The Value of Relationship-based and Market-based Contracting: Evidence from Corporate Scandals in China
Mingyi Hung
Leventhal School of Accounting Marshall School of Business
University of Southern California Los Angeles, CA 90089-0441
T.J. Wong
School of Accountancy The Chinese University of Hong Kong
Shatin, N.T., Hong Kong
Fang Zhang Department of Accountancy and Law
Hong Kong Baptist University Kowloon, Hong Kong
May 2012
Acknowledgments: The work described in this paper was fully supported by Grant No. CUHK453209 from the Research Grants Council of the Hong Kong Special Administrative Region, China. We thank Andreas Charitou, Mark DeFond, Zhaoyang Gu, Luzi Hail, Pierre Liang, Hai Lu, Joseph Piotroski, Tianyu Zhang, and workshop participants at Carnegie Mellon University, the Chinese University of Hong Kong, the University of Southern California, and the 35th Annual Congress of European Accounting Association for their helpful comments.
The Value of Relationship-based and Market-based Contracting: Evidence from Corporate Scandals in China
Abstract
This paper compares the value of relationship-based and market-based contracting in China by examining the consequence of corporate scandals, defined as enforcement actions against firms or their managers by Chinese courts and securities regulators. Since contracts are primarily conducted based on political relationships rather than market mechanisms in China, we hypothesize that scandals such as bribery of government officials that sever firms’ political ties are more damaging than scandals such as misrepresentation of financial statements that hurt firms’ market credibility. To test this hypothesis, we categorize 212 Chinese corporate scandals from 1997-2005 by whether the scandal primarily damages i) the firm’s political networks and hence its ability to conduct relationship-based contracting (relationship scandals), ii) the firm’s market credibility and thus its ability to conduct market-based contracting (market scandals), or iii) both (mixed scandals). Consistent with our hypothesis, we find that the stock market reacts more negatively to relationship and mixed scandals than to market scandals, and this result holds with or without controlling for the magnitude of scandals, average stock price and earnings performance prior to the scandals, and legal penalties imposed on the firms and individuals. In addition, firms involved with relationship and mixed scandals experience worse stock returns when they rely more on political networks. We also find that, compared to market scandals, relationship and mixed scandals lead to greater departure of political and affiliated directors and larger decreases in loans from state-owned banks.
1
The Value of Relationship-based and Market-based Contracting: Evidence from
Corporate Scandals in China
1. Introduction
Corporate misconduct is a key concern for investors and regulators worldwide. Prior
studies on corporate scandals generally focus on financial misrepresentation in market-
based economies such as the U.S. (Karpoff et al., 2008a; Keida and Philippon, 2009).
These studies find that U.S. firms experience huge losses in firm value if they are
targeted by SEC enforcement actions for financial misrepresentation, consistent with
credible accounting disclosure being critical for investors to conduct market-based arm’s
length contracting (Rajan and Zingales, 1998). The literature, however, provides little
insight on the impact of corporate scandals in relationship-based economies, such as most
emerging economies, where markets tend to be less developed. In these economies, firms
rely heavily on their owners’ and senior executives’ political and social networks to
conduct business. Thus, while corporate scandals in the U.S. destroy firm value largely
by damaging firms’ ability to conduct market-based contracting, corporate scandals in
emerging economies such as China can destroy firm value in two important ways: (1) by
shaking the market’s confidence, which hurts firms’ ability to conduct market-based
contracting, and (2) by severing firms’ political and social ties, which hurts firms’ ability
to conduct relationship-based contracting.
The purpose of this study is to provide evidence on the value of relationship-based
contracting and market-based contracting in China. We estimate the value of relationship-
based and market-based contracting using events that destroy firms’ abilities to conduct
2
future contracts (i.e., corporate scandals). To the best of our knowledge, this is the first
paper that compares the market reactions to the destruction of the two types of
contracting ability. While it is important to understand different contracting arrangements
of firms and their values, prior studies generally focus on documenting firm value decline
due to disruption in a specific type of contracting ability. Using U.S. data, Karpoff et al.
(2008a) document a significant share price decline from scandals that damage firms’
ability to engage in market-based contracting. Studies in East Asia also suggest that
severing political ties would lead to negative share price decline because weakened
political support from the government hurts firms’ ability to obtain implicit and explicit
contracts from the government (Fisman, 1998; Johnson and Mitton, 2003; Leuz and
Oberholzer-Gee, 2006).1 In addition, since firms within the same political network rely
critically on political connections for contracting, managers’ loss of political connections
also hurts their firms’ ability to engage in relationship contracting with other firms.
Focusing on corporate scandals in China offers a unique opportunity to compare the
value of relationship-based and market-based contracting for two reasons. First, as the
world’s second-largest economy, China provides a large variety of corporate scandals and
rich market depth for our empirical tests. Second, while the Chinese government has
introduced significant market reforms to facilitate market-based contracting over the past
three decades, relationship-based contracting via political networks remains crucial in
China due to strong government intervention in the corporate sector. We expect that in
China the value of relationship-based contracting is higher than that of market-based
1 Implicit contracts include government subsidies and tax breaks. In this study, we do not distinguish how
relationship and mixed scandals damage firms’ ability to engage in explicit versus implicit contracting.
3
contracting because contracts are conducted based on political relationships to a greater
extent than market mechanisms (Lin et al., 1996; DeFond et al.; 2000; Fan et al., 2008;
Wang et al., 2008; Hung et al., 2012). Thus, we hypothesize that scandals that sever
firms’ political ties are associated with greater losses in firm value than scandals that hurt
firms’ market credibility in China.
There are potential forces that act to attenuate the impact of scandals that sever firms’
political ties. Specifically, prior research suggests that while political ties enable firms to
receive significant economic benefits, they also impose substantial costs on firms. Fan et
al. (2007) find that strong government intervention can be costly to minority shareholders
(e.g. firms need to provide excessive employment and build infrastructure for the regions
they operate in). Corporate scandals are misconduct that breaks the government’s trust in
the firms involved. It is likely to significantly damage their ability to receive economic
benefits from relationship contracting but it is unlikely to free them from the strong
government intervention. To the extent that corporate scandals would reduce government
intervention to the firms as described in Fan et al. (2007), it would bias against our
results.
We test our hypothesis using a sample of 212 Chinese corporate scandals over the
1997-2005 period. As in prior studies (Karpoff et al., 2008a, b), we identify firms
engaged in misconduct using regulatory enforcement actions. Specifically, we identify
corporate scandals as enforcement actions against firms or their managers by Chinese
courts or securities regulators.2 These enforcement actions include not only financial
2 We include a firm in our sample if it has an enforcement action or investigation inquiry by the
government. Seven of our sample firms were cleared of wrongdoing at the end of investigation.
4
misrepresentation, but also asset misappropriation and bribery. Based on the type of
contracting ability these corporate scandals destroy, we classify them into one of three
categories: relationship scandals, market scandals, and mixed scandals.
We first identify 26 scandals as relationship scandals -- scandals that primarily
damage political networks and hurt firms’ ability to conduct relationship contracting.3
Examples of relationship scandals include managers bribing the government or stealing
from the state through tax evasion. These scandals do not necessarily harm a firm’s
outside shareholders or stakeholders (e.g., suppliers and customers), and in some bribery
cases may even help transfer government resources to the firm.4 However, while such
scandals may not hurt a firm’s ability to conduct market-based contracting, they damage
the firm’s political networks because the state will lose trust in the firm’s board and may
even arrest the government official that previously granted favors to the firm. This
disruption in political connections will reduce firms’ ability to engage in relationship-
based contracting.
Next, we identify 91 market scandals -- scandals that primarily damage market
confidence and hurt firms’ ability to conduct market-based contracting. One example of a
market scandal is financial misrepresentation, because accounting disclosure is critical
for outside stakeholders’ decisions in doing business with the firm. Another example is
3 We focus on political networks because prior research documents that politics plays a key role in
governance structure, auditor choice, IPOs, and debt financing of listed firms in China (Aharony et al., 2000; Fan et al., 2007, 2008; Wang et al., 2008). While social networks are also likely to be important for relationship contracting, we leave examination of social networks to future research.
4 Although these forms of misconduct, similar to all other scandals, may cause the market to question management integrity, their primary effect on the offending firm’s contracting ability is likely to come from direct offenses against the government rather than from direct offenses against outside shareholders and stakeholders.
5
managers misappropriating firm assets, for instance, through embezzlement, kickbacks,
or tunneling, because these types of actions effectively involve theft from shareholders.
Finally, we identify 95 mixed scandals, which are scandals that impair firms’ ability
to conduct both relationship-based and market-based contracting. An example of a mixed
scandal is embezzlement by managers of state-owned enterprises (SOEs). Embezzlement
by managers, that is, the theft of firm assets by managers, will create mistrust among
outside shareholders and stakeholders and hence limit the firm’s ability to conduct
market-based contracting. However, because embezzlement by SOE managers implies
theft from the government, it will also damage the firm’s political networks and hence its
ability to contract via these relationships.
Consistent with our hypothesis, we find that the stock market reacts more negatively
to relationship and mixed scandals than to market scandals. Specifically, while all types
of corporate scandals are associated with negative stock returns around the event date
(i.e., the first public disclosure of the scandal), the negative returns are more pronounced
for scandals that damage relationship-based contracting (i.e., relationship and mixed
scandals) than scandals that damage market-based contracting (i.e., market scandals). For
example, our univariate analysis finds that during a one-year event window (-6 months to
6 months, with month 0 being the event date), the average cumulative abnormal stock
return (CAR) is -30.8 percent for relationship scandals, -24.5 percent for mixed scandals,
and -8.8 percent for market scandals. In our multivariate analysis that further controls for
the magnitude of scandals, average stock returns and earnings performance during the
three years before the year of the scandals, and several other firm-level, provincial-level,
and industry-level variables, we continue to find that relationship and mixed scandals are
6
associated with worse stock returns than market scandals. These findings suggest that
loss of political networks is more damaging than loss of market credibility, consistent
with the view that political connections remain more important than market mechanisms
in conducting contracts in China.5
An important alternative explanation for our results is that the penalties of
relationship and mixed scandals may be more severe than the penalties of market
scandals. For example, relationship scandals such as bribery may be an excuse of one
clique eliminating a competing clique and therefore involve more severe legal sanctions
than market scandals. In addition, regulators may penalize firms more for cheating the
government than for cheating investors. Consequently, investors would react more
negatively when the scandal involves bribery of government officials or misappropriation
of state assets, thereby resulting in worse stock returns to relationship scandals and mixed
scandals than to market scandals. To address this concern, we perform additional analysis
after further controlling for monetary and non-monetary legal penalties imposed on firms
and individuals. This analysis finds results continue to support our hypothesis.6
We also perform several analyses to corroborate our primary findings. First, if
relationship and mixed scandals are associated with greater losses in firm value because
they damage firms’ relationship-based contracting ability, we expect relationship and
mixed scandals to result in greater loss in firm value among firms with strong
5 As reported in Section 6, the results are not sensitive to several robustness tests, including alternative event windows, alternative treatments of firms that have multiple scandals, alternative stock return measures, alternative treatments of delisted firms, excluding scandals enforced by stock exchanges, restricting firms to those with non-missing data on magnitude of scandals, and restricting firms to SOEs.
6 By controlling for individual penalties, we bias against finding our hypothesized results. This is because the severity of penalties to the offending manager could also proxy for the degree of loss in the firm’s political connections and its ability to engage in relationship-based contracting.
7
engagement in relationship-based contracting. To test this prediction, we repeat our
analysis after partitioning the scandals into strong and weak subsamples based on the
strength of a firm’s relationship-based contracting. Consistent with our prediction, we
find that firms involved with relationship and mixed scandals experience worse stock
returns when they have strong engagement in relationship-based contracting.
Additional analysis also finds that firms with relationship and mixed scandals
experience greater shock to its political networks as reflected in its board structure
subsequent to the event than firms with market scandals. We find that firms with
relationship and mixed scandals experience greater departure of top executives and
directors during the three years subsequent to the event. In addition, both relationship and
mixed scandal firms experience higher departure of political and affiliated directors.7 We
also find that relationship and mixed scandal firms appoint more new political directors,
suggesting that these firms exert more effort to rebuild their political networks. In
addition, while only mixed scandal firms experience a greater net loss of the number of
political directors, both relationship and mixed scandal firms experience a higher total
turnover (entry and exit) of political directors.
Finally, we find that firms with relationship and mixed scandals experience more
decreases in loans from state-owned banks than firms with market scandals. In addition,
relationship scandal firms have more difficulties in extending their existing loans during
the three years subsequent to the event. This result is consistent with our expectation that
7 Interestingly, our analysis does not find a significant difference in the turnover of independent directors
between firms with relationship and mixed scandals and firms with market scandals. This is consistent with the notion that independent directors are appointed for professional rather than political reasons.
8
relationship and mixed scandals damage political ties that are essential for Chinese listed
firms to obtain bank financing (Fan et al., 2008).8
The paper contributes to the literature in primarily two ways. First, our study is the
first to compare the price effects and other economic consequences of scandals that
damage relationship-based contracting versus scandals that damage market-based
contracting. Prior research on corporate scandals generally focuses on market-based
economies such as the U.S. and finds evidence of serious adverse price and economic
effects, especially for those scandals related to financial restatements (Karpoff et al.,
2008a, 2008b). Among the few studies that focus on corporate scandals in a relationship-
based economy, they generally find evidence of a significant negative price reaction and
management turnover to the scandals. For example, Chen et al. (2005) find that
enforcement actions of the CSRC (China Securities Regulatory Commission) lead to
negative market reactions and increased CEO turnover in China. Our study adds to this
literature by providing a comprehensive examination of the different market reactions to
and other economic consequences of corporate scandals that damage relationship-based
contracting and/or market-based contracting. In addition, to address alternative
explanations for our results, we perform analysis including various control variables such
as magnitude of the scandals, firm performance prior to the scandals, and the severity of
the penalties. By documenting significantly less severe consequences for scandals that
damage market-based contracting than for those that damage relationship-based
8 We do not find similar results using government subsidies, perhaps because subsidies are allocated
based on pre-determined procedures of the central and local governments.
