Scholarly article on topic 'Profits and losses from changes in fair value, executive cash compensation and managerial power: Evidence from A-share listed companies in China'

Profits and losses from changes in fair value, executive cash compensation and managerial power: Evidence from A-share listed companies in China Academic research paper on "Economics and business"

CC BY-NC-ND
0
0
Share paper
Academic journal
China Journal of Accounting Research
OECD Field of science
Keywords
{"Profits and losses from changes in fair value" / "Executive compensation" / "Managerial power"}

Abstract of research paper on Economics and business, author of scientific article — Ruiqing Shao, Chunhua Chen, Xiangzu Mao

Abstract According to optimal contracting theory, compensation contracts are effective in solving the agency problem between stockholders and managers. Executive compensation is naturally related to firm performance. However, contracts are not always perfect. Managers may exert influence on the formulation and implementation of compensation contracts by means of their managerial power. As fair value has been introduced into the new accounting standards in China, new concerns have arisen over the relationship between profits and losses from changes in fair value (CFV) and levels of executive compensation. In this study, we find that executive compensation is significantly related to CFV. However, this sensitivity is asymmetric in that increases to compensation due to profits from changes in fair value (PCFV) are higher than reductions to compensation due to losses from changes in fair value (LCFV). Furthermore, we find that managerial power determines the strength of this asymmetry.

Academic research paper on topic "Profits and losses from changes in fair value, executive cash compensation and managerial power: Evidence from A-share listed companies in China"

Contents lists available at SciVerse ScienceDirect

China Journal of Accounting Research

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

Profits and losses from changes in fair value, executive cash compensation and managerial power: Evidence from A-share listed companies in China

Ruiqing Shao a'*, Chunhua Chen a,b, Xiangzu Mao a

a Lixin Accounting Research Institute, Shanghai Lixin University of Commerce, China b Business School of Nanjing University, China

ARTICLE INFO

ABSTRACT

Article history: Received 17 April 2012 Accepted 22 November 2012 Available online 23 December 2012

JEL classification:

Keywords:

Profits and losses from changes in fair value

Executive compensation Managerial power

According to optimal contracting theory, compensation contracts are effective in solving the agency problem between stockholders and managers. Executive compensation is naturally related to firm performance. However, contracts are not always perfect. Managers may exert influence on the formulation and implementation of compensation contracts by means of their managerial power. As fair value has been introduced into the new accounting standards in China, new concerns have arisen over the relationship between profits and losses from changes in fair value (CFV) and levels of executive compensation. In this study, we find that executive compensation is significantly related to CFV. However, this sensitivity is asymmetric in that increases to compensation due to profits from changes in fair value (PCFV) are higher than reductions to compensation due to losses from changes in fair value (LCFV). Furthermore, we find that managerial power determines the strength of this asymmetry.

© 2012 China Journal of Accounting Research. Founded by Sun Yat-sen University and City University of Hong Kong. Production and hosting by

Elsevier B.V. All rights reserved.

1. Introduction

In recent years, anecdotal evidence from China's securities markets suggests changing standards of executive compensation. In 2008, China Southern (Stock code: 600029) suffered a huge loss of 5 billion yuan, which

* Corresponding author. E-mail address: shaorq@lixin.edu.cn (R. Shao).

1755-3091/$ - see front matter © 2012 China Journal of Accounting Research. Founded by Sun Yat-sen University and City University of

Hong Kong. Production and hosting by Elsevier B.V. All rights reserved.

http://dx.doi.org/10.1016/j.cjar.2012.1L002

included 124 million yuan in losses from changes in fair value (LCFV). At the same time, the level of compensation for executives1 of China Southern jumped by 50%.2 In 2010, CIMC (Stock code: 000039) earned a net profit of 3 billion yuan, an increase of 213%, of which 76.84 million yuan was profit from changes in the fair value (PCFV) of derivative financial instruments. The compensation to CIMC's executives soared nearly 10-fold.3 According to annual financial reports for 2008, Air China (Stock code: 601111) and China Eastern (Stock code: 600115) incurred losses of 7.9 million yuan and 4.2 million yuan, respectively, of which losses from crude oil swaps accounted for 78.4% in the case of Air China and 42.2% in the case of China Eastern. As a result, the total compensation of the three highest-paid executives in these companies decreased by only 50,000 yuan for Air China executives and 73,000 yuan for China Eastern executives. These cases indicate that executive compensation tends to (1) increase far more than growth in earnings from PCFV, (2) increase despite LCFV and (3) change little despite huge losses from CFV.

According to optimal contracting theory, compensation contracts are effective in solving the agency problem between stockholders and managers. Given that a compensation contract is effective, executive compensation should be directly related to corporate performance. Performance is always designed as the core element of executive compensation contracts. However, contrary to common expectations, levels of executive compensation have virtually nothing to do with corporate performance in practice and the core influence of performance on executive compensation tends to disappear. According to an investigation from the Information Centre of Guangdong Province, state-owned enterprises (SOEs) pay high salaries to executives as rewards for their individual contributions and millions of yuan flow into the executives' pockets regardless of their firms' continuous losses.4 In addition, scholars have found that perfect contracts do not exist and that the formulation, implementation and efficiency of contracts are controlled by managerial power (Hambrick and Finkelstein, 1995; Core et al., 1999; Bebchuk et al., 2002; Duffhues and Kabir, 2008; Cheng and Indjejikian, 2009). In accordance with rational assumptions about economic self-interest, it is not surprising that executives manipulate compensation contracts. The control of operating returns by executives is evident in theoretical explanations of managerial power and in studies providing empirical evidence (Cyert et al., 2002; Garvey and Milbourn, 2006; Gopalan et al., 2008; Morse et al., 2008; Fahlenbrach, 2009). China's economic system provides fertile soil for breeding managerial power. Numerous empirical studies provide evidence that managerial power does influence executive compensation in China's listed companies (Zhang and Shi, 2005; Lu, 2008; Lyu and Zhao, 2008; Quan et al., 2010).

In 2006, fair value was introduced into the new accounting standards in China and the structure of accounting performance has changed since CFV became an item of operating income. As a basis for compensation evaluation, CFV cannot be separated from the compensation evaluation system. Executives should be responsible for investment decisions that cause CFV. However, because of the asymmetric sensitivity of compensation to performance (Gaver and Gaver, 1998; Dorff, 2005; Garvey and Milbourn, 2006; Fang, 2009; Xu and Zeng, 2010), executive compensation displays its own characteristic of asymmetric sensitivity to CFV. Asymmetric sensitivity explains the different effects of CFV on executive compensation in the above-mentioned anecdotal cases. In dealing with this issue, the initial problem is to understand where the power of executives to control compensation comes from. What factors determine the strength of this asymmetric sensitivity in executive compensation? Does managerial power play a role in the asymmetric sensitivity of executive compensation to CFV?

Our sample consists of Chinese firms listed as A-share companies on the Shanghai and Shenzhen Securities Exchanges between 2007 and 2009. We analyze the effects of CFV on executive compensation from the viewpoint of contracting theory and attempt to explain the observed effects in reference to the theory of managerial power. We draw the following conclusions: (1) as a new item of performance in the income statement, CFV is positively related to executive compensation; (2) the sensitivity of executive compensation is asymmetric to CFV in that executive pay rises higher due to PCFV than it declines due to LCFV; and (3) the greater the managerial power, the more asymmetric the sensitivity of executive compensation is to CFV.

1 In this paper, compensation refers only to monetary compensation reported in company annual financial statements.

2 http://www.yrdnet.com/News/Detail-9906.aspx, April 15, 2009.

3 http://finance.cn.yahoo.com/mark/stocknews, March 23, 2011.

4 Haining Feng, Why is SASAC unable to control self-dealing in executive compensation? China Economic Times, July 17, 2008.

Our study makes the following contributions. First, we investigate the effects of both profits and losses from CFV on executive compensation. Previous studies have paid more attention to the sensitivity of executive compensation to PCFV (Hou and Jin, 2010; Xu and Zeng, 2010) and the sensitivity of executive compensation to LCFV is less commonly observed. We find a positive relationship between executive compensation and CFV, whether those changes involve profits or losses. This finding verifies the potential usefulness of contracts for controlling executive compensation in relation to CFV. In other words, the compensation contract is a valid tool to some extent. Second, we provide evidence to confirm the asymmetric sensitivity of executive compensation to CFV. Third, we investigate both the asymmetric sensitivity and the effects of managerial power on levels of executive compensation. This investigation makes contributions to two research streams. We use the theory of managerial power to explain the asymmetric sensitivity of executive compensation to CFV and we find new empirical evidence concerning the effects that managerial power has on compensation, while providing a reasonable explanation for compensation stickiness.

The rest of this paper is organized as follows. Section 2 presents the institutional background and literature review. Section 3 provides the theoretical analysis and develops testable hypotheses. Section 4 outlines the research design, including the sample selection, data sources, variable definitions and descriptive statistics. Section 5 provides the empirical results and Section 6 draws conclusions.

2. Institutional background and literature review

2.1. Institutional background

2.1.1. SOE reform and executive compensation

Since the market-oriented enterprise reform in 1978, operating performance has been taken into consideration by decision makers when designing executive compensation contracts. Along with the SOE shareholding system reform, the launching of the split share structure reform and the development of capital markets, the structure of executive compensation in SOEs underwent several major institutional transitions. In 1984, The CPC Central Committee's Decision on the Reform of the Economic System proposed the principle of distribution according to work. In 1986, Certain Provisions of the State Council on Deepening the Reform of Enterprises to Enhance the Vitality of Enterprises advanced a proposal that the personal salaries of managerial operators could be one to three times higher than those of staff and workers. In 1992, the Ministry of Labor and the Economic and Trade Office of the State Council issued a new rule named To Improve Income Allocation of SOE Operators, which for the first time required that managers' salaries be linked to their work performance. In 1995, the State-owned Assets Administration Bureau (which is in charge of assessing, maintaining and increasing the value of state-owned assets) claimed that managers' salaries must be connected to their performance assessments. In 2003, the State-owned Assets Supervision and Administration Commission (SASAC) issued the Interim Measures for the Performance Evaluation of the Person in Charge of the Central Enterprises, which both encouraged and constrained enterprise supervisors in the design of incentive contracts and assessments of operating performance. In 2004, the SASAC further issued the Interim Measures for the Compensation Management of Central Enterprise Principle, specifying that compensation should follow a central enterprise principle and be comprised of basic pay, long-term incentive pay and performance pay, with performance pay determined by operating performance. In 2009, the Ministry of Human Resources and Social Security issued its Guidance to Further Regulate the Compensation Management of Central Enterprise Principle. This policy further regulated the incentive mechanism in pay levels, pay structures, post-consumption, management and oversight, and organization and implementation. The evolving development of the executive compensation system shows that policy governing executive pay has been designed increasingly in relation to operating performance. However, it cannot be ignored that the gradually liberalized standards for executive compensation are progressively controlled by managers themselves, resulting in astronomical salaries due to state ownership and insider control.

