Sie sind auf Seite 1von 12

J. Account.

Public Policy 30 (2011) 89100

Contents lists available at ScienceDirect

J. Account. Public Policy


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

Controlling shareholders tunneling and executive


compensation: Evidence from China
Kun Wang , Xing Xiao 1
School of Economics and Management, Tsinghua University, Beijing 100084, China

a r t i c l e

i n f o

a b s t r a c t
Conict of interests between controlling shareholders and minority
shareholders could affect executive compensation contracts. In this
paper, we use data on Chinese listed companies and show that controlling shareholders tunneling reduces the pay-performance sensitivity of executive compensation. These results suggest that
while incentive payment schemes are generally adopted in Chinese
listed companies, controlling shareholders who obtain private benets from listed companies have less incentive to strengthen the
relationship between executive pay and rm performance.
2010 Elsevier Inc. All rights reserved.

1. Introduction
Performance-based incentive payment schemes provide a crucial mechanism to alleviate the
classic ownermanager conict as described by Berle and Means (1932) and Jensen and Meckling
(1976). The conventional wisdom is that the presence of a blockholder could strengthen the payperformance sensitivity of executive compensation, as a large stake in the company motivates the
blockholder to monitor managers through incentive compensation contracts (Hartzell and Starks,
2003). A well-known feature of the ownership structure of Chinese listed companies is that ownership is highly concentrated. Previous studies, however, nd that the link between executive compensation and rm performance is weak in China compared to that in developed markets such as the US
(Firth et al., 2006).
We argue that conict of interests between controlling shareholders and minority shareholders
could hamper the adoption of incentive payment schemes in China. For companies with concentrated
ownership, controlling shareholders could obtain private benets from their controlling positions
through various forms of self-dealing transactions such as selling assets, goods and services to listed
Corresponding author. Tel.: +86 10 62795167; fax: +86 10 62783540.
1

E-mail addresses: wangk@sem.tsinghua.edu.cn (K. Wang), xiaox@sem.tsinghua.edu.cn (X. Xiao).


Tel.: +86 10 62795464; fax: +86 10 62783540.

0278-4254/$ - see front matter 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jaccpubpol.2010.09.014

90

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

companies at high prices or transferring assets from listed companies to member rms under controlling shareholders control at low prices. Such transactions are referred to as tunneling by Johnson
et al. (2000). The private benets acquired from tunneling are relatively large in developing countries
that do not have effective legal enforcement and corporate governance to protect monitory shareholders interests (Shleifer and Vishny, 1997).
Tunneling is usually achieved through collusion between controlling shareholders and executives.
Since tunneling reduces rm performance (Cheung et al., 2006; Jiang et al., 2010), a strong association
between executive compensation and rm performance would strengthen executives incentives to
increase rm performance and reduce their willingness to collude with controlling shareholders.
Therefore, controlling shareholders engaging in tunneling activities have less incentive to demand
high pay-performance sensitivity in executive compensation. In addition, controlling shareholders
are partially responsible for the resulting rm performance when they tunnel resources from the companies. Performance measures then become less informative of managerial effort and thus less useful
in setting executive compensation. As a result, we predict that the pay-performance sensitivity of
executive compensation is lower for rms that are subject to tunneling by their controlling shareholders than those that are not, and progressively so as the degree of tunneling gets larger.
We focus on monetary rewards and do not study non-pecuniary compensation in the paper. Cao
et al. (2009) show that state-owned rms could substitute weak pay-performance sensitivity with
promotion-based incentives. Since non-pecuniary compensation is typically provided by controlling
shareholders, it is likely to be based on how executives help benet controlling shareholders rather
than how they create value for all shareholders. Controlling shareholders are unlikely to shift away
from pecuniary to non-pecuniary compensation to incentivize managers to improve rm performance
when they could tunnel resources from the rms. Thus, the availability of non-pecuniary compensation from controlling shareholders would further divert managers attention away from rm performance and strengthen the negative association between tunneling and pay-performance sensitivity
that we predict.
Following previous studies, we measure tunneling by the amount of cash transferred from listed
rms to their controlling shareholders (Jiang et al., 2010; Jian and Wong, 2010). Using a large sample
of Chinese listed companies, we identify a positive and signicant association between executive
compensation and rm performance, after controlling for ownership structure and related factors that
affect executive compensation (see, e.g., Core et al., 1999; Jackson, et al., 2008). More importantly, we
show that pay-performance sensitivity is signicantly negatively associated with controlling shareholders tunneling behavior, consistent with our prediction. The results remain qualitatively unchanged after we use a two-stage regression to address the concern that controlling shareholders
tunneling and pay-performance sensitivity are both affected by ownership characteristics and other
common factors.
Our study adds to the small but growing literature that examines executive compensation outside
of the United States. Although there have been extensive studies on executive compensation in the US
(Murphy, 1999), research examining the issue in other countries is scarce and especially so for developing markets. The institutional environments including legal protection and corporate governance in
developing countries are signicantly different from that in the US. Conclusions from studies of executive compensation using the US data may not be directly applicable to developing markets. In addition to cross-country studies, DeFond and Huang (2004) and Kanagaretnam et al. (2009) call for
further research into a single country other than the US.
Our study using Chinese listed companies is interesting and important for two reasons. First, China
has a relatively underdeveloped legal environment compared to that of developed countries. The
importance of the legal system for preventing expropriation by controlling shareholders has been well
emphasized by La Porta et al. (1998). Effective corporate governance, which is characteristic of many
developed economies with strong legal protection, is absent in China. So the protection of minority
shareholders is relatively weak. This implies a greater degree of tunneling and higher test power to
examine its impact on executive compensation.
Second, high concentration of ownership is prevalent among Chinese public companies. On average, the equity ownership held by the largest shareholder of a rm is more than 40% (Allen et al.,
2005). The concentrated ownership structure implies that the classical ownermanager conict is

