about their accounting information are more likely to
hire reputable underwriters, providing a signal to the market. As a result, underwriter reputation
is negatively associated with earnings management. Following Brau and Johnson (2009), we call
this the signaling hypothesis.
Other researchers have an alternative explanation for the negative correlation between under-
writer reputation and earnings management; namely, that reputable underwriters mitigate issuers
earnings management behavior to protect their reputation. Prior studies (Dunbar, 2000; Karpoff,
Lee, and Martin, 2008) find that prestigious underwriters keep their valuable reputations through
maintaining high underwriting quality. During an IPO, aside from pricing the new issue, an un-
derwriter provides such non-pricing services as analyst coverage and industry expertise (Liu and
Ritter, 2011) including the important function of certifying the issuers earnings quality (Lee and
Masulis, 2011). Underwriters are, therefore, motivated to mitigate issuer earnings management to
protect their reputation. Following Brau and Johnson (2009), we call this the mitigating hypoth-
esis. Jo, Kim, and Park (2007) study a related issue in the seasoned equity offering market and
find an inverse association between underwriter quality and issuer earnings management. They
justify the negative correlation with reputable underwriters restricting firm incentives to manage
earnings to protect their reputation and avoid litigation risk. Chang, Shi, Tam, and Wang (2010)
believe that reputable underwriters mitigate information asymmetry in IPO pricing. Fang (2005)
studies the bond offering market and documents that banks underwriting decisions in corporate
bond offerings reflect their reputational concerns.
Consequently, both hypotheses, the signaling and the mitigating hypothesis, predict a nega-
tive association between underwriter reputation and issuer pre-IPO earnings management. The
explanation can be summarized as follows: 1) firms with high earnings quality are likely to
hire reputable underwriters for signaling purposes, 2) reputable underwriters are motivated to
reduce pre-IPO earnings management to protect their reputations, and 3) these two have a joint
effect as reputable underwriters and issuers with high earnings quality tend to choose each
other.2
B. Ownership Effect
Based on the above argument, we expect a negative relationship between underwriter reputation
and pre-IPO earnings management. However, as discussed in the introductory section, evidence
documented in developed markets where private agents dominate may not be equally applicable to
transitional markets, such as China, due to the issuers different ownership backgrounds. Previous
literature has studied various differences in SOEs and NSOEs during the IPO process. Kornai
(1993) and Qian (1994) find that government leaders have an incentive to help SOEs in IPOs
as successful SOE IPOs bring more resources into the government adding to the government
leaders political capital and increasing their chances for promotion (Li and Zhou, 2005). Given
the difference between SOE and NSOE issuers, our study focuses on the effect of ownership
background on issuers incentives to signal earnings quality and underwriters incentives to
mitigate issuers earnings management.
Demand for accounting information during the IPO process is derived from two main groups
in China: 1) the investors and 2) the CSRC. Before a public offering, the investors have very little
information about the issuer. Therefore, the prospectus is the most important information source
available to public investors. Armed with only that information, investors usually adversely select
low quality issuers. To reduce adverse selection, issuers tend to provide high-quality information
in their prospectus and hire reputable underwriters to signal their earnings quality. However,
we note that the level of adverse selection is different for SOEs and NSOEs for the following
reasons. First, given their relationship with the government, SOEs are more likely to receive its
help or support than NSOEs. Government support includes not only governmental subsidies under
existing regulations, but also monopoly advantages through shareholder influence on the proposed
regulations. Furthermore, in case of financial distress, government-owned firms are more likely
to avoid liquidation or bankruptcy, thanks to government bailouts, than firms with no government
ownership. Therefore, we argue that SOEs face less risk of public investor adverse selection
than NSOEs. Consequently, when compared with NSOEs, SOE issuers are associated with lower
incentives to hire reputable underwriters to signal their earnings quality. Moreover, SOE and
NSOE issuers also face different levels of adverse selection by the CSRC. Like the US Securities
and Exchange Commission, the CSRC imposes a set of minimum disclosure requirements and
investigates issuers to determine whether they have met these before allowing them to go public.
In addition to regulatory compliance, one of the most important aspects of the investigation is
financial reporting quality. Thus, issuers are likely to be adversely selected by the CSRC for poor
reporting quality. However, due to their close relationship with the government, SOE issuers are
associated with a lower risk of adverse selection by the CSRC.3 In addition, one of the CSRCs
missions is to assist government-supported firms in receiving access to sufficient funds for further
development. Given the CSRCs critical role in determining the outcome of an IPO (see Appendix
A, Steps 6, 10, and 12), government-owned issuers are more likely to successfully initiate IPOs and
2
Brau and Johnson (2009) attempt to differentiate the two effects using a simultaneous model. They find that the negative
relation between underwriter reputation and pre-IPO earnings management is mainly driven by the signaling effect in the
US market. Section IV.B follows their approach.
3
Supporting our argument that SOEs are associated with less adverse selection bias than NSOEs, Wang, Wong, and Xia
(2008) study how specific institutional features in China affect the decision making of Chinese listed firms auditors and
find that SOEs are more likely to hire small auditors than NSOEs. The authors argue that the results can be explained by
the SOEs low demand for large auditors, government bailouts, or collusion incentives.
be approved than issuers without government ownership, regardless of their accounting quality.4
Consequently, SOE issuers are less motivated to hire prestigious underwriters than NSOEs to
signal earnings quality.
In contrast, NSOE issuers face a greater risk of adverse selection by public investors and
the CSRC. To mitigate adverse selection, they are more likely to hire reputable underwriters
to signal their earnings quality. Therefore, with the signaling hypothesis, we predict a negative
relation between underwriter reputation and earnings management for NSOE issuers. The negative
correlation is attenuated for SOE issuers.
With respect to the mitigating hypothesis, we again base our argument on different issuers. As
argued above, NSOE issuers face higher levels of adverse selection as investors and the CSRC
more concerned about the accounting information of NSOEs than that of SOEs. An underwriter
of an NSOE issuer, either to establish, maintain, or enhance its reputation, is likely to certify
earnings quality by monitoring earnings management behavior since investors and the CSRC can
identify the underwriters of issuers of high earnings quality (and, as a result, superior long-term
performance) for future underwriting tasks, which is beneficial for future business. However,
the argument may not be equally applicable to SOE issuers. Due to their lower level of adverse
selection, SOE issuers accounting information is relatively less important for investors to evaluate
and for the CSRC to approve an IPO application. As a result, underwriters of SOE issuers have
fewer legal liabilities, as well as less incentive to exert efforts in mitigating pre-IPO earnings
management. In this case, underwriters tend to be motivated to mitigate earnings manipulation
only when the issuer is an NSOE.
By the inferences from both the signaling and mitigating arguments, we hypothesize that
underwriter reputation is negatively associated with earnings management for NSOE issuers.
However, for SOE issuers, the negative relationship is not obvious and, as such, subject to
empirical evidence.
II. Data and Variables
This section explains the sample and data sources. Then, we describe the measures of the key
variables used in this study.
A. Sample and Data Sources

Our sample consists of all the IPO issuers on the Shanghai and Shenzhen Stock Exchanges from
2002 to 2008. Our sample starts in 2002 because the CSRC issued a new regulation in March 2001
(see Appendix B for a detailed description of IPO reforms) that shifted the rights of proposing
potential issuers from the government to underwriters, thus requiring that lead underwriters
provide advice to issuers with respect to corporate restructuring. In May 2009, the CSRC issued
another regulation requiring IPO pricing to be based on book building rather than regulated PE
ratios. To obtain a clear sample that excludes the effect of significant regulatory changes, we
restrict our latest sample observations to the year 2008. The relevant data are extracted from the
Wind database. Specifically, this database provides data on the year of the IPO, its underwriters,
the issuing amounts, and pre-IPO financial data. After deleting observations with missing values,
we obtain a final sample of 503 issuers from 2002 to 2008.
4
From 2006 to 2010, the CSRC rejected 57 NSOE IPO applications, in contrast with only 15 SOE IPO applications (data
manually collected from the CSRC website, http://www.csrc.gov.cn/pub/zjhpublic).
B. Measure of Underwriter Reputation
Previous studies measure underwriter reputation based on either: 1) underwriter ranking in the
IPO tombstone (Carter and Manaster, 1990) or 2) the underwriters market share (Megginson and
Weiss 1991), where a larger market share indicates a higher underwriter reputation. Since neither
IPO tombstones nor data on underwriter rankings from authoritative organizations are available
in China, we estimate underwriter reputation based on the second measure, the underwriters
market share. An underwriters market share is calculated as the percentage of the total amount the
issuers investment bank j underwrote the year prior to the IPO. That is, i UWAMTi,j / i UWAMTi ,
where UWAMTi,j is the amount of firm i underwritten by investment bank j. If a single issuer
is underwritten by more than one investment bank, the average underwriting amount is used for
each investment bank.5
This study investigates the impact of underwriter reputation on issuers with different ownership
backgrounds. As discussed in Section II, government ownership alters the motivation of issuers
to signal earnings quality and underwriters to mitigate earnings management. As a response,
underwriters may adopt different strategies when choosing potential clients. As observed in the
first row of Panel A, Table I , the top five underwriters in terms of pre-IPO market share during
our sample period are CITIC Security, China International Capital Corporation (CICC), China
Galaxy, BOC International, and Cinda Security, respectively, which also ranked in the top five in
the SOE market. However, these underwriters rarely ranked in the top five within the non-SOE
market in our sample period. For example, CITIC Security ranks first in the SOE market, but
is not an important player in the NSOE market. Panel B of Table I provides further evidence of
different strategies adopted by underwriters in developing their clients. A significantly negative
correlation between underwriter market share in SOE and NSOE issuers is found suggesting that
underwriters targeting the SOE IPO market do not focus on the NSOE IPO market and vice versa.
Chen and Xu (2011) empirically investigate underwriters different strategies of obtaining
market shares in different markets and find that they rely on government ownership to expand
market shares in the SOE IPO market, while industry specialization is more important in at-
tracting issuers in the NSOE market. Therefore, it is improper to use market shares in the whole
market to proxy for the reputation of underwriters with different strategies. Thus, we calculate
market shares separately in the different market segments. The distributions of underwriters mar-
ket shares in each market are reported in Panel C of Table I. We note that the average market share
in different market segments is around 3%. In addition, none of the median values of the market
shares in the three markets are economically significant (i.e., more than 0.5%). This is especially
true in the NSOE market, where the median value is zero indicating that half of the underwriters
in the NSOE market did not underwrite any IPO tasks the previous year. To effectively capture
the effect of underwriter reputation, we define the dummy variable, UWREP, as equal to one if
the underwriters market share in a specific market the year prior to an IPO ranks in the top 25%,
and zero otherwise.
C. Measure of Earnings Management