2.1.2. Introduction of fair value into new accounting standards and CFV as a new item of income

The new enterprise accounting standard system was issued by the Ministry of Finance on February 15, 2006 and was implemented for listed companies starting January 1, 2007. As the new accounting standard

recognizes the measurement attribute of fair value, CFV was added into company income statements as a new item for reporting profits or losses originating from changes in the fair value of assets or liabilities. According to the new standards, these changes in fair value should be reported as gains or losses in the current period's income statement. Except in the case of financial enterprises, profits and losses from changes in fair value must be reported on transactions involving the trading of financial assets or liabilities, derivative financial instruments and investment property. The introduction of fair value into the new accounting standards included setting-up a new item in the income statement, distinguishing investment gains or losses with different levels of risk, improving the relevance of accounting information and enabling better decision-making by investors or managers. However, whether these changes in accounting standards affect executive compensation is still an open question.

2.2. Literature review

Many scholars have explored the issues of CFV, executive compensation and managerial power. We categorize these studies into three topics.

2.2.1. CFV and executive compensation

From the perspective of contracting theory, Hou and Jin (2010) study the influence of CFV on executive compensation after the implementation of the new accounting standards. They find that both PCFV and LCFV tend to increase executive pay levels. Zhou et al. (2010) find that CEOs and chairmen are responsible for gains or losses from short-term investment property, but their salaries are insensitive to CFV. Also, CFOs are in charge of funding operations and their compensation is positively sensitive to CFV. Xu and Zeng (2010) document that PCFV has a positive effect on executive compensation and that the incentive effects are significantly higher than for other earnings items. LCFV, however, is not significantly sensitive to executive compensation. They confirm that an asymmetric sensitivity of compensation to performance does exist in A-share listed Chinese companies. In other words, irrational incentives to profit from PCFV contrast with motivations to avoid punishment for LCFV. Zhang et al. (2011) find that PCFV (or LCFV) entering the income statement and directly credited into capital surplus is positively related (or not relevant) to changes in executive compensation. They conclude that contracts for executive compensation in listed Chinese companies fail to properly deal with changes in fair value.

2.2.2. Sensitivity of executive compensation to performance and executive compensation stickiness

According to agency theory, compensation is naturally related to performance. However, scholars disagree

over the nature of this relationship. Taussings and Baker (1925) were among the first to focus on this relationship and they find little relationship between compensation and performance. Figler and Lutz (1991) and Tosi et al. (2000) find a weak relationship between the two. The opposite conclusion is drawn by McGuire et al. (1962), Coughlan and Schmidt (1985), Jensen and Murphy (1990) and George and Hall (1998). Conyon (2006) argues that the sensitivity of executive compensation to performance has increased year by year in the United States. However, domestic research in China has not reached such a consensus on this matter. Zhang et al. (2003), Du and Wang (2007), Lu (2008) and Jiang (2008) all find evidence that performance and compensation are significantly related. Li (2000), Wei (2000) and Chen and Liu (2003), however, provide evidence of a non-significant relationship between compensation and performance.

The notion that executive compensation is sensitive to performance does not necessarily mean that changes in performance bring changes of the same magnitude in compensation. The influence of performance on compensation differs in situations of performance growth or decline. In other words, compensation is sticky. Jensen and Murphy (1990) and Sun and Liu (2004) find that executives are reluctant to reduce their pay for reasons of personal reputation and career advancement. Gaver and Gaver (1998) find that CEO pay is significantly related to rises in operating profits and non-recurring profits, but it does not fall due to operating losses or non-recurring losses. Dorff (2005) also finds that executive compensation is more sensitive to positive earnings performance than to financial losses, which means that compensation is only influenced by good performance. Fang (2009) provides further evidence of the asymmetries in the sensitivity of executive compensation

to performance in China. Xu and Zeng (2010) also find the phenomenon of big reward, light penalty in Chinese listed companies.

2.2.3. Managerial power and executive compensation

Since Bebchuk et al. (2002) presented their theory of managerial power, this understanding has played an important role in explaining executive compensation levels, the sensitivity of compensation to performance, the changing structure of compensation and the evaluation of operating performance (Lu and Wei, 2008). Otten (2008) selects 1393 compensation contracts from 451 companies distributed in 17 countries between 2001 and 2004. The results document that the theory of managerial power is universally applicable. A significant body of empirical research documents the effects of managerial power on executive compensation. Hambrick and Finkelstein (1995) find that CEO pay grows faster in enterprises controlled by management. Core et al. (1999) find that managerial power is stronger and CEO pay is higher in enterprises with large-sized boards and with outside directors who are mostly appointed by the CEO. Bebchuk et al. (2002) present evidence that the stronger the managerial power, the stronger the managers' ability to gain from rents and control their own salaries. Cyert et al. (2002) find that the pay of CEOs who serve as chairmen of the board is 20-40% higher than that of other CEOs. Duffhues and Kabir (2008) document that executives control their own pay through managerial power. Fahlenbrach (2009) finds that increases in the managerial power of CEOs significantly improve their pay. Cheng and Indjejikian (2009) find that CEOs exert a strong influence in formulating their compensation contracts and are capable of influencing their boards on compensation assessment.

In the context of China, many domestic scholars have studied the influence of managerial power on executive compensation in listed companies, taking account of China's particular institutional background. Pan and Tong (2005) find that top managers of Chinese public companies design their own compensation contracts and assess their own performance. Zhang and Shi (2005) find that the proportion of independent directors on the board, the establishment of a compensation committee and the duality of chairman and CEO roles all have significantly positive effects on executive pay levels. Lu (2008) finds that managerial power is positively related to executive pay. Lyu and Zhao (2008) find that SOE managers with strong executive powers design their own incentive portfolios and obtain higher pay largely through bonuses. In contrast, managers with weaker power are more concerned about their salaries and manipulate earnings (or fictitious profits) to satisfy the compensation assessment requirements. Quan et al. (2010) examine SEOs and find that managers with increasing power are inclined to obtain performance pay through earnings management, which means that the stronger the managerial power, the more sensitive compensation is to manipulated performance.

3. Theoretical analysis and hypothesis development

The separation of ownership and control in modern enterprises results in an agency problem. According to agency theory, the principal assigns tasks to the agent, whose objective function is different from that of the principal. This difference in functions leads agents to defend their interests against those of the principals. The determination of executive compensation is one such agency problem in which optimal contracting theory holds. Incentive contracts are an effective way to solve the agency problem. A favorably designed executive compensation contract is supposed to be an effective mechanism to make the goals of both managers and stockholders compatible and to reduce agency costs. A well-functioning contract can also prevent executives from pursuing goals detrimental to shareholders' goal of value maximization, as such deviation would be an example of the agency problem (Jensen and Meckling, 1976; Jensen and Murphy, 1990). Managerial compensation and corporate performance are connected through an effective contracting arrangement. The more sensitive executive compensation is to corporate performance, the more closely aligned executive interests are with the interests of stockholders. The best way of designing a contract is to pay according to performance (Jenson and Murphy, 1990). In an optimal contract, executive compensation is linearly related to performance (Holmstrom and Milgrom, 1991). Performance measured by audited earnings may reduce the noise caused by market volatility in determining executive compensation (Lambert and Larcker, 1987; Sloan, 1993), as earnings are more sensitive and accurate than market performance (Xu and Zeng, 2010). Earnings better reflect management's fiduciary obligation and operating efficiency (Natarajan, 1996) and thus should be the basis of performance evaluation.

The new enterprise accounting standards issued in 2006 improve the relevance of executive and accounting performance (Hou and Jin, 2010). These standards also enhance the effectiveness of measuring executive compensation against accounting performance. According to the new standards, CFV is directly introduced into operating income as an earnings item, which leads to a great change in the structure of accounting and makes CFV an important element affecting performance reports. The investment and management of financial assets and investment property are the main contributors to CFV. Despite being theoretically determined by the market, CFV is in fact decided by managers, who decide on the options of purchasing, holding or selling financial assets. It is undeniable that shareholder wealth is directly affected by CFV arising from past decisions by the managers. As optimal contracting theory claims, effective contracts must make the managers' financial results and their pay connected to the highest degree possible. Hence, we suggest that CFV is positively related to executive compensation and propose the first hypothesis.

H1. Executive compensation is positively related to CFV.

Although compensation contracts are theoretically based on optimal contracting theory and are an ideal method for solving the agency problem between stockholders and managers, in reality the contract is often far from perfect. Three preconditions are necessary to ensure that optimal contracting theory works well. These include effective negotiation by the board, efficient constraint by the market and effective execution of power by stockholders. In many cases, these preconditions are far from being realized (Bebchuk and Fried, 2003). In theory, executive compensation is sensitive to corporate performance, but the reality in the business world is quite different (Edmans et al., 2008). Once a flawed contract has been accepted, shareholders face the risk of rent appropriation by management (Grossman and Hart, 1983; Hart and Moore, 1990).

The theory of managerial power provides a new explanation for the gap between compensation contracts and actual compensation. This theory argues that management influences levels of compensation by interfering with compensation contract design and deriving rents from company profits. The more control managers have, the stronger their rent-grabbing capacity (Bebchuk and Fried, 2003). According to the theory of managerial power, increases in the sensitivity of compensation to performance do not result in a decrease in the agency cost between shareholders and management, because corporate performance contains much noise.5 Market noise is caused both by objective factors from the effects of the macro-economic environment or industry development on corporate performance and by subjective factors such as earnings manipulation by management (Quan et al., 2010). Performance-based compensation contracts induce earnings manipulation by managers, because the contracts push mangers to make efforts toward the contract objective. Thus, while compensation contracts aim to remove one agency problem, they lead to a different agency problem.

