Impact of home country on financial reporting behavior: An analysis of restatements by foreign firms listed in the US. Harvard Business School

Transcription

1 Preliminary: Please do not quote or distribute without permission. Comments welcome Impact of home country on financial reporting behavior: An analysis of restatements by foreign firms listed in the US Harvard Business School Suraj Srinivasan 1 Aida Sijamic Wahid Gwen Yu First draft: September 12, 2011 Abstract We examine how restatement frequency of foreign firms listed in the US differs by home country characteristics. We find that foreign firms from countries with weak rule of law are less likely to restate their financials than those from strong rule of law countries, despite having higher levels of earnings management. For firms from weak rule of countries, we find no relation between restatements and level of earnings managements. This is in contrast to the restatements of firms from strong rule of law countries which show a positive relation with restatement frequency and earnings management. Also, firms from weak rule of law countries are more likely to opt for less visible disclosure methods when restating their financials, after controlling for materiality of the restatement. We interpret this finding as home country enforcement affecting the likelihood of firms to report existing accounting irregularities. These findings suggest that for cross-listed firms, less frequent restatements can be a signal of opportunistic reporting rather than high quality earnings. We gratefully acknowledge financial support from the Division of Research of the Harvard Business School. All errors are our own.

2 1. Introduction This paper examines how restatement frequency of foreign firms listed in the US differs by home country characteristics. Foreign firms listed in the US are obliged to follow the disclosure requirements set forth by the Securities and Exchange Commission and relevant laws such as the Securities Exchange Act of 1934 and the Sarbanes-Oxley Act of 2002 (Coffee [1999]; Stulz [1999]). Yet, prior studies find that reporting behavior of cross-listed firms differ systematically from that of US firms and by home country characteristics (Lang et al. [2006]; Gong et al. [2010]; Hope et al. [2011]). With concerns around the effectiveness of how foreign firms bond to the US regulatory regime (Siegel [2005]), we use the large number of restatements in recent years by both US and foreign registrants, to understand the role of the home country on reporting behavior of foreign companies listed in the U.S. We focus on restatements because they play a crucial role in assessing the reliability of financial statements. SFAS 154 states that, upon findings any material errors in the reported financials, all firms are required to issue a restatement correcting the error in a timely manner. 2 In principle, the mandatory nature of restatement disclosure indicates that the frequency of restatements should be increasing in the likelihood of errors and irregularities in the financial statements. Prior studies thus find that restatements often signal poor earnings quality (Ecker et al. [2011]) and extreme earnings management (Dechow et al. [2011]). Unlike other mandatory filings which are required on a periodic basis, however, the decision to report a restatement largely depends on the firm s compliance with the mandated rule (Heitzman et al. [2010]). Thus, the procedure leading up to an actual restatement can vary 2 APB 20 (and subsequently SFAS 154) defines errors as those arising from mathematical mistakes, accounting principle application mistakes, oversight, misuse of facts, and switching from a principle not generally accepted to the one generally accepted. Some firms also consider changes in accounting standards as restatements. Our empirical analysis however focuses only on error and accounting irregularities. 1

3 by the firms own enforcement of the rules. For example, to identify a restatement, a firm must first have an effective mechanism that detects errors in their reporting process (e.g., strong internal controls and high audit effort). Moreover, managerial incentives may lead even firms that detect errors to avoid a public restatement because restating financials can jeopardize the credibility of the reported financials and prompt investor scrutiny (Desai et al. [2006]; Srinivasan [2005]). Instead firms can smooth the error over time. The voluntary nature of the mandated reporting of restatements provides a powerful setting to examine how home country characteristics affects the reporting of foreign firms. Foreign firms listed in the US have incentives to avoid restatements because acknowledging errors in the reported financials can be particularly costly. Increased transparency in financial information is one important way by which foreign firms bond to the US market (Coffee [2002]; Hail and Leuz [2009]). Admitting errors in the reported financials can impair the firm s credibility and raise questions about the extent to which the firms financial information can be relied upon. Thus, for such foreign firms, it is possible that failure to restate financials represent opportunistic reporting rather than the absence of accounting irregularities. We conduct three sets of empirical tests to examine how restatements of foreign firms listed in the US vary with home country characteristics. First, we examine whether home country characteristics affect the likelihood of restatements. The country characteristic we examine is the level of legal enforcement measured using the rule of law index from the Worldwide Governance Indicators created by the World Bank (Kaufman et al. [2003] and used in La Porta et al. [2006]). 3 We use this as a summary indicator for the extent of enforcement in 3 Rule of law measures the extent to which agents have confidence in and abide by the rules of society as measured in year These include perceptions of the incidence of both violent and non-violent crime, the effectiveness 2

