articol hu si leung.pdf

8/15/2019 articol Hu si Leung.pdf

1/28

Top management turnover, rm performance and

government control: Evidence from China's listed

state-owned enterprises☆

Fang Hu a,⁎, Sidney C.M. Leung b

a Department of Accounting, Finance and Economics, Griffith Business School, Griffith University, Australia b Department of Accountancy, College of Business, City University of Hong Kong

Received 9 July 2010

Abstract

Using a sample of 916 Chinese listed state-owned enterprises (SOEs) from 2001 to 2005, we nd that

the likelihood of top management turnover is negatively associated with rm performance, suggesting the

existence of an effective corporate governance mechanism in an emerging economy that is highly con-

trolled by government. We also nd that the negative turnover – performance relationship is stronger when

the SOE is directly held by the central or local government, holding a monopolistic position in a local econ-

omy or in a strategic/regulated industry. The results indicate that the market-based corporate governance

mechanism that disciplines top executives as a result of poor performance is not only used in Chinese

SOEs, but is used more frequently when the governance control of SOEs is more intense. Our ndings

support the notion that government control strengthens rather than weakens the turnover – performance

governance mechanism. Our additional analysis shows that this complementary effect is stronger in

regions that lack pro-market institutions, such as investor protections and a functioning capital market.

© 2012 University of Illinois. All rights reserved.

JEL classification: M41; M48; M51

Keywords: Government control; Management turnover; Chinese SOEs; Firm performance

☆ We appreciate comments from Kelvin Lam, Grant Richardson, Bikki Jaggi, Tianyu Zhang, Qingmei Xue, Oliver

Rui, participants of the 2009 American Accounting Association annual meeting, seminar participants at the Seoul

National University, the reviewer, and participants of the 2010 Illinois International Accounting Symposium at Taipei.⁎ Corresponding author.

E-mail addresses: f.hu@grif th.edu.au (F. Hu), [email protected] (S.C.M. Leung).

0020-7063/$ - see front matter © 2012 University of Illinois. All rights reserved.

doi:10.1016/j.intacc.2012.03.006

Available online at www.sciencedirect.com

The International Journal of Accounting 47 (2012) 235 – 262

mailto:[email protected]:[email protected]:[email protected]:[email protected]://dx.doi.org/10.1016/j.intacc.2012.03.006http://dx.doi.org/10.1016/j.intacc.2012.03.006http://dx.doi.org/10.1016/j.intacc.2012.03.006http://dx.doi.org/10.1016/j.intacc.2012.03.006http://dx.doi.org/10.1016/j.intacc.2012.03.006http://dx.doi.org/10.1016/j.intacc.2012.03.006http://dx.doi.org/10.1016/j.intacc.2012.03.006http://dx.doi.org/10.1016/j.intacc.2012.03.006mailto:[email protected]:[email protected]


2/28

1. Introduction

A large body of evidence from developed markets confirms the effectiveness of corpo-

rate governance in spurring CEO turnover (Banker & Datar, 1989; Bushman, Chen, Engel,& Smith, 2004; Engel, Hayes, & Wang, 2003; Holmstrom & Milgrom, 1991; Murphy &

Zimmerman, 1993; Weisbach, 1988). However, the literature generally shows that there

are doubts over whether the corporate governance mechanisms that have been shown to

be effective in developed economies (market-based governance mechanisms) are equally

effective in state-owned enterprises (SOEs) in emerging economies because the significant

political influence that government has over such firms could lead to the emergence of a

“stakeholder ” governance model, which is characterized by a relation-based contract rather

than a market-based contract (Ball, Kothari, & Robin, 2003; Ball, Robin, & Wu, 2000). It

has been argued that SOEs have multiple objectives, including political objectives, social

goals, and bureaucrats' self interests. SOEs also have to maintain close relationships

with their major stakeholders, such as governments, banks and financial institutions, and

labor unions, thus reducing the effectiveness of market-based corporate governance mech-

anisms aligned with firm performance. Recently, some researchers argue that effective

market and governance mechanisms are an important ingredient for the success of some

emerging economies such as China (e.g., Chow, 2010).

This study examines the corporate governance of listed SOEs in China, the world's larg-

est emerging economy, which is characterized by intensive state and government control.

We address two research questions. First, is the market-based governance mechanism,

whereby top executives are punished for poor performance, effectively used in corporate businesses with intensive government control or influence? Second, and more importantly,

is the association between top management turnover and poor performance stronger or

weaker when governance control of SOEs is more intense?

These two research questions are important given the opposite views on the effects of

government control of SOEs. On the one hand, doubts exist about the effectiveness of

market-based corporate governance in an emerging economy like China that is highly con-

trolled by government (e.g., Ball et al., 2000, 2003; Bushman & Piotroski, 2006; Chan,

Lew, & Tong, 2001; Jaggi & Low, 2000; Opper & Brehm, 2007). On the other hand,

some studies suggest that government control/ownership is not necessarily less efficient

than private ownership (Caves & Christensen, 1980; Kole & Mulherin, 1997; Martin &Parker, 1995; Wortzel & Wortzel, 1989). Government control could serve as a remedy

for a lack of pro-market institutions (Frye & Shleifer, 1997; Qian, 1996; Xu, Zhu, &

Lin, 2002) and offer a helping hand to ensure corporate business activity is efficiently gov-

erned (Chang & Wong, 2004a, 2004b; La Porta, Lopez-de-Silanes, Shleifer, & Vishny,

2000; Qian, 1996; Shleifer & Vishny, 1997). The findings of this study should enrich

our understanding of whether government control and corporate governance mechanisms

are complementary or substitutive in Chinese SOEs and should provide insights into the

role of government control in corporate governance in emerging economies.

China's economy, the biggest among emerging markets, is one throughout which the

government has traditionally played an important role. SOEs account for the majority of the economy. The majority of China's listed firms are controlled by state shareholders

who retain their dominant control in the form of non-tradable state-owned shares.

236 F. Hu, S.C.M. Leung / The International Journal of Accounting 47 (2012) 235 – 262


3/28

Among the top 500 firms in China in 2007, 349 (about 70%) were SOEs. While China has

embarked on a series of reforms since the 1980s that have included privatization, corporat-

ization, and the implementation of an incentive contract system for SOE managers, the

government retains dominant control through its ownership of non-tradable state-ownedshares or through its administrative authority over important SOE decisions. 1 For instance,

decisions to appoint or remove key SOE personnel, such as the chairman of the board or

the CEO, still rest with state-controlling shareholders (Wong, Opper, & Hu, 2004).

China's SOEs provide a setting that is particularly conducive to testing the role of

government control on corporate governance because the SOEs are mainly controlled at the

jurisdictional levels and there is considerable variation in the degree of political control and

the extent of regulation across the jurisdictional regions. The large number of SOEs and the

variety in market institutions across regions allow us to perform a meaningful analysis in a

single country and capture the direct effects of government control on corporate governance.

Using a sample of 4906 firm-year observations from 916 Chinese SOEs listed on the

Shanghai and Shenzhen securities markets from 2001 to 2005, we find that CEOs or chair-

men are more likely to be replaced in firms with poor accounting performance, a result

which suggests that the market-based corporate governance mechanism is employed in

SOEs that are typically subject to government control. To test the intertwining effect of

government control on the turnover – performance relationship, we use several proxies to

measure various aspects of government control: (1) the concentration of government own-

ership, in which government control is measured through property rights; (2) the jurisdic-

tional level of government ownership, in which government control is measured by

administrative hierarchy; (3) the monopolistic position of an SOE, in which government control interests are measured because of their importance to fiscal revenue and the polit-

ical status of the local government; and (4) the strategic industry of an SOE, in which the

importance of the industry to the national economy and the extent to which it is subject to

government control through regulations are measured. We find that the negative relation-

ship between top management turnover and firm performance is significantly stronger

among SOEs subject to central or direct local government control, SOEs in a monopolistic

position, and SOEs in a strategic industry. However, we do not find that the relationship is

affected by concentrated government ownership. The overall results support the view that

the tighter the government control, the greater the degree of corporate control exercised

through the turnover – performance mechanism in China's SOEs, suggesting that thesecontrol mechanisms are complementary. Our additional analysis shows that the comple-

mentary effect of government control on the turnover – performance relationship is stronger

in firms from regions with poor investor protections and less market liberalization. The

results are consistent with the view that government control can serve as a remedy for a

lack of pro-market institutions as suggested by Xu et al. (2002) and Frye and Shleifer (1997).