9
contracting in an emerging economy such as China, we add to our understanding on the
importance of financial reporting quality and political networks for firms’ contracting
ability in such an economy.
Second, our paper complements prior studies on costs and benefits of political ties by
shedding light on whether and how the disruption of political ties by corporate scandals
impacts firm value. Fisman (1998), Johnson and Mitton (2003), and Leuz and
Oberholzer-Gee (2006) document evidence suggesting that political connections bring
economic benefits to firms. However, Fan et al. (2007) examine stock return performance
subsequent to IPOs in China and provide evidence suggesting that political ties can also
impose significant costs on the connected firms, and in equilibrium, the costs outweigh
the benefits. Our paper differs from Fan et al. by examining events related to the
destruction of political ties and documents that relationship scandals are associated with a
decline rather than rise in firm value. Our result suggests that while scandal firms likely
still bear the costs from excessive government control and intervention, they have lost
their expected benefits from political ties such as state loans or strong trusts from other
firms in future contracts.
The rest of this paper is organized as follows. Section 2 discusses prior literature and
hypothesis development. Section 3 describes the sample and the classification of the
scandals. Section 4 presents main empirical results. Section 5 reports additional analysis
and Section 6 discusses sensitivity analysis. Section 7 concludes the paper.
10
2. Prior literature and hypothesis development
2.1. Prior literature on corporate scandals
Corporate scandals are costly to investors and a key concern to regulators
worldwide.9 The vast majority of prior research on corporate scandals focuses on the
U.S. due to data availability. Because the U.S. is a market-based setting that relies
heavily on arms-length transactions, most of these studies investigate market-based
scandals involving financial misrepresentation. These corporate scandals have received
heightened attention in the wake of high-profile accounting frauds such as Enron. For
example, Karpoff et al. (2008a) find that U.S. firms experience huge losses in firm value
if they are involved in accounting manipulation. Karpoff et al. use a sample of U.S. SEC
enforcement actions for financial misrepresentation from 1978-2002 and document that
the market value of the scandal firms drops on average by 38 percent. Their evidence
suggests that the price decline is primarily a reputational penalty reflecting the firm’s
increased difficulty in contracting in the future. In addition, Desai et al. (2006) and
Karpoff et al. (2008b) find that senior management involved in the accounting scandals
are more likely to be dismissed from the firms.
Among the limited studies on corporate scandals in non-U.S. markets, most also
focus on market scandals related to financial misrepresentation. For example, Weber et
al. (2008) examine the stock and audit market effects associated with the accounting
scandal of ComROAD AG in Germany and find that the clients of KPMG (ComROAd’s
9 Various countries have passed regulations in response to corporate scandals, including the 1997 Foreign
Corrupt Practices Act and the 2002 Sarbanes-Oxley Act in the U.S., the 1998 corporate governance reform in Germany, the 2002 Code of Corporate Governance for Listed Companies in China, and the 2003 corporate law reform in Italy.
11
auditor) sustain significant negative abnormal returns during the event periods related to
the scandal. Zhang (2007) focuses on the price reaction to the announcement of
accounting scandals in China and finds evidence consistent with information spill-over
across firms in the same industry.
To the best of our knowledge, there has not been any study in the literature that
examines corporate scandals that affect firms’ relationship-based contracting. However, a
recent study by Fan et al. (2008), which documents corruption charges against
government officials and consequences for the firms connected to these officials, does
have implications for corporate scandals that involve the loss of political networks. More
specifically, while not a study of corporate scandals per se, Fan et al. examine the
changes in financing ability of Chinese listed firms that are connected with high-level
(mostly provincial) government bureaucrats suddenly charged with corruption. Fan et al.
find that firms lose their ability to raise bank debt when the government officials they
bribed or were connected with are charged with corruption. Extending Fan et al. and the
above prior studies on market scandals, we intend to use China market, where
relationship-based contracting is more prevalent than market-based contracting, and
examine if relationship scandals will have a more negative effect on firm value than that
of market-scandals.
2.2. Prior literature on political connections
Another stream of literature that is closely related to our study is the research on the
benefits of political connections. Prior studies find that politically experienced directors
are more prevalent among U.S. firms with greater reliance on government-related
revenue (Agrawal and Knoeber, 2001). In addition, politically connected directors are
12
widespread in countries with poor legal institutions (Faccio, 2006). The prevalence of
political connected firms is consistent with the theoretical and empirical work suggesting
that these firms can contract more easily with governments and thus receive significant
economic benefits, such as government subsidies, state loans, and tax breaks (Fisman,
1998; Johnson and Mitton, 2003; Leuz and Oberholzer-Gee, 2006). However, there is
evidence from the prior literature showing that close ties with government will damage
firm value as firms and managers rent seek to satisfy political and social objectives
(Shleifer and Vishny, 1994). For example, prior studies find that politically connected
firms tend to distort their lending decisions during election years, have weaker stock
performance, and exhibit poor earnings quality (Dinc, 2005; Fan et al., 2007; Chaney et
al, 2011).
Fisman (1998) is among the first that uses events that destroy these political
connections as a way to measure the value of these connections to the firms. Specifically,
he estimates the value of political connections among Indonesian firms using the stock
market reaction surrounding the news of President Suharto’s worsening health. Our study
extends Fisman’s approach by focusing on firm specific events i.e. corporate scandals
and by using different types of events (scandals) for comparing the value of political
connections versus market reputation in contracting.
2.3. Institutional background
China, the world’s second largest economy as of 2010, has become a major player in
international finance. Its three decades of economic reforms have also developed an
active capital market, with the total market capitalization of China’s domestically listed
firms (on the Shanghai and Shenzhen stock exchanges) being second only to that in the
13
U.S. as of 2009.10 However, despite this impressive economic growth, China is
commonly perceived as a country with weak legal institutions and strong government
control of the corporate sector (La Porta et al., 1998; Allen et al., 2005). For example,
the Heritage Foundation ranks China among the worst countries in terms of property
rights protection in 2011.11
During the rapid capital market development over the past three decades, listed firms
in China have often been accused of corporate misconduct such as financial
misrepresentation, bribery, and asset misappropriation (Aharony et al., 2000; Chen and
Yuan, 2004; Fan et al., 2008). In response, the government has launched a series of
market reforms to address these problems. For example, prompted by a string of
corporate scandals that emerged in 2001, the CSRC issued the Code of Corporate
Governance for Listed Companies in China in 2002, which expanded the rights of
minority shareholders, defined the duties of controlling shareholders and corporate
boards, and increased information disclosure and transparency requirements.12
However, despite the significant market reform to facilitate market-based contracting,
political networks are still prevalent and crucial to conduct relationship-based contracting
10 See “China’s stock market: Another great leap” (The Economist, August 29, 2009). According to the
World Federation of Exchanges, as of 2009 the top three countries in terms of total market capitalization of firms listed in main domestic stock exchanges are: (1) the U.S. with US$ 15 trillion, (2) China with US$ 3.6 trillion, and (3) Japan with US$ 3.3 trillion.
11 Specifically, the 2011 index of economic freedom by the Heritage Foundation ranks China 135th out of 179 countries. China receives a score of 20 (out of 100) for property rights, based on the following criteria: “20-Private property is weakly protected. The court system is so inefficient and corrupt that outside settlement and arbitration is the norm. Property rights are difficult to enforce. Judicial corruption is extensive. Expropriation is common.”
12 See Shi and Weisert (2002) for a summary discussion of these cases and the regulatory reform. The scandals also received highlighted media coverage, including the following: “China shares down at noon, hit by Guangxia scandal” (Reuters News, September 5, 2001) and “Charges are filed in landmark Chinese fraud case – Three former executives of a listed store operator accused of hiding losses” (The Wall Street Journal, August 7, 2002).
14
in China. Due to the strong government intervention in the corporate sector, explicit and
implicit contracts based on political networks – such as government contracts, state loans,
listing rights, equity issuances, government subsidies, operational rights, and other
privileges (Faccio, 2006; Faccio et al., 2006) – are widespread. In addition, a substantial
portion of listed Chinese firms are SOEs in which the government has influence over the
appointment of key executives and external auditors, and the firm’s ability to obtain state
subsidies and loans (DeFond et al., 2000; Fan et al., 2008; Wang et al., 2008; Hung et al.,
2012). Even non-state firms need to build connections with the government in order to
obtain favors (Lin et al., 1996). Consequently, Chinese firms’ ability to obtain
government contracts depends critically on their political ties rather than merits.
Furthermore, in the absence of strong legal and market institutions, most contracts are
conducted privately through relationships such as personal ties and internal
communications within political networks, rather than market mechanisms using legal
procedures and public disclosures.
2.4. Hypothesis
Based on the above arguments, we expect the value of relationship-based contracting
to be higher than that of market-based contracting in China. This is because in China
contracts are primarily conducted based on political relationships rather than market
mechanisms. Since relationships within the political networks are more essential for
contracting than legal protection and accounting transparency, we predict that scandals
that damage relationship-based contracting by severing political ties are more detrimental
to firms than scandals that damage market-based contracting by impairing market
credibility in China. This leads to the following hypothesis:
15
Hypothesis: Corporate scandals that damage relationship-based contracting are
associated with greater losses in firm value than corporate scandals that damage market-
based contracting.
3. Sample and classification of corporate scandals
3.1. Sample
Our sample includes firms with enforcement actions against their Chairman/CEO by
Chinese courts and firms with enforcement actions for financial misrepresentation by
securities regulators. We begin our investigation period in 1997 because prior to this
period the regulatory disclosure and media coverage of Chinese listed firms was
relatively poor.13 We compile our sample firms and event dates as follows: First, we
identify firms with enforcement actions against their Chairman and/or CEO by Chinese
local and central courts via news searches. We obtain the key event dates for these
scandals from various sources, including 21st Century Business Herald (for news
coverage from 2001-2005) and online search engines such as www.google.com,
www.baidu.com, and http://cn.yahoo.com/ (for news coverage prior to 2001). Second, we
identify firms with enforcement actions for financial misrepresentation by the CSRC and
stock exchanges using data sources from China Security Market and Accounting
Research (CSMAR), China Center for Economic Research (CCER), websites of the
13 The Shenzhen and Shanghai stock exchanges were set up in 1990 and 1991, respectively. The
increased media coverage of Chinese firms in 1996 was partly due to the surge in stock prices during that year, which is often referred to as the ‘1996 Oddity’ (“China Stock market in a global perspective,” Dow Jones Indexes, September 2002).
16
CSRC and stock exchanges, and firms’ annual reports and public announcements.14 We
obtain the key event dates for these firms from the following sources: public
announcements by the listed firms, monthly bulletins by the CSRC, public
announcements by the Shanghai and Shenzhen stock exchanges, and news reports from
The China Securities Journal, Securities Times, Shanghai Securities News, and other
major business and finance websites in China. 15 We next obtain stock returns and
financial data from the CSMAR database, and CEO and director profiles from the WIND
financial database and companies’ annual reports.
Our initial sample consists of 340 firms. For firms with multiple scandals, we keep
the most recent ones (deleting 100 prior cases) so our test comprises distinct firms. While
using the most recent scandal will likely bias against finding significant market reactions
to the scandal (because the market already reacted to the earlier scandal), it ensures that
our investigation of governance and financing changes subsequent to the scandal is not
confounded by additional scandals. In addition, we delete 24 firms that are subsequently
delisted because such firms do not have governance and financing data in periods
subsequent to the scandal. While excluding delisted firms is also likely to underestimate
14 We also cross check the data with the list of accounting scandals used in Zhang (2007). We thank
Zhang Peng for sharing his data. In addition to financial misrepresentation (i.e., accounting manipulation and false disclosure of financial statement items), the enforcement actions of the CRSC and stock exchanges also include charges related to various other securities violations such as delayed disclosure, market manipulation, or misleading forecasts. To ensure that our sample consists of non-trivial scandals that damage market-based contracting mechanisms, we include only financial misrepresentation because this type of scandal is the common focus of prior studies and most likely to damage firms’ reporting credibility.
15 We note that Karpoff et al. (2008a) use trigger events (such as self-disclosures of malfeasance, restatements, auditor departures, and unusual trading) as the event date of their accounting frauds sample, and that such dates usually precede the U.S. SEC investigation inquiry. We use the announcement of the investigation inquiry as our event date for the scandals involving accounting manipulation because the trigger events are rare in China. Furthermore, the information media in China is not as well developed as that in the U.S. The announcement of the CSRC or stock exchanges generally is the main public information source on enforcement actions for accounting scandals.
17
market reactions to scandals, this approach ensures that there is no major systematic
difference between the sample used for our market reactions analysis and the sample used
for our governance/financing changes analysis.16 Finally, we delete four firms that do not
have stock return data in CSMAR. These selection criteria result in a final sample of 212
firms.
