Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The monitoring role of media on executive compensation

The monitoring role of media on executive compensation China Journal of Accounting Studies, 2013 Vol. 1, No. 2, 138–156, http://dx.doi.org/10.1080/21697221.2013.802974 a, a b Hong Luo *, Baohua Liu and Weiqian Zhang a b Accounting School, Southwestern University of Finance and Economics, China; Finance School, Shanghai Institute of Foreign Trade, China This paper investigates the monitoring role of media on executive compensation. Using data on media coverage from 2007 to 2009, we find that the media can serve as an effective external monitoring mechanism on executive compensation by affect- ing executive reputation. Specifically, negative media coverage and government intervention on media can enhance the monitoring role of media. We also find that the monitoring role of media and the market reaction to media coverage are stronger for non-state-owned corporations. Although the monitoring role of media is not dif- ferent between monopolistic and non-monopolistic industries, the market reaction to media coverage is stronger for companies in non-monopolistic industries. We also find that there is less coverage on monopolistic and state-owned companies. Keywords: reputation mechanism; media coverage; executive compensation 1. Introduction The governance role of the media has attracted great attention from scholars in recent years. Researchers believe that the media influence corporate behaviors through the reputation mechanism. Specifically, media coverage on corporations may attract the attention of investors and may further affect corporate reputation, and the exposed corporations have to react to media. Thus, the media may serve as an effective tool in external monitoring (Dyck & Zingales, 2002; Dyck, Volchkova, & Zingales, 2004, 2008; Joe, Louis, & Robinson, 2009; Miller, 2006; Monks & Minow, 2004). The financial crisis, which started in 2007, has badly hit the world economy. Central governments, such as the US government, offered huge financial aids to the corpora- tions trapped in the crisis. However, the executives in these corporations are still paid huge amounts of compensation, which arouses fierce criticism from the public. In the US, as well as in Britain, France, and Germany, governments immediately issued poli- cies to restrict executive compensation. Because of media coverage and government supervision, American corporations have experienced a 50% reduction in executive compensation in 2008, and a 90% reduction in monetary compensation in 2009. Execu- tive compensation in financial corporations has also been slashed in Britain, France, Germany, and so on. Meanwhile, the Chinese media have started to pay attention to corporate executive compensation. With overwhelming media coverage concerning executive compensation appearing in public, hundreds of corporate executives dominate the headlines. The sky-high compensation in Chinese corporations also attracts fierce *Corresponding author. Email: luohong@swufe.edu.cn Paper accepted by Hong Zou. 2013 Accounting Society of China China Journal of Accounting Studies 139 public criticism. Thus, whether media coverage has an impact on executive compensa- tion in China is an interesting research question. Reputation theory provides a new perspective to study the impact of media on exec- utive compensation. Reputation theory, introduced by Kreps and Wilson (1982) and revised by Fudenberg and Levine (1992), suggests that reputation can strengthen commitment and that the role of reputation is to provide implicit incentives for the long-term interests of participants to ensure their short-term commitment behavior and thus make reputation a substitute for explicit contracts. Weigelt and Camerer (1988) and Wilson (1985) explore the impact of reputation on participant behavior, and such impact could be labeled as ‘reputation effect’ or ‘reputation mechanism’. Fama (1980) finds that the reputation mechanism has an impact on the labor market. Extant literature finds that executives will suffer from reputation losses when firms’ financial frauds are detected (Fich & Shivdasani, 2007; Srinivasan, 2005), and their companies experience financial distress (Gilson, 1990), liquidation (Harford, 2003), poor performance (Yermack, 2004) or reduction of dividends (Kaplan & Reishus, 1990). Some studies also consider reputation mechanisms in China. Zhang (2002) argues that reputation could replace legislation to serve as a low-cost and effective mechanism in maintaining transaction order. Yang and Xu (2004) study the reputation incentive mecha- nism of executives in state-owned enterprises, and find that executives could gain more utility through pursing political reputation than pursuing operating performance. Xu and Luo (2007) test the effectiveness of the reputation mechanism in investment banks through the IPO market, and find that the reputation mechanism proves to be effective due to the desire for a higher market share and better career advancement by investment bankers. Extant literature has acknowledged the governance role of media from the perspec- tive of reputation mechanisms. Specifically, media could lower the private benefits of control rights (Dyck, Volchkova, & Zingales, 2004), uncover accounting irregularities (Miller, 2006), prevent corporations from expropriating the benefits of outsiders (Dyck et al., 2008; Li & Shen, 2010), and improve the operating efficiency of the board (Joe et al., 2009). Core, Guaya, and Larckerb (2008) examine the impact of media on executive com- pensation. Using firms having media coverage on their executive compensation from 1994–2002, they find that negative media coverage is insignificantly correlated with executive compensation reduction, and insignificantly correlated with executive turn- over. They ascribe the invalid role of media to its sensationalism. Whether media govern executive compensation out of sensationalism or social con- science and whether the reputation mechanism makes a difference in executive compen- sation monitoring remain interesting questions. Chen, Li, and Li (2005) argue that the main type of reputation mechanisms is the one that is consciously built, and assured by social norms and third-party governance. From the current media coverage on executive compensation and the corresponding social reactions, we believe that media coverage may have an impact on executive compensation, and such impact may work through social norms, namely, through arousing public opinion. Based on Chinese institutional background, this paper finds that media coverage, especially official and negative media coverage, plays a governance role in affecting executive compensation of Chinese listed companies. We contribute to the literature concerning the governance role of media, and provide a new perspective on the executive incentive mechanism. The rest of paper is organized as follows. Section 2 provides theoretical analysis and hypothesis development. Section 3 discusses our sample and research design. Section 4 reports empirical results. Section 5 provides further robustness tests. Section 6 concludes. 140 Luo et al. 2. Theoretical framework and research hypotheses 2.1. The governance role of media on executive compensation The media coverage on sky-high executive compensation affects the reputation of the listed companies and their executives, and to avoid reputation losses, listed companies should take action to adjust executive compensation. Thus, the media serve as an ex post monitoring tool in executive compensation. According to traditional reputation theory (Fama, 1980; Fama & Jensen, 1983), media coverage on huge compensation payments could affect the reputation of executives, and lower their human capital value. Additionally, politicians believe that failing to act on legal rules and regulations could damage their image and endanger their political career (Besley & Prat, 2006). Media coverage of huge compensation payments can pressure authorities to issue policies to restrict executive compensation. Li and Shen (2010) argue that the monitoring role of the media is achieved by raising the attention and intervention of administrations, and media coverage can raise corporate administrative costs and prevent corporations from damaging the interests of outsiders. Paligorova (2009) and Cheng and Indjejikian (2009) suggest that government legislation could influence CEOs’ compensation prac- tices. Kuhnen and Niessen (2008) demonstrate the monitoring role of the media on executive compensation using a sample of 22,507 examples of American newspaper coverage on executive compensation in the period 1990–2006. They report that execu- tive compensation levels and structures appear to change with public attitude and social norms. The media should represent the opinion of the public, and act as a rational and objective “recorder” of social justice. The negative coverage on certain issues could positively affect the whole of society (Yao, 2008). An article published in an important Italian newspaper cited directors of a large public corporation with the worst perfor- mance in the previous three years, and, as a result, two months later the CEO resigned without explanation (Dyck & Zingales, 2002). Facts prove that the more negative and furious the coverage, the stronger the social response it can bring about. Thus, media coverage on excessive executive compensation could endanger the reputation of the executives, pressuring executives to take further measures to appease public opinion and reduce the reputational risk, and the compensation increment would be under control. Scott and Konstantin (2009) find that government control over media is often stron- ger in less democratic countries. Currently, in China, official media coverage is wider, of higher reliability, and higher authority. Official media coverage often spreads rapidly to the public and results in a larger impact on public opinion. Official media coverage not only represents its voice and opinion, but reflects the will and determination of the government. Additionally, Shleifer and Vishny (1994) proposed the government support hypothesis, suggesting that the political connections that entrepreneurs try to establish with time and money often bring huge benefits (e.g., financing convenience, tax favor, property right protection) to entrepreneurs and companies (Faccio, Masulis, & McCon- nell, 2006; Fisman, 2001). From the perspective of administrative hierarchy, the official media in China often carry corresponding administrative ranks. For instance, the China Central Television (CCTV) is subordinate to the State Administration of Radio Film and Television (SARFT) and it ranks deputy ministerial, with the president of CCTV being a vice-ministerial official. Corporations and their executives are willing to maintain a good relationship with the official media to avoid the exposure of their misconducts, and once their misconducts are reported, corporations would respond quickly to the coverage. Thus, we expect that executives are more likely to make China Journal of Accounting Studies 141 adjustment to compensation once their excessive compensation is reported by official media. Based on the analysis above, we propose Hypotheses 1 and 2 as follows. H1: Ceteris paribus, negative media coverage lowers the executive compensation growth rate. H2: Ceteris paribus, official media coverage lowers the executive compensation growth rate. 2.2. The interaction between the governance role of media on executive compensation and firm characteristics Corporate ownership may exert an influence on the governance role of media on execu- tive compensation for at least two reasons: first, the executive incentive mechanisms in state-owned enterprises (SOEs) are different from those in non-state-owned companies. State-owned listed companies not only pursue profit but also assume corresponding social responsibilities (Boycko, Shleifer, & Vishny, 1996; Lin & Liu, 2001). The execu- tives in state-owned listed companies are generally appointed by administration. Yang and Xu (2004) argue that under the current personnel appointment system, the society puts more emphasis on the political reputation of SOE executives rather than their oper- ating capability. Xin, Lin, and Wang (2007) argue that under government intervention and compensation regulation, executive compensation in state-owned companies resembles the characteristic of civil servant pay. Fang (2009) finds that the executive compensation in Chinese state-owned listed companies is very sticky, namely, the compensation is easy to increase, but hard to decrease. Thus, the governance role of media on executive compensation is limited owing to executive regulation and compensation stickiness. However, state ownership may affect the information content of media. Djankov, McLiesh, Nenova, and Shleifer (2010) argue that government control over the media is stronger in less developed countries, where the state-owned economy