This paper contributes to the existing literature in three ways. First, while most prior re-

search suggests that the corporate governance system in developed economies is driven by

1

We appreciate comments from Kelvin Lam, Grant Richardson, Bikki Jaggi, Tianyu Zhang, Qingmei Xue,Oliver Rui, participants of the 2009 American Accounting Association annual meeting, seminar participants at

the Seoul National University, the reviewer, and participants of the 2010 Illinois International Accounting

Symposium at Taipei.


http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


4/28

the pressure of the external capital market, we show that the effectiveness of the important

corporate control mechanism—management turnover decisions—could also be driven by

government control and government interests in the economic success of SOEs in China, a

country which is similar to other emerging economies in its lack of full market competitionand democratization. Our results support the view that government control is complemen-

tary to the market-based turnover – performance governance mechanism.

Second, despite criticism that the Chinese economy lacks a functioning capital market

and significant investor protection and is subject to extensive government influence and

control, China experienced remarkably rapid economic growth in the last two decades

and is expected to continue on its high growth trajectory in the next decade. Our findings

provide a partial explanation for this intriguing phenomenon. Although China's economy

lacks full market competition and pro-market institutions, our results suggest that govern-

ment control could serve as a good check-and-balance mechanism that ensures more effec-

tive corporate decisions are made and as a remedy for incomplete pro-market institutions.

Third, the findings provide a better understanding of the managerial labor market in a

transitional economy that is still subject to significant government control. It is argued

that the Chinese government selects officers and managers based on political connections

or other political considerations (the non-performance hypothesis of Opper & Brehm,

2007). Although political connections remain a focus of attention in China, the evidence

of this present study is consistent with the “ performance hypothesis” of managerial labor

in an emerging economy with strong government control. Our results suggest that observ-

able performance measures are demanded as a control mechanism in a transitional econo-

my if the government aims to establish a rule-based governance system (Chen & Wang,2007; Li, Park, & Li, 2004).

The remainder of the paper is organized as follows. Section 2 reviews the literature and

provides the background to the study. Section 3 develops hypotheses, and Section 4 pre-

sents the research design. Section 5 describes the sample and data. Section 6 reports the

results, and Section 7 documents the results of robustness tests and additional analysis.

The final section summarizes and concludes the paper.

2. Literature review and background

2.1. Management turnover and firm performance

Management turnover is an important mechanism of a firm's governance system. When

ownership and control are separated, one of the methods stakeholders use to monitor and

control management is to align the manager's incentives with firm performance. Thus,

the relationship between management turnover and firm performance is a good way of

assessing the effectiveness of a firm's governance system and has been widely discussed

in the literature (e.g., Banker & Datar, 1989; Bushman et al., 2004; Engel et al., 2003;

Holmstrom & Milgrom, 1991; Murphy & Zimmerman, 1993; Weisbach, 1988). Existing

studies, which focus on privately held firms in developed market economies, have reached

two general conclusions. First, forced management turnover is significantly negatively asso-ciated with firm accounting performance (Brickley, 2003; Coughlan & Schmidt, 1985; Engel

et al., 2003; Kaplan & Minton, 1994; Morck, Shleifer, & Vishny, 1988; Weisbach, 1988).



5/28

The negative turnover – performance relationship represents an efficient incentive mechanism

in which managers are removed because of poor performance. Second, the inverse relation

between management turnover and accounting information on firm performance is induced

by the pressure of the external capital market. Because the external capital market demandsa public, precise, and observable measure of firm efficacy, accounting information is a

useful measure in turnover or compensation decision making (Banker & Datar, 1989;

DeFond & Park, 1999; Engel et al., 2003; Holmstrom & Milgrom, 1991). However,

there is no clear evidence on the effectiveness of firm governance systems in a developing

market.

2.2. Government control and corporate governance in China's SOEs

Two different roles of government control should be considered in analyzing the objec-

tive functions of government owners in China's SOEs. The first role of government control

is to act as the largest shareholder in the SOE. The incentive problems that arise between

shareholders and managers and are most commonly discussed in the context of private

firms also arise in the governance of SOEs. In spite of their social and political goals, state

shareholders have ownership interests in the profitability of their companies (Blanchard &

Aghion, 1996; Earle, Estrin, & Leshcenko, 1996; Frydman, Pistor, & Rapaczynski, 1996;

Hellman & Schankerman, 2000). The second role government control plays is to allow the

government to exert political or regulatory influence over the SOE. Government officers or bu-

reaucrats are appointed to supervise SOEs with objectives to serve social and political interests,

such as correcting market failures and providing additional employment opportunities andsocial security to the public (Shleifer & Vishny, 1994, 1997). In addition, state ownership im-

plies that bureaucrats exercise control on behalf of the government. They are motivated by

self-interest and will thus use state-owned company resources to promote their own personal

interests (Jones, 1985; Krueger, 1990; Shleifer & Vishny, 1994, 1998). As a result, the objec-

tive function of state owners is at best a weighted average of company performance and the

attainment of political goals that often detract from company performance.

Due to the complex incentives of government and individual bureaucrats, the corporate

governance structure of SOEs reflects a mixture of various controlling methods and incen-

tive issues in which government control through state shareholdings may be employed as

an explicit control device alongside other modern corporate governance mechanisms. Most existing studies focus on the objectives of individual politicians and neglect the incentive

issues of controlling shareholders (Blanchard & Aghion, 1996; Earle et al., 1996; Frydman

et al., 1996; Hellman & Schankerman, 2000; Kato & Long, 2006). Most theoretical argu-

ments rely on the assumption that the government functions purely as a political body or

regulator that has objectives other than profitability and that these objectives encourage

the government to intervene in business activities.

Since the 1990s, the Chinese government has adopted a modern corporate governance

structure in an effort to transform Chinese SOEs into modern corporations while preserv-

ing state ownership of these enterprises. One of the modern corporate systems introduced

to Chinese SOEs is a system of incentive contracts designed to improve the firm's perfor-mance and monitor SOE managers. Performance contracts have been executed between

managers and state owners (Shirley & Xu, 2001). Such performance contracts (1) obligate



6/28

the top SOE manager to achieve a specified earnings target within a given time period and

(2) specify government rewards for achieving such goals and penalties for misconduct

(Chang & Wong, 2009; Shirley & Xu, 2001). In addition, the corporate governance struc-

ture of listed SOEs includes some major sources of political control for the benefit of stateshareholders (Chang & Wong, 2004a, 2004b). One of these is the decision-making rights

held by the government through shareholdings and control rights. A second source of po-

litical control is the authority or regulatory scrutiny exerted by government administrators

and line ministries to ensure that their shareholding benefits are maximized by political

checks and balances (Chang & Wong, 2004a, 2004b; Qian, 1996).

3. Hypothesis development

3.1. Top management turnover and firm performance in China's SOEs

As discussed in the last section, the role of government control in the governance of

SOEs is controversial. On the one hand, economists generally argue that government

control is detrimental to corporate operations because the government is obsessed with mul-

tiple objectives (the political agenda) or because there is no individual owner with strong

incentives to monitor SOE managers. Thus, the contention is that government control will

drive the company away from value maximization and will be unlikely to use market-

based governance mechanisms (Bai & Xu, 2005; Kornai, 1980; Shleifer & Vishny, 1994,

1997). Consistent with this view, Kato and Long (2006) find that relative to the listed pri-vate Chinese firms, the negative relation between CEO turnover and firm performance is

weaker for state-controlled firms. On the other hand, the government has strong incentives

as the owner of an SOE to adopt effective corporate control methods that allow it to monitor

managers on the basis of economic performance. The Chinese Central Government has em-

phasized the economic performance of Chinese SOEs since embarking on its economic re-

forms in the 1980s by promoting fiscal decentralization and yardstick competition policies.

The fiscal decentralization policy involves the decentralization of the revenue and expense

responsibilities of the Central Government to lower levels of government. This policy allows

local governments to tax local state enterprises and address social issues (Montinola, Qian, &

Weingast, 1995; Oi, 1992; Qian & Weingast, 1997). The yardstick competition policy in-volves competition among jurisdictions as regions are compared and evaluated by higher

level government bodies in accordance with standardized performance criteria, one of

which is the region's economic performance relative to its peers. The political status of the

local government or a local governor's career in the Central Government is determined by

the ranking of local economic performance (Maskin, Qian, & Xu, 2000; Qian & Xu,

1993).2 Hence, senior government officials who act on behalf of state owners and the

major SOE shareholders have strong incentives to monitor managers' performance.3

2

The Central Government rewards or punishes local of

cials on the basis of economic performance in their re-gion, which motivates them to promote the local economy ( Blanchard & Shleifer, 2001).3 CEOs of SOEs may have relatively less freedom of decisions; however, the state owner still has strong share-

holding incentives to select and monitor a CEO based on the rm's overall performance.


http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


7/28

Accordingly, government/party officials have strong incentives to encourage better corporate

earnings that promote regional economic performance and allow them to discharge their

fiscal responsibilities so that they can stand out in the political competition for career

promotion.Anecdotal evidence shows that Chinese leaders of a province or region are promoted to

a higher rank as a result of economic success in the province or region they govern. For

example, Chen Demin was the Governor of Shanxi province from 2004 to 2006 and was

promoted to become the Deputy Director of the National Development and Reform

Committee in 2006 after the GDP growth rate ranking of Shanxi rose from 18th in 2003

to first in 2006.4 In addition, poor firm performance also limits the latitude and resources

available to politicians to help them serve their own personal objectives, such as on-the-job

consumption and the accumulation of personal wealth, thus further increasing their incen-

tives to dismiss relatively poor performing managers (Chang & Wong, 2009). Based on the

fact that the state owner and government officials are most likely to act in their own self-

interest, the following hypothesis is proposed:

H1. The propensity of top management turnover in Chinese SOEs is negatively associated

with firm performance.

