报告论文下载.pdf

Enforcement and the Effectiveness of Laws and Regulations Ke Li School of Finance Shanghai University of Finance & Economics rocksonlee@yahoo.cn Lei Lu School of Finance Shanghai University of Finance & Economics lu.lei@mail.shufe.edu.cn Jun “QJ” Qian Carroll School of Management Boston College qianju@bc.edu December 2009 Preliminary, comments welcome Enforcement and the Effectiveness of Laws and Regulations Abstract We empirically examine the impact of two distinct laws, announced and implemented during the same period, on publicly listed firms in China. The first law prohibits controlling shareholders and related parties to divert assets from listed firms for ‘non-operational’ purposes, while the second aims to standardize the practice of listed firms providing loan guarantees. Relative to firms not affected by the law, firms complying with the first law experience a reduction in the ownership stakes of controlling shareholders, an increase in investment, and significantly better performance. We do not find such relationships for the second law. Our results indicate laws that are more clearly defined and easier to enforce will be more effective, especially in countries with weak institutions. Keywords: Law and finance, enforcement, controlling shareholder, diversion, China. JEL Classifications: G30, G34, K42 1. Introduction The extensive literature on law and finance has established robust associations between legal protection of investors and better financial and economic outcomes across countries. But which types of laws and regulations are more important in different countries remains an open question. For developing countries, where the costs of developing formal institutions can be enormous, introducing and enforcing laws and regulations that have a first-order impact and at reasonable costs is of considerable significance. There is some consensus that one of the key obstacles for developing financial markets and strengthening corporate governance in developing countries is powerful and entrenched controlling shareholders, who can ‘tunnel’ resources from the firms to themselves.1 While there are many possible solutions for this ‘self-dealing’ problem, a lot has to do with the enforcement of laws and regulations. Empirically it is also difficult to separate the effects of a particular law or regulation on firms and markets from other contemporaneous policies and events. In this paper, we empirically examine the impact of two distinct laws, announced and implemented during the same period, on publicly listed firms in China. As we will demonstrate, these two laws were announced and implemented quickly, and it is unlikely that firms can anticipate both the announcement dates and the implementation process. Using a difference-in-difference approach, we show that these two laws have quite different effects on firms’ governance and performance. Our results provide direct evidence that enforceability matters: laws and regulations that are more clearly defined and easier to enforce will be more effective in practice, especially in countries with weak (formal) institutions. China provides an intriguing case to study the effects of laws and legal enforcement on financial markets. Allen, Qian, and Qian (2005) show that nonbank, nonmarket financing channels, along with nonlegal mechanisms, have been fueling the growth of non-state, non-listed 1 For the literature on law and finance, see the work by La Porta, Lopez-de-Silanes, Shleifer, and Vishny (LLSV; 1997, 1998), and others. Djankov, La Porta, Lopez-de-Silanes, and Shleifer (DLLS (2008)) construct the self-dealing index and find it to be correlated with stock market development across a large sample of countries. 1 firms, but these mechanisms have not worked well for listed firms, many of which converted from State-owned Enterprises (SOEs). Despite fast growth since its inception in 1992, China’s stock market remains inefficient and dominated by insider trading, and corporate governance for listed firms is weak (Allen, Qian, and Qian, 2008). Tunneling by controlling shareholders is prevalent and takes on many forms. For example, through the holding company or related parties, controlling shareholders frequently divert assets (cash or real assets) away from listed firms, sometimes without any explanation on the use of the diverted assets. Listed firms also ‘involuntarily’ provide loan guarantees for the subsidiaries or related parties of the controlling shareholders. In 2005, the China Securities Regulatory Commission (CSRC, equivalent to the SEC in the U.S.) announced two new laws specifically designed to solve these two forms of tunneling. The first law prohibits controlling shareholders and related parties to divert assets from listed firms for ‘non-operational’ purposes. All controlling shareholders, especially state-owned entities, must repay the diverted assets by the end of 2006. The enforcement process is also transparent in that both the Shanghai and Shenzhen Stock Exchanges disclose details on both the paying and nonpaying controlling shareholders. The second law aims to ‘standardize’ the practice of listed firms providing loan guarantees, but the law does not provide specific guidelines on the implementation process. This is perhaps due to the vague nature of the law as compared to the numerous ways different combinations of loan guarantees can be provided in practice. The enforcement process of this law is much more opaque (with limited press coverage) compared to the first law. Relative to firms not affected by the law, we find that firms complying with the first law (no diversion of assets) experience a reduction in the ownership stakes of controlling shareholders. This is consistent with the hypothesis that the law makes it much more difficult for control shareholders to tunnel assets from listed firms, and as a result they reduce their ownership stakes and moved it elsewhere. These firms also experience an increase in investment and significantly better performance, as measured by return on assets (ROA), sales (ROS), 2 equity (ROE), and earnings per share (EPS). These firms also have higher (cumulative) abnormal stock returns over the period of the announcement and enforcement of the law. On the other hand, we do not find the second law (regulating loan guarantees) has any significant impact on affected firms’ governance, investment or performance. The impact of the first law is robust to controlling for the second law (as some firms are affected by both). We also find some evidence that the positive impact of either law is less pronounced for firms that are ultimately owned by the state as compared to those whose ultimate owners are not the state. This result supports the argument that playing the conflicting, dual roles of regulators and large shareholder diminishes the effectiveness of the government in each of the two roles. Our work contributes and extends the literature on law and finance. Our within-country study avoids many pitfalls of cross-country studies, and our difference-in difference approach overcomes the endogeneity problems in separating the effects of the laws and regulations of interests from other factors. More importantly, by examining the impact of two distinct laws at the same time, we provide direct evidence that enforceability is an important determinant of the effectiveness of laws and regulations. This has important implications for developing countries as formal institutions in these countries are weak, so that the legal system cannot enforce complicated laws and regulations. On the other hand, laws and regulations that have clearly defined, easy to interpret and verify ‘right’ from ‘wrong’ outcomes will be enforced much more effectively, and, as a result, have a greater impact on firms and markets. Finally, the paper contributes to the growing literature on China’s capital markets and economy by investigating the effects of different regulations on listed firms’ governance performance. This rest of the paper is organized as follows. Section 2 introduces the background of regulations in China. Section 3 presents the data and statistical analysis. Section 4 provides the effect of law on firms’ performance. Section 5 provides the relationship between law enforcement and firms’ performance. Section 6 discusses the political connections and law enforcement. Section 7 presents the mechanisms and Section 8 concludes the paper. 3 2. Background of Regulation Changes in China It is pervasive that the controlling shareholders pursue private benefits at the cost of the minority shareholders in Chinese capital markets because of the absence of laws and institutions protecting the benefits of minority shareholders. The two main channels for controlling shareholders to pursue the private benefits are the related transactions between controlling shareholders and related parties, and the asset diversions. These two problems have caused serious issues for the healthy development of Chinese capital markets. In order to solve these problems, in June 2005, the China Securities Regulatory Commission (CSRC) issued “the Notice on Solving the Problems of Asset Diversion and Illegible Loan Guarantee” (“NOTICE” hereafter), which impose the regulations on both asset diversion and loan guarantee by controlling shareholders and related parties. 