dominates. Due to the inefficiency of the government information disclosure mechanism, Chinese government officials tend to block the media, for fear that the media would damage the image of the government (Hong & Chai, 2009). Thus, since the coverage on state-owned corporations may be influenced by government attitude, we expect that the governance role of media on state-owned executive compensation would be weakened. From the perspective of monopoly, the monitoring role of media on executive compensation in monopoly corporations is likely to be limited. On one hand, Gao and Du (2010) argue that monopoly corporations with monopoly advantage can achieve better performance than competitive corporations, holding their executive capability, executive endeavor, and firm size identical. Thus, once the media questions the executive compensation in corporate monopolies, executives can resist compensation reduction on the grounds of their better corporate performance. On the other hand, monopolies may have more rigid manager selection systems owing to the lack of effective market competition. As argued by Fama (1980), external market competition and CEO market competition, which can ensure the efficiency of separation of owner- ship and management, are ineffective in monopolistic enterprises. The media coverage on a monopoly firm could also be inaccurate due to the low information transparency 142 Luo et al. of monopolistic corporation. Thus, we expect that the information content of the media coverage on a monopolistic company is lower. Based on the analysis above, we propose Hypotheses 3 and 4 below. H3: Ceteris paribus, compared with SOEs, the role of media in constraining executive compensation is more significant in non-SOEs, and the market reaction to media coverage is stronger in non-SOEs. H4: Ceteris paribus, compared with monopolistic corporations, the role of media in con- straining executive compensation is more significant for non-monopoly corpora- tions, and the market reaction to media coverage is stronger in non-monopoly corporations. 3. Sample selection and research design 3.1. Sample selection 3.1.1. Media selection The media are a platform for information dissemination, and major forms of media nowadays include newspapers, radio, television, Internet, magazines and mobile phone networks. Considering the relatively narrow coverage of radio and magazines and the difficulty of data collection of phone networks, we chose newspapers to be the source of media coverage (Jordan, 2010; Li & Shen, 2010). Zhang (2003) argues that negative reports focus on the events that are in conflict with the existing social order and moral standards. Djankov et al. (2010) divide the media using the ultimate controller theory proposed by La Porta, Lopez-de-Silanes, and Shleifer (1999). If the controller of a certain media is unique, then the media and the controller have the same identity. If the media have multiple controllers, then we deter- mine the identity based on each controller’s identity and their percentage ownership. Applying this principle, we divide the media into negative and other reports, official and non-official reports. 3.1.2. Sample selection From the media coverage on excessive executive compensation, we believe that media coverage would influence executive reputation, and the influence may be exerted through social norms (shaping public opinion) and third-party governance (e.g., any executive-compensation-limiting policy issued by the government). This paper only considers the former. The reasons are as follows: first, the executive compensation lim- iting policy only targets state-owned companies, and is limited in scope; second, Chen, Chen, and Wan (2005) and Shen and Li (2010) have already examined the effect of compensation limiting policy. Thus, our sample period is from January 12, 2007 to August 13, 2009, avoiding the confounding effect of executive compensation limiting policy. Media data are taken from the China National Knowledge Infrastructure (CNKI) database, and specifically, the China Core Newspapers Full-text database. Table 1 shows the screening process of media data. The remaining data come from the China Stock Market & Accounting Research (CSMAR) database. In order to reduce the probability of data omission, we retrieve the media coverage by both ‘topic’ and ‘keyword’. The specific retrieving processes are as follows: (1) we first use ‘topic’ as a search term and ‘compensation’ as a search word, generating 1015 China Journal of Accounting Studies 143 Table 1. Research sample screening process. Screening process 2007 2008 2009 Total 1. Media reports number by search 452 751 1485 2688 Including: by topic 267 307 441 1015 by keywords 185 444 1044 1673 2. Eliminated number of irrelevant media reports 411 699 1313 2423 Including: irrelevant to listed companies or executive compensation 307 361 843 1511 Relevant to listed companies but irrelevant to executive compensation 66 192 172 430 Irrelevant to listed companies but relevant to executive compensation 38 146 298 482 3. Number of relevant media reports 41 52 172 265 Number of listed companies involved 13 29 91 133 4. Eliminated number of defective data 2 2 0 4 5. Final number of observations 11 27 91 129 This table shows the screening process of media data. Media data are taken from the China Core Newspapers Full-text database. Our sample period is from 12 January 2007 to 13 August 2009. media coverage entries concerning executive compensation. We then use ‘keyword’ as a search term and ‘compensation’ as a search word, generating another 1673 media coverage entries concerning executive compensation. In total, we obtained 2688 media coverage entries. (2) We eliminate irrelevant media coverage and 265 relevant media coverage entries remained. Of these, we find that certain coverage may concern several listed companies, while several coverage entries may concern the same listed company. In these circumstances, we first list all the coverage in date-company order, amounting to 412 observations. Then, we merge date-company observations with year- firm observations, resulting in 133 year-firm observations. Finally, we eliminate data with missing values and end up with 129 observations as our final sample. To control for the sample selection bias, we use two methods to form the control group (i.e., firms that are not covered by media regarding executive compensation). First, we use the propensity score matching (PSM) method and identify 129 matching observations as the control group. Second, we take all the non-media reported A-share listed companies as the control group. Additionally, based on the PSM method, we use difference-in-difference estimation as a robust check of our results. 3.2. Model construction In order to use the propensity score matching method, we use model (1) to run a probit regression, and the predicted value of the regression is propensity score. For each media-reported observation, we then select a non-media reported firm/year that has the closest propensity score as the control group. PðReport ¼ 1jXÞ¼ a þ a ROA þ a Size þ a Lever þ a MB þ a Dual 0 1 2 3 4 5 P P ð1Þ þa ID þ a HERF þ a Stock þ a State þ b Ind þ c year þ e 6 7 8 9 i i i i To examine the monitoring role of media coverage, negative media coverage, and official media coverage, we construct model (2). DSalary ¼ a þ c LagMedia þ c State þ X b þ e ð2Þ 1 2 To relieve endogeneity, media coverage measures are lagged by one period. To solve sample selection bias, we estimate model (2) using 258 observations identified by 144 Luo et al. PSM, and all the A-share listed companies. There are four sets of proxy variables for media: (1) a dummy variable Report (0/1); continuous variables on the number of reports including (2) Media, (3) Media_negative and Media_other, and (4) Media_official and Media_nonofficial. Control variables include: ROA, Size, Lever, MB, Dual, ID, HERF, Stock, and Ind. Variable definitions are shown in Table 2. To examine the monitoring role of media coverage on listed companies of different firm characteristics, we run model (3). Model (3a) examines whether the impact of media coverage differs between SOEs and non-SOEs. Model (3b) examines whether the impact of media coverage is different in monopolistic companies and non-monopolistic companies. DSalary ¼ a þ c Media þ c Media  State þ c State þ X b þ e ð3aÞ 1 2 3 DSalary ¼ a þ c Media þ c Media  Mono þ c State þ c Mono þ X b þ e ð3bÞ i;t 1 2 3 4 Table 2. Variable definitions. Variables Variable definition ΔSalary Growth rate of executive compensation: ΔSalary =(Salary – Salary )/ i,t i,t i,t–1 Salary where: Salary represents the cash compensation total of the top i,t–1, i,t three executives in listed company i. Report (0/1) Dummy variable, which takes the value 1 when there is media report and 0 for otherwise. Media Number of media reports plus 1, and then take Ln. Media_negative Number of negative media reports plus 1, and then take Ln. Media_other Number of other media reports besides negative reports plus 1, and then take Ln. Media_offical Number of official media reports plus 1, and then take Ln. Media_nonofficial Number of non-official media reports plus 1, and then take Ln. Mono Dummy variable, which takes the value 1 for monopoly firms, and 0 for otherwise. Specifically, according to the China Securities Regulatory Commission (CSRC) industrial standards, we define monopoly industries as including: B-mining, C41-petroleum refining and coking, C65-black metal smelting and rolling, C67-nonferrous metal smelting and rolling, D-electricity, gas and water production and supply. State Dummy variable, which takes the value 1 for state controlled companies, and 0 for otherwise. Size Ln(total assets). ROA Return on assets = net income/ total assets. Lever Debt to assets ratio = total liability/total assets. MB Market book ratio = total assets/market value. Dual Dummy variable, which takes the value 1 if president and CEO are the same person, and 0 for otherwise. ID Number of independent directors/total number of directors. HERF Equity concentration ratio, which equals the proportion of ownership held by the largest shareholder. Stock Proportion of ownership held by executives. Ind Classified by CSRC’s 13 industries published in 2010, 2-digit codes are used in defining manufacturing industries. This table reports variable code, variable definition and computing method. China Journal of Accounting Studies 145 where State is a dummy variable that takes the value one for SOEs and zero otherwise; Mono is a dummy variable with one denoting monopolistic companies; other variables are identical with those in model (2). We use the event study methodology to examine the market reaction to media cov- erage on listed companies of different firm characteristics. We treat a media coverage entry concerning one company on one day as a single event. We also treat several media coverage entries on one company on one day as a single event. Further, we take one media coverage entry concerning different companies as different events. After the above screening process, we obtain 412 date-firm observations. Since the Chinese stock market closes on weekends, we eliminate 46 date-firm observations whose media coverage appeared on the weekends. We further eliminate 125 date-firm observations with missing values on stock return, and the remaining sample consists of 287 date-firm observations. Specifically, SOEs have 124 observations, non-SOEs have 163 observations; monopolistic companies have 25 observations, and non-monopolistic companies have 262 observations. The event window is (2, 2). The occurrence of the event (media reporting day) is defined as day 0. The estimation period is defined as (230, 30), amounting to 200 days. We use the market model to estimate normal returns, and the processes are as follows. (1) We use data in the estimation period to calculate regression coefficients: R ¼ a þ b R þ e it i mt it where R represents the return of stock i on date t; R represents the market it mt return on date t. (2) We use data in the event window to estimate the expected return of stock i: R ¼ a^ þ b R it i mt The average abnormal return would be the difference between actual returns and expected returns for sample firms: AR ¼ ðR  R Þ it it it The cumulative abnormal return would be: CARð2; 2Þ¼ AR it t¼2 According to our hypotheses, we have the following expectations. First, the cumulative abnormal return is positive in the event window period. The reason is that media coverage can help constrain excessive executive compensation, which is beneficial for shareholders. Second, the market reaction to media coverage of non-SOEs is stronger because we argue that the media coverage has more information content for non-SOEs. 146 Luo et al. Third, the market reaction to media coverage on monopolistic companies is stronger as the media coverage has more information content for monopolistic companies. We use the number of media coverage entries on executive compensation as the proxy variable for media coverage. In accordance with Aharony, Lee, and Wong (2000) and Xin and Tan (2009), we divide the sample into two subsamples: companies in com- petitive industries and companies in monopolistic industries. Specifically, monopolistic industries include petrochemical, energy and materials, and the rest belong to competi- tive industries. As in La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999), we divide the sample into two subsamples, SOEs and non-SOEs. Detailed variable definitions are shown in Table 2. 