Abstract
China Journal of Accounting Studies, 2014 Vol. 2, No. 1, 37–52, http://dx.doi.org/10.1080/21697221.2014.891069 Xi Wu* and Junsheng Zhang School of Accountancy, Central University of Finance and Economics, Beijing 100081, China Prior event studies have examined investor wealth losses due to corporate miscon- duct in China’s capital markets, using the regulator’s formal sanction announcement as the relevant event. However, an important and earlier event (namely, the regula- tory investigation announcement) has been overlooked. We find that the market reaction is a decrease of 2% around the sanction event (which is consistent with prior literature), while it reaches a decrease of about 6% around the investigation event, suggesting that focusing on the sanction event alone will substantially under- estimate investors’ wealth losses. In exploring the cross-sectional variations of inves- tor reaction to the investigation announcement event, we find that a larger magnitude of prior-year earnings management is significantly associated with more negative abnormal returns, which suggests that investors may use prior financial reporting quality in making decisions when faced with insufficient information. The implication of our evidence for civil lawsuits against corporate misrepresentation in China is also discussed. Keywords: civil lawsuits; regulatory investigation; stock market reaction 1. Introduction As the enforcers of security laws and rules, regulatory institutions in the security mar- ket play an important role in protecting investors (La Porta, Lopez-de-Silanes, Shleifer, & Vishny, 2000, 2002). In order to capture the enforcement effect (e.g., Feroz, Park, & Pastena, 1991; Chen, Firth, Gao, & Rui, 2005) and the information provided by the enforcement process (Nourayi, 1994), academic research usually measures stock price reaction over the enforcement announcement window. Prior studies on China’s security market enforcement, however, focus exclusively on the sanction announcements and their consequences. For example, Wu and Gao (2002) document that the firms involved suffer an average loss of 1.18% during the [–10, +10] event window when punished either by the China Securities Regulatory Commission (CSRC) or by the two stock exchanges (Shanghai and Shenzhen). Chen et al. (2005) also utilise such sanction events and analyse the consequent shareholder losses. They find that shareholders lose 1% to 2% wealth during the [–2, +2] event window around the sanction announce- ments. Firth, Rui, and Wu (2009) differentiate sanctions by the CSRC from those imposed by two stock exchanges. They find that firms punished by the CSRC suffer greater losses than those punished by the stock exchanges, and the losses of the former reach 2% during the [–1,+1] event window. Collectively, shareholders on average suffer losses of 1% to 2% when firms are punished by the CSRC or the stock exchanges. Although these documented wealth losses are statistically significant, they are *Corresponding author. Email: wuxi@cufe.edu.cn Paper accepted by Donghua Chen. © 2014 Accounting Society of China 38 Wu and Zhang economically minor in magnitude, especially compared with the findings from the US market. It may lead to a perception that the economic consequences of illegal corporate conduct are not serious. However, the enforcement process of the CSRC consists of a series of consecutive actions. A finalised sanction is only the last step of the process. Before that, a signifi- cant action is the investigation announcement, informing investors that a firm is sus- pected of illegal conduct and the CSRC is beginning to investigate it. Existing studies emphasise sanction announcements but have overlooked investigation announcements. Chen et al. (2005) have realised the potential significance of the investigation event but they point out that they had to drop the investigation announcements because such announcement dates were not commonly available during their sample period (i.e. 1999–2003). Firth et al. (2009) differentiate the timeliness of sanction releases by the CSRC from the timeliness of releases by the stock exchanges. Again, they did not explore the investigation events. Focusing on sanction events alone, however, may well underestimate shareholder wealth losses due to listed companies’ misconduct, and may affect investors’ future claims. A more important consequence of underestimation is that it will not be enough to deter and suppress potential misconduct by other or future listed companies, thus harming investor protection in the long term. On the other hand, we note that the accuracy of information in sanction announce- ments is obviously different from that of investigation announcements. Sanction announcements usually disclose in great detail the information about listed companies’ misconduct and detailed sanction measures. In contrast, investigation announcements only offer ambiguous and vague information about corporate misconduct, and the infor- mation is usually qualitative and uncertain. The following is a typical investigation announcement: The company received the notification from Hunan Bureau of the Chinese Securities Regu- latory Commission on June 7th, 2005, that since the company is suspected of violating security laws and regulations such as reporting false information, the CSRC decides to investigate the company since June 7th, 2005. (Stock code: 000430, announced on June 8th, 2005) The key information arising from an investigation announcement is that the com- pany has probably committed misconduct against securities laws, which is a piece of qualitative information. Investors have no idea about the nature, the means, or the extent of misconduct yet. Compared with investigation announcements, sanction announcements provide three additional pieces of information: (1) confirming prior sus- picion of corporate misconduct; (2) informing market participants about the specific