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

91

likely to be of less concern because controlling shareholders have incentives to monitor managers.
Rather, conict of interests arises between controlling shareholders and minority shareholders, making the presence of tunneling by controlling shareholders a distinct feature. Although the recently
introduced rules and guidelines on executive compensations refer to regulations and practices of US
rms,2 the unique ownership structure of Chinese listed companies is likely to have signicant impact
on the enforcement of these rules.
From minority shareholders viewpoint, lack of rm performance-based compensation (either
pecuniary or non-pecuniary) and thus lack of incentives to improve rm performance is an important
issue, even though we focus only on pecuniary compensation. If one takes controlling shareholders
incentive and ability to tunnel resources from listed companies as given, the low pay-performance
sensitivity we document might be part of an efcient contract to deal with the agency problem between controlling shareholders and managers. However, tunneling reduces overall rm value (Cheung
et al., 2006; Jiang et al., 2010) and need not be taken for granted. Therefore, such contracts are not
globally efcient as they cannot prevent controlling shareholders from expropriating the minority
shareholders.
Regulators should constrain controlling shareholders tunneling, increase the use of incentive payment schemes, and improve the overall contracting efciency. Over the years, various regulatory
actions have been taken in China, such as increased transparency of related party transactions and
strengthened board independence, encouraged adoption of US practices of incentive payment
schemes (e.g., stock options), among others. Our study suggests that a key ingredient in the effective
use of incentive payment schemes in Chinas future reforms is the reduction in controlling shareholders tunneling behavior and their role in setting executive compensation.
The rest of this paper is organized as follows. Section 2 describes the data and research design.
Section 3 discusses the empirical results and sensitivity tests, and Section 4 concludes the paper
and discusses the implications of our study for policy makers in China.
2. Sample, data, and research design
2.1. Sample and data
In China, the disclosure of executive compensation in the annual reports of listed companies has
been required since 1998 and available in the China Stock Market and Accounting Research (CSMAR)
database from 1999. Total cash compensation of the three highest paid executives in each company is
disclosed, where compensation consists of base salary, bonuses and commissions.3 We obtain data of
total compensation from 1999 to 2005 and use one third of the reported amount to measure the average
level of executive compensation. In line with prior studies, we exclude rms in the nancial industry. Our
nal sample consists of 6670 rm-year observations with executive compensation and 5150 rm-year
observations with changes in executive compensation.
Table 1 Panel A provides the summary statistics on the level and change of executive compensation
for our sample. The mean (median) compensation is 152,000 (105,000) RMB yuan for overall sample.
Compensation increased signicantly over the seven-year sample period, from 40,000 RMB yuan in
1999 to 206,000 RMB yuan in 2005. Changes in executive compensation are positive in all years, with
an average (median) of 25,000 (11,000) RMB yuan. The rising compensation in listed companies suggests that executive pay has become an important incentive scheme during the economic development in China.
CSRC required listed companies to disclose capital transfers to their related parties since 1997 in
annual reports. Jiang et al. (2010) provide compelling evidence that inter-corporate lending is used
by controlling shareholders as a tool to tunnel from listed companies. Jian and Wong (2010) suggest
that controlling shareholders transfer cash back from listed companies in the form of inter-corporate
2
For example, The Code of Corporate Governance for Listed Firms in China states that management should be selected on the basis
of competition and compensation should include incentive pay, and China government encourages adoption of stock options in
executive compensation contracts since 2006 (Act of Granting Stock Option in State-owned listed companies).
3
We focus on cash compensation in the paper because stocks and stock options were hardly used during our sample period.

92

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

Table 1
Descriptive statistics.
Level variable
Year

Obs.

Change variable
Obs.

Mean

Std. dev.

Median

Panel A: executive compensation by year (in thousand RMB yuan)


1999
394
40
32
27
2000
317
49
44
34
2001
1024
105
220
70
2002
1157
129
127
92
2003
1210
164
157
120
2004
1286
199
203
146
2005
1282
206
196
156

NA
261
297
1013
1129
1188
1262

9
27
29
33
33
11

1
3
2
2
2
3

3
12
13
15
13
5

Total

5150

25

11

6670

Mean

Std. dev

152

181

Median

105

Entire sample
Obs.
Panel B: controlling shareholders
1999
394
2000
317
2001
1024
2002
1157
2003
1210
2004
1286
2005
1282
Total

6670

Sample with TUL > 0


Mean%
tunneling
1.47
1.43
1.12
0.88
0.59
0.80
0.72
0.88

Std. dev.%

Obs.

