The quality of ﬁnancial reporting in China: An examination from an accounting restatement perspective Xia Wanga, Min Wub,⇑,1 aSchool of Business, East China Normal University, China bMorgan Stanley Asia Limited, Hong Kong a r t i c l e i n f o Article history: Received 4 October 2010 Accepted 25 July 2011 Available online 9 December 2011 JEL classiﬁcation: M41 Keywords: Earnings quality Earnings management Restatements China capital markets Regulation Disclosure abstract This study uses restatements to reveal the poor quality of pastaccountinginformationreportedwithinChina’scapital market. We show that up to a quarter of listed ﬁrms in mainlandChinaexplicitlyadmittedthepoorqualityoftheir ﬁnancial information by restating their previous ﬁnancial reports between 1999 and 2005. Many of these ﬁrms managed their earnings mainly via below-the-line items to avoid losses and promote survival, rather than to support reﬁnancing goals. Such poor-quality ﬁnancial reporting is more likely among ﬁrms that have weaker proﬁtability and a shareholder base that is state-controlled, with diffused ownership and a relatively low proportion of shares held by institutional investors. Furthermore, we ﬁnd the market to be relatively insensitive to such admissions. Investors’ reactions capture only the earnings information of the current reported year, rather than also reﬂecting the concurrently revealed correction of past ﬁnancial 1755-3091/$ - see front matter 2011 China Journal of Accounting Research. Founded by Sun Yat-sen University and City University of Hong Kong. Production and hosting by Elsevier B.V. All rights reserved. doi:10.1016/j.cjar.2011.09.001 E-mail address: [email protected] (M. Wu). 1 TheviewsexpressedhereinarethoseoftheauthoranddonotnecessarilyreﬂecttheviewsofMorganStanleyoritsafﬁliates. China Journal of Accounting Research 4 (2011) 167–196 Contents lists available at SciVerse ScienceDirect China Journal of Accounting Research journal homepage: www.elsevier.com/locate/cjar reporting. However, the equity market does not completely ignore the earnings information. Investors’ reliance on earnings is merely low relative to the mature US market. These ﬁndings demonstrate that accounting credibility in China has low value; providing poor-quality ﬁnancial information bears little cost because various market mechanisms fail to deter such behavior. Nevertheless, regulators’ ongoing efforts to enhance the quality of ﬁnancial information and disclosure among listed ﬁrms are still fruitful. The frequency of restatements decreased over our sample period, which reinforces the current regulatory prospects and strategies for further improving China’s capital markets. 2011 China Journal of Accounting Research. Founded by Sun Yat-sen University and City University of Hong Kong. Production and hosting by Elsevier B.V. All rights reserved. 1. Introduction Efﬁcient capital markets reward high-quality ﬁnancial reporting, which facilitates the efﬁcient raising and allocation of corporate capital and thus creates beneﬁts for investors. In the past decade, China’s stock market has become very popular among domestic investors, as the stock market is one of a very limited number of investment vehicles open to them. The total market value of equity invested in the Chinese stock market grew by an order of magnitude during the 1999–2007 period.2 In less than three years, from mid2005 to late-2007, the Shanghai and Shenzhen Stock Exchange Composite Index experienced a sixfold increase, after a decade of quietness. These gains generated euphoria among investors, at least until the stock market started to correct by as much as 70% a year from the end of2007.Thisriseandfallofthestockmarkethighlightedgeneralconcernsoverthequalityof ﬁnancial reporting in China. The negatively perceived turn of events in the Chinese market appears to parallel that of the US market during the same period. It is noted that an increasing number of US ﬁrms in recentyearshavehadtorestatetheirpreviousyears’ﬁnancialreports,eithervoluntarilyor when forced to do so by regulators (Scholz, 2008; Wu, 2002). The number of ﬁrms restating their previous ﬁnancial reports reached almost 300 in 2005, amounting to roughly two percent of all public companies in the United States. This number was high enough to draw appreciable attention from the media, regulators and academics. In mainland China, a similar yet more pronounced phenomenon recently emerged. We ﬁnd that a signiﬁcant proportion of listed companies on the Shanghai and Shenzhen Stock Exchanges restated their annual reports for the years from 1999 to 2005. Interestingly, in contrast to the enormous publicity received by US earnings restatements, in China restatements have received scant coverage in the Chinese media, despite the problem being more pronounced. Restatements represent clear-cut violations of accounting rules and hence an explicit admission of the poor quality of companies’ past ﬁnancial reporting. Research shows that 2 From 821 billion Chinese Yuan (approximately USD99.2 billion at USD1 =CNY8.2768) by the end of 1999 to 8555 billion Yuan (approximately USD1171.9 billion at the USD1= CNY7.30) by the end of 2007. 168 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 in the United States, the announcement of a ﬁrm’s earnings restatement usually triggers a severe decline in the stock price, thus harshly penalizing the restatement ﬁrm (Palmrose et al., 2004; Wu, 2002; Turner et al., 2001). US research also ﬁnds that after a restatement, the market relies less on earnings information to determine a ﬁrm’s stock price, reﬂecting investors’ loss of conﬁdence in the company’s ﬁnancial reports (Wu, 2002; Andersen and Yohn, 2002), and the company is forced to pay a higher cost for equity capital (Hribar and Jenkins,2004).Theseeffectsreﬂectthedamagetothecredibilityofaﬁrm’sfutureﬁnancial reports in the wake of previously released low-quality ﬁnancial information. In an emerging market such as China, however, it is highly uncertain whether the same market reaction exists, because investors’ conﬁdence in ﬁrms’ ﬁnancial reports may be low at the outset. The question also arises whether earnings restatements will reduce investors’ reliance on accounting earnings for setting stock prices, or increase the ﬁrm’s cost of equity capital, as found in the US research. Furthermore, it is possible that many of the restatementsinChinacouldbetheresultofopportunisticbehavior,althoughwithdifferentmotivations than in the United States, given the different institutional setting. Thus, China’s restatements provide an opportunity to examine the value ascribed by an emerging market to the quality and credibility of ﬁnancial reports, relative to the value assigned by a mature market. Our ﬁrst objective in this paper is to explore the characteristics of poor-quality ﬁrms— deﬁned as ﬁrms that issue restatements—relative to control ﬁrms. We also investigate the incentives to report poor-quality ﬁnancial information previously released by such ﬁrms. Inaddition,weexamine theconsequences,intermsofthestockmarketreaction, ofadmitting the publication of such information. Finally, we rationalize why Chinese companies so frequently provide poor-quality ﬁnancial reporting, only to subsequently correct it. We hope this study will offer insights to regulators on how to detect poor-quality companies andsuggest aspects thatmayimprovethequalityoflisted ﬁrms.As withmanyotherstudies on issues concerning mainland China, this study also aims to remind readers that the regulatoryandﬁnancialreportingenvironmentisvastlydifferentinChina,asarethemotivations and consequences for ﬁrms and for the market as a whole. This study offers academic researchers, regulators and investors—both domestic and international—insights into the overall quality of China’s accounting information and a further understanding of China’s increasingly important capital markets. This is perhaps the ﬁrst empirical study that directly examines the issue of accounting quality in China—a nation whose capital markets are becoming increasingly important and hence cannot be ignored in the global capital market. This study complements the broad research literatures on China, earnings quality and restatements. Previous literature (Wu, 2002; Anderson and Yohn, 2002) on the US market argues that earnings restatements are indicators of the poor quality of prior ﬁnancial reporting. Poor accountingqualityisgenerallypenalizedbythecapitalmarket,andsuchpenaltiesserveas a deterrent to companies’ delivery of poor accounting quality via accounting manipulation and similar means. The results of our study imply that such penalties do not yet exist in China. Along with the Ministry of Finance (MOF), which sets accounting standards, China’s regulatorybody, theChineseSecurities and RegulatoryCommission (CSRC), hasbeen making efforts since 1996 to enhance the regulatory environment of the nation’s capital market. Accounting regulation is an important part of this process and there have been signiﬁcant and gradual economic improvements since 1978. Meanwhile, we also realize thatsucheffortsmustbepersistentlycarriedintothefuture.Forexample,ourstudyshows X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 169 that in the absence of an effective penalty system in the market, investors generally do not distinguish between good and poor accounting quality. Hence, our study also suggests the establishment of an effective penalty system through government regulation, which appears to be a necessary step for China’s emerging market to become a complete capital market. Following this introduction,the paperconsists ofsix sections. Section 2 offers a detailed, topic-relevant background to China’s accounting and regulatory environment. Section 3 conducts a literature review on restatements and other related areas. Section 4 develops our hypotheses and outlines the empirical models. Section 5 describes the details of restatements in China. Section 6 provides the empirical tests and interprets the results. Section 7 concludes. 