As fair value is introduced into the new enterprise accounting standards, it becomes a new item in the compensation contract. According to the theory of managerial power, the assessment of fair value becomes a new method for managers to manipulate earnings and a new means for enhancing the contract's efficiency in serving managers' interests. The coexistence in China of deficiencies in property rights, deficiencies in securities market regulation, the desire for power by private owners, weak internal controls and fatherly love given to SOEs by the government (Lu, 2007a) make the valid oversight of management in Chinese listed companies almost impossible. Such absence of constraints on managerial power makes the rent-seeking motivation of managers stronger. In this way, managers can both hire and supervise themselves, becoming both designers and implementers of their compensation contracts (Wang and Wang, 2007). Executives can increasingly demand higher salaries based on the excuse of PCFV (Hou and Jin, 2010). When changes in fair value earn profits and operating performance is thereby improved, executives attribute the improvement to their own efforts and gain a louder voice in formulating their compensation plans. Likewise, executives use managerial power to excuse LCFV in a disguised or opportunistic way (Na, 2009). They attribute losses to external factors such as market changes to free themselves from obligations and do so with impunity. The sensitivity of executive compensation to corporate performance is asymmetric and is characterized by stickiness, which means that an increase in executive pay on account of performance growth is higher than a decrease in executive pay

5 Accounting performance is directly affected by managers' behavior, such as making changes in liability structure, inventory management or accounting standards (Murphy, 2000). Market performance is less vulnerable to control by managers, but noise floods the market (Wiseman and Gomez-Mejia, 1998).

on account of performance decline (Fang, 2009). Hence, we argue that the increase in executive pay due to PCFV is higher than the decrease in executive pay due to LCFV. The stronger the managerial power, the more asymmetric the sensitivity of executive pay is to CFV. The above analysis leads to the following two hypotheses:

H2. Increases in executive compensation due to PCFV are significantly higher than decreases in executive compensation due to LCFV.

H3. The stronger the managerial power, the more asymmetric the sensitivity of executive compensation is to CFV.

4. Methodology

4.1. Data and sample

Our initial sample consists of 4893 Chinese firms listed on the A-share stock markets in Shanghai and Shenzhen between 2007 and 2009. We drop observations from finance and insurance firms, and from firms with no changes in profits or losses due to changes in fair value. After merging the different sources of data and deleting observations with missing information, we obtain a final data set consisting of 1148 firm-year observations.

Data related to all variables was retrieved from the China Stock Market and Accounting Research (CSMAR) database offered by GTA Information Technology Co., Ltd.

The sample screening process and distribution are detailed in Table 1. Panel A of Table 1 reports the sample screening process and yearly distribution. It shows that the 1148 observations are evenly distributed across the

Table 1

Sample description.

2007 2008 2009 Total

Panel A: sample screening process and distribution by year

All A-share listed companies 1550 1625 1718 4893

Financial and insurance companies 36 32 39 107

Companies missing other data 1190 1199 1249 3638

Final observations 324 394 430 1148

Panel B: industry distribution by yeara

Agriculture 10 13 15 38

Mining 3 8 12 23

Manufacturing 173 218 245 636

Electric power, gas and water production and supply 12 17 14 43

Construction 4 5 8 17

Transport and storage 17 16 16 49

Information technology 31 31 28 90

Wholesale and retail 20 22 30 72

Real estate 13 19 19 51

Social service 13 11 13 37

Media and culture 3 5 3 11

Residual category 25 29 27 81

Total 324 394 430 1148

Panel C: distribution by CFV

Profits from changes in fair value 222 70 322 614

Losses from changes in fair value 102 324 108 534

Total 324 394 430 1148

a Observations from the Finance and Insurance industry are omitted from the sample.

3 years, with 324 in 2007, 394 in 2008 and 430 in 2009. Panel B of Table 1 presents the industry distribution of the sample by year. Observations from the Manufacturing and Broadcasting and Media industries account for 55% and 1% of the total sample, respectively. The unbalanced distribution of our sample in different industries is consistent with the actual industrial distribution of China's listed companies. Panel C of Table 1 reports the distribution of positive and negative CFV. There are 614 observations with PCFV and 534 with LCFV, accounting for 53.84% and 46.52% of the total sample, respectively. Due to the severe financial crisis in 2008, the 324 observations with LCFV in that year account for 82.23% of the full sample, which is far higher than the 31.48% for LCFV in 2007 and the 25.12% in 2009.

4.2. Variable definitions

4.2.1. Dependent variable: executive compensation

Executive compensation is mainly composed of monetary compensation and stock option incentives. Due to the lagged implementation of equity incentive plans in China, few companies use these types of incentives and their effect on compensation is limited (Xin et al., 2007). Executive compensation is strictly regulated by the government and few managers in Chinese listed companies hold ownership of stock (Li, 2000). Hence, previous studies show that monetary compensation is generally the same as executive compensation. Considering the different definitions of the term executive, executive compensation is commonly measured by the monetary compensation of the top three managers (Lu, 2007b; Fang, 2009; Hou and Jin, 2010; Quan et al., 2010), by compensation of the top three directors (Lu, 2007b; Fang, 2009; Hou and Jin, 2010; Quan et al., 2010), by compensation of the top three directors, supervisors or managers (Xu and Zeng, 2010) or by compensation of the chairman of the board, CEO and CFO (Zhou et al., 2010).

We examine the influence of CFV on executive compensation. To undertake a thorough investigation of this relationship, we select the first three types of executive compensation discussed above as our dependent variables, which are respectively symbolized by lntm (monetary compensation of top three managers), lndir (monetary compensation of top three directors) and lncomp (monetary compensation of top three directors, supervisors and managers). The fourth type of executive compensation is classified into four variables: monetary compensation to the chairman of the board (lnchair), CEO (lnceo), CFO (lncfo) and the total of all three (lnchair_ceo_cfo). These variables are used in the robustness tests.

4.2.2. Independent variables

We use fair_value, measured as the amount of CFV in the income statement of the annual report, to proxy for profits and losses from changes in fair value.

Operating income (oper_inco) is calculated by deducting costs and expenses from revenue. The computation is as follows: operating income = operating revenue — operating costs — business tax and surcharges — sales expenses — administration expenses — interest expense — asset impairment losses.

Considering the close relationship between equity and CFV, both fair_value and oper_inco are adjusted by shareholders' equity at the beginning of each year (Zhou et al., 2010).

Managerial power (power) tends to take on the characteristics of relativity (Quan et al., 2010) and concealment (Lu, 2008), thus it is difficult to measure managerial power reliably and effectively with a single indicator. We select three main indicators of managerial power. First, duality of the chairman of the board and the CEO position is the most obvious manifestation of concentration in managerial power (Lu, 2008). Second, ownership concentration can effectively control excessive managerial power (Chen, 2010). When ownership is highly dispersed and controlled by various major shareholders, collusion can occur between major shareholders and managers (Huang, 2006). Once the power of shareholders and managers becomes a joint force, managerial power tends to peak (Lu, 2008). Third, executive tenure reflects managerial power from another perspective. With longer tenure, CEOs have a stronger ability to control companies. We thus use three indicators, power1, power2 and power3, to proxy for the three types of one-dimensional managerial power discussed above. Power1 is a proxy for duality (power1 equals 1 if the chairman and CEO are the same person, and 0 otherwise). Power2 is a proxy for ownership concentration (power2 equals 1 if the holding ratio of the largest shareholder divided by the accumulated ratio of the top two to ten shareholders is less than 1, and 0 otherwise). Power3 is a proxy for executive tenure (power3 equals 1 if the tenure of the chairman of the board or CEO is greater than

the median, and 0 otherwise). Power is the comprehensive proxy of managerial power, quantified by the sum of Power1, Power2 and Power3. Power equals 1 when the sum exceeds 2, and 0 otherwise.

4.2.3. Control variables

In accordance with prior studies (Leone and Zimmerman, 2006; Fang, 2009; Hou and Jin, 2010; Zhou et al., 2010), we also control for other variables that are related to executive compensation. These are the number of board members (board), the proportion of independent directors on the board (inde_dir), sales revenue (lnsale), sales growth (salegrowth), asset-liability ratio (leverage), adjusted annual stock return (adj_return), per capita income in the region where the corporation is registered (lnaver_inco), the nature of the company's property rights (soe), whether the company is cross-listed in other nations (cross), turnover of the chairman (chair), turnover of the CEO (ceo), the presence of a compensation committee (comp_comm), whether the company is in a regulated industry (regulate), whether the company is in a middle region (middle) and whether the company is in a western region (west).

More detailed information on the definitions of these variables is reported in Table 2.

Table 2

Variable definitions.

Variables

Definitions

Panel A: dependent variables

Incomp Natural logarithm of total compensation of top three directors, supervisors and managers, excluding allowances received

by independent directors

lndir Natural logarithm of total compensation of top three directors, excluding allowances received by independent directors

lntm Natural logarithm of total compensation of top three managers, excluding allowances received by management

Panel B: independent variables

fair_value oper_inco

Panel C: control board inde_dir lnsale salegrowth

leverage adj_return lnaver_inco

soe cross chair ceo

comp_comm regulate middle

Profits and losses from changes in fair value, divided by stockholders' equity at the end of the previous year Operating income, or the difference between operating revenue and operating expenses (which includes operating costs, business taxes and surcharges, sales expenses, administration expenses, interest expense, and asset impairment losses) divided by stockholders' equity at the end of the previous year

Dummy variable for managerial power. If power1 + power2 + power3 P 2, power is assigned the value of 1, and 0 otherwise. Power1 is a dummy variable for duality. If the chairman and CEO are the same person, it is assigned the value of 1, and 0 otherwise. Power2 is a dummy variable for ownership concentration. If the holding ratio of the largest shareholder divided by the accumulated ratio of the top two to ten shareholders is less than 1, it is assigned the value of 1, and 0 otherwise. Power3 is a dummy variable for executive tenure. If the tenure of chairman or CEO is greater than the mean it is assigned the value of 1, and 0 otherwise

variables

Number of board members

Proportion of independent directors on the board Natural logarithm of operating revenue

Growth in sales. Equals the absolute value of the difference between operating revenue for the present year and last year, divided by operating revenue for last year Ratio of total liabilities to total assets

Adjusted annual stock return. Equals the company's annual stock return less the market annual stock return

Per capita income in the region where the corporation is registered. Equals the natural logarithm of per capita disposable

income of urban households

Dummy variable assigned the value of 1 for state-owned enterprises, and 0 otherwise

Dummy variable assigned the value of 1 for cross-listed companies, and 0 otherwise

Dummy variable assigned the value of 1 for turnover of the chairman, and 0 otherwise

Dummy variable assigned the value of 1 for turnover of the CEO, and 0 otherwise

Dummy variable assigned the value of 1 if a compensation committee is constituted, and 0 otherwise

Dummy variable assigned the value of 1 for companies in regulated industries, and 0 otherwise

Dummy variable for the middle region. If a corporation is registered in the middle region (Provinces of Shanxi, Jilin,

Heilongjiang, Anhui, Jiangxi, Henan, Hubei or Hunan), it is assigned the value of 1, and 0 otherwise

Dummy variable for the western region. If a corporation is registered in the western regions (Chongqing, Inner Mongolia,

Tibet, Xinjiang Uygur Autonomous Region, Ningxia Hui Autonomous Region, or the provinces of Guangxi, Sichuan,

Guizhou, Yunnan, Shaanxi, Gansu or Qinghai), it is assigned the value of 1, and 0 otherwise

Dummy variable assigned the value of 1 for profits from changes in fair value, and 0 otherwise

4.3. Descriptive statistics

Table 3 summarizes the descriptive statistics. As the table shows, there are no differences among the means (or medians) of lncomp, lndir and lntm. The amount of compensation ranges from 814,200 (e13 610) to 1,077,300 (e13 890) yuan, which shows an uneven distribution of executive compensation. Among the three types of measurements of executive pay, compensation of the top three directors is relatively low and compensation of the top three directors, supervisors and managers is relatively high. The mean (median) offair_value is —0.001 (0) and the standard deviation is 0.059. The mean (median) of oper_inco is 0.133 (0.054) and the standard deviation is 2.053. The mean of power is 0.243, and the third quantile is 0, which means more than three quarters of the sample companies have lower degrees of managerial power.

