Do independent directors play a political role? Evidence from independent directors’ death events Academic research paper on "Social and economic geography"

China Journal of Accounting Research xxx (2016) xxx-xxx

HOSTED BY

Contents lists available at ScienceDirect

China Journal of Accounting Research

journal homepage: www.elsevier.com/locate/cjar

Do independent directors play a political role? Evidence from independent directors' death events

Qing Yea b, Zengquan Lic*

a Center for Accounting, Finance and Institutions, Sun Yat-sen University, China b Sun Yat-sen Business School, Sun Yat-sen University, China

c School of Accountancy and The Institute of Accounting and Finance, Shanghai University of Finance and Economics, China

ARTICLE INFO

ABSTRACT

Article history: Received 12 October 2015 Accepted 1 November 2016 Available online xxxx

Keywords:

Independent directors Political connections Political role

We propose that independent directors are likely to play a political role in an institutional setting featuring weak investor protection and strong government intervention. Using Chinese data for 2001-2014, we investigate whether the stock price reaction to an independent director's death is related to his or her political connection. We find a stronger negative reaction to the death announcements of politically connected than politically unconnected independent directors. The magnitude of the reaction is positively related to directors' political rank, and cannot be offset by their firms' political capital. However, we document no evidence of politically connected independent directors exhibiting superior monitoring or consulting performance. Additional analyses show that firms are more inclined to appoint a politically connected independent director as a successor if his or her predecessor also had political connections. Our study broadens scholarly understanding of independent directors' role in a non-Anglo-American setting.

© 2016 Sun Yat-sen University. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecom-

mons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Independent directors are considered to constitute an important corporate governance mechanism by academics, practitioners and policymakers alike. The traditional view is that they enhance firm value by playing a monitoring or consulting role, although empirical tests do not unanimously support that view, partly due to the endogeneity problem (Adams et al., 2010). Another view is that independent directors are little more than window dressing, offering little real value, particularly when they fall into the vortex of public opinion during

* Corresponding author. E-mail address: zqli@mail.shufe.edu.cn (Z. Li).

http://dx.doi.org/10.1016/j.cjar.2016.11.001

1755-3091/© 2016 Sun Yat-sen University. Production and hosting by Elsevier B.V.

This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

corporate scandals. Hence, the issue of whether independent directors can play a role, and if so what role they play, is by no means settled. Further investigation is needed.

Scholars have increasingly begun to recognize the importance of the institutional environment in shaping the role of independent directors1 (Liu et al., 2012, 2015). As firms face different constraints and opportunities in different institutional settings, the roles that they need independent directors to play and the roles that such directors can play are also likely to differ across settings. However, the extant literature focuses narrowly on independent directors' monitoring and consulting role in the context of developed markets. Little attention has been paid to their role in emerging markets, even though the institutional characteristics of those markets are likely to shape that role in significant ways. In this paper, we fill this research gap by proposing and empirically testing an alternative view. We posit that in addition to their monitoring and consulting role, independent directors can also play a political role in China, an emerging market setting characterized by weak investor protection and strong government intervention, and that firms in China need them to play such a role.

China offers weak legal protection for investors. To gain private benefits of control, the largest shareholders in a firm are often able to manipulate the selection process of independent director candidates and dismiss those incumbent who are "unfriendly" and thus likely to monitor them strictly, thereby impeding the monitoring role of independent directors (Liu et al., 2013; Tang et al., 2010). China is also characterized by strong government intervention, which means that firms' business strategies are severely affected by frequent policy changes and political disturbances, thereby weakening the consulting role of independent directors. At the same time, the regulatory power underscoring government intervention creates strong demand for rentseeking. Independent directors with political connections can build a bridge for communication between firms and politicians, thereby helping firms to avoid policy risks and gain regulatory rent, through which they can play a crucial political role. In this context, a firm is likely to view politically connected independent directors primarily as a rent-seeking tool, with their potential monitoring or consulting role having little relevance.

To mitigate concern about the potential endogeneity problem, we empirically test the foregoing proposition by considering independent directors' death events from 2001 to 2014. Using this quasi-natural experiment, we investigate two issues. First, we compare the stock price reactions to the death announcements of politically connected and politically unconnected independent directors, allowing us to determine whether there is a political role that is restricted to the former. Second, we investigate whether firms are more inclined to appoint a politically connected independent director if his or her predecessor also had political connections. The China Securities Regulatory Commission (CSRC) requires that the proportion of independent directors on a board cannot be less than one-third, and most firms thus have to appoint a new such director to meet this regulatory requirement in the event of a director's death, allowing us to confirm firms' need for independent directors to play a political role.

The empirical results reveal a stronger negative stock price reaction to the death announcements of politically connected independent directors than to those of their politically unconnected counterparts. Further, the magnitude of the negative market reaction is positively related to the deceased director's political rank and cannot be offset by his or her firm's political capital, regardless of whether that capital takes the form of state-owned property rights or chairperson or CEO political connections. However, we document no evidence to indicate that politically connected independent directors perform better in their monitoring or consulting roles than their politically unconnected counterparts. Finally, in our test of the selection of successors, we find that firms are more inclined to appoint a new politically connected independent director if his or her predecessor had political connections, thus suggesting that politically connected independent directors are not randomly distributed and that firms need such directors to play a political role and actively select those capable of doing so.

This study contributes to the literature in several respects. First, it broadens the traditional perspective on the roles of independent directors. Most previous research focuses on their monitoring and consulting role, whereas this paper proposes and tests their political role by taking into account distinct features of China's institutional environment. Our findings emphasize the need for researchers to pay closer attention to the char-

1 Liu et al. (2012) document the consulting role played by independent directors with a banking background in China, an institutional setting featuring financial repression. Liu et al. (2015) investigate the consulting role of nonlocal independent directors in nonlocal merger and acquisition activity in the context of a Chinese setting characterized by local protectionism.

acteristics of different institutional contexts in which independent directors may play different roles. Second, this study complements the literature on political connections. Most extant studies in this area explore the causes and consequences of the political connections of a firm's core executives, overlooking the potentially important role of independent directors' political connections. In practice, firms expand their political capital not only by seeking core executives with political titles, but also by appointing politically connected individuals as independent directors. A recent study shows that the latter's political connection is even more powerful and valuable than those of core executives (Du et al., 2014). Researchers would thus be advised to pay greater attention to the political connections of independent directors. Third, in terms of research methodology, by considering independent directors' death events as the research setting and utilizing the CSRC's "one-third" regulatory requirement concerning the proportion of independent directors on the board, our study effectively mitigates the endogeneity problem that is a major concern in the independent director literature.

This study also has important policy implications for regulatory bodies looking to further improve the independent director system. Although the CSRC introduced the current system in 2001 to improve corporate governance and strengthen investor protection, these policy objectives are far from being fulfilled in practice. Politically connected independent directors have been captured by firms as rent-seeking tools, which suggests that regulators need to restrict the qualifications of independent directors and establish related support systems (e.g., granting small shareholders more nomination rights concerning independent directors) to help independent directors to play more effective monitoring and consulting roles, and in turn to strengthen corporate governance.

The remainder of this paper is organized as follows. Section 2 reviews the literature. Section 3 describes the institutional background and theoretical analyses, and Section 4 details the research design. A statistical summary and empirical tests are presented in Section 5, and Section 6 concludes the paper.

2. Literature review

2.1. Research on independent directors

Extant studies examine the monitoring and consulting roles of independent directors from the governance perspective. For example, Adams et al. (2010) propose that independent directors play a monitoring role by appointing, assessing and dismissing top executives, thereby reducing agency costs between management and shareholders. Borokhovich et al. (1996) find that independent directors are more inclined to hire CEOs from the outside rather than promote them internally. Core et al. (1999) show that independent directors can prevent CEOs from receiving excess compensation. Weisbach (1988) reports that independent directors can strengthen the negative relationship between CEO turnover and firm performance by dismissing unqualified CEOs. However, all of this research is carried out in developed markets, in which conflicts of interest between management and shareholders are the dominant concern. In emerging markets, where those conflicts are between large and small shareholders, independent directors' monitoring role operates primarily by curbing the expropriation of minority shareholders by large shareholders. Ye et al. (2007) find that independent directors can reduce fund embezzlement by large shareholders. They also show that independent directors are likely to vote against management proposals only when firms are performing poorly, although the proportion of independent directors on the board is not related to such dissent (Ye et al., 2011).