2. Accounting standards and regulatory background in China The securities regulator in the People’s Republic of China (PRC) is the CSRC, which is equivalent to the US Securities and Exchange Commission (SEC). Established in October 1992, the CSRC is an institution of the PRC State Council and is authorized to regulate China’s securities and futures markets. Although it did not issue its ﬁrst version of Procedure for Inspecting Listed Companies until December 21, 1996 (the 1996 Procedure), it became effective immediately. The Procedure covered the scope and procedures, and the CSRC’s responsibilities during inspections. The scope emphasized the truthfulness, completeness, accuracy and timeliness of disclosures by listed companies. In the 1990s, companies followed the old PRC accounting standards, which failed to specify how to deal with accounting errors and irregularities. Accounting treatments therefore varied widely among companies and across industries. China’s accounting reform of the late 1990s introduced the ﬁrst accounting standard: The Standard of Changes in Accounting Policies and Estimates, and Corrections of Material Accounting Errors3 (the 1999 Standard). It was issued in June 1998 by the MOF, PRC’s accounting standard setter and came into effect on January 1, 1999. Section 3 of the 1999 Standard described the restatement methods and required disclosure of the reason(s) for and total amount of the restatement. The 1999 Standard was modiﬁed slightly in January 2001, with one item added: Any abuse of changes in accounting policies or accounting estimates will be treated as material accounting mistakes and therefore restated. The 1999 Standard only required a restatement to be disclosed in the company’s forthcoming annual report. For several years, the Accounting Standards of Business Enterprises (ASBE) were adopted in parallel with the Companies Accounting System, which mentioned the correction of errors in its tenth chapter and provided technical treatments that were consistent with those of the Old Standard. Issued in October 1999, the CSRC’s Notice on Improving Financial Information Disclosure of Listed Companies (the 1999 Notice) states that: (1) listed companies should make proper lossestimatesofaccountsreceivable,inventories,investments,etc.,andshouldnotchange the method of provision and percentage of provision within the same reporting period at the companies’ will and (2) listed companies should disclose any change in accounting policies or estimates. In 2001, there was still no rule that explicitly required listed companies to disclose accounting irregularities or mistakes publicly in a timely fashion. Unlike the common 3 ThisStandardwasachapterintheAccountingStandardsofBusinessEnterprises(ASBE),whichwascompletedin2002;thus, it is also called China’s 2002 ASBE. 170 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 US practice of public disclosure of a restatement upon ﬁrst discovery by the media, China’s press remained largely silent. Hence, investors were ﬁrst informed of a restatement upon thepublicrelease ofa company’sannualreport inmajor Chinesebusinessnewspapersand the website4 designated by the CSRC. Unlike the practices of US listed ﬁrms, in which a restatement will revise any affected line items in all relevant quarter(s)’ and year(s)’ income statements and balance sheets, restatements in China under the Old Standard are not required to tabulate the corrected ﬁnancial statements of all affected years. In most cases, where only the ﬁnancial statements of the previous year (t1) are corrected, the corrected ﬁnancial statements will be found in the current year’s (t) annual report for comparison purposes. If the corrected year(s) reaches beyond the previous year (t1), then the correction will not be made in the earlier released reports, but will instead bypass the proﬁt and loss statement for year t1 and directly hit the corrected balance sheet. The overall cumulative effect would, of course, be adjusted in the beginning balance of retained earnings and other affected items on the balance sheet in the annual report of year t. Because of the subtle difference between the accounting treatments required by the US Generally Accepted Accounting Principles (GAAP) and the International Financial Reporting Standards (IFRS), we cautiously call our event ‘‘restatements’’ rather than ‘‘earnings restatements’’ because not all the affected earnings would be restated during the sample period, only the previous year’s earnings. Note also that the Old Standard required the detailed reasons for and amount of the restatement to be disclosed in the footnotes, but in practice such disclosures, especially regarding the reasons for restatement, were generally rather brief and opaque. On March 19, 2001, the CSRC issued the revised Procedure for Inspecting Listed Companies (the 2001 Procedure), which superseded the original 1996 Procedure. With the new release came a CSRC announcement that it would strengthen the inspection of listed companies’ ﬁnancial reports, corporate governance structures and independence from related parties. The 2001 Procedure required companies to correct any irregularities found in the inspection and to disclose them publicly within 30 days of an ofﬁcial notice. The two years following the release of the 2001 Procedure witnessed a tremendous effort by the CSRC, resulting in the issue of 19 chapters of Rules on Information Disclosure for Listed Companies. Chapter 19: The Correction of Financial Information and its Disclosure (Rule 19) was issued at the end of 2003. Rule 19 demands that listed companies immediately ﬁle an ofﬁcial report with the CSRC regarding any material events, including the correction of ﬁnancial statements, and submit a revised and audited annual report within 45 days if the most recent annual report is incorrect. However, due to a loophole, Rule 19 did not include a scenario for change-of-accounting estimates, and thus it had little real effect on disclosure patterns. Because a change-of-accounting estimate was not deﬁned as a material event, it did not require timely disclosure. Many companies intentionally misclassiﬁed the correction of mistakes as a change-of-accounting estimate and routinely disclosed them in the forthcoming annual report rather than providing an immediate disclosure in the form of a change-of-accounting estimate.5 4 The CSRC designates the following four newspapers for listed companies to disclose their ﬁnancial information: China Securities Journal, Shanghai Securities Journal, Securities Time, and Securities Daily. The CSRC-designated website is www.cninfo.com.cn. 5 Note that a change-of-accounting estimate does not change past accounting numbers, only future ones. The true accounting practice to ameliorate a false claim is, however, a restatement. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 171 On January 6, 2004, just one day after a press conference offering explanations for both the substantial number of companies that received a qualiﬁed auditor’s opinion and the increasednumber ofrestatements in2002, theCSRCissued the Notice on Further Improving Financial Information Disclosure of Listed Companies (the 2004 Notice). By emphasizing the 1999 Notice, the 2004 Notice clearly states that listed companies should not abuse asset impairments, change-of-accounting estimates or correction of material mistakes to manipulate ﬁnancial results. Any company doing so would be held responsible. However, the 2004 Notice did not specify the scope of responsibility that a company would bear for committing a violation. On February 15, 2006, the MOF announced that starting from 2007 annual reports, all publicly traded companies would adopt the new Accounting Standards of Business Enterprises (2006 ASBE), representing a major convergence towards the IFRS. The 2006 ASBE, Changes in Accounting Policies and Estimates and Corrections of Accounting Errors, fully adopted the practices of the IFRS, which are consistent with US GAAP. That is, from 2007 annual reports and thereafter, a restatement will revise any affected line items in income statements and balance sheets for all relevant quarter(s) and year(s). Table 1 summarizes the development of accounting regulations related to this speciﬁc issue. Table 1 Regulatory timeline. Date of issuance Effective date Issuer Name of document Abbreviation December 21, 1996 December 21, 1996 CSRC Procedure for Inspecting Listed Companies 1996 Procedure June 25, 1998 January 1, 1999 MOF ASBE: Changes in Accounting Policies, Estimates and Corrections of Material Accounting Errors The 1999 Standard October 10, 1999 October 10, 1999 CSRC Notice on Improving Financial Information Disclosure of Listed Companies 1999 Notice December 29, 2000 January 1, 2001 MOF Companies’ Accounting System 2001, Chapter 10: Accounting Adjustment; Section 3: Corrections of Accounting Errors Accounting System January 18, 2001 January 1, 2001 MOF ASBE: Changes in Accounting Policies and Estimates and Corrections of Material Accounting Errors (Revised) The 1999 Standard (revised) March 19, 2001 March 19, 2001 CSRC Procedure for Inspecting Listed Companies (Revised) 2001 Procedure December 1, 2003 December 1, 2003 CSRC Rules on Information Disclosure for Listed Companies #19: The Correction of Financial Information and its Disclosure Rule 19 January 8, 2004 January 8, 2004 CSRC Notice on Further Improving Financial Information Disclosure of Listed Companies 2004 Notice February 15, 2006 January 1, 2007 MOF New ASBE #28: Changes in Accounting Policies, Estimates and Corrections of Material Accounting Errors New Standard Notes: CSRC: China’s Securities Regulatory Commission. MOF: Ministry of Finance, People’s Republic of China. 172 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 3. Literature review 3.1. Earnings restatements in the United States Usingasampleof73ﬁrmsthatcorrectedpreviouslyreportedquarterlyearnings,Kinney and McDaniel (1989) ﬁnd that the sample ﬁrms were smaller and less proﬁtable, had a higher level of debt and lower level of growth, and faced more serious uncertainties due to the receipt of more qualiﬁed audit opinions. Analyzing 224 SEC accounting and auditing enforcement releases between 1982 and 1989, Feroz et al. (1991) ﬁnd that the SEC most often pursued overstatements of accounts receivable and inventories due to premature revenue recognition and delayed write-offs. They also ﬁnd that the disclosure of these reporting violations changed expectations of the target ﬁrm’s future earnings, as reﬂected in ﬁnancial analysts’ reduced earnings estimates after the disclosures. DeFond and Jiambalvo (1991) examine 44 earnings restatements and ﬁnd that 41 of them involved overstatement, consistent with an income-increasing motivation. They ﬁnd that earnings overstatements are negatively correlated with growth in earnings and are more likely when ﬁrms have diffuse ownership and few income-increasing GAAP alternatives available. They also ﬁnd that restating ﬁrms are less likely to have audit committees. Dechow et al. (1995) ﬁnd that an important motivation for earnings manipulation is the desire to attract external ﬁnancing at a lower cost. Firms that manipulate earnings are more likely to have: (1) boards of directors dominated by management, (2) a CEO who simultaneously serves as the chairman of the board, and (3) a CEO who is also the ﬁrm’s founder. In addition, these ﬁrms are less likely to have an audit committee and an outside blockholder. Firms that manipulate earnings experience a signiﬁcant increase in their cost of capital after the manipulation is made public. Enron’s accounting scandal in 2001 and Worldcom’s in 2002 spawned a large volume of research on earnings restatements. This research can be broadly classiﬁed into three categories: (1) descriptive and market reaction studies around restatement announcements, (2) investigation of the motivations that lead to restatements, and (3) examination of the consequences of restatements. In the ﬁrst category, Wu (2003) documents the characteristics of more than 1200 US restatements announced between 1977 and 2001. She shows that there has been a signiﬁcant increase in the number of restatements since the late 1990s and ﬁnds a signiﬁcant market reaction of more than 11% over a three-day window—a reaction that can be explained by both qualitative and quantitative information carried in the restatement announcements. Concurrent research by Palmrose et al. (2004) and Turner et al. (2001) also ﬁnd similar market reactions to the announcements. Furthermore, Lev et al. (2008) ﬁnd that restatements that eliminate or shorten the history ofearningsgrowthorpositiveearningshavesigniﬁcantlymoreadverseeffectsforinvestor valuations and the likelihood of lawsuits than other restatements. In the second category of studies, Richardson et al. (2003) suggest that capital market pressures motivate restatement companies to adopt aggressive accounting policies; that is, the typical restatement ﬁrm has been attempting to maintain a string of consecutive quartersofpositiveearningsgrowthandconsecutivepositivequarterlyearningssurprises. In addition, top executives at these ﬁrms receive a larger portion of their compensation from equity relative to leaders of non-restating ﬁrms. Richardson et al. (2003) also ﬁnd the information in accruals to be a key indicator of the earnings manipulation that leads to the restatements. Grifﬁn (2003) investigates the patterns of insider trading in restating X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 173 ﬁrms and implies that proﬁting from insider trading is one of the incentives for managers to overstate earnings. Agrawal and Chadha (2005) ﬁnd that the incidence of independent directors with a background in accounting or ﬁnance on the board or audit committee is negatively related to the probability of restatements. However, unlike Burns and Kedia (2006), they do not ﬁnd signiﬁcant deﬁciencies in other aspects of corporate governance. In the third category, several studies explore the consequences of restatements. Wu (2003) ﬁnds that the earnings response coefﬁcient (ERC) dropped dramatically following restatement, which could be interpreted as a loss of conﬁdence among investors. Hribar and Jenkins (2004) ﬁnd that accounting restatements lead to both decreases in expected future earnings and increases in the ﬁrm’s cost of equity capital. Srinivasan (2005) shows that outside directors, especially audit committee members, bear reputational costs for failures in ﬁnancial reporting. 