Mean%

Std. dev.%

4.18
4.52
3.90
3.30
2.90
3.44
3.23

120
106
310
314
202
298
261

4.82
4.27
3.69
3.24
3.50
3.45
3.55

6.44
7.01
6.40
5.71
6.35
6.48
6.42

3.48

1611

3.63

6.33

Panel A presents summary statistics of executive compensation and its changes by year. Executive compensation is measured
by total compensation of three highest paid executives divided by three. Change in executive compensation is calculated as
executive compensation in year t subtracts that in year t  1. Both variables are denoted in thousand RMB yuan.
Panel B presents summary statistics of controlling shareholders tunneling by year. TUL is measured as the amount of cash
transferred from listed companies to their controlling shareholders disclosed in the footnote of year-end balances of Other
Receivables deated by total assets of listed companies in the same scal year. Mean and std. dev. of tunneling are denoted in%
value.

lending if they prop prots into the companies earlier. Following the denition of controlling shareholders tunneling in these studies, we hand-collect the amount of cash transferred from listed companies to their controlling shareholders at the end of year, disclosed in the footnote of year-end
balances of Other Receivables in annual reports. Controlling shareholders tunneling (TUL) is measured as the ratio of these cash transfers over total assets, winsorized at the top and bottom 1%.
Descriptive statistics on the tunneling measure are reported in Panel B of Table 1 for the entire
sample and the subsample of rms with tunneling. Among the 6670 rm-year observations, nearly
a quarter (1611) transfers cash to controlling shareholders. The average tunneling is 0.88% of total assets for the entire sample and 3.63% (standard deviation 6.33%) for the subsample where tunneling
exists.
Since 1997, the Chinese government has taken various measures to protect minority investors from
expropriation by controlling shareholders though a series of regulations (see Jiang et al. (2010) for a
detailed summary on regulatory reforms and policies). We nd that average tunneling by controlling
shareholders decreased from 1.47% in 1999 to 0.72% in 2005 for the entire sample. However, due to
the concentrated ownership structure and weak legal enforcement, the expropriation of minority
shareholders is still widespread among Chinese rms. Jiang et al. (2010) conclude that regulation
alone is not enough to prevent controlling shareholders tunneling, when minority shareholders have
limited litigation channels and regulators have limited jurisdiction over the controlling entities.
2.2. Research design
Following prior literature, we estimate pooled cross-sectional regressions using a levels specication and a changes specication to examine the relationship between executive compensation and

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

93

rm performance. The changes specication controls for the differences in CEO-specic factors between rms (Murphy, 1999; Jackson et al. 2008). The two cross-sectional models are as follows:

LnCompensation b0 b1 ROA b2 D TUL b3 ROA  D TUL b4 TUL b5 ROA  TUL


b6 Control d

DCompensation b0 b1 DOI b2 D TUL b3 DOI  D TUL b4 TUL b5 DOI  TUL


b6 Control d

In the levels specication of Model (1), we use natural logarithm of executive compensation as the
dependent variable to mitigate the effect of outliers on the regression results. Performance is
measured by return on assets (ROA), calculated as earnings before interests and tax divided by total
assets.4 D_TUL is a dummy variable that equals 1 for TUL > 0, and zero otherwise. Coefcient b3 on the
interaction of ROA and D_TUL measures the difference in pay-performance sensitivity between rms with
and without controlling shareholders tunneling. Coefcient b5 on the interaction of ROA and TUL
measures whether pay-performance sensitivity changes with the degree of tunneling within rms with
tunneling. Based on earlier discussion, we predict b3 and b5 to be negative.
Prior studies nd that state ownership and ownership concentration affect the pay-performance
sensitivity as well as level of executive compensation in China (Firth et al., 2006; Kato and Long,
2006; Gu et al., 2010).5 We control for these ownership characteristics with three variables: STATE is
a dummy variable that equals 1 if listed company is ultimately controlled by the government and 0
otherwise6; LSH is controlling shareholders ownership percentage measured as the proportion of shares
owned by the largest shareholder; HERF is ownership concentration of other large shareholders measured as the Herndahl index of shareholdings of the second to fth largest shareholders. A higher level
of LSH implies controlling shareholders stronger inuence on listed companies because of representation
and voting rights on the board. A higher HERF indicates that other blockholders have stronger power to
monitor the controlling shareholders. We include these variables and their interaction with ROA in Model
(1). In addition to affecting compensation directly, state ownership and ownership concentration can
also affect tunneling itself. We address this issue in Section 3.2.
For other economic determinants of compensation documented in prior studies (e.g., Core et al.,
1999, Jackson et al., 2008), we consider: rm size (SIZE), measured as natural logarithm of total assets;
nancial leverage ratio (LEV), measured as total liabilities over total assets; market to book ratio (MB),
measured as market value of equity over book value of total equity; cash ows from operation (CFO),
measured as net operating cash ows deated by total assets; one-year shareholder return (RET), measured as the percentage change in stock price; and regional economical development (GDP), measured
as natural logarithm of GDP per capita of the province where the listed company is located. We also
use industry dummy variables and year dummy variables to capture variations across industries and
years.
To further test the pay-performance sensitivity, we use Model (2) to examine the relation between
change in executive compensation and change in rm performance. Similar to Jensen and Murphy
(1990), change in executive compensation (in thousand RMB yuan) is used as the dependent variable,
and change in operating income (DOI) is used as an independent variable, measured as the difference
between earnings before interest and tax in year t versus year t  1 (in million RMB yuan). Similar
to Model (1), we include D_TUL and TUL and their interactions with DOI to capture the impact of
tunneling on pay-performance sensitivity. Other economic factors are measured mostly in changes
except dummy variables that usually do not change across years.
4
We measure rm performance by accounting earnings because shares held by controlling shareholders in China are nontradable; thus, controlling shareholders pay more attention to accounting performance (Firth et al., 2006). See Allen et al. (2005)
for the share classication in Chinas stock market in detail.
5
Previous results on the effect of state ownership on pay-performance sensitivity appear to depend on the performance
measure used. Relative to listed rms with private or foreign controlling shareholders, those with state-owned enterprises as
controlling shareholders have higher sensitivity to accounting performance (Firth et al., 2006, Table 3) but lower sensitivity to
stock performance (Firth et al., 2006, Table 3; Kato and Long, 2006, Table 5). We use accounting performance in this paper.
6
We also use a more rened classication of types of controlling shareholders including the central government, local
government, private entities, and foreign entities. They do not affect our main results qualitatively.