4. Empirical results 4.1. Descriptive statistics and univariate analysis Table 3 reports the descriptive statistics for all the variables used. For media variables, we only consider the firms with media coverage, because the media variables equal 0 for the control sample. For other variables, we report statistics based on the full sample. Table 3 shows that the average media coverage for a certain company in three years is 4.419, among which the mean official media coverage and non-official media coverage are 3.349 and 1.070 respectively, and the mean negative media coverage and neutral/ positive media coverage are 1.969 and 2.450 respectively. We also compare ΔSalary between the media reported firms and the control sample, and the results are shown in Table 4. The mean value for ΔSalary is 0.203 in reported firms and 0.414 in matching firms, and the difference is –0.211, which is significant at the 5% level. The median of ΔSalary in the reported firm sample is 0.109 and 0.095 in the control sample, and the difference is 0.014, which is not significant. The results Table 3. Descriptive statistics of variables. Variable N Mean Median Std Min Max Media 129 4.419 2 6.660 1 48 Media_negative 129 1.969 1 3.907 0 31 Media_other 129 2.450 1 4.255 0 22 Media_offical 129 3.349 1 5.562 0 36 Media_nonofficial 129 1.070 0 2.028 0 12 ΔSalary 258 0.309 0.095 0.746 0.616 3.571 State 258 0.422 0 0.495 0 1 ROA 258 0.038 0.027 0.060 0.357 0.254 Size 258 23.80 23.58 2.290 18.19 26.82 Lever 258 0.674 0.640 0.305 0.062 2.730 MB 258 0.707 0.785 0.290 0.072 1.239 Dual 258 0.0891 0 0.286 0 1 ID 258 0.359 0.353 0.053 0.125 0.571 Stock 258 0.013 0.000 0.062 0 0.511 HERF 258 0.346 0.217 0.328 0.006 0.758 Mono 258 0.155 0 0.363 0 1 This table reports descriptive statistics about the variables used in our analysis. Except for media coverage, which was drawn from the CNKI database, other data were obtained from the CSMAR database. The sample covers the period 2007–2009. Media variables are only for the firms with media coverage, because the media variables equal 0 for the control sample. For other variables, we report statistics based on the full sample. China Journal of Accounting Studies 147 Table 4. Univariate analysis for ΔSalary. Reported firms Non-reported firms (N = 129) (N = 129) Wilcoxon Variable Mean Median Mean Median t test median test ⁄⁄ ΔSalary 0.203 0.109 0.414 0.095 0.211 (2.29) 0.014 (0.75) This table reports the univariable analyses of ΔSalary. ΔSalary =(Salary – Salary )/Salary , where: Sal- i,t i,t i,t-1 i,t–1 ary represents the cash compensation total of the top three executives in listed company i. Out of media cov- i,t erage on executive compensation, the sub-sample of reported firms include 129 firm-year observations from ⁄⁄⁄ ⁄⁄ 2007–2009. Non-reported firms make up the control sample matched by PSM. significant at 1% level, significant at 5% level, significant at 10% level (two tailed). indicate that, on average, media coverage helps lower executive compensation growth in reported firms, verifying the monitoring role of media on executive compensation. 4.2. Multivariate regression analysis 4.2.1. Regression results on the determinants of media coverage Table 5 shows the probit regression results for model (1) that is used in propensity score matching. The results suggest that the media tend to report executive compensa- tion in large and low growth corporations. Table 5. Determinants of media coverage of executive compensation. Report (0/1) ROA 0.918 (1.05) ⁄⁄⁄ Size 0.279 (6.04) Lever 0.169 (0.91) ⁄⁄ MB 0.601 (2.21) Dual 0.004 (0.03) ID 0.818 (0.96) HERF 0.055 (0.30) Stock 0.399 (0.93) State 0.057 (0.57) Constant 11.50 (0.13) Industry and Year dummies YES Pseudo R 0.227 N 4714 This table provides the results for the Probit model of determinants of media coverage of executive compensa- tion. If a company is covered by media, the dependent variable equals 1, and 0 otherwise. The regression sam- ple comprises 4714 firm-year observations of A-stock firms. The predicted value of the regression is ⁄⁄⁄ ⁄⁄ ⁄ propensity score. T-value is in brackets; significant at 1% level, significant at 5% level, significant at 10% level (two tailed). 148 Luo et al. 4.2.2. The effect of media report on salary growth Table 6 reports the regression results for model (2). Panel A reports the results using the control sample identified by the PSM, and the sample consists of 258 observations. Panel B reports the results using all non-reported A-share listed companies as an alter- native control group, and the sample consists of 4714 observations from 2007–2009. Comparing panel A and panel B, we find that the results are stronger in panel B for both the coefficients and significance levels. To be prudent, the following analysis is mainly based on panel A. In column 1, ΔSalary is negatively associated with Report (0/1), suggesting that the executive compensation growth is lower in reported firms. In column 2, ΔSalary is neg- atively associated with Media, suggesting that media coverage exerts a monitoring role on executive compensation. In column 3, ΔSalary is negatively associated with Media_negative, but its negative correlation with Media_other is insignificant, suggest- ing that negative media coverage is likely more effective in curbing executive compen- sation. Thus, Hypothesis 1 obtains some support. In column 4, ΔSalary is negatively associated with Media_official, but its negative correlation with Media_nonofficial is insignificant, suggesting that official media coverage is more effective in regulating executive compensation. Thus, Hypothesis 2 is verified. 4.2.3. Regression results from the interaction analysis The regression results for model (2) are not sensitive to the media variable being lagged by one period or not, suggesting that endogenity is not a serious problem. Thus, we do not use a lagged media variable in model (3). Table 7 shows the regression results for model (3). The sample consists of 252 observations from the PSM procedure, with 126 matching observations. In column 1, ΔSalary is negatively associated with Media, and positively associated with Media  State. Column 2 has similar results. The results above suggest that, compared with SOEs, the media coverage on non-SOEs can more effectively lower executive compensation growth. Thus, Hypothesis 3 is verified. In col- umn 3, ΔSalary is negatively associated with Media, but its correlation with Media Mono is insignificant. Column 4 has similar results. Thus, Hypothesis 4 is not supported. 4.3. Results from the event study Figure 1 reports the cumulative abnormal return (CAR) of media coverage on executive compensation in the event window (5, 5). In general, CAR increased over the event window. The results suggest that the market views media coverage on executive com- pensation positive, possibly because excessive executive compensation will be curbed. In addition, we test the CAR using the window event (2, 2), and the results are shown in Table 8. In Table 8, the CAR is significantly positive, which supports the monitoring role of media coverage on executive compensation. We further test the difference in CAR between corporations of different firm characteristics, and find that the mean value of CAR (2, 2) in non-SOEs is 0.074, and in SOEs is 0.034: the differ- ence is 0.04 (t = 1.53), which is significant at the 10% level in a one-tailed test. There- fore, there is weak evidence that the monitoring role of media is stronger in non-SOEs than in SOEs. China Journal of Accounting Studies 149 Table 6. The effect of media coverage on executive compensation. Panel A Panel B (1) (2) (3) (4) (5) (6) (7) (8) Variables ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ⁄⁄⁄ ⁄⁄⁄ Report 0.239 0.357 (2.61) (2.88) ⁄⁄ ⁄⁄⁄ Media 0.128 0.353 (2.29) (4.10) ⁄ ⁄⁄⁄ Media_negative 0.129 0.432 (1.65) (3.44) Media_other 0.064 0.138 (0.84) (1.08) ⁄ ⁄⁄ Media_offical 0.182 0.456 (1.72) (2.53) Media_nonofficial 0.047 0.195 (0.66) (1.73) State 0.110 0.117 0.125 0.129 0.010 0.006 0.005 0.005 (1.02) (1.07) (1.15) (1.18) (0.24) (0.16) (0.12) (0.13) ⁄⁄ ⁄ ⁄ ⁄ ROA 2.045 1.931 1.899 1.874 0.0268 0.026 0.026 0.026 (2.00) (1.88) (1.85) (1.82) (1.33) (1.30) (1.28) (1.28) Size 0.006 0.021 0.019 0.025 0.028 0.024 0.022 0.024 (0.12) (0.40) (0.35) (0.48) (1.62) (1.36) (1.27) (1.35) Lever 0.140 0.106 0.107 0.099 0.007 0.008 0.008 0.008 (0.73) (0.55) (0.56) (0.51) (0.89) (0.93) (0.93) (0.92) MB 0.012 0.043 0.023 0.062 0.128 0.122 0.123 0.122 (0.04) (0.14) (0.07) (0.20) (1.64) (1.57) (1.58) (1.57) ⁄⁄ ⁄⁄ ⁄⁄ ⁄ Dual 0.396 0.391 0.386 0.369 0.083 0.083 0.083 0.082 (2.12) (2.09) (2.06) (1.96) (1.54) (1.54) (1.53) (1.52) ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ID 2.187 2.158 2.139 2.071 0.760 0.753 0.743 0.750 (2.31) (2.27) (2.24) (2.17) (2.12) (2.10) (2.07) (2.09) ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ HERF 0.211 0.235 0.251 0.258 0.108 0.106 0.103 0.115 (1.29) (1.42) (1.51) (1.55) (2.41) (2.37) (2.31) (2.55) ⁄ ⁄ ⁄ ⁄ Stock 1.479 1.433 1.423 1.442 0.025 0.026 0.027 0.026 (Continued) 150 Luo et al. Table 6. (Continued). Panel A Panel B (1) (2) (3) (4) (5) (6) (7) (8) Variables ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary (1.76) (1.70) (1.69) (1.71) (0.31) (0.32) (0.34) (0.32) Constant 0.262 0.651 0.625 0.732 0.688 0.598 0.567 0.597 (0.25) (0.61) (0.58) (0.68) (1.76) (1.52) (1.44) (1.52) Industry dummies YES YES YES YES YES YES YES YES R-squared 0.167 0.161 0.161 0.163 0.023 0.024 0.025 0.025 N 258 258 258 258 4714 4714 4714 4714 This table reports the OLS regression results of the effect of media coverage on executive compensation growth. The dependent variable is ΔSalary, ΔSalary =(Salary – Sal- i,t i,t ary )/ Salary , where: Salary represents the cash compensation total of the top three executives in listed company i. We divide the regression results into panel A and panel i,t–1 i,t–1 i,t B: Panel A includes columns (1) to (4), which present the results of 129 coverage sample and control sample matched by PSM; Panel B includes columns (5) to (8), which pre- sents the results of 4714 firm-year observations. Independent variables measuring media coverage are defined as the following four groups: (1) a dummy variable Report (0/1); continuous variables on the number of reports including (2) Media; (3) Media_negative and Media_other; (4) Media_official and Media_nonofficial. Control variables include: ⁄⁄⁄ ⁄⁄ ⁄ ROA, Size, Lever, MB, Dual, ID, HERF, Stock, and Ind. Variable definitions are shown in Table 2. T value is in brackets; significant at 1% level, significant at 5% level, significant at 10% level (two tailed). China Journal of Accounting Studies 151 Table 7. The effect of media coverage: interaction analysis. (1) (2) (3) (4) Variables ΔSalary ΔSalary ΔSalary ΔSalary ⁄⁄⁄ ⁄⁄ Media 0.229 0.132 (3.02) (2.35) MediaState 0.186 (1.77) MediaMono 0.560 (1.04) ⁄⁄⁄ ⁄⁄⁄ Report 0.509 0.300 (4.08) (3.06) ⁄⁄ ReportState 0.436 (2.42) ReportMono 0.303 (0.70) ⁄⁄ State 0.188 0.283 0.076 0.064 (1.57) (2.12) (0.77) (0.65) ROA 0.075 0.121 0.048 0.082 (0.20) (0.33) (0.13) (0.22) ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ Size 0.099 0.106 0.096 0.093 (2.41) (2.59) (2.32) (2.27) Lever 0.043 0.028 0.042 0.035 (0.63) (0.42) (0.60) (0.52) ⁄⁄ ⁄⁄⁄ ⁄⁄ MB 0.509 0.545 0.517 (2.46) (2.43) (2.63) (2.43) ⁄ ⁄ ⁄ Dual 0.211 0.194 0.196 0.185 (1.87) (1.75) (1.71) (1.64) ⁄ ⁄⁄ ⁄ ⁄ ID 2.066 2.185 1.902 1.938 (1.85) (1.99) (1.70) (1.75) HERF 0.107 0.021 0.160 (0.43) (0.38) (0.08) (0.57) ⁄ ⁄ ⁄ Stock 1.349 1.239 1.543 1.566 (1.70) (1.59) (1.87) (1.92) Mono 0.124 0.206 (0.37) (0.62) ⁄ ⁄⁄ ⁄ ⁄ Constant 1.899 2.067 1.770 1.729 (1.91) (2.09) (1.77) (1.75) Industry dummies YES YES YES YES R-squared 0.186 0.212 0.179 0.194 N 252 252 252 252 This table reports the OLS regression results of the effect of media coverage in light of property right and monopolistic characteristic. The dependent variable is ΔSalary. ΔSalary =(Salary – Salary )/Salary i,t i,t i,t–1 i,t–1, where: Salary represents the cash compensation total of the top three executives in listed company i. Media i,t is the natural logarithm of coverage number plus one; Report is a dummy variable that equals 1 if a firm has been reported, and 0 otherwise. State is a dummy variable of property right, which equals 1 if a firm is state- owned, and 0 otherwise. Mono is also a dummy variable that equals 1 if a firm is in a monopoly industry, and 0 otherwise. By the PSM approach, there remain 126 firm-year observations and 126 matched observa- ⁄⁄⁄ ⁄⁄ ⁄ tions. T-value is in brackets; significant at 1% level, significant at 5% level, significant at 10% level (two tailed). The CAR (2, 2) in non-monopoly firms is 0.062 and is significant at 1%; in contrast, the CAR (2, 2) in monopoly firms is not significantly different from zero. The results indicate that the governance role of the media is stronger in non-monopolistic companies than in monopolistic companies. Overall, we find some support for H3 and H4. 