Another strand of the independent director literature focuses on these directors' consulting role. Demb and Neubauer's (1992) survey indicates that the majority of independent directors see themselves as playing an advisory role, i.e., providing professional suggestions for corporate strategy-building. Adams and Ferreira (2007) argue that independent directors' consulting role depends on inside information on the firm, and thus greater independence decreases a CEO's willingness to share information, thereby weakening that role. Coles et al. (2008) find complex firms, which have greater advising requirements than less complex firms, to have larger boards with more independent directors. Finally, Liu et al. (2015) report that nonlocal independent directors play a consulting role in nonlocal mergers and acquisitions.

Although extant studies have documented considerable evidence on the roles that independent directors play, there are some notable gaps. First, most studies focus on developed markets, thus ignoring the specific role of independent directors in emerging markets. As Williamson (2000) argues, the country-level institu-

tional environment determines and shapes corporate governance practices at the firm level. Hence, the institutional environments in which firms operate influence the role of corporate governance mechanisms. For example, audit firms play a monitoring role in developed markets. However, in China, firms hire politically connected audit firms whose partners are members of the CSRC's IPO screening committee to help them to lobby for favorable regulatory decisions (Yang, 2013). Accordingly, audit firms' role is more akin to a facilitator than a watchdog. There is clearly a need for independent director research in emerging markets to pay close attention to the specific institutional features that differ from those in developed markets, and yet work in this area remains scant. One exception is Du et al. (2014), who document the positive influence of politically connected independent directors on private firms' entry into high-barrier industries in China, an institutional context with many restrictions on the scope and depth of private economic development because of "implicit discrimination."

Second, the endogeneity problem impairs the internal validity of much independent director research, resulting in potentially erroneous conclusions. For example, it remains unclear whether independent directors add firm value. Rosenstein and Wyatt (1990) and Wang et al. (2006) find a positive relationship between the two, whereas Hermalin and Weisbach (1991) and Klein (1998) find no relationship between them, and Agrawal and Knoeber (1996) document a negative relationship. One important reason for these mixed results is that most research has failed to resolve the endogeneity in the independent director selection process. There may be unobserved firm characteristics (omitted variables) that determine who firms appoint independent directors and what role they play, and the same characteristics may influence firm value (Coles et al., 2008). Therefore, we need a quasi-experimental setting to identify the role of independent directors and its influence on firm value.

This study fills the gap in the literature and offers an improved research design. Specifically, we examine the specific institutional features (i.e., weak investor protection and strong government intervention) responsible for the different roles that independent directors play in China relative to their counterparts in developed economies. Besides, we consider a death event setting, which mitigates the endogeneity concern and allows us to establish the causal effect of politically connected independent directors on firm value and firm demand for such directors.

2.2. Research on politically connected independent directors

In the independent director literature, only a small number of studies investigate the causes and consequences of politically connected directors. Agrawal and Knoeber (2001) find that such directors are more prevalent in firms that make sales to the government, are export-oriented and engage in lobbying. Goldman et al. (2009) document a positive abnormal stock return following the announcement of the nomination of a politically connected individual to the board. They also report that when the Republican Party won the 2000 U.S. presidential election, firms connected to that party increased in value, whereas those connected to the Democratic Party decreased in value. In the Chinese setting, Du et al. (2014) show politically connected independent directors to exert a positive effect on private firms' entry into high-barrier industries.

These studies contribute useful insights to the literature on politically connected independent directors, but most conduct cross-sectional tests, and are thus subject to the endogeneity concern. Even though try to identify the casual effect of directors' political connections by utilizing a shock event of the U.S. presidential election, it also faces the challenge of some alternative explanations. For example, recent studies show politicians to have different ideology and policy orientations, and hence the outcomes of elections may affect some firms positively and others negatively (Julio and Yook, 2012; Pastor and Veronesi, 2012). Fisman et al. (2012) considered a setting clearly unrelated to policy change, former U.S. Vice-President Cheney's heart attacks, and found connections with Cheney to suffer little change in value during the event window.

The independent director death setting considered in this paper is unrelated to firm characteristics or policy change, thereby allowing us to mitigate the endogeneity problem in previous research and identify the influence of independent directors' political connections on firm value more clearly.

2.3. Research using death events

Johnson et al. (1985) constitute the seminal work using death events as a quasi-experiment to study corporate governance issues. They utilize the sudden deaths of 53 top executives to analyze the effects of management characteristics on firm value. In a subsequent paper, Slovin and Sushka (1993) analyze how ownership concentration affects firm value and control by examining the effects of the deaths of 85 inside blockholders. Hayes and Schaefer (1999) explore the influence of managers' ability on firm value by using the deaths of 29 CEOs. Nguyen and Nielsen (2010) investigate the contributions of outside directors' independence to shareholder value by examining stock price reactions to the sudden deaths of 108 directors.

The political connection research also draws on quasi-experimental death-event methodology. For example, Fisman (2001) examines the stock price reactions of firms connected to former Indonesian president Suharto when rumors of his deteriorating health spread to the market. Faccio and Parsley (2009) analyze the contribution of political connections to firm value using the deaths of 192 politicians in 35 countries. It is worth noting that the death-event setting helps researchers to identify the causal effect of certain factors on the dependent variables with few concerns over potential endogeneity. However, because death events are rare, sample sizes are often limited in this strand of the literature. This study is no exception in that regard. However, although not perfect this methodology at least provides a solution to the tricky problem of causality identification.

3. Institutional background and theoretical analysis

La Porta et al. (1998) state that due to the weak legal protection for investors' rights in code-law countries, ownership concentration is an alternative mechanism for property rights protection. Such institutional divergence is accompanied by significant differences in corporate governance practices. For example, the primary conflict of interest in code-law countries is between large and small shareholders rather than between management and shareholders.

China is typical in this respect. The country lacks a civil law tradition,2 and it still lags far behind developed countries in the establishment and enforcement of laws governing property rights protection. China has yet to introduce any mechanism for shareholder class action. Further, judicial independence and enforcement in civil infringement cases are also hindered by the government (Chen et al., 2009). As a result, the large shareholders of listed firms regularly expropriate the interests of small shareholders through fund embezzlement, related party transactions, mergers and acquisitions and other means (Li et al., 2004, 2005). Although the CSRC introduced an independent director system to the Chinese capital market in 2001, requiring that at least one-third of the boards of all listed firms be made up of independent directors to strengthen investor protection, large shareholders are still able to gain and retain private benefits of control by adjusting the board structure to restrict the number of independent directors (Liu et al., 2013). In addition, large shareholders can dismiss independent directors who vote against their proposals (Tang et al., 2010). For these reasons, scholars have increasingly begun to recognize that the weak legal protection of property rights in China seriously hinders independent directors from playing a monitoring role.

The Chinese business environment is also characterized by strong government intervention. To achieve the multiple objectives of fiscal revenue, employment, social stability and political promotion, Chinese local governments actively intervene in economic activities, resulting in frequent interruptions and distortion of firm production and operating activities. For example, Pan et al. (2008) document serious intervention in firms' merger and acquisition activities by local governments. Xu et al. (2013) show that firms' investment behavior is influenced by the political cycle of government official turnover. Xue et al. (2013) discuss a case in which power-rationing was forced upon firms by the local government as part of an energy-savings and emission reduction initiative. In such an uncertain environment characterized by frequent government interference in

2 China has not traditionally been a society ruled by law. Chinese law focuses on criminal law, and the main task of the law is not to protect the interests of individuals or corporations, particularly their economic interests, against incursions by other individuals or corporations (Bodde and Morris, 1973).

firms' strategy implementation, independent directors' consulting role with regard to strategy-building is also weakened.

At the same time, strong government intervention also means that local governments have considerable discretion in the approval and regulation of various types of commercial behavior, which can benefit or harm firms' interests. They also have numerous policy instruments at their disposal, including industrial policies, lands, quotas, credits, taxes and subsidies, through which to exert a significant impact on resource allocation, thereby directly affecting firms' survival and development. Consequently, firms have a strong motivation to build a close relationship with the government to avoid policy risks and obtain regulatory rents.

In these conditions, independent directors with political connections can easily establish a communication bridge between firms and government officials. They can also help firms to avoid policy risks and obtain regulatory rents by exploiting their influence with the government. This kind of political role3 is of course of great value for the firm, and cannot be acquired by hiring an independent director without political connections. Du et al. (2014) provide supporting evidence to show that politically connected independent directors can help firms to enter high-barrier industries and that the positive effect of their connections is even stronger than that of top executives.