3.2. Classification of scandals and sample distribution
Table 1, Panel A lists the key types of scandals that damage firms’ ability to conduct
relationship-based contracting and market-based contracting. Based on this breakdown,
Table 1, Panel B then summarizes the classification of our sample scandals based on the
type of contracting ability they destroy: (1) relationship scandals – scandals that primarily
damage relationship-based contracting but not market-based contracting, (2) mixed
scandals – scandals that damage both relationship-based and market-based contracting,
and (3) market scandals – scandals that primarily damage market-based contracting but
not relationship-based contracting.
Panel A of Table 1 shows that the major types of scandals that damage firms’ ability
to conduct relationship-based contracting are managers bribing government officials (R1)
and managers misappropriating state assets (R2), with R2 divided into three subgroups:
tax evasion (R2a), managers of SOEs misappropriating firm assets (R2b), and managers
of non-SOEs misappropriating firm assets in which the government has a minority stake
(R2c). The R1/R2 scandals potentially damage the firm’s political networks in two ways.
16 As reported in Section 6, we also perform sensitivity tests after treating multiple scandals as separate
events, keeping only the earliest event for firms with multiple scandals (and deleting all subsequent cases), and adding back delisted firms. Results from these analyses are consistent with those reported in Table 4.
18
First, the government will lose trust in the firm’s managers and board. Even if the firm
replaces the entire management team, it still has to spend time and resources to rebuild its
political networks. Second, government officials that accept bribes will also be
implicated and hence no longer able to grant favors to the firm.
It is important to note that scandals that damage relationship-based contracting (i.e.,
R1/R2 scandals) do not necessarily harm a firm’s outside shareholders or stakeholders
(e.g., suppliers or customers). Bribing government officials (R1) to acquire equity
issuance rights or obtain government contracts (implicit and explicit) may help channel
resources into the firm. In addition, while tax evasion (R2a) may reduce government
revenues, it will not directly hurt a firm’s outside shareholders or stakeholders. Only
misappropriation of firm assets by managers of SOEs (R2b) or by managers of non-SOEs
in which the government holds a minority stake (R2c) will directly hurt a firm’s outside
shareholders and stakeholders, and thus damage not only firms’ ability to conduct
relationship-based contracting but also market-based contracting.
With regards to market-based contracting, Panel A of Table 1 shows that the major
types of scandals that damage firms’ ability to conduct such contracting are financial
misrepresentation (M1) and misappropriation of firm assets (M2), with M2 divided into
three subgroups: managers of non-SOEs misappropriating firm assets in which the
government has no ownership (M2a), managers of SOEs misappropriating firm assets
(M2b), and managers of non-SOEs misappropriating firm assets in which the government
has a minority stake (M2c). The M1/M2 scandals damage firms’ ability to conduct
market-based contracting in two ways. First, because accounting disclosure is critical for
outside investors and other stakeholders to make business decisions and enforce
19
contracts, misrepresentation negatively affects firms’ contracting ability with these
market participants. Second, because of loss of valuable assets (e.g., through tunneling
firm assets) or sub-optimal investment decisions (e.g., in exchange for kickbacks),
misappropriation negatively impacts the wealth of shareholders and stakeholders.
Note that scandals that damage firms’ ability to conduct market-based contracting
(i.e., M1/M2 scandals) do not necessarily harm the firms’ ability to engage in
relationship-based contracting. Financial misrepresentation (M1) and asset
misappropriation by managers of non-SOEs in which the government has no ownership
(M2a) mainly affect market-based contracting. As pointed out in our discussion of R2b
and R2c above, only the misappropriation of firm assets by managers of SOEs (M2b) or
by managers of non-SOEs in which the government holds a minority stake (M2c) will
damage firms’ market-based as well as relationship-based contracting ability.
Turning to Panel B of Table 1, which presents the classification of our sample
scandals based on the type of contracting ability they destroy, we regard misconduct that
is a direct offense against the government as primarily damaging relationship-based
contracting, while misconduct that is a direct offense against outside shareholders and
stakeholders as primarily damaging market-based contracting. Thus, while bribery of
government officials (R1) and tax evasion (R2a) in relationship scandals raise doubts in
the market about management integrity, we do not classify them as mixed scandals
because they are not a direct offense against outside shareholders and stakeholders. Panel
B shows that among our 212 sample firms, 26 are relationship scandal firms, 95 are
mixed scandal firms, and 91 are market scandal firms. The panel also shows that the most
common type of relationship scandal, mixed scandal, and market scandal is managers
20
bribing government officials (R1, 24 firms), managers of SOEs misappropriating firm
assets (R2b=M2b, 81 firms), and misrepresentation of financial statements (M1, 58
firms), respectively. Appendix A provides examples of relationship, mixed, and market
scandals in our sample.
Table 2 presents the sample distribution. Panel A of the table reports the sample
distribution by year and type of scandal. The table shows an increasing trend in the
number of scandals, which likely reflects greater regulatory oversight in the later period.
For example, the sharp increase in the number of mixed and market scandals in 2002 is
likely due to increased CSRC enforcement actions in response to several high-profile
scandals in 2001 (Shi and Weisert, 2002). Panel B of Table 2 presents the sample
distribution by industry and type of scandal. The table shows that the manufacturing
sector, the biggest sector in Chinese economy, also has the largest number of corporate
scandals.
4. Descriptive statistics and empirical results
4.1. Univariate analysis
We employ an event study methodology to test market reactions to corporate
scandals, with the event date defined as the first public disclosure of the scandal. We
measure market reactions using cumulative abnormal returns (CARs), calculated as stock
returns minus returns of the market index on the listing stock exchange during a specified
event window. For scandals with enforcement actions against Chairmen/CEOs by courts,
we identify the event date as the date in which the press or the firm reports that the
executive is arrested or brought in for questioning (‘ShuangGui’), whichever is earlier.
21
For scandals with enforcement actions by securities regulators, we identify the event date
as the date in which the securities regulators or the firm announce the investigation
inquiry, whichever is earlier. Figure 1 presents the timeline of the events.
While most event studies use a short event window (usually two or three days), we
use relatively long event windows (from two months up to two years) for three reasons.
First, information leakage is severe in China, especially for charges against
Chairmen/CEOs. For example, if a bureaucrat is arrested for accepting bribes from a
firm, the market will expect an enforcement action against the executive of the bribing
firm. The firm’s stock price may therefore already incorporate this information prior to
the first public disclosure of the executive’s bribery charges. Second, if a firm is
temporarily suspended for trading subsequent to the disclosure of the scandal, short event
windows will fail to pick up the full price impact of the scandal.17 Third, while we
identify the event date as the first public disclosure of the scandal, the date is generally
either the date when the executive’s arrest is reported or the date when an investigation
inquiry by securities regulators is announced. A long event window will ensure that our
results are not driven by the different nature of these event dates.
Table 3 presents the market reactions to corporate scandals during various event
windows. While we rely on long event windows (from two months up) to draw our
conclusions, we report event windows that range from two days to two years surrounding
17 According to Article 157 of the Company Law, the CSRC or Chinese stock exchanges may decide to
suspend the trading of stocks for the following reasons: (1) the company's share capital level is below the listing requirement; (2) the company has failed to comply with regulations for public disclosure of its financial situation or has falsified its financial accounting statements; (3) the company is involved in major illegal activities; and (4) the company has incurred losses for the past three consecutive years. Stock trading was suspended for four to six months for eight firms in our sample.
22
the event date for completeness. Columns 2, 3, and 4 report mean CARs for relationship,
mixed, and market scandals. These columns show that all types of corporate scandals are
associated with negative stock returns during all event windows. Columns 5 and 6 of the
panel report the differences in market reactions between scandals that damage
relationship-based contracting (relationship and mixed scandals) and the benchmark
market scandals. Consistent with our prediction, these columns show that the negative
stock returns are more pronounced for scandals that damage relationship-based
contracting than those that damage market-based contracting from the two-month
window up to the two-year window. For example, looking at the one-year event window
(-6 months to 6 months, with month 0 being the event date), we find that the average
CAR is -30.8 percent for relationship scandals and -24.5 percent for mixed scandals, but
only -8.8 percent for market scandals.
We note that we do not have predictions on the difference in market reactions
between relationship and mixed scandals, the two types of scandals that damage
relationship-based contracting. While mixed scandals may be more damaging because
they destroy both relationship-based and market-based contracting ability, relationship
scandals can be more damaging because they typically also involve charges against
government officials overseeing the firm. As shown in Panel B of Table 1, 24 out of the
26 relationship scandals relate to managers bribing government officials. In contrast, only
one of the mixed scandals involves charges against government officials.18 Since the
arrest of a connected government official has a direct negative effect on firms’ political
18 In this case, the manager was charged with both bribing a government official (damages relationship-
based contracting) and accounting manipulation to conceal the bribe (damages market-based contracting).
23
networks, relationship scandals can be as damaging as if not more damaging than mixed
scandals.
Figure 2 plots the average CAR for each type of scandal from one year before the
event date to one year after. Consistent with our results in Table 3, the figure shows that
all three types of scandals are associated with negative CARs during the two-year period
surrounding the event date. In addition, the decline in firm value is most pronounced for
relationship scandals and least pronounced for market scandals.
4.2. Hypothesis test
We test our hypothesis by regressing CARs during event windows from two months
up to two years on a dummy variable indicating relationship scandals, a dummy variable
indicating mixed scandals, and several control variables. Our model includes the
magnitude of the scandal to control for the severity of the scandal. In addition, we control
for the following firm characteristics, measured prior to the scandal, that may be
associated with market reactions to corporate scandals: firm size, market-to-book, asset
tangibility, profitability, stock returns, and a dummy variable indicating whether a firm’s
majority shareholder is the government. We also include a variable controlling for a
firm’s provincial legal environment and variables indicating industry membership to
control for industry fixed effects.
Our regression model is as follows:
CAR = β0 + β1(Relationship scandal) + β2(Mixed scandal) +β3(Magnitude of scandal)+ β4(Firm size_pre)+β5(Market-to-book_pre) + β6(Tangibility_pre) + β7(Stock return_pre) +β8(ROA_pre) +β9(SOE) + β10(Legal environment) + βm(DIndustry) + ε (1)
See Appendix B for variable definitions.
24
Our hypothesis predicts β1 and β2 to be negative.
Table 4 reports the results of this analysis. Panel A of the table presents descriptive
statistics on the variables used in this analysis. Consistent with our classification that
market scandals include financial misrepresentation that materially misstate financial
statements, the panel shows that magnitudes of scandals are higher for market scandals
than for relationship and mixed scandals. For example, the average magnitude of scandals
is 15.72% for market scandals, versus 3.97% and 3.99% for relationship and mixed
scandals.19 The panel also shows that firms involved with market scandals have lower
profitability prior to the scandals than firms involved with relationship and mixed
scandals. For example, the average ROA_pre is -0.9% for firms involved with market
scandals, versus 3.9% and 1.4% for firms involved with relationship and mixed scandals.
Panel B of Table 4 presents the regression results. Consistent with our hypothesis, the
panel reports that the coefficient on the dummy indicating relationship scandals and the
coefficient on the dummy indicating mixed scandals are both significantly negative at p ≤
10% (two-tailed) in all models. Overall, this result indicates that relationship and mixed
scandals have worse market reaction than market scandals, suggesting that scandals
damaging a firm’s relationship-based contracting ability result in greater losses in firm
value than scandals damaging a firm’s market-based contracting ability.20
19 Among 212 scandals, we are unable to find information on the magnitudes of scandals for 49 firms
(23%). We assume that the magnitudes of scandals for these firms are small and assign a value of zero. To the extent that this assumption introduces measurement errors, we also perform a sensitivity test excluding firms with missing magnitude of scandals in Section 6. This analysis finds our results remain qualitatively the same.
20 We also perform a sensitivity test in which we include a dummy variable indicating a client of a Big Four (or Big Five, before the demise of Arthur Andersen) auditor in our analysis in Table 4. We do not use a dummy variable indicating a Big Ten auditor as in other China studies such as Wang et al. (2008) because the top six to ten auditor rankings changes during our sample period, which makes it difficult to use as a
25
4.3. Analysis further controlling for legal penalties
While the results from our hypothesis test are consistent with our argument that
relationship-based contracting is more valuable than market-based contracting in China,
an alternative explanation for our results is that relationship and mixed scandals are
associated with more severe legal penalties than market scandals. For example,
relationship scandals such as bribery may be an excuse of one clique eliminating a
competing clique and therefore involve more severe legal sanctions than market scandals.
In addition, regulators may impose larger penalties on firms for cheating the government
than for cheating investors. Becker and Landes (1974) suggest that a rational individual
will weigh up the expected benefit from committing the crime against the expected cost.
If scandals involve bribery of government officials or misappropriation of state assets are
punished more severely, investors would expect that these scandals would be more
serious. Consequently, the stock market would react more negatively to relationship
scandals and mixed scandals than to market scandals. While our hypothesis test includes
the magnitude of the scandal to control for the severity of the scandal, we also perform
analysis further controlling for legal sanctions against firms and individuals involved
with the scandals. We assume here that investors can foresee the outcomes even though
the legal proceedings typically take years and often fall beyond our event windows.
Panels A and B of Table 5 reports legal penalties on firms and individuals,
respectively. Each panel presents administrative sanctions and criminal sanctions across
control variable. The results (untabulated) are qualitatively similar to those reported in Table 4. In addition, the coefficient on the dummy variable indicating Big Four/Five auditor client is insignificant at conventional levels in all regression models.
