CEO DUALITY, CEO COMPENSATION, AND EARNINGS MANAGEMENT: EVIDENCE FROM CHINA

How to cite this paper: Bao, J., & Xue, C. (2023). CEO duality, CEO compensation, and earnings management: Evidence from China [Special issue]. Corporate Ownership & Control, 20 (3), 317–327. https://doi.org/10.22495/cocv20i3siart7


INTRODUCTION
Following the recent bankruptcies of business giants in the early 2000s, chief executive officer (CEO) duality, the situation in which one person serves as both the CEO and chairperson has gained significant attention (Wijethilake & Ekanayake, 2020). Agency theory and stewardship theory are two contradictory theories discussing CEO duality. Finkelstein and D'aveni (1994) propose that CEO duality is a -double-edged sword‖ because duality increases the danger of entrenchment by placing a considerable amount of formal authority in the hands of CEOs. Still, it creates a hierarchy of command that gives stakeholders a reason to feel secure. Several studies focus on this phenomenon but get mixed findings. Therefore, the role of CEO duality in the organization and its influence are still unclear and need further investigation, especially for Chinese companies. CEO duality situation accounts for over one-fourth of Chinese non-state-owned enterprises in recent years. In many state-owned enterprises, the chairpersons also serve as CEOs, which means even if they are not CEOs; they are still the controller (Jiang & Kim, 2015). Therefore, CEO duality may be more prevalent in China than what data displays because the chairperson is likely to be the de facto CEO.
Chinese listed companies have strong interest and incentives to manage their earnings, especially when they intend to avoid de-listing because of consecutive net loss reported and when they want to reach legal criteria on return on equity for raising additional equity capital (Liu & Lu, 2007). It is also a common concern that CEOs manage earnings to increase current profit and then improve their compensation. Chinese listed companies did not introduce equity-based compensation as early as other countries and mainly use cash-based compensation. Thus, limited studies investigate the relationship between equity-based compensation and earnings management in China. However, the relationship between cash-based compensation and earnings management also remains unsettled. We are also interested in when a CEO is also the chairperson, and when the CEO earns a higher level of compensation, whether the CEO will be more likely to engage in earnings management. Therefore, we expect to answer the following research questions: RQ1: Whether CEO duality increases earnings management?
RQ2: How CEO compensation influences earnings management? RQ3: Whether CEO duality can strengthen or weaken the relationship between CEO compensation and earnings management?
Previous research shows a positive relationship between CEO/chairperson dual and earnings management (Lin, 2014). In terms of CEO compensation, empirical results are mixed. For example, equity-based compensation for CEOs is a way to help resolve agency problems. However, it can also trigger conflicts of interest between shareholders and managers because it encourages executives to control profitability (Alkebsee et al., 2022). On the other hand, cash-based compensation is not affected by changes in the stock price, thus reducing the temptation to manipulate profits for the CEO (Alkebsee et al., 2022;Ye, 2014). Nevertheless, Li et al. (2011) find that executive cash-based compensation has a significant positive effect on accrual earnings management and real earnings management. Therefore, the relationship between CEO compensation and earnings management is hard to predict. We do not assume any positive or negative relationship between CEO duality and earnings management, or any moderating role of CEO duality.
In this research, we study whether CEO duality and CEO compensation are associated with earnings management. We examine both accrual earnings management and real earnings management. We measure CEO compensation in terms of cash-based compensation and equity-based compensation. We also examine the moderating effect of CEO duality in the relationship between CEO compensation and earnings management. Compared to the previous study (Alhmood et al., 2020;Hashim & Devi, 2008;Zhang et al., 2008), this study has a larger sample with 8,419 firm-year observations from 2011 to 2021. We employ the modified Jones model (Dechow et al., 1995) and Roychowdhury model (Roychowdhury, 2006) to measure accrual earnings management and real earnings management and take cash-based compensation and equity-based incentives as measures of CEO compensation. We find significant and positive relationships between CEO duality and earnings management. The results also show a positive impact of cash-based compensation on earnings management and a negative impact of equity-based incentives on real earnings management. Moreover, CEO duality can weaken the negative relationship between CEO equity-based compensation and real earnings management.
The contributions of this paper are as follows. The study is structured as follows. Section 2 reviews the relevant literature and hypotheses development. Section 3 describes data, research design, and summary statistics. Section 4 presents the main regression results and additional analysis. Section 5 contains a discussion. Section 6 highlights the study's conclusion, limitations, and implications for future research.