Based on the foregoing analysis, we posit the following expectation. If politically connected independent directors play a political role that is not available to their politically unconnected counterparts, a firm's stock price will react more negatively to the announcement of the death of a politically connected independent director than to that of his or her unconnected counterpart. In addition, if firms need independent directors to play a political role, firms are more likely to appoint an independent director with political connections if his or her predecessor was also politically connected. Stated formally, we propose the two following hypotheses.

H1. Stock prices react more negatively to politically connected independent directors' death announcements than to those of their politically unconnected counterparts.

H2. Firms are more likely to appoint a politically connected independent director as a successor if the deceased director also had political connections.

The logic of our hypothesis development is illustrated in Fig. 1. 4. Research design

4.1. Sample and data

We first search the "Corporate Announcement" section of the WIND database using keywords carrying the meaning of death from 2001 to the end of 2014,4 obtaining 251 announcements of top executive deaths. After deleting announcements of the deaths of executive directors, managers and supervisors, we obtain 55 death announcements for 46 independent directors. Panel A of Table 1 shows that the time span is large (ranging from 2003 to 2014), with the distribution of deaths varying across the years (ranging from one death to 10 deaths in a year). By reading through the relevant firm announcements and media reports, we collect and analyze information on the causes of death among the 46 independent directors. Panel B shows the majority to have died of illness, with the remainder (15%) dying in traffic accidents, being murdered or dying suddenly from disease. Panel C describes the directorships they held. Most held only one directorship, although five held two and one held five.

We obtain stock price and financial data from the China Stock Market and Accounting Research database, and independent directors' personal attributes from the WIND database. We hand-collect missing data from firms' annual reports and the online Baidu Encyclopedia.

3 We use "political role" as a neutral term with no relevance to value judgments, as in Agrawal and Knoeber (2001). It includes but is not limited to rent-seeking behavior, and hence is not necessarily illegal.

4 These keywords include siwang, qushi, shishi, cishi, wanggu, bingshi, siqu, guoshi and lishi (Chinese phonetic alphabet).

Figure 1. Logic chain of hypothesis development.

Table 1

Distribution of independent directors' death year, death causes and directorships.

N (person)

Percent (%)

N (directorship)

Percent (%)

Panel A: Year distribution of independent directors' deaths

2003 1

2006 1

2007 1

2008 6

2009 1

2010 6

2011 10

2012 2

2013 10

2014 8 Total 46

Panel B: Cause of death

Traffic accident

Murder

Sudden disease

Illness

Panel C: Directorships 1 2 5

2.17 2.17 2.17 13.04 2.17 13.04 21.74 4.35 21.74 17.39 100.00

4.35 2.17 8.7 84.78 100.00

86.96 10.87 2.17 100.000

6 1 7 11 2 12 13 55

2 2 4 47 55

1.82 1.82 1.82 10.91 1.82 12.73 20 3.64 21.82 23.64 100.00

3.64 3.64 7.27 85.45 100.00

4.2. Model specification and variable definitions

Following Nguyen and Nielsen (2010), we construct model (1) to test hypothesis 1:

car = a0 + a1pc + a2edu + afemale + a4age + a5tenure + a6difprov + a7 bank + a8law + a9acc

+ a10cstgy + a 11cjaudt + a12c_nomn + a12c_comp + s. (1)

In model (1), the dependent variable is car (cumulated abnormal return), which is the market-adjusted return calculated as the difference between the firm return and market return.5 In the data collection process, we find that some firms did not receive notice of an independent director's death from his or her family, and thus did not release that information, until several days after his or her death (the mean [median] delay is 3.87 [3] days). To fully cover the market reaction to the death news, we set the event window to (—5, +5) around the announcement date.6

With reference to Li et al. (2006) and Xu et al. (2013), we define politically connected independent directors as those with work experience in the party, government or military hierarchies and/or membership in the People's Congress or Chinese People's Political Consultative Conference (CPPCC) (the dummy variable pc is equal to 1). We expect the sign of pc to be negative, suggesting that the stock price reacts more negatively to the death announcement of a politically connected independent director than to that of politically unconnected counterparts.

In addition, model (1) also includes a number of control variables measuring the human capital of independent directors, including education (edu), banking experience (bank), law background (law) and accounting background (acc), with their signs expected to be negative. Also, because female independent directors are known to monitor management more effectively (Adams and Ferreira, 2009), we control for female, and expect its sign to be negative. Because the young are more likely to be concerned about reputation than the old, we control for age, and expect its sign also to be negative. In addition, we control for tenure, and predict its sign to be positive, because an independent director with a shorter tenure is expected to be more independent. Also, because nonlocal independent directors (difprov) can play an advisory role but may monitor management less effectively due to a lack of local information (Liu et al., 2015), we control for difprov, but make no prediction concerning its sign. Finally, because crucial board functions provide a valuable service to shareholders, we control for membership of strategy (c_stgy), auditing (c_audt), nomination (c_nomn) and compensation (c_comp) committees, and expect their signs to be negative. Table 2 reports the variable definitions.

We construct model (2) to test hypothesis 2:

newpc1 = b0 + b1pc + b2size + b3roa + b4board + b4independence + i, (2)

where the dependent variable newpcl is equal to 1 if the successor independent director is politically connected (i.e., has work experience in the party, government or military hierarchies or membership in the People's Congress or CPPCC). We expect the sign of newpcl to be positive. With reference to Agrawal and Knoeber (1996) and Coles et al. (2008), we also control for such firm characteristics as firm size (size) and performance (roa) and such governance variables as board size (board) and board independence (independence).

5. Empirical analysis

5.1. Descriptive statistics and univariate tests

Table 3 reports the descriptive statistics. To avoid the influence of outliers, we winsorize the top and bottom 1% of each continuous variable. Table 3 shows the mean (median) of the (—5, +5) cumulated abnormal return to be —2.1% (—1.7%), and significantly negative at the 10% level. The average education is above

5 In a robustness check, we calculate car by the capital asset pricing model (CAPM), but the results remain the same.

6 In a robustness check, we reset the event window to (—3, +3) around the announcement date. The results remain the same, although the magnitude of car decreases.

Table 2

Variable definitions.

Variable Definition

car Market-adjusted return calculated as the difference between the firm return and market return over the (—5, +5) days

around the announcement date

pc A dummy variable equal to 1 if the independent director has work experience in the party, government or military

hierarchies or is a member of the People's Congress or CPPCC, and to 0 otherwise edu A discrete variable equal to 0 if the independent director holds a college degree or below, to 1 if he or she holds a

Bachelor's degree and to 2 if he or she holds a Master's degree or higher female A dummy variable equal to 1 if the independent director is female, and to 0 otherwise

age Age of independent director (in natural logarithm form when entered into the regression)

tenure Tenure of independent director (in the form of the natural logarithm of (1 + tenure) when entered into the regression)

difprov A dummy variable equal to 1 if the independent director is nonlocal, and to 0 otherwise

bank A dummy variable equal to 1 if the independent director has banking experience, and to 0 otherwise

law A dummy variable equal to 1 if the independent director has law background and zero otherwise

acc A dummy variable equal to 1 if the independent director has an accounting background, and to 0 otherwise

c_stgy A dummy variable equal to 1 if the independent director serves on the strategy committee, and to 0 otherwise

c_audt A dummy variable equal to 1 if the independent director serves on the auditing committee, and to 0 otherwise

c_nomn A dummy variable equal to 1 if the independent director serves on the nomination committee, and zero otherwise

c_comp A dummy variable equal to 1 if the independent director serves on the compensation committee and to 0 otherwise

soe A dummy variable equal to 1 if the listed firm if ultimately controlled by the government

ceopc A dummy variable equal to 1 if the firm' chairperson or CEO has work experience in the party, government or military

hierarchies or is a member of the People's Congress or CPPCC, and to 0 otherwise absence Number of an independent director's board meeting absences in a year

otherrec Other receivables scaled by total assets

TFP Total factor productivity of the firm

turnover Asset turnover ratio calculated as sales scaled by total assets

newpc1 A dummy variable equal to 1 if the successor independent director has work experience in the party, government or

military hierarchies or is a member of the People's Congress or CPPCC, and to 0 otherwise newpc2 A dummy variable equal to 1 if the successor independent director has work experience in the party, government or

military hierarchies, is a member of the People's Congress or CPPCC or is a leader of an industry association, and to 0 otherwise

size Natural logarithm of total assets

roa Return on assets

lev Long-term loans scaled by total assets

board Number of board members

independence Proportion of independent directors on the board

meets Number of board meetings in a year (in the form of the natural logarithm of (1 + meet) when entered into the

regression)

block Largest shareholder's holding

dual A dummy variable equal to 1 if the CEO also holds the position of board chair

mgtshare Managerial shareholding

the undergraduate level. The sample includes only two female independent directors (3.6%). The average age of the sample independent directors is 60, and their mean length of tenure is four years. Twenty-seven percent are nonlocal, 3.6% have banking experience, 12.7% a law background and 21.8% an accounting background. The membership percentage of various board committee is approximately 40%.

