Corporate Governance and Financial Reporting Quality: Evidence from the Split Share Structure Reform in China

Transcription

1 Corporate Governance and Financial Reporting Quality: Evidence from the Split Share Structure Reform in China Yan Luo, Xiaolin Qian, Jinjuan Ren Abstract We examine the changes in financial reporting quality after the split share structure reform in China. The non-tradable shares of controlling shareholders were prohibited from trading before the reform, resulting in severe incentive misalignment and controlling shareholders strong motivations to manipulate earnings. The reform removes this market friction by converting non-tradable shares into tradable shares, therefore aligns the interests of controlling and minority shareholders. We measure financial reporting quality based on five proxies including discretionary accrual, earnings smoothness, earnings persistence, value relevance, and conservatism. The empirical results show that firms financial reporting quality increases significantly after the reform, especially for firms with weaker pre-reform corporate governance and non-soes. The findings imply that corporate governance causally affects firms information environment, and shed light on the consequences of privatization in emerging markets. Key words: Corporate governance, Financial reporting quality Corresponding author. School of Management, Fudan University, Shanghai, P.R.China; tel: (86-21) ; fax: (86-21) ; Faculty of Business Administration, University of Macau, Macau; tel: (853) ; fax: (853) ; Faculty of Business Administration, University of Macau, Macau; tel: (853) ; January 15, 2015

2 Corporate Governance and Financial Reporting Quality: Evidence from the Split Share Structure Reform in China Abstract We examine the changes in financial reporting quality after the split share structure reform in China. The non-tradable shares of controlling shareholders were prohibited from trading before the reform, resulting in severe incentive misalignment and controlling shareholders strong motivations to manipulate earnings. The reform removes this market friction by converting non-tradable shares into tradable shares, therefore aligns the interests of controlling and minority shareholders. We measure financial reporting quality based on five proxies including discretionary accrual, earnings smoothness, earnings persistence, value relevance, and conservatism. The empirical results show that firms financial reporting quality increases significantly after the reform, especially for firms with weaker pre-reform corporate governance and non-soes. The findings imply that corporate governance causally affects firms information environment, and shed light on the consequences of privatization in emerging markets. Key words: Corporate governance, Financial reporting quality January 15, 2015

3 1. Introduction In this paper, we examine the influence of corporate governance on firms financial reporting quality in the context of the 2005 privatization event in China, which is also called the split share structure reform. A wave of failures of corporate governance as well as astonishing financial frauds results in a decline in investors confidence in the integrity of accounting information, which has caused much attention and is calling for remedies. In response to these failures, the U.S. passed the Sarbanes-Oxley Act, which was believed to contain some of the most far reaching changes to the business world in the U.S.. And the expectation that corporate governance and financial reporting quality are strongly related underlies several provisions of the Act (Hoitash, Hoitash, Bedard (2009)). The corporate governance practice and information quality raise even more concerns in emerging markets with weak legal protection of investors rights. Whether the privatization process in emerging markets like China, where improving corporate governance is an important consideration, helps to improve financial reporting quality is a question worth exploring, but has received inadequate investigation. In this paper, we attempt to fill this void by examining the influence of China s split share structure reform on the financial reporting quality of Chinese listed firms. Numerous studies have investigated the credibility or quality of earnings associated with corporate governance (e.g., Warfield, Wild, and Wild (1995), Beasley (1996), Klein (2002), Fan and Wong (2002), Gul et al. (2003), Vafeas (2005), Farber (2005), Larcker et al. (2007), Armstrong, Jagolinzer, and Larcker (2010b)). The evidence, however, is inconclusive. Moreover, very few papers present convincing evidence on the causal effect between the two. The difficulty arises as corporate governance and financial reporting are both endogenous firm-choices. As a result, financial reporting quality could affect corporate governance and at the same time be affected by it. The limited variation in governance over time further adds to the difficulty in identifying 2

4 the causal relationship. Hence, even though prior studies have documented various correlations between governance and financial reporting quality, researchers might erroneously conclude the causal relation between them (Brickley and Zimmerman (2010)). Armstrong, Guay, and Weber (2010) therefore encourage researchers to make use of exogenous shocks to develop research settings where the causality direction could be identified. We take advantage of a unique exogenous setting to investigate the causal effect of corporate governance on firms financial reporting quality: the split share structure reform in China. Prior to the Chinese split-share structure reform, all listed Chinese firms had two types of shareholders: the shareholders of tradable shares and the shareholders of non-tradable shares. As a result of the split share structure, about two-thirds of domestically listed A-shares were not tradable. The non-tradable shares, which constitute a significant proportion of stocks in the market, are largely held by the state and legal persons. The tradable shares, on the other hand, are held by individual and institutional investors. The split-share structure results from the partial privatization of stated owned enterprises (SOE) by Chinese government starting in 1978, which allows employees and domestic institutions to own shares of SOEs. However, to conform to the communist public ownership, the majority of the shares of SOEs are concentrated in the hands of the government and are prohibited from public trading. The split-share structure creates heavy concentration of ownership in the hands of the non-tradable shareholders. The Chinese government has long realized that the predominance of non-tradable shares hinders the development and expansion of Chinese stock market, as well as the transformation of centralized economies into market economies. On April 29, 2005, the Chinese government launched the split share structure reform to convert the nontradable shares into shares that could be traded freely during a specified time period. According to China Securities Regulatory Commission (CSRC hereafter), by the end of 2007, over 97% of Chinese A-share listed firms, in terms of market capitalization, 3