4 a country that can drive firm behavior by impacting factors such as auditor effort, investor protection, managerial self-dealing, etc. Since foreign firms listed in the US are all held to the same reporting standards, auditor monitoring requirements, and regulatory and private enforcement, we expect home country enforcement to have little effect on the restatement probability of foreign firms. However, if home country enforcement continues to affect the likelihood of accounting irregularities and/or the subsequent reporting of such irregularities, it will have a systematic effect on restatement frequency by foreign firms. In our second set of empirical tests, we distinguish between two different interpretations of frequent restatements, (1) as a signal of poor earnings quality, and (2) as a signal of prudent restatement reporting, by associating frequency of restatements with the properties of reported earnings. If firms correctly report their accounting irregularities, the frequency of restatements will be positively associated with proxies for firm s earnings management. In contrast, if restatements are likely to be concealed, we expect to find a less significant relationship between restatement frequency and earnings management. We therefore examine whether the relationship between level of earnings management and tendency to restate financials differs systematically with the strength of home country enforcement. Our prediction is that the association between restatement probability and level of earnings management will increase with home country rule of law. Finally, we examine whether home country enforcement affects the disclosure method of restatements, conditional on reporting a restatement. We hypothesize that firms from weak enforcement countries will opt for less visible disclosure choices, after controlling for the and predictability of the judiciary, and the enforceability of contracts. The source for the data is Kaufmann, Kraay and Mastruzzi (2003) 3

5 materiality of the restatement. 4 In particular, we test whether firms from weak rule of law countries are more likely to choose less visible disclosure methods those filed using regularly scheduled filings (10-K or equivalent) without any prominent indication of a restatement (such as, late filing notice, press release, 6-K or 8-K) after controlling for the restatement magnitude. Our empirical tests yield three major findings. First, we document that the frequency of restatements varies with the level of home country rule of law. We find that foreign firms from weak rule of law countries are less likely to restate (7.6% of firm years are restated) compared to a matched sample of US firms (14.6% of firm years) the difference being statistically significant (p-value <0.001), despite having higher levels of earnings management on average. This is in contrast to foreign firms from strong rule of law countries which show a smaller difference in their frequency to restate compared to the matched US firms (11.4% vs. 15.8% of firm years). Second, we find that the association between restatement frequency and earnings management increases with the level of home country rule of law effectiveness. Firms from weak rule of law countries show no evidence of more frequent restatements when level of earnings management is high. This is in contrast to firms from strong rule of law countries (and the matched sample of US firms) that show the expected strong positive relationship between restatements and earnings management. We interpret this as home country enforcement increasing a firms tendency to report existing accounting irregularities. 5 4 Following Myers et al., (2010), less visible methods are those announced in the regularly scheduled financial statements or in the amended statements. Such restatements are also known as stealth restatements in the popular press and know to provide information with minimum publicity by quietly including fixes in their regular filings. Other visible announcement methods include filing a separate form (e.g., form 8-K or 6-K) with the SEC or a press release. 5 In untabulated analysis, we examine whether firm-level governance can mitigate the effect of home country enforcement. In particular, we examine whether the effect of home country enforcement on restatement frequency is 4