nature, means, and extent of corporate misconduct; and (3) announcing detailed puni- tive measures. Therefore, the average magnitude of market reaction (or investors’ loss) around sanction announcement events (which is –2% as documented by prior studies), can be viewed as an overall magnitude of the reaction to the above three pieces of additional information. Taking investigation announcements into account allows us to more accu- rately estimate investor reaction to such an abstract and qualitative event as potential corporate misconduct. We collect all available CSRC investigation announcements issued from 2002 to 2011 as the sample, and examine the market reaction or investor losses around such announce- ments. We find that the average cumulative abnormal return of investigated companies China Journal of Accounting Studies 39 during the [0, +1] investigation announcement window is about –6%, which is much larger in magnitude than that around the sanction announcement window. The evidence suggests that the market reaction to sanction announcements alone significantly underesti- mates investors’ wealth losses due to listed companies’ violations against regulations. The remainder of this paper is organised as follows. Section 2 reviews related litera- ture on the market reaction to securities regulatory enforcements. Section 3 introduces the research design. Section 4 presents the empirical results. Section 5 discusses the implications of our evidence for China’s civil lawsuit system against falsified financial reporting in securities markets. Section 6 concludes. 2. Literature Review 2.1. US-based studies Prior literature has examined the enforcement actions of the Securities and Exchange Commission (SEC) when analysing regulators’ enforcement events. The SEC issues Accounting and Auditing Enforcement Releases (AAERs) in which corporate account- ing and auditing malpractices are reported. Companies subject to AAERs are usually involved in misstating (often overstating) accounting earnings. Feroz et al. (1991) examine the market reaction to 224 AAERs released by the SEC from April 1982 to April 1989, among which there are 58 investigation announcements. They find that the cumulative abnormal return (CAR) around the investigation announcements reaches – 7.5% for the event window [–1, 0], while the CAR around the outcome announcements is insignificant. Nourayi (1994) uses 89 companies listed on the NYSE or AMEX that were prose- cuted by the SEC from 1977 to 1984 as the sample for analysis of investor reaction to the litigation announcements. He finds that when the financial media release the news before the SEC makes the announcement, companies suffer a greater negative abnormal return of –1.883%, while the average abnormal return is –0.091% for cases where the SEC announces the litigation earlier than the media. Dechow, Sloan, and Sweeney (1996) also analyse the market reaction to a sample of AAERs. They find that the CAR reaches –50% over a longer event window of [–120, +120] around the date when companies are suspected of manipulating earnings. More- over, the bid-ask spreads of involved companies also rise significantly. These results suggest that corporate accounting misconduct does substantial harm to shareholder wealth. In Dechow et al. (1996) there are three important time points: the first is the date when the media publicly alleges a subject company to overstate earnings; the second is that when the SEC initiates the investigation process; and the third is that when the SEC publicly releases an AAER. Dechow et al. (1996) use the first date as their event date. Miller (2006) also uses the media disclosure date as the original event when exam- ining AAERs. He finds that the mean and median of market-adjusted abnormal return is –6.3% and –2.9% respectively on the media disclosure day. Karpoff, Lee, and Martin (2008) describe the detailed process of SEC investigation. They decompose the process into: the event-triggering date; the informal-inquiry date; the formal-investigation date; and the sanction date. They find that although the penal- ties imposed on firms by the legal system average only $23.5 million per firm, the rep- utational loss in the capital market (measured by CAR) is over 7.5 times. The mean abnormal return is –25.24% around the event-triggering date, –14.41% around the investigation date, and –6.56% around the sanction date. Their evidence suggests that the earlier the event date, the more negative the stock market reaction. 40 Wu and Zhang 2.2. China-based studies and research opportunities The foregoing studies based on the US market reveal that researchers need to analyse a series of important announcement events in order to estimate accurately any investor losses due to corporate misconduct. Existing studies based on the Chinese market, how- ever, only examine market reaction to the much later sanction events. Wu and Gao (2002) examine market reactions to sanction announcements from the CSRC and Shanghai and Shenzhen stock exchanges from 1999 to 2000, and find that the average CAR is –1.18% over a 21-day event window beginning from ten days before and ending with ten days after the sanction announcement. Chen et al. (2005) identify 169 firms disciplined by the CSRC and the two stock exchanges over the period 1999–2003 (among which 49 are punished by the CSRC). They find that the average CAR ranges from –1% to –2% within the [–2, +2] event window. They conclude that the CSRC is not a ‘tiger without teeth’, but actually plays a monitoring role. Sun and Zhang (2006) also examine firms punished by the CSRC and two stock exchanges during the period 1990–2002 and find the average CAR around sanction announcements is –1.4%. These previous studies on China combine sanctions by CSRC and those by the stock exchanges. However, these two types of sanctions are quite different. The CSRC imposes an administrative sanction, while the sanctions of the two stock exchanges