Chang and Wong (2009) Journal of Corporate Finance, 15 (2009): 230– 244

1. Governance with multiple
objectives: Evidence from
top executive turnover
in China
Chang and Wong (2009)
Journal of Corporate Finance,
15 (2009): 230– 244
中山大学
连玉君
arlionn@163.com

2. Introduction
 Corporate governance: ??
 Existing studies on managerial turnover mostly focus on privately-
controlled firms
 Existing studies assume shareholders have the same incentive
structure to discipline managers under all circumstances.
 We provide empirical evidence on the monitoring of managers in
state-owned firms
 We assume shareholders have a time-varying objective function that
depends on firm performance, profit- and loss-making firms
separately

3. Introduction
 State-owned firms
 State shareholders are agents of the government
 Social , political (employment opportunities) v.s. personal
objectives (personal wealth, job security) v.s. Firm performance
 Different situations lead to different levels of importance attached
to the objectives and then different importance of factors (e.g.,
performance, social and political factors) in determining
managerial turnover

4.  Pre-turnover performance  managerial turnover 
post-turnover performance
 State shareholders attach greater importance to firm performance when
facing financial losses than in times of profit;
 Strong incentives to fire the incumbent managers and to hire and
monitor the new managers with the ability to improve that performance
when facing financial losses
 Reasons:
 Greater government pressure as financial loss is a burden on
government budgets (subsidy) and on state-owned banks (low cost
loans);
 Less resources for state shareholders to serve personal objectives
Introduction

5. Hypothesis
 Loss-making firms:
Poor performance  CEO turnover  Better performance
 More significant negative relationship between pre-turnover
profitability and CEO turnover
 Managerial turnover in those firms is more likely to be followed
by significant improvements in performance

6. Model: Pre-turnover performance and managerial
turnover
 Logit regression model
 Pr (Forced CEO turnover =1)
= F(Performance ; Control Variables )
Where F(z) = ez/(1 +ez)

7.  Control-group
 Using firm performance in both year 0 and year −1 as the benchmarks
 Match each firm that has a CEO turnover to one in the same industry
whose firm performance in the corresponding year was within +/− 20% of
the sample firm's performance but with no turnover occurring in the event
year and in the three years preceding the turnover.
 If multiple firms satisfy these criteria, then we include the firm whose
asset size is closest to that of the turnover firm as the control firm
Managerial turnover and post-turnover performance

8. Variables
 Performance measures:
 ROA,
 Industry-adjusted return on assets (IROA),
 Three-year moving average of ROA (MROA)
 Three-year moving average of IROA

9. Variables
 Control Variables:
 CEOs' age (Age),
 number of years a CEO has served in a listed firm (Tenure ) ,
 existence of a duality structure (Duality),
 Leverage , size,
 ownership nature of the largest(state shareholders or legal
person shareholders)

10. Variables
 CEO turnover
 Persons holding the formal title of either General Manager or Chief
Executive are identified as CEOs
 If a firm underwent two or more turnovers in the same year, only 1 is
recorded.
 Forced CEO turnover
 Exclude retirement, health (including death), corporate governance reform,
change in controlling shareholder, legal disputes;
 Exclude that take up a position that is better than their previous managerial
position；voluntary departures； tenure less than one year.
 Include no information available ；retirement before 55

11. Data
 Sample range: 1995-2001
 Exclude private shareholders as the ultimate controlling
shareholders；
 exclude financial firms and listed only in the B-share market
 Exclude any firm listed for less than six months and firms with
negative equity
 Eliminate observations with missing values
 ROA and Leverage are Winsorized at the 1st and 99th percentiles

12. Turnover rate: overall

13. Turnover rate：by reasons

14. Turnover rate：by Destinations
Note: 这个工作量非常大，我们在 Follow 的过程中，没有做这个。

15. Summary statistics

16. Data insight

19. Notes
 T-statistics for MROA (MIROA) are potentially overstated, as there is
a lack of independence across observations for a given CEO. Using
the Huber/White/sandwich robust method with adjustment for
within-cluster correlations for each CEO
 Pearson correlation test , correlations<0.5
 Ensure multicollinearity is not a problem, we calculate the variance
inflation factors (VIF)for each independent variable. These VIFs
never exceed 2, which suggest no serious multicollinearity problems

20. Robustness
 Different incentives to serve political and social objectives
when their firms are making profits;
1. Central government VS Local governments:
 Shareholders of firms owned by central government place greater
emphasis on social and political objectives than those of firms owned by
local governments
 Fiscal decentralization induces local governments to place greater
emphasis on economic performance relative to political objectives

21. Robustness
2、Firms in regions with different fiscal conditions：
 Bai et al. (2000) show local governments in regions with poor fiscal
conditions have a greater incentive to use SOEs to serve their social
and political objectives.
 Chen et al. (2004) show listed Chinese firms appoint more politically
connected CEOs if they are located in regions with larger fiscal deficits.
 LOCAL：a dummy variable that equals 1 if a listed firm is owned by a
local government and 0 if it is owned by the central government；
 EFICIT：dummy variable that equals 1 if the province‘s budgetary
deficit is larger than the median level of the country；their interaction
terms with performance measures

22. Robustness

23. Robustness
 Use either ages 60 or 65 as the benchmark for the classification
of forced retirement
 Include turnovers that are associated with legal disputes as
forced turnovers.
 Findings
 Consistent results are obtained with these alternative classification
schemes.

24. Robustness
 Frequency of board meetings
 Involvement of a board chairperson in the management：a dummy
variable that equals 1 if a chairperson receives a salary and 0 if he
or she receives only an honorarium
 Findings
 A greater frequency of board meetings is positively and significantly related to
forced turnovers;
 Existence of a chairperson who receives a salary is negatively related to the
turnover rate
 All of the other results pertaining to the performance–turnover relationship are
retained

25. Managerial turnover and post-performance results

26. Contributions
 Provides a useful addition to the existing literature on monitoring
activities in state-owned firms
 Relevant to not-for-profit and collectively owned organizations,
which are similarly characterized by the absence of dominant
private owners and the presence of multiple objectives
 Have some implications for the performance –turnover
relationship in private firms
 Whether private firms also have different turnover–performance
sensitivities when they experience different levels of financial
performance is another issue worthy of future investigation