3.2. Government control and the association between top management turnover and firm

accounting performance

Because the government plays both a political role and a shareholder's role in state-controlled business, government control could have both benefits and costs for the corpo-

rate governance of SOEs. Strong government control could serve as a substitute for or a

complement to the governance mechanism of top manager turnover based on poor perfor-

mance. We discuss these possible effects below.

3.2.1. The substitutive effect

Stronger government control could weaken the manager turnover/performance governance

mechanism for three reasons. First, as the government looks to pursue its political and social ob-

jectives, it may rely less on the economic performance of SOE managers. Second, government

control mainly relies on information obtained through bureaucratic channels rather than on in-

formation provided by the capital market (e.g., the stock market, rating companies, invest-

ment banks, or consulting companies). Information on firm performance used to evaluate

managers would be noisier with greater government control. Third, bureaucrats from the

Party/government departments who exercise government control are not in a position and

have no incentive to appoint the best candidates to top management positions because deci-

sions on such appointments are mainly politically motivated. Hence, tight government control

is likely to result in SOEs retreating from efficient corporate control mechanisms.

4

Another example of the promotion of government of

cials based on provincial economic performance is BoXilai, the governor of Liaoning province from 2001 to 2004. He was promoted to e General Director of China's

Department of Commerce in 2004 after the nationwide GDP growth rate ranking of Liaoning province rose from

29th in 2001 to 14th in 2004.


http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


8/28

3.2.2. The complementary effect

On the other hand, strong government control could strengthen the manager turnover/

performance governance mechanism. Unlike in developed economies where there is strong

legal protection for investors, where creditors have priority of payments over shareholders,and where managers and directors can be sued for breach of fiduciary duty (La Porta et al.,

2000; Shleifer & Vishny, 1997), all of these investor safeguards are weak in a developing

economy like China's because of deficiencies in the legal system and law enforcement.

Managers are also subject to less pressure from the external managerial labor market and

face less legal risk for misconduct. Thus, managers have strong incentives to misappropri-

ate state assets rather than to achieve wealth maximization for shareholders. Given man-

agers' self-interest incentives, the exercise of tight government control by adhering to

the practice of rewarding or penalizing top managers according to firm performance serves

as an important mechanism for disciplining SOE managers and discouraging them from

engaging in behavior that reduces the amount of corporate resources over which the gov-

ernment or has stronger interests (Brada, 1996; Tang, Chow, & Lau, 1999).

In summary, the substitutive effect of government control weakens the performance-

turnover governance mechanism, and the complementary effect of government control

enhances the performance-turnover government mechanism.

The adoption of fiscal decentralization and yardstick competition policies by the

Chinese government and competition among government officials for career promotion

strongly motivate senior government officials to push for better SOE profitability and to

monitor the performance of top managers. Therefore, we expect government control has

a complementary effect and enhances the use of the market-based performance-turnover mechanism. Hence, the second hypothesis is stated as follows:

H2. The negative association between top management turnover and firm performance is

stronger in SOEs with tighter government control.

4. Research design

4.1. Measures

4.1.1. Top management turnover and firm performance

Top management turnover (Turnover ) is an indicator variable equal to 1 if either the chair-

man or CEO (or both) leaves the firm for reasons other than retirement, health, change of

ownership, or the end of a temporary appointment as disclosed in the Chinese Stock Market

and Accounting Research (CSMAR) database,5 and 0 otherwise. Firm performance is mea-

sured by return on assets lagged one period ( ROAt −1), a proxy of previous firm performance

(Chang & Wang, 1994; Engel et al., 2003; Kato & Long, 2006; Weisbach, 1988).

5 Chinese Stock Market and Accounting Research (CSMAR) database reports 12 reasons for top management turn-

over as follows: 1 = rotation; 2 = retirement; 3 = expiration of the term; 4 = change of ownership; 5 = resignation; 6 =re; 7 = health; 8 = personal reasons; 9 = improvement of corporate governance; 10 = criminalities; 12 = the end of a

temporary appointment. There is no reason 11 for top management turnover in the CSMAR database. Reasons 2, 4, 7,

and 12 are considered natural turnover and, therefore, are excluded.


http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


9/28

4.1.2. Government control

Four metrics are used to measure the extent of government control or involvement

through property rights in or regulation of SOEs. They are described as follows:

1. The concentration of government ownership: Because the attributes of ownership deter-

mine government shareholder interests in firm performance and internal managerial

efforts (Boeker, 1992; Boeker & Goodstein, 1993; Denis, Denis, & Sarin, 1997;

Salancik & Pfeffer, 1980), the extent of government control through ownership is mea-

sured by its concentration.

2. The jurisdictional level of government control : Under China's planning system, in

which blocks or regions are the foundation of the planning system (Li et al., 2004;

Maskin et al., 2000), government ownership is affiliated with the following jurisdiction-

al levels of government control—central, provincial, municipal, and county. Chinese

SOEs could be owned and controlled directly by central government and local (provin-

cial, municipal, and county) governments or indirectly through other SOEs, legal repre-

sentatives, or government-related entities. The government has a greater ability to

intervene in the firm's management in SOEs that are directly held by central and

local government than in SOEs that are indirectly controlled by the government ( Fan,

Wong, & Zhang, 2007). SOEs directly governed by the central or local government

are expected to be subject to tighter government control than those indirectly held by

the government through other SOEs or government-related entities. Thus, we expect

the association between management turnover and firm performance to be stronger in

SOEs directly held by central or local government.3. The monopolistic position of an SOE : A firm's monopolistic position is associated with

a market share advantage derived from its ability to obtain concessions, licenses, or

sizeable government contracts. The Chinese discourse on SOEs contains a number of

terms that denote the importance of certain companies and sectors in terms of their mar-

ket power, such as “ backbone enterprises,” “ pillar enterprises,” and “giant enterprises”;

SOEs described in these terms are normally large firms that hold a monopolistic posi-

tion in the local economy, are impor tant for generating fiscal revenue, and play a

vital role in economic development .6 Since SOEs with a monopolistic position are

more important to the local economy and regional development, such firms are likely

to be subject to political scrutiny and government control (Faccio, 2007).4. Strategic SOEs: The Chinese government has identified seven sectors that are consid-

ered strategically important because they are related to national or economic security,

and the government maintains absolute control in these sectors through either sole own-

ership or an absolute controlling stake (Mattlin, 2007). These seven sectors are defense,

power generation and distribution, oil and petrochemicals, telecommunications, coal,

aviation, and shipping.7 For SOEs operating in these strategic sectors, regulations

6 The Chinese government only recently specically dened those enterprises that it considers critical or

important.7

SASAC Chair Li Rongrong revealed that these seven sectors are considered strategically important as they arerelated to national or economic security and that the government has to maintain absolute control over rms in

these sectors through either sole ownership or an absolute controlling stake. Many of the biggest and most monop-

olistic Chinese companies operate within these seven industries.


http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


10/28

dictate that firms' decision-making procedures are supplanted by the administrative pro-

cess. The regulatory process that takes place in such enterprises exposes firms' decision

making to more political forces and government control and is likely to result in the im-

position of political outcomes in place of private decisions or market outcomes(Hadlock & Lee, 2002). The strategic status of the industry in which a firm operates

thus reflects the extent to which the firm is subject to government control through reg-

ulations and personnel control.

4.2. Model

To examine the sensitivity of management turnover to accounting measures of firm per-

formance, we apply a logistic model to test H1 as follows.