3.2. Earnings management and restatements in China China research often focuses on earnings management, which offers a rich background for the restatement issue. Typical incentives that are found in the United States to manipulate earnings are almost nonexistent in China. For example, demand for ﬁnancing, especially equity ﬁnancing, is huge in China, whereas incentives to meet or beat analysts’ expectations are minimal. The compensation plans of China’s listed companies are rarely incentive-based,thusmanagerscannotmanipulateearningstoinﬂatestockpriceswithintent to beneﬁt their own compensation. Chinese companies also do not face pressure from debt covenant constraints. Earnings management, nevertheless, usually occurs when companies are conducting their IPOs. Aharony et al. (2000) suggest that state-owned enterprises in unprotected industries may manage accounting accruals to boost earnings and/ or list those business units with temporarily high proﬁts resulting from high accounting accruals during the process of ﬁnancial packaging. Earnings management also takes place when listed companies conduct secondary issuances or rights issuances. Given that listed companies are required to achieve a minimum average return on equity (ROE) of 10% for the three years prior to secondary issuances or rights issuances, and given the reality that the CSRC has limited resources to monitor all applicants closely, Chen and Yuan (2004) show that many ﬁrms were able to gain approval for rights issues through earnings management and subsequently performed worse than those that did not employ such practices. Thus, capital resources might have been better allocated had the regulators examined the management of earnings more closely. Listed companies also massage earnings to avoid consecutive losses, which would result in them being tagged with ‘‘special treatment’’6 (ST) (Lu, 1999) or, worse, ‘‘particular transfer’’7 (PT). In addition, 6 Special Treatment (ST) has been adopted since April 22, 1998, as a signal for listed companies experiencing any of the following abnormal ﬁnancial or other abnormal situations: (1) two consecutive years of losses; (2) stockholders’ equity falling below the nominal value in the most recent year (in China, the nominal value per share is stipulated as 1 Yuan for all listed companies);(3)independent auditorissuanceofa qualiﬁed opinionorrefusalofissuingopinion;(4)stockholders’equity,netof the auditor fee and unrecognized portion by concerned parties, falling below the nominal value by the end of the most recent year; (5) two consecutive years of losses following the restatement of a previous year’s result in the most recent year’s annual report;or(6)anyﬁnancialsituationtheCSRCdeemsabnormal.Otherabnormalsituationsincludediscontinuationofoperations due to natural disaster or other signiﬁcant event, possible punitive and compensatory damages from lawsuits exceeding net assets, etc. A cap of 5% of the stock price movement (increase or decrease) applies to ST stocks. 7 Particular Transfer (PT), effective since July 9, 1999, is designed for listed companies that experience three consecutive years of losses. The daily trading will be suspended and substituted with once-a-week special transfers among investors. A cap of 5% will be imposed on increasing stock prices, but no stop limit for declining price. 174 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 earnings management is commonly conducted through ‘‘below-the-line’’ non-operating items (Chen and Yuan, 2004). Finally, work by Wang and Zhang (2004) shows that an increasing number of ﬁrms issued restatements between 1999 and 2002. 4. Hypotheses and empirical models Having elaborated on China’s unique institutional background on companies’ incentives tomanipulateearningsinSection3,weﬁrstofferthefollowingtwohypothesesonthetwo major incentives for Chinese companies. The ﬁrst addresses ﬁnancing needs, whereas the second concerns survival. Hypothesis 1. Restatement ﬁrms tend to be those with strong ﬁnancing incentives during or before the restated year. Speciﬁcally, restatement ﬁrms tend to be those that offered secondary or rights issues during or before the restated year. Hypothesis 2. Restatement ﬁrms tend to be those under delisting pressure during or before the restated year. Notethatthesetwohypothesescannotco-existinthesameﬁrm,becausethepreviously mentioned 10% proﬁtability requirement for reﬁnancing is far above the break-even point, and companies meeting this criterion are in little danger of delisting. Next, we examine which factors collectively inﬂuence companies to issue restatements in China’s A-share market. Our testing variables consist of three categories: corporate governance, motivations and ﬁrm performance. The reform of Chinese stateowned enterprises (SOE) has been relatively successful in solving short-term but not long-term managerial incentive problems, and has also failed to adequately address the issue of management selection. The latter problem arises from the fact that managers of SOEs are selected by bureaucrats rather than capitalists. Bureaucrats have the authority to select managers but do not need to bear the consequences of that selection. Thus, they have no proper incentives to ﬁnd and appoint high-caliber managers, which negatively affects the quality of ﬁnancial reporting (Zhang, 1999). Highly concentrated ownership, which is common in East Asia, can lead to an entrenchment effect and reduce the rights of minority shareholders. Decisions made by controlling owners are often not contestable under the region’s weak legal systems or by ineffective corporate governance mechanisms, such as boards of directors and the market for corporate control (Shleifer and Vishny, 1997; La Porta et al., 1999). Consequently, controlling owners are perceived to report accounting information for self-interested purposes, causing the reported earnings to lose credibility with outside investors in East Asia (Fan and Wong, 2002). Additionally, involvement by institutional investors will enhance corporate governance. In Chinese research, return on assets (ROA) is widely adopted as the prime performance indicator, rather than ROE, as ROE is often manipulated due to the CSRC basing various thresholds on it. Hence, we propose the following hypothesis. Hypothesis 3. Restatements tend to occur more frequently among ﬁrms with poor governance, more concentrated ownership and poorer ﬁnancial performance. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 175 We propose the following model to test our ﬁrst three hypotheses: Restatementi ¼aþb1SOEi þb2L Sharei þb3IIHþb4LOSSþb5LOSS STþb6RSI þb7DAþb8ROAþb9LEVþb10Sizei þei ð1Þ Here, Restatementi is a dummy variable for ﬁrm i, which takes on the value 1 for restatement ﬁrms and 0 for other listed companies. For the sake of parsimony, we use three variables as proxies for corporate governance. The ﬁrst variable, SOE, is a dummy variable that indicates whether a company is an SOE or not. The second variable is the largest shareholder’s ownership proportion, L_Share. Given that the Chinese government is usually the largest shareholder of a China-listed ﬁrm, we adopt a third variable, the proportion of top-10 institutional investors’ share-holding, IIH. Motivating factors for ﬁrm i are represented by the delisting pressure (LOSS and LOSS_ST) and the need for equity ﬁnancing (RSI).Corporateperformanceandﬁnancialcharacteristicsarerepresentedbydiscretionary accruals (DA), ROA and leverage (LEV). Firm size (Size) is our control variable. It is surprising to observe that the considerable number of restatements that occurred in China attracted so little attention. We conjecture that it may be because the market is suspicious of the credibility of ﬁnancial reports and attaches an almost-independent value to listed companies. Stock prices in China rarely reﬂect the value of companies, and stock price changesare rarely an effective reﬂection of thechange in available information. Market irregularities were not uncommon among ﬂedgling companies during our sample period and include such practices as insider trading and institutional manipulation of stock prices (CSRC, 2008). Accounting reporting does not serve as central a role in China’s capital market as it does in a mature market, and ﬁnancial reports that overstate or poorly state a Chinese ﬁrm’s true status may have limited effect on the ﬁrm’s stock valuation. Hence, we arrive at Hypotheses 4 and 5, along with their corollaries: Hypothesis 4. The stock market fails to punish the poor quality of ﬁnancial reports. Within this general hypothesis, we construct the three following corollaries. Hypothesis 4a. The stock market fails to punish the poor quality of ﬁnancial reports following restatement announcements. Weconjecturethatthestockmarketdoesnotreactsigniﬁcantlytotheannouncementof earnings restatements, which means there is no penalty for restatements. For this hypothesis,wefocusontheshort-termstockpricereactiontotherestatementannouncements.In our analysis, we examine various windows up to two weeks before and after the announcements for any information leakage or delay to the stock market: (10,6), (5,2), (1,+1), (+2,+5) and (+6,+10) days around the restatement dates (disclosure dates for enforced restatements and annual report announcement dates for voluntary restatements). Buy-and-hold market-adjusted cumulative abnormal returns (CARs) are used as the return metric. Hypothesis 4b. The stock market does not anticipate the poor quality of ﬁnancial reports before the restatement announcements. We conduct a long-term event study, which is designed to examine the year preceding the restatement announcements. The purpose of this hypothesis is to detect whether any information has been leaked to the stock market, either through insider trading or 176 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 analysts’ warnings, during the period of the ﬁnancial report being restated. Again, buyand-hold market-adjusted CARs are used as the return metric. Failure to reject the ﬁrst two null hypotheses would indicate that the stock market in China does not punish poor ﬁnancial reporting. Hypothesis 4c. The stock market does not penalize the poor quality of accounting that is uncovered by regulatory inspection. As mentioned earlier, the restatements disclosed according to the regulatory