5 has completed the reform. One key feature of the split share structure reform is that it converts non-tradable shares into tradable shares, which significantly improves stock market liquidity as well as the market s pricing function. Moreover, it allows the nontradable shareholders to benefit from the improvement in firm performance, therefore aligns the interests between controlling and minority shareholders. The split share structure reform provided an exogenous shock to firms corporate governance system. And it only affected the tradability of previously non-tradable shares, and therefore the incentives of controlling shareholders, but has no influence on factors unrelated to governance, such as operating, financing, or investing (Chen et al. (2012)). Thus, it offers an opportunity for us to examine whether improvement in corporate governance helps to promote financial reporting quality. We expect the split share structure reform to affect firms financial reporting quality through altering controlling shareholders incentives. The shareholders of nontradable shares, which used to account for around 70% of total shares outstanding in the market, are typically controlling shareholders. These controlling shareholders play a significant governance role as they have both the incentives and ability to monitor the managers. Nevertheless, the nontradability of their shares creates severe agency problems as the interests of controlling and minority shareholders are not aligned. Before the split share structure reform, the wealth of controlling shareholders is irrelevant of firms market value, therefore they have less incentives to maximize shareholders value. Moreover, as they cannot benefit from share price appreciation, the only means that they could realize a return on their investments is to extract gains from minority shareholders through expropriation or tunneling. The reform worked to reduce controlling shareholders incentives to take selfinterested actions through aligning their incentives with those of the minority shareholders and reducing agency conflicts. It presented an exogenous shock to corporate governance that increases controlling shareholders incentives to be concerned about share prices and to monitor managers behavior to enhance firm value, as their per- 4

6 sonal wealth is now closely tied to firm value. We thus expect controlling shareholders to be less likely to operate in secrecy but to be more active in promoting financial reporting quality, which would add to firm value through lowering information risk and cost of capital, improving investment efficiency, reducing market friction and price delay, and so on (e.g., Francis, LaFond, and Olsson (2004), Biddle, Hilary, and Verdi (2009), Callen, Khan, and Lu (2013)). The reform could also help to improve financial reporting quality through evoking the role of the stock market in monitoring the behavior of firm managers. Holmstrom and Tirole (1993) contend that market liquidity affects the value of market monitoring. Informed investors (other than controlling shareholders), such as speculators, will spend more time on monitoring only in liquid markets, where they could disguise their private information and make money on it. They further conjecture that a decrease in ownership concentration, which will result in a more liquid markets, could add to the value of market monitoring. By converting non-tradable shares into tradable shares, the reform is expected to increase stock market liquidity. As a result, speculators would have stronger incentives to monitor firms behaviors, and increasing the transparency of reported financial information. However, as discussed later, the reform in China imposes certain lock up periods on controlling shareholders to prolong their share sale process, to avoid significant price drop following a sharp increase in share supply. Therefore, although all nontradable shares are valued at the market price and are literally tradable after the reform, shareholders of these shares could not sell these shares within a short period of time. The lock up period requirement further tightens the alignment between controlling and minority shareholders, and we expect the monitoring effect from controlling shareholders to dominate the monitoring effect from other informed investors in the market, such as the speculators. In the empirical investigation, we measure financial reporting quality from five perspectives: discretionary accruals, earnings smoothness, earnings persistence, timely 5

7 loss recognition, and earnings response coefficient. The first four proxies focus on earnings properties, while the last one measures investor responsiveness to earnings. Our sample includes 2,066 firms during the period from 2000 to Application of the five proxies in the empirical investigation produces consistent results: the financial reporting quality of Chinese firms improves significantly after the split share structure reform. More specifically, the improvement is more evident for firms with weaker pre-reform corporate governance and non-soes. Our study contributes to the literature on corporate governance and financial reporting quality. Although there are many studies devoted to this topic, very few present convincing evidence on the causal effect between the two. One fundamental problem, as aforementioned, is that both corporate governance and financial reporting can be endogenous to firms operating environments that are unobservable, and could be affecting each other at the same time. Thus, as suggested by Armstrong et al. (2010), it is desirable to examine the casual effect in a setting with exogenous shocks. The split share structure reform in China provided a exogenous shock to firms corporate governance system, thus enable us to test the causality from corporate governance to financial reporting quality. By providing evidence that firms financial reporting quality is improved significantly after the reform, our study supports the conjecture that better corporate governance leads to higher financial reporting quality. This study also helps us to better understand the impact of privatization on the financial markets, and the findings in this paper could have implications for privatization in other emerging markets. Over the past decades, privatization has swept across developing countries, transferring decision-making responsibility from government to private hands. The removal of market frictions, following privatization, is expected to be associated with improved resources allocation and market efficiency. However, there are also cases where privatization, e.g., in Eastern Europe and Russia, privatization does not lead to successful outcomes. Therefore, understanding the consequences of privatization is of particular importance to countries considering 6