26
the three types of scandals.21 Panel A shows that for administrative sanctions on firms,
monetary penalties are generally trivial and non-monetary penalties range from minor
criticism to trading and operating suspension. The panel also shows that for criminal
sanctions on firms, monetary penalties are greater for relationship and mixed scandals
than for market scandals. For example, the average criminal monetary penalty on firms is
$141.8 US million for relationship scandals, $17.1 US million for mixed scandals, and 0
for market scandals. Panel B shows that for legal sanctions on individuals, mixed
scandals generally are associated with the most severe penalties while market scandals
are associated with the least severe penalties. For example, 49 of the mixed scandals
involve jail time, versus 10 and 1 for relationship and market scandals.22
We also perform analysis to assess the average total dollar loss attributable to loss in
contracting abilities for each type of scandal. We measure total dollar loss in contracting
ability during each event window as the total dollar loss during the event window for all
firms minus the total monetary penalties on firms and the total valuation effect of
accounting readjustments. We calculate a firm’s dollar loss during each event window as
its CAR during the event window multiplied by its market capitalization prior to the
event window. Following Karpoff et al. (2008a), we estimate the valuation effect of
accounting readjustment as the book value of the write-off multiplied by industry median
market-to-book ratios prior to the scandal.
21 We do not include civil sanctions because civil litigation against corporate misconduct is difficult and
extremely rare in China. 22 While not reported in the table, seven of our sample firms were cleared of wrongdoing at the end of
investigation.
27
Panel C of Table 5 presents the results of this analysis. The top row of the panel
reports the total monetary penalties on firms and the second row reports the total
valuation effect of accounting readjustments. The remaining rows in Panel C report for
each event window, the total dollar loss, the total dollar loss in contracting ability, and the
average dollar loss in contracting ability for each event window. The panel indicates that
the average dollar loss in contracting ability is the highest for relationship scandals and
the lowest for market scandals. This finding is consistent with our hypothesis that
scandals damaging relationship-based contracting are more detrimental to firms than
scandals damaging market-based contracting.
Panel D of Table 5 reports results re-estimating equation (1) after further controlling
for both monetary and non-monetary penalties imposed on firms and individuals. We
note that by controlling for individual penalties, we potentially bias against finding our
hypothesized results. This is because we expect that relationship and mixed scandals hurt
a firm’s contracting ability through the political persecution against their managers. That
is, relationship or mixed scandals damage the contracting ability of a firm by destroying
its political connections possessed by its managers. Nonetheless, the panel reports that the
coefficient on the dummy indicating relationship scandals and the coefficient on the
dummy indicating mixed scandals remain significantly negative at p ≤ 10% (two-tailed)
in all models. Thus, the analysis after controlling for legal penalties continues to find
results supporting our hypothesis.
Panel D of Table 5 also shows that while the signs of the coefficients on legal
penalties are generally negative, most of them are insignificant at conventional levels
except for the monetary criminal penalties on firms, trading suspensions, criticism on
28
individuals, and restriction of employment in securities markets. While it is somewhat
surprising that the coefficient on the dummy variables indicating imprisonment and death
penalty are insignificant, this is likely due to that those penalties typically involve lengthy
legal proceedings that last several years and are difficult to predict during our event
windows.
5. Additional analysis
5.1. Market reaction to corporate scandals partitioned by the extent of a firm’s
engagement in relationship-based contracting
The results from our hypothesis test suggest that loss of government connections
leads to greater decreases in share prices than loss of market confidence on reported
numbers in China. To gain further evidence on this inference, we examine the association
between market reactions and relationship scandals partitioned by the strength of a firm’s
engagement in relationship-based contracting. We expect that relationship scandals will
result in greater losses in firm value among firms with stronger relationship-based
contracting.
To test this prediction, we first create a binary measure for each of the five proxies
that potentially capture the importance of relationship-based contracting for a firm, where
a value of one indicates strong relationship-based contracting, and zero otherwise. The
five proxies and their associated binary variables are as follows:
Political connection of Chairman/CEO. Following prior studies (Fan et al., 2007), we
define an executive as politically connected if he/she is a current or former officer of
29
the central government, a local government, or the military. The binary variable equals
one if a firm’s Chairman or CEO is politically connected, and zero otherwise.
Political connection of corporate board. We define a board member as politically
connected if he/she is a current or former officer of the central government, a local
government, or the military. The binary variable equals one if the percentage of
politically connected board members for a firm is greater than the sample firm-level
median, and zero otherwise.
Loan from state-owned banks. Bank loans are a major financing source in China and
generally consist of state loans that require strong political connections in order to be
approved (Fan et al. 2008). The binary variable equals one if a firm’s loans from
state-owned banks divided by total assets is above the sample firm-level median, and
zero otherwise.
Government subsidy. Prior studies show that firms can benefit financially through
political ties in the form of government subsidies (Faccio et al., 2006). The binary
variable equals one if a firm’s government subsidies divided by total assets is above
the sample firm-level median, and zero otherwise.
Legal environment. Rajan and Zingales (1998) argue that relationship-based contracting
is more prevalent in less developed environments. We capture the development level
of a firm’s province using the legal index in the 2005 National Economic Research
Institute (NERI) Index of Marketization of China’s provinces. The index is based on
the average of the following three components (after normalizing each component to a
range of 0-10): (1) the number of lawyers as a percentage of the province’s
population; (2) the efficiency of local courts, as captured by the percentage of lawsuits
30
pursued by the courts; and (3) the extent of property rights protection, as captured by
the number of patents granted per research and development personnel. The binary
variable equals one if a firm’s legal index is equal to or below the sample province-
level median, and zero otherwise.
After coding each of the five binary variables as described above, we create a
summary measure for each firm using the sum of these five binary variables. Finally,
based on the summary measure, we classify firms with a score above the sample firm-
level median as having strong relationship-based contracting, and firms with a score
equal to or below the median as having weak relationship-based contracting.
Table 6 presents the results of this analysis. Panel A of the table reports descriptive
statistics for the proxies used to capture the strength of relationship-based contracting.
Panel B reports the results for the partitioning analysis. We find that the coefficients on
the dummies indicating relationship scandals and mixed scandals are significantly
negative at p ≤ 5% (two-tailed) among the subsample of firms with strong relationship-
based contracting in all event windows, but are generally insignificant at conventional
levels among the subsample of firms with weak relationship-based contracting. In
addition, the difference in the coefficient on the dummy indicating relationship scandals
between the strong and the weak subsamples is significant at p ≤ 5% (two-tailed) in all
event windows. The difference in the coefficient on the dummy indicating mixed
scandals between the strong and the weak subsamples is significant at p ≤ 5% (two-
tailed) in all event windows except for the longest window. Overall, this analysis
corroborates our primary results by finding that firms involved with relationship and
31
mixed scandals experience worse stock returns when they are more engaged in
relationship-based contracting.
5.2. Impact of scandals on board structure
The results from our hypothesis test suggest that compared to market scandals,
relationship and mixed scandals are more damaging in China because they destroy firms’
political networks. To provide further corroborating evidence, we examine the impact of
relationship and mixed scandals on firms’ board structures. We expect firms with
relationship and mixed scandals to experience a greater shock to their political networks
as reflected by changes in their board structures subsequent to the event.
To test the impact of scandals on board structures due to loss of political networks,
we begin our analysis by examining departure of all directors as well as departure of the
following subgroups of directors: Chairman or CEO, political directors (i.e., directors
who are politically connected), and affiliated directors (i.e., directors who have personal
affiliations with the Chairman or CEO during the event year). We define personal
affiliations as being a relative, laoxiang, former classmate in college, or former colleague
in previous employment.23 We expect firms with relationship and mixed scandals to
experience more departure among these groups of directors. For completeness, we also
examine the departure of independent directors (i.e., non-executive directors without
business affiliations with the firm) although we do not have a prediction for this group of
directors. While prior U.S. studies show that independent directors are more likely to be
23 Laoxiang means individuals from the same hometown. Laoxiang is an important type of relationship in
China because people tend to take care of their laoxiang that are outside their hometown. We treat a director as a liaxiang of the Chairman or CEO if the director came from the same city (市), township (乡), county (县), or town (镇) and the location of the company is outside their hometown.
32
terminated subsequent to accounting restatements (Srinivasan, 2005), we do not expect
relationship and mixed scandals to be differentially harmful to these directors than market
scandals because independent directors are appointed for professional rather than political
reasons.
To test our prediction we regress departure of directors on a dummy variable
indicating relationship scandals, a dummy variable indicating mixed scandals, and several
control variables. We measure departure of directors as the percentage of directors
leaving the firm during the three years subsequent to the event. In addition, our regression
models include various control variables that are likely to affect changes in firms’
governance structures. First, we control for changes in firm characteristics: firm size,
market-to-book assets, asset tangibility, and profitability. We measure the changes in the
average values of these variables from the three years before to three years after the event
(excluding the event year). Second, we include a dummy variable indicating SOE (Fan et
al., 2008). Third, we control for a firm’s legal environment because prior studies suggest
that legal institutions affect governance (La Porta et al., 1998). Fourth, we include
dummy variables indicating industry membership and years to control for industry and
year fixed effects. We winsorize all scaled variables at the top and bottom 1% of their
distributions. Our regression model is as follows:
Departure of directors =β0 + β1(Relationship scandal) + β2(Mixed scandal) + β3(Magnitude of scandal) + β4(∆Firm size) +
β5(∆Market-to-book) + β6(∆Tangibility) +β7(∆ROA) + β8(SOE) + β9(Legal environment)+ βm(DIndustry)+βn(DYear)+ε (2) See Appendix B for variable definitions.
33
Table 7 reports the results of this analysis. Panel A of the table presents descriptive
statistics on the additional variables used in this analysis. The panel shows that on
average, 48 percent of directors depart the scandal firms in the three years subsequent to
the event. This finding suggests that corporate scandals in China are associated with great
disruption in firms’ governance structures. Panel B of Table 5 presents the results on the
impact of relationship or mixed scandals on board structures. We find that relative to
market scandal firms, relationship and mixed scandal firms have higher departure rates
among top executives (Chairman or CEO) and all directors during the three years
subsequent to the event. We also find that both relationship and mixed scandal firms
experience higher departure of political and affiliated directors, although only mixed
scandal firms experience higher departure of political Chairman/CEO. Interestingly, we
do not find a significant difference in the departure of independent directors between
firms with relationship and mixed scandals and firms with market scandals, consistent
with the notion that these directors are typically appointed for professional rather than
political reasons.24
In addition, we perform analysis examining the appointment of new political directors
and the overall turnover and net loss of these directors. For completeness, we also report
results for the appointment of political Chairman/CEO and the overall turnover and net
loss of political Chairman/CEO. We expect firms with relationship and mixed scandals
24 We also perform sensitivity analysis in which we further control for Chairman/CEO departure in the models using non-Chairman/CEO departure as the dependent variable. The results (untabulated) are generally consistent with those reported in Panel B of Table 7. Specifically, we find that the coefficient on the dummy variables indicating relationship scandals and mixed scandals continue to be significantly positive at p ≤ 5% (two-tailed) in models using departure of directors, departure of political directors, and departure of affiliated directors as the dependent variable, with one exception – the coefficient on the dummy indicating relationship scandals becomes insignificant in the model using departure of directors as the dependent variable.
34
to exert more efforts to recruit political directors to repair their political networks. Panel
C of Table 7 presents the results on the political realignment subsequent to the scandals
among political Chairmen/CEOs and among political directors. Consistent with our
expectation, the results show that relationship and mixed scandal firms appoint more new
political directors. While only mixed scandal firms experience a greater net loss of
political directors, both relationship and mixed scandal firms experience an overall higher
turnover (entry and exit) of political directors. The results are weaker for political
Chairmen/CEOs, likely because the changes in political Chairman/CEOs are relatively
unusual and require longer planning and selection processes. Overall, these findings
corroborate the results from our hypothesis test that relationship and mixed scandals
experience a greater shock to the political networks via their boards subsequent to the
event.
5.3. Impact of scandals on loans from state-owned banks
In addition to changes in board structures, we expect another important consequence
of the scandals is the loss of financing from the state. Specifically, a major financing
source for Chinese firms is loans from state-owned banks (Chen et al., 2010). If
relationship and mixed scandals result in greater losses of political ties, we expect firms
involved with these scandals to experience larger decreases in loans from state-owned
banks (Faccio, 2006; Fan et al., 2008).
To test the impact of scandals on loans from state-owned banks, we manually collect
loan information from the footnotes of financial statements. In addition to the amount and
sources of loans, we also collect the incidents of overdue borrowing. Many borrowers in
China rely on short-term debt to finance long-term projects and have to renew their
35
borrowings year by year. If a bank is concerned about a firm’s credit worthiness, it would
refuse to renew the loan, thereby leading to overdue borrowing for the firm. In China,
“overdue” borrowing is an important trigger and a hard indicator that the borrower's
credit worthiness will be downgraded.25
Table 8 reports the results of this analysis. Panel A provides descriptive statistics on
the additional variables used in this analysis. In addition to changes in loans and overdue
borrowing from state-owned banks, we also report additional variables including loans
from local government and accounts payables. Panel A of the table shows that, on
average, loans from state-owned banks increase subsequent to the public release of
scandals, this is likely due to the increasing trend of short-term debt during our sample
period.26
Panel B of Table 8 presents the results regressing changes in financing from state-
owned banks on a dummy variable indicating relationship scandals, a dummy variable
indicating mixed scandals, and the same control variables in equation (2). 27 The panel
shows that firms with relationship and mixed scandals experience more decreases in
loans from state-owned banks, as well as in additional measures including loans from
local governments or accounts payable. In addition, the panel shows that firms with
relationship scandals experience more increases in overdue borrowing from state-owned
banks. These findings are consistent with our hypothesis that relationship and mixed
25 We thank David Wu from PricewaterhouseCoopers China for sharing this insight. 26 Fan et al. (2008) also find that short-term debt increases among their sample firms involved with 23
high-level government officer corruption cases over the 1995 to 2003 period. 27 The number of observations for the analysis in Table 8 is slightly smaller than that in Table 4 because
of additional data requirements on bank loan disclosure.