Reported earnings consist of cash flow from operations and accruals resulting from accrual basis
accounting. While managers exert limited discretion in managing cash flows, they manipulate
earnings through accruals. Accruals can be divided into discretionary and non-discretionary
5
Note that underwriter reputation is measured by the market share one year prior to the IPO year, instead of the IPO
year, as underwriter reputation was established before the occurrence of a particular underwriting task. Section IV uses
alternative measures of underwriter reputation with statistically similar results.
Table I. Underwriting Markets Separated by Issuer Ownership
The sample period is from 2002 to 2008. All A-share IPO issuers on the Shanghai and Shenzhen Exchanges
are included with N = 503. An underwriters market share is measured as the percentage of the amount
underwritten with respect to the total underwriting amount of the whole market or of each segment market
one year prior to the IPO year. An issuer is identified as an SOE if its ultimate shareholder is the government;
otherwise, it is identified as an NSOE. Panel A lists the top five underwriters in terms of market share in
the whole, SOE, and NSOE markets, respectively. Panel B reports the pairwise correlation of underwriter
market shares in the whole, SOE, and NSOE markets. Panel C describes the distribution of underwriter
market shares.
Panel A. Top Five Underwriters by Market Segment
1 2 3 4 5
WHOLE CITIC CICC China BOC Cinda
Security Galaxy International Security
SOE CITIC CICC China BOC Cinda
Security Galaxy International Security
NSOE GF Security Guosen Security Pingan Security Huatai Security China Merchants
Panel B. Pairwise Correlation of Underwriter Market Shares by the SOE, NSOE, and Whole IPO Markets
NSOE WHOLE
SOE 0.082 0.802
Whole 0.0083 1.000
Panel C. Distribution of Underwriters Market Shares by Market Segment the Year before IPO
Percentile 25% 50% 75% Largest Mean

Whole 0 0.005 0.019 0.493 0.021
SOE 0 0.003 0.018 0.493 0.030
NSOE 0 0 0.044 0.256 0.036
Significant at 0.01 level.

accruals. Consistent with prior research (Dechow, Sloan, and Sweeney, 1995; Jo, Kim, and Park,
2007), we use discretionary accruals to measure earnings management.6 Since non-discretionary
accruals are determined by a firms economic fundamentals, managers typically do little to
manipulate them.
Bartov, Gul, and Tsui (2001) indicate that the cross-sectionally modified Jones model is
superior to the time series Jones model in detecting earnings management. Therefore, we use a
cross-sectionally modified Jones model to measure discretionary accruals. Following Hribar and
6
Some studies (Chen and Yuan, 2004) use nonoperating income to measure earnings management. However, as indicated
by Chen and Yuan (2004), Chinese regulators may scrutinize firms that use nonoperating income. In addition, their ability
to do so has improved over time. Therefore, it is likely that excess nonoperating income will be particularly examined
by the CSRC and will then be corrected during the IPO process. When compared to nonoperating income, discretionary
accruals are relatively difficult for regulators to detect. Thus, we use discretionary accruals, rather than nonoperating
income, to measure earnings management. Section IV employs alternative measures to proxy for earnings management
and the results remain unchanged.
7
Collins (2002), we calculate total accruals (TAC) using a cash flow statement approach:
TAC = Net income Cash flow from operations. (1)
To calculate discretionary accruals using the modified Jones model, we take the residual term
from estimating the cross-sectional regression:
TAC = 0 + 1 PPE + 2 (REV AR) + , (2)
where PPE is property, plant, and equipment, REV is the change in sales, and AR is the change
in accounts receivable. All the variables are deflated by the beginning balance of the years total
assets. The explanatory variable PPE is included as the greater the investment in PPE, the more
depreciation charged resulting in negative accruals. REV AR is included to control for sales
on credit, which is another major determinant of accrual.
Equation (2) is estimated by first excluding all IPO firms in year t, and then it is regressed
by industry year. Any industry with fewer than eight observations is removed. We obtain the
estimates of 0 , 1 , and 2 after running the regression on Equation (2). The terms 0 , 1 , and
2 are then used as firm-specific parameters to estimate the non-discretionary accruals of each
IPO firm. The residual term () is the discretionary accrual (DAC) for each individual IPO firm
that is used as the measure of earnings management.
D. Descriptive Statistics
Panel A of Table II presents the descriptive statistics of the variables used in this study. To
mitigate the concern that the results may be driven by extreme values, we winsorize all financial
reporting data at 1%. We observe that for issuers from 2002 to 2008, the mean and median values
of our dependent variable, discretionary accruals (DAC), are 0.088 and 0.080, respectively. A
mean (median) value of 0.151 (0.140) is reported for the pre-IPO operating cash flows deflated
by total assets (CFO). The pre-IPO mean (median) return on assets (ROA) is 0.151 (0.138) and the
mean value of total assets (ASSETS) for issuers in their pre-IPO year is RMB 44.40 billion, while
the median value of ASSETS is RMB 0.484 billion. The right-skewed distribution is caused by a
maximum value of RMB 6.457 trillion (not reported here). The total issuing amount (ISSAMT)
has a mean (median) value of RMB 1.623 (0.307) billion. A mean of 8.5% of our sample is audited
by a Big Four auditor (AUDITOR).8 Three variables, STATESH, MGMTSH, and LARGESH, are
used to identify the issuers ownership structure. Specifically, STATESH is a dummy variable
that is equal to one if the firm is ultimately controlled by the government, and zero otherwise;
MGMTSH is equal to one if top management holds shares, and zero otherwise; and LARGESH
measures the percentage of shares owned by the largest shareholder. About half (49.4%) of our
sample firms are SOEs. An average of 5.8% of our sample firms has top management ownership.
The mean value of LARGESH is 0.430 suggesting that the largest shareholder holds 43% of the
shares, on average. The variable GSALES, which measures the percentage change in sales from
7
In addition to the argument of Hribar and Collins (2002), we find that using the balance sheet approach to calculate
total accruals in an IPO dataset results in fewer observations than using the cash flow statement approach. This is because
some issuers disclose pre-IPO financial data for only one year and the cash flow statement approach has fewer data
requirements.
8
The percentage of firms audited by Big Four auditors is relatively low in China when compared to the US market
due to other, powerful Chinese certified public accountant firms (commonly referred to as Top Ten auditors). Defining
AUDITOR as either Big Four or Top Ten produces statistically similar empirical results.
Table II. Descriptive Statistics
are included with a total of 503 observations. The variable DAC is total discretionary accruals based on
the modified Jones model; CFO is operating cash flow divided by total assets in the year before the IPO;
ROA is the return on total assets in the year before the IPO; ASSETS is total assets at year-end before the
IPO in RMB billion; ISSAMT is the total issuing amount of an issuer measured in RMB billion; AUDITOR
is a dummy variable that is equal to one if an issuer is audited by a Big Four auditor, and zero otherwise;
STATESH is a dummy variable equal to one if the issuers ultimate controller is the government, and zero
otherwise; MGMTSH is a dummy variable equal to one if top management holds shares, and zero otherwise;
LARGESH is the percentage of shares owned by the largest shareholder; GSALES is the percentage change
in sales; and PB is the market-to-book ratio at the end of the IPO year. All financial reporting data are
winsorized at 1%. An issuer is identified as an SOE if its ultimate shareholder is the government; otherwise,
it is identified as an NSOE. Panel A reports descriptive statistics for all of the variables. Panel B presents
the means of all of the variables by year. Panel C compares SOE and NSOE issuers. Panel D divides the
sample into deciles based on the value of DAC. An underwriter is classified as reputable if its market share
on the SOE or NSOE market in the year prior to the IPO is ranked among the top 25%.
Panel A. Descriptive Statistics of Variables

N Mean Q25 Median Q75 Std. Dev.
DAC 503 0.088 0.000 0.080 0.170 0.157
CFO 503 0.151 0.080 0.140 0.220 0.109
ROA 503 0.151 0.099 0.138 0.187 0.078
ASSETS (RMB billion) 503 43.40 0.271 0.484 0.909 435.000
GSALES 503 1.267 0.540 0.920 1.640 1.230
ISSAMT (RMB billion) 503 1.632 0.221 0.307 0.485 6.489
AUDITOR 503 0.085 0 1 1 0.279
STATESH 503 0.494 0 0 1 0.500
MGMTSH 503 0.058 0 0 0 0.233
LARGESH 503 0.430 0.300 0.430 0.540 0.164
PB 503 3.869 2.210 3.125 4.640 2.602
Panel B. Means of Variables by Year