8 transitions of economies. The share issue privatization in China is the largest one in history so far, and has caused much attention worldwide. Liao et al. (2014) and Chen et al. (2012) have examined how the reform enhances the risk-sharing function of the market and affects corporate cash holdings through improving corporate governance. This paper adds to these studies by evaluating the success of the reform in terms of the improvement in financial reporting quality. The findings are compatible with the study of Chen et al. (2012), which documents that controlling shareholders are less likely to engage in tunneling activities after the reform. We further show that they have stronger incentives to participate in value-enhance activities after the reform, such as monitoring firm managers against opportunistic behavior, as the reform ties the wealth of controlling shareholders to firm market value. The reform therefore benefits investors by improving information quality, which is crucial to the development of financial markets. The remainder of the article is organized as follows. Section 2 reviews related literature. Section 3 describes institutional background and develops our research hypotheses. Section 4 introduces data and measurements. Section 5 present empirical results. Section 6 concludes the paper. 2. Literature review 2.1. Non-tradable shares, misaligned incentives, and corporate governance Before the split share structure reform in China, tradable and non-tradable shareholders have the same cash flow rights and voting rights, but the latter were unable to dispose their shares in open market transactions, which constituted a significant market friction. Under the split-share structure, the non-tradable shares were priced according to the book value of firm assets rather than market value of the firm. The different pricing mechanisms for tradable and non-tradable shares create significant incentive misalignment between controlling and minority shareholders. 7

9 The misaligned incentives of corporate insiders and outsiders could cause severe agency problems, and resulting in controlling shareholders tendency to purse their private benefits through expropriating the interests of minority shareholders (Shleifer and Vishny (1997), La Porta, Lopez-De-Silanes, and Shleifer (1999), Johnson, La Porta, Lopez de Silanes, and Shleifer (2000)). Some studies have discussed the corporate governance issues emerged under the split share structure in China. For instance, Liao et al. (2014) point out that as the interests of tradable and dominant non-tradable shareholders were fundamentally divided before the split share structure reform, the controlling shareholders were motivated to raise money relentlessly through seasoned cash offerings and extract value through related-party transactions, such as assets sales and product purchases with entities owned by the controlling shareholders, corporate lending, and so on. Chen et al. (2012) document that the cash holdings is high for listed firms before the reform, as controlling shareholders could only realize gains from cash distributions, including cash obtained from related transactions and tunneling. However, the cash holdings is significantly reduced after the reform, as controlling shareholders turn to care about share values and take value-maximizing actions for the firm, such as reducing their expropriating behaviors Corporate governance and financial reporting quality As controlling shareholders effectively control a firm, they control the firm s financial information production. Within firms with weak corporate governance, where the conflict of interests between controlling and minority shareholders is severe and controlling shareholders are strongly motivated to behave in secret in order to hide their private benefits of control, the information environment is expected to be poor and the credibility of financial information is expected to be low. Prior studies on the relation between corporate governance and financial reporting quality mainly evaluate the effectiveness of corporate governance based on factors such as characteristics of the board, managerial share ownership, managerial compensa- 8

10 tion, and managerial change. The evidence, however, is weak or mixed. For instance, some studies document that more dependent boards and high quality audit committee is associated with better information quality (e.g., Beasley (1996), Klein (2002), Vafeas (2005), Farber (2005)). However, Larcker, Richardson, and Tuna (2007) employ 14 indices for corporate governance factors, and find that they exhibit a mixed association with abnormal accruals and little relation with accounting restatements. The relation between managerial ownership or compensation and financial reporting quality is also inconclusive. Some find that a greater managerial ownership or equitybased managerial compensation helps to improve financial information quality (e.g., Warfield et al. (1995), Gul et al. (2003), Armstrong et al. (2010b)), while others find contradicting evidence (Smith (1976), Dhaliwal, Salamon, and Smith (1982), Lafond and Roychowdhury (2008), Cheng and Warfield (2005), Bergstresser and Philippon (2006), Efendi, Srivastava, and Swanson (2007)). Armstrong et al. (2010a) assert that studies on the relation between corporate governance and financial reporting quality are likely to be subject to the causality concerns. Establishing the direction of causality in the relation between financial reporting quality and corporate governance could be difficult because of the endogenous nature of firms economic characteristics and financial information environment. The causal relation could go in both direction: corporate governance could affect financial information quality and at the same time be affected by it. Brickley et al. (2010) thus contend that researchers are prone to erroneously conclude that an observed statistical relation between endogenous governance and financial reporting quality measures is causal. Armstrong et al. (2010a) further call for more careful consideration of endogeneity and causality when investigating governance-related issues, and suggest that one solution is to work with exogenous shocks to firms governance structure. 9