152 Luo et al. 0.02 0.015 0.01 0.005 -5 -4 -3 -2 -1 0 1 2 3 4 5 day Figure 1. Market reaction to media coverage on executive compensation. This figure reports the cumulative abnormal return (CAR) of media coverage on executive compensation in the event window (5, 5). Table 8. CAR of media coverage in window (2, 2). Sample Number CAR(2,2) ⁄⁄⁄ Full sample 287 0.057 (4.37) ⁄⁄⁄ State-owned firms 124 0.034 (3.69) ⁄⁄⁄ Non-state-owned firms 163 0.074 (3.42) Monopoly firms 25 0.001 (0.09) ⁄⁄⁄ Non-monopoly firms 262 0.062 (4.40) This table reports the result of CAR of media coverage in window (2, 2). Excluding the insufficient data on stock returns and removing events happening on weekends, 287 firm-date observations remain. With such a sample, we conduct tests according to property right and monopolistic characteristic. T-value is in brackets; ⁄⁄⁄ ⁄⁄ ⁄ significant at 1% level, significant at 5% level, significant at 10% level (two tailed). 5. Robustness checks To examine the effect of media coverage on executive compensation, we also use the difference-in-difference (DID) method on the PSM sample to test the robustness of the results. The DID model is as follows: Salary ¼ a þ a Treat þ a Post þ a Treat  Post þ X b þ e ð4Þ 0 1 2 3 If a company is reported by media, Treat equals 1 and 0 otherwise. Post equals 1 for years in or after the media coverage year and 0 otherwise. Control variables include ROA, Size, Lever, MB, Dual, HERF, Stock, ID and State. We use the PSM method to identify a matching sample: in time matching, we use the first media coverage year as the cutoff point, and search for the reported firms before and after the media coverage year. The matching sample is drawn from the same period as the reported sample. After matching, we end up with 612 firm-year observations. Table 9 reports the DID regression results. The dependent variable is the natural logarithm of executive compensation. The results show that Salary is negatively corre- lated with Treat x Post, which shows that media coverage has a limiting effect on exec- utive compensation growth, which is robust to alternative estimation. CAR China Journal of Accounting Studies 153 Table 9. Robustness test: Regression results from the DID estimation. Variables Salary ⁄⁄⁄ Treat 0.373 (4.10) Post 0.183 (1.93) ⁄⁄⁄ TreatPost 0.443 (3.60) ⁄⁄⁄ State 0.322 (4.16) ⁄⁄⁄ ROA 2.686 (5.10) ⁄⁄⁄ Size 0.243 (10.56) Lever 0.143 (0.63) MB 0.211 (1.20) Dual 0.176 (1.48) ID 0.547 (0.83) HERF 0.189 (0.81) Stock 0.512 (0.89) ⁄⁄⁄ Constant 6.406 (6.89) Industry dummies YES R-squared 0.629 N 612 This table reports the results of robustness check by means of DID estimation. The independent variable is Salary. Salary is the natural logarithm of executive compensation. If a company is reported by media, Treat equals 1, and 0 otherwise. Post equals 1 for years in or after the media coverage year and 0 otherwise. T-value ⁄⁄⁄ ⁄⁄ ⁄ is in brackets; significant at 1% level, significant at 5% level, significant at 10% level (two tailed). 6. Conclusions and limitations We study whether media coverage constrains executive compensation in China, and we find that media coverage has a governance role on executive compensation via increas- ing reputation risk, and negative and official reports are more effective in constraining executive compensation. Meanwhile, the effect of media governance is conditional on some firm characteristics. For example, media coverage works more effectively in non- SOEs and results in more positive market reactions. Media reports result in a stronger market reaction to non-monopoly companies. One limitation of our paper is that we only focus on cash compensation but ignore stock ownership and stock option in our analysis, although they are not a major compo- nent of executive compensation in China. Acknowledgements We would like to thank the support from the National Social Science Foundation project (10XJL0012) ‘Research on Executive Compensation Mechanism in Monopoly Enterprises: With Compensation Cap Regulation’, the National Natural Science Foundation project (70802050) 154 Luo et al. ‘Research on Effectiveness of Independent Director Mechanism: Appointment, Reputation and Economic Consequences’, and Southwestern University of Finance and Economics (SWUFE) university project (2010XG076). This paper has benefited from comments by Zhi Jin, Xuesong Tang, Hongtao Tan, Li Ji, Danlu Bu and Xue Wang from SWUFE, Kangtao Ye and Xinjiao Guan from Renmin University of China, and the China Accounting Academic Leading Talent workshop. We thank the editors (Hong Zou and Jason Xiao) and reviewers for their hard work. Notes 1. Greg Farrell, ‘US Bank Chiefs Face $500,000 Limit’, Financial Times, 5 February 2009 (available at http://www.ft.com/cms/s/0/b06b7572-f2ed-11dd-abe6-0000779fd2ac. html#axzz2RknQttRK). Davis Polk and Wardell, “New Executive Compensation Restrictions under the Emergency Economic Stabilization Act of 2008’, 6 February 2009 (available at http://www.dpw.com/1485409/clientmemos/2009/02.05.09.ec.pdf). 2. Irrelevant media reports refer to indirect compensation reports, which focus on industrial compensation, average staff salaries, executive compensation policies, compensation systems or overseas compensation introduction. 3. When calculating ‘Salary’, we use the cash compensation total of the top three executives in listed companies, and exclude stock options and executive shares. The reason is that the stock incentive plan started late in China with an average lower level, and 0 shareholdings are quite common (Jiang & Huang, 2011). Lv and Zhao (2008) show that executive ownership as well as stock options do not seem to have significant incentive effects. We therefore expect that the omission of executive ownership and stock options should not result in a significant bias. References Aharony, J., Lee, C., & Wong, T. (2000). Financial packaging of IPO firms in China. Journal of Accounting Research, 38, 103–126. Besley, T., & Prat, A. (2006). Handcuffs for the grabbing hand? Media capture and government accountability. American Economic Review, 96, 720–736. Boycko, M., Shleifer, A., & Vishny, R. A. (1996). Theory of privatization. The Economic Jour- nal, 106, 309–319. Chen, Donghua, Chen, Xinyuan, & Wan, Hualin. (2005). Regulation and non-pecuniary compen- sation in Chinese SOEs. Economic Research Journal, 2,92–101. Chen, Yan, Li, Yanshu, & Li, Yong. (2005). Reputation mechanism and the deficiency of financial creditability. China Industrial Economy, 8,73–80. Cheng, S., & Indjejikian, R. (2009). The Market for corporate control and CEO compensation: Complements or substitutes? Contemporary Accounting Research, 26, 933–968. Core, J. E., Guaya, W., & Larckerb, D. F. (2008). The power of the pen and executive compensa- tion. Journal of Financial Economics, 88,1–25. Djankov, S., McLiesh, C., Nenova, T., & Shleifer, A. (2010). Who owns the media? NBER work- ing Paper No. 8288. Dyck, A., Volchkova, N., & Zingales, L. (2004). Private benefits of control: An international comparison. The Journal of Finance, 59, 537–600. Dyck, A., Volchkova, N., & Zingales, L. (2008). The corporate governance role of the media: Evidence from Russia. The Journal of Finance, 63, 1093–1135. Dyck, A., & Zingales, L. (2002). The corporate governance role of the media. NBER Working Paper no. 9309. Faccio, M., Masulis, R. W., & McConnell, J. J. (2006). Political connections and corporate bail- outs. Journal of Finance, 61, 2597–2635. Fama, E. F. (1980). Agency problems and the theory of the firm. Journal of Political Economy, 88, 288–307. Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. Journal of Law and Economics, 26, 301–325. Fang, Junxiong. (2009). Is there executive compensation stickiness in Chinese listed companies? Economic Research Journal, 3,110–124. China Journal of Accounting Studies 155 Fich, E. M., & Shivdasani, A. (2007). Financial fraud, director reputation, and shareholder wealth. Journal of Financial Economics, 86, 306–336. Fisman, R. (2001). Estimating the value of political connections. The American Economic Review, 91, 1095–1102. Fudenberg, D., & Levine, D. K. (1992). Maintaining a reputation when strategies are imperfectly observed. The Review of Economic Studies, 59, 561–579. Gao, Minghua, & Du, Wencui. (2010). The wage for manager in monopoly enterprise: Shortage or overage? Academia Bimestris, 3, 162–168. Gilson, S. C. (1990). Bankruptcy, boards, banks, and blockholders evidence on changes in corpo- rate ownership and control when firms default. Journal of Financial Economics, 27, 355–387. Harford, J. (2003). Takeover bids and target directors’ incentives: The impact of a bid on direc- tors’ wealth and board seat. Journal of Financial Economics, 69,51–83. Hong, Li., & Chai, Chunying. (2009). On the study of interactive relationship of media, the pub- lic and government in the formation of public opinion. Press Circles, 2,74–75. Jiang, Fuxiu., & Huang, Jicheng. (2011). Managerial incentive, debt and firm value. Economic Research Journal, 5,46–60. Joe, J. R., Louis, H., & Robinson, D. (2009). Managers’ and investors’ responses to media expo- sure of board ineffectiveness. Journal of Financial and Quantitative Analysis, 44, 579–605. Jordan, Otten. (2010). Media attention and executive pay in the Netherlands. Erasmus Institute Monitoring & Compliance, 4,1–45. Kaplan, S. N., & Reishus, D. (1990). Outside directorships and corporate performance. Journal of Financial Economics, 27, 389–410. Kreps, D. M., & Wilson, R. (1982). Reputation and imperfect information. Journal of Economic Theory, 27, 253–279. Kuhnen, C. M., & Niessen-Ruenzi, A. (2008). Is executive compensation shaped by public atti- tudes? CFR Working Paper no. 08–09. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. (1999). The quality of government. Journal of Law, Economics, and Organization, 15, 222–279. Li, Peigong, & Shen, Yifeng. (2010). The corporate governance role of media: Empirical evidence from China. Economic Research Journal, 4,14–27. Lin, Yifu, & Liu, Peilin. (2001). Autogenic ability and SOE reform. Economic Research Journal, 9,60–70. Lv, Changjiang, & Zhao, Yuheng. (2008). A study on the effect of the incentive given to manag- ers of state-owned enterprises. Management World, 11, 188. Miller, G. S. (2006). The press as a watchdog for accounting fraud. Journal of Accounting Research, 44, 1001–1033. Monks, A. G., & Minow, N. (2004). Corporate governance (3rd ed.). Malden, MA: Blackwell. Paligorova, T. (2009). Agency conflicts in the process of securitization. Bank of Canada Review, 9,33–48. Scott, G., & Konstantin, S. (2009). Government control of the media. CEFIR /NES Working Paper no. 126. Shen, Yifeng., & Li, Peigong. (2010). Empirical research on the impact of ‘Pay Ceiling Order’ on the CEO compensation and Pay-for-Luck phenomenon. China Industrial Economics, 11, 130–139. Shleifer, A., & Vishny, R. W. (1994). Politicians and firms. The Quarterly Journal of Economics, 109, 995–1025. Srinivasan, S. (2005). Consequences of financial reporting failure for outside directors: Evidence from accounting restatements and audit committee members. Journal of Accounting Research, 43, 291–334. Weigelt, K., & Camerer, C. (1988). Reputation and corporate strategy: A review of recent theory and applications. Strategic Management Journal, 9, 443–454. Wilson, R. (1985). Reputations in games and markets. In game-theoretic models of bargain. New York, NY: Cambridge University Press. Xin, Qingquan., Lin, Bin, & Wang, Yanchao. (2007). Government control: Executive compensa- tion and capital investment. Economic Research Journal, 8,110–122. Xin, Qingquan, & Tan, Weiqiang. (2009). Marketization reform: Corporate performance and SOE executive compensation. Economic Research Journal, 11,68–81. 156 Luo et al. Xu, Haoping, & Luo, Wei. (2007). Reputation effect of investment banks: Research from aspects of market share and service quality. Economic Research Journal, 2, 124–136. Yang, Yada, & Xu, Hong. (2004). Manager reputation motivation mechanism in SOEs. Economic Theory and Business Management, 4,47–49. Yao, Renan. (2008). Value the positive effect on negative reports. Information of Economics and Law, 6,12–13. Yermack, D. (2004). Remuneration, retention, and reputation incentives for outside directors. Journal of Finance, 59, 2281–2308. Zhang, Wei (2003). Comparative journalism: Methodology and textual research. Guangzhou: Southern Daily Press. Zhang, Weiying. (2002). Reputation foundation of legal system. Economic Research Journal, 1,3–13. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png China Journal of Accounting Studies Taylor & Francis

The monitoring role of media on executive compensation

Download PDF
19 pages

Loading next page...