2002 2003 2004 2005 2006 2007 2008
N 66 63 99 14 70 115 76
DAC 0.009 0.007 0.021 0.233 0.142 0.168 0.142
CFO 0.119 0.158 0.132 0.062 0.165 0.158 0.199
ROA 0.143 0.153 0.149 0135 0.126 0.148 0.191
ASSETS 3.006 1.386 0.753 2.682 171.282 539.990 4.836
GSALES 0.927 1.223 1.145 1.017 1.365 1.410 1.498
ISSAMT 0.582 0.467 0.296 0.377 2.346 11.573 1.361
AUDITOR 0.106 0.032 0.010 0.214 0.157 0.122 0.658
STATESH 0.803 0.667 0.515 0.429 0.471 0.405 0.224
MGMTSH 0 0 0.010 0 0.057 0.086 0.184
LARGESH 0.457 0.443 0.385 0.444 0.410 0.435 0.461
PB 3.464 2.778 2.294 2.023 3.254 7.215 2.992
(Continued)
Table II. Descriptive Statistics (Continued)
Panel C. Comparison of Means for the SOE and NSOE IPO Markets
SOE NSOE Difference (t-value)
N 248 255
DAC 0.050 0.124 5.463
CFO 0.141 0.161 2.041
ROA 0.129 0.173 6.587
ASSETS 95.400 0.556 5.966
GSALES 1.159 1.372 1.945
ISSAMT 2.962 0.333 4.529
AUDITOR 0.145 0.027 4.819
MGMTSH 0.020 0.094 3.596
LARGESH 0.472 0.388 5.908
PB 3.528 4.193 2.886
Panel D. Distribution of Earnings Management

10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
DAC 0.100 0.030 0.020 0.050 0.080 0.120 0.150 0.190 0.270 0.590
% of SOEs 0.667 0.580 0.680 0.667 0.580 0.500 0.353 0.280 0.300 0.333
By reputable 0.382 0.207 0.412 0.559 0.413 0.240 0.220 0.500 0.267 0.294
underwriter: SOE
market
By reputable 0.412 0.238 0.312 0.294 0.333 0.520 0.394 0.333 0.343 0.235
underwriter: NSOE
market


year t 1 to year t, has a mean of 1.267, and PB, which is the market-to-book ratio at the end of
the IPO year, exhibits a mean of 3.869. All of the variables except ASSETS exhibit normal values
of the standard deviation indicating that our sample is not skewed.
Our sample distribution by year is reported in Panel B of Table II. The number of observations
ranges from 14 in 2005 to 115 in 2007, with the minimum of IPO issuers in 2005 attributed to a one
year suspension of CSRCs IPO approvals.9 Our dependent variable, DAC, increases gradually
before the suspension and maintains a relatively constant growth rate after the suspension of
IPOs. We observe from Panel B of Table II that the mean values of ASSETS in 2006 and 2007
are high. The reason is that the four largest Chinese banks (Bank of China, Industrial and
Commercial Bank of China, Bank of Communications, and China Construction Bank) went
public in 2006 and 2007. We also observe an overall decrease in the mean value of STATESH
over our sample period, from 80.3% in 2002 to 22.4% in 2008, demonstrating a declining
number of SOE issuers. This reinforces the privatization trend of reform in Chinas capital
market.
Panel C of Table II describes the mean values of all of the variables for SOE and NSOE issuers
separately. A univariate test is conducted to compare the difference between the two markets
9
The CSRC closed the Chinese IPO market from June 7, 2005 to June 19, 2006 to carry out the split share structure
reform. See Chen, Jin, and Yuan (2011) for a detailed discussion of the reform.
using a t-test. We observe that NSOE issuers are associated with higher DAC values than SOE
issuers. With respect to control variables, the following is noteworthy. When compared with
NSOE issuers, SOE issuers have less operating cash flow (CFO), exhibit poorer performance
(ROA), lower growth potential (GSALES, PB), less management shareholding (MGMTSH), and
higher levels of ownership concentration (LARGESH), finance greater amounts (ISSAMT), and
are larger (ASSETS) and more likely to use Big Four auditors (AUDITOR).
While Panel C of Table II suggests the lower earnings management of SOEs, Panel D of
Table II provides more insight into the distribution of earnings management. Issuers are divided
into deciles based on DAC, and the second row reports the percentage of SOE issuers in each
decile. Although the lowest DAC decile is comprise of two-thirds SOEs, the highest DAC decile
still consists of one-third SOE issuers. This distribution suggests that earnings management is
also important to SOEs.10 The third and fourth rows of Panel D of Table II report the percentages
of issuers underwritten by reputable underwriters in each decile for the SOE and NSOE markets,
respectively. While no clear pattern is observed for the SOE market (third row), the following
is found for the NSOE market. First, the percentage of issuers underwritten by reputable under-
writers in the lowest DAC decile is relatively high (0.412), which is consistent with the signaling
hypothesis that issuers are likely to hire a reputable underwriter to signal high earnings quality. In
addition, the percentage of issuers underwritten by reputable underwriters is highest in the sixth
decile, which may be the result of the mitigating effect where most reputable underwriters monitor
their clients earnings management behavior at an acceptable level that is just above the median.
Table III reports the Pearson correlation coefficients between the variables used in this study.
The upper part of Table III reports the correlations between the control variables. These cor-
relation coefficients are generally within a normal range suggesting that our variables are free
of multicollinearity problems. Moreover, to further ensure that our empirical model is not sig-
nificantly affected by multicollinearity problems, we check the variance inflation factors (VIFs)
of our regression. Our test indicates that the VIFs of all the independent variables are less than
3.1. The lower part of Table III presents the correlations between underwriter reputation in the
different market segments and the control variables. Here UWREP_WM, UWREP_SOE, and
UWREP_NSOE represent underwriter reputation in the whole IPO market, the SOE IPO market,
and the NSOE IPO market, respectively. We note that the correlations between UWREP_SOE
and the control variables and between UWREP_NSOE and the control variables are generally
opposite indicating the significant differences between the roles of underwriter reputation in the
SOE and NSOE IPO markets. This finding supports our argument that investment banks may
adopt different strategies in different market segments.
III. Model Specification and Empirical Results
A. Model Specification
This section employs multivariate analysis to empirically test our hypothesis. We seek to an-
swer the research question of whether reputable underwriters can effectively mitigate earnings
10
This is intuitive since IPO prices are set through a regulated PE ratio with an upper bound during our sample period.
Both SOEs and NSOEs have incentives to manage earnings to raise a maximum amount of equity capital. For example,
before the China Railway Group went public in 2007, it wrote off its over accrued employee benefits. As a result, its
pre-IPO earnings increased by RMB 1.173 billion. In comparison, the net income reported in the prospectus was RMB
1.544 billion. The focus of our paper, however, is not the type of firm, SOE or NSOE, associated with higher earnings
management but, the different incentives of the issuers and their underwriters.
Table III. Pearson Correlations of Independent Variables
are included with a total number of 503 observations. The variable AUDITOR is a dummy variable equal to
one if an issuer is audited by a Big Four auditor, and zero otherwise; CFO is operating cash flow divided
by total assets in the year before the IPO; LNISSAMT is the natural log of an issuers total issuing amount,
measured in RMB billion; STATESH is a dummy variable equal to one if the issuers ultimate controller
is the government, and zero otherwise; MGMTSH is a dummy variable equal to one if top management
holds shares, and zero otherwise; LARGESH is the percentage of shares owned by the largest shareholder;
GSALES is the percentage change in sales; and PB is the market-to-book ratio at the end of the IPO year.
All financial reporting data are winsorized at 1%. The variables UWREP SOE and UWREP NSOE are
dummies that are equal to one if the ranking of the underwriters market share on the SOE or NSOE market,
respectively, in the year prior to the IPO is among the top 25%, and zero otherwise.
A B C D E F G H

AUDITOR (A) 1 0.007 0.670 0.210 0.076 0.103 0.057 0.031
CFO (B) 1 0.1041 0.093 0.008 0.017 0.107 0.136
LNISSAMT (C) 1 0.328 0.064 0.211 0.086 0.035
STATESH (D) 1 0.159 0.255 0.092 0.127
MGMTSH (E) 1 0.049 0.079 0.035
LARGESH (F) 1 0.029 0.004
GSALES (G) 1 0.282
PB(H) 1
UWREP_SOE 0.334 0.039 0.418 0.251 0.063 0.088 0.022 0.090
UWREP_NSOE 0.097 0.005 0.099 0.202 0.091 0.023 0.0327 0.099