2.1 Two Cases: Asset Diversion and Related-Party Loan Guarantee From the following two cases of Beer Flower Company and Monkey King Co. Ltd., we can find the serious impact of above two problems. The company of Beer Flower was run in the good condition until the firm’s financial statement disclosed its loan guarantee of 799.8 millions, for which 354.8 millions are lend to its subsidiaries on November 4th 2003, and the undisclosed loan guarantee is 987.8607 millions. It means that the total loan guarantee is 1.7877 billions, which exceeds 80 percent of its total assets of 2.2096 billion, while its net assets is only 601.2262 millions. Since its loan guarantee was publicly released, its stock price fell drastically from 17RMB to 5RMB in 13 trading days. The firm of Beer Flower was finally merged by Sichuan Blue Arrow Company. The Monkey King Co. Ltd., formed on November 18th 1992, was a subsidiary of Monkey King Group. The Monkey King A-share was issued in Shenzhen Stock Exchange on November 30th 1993. Its total assets rose to 300 millions of which 110 millions was raised from the stock market, and the prime performing revenue was 394 millions at the same year. As the largest 4 shareholder of Monkey King Co. Ltd, the Monkey King group started to divert the assets of the Monkey King Co. Ltd for ‘non-operational purposes’ since 1994 and the accumulated amount was 890 millions in 1999. In addition, the Monkey King Group borrowed 370 millions from the bank with the related-party loan guarantee from the Monkey King Co. Ltd. as the collateral. The performance of Monkey King Co. Ltd. became worse and the prime performing revenue was only 422.9 millions in 2000. The financial statement showed that its net asset is -376 millions with loss of 689 millions in 2000 comparing to the gain of 328 millions in 1993. The revenue per share and the return on common stockholders’ equity dropped from 0.57 and 19.56% in 1993 to -2.28 and -183.16% in 2000, respectively. The Money King Group disclaimed the bankruptcy on March 1st 2001, and then the Monkey King A-share was treated specially as “ST Monkey King”. Due to the bankruptcy of Monkey King Group, the ST Monkey King suffers the loss of 890 millions and the liability of 244 millions, it has to disclaim the bankruptcy in 2001 2.2 Regulations of CRSC The CSRC’s regulations on asset diversion and loan guarantee by both controlling shareholders and related parties provide a natural experiment to study the effects of law and its enforcement on corporate governance and firms’ performance. First, the exogeneity of laws (or regulations) relaxes the problem of endogeneity of corporate governance. In June 2005, the CSRC issued the “NOTICE” to solve the problems of asset diversion and illegal loan guarantee, and required its branches to evaluate the listed firms according to the “NOTICE”, and especially to completely solve the problem of the diversion of assets for ‘non-operational purposes’ by the end of 2005. In November 2005, the State Council authorized the CSRC to announce “the Official Suggestions on Improving the Quality of Listed Firms”. It requires that controlling shareholders or firms’ eventual owners are strictly prevented from diverting firms’ assets for ‘non-operational purposes’ and all controlling shareholders, especially for state-owned firms and eventual owners, must payback the diverted assets by the 5 end of 2006. In 2006, both Shanghai and Shenzhen Stock Exchanges required all listed firms with asset diversion must report their payback plan, the exact amount of payback, and the schedule in their 2005 annual financial statements to guarantee that the asset diversion disappeared completely by the end of 2006. The following example shows the impact of above regulations on corporate governance. The assets of 600 millions of Shanghai Broadband Technology were diverted by its largest shareholder, Nanjing SVT. Its Chairman of the board, Jie Zhang, was responsible for the asset diversion and was arrested on a charge of asset diversion in July 2005. Until the end of 2006, the responsibility persons of 10 firms (among 17) who didn’t payback firms’ diverting asset were investigated by the judicial authority and 399 firms have solved the problem of asset diversion with the total assets of 39 billions. In sum, the regulations imposed by the CRSC prevent controlling shareholders or firms’ actual owners from pursing private benefits at the cost of minority shareholders, and improve firms’ corporate governance. Second, the two regulations announced by the CSRC have important differences in the enforcement, which helps us to understand the effects of law enforcement on the relationship of law with corporate governance and then with firms’ performance. Although the two regulations were announced by the CSRC simultaneously, they have significant different requirements from the CSRC. The second item of the “NOTICE” clearly demonstrates that the emphasis should be placed on solving the problem of asset diversion by controlling shareholders and illustrates the detailed procedures, while the regulation on loan guarantee only requires the companies to standardize their behavior of loan guarantee and did not gives the detailed explanation. Corresponding to the different requirements from the CSRC, the two regulations were exercised with different enforcement. In 2006, both the Shanghai and Shenzhen Stock Exchanges frequently announced the “Notice of Asset Diversion by Controlling Shareholders and their Subsidiaries” and disclosed their commitment of asset payback. Not only the names of listed firms and the amounts of asset diversion, but the list of legal persons was publicly 6 announced. However, the exchanges and news media rarely mentioned the information of the related-party loan guarantee. On June 28th 2006, the “Crime of Asset Diversion of Listed Firms” was added to the “Amendment 6 to the Criminal Law of the People's Republic of China”. It clearly demonstrates that “the member of board, the supervisors, and the senior administrators of listed firms will be charged of three to seven-year imprisonment and are imposed a fine if they manipulate the performance of listed firms and the manipulation causes firms to suffer serious loss. The controlling shareholders or eventual owners of the firms will be penalized the same charge if they incite the above behaviors.” 3. Methodology and Data 3.1 Data The data includes five types of information: accounting variables, stock price and return, ownership structure, regulations, and other firm characteristics. In order to avoid the bias caused by outliers, we winsorize all data at 1% level. First, the accounting data include the return on assets (ROA) defined as the earnings before interest and tax (EBIT) by lagged total assets at year t for firm i; the return on sales (ROS) defined as the earnings before interest and tax (EBIT) by lagged total sales at year t for firm i; the return on equity (ROE) defined as the net profit by lagged equity at year t for firm i; the earnings per share (EPS) defined as the profit by outstanding shares at year t for firm i; the investment defined as the capital expenditure by lagged total assets at year t for firm i; total assets, total liability, and tangible assets, and they are obtained from the CSMAR database. Second, the stock price and return is obtained from the WIND database. Third, the data on the change of control right, the share ratio hold by shareholders, the turnovers of CEO, and the reform of share split structure (SPLIT) are obtained from the CSMAR database. The variable of SPLIT equals 1 starting from the year of structural reform, otherwise equals 0. 7 Fourth, the policy change is collected manually. The information about asset diversion by controlling shareholders or their related parties is obtained from the 2005 firms’ annual report and the website of Shanghai Stock Exchange. The Shanghai Stock Exchange disclosed the detailed information about asset diversion for all firms listed on the Shanghai Stock Exchange. For firms listed on the Shenzhen Stock Exchange, we manually collect the data from the abstract of firms’ annual report. Finally, the other firm characteristics (e.g., special treatment (ST), relationship with the government, issuance of H-Shares, and industry classification) are obtained from the CCER database and the WIND database. We choose all A-share listed firms in Shanghai and Shenzhen Stock Exchanges from 2003 through 2007, excluding all financial firms and firms listed after 2004. 3.2 Descriptive Statistics Table 1 presents the summary statistics of firms’ characteristics for the period 2003 through 2007 by categorizing all firms into the treatment group (TREAT) and the control group (CONTL). In Panel A, the controlling shareholders in the firms of the treatment group diverted firms’ assets before 2006, while the firms in the control group did not. The third column shows the means of various variables for full samples, and the fifth and seventh columns report the means for the two groups. The values of the last column are the difference between two groups. We find that there exist significant differences between two groups in the return on assets (ROA), return on equity (ROE), earnings per share (EPS), investment, leverage, size, percentage change of stock prices, and turnover of CEO at 1% confidence level. The concentration of the largest shareholders is significantly different at 5% confidence level, the concentration of the first three shareholders and the ratio of tangible assets to total assets are significantly different at 10% confidence level, while the return on sales (ROS) and the turnover of control right are not statically significantly different. 8 In Panel B, the firms of the treatment group have the related-party loan guarantee before 2006, while the firms in the control group did not. The analysis is very similar to that for the firms’ assets illustrated in the Panel A, it has significant difference in firms’ characteristics between the treatment and control groups, which can be observed from the last column. In particular, the means of ROA, EPS, investment, leverage, concentration of largest shareholder, and concentration of first three shareholders are significantly different at 1% confidence level. Table 2 presents the distribution of the asset diversion and the related-party loan guarantee for full samples (Panel A), classified by industry (Panel B) and the type of controlling shareholders (Panel C), respectively. The Panel A shows that, 1908 of 5579 firms (around 34.2%) diverted firms’ assets before 2006, and 3536 (63.4%) firms have related-party loan guarantee. Moreover, the controlling shareholders of 1322 (around 23.7%) among 5579 firms diverted firms’ assets and have related-party loan guarantee simultaneously before 2006. The Panel B reports the distribution of the asset diversion and the related-party loan guarantee across nine industries. For the asset diversion, the agriculture has the highest percentage of 50.8% and the trade has the lowest with 21.4%. In general, it doesn’t have significant difference across industries. For the related-party loan guarantee, the distribution across industries is more even with the highest of 76.1% for conglomerate and the lowest of 45.5% for transportation. The Panel C shows the distribution between government related and non-government related firms. Of 3381 government-related firms, 1217 (around 31.8%) and 2343 (around 61.2%) have asset diversion and related-party loan guarantee before 2006, respectively. Of 1754 non-government related firms, 688 firms (around 39.2%) and 1183 (around 67.4%) firms have asset diversion and related-party loan guarantee before 2006, respectively. We do not find the significant difference in the distribution between government- and non-government-related firms. 