36
scandals damage political ties that are essential for Chinese listed firms to obtain loans
(Fan et al., 2008).
6. Sensitivity tests
6.1. Alternative event windows
Our primary analysis in Table 4 uses long windows from two months up to two years
surrounding the event because it is difficult to identify when the market learns about the
scandals in China. One concern from starting the event window several months prior to
the disclosure date is that firms may commit scandals after experiencing poor returns,
thereby confounding the interpretation of our results. To address this concern, we repeat
our analysis in Panel B of Table 4 after using alternative windows starting at one month
prior to the public disclosure of the scandal: (-1, 1) months, (-1, 6) months, and (-1, 12)
months. Panel A of Table 9 presents the results of this analysis. It shows that the results
from this analysis are qualitatively the same as those reported in Table 4.
6.2. Alternative treatments of firms with multiple scandals
In our primary analysis in Table 4, we only keep the most recent scandal for firms
with multiple scandals. To assess the robustness of our results to this research design
choice, we repeat our analysis in Panel B of Table 4 after using the following alternative
treatments of firms with multiple scandals: (1) including all scandals as separate events,
and (2) keeping the earliest scandal. Panel B of Table 9 presents the results of this
analysis. It shows that the results from this analysis are qualitatively the same as those
reported in Table 4.
37
6.3. Alternative return measures
One concern with using CARs as a measure of market reaction is that it may be
biased in capturing long-term abnormal returns (Barber and Lyon, 1997). We therefore
repeat our analysis in Panel B of Table 4 after using buy-and-hold abnormal returns to
calculate long-term abnormal returns (i.e., for event windows longer than one month).
Panel C of Table 9 presents the results of this analysis. It shows that the results from this
analysis are qualitatively the same as those reported in Table 4.
6.4. Alternative treatments of delisted firms
Our analysis in Table 4 excludes firms that are subsequently delisted to keep our
firms consistent with those used in subsequent analysis on board and financing changes.
Since delisting indicates severe losses in shareholder value, this approach underestimates
the market reactions to corporate scandals.28 We assess whether our results are sensitive
to the treatment of delisted firms by repeating our analysis in Panel B of Table 4 after
adding back firms that are delisted subsequent to the scandals. Panel D of Table 9
presents the results of this analysis. We report the results with and without controlling for
a dummy indicating delisting. The panel shows that the results from this analysis are
qualitatively the same to those reported in Table 4.
6.5. Excluding scandals enforced by the Chinese stock exchanges
28 In our sample, 24 firms delisted one or two years after the disclosure of the scandal. Among these
firms, five are relationship scandal firms, 12 are mixed scandal firms, and seven are market scandal firms. In additional analysis regressing a dummy indicating delisting on a dummy indicating relationship scandals, a dummy indicating mixed scandals, and control variables, we find the dummy variables indicating relationship and mixed scandals to be insignificant at conventional levels. Thus, we do not find evidence suggesting that firms involved with relationship scandals or mixed scandals are more likely to be delisted.
38
A potential alternative explanation for the less negative stock returns associated with
market scandals is that they may be enforced by the Chinese stock exchanges, which do
not have inspection rights of listed companies and hence tend to take enforcement actions
against relatively minor offenses. To address this concern, we repeat our analysis in Panel
B of Table 4 after excluding from our sample five scandal cases enforced by the stock
exchanges. Panel E of Table 9 presents the results of this analysis. It shows that the
results from this analysis are qualitatively the same to those reported in Table 4, except
that the coefficient on the dummy indicating relationship scandals becomes insignificant
in the longest event window.
6.6. Restricting firms to those with non-missing data on magnitude of scandals
Our analysis in Table 4 assumes zero for missing information on magnitudes of
scandals. To the extent that this assumption introduces measurement errors, we perform
analysis after excluding firms with missing data on magnitudes of scandals. Panel F of
Table 9 presents the results of this analysis. It shows that the results from this analysis are
qualitatively the same to those reported in Table 4.
6.7. Restricting firms to SOEs
To explore whether our results in Table 4 are sensitive to restricting the sample to
SOEs, we repeat our analysis in Panel B of Table 4 after deleting 37 non-SOEs. Panel G
of Table 9 presents the results of this analysis. It shows that the results from this analysis
are qualitatively the same to those reported in Table 4.
7. Conclusion
39
This paper examines the value of relationship-based contracting and market-based
contracting in China. Using a sample of enforcement actions by the Chinese courts and
securities regulators from 1997-2005, we categorize each corporate scandal by whether it
primarily damages a firm’s ability to conduct relationship-based contracting (relationship
scandals), both relationship-based and market-based contracting (mixed scandals), or
market-based contracting (market scandals). We document that scandals damaging firms’
political networks and thus ability to conduct relationship-based contracting (relationship
and mixed scandals) lead to more negative stock returns than scandals damaging firms’
ability to conduct market-based contracting (market scandals). In addition, firms that are
more engaged in relationship-based contracting experience worse stock return
performance when they are involved with relationship and mixed scandals. Finally, we
document that relationship and mixed scandals lead to greater disruption in scandal firms’
political networks as reflected by the turnover of political directors and changes in
financing from state-owned banks subsequent to the scandals.
We caution that our study is not designed to be prescriptive and our analysis does not
consider the optimal level of relationship-based versus market-based contracting. Rather,
our goal is to provide an empirical assessment of the value of relationship-based and
market-based contracting in China. Overall, our study is among the first to investigate the
price effects and other economic consequences of corporate scandals in emerging
markets. By using China as a unique setting where both relationship scandals and market
scandals are common, we add to our understanding of the value of relationship- and
market-based contracting.
40
References
Agrawal, A., and Knoeber, C. R. 2001. Do some outside directors play a political role? Journal of Law and Economics 44, 179–198.
Aharony, J., Lee, J., and Wong, T.J. 2000. Financial package of IPO firms in China,
Journal of Accounting Research 38, 103–126. Allen, F., Qian, J., and Qian, M. 2005. Law, finance, and economic growth in China.
Journal of Financial Economics 77, 57–116. Barber, B., and Lyon, J.D. 1997. Detecting long-run abnormal stock returns: the
empirical power and specification of test statistics. Journal of Financial Economics 43, 341–372.
Becker, G.S., and Landes, W.M. 1974. Essays in the Economics of Crime and
Punishment. National Bureau of Economic Research. New York. Billett, M., King, T., and Mauer, D. 2007. Growth opportunities and the choice of
leverage, debt maturity, and covenants. The Journal of Finance 62, 697–730. Chaney, P.K., Faccio, M., and Parsley, D. 2011. The quality of accounting information in
politically connected firms. Journal of Accounting and Economics 51, 58–76. Chen, C., Chen, J.Z., Lobo, G.J., and Wang, Y. 2010. Association between borrower and
lender state ownership and accounting conservatism. Journal of Accounting Research 48, 973–1014.
Chen, G., Firth, M., Gao, D., and Rui, O. 2005. Is China’s securities regulatory agency a
toothless tiger? Evidence from enforcement actions. Journal of Accounting and Public Policy 24, 451–488.
Chen, K., and Yuan, H. 2004. Earnings management and capital resource allocation:
Evidence from China's accounting-based regulation of rights issues. The Accounting Review 74, 645–665.
DeFond, M.L., Wong, T.J., and Li, S. 2000. The impact of improved auditor
independence on audit market concentration in China. Journal of Accounting and Economics 28, 269–305.
Desai, H., Hogan, C.E., and Wilkins, M.S. 2006. The reputational penalty for aggressive
accounting: Earnings restatements and management turnover. The Accounting Review 81, 83–112.
41
Dinc, I. S. 2005. Politicians and banks: political influences on government-owned banks in emerging countries. Journal of Financial Economics 77, 453–479.
Faccio, M. 2006. Politically connected firms. American Economic Review 96, 369–386. Faccio M., Masulis, R., and McConnell, J.J. 2006. Political connections and corporate
bailouts. The Journal of Finance 61, 2597–2635. Fan, J., Rui, O., and Zhao, M. 2008. Public governance and corporate finance: evidence
from corruption cases. Journal of Comparative Economics 36, 343–364. Fan, J., Wong, T.J., and Zhang, T. 2007. Politically connected CEOs, corporate
governance, and post-IPO performance of China’s newly partially privatized firms. Journal of Financial Economics 84, 330–357.
Fisman, R. 1998. Estimating the value of political connections. American Economic
Review 91, 1095–1102. Hung, M., Wong, T.J., and Zhang, T. 2012. Political considerations in the decision of
Chinese SOEs to list in Hong Kong. Journal of Accounting and Economics 53, 435–449
Johnson, S., and Mitton, T. 2003. Cronyism and capital controls: evidence from
Malaysia. Journal of Financial Economics 67, 351–382. Karpoff, J., Lee, D.S., and Martin, G. 2008a. The cost to firms of cooking the books.
Journal of Financial and Quantitative Analysis 43, 581–612. Karpoff, J., Lee, D.S., and Martin, G. 2008b. The consequences to managers for financial
misrepresentation. Journal of Financial Economics 88, 193–215. Kedia, S., and Philippon, T. 2009. The economics of fraudulent accounting. Review of
Financial Studies 22, 2169–2199. La Porta, R., Lopez-de-Silanes, R., Shleifer, A., and Vishny, R.W. 1998. Law and
finance. Journal of Political Economy 106, 1113–1155. Leuz, C., and Oberholzer-Gee, F. 2006. Political relationships, global financing, and
corporate transparency: evidence from Indonesia. Journal of Financial Economics 81, 411–439.
Lin, J. Y., Cai, F., and Li, Z. 1996. The China Miracle: Development Strategy and
Economic Reform. Chinese University Press, Hong Kong.
42
Rajan, R. and Zingales, L. 1998. Which capitalism? Lessons from the East Asian Crisis, Journal of Applied Corporate Finance 11, 40–48.
Shi, S. and Weisert, D. 2002. Corporate governance with Chinese characteristics. The
China Business Review 29, 40. Shleifer, A. and Vishny, R. W. 1994. Politicians and firms. Quarterly Journal of
Economics 109, 995–1025. Srinivasan, S. 2005. Consequences of financial reporting failure for outside directors:
Evidence from accounting restatements and audit committee members. Journal of Accounting Research 43, 291–334.
Wang, Q., Wong, T.J., and Xia, L. 2008. State ownership, institutional environment and
auditor choice: evidence from China. Journal of Accounting and Economics 46, 112–134.
Weber, J., Willenborg, M., and Zhang, J. 2008. Does auditor reputation matter? The case
of KPMG Germany and Comroad AG. Journal of Accounting Research 46, 941–972. Zhang, P. 2007. An empirical study of intra-industry effect of corporate scandal
announcements in China. Working paper, The Chinese University of Hong Kong.
43
Appendix A Examples of relationship, mixed, and market scandals
Relationship scandals Example of R1 in Panel B of Table 1 -- LITTLE SWAN CO., LTD. (Company code: 000418) Little Swan is the first automatic washing machine manufacturer in China and the third largest washing machine manufacturer in the world at the end of 2007. It was listed in the Shenzhen Stock Exchange in 1997. To get the listing quota, the CEO of Little Swan, Zhu Dekun, bribed the head of State Reform Commission in the Jiangsu province, Li Sanyuan, by giving him 10,000 shares of employee shares (worth RMB 23.48 per share). Li Sanyuan was investigated by Jiangsu Provincial People’s Procuratorate on May 28, 2001 and was sentenced 11 years of imprisonment and had his property worth of RMB 50,000 confiscated. Example of R2a in Panel B of Table 1 -- ERDOS Group (Company code: 600295) Erdos Group is a leading cashmere manufacturer. It has an annual production capacity of 10 million pieces, which accounts for 30 percent of the international cashmere market. One of its wholly owned subsidiary illegally evaded value added tax by issuing fake invoices in 2003 and 2004. The money involved in this tax fraud is more than RMB 40 millions. On 14th January, 2005, Erdos suspended its stock trading on the Shanghai Stock Exchange and announced the enforcement action by the State Taxation Administration.
Mixed scandals Example of R2b/M2b in Panel B of Table 1 -- SHENZHEN ENERGY GROUP (Company code: 000027) Shenzhen Energy Group is the first state-owned utility company that was listed in the Shenzhen Stock Exchange in 1993 and is one of the largest companies in the Guangdong province. Lao Derong, the CEO of Shenzhen Energy Group, received RMB 7,780,000, HK$ 500,000, and US$ 139,000 from the company’s suppliers and construction agents from 1994 to 2002. She was sentenced to life imprisonment with charges of bribery, embezzlement, and abuse of power in 2003. Example of R2c/M2c in Panel B of Table 1 -- SHENZHEN DAWNCOM BUSINESS TECHNOLOGY AND SERVICE CO., LTD. (Company code: 000863) Shenzhen Dawncom Business Technology and Service is a privately controlled listed company with its largest shareholder (Heguang Group) controlling 28.55% of the shares at the end of 2004. The government also maintains a minority stake of 8.5% of untradeable shares. Wu Li was the Chairman of both the listed company and the parent company. On March 9, 2005, the listed company suddenly announced that Wu had resigned from his position. One day later, the company announced that it was under the investigation of the CSRC. According to the CSRC investigation report, Wu tunneled a significant amount of assets (worth RMB 0.31 billion in total) from the listed company through related party loans and guarantees before his resignation. Wu escaped to New Zealand in July 2004 and is still at large.