manipulation in different market segments. As discussed in Section I, issuers signal earnings qual-
ity by hiring prestigious underwriters and/or underwriters mitigate pre-IPO earnings management
to protect their reputations. The correlation between underwriter reputation and issuer earnings
management may suffer from an endogeneity problem as the choice of reputable underwriters
may not be an exogenous event. To address this issue, we employ a two-stage least squares (2SLS)
approach. Specifically, in the first stage, we estimate a probit underwriter choice model in which
the ex ante likelihood of choosing a prestigious underwriter, denoted by UWREP, is regressed on
a set of variables that are deemed to affect the choice of a reputable underwriter:
UWREP = 0 + 1 UW GOV + Controls, (3)
where UWREP is a dummy variable that equals one if the market share of a specific underwriter
in the year prior to the IPO is in the upper 25% of the sample, and zero otherwise. Here UW_GOV
is employed as the instrumental variable, it being a dummy variable that equals one if the
underwriter is ultimately owned by the government, and zero otherwise. Government ownership
of underwriters is relevant to underwriters strategies in choosing clients and to clients choices
of underwriters (Chen and Xu, 2011). We argue that UW_GOV is a good candidate for the
instrumental variable since the ownership structure of underwriters is largely exogenous and can
affect the decision to choose a particular underwriter, which in turn affects UWREP.
In the second stage, we empirically test the association between issuer pre-IPO earnings
management and underwriter reputation. To do so, we estimate the following regression using
the predicted value of Equation (3) to proxy for underwriter reputation and linking that with
DAC:
DAC = 0 + 1 UWREP hat + Controls. (4)
We also include a set of control variables deemed to influence pre-IPO earnings management,
industry, and year fixed effects. In this equation, DAC is a proxy for earnings management,
as estimated by the modified Jones model specified in Equation (2). A negative coefficient
for UWREP_hat is consistent with the argument that prestigious underwriters alleviate earnings
management. According to our hypothesis, we expect a more negative coefficient for UWREP_hat
for NSOE issuers than for SOE issuers.
Other variables are included in the regression to control for issuer-specific factors that may
affect an issuers pre-IPO earnings management behavior. First, high-quality auditors are expected
to detect and mitigate earnings management. Thus, we include a dummy variable that is equal
to one if the issuer is audited by a Big Four auditor, and zero otherwise (AUDITOR). However,
auditors roles in China are often questioned because of government relationships and institutional
features (Wang, Wong, and Xia, 2008), especially given the leading role played by underwriters
(see Appendix A). Therefore, we expect that when compared with the role of underwriters,
auditors, in reducing earnings management during the IPO process in China, have only a second-
order effect. Additionally, CFO, operating cash flow deflated by total assets in the year prior
to the IPO, is used to control for real performance in terms of cash flow. We predict that firms
with higher levels of cash performance are less likely to use accruals to improve book earnings
(i.e., a negative association exists between DAC and CFO). Moreover, LNISSAM, the natural
logarithm of an issuers total issuing amount, is employed to control for issuance size. In addition,
two variables are included to control for an issuers governance structure: 1) MGMTSH is equal
to one if top management holds shares, and zero otherwise, and 2) LARGESH measures the
percentage of shares owned by the largest shareholder. Also, since high growth firms naturally
have greater accruals, we include two more variables to control for an issuers growth potential:
1) GSALES, calculated as the percentage change in sales from year t 1 to year t, and 2) PB,
the market-to-book ratio. Finally, industry and year dummies are included to control for industry
and year fixed effects. The same set of control variables as mentioned earlier, AUDITOR, CFO,
LNISSAMT, MGMTSH, LARGESH, GSALES, PB, and industry and year fixed effects, is also
included in the first-stage probit model.
B. Regression Results
We estimate Equations (3) and (4) for SOE and NSOE issuers, respectively. The results for the
2SLS approach are reported in Table IV. All t-statistics are corrected for IPO-year clustering.11
Section A (B) describes the results for the SOE (NSOE) subsample, while Column 1 (2) reports
the results of the first (second) stage. We find that in the first-step regression, the coefficient of
UW_GOV is significantly related to UWREP, with a negative sign for the NSOE market indicating
that NSOE issuers are less likely to be underwritten by government-related investment banks.
According to Chen and Xu (2011), the negative relationship is driven by underwriter strategy.
Government-owned underwriters take on SOEs as their main target clients and throw less effort
11
We also estimate the 2SLS model with t-statistics adjusted for heteroskedasticity to ensure that our results are robust to
different econometric specifications. Although not presented in the paper for brevity, the results are statistically similar
to the main results reported in Table IV.
Table IV. 2SLS Regressions on Earnings Management
are included with 503 total observations. In the first stage, the following probit model is estimated:
UWREP = 0 + 1 UW GOV + Controls.
In the second stage, the following equation is estimated:
DAC = 0 + 1 UWREP hat + Controls,
where DAC is total discretionary accruals based on the modified Jones model and UWREP_SOE
and UWREP_NSOE are dummies equal to one if the ranking of the underwriters market share on the SOE
or NSOE market, respectively, in the year prior to the IPO is among the top 25%, and zero otherwise. The
predicted values of these variables in the first stage are used in the second stage regression. The variable
UW_GOV is equal to one if the underwriter is owned by the government, and zero otherwise; AUDITOR
is a dummy variable equal to one if an issuer is audited by a Big Four auditor, and zero otherwise; CFO is
operating cash flow divided by total assets in the year before the IPO; LNISSAMT is the natural log of the
issuers total issuing amount, measured in RMB billion; MGMTSH is a dummy variable equal to one if top
management holds shares, and zero otherwise; LARGESH is the percentage of shares owned by the largest
shareholder; GSALES is the percentage change in sales; and PB is the market-to-book ratio at the end of the
IPO year. All financial reporting data are winsorized at 1%. All of the t-values are corrected for IPO-year
clustering and are presented in parentheses.
Section A: SOE Section B: NSOE
(1) 1st Stage (2) 2nd Stage (3) 1st Stage (4) 2nd Stage
UWREP_SOE 0.040
(0.410)
UWREP_NSOE 0.132
(3.820)
UW_GOV 0.785 1.026
(1.540) (6.060)
AUDITOR 0.529 0.055 0.530 0.091
(1.390) (1.110) (0.560) (1.630)
CFO 0.449 0.487 0.573 0.388
(0.520) (2.210) (0.820) (2.130)
LNISSAMT 0.905 0.000 0.301 0.064
(5.890) (0.010) (2.660) (3.760)
MGMTSH 0.000 0.000 0.091 0.016
(0.000) (0.000) (0.580) (2.400)
LARGESH 0.677 0.062 0.384 0.004
(0.740) (1.310) (0.540) (0.200)
GSALSES 0.097 0.018 0.026 0.002
(0.630) (2.000) (0.560) (0.440)
PB 0.011 0.007 0.001 0.024
(0.150) (1.490) (0.040) (4.120)
INTERCEPT 17.781 0.061 6.968 1.323
(5.740) (0.150) (2.810) (2.880)
INDUSTRY Controlled Controlled Controlled Controlled
YEAR Controlled Controlled Controlled Controlled
N 248 248 255 255
Pseudo R2 0.297 0.248
Adj. R2 0.436 0.506


into their NSOE clients. Consequently, reputable underwriters are less likely to exist in the NSOE
market.
With respect to the results of the second-step regression, the coefficient of UWREP for the
NSOE issuer subsample (Column 4) is significantly negative (0.132 with a t-value of 3.820)
suggesting that underwriter reputation can effectively reduce pre-IPO earnings manipulation if
the issuer is an NSOE issuer. In contrast, the estimate of UWREP for the SOE issuer subsample
(Column 2) is not significantly different from zero. As argued in Section I, these results can
be attributed to either: 1) SOE issuers having fewer incentives than NSOE issuers to signal
earnings quality or 2) the underwriters of NSOE issuers are more motivated to mitigate earnings
management than those of SOE issuers since NSOE issuers face higher levels of adverse selection
by public investors and/or the CSRC.
The following results from the second stage are noteworthy regarding the control variables.
The variable CFO is negatively associated with earnings management at less than the 10% level
in all model specifications, which is consistent with our prediction and prior studies (Dechow and
Dichev, 2002). LNISSAMT is significantly positive if the issuer is in the NSOE market segment,
confirming a positive correlation between financing amounts and pre-IPO earnings management.
The coefficient of GSALES (PB) is significantly positive in the SOE (NSOE) market segment
indicating that issuers with greater growth potential are associated with higher discretionary
accruals. The sign on AUDITOR is negative, but not significant confirming the concern over the
auditors role in mitigating pre-IPO earnings manipulation. Finally, with respect to the relationship
between the issuers governance structure and earnings management, we find that the coefficient
of MGMTSH is significantly negative in the NSOE sample, while that of LARGESH is not
significant in either the SOE or NSOE sample.
IV. Robustness
A. Alternative Measures of Discretionary Accruals

Ball and Shivakumar (2008) raise concerns over the use of the Jones model to measure earnings
management in the IPO setting, as Teoh, Wong, and Rao (1998b) did. Ball and Shivakumar (2008)
argue that the application of the Jones model in Teoh, Wong, and Rao (1998b) may be biased in
the IPO setting due to the use of: 1) balance sheet data to calculate total accruals, 2) DAC in the
IPO year to measure pre-IPO earnings management, and 3) variables deflated by pre-IPO total
assets in the IPO year. However, our study is less likely to suffer from these concerns due to the
following. First, instead of the balance sheet approach, we use the cash flow statement approach
to calculate total accruals following Hribar and Collins (2002). Additionally, since our focus is on
pre-IPO earnings management, total accruals calculated by Equation (1) are accruals in the year
prior to the IPO rather than in the IPO year. Therefore, the estimation of total accruals does not
involve the use of data from the post-IPO period. As a result, deflating total accruals by pre-IPO
total assets does not suffer from the problems raised by Ball and Shivakumar (2008).
This section examines the robustness of our results to alternative measures of discretionary
accruals. Specifically, we follow the approach of Ball and Shivakumar (2006) that emphasizes
the role of accruals in the timely recognition of gains and losses.12 The intuition is that this role
12
We do not follow Ball and Shivakumar (2008) in using a positive (negative) operating cash flow to measure economic
gains (losses) in the IPO setting as our sample only contains three of 503 observations (0.59%) that have negative cash
flows. Since Ball and Shivakumars (2008) approach originates from their 2006 work, we adopt a suggested measure from
their 2006 paper that shares a similar theoretical background as their 2008 work.
increases the volatility of accruals, and volatile accruals can be viewed as having low earnings
quality in the modified Jones model. Furthermore, since accounting recognition is conditionally
conservative in that losses are recognized in a timelier manner than gains (Basu, 1997), the
relationship between proxies for economic gains and losses and accruals is nonlinear. However,
Guary (2006) points out that the recognition of gains is as important as that of losses from the
perspective of either the cost of accounting accruals or the contract. As a result, conservatism is
not a concern in estimating discretionary accruals. To determine whether our results are robust
to alternative measures of discretionary accruals, we adopt the model of Ball and Shivakumar
(2006), both with and without conditional conservatism concerns, to estimate discretionary
accruals.
Following Ball and Shivakumar (2006), we use industry median-adjusted operating cash flow
to measure economic gains and losses. When the conditional conservatism concern is included
in the model, the following piecewise linear regression is estimated:
TAC = 0 + 1 REV + 2 PPE + 3 ADJCFO + 4 DADJCFO

+ 5 ADJCFO DADJCFO + , (5)
where TAC is the total accruals as calculated in Equation (1); REV is the change in sales; PPE
is plant, property, and equipment; ADJCFO is the industry median-adjusted operating cash flow;
and DADJCFO is a dummy variable equal to one if ADJCFO is negative, and zero otherwise. All
of the variables are scaled by total assets at the beginning of the year. As in the main test, the year
of interest here is one year prior to the IPO event. Therefore, deflating the variables by total assets
at the beginning of year does not lead to the concern raised by Ball and Shivakumar (2008).
Next, we estimate discretionary accruals without considering conservatism concerns. The
correlation between discretionary accruals and the proxies for gains or losses becomes linear and
Equation (5) can be rewritten as:
TAC = 0 + 1 REV + 2 PPE + 3 ADJCFO + . (6)
The discretionary accruals, DAC, with and without concerns of conservatism, are estimated as
the residual terms of Equations (5) and (6), respectively. We then re-estimate Equation (3) using
these two alternative measures of discretionary accruals. The results are reported in Table V.
Sections A and B of Table V report the results using DAC with and without conservatism
concerns as the dependent variable, respectively. As demonstrated, the coefficient of UWREP in
the SOE market is not significant in either Section A or B. In contrast, the coefficient of UWREP
in the NSOE market is significantly negative for both sections. This finding is consistent with
the main results reported in Table IV, although the results become less significant.13 Therefore,
our results are robust to alternative measures of discretionary accruals.
B. Signaling or Mitigation Hypothesis?