11 2.3. Our study in the context of existing literature We examine the effect of corporate governance on firms financial reporting quality in the context of the split share structure reform carried out in 2005 in China. This setting enables us to address the endogeneity issue and establish the direction of causality. The split share structure reform converted non-tradable shares to tradable shares, thereby align controlling shareholders interests with firm value. Such a change results from the government s attempt to remove a substantial market friction, and thus provides an exogenous shock to the corporate governance of affected firms by altering controlling shareholders incentives. We take advantage of this opportunity to test the causal impact of corporate governance on firms financial reporting quality. 3. Institutional background and hypothesis development 3.1. Institutional background China s stock exchanges, including both Shanghai stock exchange and Shenzhen stock exchange, were featured by a split share structure after established in the early 1990s. Under the split share structure, the shares of Chinese listed firms were classified into tradable and non-tradable shares (or restricted shares). The split share structure was a legacy of China s initial SOE partial privatization, and the original purpose for the creation of non-tradable shares was to maintain state control over listed companies. The non-tradable shares could be further classified into state shares and legal person shares. The state shares refers to shares held by governmental agencies or authorized institutions on behalf of the State, while the legal person shares refer to shares of a joint stock company owned by another (often governmentrelated) company or institution with a legal person status. Non-tradable shares enjoy the same voting and cash-flow rights as the tradable shares, but could not be freely traded on the stock markets. According to the statistics reported by CSRC, as of the end of 2004, 64% of the total billion shares (denominated in RMB) issued 10

12 by Chinese listed companies in the domestic markets were non-tradable, and 74% of these non-tradable shares were state-owned. The existence of non-tradable shares seriously jeopardized the development of Chinese stock markets. It imposed significant liquidity problem and distorted the pricing mechanism of the capital markets, as the non-tradable shares, which account for two-thirds of shares outstanding on average, were priced based on book value of assets rather than firm performance. This also caused serious corporate governance problems. As the holders of non-tradable shares, who were dominant shareholders, could not benefit from capital appreciation, they have little incentives to monitor the firm. Furthermore, they were found to behave opportunistically in seeking for private benefits at the expense of other stakeholders through tunneling in the form of relatedparty transactions or corporate loans to large shareholders (Chen et al. (2012), Liao et al. (2014)). Recognizing the negative impacts imposed by the split share structure, the CSRC initiated the split-share structure reform aiming to convert all non-tradable shares into tradable shares. Two pilot programs involving four and 42 listed companies were launched in May and June of 2005, respectively, The reform was further expanded to all listed firms in August 2005, and was completed for firms accounting for over 97% of total Chinese domestic market capitalization by the end of 2007 (Liao et al. (2014)). The timeline of a typical reform process is described in detail in prior studies, e.g., Liao et al. (2014) and Chen et al. (2012). First, the holders of non-tradable shares propose a reform plan that contains a compensation scheme to tradable shares. The compensation is to be paid by the non-tradable shareholders to tradable shareholders to make up for their potential losses following a significant increase in share supply and thereafter stock price drops, and could be in the form of cash, warrants, and most often, additional shares. The board facilitates the negotiations between the nontradable and tradable shareholders, and the trading of the firm s stock is suspended 11

13 during the negotiation period. After rounds of negotiation, the board makes an announcement of the finalized plan and a special shareholder meeting is called for tradable shareholder to vote on the proposal. If the reform plan is approved by at least two-thirds of tradable shareholders, it will be submitted to CSRC. The reform plan becomes effective after obtaining approval from the CSRC, and the trading resumes on the next day. Compulsory lockups were imposed to prolong the sale process of non-tradable shares with the purpose of avoiding a surge in share supply, which might destabilize the market. Non-tradable shareholders are restricted to sell their shares over a 12-month period after the implementation day of the reform plan. Moreover, nontradable shareholders with more than 5% of the firm s total shares outstanding could only sell up to 5% of total shares outstanding 12 months after the lockup, and up to 10% of total shares 24 months after the lockup Hypothesis development We expect the the split share structure reform to affect firms financial reporting quality by altering the incentives of controlling shareholders and improve firms corporate governance. As the shares of controlling shareholders are valued at the market price after the reform, their wealth becomes closely related to firm value. Therefore, controlling shareholders are expected to have less incentives to expropriate minority shareholders and less likely to hide their benefits of control through reporting untruthful financial information. On the contrary, they have stronger incentives to engage in firm value-enhancing activities, such as improving firms information quality, which will help to reduce information risk and thereby reduce firms cost of capital. The reform could also affect firms financial reporting quality through increasing the liquidity of stocks in the market and thereby the value of market monitoring. Arbitrageurs, who could hide their information while make a profit in a liquid market, are willing to spend more effort to collect information and monitor firms behavior. The 12