inspection outcomes represent required restatements. We test whether the market penalizes enforced restatements by examining the market reaction around the disclosure date of the enforced restatements. We examine a window of (6,+6) months around the disclosure date to allow a reasonable period for routine regulatory procedures before and after the formal announcement. Hypothesis 5. The stock market attaches minimal value to ﬁnancial information. We extend this general hypothesis into two detailed sub-hypotheses, as follows. Hypothesis 5a. Stock price changes poorly reﬂect revelations of low-quality earnings. We conduct two sets of returns/earnings tests on both the level of and change in earnings information. CARi ¼a0 þa1EPSt þa2AdjEPSt1 þei ð2Þ CARi ¼a0 þa1UnEPSt þa2Magt1 þa3Control variablesþei ð3Þ In Eq. (2), CARi is (11,+1) months for a restating company. EPSt is the earnings per share in the reported year t. AdjEPSt1 is the adjusted earnings for the previous year t1; that is, the true earnings per share according to the restatement. We use Eq. (2) to test whether the market comprehensively reﬂects the value of the company by reacting to both current annual earnings and past earnings, which can be naïvely adjusted by the given corrected amount. We conjecture that the market does not react signiﬁcantly, at least to the adjusted past earnings. In Eq. (3), CARi is (11,+1) months for a restating company. UnEPSt is the surprise reported year t’s earnings per share, which is measured by the difference between the current year’s earnings and the expected earnings, represented by the originally reported earnings of t1, given that barely any systematic analyst’s forecasts exist in China so far. Magt1 is the surprise of the past year, t1’s, earnings, which is the per-share scaled restatedmagnitude. Weuse Eq.(3) totest whetherthemarketreacts fullytothechangein the accounting information setting, which includes two surprises: the surprise of current earnings and the surprise of past earnings. We calculate the surprise—or the unexpected part—of the current earnings as the scaled difference between current earnings and oneyear-prior earnings, taking into account the limited scope of analysts’ forecasting in China. Werepresentthesurpriseofpastearningsbythescaledmagnitudeoftherestatement.We conjecturethat,at minimum,themarket does notreactsigniﬁcantly tothesurprise ofpast earnings. Hypothesis 5b. The equity market’s reliance on earnings information is minimal. To test this hypothesis, we conduct the following tests on our sample, using the ERC as a measure of reliance. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 177 CARi ¼aþbUEi þei ð4Þ CARi ¼aþb1UEi þb2UEiTi þei ð5Þ Eq. (4) is tested separately before and after the sample ﬁrms revealed their poor quality. Eq. (5) is the pooled test with a dummy variable Ti representing the period after restatement. We expect all bs in both (4) and (5) to be insigniﬁcant. 5. Description of restatements in China 5.1. Data selection Due to the absence of a restatement database, we manually collected our sample from domestic companies that issued A-shares8 on the Shanghai Stock Exchange and/or the ShenzhenStockExchange.Wesearchedtheannualreportsoftheselistedcompaniesandcollected information from the Correction of Material Accounting Mistakes section in the footnotes. Our sample period runs from January 1, 1999, when the Old Standard for restatements became effective, to December 31, 2005, just weeks before the announcement ofthe 2006 ASBE.As such,the2005annualreportswerethelastannualreportsﬁled bylisted companies under the 2002 ASBE. Excluding restatements due to mergers and acquisitions, dividend distributions and changes in accounting policies and accounting estimation, 1092 restatement announcements were identiﬁed due to accounting misrepresentation, irregularities, fraud or errors during the study period. Market and accounting data are from the CSMAR database. Institutional investors’ information is from the Genius database. 5.2. Data description Table 2 shows that during 1999–2000, 21 companies each year restated their ﬁnancial reports, representing approximately 2% of listed companies. However, the number of restating ﬁrms soared in 2001 and 2002 to 282 and 250, respectively, amounting to 24.96% and 20.85% of public companies. Such a dramatic increase is perceived to be a natural delay following the adoption of the Old Standard. The number of restating ﬁrms dropped in subsequent years, but still comprised more than 10% of listed companies. The reduction may be a direct reﬂection, not only of the restatements of previous years, but also of the 2001 Rules and the CSRC’s intensive and extensive inspection throughout all provinces after observing the increase in qualiﬁed auditors’ reports. What is striking and puzzling is the large proportion of ﬁrms that issued restatements in China. In the United States, although the number of restatements increased during the time frame of our study, the proportion remained low and steady at around 2% by the end of 2005 (Wu, 2003). On April 4, 2001, the CSRC issued the Index of Listed Companies’ Industry Classiﬁcation, whichserves as theindustryclassiﬁcation standard. Theinformation inTable 3adopts this standard for the distribution of restatement companies by industry. 8 Two types of shares of domestic companies are traded on the Shanghai and Shenzhen stock exchanges: A-shares and Bshares. A-shares are traded in Chinese Yuan and B-shares are traded in US dollars. The stock market is dominated by A-share trading. Only 106 B-shares were traded in January 1999; since the beginning of December 2005 the number has been steady at 109. 178 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 Table2 Yearlyrestatementcharacteristicsbystockexchange:1999–2005. Annual report year(t) Numberof restatementﬁrms listedonShenzhen StockExchange NumberofAshareslistedon ShenzhenStock Exchange Percentage (%) Numberof restatementﬁrms listedonShanghai StockExchange NumberofAshareslistedon ShanghaiStock Exchange Percentage (%) Total numberof restatements Total number ofAShares Percentage (%) 1999144503.1174711.49219212.28 2000134512.8885591.432110102.08 200113949428.1414363622.48282113024.96 200211249422.6713870519.57250119920.85 20039249118.7410677013.77198126115.70 20048452615.97718278.59155135311.46 20057653414.238982410.80165135812.15 Total5305621092 Note:t+1istheyearwhentheannualreportforyeartisreleased,inwhichtherestatementofearlieryear(s)isdisclosed. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 179 More than half of the restatements were from ﬁrms in the manufacturing sector, with Machinery,EquipmentandMeter,PetroleumandChemicalProducts,andRubberandPlastics making double-digit contributions to the percentage of all restatement ﬁrms. Following its issuance of the 2001 Rules, the CSRC started to inspect listed companies’ ﬁnancial reports, corporate governance structures and independence from related parties. The Rules require companies to correct any irregularities found in the inspection and to disclosethempubliclywithin30 daysofofﬁcialnotice.Becauseanaccountingrestatement is one possible result of the tightened regulatory inspection, such disclosure makes it possible for us to identify the restatements initiated by the inspection versus those disclosed by ﬁrms voluntarily. Table 4, Panel A, shows the number of enforced restatements disclosed by ﬁrms each year as a result of the inspection ﬁndings, along with those resulting from voluntary disclosure, from 1999 to 2005. The enforced restatements comprise only 6% of the sample and are largely concentrated in the 2001–2003 period. This is a much lower proportion than in the US, where roughly one-quarter of restatements were instigated by the SEC or other regulators. However, the number of voluntary restatements also increased in China during those years, which could beinterpretedasaperceptionbyﬁrmsoftheseriousintentofthe2001 Rules,leadingthem to clean upvoluntarilyrather than becaught by theCSRC. Nevertheless, wedo notexclude the possibility that we may have failed to identify a complete set of enforced restatements due to the simple, possibly incomplete, disclosure by listed companies in general. From what we read of companies’ statements, no ﬁrm ever mentioned that the restatement was suggested by its auditor. Table 3 Industry distribution. Number of ﬁrm observations Percentage of total ﬁrms (%) Agriculture, Forestry, Fishing and Hunting A 36 3.30 Mining B 11 1.00 Manufacturing C 610 55.84 Food, Beverage C0 50 4.56 Textile, Apparel, Leather C1 36 3.28 Wood Product C2 3 0.27 Paper, Printing C3 23 2.10 Petroleum, Chemical Product, Rubber, Plastics C4 123 11.22 Electronic Equipment C5 31 2.83 Metal, Nonmetallic Mineral Product C6 93 8.49 Machinery, Equipment, Meter C7 163 14.93 Medicine, Biologic Products C8 77 7.05 Other manufacturing C9 11 1.00 Electricity, Gas, Water Supply D 64 5.84 Construction E 22 2.01 Transport, Storage F 28 2.55 Information, Technology G 61 5.57 Wholesale and Retail Trade H 80 7.30 Real Estate J 38 3.47 Social Services K 29 2.66 Transmission, Culture L 8 0.73 Conglomerate M 105 9.58 Total 1092 100 Note: The classiﬁcation is based on the Index of Industrial Distribution of Listed Companies, issued by the CSRC on April 3, 2001. 180 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 Table 4 Description of restatements, by year of annual report (t). 1999 2000 2001 2002 2003 2004 2005 Total Panel A: Enforcement by year Number of Voluntary Restatements 19 20 265 231 185 146 159 1025 Percentage (%) 90.48 95.24 93.97 92.40 93.43 94.19 96.36 93.86 Number of CSRC-enforced Restatements 2 1 17 19 13 9 6 67 Percentage (%) 9.52 4.76 6.03 7.60 6.57 5.81 3.64 6.14 Total 21 21 282 250 198 155 165 1092 1 2 3 4 5 6 Total number of ﬁrms Panel B: Frequency of restatements, by number of years restatements occurred Number of Firms 354 170 67 34 11 1 637 Percentage (%) 55.57 26.69 10.52 5.34 1.73 0.16 100.00 Year t1 only Years before t1 only Both year t1 and previous years No disclosure Total Panel C: Distribution of restated years Number of observations 494 125 451 22 1092 Percentage (%) 46.17 11.68 42.15 2.01 100.00 Reason for classiﬁcation Number of restatements in annual report year (t) 1999 2000 2001 2002 2003 2004 2005 Total Panel D: Reasons for restatements by year 1 Mistakes from Subsidiaries 2 7 86 83 85 62 78 403 10% 33% 30% 33% 42% 39% 47% 38% 2 Tax Miscalculation 3 5 87 93 72 57 52 369 14% 24% 31% 37% 36% 36% 32% 34% 3 Costs and Expenses 11 5 80 51 30 26 21 224 52% 24% 28% 20% 15% 17% 13% 20% 4 Depreciation and Provisions 2 3 61 37 36 23 27 189 10% 14% 22% 15% 18% 15% 16% 17% 5 Revenue Recognition 2 1 21 12 11 13 13 73 10% 5% 7% 5% 5% 8% 8% 7% 6 Subsidies Revenue and Tax Refunds 0 1 17 10 7 3 5 43 0 5% 6% 4% 3% 2% 3% 4% 7 Other Mistakes and Misclassiﬁcations 10 12 113 74 56 45 49 359 48% 57% 40% 29% 28% 29% 30% 33% Downward restatements Upward restatements No impact No disclosure Total Panel E: Restatement effect on retained earnings Number of observations 838 181 51 22 1092 Percentage (%) 76.74 16.58 4.67 2.01 100 Obs. Mean Min. 25% Median 75% Max. Std Dev Panel F: Restated amount (in millions of Chinese Yuan, USD1 = CNY8.2768) Restated Amount 1068 13.811 815.467 11.147 2.383 0.172 863.188 62.267 Scaled Restated Amount 1063 0.023 7.768 0.008 0.002 0.000 0.158 0.250 (continued on next page) X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 181 Table 4, Panel B, reports that 637 companies produced a total of 1092restatements from 1999 to 2005. More than half of the sample companies restated just once during the sample period, although some restated several times. One