As shown in Table 3, about 60% of the sample firms are owned by the state, 16% (17%) experienced turnover of the chairman or CEO, 97% of the firms have constituted a compensation committee, 19% of the firms are located in the middle region, 13% of the firms are located in the western region and 68% are in the eastern region. The mean and median of adj_return are 0.261 and 0.038 (the great difference between these returns may result from the severe financial crisis in 2008). The mean (median) ratio of total liabilities to total assets is 64.3% (51.9%), which indicates a reasonable capital structure for the sample companies. The number of board members ranges from 4 to 16 with a mean (median) of 9.230 (9) board members. The mean (median) of inde_dir is 0.366 (0.333), which is consistent with the regulations of the China Securities Regulatory Commission. The minimum of inde_dir is 0.222, which indicates that some companies in the sample do not meet the required threshold of one-third independent members. The standard deviation of inde_dir is 0.053, which implies a slight change in the ratio of independent directors on the board over the time of the survey.

In general, all variables are normally distributed with little difference between the mean and the median (excluding oper_inco, salegrowth and adj_return). In addition, according to the standard deviation and first and third quantile statistics, there is adequate variation in the variables during the sample period.

Descriptive statistics of the three executive compensation variables categorized by positive or negative CFV are shown in Table 4. As indicated in this table, compensation for all types of executives rises year by year and the growth in compensation in companies suffering LCFV is relatively higher than in companies earning PCFV. The three executive compensation variables show little difference in either their means or medians. In general, executive compensation does not change with changes in fair value. There is little difference in

Table 3 Descriptive statistics.

Variable N Mean Std Min P25 P50 P75 Max

lncomp 1148 13.890 0.799 11.230 13.410 13.870 14.360 16.650

lndir 1148 13.610 0.931 4.754 13.050 13.610 14.200 16.600

lntm 1146 13.760 0.809 10.360 13.230 13.770 14.230 16.530

fair_value 1148 -0.001 0.059 -1.794 -0.001 0 0.002 0.243

oper_inco 1148 0.133 2.053 -3.908 -0.006 0.054 0.147 68.560

power 1148 0.243 0.429 0 0 0 0 1

board 1148 9.230 1.866 4 9 9 10 16

inde_dir 1148 0.366 0.053 0.222 0.333 0.333 0.385 0.714

lnsale 1148 21.290 1.640 9.310 20.340 21.140 22.140 28.000

salegrowth 1148 1.778 44.450 -0.995 -0.070 0.098 0.290 1497.000

leverage 1148 0.643 4.203 0.018 0.371 0.519 0.668 142.700

adj_return 1148 0.261 0.891 -1.592 -0.130 0.038 0.497 7.242

lnaver_inco 1148 9.876 0.299 9.293 9.619 9.912 10.130 10.390

soe 1148 0.602 0.490 0 0 1 1 1

cross 1148 0.063 0.243 0 0 0 0 1

chair 1148 0.163 0.369 0 0 0 0 1

ceo 1148 0.170 0.376 0 0 0 0 1

comp_comm 1148 0.968 0.177 0 1 1 1 1

regulate 1148 0.037 0.190 0 0 0 0 1

middle 1148 0.190 0.392 0 0 0 0 1

west 1148 0.134 0.341 0 0 0 0 1

Table 4

Executive compensation and CFV (in 10 thousands of yuan).

2007 2008 2009 Total

Compensation of the top three directors, supervisors and managers

Profits from changes in fair value

Mean 141.98 145.85 161.97 152.90

Median 99.80 118.35 111.64 107.45

Std 159.28 104.53 171.51 160.84

Losses from changes in fair value

Mean 117.70 155.01 179.84 152.90

Median 77.55 108.65 119.02 104.44

Std 179.56 191.08 164.83 184.60

Compensation of the top three directors

Profits from changes in fair value

Mean 111.36 118.74 135.09 124.65

Median 75.00 86.67 90.10 83.24

Std 133.36 100.32 161.54 146.06

Losses from changes in fair value

Mean 98.26 124.32 138.19 122.15

Median 58.28 82.51 90.85 78.90

Std 169.23 161.12 133.26 157.76

Compensation of the top three managers

Profits from changes in fair value

Mean 119.64 130.38 136.09 129.49

Median 87.83 104.22 98.00 95.28

Std 118.92 95.92 127.06 121.02

Losses from changes in fair value

Mean 104.84 133.26 165.28 134.31

Median 72.71 96.60 112.95 95.39

Std 155.06 153.41 158.65 155.67

compensation in relation to changes in fair value, either in the mean or the median. In 2007, the executive compensation in companies earning PCFV is a little higher and the executive compensation in companies suffering LCFV is also a little higher. It is obvious that executive compensation in companies suffering LCFV is not less than executive compensation in companies earning PCFV, which means that LCFV has no significant influence on executive compensation.

4.4. Correlation analysis

Table 5 presents the correlation matrix for our sample. Spearman correlation coefficients are reported in the upper right corner and Pearson correlation coefficients in the bottom left corner.

Table 5 shows that there is no significant correlation between executive compensation and CFV (fair_va-lue). Operating income (oper_inco) is positively correlated with executive compensation, which implies that operating income is a reasonable indicator to evaluate performance. Managerial power (power) is significantly correlated with executive compensation, which indicates that the greater the managerial power, the higher the executive compensation. Other variables such as board, inde_dir, lnsale, lnaver_inco, cross, comp_comm (not significant in the Spearman coefficient), middle and west are significantly correlated with executive compensation. These significant correlations indicate that bigger boards, more independent directors, greater sales and higher per capita income, cross-listing and compensation committees all contribute to higher executive compensation. Table 5 also reports that executive compensation in firms located in the middle and western regions is lower than that of firms in eastern regions. In addition, the coefficients in Table 5 suggest that the correlations between independent variables are reasonable. We further compute variance inflation factors (VIFs) and find that there are no potential multicollinearity problems among the variables listed in Table 5.

Table 5

Correlation coefficient matrix.

incomp indir intm fair_vaiue oper_inco power board inde_dir insaie saiegrowth ieverage adj_return inaver_inco soe cross chair ceo comp_comm reguiate middie west

lncomp 1 0.887* 0.970* 0.037 0.318* 0.125* 0.128* 0.124* 0.451* 0.110 0.019 -0.030 0.314* 0.048 0.256* -0.050 -0.022 0.082 -0.044 -0.164* -0.230*

lndir 0.855* 1 0.834* 0.033 0.303* 0.128* 0.158* 0.056 0.408* 0.107 0.035 -0.015 0.229* -0.032 0.214* -0.089 -0.033 0.078 -0.071 -0.144* -0.176*

lntm 0.960* 0.800* 1 0.025 0.311* 0.144* 0.136* 0.116 0.456* 0.103 0.025 -0.033 0.332* 0.096 0.265* -0.032 -0.027 0.076 -0.033 -0.167* -0.242*

fair_value -0.001 0.046 -0.003 1 0.032 -0.012 -0.030 0.007 -0.017 0.020 -0.005 0.066 0.043 -0.043 -0.029 0.026 0.004 -0.041 -0.0202 -0.032 -0.004

operinco 0.121* 0.114* 0.099* -0.034 1 0.082 0.020 -0.024 0.332* 0.449* -0.041 0.119 -0.036 -0.052 0.099 -0.060 0.022 0.047 -0.0531 0.015 -0.019

power 0.116* 0.111* 0.132* 0.005 -0.017 1 -0.040 -0.006 -0.092 0.015 -0.112 0.030 0.011 -0.176* -0.038 0.015 -0.040 -0.069 0.0271 -0.032 -0.037

board 0.129* 0.150* 0.136* -0.007 -0.005 -0.033 1 -0.167* 0.267* 0.039 0.105 -0.062 0.003 0.210* 0.172* 0.037 0.062 0.034 0.0819 0.004 0.006

inde_dir 0.095* 0.031 0.090* -0.093* -0.026 -0.017 -0.217* 1 0.061 0.026 -0.014 0.009 0.045 -0.007 0.106 -0.013 -0.008 0.039 0.0007 0.006 -0.067

Insale 0.438* 0.382* 0.470* -0.066 0.030 -0.091* 0.314* 0.093* 1 0.193* 0.349* -0.018 0.139* 0.242* 0.370* -0.016 0.044 0.097 -0.0244 -0.052 -0.078

salegrowth -0.006 -0.006 -0.012 0.014 0.001 -0.018 -0.067 0.015 -0.030 1 0.073 0.101 -0.163* 0.047 0.059 -0.004 0.030 0.018 0.0384 0.073 0.0595

leverage -0.015 -0.009 -0.010* -0.003 -0.000 -0.022 -0.064 0.019 -0.205* -0.002 1 0.025 -0.044 0.095 0.092 0.048 0.002 0.007 0.0438 -0.003 0.077

adj_return -0.039 -0.029 -0.046 0.055 -0.003 0.035 -0.037 -0.009 -0.018 0.160* 0.012 1 -0.03 -0.028 -0.081 0.057 0.029 -0.010 -0.0219 0.051 0.038

lnaver_inco 0.296* 0.210* 0.317* -0.052 -0.039 0.015 0.008 0.038 0.161* 0.003 -0.027 -0.061 1 -0.034 0.0841 0.023 0.016 0.038 -0.0408 -0.515* -0.454*

soe 0.027 -0.033 0.089* -0.043 0.010 -0.174* 0.196* 0.005 0.275* -0.033 -0.033 -0.022 -0.037 1 0.173* 0.077 0.021 0.044 0.0758 0.066 0.046

cross 0.246* 0.201* 0.258* -0.125* -0.002 -0.038 0.188* 0.152* 0.448* -0.009 -0.002 -0.025 0.088* 0.174* 1 0.013 0.0072 0.047 0.0056 -0.043 -0.049

chair -0.054 -0.090* -0.037 -0.005 0.061 0.014 0.027 -0.018 -0.021 0.080* -0.009 0.051 0.021 0.074 0.012 1 0.216* -0.027 0.0127 -0.014 -0.013

ceo -0.014 -0.024 -0.022 0.000 0.070 -0.040 0.054 -0.026 0.044 -0.013 -0.014 0.018 0.018 0.022 0.007 0.215* 1 0.030 -0.0405 -0.012 -0.021

comp_comm 0.086* 0.078* 0.085* -0.028 0.016 -0.069 0.015 0.052 0.098* 0.007 0.005 -0.099* 0.035 0.043 0.047 -0.026 0.030 1 0.0101 0.013 -0.001

regulate -0.052 -0.065 -0.038 0.006 -0.010 0.027 0.091* -0.001 -0.016 -0.007 -0.004 -0.006 -0.044 0.076* 0.006 0.012 -0.040 0.010 1 0.150* -0.064

middle -0.150* -0.122* -0.154* 0.001 0.058 -0.031 -0.003 -0.000 -0.053 -0.015 -0.015 0.025 -0.535* 0.067 -0.043 -0.015 -0.012 0.013 0.150* 1 -0.190*

west -0.229* -0.178* -0.244* 0.026 -0.007 -0.038 0.020 -0.029 -0.090* -0.013 0.080* 0.055 -0.463* 0.043 -0.049 -0.008 -0.022 -0.000 -0.064 -0.191* 1