9 3.3 Methodology Following Bertrand and Mullainathan (2003), we use the difference-in-difference method to examine the effect of law and its enforcement on corporate governance, and firms’ behavior and performance. y it  a  bPost it * Event it   cControls it  Firmi  Yeart   it (1) where i denotes firm i, t denotes year t, y denotes dependent variables including the return on assets (ROA), the return on sales (ROS), the return on equity (ROE), the earnings per share (EPS), the investment, and the ownership change; Firm and Year denote the firm- and year-fixed effect, respectively; Controls denote control variables including logarithmic assets, ratio of tangible assets to total assets, firm leverage (the ratio of book value of debt to book value of assets), dummy variables of split share structure reform, government, ST, turnover of CEO, and turnover of control right, change of stock price, concentration of ownership by the largest shareholder, concentration of ownership by first three largest shareholders. Post, a dummy variable, denotes whether the regulations are exercised. If the observations occurred before 2006, Post equals 0 otherwise 1. Event, a dummy variable, denotes whether the regulations affect firm i at year t. The dummy variable, Event, equals to 1 for firms of the treatment group, otherwise 0 for firms of the control group. Thus the coefficient of Post*Event reflects the impact of regulations on the relative performance of the treatment group to the control group. Bertrand et al. (2004) claim that the difference-in-difference method suffers the serial correlation, which will increase the statistic significance. To make the results robust, we cluster the errors at the firm level (e.g., Petersen (2005) and Bertrand et al. (2004)). 4. The Effect of Law on Firms’ Performance In this section, we will investigate the effect of the regulation of asset diversion by controlling shareholders on firms’ performance. 10 4.1 Preliminary Analysis Figure 1 plots the changes in the ROA when the regulation of asset diversion by controlling shareholders was announced in 2005. It shows the different patterns of median ROA for the treatment group and the control group. We observe that the firm’s performance of the treatment group was increased more than the control group. It implies that the difference in ROA tends to be narrowed when the regulation was imposed on firms in 2005. Before the regulation of asset diversion was exercised, the ROA for the treatment group decreased from 3.3% in 2003 to 2.1% in 2005 with the decrease of 1.2%, and the ROA for the control group decreased from 6% in 2003 to 4.7% in 2005 with the decrease of 1.3%. These two groups have the similar patterns. However, the situation changed since the regulation was exercised in 2005, the ROA of the treatment group increased from 2.1% in 2005 to 5.7% in 2007 with the increase of 3.6%, while the ROA of the treatment group increased from 4.7% in 2005 to 6.5% in 2007 with the increase of 1.8% only. To some degree, it means that the regulation of asset diversion improves the corporate governance of the treatment group, and then causes an increase of 1.8% (=3.6%-1.8%) in the ROA. 4.2 Empirical Results In this section, we investigate the effect of the regulation of asset diversion on firms’ performance by running model (1). Table 3 reports the regression results. The estimated coefficient for Post*Asset shows that, the ROA of the treatment group relative to the control group was increased. The first column demonstrates that when controlling shareholders were prohibited from diverting assets in 2005 the relative ROA of treatment group significantly increases by 3.3% at the 1% confidence level. The possible reason is that the regulation reduces the possibility for controlling shareholders to pursue private benefits at the cost of firms’ performance. When we control firms’ characteristics, e.g., size, leverage and ratio of tangible assets to total assets, we find that the coefficient of Post*Asset increases to 3.9% and it is still 11 significant at 1% confidence level which can be found in column 2. In 2005, the CRSC started the reform of split share structure to eliminate the non-tradable shares, which accounted for two third of total shares. Although it has systematic influence on the Chinese capital market, we rerun model (1) by controlling its effect completely. The column 3 shows that, by controlling the reform of split share structure, the relative ROA of the treatment group increases 3% with the significance at 1% confidence level due to the regulation of asset diversion by controlling shareholders. One interesting by-product of the regression is that the coefficient of SPLIT is significantly negative of -2.6% at 1% confidence level. It means that the reform of split share structure decreases firms’ performance (e.g., ROA). One possible explanation is that firms manipulated the performance before the reform to satisfy the requirements imposed by the CSRC, while the performance was adjusted back to the actual level after the reform. Started from 1998, the stocks of listed firms with either financial distress or other anomalies were treated by the CSRC as the special stocks (in short, “ST” stocks). Since 2001, the CSRC claimed that firms with 3-year consecutive negative profits have the possibility to be delisted from the exchanges and their stocks were prohibited from being traded. In column 4, we report the regression results by controlling the property of firms’ eventual owners (e.g., state-owned or not), GOV, and the property of stocks, ST (special treatment). It shows that firms’ ROA is not significantly affected by the property of eventual owners (GOV), while it is positively and significantly affected by the property of stocks (ST). The reasons are as follows. One the one hand, if the firm has connections with the government, it has to suffer the policy burden and its performance will be distorted; on the other hand, firms with connections to the government might be financially sponsored by the government (e.g., lower borrowing interest rate) and then have higher performance. Therefore, the effect of the property of firms’ eventual owners on firms’ performance is ambiguous. It also shows that the ROA of the treatment group is increased 2.6% relative to the control group by controlling the GOV and the ST. 12 In column 5, we control the change in stock prices and the corporate governance variable (e.g., turnover of CEO, turnover of control right, ownership of the largest shareholder, and ownership of first three shareholders) since they might be endogenously determined by firms’ performance. For example, the controlling shareholders have more incentive to supervise managers as their ownership increases and then firms’ performance is increased, while the controlling shareholders might increase their ownerships because they have private information that the firms’ performance has improved (but has not been released to the public). The result shows that the ROA of the treatment group relative to the control group significantly increases 2.9% at 1% confidence level. In sum, we find that, by prohibiting controlling shareholders from diverting firms’ assets, the relative ROA of the treatment group relative to the control group is significantly increased. We demonstrate that the regulation on controlling shareholders reduces the possibility for them to pursue private benefits at the cost of minority shareholders and then the firms’ performance is increased. 4.3 Robust check 4.3.1 Regression without 2005 year data One potential problem with the above results is that the regulation announced in 2005 only has effect on firms’ current performance, while it has no effect on firms’ corporate governance and then does not affect firms’ future performance. To eliminate this possibility, we remove the data of 2005 and rerun the regression (1) to investigate the relationship between regulation and firms’ performance. Table 4 reports the regression results. We find that, by preventing controlling shareholders from diverting firms’ assets, the ROA of the treatment group relative to the control group is significantly increased about 3% at 1% confidence level when control controlling firms’ characteristics and other variables. This is consistent with the above results. 13 4.3.2 Alternative Measures for Firms’ Performance We use the alternatives to measure firms’ performance, e.g., ROE, ROS, and EPS. Table 5 presents the regression results. The first three columns show the effect of the regulation of asset diversion on firms’ ROE. We find that most of the results are consistent with the analysis for the ROA although they have minor differences. The coefficient of Post*Asset increases from 3.3% for the ROA to 9.1% for the ROE when we only include the Post*Asset in the regressions, and it is significant at 10% confidence level when we control firms’ characteristics and other variables. The columns 4-6 show that the regulation preventing controlling shareholders from diverting firms’ assets generates an extra 26.2% in ROS for the treatment group relative to the control group when we only use the Post*Asset to explain the ROS. When we include firms’ characteristics and other control variables, the relative change for the treatment group is still positive although it becomes insignificant. The possible reason is the negative effect of the ratio of tangible assets to the total assets (Tangible) offsets the positive effect of the regulation. The columns 7-9 report the regression results for the EPS. The first row for Post*Asset shows, on average, the relative EPS of the treatment group is increased 5% by imposing the regulation on controlling shareholders. Although the significance level decreases, it is still positive and significant at 5% confidence level for columns 7 and 9 and significant at 10% confidence level for column 8. In general, when we use the alternatives to measure firms’ performance, the regulation of preventing controlling shareholders from diverting firms’ assets increases the relative performance of the treatment group. 