ﬁrm, astonishingly, restated six times in seven years. In all, 494 (46.17%) of the restatements were to correct the prioryear’s ﬁling,9 and 451 (42.15%) of the restatements were to revise the ﬁlings of both the prior-year’s ﬁling and that of earlier year(s) (Panel C). However, we cannot discern the number of years these ﬁrms restated, as such information was not provided in their annual reports. The remainder of the 1092 restatements (147) were to correct mistakes made before the most recent year. Twenty-two observations did not disclose which year(s) they restated. We notice that the reasons for restatements are quite different from those in the United States. We classify reasons for restatements into eight categories in Table 4, Panel D. Most restatements involved more than one reason; therefore, the sum of each column exceeds the annual sample size. The correction of mistakes in subsidiaries when consolidated ﬁnancial reports are prepared tops our list. This type of restatement is more technical, because parent companies cannot fully control the subsidiaries’ ﬁnancial reports to the same extent to which they control their own. The more subsidiaries a parent company has, the more difﬁcult the process of management may be and the greater the likelihood of error and subsequent restatement. Unfortunately, no database contains such information, so we do not have theopportunitytotestthishypothesis.Withoutdetaileddisclosure,weareunabletomeasure its effect on the magnitude of overall restatements. Incorrect tax estimation, which is a rarity among US ﬁrms, is frequently observed in China. Approximately three-quarters of companies in this category underestimated the tax paid, which led us to suspect that such underestimation is more intentional than unintentional. Misstating, mostly under-misstating, the cost of goods sold or operating expenses is quite common in China, whereas manipulation of recognized revenue is not a major vehicle: only 7% of restatements involve inﬂating revenue or recognizing revenue earlier (in the United States, nearly 40% of the restatements are due to the revenue recognition problem). Table 4 (continued) 1999 2000 2001 2002 2003 2004 2005 Total Panel G: Change of external auditors, by year (t) of annual report Number of Firms with Auditor Change in year t 6 5 84 20 33 29 33 210 Percentage (%) 29 24 30 8 16 18 20 19 Number of Firms with Auditor Change in year t + 1 6 4 29 21 14 24 19 117 Percentage (%) 29 19 10 9 7 16 12 11 Notes: Panel D: The sum of the percentage numbers could exceed 100% in each year as most restatements involve more than one reason. Panel F: Restated amount: the amount of change in retained earnings. If the restatement decreased retained earnings, the amount is negative; if it increased, the amount is positive. Scaled restated amount: scaled by total assets at the beginning of year t. Panel G: We lost 14 observations in year t +1; they are not included in the percentage calculation. In annual reports, there is no disclosure of whether the audit ﬁrm quit or was ﬁred. 9 Chinese companies’ ﬁscal year is stipulated to be the same as the calendar year. 182 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 Duringoursampleyears,asmanyas17%ofrestatementsweretocorrectdepreciationor various provision estimates, which is not particularly surprising considering the existence ofmanytypesofprovisionsinChina.In1999,fourtypesofprovisionsrequiredestimation: bad-debt reserve, provision for inventory impairment, provision for short-term investment impairment and provision for long-term investment impairment. Later in 2001, four more types of provisions were added: provision for ﬁxed-assets impairment, provision for intangible-assets impairment,provision for construction-in-progress impairment and provision for entrusted-loans impairment. The estimation of such provisions demands significant professional judgment and becomes quite a challenge to accounting professionals with limited experience10 in a ﬂedging capital market. Meanwhile, various kinds of provisions also offer room for manipulation, which is reﬂected in the CSRC’s 2004 Notice, emphasizing that listed companies should not abuse asset impairments, change-of-accounting estimates or correction-of-material mistakes for the purpose of manipulation. For many years, a tax refund policy has served as a major incentive to encourage export in mainland China. In July 2000, the MOF issued the Rules of Accounting Treatments for Tax Refunds, etc., which clearly demanded the adoption of cash accounting, rather than accrual accounting, for income from subsidies and tax refunds to prevent companies from manipulating earnings via such vehicles. However, a few listed companies continue to openly violate the tax refund rules and adopt homemade recognition practices. Because the simple and general disclosures in one-third of the observations make these restatingcompanies difﬁculttocategorize, wecombinethislargefractionintoa singlecategory that includes unspeciﬁed mistakes, oversights and reclassiﬁcations. Table 4, Panels E and F, shows that more than three-quarters of the events are related to an earlier overstatement of retained earnings; the remaining quarter are either underreported or had no effect on retained earnings. The amount of restatements varies widely, from reducing retained earnings by CNY815 million (USD98.43 million) to increasing retained earnings by CNY863 million (USD104.23 million), with an average decrease of 2.3% of total assets, which is comparable to US percentages. Table 4, Panel G, shows that almost 20% of the ﬁrm-observations changed auditors during the annual report year and 11% during the event year when restatements were disclosed. However, with limited disclosure in annual reports, we are unable to distinguish whether such changes were due to the auditing ﬁrm leaving the employment of the client company voluntarily or under duress. 6. Empirical analysis 6.1. Firm characteristics and potential motivations We nowanalyze thecharacteristics of our sample andseek the potential motivations for these companies to ﬁle incorrect ﬁnancial reports. Table 5, Panel A, reveals that restatement ﬁrms have slightly lower mean total assets compared with other listed companies, but most observations fall within the comparable size range. Table 5, Panel B, compares various company characteristics in the restated year (year t1) of the restatement sample to the control sample. As a control, we used the full set 10 China only embarked on the development of the CPA profession in the early 1980s. The Chinese Institute of Certiﬁed Public Accountants (CICPA) was established in November 1988, and the ﬁrst CPA exam was conducted in 1991. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 183 Table 5 Restatement ﬁrm characteristics. Obs. Mean Min. 25% Median 75% Max. Std Dev Panel A: Firm size (in millions of Chinese Yuan) Restatement Firm Size 1086 1994.15 42.01 725.51 1217.12 2315.29 31426.19 2471.01 Firm size of all A-share Companies 8242 2611.84 21.51 748.89 1270.47 2413.29 520,572 11,635 Mean Median Restatement sample N =911 Control sample N = 3058 t-Stat Restatement sample N = 911 Control sample N = 3058 Wilcoxon Z-stat Panel B: Comparison between the restatement sample and control sample E/P 0.000 0.022 7.39* 0.012 0.022 13.60* B/P 0.502 0.558 7.46* 0.498 0.562 7.43* Income 0.010 0.046 9.61* 0.027 0.050 13.08* OpInc 0.009 0.043 10.33* 0.022 0.045 13.60* EPS 0.055 0.198 7.58* 0.119 0.210 13.92* EG 0.013 0.002 2.12 0.003 0.002 7.84* LEV 0.552 0.440 7.35* 0.506 0.429 10.87* UI 0.005 0.006 3.32* 0.002 0.003 3.03* Age 5.850 5.290 5.48* 6.000 5.000 6.31* L_Share 40.450 45.970 8.60* 37.920 46.230 8.73* SOE 0.670 0.650 0.87 1.000 1.000 0.87 II 0.368 0.436 3.70* 0.000 0.000 3.70* IIH 0.005 0.008 5.07* 0.000 0.000 4.55* TA 0.035 0.019 3.43* 0.022 0.014 3.70* DA 0.016 0.002 3.04* 0.002 0.004 3.34* ROA 0.001 0.027 5.07* 0.022 0.037 11.27* CROA 0.007 0.033 5.92* 0.022 0.039 9.94* ROE 0.019 0.021 2.02 0.048 0.066 9.14* LOSS_ST 0.068 0.022 5.62* 0.000 0.000 7.27* LOSS 0.162 0.081 6.15* 0.000 0.000 7.17* RSI 0.290 0.303 0.76 0.000 0.000 0.76 Notes: Panel A: Firm size: the total assets at the end of year t. Firm size of all A-share companies: the size of A-share listed companies across the board from 1999 to 2005. Panel B: All are measured at the end of year t1, the restated year). E/P: Fiscal Operating Earnings/Market Capitalization. B/P: Book Value/Market Capitalization. Income: Net income before Extraordinary Items/mean Total Assets. OpInc: Operating Income/mean Total Assets. EPS: Earnings per share. EG: Earnings Growth, (Net Incomet1Net Incomet2)/Total Assets. LEV: Leverage, Total Debt/Stockholders’ Equity. UI: Below-the-line items, (income from investment +non-operating income+ subsidies)/Total Assets. Age: The number of years ﬁrms have been listed on the stock exchange. L_Share: The percentage stock holding by the largest stakeholder. SOE: Dummy variable. State=1 if a state-owned enterprise, and 0 otherwise. II: Institutional investor among top 10 shareholders. If there is an institutional investor, II = 1, and 0 otherwise. IIH: Institutional investor’s holding proportion among top 10 shareholders. TA: Total accruals=(Operating IncomeCash Flow from Operation)/Total Assets. DA: Discretionary accruals, calculated with modiﬁed Jones Model. ROA: Return on Assets =(EBITTax)/Total Assets. CROA: Return of Operating Income on Assets. ROE: Return on Equity= Net Income/Shareholders’ Equity. LOSS: Dummy variable. LOSS= 1 when there was a loss in year t1, and 0 otherwise. LOSS_ST: Dummy variable. LOSS_ST= 1 when there was a loss in both years t1 and t2, and 0 otherwise. RSI: Dummy variable. RSI = 1 if rights or secondary issuance occurred in t1, t2, or t3, and 0 otherwise. ⁄Statistical signiﬁcance at the 10% level for two-tailed tests. Statistical signiﬁcance at the 5% level for two-tailed tests. * Statistical signiﬁcance at the 1% level for two-tailed tests. 