94

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

Table 2
Firm performance and characteristics.
Variables

Entire sample
Obs.

Mean

TUL > 0
Std. dev.

Median

TUL = 0

Obs.

Mean

Std. dev.

Median

Panel A: descriptive statistics on level of variables


ROA
6670
0.01
0.10
0.03
STATE
6670
0.75
0.43
1.00
LSH
6664
0.43
0.17
0.42
HERF
6664
0.02
0.03
0.01
SIZE
6670 21.09
0.92
21.01
LEV
6670
0.48
0.18
0.48
MB
6665
3.37
3.31
2.48
CFO
6669
0.05
0.08
0.05
RET
6665 0.17
0.28
0.22
GDP
6523
9.49
0.67
9.46

Obs.

Mean

Std. dev.

1611
1611
1610
1610
1611
1611
1611
1611
1611
1585

0.01
0.80
0.46
0.02
21.07
0.48
3.42
0.04
0.17
9.34

Panel B: descriptive statistics on change of variables


DOI (million) 6351
8.75 130.35
2.52
DLSH
5998 0.01
0.04
0.00
DHERF
5998
0.00
0.01
0.00
DSIZE
6351
0.08
0.22
0.07
DLEV
6351
0.04
0.12
0.02
DMB
6341 0.71
3.10
0.55
DCFO
5339
0.00
0.10
0.00
DRET
5334 0.02
0.40
0.00
DGDP
6523
0.12
0.06
0.11

1611
1489
1489
1611
1611
1611
1295
1295
1585

3.22
0.01
0.00
0.06
0.04
0.60
0.00
0.04
0.12

Median

0.11
0.40
0.17
0.03
0.88
0.18
3.29
0.08
0.29
0.61

0.02
1.00
0.46
0.00
20.99
0.48
2.63
0.04
0.24
9.29

5059
5059
5054
5054
5059
5059
5054
5058
5054
4938

0.02
0.74
0.42
0.02
21.09
0.47
3.35
0.05
0.17
9.54

0.10
0.44
0.17
0.03
0.93
0.18
3.32
0.08
0.28
0.68

0.03
1.00
0.40
0.01
21.02
0.48
2.44
0.05
0.22
9.50

138.37
0.04
0.01
0.22
0.12
3.18
0.10
0.42
0.05

2.44
0.00
0.00
0.06
0.03
0.53
0.00
0.00
0.11

4740
4509
4509
4740
4740
4730
4044
4039
4938

12.57
0.01
0.00
0.09
0.03
0.75
0.00
0.01
0.12

127.47
0.04
0.01
0.22
0.11
3.08
0.10
0.39
0.06

4.20
0.00
0.00
0.08
0.02
0.56
0.00
0.00
0.12

Variables are dened as follows: ROA = return on assets, measured as income before interest and tax deated by total assets;
STATE = a dummy variable that equals one if listed company is ultimately controlled by the state and zero otherwise; LSH = the
percentage of shares owned by the largest shareholder; HERF = the Herndahl index of shareholdings of the second to fth
largest shareholders; SIZE = log of total assets; LEV = nancial leverage, measured as total liability over total assets;
MB = market-to-book ratio, measured as market value of equity over book value of equity; CFO = operating cash ows scaled by
total assets; RET = one-year shareholder return; measured as the percentage change in stock price; GDP = log of the per capita
GDP of province where listed company is located. DOI = change in operating income, dened as the difference between earning
before interest and tax in year t and year t  1, denoted in million RMB yuan. The change variable of LSH, HERF, SIZE, LEV, MB,
CFO, RET and GDP are denoted as DLSH, DHERF, DSIZE, DLEV, DMB, DCFO, DRET and DGDP, respectively. All the change variables
are measured as difference of the variable between year t and t  1.

Data on accounting prots and rm characteristics are obtained from CSMAR database. All continuous variables are winsorized at the top and bottom 1%. Table 2 presents the descriptive statistics of
the main variables for the entire sample and the subsamples with and without controlling shareholders tunneling. Regardless of the measures used, rm performance is worse in rms with TUL > 0. Taking ROA as an example, the average ROA is 1% for the entire sample, and negative 1% and positive 2% for
rms with TUL > 0 and TUL = 0, respectively. Eighty percent of rms are ultimately controlled by government (STATE = 1) in the subsample with tunneling, compared to 74% in the sample without tunneling. Other rm characteristics are not signicantly different between the two subsamples. For
example, both groups have an average leverage of about 0.50.

3. Empirical results
3.1. Primary regression results
Table 3 Panel A reports the regression results for the levels specication of Model (1). t-Statistics in
all regressions are based on standard errors clustered by rm. Coefcients on industry and year dummy variables are not reported. We rst estimate the model without considering controlling shareholders tunneling in column (1). The coefcient on ROA is positive and signicant at the 10% level,
suggesting that executive compensation is sensitive to rm performance. State-owned rms have
higher pay-performance sensitivity than other rms, consistent with the ndings of Firth et al.