 
/lp/taylor-francis/the-monitoring-role-of-media-on-executive-compensation-DJ4eJqczLY

References (57)

Publisher
Taylor & Francis
Copyright
Copyright Accounting Society of China
ISSN
2169-7221
eISSN
2169-7213
DOI
10.1080/21697221.2013.802974
Publisher site
See Article on Publisher Site

Abstract

China Journal of Accounting Studies, 2013 Vol. 1, No. 2, 138–156, http://dx.doi.org/10.1080/21697221.2013.802974 a, a b Hong Luo *, Baohua Liu and Weiqian Zhang a b Accounting School, Southwestern University of Finance and Economics, China; Finance School, Shanghai Institute of Foreign Trade, China This paper investigates the monitoring role of media on executive compensation. Using data on media coverage from 2007 to 2009, we find that the media can serve as an effective external monitoring mechanism on executive compensation by affect- ing executive reputation. Specifically, negative media coverage and government intervention on media can enhance the monitoring role of media. We also find that the monitoring role of media and the market reaction to media coverage are stronger for non-state-owned corporations. Although the monitoring role of media is not dif- ferent between monopolistic and non-monopolistic industries, the market reaction to media coverage is stronger for companies in non-monopolistic industries. We also find that there is less coverage on monopolistic and state-owned companies. Keywords: reputation mechanism; media coverage; executive compensation 1. Introduction The governance role of the media has attracted great attention from scholars in recent years. Researchers believe that the media influence corporate behaviors through the reputation mechanism. Specifically, media coverage on corporations may attract the attention of investors and may further affect corporate reputation, and the exposed corporations have to react to media. Thus, the media may serve as an effective tool in external monitoring (Dyck & Zingales, 2002; Dyck, Volchkova, & Zingales, 2004, 2008; Joe, Louis, & Robinson, 2009; Miller, 2006; Monks & Minow, 2004). The financial crisis, which started in 2007, has badly hit the world economy. Central governments, such as the US government, offered huge financial aids to the corpora- tions trapped in the crisis. However, the executives in these corporations are still paid huge amounts of compensation, which arouses fierce criticism from the public. In the US, as well as in Britain, France, and Germany, governments immediately issued poli- cies to restrict executive compensation. Because of media coverage and government supervision, American corporations have experienced a 50% reduction in executive compensation in 2008, and a 90% reduction in monetary compensation in 2009. Execu- tive compensation in financial corporations has also been slashed in Britain, France, Germany, and so on. Meanwhile, the Chinese media have started to pay attention to corporate executive compensation. With overwhelming media coverage concerning executive compensation appearing in public, hundreds of corporate executives dominate the headlines. The sky-high compensation in Chinese corporations also attracts fierce *Corresponding author. Email: luohong@swufe.edu.cn Paper accepted by Hong Zou. 2013 Accounting Society of China China Journal of Accounting Studies 139 public criticism. Thus, whether media coverage has an impact on executive compensa- tion in China is an interesting research question. Reputation theory provides a new perspective to study the impact of media on exec- utive compensation. Reputation theory, introduced by Kreps and Wilson (1982) and revised by Fudenberg and Levine (1992), suggests that reputation can strengthen commitment and that the role of reputation is to provide implicit incentives for the long-term interests of participants to ensure their short-term commitment behavior and thus make reputation a substitute for explicit contracts. Weigelt and Camerer (1988) and Wilson (1985) explore the impact of reputation on participant behavior, and such impact could be labeled as ‘reputation effect’ or ‘reputation mechanism’. Fama (1980) finds that the reputation mechanism has an impact on the labor market. Extant literature finds that executives will suffer from reputation losses when firms’ financial frauds are detected (Fich & Shivdasani, 2007; Srinivasan, 2005), and their companies experience financial distress (Gilson, 1990), liquidation (Harford, 2003), poor performance (Yermack, 2004) or reduction of dividends (Kaplan & Reishus, 1990). Some studies also consider reputation mechanisms in China. Zhang (2002) argues that reputation could replace legislation to serve as a low-cost and effective mechanism in maintaining transaction order. Yang and Xu (2004) study the reputation incentive mecha- nism of executives in state-owned enterprises, and find that executives could gain more utility through pursing political reputation than pursuing operating performance. Xu and Luo (2007) test the effectiveness of the reputation mechanism in investment banks through the IPO market, and find that the reputation mechanism proves to be effective due to the desire for a higher market share and better career advancement by investment bankers. Extant literature has acknowledged the governance role of media from the perspec- tive of reputation mechanisms. Specifically, media could lower the private benefits of control rights (Dyck, Volchkova, & Zingales, 2004), uncover accounting irregularities (Miller, 2006), prevent corporations from expropriating the benefits of outsiders (Dyck et al., 2008; Li & Shen, 2010), and improve the operating efficiency of the board (Joe et al., 2009). Core, Guaya, and Larckerb (2008) examine the impact of media on executive com- pensation. Using firms having media coverage on their executive compensation from 1994–2002, they find that negative media coverage is insignificantly correlated with executive compensation reduction, and insignificantly correlated with executive turn- over. They ascribe the invalid role of media to its sensationalism. Whether media govern executive compensation out of sensationalism or social con- science and whether the reputation mechanism makes a difference in executive compen- sation monitoring remain interesting questions. Chen, Li, and Li (2005) argue that the main type of reputation mechanisms is the one that is consciously built, and assured by social norms and third-party governance. From the current media coverage on executive compensation and the corresponding social reactions, we believe that media coverage may have an impact on executive compensation, and such impact may work through social norms, namely, through arousing public opinion. Based on Chinese institutional background, this paper finds that media coverage, especially official and negative media coverage, plays a governance role in affecting executive compensation of Chinese listed companies. We contribute to the literature concerning the governance role of media, and provide a new perspective on the executive incentive mechanism. The rest of paper is organized as follows. Section 2 provides theoretical analysis and hypothesis development. Section 3 discusses our sample and research design. Section 4 reports empirical results. Section 5 provides further robustness tests. Section 6 concludes. 140 Luo et al. 2. Theoretical framework and research hypotheses 2.1. The governance role of media on executive compensation The media coverage on sky-high executive compensation affects the reputation of the listed companies and their executives, and to avoid reputation losses, listed companies should take action to adjust executive compensation. Thus, the media serve as an ex post monitoring tool in executive compensation. According to traditional reputation theory (Fama, 1980; Fama & Jensen, 1983), media coverage on huge compensation payments could affect the reputation of executives, and lower their human capital value. Additionally, politicians believe that failing to act on legal rules and regulations could damage their image and endanger their political career (Besley & Prat, 2006). Media coverage of huge compensation payments can pressure authorities to issue policies to restrict executive compensation. Li and Shen (2010) argue that the monitoring role of the media is achieved by raising the attention and intervention of administrations, and media coverage can raise corporate administrative costs and prevent corporations from damaging the interests of outsiders. Paligorova (2009) and Cheng and Indjejikian (2009) suggest that government legislation could influence CEOs’ compensation prac- tices. Kuhnen and Niessen (2008) demonstrate the monitoring role of the media on executive compensation using a sample of 22,507 examples of American newspaper coverage on executive compensation in the period 1990–2006. They report that execu- tive compensation levels and structures appear to change with public attitude and social norms. The media should represent the opinion of the public, and act as a rational and objective “recorder” of social justice. The negative coverage on certain issues could positively affect the whole of society (Yao, 2008). An article published in an important Italian newspaper cited directors of a large public corporation with the worst perfor- mance in the previous three years, and, as a result, two months later the CEO resigned without explanation (Dyck & Zingales, 2002). Facts prove that the more negative and furious the coverage, the stronger the social response it can bring about. Thus, media coverage on excessive executive compensation could endanger the reputation of the executives, pressuring executives to take further measures to appease public opinion and reduce the reputational risk, and the compensation increment would be under control. Scott and Konstantin (2009) find that government control over media is often stron- ger in less democratic countries. Currently, in China, official media coverage is wider, of higher reliability, and higher authority. Official media coverage often spreads rapidly to the public and results in a larger impact on public opinion. Official media coverage not only represents its voice and opinion, but reflects the will and determination of the government. Additionally, Shleifer and Vishny (1994) proposed the government support hypothesis, suggesting that the political connections that entrepreneurs try to establish with time and money often bring huge benefits (e.g., financing convenience, tax favor, property right protection) to entrepreneurs and companies (Faccio, Masulis, & McCon- nell, 2006; Fisman, 2001). From the perspective of administrative hierarchy, the official media in China often carry corresponding administrative ranks. For instance, the China Central Television (CCTV) is subordinate to the State Administration of Radio Film and Television (SARFT) and it ranks deputy ministerial, with the president of CCTV being a vice-ministerial official. Corporations and their executives are willing to maintain a good relationship with the official media to avoid the exposure of their misconducts, and once their misconducts are reported, corporations would respond quickly to the coverage. Thus, we expect that executives are more likely to make China Journal of Accounting Studies 141 adjustment to compensation once their excessive compensation is reported by official media. Based on the analysis above, we propose Hypotheses 1 and 2 as follows. H1: Ceteris paribus, negative media coverage lowers the executive compensation growth rate. H2: Ceteris paribus, official media coverage lowers the executive compensation growth rate. 2.2. The interaction between the governance role of media on executive compensation and firm characteristics Corporate ownership may exert an influence on the governance role of media on execu- tive compensation for at least two reasons: first, the executive incentive mechanisms in state-owned enterprises (SOEs) are different from those in non-state-owned companies. State-owned listed companies not only pursue profit but also assume corresponding social responsibilities (Boycko, Shleifer, & Vishny, 1996; Lin & Liu, 2001). The execu- tives in state-owned listed companies are generally appointed by administration. Yang and Xu (2004) argue that under the current personnel appointment system, the society puts more emphasis on the political reputation of SOE executives rather than their oper- ating capability. Xin, Lin, and Wang (2007) argue that under government intervention and compensation regulation, executive compensation in state-owned companies resembles the characteristic of civil servant pay. Fang (2009) finds that the executive compensation in Chinese state-owned listed companies is very sticky, namely, the compensation is easy to increase, but hard to decrease. Thus, the governance role of media on executive compensation is limited owing to executive regulation and compensation stickiness. However, state ownership may affect the information content of media. Djankov, McLiesh, Nenova, and Shleifer (2010) argue that