are non-administrative and self-disciplinary. Therefore the former is much more severe in nature. Moreover, the approach to disseminating the disciplinary information varies across these two types of sanctions (Firth et al., 2009). Firth et al. (2009) disentangle them and find that CAR in the 3-day event window [–1, +1] is –2% for announcements about the CSRC-initiated sanctions, and –1.9% to –1.6% for sanction announcements from two stock exchanges. Apparently, the stock market appears to react similarly to these two types of sanction announcements, which will lead to a perception that inves- tors do not suffer huge losses if corporate management violates security laws and rules and is then punished by the CSRC. The enforcement process of the CSRC normally involves official administrative pro- cedures and time-consuming investigations. Before the sanction document is finalised, there exists an important event, i.e. the initiation of an investigation. Although in the early stage of capital market there were few publicly disclosed investigation announce- ments, we observe that in more recent years a considerable number of listed companies released the investigation announcement soon after they were investigated by regulatory agencies. If one overlooks the market reaction around investigation announcements, the estimation of investor wealth losses due to corporate misconduct is very likely insuffi- cient. Therefore, by focusing on administrative sanctions issued by the CSRC rather than non-administrative condemnation by stock exchanges, we examine the market reaction to investigation announcements. 3. Research Design 3.1. Sample selection To identify all investigation announcements in China’s stock market, we first collect all related violation and sanction announcements from Wind, CSMAR and SINOFIN databases up to December 31, 2011. Among these announcements, we find that the number of investigation announcements began to increase since 2002. We only identify China Journal of Accounting Studies 41 two investigation announcements for 2001. In these two announcements there was additional information about other confirmed corporate illegal behaviour, indicating that these investigation events were confounded by other events. Therefore, we start our sample period from the year 2002. Up to the end of 2011, we obtain 232 investigation announcements. Table 1 shows our sample selection process. For firms that disclose the same inves- tigation event several times over a short period, we only retain the initial announce- ment. For a few firms that have experienced multiple investigations and sanctions during our sample period, we recognise if there is a later investigation announcement that occurs after a sanction announcement. If so, we keep the later investigation announcement as a different one from an earlier investigation event. Through the first selection procedure, we drop 15 repeated investigation announcements. Firms in 57 observations have no stock trading data at the announcement day. To ensure our analyses are not confounded by other events, we drop these observations. In order to use a market model to calculate abnormal returns, we require each observation to have data of at least 120 trading days until 31 days before the announce- ment day. Three observations among the remaining 160 observations fail to meet the criterion. Finally, we obtain 157 observations as our main sample. Panel A of Table 2 presents sample distribution across years. During the sample per- iod, the year with the fewest observations is 2011 (N=5) while the year with the most is 2005 (N=37). Panel B of Table 2 lists the sample distribution across industries. Among 13 industries, manufacturing has the most observations (53.5%) and each of three other industries (information technology industry, agriculture, forestry, animal husbandry, and fishing industry, and comprehensive industry) accounts for more than 5% of the total observations. 3.2. Measurement of market reactions We measure market reactions (or investor losses) using cumulative abnormal returns over several trading days around the investigation announcement day (t=0). Following prior studies (Chen et al., 2005), we use the risk-adjusted method to calculate daily abnormal returns (AR) and cumulative abnormal returns (CAR). First, we estimate daily abnormal returns using a 200-day estimation window [– 230, –31]. We require each observation to have data for at least 120 trading days. The market model is as follows: R ¼ a þ b R þ e (1) it i i mt it where R is the daily return of firm i on date t,and R is the equally-weighted market it mt return on date t (both considering cash dividend investment). The residual (ε ) of the it market model is the abnormal return of firm i on date t: Table 1. Sample selection. No. Initial observations (2002.1.1–2011.12.31) 232 Less: repeated investigation announcements 15 investigation announcements without trading data at the announcement date 57 investigation announcements without trading data of at least 120 trading days 3 Final sample observations 157 42 Wu and Zhang Table 2. Sample distribution across years and industries. Panel A: Distribution of investigation announcements across years Year No. % 2002 9 5.73 2003 12 7.64 2004 15 9.55 2005 37 23.57 2006 26 16.56 2007 13 8.28 2008 13 8.28 2009 12 7.64 2010 15 9.55 2011 5 3.18 Total 157 100.00 Panel B: Distribution of investigation announcements across industries Industry No. % A. Agriculture, Forestry, Husbandry & Fishery 12 7.64 B. Mining 1 0.64 C. Manufacturing 84 53.50 D. Electricity, Gas and Water production 4 2.55 E. Construction 5 3.18 F. Transportation and Warehousing 5 3.18 G. Information Technology 16 10.19 H. Wholesale and Retail Trade 7 4.46 I. Finance and Insurance 1 0.64 J. Real Estate 5 3.18 K. Services 4 2.55 L. Communication and Culture 1 0.64 M. Comprehensive 12 7.64 Total 157 100.00 e ¼ R ða^ þ b R Þ (2) it it i mt Then, we compute the