95

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100


Table 3
Relation between controlling shareholders tunneling and pay-performance sensitivity of executive compensation.
Entire sample
(1)

TUL > 0
(2)

TUL = 0
(3)

Panel A: regressions of executive compensation on performance and tunneling


Intercept
2.2720***
1.4905*
2.5078***
(5.39)
(1.93)
(5.54)
0.1448
1.0300*
ROA
0.8505*
(1.71)
(0.21)
(1.73)
STATE
0.0476
0.0532
0.0488
(1.31)
(0.87)
(1.25)
ROA  STATE
0.9813***
0.3375
1.4031***
(3.61)
(0.88)
(4.13)
LSH
0.5937***
0.5146***
0.5757***
(4.78)
(2.74)
(4.09)
ROA  LSH
0.3055
1.2839
0.2166
(0.23)
(0.77)
(0.13)
HERF
0.8882
1.3529
0.6828
(1.49)
(1.45)
(1.01)
ROA  HERF
5.9780
2.4312
4.9463
(1.09)
(0.29)
(0.73)
SIZE
0.2786***
0.3128***
0.2710***
(13.54)
(8.25)
(12.14)
LEV
0.3316***
0.3961**
0.3050***
(3.25)
(2.30)
(2.75)
MB
0.0108**
0.0195**
0.0081*
(2.48)
(2.25)
(1.65)
CFO
0.2793*
0.3658
0.2093
(1.89)
(1.49)
(1.24)
RET
0.0121
0.0031
0.0274
(0.36)
(0.04)
(0.71)
GDP
0.3077***
0.3202***
0.2970***
(12.10)
(7.73)
(10.90)
D_TUL
ROA  D_TUL

Entire sample
(4)

Entire sample
(5)

2.3171***
(5.49)
1.0035**
(2.03)
0.0465
(1.28)
1.0179***
(3.75)
0.5726***
(4.60)
0.2951
(0.22)
0.8918
(1.50)
5.6887
(1.04)
0.2788***
(13.54)
0.3238***
(3.19)
0.0104**
(2.40)
0.2543*
(1.71)
0.0142
(0.42)
0.3033***
(11.90)
0.0720***
(2.61)
0.6124***
(2.62)

2.3598***
(5.53)
1.0274**
(2.08)
0.0472
(1.30)
1.0493***
(3.88)
0.5672***
(4.56)
0.2161
(0.16)
0.8591
(1.45)
6.0015
(1.10)
0.2775***
(13.41)
0.3215***
(3.16)
0.0110**
(2.52)
0.2340
(1.59)
0.0143
(0.43)
0.3017***
(11.84)
0.0341
(1.16)
0.6862***
(2.87)
1.1675**
(2.48)
0.3103***
(4.00)

6512
0.4519

6512
0.4540

TUL
ROA  TUL
Observations
Adjusted R2

6512
0.4501

1584
0.4483

4928
0.4401

Panel B: regressions of change of executive compensation on change in rm performance and tunneling


Intercept
2.5978
17.4194
1.8419
3.8508
(0.41)
(1.59)
(0.24)
(0.61)
*
**
DOI
0.0349
0.0207
0.0553
0.0465**
(1.68)
(0.61)
(2.15)
(2.17)
STATE
3.0185
6.9074
2.3212
3.3056
(1.23)
(1.53)
(0.80)
(1.35)
DOI  STATE
0.0363
0.0607*
0.0219
0.0366*
(1.63)
(1.73)
(0.79)
(1.65)
DLSH
0.2978
0.7383
0.1679
0.3413
(0.93)
(1.08)
(0.46)
(1.07)
DOI  DLSH
0.6105***
1.1563***
0.0637
0.6546***
(2.84)
(3.06)
(0.23)
(3.03)
DHERF
2.6872**
2.4647
2.2406
2.8441**
(2.12)
(0.94)
(1.53)
(2.24)
DOI  DHERF
0.7144
3.6386***
2.3550**
0.8969
(1.19)
(4.61)
(2.58)
(1.48)
DSIZE
0.3311***
0.2048**
0.3643***
0.3244***
(6.50)
(2.34)
(5.95)
(6.35)

4.0173
(0.63)
0.0458**
(2.14)
3.2631
(1.33)
0.0375*
(1.69)
0.3427
(1.07)
0.6530***
(3.03)
2.8557**
(2.25)
0.8728
(1.44)
0.3236***
(6.25)
(continued on next page)

96

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

Table 3 (continued)

DLEV
DMB
DCFO
DRET
DGDP

Entire sample
(1)

TUL > 0
(2)

TUL = 0
(3)

Entire sample
(4)

Entire sample
(5)

0.2446**
(2.58)
0.1382
(0.37)
1.7574
(0.16)
5.2768*
(1.65)
0.3618
(1.57)

0.2623
(1.63)
0.4587
(0.75)
11.6958
(0.62)
1.2773
(0.23)
0.9082*
(1.92)

0.2522**
(2.20)
0.3155
(0.70)
2.5467
(0.20)
6.1604
(1.60)
0.2584
(0.98)

0.2403**
(2.53)
0.1233
(0.33)
1.0343
(0.10)
5.4866*
(1.71)
0.3398
(1.47)
6.1559**
(2.47)
0.0393**
(2.20)

0.2431**
(2.54)
0.1371
(0.37)
1.1092
(0.10)
5.5038*
(1.72)
0.3408
(1.48)
5.5781***
(2.02)
0.0368**
(2.02)
17.2048
(0.51)
0.0133
(0.95)

5004
0.0481

1216
0.0663

3788
0.0510

5004
0.0499

5004
0.0497

D_TUL

DOI  D_TUL
TUL

DOI  TUL
Observations
Adjusted R2

All variables except D_TUL are dened in Table 1 and 2. D_TUL is a dummy variable that equals 1 if controlling shareholders
tunneling exists (TUL is positive) and 0 otherwise. In Panel B, the coefcients of DLSH, DHERF, DSIZE, DLEV, DGDP is deated by
100 for presentation in the table. We estimate regression with year xed effects and industry xed effects (unreported). The tstatistics are in parentheses, adjusted for clustering by rm.
*
Denote signicance at the 10% level.
**
Denote signicance at the 5% level.
***
Denote signicance at the 1% level.