Up to now, our empirical results have suggested that underwriter reputation is negatively
associated with earnings management for NSOE issuers. We argue that this negative association
is driven by the signaling hypothesis, the mitigation hypothesis, or a joint effect of the two. Our
13
When compared with the results in Table IV, the coefficients of CFO become positive in the NSOE IPO market. This
may result from the inclusion of CFO as an explanatory variable in estimating discretionary accruals in Equations (5)
and (6).
Table V. Second-Step Regressions on DAC with and without Conditional
Conservatism Concerns
are included. The total number of observations is 446. Section A presents the results when the dependent
variable is DAC with conservatism concerns. Section B reports the results when the dependent variable is
DAC without conservatism concerns. The DAC values with and without conservatism concerns are measured
as the residual terms of the following regressions: with conditioned conservatism concerns:
TAC = 0 + 1 REV + 2 PPE + 3 ADJCFO + 4 DADJCFO + 5 ADJCFO DADJCFO + ,
and without conditioned conservatism concerns:
TAC = 0 + 1 REV + 2 PPE + 3 ADJCFO + ,
where TAC is total accruals; ADJCFO is the industry median-adjusted operating cash flow prior to
the IPO; and DADJCFO is a dummy variable equal to one if ADJCFO is negative, and zero otherwise. All
the above variables are scaled by total assets at the beginning of the year. The independent variables include
UWREP_SOE and UWREP_NSOE, which equal one if the ranking of the underwriters market share in the
SOE or NSOE market, respectively, in the year prior to the IPO is among the top 25%, and zero otherwise;
AUDITOR, a dummy variable equal to one if an issuer is audited by a Big Four auditor, and zero otherwise;
CFO, operating cash flow divided by total assets in the year before the IPO; LNISSAMT, the natural log of
an issuers total issuing amount, measured in RMB billion; MGMTSH, a dummy variable equal to one if top
management holds shares, and zero otherwise; LARGESH, the percentage of shares owned by the largest
shareholder; GSALES, the percentage change in sales; and PB, the market-to-book ratio at the end of the
IPO year. All financial reporting data are winsorized at 1%. The t-values are presented in parentheses.
Section A: DAC with Section B: DAC without
conservatism concerns conservatism concerns
SOE NSOE SOE NSOE
UWREP_SOE 0.054 0.006
(1.030) (0.109)
UWREP_NSOE 0.051 0.057
(1.628) (1.820)
AUDITOR 0.023 0.010 0.040 0.007
(1.309) (0.373) (2.339) (0.260)
CFO 0.063 0.084 0.099 0.079
(1.640) (1.949) (2.664) (1.806)
LNISSAMT 0.008 0.011 0.005 0.010
(0.573) (1.216) (0.371) (1.044)
MGMTSH 0.000 0.017 0.000 0.022
0.000 (1.160) 0.000 (1.464)
LARGESH 0.029 0.044 0.064 0.050
(1.033) (1.506) (2.320) (1.686)
GSALES 0.001 0.004 0.002 0.001
(0.212) (1.284) (0.554) (0.191)
PB 0.004 0.011 0.004 0.010
(1.492) (4.290) (1.386) (3.701)
INTERCEPT 0.077 0.260 0.120 0.252
(0.311) (1.398) (0.495) (1.340)
INDUSTRY Controlled Controlled Controlled Controlled
YEAR Controlled Controlled Controlled Controlled
N 212 234 212 234
Adj R2 0.403 0.301 0.375 0.274


main test does not explicitly differentiate between the two effects for the following two reasons.
First, the incentives of issuers that choose reputable underwriters and reputable underwriters
that certify earnings management are unobservable. Without a direct empirical proxy for these
incentives, we avoid drawing a quick conclusion regarding which hypothesis dominates. Moreover,
both hypotheses lead to the same conclusion. Therefore, it is difficult to differentiate between
them given the existing evidence. This section attempts to determine which effect dominates in
determining the negative correlation between DAC and UWREP using the approach suggested
by Brau and Johnson (2009). These authors examine the causality of earnings management and
the performance of financial intermediaries using simultaneous regression models. Specifically,
we differentiate between the two hypotheses with the following two simultaneous models: 1) the
mitigation hypothesis model:
1st step: UWREP = + Instrument + Controls, (7)
2nd step :DAC = + UWREP hat + Controls, (8)
and 2) the signaling hypothesis model:
1st step: D AC = + Instrument + Controls, (9)
2nd step: UWREP = + DAC hat + Controls. (10)
Our main test follows a 2SLS approach, whereas in the first stage [Equation (3)], a probit
underwriter choice model is estimated and UW_GOV is employed as the instrumental variable.
In the second stage [Equation (4)], the predicted value of UWREP from the first stage is applied
as the test variable. Thus, our main test is substantially the same as the mitigation hypothesis
model [Equations (7) and (8)]. To differentiate between the signaling and mitigation hypotheses,
we continue to test the signaling hypothesis model [Equations (9) and (10)]. Specifically, in the
first-step regression with discretionary accruals as the dependent variable, we use operating cash
flow (CFO) and growth of sales (GSALES) as the instrumental variables. The variable CFO and
the accounting accruals are both indicators of gains and losses. As such, DAC is influenced by
CFO. In addition, firms with higher sales growth are usually associated with higher accruals,
especially IPO firms (Ball and Shivakumar, 2008). Alternatively, there is no compelling argument
that the choice of reputable underwriters is affected by CFO or GSALES. The predicted value of
DAC from Equation (9) is then included as an independent variable in Equation (10).
The results of the second step are reported in Table VI, with Sections A and B presenting
the results for the SOE and NSOE markets, respectively.14 The coefficients of the predicted
value of DAC (DAC_hat) are insignificant in both the SOE and NSOE markets, providing no
evidence for the signaling hypothesis. Together, with the main results provided in Table IV, we
find that the negative relationship between DAC and UWREP is largely driven by the mitigation
hypothesis instead of the signaling hypothesis. However, since the two hypotheses are difficult to
differentiate between, by nature, we acknowledge that this conclusion is drawn conditionally upon
the appropriateness of a simultaneous regression model when addressing the issue. Our findings
are different from that of Brau and Johnson (2009), who support the signaling hypothesis using
US data. This may be due to either: 1) increasing legal liabilities imposed on underwriters, as
14
The results of the first-stage model are not reported in the paper for brevity, but are available upon request.
Table VI. 2SLS with the Signaling Hypothesis Model
are included. The total number of observations is 503. Section A reports the results for the SOE market and
Section B reports the results for the NSOE market. In the first step, the following model is estimated:
DAC = + Instrument + Controls.
In the second step, the following probit model is estimated:
UWREP = + DAC hat + Controls.
DAC is the total discretionary accruals based on the modified Jones model; Instrument includes CFO and
GSALES; and UWREP is either UWREP_SOE or UWREP_NSOE, where UWREP_SOE and UWREP_NSOE
are dummies that are equal one if the ranking of the underwriters market share on the SOE or NSOE market,
respectively, in the year prior to the IPO is among the top 25%e, and zero otherwise. The predicted value
of DAC from the first step is used in the second step regression. The variable UW_GOV is equal to one if
the underwriter is owned by the government, and zero otherwise; AUDITOR is a dummy variable equal to
one if an issuer is audited by a Big Four auditor, and zero otherwise; CFO is operating cash flow divided
by total assets in the year before the IPO; LNISSAMT is the natural log of the issuers total issuing amount,
measured in RMB billion; MGMTSH is a dummy variable equal to one if top management holds shares,
and zero otherwise; LARGESH is the percentage of shares owned by the largest shareholder; GSALES is
the percentage change in sales; and PB is the market-to-book ratio at the end of the IPO year. All financial
reporting data are winsorized at 1%.

Coefficient z-Value Coefficient z-Value
DAC_hat 0.581 0.260 1.694 0.62
UW_GOV 0.779 1.260 1.017 4.120
AUDITOR 0.465 1.010 0.695 0.900
LNISSAMT 0.866 4.760 0.198 0.790
MGMTSH 0.000 0.000 0.120 0.350
LARGESH 0.403 0.520 0.420 0.630
PB 0.029 0.340 0.045 0.590
INTERCEPT 17.166 4.660 4.796 0.940
INDUSTRY Controlled Controlled
YEAR Controlled Controlled
N 248 255
Pseudo R2 0.294 0.204
discussed in introductory section, and/or 2) the relatively weak incentives of issuers to signal their
reporting quality due to the strong demand for stocks from equity market investors.15
C. Alternative Proxies of Underwriter Reputation