Logit Turnover it ð Þ ¼ α þ β 1 Perf ormanceit −1 þXn

j¼2

β j CV it −1 j þ ε it ð1Þ

Where:

Turnover a dummy variable equal to 1 if either the chairman or CEO, or both, is (are)

forced to leave the firm (other than for natural reasons disclosed in the CSMAR

database), and 0 otherwise;

Performance an accounting measure of previous firm performance: return on assets

lagged one period ( ROAt −1);

The definitions of CV , the control variables, are based on previous studies (Berry,

Bizjak, Lemmon, & Naveen, 2000; Clayton, Hartzell, & Rosenberg, 2003; Fan & Wang,

2007; Smith & Watts, 1992) and are listed as follows:

Log (asset ) the log of total assets, which is used to measure and control for rm size. Berry

et al. (2000) suggest that an increase in size increases management entrenchment

so that top executives are less likely to be found to be “incompetent ”;

Leverage equal to the ratio of total debt to total assets and used to control for the firm's

financial risk. Gilson (1990) provides evidence that management turnover ismore prevalent in financially distressed firms;

Indus _ MB the industry median market-to-book ratio by 2-digit SIC code, a surrogate for

the investment opportunity set (IOS). Prior research indicates that the IOS may

also be associated with management turnover because firms with a high IOS

demand high-quality managers (Smith & Watts, 1992);

Log (employee) the log of the number of employees. Some researchers argue that manage-

ment in SOEs is also evaluated on the basis of social duties, such as maintaining

and increasing employment (Fan et al., 2007). Hence, the government is likely to

replace top executives if the unemployment rate is too high;

Log (GDP ) the log of provincial GDP, a surrogate for regional variety. This is associatedwith management turnover because of differences in local labor markets and local

economic development (Fan et al., 2007); and


http://-/?-http://-/?-


11/28

Dual a dummy variable equal to 1 for a manager who simultaneously holds the chairman

of the board and CEO positions. Prior studies suggest that the dual position of chair-

man and CEO results in greater management power and entrenchment as executives

who simultaneously occupy both positions are less likely to leave their company.

The first hypothesis predicts that β 1 will be negative and is consistent with the disciplin-

ary mechanism in which incompetent managers are dismissed when performance is poor.

To test the second hypothesis about the interaction of government control and corporate

control, we extend Model 1 as follows.

Logit Turnover it ð Þ ¼ α þ β 1 GC it −1 þ β 2 Perf ormanceit −1þ

β 3 Perf ormanceit −1 GC it −1 þXn

j¼4

β jCV it −1 j þ ε it ð2Þ

Where:

GC is one of four measures of government control, which include concentration of

government ownership, jurisdictional level of government ownership, the monop-

olistic position of an SOE, and whether or not the firm is a strategic SOE. These

measures are defined as follows. Concentration of government ownership (Own)

is a dummy variable equal to 1 if the proportion of shares owned by the govern-

ment is above the median level for the year, and 0 otherwise. Jurisdictional level

of government ownership (Central ) is a dummy variable equal to 1 for SOEs that are held directly by the central government or local government, and 0 for those

SOEs indirectly owned by legal representatives or government-related entities

(Fan et al., 2007). The monopolistic position of an SOE ( LARGE_MS ) i s a

dummy variable equal to 1 if firm i's market sales are above the median level

for the year of all listed firms in the province in which it is located, and 0 other-

wise. Finally, strategic SOE (Strategic) is a dummy variable equal to 1 for an SOE

in a regulated or strategic industry based on the SIC code issued by the China

Securities Regulatory Commission (CSRC), China's equivalent to the SEC, and

0 otherwise. These regulated or strategic industries are described earlier in this

paper. All other variables are as defined in Model 1.

Hypothesis 2 focuses on the estimate of the interaction termβ 3, the coefficient of the in-

teraction terms for government control and the turnover – performance relationship. If β 3 is

negative, then there is a stronger inverse relationship between top management turnover and

firm performance, indicating that stronger government control strengthens the turnover –

performance relationship. In contrast, if β 3 is positive, then there is a weaker inverse relation-

ship between top management turnover and firm accounting performance.

5. Data and sample

The sample includes Chinese SOEs that were listed on the domestic Shanghai and

Shenzhen stock exchanges during the 2001–2005 period. We identify the ultimate controller


http://-/?-http://-/?-


12/28

by examining the ownership structures of the sample firms as disclosed in their annual reports.

The management turnover data are downloaded from the CSMAR dataset. To be included in

the sample, firms must have sufficient accounting data available in the CSMAR dataset. In

identifying non-natural turnover of top managers, we exclude turnover events in which atop manager left because of retirement, health reasons, change of ownership, or the end of

a temporary appointment to the position, all of which are natural reasons for turnover. This

screening procedure ensures that only non-natural top management turnovers are captured

in our analysis. To mitigate the outlier effect, we truncate the top and bottom 1% of the dis-

tribution for return on assets, total assets, net sales, total debt, and earnings before interest

and taxes (EBIT). The final sample consists of 918 unique SOE firms and 4,096 firm-year

observations.

Table 1 reports the descriptive statistics for the sample firms. For each variable used in

the regression model, we report the mean, median, and standard deviation across two sets

of sample firms: one being the sub-sample in which top management changed during the

Table 1

Descriptive statistics for firms by sub-sample. The turnover group, assigned the value of 1, comprises firms in

which there was a change in either the chairman or the CEO during the 2001 –2005 period; the non-turnover

group, assigned the value of 0, comprises firms in which there was no change in top management. For variable

definitions, refer to Sections 4.1 and 4.2. pb1%, **= pb5%, *=pb10%.

Variable N Turnover Mean Median Std. Dev. Diff. (1–0) (t -test)

ROAt 1222 1 0.009 0.019 0.067 − 0.012

2874 0 0.021 0.025 0.058 (5.26***) ROAt − 1 1222 1 0.008 0.024 0.164 − 0.012

2874 0 0.020 0.030 0.119 (2.27**)

Own 1222 1 0.391 0.390 0.148 0.002

2874 0 0.389 0.385 0.144 (− 0.20)

High _ Own 1222 1 0.404 0 0.491 0.019

2874 0 0.385 0 0.487 (− 1.16)

Central 1222 1 0.258 0 0.438 0.028

2874 0 0.230 0 0.421 (− 1.86*)

MS 1222 1 0.290 0.093 0.363 − 0.013

2874 0 0.303 0.110 0.362 (1.05)

LARGE _ MS 1222 1 0.516 1 0.500 − 0.040

2874 0 0.556 1 0.497 (2.35**)Strategic 1222 1 0.555 1 0.497 − 0.030

2874 0 0.585 1 0.493 (1.76*)

Log (assets) 1222 1 21.14 21.11 0.818 − 0.10

2874 0 21.24 21.18 0.791 (3.60***)

Leverageit 1222 1 0.509 0.491 0.339 0.029

2874 0 0.480 0.475 0.240 (− 2.77***)

Indus _ MB 1222 1 2.507 2.259 1.037 0.045

2874 0 2.462 2.243 0.942 (− 1.32)

Log (employ) 1222 1 7.380 7.430 1.226 − 0.074

2874 0 7.454 7.565 1.170 (1.78*)

Log (GDP ) 1222 1 8.538 8.596 0.837 − 0.025

2874 0 8.563 8.596 0.799 (0.87) Dual 1222 1 0.061 0 0.240 − 0.041

2874 0 0.102 0 0.302 (4.53***)



13/28

year and the other sub-sample consisting of firms in which there was no turnover in top

management. The first two variables reported in Table 1 are accounting measures of

firm performance used in previous studies (Chang & Wang, 1994; Engel et al., 2003;

Kato & Long, 2006; Weisbach, 1988). Although the firms with management turnover onaverage demonstrate both positive current performance ( ROAit ) and positive lagged perfor-

mance ( ROAit −1), they significantly underperform the non-turnover group (t-value=5.26,

significant at the 1% level for ROAit ; and t-value=2.27, significant at the 5% level for

ROAit-1). The next two variables measure concentration of government ownership. The

t -test for the difference between the two sub-samples shows that the ownership proportion

for the management-turnover group is similar to that of the non-turnover group. The next

variable reported in Table 1 is the jurisdictional level of government ownership. We find

that management turnover is more likely to occur in SOEs with central government

ownership/control. A comparison of SOE market shares in terms of sales shows that the

management-turnover group has a somewhat lower market share than the non-turnover

group. The results also shows that the frequency with which management-turnover

firms appear in a strategic industry is significantly lower than that of non-turnover

firms, which is consistent with the argument that strategic or regulated industries are sub-

ject to less competitive labor market conditions.

6. Empirical results

6.1. Results for the relationship between top management turnover and firm performance

(H1)

Table 2 presents the correlation matrix of the independent variables included in the logit

analysis. Although many of the measures are significantly correlated, the magnitude of the

correlations (absolute valueb0.4) does not indicate potential problems of multicollinearity.