average abnormal return on date t for the full sample: AR ¼ e (3) t it Finally, we compute the cumulative abnormal return over the [t1，t2] event window: t2 CAR ¼ AR (4) ½t1;t2 t t1 4. Empirical Results 4.1. Market reactions to investigation announcements Table 3 presents the market reaction around the investigation announcement. Panel A and Panel B show that the negative market reaction occurs mainly on the announce- ment day and the first day after the announcement. China Journal of Accounting Studies 43 The mean AR is –3.83% on the investigation announcement day (t = 0). A one sample t-test shows that the AR is negative and statistically significant (p < 0.001), and 86.6% observations in the full sample have a negative AR. The mean AR is –1.20% on the first day after investigation announcement. A one sample t-test shows a significantly negative result ( p < 0.001), and 68.8% observations have a negative AR. On the other trading days (t = –3, –2, –1, 2, and 3) around the investigation announcement, we fail to find any negative market reaction statistically significant at the 5% level. Since negative market reactions mainly occur on t = 0 and t = 1, we compute cumulative abnormal returns (CAR ) over the 2-day event window [0, +1] to [0，+1] measure the whole market reaction around the investigation announcement. Panel C shows that mean CAR is –5.48%, which is significant at the p < 0.001 level. [0，+1] Among the full sample, 87.9% of observations have a negative CAR. Table 3. Market reactions to investigation announcements. Panel A: Average abnormal return around investigation announcements No. Mean t-stat. % (AR<0) p-value AR 157 –0.0001 –0.051 57.3 0.079 –3 AR 157 0.0014 0.523 53.5 0.425 –2 AR 157 –0.0050 –1.887 55.4 0.201 –1 *** *** AR 157 –0.0383 –12.982 86.6 0.000 *** *** AR 157 –0.0120 –4.209 68.8 0.000 AR 156 –0.0005 –0.214 54.5 0.298 AR 156 0.0021 0.785 50.6 0.936 Panel B: The trend of average abnormal return around investigation announcements 0.01 t=-3 t=-2 t=-1 t=0 t=1 t=2 t=3 -0.01 -0.02 AR -0.03 -0.04 -0.05 Panel C: Cumulative abnormal return around investigation announcements No. Mean t-stat. % (CAR<0) p-value *** *** CAR 157 –0.0548 –12.570 87.9 0.000 [0，+1] *** * and denote significance at 1% and 10% level (two-tailed), respectively. AR denotes average daily abnormal return. CAR denotes cumulative average abnormal returns. t=0 denotes the investigation announcement day. 44 Wu and Zhang 4.2. Magnitude of market reactions to investigation announcements: A comparison with sanction announcements Prior studies find that the market reaction is –2% around the sanction announcement events of Chinese listed firms (e.g. Chen et al., 2005; Firth et al., 2009). In contrast, Panel C of Table 3 indicates that the market reaction reaches about –5.5% around investigation announcement events. To provide a stricter comparison, we identify 50 observations among our sample of investigation announcements, each of which also has a corresponding sanction announcement. Untabulated statistics show that, for these 50 observations, the mean (median) interval between the investigation announcement day and the sanction announcement day is 734 (715) days, i.e. around 2 years. Table 4 provides a comparison between the market reaction around investigation announce- ments and that around sanction announcements. In Table 4, the market reaction to investigation announcements of the above-men- tioned 50 observations is –6.42% ( p < 0.001), while the market reaction to sanction announcements is –2.02% ( p = 0.034), which is consistent with prior studies (Chen et al., 2005; Firth et al., 2009). Both T-test and Wilcoxon test show that the negative market reaction around investigation announcements is significantly stronger than that around sanction announcements ( p < 0.001). The results in Table 4 indicate that although the stock market reacts negatively to sanction announcements, as documented in prior literature and confirmed in our study, investigation announcements cause much more negative (and much earlier) market reac- tions. In other words, corporate illegal behaviour has resulted in great investor wealth losses at the investigation phase. At the sanction stage however, investor loss is lower. 4.3. Toward an understanding of cross-sectional variation of market reactions to investigation announcements We find that in investigation announcements the specific nature, items and amounts of falsified statements are not usually disclosed, because such announcements are by defini- tion of an investigative (rather than conclusive) nature. They are thus different from the sanction announcements on which prior literature has focused. However, to explain the variability of market reaction around investigation announcements across firms, we explore the potential influence of corporate financial information quality (proxied by the magnitude of earnings management and the type of auditor’s opinion) for the immedi- ately prior fiscal year. The rationale is that investors may use recent financial reporting quality in making investment decisions when faced with insufficient information about investigation announcements. The lower is the prior financial reporting quality, the more Table 4. Market reactions around investigation versus sanction announcements. CAR No. Mean t-stat. %(CAR<0) p-value [0，+1] *** *** Investigation event 50 –0.0642 –9.498 94.0 0.000 ** *** Sanction event 50 –0.0202 –2.181 72.0 0.003 Matched sample t-test Wilcoxon test CAR t-stat. p-value t-stat. p-value [0，+1] *** *** investigation vs. sanction –3.832 0.000 –4.453 0.000 *** ** * , and denote significance at 1%, 5% and 10% level (two-tailed), respectively. CAR denotes cumulative average abnormal returns. t=0 denotes the investigation announcement day or sanction announcement day. China Journal of Accounting Studies 45 negatively investors may react. That is, the market reaction around investigation announcements is negatively related to the