Note: The upper right corner reports Spearman correlation coefficients and the bottom left corner reports Pearson correlation coefficients. Significant at the 1% level.

i Ö o

5. Empirical analysis

5.1. Hypothesis 1

Hypothesis 1 predicts that executive compensation is positively related to CFV. Following Dechow et al. (2010), we use the following regression model.

Model (1):

comp = a0 + a1 fair-value + a2operdnco + a3power + a4board + a5inde^dir + a6 insale + a7salegrowth + a8leverage + a9adjj-eturn + a10 lnaveJnco + a11soe + a12cross + a13chair + a14ceo + a15comp_comm + a16regulate + a17middle + a18west + e

In this model, comp refers to the terms lntm, lncomp and lndir, and the regression coefficient of fair_value is our primary concern. If a1 is significantly positive, then executive compensation is positively related to CFV. That is to say, CFV has a positive effect on executive compensation.

Table 6 reports the regression results from Model (1). The dependent variables in columns 1-3 are the compensation of the top three managers (lntm), the top three directors, supervisors, and managers (lncomp), and the top three directors (lndir), respectively. Table 6 shows that the coefficient on fair_value is significantly positive for lntm, lncomp and lndir, which indicates that all three types of executive compensation are significantly related to CFV (a1 > 0, t-values are 3.49, 4.14, and 7.96,6 respectively). Therefore, Hypothesis 1 is supported. Concerning lntm, the regression coefficient means that if CFV increases by one unit, compensation of top managers increases by 69.7%. The probable reason for these results is that the board or compensation committee considers the effects of CFV on executive compensation when designing the compensation plan. As past investment decisions do affect shareholders' wealth and the compensation contract may be based on accounting performance, executives may use CFV as an excuse to ask for higher salaries (Hou and Jin, 2010).

The results in Table 6 are not the same as the findings of Hou and Jin (2010) and of Xu and Zeng (2010). Hou and Jin (2010) find that a positive relationship between executive compensation and CFV exists only when profits are earned from changes in fair value and a negative relationship exists when losses are suffered from changes in fair value. Xu and Zeng (2010) argue that the positive relationship between executive compensation and CFV is significant only in companies with PCFV and no significant relationship exists in sample companies suffering LCFV. We argue that several reasons contribute to these differences in results. First, our sample period is from 2007 to 2009, rather than the 2007-2008 period used by Hou and Jin (2010) and Xu and Zeng (2010). Second, due to the global financial crisis in 2008, more than 80% of the sample firms suffered LCFV in 2008, as shown in Table 3. This crisis may have had a great influence on the regression results for companies suffering from LCFV. In 2009, the securities markets recovered gradually and CFV became more normal. After the experience of the financial crisis, the boards or the compensation committees may have started to include CFV in compensation evaluation systems. Thus, CFV is also significantly related to executive compensation when companies suffer LCFV.

Table 6 also reports that the coefficient on oper_inco is significantly positive at the 1% level, thus showing that executive compensation is sensitive to operating income. However, this sensitivity is lower than the sensitivity of executive compensation to CFV (a1 = 0.697, a2 = 0.042). There are two explanations for this. One is that income from operations and from investment are differentiated in the design of compensation contracts, with investment income from CFV showing a stronger relationship with the level of compensation.7 The other

6 The significant correlation between executive compensation and CFV is not found in Table 5. However, CFV has a significant influence on executive compensation after other variables are controlled for in Table 6. Thus, we infer that one or more control variables must significantly change the coefficient offair_value in the multiple regression analysis. To identify the relevant variables, we enter the control variables one by one into the regression equation of Model (1) and find that a significant positive relationship between executive compensation and CFV emerges after lnsale and cross are added into the regression. The addition of other control variables fails to change the coefficient of fair_value.

7 If the company earns one more dollar through business operations, the manager may get one more cent. If the company earns one more dollar through investment (or trading financial assets), the manager may get ten more cents. Although the above explanation may be plausible in theory, further exploration is required to determine whether this reflects reality in practice.

Table 6

Regression of executive compensation on CFV.

Intm Incomp Indir

fair_value 0.697*** 0.680*** 1.303***

(3.49) (4.14) (7.96)

oper_inco 0.042*** 0.049*** 0.055***

(8.80) (9.87) (10.40)

power 0.322*** 0.271*** 0.281***

(6.96) (5.87) (5.12)

board 0.010 0.015 0.035**

(0.84) (1.14) (2.27)

inde_dir 0.671* 0.816** 0.123

(1.84) (2.19) (0.27)

lnsale 0.204*** 0.196*** 0.212***

(13.06) (11.47) (9.60)

salegrowth 0.000 0.000 0.000

(0.74) (1.38) (0.98)

leverage 0.001 0.016*** 0.018***

(0.40) (10.42) (9.20)

adj_return -0.015 -0.014 -0.011

(-0.67) (-0.59) (-0.34)

lnaver_inco 0.328*** 0.263** 0.045

(3.01) (2.53) (0.37)

soe 0.029 -0.078* -0.200**

(0.67) (-1.75) (-3.88)

comp_comm 0.212* 0.217* 0.267**

(1.89) (196) (2.10)

cross 0.158 0.160 0.149

(1.51) (1.55) (1.23)

chair -0.085 -0.102* -0.198**

(-1.45) (-1.81) (-2.88)

ceo -0.091 -0.069 -0.089

(-1.60) (-1.22) (-1.31)

regulate -0.136 -0.173* -0.266**

(-1.34) (-1.71) (-2.24)

middle -0.194** -0.201*** -0.266**

(-2.58) (-2.72) (-3.26)

west -0.402*** -0.387*** -0.438**

(-4.32) (-4.37) (-4.34)

Constant 5.642*** 6.555*** 8.226***

(5.19) (6.37) (6.94)

N 1146 1148 1148

Adj. R2 0.337 0.308 0.245

F 84.544 29.687 26.310

Notes: t Statistics are in parentheses. Significant at the 10% level. Significant at the 5% level.

*** Significant at the 1% level.

explanation may be bias in the regression coefficient of fair_value, in other words, some variables are omitted in Model (1). Changes in fair value may reflect differences in profit models or other characteristics that affect the level of executive compensation.8 Power is significantly positively related to executive compensation at the 1% level, suggesting that the degree of managerial power plays an important role in determining executive compensation and the greater the managerial power, the higher the executive compensation.

8 We test this explanation using a fixed effects regression, and find that the regression coefficient of fair_vaue is no higher than that of oper_inco, thus indicating that Model (1) does omit some unobservable factors, which may have a certain influence on the levels of executive compensation.

The regression results of other variables are shown as follows. Board (only in column 3), inde_dir (columns 1 and 2), Insale, leverage (columns 2 and 3) and comp_comm are all significantly positively related to executive compensation, which indicates that companies with bigger boards, more independent directors, higher sales, higher leverage, higher per capita income, and compensation committees all tend to set higher pay for executives. Soe, chair (columns 2 and 3), regulate (columns 2 and 3), middle and west are all significantly negatively related to executive compensation, which indicates that companies that are state owned, have experienced chairman turnover, are in regulated industries and are in the middle or western regions tend to pay less to executives. These finding are consistent with other studies (Fang, 2009; Hou and Jin, 2010; Zhou et al., 2010; Xu and Zeng, 2010).

5.2. Hypothesis 2

Hypothesis 2 predicts that the rise in executive compensation due to PCFV is significantly higher than the decrease in executive compensation due to LCFV. Following Fang (2009), Model (2) is given as

comp = b0 + b1fair^value + b2dum + b3dum * fairjvalue + fi4operJnco + b5power + b6board

+ Pqinde-dir + b8lnsale + $9salegrowih + b10leverage + b11 adj-return + b12lnave + b13soe + b14cross + b15chair + b16ceo + b17comp_comm + b18 regulate + fi^middle + b20west + e

In this model, compensation (comp) refers to the terms lntm, lncomp and lndir. Dum is a dummy variable assigned the value of 1 for PCFV, and 0 otherwise. Our main interest is b3, the coefficient on the interaction term between dum and fair_value. If b3 is significantly greater than 0, this means that the increase in executive compensation due to PCFV is greater than the decrease in executive compensation due to LCFV, then we can draw the conclusion that the sensitivity is asymmetric.

Table 7 reports the results concerning the asymmetry of executive compensation on CFV. The coefficient on dum * fair_value is significantly positive in columns 1 and 2 at the 1% and 10% levels, respectively, which means that the sensitivity of executive compensation to CFV is strongly asymmetric. A probable explanation is that executives with strong managerial power may ask for higher compensation due to PCFV and find excuses to explain LCFV in an opportunistic way, to reduce losses to their personal salaries (Hou and Jin, 2010). In column 3, the coefficient on dum * fair_value is negative but not significant, which demonstrates that the sensitivity of directors' compensation to CFV is not characterized by stickiness. A probable explanation is that most directors do not participate directly in the operation and management of companies, and have no ability to further their personal interests on account of CFV.

The regression results of other variables in Model (2) are consistent with those in Model (1).

5.3. Hypothesis 3

Hypothesis 3 predicts that the stronger the managerial power, the more asymmetric the sensitivity of executive compensation is to CFV. Based on Model (1) and Model (2), we build Model (3) to test H3.