In our main test, underwriter reputation is proxied by a dummy variable that is based on the
pre-IPO market share in each market segment, where market share is measured by the percentage
15
An example that supports this conjecture is the IPO subscription ratio. During our sample period, the mean IPO
subscription ratio in China was only 0.5%. This relatively low subscription ratio suggests high investor demand for stocks.
of the total underwriting amount. To determine whether our results are driven by the approach in
which we measure underwriter reputation, we re-calculate the market share of an investment bank
using alternative proxies. First, instead of the dummy variable UWREP, we use the continuous
value of market share to measure underwriter reputation. In addition, we calculate the market
share based on the number (instead of the underwriting amount) of clients underwritten by a
particular investment bank in the year prior to the IPO. Furthermore, we employ the three-year-
ahead cumulative market share to proxy for underwriter reputation. Equation (4) is re-estimated
using the ordinary least squares (OLS) regression with three alternative measures of underwriter
reputation, and the results are reported in Table VII. Consistent with the findings documented
in the previous tables, the coefficient of UWREP using the first two measures is significantly
negative in the NSOE market, while no significant coefficient is found in the SOE market.
The coefficient of UWREP using the three-year-ahead cumulative market share demonstrates an
expected, but insignificant negative sign, possibly since the three-year cumulative market share
negates the most recent underwriter growth and performance. In short, Table VII reports that our
results are generally robust to different measures of underwriter reputation.
D. Cross-Market Test
A competing interpretation of our results argues that the negative correlation in the NSOE
market and the insignificant relation in the SOE market may simply be the result of the competitive
advantages in monitoring the earnings quality of reputable underwriters in the NSOE market. To
address this competing hypothesis, we conduct a cross-market test to determine whether reputable
underwriters in the NSOE market are still effective in mitigating earnings management in the SOE
market. For example, if a prestigious investment bank in the NSOE market underwrites an SOE
issuer, will it still be able to detect and mitigate earnings management as it does for an NSOE
issuer? Equation (4) is then re-estimated on a cross-market basis using the OLS approach. If
government ownership moderates the signaling function of issuers and/or the mitigating function
of underwriters, as we discussed, we do not expect reputable underwriters in the NSOE market
to effectively reduce earnings management behavior in the SOE market.
The results of our cross-market test are presented in Table VIII. The coefficient of
UWREP_NSOE is not significant at the conventional level for the SOE market, implying that a
reputable underwriter in the NSOE market is no longer effective if it underwrites an SOE issuer.
Meanwhile, reputable underwriters in the SOE market do not work effectively in the NSOE mar-
ket either. We interpret our results as signifying that underwriters in the SOE market, with SOE
issuers as their major clients, do not develop their ability when certifying the earnings quality of
NSOE issuers. This result supports our argument that underwriters adopt different strategies in
different market segments validating our approach to measure UWREP separately in each market
segment.
In addition to the above robustness checks, we also conduct the following tests to determine
whether our main results are sensitive to alternative econometric specification and alternative
sampling. First, we estimate Equation (4) using the OLS approach with t-statistics corrected for
IPO-year clustering and heteroskedasticity, respectively. Additionally, to eliminate the influences
of different regulatory environments regarding underwriter behavior, we remove any observations
cross-listed on other stock exchanges. Further, we delete any issuers that belong to the financial
sector due to their different accounting treatment. Moreover, we delete all of the observations
from 2005, which is a special year as the CSRC closed the IPO market for half of the year. As
a result, only a relatively small number of issuers went public in 2005. To rule out any possible
bias in the small sample when measuring earnings management and underwriter reputation, we
Table VII. Robustness Tests for Alternative Measurements of Underwriter
Reputation
are included. The total number of observations is 503. The dependent variable is DAC, total discretionary
accruals, based on the modified Jones model. The variables UWREP_SOE and UWREP_NSOE are measured
by three proxies: 1) Measurement 1 is the continuous variable of market share based on the amount
underwritten in the year before the IPO, 2) Measurement 2 is the market share based on the number of IPOs
underwritten in the year prior to the IPO, and 3) Measurement 3 is the cumulative market share based on
the amount underwritten in the three years prior to the IPO. The variable AUDITOR is a dummy variable
equal to one if an issuer is audited by a Big Four auditor, and zero otherwise; CFO is the operating cash
flow divided by the total assets in the year before the IPO; LNISSAMT is the natural log of the issuers total
issuing amount, measured in RMB billion; MGMTSH is a dummy variable equal to one if top management
All financial reporting data are winsorized at 1%. The t-values are presented in parentheses.
Measurement 1 Measurement 2 Measurement 3

SOE NSOE SOE NSOE SOE NSOE
UWREP_SOE 0.020 0.133 0.163
(0.172) (0.607) (0.937)
UWREP_NSOE 0.246 0.305 0.237
(1.813) (1.983) (1.124)
AUDITOR 0.057 0.099 0.058 0.104 0.032 0.056
(1.744) (1.856) (1.754) (1.962) (0.656) (0.928)
CFO 0.474 0.364 0.475 0.367 0.239 0.308
(6.314) (4.332) (6.332) (4.368) (2.168) (3.326)
LNISSAMT 0.014 0.062 0.015 0.056 0.008 0.048
(1.119) (3.366) (1.284) (3.090) (0.432) (2.477)
MGMTSH 0.089 0.006 0.086 0.006 0.056 0.003
(1.523) (0.191) (1.465) (0.192) (0.862) (0.102)
LARGESH 0.082 0.011 0.083 0.009 0.087 0.007
(1.594) (0.198) (1.613) (0.163) (1.313) (0.106)
GSALES 0.013 0.002 0.013 0.002 0.020 0.006
(1.569) (0.318) (1.562) (0.347) (1.683) (0.761)
PB 0.006 0.025 0.006 0.025 0.010 0.023
(1.145) (4.635) (1.131) (4.616) (1.504) (4.143)
INTERCEPT 0.206 1.262 0.215 1.133 0.075 0.855
(0.808) (3.477) (0.922) (3.162) (0.205) (2.217)
INDUSTRY Controlled Controlled Controlled Controlled Controlled Controlled
YEAR Controlled Controlled Controlled Controlled Controlled Controlled
N 248 255 248 255 141 213
Adj R2 0.489 0.513 0.489 0.514 0.490 0.474


delete these observations. While the results for the above robustness checks are not reported here
for brevity, they are statistically similar to our main results.
In short, the results presented in this section suggest that the main results presented in Table IV
are robust to alternative measures of earnings management and underwriter reputation, different
Table VIII. Cross-Market Test
are included. The total number of observations is 503. Section A reports the results for the SOE market,
while Section B reports the results for the NSOE market. The dependent variable is DAC, total discretionary
accruals, based on the modified Jones model. The variables UWREP_SOE and UWREP_NSOE are dummies
that equal one if the ranking of the underwriters market share on the SOE market or the NSOE market,
respectively, in the prior year is among the top 25%, and zero otherwise; AUDITOR is a dummy variable
equal to one if the issuer is audited by a Big Four auditor, and zero otherwise; CFO is the operating cash
flow divided by the total assets in the year before the IPO; LNISSAMT is the natural log of the issuers total
issuing amount, measured in RMB billion; MGMTSH is a dummy variable equal to one if top management
All financial reporting data are winsorized at 1%.
Coefficient t-Value Coefficient t-Value

UWREP_SOE 0.023 0.910
AUDITOR 0.053 1.730 0.111 2.070
CFO 0.469 6.240 0.370 4.370
LNISSAMT 0.013 1.140 0.054 2.990
LARGESH 0.083 1.610 0.025 0.450
MGMTSH 0.092 1.590 0.010 0.350
GSALES 0.013 1.530 0.002 0.380
PB 0.006 1.150 0.023 4.420
INTERCEPT 0.173 0.750 1.150 3.200
INDUSTRY Controlled Controlled
Year Controlled Controlled
N 248 255
Adj R2 0.401 0.431%


econometric specifications, and alternative sampling. In addition, we rule out the possibility of
alternative explanations.
V. Test of Underwriter Reputation on Post-IPO Performance
Until now, we obtained consistent evidence that a reputable underwriter can effectively restrain
an issuers pre-IPO earnings management behavior in the NSOE market. Prior studies provide
evidence from the US IPO market regarding the effect of underwriter reputation on post-IPO
performance (Hansen and Torregrosa, 1992; Michaely and Shaw, 1994; Carter, Dark, and Singh,
1998; Loughran and Ritter, 2004) and the effect of earnings management on post-IPO performance
(Teoh, Wong, and Rao, 1998a). Jo, Kim, and Park (2007) examine US seasoned equity offerings
and find a positive correlation between underwriter reputation and postissue performance after
controlling for earnings quality. Following their study, we continue to investigate the effect
of hiring a reputable underwriter on post-IPO performance in different market segments after
controlling for earnings quality. First, we calculate monthly abnormal returns as a particular
issuers monthly return minus the monthly value-weighted market return. The 12-month postissue
cumulative abnormal return (POSTCAR) is then calculated as the sum of the 12 consecutive
monthly abnormal returns using the month immediately after the initial listing as the first month.
Although these results are not reported here for brevity, we find that issuers experience an
average (median) postissue 12-month cumulative abnormal return of 9.5% (11.6%), which is
consistent with the previous findings of the long-term underperformance of IPO firms (Ritter,
1991; Brav and Gompers, 1997; Gompers and Lerner, 2003). A yearly analysis suggests that
issuers experienced the lowest (highest) negative postissue returns in 2003 (2006). Conducting a
univariate comparison between SOE and NSOE issuers, we find that the SOE issuers experience
more negative postissue returns than NSOE issuers, although the difference is not significant.
Our multivariate analysis is based on the following model:
POSTCAR = 0 + 1 UWREP + 2 DAC PRED + Controls. (11)
We include ROA, LNISSAMT, MGMTSH, and LARGESH as the control variables, defined
in the same way as in Equation (4). Industry fixed effects are also included in the model.16
We are interested in the sign of 1 , which captures the effect of underwriter reputation on
postissue performance. Including both UWREP and DAC in the regression, we examine the
mechanism of how reputable underwriters affect post-IPO performance. Specifically, we seek
to determine whether reputable underwriters affect post-IPO performance through monitoring
earnings management or through other marginal effects of their reputation. Since DAC is affected
by UWREP, as documented in our previous analysis, we use the predicted value of the OLS model
of Equation (4), DAC_PRED, in this regression.17
Panel A of Table IX portrays the univariate analysis of POSTCAR for issuers underwritten
by different underwriters. It indicates that in the NSOE issuer market, issuers underwritten
by prestigious underwriters obtain higher post-IPO returns than others. However, underwriter
reputation does not lead to a significant difference in the POSTCAR values of the SOE market.
When Equation (11) is estimated by including UWREP only, as in Columns (1) and (4) in
Panel B of Table IX, we find that the coefficient of UWREP is not significant in the SOE market.
In contrast, it is significantly positive in the NSOE subsample providing support for our research
question as to whether prestigious underwriters can improve postissue performance in different
market segments. We interpret these results as the ability of reputable underwriters to effectively
monitor earnings quality in the NSOE market resulting in higher post-IPO returns. We discover
that DAC_PRED alone is significantly negative in both the SOE and NSOE markets suggesting
that low earnings quality deteriorates post-IPO performance in both markets, as indicated in
Columns (2) and (5) in Panel B of Table IX. Interestingly, when we include both UWREP and
DAC_PRED in the regression [Columns (3) and (6)], we find that the coefficients of DAC_PRED
in both markets remain significantly negative and that UWREP is significantly positive only in the
NSOE market. This result suggests that the positive effect of underwriter reputation on post-IPO
performance exists beyond enhancing the earnings quality in the NSOE market, while the effect
of underwriter reputation is not as obvious in the SOE market. As for the control variables, we
determine that the coefficient of financial performance (ROA) is significant and positive only
16
We do not include year dummies in Equation (11) because the calculation of POSTCAR already captures the year fixed
effects.
17
Here DAC_PRED is different from DAC_hat in Equation (10); the former is predicted by Equation (4), while the latter
is based on Equation (9).
Table IX. Analyses of Post-IPO Performance
are included. The total number of observations is 503. The variable POSTCAR is the market-adjusted
cumulative return 12 months after the month of the IPO; UWREP_SOE and UWREP_NSOE are dummies
that are equal to one if the ranking of the underwriters market share in the SOE or NSOE market, respectively,
in the prior year is among the top 25%, and zero otherwise; DAC_PRED is the predicted value of the OLS
regression in Table VI; ROA is the return on assets; LNISSAMT is the natural log of the issuers total issuing
amount, measured in RMB billion; MGMTSH is a dummy variable equal to one if top management holds
shares, and zero otherwise; and LARGESH is the percentage of shares owned by the largest shareholder.
All financial reporting data are winsorized at 1%. Panel A compares the POSTCAR values of issuers by
reputable underwriters and other underwriters for the SOE and NSOE markets, respectively. Panel B reports
the multivariate regression results. Sections A and B report the results for the SOE and NSOE markets,
respectively. The t-values are presented in parentheses.
Panel A. Univariate Analysis of POSTCAR
SOE NSOE
By Reputable Underwriters 0.078 0.013
By Other Underwriters 0.155 0.107
Difference (t-value) 1.530 1.976
Panel B. Multiple Variables Analysis of POSTCAR