We conduct univariate tests comparing politically connected independent directors with their politically unconnected counterparts. As shown in Table 4, the deaths of the former are associated with a stock price drop of more than 4% (mean = —4.2%, significant at the 1% level), whereas the stock price reaction to the deaths of politically unconnected directors is roughly 0 (mean = 0.3%, non-significant), and the difference between the two groups is significant at the 5% level. There is no significant between-group difference in education, age, sex, tenure or localness, but a significant difference in accounting and law background, and auditing and nomination committee membership. Although these univariate tests support hypothesis 1, to ensure that the between-group difference in the stock price reaction can be attributed to political connections, we also conduct multiple regression analysis with more control variables included.

Fig. 2 depicts the cumulated abnormal returns of the two groups during the event window.

Table 3

Descriptive statistics of full sample.

Variable N Mean Median Std.dev. Min Q1 Q3 Max

car 55 -0.021 -0.017 0.080 -0.168 -0.055 0.004 0.199

pc 55 0.491 0 0.505 0 0 1 1

edu 55 1.291 1 0.658 0 1 2 2

female 55 0.036 0 0.189 0 0 0 1

age 55 60.927 65 11.687 39 51 69 78

tenure 55 3.818 4 2.373 1 2 5 13

difprov 55 0.273 0 0.449 0 0 1 1

bank 55 0.036 0 0.189 0 0 0 1

law 55 0.127 0 0.336 0 0 0 1

acc 55 0.218 0 0.417 0 0 0 1

c_stgy 55 0.364 0 0.485 0 0 1 1

c_audt 55 0.400 0 0.494 0 0 1 1

c_nomn 55 0.418 0 0.498 0 0 1 1

c_comp 55 0.382 0 0.490 0 0 1 1

Table 4

Univariate comparison.

Variable pc =1 pc = 0 Difference of mean P-value Difference of median P-value

Mean Median Mean Median

car -0.046 -0.042 0.003 -0.007 —0.048 0.023 —0.045 0.027

edu 1.259 1 1.321 1 —0.062 0.730 —0.321 0.595

female 0.037 0 0.036 0 0.001 0.980 —0.036 0.979

age 62.037 67 59.857 64.5 2.180 0.494 7.143 0.458

tenure 3.704 3 3.929 4 —0.225 0.729 —0.929 0.371

difprov 0.185 0 0.357 0 —0.172 0.158 —0.357 0.156

bank 0.000 0 0.071 0 —0.071 0.163 —0.071 0.161

law 0.222 0 0.036 0 0.187 0.039 —0.036 0.040

acc 0.074 0 0.357 0 —0.283 0.010 —0.357 0.012

c_stgy 0.370 0 0.357 0 0.013 0.921 0.000 0.920

c_audt 0.185 0 0.607 1 —0.422 0.001 — 1.000 0.002

c_nomn 0.296 0 0.536 1 —0.239 0.074 — 1.000 0.075

c_comp 0.333 0 0.429 0 —0.095 0.477 0.000 0.471

-5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

politically connected —a--- politically unconnected

Figure 2. Stock price reaction to independent directors' death announcements.

5.2. Regression analysis of stock price reaction to independent directors' death announcements

Here, we first report the main results of multiple regression analysis, and then those of several robustness checks. Further, we also examine whether firms' political connections can offset the loss of the political connections of independent directors. Next, we analyze the potential channels through which politically connected independent directors play a political role. Finally, we examine the monitoring and consulting roles of politically connected independent directors.

5.2.1. Main results

Table 5 reports the main results of multiple regression analysis on the stock price reactions to the deaths of independent directors. In column 1, without control variables, the coefficient of pc is —0.048 and significant at the 1% level, which is consistent with the univariate comparison in Table 4, suggesting that stock prices drop more than 4.8% when the independent directors concerned are politically connected. In column 2, in which the personal attribute variables are included, the coefficient of pc becomes more strongly negative and remains significant at the 1% level. There are two implications of these results. First, the difference in the stock price reaction to politically connected and unconnected independent directors can be attributed to political connec-

Table 5

Regression results of stock price reaction to an independent director's death announcement.

Variables (1) (2) (3)

car Car car

pc -0.048** —0.079***

(-2.33) (—3.48)

govrank —0.017***

(—2.94)

edu —0.003 —0.005

(—0.22) (—0.34)

female 0.105* 0.077

(1.76) (1.41)

age —0.004 —0.011

(—0.06) (—0.15)

tenure 0.049** 0.043*

(2.45) (1.93)

bank —0.014 —0.004

(—0.49) (—0.14)

law —0.025 0.037

(—0.69) (0.78)

acc —0.047 —0.042

(—1.66) (—1.44)

difprov —0.038 —0.027

(—1.39) (—1.05)

c_stgy —0.024 —0.029

(—0.92) (—1.08)

c_audt —0.041 —0.050

(—1.02) (—1.24)

c_nomn —0.020 —0.010

(—0.60) (—0.31)

c_comp 0.028 0.016

(1.02) (0.57)

N 55 55 50

Adj. R2 0.076 0.221 0.171

All variables are defined in Table 2. The effect of the constant term is omitted. Values of robust t-statistics are in parentheses.

Indicates significance at the 1% level. Indicates significance at the 5% level. Indicates significance at the 10% level.

tions because column 2 includes other control variables. Second, after controlling for the other attributes of independent directors, the importance of political connections becomes more prominent. These results further support hypothesis 1.

With regard to the control variables, the coefficient of female is positive and marginally significant, suggesting that the market reacts more negatively to the death of male independent directors. The coefficient of tenure is positive and significant at the 5% level, consistent with the prediction that independent directors with a shorter tenure are more independent in monitoring management (Nguyen and Nielsen, 2010). The coefficient of acc is negative and marginally significant, which suggests that an accounting background matters as a form of human capital. None of the other variables is significant, suggesting they are not the main causes of the differing stock price reactions to the death announcements of the two types of independent directors.

In addition, we also examine whether the stock price decline is related to independent directors' political rank. Because high-ranking officials have more power and influence and cultivate higher-level contacts than their low-ranking counterparts, we replace the dummy variable pc with the ordinal variable gov_rank, which equals 0 as a benchmark for politically unconnected independent directors, and assigns values of 1-6 to their politically connected counterparts in accordance with their political rank7 in the order of under deputy division level, deputy division level, division level, deputy departmental level, departmental level and vice-ministerial level. We remove observations with membership in the People's Congress or CPPCC but no work experience in the party, government or military hierarchies because we cannot determine their political rank. As shown in column 3 of Table 5, the coefficient of gov_rank is —0.017 and significant at the 1% level, suggesting that the higher an independent director's political rank, the greater the stock price decline resulting from his or her death, thus providing further support for hypothesis 1.

5.2.2. Robustness checks

We conduct several robustness tests for hypothesis 1. First, if firms announce other major events in the same event window as directors' deaths, the stock price reaction will be contaminated by those confounding events. Therefore, we reexamine the results after excluding contaminated observations. Table 6, column 1 shows the magnitude and significance level of the pc coefficient to remain unchanged, suggesting that the potential noise of confounding announcements does not exert a serious effect on our main results. Given our limited sample size, we retain all observations in the following analyses to increase the test power.

Second, the stock price reaction to independent directors' deaths may depend on their importance relative to other incumbent independent directors. Hence, in column 2, we include numpc_indir, the number of incumbent independent directors with political connections, in the regression. It can be seen that the coefficient of numpc_indir is positive, but insignificant, whereas that of pc is still significantly positive at the 1% level. In column 3, we include the interaction term pc*numpc_indir, but the result does not change. Column 4 adds num-high_indir, the number of incumbent politically connected independent directors whose political rank is the same as or higher than that of the deceased. The sign, magnitude and significance remain almost the same, although the coefficient of numhigh_indir is positive but insignificant.

Third, we use CAPM to calculate the cumulated abnormal return as the dependent variable.8 As shown in column 5, the coefficient of pc is now —0.062, and significant at the 1% level.