Model (3):

comp = y0 + yfair.value + y2operJnco + y3dum * fair.value + y4dum * fair.value + y5power * dum + yfairjvalue * power + y7power * dum * fair_value + y8board + y9inside_dir + y10lnsale + y11salegrowth + y12leverage + y 13adj-return + y14lnave + y15soe + y16cross + y17chair + y18ceo + y19comp_comm + y20regulate + y21middle + y22west + e

In this model, comp refers to the terms lntm, lncomp and lndir. Dum is a dummy variable assigned the value of 1 for PCFV and 0 otherwise. Power represents the degree of managerial power, and equals 1 for strong managerial power and 0 otherwise. Our main interest is y7, the regression coefficient on power -* dum * fair_value. If y7 is significantly greater than 0, Hypothesis 3 is supported.

Table 7

Regression of the asymmetry of executive compensation on CFV.

intm 1 incomp 2 indir 3

fair_value 0.433*** 0.482*** 1.246***

(3.41) (3.96) (8.61)

oper_inco 0.041*** 0.049*** 0.055***

(9.18) (10.15) (10.44)

dum -0.006 0.015 0.040

dum fair_value (-0.15) (0.37) (0.79)

2.852*** 1.835* -0.029

(2.65) (1.76) (-0.02)

power 0.323*** 0.272*** 0.283***

(6.97) (5.88) (5.12)

board 0.011 0.015 0.034**

(0.94) (1.19) (2.25)

inde_dir 0.628* 0.783** 0.111

(1.74) (2.11) (0.25)

lnsale 0.204*** 0.196*** 0.212***

(13.02) (11.43) (9.57)

salegrowth 0.000 0.000 0.000

(0.55) (1.26) (0.98)

leverage 0.001 0.016*** 0.018***

(0.44) (10.28) (9.21)

adj_return - 0.018 -0.017 -0.013

(-0.82) (-0.73) (-0.40)

lnaver_inco 0.330*** 0.261** 0.038

(3.04) (2.52) (0.32)

soe 0.033 -0.075* -0.199**

(0.76) (-1.67) (-3.84)

comp_com 0.222** 0.226** 0.271**

(2.02) (2.06) (2.14)

cross 0.140 0.148 0.150

(1.33) (1.43) (1.23)

chair -0.089 -0.105* -0.199**

(-1.52) (-1.86) (-2.90)

ceo -0.089 -0.068 -0.089

(-1.57) (-1.20) (-1.30)

regulate -0.136 -0.171* -0.263**

(-1.35) (-1.70) (-2.21)

middle -0.186** -0.197*** -0.269**

(-2.47) (-2.68) (-3.31)

west -0.403*** -0.388*** -0.440**

(-4.31) (-4.36) (-4.35)

Constant 5.621*** 6.561*** 8.265***

(5.17) (6.39) (6.96)

N 1146 1148 1148

Adj. R2 0.339 0.309 0.244

F 71.035 27.216 24.734

Notes: t Statistics are in parentheses.

Significant at the 10% level.

Significant at the 5% level.

*** Significant at the 1% level.

Table 8 reports the results of Model (3). The variable in column 1 is lntm, for which the regression coefficient on power * dum * fair_value is positive and significant almost at the 10% level. This result indicates that managerial power does have a significant influence on the stickiness of executive compensation and the stronger the managerial power, the more asymmetric the sensitivity of executive compensation is to CFV. Hypoth-

Table 8

Regression on whether managerial power influences the stickiness of executive compensation.

lntm 1 lncomp 2 lndir 3

fair_value 0.456*** 0.507*** 1.257***

(3.94) (4.49) (8.73)

oper_inco 0.041*** 0.049*** 0.055***

(8.99) (9.98) (10.35)

power 0.334*** 0.257*** 0.280***

(5.07) (3.90) (3.46)

dum 0.010 0.017 0.044

dum fair_value (0.21) (0.37) (0.72)

2.086* 1.192 -0.395

(1.78) (1.05) (-0.31)

power dum -0.072 -0.014 -0.019

fair_value power (-0.73) (-0.14) (-0.17)

-0.789 -0.787 -0.345

power dum fair_value (-0.63) (-0.66) (-0.29)

5.280+ 4.471 2.474

(1.60) (1.36) (0.64)

board 0.011 0.015 0.034**

(0.92) (1.18) (2.24)

inde_dir 0.610* 0.784** 0.107

(1.68) (2.10) (0.24)

lnsale 0.203*** 0.196*** 0.212***

(12.99) (1141) (9.55)

salegrowth 0.000 0.000 0.000

(0.57) (1.28) (0.98)

leverage 0.001 0.016*** 0.018***

(0.39) (10.26) (9.21)

adj_return -0.017 -0.016 -0.012

(-0.78) (-0.67) (-0.38)

lnaver_inco 0.328*** 0.259** 0.037

(3.03) (2.51) (0.31)

soe 0.033 -0.075* -0.199**

(0.75) (-1.68) (-3.84)

comp_com 0.240** 0.243** 0.281**

(2.20) (2.21) (2.17)

cross 0.144 0.152 0.152

(1.36) (1.46) (1.24)

chair -0.085 -0.103* -0.197**

(-1.45) (-1.80) (-2.85)

ceo -0.088 -0.066 -0.088

(-1.55) (-1.17) (-1.29)

regulate -0.136 -0.168* -0.262**

(-1.35) (-1.66) (-2.19)

middle -0.187** -0.198*** -0.270**

(-2.49) (-2.70) (-3.31)

west -0.401*** -0.388*** -0.440**

(-4.32) (-4.38) (-4.35)

Constant 5.627*** 6.566*** 8.267***

(5.19) (6.40) (6.96)

N 1146 1148 1148

Adj. R2 0.339 0.308 0.242

F 63.080 23.923 22.631

Notes: t Statistics are in parentheses. Significant at the 10% level. Significant at the 5% level. Significant at the 1% level. + Significant almost at the 10% level (p-value is 0.111).

Table 9

Robustness test on the relationship between executive compensation and CFV.

lnchair lnceo lncfo lnchair_ceo_cfo

fair_value -0.510 2.170* 0.910*** 0.992***

(-0.33) (1.88) (3.30) (4.07)

oper_inco 0.057*** 0.056*** -0.004 0.057***

(10.10) (13.14) (-1.02) (16.85)

power 0.465*** 0.293*** 0.257*** 0.377***

(5.94) (4.95) (4.45) (6.21)

board 0.003 0.011 -0.025* 0.010

(0.14) (0.69) (-1.75) (0.60)

inde_dir 1.294 1.007** 0.387 0.637

(1.64) (2.12) (0.81) (1.08)

lnsale 0.220*** 0.198*** 0.187*** 0.141***

(4.72) (9.06) (9.12) (4.77)

salegrowth 0.001*** 0.000 -0.003 0.000**

(3.35) (0.77) (-1.30) (2.53)

leverage -0.419** -0.274** -0.281** -0.320**

(-2.11) (-2.01) (-2.21) (-2.49)

adj_return -0.011 0.016 -0.010 -0.016

(-0.24) (0.59) (-0.35) (-0.51)

lnaver_inco 0.269 0.318** 0.521*** 0.213

(1.31) (2.41) (3.62) (1.50)

soe -0.118 -0.023 0.179*** -0.088

(-1.23) (-0.42) (3.34) (-119)

comp_comm 0.324 0.229 0.041 0.359**

(1.47) (1.35) (0.25) (2.27)

cross 0.380* 0.181 0.236* 0.276**

(1.92) (1.64) (1.93) (2.16)

chair -0.275* 0.029 -0.105 -0.278***

(-1.74) (0.40) (-1.27) (-3.49)

ceo 0.144 -0.420*** -0.144* -0.111

(1.31) (-5.53) (-1.92) (-1.57)

regulate -0.034 -0.067 0.064 -0.007

(-0.20) (-0.54) (0.49) (-0.06)

middle -0.234 -0.156* -0.015 -0.134

(-1.29) (-1.68) (-0.15) (-1.04)

west -0.195 -0.428*** -0.267** -0.303***

(-1.20) (-4.13) (-2.44) (-2.62)

Constant 4.742** 4.855*** 3.272** 7.795***

(2.08) (3.59) (2.20) (4.71)

N 800 843 810 1111

Adj. R2 0.151 0.261 0.271 0.141

F 22.746 26.504 16.592 26.600

Note: Lnchair is the natural logarithm of the chairman's compensation (if the chairman and CEO are the same person, the chairman's value is used). Lnceo is the natural logarithm of CEO compensation (if the CEO and CFO are the same person, the CEO's value is used). Lncfo is the natural logarithm of CFO compensation. Ln_chair_ceo_cfo is the natural logarithm of the total compensation of the chairman, CEO and CFO. The t statistics are in parentheses. Significant at the 10% level. Significant at the 5% level. Significant at the 1% level.

esis 3 is partially supported. The regression coefficients on the cross-multiple items in columns 2 and 3 are both positive, but not significant, which implies that the level of directors' compensation is influenced by managerial power, but the influence is not significant. A possible explanation is that managerial power originates from management, and if most directors have no position in management, the management lacks motivation to seek higher compensation for directors.

The regression results of other variables in Model (3) are consistent with those in Model (1).