4.3.3 Matching We observe from the Table 1 that the treatment group and the control group have significant differences in firms’ characteristics. This might lead to overestimate the effect of the regulation on firms’ performance. In the following, we use the matching method of Propensity 14 Score (Abadie and Imbens, 2002) to solve the potential problem. First, we calculate the changes in performance variables (e.g., ROA, ROE, ROS and EPS) for two groups when imposing the regulation on firms in 2005 --- the difference between the average performance for years 2006-2007 and that for years 2003-2005. We denote them by FD_ROA, FD_ROE, FD_ROS, and FD_EPS, respectively, and then we calculate the difference in difference between the treatment and control groups. Second, we calculate the propensity scores for matching firms. We select one firm from the treatment group and some firms from the control group, these firms have the closest characteristics in the same industry, e.g., total assets, loan guarantee, sales, leverage, ratio of tangible assets to total assets, and actual controlling shareholders. Table 6 reports the results using the matching method of propensity score. The estimates represent the increases in firms’ performance of the treatment group relative to the control group when the regulation on asset diversion by controlling shareholders was strictly exercised in 2005. In model (1), the matching firms selected from the control group have the most closed total assets. N denotes the numbers of firms selected from the control group to match the firm of the treatment group. The matching variables are total assets and whether firms have loan guarantee in model (2), and total assets and sales in model (3). We add the leverage and the ratio of tangible assets to total assets to model (3) to generate the model (4), and add whether firms have loan guarantee and the property of eventual controlling shareholders to model (4) to generate model (5). Panel A presents the estimated results for the FD_ROA. The estimate is 0.032 for N=1 in model (1) with the significance of 1% confidence level. It means that, when we choose one firm with the closest total assets from the control group to compare with the firm of the treatment group, the relative ROA of the treatment group to the control group increases 3.2% when preventing controlling shareholders from diverting firms’ assets. When we either increase the number of firms selected from the control group (e.g., N=2, 4), or add more matching variables 15 (e.g., models (2)-(5)), we have the similar conclusion. Moreover, we also examine the effect on the ROE, the ROS and the EPS, the results are the same although they are less significant than the ROA. 4.3.4 CAR Analysis Table 7 reports the cumulative abnormal returns (CARs) sorted on firms’ characteristics. Panels A-E provides the CARs around the regulation announcement of asset diversion for different groups (e.g., full sample, firms excluding H shares or ST shares, state-owned enterprises, and non state-owned enterprises). For full samples of Panel A, the performance of the treatment group relative to the control group is positive and significant at 1% level (except for the window [-5, 12] with 10% confidence level). When we exclude the firms with H shares or ST shares illustrated in Panels B and C, the above results become more significant. Even when we separate the firms in into state-owned enterprises (SOEs) and non state-owned enterprises (non-SOEs), the above results don’t change qualitatively. 4.3.5 Other Robustness Checks Typically, the ST firms have poor performances and their corporate governance is relative weak. Since the firms with H-Shares are listed in Hong Kong, the must satisfy the Hong Kong exchange regulatory requirements. We delete firms with either special treatment (ST) or H-shares and rerun the above regressions. We find that these two factors do not change the results qualitatively although the results are not reported din this paper. 5. The Effect of Law Enforcement on Firms’ Performance In 2005, the CSRC announced two regulations to prevent controlling shareholders from pursuing private benefits at the cost of minority shareholders, while they have significant different enforcement. The CSRC emphasized that controlling shareholders are prevented from 16 diverting firms’ assets. The two stock exchanges, the news media and the judicial institutions responded strongly to this regulation. The firms whose assets were diverted by controlling shareholders in 2005 are strictly required by the “NOTICE” to payback by the end of 2006, otherwise they will be charged of three to seven-year imprisonment by the “Amendment 6 to the Criminal Law of the People's Republic of China”. In contrast, none of them paid much attention to the regulation on related-party loan guarantee, and firms which have related-party loan guarantee in 2005 are only required to standardize their behavior of loan guarantee. This provides us an opportunity to investigate whether the regulations with weak law enforcement improve firms’ performance and whether the difference in law enforcement has different impacts on firms’ performance. Table 8 reports the regression results. The coefficients of Post*Asset and Post*Loan represent the effect of the regulations of asset diversion and related-party loan guarantee on the relative ROA of the treatment group to the control group, respectively. The columns (1)-(4) provide the results for the impact of the regulation of related-party loan guarantee, we find that the coefficients of Post*Loan are not significant at 10% confidence level and they can be negative when we control firms’ characteristics and other variables. It means that the regulation on firms’ related-party loan guarantee does not significantly affect the relative ROA of the treatment group. To robustly check the above conclusion, we rerun the regressions for alternative measures of firms’ performance, e.g., ROE, ROS, and EPS, and the results are consistent with the analysis for the ROA. To investigate whether the difference in law enforcement has different impacts on firms’ performance, we add the regulation with more strict enforcement – preventing controlling shareholders from diverting firms’ assets -- to the above regression models. The columns 5-8 report the effect of regulations with different law enforcement. On average, the coefficients of Post*Asset, the impact of the regulation with stronger law enforcement on firms’ ROA, are positive and significant at 1% confidence level, while the coefficients of Post*Loan, the impact 17 of the regulation with the weaker law enforcement on firms’ ROA, are insignificant and even become negative. We also run the regressions for alternative measures of firms’ performance, e.g., ROE, ROS, and EPS, and the results do not change qualitatively. In summary, the regulations with the more strict law enforcement improve the relative performance of the treatment group to the control group, while the regulations with the weak law enforcement are unrelated with firms’ performance. 6. Political Connections and Law Enforcement If the regulations are strictly exercised by the authorities, all firms whether they have connections with the government or not have to obey and then there does not exist the significant difference between their performances. However, if the regulation is not strictly exercised by the authorities, firms with connections to the government have more incentive to seek for the help from the government, and then there might exist the significant difference between their performances. In this section, we will investigate the role of the connections to the government on the relationship between law enforcement and firms’ performance. 6.1 Preliminary Analysis First, we analyze the regulation on preventing controlling shareholders from diverting firms’ assets. Figure 2 plots the different patterns in the ROA for the treatment group and the control group by separating into two classes: firms with connections to the government (e.g., government is the eventual controlling shareholder) and firms without connections to the government (e.g., the eventual controlling shareholder is not the government). The upper picture shows the changes in the ROA for firms with connections to the government. We find that the treatment group and the control group have the similar decreases in ROA from 2003 to 2005, while the ROA of the control group increases 1.8% (e.g., from 4.6% to 6.4%) and the treatment group increases 2.6% (e.g. from 2.5% to 5.1%) from 2005 to 2007. It means that, for the firms 18 with connections to the government, the regulation preventing controlling shareholders from diverting firms’ assets improves the ROA of 0.8% for the treatment group relative to the control group. The lower picture shows the changes in the ROA for firms without connections to the government. Similar to the analysis to the firms with connections to the government, the treatment group and the control group have the similar decreases in ROA from 2003 to 2005, while the ROA of the control group was increased 2.4% (e.g., from 4.7% to 7.1%) and the increase for the treatment group is 5.7% (e.g. from 0.5% to 6.2%) from 2005 to 2007. Therefore, for the firms without connections to the government, the regulation preventing controlling shareholders from diverting firms’ assets improves the performance of 3.3% for the treatment group relative to the control group. Combing the two pictures, we find that the firms without connections to the government have a higher relative increase in the ROA of 2.5% than firms with connections to the government. Second, we analyze the regulation on preventing controlling shareholders from related-party loan guarantee. Figure 3 plots the different patterns in the ROA for the treatment group and the control group by separating into two classes: with or without connections to the government. The upper picture shows the changes in the ROA for firms with connections to the government. We find that the ROA of the control group was increased 1.4% (e.g., from 4.5% to 5.9%) and the increase for the treatment group is 2.2% (e.g. from 3.8% to 6.0%) from 2005 to 2007. It means that, for the firms with connections to the government, the regulation preventing controlling shareholders from related-party loan guarantee improves the performance of 0.8% for the treatment group relative to the control group. The lower picture shows the changes in the ROA for firms without connections to the government. Similar to the analysis to the firms with connections to the government, the ROA of the control group was increased 2.6% (e.g., from 3.9% to 6.5%) and the increase for the treatment group is 3.7% (e.g. from 3.1% to 6.8%) from 2005 to 2007. Therefore, for the firms without connections to the government, the regulation preventing controlling shareholders from related-party loan guarantee improves the performance 19 of 1.1% for the treatment group relative to the control group. Combing the two pictures, we find that the firms without connections have a higher increase in the ROA of 0.3% than firms with connection to government. In summary, although the two regulations have different positive effect on firms’ performance, the effect is more significant for firms with less connection to the government. 