government control over the media is stronger in less developed countries, where the state-owned economy dominates. Due to the inefficiency of the government information disclosure mechanism, Chinese government officials tend to block the media, for fear that the media would damage the image of the government (Hong & Chai, 2009). Thus, since the coverage on state-owned corporations may be influenced by government attitude, we expect that the governance role of media on state-owned executive compensation would be weakened. From the perspective of monopoly, the monitoring role of media on executive compensation in monopoly corporations is likely to be limited. On one hand, Gao and Du (2010) argue that monopoly corporations with monopoly advantage can achieve better performance than competitive corporations, holding their executive capability, executive endeavor, and firm size identical. Thus, once the media questions the executive compensation in corporate monopolies, executives can resist compensation reduction on the grounds of their better corporate performance. On the other hand, monopolies may have more rigid manager selection systems owing to the lack of effective market competition. As argued by Fama (1980), external market competition and CEO market competition, which can ensure the efficiency of separation of owner- ship and management, are ineffective in monopolistic enterprises. The media coverage on a monopoly firm could also be inaccurate due to the low information transparency 142 Luo et al. of monopolistic corporation. Thus, we expect that the information content of the media coverage on a monopolistic company is lower. Based on the analysis above, we propose Hypotheses 3 and 4 below. H3: Ceteris paribus, compared with SOEs, the role of media in constraining executive compensation is more significant in non-SOEs, and the market reaction to media coverage is stronger in non-SOEs. H4: Ceteris paribus, compared with monopolistic corporations, the role of media in con- straining executive compensation is more significant for non-monopoly corpora- tions, and the market reaction to media coverage is stronger in non-monopoly corporations. 3. Sample selection and research design 3.1. Sample selection 3.1.1. Media selection The media are a platform for information dissemination, and major forms of media nowadays include newspapers, radio, television, Internet, magazines and mobile phone networks. Considering the relatively narrow coverage of radio and magazines and the difficulty of data collection of phone networks, we chose newspapers to be the source of media coverage (Jordan, 2010; Li & Shen, 2010). Zhang (2003) argues that negative reports focus on the events that are in conflict with the existing social order and moral standards. Djankov et al. (2010) divide the media using the ultimate controller theory proposed by La Porta, Lopez-de-Silanes, and Shleifer (1999). If the controller of a certain media is unique, then the media and the controller have the same identity. If the media have multiple controllers, then we deter- mine the identity based on each controller’s identity and their percentage ownership. Applying this principle, we divide the media into negative and other reports, official and non-official reports. 3.1.2. Sample selection From the media coverage on excessive executive compensation, we believe that media coverage would influence executive reputation, and the influence may be exerted through social norms (shaping public opinion) and third-party governance (e.g., any executive-compensation-limiting policy issued by the government). This paper only considers the former. The reasons are as follows: first, the executive compensation lim- iting policy only targets state-owned companies, and is limited in scope; second, Chen, Chen, and Wan (2005) and Shen and Li (2010) have already examined the effect of compensation limiting policy. Thus, our sample period is from January 12, 2007 to August 13, 2009, avoiding the confounding effect of executive compensation limiting policy. Media data are taken from the China National Knowledge Infrastructure (CNKI) database, and specifically, the China Core Newspapers Full-text database. Table 1 shows the screening process of media data. The remaining data come from the China Stock Market & Accounting Research (CSMAR) database. In order to reduce the probability of data omission, we retrieve the media coverage by both ‘topic’ and ‘keyword’. The specific retrieving processes are as follows: (1) we first use ‘topic’ as a search term and ‘compensation’ as a search word, generating 1015 China Journal of Accounting Studies 143 Table 1. Research sample screening process. Screening process 2007 2008 2009 Total 1. Media reports number by search 452 751 1485 2688 Including: by topic 267 307 441 1015 by keywords 185 444 1044 1673 2. Eliminated number of irrelevant media reports 411 699 1313 2423 Including: irrelevant to listed companies or executive compensation 307 361 843 1511 Relevant to listed companies but irrelevant to executive compensation 66 192 172 430 Irrelevant to listed companies but relevant to executive compensation 38 146 298 482 3. Number of relevant media reports 41 52 172 265 Number of listed companies involved 13 29 91 133 4. Eliminated number of defective data 2 2 0 4 5. Final number of observations 11 27 91 129 This table shows the screening process of media data. Media data are taken from the China Core Newspapers Full-text database. Our sample period is from 12 January 2007 to 13 August 2009. media coverage entries concerning executive compensation. We then use ‘keyword’ as a search term and ‘compensation’ as a search word, generating another 1673 media coverage entries concerning executive compensation. In total, we obtained 2688 media coverage entries. (2) We eliminate irrelevant media coverage and 265 relevant media coverage entries remained. Of these, we find that certain coverage may concern several listed companies, while several coverage entries may concern the same listed company. In these circumstances, we first list all the coverage in date-company order, amounting to 412 observations. Then, we merge date-company observations with year- firm observations, resulting in 133 year-firm observations. Finally, we eliminate data with missing values and end up with 129 observations as our final sample. To control for the sample selection bias, we use two methods to form the control group (i.e., firms that are not covered by media regarding executive compensation). First, we use the propensity score matching (PSM) method and identify 129 matching observations as the control group. Second, we take all the non-media reported A-share listed companies as the control group. Additionally, based on the PSM method, we use difference-in-difference estimation as a robust check of our results. 3.2. Model construction In order to use the propensity score matching method, we use model (1) to run a probit regression, and the predicted value of the regression is propensity score. For each media-reported observation, we then select a non-media reported firm/year that has the closest propensity score as the control group. PðReport ¼ 1jXÞ¼ a þ a ROA þ a Size þ a Lever þ a MB þ a Dual 0 1 2 3 4 5 P P ð1Þ þa ID þ a HERF þ a Stock þ a State þ b Ind þ c year þ e 6 7 8 9 i i i i To examine the monitoring role of media coverage, negative media coverage, and official media coverage, we construct model (2). DSalary ¼ a þ c LagMedia þ c State þ X b þ e ð2Þ 1 2 To relieve endogeneity, media coverage measures are lagged by one period. To solve sample selection bias, we estimate model (2) using 258 observations identified by 144 Luo et al. PSM, and all the A-share listed companies. There are four sets of proxy variables for media: (1) a dummy variable Report (0/1); continuous variables on the number of reports including (2) Media, (3) Media_negative and Media_other, and (4) Media_official and Media_nonofficial. Control variables include: ROA, Size, Lever, MB, Dual, ID, HERF, Stock, and Ind. Variable definitions are shown in Table 2. To examine the monitoring role of media coverage on listed companies of different firm characteristics, we run model (3). Model (3a) examines whether the impact of media coverage differs between SOEs and non-SOEs. Model (3b) examines whether the impact of media coverage is different in monopolistic companies and non-monopolistic companies. DSalary ¼ a þ c Media þ c Media  State þ c State þ X b þ e ð3aÞ 1 2 3 DSalary ¼ a þ c Media þ c Media  Mono þ c State þ c Mono þ X b þ e ð3bÞ i;t 1 2 3 4 Table 2. Variable definitions. Variables Variable definition ΔSalary Growth rate of executive compensation: ΔSalary =(Salary – Salary )/ i,t i,t i,t–1 Salary where: Salary represents the cash compensation total of the top i,t–1, i,t three executives in listed company i. Report (0/1) Dummy variable, which takes the value 1 when there is media report and 0 for otherwise. Media Number of media reports plus 1, and then take Ln. Media_negative Number of negative media reports plus 1, and then take Ln. Media_other Number of other media reports besides negative reports plus 1, and then take Ln. Media_offical Number of official media reports plus 1, and then take Ln. Media_nonofficial Number of non-official media reports plus 1, and then take Ln. Mono Dummy variable, which takes the value 1 for monopoly firms, and 0 for otherwise. Specifically, according to the China Securities Regulatory Commission (CSRC) industrial standards, we define monopoly industries as including: B-mining, C41-petroleum refining and coking, C65-black metal smelting and rolling, C67-nonferrous metal smelting and rolling, D-electricity, gas and water production and supply. State Dummy variable, which takes the value 1 for state controlled companies, and 0 for otherwise. Size Ln(total assets). ROA Return on assets = net income/ total assets. Lever Debt to assets ratio = total liability/total assets. MB Market book ratio = total assets/market value. Dual Dummy variable, which takes the value 1 if president and CEO are the same person, and 0 for otherwise. ID Number of independent directors/total number of directors. HERF Equity concentration ratio, which equals the proportion of ownership held by the largest shareholder. Stock Proportion of ownership held by executives. Ind Classified by CSRC’s 13 industries published in 2010, 2-digit codes are used in defining manufacturing industries. This table reports variable code, variable definition and computing method. China Journal of Accounting Studies 145 where State is a dummy variable that takes the value one for SOEs and zero otherwise; Mono is a dummy variable with one denoting monopolistic companies; other variables are identical with those in model (2). We use the event study methodology to examine the market reaction to media cov- erage on listed companies of different firm characteristics. We treat a media coverage entry concerning one company on one day as a single event. We also treat several media coverage entries on one company on one day as a single event. Further, we take one media coverage entry concerning different companies as different events. After the above screening process, we obtain 412 date-firm observations. Since the Chinese stock market closes on weekends, we eliminate 46 date-firm observations whose media coverage appeared on the weekends. We further eliminate 125 date-firm observations with missing values on stock return, and the remaining sample consists of 287 date-firm observations. Specifically, SOEs have 124 observations, non-SOEs have 163 observations; monopolistic companies have 25 observations, and non-monopolistic companies have 262 observations. The event window is (2, 2). The occurrence of the event (media reporting day) is defined as day 0. The estimation period is defined as (230, 30), amounting to 200 days. We use the market model to estimate normal returns, and the processes are as follows. (1) We use data in the estimation period to calculate regression coefficients: R ¼ a þ b R þ e it i mt it where R represents the return of stock i on date t; R represents the market it mt return on date t. (2) We use data in the event window to estimate the expected return of stock i: R ¼ a^ þ b R it i mt The average abnormal return would be the difference between actual returns and expected returns for sample firms: AR ¼ ðR  R Þ it it it The cumulative abnormal return would be: CARð2; 2Þ¼ AR it t¼2 According to our hypotheses, we have the following expectations. First, the cumulative abnormal return is positive in the event window period. The reason is that media coverage can help constrain excessive executive compensation, which is beneficial for shareholders. Second, the market reaction to media coverage of non-SOEs is stronger because we argue that the media coverage has more information content for non-SOEs. 146 Luo et al. Third, the market reaction to media coverage on monopolistic companies is stronger as the media coverage has more information content for monopolistic companies. We use the number of media coverage entries on executive compensation as the proxy variable for media coverage. In accordance with Aharony, Lee, and Wong (2000) and Xin and Tan (2009), we divide the sample into two subsamples: companies in com- petitive industries and companies in monopolistic industries. Specifically, monopolistic industries include petrochemical, energy and materials, and the rest belong to competi- tive industries. As in La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999), we divide the sample into two subsamples, SOEs and non-SOEs. Detailed variable definitions are shown in Table 2. 