Table 10

Robustness test on the asymmetry of executive compensation to CFV.

lntm D = 1 D=0 lncomp D=1 D=0 lndir D = 1 D=0

fair_value 3.004*** 0.308** 2.045** 0.429*** 0.675 1.192***

(2.76) (2.05) (2.01) (2.79) (0.54) (6.94)

oper_inco 0.150 0.041*** 0.152 0.049*** 0.223 0.055***

(116) (13.23) (1.11) (12.85) (1.38) (14.32)

power 0.293*** 0.331*** 0.273*** 0.261*** 0.281*** 0.292***

(4.46) (5.10) (4.13) (3.97) (3.75) (3.31)

board 0.020 0.002 0.024 0.004 0.043** 0.027

(1.27) (0.09) (1.39) (0.23) (2.07) (1.17)

inde_dir 0.863* 0.455 1.041** 0.655 0.564 -0.167

(1.73) (0.84) (2.01) (119) (0.91) (-0.25)

lnsale 0.195*** 0.221*** 0.191*** 0.198*** 0.209*** 0.208***

(8.89) (9.47) (8.31) (7.11) (7.87) (5.70)

salegrowth 0.000 -0.005*** 0.000** -0.005** 0.000** -0.009**

(1.30) (-3.18) (2.03) (-2.28) (2.00) (-3.20)

leverage 0.001 -0.309** 0.017*** -0.011 0.019*** 0.037

(0.77) (-2.37) (8.52) (-0.06) (8.37) (0.17)

adj_return -0.040* 0.051 -0.033 0.032 -0.012 -0.010

(-1.68) (1.00) (-1.26) (0.64) (-0.37) (-0.11)

lnaver_inco 0.350** 0.306* 0.290** 0.249 0.171 -0.076

(2.54) (1.81) (2.18) (1.53) (1.17) (-0.38)

soe 0.045 0.016 -0.059 -0.090 -0.183*** -0.195**

(0.76) (0.24) (-0.97) (-1.36) (-2.63) (-2.40)

comp_com 0.120 0.335** 0.170 0.282** 0.251 0.228

(0.81) (2.31) (1.12) (2.13) (1.53) (1.21)

cross 0.092 0.200 0.132 0.182 0.091 0.237

(0.80) (1.14) (1.14) (1.08) (0.61) (1.24)

chair 0.015 -0.183** -0.026 -0.168* -0.095 -0.281**

(0.18) (-2.15) (-0.34) (-1.96) (-1.05) (-2.70)

ceo -0.203*** -0.017 -0.184*** 0.028 -0.214** 0.009

(-2.86) (-0.19) (-2.63) (0.30) (-2.54) (0.09)

regulate -0.193 -0.068 -0.250 -0.088 -0.290 -0.229*

(-1.11) (-0.56) (-1.43) (-0.73) (-1.45) (-1.65)

middle -0.182* -0.186* -0.176* -0.209* -0.220* -0.312**

(-1.78) (-1.68) (-1.71) (-1.94) (-1.94) (-2.65)

west -0.544*** -0.243* -0.530*** -0.237* -0.563*** -0.307**

(-4.57) (-1.72) (-4.77) (-1.73) (-4.27) (-2.04)

Constant 5.550*** 5.657*** 6.282*** 6.720*** 6.841*** 9.668***

(3.96) (3.37) (4.69) (4.26) (4.49) (5.18)

N 614 532 614 534 614 534

Adj. R2 0.355 0.332 0.319 0.298 0.276 0.213

F 169.124 24.797 19.388 20.962 15.313 25.777

F-test 322.14 110.67 9.07

p-Value 0.0000 0.0000 0.0027

Note: D is a dummy variable assigned the value of 1 for PCFV or assigned the value of 0 for LCFV. t statistics are in parentheses. Significant at the 10% level. Significant at the 5% level. Significant at the 1% level.

5.4. Robustness tests

5.4.1. Robustness test on the relationship between executive compensation and CFV

We use the compensation of the chairman (lnchair), CEO (lnceo), CFO (Incfo) and the total of these three (lnchair_ceo_cfo) as executive compensation variables to re-test Hypothesis 1. The results are shown in Table 9.

Table 11

Robustness test on whether managerial power influences the stickiness of executive compensation.

Intm lncomp Indir

power = 1 power = 0 power = 1 power = 0 power = 1 power = 0

fair_value -0.815 0.403*** -0.581 0.456*** 0.779 1.219***

(-0.64) (3.39) (-0.47) (3.98) (0.62) (8.29)

oper_inco 0.382* 0.038*** 0.392* 0.047*** 0.419 0.053***

(1.66) (14.43) (1.67) (16.18) (1.34) (16.38)

dum -0.036 0.013 0.024 0.019 0.048 0.046

(-0.42) (0.27) (0.28) (0.40) (0.50) (0.74)

dum fair_value 7.575*** 2.131* 6.039** 1.173 2.244 -0.439

(2.66) (1.78) (2.12) (1.02) (0.66) (-0.33)

board 0.019 0.010 0.045 0.009 0.061* 0.028

(0.71) (0.75) (1.63) (0.64) (196) (1.59)

inde_dir 2.971*** 0.010 2.970*** 0.241 1.559 -0.282

(4.17) (0.03) (3.60) (0.60) (1.50) (-0.57)

lnsale 0.264*** 0.182*** 0.267*** 0.175*** 0.296*** 0.185***

(6.84) (10.32) (6.76) (9.04) (5.96) (7.29)

salegrowth -0.009 0.000 -0.009* 0.000 -0.017** 0.000

(-1.56) (0.67) (-1.66) (1.10) (-2.46) (1.01)

leverage -0.333* -0.001 -0.485*** 0.014*** -0.439** 0.015***

(-1.77) (-0.59) (-2.78) (8.29) (-2.19) (6.98)

adj_return -0.021 -0.014 -0.032 -0.007 -0.007 -0.015

(-0.52) (-0.51) (-0.78) (-0.25) (-0.15) (-0.36)

lnaver_inco 0.406** 0.317** 0.295 0.264** -0.033 0.093

(2.00) (2.55) (1.45) (2.21) (-0.14) (0.66)

soe 0.091 0.045 0.012 -0.079 -0.096 -0.208***

(0.97) (0.89) (0.13) (-1.50) (-0.88) (-3.38)

comp_com 0.304 0.194* 0.313 0.190* 0.349 0.239*

(1.43) (1.79) (1.39) (1.88) (1.39) (1.74)

cross -0.304 0.261** -0.268 0.254** -0.200 0.245*

(-1.56) (2.23) (-1.32) (2.20) (-0.97) (1.76)

chair 0.065 -0.099 0.066 -0.122* -0.127 -0.183**

(0.49) (-1.51) (0.61) (-1.82) (-0.92) (-2.26)

ceo -0.228** -0.058 -0.200* -0.038 -0.241* -0.057

(-2.03) (-0.88) (-1.87) (-0.58) (-1.81) (-0.71)

regulate -0.061 -0.135 -0.116 -0.153 -0.267 -0.233*

(-0.28) (-1.24) (-0.50) (-1.43) (-0.95) (-1.88)

middle -0.173 -0.195** -0.137 -0.217** -0.209 -0.283***

(-1.14) (-2.26) (-0.91) (-2.57) (-1.25) (-2.97)

west -0.473** -0.360*** -0.405** -0.353*** -0.597*** -0.356***

(-2.18) (-3.52) (-2.28) (-3.46) (-2.82) (-3.14)

Constant 3.035 6.441*** 4.018* 7.267*** 6.815** 8.525***

(1.36) (5.27) (1.78) (6.33) (2.58) (6.36)

N 279 867 279 869 279 869

Adj. R2 0.426 0.310 0.432 0.274 0.392 0.199

F 11.317 128.312 10.914 32.266 9.956 29.201

F-test 20.65 17.91 4.11

p-Value 0.0000 0.0000 0.0428

Note: t statistics are in parentheses.

Significant at the 10% level.

Significant at the 5% level.

Significant at the 1% level.

From the robustness test results reported in Table 9, we find that lnceo, lncfo and lnchair_ceo_cfo are sig-

nificantly positively related to CFV. The results are consistent with the conclusions of Zhou et al. (2010).9

9 Zhou et al. (2010) find that the CEO and chairman are not responsible for CFV and their compensation has no significant relationship with CFV. The CFO, however, is in charge of capital operations, and this officer's compensation is significantly related to CFV.

5.4.2. Robustness test on the asymmetry of executive compensation to CFV

We divide the sample into two sub-groups: one group comprises companies earning PCFV (D = 1), and the other comprises companies suffering LCFV (D = 0). We then conduct an F-test to examine Hypothesis 2, by testing the difference in the coefficients of fair_value between the two groups. The robustness results shown in Table 10 remain unchanged compared with those of Table 7.

5.4.3. Robustness test on whether managerial power influences the stickiness of executive compensation

We further categorize the sample into two sub-groups according to managerial power, one group comprising companies with strong managerial power (power = 1) and the other group comprising companies with weak managerial power (power = 0). We conduct an F-test to verify Hypothesis 3 by testing the difference in the coefficient for dum * fair_value between the two groups and find that the results remain unchanged (see Table 11).

5.4.4. Eliminating the effects of the financial crisis

In 2008, domestic companies suffered severely from the global financial crisis. Taking into consideration the noise caused by this crisis on the value relevance of financial reports, earnings management and accounting conservatism, we drop the observations from 2008 and test the three hypotheses again. The results remain largely unchanged.

5.4.5. Eliminating cross-listed companies

Zhang (2011) finds that the sensitivity of compensation in cross-listed companies is significantly higher than that in mainland-listed companies. Therefore, we remove 72 cross-listed companies from the sample and test the three hypotheses again. The results remain largely unchanged.

6. Conclusions and limitations

6.1. Conclusions

Examining A-share listed companies on the Shanghai Securities Exchange and Shenzhen Securities Exchange from 2007 to 2009, we test the sensitivity of executive compensation to CFV. We draw the following conclusions: (1) CFV is positively related to executive compensation, and (2) the sensitivity of executive compensation is asymmetric to CFV. That is, executive pay rises higher due to PCFV than it declines due to LCFV. Further examination reveals that the greater the degree of managerial power, the more asymmetric the sensitivity of executive compensation is to CFV. These findings suggest the following policy implications.

With the implementation of the new accounting standards, changes in accounting measurement attributes have influenced the levels of executive compensation. The effects of CFV on executive compensation should therefore be considered in the design of compensation contracts and changes in fair value that come from management effort, from opportunism or from changes in the market environment should be reasonably distinguished.

This study offers some insights into the motives and methods of compensation manipulation using CFV. It suggests that in the design of compensation contracts, more attention should be paid to the dangers of management opportunism. The role that CFV plays in the compensation contract should be cautiously balanced.

If corporate governance and internal control mechanisms are perfected and the power of management to manipulate compensation is suppressed, then compensation contracts based on performance will function better and a rational compensation system may truly come into being.

6.2. Limitations

Despite its potential contributions, our study also has several limitations, which suggest possible directions for further research.

We excluded observations in which CFV was reported as 0 in annual reports, which may have led to bias in sample selection, thus limiting the generalizability of our conclusions.

The measurement of managerial power in our study only includes duality, equity disparity and managers' tenure, which are probably insufficient to accurately measure the real conditions of managerial power in Chinese listed companies.

We ignore perks in the calculation of executive compensation. Perks are common and even a major component of compensation in listed Chinese companies, especially in SOEs. Chen et al. (2005) find that perks are a substitute for bonuses for SOE managers who are under compensation regulation. Excluding perks from executive compensation may influence the results.

The effect of earnings management on CFV is also beyond our consideration. Ye et al. (2009) find that managers commonly recognize a large proportion of corporate financial assets as available for sale to reduce the effect of CFV on their income statements. In that case, it becomes difficult to account for managements' real intentions in holding financial assets and this factor may have affected our results.

Our study provides evidence on the sensitivity of executive compensation to CFV. In practice, however, there are many other factors that may affect executive compensation. Better handling of these issues will involve a complex process of developing standards, dealing with competing interests in the process of making contracts and better implementation and re-evaluation of compensation contracts.