We test the first hypothesis by performing a likelihood analysis in which we regress the

likelihood of top management turnover on the accounting performance measure and sever-

al control variables in Model 1. The dependent variable is coded 1 for management-

turnover firm-year observations, and 0 for non-turnover observations. Consistent with

the hypothesis that the frequency of management turnover is negatively associated with

the firm's prior accounting performance, the results in Table 3 show that the coefficient on firm accounting performance is significantly negative (− 1.06; p-value=0.033). Regard-

ing the control variables in the regression, firm size has a negative but insignificant influ-

ence on the likelihood of management turnover. Management turnover is more frequent in

firms that face higher financial risks (i.e., firms with higher debt-to-asset ratios, as evi-

denced by Gilson, 1990). The industrial market-to-book (MB) ratio has a significantly neg-

ative influence on the likelihood of management turnover, implying that growing firms

have less management turnover. As expected, the existence of a top manager in a dual

role has a significantly negative influence on management turnover. This is consistent

with the results of prior studies showing that the appointment of the same individual to

the positions of chairman and CEO results in less management turnover because such in-dividuals become entrenched in their positions. Regional variety ( Log (GDP )) and social

duties ( Log (employ)) have no significant influence on senior management turnover.


http://-/?-http://-/?-


14/28

Table 2

Pearson correlations between independent variables (p-value in parentheses).

ROA ROA Log Variable

(t)(t − 1) Own Central MS Strategic (asset) Leverage

ROA(t) 1

ROA(t-1) 0.19 1

(b .01***)

Own 0.12 0.06 1

(b .01***) (0.00***)

Central 0.03 − 0.01 0.07 1

(0.04**) (0.51) (b .01***)

MS 0.06 0.07 0.02 − 0.10 1

(b .01***) (b .01***) (0.16) (b .01***)

Strategic 0.07 0.03 0.10 0.14 0.06 1(b .01***) (0.09*) (b .01***) (b .01***) (b .01***)

Log(asset) 0.21 0.13 0.12 0.00 0.16 0.03 1

(b .01***) (b .01***) (b .01***) (0.98) (b .01***) (0.03**)

Leverage − 0.30 − 0.31 − 0.11 − 0.01 − 0.03 − 0.06 0.01 1

(b .01***) (b .01***) (b .01***) (0.45) (0.06*) (0.00***) (0.46)

Indus_MB 0.08 0.03 − 0.00 0.04 − 0.03 − 0.07 − 0.13 − 0.09

(b .01***) (0.05**) (0.97) (0.01***) (0.09*0 (b .01***) (b .01***) (b .01***)

Log(GDP) 0.03 − 0.00 0.02 − 0.00 − 0.07 − 0.05 0.12 0.06

(0.03**) (0.76) (0.28) (0.94) (b .01***) (b .01***) (b .01***) (0.00***)

Log(employ) 0.05 0.02 0.08 − 0.03 0.22 0.15 0.42 0.04

(0.00***) (0.12) (b .01***) (0.06**) (b .01***) (b .01***) (b .01***) (0.02**)

Dual − 0.03 − 0.04 − 0.06 − 0.05 − 0.02 − 0.07 − 0.02 0.02

(0.03**) (0.01***) (b .01***) (b .01***) (0.30) (b .01***) (0.22) (0.14)

Note: *** = pb1%, **= pb5%, *=pb10%.


15/28

6.2. Results for the interaction of government control and the turnover – performance

relationship (H2)

We next examine whether the inverse relationship between top management turnover

and firm accounting performance is stronger or weaker in firms with tighter government

control. We test the second hypothesis by estimating a logit model to which we add an in-

teraction term between the firm performance variable and the variable proxied for govern-

ment control. The four proxies we employ for the extent of government control are:concentration of government ownership, jurisdictional level of government ownership,

the monopolistic position of the SOE, and whether or not the SOE is in a strategic industry.

The first column of Table 4 reports the results for the interaction term including concentra-

tion of government ownership. As expected, the coefficient of Performanceit −1∗ Ownit −1 is

negative and is not statistically significant.

The second column of Table 4 reports the results for the interaction term includ-

ing jurisdictional level of government ownership. For the interaction term, including

the dummy variable for SOEs directly governed by the central or local government

(Central ) and previous firm performance, Column 2 shows that the coefficient of

Performanceit −1* Central it −1 is negative and significant (− 2.78; p value = 0.029). This find-ing suggests that when an SOE is directly controlled by the central or local government, rather

than indirectly controlled by other state enterprises, the inverse relationship between top

Table 3

The relationship between management turnover and firm performance (Model 1).

Variable Expected sign

Intercept − 0.54 7.74

(0.014***) (b .0001***)

Performanceit −1 - −0.58 −1.06

(0.030**) (0.033**)

Log (assetsit − 1) - − 0.07

(0.236)

Leverageit − 1 + 0.40

(0.018**)

Indus _ MBit −1 + − 1.12

(b .0001***)

Log (GDP it −1) +/ − 0.01

(0.809) Log (employit − 1) - − 0.03

(0.519

Dual it − 1 - − 0.51

(b .0001***)

Year _ dummy Yes Yes

Indus _ dummy Yes Yes

Obs. # 4096 4096

Pseudo-R-sq 0.014 0.12

See Sections 4.1 and 4.2 for variable definitions. The dependent variable is the likelihood of management turn-

over. Performanceit − 1 is measured by accounting measures, ROA is lagged one period. (P-value in parentheses:

*** pb1%, ** pb5%, *b10%).


http://-/?-http://-/?-


16/28

management turnover and firm performance is stronger and performance-based turnover de-

cisions are enforced more effectively.

The third column of Table 4 reports the results for the interaction term including

the monopolistic position of SOEs. The coefficient on the interaction term Performanceit −1* LARGE _ MS it −1 is negative and significant (− 2.25; p-value = 0.051).

This finding suggests that the government has a greater interest in the performance

of SOEs, which have a greater market share in the local economy (based on provin-

cial unit) because they make an important contribution to the fiscal revenue of the

Table 4

The influence of government control (GC it − 1) on the relationship between top management turnover and firm

accounting performance measures with control variables (Model 2).

Variable Expectedsign

(1) (2) (3) (4)GC = government

ownership (%)

above median

GC = central

government/Direct local

government control

GC = monopolistic

position by market

share

GC = strategic

industries

Intercept 7.76 3.42 6.87 7.13

(b .0001***) (0.008***) (b .0001***) (b .0001***)

Performanceit − 1 - − 1.03 1.52 − 0.48 0.02

(0.132) (0.214) (0.408) (0.977)

GC it − 1 +/ − 0.04 − 0.05 − 0.08 − 0.13

(0.650) (0.672) (0.469) (0.109)

Performanceit −1*

GC it −

1

− − 0.95 −2.58 −2.25 −3.56

(0.211) (0.049**) (0.051**) (0.001***)

Log (assetsit − 1) − − 0.07 − 0.14 − 0.03 − 0.06

(0.229) (0.020**) (0.704) (0.313)

Leverageit − 1 + 0.41 0.32 0.40 0.43

(0.017**) (0.045**) (0.018**) (0.013***)

Indus _ MBit −1 + − 1.12 − 0.29 − 1.12 − 1.02

(b .0001***) (b .0001***) (b .0001***) (b .0001***)

Log (GDP it −1) +/ − 0.01 − 0.03 0.006 0.02

(0.812) (0.503) (0.915) (0.641)

Log (employit −1) − − 0.03 0.02 − 0.02 − 0.02

(0.518) (0.654) (0.621) (0.617)

Dual it − 1 − − 0.51 − 0.62 − 0.50 − 0.51(b .0001***) (b .0001***) (0.001***) (b .0001***)

Year _ dummy Yes Yes Yes Yes

Indus _ dummy Yes Yes Yes Yes

Obs. # 4096 4096 4096 4096

Pseudo R-sq. 0.12 0.04 0.08 0.12

See Sections 4.1 and 4.2 for variable definitions. The dependent variable is the likelihood of management turn-

over. Performanceit − 1 is measured by ROA lagged one period. GC it − 1 is defined as: (1) the concentration of

government ownership, measured by a dummy equal to 1 if the proportion of ownership is above the median

level for the year; (2) jurisdictional government control, measured by a dummy variable equal to 1 if the SOE

is affiliated with the central government/directly held by the local government; (3) monopolistic position of

SOE, measured by a dummy variable equal to 1 if the SOE's market sales are above the median level of salesin the provincial unit where it is located; (4) the strategic industries, measured by a dummy variable equal to 1

if the SOE is in one of seven industries that are strategically regulated by the government. (P-value in parentheses:

*** pb1%, ** pb5%, *b10%).



17/28

local economy. Thus, government officials have strong incentives to ensure that top

management turnover will be tied more closely to firm performance in SOEs with a

monopolistic position.