6.2 Empirical Results Table 9 reports the regression results for the role of political connections on the relationship between law enforcement and firms’ performance (e.g., ROA). The estimated coefficients of Post*Asset and Post*Loan denote the effects of the regulations of preventing controlling shareholders from diverting firms’ assets (with strong law enforcement) and from related-party loan guarantee (with weak law enforcement) on the relative ROA of the treatment group to the control group for the firms without connections to the government, respectively. The coefficients of Gov*Post*Asset and Gov*Post*Loan denote the differences in effects between firms with connections to the government and those without connections to the government. The first three columns present the effect of the regulation on the ROA under the stronger law enforcement. The coefficient for the Post*Asset is 5.1% in the first column and is significant at 1% confidence level. When we include the variables of firms’ characteristics and other variables, the coefficients of Post*Asset are still positive with 4.3% and 4.1%, and are significant at 1% confidence level as illustrated in columns 2 and 3. It means that the ROA of firms without connections to the government is increased by 5.1% when imposing the regulation on controlling shareholders from diverting firms’ assets. The coefficient of Gov*Post*Asset is -3.2% and is only significant at 10% confidence level, while it becomes negatively insignificant at 10% confidence level when we include the control variables. This implies that the regulation on controlling shareholders from diverting firms’ assets reduces the ROA of firms with 20 connections to the government but the negative effect is insignificant. We conclude that the political connections do not significantly affect the relationship between regulations and firms’ performance under the strong law enforcement. Columns 4-6 report the effect of regulations on firms’ performance under the weak law enforcement. The coefficients for Post*Loan are positive and insignificant at 10% confidence level. It is 1.6% when we exclude other variables and becomes 1.5%% and 0.7% when include control variables. It means that the regulation on related-party loan guarantee generates a positive effect on firms’ performance but it is not significant. The coefficient for Gov*Post*Loan in column 4 is -2.5% with significance at 5% confidence level. This implies that the connections of firms to the government significantly offset the effect of regulations on firms’ performance under the weak law enforcement (e.g., the regulation on related-party loan guarantee is not strictly required by the CSRC). This result is robust when including variables of firms’ characteristics and other variables which can be observed in columns 5 and 6. The last four columns report the regression results when including the two regulations. The results for the regulation on preventing controlling shareholders from diverting firms’ assets do not change qualitatively. The coefficients for Post*Asset are positive and significant at 1% confidence level, while the coefficients for Gov*Post*Asset are insignificant even though become negative when include more control variables. These imply that, even controlling the effect of the regulation on related-party loan guarantee, the regulation on preventing controlling shareholders from diverting firms’ assets still has positive and significant on firms’ performance, and the political connections do not significantly affect the relationship between regulations and firms’ performance under strong law enforcement. For the regulation on related-party loan guarantee which has weak law enforcement, the coefficients of Post*Loan are still insignificant consistent with the results in columns 4-6, even they become negative. The coefficients of Gov*Post*Loan are slightly different, they become insignificant in most of the cases even they are still negative (e.g., -1.5% or -0.7%). The possible reason is that the regulation on preventing 21 controlling shareholders from diverting firms’ assets is strictly enforced by the CSRC, while the regulation on related-party loan guarantee is not and then the first one dominates the second one in affecting the difference in firms’ performance between firms with and without connections to the government. In sum, we conclude that the political connections negatively affect the effects of the regulations on firms’ performance, especially when the regulations are not strongly enforced by the authorities. 7. Mechanisms From the above analysis, we conclude that the regulations can improve firms’ corporate governance and then have positively effects on firms’ performance, and the relationship is affected by the law enforcement. In this section, we provide the mechanism from two perspectives. First, we study the effect of the regulations on the changes of ownerships held by controlling shareholders under different law enforcement. Second, we study firms’ investment change caused by regulations under different law enforcement. 7.1 Changes in Ownerships of Controlling Shareholders The regulations prohibit controlling shareholders from stealing the private profits at the cost of minority shareholders. In this section, we examine whether the controlling shareholders reduce their ownerships of the firm in response to the regulations. Table 10 reports the regression results of the changes in ownerships held by the largest shareholder affected by the regulations under different law enforcement. The first four columns provide the influence of prohibiting controlling shareholders from diverting firms’ assets on the ownerships held by the largest shareholder. The first column shows that the largest shareholder reduces 0.8% ownerships of the firm in response to the regulation. One possible reason is that the regulation decreases the possibility to pursue the private benefits and the best strategy is to 22 reduce the ownerships of the firm. An alternative explanation is that the controlling shareholders have the incentive to decrease the ownerships of the firm, while they are not allowed because of the share split structure. The reform of share split structure started from 2005 makes this possible. To remove this possibility, we solve the potential problem from two sides. First, the reform influenced all listed firms including the treatment and control group. Second, we control the effect of the reform of share split structure and run the above regression again. The second column shows that the largest shareholder reduces 1.5% of the ownership in response to the regulation and it is significant at 1% confidence level. An interesting byproduct is that the reform of share split structure caused the largest shareholder to reduce 6.9% of the ownership. The reform of share split structure explains more of the ownership decreases by controlling shareholders than the regulation on asset diversion by controlling shareholders although both are significant at 1% confidence level. When we control the variables of firms’ characteristics and other variables, the regression results are illustrated in columns 3 and 4, and we find that the above conclusion does not change qualitatively The last four columns report the effect of regulations on the ownership changes by the largest shareholder under different law enforcement. The fifth column shows that the regulation on asset diversion significantly decreases 0.8% ownerships of the largest shareholder while the effect of the regulation of related-party loan guarantee is 0.3% and it is not significant. As we discussed in the section 6, the regulation on asset diversion has stronger law enforcement than the regulation on related-party loan guarantee. Therefore, the fifth column demonstrates that the regulations with strong law enforcement have more significant effects on the ownership by controlling shareholders than the regulations with weak law enforcement. When we control the variables of firms’ characteristics and other variables, the conclusion does not change. The results are reported in columns 6-8. In summary, the controlling shareholders reduce their ownerships of the firm when the regulations prevent them from stealing firms’ profits, and then firms’ performance is increased. 23 In addition, this effect is more significant under the stronger law enforcement. 7.2 The Effect of Law and its Enforcement on Firms’ Investments Albuquerque and Wang (2008) find that, when the minority shareholders cannot be protected well with the weak governance, the controlling shareholders have the incentive to invest to pursue private benefits at the cost of minority shareholders. However, the models of Albuquerque and Hopenhayn (2004) and Demarzo and Fishman (2007) generate underinvestment which relaxes the conflicts between the investor and the manager. The following analysis demonstrates that the underinvestment exists in the firms of the treatment group and they increase their investments to improve firms’ performance when imposed the regulations in the Chinese market. 