Acknowledgments

We thank Professor Zhihong Chen and the anonymous referees for their valuable comments and constructive advice. We are particularly indebted to Dequan Jiang for his selfless help in data processing and analysis. We also thank Junly Yu, Shangkun Liang, Professor Xiongsheng Yang, Donghua Chen and Kai Zhu for their patient help. However, we are completely responsible for any flaws in the study. Our study has been supported by the Leading Academic Discipline Project from the Education Committee of Shanghai (Project No. J51701), the National Natural Science Foundation of China (Project No. 70902063), the Humanities and Social Sciences Foundation of Ministry of Education of China (Project Nos. 09YJC630157 and 12YJA790197), the Key Research and Innovation Project of Shanghai Municipal Education Commission (11ZS86) and the IAPHD Project of Nanjing University.

References

Bebchuk, L.A., Jesse, F., David, W., 2002. Managerial power and rent extraction in the design of executive compensation. University of

Chicago Law Review 69, 751-846. Bebchuk, L.A., Fried, J.M., 2003. Executive compensation as an agency problem. Journal of Economic Perspectives 3, 71-92. Chen, D.H., Chen, X.Y., Wan, H.L., 2005. Regulation and non-pecuniary compensation in Chinese SOEs. Economic Research Journal 02, 92-101, in Chinese.

Chen, J., 2010. Factors influencing top-management incentives of listed companies: empirical evidence from Chinese stock market. Journal

of Shanxi Finance and Economics University 09, 70-79 (in Chinese). Chen, X.M., Liu, S.M., 2003. An empirical study on the structural difference among the compensation of managers in Chinese public

companies. Economic Research Journal 08, 55-63 (in Chinese). Cheng, S., Indjejikian, R., 2009. Managerial influence and CEO performance incentives. International Review of Law and Economics 29 (2), 115-126.

Conyon, M.J., 2006. Executive compensation and incentives. The Academy of Management Perspective 20 (1), 25-44.

Core, J., Holthausen, R., Larker, D., 1999. Corporate governance, chief executive compensation and firm performance. Journal of

Financial Economics 51, 371-406. Coughlan, A.T., Schmidt, R.M., 1985. Executive compensation, management turnover, and firm performance: an empirical investigation.

Journal of Accounting and Economics 7 (1-3), 43-66. Cyert, R., Kang, S., Kumar, P., 2002. Corporate governance, takeover, and top-management compensation: theory and practice.

Management Science 48 (4), 453-469. Dechow, P.M., Linda, A.M., Catherine, S., 2010. Fair value accounting and gains from asset securitizations: a convenient earnings

management tool with compensation side-benefits. Journal of Accounting and Economics 49 (1-2), 2-25. Dorff, M.B., 2005. Does one hand wash the other? Testing the managerial power and optimal contracting theories of executive

compensation. Journal of Corporation Law 30, 255-307. Du, X.Q., Wang, L.H., 2007. Empirical research on correlation between compensation schemes of management and the change of

performance of public listed companies. Accounting Research 01, 53-65 (in Chinese). Duffhues, P., Kabir, R., 2008. Is the pay-performance relationship always positive? Evidence from the Netherlands. Journal of Multinational Financial Management 18 (1), 45-60.

Edmans, A., Gabaix, X., Landier, A., 2008. A Multiplicative Model of Optimal CEO Incentives in Market Equilibrium. Working Paper, University of Pennsylvania.

Fahlenbrach, R., 2009. Shareholder rights, boards, and CEO compensation. Review of Finance 13 (1), 81-113.

Fang, J.X., 2009. Is top management compensation of Chinese public companies sticky? Economic Research Journal 03, 110-124 (in Chinese).

Figler, R.A., Lutz, C.R., 1991. Do the excellent firms effectively match CEO compensation to corporate performance? Journal of Managerial Issues 111 (4), 445-457.

Garvey, G.T., Milbourn, T.T., 2006. Asymmetric benchmarking in compensation: executive are rewarded for good luck but not penalized for bad. Journal of Financial Economics 82 (1), 197-225.

Gaver, J., Gaver, K., 1998. The relation between nonrecurring accounting transactions and CEO cash compensation. The Accounting Review 73, 235-253.

George, P.B., Hall, B.J., 1998. CEO Incentives and Firm Size. NBER Working Paper No. 6868.

Gopalan, R., Milbourn, T., Song, F., 2008. Strategic Flexibility and the Optimality of Pay for Luck. Working Paper. Washington University in St. Louis.

Grossman, S., Hart, O., 1983. Implicit contracts under asymmetric information. Quarterly Journal of Economics 98, 123-156.

Hambrick, D.C., Finkelstein, S., 1995. The effects of ownership structure on conditions at the top: the case of CEO pay raises. Strategic Management Journal 16, 175-193.

Hart, O., Moore, J., 1990. Property rights and the nature of the firm. Journal of Political Economy 98, 1119-1158.

Holmstrom, B., Milgrom, P., 1991. Multitask principal-agent analyses: incentive contracts, asset ownership, and job design. Journal of Law, Economics, and Organization 7, 24-52.

Hou, Q.C., Jin, Q.L., 2010. New Accounting Standard, Corporate Accounting Performance and Executive Compensation. Working Paper (in Chinese).

Huang, Z.Z., 2006. Share-holding ratio, tunneling strategy and totally circulating. Nankai Business Review 09, 58-65 (in Chinese).

Jensen, M.C., Murphy, K.J., 1990. Performance pay and top management incentives. Journal of Political Economy 98 (2), 225-264.

Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3 (4), 305-360.

Jiang, W., 2008. Agent cost of liability and executive compensation: empirical analysis on listed companies in China. Economic Science 04, 110-123 (in Chinese).

Lambert, R., Larcker, D.F., 1987. An analysis of the use of accounting and market measures of performance in executive compensation contracts. Journal of Accounting Research 25, 85-125.

Leone, J.W., Zimmerman, J., 2006. Asymmetric sensitivity of CEO cash compensation to stock returns. Journal of Accounting and Economics 42, 167-192.

Li, Z.Q., 2000. Incentive mechanism and enterprise performance: an empirical study based on listed companies. Accounting Research 01, 24-30 (in Chinese).

Lu, R., 2007a. Analysis of the agent problem of compensation incentives in the perspective of managerial power. Modern Management Science 07, 40-42 (in Chinese).

Lu, R., 2007b. Managerial power, compensation gap and performance. South China Journal of Economics 07, 60-70 (in Chinese).

Lu, R., 2008. Managerial power, compensation and sensitivity of performance: empirical evidence from listed companies in China. Contemporary Finance and Economics 7, 107-112 (in Chinese).

Lu, R., Wei, M.H., 2008. Research developments in managerial power approach of compensation determination. Economic Management 01, 90-93 (in Chinese).

Lyu, C.J., Zhao, Y.H., 2008. A study on the effect of the incentive given to managers of state-owned enterprises. Management World 11, 99-109 (in Chinese).

McGuire, J.W., Chiu, J.S.Y., Elbing, A.O., 1962. Executive incomes, sales and profits. The American Economic Review 52 (4), 753-761.

Morse, A., Nanda, V.K., Seru, A., 2008. Are Incentive Contracts Rigged by Powerful CEOs? Working Paper SSRN No. 687504.

Murphy, K.J., 2000. Performance standards in incentive contracts. Journal of Accounting & Economics 30 (3), 245-278.

Na, C.H., 2009. Managerial Power, Discretionary Investment and Executive Compensation: Empirical Study based on Chinese Listed Companies. Jinan University (in Chinese).

Natarajan, R., 1996. Stewardship value of earnings components: additional evidence on the determinants of executive compensation. The Accounting Review 71, 1-22.

Otten, J.A., 2008. Theories on Executive Pay: A Literature Overview and Critical Assessment. MPRA Working Paper No. 6969.

Pan, F., Tong, W.H., 2005. The Incentive Contract of Top Managers: Existing Condition Analysis - Evidence from China. China Accounting Society of China Annual Academic Symposiums, 2005 (in Chinese).

Quan, X.F., Wu, S.N., Wen, F., 2010. Managerial power, private income and compensation rigging. Economic Research Journal 11, 7386 (in Chinese).

Sloan, R., 1993. Accounting earnings and top executive compensation. Journal of Accounting and Economics 16, 55-100.

Sun, Z., Liu, H., 2004. The expense "Stickiness" behavior of Chinese listed companies. Economic Research Journal 12, 26-34 (in Chinese).

Taussings, F.W., Baker, W.S., 1925. American corporations and their executives: a statistical inquiry. Quarterly Journal of Economics 3, 1-51.

Tosi, H.L., Werner, S., Katz, J.P., Gomez-Mejia, L.R., 2000. How much does performance matter? A meta-analysis of CEO pay studies. Journal of Management 26 (2), 301-339.

Wang, K.M., Wang, Z.C., 2007. Executive controlling power, compensation, and earnings management: empirical study on listed companies in China. Management World 07, 111-119 (in Chinese).

Wei, G., 2000. Incentives for top-management and performance of listed companies. Economic Research Journal 03, 32-39 (in Chinese).

Wiseman, R.M., Gomez-Mejia, L.R., 1998. A behavioral agency model of managerial risk taking. The Academy of Management Review 23 (1), 133-153.

Xin, Q.Q., Lin, B., Wang, Y.C., 2007. Government control, executive compensation and capital investment. Economic Research Journal 08, 110-112 (in Chinese).

Xu, J.C., Zeng, X.Y., 2010. The CEOs compensations contract and fair value accounting in China. Accounting Research 03, 12-19 (in Chinese).

Ye, J.F., Zhou, L., Li, D.M., Guo, L., 2009. Management motivation, accounting policy options and earnings management: an empirical study on the classification of financial assets for listed companies under the new accounting standards. Accounting Research 03, 25-30 (in Chinese).

Zhang, B.W., Shi, J.T., 2005. Board characteristic, top managers compensation and pay-performance sensitivity in Chinese listed companies: an empirical analysis. Policy-making Reference 04, 32-39 (in Chinese).

Zhang, J.R., Zhang, F.D., Zhou, H.F., 2011. A study on the different effects of two types of the changes in fair value on executive's compensations: empirical evidence from Chinese A-share listed firms in 2007-2008. Accounting Research 10, 63-68 (in Chinese).

Zhang, J.R., Zhao, J.W., Zhang, J., 2003. Empirical study on the relevance of top management incentive and operating performance of listed companies. Accounting Research 09, 29-34 (in Chinese).

Zhang, L., 2011. Study on the Correlation between Executive Pay and Corporate Performance of Cross-listed Companies and Non-cross-listed Companies. Nanjing Normal University (in Chinese).

Zhou, H.F., Xin, Q.Q., Zhang, J.R., 2010. Fair value measurement and executive compensation contract. Economic Science 05, 102-110 (in Chinese).