Fourth, we exclude independent directors with multiple directorships to mitigate concern that these observations are driving our results. As shown in column 6, when the independent directors with five directorships are excluded, the coefficient of pc is —0.070 and significant at the 1% level. When those with two directorships are further excluded, the coefficient of pc becomes —0.087 and still significant at 1% (unreported). Hence, we can exclude the possibility that observations with multiple directorships are driving the results.

Fifth, when we shorten the event window from (—5, +5) to (—3, +3), as shown in column 7, the coefficient of pc becomes significantly negative at the 5% level and decreases in magnitude to —0.027, which indicates that the effects of independent directors' deaths are not fully covered in a shorter event window. Fig. 2 also shows the gap between the cumulated abnormal returns of the deaths of politically connected and unconnected

7 In our sample, the vice-ministerial level is the highest political rank.

8 The estimation window is (-220, -20).

Table 6

Robustness check of stock price reaction to an independent director's death announcement.

Variable (1) (2) (3) (4) (5) (6) (7)

car car car car car car car

Pc -0.084** -0.084*** -0.078** -0.088*** -0.062*** -0.070*** -0.027**

(-2.70) (-3.71) (-2.58) (-3.78) (-2.86) (-2.82) (-2.11)

edu 0.016 0.001 -0.000 0.003 0.003 -0.004 0.000

(0.83) (0.05) (-0.01) (0.19) (0.20) (-0.26) (0.03)

female -0.017 0.114* 0.115* 0.106* 0.088 0.061 0.049

(-0.33) (196) (1.94) (1.82) (1.52) (1.10) (1.58)

age 0.035 -0.027 -0.033 0.002 0.068 -0.028 -0.102**

(0.33) (-0.47) (-0.53) (0.04) (1.05) (-0.41) (-2.66)

tenure 0.059* 0.061*** 0.059** 0.056** 0.036* 0.047** 0.036**

(2.06) (2.91) (2.70) (2.60) (1.97) (2.05) (2.63)

bank -0.024 -0.032 -0.010 -0.050* -0.005 0.007

(-0.77) (-0.89) (-0.35) (-1.72) (-0.15) (0.22)

law -0.084 -0.029 -0.031 -0.020 -0.006 0.042 -0.034

(-1.12) (-0.85) (-0.88) (-0.56) (-0.15) (0.91) (-1.22)

acc -0.057 -0.050* -0.050* -0.051* -0.043 -0.044 -0.026

(-1.52) (-1.85) (-1.84) (-1.86) (-1.63) (-1.53) (-1.45)

difprov -0.074* -0.036 -0.037 -0.033 -0.025 -0.036 -0.042***

(-1.77) (-1.40) (-1.42) (-1.23) (-0.92) (-1.27) (-3.31)

c_stgy -0.015 -0.024 -0.024 -0.026 -0.022 -0.021 -0.016

(-0.32) (-0.90) (-0.88) (-0.97) (-0.87) (-0.75) (-1.15)

c_audt -0.100** -0.038 -0.037 -0.036 -0.022 -0.048 -0.022

(-2.17) (-0.95) (-0.90) (-0.90) (-0.57) (-1.09) (-0.96)

c_nomn 0.046 -0.022 -0.024 -0.019 -0.019 -0.007 -0.015

(0.92) (-0.67) (-0.69) (-0.56) (-0.62) (-0.18) (-0.74)

c_comp 0.041 0.021 0.023 0.016 0.013 0.018 0.023

(1.14) (0.76) (0.86) (0.60) (0.51) (0.64) (1.30)

numpc_indir 0.018 0.025

(1.24) (1.25)

pc*numpc_indir -0.010

(-0.37)

numhigh_indir 0.028

(1.39)

N 34 55 55 55 55 50 55

Adj. R2 0.192 0.234 0.216 0.229 0.164 0.140 0.195

All variables are defined in Table 2. The effect of the constant term is omitted. Values of robust t-statistics are in parentheses. Indicates significance at the 1% level. Indicates significance at the 5% level. Indicates significance at the 10% level.

independent directors to widen over time. Because the announcement date lags the death date by 3.87 days on average, a time window of (—5, +5) is appropriate.

Finally, because our death sample is small, there may be concern that the results are sensitive to outliers. To mitigate that concern, we conduct a median regression. However, the result remains almost the same, i.e., the coefficient of pc is still significantly negative at the 5% level (unreported).

The foregoing robustness checks confirm the validity of the empirical results concerning hypothesis 1.

5.2.3. Further analysis of the role of firms' political connections

An important issue worth further exploration is the relationship between the political connections of independent directors and firms. To shed light on this issue, we focus on the question of whether firms' political capital can mitigate or even entirely counteract the negative market reaction to the deaths of politically connected independent directors. To answer that question it is necessary to discuss the types of and motivations for firms' political connections. Theoretically, firms can have two types of political connections: impersonal

political connections, which are determined by the state-owned property rights of the firm, and personal political connections, which are attached to firms' core executives (e.g., chairperson or CEO). From the perspective of property rights, state-owned enterprises (SOEs) by their nature have political connections and enjoy policy favoritism with respect to land, credit, quotas and access, whereas private firms are subject to government discrimination. It thus seems plausible to draw the conclusion that independent directors' political connections are more important for private firms than for SOEs. However, once we take into account the central government's "seize big, drop small" and "SOE strategic restructuring" reform strategies, it becomes clear that property rights are endogenous, i.e., larger firms and firms in monopoly industries are more likely to be controlled by the government (Xia and Chen, 2007). These SOEs undoubtedly face higher political costs due to their monopoly status, and demand more policy favoritism to maintain that status, and hence politically connected independent directors also play an indispensable role in SOEs. It is not surprising that SOEs who command considerable political capital still need to hire high-ranking ministerial-level officials as independent directors. As a result, the question of the type of firm to which politically connected independent directors are more important remains an empirical one.

Similarly, the personal political connections attached to firms' core executives are also endogenous, depending on the strength of the firms' need for such connections. Li et al. (2006) find that entrepreneurs are most likely to establish political connections with the government when they seek to protect their property rights, avoid unfavorable regulations and/or seek preferential policies and rents. Thus, at first glance, it seems that politically connected independent directors are more important for firms whose chairpersons or CEOs have no political connections. When we take into account that whether those executives have such connections may reflect their firms' need for them, however, it is clear that politically connected independent directors are also important for these firms. Consequently, whether top executives' political connections can mitigate

Table 7

Further analysis of the role of firms' political connections.

Variable (1) (2) (3)

car car car

pc*soe -0.086**

(-2.70)

pc*(1-soe) -0.076***

(-3.32)

pc*ceopc -0.080**

(-2.57)

pc*(1-ceopc) -0.083**

(-2.11)

pc*soe*ceopc -0.088**

(-2.24)

pc*soe*(1-ceopc) -0.091*

(-1.72)

pc* (1-soe)*ceopc -0.082**

(-2.09)

pc* (1-soe)* (1-ceopc) -0.073*

(-1.88)

soe 0.017 0.017

(0.64) (0.61)

ceopc 0.011 0.011

(0.29) (0.31)

control variables yes yes yes

N 55 55 55

Adj. R2 0.188 0.187 0.126

All variables are defined in Table 2. The effect of the constant term is omitted. Values of robust t-statistics are in parentheses. *** Indicates significance at the 1% level. ** Indicates significance at the 5% level. * Indicates significance at the 10% level.

Table 8

Identification of the channels of independent directors' political role.