7.2.1 Preliminary Analysis Since the regulation on the related-party loan guarantee has insignificant effect on firms’ performance, we mainly study the changes in investment caused by the regulation on asset diversion. Figure 4 plots the pattern of median investment (above) and industry-adjusted (below) for the treatment and the control group. The above picture shows that the treatment group and the control group have the similar patters before 2005 when the regulation was announced, while they show the difference after 2005. The median investment of the treatment group increases from 2.1% in 2005 to 2.7% in 2007 with the increase of 0.6%, while the median investment of the control group decreases from 4.6% in 2005 to 4.2% in 2005 with the decrease of -0.4%. Therefore, the investment of the treatment group is increased 1% relative to the control group caused by the regulation of asset diversion. As we mentioned, the controlling shareholders with the weak investor protection have the incentive to overinvest. However, this doesn’t apply for Chinese market. The below picture plots 24 the changes of industry-adjusted investment and demonstrates that the median investment of the treatment group is below the average level of investment when adjusted the industry effect. The industry-adjusted median investment of the treatment group changes from -1.5% in 2005 to -0.6% in 2007 with the increase of 0.5%, and it changes for the control group from 0.6% in 2005 to 0.2% in 2007 with the decrease of 0.4%. It means that the treatment group has underinvestment relative to the control group, they have the incentive to increase their investment and then firms’ relative performance of the treatment group to the control group is increased. 7.2.2 Empirical Results We examine the effect of regulations on firms’ investment under different law enforcement. Table 11 reports the regression results. The first column show that the average investment for the treatment group relative to the control group is significantly increased 1.2% at 1% confidence level when imposed the regulation on asset diversion. When we control the variables of firms’ characteristics and other variables, the results don’t have significant change. To investigate the role of law enforcement on the relationship between regulations and firms’ investment, we include the regulation on related-party loan guarantee which has weak law enforcement into the regressions and report the results in the last three columns. Unlike the results for the regulation on asset diversion which has the strict law enforcement, the regulation with the weak law enforcement significantly decreases the relative investment of the treatment group relative to the control group 1% at 5% confidence level as showed in columns 4 and 5. When we include more control variables including firms’ characteristics, its effect becomes significant even it is still negative. In summary, the regulation with the strict law enforcement significantly increases the relative firms’ investment of the treatment group relative to the control group, while the regulation with the weak law enforcement does not have positive effect on firms’ investment. 25 8. Conclusion This paper empirically studies the relationship between the regulations on controlling shareholders and firms’ performance. Using a natural experiment in Chinese market, we find that firms’ performance is significantly increased when controlling shareholders are prohibited from diverting firms’ assets and related-loan loan guarantee. In particular, this relationship becomes stronger under the strict law enforcement. We provide two mechanisms to explain these results. When imposed the regulations, the controlling shareholders reduce their ownerships of the firm and at the same time have the incentive to increase their investment, and then firms’ performance is increased. These results are also stronger for the regulations with the strict law enforcement. 26 References Abadie, A. and G. Imbens. 2002. “Simple and Bias-Corrected Matching Estimators. Working paper, Department of Economics,” Working Paper at UC Berkeley. Albuquerque, R. and H. Hopenhayn, 2004, “Optimal Lending Contracts and Firm Dynamics,” Review of Financial Studies 71, 285-315. Albuquerque, R and N. Wang, 2008, “Agency Conflicts, Investment, and Asset Pricing”, Journal of Finance 63, 1-40. Allen, F. J. Qian, and M. Qian, 2005, “Law, Finance, and Economic Growth in China,” Journal of Financial Economics 77, 57-116. Allen, F. J. Qian, and M. Qian, 2008, “China’s Financial System: Past, Present, and Future,” Ch. 14 of China’s Great Economic Transformation, L. Brandt and T. Rawski, eds., Cambridge University Press Atanasov, Vladimir, 2005, “How Much Value can Blockholders Tunnel? Evidence from the Bulgarian Mass Privatization Auctions” Journal of Financial Economics, 76, 191-234. Barclay, M. J., and C. G. Holderness, “Private Benefits from Control of Public Corporations”, Journal of Financial Economics, 25, 371-395. Barclay, M. J., C. G. Holderness, and J. Pontiff, 1993, “Private Benefits from Block Ownership and Discounts on Closed-end Assets”, Journal of Financial Economics, 33, 263-291. Bebchuk, Lucian A., and Alma Cohen, 2005, “The Costs of Entrenched Boards”, Journal of Financial Economics, 78, 409-433. Bertrand, Marianne, Esther Duflo, and Sendhil Mullainathan, 2004, “How Much Should We Trust Differences-in-Differences Estimates?” Quarterly Journal of Economics, 119, 249-275. Bertrand M. and S. Mullainathan, 2003, “Enjoying the Quiet Life? Corporate Governance and Managerial Preferences”, Journal of Political Economy, 111, 1043-1075. Brockman, Paul, and Dennis Y. Chung, 2003, “Investor Protection and Firm Liquidity”, Journal of Finance, 58, 921-937. Claessens, Stijn, Simeon Djankov, and Larry H. P. Lang, 2000, “The separation of ownership and control in East Asian Corporations”, Journal of Financial Economics, 58, 81-112. Claessens, Stijn, Simeon Djankov, Joseph P. H. Fan, and Larry H. P. Lang, 2002,” Disentangling the Incentive and Entrenchment Effects of Large Shareholdings”, Journal of Finance, 57, 2741-2771. 27 Cremers, K. J. Martijn and Vinay B. Nair, 2005, “Governance Mechanisms and Equity Prices”, Journal of Finance, 60, 2859-2894. Demarzo, P.M. and M.J. Fishman, 2007, “Agency and Optimal Investment Dynamics,” Review of Financial Studies 20, 151-188. Dyck, Alexander and Luigi Zingales, 2004, “Private Benefits of Control: An International Comparison”, Journal of Finance, 59, 537-600. Gompers, P., J. Ishii and A. Metrick, 2003, “Corporate Governance and Equity Prices”, Quarterly Journal of Economics 118 107–155. Jensen, M., and W., Meckling, 1976, “Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure”, Journal of Financial Economics, 3, 305–360. Johnson, S., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer, 2000, “Tunneling”, American Economic Review, 90, 22–27. La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny, 1997, “Legal Determinants of External Finance”, Journal of Finance 52, 1131-1150. La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny, 1998, “Law and Finance”, Journal of Political Economy 106, 1113-1155. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 1999, “Corporate Ownership around the World”, Journal of Finance, 54, 471–518. La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny, 2000, “Investor Protection and Corporate Governance”, Journal of Financial Economics 58, 3-27. La Porta, R., F. Lopez-De-Silanes, A. Shleifer and R. Vishny, 2002, “Investor Protection and Corporate Valuation”, Journal of Finance, 57, 1147–1170. Lemmon Michael L., and Karl V. Lins, 2003, “Ownership Structure, Corporate Governance, and Firm Value: Evidence from the East Asian Financial Crisis”, Journal of Finance, 58, 1445-1468. Lins, Karl V., 2003, “Equity Ownership and Firm Value in Emerging Markets”, Journal of Financial and Quantitative Analysis, 38, 159-184. Meyer, Bruce D., 1995, “Natural and Quasi-experiments in Economics”, Journal of Business & Economic Statistics, 13, 151–161. Nenova, Tatiana, 2003, “The Value of Corporate Voting Rights and Control: A Cross-country Analysis”, Journal of Financial Economics, 68,325-351. Yermack, D., 1996, “Higher Market Valuation for Firms with a Small Board of Directors”, Journal of Financial Economics, 40, 185–211 28 0.09 0.08 median of control group 0.07 ROA 0.06 0.05 median of treated group 0.04 0.03 0.02 0.01 year 0 2003 2004 2005 2006 2007 Treated and control graphs for ROA (fund occupation) Figure 1: Median ROA (return on assets) for firms in which the controlling shareholders diverted assets before 2006 (treatment group) and for firms in which the controlling shareholders didn’t (control group). 29 ROA 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 median of control group median of treated group 2003 2004 2005 2006 2007 year Government Treated and control graphs for ROA (fund occupation) 0.08 0.07 median of control group 0.06 ROA 0.05 median of treated group 0.04 0.03 0.02 0.01 0 2003 2004 2005 2006 2007 YEAR Non-government treated and control graphs for ROA (fund occupation) Figure 2: Median ROA (return on assets) for firms in which the controlling shareholders diverted assets before 2006 (treatment group) and for firms in which the controlling shareholders didn’t (control group). The top picture is for firms in which government is the eventual controlling shareholder, and the bottom one is for firms in which government is not the eventual controlling shareholder. 30 ROA 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 median of control group median of treated group 2003 2004 2005 2006 2007 year ROA Government treated and control graphs for ROA (loan guarantee) 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 median of control group median of treated group 2003 2004 2005 2006 2007 year Non-government treated and control graphs for ROA (loan guarantee) Figure 3: Median ROA (return on assets) for firms with related-party loan guarantee before 2006 (treatment group) and for firms without related-party loan guarantee (control group). The top picture is for firms in which government is the eventual controlling shareholder, and the bottom one is for firms in which government is not the eventual controlling shareholder. 31 0.07 median of control group investment 0.06 0.05 median of treated group 0.04 0.03 0.02 0.01 year 0 2003 2004 2005 2006 2007 Treated and control graphs for investment(fund occupation) 0.01 median of control group investment 0.005 0 -0.005 2003 2004 2005 2006 2007 median of treated group -0.01 -0.015 year -0.02 Treated and control graphs for investment (adjusted by industry median) Figure 4: Median investments for firms in which the controlling shareholders diverted assets before 2006 (treatment group) and firms in which the controlling shareholders didn’t (control group). The investment in the top picture is not adjusted by industry median, and the investments in the bottom one is adjusted by industry median. 