6 Finally, we find that, conditional on restatements, firms from weak rule of law countries are more likely to disclose their restatements with less publicity after controlling for other likely determinants of disclosure medium. For example, the percentage of restatements announced using a more visible disclosure method is lower for firms from weak rule of law countries (28.2%) than the matched US firms (63.8%) or firms from strong rule of law countries (44.0%). The finding holds after controlling for the magnitude of the restatements and other firm and country characteristics. Our findings suggest that home country enforcement continues to affect reporting behavior of foreign firms listed in the US despite all firms being subject to the same US regulatory requirements and enforcement mechanisms. We find that foreign firms are less likely to restate their financials relative to a matched sample of US firms and the reluctance to restate is greater when the firm is from a weak rule of law country. In economic terms, we find that firms from weak rule of law countries are 1.73 times less likely to restate their financials than those from strong rule of law countries, after controlling for other determinants of restatement probability. Additional analysis suggests that the lower frequency of restatement for firms from weak rule of law countries is due to reluctance to admit exiting errors/irregularities than to lack of such accounting irregularities. Thus, for foreign firms listed in the US, and especially for those from weak rule of law countries, the absence of restatements likely represents opportunistic reporting rather than high quality earnings. reduced when firms have better governance (e.g., audit effort, board size, less insider ownership, and greater institutional ownership). We fail to find evidence that firm-level governance tightens the link between earnings management and restatement probability. However, such firm-level findings should be interpreted with caution because cross-listed firms are those that select into the US market, thus, tend to have less variation in such monitoring or firm-level governance. 5

7 Our paper contributes to three streams of literature. The first is the literature that examines the causes and consequences of restatements. These studies generally conclude that restatements represent poor earnings quality and firms suffer capital market consequences as a result (Palmrose et al. [2004]; Plumlee and Yohn [2010]). These studies, however, largely focus on US firms. To our knowledge, our study is the first to examine restatements by foreign firms listed in the US. Contrary to domestic firms, our findings suggest that for foreign firms, restatement reporting can be opportunistic and thus only weakly related to firm s earnings quality. More broadly, our findings suggest that when firms face greater pressure for transparent reporting (e.g., foreign firms listed in the US), using restatements as an indicator of poor earnings quality may be less appropriate. Second, our findings have implications for understanding how foreign firms listed in the US bond to the US financial reporting regime. The stringent disclosure rules required by US securities laws and resulting transparency are considered important reasons why cross-listed firms gain access to the US equity market. Prior studies, however, shows that reporting behavior of foreign firms listed in the US continues to differ from that of the US firms. Studies often find that home country characteristics continue to drive differences in the level of enforcement (Siegel [2005]) and reporting incentives (Lang et al [2006]). Our findings further this evidence by suggesting that home country characteristics continues to largely influence the likelihood of restatements of foreign firms listed in the US. Third, our findings have regulatory implications. Disclosure of accounting problems can be an early warnings signal of irregularities inside a firm, providing a basis for SEC investigation and investor scrutiny (Feroz et al, [1991]; Hennes et al. [2008]; Files [2009]). 6 Consequently, 6 The SEC describes restatements as "the most visible indicator of improper accounting and source of new investigations" (Schroeder [2001]). 6

8 firms which do not restate existing irregularities in their financials manage to minimize risk of SEC investigation and private securities litigation. In the absence of alternative mechanism that can trigger effective investigations, our study implies that foreign firms listed on US exchanges are disproportionately under-scrutinized by the US private and public enforcement mechanisms. The remainder of the paper is organized as follows. Section 2 reviews the literature and develops our hypotheses. Section 3 describes the data and empirical test; section 4 presents our results. In section 5, we conclude. 2. Hypothesis development and institutional details In this section, we describe our main hypothesis that home country enforcement will affect the restatement likelihood of foreign firms listed in the US. We first review the literature on reporting behavior of cross-listed firms. Next, we discuss the manner in which home country enforcement will affect restatements of foreign firms listed in the US. 2.1 Home country enforcement and reporting by foreign firms listed in the US Under US securities laws, a foreign private issuer listed on a major US stock exchange (e.g., NYSE, NASDAQ) is held to similar reporting requirement as the domestic registrants (Coffee [2002]). 7 Foreign registrants are required to make ongoing filings with the SEC under the Securities Exchange Act of 1934 and become subject to SEC oversight and enforcement actions. Prior research considers this commitment to ongoing disclosure and the enforcement of securities laws as allowing foreign firms listed in the US to reap the benefits of listing on US capital market (Reese and Weisbach [2002]; Karolyi [2006]). 7 A foreign private issuer is any issuer (other than a foreign government) incorporated or organized under the laws of a jurisdiction outside of the US. 7