The final column of Table 4 reports the results for the interaction term includingthe strategic importance of SOEs. The results show that the coefficient of

Performanceit −1* Strategicit −1 is negative and significant (− 3.56; p-value= 0.001). This

finding suggests that for SOEs in the seven strategic industrial sectors—defense, power

generation and distribution, oil and petrochemicals, telecommunications, coal, aviation, and

shipping—all of which are subject to more stringent regulatory processes and control, the

inverse relationship between top management turnover and firm performance is stronger.

Overall, there is clear evidence that the government and government officials represent-

ing the state interests in Chinese SOEs care more about the performance of SOEs with

tighter government control and that the effective corporate governance mechanism of

replacing top management for poor performance is more often used in SOEs with stronger

government control.

7. Additional tests

The objectives of this section are to provide additional evidence in support of our find-

ings and shed additional insights into the effects of government control on the use of the

turnover – performance governance mechanism.

7.1. Regional institutional factors

Prior studies (Chang & Wong, 2004a, 2004b; La Porta et al., 2000; Qian, 1996; Shleifer

& Vishny, 1997) suggest that corporate control and government control are more necessary

in regions with poor investor protection or undeveloped market institutions. We perform

additional analysis to evaluate whether the complementary effect of government control

on the turnover – performance government mechanism is stronger in regions with different

institutional environments by stratifying the full sample according to regional institutional

factors and re-running Model 2 on alternative sub-samples. We partition the sample by the

yearly median value of the ratio of stock market capitalization to local GDP in a given re-

gion (province), the number of firms listed in a given region (province), and the number of initial public offerings of equity in a given region (province). La Porta and Shleifer (1997)

use a sample of 49 countries and provide evidence that these three variables measuring the

size, breadth, and valuation of capital markets are positively associated with investor pro-

tection. We also partition the sample by the yearly median value of the comprehensive

marketization index of regions/provinces in China investigated by Fan and Wang (2007).

Table 5 reports the results of the sub-sample regressions. Panel A partitions the sample

by the yearly median value of the ratio of stock market capitalization to local GDP in a

given region (province). The results show that the interaction of government control and

firm performance ( Performanceit −1* GC it −1) is significantly negative if the ratio of the

stock market capitalization of firms in the region to local GDP is low where government control is measured by the concentration of government ownership (see Column 1) or cen-

tral/direct local government control (see Column 2). However, the inverse relation does not



18/28

Table 5

Regression results of the interaction effects of government control and turnover – performance relationship in China's SOEs us

tutional factors used as proxies for investor protection or market development.

Panel A: Partition by the

stock market capitalization

in a given region (La Porta

& Shleifer, 1997)

Expected

sign

GC(1) GC(2) GC(3)

Low High Low High Low H

Intercept − 0.40 9.11 − 0.82 9.07 − 2.06 8

(0.824) (b .0001***) (0.649 (b .0001***) (0.330) (0

Performanceit − 1 − 0.06 − 0.49 (1.78 1.26 − 1.39 0

(0.939) (0.331) (0.421) (0.382) (0.108) (0

GC it −1 +/ − 0.20 0.04 − 0.31 0.29 − 0.21 −

(0.086*) (0.745) (0.072*) (0.129) (0.131 (0

Performanceit − 1* GC it − 1 −/+ − 3.91 0.58 − 3.92 − 1.79 0.78 −

(0.009***) (0.554) (0.094*) (0.236) (0.530) (0

Log (assetsit − 1) − 0.008 − 0.36 0.03 − 0.36 0.09 −

(0.926) (0.0002***) (0.686) (0.0002***) (0.395) (0

Leverageit − 1 + 0.43 − 0.16 0.39 − 0.15 0.44 −

(0.029**) (0.677) (0.042**) (0.691) (0.027**) (0

Indus _ MBit − 1 + − 0.30 − 0.38 − 0.24 − 0.40 − 0.28 −

(0.003***) (0.0003***) (0.019**) (0.0002***) (0.005***) (

Log (GDP it −1) +/ − 0.08 − 0.28 0.08 − 0.28 0.09 −

(0.279) (0.0004***) (0.308) (0.0006***) (0.248) (0

Log (employit − 1) − − 0.08 0.22 − 0.07 0.21 − 0.08 0

(0.142) (0.002***) (0.190) (0.003***) (0.177) (0 Dual it − 1 − − 0.50 − 0.86 − 0.56 − 0.89 − 0.53 −

(0.012***) (0.002***) (0.005***) (0.002***) (0.007***) (0

Year _ dummy Yes Yes Yes Yes Yes Y

Indus _ dummy Yes Yes Yes Yes Yes Y

Obs. # 2344 1718 2344 1718 2344 1

Pseudo R-sq. 0.04 0.09 0.04 0.09 0.04 0


19/28

Panel B: Partition by the

number of rms listed in

a given region (La Porta

& Shleifer, 1997)

Expected

sign

GC(1) GC(2) GC(3)


Intercept 6.46 2.69 6.22 2.66 2.19 3.

(0.003***) (0.100*) (0.005***) (0.104*) (0.406) (0 Performanceit −1 − 1.00 − 0.56 2.82 0.76 0.58 −

(0.410) (0.238) (0.277) (0.514) (0.624) (0

GC it − 1 +/ − 0.26 0.01 − 0.20 − 0.08 − 0.46 0.

(0.085*) (0.923) (0.325) (0.620) (0.012****) (0

Performanceit −1* GC it −1 −/+ − 6.96 − 0.08 − 4.61 − 1.69 − 3.27 0.

(0.003***) (0.919) (0.104*) (0.179) (0.119) (0

Log (assetsit − 1) − − 0.31 − 0.07 − 0.31 − 0.07 − 0.12 −

(0.004***) (0.338) (0.005***) (0.376) (0.343) (0

Leverageit − 1 + − 0.04 0.47 − 0.19 0.46 0.03 0.

(0.920) (0.015**) (0.650) (0.016**) (0.944) (0

Indus _ MBit −1 + − 0.23 − 0.32 − 0.19 − 0.32 − 0.21 −

(0.053**) (0.0003***) (0.106) (0.0004***) (0.083*) (0

Log (GDP it − 1) +/ − − 0.10 − 0.11 − 0.08 − 0.12 − 0.09 −

(0.214) (0.164 (0.317) (0.131) (0.289) (0

Log (employit − 1) − 0.19 − 0.04 0.21 − 0.04 0.22 −

(0.015**) (0.417) (0.008***) (0.458) (0.006***) (0

Dual it − 1 − − 0.45 − 0.80 − 0.48 − 0.81 − 0.40 −

(0.057*) (0.0003***) (0.042**) (0.0003***) (0.092*) (0



Obs. # 1515 2547 1515 2547 1515 25

Pseudo R-sq. 0.07 0.05 0.07 0.05 0.07 0.


20/28

Table 5 (continued )

Panel C: Partition by the

number of initial public

offerings of equity in a

given region (La Porta &

Shleifer, 1997)

Expected

sign

GC(1) GC(2) GC(3)

Low High Low High Low

Intercept 0.18 6.20 − 0.18 5.92 0.30

(0.930) (0.009***) (0.933) (0.013***) (0.903)

Performanceit − 1 − − 0.70 0.07 5.69 0.31 − 1.09

(0.295) (0.936) (0.091*) (0.770) (0.246)

GC it − 1 +/ − 0.12 0.25 − 0.03 − 0.11 0.11

(0.368) (0.096*) (0.907) (0.618) (0.536)

Performanceit − 1* GC it − 1 −/+ − 3.87 0.09 − 9.01 − 0.30 − 3.39 (0.026**) (0.942) (0.010***) (0.817) (0.086*)

Log (assetsit − 1) − − 0.01 − 0.28 0.02 − 0.26 − 0.02

(0.893) (0.014***) (0.857) (0.023***) (0.874)

Leverageit −1 + 0.46 0.02 0.45 − 0.04 0.50

(0.032**) (0.962) (0.038**) (0.922) (0.028**)

Indus _ MBit − 1 + − 0.27 − 0.36 − 0.23 − 0.34 − 0.26

(0.009***) (0.003***) (0.030**) (0.006***) (0.012***)

Log (GDP it − 1) +/ − 0.01 − 0.20 − 0.02 − 0.22 0.02

(0.920) (0.093*) (0.842) (0.078*) (0.822)

Log (employit − 1) − − 0.005 0.10 − 0.01 0.11 − 0.02

(0.945) (0.173) (0.906) (0.165) (0.764)

Dual it −1 − − 0.45 − 0.86 − 0.51 − 0.89 − 0.45

(0.062*) (0.009***) (0.036**) (0.007***) (0.065*)

Year _ dummy Yes Yes Yes Yes Yes

Indus _ dummy Yes Yes Yes Yes Yes

Obs. # 1756 1360 1756 1360 1756

Pseudo R-sq. 0.06 0.08 0.07 0.08 0.06

Table 5 (continued )