184 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 of listed companies that did not ﬁle restatements and relied on their test year’s information to calculate the results. The reason for such a ‘‘naïve’’ matching method is the relatively small number of companies listed on either the Shanghai or Shenzhen stock exchanges by the end of 2005 (just over 1300 in all). If we adopted the conventional method of matching by industry and ﬁrm size among the non-restatement companies, then there would be very few comparable companies and we would be unable to construct an effective matched sample. Table5,PanelB,showsthattherestatementcompaniesgenerallyhavelowervaluations, poorer performance,higherleverage and greaterlosses comparedwiththecontrolsample. The largest shareholders of the sample ﬁrms hold a lower proportion of shares than those of the control sample. There is no signiﬁcant difference between the two samples in terms of the number of SOEs. Notably, we observe that restatement ﬁrms have been listed signiﬁcantly longer on the stock exchange, with a mean age of almost 6 years compared with 5.29 years for control ﬁrms. Their ROA, core ROA and ROE are all signiﬁcantly lower than those of the control sample. Meanwhile, the benchmark for determining whether a ﬁrm can offer rights or a secondary issue, ROE, is far from the 10% threshold. The mean ROE is slightly below 0. This result contradicts our expectation in Hypothesis 1. Restatement companies have a higher frequency of single-year losses and consecutive losses, consistent with our expectation in Hypothesis 2. Although Table 5, Panel B, shows that approximately 29.6% of the sample had rights or secondary issuances in the two years prior to and in the most-recent restated year, combining all of the results, we would not claim that companies’ manipulation during the restated year is aimed principally at achieving equity ﬁnancing needs, but is more likely to be associated with a struggle against poor performance and delisting pressures. ConsistentwiththeChinaresearchliterature,weﬁndthatbothtotalaccrualsanddiscretionary accruals are generally negative during the most recent restated year. Perplexingly, however, we also observe that both types of accruals in the test sample are signiﬁcantly lower than in the control sample. No explanation for this ﬁnding comes to mind. 6.2. Determinants of restatements Here,weexaminewhetherthecompanycharacteristicsduringthecorrectedyear(t1) can collectively explain the restatement phenomenon in China. These characteristics may offer some guidance to regulators on which aspects to explore when seeking out poorquality ﬁrms. Table 6 provides the Pearson correlations for all of the potential test variables. The performance variables are highly correlated with one another. We leave ROE in the table, as it is an important threshold in China’s capital market. However, ROE is widely manipulated in the Chinese market due to its role as a threshold. We omit it from our ofﬁcial empirical model (Eq. (2)) and adopt only ROA in the ﬁnal test. We similarly omit earnings growth from our ofﬁcial empirical tests. Naturally, ROA is highly correlated with LOSS, LOSS_ST and LEV, which we will control for in the regressions. Table 7 examines the company characteristics and incentives that collectively lead to a restatement. The ﬁnal regression is conducted on the complete sample and includes all variables. It shows that companies with lower proﬁtability, higher leverage, lower holdings by institutional investors, lower ownership concentration and SOEs are more likely to be restatement ﬁrms. Because the incentive variables for delisting pressure, LOSS and X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 185 Table6 Pearsoncorrelations. RDAROAROEEGLOSS-STLOSSLEVL_ShareSOERSISize R1 DA0.0481 ROA0.0860.679***1 ROE0.0270.3560.5691 EG0.0170.3220.4530.3771 LOSS_ST0.0800.3820.4720.2910.0921 LOSS0.1140.3540.4980.3320.2570.5331 LEV0.1380.1500.3480.1950.0230.1720.2191 L_Share0.1350.0580.1210.0850.0030.0790.1280.1341 SOE0.0140.0000.0300.0240.0070.0090.0320.0130.2681 RSI0.0120.0930.072*0.045***0.0290.0840.1150.1080.0010.0381 Size0.0590.0460.1460.0770.0020.1400.1730.1710.2130.1420.1431 IIH0.0690.0620.1080.0480.0260.0620.1150.0430.0110.0250.030*0.171*** Notes:SeeTable5notesforvariabledeﬁnitions. R:Adummyvariable.R=1ifrestatement,otherwise0foralllistedcompanies. Statisticalsigniﬁcanceatthe10%level. Statisticalsigniﬁcanceatthe5%level. Statisticalsigniﬁcanceatthe1%level. 186 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 Table7 Determinantsofrestatements. DependentVariablesModel1Model2Model3Model4Model5Model6 (predictedsign)CoefﬁcientPr>ChiCoefﬁcientPr>ChiCoefﬁcientPr>ChiCoefﬁcientPr>ChiCoefﬁcientPr>ChiCoefﬁcientPr>Chi Intercept0.2080.8290.4420.6350.1960.8330.1510.8760.1780.8540.0980.919 SOE(+)0.3200.0000.2960.0010.2930.0000.3210.0000.3220.0000.3220.000 L_Share(+)0.017<.0000.019<.0000.019<.0000.017<.0000.016<.0000.017<.000 IIH()8.4400.0039.2520.0019.8730.0007.9680.0057.9260.0047.9160.005 LOSS(+)0.3650.0130.534<.0000.1650.354 LOSS_ST(+)0.0530.8390.4690.0390.1070.6860.1610.549 RSI(+)0.0550.5340.0380.6650.0570.5090.0440.6240.0430.6290.0470.600 DA(+)0.1420.7470.2180.6020.2700.5230.2210.6410.2050.6670.2030.183 ROA()0.5010.2010.8430.0452.0310.0022.1420.0021.7140.039 LEV(+)1.579<.0001.447<.0001.449<.0001.452<.000 Size(?)0.0780.1050.0060.8960.0140.7620.0690.1540.0690.1470.0680.163 Chi-square162.63130.91120.07165.98166.15166.99 Numberofobservations3928 Notes:Allvariablesareforyeart1. Boldnumbersindicatestatisticalsigniﬁcanceatthe5%level. SeeTable5forvariabledeﬁnitions. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 187 LOSS_ST, are highly positively correlated with the performance variables, ROA and leverage, we test the model alternatively with and without ROA or LEV. The results imply that the delisting pressure is an incentive to manipulate earlier years’ earnings and contributes to later restatement. All tests fail to demonstrate that ﬁrms’ ﬁnancing needs serve as an incentive for manipulation in our sample. We are surprised to ﬁnd that our test produces the opposite sign from that expected for ownership concentration; that is, the lower the proportion of the largest shareholder’s stake, or the lower the ownership concentration, the higher the probability that a listed company will restate. We interpret this result as follows: in China, the ‘‘shell’’ of a listed company is a very valuable and limited resource due to the slow CSRC approval process. When a company faces proﬁtability stress or potential losses, the largest shareholder will pump proﬁts into the company by arranging, for example, related-party transactions, a very common practice in China. However, companies with lower ownership concentration will not beneﬁt from their largest shareholders when experiencing the same pressure, as the largest shareholders have insufﬁcient incentive to rescue the ailing company. Such companies then either face being delisted, or resort to accounting manipulation, which is later reversed through restatement. Such conjecture is supported by the signiﬁcant negative correlation between ownership concentration and the delisting pressure variables, LOSS and LOSS_ST. Our control variable of ﬁrm size is negatively but not signiﬁcantly related to restatements. We also conduct extra tests within subsamples, limiting the observations to downward restatements and core earnings related to restatements. The results, which we do not include in the tables, are consistent with our primary tests. Combining all factors, one can understand why a large SOE, whose largest shareholder is the government, will remain above the ﬁnancial fray and avoid the need to restate: it is simply because the government will not let that happen. Table 8 Cumulative Abnormal Returns (CAR) around restatement announcements. Full sample 1999–2005. Period N Mean Median Std Dev t-Stats P-value Signed rank P-value Panel A: Short-term CAR with available return data Day (10,6) 1088 0.0034 0.0058 0.0572 0.1361 <.0001 Day (5,2) 1087 0.0039 0.0051 0.0468 0.0124 0.0009 Day (1,+1) 1087 0.0001 0.0048 0.0573 0.7802 0.0003 Day (+2,+5) 1087 0.0008 0.0031 0.0537 0.4227 0.3721 Day (+6,+10) 1085 0.0037 0.0002 0.0558 0.0361 0.1314 Panel B: Long window—periods leading up to the announcement Day (251,211) 1007 0.0104 0.0126 0.1053 0.0024 0.0011 Day (210,170) 1012 0.0107 0.0147 0.1049 0.0013 <.0001 Day (169,128) 1020 0.0057 0.0113 0.1183 0.1275 0.0041 Day (127,86) 1023 0.0114 0.0176 0.1189 0.0022 <.0001 Day (85,44) 1030 0.0162 0.0164 0.1188 <.0001 <.0001 Day (43,22) 1089 0.0048 0.0017 0.0821 0.0714 0.4887 Day (21,11) 1089 0.0018 0.0059 0.0713 0.3442 0.0001 Notes: Signiﬁcance levels are for two-tailed tests. A market-adjusted returns model is adopted. We obtained our data from the CSMAR database. Individual stock’s daily abnormal return is calculated as ARit = RitRmt. Rmt is the market return, represented by the A-share composite index daily return. Sample daily abnormal returns are calculated as ARt ¼ 1 NPN i ARit. Portfolio CARs are calculated as CARBE ¼PE t¼BARt. B and E, respectively, represent the beginning and ending days around event day 0. 188 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 6.3. Consequences of revealing poor earnings quality Even though ﬁnancial restatements in China have become a prevailing phenomenon, it remains largely below the radar, which seems perplexing. We now test Hypothesis 4 and offer explanations. Following most research on restatements in the US, we conduct event tests and let their results tell the story. Table 8, Panel A, shows the short-term CARs for our restatement sample. Contrary to the strikingly negative results from US data (CARs of approximately 11%), China’s stock market reacts very weakly, barely registering any reaction to restatement announcements. The mean CAR values for the 3-day period around the announcement (1,+1) are not signiﬁcant, although the median CARs are signiﬁcant, with a market reaction of 0.48%. During the week ahead of the annual report release, however, the market reacts in a signiﬁcantly negative way, a drop of roughly half of one percent. These results show that investors in China’s stock market fail to punish the poor quality of ﬁnancial reports following restatement announcements as severely as investors in the US stock market. Nevertheless, investors may perceive such poor-quality accounting ahead of the restatements due to possible information leakage. Hence, we look back over a longer period prior to the release of the annual report and ﬁnd that the results are quite mixed. We divide the timeframe into several periods for examination. We ﬁnd there is a slight decline one year before the release of the report. However, the CARs of some timeframes are signiﬁcantly negative, whereas others are not. Some are signiﬁcant for the median, but not the mean. Overall, the magnitude is limited and the mixed results do not permit us to reject Hypothesis 5, which states that investors do not anticipate the poor quality of ﬁnancial reports. Long-term tests up to six months before the restatement announcements are presented in Table 9, Panel C; these fail to reveal a post-announcement drift, which implies that the poor qualityofﬁnancial reports negates thepotential for themarketto digest theinformation and eventually react in a rational, albeit delayed, fashion. Table 9 CumulativeAbnormalReturns(CAR)aroundrestatementannouncements.CSRC-enforcedsubsample1999– 2005. Period N Mean Median Std Dev t-Stats P-value Signed rank P-value Panel A: Short-term CAR with available return data Day (10,6) 67 0.006 0.000 0.047 0.289 0.666 Day (5,2) 67 0.001 0.004 0.039 0.776 0.384 Day (1,+1) 67 0.011 0.010 0.042 0.028 0.014 Day (+2,+5) 67 0.007 0.004 0.055 0.273 0.482 Day (+6,+10) 67 0.008 0.000 0.058 0.237 0.955 Day (+11,+21) 67 0.013 0.012 0.074 0.134 0.085 Panel B: Long window—periods leading up to the announcement Day (251,211) 63 0.008 0.012 0.067 0.231 0.024 Day (210,170) 64 0.009 0.017 0.062 0.189 0.018 Day (169,128) 65 0.005 0.010 0.110 0.127 0.026 Day (127,86) 67 0.002 0.013 0.098 0.767 0.000 Day (85,44) 67 0.014 0.015 0.109 0.000 0.000 Day (43,22) 67 0.003 0.001 0.076 0.712 0.879 Day (21,11) 67 0.009 0.018 0.059 0.177 0.017 Notes: Refer to those for Table 8. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 189 Our analysis also indicates that the stock market only slightly penalizes the poor quality of accounting uncovered by the CSRC’s inspection following restatement announcements. InTable10,PanelA,theCARsof1%aresigniﬁcantbothforthemeanandthemedianonly for the (1,+1) period around the event date. Usually, the regulatory inspection will take a few months to conclude, leading us to look back over a longer period preceding the CSRC’s decision. During the six months preceding the restatement in Table 10, Panel B, the CAR values (mean 1.4%, median 1.5%) are signiﬁcant only between the fourth month and the second month. The CARs from the remaining time frames are mostly negative and insigniﬁcant. Overall, the stock market reaction offers an ambiguous and different picture of restatements from that of the US stock market. We can claim that the stock market in China either fails to punish poor ﬁnancial reporting or does not punish it enough. 