44
Appendix A, continued Market scandals Example of M1 in Panel B of Table 1--USTC CHUANGXIN CO., LTD. (Company code: 600551) USTC Chuangxin was listed in the Shanghai Stock Exchange in 2001. The company reported inflated net income of RMB 10,045,600 and RMB 7,360,700 for 2001 and 2002 through manipulation of revenues and expenses. The inflated net income is approximately 480% and 410% of its restated numbers for 2001 and 2002. After the restatement, the company failed to satisfy CSRC’s IPO requirement. The CSRC issued an administrative proceeding to criticize the CEO and imposed a fine of RMB 200,000 for the CEO and a fine of RMB 400,000 for the firm in September 2005. Example of M2a in Panel B of Table 1-- ZHEJIANG UNITED ELECTRONIC INDUSTRY CO., LTD. (Company code: 000925) Zhejiang United Electronic Industry is a privately controlled listed company and its largest shareholder controlled 28.44% of shares in February 2003. In 2003 and 2004, the listed company provided RMB 0.34 billion loan guarantee and RMB 0.25 billion related party loans to its largest shareholder and other related parties controlled by the largest shareholder. Both the CSRC and the Shenzhen Stock Exchange issued rectification notice and administrative proceeding to criticize the senior managers and board directors in April and June 2005. As the largest shareholder failed to make repayment in the following years, the listed company became seriously insolvent with negative net assets of RMB 0.28 billion (or a negative net asset per share of RMB 3.1) in the first half of the 2007. The company restructured in September 2007.
45
Appendix B Variable definitions
Variables of interest CAR = Cumulative abnormal return, calculated by cumulating stock returns minus returns
of the market index on the listing stock exchange in various event windows. Relationship scandal = A dummy variable equal to 1 if the firm is involved in a
relationship scandal and 0 otherwise. Mixed scandal = A dummy variable equal to 1 if the firm is involved in a mixed scandal
and 0 otherwise. Control variables Magnitude of scandal = The amount of bribery, tax evasion, misappropriation, or
financial misrepresentation, divided by total assets prior to the scandal. Firm size_pre = Natural logarithm of average total assets (in RMB) during the three years
prior to the event. Market-to-book_pre = Average ratio of market value to book value of assets during the
three years prior to the event. Tangibility_pre = Average ratio of fixed assets to total assets during the three years prior to
the event. Stock return_pre = Average yearly stock returns during the three years prior to the event. ROA_pre = Average ratio of net income to total assets during the three years prior to the
event. SOE = A dummy variable equal to 1 if the firm is owned by the government, and 0
otherwise. Legal environment = An index that captures the legal development level of each
province, based on the 2005 National Economic Research Institute (NERI) Index of Marketization of China’s provinces. The index is based on the average of the following three components (after normalizing each component to a range of 0-10): (1) the number of lawyers as a percentage of the province’s population; (2) the efficiency of local courts, as captured by the percentage of lawsuits pursued by the courts; and (3) the extent of property rights protection, as captured by the number of patents granted per research and development personnel.
Industry dummy = Dummy variables indicating industry sector membership based on the CSRC classification.
Conditional variables Political Chairman/CEO = A dummy variable equal to 1 if the Chairman or CEO is
politically connected, defined as a current or former officer of the central government, a local government, or the military.
Political director = Average percentage of politically connected board members during the three years prior to the event year. A board member is defined as politically connected if he/she is a current or former officer of the central government, a local government, or the military.
46
Appendix B, continued Strong relationship-based contracting = A dummy variable equal to 1 if the firm has
higher than the median value of the summary measure of the following five binary variables: (1) whether its Chairman/CEO is politically connected, (2) whether its percentage of politically connected directors is above the sample firm-level median, (3) whether its loans from state-owned banks is above the sample firm-level median, (4) whether its government subsidy is above the sample firm-level median, and (5) whether the legal development of its province is above the sample province-level median.
Changes in board structures Departure of directors = Accumulated departure rate of directors in the three years
subsequent to the event period. We calculate departure rate as the number of directors leaving the firm divided by the size of the board. In addition to overall departure of the board, we examine the following subgroups of directors that depart the firm: (1) Chairman/CEO; (2) political directors – directors who are politically connected, defined as a current or former government official; (3) affiliated directors – directors who have a personal affiliation with the Chairman/CEO based on one of the following relationships: relatives, laoxiang (individuals from the same hometown), former classmate in college, or former colleague in the previous employer; and (4) independent directors – directors who are not an employee of the listed company or related companies of the listed company, who are independent of company shareholders and management, and who are free from any business or other significant relationships.
Entry of directors = Accumulated entry rate of new directors in the three years subsequent to the event period. We calculate entry rate as the number of new directors joining the firm divided by the size of the board. In addition to examining the entry of new directors that are politically connected, we also examine the entry of new Chairman/CEO that is politically connected.
Net loss of directors = Accumulated departure rates of directors minus accumulated entry rates of directors subsequent to the event period.
Total turnover of directors = Accumulated departure rates of directors plus accumulated entry rates of directors subsequent to the event period.
Change in loans from state-owned banks ∆ Loans from SB = Change in average ratio of loans from state-owned banks divided by
total assets, from the three years before to three years after the event (excluding the event year).
∆(Loans from SB and government) = Change in average ratio of loans from state-owned banks plus loans from local government, divided by total assets, from the three years before to three years after the event (excluding the event year).
∆(Loans from SB and accounts payable) = Change in average ratio of loans from state-owned banks plus accounts payable, divided by total assets, from the three years before to three years after the event (excluding the event year).
47
Appendix B, continued ∆(Overdue borrowing) = Changes in the incidents of borrowing that is indicated as
overdue, from the three years before to three years after the event (excluding the event year). The incident of overdue borrowing is measured as the sum of two dummy variables, with one indicating if a short-term debt is overdue and the other indicating if a long-term debt is overdue.
Additional control variables ∆Firm size = Change in nature logarithm of average total assets in RMB from the three
years before to three years after the event (excluding the event year). ∆Market-to-book = Change in average ratio of market value to book value of assets from
the three years before to three years after the event (excluding the event year). ∆Tangibility = Change in average ratio of fixed assets to total assets from the three years
before to three years after the event (excluding the event year). ∆ROA = Change in average ratio of net income to total assets from the three years before
to three years after the event (excluding the event year). Year dummy = Dummy variables indicating years.
48
Figure 1 Timeline of enforcement actions
For enforcement actions against CEOs/Chairmen by courts, the event date is the date on which the press or the firm reports that the executive is arrested or brought in for questioning (‘ShuangGui’), whichever is earlier.
For enforcement actions by the CSRC and the stock exchanges, the event date is the date on which the securities regulators or the firm announce the investigation inquiry, whichever is earlier.
Violation period Enforcement period
Event date: First public disclosure of the scandal
49
Figure 2 Cumulative abnormal returns (CARs) for different types of scandals
Relationship scandals are scandals that primarily damage political networks and hurt firms’ ability to conduct relationship contracting.
Market scandals are scandals that primarily damage market confidence and hurt firms’ ability to conduct market-based contracting.
Mixed scandals are scandals that damage firms’ ability to conduct both relationship-based and market-based contracting.
‐0.400
‐0.350
‐0.300
‐0.250
‐0.200
‐0.150
‐0.100
‐0.050
0.000
0.050
‐12 ‐10 ‐8 ‐6 ‐4 ‐2 0 2 4 6 8 10 12
Cumulative Abnormal Returns by Different Types of Scandals
Mixed Scandals Relationship Scandals Market Scandals
50
Table 1 Classification of scandals based on contracting characteristics
Panel A: Key types of scandals that damage relationship-based and market-based contracting 1. Scandals that damage firms’ ability to conduct relationship-based contracting (i.e., contracting ability
with the government and political networks) R1. Managers bribing government officials R2. Managers misappropriating state assets
a. Tax evasion b. Managers of SOEs misappropriating firm assets c. Managers of non-SOEs misappropriating firm assets in which the government has a minority
stake 2. Scandals that damage firms’ ability to conduct market-based contracting (i.e., contracting ability with market participants such as outside shareholders, suppliers, and customers)
M1. Financial misrepresentation M2. Managers misappropriating firm assets
a. Managers of non-SOEs misappropriating firm assets in which the government has no ownership b. Managers of SOEs misappropriating firm assets c. Managers of non-SOEs misappropriating firm assets in which the government has a minority
stake Panel B: Classification of scandals for our sample firms Category Description N Relationship scandals
Scandals that primarily damage firms’ ability to conduct relationship-based contracting
1. Managers bribing government officials [R1] -Bribing CSRC officials for IPOs, SEOs, and relationship building -Bribing government officials to obtain loans or projects 2. Tax evasion [R2a]
17 7 2 26
Mixed scandals Scandals that impair firms’ ability to conduct both relationship-based and market-based contracting
1. Managers of SOEs misappropriating firm assets [R2b=M2b] -Embezzlement -Taking kickbacks -Others (abuse of power for private gains, forgery etc.) 2. Managers of non-SOEs in which government maintains a minority
stake misappropriating firm assets [R2c=M2c] 3. Managers bribing government officials and manipulating accounting
numbers to conceal the bribe [R1+M1]
44 18 19 13 1 95
Market scandals
Scandals that primarily damage firms’ ability to conduct market-based contracting
1. Financial misrepresentation [M1] -Accounting manipulations to inflate earnings
-False accounting disclosure 2. Managers of non-SOEs misappropriating firm assets [M2a] -Tunneling -Excessive related party loans and guarantee
33 25 23 10 91
51
Table 2 Sample distribution by year and industry
Panel A: Sample distribution by year
Year Relationship scandals Mixed scandals Market scandals N % N % N %
1997 0 0.00% 1 1.05% 2 2.20% 1998 1 3.85% 0 0.00% 2 2.20% 1999 1 3.85% 7 7.37% 6 6.59% 2000 1 3.85% 7 7.37% 9 9.89% 2001 8 30.77% 8 8.42% 10 10.99% 2002 2 7.69% 11 11.58% 21 23.08% 2003 2 7.69% 12 12.63% 13 14.29% 2004 5 19.23% 16 16.84% 12 13.19% 2005 6 23.08% 33 34.74% 16 17.58% Total 26 100.00% 95 100.00% 91 100.00%
Panel B: Sample distribution by industrya
a Based on the industry classification by the CSRC.
Industry Relationship scandals Mixed scandals Market scandals N % N % N %
1 Agriculture 0 0.00% 2 2.11% 7 7.69% 2 Natural resources 0 0.00% 0 0.00% 1 1.10% 3 Manufacturing 15 57.69% 51 53.68% 46 50.55% 4 Utilities 1 3.85% 8 8.42% 0 0.00% 5 Construction 1 3.85% 2 2.11% 1 1.10% 6 Transportation 0 0.00% 7 7.37% 3 3.30% 7 Information technology 3 11.54% 5 5.26% 6 6.59% 8 Wholesale and retail 1 3.85% 4 4.21% 5 5.49% 9 Finance and insurance 0 0.00% 1 1.05% 0 0.00% 10 Real estate 1 7.69% 3 3.16% 5 5.49% 11 Services 1 7.69% 5 5.26% 5 5.49% 12 Communication 0 0.00% 1 1.05% 1 1.10% 13 Others 3 11.54% 6 6.32% 11 12.09% Total 26 100.00% 95 100.00% 91 100.00%
52
Table 3 Mean cumulative abnormal returns (CARs) during various event windows a
Event window Relationship scandals
(A) Mixed scandals
(B) Market scandals
(C) Difference
(A)-(C) Difference
(B)-C) (-1, 1) day -0.035 -0.026 -0.027 -0.008 0.001 [-2.50]** [-4.29]*** [-3.73]*** [-0.55] [0.10] (-5, 5) days -0.034 -0.066 -0.042 0.008 -0.023 [ -1.86]* [ -4.73]*** [-3.80]*** [0.36] [-1.29] (-10, 10) days -0.064 -0.124 -0.060 -0.004 -0.063 [ -2.84]*** [ -5.27]*** [-3.97]*** [-0.13] [-2.24]** (-15, 15) days -0.078 -0.148 -0.060 -0.018 -0.088 [ -3.56]*** [-5.23]*** [-3.75]*** [-0.56] [-2.67]*** (-1, 1) month -0.173 -0.151 -0.045 -0.128 -0.106 [ -3.22]*** [-6.25]*** [-3.01]*** [-3.22]*** [-3.71]*** (-2, 2) months -0.185 -0.182 -0.039 -0.146 -0.143 [ -3.59]*** [-6.01]*** [-2.07]** [-3.27]*** [-3.99]*** (-6, 6) months -0.308 -0.245 -0.088 -0.220 -0.157 [-4.14]*** [-6.77]*** [-3.31]*** [-3.47]*** [-3.44]*** (-12, 12) months -0.373 -0.313 -0.158 -0.215 -0.155 [-3.57]*** [-6.24]*** [-4.17]*** [-2.40]** [-2.45]** N 26 95 91 a***, **, * Indicates significance at the 1%, 5%, and 10% two-tailed level, respectively. t-statistics in brackets.