14 stronger market monitoring could help to improve firms financial reporting quality. However, as aforementioned, although the reform converts non-tradable shares into tradable shares, certain lock up periods are posted to prolong the share sale process of controlling shareholders. Thus, we expect the first channel to dominate the second channel in improving the financial reporting quality in the short run after the reform. Following these reasoning, we propose the first hypothesis as follows: H1: Firms financial reporting quality increases after the split share structure reform. We also examine whether there s any cross-sectional difference in the changes in firms financial reporting quality after the reform. We expect the improvement in financial reporting quality to be greater in firms with weaker corporate governance before the reform, which are expected to experience greater changes through the reform. We measure corporate governance using monitoring from other large (noncontrolling) shareholders or the percentage of non-tradable shares. Firms with less monitoring from other large shareholders or a higher percentage of non-tradable shares are considered to have weaker corporate governance. Our second hypothesis is as follows: H2: The improvement in financial reporting quality is greater for firms with weaker corporate governance than for those with stronger corporate governance before the reform. We also expect the impact of the reform to differ for SOEs and non-soes. The controlling shareholders of SOEs are government agencies, who might have substantial nonpricing considerations, such as meeting certain political or social welfare objectives (Chen et al. (2012)). Therefore, the alignment of interests through strengthening the relation between shareholder wealth and firm value to have less influence over the incentives of the controlling shareholders of SOEs. We thus expect the improvement in financial reporting quality to be more evident for non-soes than SOEs after the reform. We summarize this third prediction as follows: 13

15 H3: The improvement in financial reporting quality is greater for non-soes than for SOEs. 4. Data and Measures 4.1. Data We collect all variables from the CSMAR database. We start with all non-financial firms listed on the Shanghai or Shenzhen stock exchange during the period from 2000 to For a firm to be included in our sample, it must have data available for both the dependent and independent variables used in our most comprehensive model specification. Our final sample includes 16,538 firm-year observations and 2,066 unique firms. Among these firms, 231 firms completed the conversion of non-tradable shares to tradable shares in 2005, 901 in 2006, 113 in 2007, 33 in 2008, 18 in 2009, and 2 in The rest 768 firms either did not have non-tradable shares before the reform or were listed after the reform, or had not completed the conversion by the end of The inclusion of these firms helps to provide a benchmark about the financial reporting quality for firms that do not go through the reform Measures for financial reporting quality and control variables We measure firms financial reporting quality from five perspectives: discretionary accrual, earnings smoothness, earnings persistence, value relevance, and conservatism. The five measures are commonly used in the literature on financial reporting quality. Discretionary accrual, or DisAccrual i,t is measured following Dechow, Slown, and Sweeney (1995), and is defined as the residual from Eq. (1): Accrual i,t 1 = β 0 +β 1 +β 2 ( Rev i,t Rec i,t )+β 3 P P E i,t +ɛ i,t, (1) Asset i,t 1 Asset i,t 1 Asset i,t 1 Asset i,t 1 where Accrual i,t is total accrual, Asset i,t 1 is total asset in the previous year, Rev i,t and Rec i,t are the changes in revenue and accounts receivable from year t 1 to 14

16 t, and P P E i,t is property, plant, and equipment. The cross-sectional regression is performed each year, and a higher DisAccrual i,t indicates a lower level of financial reporting quality. Earnings smoothness, or Smooth i,t, is defined as the standard deviation of quarterly earnings relative to the standard deviation of quarterly cash flow from operation over the past two years. A high ratio indicates less smoothing of the earnings stream relative to cash flows, thus Smooth i,t is an inversed indicator of earnings smoothness. Higher earnings smoothness is suspected to be associated with greater earnings management, and has been documented to be related with determinants of lower financial reporting quality, such as low quality GAAP, less enforcement, and poor shareholder rights, in prior studies (e.g., Leuz, Nanda, and Wysocki (2003)). We follow previous research (e.g., Lev (1983), Ali and Zarowin (1992), Francis et al. (2004)) to measure earnings persistence (P ersistence i,t ) using the slope coefficient estimate from an autoregressive model of order one. Specifically, it is measured by β 1 from Eq. (2): EP S i,t+1 = β 0 + β 1 EP S i,t + ɛ i,t (2) where EP S i,t+1 and EP S i,t are earnings per share of firm i in year t + 1 and t, respectively. And persistent earnings is viewed as desirable. Value relevance is captured by earnings response coefficient, or ERC i,t. We follow the literature to measure ERC i,t by examining the ability of changes in earnings to explain the changes in stock returns. Specifically, we perform the following regression: Ret i,t = β 0 + β 1 EP S i,t + ɛ i,t, (3) where Ret i,t is cumulative monthly returns of firm i during year t, and EP S i,t is the change in earnings per share from year t 1 to year t. ERC i,t is measured by β 1 15