(2006). Ownership concentration of controlling shareholders appears to negatively affect the level of
compensation but not pay-performance sensitivity. In addition, compensation decreases in nancial
leverage and increases in rm size, market to book ratio, cash ow from operation, and regional economic development.
When we separate the sample into subsamples with and without controlling shareholders tunneling in columns (2) and (3), we nd that the pay-performance sensitivity is signicantly lower
for rms with tunneling than for rms without tunneling. Specically, the coefcient of ROA is
0.1448 (statistically insignicant) in column (2) and 1.0300 (signicant at the 10% level) in column
(3). This nding is consistent with our prediction that the ability of controlling shareholders to tunnel resources from the rms reduces the association between executive compensation and rm performance. The results in columns (2) and (3) are conrmed by those in column (4) where we use the
entire sample but allow the pay-performance sensitivity to differ between rms with and without
tunneling. The coefcient on the interaction of D_TUL and ROA is 0.6124 and signicant at the
1% level, suggesting that, on average, pay-performance sensitivity of rms with tunneling is lower
than that of rms without tunneling by more than half. In column (5) we estimate the full Model
(1). The coefcient on ROA  D_TUL and ROA  TUL are both negative and signicant at the 1% level.
The negative coefcient on ROA  TUL indicates that, within rms with tunneling, pay-performance
sensitivity further decreases in the degree of tunneling.7
We also nd that controlling shareholders tunneling itself has a negative impact on executive compensation. For example, the coefcient on D_TUL in column (4) is 0.0720 and the coefcient on TUL is
1.1675 in column (5). There are two possible reasons for this result. First, tunneling is negatively related to executive ability. When controlling shareholders seek private benets from listed companies,
they tend to hire docile executives instead of capable managers. As a result, these executives get less well
7

If we drop D_TUL and D_TUL  ROA from column (5), coefcient on TUL  ROA remains similar and more signicant.

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

97

paid. Second, executives that help controlling shareholders tunnel get compensated in non-pecuniary
form such as promotion within controlling shareholders groups, which we do not address in this paper.
The results for the changes specication of Model (2) are similar and presented in Table 3 Panel B.
The association between DOI and change in executive compensation for entire sample is 0.0349 and
signicant at the 10% level (column (1)). This implies that if operating income improves by one standard deviation (130.35 million), executive compensation would increase by about 4500 RMB yuan. For
the subsample of rms with controlling shareholders tunneling (column (2)), increases in operating
income are not signicantly associated with increases in compensation. In contrast, for the subsample
of rms without tunneling (column (3)), one standard deviation increase in operating income is associated with a signicant compensation increase of about 7000 RMB yuan.
The results in columns (4) and (5) using the entire sample are consistent with those of columns (2)
and (3). The coefcient on interaction of rm performance and D_TUL is negative and signicant at the
5% level. As long as controlling shareholders tunnel resources from the rms, pay-performance sensitivity is reduced. Specically, the pay-performance sensitivity is lower by 0.0393 for rms with tunneling. The coefcient on the interaction of rm performance and TUL has the predicted negative sign,
although it is insignicant, suggesting that the act of tunneling itself, rather than the degree of tunneling reduces the pay-performance sensitivity. Taken together, our ndings support the prediction that
expropriation of controlling shareholders results in signicantly weakened link between executive
payment and rm performance.
3.2. Controlling for determinants of controlling shareholders tunneling
Controlling shareholders tunneling and pay-performance sensitivity might be affected by some
common factors such as ownership structure and ownership type. For example, more concentrated
ownership of controlling shareholders makes it easier for them to tunnel resources from the rms.
When government is the controlling shareholder, related party transactions are relatively common because most of the listed companies are carved out from their state-owned groups (Aharony et al.,
2010). The common factors could lead to an association between tunneling and executive compensation rather than tunneling per se. To address this concern, we adopt a two-stage regression procedure.
First, we model controlling shareholders tunneling as a function of economic determinants proposed
in prior studies. Then we examine the association between pay-performance sensitivity and the unexplained tunneling, which captures the impact of tunneling on executive compensation contract unrelated to ownership characteristics. The model we employ for tunneling is as follows:

TUL b0 b1 LagTUL b2 LagROA b3 LSH b4 STATE b5 SIZE b6 LEV b7 MKT e

where LagTUL is the TUL in year t  1; LagROA is ROA in year t  1; MKT is a marketization index measuring the development of the regional market in which the rm is located created by Fan and Wang
(2003), where higher values indicate greater regional market development. Other variables are dened the same as before. Industry and year dummies are also included. The explanatory variables
in Model (3) are primarily based on Jiang et al. (2010) and Jian and Wong (2010).
Panel A of Table 4 presents the regression coefcients for Model (3). The coefcient on LagTUL is
signicantly positive, suggesting that controlling shareholders tunneling is persistent overtime. Similar to Jiang et al. (2010), controlling shareholders tunneling decreases with prior rm performance,
rm size and market development. Findings on the impact of state ownership are inconclusive in prior
studies. Jiang et al. (2010, Table VI) nd that state-owned rms are less tunneled by controlling shareholders compared to non-stated-owned rms, whereas Jian and Wong (2010, Table 6) nd the opposite. We nd that STATE is not signicantly related to controlling shareholders tunneling after
controlling for other factors.8 The regression yields R2 of 30.71%, implying that the model explains a signicant portion of the variation in controlling shareholders tunneling.
8
One possible reason for the inconclusive results is that tunneling is measured differently. Jian and Wong (2010) use unexpected
cash lending to the largest shareholders, while Jiang et al. (2010) use raw values of cash lending to all other parties (include all
related parties and non-related parties). We measure tunneling as cash lending to controlling shareholders only. We also use a
more rened classication of ownership types including ownership by the central government, local government, private entities
and foreign entities. None of them is signicant.

98

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

Table 4
2SLS sensitivity tests.
Coefcients
Panel A: Regression of controlling shareholders tunneling on economic determinants
Intercept
0.0305***
LAG_TUL
0.5394***
LAG_ROA
0.0037**
LSH
0.0099***
STATE
0.0010
SIZE
0.0016***
LEV
0.0113***
MKT
0.0006***

t-Stat
(3.34)
(51.09)
(2.13)
(4.08)
(1.04)
(3.43)
(5.05)
(3.46)

Observations
6651
0.3071
Adjusted R2
Panel B: Regressions of executive compensation on performance and unexplained tunneling

Intercept
PERF
STATE
PERF  STATE
LSH
PERF  LSH
HERF
PERF  HERF
SIZE
LEV
MB
CFO
RET
GDP
UNEXP_TUL
PERF  UNEXP_TUL
Observations
Adjusted R2

Level test (1)


PERF = ROA

Change test (2)


PERF = DOI

2.2245***
(8.86)
0.9875***
(3.19)
0.0500**
(2.33)
1.0328***
(5.55)
0.6115***
(9.75)
0.2716
(0.43)
0.9155***
(2.62)
5.7919
(1.36)
0.2816***
(23.92)
0.3636***
(6.09)
0.0106***
(3.28)
0.2059**
(2.05)
0.0191
(0.51)
0.3122***
(21.57)
0.0316
(1.50)
0.5432***
(-2.89)

1.4613
(0.23)
0.0420**
(1.97)
3.0822
(1.26)
0.0364*
(1.67)
0.2952
(0.93)
0.6028***
(2.80)
2.6517**
(2.09)
0.7035
(1.17)
0.3393***
(6.63)
0.2442**
(2.57)
0.1076
(0.29)
2.0824
(0.19)
5.2081
(1.62)
0.3418
(1.48)
4.0366
(1.52)
0.0261
(-1.43)

6512
0.4508

5004
0.0486

Panel A presents cross-sectional regression coefcients of TUL on economic determinants. Regressions coefcients of executive
compensation on rm performance and unexpected tunneling are presented in Panel B. We report regression estimates of level test
as specied in Model (1) in Column (1) of Panel B (PREF is measured as ROA). The estimates of change test as specied in Model (2)
are reported in Column (2) of Panel B (PERF is measured as DO). Levels (change) of control variables are used accordingly and their
coefcient estimates are reported in Column 1 (2), except STATE. All variables except MKT and UNEXP_TUL are dened in Table 1 and
2. MKT is a comprehensive index measuring the development of the regional market in which the rm is located, created by Fan and
Wang (2003). The higher values of MKT indicate greater regional market development. UNEXP_TUL is a dummy variable that equals
1 if the residual TUL based on the rst stage regression in Panel A is greater than mean and 0 otherwise. We estimate regression
with year xed effects and industry xed effects (unreported). The t-statistics are in parentheses, adjusted for clustering by rm.
*
Denote signicance at the 10% level.
**
Denote signicance at the 5% level.
***
Denote signicance at the 1% level.

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

99

In the second stage, we construct a dummy variable denoted by UNEXP_TUL, which equals 1 if
unexplained tunneling or the residuals from Model (3) are greater than mean and 0 otherwise.9 We
re-estimate Models (1) and (2) replacing D_TUL with UNEXP_TUL, and report the results in Panel B of
Table 4. The coefcient on the interaction of rm performance and UNEXP_TUL is negative and at least
marginally signicant at the 10% level (one-tailed test). For example, the coefcient of ROA  UNEXP_TUL
in column (1) is 0.5432, comparable to that on ROA  D_TUL in column (4) of Table 3 Panel A. Overall,
above-average tunneling reduces the pay-performance sensitivity by one quarter or more. These results
indicate that controlling shareholders tunneling reduces pay-performance sensitivity of executive
compensation even after controlling for the inuence of ownership structure, ownership type as well
as other related factors. It does not appear that the impact of controlling shareholders tunneling on
pay-performance sensitivity is all attributable to ownership characteristics.