CEO duality and earnings management
Earnings management refers to using managerial discretion in transaction structuring to alter financial reports, either to deceive stakeholders about the firm's performance or to profit from a contractual outcome that depends on accounting numbers (Healy & Wahlen, 1999). Therefore, earnings management may be a kind of agency cost if managers issue financial reports that do not accurately depict the company's economic situation and shareholders subsequently make unfavourable investment decisions (Davidson et al., 2004). Earnings management can be divided into accrual earnings management and real earnings management.
Accrual earnings management involves manipulating accounting entries to alter the number of reported earnings in a particular accounting period. On the other hand, real earnings management involves manipulating a company's actual operations to alter reported earnings (Roychowdhury, 2006).
The agency theory suggests that there may be a conflict of interest between the firm and the manager, as the manager may have incentives that are not aligned with the firm's interests. The managers may prioritize their interests, such as maximizing their private benefits and minimizing personal risk, rather than the interests of their firms (Boyd, 1995;Fama & Jensen, 1983;Jensen & Meckling, 1976;Ross, 1973). To mitigate the agency problem, the companies shall separate the board from the CEO to perform its monitoring function effectively (Jensen, 1993). However, CEO duality is a common phenomenon in corporate governance, which is a practice in which the chairperson of the board also holds the position of CEO (Dalton et al., 2007;Davidson et al., 2004). The board of directors serves as the common apex of decision control systems in firms, exerting top-level decision control rights, including selecting and removing CEOs as well as approving and overseeing significant decisions (Fama & Jensen, 1983). Thus, CEO duality would weaken the board's role of supervision and control over the practices and performances of the CEO, and affect firm performance adversely (Boyd, 1995).
Agency theory indicates that the combination of CEO and chairperson increases CEO control over the board, impairs the independence of the board, and reduces the effectiveness of the board as an internal supervising instrument in ensuring efficient control, advisory, and monitoring procedures (Bliss, 2011;Daily & Dalton, 1997;Giannarakis, 2014). Accounting manipulations are more frequent when CEO power is excessively high (Feng et al., 2011;Klein, 2002). As a result, the CEO who is also the chairperson is more likely to manage earnings (Baker et al., 2019; Nuanpradit, 2019). Lin (2014) investigates 278 companies listed on the Shanghai Stock Exchange (SSE) and Shenzhen Stock Exchange (SZSE) and finds a positive relationship between CEO duality and earnings management. He also finds that the board's independence level can moderate this positive association.
On the other hand, stewardship theory holds divergent views from agency theory (Donaldson & Davis, 1991). It suggests that managers are responsible for acting as stewards of the resources and assets entrusted to them by shareholders or other stakeholders. The theory assumes that managers are motivated by a sense of responsibility and duty to the organization and its stakeholders, rather than by the pursuit of personal gain. Different from the assumptions in agency theory that CEO are opportunistic agents who will enhance personal interest at the expense of shareholders' interest, stewardship theory believes that there is no interest conflict between managers and owners, and the ideal organizational structure gives managers the freedom to act as they do in the owners' best interests to achieve coordination most successfully (Donaldson, 1990). As a result, CEO duality shall be negatively associated with opportunistic behaviours (Donaldson, 1990;Donaldson & Davis, 1991).
Limited evidence is found to support stewardship theory in earnings management literature. Alhmood et al. (2020) examine listed companies on the Amman Stock Exchange (ASE) and indicated that dual CEOs are negatively related to real earnings management, which is consistent with stewardship theory. As a result, we propose our hypotheses as follows: H1a: CEO duality is positively associated with earnings management.
H1b: CEO duality is negatively associated with earnings management.

CEO compensation and earnings management
CEO incentives, which typically refer to their compensation, are critical in corporate governance. These incentives are intended to motivate CEOs to achieve the organization's goals and align their interests with those of the company and its shareholders. The design of CEO compensation contracts can effectively align CEO incentives with those of the owners ( H2a: CEO cash-based compensation is not related to earnings management.
H2b: CEO equity-based incentives are not related to earnings management.
CEOs with more power receive higher overall remuneration than less powerful CEOs (Song & Wan, 2019). Baker et al. (2019) investigate the effect of CEO power on the earnings management outcome. They find that powerful CEOs prefer accruals earnings management and inhibit CFOs' earnings management in only the pre-Sarbanes-Oxley Act (SOX) period. Due to the relationship between incentives and earnings management is not clear in general, we propose the following hypothesis: H3: CEO duality has no moderating role in the relationship between CEO incentives and earnings management.