(1) (2) (3) (4) (5) (6)
UWREP_SOE 0.005 0.002
(0.088) (0.037)
(1.652) (1.712)
DAC_PRED 0.521 0.519 0.713 0.711
(1.802) (1.769) (2.507) (2.511)
ROA 0.116 0.022 0.020 2.030 2.012 1.960
(0.310) (0.052) (0.045) (4.965) (4.885) (4.765)
LNISSAMT 0.071 0.078 0.078 0.005 0.001 0.003
(2.478) (2.847) (2.637) (0.084) (0.017) (0.055)
MGMTSH 0.027 0.047 0.048 0.089 0.044 0.056
(0.145) (0.262) (0.264) (0.927) (0.462) (0.584)
LARGESH 0.082 0.089 0.089 0.227 0.144 0.184
(0.517) (0.545) (0.537) (1.166) (0.746) (0.955)
INTERCEPT 1.525 1.711 1.705 0.605 0.646 0.568
(2.672) (3.020) (2.886) (0.524) (0.564) (0.497)
INDUSTRY Controlled Controlled Controlled Controlled Controlled Controlled
N 248 248 248 253 253 253
Adj R2 0.144 0.162 0.162 0.186 0.204 0.214


in the NSOE market, while it is insignificant in the SOE market, which is consistent with our
argument that investors emphasize operating performance less when evaluating an SOE issuer.
Size (LNISSAMT) is significantly positive in the SOE market, but insignificant in the NSOE
market indicating that the financing amount is no longer a determinant of the postissue returns
of IPO firms without a government background.
In short, this section examines the underwriters role in postissue performance. Consistent with
the argument that underwriter reputation is effective only in the NSOE market, we document
a significantly positive impact of underwriter reputation in the NSOE market only, even after
controlling for earnings quality. No significant correlation between underwriter reputation and
post-IPO performance is found in the SOE market.
VI. Conclusions
Using a sample of 503 IPO issuers from 2002 to 2008, we find that underwriter reputation is
negatively related to pre-IPO earnings management for NSOE issuers only. Moreover, underwriter
reputation helps improve post-IPO performance, and the effect is beyond enhancing earnings
quality in the NSOE market. In contrast, the effect of underwriter reputation on either pre-IPO
earnings quality or post-IPO performance is not significant in the SOE market.
Evidence documented in this study reveals that in the NSOE market, earnings information is
important for public investors of NSOEs. As such, investment banks are motivated to mitigate
pre-IPO earnings management to build up their reputations. However, in the SOE market, issuers
face a lower level of adverse selection by both investors and the CSRC. As a result, underwriters
have less incentive to mitigate earnings management. While prior studies provide evidence of
the relationship between underwriter reputation and earnings management in the developed
market, they typically consider the market as a homogeneous whole. Our study helps us to
better understand the role of underwriters in the transitional market. Moreover, it emphasizes the
importance of the adverse selection bias faced by issuers in determining the role of underwriters
in earnings management highlighting the importance of future research on the Chinese IPO
market.
Appendix A: IPO Process: The Role of Investment Banks (IBs) and

the CSRC
1. Hire IB for restructuring
Improvement
2. IB forms a restructuring team
3. Implement restructuring
4. Set up a new shareholding corporation
5. IB assists firm with regulatory compliance
Unsuccessful
6. CSRCs preliminary evaluation

of the regulatory compliance
7. IB prepares for registration statements
8. Sponsors due diligence and