corporate recent financial reporting quality. To test this conjecture, we design the following OLS regression model: CAR ¼ a þ a ABSDA þ a MOD þ a LTA þ a LEV þ a ROA þ a BIGAUD ½0;þ1 0 1 2 3 4 5 6 þ a SOE þ a DETAIL þ a POST07 þ a MANUF þ e 7 8 9 10 (5) In equation (5), the dependent variable is the market reaction around the investiga- tion announcement (CAR). We employ the absolute value of discretionary accruals (ABSDA) and auditor’s opinion type (MOD), in both cases for the latest fiscal year before the investigation announcement, to assess financial reporting quality. Discretion- ary accruals are estimated based on the modified Jones model by each year and indus- try, while we require that each year-industry combination has at least eight observations. MOD is a dummy variable coded 1 for a modified auditor’s opinion. A higher value of ABSDA or MOD taking the value of 1 suggests that the firm has lower financial reporting quality prior to an investigation announcement and is expected to be associated with lower CAR around the announcement event. We include a number of control variables: size (LTA, calculated as the natural loga- rithm of total assets at the end of the latest fiscal year prior to the investigation announce- ment); leverage ratio (LEV, calculated as total liabilities over total assets); financial performance (ROA, calculated as net income over total assets); audit firm size (BIGAUD, equal to 1 when the audit firm is one of the Big 4 or top 10 local audit firms based on cli- ent firm asset size, and 0 otherwise); and the nature of controlling shareholder (SOE, a dummy variable coded 1 if the controlling shareholder is government or state-owned enterprises). We include a dummy variable DETAIL to capture the difference among investigation announcements regarding the specificity of disclosure. By reading each of the investigation announcements, we identify 23 announcements that mention some details of violations (such as failure to file annual reports timely and insider trading), while in other announcements we do not find any detailed disclosure about violations except a generic mention of possible law-breaking or falsified representations. Therefore, we include DETAIL in the model, which equals 1 for the 23 detailed announcements and 0 for others. Finally, we include year and industry dummies. Considering our sample size is quite small, we use the dummy POST07 (coded 1 for post-2007 observations, and 0 otherwise) and the dummy MANUF (coded 1 for observations in manufacturing industry, and 0 otherwise). They are expected to capture the effect of a change in financial report- ing regime due to the effectiveness of Chinese new accounting standards since 2007 and the potential difference between manufacturing and other industries, respectively. As there are missing values for some explanatory variables, the sample size is reduced to 152 for the regression of model (1). Untabulated descriptive statistics show that for the latest year prior to investigation announcements, 79 observations (52%) receive a modified auditor’s opinion, 56 observations (36.8%) are audited by large audit firms, and 68 observations (44.7%) are controlled by the State. Table 5 presents the regression result of equation (5). The results show that the coefficient on ABSDA is significantly negative (t = – 2.65, p < 0.01), which suggests that the market reaction around investigation announce- ments is more negative for companies that have a greater magnitude of earnings man- agement prior to investigation. The coefficient on MOD, the other proxy of financial reporting quality, is also negative (p = 0.121). As for control variables, the coefficient 46 Wu and Zhang on ROA is significantly negative, indicating that the market reaction to investigation announcements is more negative for better-performing firms, which indicates that inves- tors may be more suspicious about companies with a higher prior-year book profit. The coefficient on DETAIL is not significantly different from zero, suggesting that the mar- ket reaction we observed around the investigation announcement is not driven by the more detailed announcements. To sum up, the results of Table 5 are consistent with our conjecture that investors use prior financial reporting quality in making decisions when faced with insufficient information from regulators. 4.4. Robustness checks In our main test, we compute market return on an equal-weight basis. As a robustness check, we compute the market return on an outstanding-equity-weight or total-equity- weight basis. Untabulated results show that our main findings remain qualitatively unchanged. Second, we use market-adjusted abnormal return instead of risk-adjusted when cal- culating CAR. According to the market-adjusted method, we regard market return as the firm’s normal return and use the difference between daily firm return and market return to proxy for abnormal return. This alternative approach is relatively simple and widely used in prior studies (e.g. Chen et al., 2005; Firth et al., 2009; Karpoff et al., 2008). Karpoff et al. (2008) only use this method to calculate abnormal return. Untabulated results show that our main findings are qualitatively similar when using the alternative approach to calculating the abnormal return. Table 5. OLS regression result of the market reaction to investigation announcements. Dep. var.: CAR Coefficient t-stat. [0，+1] *** ABSDA –0.107 –2.65 MOD –0.014 –1.56 LTA –0.008 –1.54 LEV 0.010 0.63 *** ROA –0.111 –3.03 BIGAUD –0.000 –0.00 SOE –0.000 –0.03 DETAIL 0.007 0.56 POST07 0.011 1.03 MANUF –0.001 –0.11 Constant 0.037 0.59 N 152 Adj. R 0.023 *** denotes significance at 1% level (two-tailed). Variable definition: CAR is cumulative abnormal returns at the [0, +1] window near investigation announce- ment day. ABSDA is the absolute value of discretionary accruals, estimated from the modified Jones model by year and industry. MOD equals 1 for modified