4. Empirical results 4.1. Descriptive statistics and univariate analysis Table 3 reports the descriptive statistics for all the variables used. For media variables, we only consider the firms with media coverage, because the media variables equal 0 for the control sample. For other variables, we report statistics based on the full sample. Table 3 shows that the average media coverage for a certain company in three years is 4.419, among which the mean official media coverage and non-official media coverage are 3.349 and 1.070 respectively, and the mean negative media coverage and neutral/ positive media coverage are 1.969 and 2.450 respectively. We also compare ΔSalary between the media reported firms and the control sample, and the results are shown in Table 4. The mean value for ΔSalary is 0.203 in reported firms and 0.414 in matching firms, and the difference is –0.211, which is significant at the 5% level. The median of ΔSalary in the reported firm sample is 0.109 and 0.095 in the control sample, and the difference is 0.014, which is not significant. The results Table 3. Descriptive statistics of variables. Variable N Mean Median Std Min Max Media 129 4.419 2 6.660 1 48 Media_negative 129 1.969 1 3.907 0 31 Media_other 129 2.450 1 4.255 0 22 Media_offical 129 3.349 1 5.562 0 36 Media_nonofficial 129 1.070 0 2.028 0 12 ΔSalary 258 0.309 0.095 0.746 0.616 3.571 State 258 0.422 0 0.495 0 1 ROA 258 0.038 0.027 0.060 0.357 0.254 Size 258 23.80 23.58 2.290 18.19 26.82 Lever 258 0.674 0.640 0.305 0.062 2.730 MB 258 0.707 0.785 0.290 0.072 1.239 Dual 258 0.0891 0 0.286 0 1 ID 258 0.359 0.353 0.053 0.125 0.571 Stock 258 0.013 0.000 0.062 0 0.511 HERF 258 0.346 0.217 0.328 0.006 0.758 Mono 258 0.155 0 0.363 0 1 This table reports descriptive statistics about the variables used in our analysis. Except for media coverage, which was drawn from the CNKI database, other data were obtained from the CSMAR database. The sample covers the period 2007–2009. Media variables are only for the firms with media coverage, because the media variables equal 0 for the control sample. For other variables, we report statistics based on the full sample. China Journal of Accounting Studies 147 Table 4. Univariate analysis for ΔSalary. Reported firms Non-reported firms (N = 129) (N = 129) Wilcoxon Variable Mean Median Mean Median t test median test ⁄⁄ ΔSalary 0.203 0.109 0.414 0.095 0.211 (2.29) 0.014 (0.75) This table reports the univariable analyses of ΔSalary. ΔSalary =(Salary – Salary )/Salary , where: Sal- i,t i,t i,t-1 i,t–1 ary represents the cash compensation total of the top three executives in listed company i. Out of media cov- i,t erage on executive compensation, the sub-sample of reported firms include 129 firm-year observations from ⁄⁄⁄ ⁄⁄ 2007–2009. Non-reported firms make up the control sample matched by PSM. significant at 1% level, significant at 5% level, significant at 10% level (two tailed). indicate that, on average, media coverage helps lower executive compensation growth in reported firms, verifying the monitoring role of media on executive compensation. 4.2. Multivariate regression analysis 4.2.1. Regression results on the determinants of media coverage Table 5 shows the probit regression results for model (1) that is used in propensity score matching. The results suggest that the media tend to report executive compensa- tion in large and low growth corporations. Table 5. Determinants of media coverage of executive compensation. Report (0/1) ROA 0.918 (1.05) ⁄⁄⁄ Size 0.279 (6.04) Lever 0.169 (0.91) ⁄⁄ MB 0.601 (2.21) Dual 0.004 (0.03) ID 0.818 (0.96) HERF 0.055 (0.30) Stock 0.399 (0.93) State 0.057 (0.57) Constant 11.50 (0.13) Industry and Year dummies YES Pseudo R 0.227 N 4714 This table provides the results for the Probit model of determinants of media coverage of executive compensa- tion. If a company is covered by media, the dependent variable equals 1, and 0 otherwise. The regression sam- ple comprises 4714 firm-year observations of A-stock firms. The predicted value of the regression is ⁄⁄⁄ ⁄⁄ ⁄ propensity score. T-value is in brackets; significant at 1% level, significant at 5% level, significant at 10% level (two tailed). 148 Luo et al. 4.2.2. The effect of media report on salary growth Table 6 reports the regression results for model (2). Panel A reports the results using the control sample identified by the PSM, and the sample consists of 258 observations. Panel B reports the results using all non-reported A-share listed companies as an alter- native control group, and the sample consists of 4714 observations from 2007–2009. Comparing panel A and panel B, we find that the results are stronger in panel B for both the coefficients and significance levels. To be prudent, the following analysis is mainly based on panel A. In column 1, ΔSalary is negatively associated with Report (0/1), suggesting that the executive compensation growth is lower in reported firms. In column 2, ΔSalary is neg- atively associated with Media, suggesting that media coverage exerts a monitoring role on executive compensation. In column 3, ΔSalary is negatively associated with Media_negative, but its negative correlation with Media_other is insignificant, suggest- ing that negative media coverage is likely more effective in curbing executive compen- sation. Thus, Hypothesis 1 obtains some support. In column 4, ΔSalary is negatively associated with Media_official, but its negative correlation with Media_nonofficial is insignificant, suggesting that official media coverage is more effective in regulating executive compensation. Thus, Hypothesis 2 is verified. 4.2.3. Regression results from the interaction analysis The regression results for model (2) are not sensitive to the media variable being lagged by one period or not, suggesting that endogenity is not a serious problem. Thus, we do not use a lagged media variable in model (3). Table 7 shows the regression results for model (3). The sample consists of 252 observations from the PSM procedure, with 126 matching observations. In column 1, ΔSalary is negatively associated with Media, and positively associated with Media  State. Column 2 has similar results. The results above suggest that, compared with SOEs, the media coverage on non-SOEs can more effectively lower executive compensation growth. Thus, Hypothesis 3 is verified. In col- umn 3, ΔSalary is negatively associated with Media, but its correlation with Media Mono is insignificant. Column 4 has similar results. Thus, Hypothesis 4 is not supported. 4.3. Results from the event study Figure 1 reports the cumulative abnormal return (CAR) of media coverage on executive compensation in the event window (5, 5). In general, CAR increased over the event window. The results suggest that the market views media coverage on executive com- pensation positive, possibly because excessive executive compensation will be curbed. In addition, we test the CAR using the window event (2, 2), and the results are shown in Table 8. In Table 8, the CAR is significantly positive, which supports the monitoring role of media coverage on executive compensation. We further test the difference in CAR between corporations of different firm characteristics, and find that the mean value of CAR (2, 2) in non-SOEs is 0.074, and in SOEs is 0.034: the differ- ence is 0.04 (t = 1.53), which is significant at the 10% level in a one-tailed test. There- fore, there is weak evidence that the monitoring role of media is stronger in non-SOEs than in SOEs. China Journal of Accounting Studies 149 Table 6. The effect of media coverage on executive compensation. Panel A Panel B (1) (2) (3) (4) (5) (6) (7) (8) Variables ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ⁄⁄⁄ ⁄⁄⁄ Report 0.239 0.357 (2.61) (2.88) ⁄⁄ ⁄⁄⁄ Media 0.128 0.353 (2.29) (4.10) ⁄ ⁄⁄⁄ Media_negative 0.129 0.432 (1.65) (3.44) Media_other 0.064 0.138 (0.84) (1.08) ⁄ ⁄⁄ Media_offical 0.182 0.456 (1.72) (2.53) Media_nonofficial 0.047 0.195 (0.66) (1.73) State 0.110 0.117 0.125 0.129 0.010 0.006 0.005 0.005 (1.02) (1.07) (1.15) (1.18) (0.24) (0.16) (0.12) (0.13) ⁄⁄ ⁄ ⁄ ⁄ ROA 2.045 1.931 1.899 1.874 0.0268 0.026 0.026 0.026 (2.00) (1.88) (1.85) (1.82) (1.33) (1.30) (1.28) (1.28) Size 0.006 0.021 0.019 0.025 0.028 0.024 0.022 0.024 (0.12) (0.40) (0.35) (0.48) (1.62) (1.36) (1.27) (1.35) Lever 0.140 0.106 0.107 0.099 0.007 0.008 0.008 0.008 (0.73) (0.55) (0.56) (0.51) (0.89) (0.93) (0.93) (0.92) MB 0.012 0.043 0.023 0.062 0.128 0.122 0.123 0.122 (0.04) (0.14) (0.07) (0.20) (1.64) (1.57) (1.58) (1.57) ⁄⁄ ⁄⁄ ⁄⁄ ⁄ Dual 0.396 0.391 0.386 0.369 0.083 0.083 0.083 0.082 (2.12) (2.09) (2.06) (1.96) (1.54) (1.54) (1.53) (1.52) ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ ID 2.187 2.158 2.139 2.071 0.760 0.753 0.743 0.750 (2.31) (2.27) (2.24) (2.17) (2.12) (2.10) (2.07) (2.09) ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ HERF 0.211 0.235 0.251 0.258 0.108 0.106 0.103 0.115 (1.29) (1.42) (1.51) (1.55) (2.41) (2.37) (2.31) (2.55) ⁄ ⁄ ⁄ ⁄ Stock 1.479 1.433 1.423 1.442 0.025 0.026 0.027 0.026 (Continued) 150 Luo et al. Table 6. (Continued). Panel A Panel B (1) (2) (3) (4) (5) (6) (7) (8) Variables ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary ΔSalary (1.76) (1.70) (1.69) (1.71) (0.31) (0.32) (0.34) (0.32) Constant 0.262 0.651 0.625 0.732 0.688 0.598 0.567 0.597 (0.25) (0.61) (0.58) (0.68) (1.76) (1.52) (1.44) (1.52) Industry dummies YES YES YES YES YES YES YES YES R-squared 0.167 0.161 0.161 0.163 0.023 0.024 0.025 0.025 N 258 258 258 258 4714 4714 4714 4714 This table reports the OLS regression results of the effect of media coverage on executive compensation growth. The dependent variable is ΔSalary, ΔSalary =(Salary – Sal- i,t i,t ary )/ Salary , where: Salary represents the cash compensation total of the top three executives in listed company i. We divide the regression results into panel A and panel i,t–1 i,t–1 i,t B: Panel A includes columns (1) to (4), which present the results of 129 coverage sample and control sample matched by PSM; Panel B includes columns (5) to (8), which pre- sents the results of 4714 firm-year observations. Independent variables measuring media coverage are defined as the following four groups: (1) a dummy variable Report (0/1); continuous variables on the number of reports including (2) Media; (3) Media_negative and Media_other; (4) Media_official and Media_nonofficial. Control variables include: ⁄⁄⁄ ⁄⁄ ⁄ ROA, Size, Lever, MB, Dual, ID, HERF, Stock, and Ind. Variable definitions are shown in Table 2. T value is in brackets; significant at 1% level, significant at 5% level, significant at 10% level (two tailed). China Journal of Accounting Studies 151 Table 7. The effect of media coverage: interaction analysis. (1) (2) (3) (4) Variables ΔSalary ΔSalary ΔSalary ΔSalary ⁄⁄⁄ ⁄⁄ Media 0.229 0.132 (3.02) (2.35) MediaState 0.186 (1.77) MediaMono 0.560 (1.04) ⁄⁄⁄ ⁄⁄⁄ Report 0.509 0.300 (4.08) (3.06) ⁄⁄ ReportState 0.436 (2.42) ReportMono 0.303 (0.70) ⁄⁄ State 0.188 0.283 0.076 0.064 (1.57) (2.12) (0.77) (0.65) ROA 0.075 0.121 0.048 0.082 (0.20) (0.33) (0.13) (0.22) ⁄⁄ ⁄⁄ ⁄⁄ ⁄⁄ Size 0.099 0.106 0.096 0.093 (2.41) (2.59) (2.32) (2.27) Lever 0.043 0.028 0.042 0.035 (0.63) (0.42) (0.60) (0.52) ⁄⁄ ⁄⁄⁄ ⁄⁄ MB 0.509 0.545 0.517 (2.46) (2.43) (2.63) (2.43) ⁄ ⁄ ⁄ Dual 0.211 0.194 0.196 0.185 (1.87) (1.75) (1.71) (1.64) ⁄ ⁄⁄ ⁄ ⁄ ID 2.066 2.185 1.902 1.938 (1.85) (1.99) (1.70) (1.75) HERF 0.107 0.021 0.160 (0.43) (0.38) (0.08) (0.57) ⁄ ⁄ ⁄ Stock 1.349 1.239 1.543 1.566 (1.70) (1.59) (1.87) (1.92) Mono 0.124 0.206 (0.37) (0.62) ⁄ ⁄⁄ ⁄ ⁄ Constant 1.899 2.067 1.770 1.729 (1.91) (2.09) (1.77) (1.75) Industry dummies YES YES YES YES R-squared 0.186 0.212 0.179 0.194 N 252 252 252 252 This table reports the OLS regression results of the effect of media coverage in light of property right and monopolistic characteristic. The dependent variable is ΔSalary. ΔSalary =(Salary – Salary )/Salary i,t i,t i,t–1 i,t–1, where: Salary represents the cash compensation total of the top three executives in listed company i. Media i,t is the natural logarithm of coverage number plus one; Report is a dummy variable that equals 1 if a firm has been reported, and 0 otherwise. State is a dummy variable of property right, which equals 1 if a firm is state- owned, and 0 otherwise. Mono is also a dummy variable that equals 1 if a firm is in a monopoly industry, and 0 otherwise. By the PSM approach, there remain 126 firm-year observations and 126 matched observa- ⁄⁄⁄ ⁄⁄ ⁄ tions. T-value is in brackets; significant at 1% level, significant at 5% level, significant at 10% level (two tailed). The CAR (2, 2) in non-monopoly firms is 0.062 and is significant at 1%; in contrast, the CAR (2, 2) in monopoly firms is not significantly different from zero. The results indicate that the governance role of the media is stronger in non-monopolistic companies than in monopolistic companies. Overall, we find some support for H3 and H4. 