6.4. Usefulness of accounting information Puzzled by the market’s failure to react to the restatement phenomenon, we suspect that investors may attach a different value to accounting information in China, and thus may partially or completely ignore such information. Restatements are disclosed in ﬁrms’ annual reports, which also offer a great deal of other information, mainly relating to the current reported year. Buried within, the revealing of restatements may simply be missed by investors. Hence, we test two dimensions of the return/earnings relationship: the level of and the change in information. Table 11, Panel A, shows how the stock market reacts to the current year’s earnings and the past year’s corrected earnings. We obtain the latter by adding the restated amount back onto the originally reported earnings for those sample ﬁrms that restate at least the previous year’s results. The results show that the market does respond to accounting information: its reaction is clearly and signiﬁcantly related to the current year’s earnings. However, it ignores the corrected past year’s earnings, which is also new information re Table 10 Cumulative Abnormal Returns (CAR) around restatement announcements. Subsample of core earnings reasons 1999–2005. Period N Mean Median Std Dev t-stats P-value Signed rank P-value Panel A: Short-term CAR with available return data Day (10,6) 376 0.004 0.005 0.054 0.169 0.029 Day (5,2) 375 0.003 0.005 0.053 0.214 0.089 Day (1,1) 375 0.004 0.006 0.057 0.164 0.004 Day (+2,+5) 374 0.003 0.001 0.063 0.367 0.765 Day (+6,+10) 372 0.005 0.000 0.061 0.131 0.104 Day (+11,+21) 369 0.007 0.004 0.075 0.083 0.156 Panel B: Long window—periods leading up to the announcement Day (251,211) 362 0.000 0.002 0.046 0.432 0.284 Day (210,170) 362 0.011 0.015 0.107 0.061 0.019 Day (169,128) 365 0.014 0.019 0.125 0.032 0.011 Day (127,86) 365 0.020 0.036 0.187 0.038 0.000 Day (85,44) 364 0.026 0.025 0.117 <0.000 <0.000 Day (43,22) 376 0.006 0.002 0.088 0.216 0.592 Day (21,11) 376 0.001 0.005 0.087 0.898 0.205 Notes: Refer to those for Table 8. 190 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 leased to the market. These results imply that accounting information is partially useful to the equity market. Investors seem to pay attention, naïvely, only to the current year’s accounting information while ignoring information about the past. Table 11, Panel B, presents the test results on the market reaction to the change in information. Such change has two components in our tests: the change in the current year’s earnings, proxied by unexpected earnings, and the change in past earnings, represented by the magnitude of the restatement. We use a dummy variable, Loss, to control for the major characteristics of the current year’s earnings, along with variables describing the restatement characteristics, and ﬁnd the marginal signiﬁcance of the magnitude disap Table 11 Usefulness of earnings information. CARi = a0 + a1EPSt +a2AdjEPSt1 +e Coefﬁcient t-Stat Panel A: Returns on earnings Intercept 0.007 0.98 EPSt 0.142 9.23 AdjEPS 0.010 0.93 R2 9.18% F value 43.33 Number of observations 838 CARi Model 1 Model 2 Model 3 Model 4 Coefﬁcient t-Stat Coefﬁcient t-Stat Coefﬁcient t-Stat Coefﬁcient t-Stat Panel B: Returns on earnings surprises Intercept 0.006 0.86 0.066 0.70 0.061 0.65 0.079 0.83 Un_EPS 0.051 4.94 0.030 2.82 0.031 2.87 0.030 2.83 Magnitude 0.051 1.87 0.027 0.98 0.031 1.12 0.027 0.95 Loss 0.105 5.82 0.107 5.92 0.103 5.70 CSRC 0.017 0.61 0.017 0.62 0.015 0.54 Tax 0.013 0.86 Subsidiary 0.011 0.75 Revenue 0.035 1.23 Core 0.006 0.41 0.006 0.41 Firm_Size 0.007 0.89 0.007 0.84 0.008 1.01 R2 2.57% 5.73% 5.86% 5.70% F value 14.38 11.02 11.26 8.48 Number of observations 1015 989 989 989 Notes: CARi: Cumulative abnormal returns of (5,+1) months around the restatement event date. EPSt: EPS for year t. AdjEPSt1: True EPS for year t1, naively adjusted by restated amount. Un_EPS: Unexpected earnings per share (EPS): the reported EPS subtracts’ expected EPS, which is last year’s EPS. Magnitude: Restated amount per share. Loss: A dummy variable. Loss =1 if the company experienced a loss in year t, and 0 otherwise. CSRC: A dummy variable. CSRC = 1 if the restatement is CSRC-enforced, and 0 otherwise. Tax: A dummy variable. Tax =1 if the restatement is tax related, and 0 otherwise. Subsidiary: A dummy variable. Subsidiary =1 if the restatement is a mistake from a subsidiary, and 0 otherwise. Revenue: A dummy variable. Revenue = 1 if the restatement is revenue recognition related, and 0 otherwise. Core: A dummy variable. Core = 1 if the restatement reason(s) is/are related to revenue, cost & operating expenses, and/or depreciation and provision, and 0 otherwise. Firm size: The log form of total ﬁrm assets. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 191 pears. The change in the current year’s earnings and the loss character of its earnings are captured signiﬁcantly by investors, whereas the change in the previous year’s earnings information via outright admission of poor-quality earnings for the previous period is completely ignored. These results are consistent with those in Table 11, Panel A. Next, we test the credibility of accounting information to investors using annual data. The dependent variable is unexpected annual earnings, which we compute by subtracting theprioryear’searningspersharefromthecurrentyear’searningspershare.Theindependent variable is the corresponding CAR for (11,+1) months around the annual report date. As demonstrated in Table 12, before the restatement, the ERC of b1 is signiﬁcantly positive (0.356), and after the restatement it drops to 0.231. The drop is signiﬁcantly negative at the10% level, as shownby b2 =0.127 for dummyUE in thepooled sample. This can be interpreted as the market’s acknowledgement that revealing poor accounting qualityisbadnews,althoughtheERCsinChinaarelow.Ourtestsshowthattheyarebelow0.4, which is much lower than in the United States, where they are above 1. Such a contrast implies that investors in China generally attach much lower value to accounting information. 6.5. Robustness tests We also conduct robustness tests by taking out observations due to miscalculations, typos and postings by mistake due to human error. All statistical results remain consistent with earlier tests.11 The CARs for the year-long window leading up to the annual report announcement increase slightly from around 6% to around 10%, which again indicates that investors do not punish poor ﬁnancial reporting as much as they should. 7. Conclusion China’scapitalmarkethasbeenrapidlyimprovingsincetheendofthe1970s,howeverit has yet to mature and become as efﬁcient as developed markets. Our accounting-based research yields a variety of test results that collectively indicate that low-quality accounting Table 12 Earnings credibilityearnings response coefﬁcients. CARi ¼aþbUEi þei (4) CARi ¼aþb1UEi þb2UEiTi þei (5) t1 t Pooled Coefﬁcient t-Stat Coefﬁcient t-Stat Coefﬁcient t-Stat Intercept 0.001 0.14 0.058 5.25 0.028 4.14 UE 0.376 8.21 0.244 5.71 0.392 7.11 UET 0.136 2.04 F value 67.44 32.61 48.63 Adj R2 0.072 0.036 0.053 Number of observations 857 857 1715 Notes: CARi: Cumulative returns for (11,+1) months around annual report date. UEi: Unexpected earnings of year i: the difference between the reported earnings and the expected earnings, i.e., the prioryear’s earnings, scaled by the stock price of the day before the announcement date. Ti: Dummy variable. T =1 if UEi is for year t, and T =0 if UEi is for year t1. 11 We do not list these tables in the paper. They are available upon request. 192 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 reporting persists in China. Moreover, accounting reporting has yet to play the signiﬁcant role in investment behavior and philosophy that it does in mature markets. Various aspects of capitalmarkets canoffer insights intoour testresults on marketreaction. We now examine these in turn. Short-selling,animportantstockmarketmechanism,wasnotavailableeitherdirectlyor indirectly to investors during our sample period.12 Investors’ inability to sell short deters a stock’s ability to approach an efﬁcient price in a timely manner. The investor structure is disproportionate relative to that in mature markets. The scale ofinstitutionalinvestorsissmall. Individualinvestors,especiallymediumandsmallinvestors, account for a signiﬁcant proportion of trading accounts and trading turnover. Shortterm speculation dominates long-term investment (CSRC, 2008). Our test results imply that accounting information has yet to play a signiﬁcant role in investment behavior and philosophy. Developing institutional investment and improving investors’ education are suggested to be crucial tasks for the government and regulators. We call for effective regulation on the disclosure of ﬁnancial reporting. Poor disclosure causes market inefﬁciency in China. Throughout our research, our attention was caught by the limited amount of information disclosed. For example, when a restatement involves more than one year, there is no indication of the number of prior years that it affects, and when there is more than one reason for restatement, there is no guidance on the relative weight of the reasons. Although the 2004 Notice deﬁned the need for timely and separate disclosure of signiﬁcant events, including restatements, the speciﬁcation is not reliably followed. We also call for more vigorous regulatory and administrative enforcement. The CSRC does not have the power to determine appropriate monetary penalties for regulation-violating corporate behavior. The executor of the justice system, the court, is legally equipped with the power to decide the size of monetary penalties, based on the Law of the People’s RepublicofChinaonAdministrativePenalty13 andtheSecuritiesLawofthePeople’sRepublic of China.14 Nevertheless, the penalty’s cap is so low—CNY 600,00015—that it cannot act as a real ﬁnancial deterrent to violating companies. Alternative punishments come from the regulatory body of the CSRC, which could either exert pressure on a listed company’s future application for rights or secondary issuances, or publicly criticize violating companies on the stock exchange. Disallowing reﬁnancing would be a substantial discouragement to a company with ﬁnancing needs. However, as our tests show, the ﬁnancial status of companies thatissuerestatementsgenerallyforbidsthemfromeffectivelyapplyingforequityﬁnancing. Financing needs may not even appear in their timetable because survival has a higher priority than ﬁnancing. As for the option of public criticism, it does not inﬂict much real imme 12 Investors still cannot short-sell stocks directly, but can do so via put options on a limited number of companies. The ﬁrst (call) option in China was of Baosteel Co., Ltd., listed and traded on the Shanghai Stock Exchange on August 22, 2005, whereas the ﬁrst put option was not issued until May 30, 2006, on China Kweichow Maotai Distillery Co., Ltd. By the end of 2007, there were only 27 options issued in total, 21 of which had already expired with six still outstanding; 10 of the 27 were put options. (Shanghai Stock Exchange: http://www.sse.com.cn) 13 The Law was passed at the fourth Session of the Eighth National People’s Congress and promulgated by Order No. 63 of the President of the People’s Republic of China on March 17, 1996. It became effective on October 1, 1996. 14 The Securities Law was passed at the sixth Meeting of the Standing Committee of the Ninth National People’s Congress on December 29, 1998, revised at the 18th Meeting of the Standing Committee of the Tenth National People’s Congress of the People’sRepublicofChinaonOctober27,2005,accordingtotheDecisiononRevisingtheSecuritiesLawofthePeople’sRepublic of China made at the 11th meeting of the Standing Committee of the Tenth People’s Congress on August 28, 2004. 