auditor’s opinion and 0 otherwise. LTA is the natural loga- rithm of total assets at the end of latest fiscal year prior to an investigation announcement. LEV is total liabili- ties over total assets. ROA is net income over total assets. BIGAUD equals 1 for firms audited by Big 4 auditors or top 10 top local auditors and 0 otherwise. SOE equals 1 for firms controlled by governments or state-owned enterprises, and 0 otherwise. DETAIL equals 1 for investigation announcements that mention some details of violations (such as failure to file annual reports timely and insider trading), and 0 otherwise. POSTO7 equals 1 for post-2007 observations and 0 otherwise. MANUF equals 1 for observations in the man- ufacturing industry and 0 otherwise. China Journal of Accounting Studies 47 5. Further discussion In this section we discuss the implications of our evidence for civil lawsuits against cor- porate falsified statements in China. On January 9, 2003, the Supreme Court promul- gated the ‘Several Rules on Judging Civil Compensation Cases against Falsified Representation in Securities Market’ (hereinafter referred to as the ‘Rules’). Part IV of the Rules, titled ‘Defining Falsified Representation’, specifies how to identify whether there is a causal relationship between the misrepresentation and the injurious results. Specifically, Article 18 of the Rules states: Under the following circumstances, the People’s Court shall determine that a causal rela- tionship exists between the falsified representation and the injurious results: (1) the securi- ties in which the investor invested are directly associated with the falsified representation; (2) the investor buys the securities on or after the date on which the falsified representation is made while before it is uncovered or corrected; and (3) the investor incurs losses as a result of selling securities or as a result of continuing to hold the securities, on or after the date on which the falsified representation is uncovered or corrected. Accordingly, it is crucial to define the date on which the falsified representation is uncovered. Article 20 of the Rules specifies that such a date refers to the first date for the falsified representation to be publicly disclosed in nationwide media such as newspa- pers, periodicals, radio or television. In China’s actual judicial practice, however, differ- ent local courts have different definitions for the exposure date since the Supreme Court does not clarify the details of the definition. This causes a huge difference in practice, even leading to ‘contrary recognitions for similar situations’ (Mo, 2011, pp. 1–2). In particular, when a regulatory agency initiates an investigation and the subject company discloses the investigation event, it can be controversial whether the announcement date should be deemed to be the exposure date. Panel A of Figure 1 depicts the timeline of misrepresentation-related events. Suppose that investors A and B buy a company’s shares after the misrepresentation date but before the investigation announcement date. Investor A sells the shares after the investigation announcement date and suffers wealth losses. Investor B sells the shares after the formal sanction announcement date and suffers wealth losses as well. Panel B describes the traditional situation where the formal sanction announcement date is treated as the misrepresentation exposure date. Because the investigation announcement does not indicate formal confirmation and sanctions from regulatory agencies on a listed company for misconduct, it appears to be no problem in the legal sense that we do not take the investigation announcement date as the exposure date even if investors know that the misrepresentation may be likely to exist. However, if the court defines the exposure date as in Panel B, the causal relationship between wealth losses of investor A and corporate misrepresentation will be challenged. This is because investor A sells shares prior to ‘the exposure date’, thus it may not be appli- cable to the third situation in Article 18 of the Rules to define a causal relationship. Panel C illustrates the situation where the investigation announcement date is trea- ted as the misrepresentation exposure date. Based on the evidence in this paper, inves- tors re-adjust their expectation for companies upon investigation announcement and suffer huge wealth losses, and the losses are more significant than those around the sanction announcement. The evidence suggests that: (1) investors realise that corporate misrepresentation very likely exists; (2) investors suffer considerable wealth losses around the investigation announcement; and (3) we can reasonably infer that investor 48 Wu and Zhang Figure 1. Alternatives to defining the date of exposure to misrepresentation. wealth losses are due to corporate misrepresentation implied in the investigation announcement, and consequently there is a causal relationship between potential mis- representation and investor wealth losses around the investigation announcement. Therefore, the situation in Panel C is supported by our empirical evidence. Moreover, the situation also exists in judicial practice (e.g. Mo, 2011). In this situation, investor A’s wealth losses satisfy the third condition of Article 18 of the Rules and are judged as having a causal relationship with misrepresentation. Investor losses due to corporate misrepresentation, however, should only cover the period from the perpetration of mis- representation to misrepresentation exposure (Guo, 2003). Therefore, for those investors who make a decision based on the formal sanction announcements and sell shares after a formal sanction (as Investor B does in Panel C), losses may not be acknowledged to have a causal relationship with the misrepresentation. Moreover, if the investigation China Journal of Accounting Studies 49 announcement date is treated as the exposure date, the (usually long) time lag between the investigation