4. Conclusion
We show in this paper that the pay-performance sensitivity of executive compensation is lower in
rms where controlling shareholders tunnel resources for private benets compared to other rms.
Large quantities of tunneling transactions reduce rm performance and controlling shareholders demand on performance-based incentive payment schemes. They are also largely beyond the control of
executives, making performance measures less informative of managerial effort and less useful in setting executive compensation. In addition, controlling shareholders could provide non-pecuniary compensation such as promotion to further induce executives cooperation in tunneling activities. Given
that controlling shareholders have the incentive and ability to expropriate from the rm, the low
pay-performance sensitivity could be an efcient contract in dealing with the management agency
problem. From minority shareholders perspective, however, such contracts respond to controlling
shareholders maximization of their own private benets at the expense of minority shareholders.
Thus, they are not necessarily globally efcient in the sense of maximizing total shareholders wealth.
Our paper provides an alternative explanation for the weak association between executive pay and
performance in China documented in previous studies, that is, the widespread pursuit of private benets by controlling shareholders. The results suggest that the alignment of interest between controlling shareholders and minority shareholders is an important factor that determines the adoption of
incentive payment schemes.
The policy implication of our study is that regulation of executive compensation must be considered in light of other changes to the current system. In addition to encouraging the adoption of Western practices of incentive payment schemes such as granting stock options, the government should
reinforce the protection of minority shareholders from controlling shareholders expropriation. Controlling shareholders tunneling reduces not only the use of performance-based incentive schemes,
but also the effectiveness of such schemes even if they are imposed on listed companies. Executives
may not care much about rm performance after all if controlling shareholders are able to provide
non-pecuniary compensation to executives based on how they tunnel for the controlling shareholders.
The board and especially independent directors on the board should strengthen their role in reducing
controlling shareholders tunneling behavior. Likewise, compensation committee and especially independent directors in the committee should be more active in mitigating controlling shareholders role
in setting executive compensation.

Acknowledgements
We appreciate the helpful comments from Qi Chen, Zhaoyang Gu, two anonymous referees and
workshops participants at Tsinghua University. We acknowledge funding from the Projects
9
We also use the continuous measure of the residuals from Model (3). The coefcient on the interaction of this measure and
rm performance is even larger than the counterpart in Table 3 and similarly signicant. The larger coefcient is likely due to the
fact the residuals are smaller in magnitude than the original measure.

100

K. Wang, X. Xiao / J. Account. Public Policy 30 (2011) 89100

70872055 and 70902004 supported by National Natural Science Foundation of China, and support
from Center for Corporate Governance of Tsinghua University. All the remaining errors are ours.
References
Aharony, J., Wang, J., Yuan, H., 2010. Tunneling as an incentive for earnings management during the IPO process in China.
Journal of Accounting and Public Policy 29 (1), 126.
Allen, F., Qian, J., Qian, M., 2005. Law, nance, and economic growth in China. Journal of Financial Economics 77, 57116.
Berle, A., Means, G., 1932. The Modern Corporation and Private Property. Macmillan, New York, N.Y.
Cao, J., Lemmon, M., Pan, X., Tian, G., 2009. Political Promotion, CEO Compensation, and their Effect on Firm Performance.
Working Paper.
Cheung, Y., Rau, P., Stouraitis, A., 2006. Tunneling, propping, and expropriation: evidence from connected party transactions in
Hong Kong. Journal of Financial Economics 82 (2), 343386.
Core, J., Holthausen, R., Larcker, D., 1999. Corporate governance, chief executive ofcer compensation and rm performance.
Journal of Financial Economics 51, 493509.
DeFond, M., Huang, M., 2004. Investor protection and corporate governance: evidence from worldwide CEO turnover. Journal of
Accounting Research 42, 269312.
Fan, G., Wang, X., 2003. The report on the relative progress of marketization of each region in China. The Economic Science Press
(in Chinese).
Firth, M., Fung, P., Rui, O., 2006. Corporate performance and CEO compensation in China. Journal of Corporate Finance 12, 693
714.
Gu, Z., Wang, K., Xiao, X., 2010. Government Control and Executive Compensation: Evidence from China. Working Paper,
University of Minnesota and Tsinghua University.
Hartzell, J., Starks, L., 2003. Institutional investors and executive compensation. Journal of Finance 58 (6), 23512374.
Jackson, S., Lopez, T., Reitenga, A., 2008. Accounting fundamentals and CEO bonus compensation. Journal of Accounting and
Public Policy 27 (5), 374393.
Jensen, M., Meckling, W., 1976. Theory of the rm: managerial behavior, agency costs, and ownership structure. Journal of
Financial Economics 3, 305360.
Jensen, M., Murphy, K., 1990. Performance pay and top-management incentives. Journal of Political Economy 98, 225264.
Jian, M., Wong, T., 2010. Propping through related party transactions. Review of Accounting Studies 15 (1), 70105.
Jiang, G., Lee, C., Yue, H., 2010. Tunneling through intercorporate loans: The China experience. Journal of Financial Economics 98
(1), 120.
Johnson, S., La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2000. Tunneling. The American Economic Review 90 (2), 2227.
Kanagaretnam, K., Mathieu, R., Shehata, M., 2009. Usefulness of comprehensive income reporting in Canada. Journal of
Accounting and Public Policy 28 (4), 349365.
Kato, T., Long, C., 2006. Executive compensation, rm performance, and corporate governance in China: evidence from rms
listed in the Shanghai and Shenzhen stock exchanges. Economic Development and Cultural Change 54 (4), 945983.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and nance. Journal of Political Economy 106, 11131155.
Murphy, J.K., 1999. Executive Compensation. Handbook of Labor Economics, Part 2, vol. 3, pp. 24852563 (Chapter 38).
Shleifer, A., Vishny, R., 1997. A survey of corporate governance. Journal of Finance 52, 737783.

Das könnte Ihnen auch gefallen