recommendations
Rejected by CSRC
9. Lead underwriters internal assessment
10. CSRCs pre-evaluation
11. Revise and resubmit
12. CSRCs final approval
13. Public announcement
14. IPO
Appendix B: IPO process reforms in China
1990-2000: China implements an administrative approval regime for public offerings, often
referred to as the quota system. Under this quota system, the CSRC first sets the amount of
total shares that can be issued to the public. Then, central and provincial governments negotiate
the amount of quotas they can receive and make potential issuer recommendations. The IPO
candidates recommended must go through the issuing process and obtain approval from the
CSRC.
2001-present: This approval system was adopted in 2001. Under it, the CSRC no longer sets
quota restrictions on the amount of shares that can be issued. Instead, it imposes strict regulations
on the quality of and IPO procedures for potential issuers. Underwriters propose IPO candidates,
but the applications must be approved by the CSRC. Since the adoption of this approval system,
the CSRC has undertaken reforms to strengthen underwriter responsibilities. These reforms can
be divided into the following two stages.
2001-2003: The maximum number of in process issues recommended by a single underwriter,
called channels, is determined by the CSRC. The CSRC maintains the right to assign channels and
to give final approval regarding issuers qualifications and offering prices. The choice of potential
issuers, proceeds, and offering prices were transferred from the CSRC to the underwriters.
2004-present: The representative sponsor system was adopted in 2004, where a sponsor is
employed by the investment banks. Under this system, underwriters play the dual role of under-
writer and sponsors representative. Underwriters advise regarding issuers corporate governance,
financial systems, and regulatory compliance, as well as corporate restructuring. The lead under-
writer is responsible for the truthfulness of disclosure. In addition, underwriters are responsible
for post-IPO financial performance with the post-IPO monitoring responsibility lasting up to two
years after the IPO.
References
Aharony, J., J. Lee, and T.J. Wong, 2000, Financial Packaging of IPO Firms in China, Journal of Accounting
Research 38, 103-126.
Aharony, J., J. Wang, and H. Yuan, 2010, Tunneling as an Incentive for Earnings Management During the
IPO Process in China, Journal of Accounting and Public Policy 29, 1-26.
Ball, R. and L. Shivakumar, 2006, The Role of Accruals in Asymmetrically Timely Gain and Loss
Recognition, Journal of Accounting Research 44, 207-242.
Ball, R. and L. Shivakumar, 2008, Earnings Quality at Initial Public Offerings, Journal of Accounting and
Economics 45, 324-349.
Bartov, E., F.A. Gul, and J. Tsui, 2001, Discretionary-Accruals Models and Audit Qualifications, Journal
of Accounting and Economics 30, 421-452.
Basu, S., 1997, The Conservatism Principle and the Asymmetric Timeliness of Earnings, Journal of
Accounting and Economics 24, 3-37.
Bharat, J.A. and M. Kini, 1999, On Investment Banker Monitoring in the New Issue Market, Journal of
Banking & Finance 23, 49-84.
Bloomberg, 2010, China Overtakes Japan as Worlds Second-Biggest Economy, Available at: http://www.
bloomberg.com/news/2010-08-16/china-economy-passes-japan-s-in-second-quarter-capping-three-
decade-rise.html.
Borisova, G. and W.L. Megginson, 2011, Does Government Ownership Affect the Cost of Debt? Evidence
from Privatization, Review of Financial Studies 24, 2693-2737.
Brandt, L. and H. Li, 2003, Bank Discrimination in Transition Economies: Ideology, Information or
Incentives? Journal of Comparative Economics 31, 387-413.
Brau, J.C. and S. Fawcett, 2006, Initial Public Offerings: An Analysis of Theory and Practice, Journal of
Finance 61, 399-436.
Brau, J.C. and P.M. Johnson, 2009, Earnings Management in IPOs: Post-Engagement Third-Party Mitiga-
tion or Issuer Signaling? Advances in Accounting 25, 1-11.
Brav, A. and P. Gompers, 1997, Myth or Reality? The Long-Run Underperformance of Initial Public
Offerings: Evidence from Venture and Nonventure Capital-Backed Companies, Journal of Finance 52,
1791-1821.
Carter, R., F.H. Dark, and J.K. Singh, 1998, Underwriter Reputation, Initial Returns, and the Long-Run
Performance of IPO Stocks, Journal of Finance 53, 285-311.
Carter, R. and S. Manaster, 1990, Initial Public Offering and Underwriter Reputation, Journal of Finance
45, 1045-1068.
Chang, X., H.L. Shi, L. Tam, and G. Wong, 2010, The Cross-Sectional Determinants of Post-IPO Stock
Performance: Evidence from China, Accounting & Finance 50, 581-603.
Chen, C., Q. Jin, and H. Yuan, 2011,Agency Problems and Liquidity Premium: Evidence from Chinas
Stock Ownership Reform, International Review of Financial Analysis 20, 76-87.
Chen, C. and H. Xu, 2011, Ownership, Underwriting Quality, and Investment Bank Market Share: Evidence
from Chinese IPO Market, Fudan University Working Paper.
Chen, K. and H. Yuan, 2004, Earnings Management and Capital Resource Allocation: Evidence from
Chinas Accounting-Based Regulation of Rights Issues, Accounting Review 79, 645-665.
China Securities Regulatory Commission (CSRC), 2010, Security Market Monthly Data, Available at:
http://www.csrc.gov.cn/pub/newsite/sjtj/zqscyb/.
Hribar, P. and D.W. Collins, 2002, Errors in Estimating Accruals: Implications for Empirical Research,
Journal of Accounting Research 40, 105-134.
Dai, N., H. Jo, and J.D. Schatzberg, 2010, The Quality and Price of Investment Banks Service: Evidence
from the PIPE Market, Financial Management 39, 585-612.
Darrough, M. and S. Rangan, 2005, Do Insiders Manipulate Earnings When They Sell Their Shares in an
Initial Public Offering? Journal of Accounting Research 43, 1-33.
Dechow, P. and I. Dichev, 2002, The Quality of Accruals and Earnings: The Role of Accrual Estimation
Errors, Accounting Review 77, 35-59.
Dechow, P., S. Kothari, and R. Watts, 1998, The Relation Between Earnings and Cash Flows, Journal of
Dechow, P., R. Sloan, and A. Sweeney, 1995, Detecting Earnings Management, Accounting Review 70,
193-225.
Dong, M., J. Michel, and J.A. Pandes, 2011, Underwriter Quality and Long-Run IPO Performance,
Financial Management 40, 219-251.
DuCharme, L., P. Malatesta, and S. Sefcik, 2004, Earnings Management, Stock Issues and Shareholder
Lawsuits, Journal of Financial Economics 71, 27-49.
Dunbar, C.G., 2000, Factors Affecting Investment Bank Initial Public Offering Market Share, Journal of
Financial Economics 55, 3-41.
Fang, L.H., 2005, Investment Bank Reputation and the Price and Quality of Underwriting Services,
Journal of Finance 60, 2729-2761.
Gompers, P. and J. Lerner, 2003, The Really Long-Run Performance of Initial Public Offerings: The
Pre-Nasdaq Evidence, Journal of Finance 58, 1355-1392.
Guary, W., 2006, Discussion of the Role of Accruals in Asymmetrically Timely Gain and Loss Recognition,
Journal of Accounting Research 44, 1-13.
Hansen, R. and P. Torregrosa, 1992, Underwriter Compensation and Corporate Monitoring, Journal of
Finance 47, 1537-1555.
Ho, C.-C., C.L. Huang, C.T. Lin, and G.Y.C. Lin, 2010, Managing News Coverage Around Initial Public
Offerings, Financial Management 39, 187-225.
Huang, Y., 2003, Selling China: Foreign Direct Investment During the Reform Era, Cambridge, MA,
Cambridge University Press.
Jo, H., Y. Kim, and M.S. Park, 2007, Underwriter Choice and Earnings Management: Evidence from
Seasoned Equity Offerings, Review of Accounting Studies 12, 23-59.
Kao, J., D. Wu, and Z. Yang, 2009, Regulations, Earnings Management and Post-IPO Performance: The
Chinese Evidence, Journal of Banking & Finance 33, 63-76.
Karpoff, J.M., D.S. Lee, and G. Martin, 2008, The Cost to Firms of Cooking the Books, Journal of
Financial and Quantitative Analysis 43, 581-611.
Kim D., D. Palia, and A. Saunders, 2010, Are Initial Returns and Underwriting Spreads in Equity Issues
Complements or Substitutes? Financial Management 39, 1403-1423.
Kornai, J., 1993, The Evolution of Financial Discipline Under the Post-Socialist System, Kyklos 46,
315-336.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny, 1998, Law and Finance, Journal of Political
Economy 106, 1113-1155.
Lee, G. and R.W. Masulis, 2011, Do Underwriters or Venture Capitalists Restrain Earnings Management
by IPO Issuers? Journal of Corporate Finance 17, 982-1000.
Li, H. and L. Zhou, 2005, Political Turnover and Economic Performance: The Discipline Role of Personnel
Control in China, Journal of Public Economics 89, 1743-1762.
Li, Y., 2009, A Comparison of Chinas State-Owned Enterprises and Their Counterparts in the United
States: Performance and Regulatory Policy, Public Administration Review 69, 46-52.
Lin, X. and R. Germain, 2003, Organizational Structure, Context, Customer Orientation, and Perfor-
mance: Lessons from Chinese State-Owned Enterprises, Strategic Management Journal 24, 1131-
1151.
Liu, J., D. Nissim, and J. Thomas, 2002, Equity Valuation Using Multiples, Journal of Accounting Research
40, 135-172.
Liu, X. and J. Ritter, 2011, Local Underwriter Oligopolies and IPO Underpricing, Journal of Financial
Economics 102, 579-601.
Loughran, T. and J. Ritter, 2004, Why Has IPO Underpricing Changed Over Time? Financial Management
33, 5-37.
Megginson, W. and K. Weiss, 1991, Venture Capitalist Certification in Initial Public Offerings, Journal
of Finance 46, 879-904.
Michaely, R. and W.H. Shaw, 1994, The Pricing of Initial Public Offerings: Tests of Adverse Selection and
Signaling Theories, Review of Financial Studies 7, 279-319.
Qian, Y., 1994, A Theory of Shortage in Socialist Economies Based on the Soft Budget Constraint,
American Economic Review 84, 145-156.
Ritter, J., 1991, The Long-Run Performance of Initial Public Offerings, Journal of Finance 46, 3-27.
Teoh, S.H., I. Welch, and T.J. Wong, 1998a, Earnings Management and Long-Run Market Performance of
Initial Public Offerings, Journal of Finance 53, 1935-1974.
Teoh, S.H., T.J. Wong, and G.R. Rao, 1998b, Are Accrual During Initial Public Offerings Opportunistic?
Review of Accounting Studies 3, 175-208.
Titman, S. and B. Trueman, 1986, Information Quality and the Valuation of New Issues, Journal of
Wang, Q., T.J. Wong, and L. Xia, 2008, State Ownership, the Institutional Environment, and Auditor
Choice: Evidence from China, Journal of Accounting and Economics 46, 112-134.

Chen - 2013 - Underwriter Reputation Issuer Ownership and Pre-Ipo Earnings Management Evidence From China - FM

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Chen - 2013 - Underwriter Reputation Issuer Ownership and Pre-Ipo Earnings Management Evidence From China - FM

Hochgeladen von

Copyright:

Verfügbare Formate

Underwriter Reputation, Issuer

Ownership, and Pre-IPO Earnings

I. Related Literature and Hypothesis

A. Earnings Management and Underwriter Reputation during IPOs

II. Data and Variables

A. Sample and Data Sources

C. Measure of Earnings Management

Percentile 25% 50% 75% Largest Mean

Significant at 0.01 level.

TAC = Net income Cash flow from operations. (1)

TAC = 0 + 1 PPE + 2 (REV AR) + , (2)

Panel A. Descriptive Statistics of Variables

Panel B. Means of Variables by Year

Panel D. Distribution of Earnings Management

Significant at 0.01 level.

III. Model Specification and Empirical Results

Significant at 0.01 level.

UWREP = 0 + 1 UW GOV + Controls, (3)

DAC = 0 + 1 UWREP hat + Controls. (4)

Significant at 0.01 level.

A. Alternative Measures of Discretionary Accruals

TAC = 0 + 1 REV + 2 PPE + 3 ADJCFO + 4 DADJCFO

TAC = 0 + 1 REV + 2 PPE + 3 ADJCFO + . (6)

B. Signaling or Mitigation Hypothesis?

Significant at 0.01 level.

1st step: UWREP = + Instrument + Controls, (7)

2nd step :DAC = + UWREP hat + Controls, (8)

and 2) the signaling hypothesis model:

1st step: D AC = + Instrument + Controls, (9)

2nd step: UWREP =  + DAC hat + Controls. (10)

DAC = + Instrument + Controls.

In the second step, the following probit model is estimated:

UWREP =  + DAC hat + Controls.

Section A: SOE Section B: NSOE

Significant at 0.01 level.

C. Alternative Proxies of Underwriter Reputation

Measurement 1 Measurement 2 Measurement 3

Significant at 0.01 level.

Section A: SOE Section B: NSOE

Coefficient t-Value Coefficient t-Value

Significant at 0.01 level.

V. Test of Underwriter Reputation on Post-IPO Performance

POSTCAR = 0 + 1 UWREP + 2 DAC PRED + Controls. (11)

Panel A. Univariate Analysis of POSTCAR

Panel B. Multiple Variables Analysis of POSTCAR

Significant at 0.01 level.

Appendix A: IPO Process: The Role of Investment Banks (IBs) and

1. Hire IB for restructuring

4. Set up a new shareholding corporation

5. IB assists firm with regulatory compliance

6. CSRCs preliminary evaluation

7. IB prepares for registration statements

8. Sponsors due diligence and

10. CSRCs pre-evaluation

11. Revise and resubmit

12. CSRCs final approval

13. Public announcement

Appendix B: IPO process reforms in China

Das könnte Ihnen auch gefallen

2nd step: UWREP = + DAC hat + Controls. (10)

UWREP = + DAC hat + Controls.