21/28

Panel D: Partition by

regional market index

(Fan & Wang, 2007)

Expected

sign

GC(1) GC(2) GC(3)


Intercept 7.43 2.19 7.98 2.15 5.68 1

(0.006***) (0.174) (0.003***) (0.182) (0.075*) (

Performanceit − 1 − 0.42 − 0.43 1.52 1.39 − 0.24 −(0.706) (0.338) (0.488) (0.324) (0.835) (

GC it −1 +/ − 0.06 0.15 − 0.04 − 0.04 − 0.27 −

(0.715) (0.133) (0.861) (0.776) (0.202) (

Performanceit − 1* GC it − 1 −/+ − 5.45 − 0.39 − 2.73 − 2.46 − 0.60 −

(0.030**) (0.619) (0.268) (0.101) (0.785) (

Log (assetsit − 1) - − 0.28 − 0.10 − 0.30 − 0.10 − 0.19 −

(0.038**) (0.143) (0.026**) (0.182) (0.227) (

Leverageit − 1 + 0.05 0.40 0.03 0.37 0.11 0

(0.918) (0.027**) (0.956) (0.039**) (0.805) (

Indus _ MBit − 1 + − 0.40 − 0.29 − 0.41 − 0.28 − 0.40 −

(0.002***) (0.001***) (0.002***) (0.001***) (0.002***) (

Log (GDP it −1) +/ − − 0.17 0.02 − 0.18 0.006 − 0.18 0

(0.082*) (0.799) (0.064*) (0.941) (0.061*) (

Log (employit − 1) − 0.12 − 0.01 0.13 − 0.006 0.14 −

(0.214) (0.865) (0.191) (0.896) (0.166) (

Dual it − 1 − − 0.71 − 0.57 − 0.75 − 0.59 − 0.71 −

(0.022**) (0.003***) (0.017**) (0.002***) (0.023**) (



Obs. # 1098 2943 1098 2943 1098 2

Pseudo R-sq. 0.07 0.04 0.07 0.04 0.07 0

The sample is partitioned alternatively by the yearly median value of the ratio of stock market capitalization to local GDP in a g

listed in a given region (province), the number of initial public offerings of equity in a given region (province) ( La Porta & S

ketization index of regions/provinces in China investigated by Fan and Wang (2007). The dependent variable is the likelihood o

measured by ROA lagged one period. GC it − 1 is defined as: (1) the concentration of government ownership, measured by a dumm

is above the median level for the year; (2) jurisdictional government control, measured by a dummy variable equal to 1 if the S

or is directly held by a local government; (3) monopolistic position of the SOE, measured by a dummy variable equal to 1 if th

level of sales in the provincial unit where it is located; and (4) the strategic industries, measured by a dummy variable equal to 1

are strategically regulated by the government. (P-value in parentheses: *** pb1%, ** pb5%, *b10%).


22/28

persist or is not significant if the firms are in regions with a high ratio of the stock market

capitalization to local GDP. In Columns 3 and 4, where the government control is

measured by the market shares of firms and strategic industries, respectively, the results

for the low and high sub-samples are not significantly different. Panel B partitions the sample by the yearly median value of the number of firms listed in a given region (province),

and Panel C divides the sample by the yearly median value of the number of initial public

offerings of equity in a given region (province). The results in both panels show that the

interaction of government control and firm performance ( Performanceit −1* GC it −1) is per-

sistently negative in regions with a low number of listed firms for all measures of govern-

ment control. However, the inverse relation does not exist or is not significant for regions

with a high number of listed firms. Panel D partitions the sample by the yearly median

value of the comprehensive marketization index of regions/provinces in China (Fan &

Wang, 2007). The results show that the interaction of government control and firm perfor-

mance ( Performanceit −1* GC it −1) is significantly negative if the firms are from regions

with low marketization indices, where government control is measured by the concentration

of government ownership (see Column 1). However, the inverse relation is not significant if

the firms are from regions with high marketization indices. In Column 2 and Column 3,

where government control is measured by central/direct local government control and the

market shares of firms, respectively, the results for the low and high sub-samples are not

significantly different. In Column 4, where government control is measured by whether

or not the firm is in a strategic industry, the interaction term Performanceit −1* GC it −1 is sig-

nificantly negative regardless of whether the firms are from a region with a high or low

marketization index. However, the economic effect of the interaction is stronger in thelow sub-sample than in the high sub-sample (− 4.39 in the low sub-sample vs. − 2.30 in

the high sub-sample). Overall, the sub-sample regression results corroborate the results in

Table 4, providing further support for the argument that the complementary relation be-

tween government control and corporate control is more prominent in regions with poor in-

vestor protection or poor market institutions.

7.2. Endogeneity concern

Government administrators control not only the appointment/dismissal of top managers but also the approval of large investment projects and asset disposals; therefore, it is pos-

sible that top management turnover and firm performance are both endogenously deter-

mined by omitted institutional factors. To address this potential problem, we follow Fan

et al. (2007) and re-run Model 1 as in Table 3 on sub-samples alternately stratified by

firm and regional institutional factors to evaluate whether the documented relations persist

in the sub-sample regressions. Specifically, we partition the sample by the sample median

value of (1) local (provincial) GDP, (2) local fiscal deficit levels, (3) local unemployment

rates, and (4) firm return on sales (ROS), respectively.

Table 6 reports the results of the sub-sample regressions. Top management turnover is

associated with significantly negative firm performance regardless of whether the firms arefrom regions with high or low GDP, healthy or poor fiscal conditions, high or low employ-

ment rates, or whether the SOEs have high or low ROS. The sub-sample regression results



23/28

Table 6

Results of the sub-sample regressions. We re-run Model 1 (as in Table 3) on sub-samples alternately stratified by firm and regio

the predicted relations persist in the sub-sample regressions (Fan et al., 2007). We partition the sample by the sample median vfiscal deficit levels, (3) local unemployment rates, and (4) firm return on sales (ROS). The negative relations between top ma

(ROA) are robust in the sub-samples.

Expected

sign

(1) GDP (2) Fiscal decit (3) Unemployment

Low High Low High Low Hi

Intercept 30.48 22.98 22.21 27.18 27.59 21

(0.906) (0.928) (0.893) (0.904) (0.873) (0

Performanceit −1 - − 4.42 − 5.00 − 5.40 − 5.04 − 4.29 − 5

(0.086*) (b .001***) (0.004***) (0.029**) (0.009***) (0

Log (assetsit − 1) − 0.53 − 0.19 − 0.19 − 0.41 − 0.49 − 0

(0.019**) (0.059*) (0.060*) (0.042**) (b .001***) (0

Leverageit − 1 + − 0.30 0.26 0.46 − 0.69 − 0.05 0.

(0.740) (0.541) (0.300) (0.381) (0.926) (0

Indus _ MBit −1 − 2.06 − 1.17 − 1.19 − 1.90 − 1.49 −

(b .001***) (b .001***) (b .001***) (b .001***) (b .001***) (b

Log (GDP it − 1) 0.00 0.00 + 0.00 0.00 0.00 0.

(0.522) (0.148) (0.080*) (0.362) (0.194) (0

Log (employit − 1) 0.21 − 0.04 − 0.06 0.23 0.188 − 0

(0.149) (0.546) (0.343) (0.081*) (0.038**) (0

Dual it − 1 - 0.46 − 0.43 − 0.36 0.13 − 0.60 0.

(0.238) (0.065*) (0.125) (0.715) (0.028**) (0

Year _ dummy Yes Yes Yes Yes Yes Ye

Indus _ dummy Yes Yes Yes Yes Yes YeObs. # 971 3125 2945 1151 2285 18

Pseudo R-sq. 0.18 0.12 0.12 0.17 0.16 0.


24/28

confirm that the results in Table 3 are not affected by the regional institutional factors, thus

providing additional support for the robustness of our findings.

7.3. Management turnover issue

We identified forced top management turnover by excluding natural reasons for turn-

over, such as retirement, change of ownership, health, and the end of a temporary ap-

pointment. The management turnover data in the analysis contain changes in top

management personnel at the expiration of the term of office (CSMAR turnover data item

number 3). It is possible that management turnover at the end of the term of office could be

a result of forced departure or for other reasons. As a sensitivity test, we excluded the manage-

ment turnover at the end of term from our analysis. The results are qualitatively the same as

reported in Tables 3 and 4. We also conducted a sensitivity test to control the competence of top managers by including their education levels in the regression model and still find consis-

tent results.