17 from the regression, and a higher β 1 indicates greater value relevance. Lastly, we measure conservatism of financial reporting (Conservatism i,t ) by examining whether firms recognize losses on a timely basis. We follow France et al. (2004) to derive Conservatism i,t based on Eq. (4): EP S i,t = β 0 + β 1 Neg i,t + β 2 Ret i,t + β 3 Neg i,t Ret i,t + ɛ i,t, (4) where EP S i,t and Ret i,t are earnings per share and cumulative monthly return of firm i in year t, respectively, and Neg i,t is an indicator that equals one if Ret i,t is negative, and zero otherwise. The coefficient estimate of β 3 is used to measure conservatism, and a higher β 3 suggests greater conservatism. A set of control variables are included in the empirical analysis, including the logarithm of total assets (LogT A i,t ), the market-to-book ratio (M2B i,t ), cumulative monthly return over the year (Ret i,t ), the standard deviation of monthly return over the year (RetStd i,t ), the leverage ratio defined as total liability divided by total assets (Leverage i,t ), the number of years since the firm became listed (Age i,t ), an indicator that equals one of the firm pays dividend during the year and zero otherwise (Dividend i,t ), the number of non-tradable shares divided by total number of shares outstanding (NT S i,t ), an indicator that equals one if the firm is an SOE (SOE i,t ), and an indicator that equals one if the firm has losses over the preceding two years (ST i,t ). 1 ST i,t is included as a firm on Chinese stock exchanges would get a special treatment designation if it posts losses for two consecutive years, and could face the risk of being suspended from trading if it continues to make loss for one more year. Table 1 provides definitions for all the variables. [Insert Table 1 here] We report summary statistics of the sample characteristics in Panel A of Table 1 We define a firm as an SOE if its ultimate controlling party is a government agency. 16

18 2, and the Pearson correlation among the five financial reporting quality measures in Panel B. The mean and median of DisAccrual are both close to zero, while those for Smoothness are and 0.496, respectively. The average of P ersistence and ERC for the sample firms over the sample period is and 0.855, respectively. Conservatism has the greatest standard deviation of , and a negative mean of Panel B suggests that the correlations among the five financial reporting quality measures are small in economic terms, and only the correlation between P ersistence and ERC is significantly different from zero at at least 10% level. The evidence implies that there s little overlap among the five measures, which is not surprising as the five measures capture financial reporting quality from different perspectives. [Insert Table 2 here] 5. Empirical Analysis 5.1. The influence of reform on firms financial reporting quality We follow Chen et al. (2012) to apply a difference-in-difference approach to investigate the influence of the split share structure reform on firms financial reporting quality. The conversion of non-tradable shares to tradable shares took place at different calendar times, giving us the opportunity to examine the changes in financial reporting quality for both affected and non-affected firms simultaneously. Firms that have completed the conversion are taken as the treatment group, while firms that have not completed the conversion are treated as the control group. To investigate how the reform affects firms financial reporting quality using the five earnings quality measures, we perform the following five regressions separately based on all sample firms: DisAccrual i,t = β 0 + β 1 Reform i,t + β 2 X i,t + ɛ i,t, (5) Smoothness i,t = β 0 + β 1 Reform i,t + β 2 X i,t + ɛ i,t, (6) 17

19 EP S i,t+1 = β 0 + β 1 EP S i,t Reform i,t + β 2 Reform i,t + β 3 EP S i,t + β 4 X i,t + ɛ i,t, (7) Ret i,t = β 0 + β 1 EP S i,t Reform i,t + β 2 Reform i,t + β 3 EP S i,t + β 4 X i,t + ɛ i,t, (8) EP S i,t+1 = β 0 + β 1 Ret i,t Neg i,t Reform i,t + β 2 Reform i,t + β 3 Ret i,t + β 4 Neg i,t + β 5 Ret i,t Neg i,t + β 6 X i,t + ɛ i,t, (9) where Reform i,t is a dummy that equals one if firm i has completed the conversion by the end of year t, and zero otherwise. X i,t is a set of control variables, including LogT A i,t, M2B i,t, Ret i,t, RetStd i,t, Leverage i,t, Age i,t, Dividend i,t, ST i,t, NT S i,t, and SOE i,t. In all the regressions, industry and year fixed effects are both controlled, and t-values are calculated based on standard errors clustered by firm. As the reform timing differs across firms, Reform i,t or its interaction terms with other variables helps to identify the causal effect of the split share structure reform on firms financial reporting quality. If a firm s financial reporting quality increases after the reform, we would expect both discretionary accruals and earnings smoothness to decrease after the conversion, resulting in a negative β 1 in Eq. (5) and a positive β 1 in Eq. (6). The coefficients on the interaction terms of EP S i,t Reform i,t in Eq. (7) and EP S i,t Reform i,t in Eq. (8) measure the changes in earnings persistence and value relevance after the reform, respectively, and an improvement in earnings persistence and value relevance would lead to a positive coefficient on both interaction terms. Lastly, the change in conservatism after the reform is captured by the coefficient on Ret i,t Neg i,t Reform i,t, and a positive coefficient suggests that incurred losses in earnings is recognized on a more timely basis. The regression results are reported in Table 3. [Insert Table 3 here] 18