Data and sample selection
We collect CEO duality, CEO compensation, and financial data of Chinese listed companies from Chinese Securities Market and Accounting Research (CSMAR) databases. We start with the sample including all listed firms for the period of 2011-2021, and exclude financial industry firms since their special accounting requirements make the estimation of accruals and abnormal accruals difficult (Klein, 2002). We delete firms that belong to industries that have fewer than 10 companies and remove samples that miss variables. Each firm is required to have at least eight years of observations in our sample. The final sample contains 8,419 firmyear observations with 881 unique firms in 14 industries 1 . All the firms are A-shares listed on the SSE and the SZSE and mainly operate in China. However, the major shareholders may be the state or individuals of Chinese or foreign nationality.

Accrual earnings management
We use the absolute value of discretionary accruals to measure the degree of accrual earnings management. We employ a cross-sectional modified Jones model to estimate discretionary accruals (Dechow et al., 1995). Specifically, we first use Eq. (1) to estimate the parameters , , by industry and year, and then apply three parameters to Eq. (2) to calculate discretionary accruals. Then we obtain the absolute value of discretionary accruals. In order to eliminate heteroscedasticity, all variables in the accrual expectation model are deflated by lagged total assets (Hashim & Devi, 2008). Following Liu et al. (2014), we define total accruals as operating profit minus net cash flow from operations.
We first estimate discretionary accruals based on the following cross-sectional model by industry: (1) where, represents the total accruals for firm i in year t; is total assets for firm i at the end of the previous year ; change in revenue for firm i between year t and the previous year ; and is total property, plant and equipment for firm i in year t.
The estimated coefficients from Eq. (1) are used to estimate the discretionary accruals ( ) for firm i in year t: We use the absolute value of discretionary accruals, , to proxy for accrual earnings management. 1 The 14 industries include: 1) farming, forestry, animal husbandry and fishery industry, 2) mining industry, 3) manufacturing industry, 4) production and supply of electric power, gas and water, 5) construction industry, 6) wholesale and retail trade industry, 7) traffic, storage and mail business industry, 8) information transfer, software and computer service industry, 9) realty business industry, 10) leasehold and business service industry, 11) scientific research and technical service industry, 12) water conservancy, environment and public institution management industry, 13) cultural, physical and entertainment industry, and 14) comprehensive industry. The industry classification is based on the China Securities Regulatory Commission (CSRC) industry classification 2012 version.

Real earnings management
We measure real earnings management by employing the model in Roychowdhury (2006). We consider three aspects for the abnormal level of real activities: 1) abnormal level of cash flow from operations, 2) abnormal production costs, and 3) abnormal discretionary expenses.
The normal levels of cash flow from operations, production costs, and discretionary expenses are measured as: where, is net cash flow from operations for firm in year ; is revenue for firm in year ; is product cost for firm in year (defined as the cost of goods sold in year plus the change in inventory between year t and the previous year ); is discretionary expenditures for firm in year (defined as selling and administration expense).
We estimate the cross-sectional model on the industry level. The estimated residuals from Eq. (3) capture the abnormal level of cash flows from operations ( ). The estimated residuals from Eq. (4) capture the abnormal level of production costs ( ). And the estimated residuals from Eq. (5) capture the abnormal level of discretionary expenses ( ). We aggregate the three abnormal level measures of real activities into one measure as follows: and use the absolute value of it as to capture the level of real earnings management.

Independent variables
The proxy for CEO duality is DUALITY, which equals one if CEO chairs the board and zero otherwise. CEO compensation can be divided into cash-based compensation and equity-based incentives. Cash-based compensation (CSC) is calculated as the logarithm of the total cash compensation of the CEO. For equity-based incentives (CEI), we measure it by using the ratio between the number of shares owned by the CEO and the total shares of the company in the current year.
Following  Table 1. The number of shares owned by the CEO divided by the total shares of the company in the current year.

Discretionary accrual
The absolute value of discretionary accrual; the higher ABSDA, the higher the degree of accrual earning management.

Real earnings management
The absolute value of real earnings management; the higher ABSRM, the higher degree of real earnings management.