Firm name

Primary business

Registration location Independent director's position

Political role

TO L_H

c^ cr.

i—» .-o £ i § n

000301 Eastern Market

000612 Jiaozuo Wanfang

000686 Northeastern

Securities

000997 New Continent

002117 Donggang

002152 Guangdian Currency

002695 Huangshanghuang

600069 Yinge Investment

600094 Daming City

600120 Zhejiang Dongfang

600318 Chaodong

601390 China Railway

601965 China Qiyan

601989 China Shipbuilding

Real estate, thermal electricity, gasoline

Aluminum production

Securities brokerage, underwriting

Communications, finance, government information system Commercial note printing Automatic currency processing equipment and software Meat processing

Papermaking Real estate

Import and export trade

Cement production and sales

Infrastructure construction

Automotive technology services and detection Military trade

Wujiang, Jiangsu

Jiaozuo, Henan

Changchun, Jilin

Mawei District, Fuzhou, Fujian Jinan, Shandong Guangzhou, Guangdong Nanchang, Jiangxi

Luohe, Henan Shanghai and Fujian

Hangzhou, Zhejiang

Hefei, Anhui

Beijing

Chongqing

Beijing

Deputy Mayor of Wujiang Government

Deputy Director of Henan Environmental

Protection Bureau

Deputy Minister of Finance

Deputy Mayor of Fuzhou, Director of Mawei Development District

Vice President of Jinan Intermediate Court Director of Guangdong Electronic Machinery Department

Executive of Frozen Meat Processing Department of State Internal Trade Bureau Director of High-tech District in Luohe Deputy Secretary-general of Fujian Provincial Government

Deputy Director of Zhejiang Economic System Reform Committee

Deputy Secretary-general of Anhui Provincial Government

Director of Construction Management, Department of Ministry of Construction Director of Automobile Department of Ministry of Machinery Industry

Chief Engineer of Naval Equipment Department

Land, bank loan,

approval

Environmental

supervision

Preferential policy,

regulation

Government orders, tax benefits Legal disputes Preferential policy

Preferential policy, regulation Preferential policy Land, bank loan, approval

Regulation, approval

Preferential policy, subsidy

Preferential policy, approval, regulation Preferential policy, approval, regulation Military orders

or even offset the important role played by independent directors' political connections remains an open question.

To test these ideas, we construct the variable soe, which equals 1 if the firm is ultimately controlled by the government, and the variable (1-soe), which is equal to 1 if the firm is ultimately controlled by a private entity. We also construct ceopc, which equals 1 if the firm's chairperson or CEO has political connections, and (1-ceopc), which equals 1 if he or she has no such connections. We then add soe, ceopc and their interaction term with pc to model (1). Column 1 of Table 7 shows that the coefficients of pc*soe and pc*( 1-soe) are both significantly negative, with no significant difference between them, suggesting that independent directors' political connections are equally important for SOEs and private firms. Column 2 reports the coefficients of pc*ceopc andpc*(1-ceopc) both to be significantly negative, with no significant difference between them, indicating that independent directors' political connections are important for firms regardless of whether their top executives also have political connections. Finally, in column 3, we take into account both impersonal and personal political connections. It can be seen that the coefficients of the interaction term are all significantly negative, with the difference between any two non-significant, which suggests that a firm's political connections, regardless of the form they take, cannot offset the role played by those of independent directors.

5.2.4. Channels through which politically connected independent directors play a political role

In the previous section, we propose that politically connected directors can play a political role by exploiting their contacts and influence in the government. Here, we further investigate the channels through which they play such a role. However, due to sample size limitations, we can make only preliminary analysis based on information on firms' primary business and registered location. Table 8 reports such information, from which we summarize three characteristics.9

First, independent directors can play a political role through a rich variety of channels. Because firms face different situations, the specific political roles they need independent directors to play are accordingly different. As shown in Table 8, they can play such a role by helping firms to avoid environmental regulations, resolve legal disputes, facilitate government approval, acquire land and other resources, and seek preferential policies, tax benefits and subsidies.

Second, firms precisely target independent director candidates who can meet specific needs rather than hire them blindly. For example, firms in the real estate sector require land, capital and approval, and thus seek out government officials such as vice mayors and secretary generals who have relatively comprehensive functions in these aspects. Similarly, because firms in heavily polluting industries face severe regulations, they are inclined to hire government officials who are in charge of environmental supervision. If firms want to obtain government or military orders, they will seek out government officials or military leaders who can facilitate such deals.

Finally, firms appoint their politically connected independent directors considering and following the territorial jurisdiction and industry jurisdiction principle. As shown in Table 8, the office location of most politically connected independent directors is the same as their firms' registered location or main business location, which can be explained by China's territorial jurisdiction principle, i.e., politically connected independent directors have a say only in the territory in which their office is located. When a firm's business is affected by industry policies more than regional policies, it is likely to turn to an official in relevant state ministries or commissions in charge of the corresponding industry, following the principle of industry jurisdiction.

Again, the foregoing discussion constitutes only a preliminary identification of the possible channels through which politically connected independent directors play a political role. We look forward to future research engaging in a more systematic examination of this issue.

5.2.5. Analysis of the monitoring and consulting performance of politically connected independent directors

We also examine the monitoring and consulting performance of politically connected independent directors

to determine whether the negative stock price reaction to their deaths is the result of their better such performance relative to that of their politically unconnected counterparts. Although the behavior of independent

9 For academic discussion alone, we do not disclose the names of independent directors.

Table 9

Analysis of monitoring and consulting performance of politically connected independent directors.

Variable (1) (2) (3) (4)

absence otherrec TFP turnover

Pc 0.802* 0.646* 0.091 0.136

(1.86) (1.84) (1.41) (0.97)

control variables yes yes yes yes

N 109 107 97 105

Adj./Pseudo R2 0.097 0.108 0.076 0.092

All variables are defined in Table 2. The effect of the constant term is omitted. Values of robust t-statistics are in parentheses. «, Indicates significance at the 1% level. ™ Indicates significance at the 5% level. Indicates significance at the 10% level.

directors cannot be observed directly, several variables help us to evaluate their monitoring and consulting performance. For example, if politically connected independent directors take part in firms' affairs diligently, they will actively attend board meetings. If they monitor their firms strictly, large shareholders' expropriation of the interests of minority shareholders will be more effectively constrained. If they provide effective consultation, their firms will display greater productivity and operational efficiency. If we fail to observe these phenomena, or observe their opposite, then we cannot conclude that politically connected independent directors do a better job of monitoring and consulting than their unconnected counterparts, and thus cannot attribute the negative stock price reactions following their deaths to their monitoring or consulting performance rather than to their political connections.

Accordingly, we construct models (3)-(6) to examine whether politically connected independent directors play a superior monitoring or consulting role by examining their board meeting attendance, fund embezzlement by related parties and firm productivity and operating efficiency:

absence = h0 + h1pc + d2meets + d3difprov + i (3)

otherrec = y0 + y1pc + y2size + y3 lev + y4 roa + y5soe + y6 block + m (4)

TFP = k0 + k1pc + k2 size + k3lev + k4soe + k5block + k6dual + k7mgtshare + f (5)

turnover = p0 + p1size + p2lev + p3 soe + p4block + p5dual + p6mgtshare + h (6)

Model (3) examines whether politically connected independent directors are absent from board meetings less often than their politically unconnected counterparts in the three-year window prior to their deaths. The dependent variable absence is defined as the number of absences in each year. Because independent directors should attend board meetings regardless of their firm's condition, we do not include firm characteristics as control variables. However, because the likelihood of absences is greater when the board has numerous meetings and/or the directors are nonlocal, we control for the number of board meetings (meets) and the nonlo-calness of independent directors (difprov). Because absence is a discrete variable, we use a negative binomial model to run the regression.

Model (4) examines whether politically connected independent directors can deter large shareholders from embezzling funds more effectively than their unconnected counterparts. Following Ye et al. (2007) and Jiang et al. (2010), we include firm size (size), leverage (lev), operating performance (roa), property rights (soe) and largest shareholder ownership (block) as control variables.

Finally, models (5) and (6) test whether politically connected independent directors provide more valuable advice that improves firms' productivity and operating efficiency relative to independent directors without such connections. The dependent variables are firms' total factor productivity (TFP)10 and asset turnover ratio

10 Following Giannetti et al. (2015), we regress the natural logarithm of sales on the natural logarithm of total assets, natural logarithm of the total number of employees and natural logarithm of cash payments for raw materials and services by year and industry. A firm's TFP is computed as the residual of this regression. If the number of firms in a given industry in a given year is fewer than 20, we delete these observations.

(turnover). With reference to Giannetti et al. (2015), we control for such firm characteristics as size (size) and leverage (lev) and such governance variables as property rights (soe), largest shareholder holding (block), CEO duality (dual) and managerial ownership (mgtshare).

Table 9 reports the regression results for all four models. To save space, we report only the coefficient ofpc. The significantly positive coefficient of pc in column 1 suggests that politically connected independent directors are more likely to be absent from board meetings and less likely to diligently attend to firm affairs than their politically unconnected counterparts. Column 2 also shows that pc is also significantly positive, indicating that large shareholders embezzle more funds from firms with politically connected independent directors, and thus that these directors deter expropriation less effectively than their unconnected counterparts. Columns 3 and 4 report a positive but insignificant coefficient for pc, suggesting that politically connected independent directors do not provide superior advice that aids firms' decision-making and operations.