7.4. Alternative firm performance measures

An alternative measure of firm performance used in the prior literature is market perfor-

mance (stock price). We re-run our tests using stock price performance (cumulative retur n

based on 12 monthly returns) and, alternatively, Tobin's Q to measure firm performance.8

The unreported results show no significant association between management turnover and

firm market performance in SOEs. This result is consistent with prior studies using SOEs

as the sample. For example, Chang and Wong (2004a, 2004b) find that managerial change

and selection in Chinese SOEs are not significantly associated with stock price perfor-

mance because state-owned shares are usually non-tradable and can be transferred only

on receiving administrative approval. SASAC officials and other bureaucrats who exercise

control on behalf of the government are unable to capture any capital gains for themselves

when enterprise shares are not transferred. Consequently, state owners may be less con-

cerned about long-term shareholder value as reflected in stock prices. Firth, Fung, and

Rui (2006) also show that chairman turnover is related to a firm's profitability but not to

its stock returns in Chinese-listed firms. But they don't provide any explanations. Peng

and Luo (2000) (p. 492) suggest that ROA is a preferred measure for Chinese firm perfor-

mance because accounting for assets in China is typically more accurate than accounting

for equity as the equity contributions for a majority of Chinese firms are historically ambig-

uous, thus rendering it difficult to compute the stock return.

We also conduct robustness tests by using measures of adjusted firm performance such

as ROA adjusted by industrial mean ROA. The unreported results are consistent with the

results derived using firm-specific ROA and imply that the stat e owner normally uses earn-

ings as a criterion in making management turnover decisions.9

8

Tobin

0

sQ ¼

Equity Market ValueþLiabilitiesBook Values

EquityBook Values , wherein the non-tradable state shareholdings are valued bythe market price of tradable state shareholdings.9 This may be because the government can solicit accounting information (Laffont, 1994; Stigler, 1971) and

earnings are a single summary of rm performance that is easily understood.


http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


25/28

7.5. Redefining the concentration of government ownership

Although the result for the interaction including concentration of government owner-

ship is not significant, government ownership is a critical method of government control.We, therefore, redefine concentration of government ownership to check on the robustness

of the earlier finding. First, we re-run the model by using an indicator variable for a high

level of government ownership, setting the cutoffs for a high degree of government own-

ership at the 20%, 50%, or 75% shareholding level. The interaction results remain insignif-

icant. Second, we partition the sample into two sub-samples according to whether the firm

has a high or low proportion of government ownership and run Model 1 using these sub-

samples. For both the high and low government ownership level sub-samples, the likeli-

hood of management turnover is significantly inversely associated with the two accounting

measures of firm performance.

8. Conclusions

This study examines the turnover – performance relationship in Chinese SOEs by first

addressing the question: “Is corporate control through performance-based turnover deci-

sions effective in businesses subject to a high degree of government control, such as

Chinese SOEs, which operate in an institutional environment characterized by a mixture

of capitalism and socialism?” Our findings show that the likelihood that top SOE manage-

ment will be replaced increases when firm accounting performance is poorer. The evi-

dence presented here suggests that despite insufficient external market pressure,government owners of SOEs adopt a control mechanism that ties top management turn-

over to firm performance given the current interests of the Chinese government in eco-

nomic performance.

Second, we investigate the intertwining role government control plays in the corporate

governance of Chinese SOEs. We measure government control by four metrics: concentra-

tion of government ownership, jurisdictional level of government control, the monopolistic

position (or market share) of the firm, and whether the firm is in a strategic industry. Over-

all, the findings show that stronger government control enhances the inverse relationship

between top management turnover and firm accounting performance. The results of our ad-

ditional tests provide further evidence that the complementary relation between govern-ment control and corporate control is more prominent in regions with poor investor

protection or poor market institutions. Our findings are important given the debate on

whether privatization is an essential step for improving the performance of SOEs. Chinese

SOEs have been criticized for their potentially inefficient decisions and the ineffectiveness

of their governance systems. Despite the criticism of government influence in Chinese

enterprises, our results suggest that government control in SOEs could play a positive

role in ensuring the establishment of effective internal governance structures and provides

a partial explanation for the success of China's economic reforms.

A caveat of the study is that due to lack of available data we are unable to evaluate

departed top managers' competence and whether their departures are for promotion or demotion. Future research could examine the turnover – performance relationship in Chinese

SOEs by following the careers of departed management and determining whether the



26/28

departed chairman or CEO's next position is a promotion or demotion. This approach will

provide a deeper understanding of the managerial labor market in Chinese SOEs and the

interaction between the promotion/demotion of top management, firm performance, and

government control.

References

Bai, C., & Xu, L. (2005). Incentives for CEOs with multitasks: Evidence from Chinese state-owned enterprises.

Journal of Comparative Economics, 33, 511–539.

Ball, R., Kothari, S. P., & Robin, A. (2003). The effect of international institutional factors on properties of

accounting earnings. Journal of Accounting and Economics, 29, 1–51.

Ball, R., Robin, A., & Wu, J. (2000). Accounting standards, the institutional environment and issuer incentives:

Effect on timely loss recognition in China. Asia Pacific Journal of Accounting and Economics, 7 , 71–96.

Banker, R. D., & Datar, S. M. (1989). Sensitivity, precision, and linear aggregation of signals for performanceevaluation. Journal of Accounting Research, 27 , 21–39.

Berry, T. K., Bizjak, J. M., Lemmon, M. L., & Naveen, L. (2000). CEO turnover and firm diversification. Working

Paper.

Blanchard, O., & Aghion, P. (1996). On insider privatization. European Economic Review, 40(3–5), 759–766.

Blanchard, O., & Shleifer, A. (2001). Federalism with and without political centralization: China vs. Russia in

transitional economics: How much progress? IMF Staff Paper , 48, 171–179.

Boeker, W. (1992). Power and managerial dismissal: Scapegoating at the top. Administrative Science Quarterly,

37 , 400–421.

Boeker, W., & Goodstein, J. (1993). Performance and successor choice: The moderating effects of governance

and ownership. The Academy of Management Journal , 36 (1), 172–186.

Brada, J. C. (1996). Privatization is transition—Or is it? Journal of Economic Perspective, 10(2), 67–86.

Brickley, J. A. (2003). Empirical research on CEO turnover and firm-performance: A discussion. Journal of Accounting and Economics, 36 , 227–233.

Bushman, R., Chen, Q., Engel, E., & Smith, A. (2004). Financial accounting information, organizational complex-

ity and corporate governance systems. Journal of Accounting and Economics, 37 (2), 167–201.

Bushman, R., & Piotroski, J. D. (2006). Financial reporting incentives for conservative accounting: The influence

of legal and political institutions. Journal of Accounting and Economics, 42, 107–148.

Caves, D. W., & Christensen, L. R. (1980). The relative efficiency of public and private firms in a competitive

environment: The case of Canadian railroads. Journal of Political Economy, 88, 958–976.

Chan, K. H., Lew, A. Y., & Tong, M. Y. J. W. (2001). Accounting and management controls in the classical

Chinese novel: A dream of the red mansions. The International Journal of Accounting , 36 , 311–327.

Chang, C., & Wang, Y. (1994). The nature of the township enterprises. Journal of Comparative Economics, 19,

434–452.

Chang, E. C., & Wong, S. M. L. (2004). Political control and performance in China's listed firms. Journal of

Comparative Economics, 36 , 617–636.

Chang, E. C., & Wong, S. M. L. (2004). Chief executive officer turnovers and the performance of China's listed

enterprises. Working Paper.

Chang, E. C., & Wong, S. M. L. (2009). Governance with multiple objectives: Evidence from top executive turn-

over in China. Journal of Corporate Finance, 15, 230–244.

Chen, K. C. W., & Wang, J. (2007). Accounting-based regulation in emerging markets: The case of China's

seasoned-equity offerings. The International Journal of Accounting , 42(3), 221–236.

Chow, G. (2010). Embracing an outgoing China. Hong Kong Economic Journal Monthly, 394, 10–13 (January).

Clayton, M. J., Hartzell, J. C., & Rosenberg, J. V. (2003). The impact of CEO turnover on equity volatility.

Federal Reserve Bank of New York, Staff Report No. 166 .

Coughlan, A. T., & Schmidt, R. M. (1985). Executive compensation, management turnover, and firm perfor-mance. Journal of Accounting and Economics, 7 , 43–66.

DeFond, L. M., & Park, C. W. (1999). The effect of competition on CEO turnover. Journal of Accounting and

Economics, 27 , 35–56.



27/28

Denis, D. J., Denis, D. K., & Sarin, A. (1997). Ownership structure and top executive turnover. Journal of

Financial Economics, 45(2), 193–221.

Earle, J., Estrin, S., & Leshcenko, L. (1996). Ownership structure, patterns of control, and enterprises behavior in

Russia. Enterprise restructuring and economic policy in Russia (pp. 205–252). : The World Bank.

Engel, E., Hayes, R. M., & Wang, X. (2003). CEO turnover

articol hu si leung.pdf

Documents