SIZE
Company size Natural logarithm of total assets. GROWTH Sales growth Sales growth over the last year.
BM Book-to-market ratio Book value of equity divided by the market value of equity.

Leverage
Book value of all liability divided by total assets.

CFO
Cash flow from operations Cash flow from operation in the current year divided by lagged total assets.

ROA
Return on assets Earnings before interest and tax (EBIT) divided by total book assets.

TOV
Asset turnover Equal to the ratio of main business revenue to average total assets at the beginning and end of the year.

Accounting flexibility
Equal to the ratio of net operating assets at the end of the previous year to the operating income of the previous year, the larger the value is in the table, the less room to manipulate accrual earnings is shown SOE State-owned enterprise Dummy variables. Equals one if the firm is a stated-owned enterprise and zero otherwise.

Ownership concentration
Equal to the ratio of the number of shares held by the top 10 shareholders to the total number of shares.

BSIZE
Board size The number of directors on the board.

Independent directors
The number of independent directors on the board divided by the total number of directors.

Female directors
The number of female directors on the board. Note: DUALITY, CSC, and CEI are used as independent variables. ABSDA and ABSRM are used as dependent variables. Other firmspecific variables are used as control variables.

Model specification
To investigate the relation between CEO duality, CEO compensation and earnings management, we use the regression model where is either or : ( ) To investigate the moderating effect of CEO duality, we employ the following regression model: Equation (7) is designed to test H1a, H1b, H2a, and H2b. Equation (8) is designed to test H3. We introduce the interaction of DUALITY and CEI to examine the moderating role of CEO duality.     Table 3 presents the correlations between variables. CEO duality is significantly positively correlated with discretionary accruals and real earnings management at a 5% level. CEO duality has an evident relationship with both CEO cash-based compensation and equity-based incentives positively, suggesting significant compensation differences between CEO and CEO who also serves as the chairperson. CEO cash-based compensation is positively related to both earnings management methods, which implies that a higher level of cash-based compensation may encourage CEOs to manage earnings to get better firm performance. There is no significant association between CEO equity-based incentives and earnings management.

Summary statistics and correlations
We also conduct a collinearity analysis using the variance inflation factor (VIF). When VIF < 10, there is no multicollinearity problem. In this study, VIF = 1.94. Therefore, it will not lead to multicollinearity concerns. Table 4 presents the OLS estimates of Eq. (7) with year-fixed effect. Columns (1) and (2) show that the estimated coefficient of DUALITY is positive and significant at 10% and 1%, respectively, suggesting that CEOs who also serve as chairpersons on the board tend to manage earnings, especially real earnings. The result is consistent with H1a that CEO duality is positively associated with earnings management.

RESULTS
Columns (3) and (4) report the results from regressions that include CEO compensation. The estimated coefficients of CSC are significant at a 1% level, indicating that CEOs with higher cashbased compensation are more likely to engage in earnings management. The estimated coefficient of CEI in column (5) is negative and insignificant, while it in column (6) is negative and significant at a 1% level. The results suggest that equity-based compensation allies CEOs' interest with the shareholders, leading to less earnings management, especially less real earnings management which has a negative impact on firms' future operations (Gunny, 2005). As a result, we reject H2a and H2b. Table 5 reports the regression results of Eq. (8). The coefficient of DUALITY × CEI is insignificant in column (1) but positive and significant at 1% level in column (2). The finding supports that CEO duality can weaken the interest alignment and strengthen the agency problem. It can be explained that CEO duality leads to fewer monitors from the board and then provides space for CEO/chairperson to manage earnings by the real earnings management approach. Therefore, we reject H3. However, we do not find any moderating role of duality in the relationship between equity-based incentives and accrual earnings management.   Table 6 presents the descriptive statistics of SOE companies and non-SOE companies. The means show that CEO duality occurs in 9.6% of state-owned enterprises and 29.3% of private companies, suggesting that CEO duality in stateowned enterprises is less frequent than in private companies.  Table 7 reports the regression results concerning the effects of state ownership. The coefficients of SOE are negative and significant in columns (1), (2), (5), and (6), suggesting that there are lower levels of both accrual earnings management and real earnings management in state-owned companies. Columns (3) and (4) report insignificant coefficients of CSC×SOE. Column (5) shows a positive and significant coefficient of CEI×SOE, indicating that CEO equity-based incentives promote accruals earning manipulations in state-owned enterprises. Real earnings management requires higher costs and can harm the growth of companies in the future because they often decrease capital investments, raise price concessions, and slash investment in research and development and funding in marketing (Enomoto et al., 2015;Wang & Yung, 2011). State-owned enterprises engage more in accrual earnings management than real earnings management probably since CEOs in state-owned enterprises do not have power as high as in private enterprises and they are more difficult to implement real earnings management which can help them get better performance in the short-term but at the expense of companies' future development. Therefore, a higher level of equity-based compensation can motivate CEO in state-owned enterprises to manage earnings through accruals rather than through real activities.  (1) and (2) report the regression results, where DUALITY × SOE is the variable in interest, and ABSDA and ABSRM are dependent variables respectively (same in columns (3) to (6)). Robust t-statistics in parenthesis. *, **, and *** represent the statistical significance at the 0.1, 0.05, and 0.01 levels respectively.