53
Table 4 Analysis regressing market reactions (CARs) on types of scandals and control variables
Panel A: Descriptive statistics (N=212)
Variables Relationship scandal
(N=26)Mixed scandal
(N=95)Market scandal
(N=91) CAR (-1,1) month Mean -0.173 -0.151 -0.045 Median -0.103 -0.096 -0.046 Std. dev. 0.263 0.229 0.138 CAR (-2,2) months Mean -0.185 -0.182 -0.039 Median -0.123 -0.148 -0.009 Std. dev. 0.258 0.282 0.175 CAR (-6,6) months Mean -0.308 -0.245 -0.088 Median -0.226 -0.209 -0.111 Std. dev. 0.372 0.347 0.244 CAR (-12,12) months Mean -0.373 -0.313 -0.158 Median -0.330 -0.326 -0.187 Std. dev. 0.522 0.467 0.345 Magnitude of scandal (%) Mean 3.972 3.987 15.715 Median 0.007 0.124 3.851 Std. dev. 15.817 16.808 30.898 Firm size pre Mean 21.081 21.077 20.648 Median 21.122 20.893 20.667 Std. dev. 0.850 0.955 0.806 Market-to-book pre Mean 3.222 2.641 3.089 Median 3.122 2.118 2.808 Std. dev. 1.770 1.371 1.388 Tangibility pre Mean 0.291 0.430 0.348 Median 0.249 0.402 0.332 Std. dev. 0.186 0.251 0.182 Stock return pre Mean 0.165 0.052 0.051 Median 0.065 -0.017 0.065 Std. dev. 0.405 0.398 0.291 ROA pre Mean 0.039 0.014 -0.009 Median 0.031 0.022 0.004 Std. dev. 0.049 0.062 0.068 SOE Mean 0.692 0.863 0.824 Median 1.000 1.000 1.000 Std. dev. 0.471 0.346 0.383 Legal environment Mean 5.967 5.863 5.637 Median 6.504 5.396 5.438 Std. dev. 2.047 1.912 1.864
54
Table 4, continued
Panel B: Regression of CARs on types of scandals and control variables a
CAR
(-1, 1) month CAR
(-2, 2) months CAR
(-6, 6) months CAR
(-12, 12) months Relationship scandal -0.098 -0.120 -0.230 -0.283 [-2.22]** [-2.31]** [-3.40]*** [-3.05]*** Mixed scandal -0.131 -0.170 -0.223 -0.254 [-4.48]*** [-4.93]*** [-4.95]*** [-4.12]*** Magnitude of scandal 0.000 0.001 -0.000 -0.001 [0.18] [0.86] [-0.49] [-0.51] Firm size pre -0.046 -0.049 -0.057 -0.037 [-2.43]** [-2.20]** [-1.96]* [-0.92] Market-to-book pre -0.041 -0.048 -0.071 -0.096 [-3.14]*** [-3.12]*** [-3.56]*** [-3.50]*** Tangibility pre 0.092 0.083 0.134 0.065 [1.30] [0.99] [1.23] [0.43] Stock return pre -0.029 -0.066 -0.014 0.138 [-0.65] [-1.25] [-0.20] [1.46] ROA pre 0.561 0.789 1.462 1.487 [2.48]** [2.95]*** [4.19]*** [3.11]*** SOE 0.136 0.135 0.084 -0.040 [3.74]*** [3.14]*** [1.50] [-0.52] Legal environment 0.009 0.008 0.002 0.020 [1.23] [0.93] [0.18] [1.27] Industry dummy Included Included Included Included N 212 212 212 212 Adj. R2 0.167 0.180 0.188 0.136
a***, **, * Indicates significance at the 1%, 5%, and 10% two-tailed level, respectively. t-statistics in brackets. Variable definitions: See Appendix B.
55
Table 5 Legal penalty and total valuation loss
Panel A: Legal penalties imposed on firms Relationship scandal
(N=26) Mixed scandal
(N=95) Market scandal
(N=91) Administrative sanction Monetary penalty (in US$ million)a Mean 0.056 0.051 0.071 Median 0.048 0.048 0.048 Std. dev. 0.045 0.015 0.134 Non-monetary penalty Criticism 6 17 35 Rectification 1 6 22 Trading suspension 0 0 1 Operating suspension 1 0 1 Criminal sanction Monetary penalty (in US$ million)a Mean 141.832 17.147 0.00 Median 3.554 12.121 0.00 Std. dev. 262.974 22.941 0.00
Panel B: Legal penalties imposed on individuals Relationship scandal
(N=26) Mixed scandal
(N=95) Market scandal
(N=91) Administrative sanction Monetary penalty (in US$ million)a Mean 0.038 0.254 0.033 Median 0.024 0.055 0.028 Std. dev. 0.032 0.367 0.025 Non-monetary penalty Criticism 9 15 65 Bar from securities market 5 5 3 Criminal sanction Monetary penalty (in US$ million)a Mean 0.149 0.682 0.00 Median 0.070 0.025 0.00 Std. dev. 0.210 1.502 0.00 Non-monetary penalty Imprisonment 10 49 1 Death penalty 0 10 0 Others (pending investigation,
secret hearing, fled the country) 6 25 0
56
Table 5, continued
Panel C: Total dollar losses attributable to loss in contracting ability (in US$ million)a,b
Relationship scandal (N=26)
Mixed scandal (N=95)
Market scandal (N=91)
Total monetary penalties imposed on firm (A) 709.438 86.297 3.422 Total valuation effect of accounting readjustments (B) n.a. 1.484 1,611.239 (-1, 1) month Total dollar loss (C) 3,748.851 7,192.381 607.943 Total dollar loss in contracting ability (C-A-B) 3,039.413 7,104.600 -1,006.718 Average dollar loss in contracting ability ((C-A-B)/N) 116.901 74.785 -11.063 (-2, 2) months Total dollar loss (C) 4,095.378 8,916.890 745.775 Total dollar loss in contracting ability (C-A-B) 3,385.940 8,829.109 -868.886 Average dollar loss in contracting ability ((C-A-B)/N) 130.228 92.938 -9.548 (-6, 6) months Total dollar loss (C) 6,629.036 11,709.377 1,549.916 Total dollar loss in contracting ability (C-A-B) 5,919.598 11,621.596 -64.745 Average dollar loss in contracting ability ((C-A-B)/N) 227.677 122.333 -0.711 (-12, 12) months Total dollar loss (C) 7,180.354 15,007.927 3,507.541 Total dollar loss in contracting ability (C-A-B) 6,470.916 14,920.146 1,892.880Average dollar loss in contracting ability ((C-A-B)/N) 248.881 157.054 20.801
57
Table 5, continued Panel D: Regression of CARs on types of scandals and control variables, after further controlling for monetary and non-monetary penalties on firms and individuals
CAR
(-1, 1) month CAR
(-2, 2) months CAR
(-6, 6) months CAR
(-12, 12) months Relationship scandal -0.087 -0.118 -0.217 -0.276 [-1.75]* [-2.01]** [-2.85]*** [-2.63]*** Mixed scandal -0.163 -0.211 -0.274 -0.304 [-3.95]*** [-4.38]*** [-4.37]*** [-3.51]*** Penalty on firms Monetary administration penalty -0.471 -11.911 -25.133 9.691 [-0.01] [-0.20] [-0.33] [0.09] Monetary criminal penalty -0.709 -0.903 -1.439 -1.632 [-1.76]* [-1.91]* [-2.35]** [-1.92]* Criticism -0.002 0.008 0.012 0.052 [-0.07] [0.21] [0.26] [0.78] Rectification -0.043 -0.051 -0.062 -0.061 [-1.10] [-1.12] [-1.04] [-0.74] Trading suspension -0.440 -0.790 -0.758 -1.648 [-1.34] [-2.05]** [-1.52] [-2.39]** Operating suspension 0.134 0.149 0.301 0.880 [0.49] [0.47] [0.73] [1.55] Penalty on individuals Monetary administration penalty 59.526 76.646 38.294 168.268 [1.15] [1.27] [0.49] [1.55] Monetary criminal penalty 26.851 45.049 9.776 83.232 [0.83] [1.19] [0.20] [1.22] Criticism imposed on individual -0.047 -0.057 -0.112 -0.141 [-1.39] [-1.42] [-2.15]** [-1.96]* Bar from securities market -0.136 -0.105 -0.203 -0.316 [-2.11]** [-1.40] [-2.07]** [-2.34]** Imprisonment 0.003 -0.001 0.005 0.019 [0.06] [-0.01] [0.09] [0.24] Death penalty 0.076 0.076 0.024 -0.138 [1.15] [0.98] [0.24] [-0.99] Control variables Magnitude of scandal 0.001 0.001 0.001 -0.000 [0.83] [1.29] [0.90] [-0.03] Firm size pre -0.025 -0.027 -0.022 -0.003 [-1.27] [-1.17] [-0.72] [-0.08] Market-to-book pre -0.034 -0.040 -0.056 -0.086 [-2.59]** [-2.60]*** [-2.85]*** [-3.16]*** Tangibility pre 0.046 0.028 0.042 -0.014 [0.65] [0.33] [0.38] [-0.09] Stock return pre -0.041 -0.085 -0.030 0.115 [-0.91] [-1.62] [-0.43] [1.21] ROA pre 0.554 0.808 1.397 1.597 [2.44]** [3.05]*** [4.05]*** [3.35]*** SOE 0.113 0.114 0.053 -0.073 [3.05]*** [2.62]*** [0.94] [-0.93] Legal environment 0.002 -0.001 -0.011 0.005 [0.23] [-0.11] [-0.94] [0.34] Industry dummy Included Included Included Included N 212 212 212 212 Adj. R2 0.201 0.229 0.241 0.178
58
Table 5, continued
a We calculate a firm’s total loss during each event window as its CAR during the event window multiplied by its market capitalization prior to the event window. The numbers are converted to US dollar based on the average exchange rate between US$ and RMB during our sample period, 8.25:1. b***, **, * Indicates significance at the 1%, 5%, and 10% two-tailed level, respectively. t-statistics in brackets. Variable definitions: See Appendix B.