To summarize, there is no evidence showing politically connected independent directors to exhibit better monitoring or consulting performance than their politically unconnected counterparts. Therefore, to some extent, we can rule out the alternative explanation that the more negative stock price reaction to the former's death is the result of their more effective monitoring or consulting role. The foregoing results thus lend support

Table 10

Analysis of the political connections of successor independent directors.

Variable (1) newpcl (2) newpcl (3) newpcl (4) newpcl (5) newpc2 (6) newpc2 (7) newpc2 (8) newpc2

pc 0.885* 1.368***

(1.66) (2.60)

pc*soe 0.744 0.741

(1.02) (0.90)

pc*(1-soe) 1.005 2.057***

(1.37) (2.79)

pc*ceopc 0.020 0.797

(0.02) (0.95)

pc*(1-ceopc) 1.400* 1.745***

(1.89) (2.58)

pc*soe*ceopc 0.225 0.673

(0.21) (0.62)

pc*soe*(1-ceopc) 0.627 0.514

(0.63) (0.47)

pc* (1-soe) *ceopc 0.988

(0.86)

pc* (1-soe) *(1-ceopc) 1.922** 2.938***

(2.06) (3.67)

size -0.278 -0.325 -0.254 -0.369 -0.371 -0.398* -0.361 -0.484*

(-1.19) (-1.24) (-1.06) (-1.41) (-1.58) (-1.65) (-1.51) (-1.92)

roa 0.408 0.298 -1.535 -0.287 -1.289 -1.405 -2.611 -1.199

(0.08) (0.05) (-0.28) (-0.05) (-0.21) (-0.25) (-0.41) (-0.21)

board 0.138 0.155 0.199 0.271 0.160 0.292* 0.219 0.396**

(0.89) (1.02) (1.18) (1.59) (0.98) (1.67) (1.21) (2.07)

independence 5.682* 5.943* 6.427* 7.317 0.089 0.764 0.678 1.050

(1.70) (1.68) (1.79) (1.56) (0.03) (0.20) (0.19) (0.27)

soe 0.302 0.457 0.078 0.201

(0.37) (0.52) (0.09) (0.23)

ceopc 0.629 0.614 0.286 0.230

(0.86) (0.83) (0.41) (0.31)

N 34 34 34 32 34 34 34 34

Pseudo R2 0.106 0.109 0.139 0.171 0.140 0.191 0.157 0.242

All variables are defined in Table 2. The effect of the constant term is omitted. Values of robust t-statistics are in parentheses.

Indicates significance at the 1% level. ** Indicates significance at the 5% level. * Indicates significance at the 10% level.

to the proposition that it is politically connected independent directors' political role that contributes to the negative market reactions to their death announcements. However, the validity of these results may be hampered by the endogeneity problem, as they are based on cross-sectional tests. It is therefore necessary to emphasize that this evidence is suggestive rather than conclusive.

5.3. Regression analysis of successors' political connections

We now examine hypothesis 2, that is, whether firms are more likely to appoint a politically connected independent director if his or her predecessor was also politically connected. Given that the CSRC requires that at least one-third of the boards of all Chinese listed firms be made up of independent directors, a firm has to appoint a new independent director to succeed the deceased independent director if it fails to meet the one-third requirement. Death events in this regulatory environment provide us with a quasi-experimental setting to explore whether politically connected directors are randomly distributed or purposefully selected and appointed by firms who need them to play a political role. If the former is the case, then the political connections of a predecessor will not predict those of his or her successor. If the latter is the case, however, they will because the firm will still need an independent director with political connections to fill the vacancy.

In collecting data on independent directors who succeeded directors who had died, we find one firm that had been delisted, 14 firms that changed their overall boards of directors and two that did not appoint successors as they were already compliant with the CSRC's one-third requirement. In these situations, it is impossible to obtain information on successors or accurately identify which successor corresponds to which predecessor. After deleting these firms, we obtain 34 successors. We next collect information on these successors' political connections. If a successor is politically connected (i.e., has work experience in the party, government or military hierarchies or is a member of the People's Congress or CPPCC), we assign newpcl a value of 1, and otherwise 0.

Table 10 reports the regression results for model (2). In column 1, the coefficient of pc is significantly positive at the 10% level, which means that there is a 29% greater likelihood of the firm appointing a politically connected independent director as a successor if the deceased predecessor was also politically connected. We can thus reject the notion of a random distribution of politically connected independent directors and support the supposition that they are intentionally selected and appointed by firms to play a political role. In columns 2-4, we further analyze the moderating role of firms' political connections, i.e., state-owned property rights and core executives' political connections. The results show that the coefficient of pc*(1-ceopc) is significantly positive in column 3 and the coefficient of pc* (1-soe) *(1-ceopc) is significantly positive in column 4,11 suggesting that the demand for politically connected independent directors is mainly concentrated in firms without political connections.

It is worth noting that a firm may fail to appoint a politically connected independent director even though it intends to make such an appointment because there is a limited supply of director candidates capable of playing the specific role required by the firm. As a result, as shown in column 1 of Table 10, the significance level of the coefficient of pc is just 10% rather than 1%. In these circumstances, firms may hire their second-best choice by seeking out the leaders of industry associations. Because most industry associations in China are closely affiliated with the government, and some even have administrative power, their leaders are often regarded as quasi-government officials. Thus, they are also able to play a political role to some extent. To analyze the possibility that these leaders may be hired as independent directors for their political role, we broaden our definition of a successor director's political connections (newpc2) to encompass the leadership of industry associations.

As reported in columns 5-8 of Table 10, when we use newpc2 as the dependent variable, the coefficients of pc increase in both magnitude and significance. For instance, in column 5, the coefficient of pc is significantly positive at the 1% level, which means that there is a 50% greater likelihood of a firm appointing a (broadly defined) politically connected independent director as a successor if his or her predecessor was also politically connected. In columns 6 and 7, the coefficients ofpc*(1-soe) andpc*(1-ceopc) are both significantly positive,

11 In column 4, pc*(1-soe)*ceopc is dropped due to the perfect predictability of the dependent variable.

whereas those of pc*soe and pc*(1-ceopc) are both positive but insignificant, indicating that the appointments of politically connected independent directors as successors are concentrated primarily in firms without political connections (i.e., private firms or firms whose core executives have no political connections). Finally, in column 8, the coefficient ofpc*(1-soe)*(1-ceopc) is significantly positive at the 1% level, which provides further confirmation.

In summary, the foregoing analysis shows that firms in China require politically connected independent directors to play a political role. When such a director dies, the firm will try to appoint another independent director who is also politically connected to succeed him or her. However, because of the limited supply of suitable candidates, a firm may seek out a quasi-government official, i.e., the leader of an industry association, as an alternative choice. Compared with his or her predecessor, such a director is expected to be less powerful and to have less influence. It is thus reasonable to expect the firm's stock price to react negatively to the death announcement of a politically connected independent director.

6. Conclusion

We propose that in addition to their traditional monitoring and consulting roles, independent directors can also play a political role in institutional contexts featuring weak investor protection and strong government intervention. Using Chinese data from 2001 to 2014, we test this proposition by investigating whether the market reaction to news of an independent director's death is related to his or her political capital. We find a more negative stock price reaction to the death announcements of politically connected independent directors than to those of their unconnected counterparts. The magnitude of the negative market reaction is positively related to the deceased director's political rank and cannot be offset by the firm's political capital. However, we find no evidence of superior monitoring or consulting performance among politically connected independent directors relative to their unconnected counterparts. Additional analysis also shows that firms are more likely to appoint a new independent director with political connections if his or her predecessor was also politically connected. Taken together, the evidence reported herein suggests that politically connected independent directors do play a political role in China and that firms value that role.

The academic contributions of this study are as follows. First, by taking into account specific institutional features of China that differ from those of developed markets, it broadens the traditional view of the role of independent directors, emphasizing that researchers need to pay closer attention to the characteristics of the institutional contexts in which independent directors operate, as they may play very different roles depending on the context. Second, the paper adds complementary evidence to the political connections literature. Most of that literature focuses on the political connections of a firm's core executives (i.e., chairperson or CEO), whereas we demonstrate that the political connections of independent directors are also of considerable importance. Third, in methodological terms, by using independent directors' death events as the research setting, the paper mitigates the endogeneity problem that is a major concern in independent director research and helps to resolve several disputes in the literature.

This study also has important policy implications for regulatory bodies seeking to improve the independent director system. Although the CSRC introduced the current system in 2001 to improve corporate governance and strengthen investor protection, these policy objectives are far from being achieved in practice. Politically connected independent directors have been captured by firms as rent-seeking tools. Therefore, the regulatory bodies need to restrict the qualifications of independent directors and establish additional rules (e.g., granting small shareholders more nomination rights over independent directors) to ensure that these directors play a more effective monitoring and consulting role.