announcement and the formal sanction announcement may well exceed the ‘reasonable period’ of stock trading after the exposure date, and becomes unfavourable toward defining the causal relationship. The foregoing discussion suggests that the investigation announcement may create a dilemma in civil lawsuits against corporate misrepresentation in China’s stock mar- kets. The legislative institutions need to consider whether the investigation announce- ment should be treated as the date of exposure to misrepresentation, and make clearer specifications about the situations where investors should be aware of the existence of corporate misrepresentation, which could be useful to judge the causal relationship between corporate misrepresentation and investor wealth losses in a fair and uncontro- versial manner. Considering the significantly negative market reaction to investigation announce- ments we document in this study, it is suggested that regulatory agencies should be cautious when making the decision to initiate a corporate investigative process. They should maintain similar standards for the processes of initiation and inquiry across sub- ject companies, and should stipulate clearer rules on the disclosure of investigation events. 6. Conclusion This study contributes to the literature and the practice relating to Chinese stock market regulation in three ways. First, we extend prior studies by further evaluating investor wealth losses due to corporate misconduct. Prior event studies show that companies suffer a fall of 2% of market value around formal sanction announcements. We find that the investigation announcement, a much earlier event prior to the sanction announcement, leads to sig- nificantly more investor losses (about –6% of CAR). Our results indicate that prior studies focusing on the sanction announcement underestimate investor losses due to corporate misconduct. After considering the reaction around the investigation announce- ment, we conclude that administrative enforcement actions by the CSRC are associated with more investor losses, compared with the losses associated with non-administrative condemnation by the two stock exchanges. Second, our research setting allows us to make a relatively clean estimate of how investors react to the qualitative nature of corporate misconduct (without blending information about the specific acts, approaches, and magnitude). Since investigation announcements only contain general and qualitative information about corporate mis- conduct, our results of –6% CAR over the two-day event window [0,+1] are relatively pure market reactions to the corporate misconduct itself. In contrast, investor losses resulting from incremental information in the sanction announcement, such as miscon- duct confirmation, specific violations, means, amounts, punitive measures, and so on, is only –2%, which is much less than that caused by qualitative information in the inves- tigation announcement. In exploring the cross-sectional variations of investor reaction around the investigation announcement event, we find that the market reaction is nega- tively correlated with corporate earnings management prior to the investigation announcement. This means that investors may use prior financial reporting quality in making decisions when faced with insufficient information. Third, this study has important implications for defining the causal relationship between listed companies’ misrepresentation and investor losses. On one hand, from 50 Wu and Zhang the legal perspective, investigation announcement events do not have an official and legitimate sense in identifying misrepresentation. On the other hand, our evidence shows that investors suffer heavy losses around investigation announcement events. Therefore, it is necessary for legislators to clarify the role that an investigation announcement event plays in the judicial process, and for regulators to improve and unify the investigation standards, procedures and disclosure requirements across differ- ent companies. Future studies could proceed by exploring factors that help further explain cross- sectional variations in market reaction around investigation announcements, which lack detailed information about the violation. In addition, when faced with uncertain and qualitative information, to what extent is the investor reaction rational (or irrational)? This may improve our understanding of the capital asset pricing model with blurry information. Moreover, since investors react to investigation announcements by adjust- ing the expectation of corporate value to some extent, related research questions may emerge regarding how other information users (such as creditors, auditors, and those charged with governance) react to such events. Will other information users’ reaction to an investigation announcement be different from their reaction to a sanction announce- ment documented in prior studies (e.g., Chen et al., 2005; Chen, Zhu, & Wang, 2011; Zhu & Wu, 2009)? It is a challenging task to give a comprehensive evaluation of investor wealth losses due to listed companies’ misconduct (Chen, Zhang, & Li., 2008). Admittedly, we are not able to analyse investor reaction to sporadic information releases prior to investiga- tion announcements, which may include media and regulatory reports. It is not unusual to see a regulatory investigation triggered by an individual or a newspaper questioning a company. In addition, during the long process from the initiation of investigation to formal sanction, there may be some unobserved events leading to a negative market reaction. Our estimation of the market reaction to investigation announcements does not reflect investor losses from those unobserved events. In this sense, our estimation for investor losses is still conservative. Acknowledgements We are grateful for helpful comments from two anonymous reviewers, Donghua Chen (Associate editor), Jason Xiao (Joint-editor), Oliver Rui and Meng Chen. Wenjing Chen, Hao Zhang, and Meng Chen provided able assistance in data collection. We acknowledge financial support from the Ministry of Education ‘New Century Talents Program’ (NCET-11-0754), the Beijing Munici- pal Commission of Education ‘Joint Construction Project’, the Beijing Municipal Commission of Education ‘Pilot Reform of Accounting Discipline Clustering’, CUFE ‘211 Construction’ Project, and CUFE Innovative Research Team Project ‘Empirical Accounting and Auditing’. Notes 1. See the next section of the literature review for a more detailed discussion. 