59
Table 6 Analysis partitioned by the strength of relationship-based contracting
Panel A: Descriptive statistics on conditional variables (N=212) Variables Mean Q1 Median Q3 Std. dev. Political Chairman/CEO 0.324 0.000 0.000 0.750 0.415 Political directors 0.190 0.000 0.143 0.290 0.199 Loans from state-owned banks 0.278 0.173 0.287 0.397 0.158 Government subsidy 0.005 0.000 0.001 0.004 0.014 Legal environment 5.972 4.566 5.733 7.446 1.939 Strong relationship-based contracting 0.509 0.000 1.000 1.000 0.501
60
Table 6, continued
Panel B: Regression of CARs on types of scandals and control variables, conditional on the strength of relationship-based contracting a Weak relationship-based contracting Strong relationship-based contracting CAR
(-1, 1) month
CAR (-2, 2)
months
CAR (-6, 6)
months
CAR (-12, 12) months
CAR (-1, 1) month
CAR (-2, 2)
months
CAR (-6, 6)
months
CAR (-12, 12) months
(1) (2) (3) (4) (5) (6) (7) (8) Relationship scandal (β1) 0.007 0.005 -0.061 -0.046 -0.169 -0.189 -0.295 -0.418 [0.12] [0.07] [-0.64] [-0.30] [-2.80]*** [-2.54]** [-3.14]*** [-3.74]*** Mixed scandal (β2) 0.014 -0.024 -0.111 -0.182 -0.254 -0.293 -0.305 -0.295 [0.38] [-0.52] [-1.74]* [-1.76]* [-6.39]*** [-5.99]*** [-4.92]*** [-3.99]*** Magnitude of scandal 0.001 0.002 0.002 0.002 -0.000 -0.001 -0.003 -0.003 [1.53] [2.51]** [2.02]** [0.94] [-0.34] [-0.66] [-2.32]** [-1.77]* Firm size pre -0.013 -0.004 0.027 0.064 -0.044 -0.068 -0.134 -0.080 [-0.58] [-0.13] [0.71] [1.03] [-1.48] [-1.87]* [-2.92]*** [-1.47] Market-to-book pre -0.036 -0.036 -0.076 -0.116 -0.038 -0.060 -0.074 -0.095 [-2.11]** [-1.73]* [-2.60]** [-2.47]** [-1.95]* [-2.54]** [-2.45]** [-2.65]*** Tangibility pre 0.006 0.047 0.125 -0.007 0.260 0.127 0.120 0.177 [0.07] [0.49] [0.95] [-0.03] [2.19]** [0.87] [0.65] [0.81] Stock return pre 0.055 0.045 0.148 0.227 -0.075 -0.089 -0.036 0.135 [0.98] [0.64] [1.52] [1.44] [-1.15] [-1.11] [-0.35] [1.12] ROA pre 0.591 0.743 1.378 1.851 0.282 0.586 1.389 0.889 [1.99]* [2.01]** [2.69]*** [2.23]** [0.88] [1.49] [2.79]*** [1.50] SOE -0.001 0.045 -0.016 -0.068 0.228 0.171 0.128 -0.036 [-0.01] [0.69] [-0.17] [-0.46] [4.88]*** [2.96]*** [1.76]* [-0.42] Industry dummy Included Included Included Included Included Included Included Included N 105 105 105 105 107 107 107 107 Adj. R2 -0.066 -0.015 0.097 0.009 0.422 0.358 0.310 0.305 Diff in the coeff. on β1 (5)-(1) (6)-(2) (7)-(3) (8)-(4) Chi-square [5.40]** [5.75]** [4.77]** [5.10]** Diff in the coeff. on β2 (5)-(1) (6)-(2) (7)-(3) (8)-(4) Chi-square [27.74]*** [17.07]*** [5.46]** [0.97]
a***, **, * Indicates significance at the 1%, 5%, and 10% two-tailed level, respectively. t-statistics in brackets. Variable definitions: See Appendix B
61
Table 7 Impact of scandals on governance
Panel A: Descriptive statistics on additional variables (N=212) Variables Mean Q1 Median Q3 Std. dev. Departure of Chairman/CEO 0.093 0.056 0.080 0.118 0.078 Departure of directors 0.480 0.200 0.424 0.677 0.369 Departure of political Chairman/CEO 0.024 0.000 0.000 0.000 0.050 Departure of political directors 0.145 0.000 0.091 0.200 0.175 Departure of affiliated directors 0.146 0.000 0.071 0.231 0.197 Departure of independent directors 0.069 0.000 0.000 0.100 0.103 Entry of political Chairman/CEO 0.022 0.000 0.000 0.000 0.050 Net loss of political Chairman/CEO 0.002 0.000 0.000 0.000 0.051 Total turnover of political Chairman/CEO 0.046 0.000 0.000 0.077 0.086 Entry of political directors 0.103 0.000 0.077 0.167 0.119 Net loss of political directors 0.042 0.000 0.000 0.095 0.143 Total turnover of political directors 0.247 0.059 0.182 0.333 0.263 ∆ Firm size -0.044 -0.354 -0.021 0.298 0.549 ∆ Market-to-book -0.736 -1.361 -0.538 -0.069 1.252 ∆ Tangibility 0.108 0.024 0.076 0.173 0.167 ∆ ROA -0.049 -0.091 -0.021 0.014 0.101
62
Table 7, continued
Panel B: Regressions with the dependent variable being departure of different types of directors a
Departure of
Chairman/CEO Departure of
directors
Departure of political
Chairman/CEO
Departure of political directors
Departure of affiliated directors
Departure of independent
directors Relationship scandal 0.039 0.172 0.008 0.078 0.245 0.025 [2.87]*** [2.14]** [0.40] [2.01]** [5.84]*** [1.02] Mixed scandal 0.047 0.211 0.028 0.137 0.210 0.009 [5.27]*** [3.99]*** [2.18]** [5.41]*** [7.61]*** [0.54] Magnitude of scandal -0.000 -0.000 -0.000 0.000 0.000 0.000 [-1.67]* [-0.31] [-0.98] [1.27] [0.50] [0.14] ∆ Firm size -0.011 -0.103 -0.020 -0.016 -0.042 -0.020 [-1.14] [-1.82]* [-1.51] [-0.60] [-1.44] [-1.17] ∆ Market-to-book 0.001 -0.010 -0.008 -0.005 -0.004 0.001 [0.17] [-0.46] [-1.53] [-0.49] [-0.38] [0.18] ∆ Tangibility 0.024 0.175 0.034 0.045 0.044 0.084 [0.86] [1.05] [0.86] [0.56] [0.51] [1.65]* ∆ROA 0.041 0.078 0.074 -0.076 -0.031 -0.100 [0.89] [0.28] [1.13] [-0.58] [-0.21] [-1.21] SOE -0.015 -0.186 0.000 -0.042 0.053 -0.020 [-1.42] [-3.05]*** [0.01] [-1.44] [1.66]* [-1.08] Legal environment -0.004 0.014 -0.001 0.009 0.007 0.007 [-1.62] [1.07] [-0.42] [1.41] [0.98] [1.69]* Industry dummy Included Included Included Included Included Included Year dummy Included Included Included Included Included Included N 212 212 212 212 212 212 Adj. R2 0.542 0.271 0.222 0.253 0.302 0.145
63
Table 7, continued
Panel C: Regressions with the dependent variable being entry of political directors, net loss of political directors, and total turnover of political directors
Realignment of political chairman/CEOs Realignment of political directors
Entry of political
Chairman/CEO
Net loss of political
Chairman/CEO
Total turnover of political
Chairman/CEO
Entry of political directors
Net loss of political directors
Total turnover of political directors
Relationship scandal -0.002 0.012 0.008 0.077 0.001 0.154 [-0.18] [0.91] [0.40] [2.71]*** [0.02] [2.65]*** Mixed scandal 0.018 -0.008 0.028 0.067 0.070 0.204 [2.29]** [-0.93] [2.18]** [3.58]*** [3.10]*** [5.32]*** Magnitude of scandal -0.000 -0.000 -0.000 0.000 0.000 0.000 [-0.63] [-0.31] [-0.98] [0.38] [1.10] [1.03] ∆ Firm size -0.011 0.001 -0.020 -0.034 0.018 -0.050 [-1.29] [0.11] [-1.51] [-1.70]* [0.73] [-1.22] ∆ Market-to-book 0.000 -0.009 -0.008 -0.012 0.007 -0.017 [0.03] [-2.33]** [-1.53] [-1.51] [0.70] [-1.06] ∆ Tangibility -0.004 0.042 0.034 0.002 0.042 0.047 [-0.17] [1.57] [0.86] [0.04] [0.59] [0.39] ∆ROA 0.043 -0.013 0.074 -0.026 -0.050 -0.102 [1.07] [-0.28] [1.13] [-0.27] [-0.43] [-0.52] SOE 0.004 -0.008 0.000 0.002 -0.044 -0.040 [0.45] [-0.81] [0.01] [0.09] [-1.69]* [-0.91] Legal environment -0.003 0.004 -0.001 0.004 0.006 0.013 [-1.45] [2.01]** [-0.42] [0.74] [0.96] [1.30] Industry dummy Included Included Included Included Included Included Year dummy Included Included Included Included Included Included N 212 212 212 212 212 212 Adj. R2 0.124 0.017 0.222 0.117 0.101 0.243
a***, **, * Indicates significance at the 1%, 5%, and 10% two-tailed level, respectively. t-statistics in brackets. Variable definitions: See Appendix B.
64
Table 8 Impact of scandals on financing
Panel A: Descriptive statistics on additional variables Variables N Mean Q1 Median Q3 Std. dev. ∆ Loans from SB 171 0.036 -0.046 0.013 0.098 0.132 ∆ (Loans from SB + government) 171 0.037 -0.046 0.010 0.109 0.133 ∆ (Loans from SB + accounts payable) 171 0.058 -0.024 0.046 0.123 0.143 ∆ Overdue borrowing from SB 212 0.258 0.000 0.000 0.500 0.540
Panel B: Regressions with the dependent variable being changes in loans from state-owned banks and other short-term financing
∆ Loans from SB ∆ (Loans from SB +
government) ∆ (Loans from SB + accounts payable)
∆ Overdue borrowings from SB
Relationship scandal -0.081 -0.079 -0.090 0.337 [-2.35]** [-2.23]** [-2.45]** [2.72]*** Mixed scandal -0.041 -0.042 -0.044 0.094 [-1.88]* [-1.86]* [-1.87]* [1.15] Magnitude of scandal 0.000 0.000 0.000 0.000 [1.04] [0.96] [1.53] [0.57] ∆ Firm size -0.041 -0.041 -0.025 -0.145 [-1.70]* [-1.64] [-0.97] [-1.67]* ∆ Market-to-book -0.014 -0.015 -0.011 -0.007 [-1.62] [-1.74]* [-1.15] [-0.19] ∆ Tangibility 0.113 0.099 0.080 0.293 [1.65] [1.41] [1.09] [1.14] ∆ ROA -0.550 -0.551 -0.654 -1.098 [-4.96]*** [-4.84]*** [-5.52]*** [-2.61]*** SOE 0.022 0.023 0.001 -0.043 [0.90] [0.92] [0.03] [-0.46] Legal environment -0.002 -0.002 -0.002 -0.036 [-0.30] [-0.32] [-0.35] [-1.72]* Industry dummy Included Included Included Included Year dummy Included Included Included Included N 171 171 171 212 Adj. R2 0.227 0.207 0.257 0.192
a***, **, * Indicates significance at the 1%, 5%, and 10% two-tailed level, respectively. t-statistics in brackets. Variable definitions: See Appendix B
65
Table 9
Sensitivity analysisa
Panel A: Alternative event windows: (-1, 2) months, (-1,6) months, and (-1, 12) months
CAR
(-1, 2) month CAR
(-1, 6) months CAR
(-1, 12) months Relationship scandal -0.085 -0.160 -0.230 [-1.68]* [-2.75]*** [-2.80]*** Mixed scandal -0.152 -0.180 -0.228 [-4.52]*** [-4.63]*** [-4.17]*** Control variables Included Included Included N 212 212 212 Adj. R2 0.155 0.129 0.126 Panel B: Alternative treatment of firms with multiple scandals
Including all scandals CAR
(-1, 1) month CAR
(-2, 2) months CAR
(-6, 6) months CAR
(-12, 12) months Relationship scandal -0.120 -0.140 -0.261 -0.248 [-2.92]*** [-2.95]*** [-4.19]*** [-2.83]*** Mixed scandal -0.140 -0.172 -0.254 -0.261 [-5.09]*** [-5.41]*** [-6.10]*** [-4.44]*** Control variables Included Included Included Included N 250 250 250 250 Adj. R2 0.196 0.204 0.269 0.215 Keeping the earliest scandal
CAR (-1, 1) month
CAR (-2, 2) months
CAR (-6, 6) months
CAR (-12, 12) months
Relationship scandal -0.098 -0.120 -0.230 -0.283 [-2.22]** [-2.31]** [-3.40]*** [-3.05]*** Mixed scandal -0.131 -0.170 -0.223 -0.254 [-4.48]*** [-4.93]*** [-4.95]*** [-4.12]*** Control variables Included Included Included Included N 212 212 212 212 Adj. R2 0.167 0.180 0.188 0.136
Panel C: Alternative return measures (BHAR)
CAR
(-1, 1) month CAR
(-2, 2) months CAR
(-6, 6) months CAR
(-12, 12) months Relationship scandal -0.076 -0.100 -0.160 -0.110 [-2.04]** [-2.39]** [-2.70]*** [-1.31] Mixed scandal -0.105 -0.125 -0.170 -0.177 [-4.24]*** [-4.48]*** [-4.35]*** [-3.17]*** Control variables Included Included Included Included N 212 212 212 212 Adj. R2 0.115 0.122 0.125 0.047
66
Table 9, continued Panel D: Alternative treatment of delisted firms Without controlling for delisting
CAR (-1, 1) month
CAR (-2, 2) months
CAR (-6, 6) months
CAR (-12, 12) months
Relationship scandal -0.068 -0.090 -0.179 -0.268 [-1.65]* [-1.85]* [-2.73]*** [-2.92]*** Mixed scandal -0.117 -0.156 -0.196 -0.263 [-4.14]*** [-4.68]*** [-4.35]*** [-4.18]*** Control variables Included Included Included Included N 236 236 236 236 Adj. R2 0.166 0.172 0.180 0.222
Controlling for delisting CAR
(-1, 1) month CAR
(-2, 2) months CAR
(-6, 6) months CAR
(-12, 12) months Relationship scandal -0.077 -0.096 -0.189 -0.268 [-1.85]* [-1.96]* [-2.85]*** [-2.89]*** Mixed scandal -0.122 -0.159 -0.201 -0.263 [-4.29]*** [-4.76]*** [-4.44]*** [-4.15]*** Delisting 0.065 0.049 0.072 0.004 [1.44] [0.93] [1.00] [0.04] Control variables Included Included Included Included N 236 236 236 236 Adj. R2 0.170 0.172 0.180 0.218
Panel E: Excluding scandals enforced by stock exchanges
CAR
(-1, 1) month CAR
(-2, 2) months CAR
(-6, 6) months CAR
(-12, 12) months Relationship scandal -0.089 -0.110 -0.231 -0.287 [-1.99]** [-2.08]** [-3.27]*** [-3.02]*** Mixed scandal -0.125 -0.163 -0.224 -0.258 [-4.19]*** [-4.60]*** [-4.74]*** [-4.07]*** Control variables Included Included Included Included N 207 207 207 207 Adj. R2 0.169 0.177 0.195 0.135
67
Table 9, continued
Panel F: Restricting sample to firms with non-missing data on magnitude of scandals
CAR
(-1, 1) month CAR
(-2, 2) months CAR
(-6, 6) months CAR
(-12, 12) months Relationship scandal -0.075 -0.088 -0.201 -0.293 [-1.31] [-1.28] [-2.26]** [-2.45]** Mixed scandal -0.084 -0.153 -0.201 -0.304 [-2.09]** [-3.21]*** [-3.21]*** [-3.62]*** Magnitude of scandal -0.001 0.000 -0.000 -0.001 [-0.88] [0.40] [-0.48] [-0.82] Other control variables Included Included Included Included N 139 139 139 139 Adj. R2 0.087 0.065 0.052 0.072
Panel G: Restricting sample to SOEs
CAR
(-1, 1) month CAR
(-2, 2) months CAR
(-6, 6) months CAR
(-12, 12) months Relationship scandal -0.045 -0.067 -0.168 -0.212 [-1.00] [-1.20] [-2.16]** [-1.89]* Mixed scandal -0.085 -0.122 -0.213 -0.251 [-3.04]*** [-3.47]*** [-4.41]*** [-3.58]*** Control variables Included Included Included Included N 175 175 175 175 Adj. R2 -0.008 0.031 0.112 0.072
a See Table 4 for the list of control variables