32 Table 1: Summary of statistics (2003-2007) The table reports the firm characteristics by categorizing into the treatment group and the control group. In Panel A, controlling shareholders in the treatment group diverted firms’ assets at in 2005, while the firms in the control group did not. In Panel B, firms in the treatment group have the related-party loan guarantees in 2005, while the firms in the control group did not. The stock price and related-party loan guarantee are obtained from the WIND database; the information about actual controlled shareholders is obtained from the CCER database; the diversion of assets (by controlling shareholders) is manually collected; and the other financial data are obtained from the CSMAR database. ROA=EBIT/assets; ROS=EBIT/sales; ROE=EBIT/equity= EBIT/(Asset-Debt); EPS=earnings/the total number of share; Leverage=debt/total asset; Tangible=fixed assets/total assets. ***, **, and * denote significance at 1%, 5%, and 10%, respectively. Full Samples N Mean Treatment Group % of total Mean Control Group Difference % of total Mean Mean Panel A: Diversion of assets ROA 5599 0.052 34.08 0.026 65.92 0.066 -0.040*** ROS 5589 0.138 34.00 0.117 66.00 0.149 -0.033 ROE 5574 0.045 34.14 0.007 65.86 0.064 -0.057*** EPS 5599 0.157 34.08 -0.007 65.92 0.242 -0.249*** Investment 5568 0.067 33.87 0.049 66.13 0.076 -0.027*** Leverage 5598 0.569 34.07 0.670 65.93 0.517 0.153*** Log(Assets) 5599 21.24 34.08 21.00 65.92 21.37 -0.373*** Tangible 5598 0.299 34.07 0.293 65.93 0.302 -0.009* Change of stock price (%) 5593 51.13 34.02 45.37 65.98 54.11 -8.73*** Turnover of CEO 5599 0.333 34.08 0.387 65.92 0.305 0.082*** Turnover of Control Right 5599 0.039 34.08 0.039 65.92 0.039 0.000 5599 0.390 34.08 0.383 65.92 0.393 -0.009** 5596 0.515 34.06 0.510 65.94 0.518 -0.007* Concentration of largest shareholder Concentration of first three largest shareholders Panel B: Related-party loan guarantees ROA 5599 0.052 63.15 0.046 36.85 0.063 -0.018*** ROS 5589 0.138 63.16 0.133 36.84 0.147 -0.014 ROE 5574 0.045 63.35 0.035 36.65 0.062 -0.027** EPS 5599 0.157 63.15 0.114 36.85 0.231 -0.117*** Investment 5568 0.067 63.09 0.064 36.91 0.071 -0.008*** Leverage 5598 0.569 63.15 0.624 36.85 0.476 0.148*** Log(Assets) 5599 21.24 63.15 21.24 36.85 21.26 -0.022 Tangible 5598 0.299 63.15 0.288 36.85 0.319 -0.031*** Change of stock price (%) 5593 51.13 63.11 49.94 36.89 53.17 -3.24 Turnover of CEO 5599 0.333 63.15 0.333 36.85 0.331 0.002 Turnover of Control Right 5599 0.039 63.15 0.038 36.85 0.041 0.003 5599 0.390 63.15 0.369 36.85 0.424 -0.054*** 5596 0.515 63.17 0.498 36.83 0.545 -0.047*** Concentration of largest shareholder Concentration of first three largest shareholders 33 Table 2 Distributions of diversion of assets and related-party loan guarantees（2003-2007） The table presents the distribution of diversion of assets and related-party loan guarantees for full samples (Panel A), classified by industry (Panel B) and classified by the type of controlling shareholders (Panel C). Panel A Full samples Treatment (asset diversion) Control Total Treatment (loan guarantee) 1322 2214 3536 Control 586 1477 2063 Total 1908 3691 Panel B Classifications by industry Asset diversion Related-party loan guarantee Number Treatment % Treatment % Agriculture 122 62 50.8 87 71.3 Manufacturing 3428 1239 36.1 2106 61.4 Housing 103 37 35.9 65 63.1 Transportation 248 62 25 113 45.5 Information 378 134 35.4 245 64.8 Trade 419 90 21.4 289 68.9 Real estate 232 50 21.5 143 61.6 Services & Culture 225 74 32.9 150 66.7 Conglomerate 444 160 36 338 76.1 Total 5599 1908 34.1 3536 63.1 Panel C: Classifications by controlling shareholders Asset diversion Related-party loan guarantee Number Treatment % Treatment % Government 3831 1217 31.8 2343 61.2 Non-Government 1754 688 39.2 1183 67.4 Total 5585 1905 34.1 3526 63.1 34 Table 3: The effect of law on the ROA The table reports the regression results for the following model ROAit  a( Postit * Fundit )  b(controlsit )  Firmi  Yeart   it where Post*Asset denotes the asset diversion by controlling shareholders in 2005, and controls include variables of log(assets), tangible, square of log(assets), leverage, log(sales), percentage change of stock price, government (the actual controller type), ST share, turnover of CEO, the change of the control right, the share ratio hold by the first large shareholder, and the share ratio hold by the first three large shareholders; Firm and Year denote firm- and year-fixed effect, respectively; The variable of SPLIT equals 1 starting from the year of structural reform, otherwise equals 0. The numbers in the brackets are the standard deviations. ***, **, and * mean that the differences between affected and unaffected groups are significance at the confidence levels of 1%, 5%, and 10%, respectively. (1) Post*Asset 0.033 (2) *** (0.008) log(assets) 0.039 (3) *** (0.009) Leverage (0.007) 0.026 (5) *** (0.008) (6) *** 0.029*** (0.007) (0.008) 0.029 0.015 0.026* (0.013) (0.013) *** -0.083*** (0.033) (0.031) -0.021 -0.035 -0.103 Tangible 0.030 (4) *** (0.024) (0.024) *** -0.011 (0.010) (0.009) -0.026 Split Gov -0.018 -0.016 (0.012) (0.010) *** 0.061*** 0.043 ST (0.010) (0.010) Change of 0.026 *** 0.024*** stock price (%) (0.003) (0.003) -0.005 -0.005* (0.003) (0.003) 0.001 -0.003 (0.011) (0.012) 0.072 0.083* (0.046) (0.046) Ownership 0.175 *** 0.116** (first three largest) (0.052) (0.054) Turnover of CEO Turnover of control right Ownership (largest) Intercept Y Y Y Y Y Y Firm fixed effect Y Y Y Y Y Y Year fixed effect Y Y Y Y Y Y R-Square 0.399 0.413 0.402 0.408 0.441 0.469 adj R-Square 0.244 0.262 0.248 0.255 0.297 0.331 Number of obs 5599 5598 5599 5585 5590 5575 35 Table 4: Robust check: regression without 2005 year data The table reports the regression results for the following model ROAit  a( Postit * Fundit )  b(controlsit )  Firmi  Yeart   it excluding the year of 2005. where Post*Asset denotes the asset diversion by controlling shareholders in 2005, and controls include variables of log(assets), tangible, square of log(assets), leverage, log(sales), percentage change of stock price, government (the actual controller type), ST share, turnover of CEO, the change of the control right, the share ratio hold by the first large shareholder, and the share ratio hold by the first three large shareholders; Firm and Year denote firm- and year-fixed effect, respectively. The numbers in the brackets are the standard deviations. ***, **, and * mean that the differences between affected and unaffected groups are significance at the confidence levels of 1%, 5%, and 10%, respectively. Post*Asset (1) (2) (3) (4) (5) (6) 0.029*** 0.034*** 0.024*** 0.023*** 0.025*** 0.024*** (0.009) (0.010) (0.008) (0.008) (0.008) (0.009) log(assets) 0.013 0.023 (0.014) (0.015) ** -0.068* (0.038) (0.036) -0.009 -0.025 -0.091 Tangible Leverage (0.026) (0.026) ** -0.019 (0.015) (0.014) -0.034 Split Gov ST -0.021 -0.018 (0.014) (0.012) 0.033*** 0.049*** (0.011) (0.012) Change of *** 0.025 0.023*** stock price (%) (0.003) (0.003) -0.005 -0.006 (0.004) (0.004) -0.002 -0.006 (0.014) (0.015) 0.077 0.090* (0.053) (0.053) Ownership *** 0.187 0.122** (first three) (0.059) (0.060) Turnover of CEO Turnover of control right Ownership (largest) Intercept Y Y Y Y Y Y Firm fixed effect Y Y Y Y Y Y Year fixed effect Y Y Y Y Y Y R-Square 0.392 0.400 0.395 0.399 0.436 0.455 adj R-Square 0.183 0.194 0.188 0.192 0.240 0.264 Number of obs 4468 4467 4468 4454 4459 4444 36 Table 5: Robust check: alternative measures of firm performance The table reports the regression results for the ROE, ROS and EPS. Post*Asset denote asset diversion by controlling shareholders (existed by the end of 2005) after regulation issued in 2005, and controls include variables of log(assets), tangible assets , leverage, log(sales), Split share structure reform, percentage change of stock price, government (actual controller type), ST share, turnover of CEO, change of control right, share ratio hold by largest shareholder, and share ratio hold by first three large shareholders; Firm and Year denote firm- and year-fixed effect, respectively. The numbers in the brackets are the standard deviations. ***, **, and * denote significance at 1%, 5%, and 10% levels, respectively. Post*Asset (1) 0.091*** (0.034) ROE (2) 0.090*** (0.033) log(assets) Tangible Leverage Split Gov ST *** Change of stock price (%) Turnover of CEO Turnover of control right Ownership (largest) Ownership (first three) Intercept Firm fixed Year fixed R-Square adj R-Square Number of obs Y Y Y 0.207 0.003 5574 0.001 (0.000) -0.011 (0.016) 0.019 (0.050) 0.066 (0.150) 0.359** (0.182) Y Y Y 0.216 0.013 5565 (3) 0.059* (0.032) -0.034 (0.037) -0.072 (0.112) -0.036 (0.053) 0.041 (0.040) -0.033 (0.057) 0.213*** (0.057) 0.001*** (0.000) -0.017 (0.016) 0.002 (0.049) 0.077 (0.155) 0.433** (0.179) Y Y Y 0.228 0.026 5551 (4) *** ROS (5) (6) EPS (8) (7) 0.262 *** 0.229 0.074 (0.088) (0.082) (0.069) ** (9) 0.057 0.046 * 0.056** (0.029) (0.027) (0.026) 0.124*** -0.240 (0.179) (0.038) -0.136 -0.361*** (0.449) (0.101) 0.397 -0.208*** (0.380) (0.063) -0.022 -0.041 (0.076) (0.030) -0.048 -0.057 (0.118) (0.037) 0.140 0.284*** (0.089) (0.040) 0.147*** 0.156*** 0.086*** 0.080*** (0.055) (0.055) (0.010) (0.010) 0.025 0.008 *** -0.046 -0.048*** (0.034) (0.032) (0.013) (0.012) 0.048 0.015 -0.008 -0.025 (0.133) (0.125) (0.033) (0.034) ** ** 0.014 0.056 (0.347) (0.320) (0.162) (0.165) 0.602 0.481 0.669*** 0.436** (0.515) (0.486) (0.181) (0.201) Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y 0.185 0.201 0.221 0.537 0.561 0.593 -- -- 0.017 0.419 0.448 0.487 5589 5580 5565 5599 5590 5575 0.686 37 0.733 Table 6: Robust check: propensity-score matching estimates FD_ROA=Average ROA before 2005(including 2005) - Average ROA after 2005.ROA and ROS are EBIT scaled by assets and sales respectively, ROE and EPS are net profits scaled by equity and number of shares respectively. N is the number of the matched. Model 1 estimates the propensity score based on assets for each industry in 2005. Model 2 estimates the propensity score based on assets and loan-guarantee for each industry in 2005. Model 3 estimates the propensity score based on assets and sales for each industry in 2005. Model 4 estimates the propensity score based on assets, sales, leverage and tangible for each industry in 2005. Model 5 estimates the propensity score based on assets, sales, leverage, tangible, loan-guarantee and actual controller for each industry in 2005.Standard Error in parentheses, ***, **, and * mean that the ATE between affected and unaffected groups are significance at the confidence levels of 1%, 5%, and 10%, respectively. Model (1) Model (2) Model (3) Model (4) Model (5) Panel A: FD_ROA (N =1) (N =2) (N =4) *** 0.023*** 0.035*** 0.017** 0.017** (0.008) (0.008) (0.008) (0.007) (0.007) *** *** *** ** 0.021*** 0.032 0.030 0.026 0.032 0.017 (0.008) (0.008) (0.008) (0.007) (0.007) *** *** *** *** 