20 In column (1), DisAccrual i,t is regressed on Reform i,t with the control of a set of firm characteristics. The coefficient on Reform i,t is negative, as expected, but not significantly different from zero. In column 2, where Smoothness i,t is used as the dependent variable, the coefficient on Reform i,t turns to be significantly positive, suggesting that earnings smoothness decreases after the reform. The results are much stronger in columns (3) to (5). The coefficients on EP S i,t Reform i,t, EP S i,t Reform i,t, and Ret i,t Neg i,t Reform i,t are all positive, and are significantly different from zero at the 5% level in column (3), and at the 1% level in columns (4) and (5). The results in Table 3 suggest that after the split share structure reform, on average, there s an decrease in firms discretionary accrual (although not statistically significant) and earnings smoothness, and an increase in firms earnings persistence, value relevance, and conservatism. The results confirm that the reform helps to improve firms financial reporting quality, rendering support to our Hypothesis Cross-sectional variation in the influence of reform on firms financial reporting quality Hypotheses 2 and 3 expect the influence of the split share structure reform on firms financial reporting quality to differ across firms. We examine the variation conditional on the strength of corporate governance before the reform, and whether the firm is an SOE Pre-reform corporate governance We use two corporate governance measures, which are the monitoring intensity by large (non-controlling) shareholders (Monitor i,t ) and the percentage of non-tradable shares (NT S i,t ). Monitor i,t is the Herfindahl index for the concentration of shares among the second to the fifth largest shareholders of firm i, constructed following Chen et al. (2012). A higher Monitor i,t indicates stronger concentration of shares among large shareholders, who have more stake in the firm and stronger ability to intervene managers behavior. Thus, firms with a higher Monitor i,t before the reform 19

21 is expected to have stronger pre-reform monitoring, as large shareholders hold more shares in a more concentrated way. We therefore expect the influence of reform on firms financial reporting quality to be weaker in firms with a higher Monitor i,t before the reform, but stronger in firms with a lower pre-reform Monitor i,t. We divide sample firms into two groups based on Monitor i,t in 2004, one year before the government started the split share structure reform. Firms with a Monitor i,t above the median are grouped in the high Monitor i,t group, which are expected to have stronger corporate governance before the reform and thus smaller changes in financial reporting quality after the reform. Firms with a Monitor i,t below the median are grouped in the low Monitor i,t group, which are expected to experience a greater increase in financial reporting quality after the reform. We perform the regressions of Eq. (5) to (9) for these two subsamples, respectively, and report the results in Table 4. [Insert Table 4 here] Columns (1), (3), (5), (7), and (9) show the results for high pre-reform Monitor i,t firms, and columns (2), (4), (6), (8), and (10) show the results for low pre-reform Monitor i,t firms. Column (1) to (2) reveal that there s no significant difference in the changes in discretionary accrual and earnings smoothness after the reform between high and low Monitor i,t firms. The coefficients on EP S i,t Reform i,t, EP S i,t Reform i,t, and Ret i,t Neg i,t Reform i,t are all significantly positive at the 1% level in columns (6), (8), and (10), but are indistinguishable from zero in columns (5), (7), and (9). The results suggest that earnings persistence, value relevance, and conservatism all increase significantly after the reform for low pre-reform Monitor i,t firms, but not for high pre-reform Reform i,t firms. Overall, the results in Table 4 imply that the influence of reform on firms financial reporting quality is stronger for firms with weaker pre-reform monitoring from other large shareholders, but weaker for firms with stronger pre-reform monitoring intensity. 20

22 Our second corporate governance measure is the proportion of non-tradable shares before the reform. Firms with a higher proportion of non-tradable shares before the reform are considered to have more corporate governance problems and are expected to go through a more dramatic change in share structure during the reform. We therefore expect the changes in firm s financial reporting quality after the reform to be stronger for these firms. We calculate the proportion of non-tradable shares relative to total shares outstanding, or NT S i,t, for all sample firms in 2004, and divide the sample firms into two groups according to whether NT S i,t is above or below the sample median. We then examine the impact of the reform on high and low NT S i,t firms after the reform, and report the results in Table 5. [Insert Table 5 here] The first four columns show that there s no significant difference in the change in DisAccrual i,t and Smoothness i,t for high and low NT S i,t firms. Columns (5) and (6) show that both types of firms experience an increase in earnings persistence. The magnitude of the coefficient on EP S i,t Reform i,t implies that after the reform, the correlation between EP S i,t and EP S i,t+1 is increased by for firms with more pre-reform non-tradable shares, and increased by for firms with fewer prereform non-tradable shares. Columns (7) and (8) show that earnings value relevance is improved significantly for low NT S i,t firms after the reform, but the change is indistinguishable from zero for high NT S i,t firms, which is not consistent with our expectation. The last two columns present evidence suggesting that the conservatism increases for both types of firms, and the improvement is stronger for firms with more pre-reform non-tradable shares. Overall, the results in Tables 4 to 5 suggest that the improvement in financial reporting quality, especially in terms of earnings persistence, value relevance, and conservatism, is more evident in firms with a lower pre-reform Monitor i,t, consistent 21