Despite its important research and policy implications, this study also has several limitations. First, although death events help us to identify causal relationships and avoid the endogeneity problem, our sample size is also constrained by the rarity of those events. Second, although we try our best to identify some of the channels through which politically connected independent directors play a political role, sample limitations render our identification and induction highly preliminary. Hence, further research with a larger sample and more systematic examination is needed.

Acknowledgment

We would like to thank Xingqiang Du (the discussant), the anonymous referee, and participants of the CJAR annual conference (Changsha, 2015). We also acknowledge financial support from the National Natural Science Foundation of China (Nos. 71302105, 71372041, and 71572100), the MOE Project for Key Research Institutes of Humanities and Social Science (No. 13JJD790019), and the MOE Foundation for Humanities and Social Science (No. 13YJA790057). All remaining errors and omissions are our own.

References

Adams, R.B., Ferreira, D., 2007. A theory of friendly boards. J. Financ. 62 (1), 217-250.

Adams, R.B., Ferreira, D., 2009. Women in the boardroom and their impact on governance and performance. J. Financ. Econ. 94 (2), 291-309.

Adams, R.B., Hermalin, B.E., Weisbach, M.S., 2010. The role of boards of directors in corporate governance: a conceptual framework

and survey. J. Econ. Lit. 48 (1), 58-107. Agrawal, A., Knoeber, C.R., 1996. Firm performance and mechanisms to control agency problems between managers and shareholders. J.

Financ. Quant. Anal. 31 (03), 377-397. Agrawal, A., Knoeber, C.R., 2001. Do some outside directors play a political role? J. Law Econ. 44 (1), 179-198. Bodde, D., Morris, C., 1973. Law in Imperial China. University of Pennsylvania Press, Philadelphia, PA.

Borokhovich, K.A., Parrino, R., Trapani, T., 1996. Outside directors and CEO selection. J. Financ. Quant. Anal. 31 (3), 337-355. Chen, X., Li, M., Rui, M., Xia, L., 2009. Judiciary independence and the enforcement of investor protection laws: market responses to the

"1/15" notice of the Supreme People's Court of China. China Econ. Quart. 9 (1), 1-28 (in Chinese). Coles, J.L., Daniel, N.D., Naveen, L., 2008. Boards: does one size fit all? J. Financ. Econ. 87 (2), 329-356.

Core, J.E., Holthausen, R.W., Larcker, D.F., 1999. Corporate governance, chief executive officer compensation, and firm performance. J.

Financ. Econ. 51 (3), 371-406. Demb, A., Neubauer, F.F., 1992. The Corporate Board: Confronting the Paradoxes. Oxford University Press, Oxford. Du, X., Zeng, Q., Du, Y., 2014. Do politically connected independent directors help Chinese listed private firms enter high-barrier

industries? China Account. Financ. Rev. 16 (4), 90-146 (in Chinese). Faccio, M., Parsley, D.C., 2009. Sudden deaths: taking stock of geographic ties. J. Financ. Quant. Anal. 44 (3), 683-718. Fisman, D., Fisman, R.J., Galef, J., Khurana, R., Wang, Y., 2012. Estimating the value of connections to vice-president Cheney. B.E. J.

Econ. Anal. Pol. 13 (3), 1-20. Fisman, R., 2001. Estimating the value of political connections. Am. Econ. Rev. 91 (4), 1095-1102.

Giannetti, M., Liao, G., Yu, X., 2015. The brain gain of corporate boards: evidence from China. J. Financ. 70 (4), 1629-1682. Goldman, E., Rocholl, J., So, J., 2009. Do politically connected boards affect firm value? Rev. Financ. Stud. 22 (6), 2331-2360. Hayes, R.M., Schaefer, S., 1999. How much are differences in managerial ability worth? J. Account. Econ. 27 (2), 125-148. Hermalin, B.E., Weisbach, M.S., 1991. The effects of board composition and direct incentives on firm performance. Financ. Manage. 20 (4), 101-112.

Jiang, G., Lee, C.M., Yue, H., 2010. Tunneling through intercorporate loans: the China experience. J. Financ. Econ. 98 (1), 1-20. Johnson, W.B., Magee, R.P., Nagarajan, N.J., Newman, H.A., 1985. An analysis of the stock price reaction to sudden executive deaths:

implications for the managerial labor market. J. Account. Econ. 7 (1), 151-174. Julio, B., Yook, Y., 2012. Political uncertainty and corporate investment cycles. J. Financ. 67 (1), 45-83. Klein, A., 1998. Firm performance and board committee structure. J. Law Econ. 41 (1), 275-304. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. J. Polit. Econ. 106 (6), 1113-1155. Li, H., Meng, L., Zhang, J., 2006. Why do entrepreneurs enter politics? Evidence from China. Econ. Inq. 44 (3), 559-578. Li, Z., Sun, Z., Wang, Z., 2004. Tunneling and ownership structure of a firm: evidence from controlling shareholder's embezzlement of

listed company's funds in China. Account. Res. 12, 3-13 (in Chinese). Li, Z., Yu, Q., Wang, X., 2005. Tunneling, propping and M&A: evidence from Chinese listed companies. Econ. Res. J. 1, 95-105 (in Chinese).

Liu, C., Li, S., Sun, L., 2015. Do independent directors have the consulting function? An empirical study on the function of independent

directors in strange land. Manage. World 3, 124-136 (in Chinese). Liu, H., Tang, S., Lou, J., 2012. Independent directors: monitoring or consulting? A study on the impact of independent directors' banking

background on firms' loan financing. Manage. World 1, 141-156 (in Chinese). Liu, H., Zhao, L., Liu, F., 2013. Private benefits of control and dynamics of board structure adjustment: evidence from the CSRC's

compulsory introduction of independent directors. China Account. Financ. Rev. 15 (3), 1-66 (in Chinese). Nguyen, B.D., Nielsen, K.M., 2010. The value of independent directors: evidence from sudden deaths. J. Financ. Econ. 98 (3), 550-567. Pan, H., Xia, X., Yu, M., 2008. Government intervention, political connections and the mergers of local government-controlled

enterprises. Econ. Res. J. 4, 41-52 (in Chinese). Pastor, L., Veronesi, P., 2012. Uncertainty about government policy and stock prices. J. Financ. 67 (4), 1219-1264. Rosenstein, S., Wyatt, J.G., 1990. Outside directors, board independence, and shareholder wealth. J. Financ. Econ. 26 (2), 175-191.

Slovin, M.B., Sushka, M.E., 1993. Ownership concentration, corporate control activity, and firm value: evidence from the death of inside

blockholders. J. Financ. 48 (4), 1293-1321. Tang, X., Shen, H., Du, J., 2010. The monitoring incentive of independent directors: evidence from independent directors' opinions.

Manage. World 9, 138-149 (in Chinese). Wang, Y., Zhao, Z., Wei, X., 2006. Does independence of the board affect firm performance? Econ. Res. J. 5, 62-73 (in Chinese). Weisbach, M.S., 1988. Outside directors and CEO turnover. J. Financ. Econ. 20, 431-460.

Williamson, O.E., 2000. The new institutional economics: taking stock, looking ahead. J. Econ. Lit. 38 (3), 595-613. Xia, L., Chen, X., 2007. Marketization, SOE reform strategy, and endogenously determined corporate governance structure. Econ. Res. J. 7, 82-95 (in Chinese).

Xu, Y., Qian, X., Li, W., 2013. Political uncertainty, political connections, and private firms' investments: evidence from the turnover of

the secretaries of municipal party committees. Manage. World 5, 116-130 (in Chinese). Xue, S., Hong, Y., Chen, X., 2013. Ownership property, political connections, and power rationing of local governments: evidence from

the nonferrous metals industry. J. Financ. Res. 3, 130-142 (in Chinese). Yang, Z., 2013. Do political connections add value to audit firms? Evidence from IPO audits in China. Contemp. Account. Res. 30 (3), 891-921.

Ye, K., Lu, Z., Zhang, Z., 2007. Can independent directors deter the tunneling of large shareholders? Econ. Res. J. 4, 101-111 (in Chinese). Ye, K., Zhu, J., Lu, Z., Zhang, R., 2011. The independence of independent directors: evidence from board voting behavior. Econ. Res. J. 1, 126-139 (in Chinese).