0.030 (0.008) 0.026 0.028 0.019 0.023*** (0.007) (0.007) (0.007) (0.007) Panel B: FD_ROS (N =1) (N =2) (N =4) * 0.203* 0.211* 0.033 0.064 (0.108) (0.172) (0.114) (0.108) (0.104) ** ** * 0.187 0.045 0.105 0.192 0.219 0.214 (0.106) (0.110) (0.100) (0.100) (0.100) ** ** * 0.158 0.094 0.121 (0.096) (0.098) (0.099) 0.212 0.227 (0.108) (0.110) Panel C: FD_ROE (N =1) (N =2) (N =4) ** 0.067 0.101* 0.078 0.075 (0.062) (0.062) (0.058) (0.054) (0.052) * 0.065 0.087 0.065 0.115** (0.058) (0.059) (0.054) (0.055) (0.054) ** ** ** * 0.086 0.133** (0.057) (0.052) (0.053) 0.129 0.096 0.124 0.121 (0.059) (0.058) 0.146 Panel D: FD_EPS (N =1) (N =2) (N =4) *** 0.047 0.102*** 0.033 0.023 (0.035) (0.034) (0.033) (0.030) (0.030) *** ** *** 0.025 0.040 0.090 0.088 0.064 0.081 (0.032) (0.031) (0.031) (0.028) (0.027) ** * ** 0.023 0.041 (0.027) (0.027) 0.071 0.060 0.073 (0.031) (0.031) (0.030) 38 Table 7: Cumulative abnormal returns sorted on firms’ characteristics The table reports the average cumulative abnormal returns (CAR) for various windows. The announcement of regulations of asset diversion occurred in June 2005, the CARs are calculated from m months (negative) before and n months (positive) after the regulation announcements. The final date for the regulation to be executed is December 2005, which corresponding to 6 months after the regulation announcement. A two-tail t-test is executed, ***, **, and * denote significance at 1%, 5%, and 10% levels, respectively. Asset Diversion Treatment (1) Control (2) (1)-(2) [0, 8] 0.026 -0.046 0.072*** [0, 12] 0.075 -0.028 0.103*** [-5, 12] 0.003 -0.046 0.049* [-8, 12] 0.068 -0.011 0.078*** Panel A: Full Samples Panel B: Excluding firms with H shares [0, 8] 0.026 -0.041 0.067*** [0, 12] 0.077 -0.017 0.095*** [-5, 12] 0.005 -0.040 0.045 [-8, 12] 0.069 -0.005 0.074*** [0, 8] -0.019 -0.062 0.043* [0, 12] 0.037 -0.051 0.088** [-5, 12] 0.005 -0.060 0.065** [-8, 12] 0.063 -0.019 0.082*** Panel C: Excluding with ST shares Panel D: State-Owned Enterprises (SOEs) [0, 8] 0.045 -0.040 0.085*** [0, 12] 0.119 -0.021 0.141*** [-5, 12] 0.047 -0.043 0.090*** [-8, 12] 0.098 -0.010 0.107 Panel E: None State-Owned Enterprises (Non-SOEs) [0, 8] -0.008 -0.061 0.052* [0, 12] -0.006 -0.043 0.037 [-5, 12] -0.078 -0.054 -0.024 [-8, 12] 0.012 -0.013 0.025 39 Table 8: The effect of law enforcement on corporate performance (ROA) The table reports the regression results for the following model ROAit  a( Postit * Fund )  b( Postit * Loan)  c(controlsit )  Firmi  Yeart   it where Post*Asset and Post*Loan denote asset diversion by controlling shareholders and related-party loan guarantee in 2005, and controls include variables of log(assets), tangible assets , leverage, log(sales), Split share structure reform, percentage change of stock price, government (actual controller type), ST share, turnover of CEO, change of control right, share ratio hold by largest shareholder, and share ratio hold by first three large shareholders; Firm and Year denote firm- and year-fixed effect, respectively. The numbers in the brackets are the standard deviations. ***, **, and * denote significance at 1%, 5%, and 10% levels, respectively. (1) (2) (3) (4) 0.001 0.003 -0.002 -0.004 (0.007) (0.007) (0.006) (0.006) Post*Asset Post*Loan 0.008 log(assets) Tangible Leverage 0.020 (0.013) (0.013) -0.105*** -0.085*** (0.033) (0.031) -0.020 -0.035 (0.024) (0.024) (6) 0.039*** (0.009) 0.001 (0.007) 0.016 (0.013) -0.103*** (0.033) -0.021 (0.024) (7) 0.029*** (0.007) -0.005 (0.006) -0.013 Split (0.009) -0.015 Gov (0.010) 0.064*** ST (0.010) Change of 0.026*** 0.025*** stock price (%) (0.003) (0.003) -0.005 -0.006* (0.003) (0.003) 0.002 -0.003 (0.011) (0.012) 0.068 0.080* (0.047) (0.047) *** 0.183 0.122** Turnover of CEO Turnover of control right Ownership (largest) Ownership (first three) Intercept Firm fixed effect Year fixed effect R-Square adj R-Square Number of obs (5) 0.033*** (0.008) -0.002 (0.007) (0.054) (0.054) Y Y Y Y Y Y Y Y Y Y Y Y 0.393 0.406 0.437 0.465 0.237 0.253 0.292 0.326 5599 5598 5590 5575 40 Y Y Y 0.399 0.244 5599 Y Y Y 0.413 0.262 5598 0.026*** (0.003) -0.005 (0.003) 0.001 (0.011) 0.073 (0.046) 0.174*** (0.052) Y Y Y 0.441 0.297 5590 (8) 0.029*** (0.008) -0.005 (0.006) 0.025* (0.013) -0.083*** (0.031) -0.035 (0.024) -0.010 (0.009) -0.016 (0.010) 0.061*** (0.010) 0.024*** (0.003) -0.005* (0.003) -0.003 (0.012) 0.084* (0.046) 0.115** (0.054) Y Y Y 0.469 0.331 5575 Table 9: Political connections and law enforcement The table reports the regression results for the following model ROAit  a( Postit * Fund )  b( Postit * Loan)  c(Government * Postit * Fund )  d (Government * Postit * Loan)  (controlsit )  Firmi  Yeart   it where Post*Asset and Post*Loan denote asset diversion by controlling shareholders and related party loan guarantees in 2005, Government equals 1 if (actual) controlling shareholder is government, otherwise equals 0, Gov*Post*Asset and Gov*Post*Loan mean that firm existed asset diversion and related-party loan guarantee, and is finally controlled by government before 2005. Post*Asset Gov*Post*Asset (1) 0.051*** (0.015) -0.032* (0.017) (2) 0.043*** (0.013) -0.024 (0.016) (3) 0.041*** (0.014) -0.020 (0.015) Post*Loan Gov*Post*Loan (4) (5) (6) 0.016 (0.010) -0.025** (0.010) 0.015 (0.010) -0.025** (0.010) 0.007 (0.010) -0.016* (0.009) 0.026** (0.013) -0.081** (0.031) -0.037 (0.024) -0.011 (0.009) -0.012 (0.010) 0.060*** (0.010) log(assets) Tangible Leverage Split Gov ST (7) 0.045*** (0.015) 0.007 (0.010) -0.021 (0.018) -0.015 (0.010) (8) 0.053*** (0.016) 0.011 (0.010) -0.024 (0.017) -0.015* (0.009) 0.018 (0.013) -0.096** (0.033) -0.028 (0.024) (9) 0.040*** (0.014) -0.001 (0.010) -0.020 (0.017) -0.007 (0.009) (10) 0.039*** (0.014) -0.000 (0.010) -0.016 (0.016) -0.007 (0.009) 0.026* (0.013) -0.081** (0.031) -0.037 (0.024) -0.011 (0.009) -0.012 (0.010) 0.060*** (0.010) -0.028** (0.010) Change of 0.025*** 0.024*** 0.025*** 0.025*** 0.024*** stock price (%) (0.003) (0.003) (0.003) (0.003) (0.003) -0.005 (0.003) 0.000 (0.011) 0.066 (0.046) 0.174*** (0.052) Y Y Y 0.444 0.299 5576 * -0.005 (0.003) 0.002 (0.011) 0.058 (0.048) 0.186*** (0.054) Y Y Y 0.439 0.294 5576 -0.005 (0.003) 0.000 (0.011) 0.064 (0.047) 0.174*** (0.052) Y Y Y 0.444 0.299 5576 -0.005* (0.003) -0.003 (0.012) 0.076 (0.047) 0.116** (0.054) Y Y Y 0.470 0.331 5575 Turnover of CEO Turnover of control right Ownership (largest) Ownership (first three) Intercept Firm fixed effect Year fixed effect R-Square adj R-Square Number of obs Y Y Y 0.402 0.248 5585 -0.005 (0.003) -0.003 (0.012) 0.078* (0.046) 0.116** (0.054) Y Y Y 0.470 0.331 5575 Y Y Y 0.397 0.242 5585 Y Y Y 0.401 0.247 5585 41 Y Y Y 0.402 0.248 5585 Y Y Y 0.418 0.268 5584 Table 10 The effect of law and its enforcement on the ownership change of the largest shareholder The table reports the regression results for the ownership change of the largest shareholder. Post*Asset and Post*Loan denote asset diversion by controlling shareholders and related-party loan guarantee in 2005, and controls include variables of log(assets), tangible assets , leverage, log(sales), Split share structure reform, percentage change of stock price, government (actual controller type), ST share, turnover of CEO, change of control right, share ratio hold by largest shareholder, and share ratio hold by first three large shareholders; Firm and Year denote firm- and year-fixed effect, respectively. The numbers in the brackets are the standard deviations. ***, **, and * denote significance at 1%, 5%, and 10% levels, respectively. Post*Asset (1) -0.008* (0.005) (2) -0.015*** (0.004) (3) -0.008** (0.003) (4) -0.007** (0.003) Post*Loan log(assets) Tangible Leverage -0.069*** Split (0.005) Gov ST Change of stock price (%) Turnover of CEO Turnover of control right Ownership (largest) Ownership (first three) Intercept Firm fixed Year fixed R-Square adj R-Square Number of obs Y Y Y 0.274 0.030 4526 Y Y Y 0.339 0.117 4526 -0.001 (0.002) -0.002 (0.002) -0.000 (0.009) 0.644*** (0.031) 0.036 (0.032) Y Y Y 0.561 0.412 4517 0.017*** (0.004) 0.024* (0.013) 0.001 (0.004) -0.033** (0.005) 0.002 (0.005) 0.001 (0.005) 0.000 (0.002) -0.001 (0.002) 0.000 (0.009) 0.647*** (0.030) -0.048 (0.035) Y Y Y 0.577 0.431 4505 42 (5) -0.008* (0.005) 0.003 (0.004) (6) -0.015*** (0.004) 0.001 (0.004) (7) -0.008** (0.003) -0.002 (0.003) -0.069*** (0.005) Y Y Y 0.274 0.030 4526 Y Y Y 0.339 0.117 4526 -0.001 (0.002) -0.002 (0.002) -0.000 (0.009) 0.645*** (0.031) 0.036 (0.032) Y Y Y 0.561 0.412 4517 (8) -0.007** (0.003) -0.001 (0.003) 0.017*** (0.004) 0.024* (0.013) 0.001 (0.004) -0.033** (0.005) 0.002 (0.005) 0.001 (0.005) 0.000 (0.002) -0.001 (0.002) 0.000 (0.009) 0.647*** (0.030) -0.048 (0.035) Y Y Y 0.577 0.431 4505 Table 11: The effect of law and its enforcement on investment The table reports the regression results for the investment. Post*Asset and Post*Loan denote asset diversion by controlling shareholders and related-party loan guarantee in 2005, and controls include variables of log(assets), tangible assets , leverage, log(sales), Split share structure reform, percentage change of stock price, government (actual controller type), ST share, turnover of CEO, change of control right, share ratio hold by largest shareholder, and share ratio hold by first three large shareholders; Firm and Year denote firm- and year-fixed effect, respectively. The numbers in the brackets are the standard deviations. ***, **, and * denote significance at 1%, 5%, and 10% levels, respectively. Post*Asset (1) 0.012*** (0.005) (2) 0.012*** (0.004) (3) 0.022*** (0.005) Post*Loan (5) (6) *** 0.022*** (0.004) (0.005) ** -0.005 0.013 -0.010 (0.005) 0.037*** (0.006) -0.081*** (0.021) 0.018*** (0.005) -0.002 (0.005) -0.001 (0.005) -0.006 (0.004) log(assets) Tangible Leverage Split Gov ST Change of 0.008 stock price (%) Turnover of CEO Turnover of control right Ownership (largest) Ownership (first three) Intercept Firm fixed effect Year fixed effect R-Square adj R-Square Number of obs (4) 0.013*** (0.004) -0.010** (0.005) Y Y Y 0.544 0.426 5568 *** (0.005) 0.037*** (0.006) -0.080*** (0.021) 0.018*** (0.005) -0.002 (0.005) -0.001 (0.005) -0.006 (0.004) *** 0.006 *** 0.008 0.007*** (0.002) (0.002) (0.002) (0.002) -0.004** (0.002) -0.008* (0.005) 0.007 (0.034) 0.065* (0.035) Y Y Y 0.550 0.433 5559 -0.003* (0.002) -0.007 (0.005) 0.006 (0.033) 0.040 (0.035) Y Y Y 0.572 0.459 5544 -0.004** -0.003 (0.002) (0.002) 43 * -0.007 (0.005) (0.005) 0.010 0.007 (0.034) (0.033) 0.062* 0.039 (0.035) (0.035) Y Y Y Y Y Y 0.551 0.572 0.434 0.459 5559 5544 -0.008 Y Y Y 0.544 0.427 5568