DISCUSSION
Agency theory suggests a potential conflict of interest between shareholders and managers. An independent board can monitor the CEO more effectively to mitigate the agency's problem. However, when the CEO serves as the chairman of the board, the monitoring role of the board is weakened. On the other hand, stewardship theory posits that individuals in leadership positions, such as CEOs, are motivated by intrinsic factors and have an inherent desire to act in the best interests of the organization and its stakeholders. As a result, CEO duality can align the interests of shareholders and CEOs. Empirical analyses of CEO duality and earnings management have yielded mixed results.
Our study seeks to provide additional evidence by studying the Chinese listed companies since they have strong incentives to manage earnings to avoid de-listing. We consider both accrual earnings management and real earnings management. We also consider the incentives of CEOs by measuring their cash-based compensation and equity-based compensation. We find that CEO duality is positively associated with both accrual earnings management and real earnings management. Specifically, our results suggest a more significant and consistent positive association between CEO duality and real earnings management. This may imply that it is less costly for more powerful CEOs to engage in real earnings management. Future research can further investigate the trade-off between accrual earnings management and real earnings management. While some argue that CEO duality can lead to efficiency and better decision-making, the controversial issues surrounding it underscore the ongoing debate in corporate governance about the optimal structure for separating leadership and oversight roles within an organization.
We also find that CEO cash-based compensation is positively associated with earnings management, while equity-based compensation is negatively associated with earnings management. The results suggest that equity-based compensation better align the interests of CEOs and shareholders. Although equity-based compensation may serve as a mechanism to mitigate agency problems by linking CEO compensation to stock price performance, it may also influence risk-taking behavior and the propensity to engage in earnings management by CEOs who engage in aggressive accounting practices to achieve higher stock prices and maximize their equity-based compensation. Our evidence supports the former. However, we also find that CEO duality can induce the likelihood of earnings management in the presence of equity-based compensation.

CONCLUSION
This paper mainly focuses on the relationship between CEO duality and earnings management, and the relationship between CEO compensation and earnings management. Using a sample of 8,419 firmyear observations covering listed companies in China from 2011 to 2021, we find that CEO duality is positively associated with both accrual and real earnings management. It provides evidence for agency theory in which CEO duality can weaken the board's monitoring role and offer greater power to dual CEO. The result also suggests that higher cash-based compensations provide managers incentives in both accrual earnings management and real earnings management, while higher equitybased incentives inhibit real earnings management. Equity-based incentives can align the interest of managers with the interest of shareholders, but cash-based compensations motivate managers to manage earnings. The moderating role of duality is significant in the relationship between equity-based incentives and real earnings management. We further find that CEO equity-based incentives promote accruals earning manipulations but not real earnings management in state-owned enterprises.
Our study contributes to the existing literature by providing additional evidence on CEO duality, CEO compensation, and earning management (Armstrong et  . We investigate the moderating role of CEO duality and find that it provides additional incentives for managers to engage in real earnings management rather than accrual earnings management. We also show that state-owned companies are less likely to manage earnings. However, if CEOs of state-owned companies are more potent by serving the chair, they are more likely to manage earnings through accruals rather than real activities.
Our study has the following limitations. First, although our findings can be generalized to listed companies in China, caution should be exercised in generalizing the conclusions to companies in other countries. Second, we consider control variables with reference to previous research, but there will always be other factors that may be correlated with earnings management that we ignore. Third, we use earnings management models used by other researchers and do not make any modifications according to the Chinese situation. Therefore, such models may be subject to measurement errors when we study Chinese companies.