152 Luo et al. 0.02 0.015 0.01 0.005 -5 -4 -3 -2 -1 0 1 2 3 4 5 day Figure 1. Market reaction to media coverage on executive compensation. This figure reports the cumulative abnormal return (CAR) of media coverage on executive compensation in the event window (5, 5). Table 8. CAR of media coverage in window (2, 2). Sample Number CAR(2,2) ⁄⁄⁄ Full sample 287 0.057 (4.37) ⁄⁄⁄ State-owned firms 124 0.034 (3.69) ⁄⁄⁄ Non-state-owned firms 163 0.074 (3.42) Monopoly firms 25 0.001 (0.09) ⁄⁄⁄ Non-monopoly firms 262 0.062 (4.40) This table reports the result of CAR of media coverage in window (2, 2). Excluding the insufficient data on stock returns and removing events happening on weekends, 287 firm-date observations remain. With such a sample, we conduct tests according to property right and monopolistic characteristic. T-value is in brackets; ⁄⁄⁄ ⁄⁄ ⁄ significant at 1% level, significant at 5% level, significant at 10% level (two tailed). 5. Robustness checks To examine the effect of media coverage on executive compensation, we also use the difference-in-difference (DID) method on the PSM sample to test the robustness of the results. The DID model is as follows: Salary ¼ a þ a Treat þ a Post þ a Treat  Post þ X b þ e ð4Þ 0 1 2 3 If a company is reported by media, Treat equals 1 and 0 otherwise. Post equals 1 for years in or after the media coverage year and 0 otherwise. Control variables include ROA, Size, Lever, MB, Dual, HERF, Stock, ID and State. We use the PSM method to identify a matching sample: in time matching, we use the first media coverage year as the cutoff point, and search for the reported firms before and after the media coverage year. The matching sample is drawn from the same period as the reported sample. After matching, we end up with 612 firm-year observations. Table 9 reports the DID regression results. The dependent variable is the natural logarithm of executive compensation. The results show that Salary is negatively corre- lated with Treat x Post, which shows that media coverage has a limiting effect on exec- utive compensation growth, which is robust to alternative estimation. CAR China Journal of Accounting Studies 153 Table 9. Robustness test: Regression results from the DID estimation. Variables Salary ⁄⁄⁄ Treat 0.373 (4.10) Post 0.183 (1.93) ⁄⁄⁄ TreatPost 0.443 (3.60) ⁄⁄⁄ State 0.322 (4.16) ⁄⁄⁄ ROA 2.686 (5.10) ⁄⁄⁄ Size 0.243 (10.56) Lever 0.143 (0.63) MB 0.211 (1.20) Dual 0.176 (1.48) ID 0.547 (0.83) HERF 0.189 (0.81) Stock 0.512 (0.89) ⁄⁄⁄ Constant 6.406 (6.89) Industry dummies YES R-squared 0.629 N 612 This table reports the results of robustness check by means of DID estimation. The independent variable is Salary. Salary is the natural logarithm of executive compensation. If a company is reported by media, Treat equals 1, and 0 otherwise. Post equals 1 for years in or after the media coverage year and 0 otherwise. T-value ⁄⁄⁄ ⁄⁄ ⁄ is in brackets; significant at 1% level, significant at 5% level, significant at 10% level (two tailed). 6. Conclusions and limitations We study whether media coverage constrains executive compensation in China, and we find that media coverage has a governance role on executive compensation via increas- ing reputation risk, and negative and official reports are more effective in constraining executive compensation. Meanwhile, the effect of media governance is conditional on some firm characteristics. For example, media coverage works more effectively in non- SOEs and results in more positive market reactions. Media reports result in a stronger market reaction to non-monopoly companies. One limitation of our paper is that we only focus on cash compensation but ignore stock ownership and stock option in our analysis, although they are not a major compo- nent of executive compensation in China. Acknowledgements We would like to thank the support from the National Social Science Foundation project (10XJL0012) ‘Research on Executive Compensation Mechanism in Monopoly Enterprises: With Compensation Cap Regulation’, the National Natural Science Foundation project (70802050) 154 Luo et al. ‘Research on Effectiveness of Independent Director Mechanism: Appointment, Reputation and Economic Consequences’, and Southwestern University of Finance and Economics (SWUFE) university project (2010XG076). This paper has benefited from comments by Zhi Jin, Xuesong Tang, Hongtao Tan, Li Ji, Danlu Bu and Xue Wang from SWUFE, Kangtao Ye and Xinjiao Guan from Renmin University of China, and the China Accounting Academic Leading Talent workshop. We thank the editors (Hong Zou and Jason Xiao) and reviewers for their hard work. Notes 1. Greg Farrell, ‘US Bank Chiefs Face $500,000 Limit’, Financial Times, 5 February 2009 (available at http://www.ft.com/cms/s/0/b06b7572-f2ed-11dd-abe6-0000779fd2ac. html#axzz2RknQttRK). Davis Polk and Wardell, “New Executive Compensation Restrictions under the Emergency Economic Stabilization Act of 2008’, 6 February 2009 (available at http://www.dpw.com/1485409/clientmemos/2009/02.05.09.ec.pdf). 2. Irrelevant media reports refer to indirect compensation reports, which focus on industrial compensation, average staff salaries, executive compensation policies, compensation systems or overseas compensation introduction. 3. When calculating ‘Salary’, we use the cash compensation total of the top three executives in listed companies, and exclude stock options and executive shares. The reason is that the stock incentive plan started late in China with an average lower level, and 0 shareholdings are quite common (Jiang & Huang, 2011). Lv and Zhao (2008) show that executive ownership as well as stock options do not seem to have significant incentive effects. We therefore expect that the omission of executive ownership and stock options should not result in a significant bias. References Aharony, J., Lee, C., & Wong, T. (2000). Financial packaging of IPO firms in China. Journal of Accounting Research, 38, 103–126. Besley, T., & Prat, A. (2006). Handcuffs for the grabbing hand? Media capture and government accountability. American Economic Review, 96, 720–736. Boycko, M., Shleifer, A., & Vishny, R. A. (1996). Theory of privatization. The Economic Jour- nal, 106, 309–319. Chen, Donghua, Chen, Xinyuan, & Wan, Hualin. (2005). Regulation and non-pecuniary compen- sation in Chinese SOEs. Economic Research Journal, 2,92–101. Chen, Yan, Li, Yanshu, & Li, Yong. (2005). Reputation mechanism and the deficiency of financial creditability. China Industrial Economy, 8,73–80. Cheng, S., & Indjejikian, R. (2009). The Market for corporate control and CEO compensation: Complements or substitutes? Contemporary Accounting Research, 26, 933–968. Core, J. E., Guaya, W., & Larckerb, D. F. (2008). The power of the pen and executive compensa- tion. Journal of Financial Economics, 88,1–25. Djankov, S., McLiesh, C., Nenova, T., & Shleifer, A. (2010). Who owns the media? NBER work- ing Paper No. 8288. Dyck, A., Volchkova, N., & Zingales, L. (2004). Private benefits of control: An international comparison. The Journal of Finance, 59, 537–600. Dyck, A., Volchkova, N., & Zingales, L. (2008). The corporate governance role of the media: Evidence from Russia. The Journal of Finance, 63, 1093–1135. Dyck, A., & Zingales, L. (2002). The corporate governance role of the media. NBER Working Paper no. 9309. Faccio, M., Masulis, R. W., & McConnell, J. J. (2006). Political connections and corporate bail- outs. Journal of Finance, 61, 2597–2635. Fama, E. F. (1980). Agency problems and the theory of the firm. Journal of Political Economy, 88, 288–307. Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. Journal of Law and Economics, 26, 301–325. Fang, Junxiong. (2009). Is there executive compensation stickiness in Chinese listed companies? Economic Research Journal, 3,110–124. China Journal of Accounting Studies 155 Fich, E. M., & Shivdasani, A. (2007). Financial fraud, director reputation, and shareholder wealth. Journal of Financial Economics, 86, 306–336. Fisman, R. (2001). Estimating the value of political connections. The American Economic Review, 91, 1095–1102. Fudenberg, D., & Levine, D. K. (1992). Maintaining a reputation when strategies are imperfectly observed. The Review of Economic Studies, 59, 561–579. Gao, Minghua, & Du, Wencui. (2010). The wage for manager in monopoly enterprise: Shortage or overage? Academia Bimestris, 3, 162–168. Gilson, S. C. (1990). Bankruptcy, boards, banks, and blockholders evidence on changes in corpo- rate ownership and control when firms default. Journal of Financial Economics, 27, 355–387. Harford, J. (2003). Takeover bids and target directors’ incentives: The impact of a bid on direc- tors’ wealth and board seat. Journal of Financial Economics, 69,51–83. Hong, Li., & Chai, Chunying. (2009). On the study of interactive relationship of media, the pub- lic and government in the formation of public opinion. Press Circles, 2,74–75. Jiang, Fuxiu., & Huang, Jicheng. (2011). Managerial incentive, debt and firm value. Economic Research Journal, 5,46–60. Joe, J. R., Louis, H., & Robinson, D. (2009). Managers’ and investors’ responses to media expo- sure of board ineffectiveness. Journal of Financial and Quantitative Analysis, 44, 579–605. Jordan, Otten. (2010). Media attention and executive pay in the Netherlands. Erasmus Institute Monitoring & Compliance, 4,1–45. Kaplan, S. N., & Reishus, D. (1990). Outside directorships and corporate performance. Journal of Financial Economics, 27, 389–410. Kreps, D. M., & Wilson, R. (1982). Reputation and imperfect information. Journal of Economic Theory, 27, 253–279. Kuhnen, C. M., & Niessen-Ruenzi, A. (2008). Is executive compensation shaped by public atti- tudes? CFR Working Paper no. 08–09. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. (1999). The quality of government. Journal of Law, Economics, and Organization, 15, 222–279. Li, Peigong, & Shen, Yifeng. (2010). The corporate governance role of media: Empirical evidence from China. Economic Research Journal, 4,14–27. Lin, Yifu, & Liu, Peilin. (2001). Autogenic ability and SOE reform. Economic Research Journal, 9,60–70. Lv, Changjiang, & Zhao, Yuheng. (2008). A study on the effect of the incentive given to manag- ers of state-owned enterprises. Management World, 11, 188. Miller, G. S. (2006). The press as a watchdog for accounting fraud. Journal of Accounting Research, 44, 1001–1033. Monks, A. G., & Minow, N. (2004). Corporate governance (3rd ed.). Malden, MA: Blackwell. Paligorova, T. (2009). Agency conflicts in the process of securitization. Bank of Canada Review, 9,33–48. Scott, G., & Konstantin, S. (2009). Government control of the media. CEFIR /NES Working Paper no. 126. Shen, Yifeng., & Li, Peigong. (2010). Empirical research on the impact of ‘Pay Ceiling Order’ on the CEO compensation and Pay-for-Luck phenomenon. China Industrial Economics, 11, 130–139. Shleifer, A., & Vishny, R. W. (1994). Politicians and firms. The Quarterly Journal of Economics, 109, 995–1025. Srinivasan, S. (2005). Consequences of financial reporting failure for outside directors: Evidence from accounting restatements and audit committee members. Journal of Accounting Research, 43, 291–334. Weigelt, K., & Camerer, C. (1988). Reputation and corporate strategy: A review of recent theory and applications. Strategic Management Journal, 9, 443–454. Wilson, R. (1985). Reputations in games and markets. In game-theoretic models of bargain. New York, NY: Cambridge University Press. Xin, Qingquan., Lin, Bin, & Wang, Yanchao. (2007). Government control: Executive compensa- tion and capital investment. Economic Research Journal, 8,110–122. Xin, Qingquan, & Tan, Weiqiang. (2009). Marketization reform: Corporate performance and SOE executive compensation. Economic Research Journal, 11,68–81. 156 Luo et al. Xu, Haoping, & Luo, Wei. (2007). Reputation effect of investment banks: Research from aspects of market share and service quality. Economic Research Journal, 2, 124–136. Yang, Yada, & Xu, Hong. (2004). Manager reputation motivation mechanism in SOEs. Economic Theory and Business Management, 4,47–49. Yao, Renan. (2008). Value the positive effect on negative reports. Information of Economics and Law, 6,12–13. Yermack, D. (2004). Remuneration, retention, and reputation incentives for outside directors. Journal of Finance, 59, 2281–2308. Zhang, Wei (2003). Comparative journalism: Methodology and textual research. Guangzhou: Southern Daily Press. Zhang, Weiying. (2002). Reputation foundation of legal system. Economic Research Journal, 1,3–13.

Journal

China Journal of Accounting Studies – Taylor & Francis

Published: Jun 1, 2013

Keywords: reputation mechanism; media coverage; executive compensation

There are no references for this article.