23 with our Hypothesis 2. But the evidence is mixed when we examine the cross-sectional difference conditional on pre-reform proportion of non-tradable shares SOEs vs. Non-SOEs We next investigate the influence of the split share structure reform on firms financial reporting quality conditional on whether the firm is an SOE or not. The controlling shareholder of SOE is a government agency, for whom maximizing shareholder s wealth might be less a concern compared to non-soes. The interest alignment brought by the reform, therefore, is expected to exert a stronger influence on non-soes than SOEs. As a consequence, we expect the improvement in financial reporting quality after the reform to be more evident among non-soes than for SOEs. The results for the subsample tests are reported in Table 6. [Insert Table 6 here] The coefficient on Reform i,t is negative in both columns (1) and (2), where DisAccrual i,t is used as the dependent variable, but not statistically significant. In columns (3) and (4), where Smoothness i,t is employed as the dependent variable, the coefficient on Reform i,t is positive for non-soes but negative for SOEs, though insignificant in both columns. Columns (5) and (6) suggest that earnings persistence increases significantly after the reform for both SOE and non-soe firms, where the coefficient on EP S i,t Reform i,t is significantly different from zero at the 5% level in both columns, but the magnitude is twice as large for non-soes (0.238) than for SOEs (0.109). When earnings value relevance is examined, the change in earnings response coefficient after the reform is only marginally significant for SOEs in column (7), but significantly positive at the 5% level for non-soes in column (8). The last two columns of Table 6 show that for both SOE and non-soe firms, earnings conservatism is improved after the reform, but the improvement is much stronger in both statistic and economic 22

24 terms for non-soe firms. The evidence is consistent with our Hypothesis 3 that the improvement in financial reporting quality is stronger for non-soe than for SOEs. 6. Conclusions In this paper, we examine the change in financial reporting quality of Chinese listed firms after the split share structure reform, which converts non-tradable shares held by controlling shareholders into tradable shares. Before the reform, controlling shareholders hold non-tradable shares that could not be freely traded on stock exchanges, resulting in their strong incentives to manipulate earnings in order to hide their own private benefits of control and poor financial reporting quality in Chinese stock markets. The reform eliminates the significant market friction by converting non-tradable shares into tradable shares, and ties the wealth of controlling shareholders to the market value of the shares. As a consequence, controlling shareholders are more willing to engage value-enhancing activities after the reform and have less incentives to manipulate earnings. Our empirical evidence supports our conjecture. We find a decrease in discretionary accrual (though not significant) and a decrease in earnings smoothness after the reform, suggesting that firms are less likely to manipulate their earnings after the conversion of non-tradable shares into tradable shares. Moreover, earnings persistence, value relevance, and conservatism all increase significantly after the reform, confirming that the reform exerts a positive influence on firms financial reporting quality. Subsample tests further suggest that the improvement in financial reporting quality after the reform is stronger for firms with weaker pre-reform corporate governance and non-soes. Our study contributes to the literature in several ways. Many studies have examined the relation between corporate governance and financial reporting quality. However, it is difficult to determine the direction of causality due to the endogenous nature of firm characteristics and financial information environment (e.g., Armstrong 23

25 et al. (2010a)). The Chinese split share structure reform provides a good opportunity to test the causal impact of ownership structure on financial reporting quality. The change in firms ownership structure results from the government s attempt to eliminate a market friction, thus was exogenous to the affected firms. By showing that firms financial reporting quality increases significantly after the reform, we provide evidence showing that corporate governance causally affects financial reporting quality. Moreover, our study adds to the literature on the consequences of privatization, especially in countries undergoing economy transitions. The split share structure reform results from the share issue privatization in China, which is the largest privatization in history so far. Liao et al. (2014) and Chen et al. (2012) have examined the changes in risk-sharing function and corporate cash holdings after the reform. We add to these studies by showing that the reform also works to improve financial reporting quality, which is a key issue in the financial markets. 24

26 References [1] Armstrong, Christopher S., Wayne R. Guaya, and Joseph P. Weber, 2010a, The role of information and financial reporting in corporate governance and debt contracting, Journal of Accounting and Economics, 50, [2] Armstrong, Christopher S., Alan D. Jagolinzer, and David F. Larcker, 2010b, Chief executive officer equity incentives and accounting irregularities, Journal of Accounting Research, 48, [3] Beasley, Mark S., 1996, An empirical analysis of the relation between the board of director composition and financial statement fraud, The Accounting Review, 71, [4] Bergstresser, Daniel, and Thomas Philippon, 2006, CEO incentives and earnings management, Journal of Financial Economics, 80, [5] Biddle, Gary C., Gilles Hilary, and Rodrigo Verdi, 2009, How does financial reporting quality relate to investment efficiency?, Journal of Accounting and Economics, 48, [6] Callen, Jeffrey L., Mozaffar Khan, and Hai Lu, 2013, Accounting quality, stock price delay, and future stock returns, Contemporary Accounting Research, 30, [7] Chen, Qi, Xiao Chen, Katherine Schipper, Yongxin Xu, and Jian Xue, 2013, The Sensitivity of Corporate Cash Holdings to Corporate Governance, Review of Financial Studies, forthcoming. [8] Cheng, Qiang, and Terry D. Warfield, 2005, Equity incentives and earnings management, The Accounting Review, 80,