15 CNY600,000USD72,464 during our sample period. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 193 diate damage. There are insufﬁcient regulatory and administrative penalties in China and they need to be enforced and substantiated. Class action lawsuits, a commonly practiced US legal vehicle to inﬂict damage, are a rarity in China. First, from our tests on stock market reaction, a company that reveals mistakes in previously released ﬁnancial reports incurs little if any damage. In fact, there is hardly any damage to seek. Second, although there is damage from the decline in the individual stock price, China’s courts simply will not accept damage cases as, to the courts, such cases are not as important as numerous other civil lawsuits. Implicitly, such reality encourages companies to report poor-quality ﬁnancial statements because later discovery and restatement will be virtually costless. The introduction of a legal procedure for processing cases, along with a punitive legal system to deal with violations of accounting rules, should provide a valuable complement to the current structure of China’s capital market. China’s credit market falls far behind the mature markets worldwide. The scale of the credit market, especially the corporate bond market, is quite small16 (CBRC, 2006). Bank loansarethemajorformofcorporatedebt.Theprocessofintroducingafreelendingrateinto the market from the People’s Bank of China (PBOC) offers a few crucial implications for the debt capital market. As with reforms in many other aspects of the economy, the introduction of a free lending rate is also a gradual process. The ﬂoating range of ﬁnancial institutions’ lending interest rate is completely regulated by the PBOC. Before January 1, 2004, the ceiling was capped at 10% above the base rate.17 It was relaxed to 70% for the next nine months before beingtotallyfreed.Under the regulatedlendingsystem, therate acompany receiveddid not necessarily reﬂect the rate it should have received according to the company’s overall risk–risk that includes accounting quality as a crucial component. Speciﬁcally, poor accounting quality was not penalized fairly with an appropriately higher lending rate. The low additional cost from loan borrowings cannot effectively prevent companies from providing lowquality ﬁnancial reports. Unfortunately, we are unable to test this hypothesis directly, because interest expense is combined with other operating expenses in all existing databases. During most of our sample period, commercial banks did not have sufﬁcient incentive to distinguish clearly among companies with different credit levels until the big four state banks and top-tier banks were listed on the Hong Kong Stock Exchange. This began only in mid2005 after shaking off a signiﬁcant amount of non-performing loans, following international banking rules and accounting standards, and offering executive stock option plans. From examining accounting restatements in our sample of China’s A-share listings, we conclude that companies with weak proﬁtability, a state-controlled shareholder and diffused ownership tend to report poor ﬁnancial statements and later restate. In an emerging marketsuch asChina’s,however, weﬁndthat thestock marketdoes not reactsigniﬁcantly to restatements forced by low-quality accounting. The stock market is only able to digest partial accounting information and accounting credibility is low. These ﬁndings, together with the inefﬁcient debt market, weak regulatory system and legal punitive system discussed above, underscore why restatements were such a widespread phenomenon during our sample period. We show that in China, accounting credibility has lower value and an 16 By the end of 2006, the total value of the credit market was 28.7% of GDP, compared to 188% in the United States. The value of corporate bonds was 1.44% of GDP versus 125.72% in the United States. 17 In 1998, the ceiling was increased to 20% above the base rate for small enterprises, and to 30% for medium and small enterprises in 1999.The ﬂoor lending rate has remainedsteady at 10% below the base rate. For example, on January1, 2004, the PBOC’s one-year base lending rate was 5.31%; the range of lending rates would therefore be 4.78–9.03%. 194 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 accounting misstatement is much less costly than in a mature market, such as that of the United States, because the market mechanism fails to deter ﬁrms’ misstatement behavior. We advocate thereinforcement of market regulation and supervision, strengthening of the legal system, further improvement of free-market mechanisms and continuous investor education in China’s capital markets. Webelieveourstudy tobeverytimely,given thatitechoestherecentspiritandreforms of the Chinese government and regulators. The release of the Opinions of the State Council on Promoting the Reform, Opening and Steady Growth of Capital Markets (the Opinions) on January 31, 2004, sees the role of capital markets as sovereign and strategic for national economic development. The CSRC concurrently implemented a series of reforms to reﬁne market infrastructure and functionality—reforms that include improving the quality of listedcompaniesandfacilitatinginstitutionalinvestors’entranceintothecapitalmarket.18 Interestingly, we ﬁnd that our sample size starts to decrease during the ﬁnal two to three years of our 1999–2005 sample period. This coincides with: (1) the CSRC’s adoption of a decentralized supervision system in 2004, a measure intended to improve the quality of listed companies and whose regulatory efﬁciency was immediately evident (CSRC, 2008); (2) the CSRC’s Rule 19, introduced at the end of 2003, and the 2004 Notice on Further Improving Financial Information Disclosure of Listed Companies; (3) the CBRC’s introduction of a free lending-rate system in 2004; and (4) the listing of big banks in Hong Kong since 2005, following various international industry standards. Accounting quality is improving, in terms of the decreasing number of ﬁrm restatements, as a result of regulatory efforts and a more extensivefree-marketmechanism.WealsoexpectthatwhenChina’scapitalmarketachieves maturity in the near future, investors’ behavior will change commensurately towards that of investorsinmaturemarkets.Speciﬁcally,inrelationtoourstudiedcases,investorswilleffectively differentiate bad accounting quality from good practices. Acknowledgements We appreciate the comments of seminar participants at the Chinese University of Hong Kong, HongKongPolytechnic University,CityUniversity ofHongKong,Hong KongUniversity of Science & Technology–Seoul National University–Singapore Management University Tri-School Research Camp, Peking University, Tshinghua University, Fudan University, Shanghai University of Finance and Economics, Nanjing University, University of Edinburgh and the 4th Symposium of China Journal of Accounting Research. Specially, we are indebted to Baruch Lev, James A. Ohlson, Katherine Schipper, T.J. Wong and Richard Tafﬂer for their valuable comments and advice. We would also like to thank Dr. Bin Qi, Director of the China Securities and Regulatory Commission (CSRC) Research Center and Lin Zhu, General Manager of the Bank of China (Headquarters) Risk Management Department, for sharing their expertise on China’s capital markets. Xia Wang thanks China’s National Science Foundation (Project Number: 70872107) for their ﬁnancial support of this work. References Agrawal,A., Chadha,S., 2005. Corporate governanceand accounting scandals.JournalofLawand Economics 48 (2), 371–406. 18 Other measures include the initiation of the non-tradable share reform, restructuring of securities companies, reform of the share issuance system and the introduction of a free market for investment funds. X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196 195 Aharony, J., Lee, C.W.J., Wong, T.J., 2000. Financial packaging of IPO ﬁrms in China. Journal of Accounting Research 38 (1), 103–126. Anderson, K.L., Yohn, T.L., 2002. The Effect of 10K Restatements on Firm Value, Information Asymmetries, and Investors’ Reliance on Earnings. Working Paper. Georgetown University. Burns, N., Kedia, S., 2006. The impact of performance-based compensation on misreporting. Journal of Financial Economics 79, 35–67. China Banking and Regulatory Commission (CBRC), 2006. China Financial Market Development Report. Chen, K.C.W., Yuan, H., 2004. Earnings management and resource allocation: evidence from China’s accounting-based regulation of rights issues. The Accounting Review 79 (3), 645–665. China Securities and Regulatory Commission (CSRC), 2008. China Capital Market Development Report. Bilingual version. Dechow, P., Sloan, R., Sweeney, A., 1995. Detecting earnings management. The Accounting Review 70, 193– 225. DeFond, M., Jiambalvo, J., 1991. Incidence and circumstances of accounting errors. The Accounting Review 66, 643–655. Fan, J.P.H., Wong, T.J., 2002. Corporate ownership structure and the informativeness of accounting earnings in East Asia. Journal of Accounting and Economics 33, 401–425. Feroz, E.H., Park, K., Pastena, V.S., 1991. The ﬁnancial and market effects of the SEC’s accounting and auditing enforcement releases. Journal of Accounting Research 29 (Supplement), 107–148. Grifﬁn, P., 2003. League of their own? Financial analysts’ responses to restatements and corrective disclosures. Journal of Accounting, Auditing & Finance 18 (4), 479–518. Hribar,P.,Jenkins,N.T.,2004.Theeffectofaccountingrestatementsonearningsrevisionsandestimatedcostof capital. Review of Accounting Studies 9, 337–356. Kinney Jr., W.R., McDaniel, L.S., 1989. Characteristics of ﬁrms correcting previously reported quarterly earnings. Journal of Accounting and Economics 11, 71–93. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., 1999. Corporate ownership around the world. Journal of Finance 54, 471–518. Lev, B., Ryan, S., Wu, M., 2008. Rewriting earnings history. The Review of Accounting Studies 13, 419–451. Lu, J., 1999. An empirical study on the earnings management of loss listed companies in China. Accounting Research 9, 25–35 (in Chinese). Palmrose,Z.,Richardson,V.,Scholz,S.,2004.Determinantsofmarketreactionstorestatementannouncements. Journal of Accounting and Economics 37, 59–89. Richardson, S., Tuna, I., Wu, M., 2003. Predicting Earnings Management: The Case of Earnings Restatements. Working Paper. University of Pennsylvania. Scholz, S., 2008. The Changing Nature and Consequences of Public Company Financial Restatements 1997– 2006. A Department of the Treasury Survey. April 2008. Shleifer, A., Vishny, R., 1997. A survey of corporate governance. Journal of Finance 52, 737–783. Srinivasan, S., 2005. Consequences of ﬁnancial reporting failure for outside directors: evidence from accounting restatements and audit committee members. Journal of Accounting Research 43, 291–334. Turner, L., Dietrich, J.R., Anderson, K.L., Bailey, A.D., 2001. Accounting Restatements. Working Paper. Ohio State University. Wang, X., Zhang, W., 2004. Incidence of accounting errors in china’s listed companies. Accounting Research (in Chinese). Wu, M., 2002. Earnings Restatements: A Capital Market Examination. Doctoral Dissertation. New York University. Wu, M., 2003. Earnings Restatements: A Capital Market Perspective. Working Paper. New York University. Zhang, W., 1999. China’s SOE Reform: A Corporate Governance Perspective. Working Paper. Peking University. 196 X. Wang, M. Wu/China Journal of Accounting Research 4 (2011) 167–196