2. Wealth losses referred to in this study is a concept from the field of finance, i.e. investors’ reaction to stock price measured by a traditional event study. The concept of wealth losses does not refer to the legal definition in Provision 18 in ‘Several Rules on Judging Civil Compensation Cases against False Representation in Security Market’ (hereinafter referred to as the ‘Rules’) promulgated by the Supreme People’s Court on 9 January 2003. We thank an anonymous reviewer for pointing this out. 3. See Chen, Firth, Gao, and Rui (2005, Appendix A) for examples. China Journal of Accounting Studies 51 4. In addition, their univariate analyses show that being punished by the CSRC and the two exchanges leads to higher probabilities of auditor turnover, chief executive officer turnover, auditor issuance of modified opinion, and increased bid-ask spread (indicating an increase in costs of capital). 5. Sanctions initiated by stock exchanges do not involve administrative investigation. 6. For example, in the civil lawsuit case against securities misrepresentation discussed in Mo (2011), the plaintiff bought the stock during the period between the investigation announce- ment date and the formal sanction date. The plaintiff claimed that the sanction announcement date should be the misrepresentation exposure date, while the local court treated the investi- gation announcement date, which only disclosed uncertain information, as the misrepresenta- tion exposure date. Thus, the controversial definition of exposure date became the key issue in this case. Mo (2011, p. 18) comments that ‘the local court’sdefinition of exposure date violates ... the principle of disclosing definitive information.’ 7. We do not exclude the possibility that the significantly negative market reaction around the investigation announcement is partly due to irrational behaviour (such as panic selling) since the information is quite vague. But the Rules, when defining the causal relationship, concern whether losses incur after the misrepresentation exposure, rather than whether losses are rational or panic. 8. In addition, Panel C does not illustrate the debatable case discussed in Mo (2011), where investors buy shares after the investigation announcement but before the formal sanction announcement. In that case, investors claim their losses on the grounds that the investigation announcement has not officially confirmed the existence of corporate misrepresentation. 9. This issue may well exist for sanction announcement, but to a lesser extent, because its infor- mation is much clearer than that of an investigation announcement. References Chen, D., Zhang, T., & Li, X. (2008). Legal environment, government regulation, and implicit con- tract: Evidence from the scandals of Chinese listed companies. Economic Research Journal, (3), 60–72. (in Chinese). Chen, G., Firth, M., Gao, N., & Rui, O. (2005). Is China’s securities regulatory agency a tooth- less tiger? Evidence from enforcement actions. Journal of Accounting and Public Policy, 24 (6), 451–488. Chen, Y., Zhu, S., & Wang, Y. (2011). Corporate fraud and bank loans: Evidence from China. China Journal of Accounting Research, 4(3), 155–165. Dechow, P., Sloan, R., & Sweeney, A. (1996). Causes and consequences of earnings manipula- tion: An analysis of firms subject to enforcement actions by the SEC. Contemporary Accounting Research, 13(1), 1–36. Feroz, E., Park, K., & Pastena, V. (1991). The financial and market effects of the SEC’s account- ing and auditing enforcement releases. Journal of Accounting Research, 29(Supplement), 107–142. Firth, M., Rui, O., & Wu, X. (2009). The timeliness and consequences of disseminating public information by regulators. Journal of Accounting and Public Policy, 28(2), 118–132. Guo, F. (2003). The cognizance and compensation of misrepresentation. China Legal Science, (2), 95–99. (in Chinese). Karpoff, J., Lee, D., & Martin, G. (2008). The cost to firms of cooking the books. Journal of Financial and Quantitative Analysis, 43(3), 581–611. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. (2000). Investor protection and cor- porate governance. Journal of Financial Economics, 58(1–2), 3–27. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. (2002). Investor protection and cor- porate valuation. Journal of Finance, 57(3), 1147–1170. Miller, G. (2006). The press as a watchdog for accounting fraud. Journal of Accounting Research, 44(5), 1001–1033. Mo, T. (2011). A study on the misrepresentation exposure date. MS Thesis, Southwest University of Political Science and Law (in Chinese). Nourayi, M. (1994). Stock price responses to the SEC’s enforcement actions. Journal of Accounting and Public Policy, 13(4), 333–347. 52 Wu and Zhang Sun, P., & Zhang, Y. (2006). Is there penalty for crime: corporate scandal and management turn- over in China. Working paper, Peking University. Wu, L., & Gao, Q. (2002). Market reactions to regulatory sanction announcements. Economics Science, (3), 62–73. (in Chinese). Zhu, C., & Wu, L. (2009). The impact of illegal behavior of public firms on auditors: Evidence from sanction announcements issued by securities regulator. Auditing Research, (4), 42–51. (in Chinese).
Journal
China Journal of Accounting Studies
– Taylor & Francis
Published: Jan 2, 2014
Keywords: civil lawsuits; regulatory investigation; stock market reaction