9 Prior studies, however, find that the quality of information disclosed by foreign firms listed in the US is not always on par with that of the US firms. For example, Lang et al. [2006] show that properties of reported earnings of foreign issuers show more evidence of earnings management than the earnings of US domiciled firms. More importantly, that paper finds that the difference in the quality of reported earnings is systematically related to home country characteristics such as investor protection and legal enforcement. Similarly, studies that examine other disclosure behavior of foreign firms listed in the US find that the quality of their information continues to be influenced by home country factors. For example, foreign firm from weak investor protection countries are less likely to report incidents of internal control weaknesses (Gong et al. [2010]) and less likely to provide voluntary disclosures such as management forecasts (Hope et al. [2011]). These findings suggest that US regulation and monitoring by the SEC does not supersede the effect of the home country s regulatory environment. Accounting restatements (i.e., restating previous errors and irregularities in reported financials) are a mandatory reporting required under US GAAP when firms are faced with a situation requiring correction of prior financial statements. All firms registered with the SEC, foreign or domestic, are required to correct inaccurate information disclosed in existing financial statements in a timely manner. 8 While correcting a past misstatement is a material and mandated filing, the decision to report a restatement depends on the firm s compliance with the mandated rule (Heitzman et al. [2010]). Thus, this implicit variation provides an interesting setting to examine whether home country characteristics affects the reporting behavior of foreign firms listed in the US. 8 See Accounting Principles Board Opinion 20, Statement of Financial Accounting Standards (SFAS) No. 16 and SFAS No. 154 among others. 8

10 All firms have an incentive to avoid reporting restatements, because truthful reporting of a past misstatement, whether intentional or otherwise, undermines the credibility of the reported financials and has firm level and managerial consequences (Collins et al. [2009]; Desai et al. [2006]). Foreign firms are especially sensitive to this incentive as one reason to list in the US is posited to be to bond to a higher quality financial reporting regime (Hail et al. [2009]). At the same time, prior studies find that foreign cross-listed firms show greater tendency of earnings management which can in turn increase the likelihood of restatements (DeFond and Jiambalvo [1991]; Lang et al. [2006]). Thus, before providing our primary tests regarding cross-sectional variations in restatement frequency of foreign firms, we first examine whether foreign firm s restatement frequency differs from that of the US firms and whether the difference exhibits systematic patterns by home country enforcement. H1: The probability of restatements of foreign firms cross-listed in the US will vary by level of home-country enforcement. There are several reasons to expect restatement frequency to vary systematically by home country characteristics. Restatements often represent extreme examples of poor quality earnings (Palmrose et al. [2004]; Dechow et al. [2011]). Leuz et al. (2003) show that foreign firms exhibit more evidence of earnings management especially in countries with weak enforcement. Also, Lang et al. (2006) find that even after cross-listing, the accounting quality of foreign firms does not measure up to that of U.S. firms. Therefore, if foreign firms listed in the US continue to have low quality accounting and more so when firms are from weak rule of law countries, the likelihood of restatements for firms from this group of countries will be higher. Prior literature on cross-listings suggests that that despite foreign registrants being subject to similar standards as US firms, the extent of bonding to US regulatory and governance regime continues to differ systematically across countries (Fuerst [1998]; Frost and Pownall [1994]). 9

11 This is partly due to the cost of enforcement by the SEC or private litigation against foreign companies relative to US companies (Siegel [2005]). The difference can also arise due to differences in the level of domestic infrastructure for expert intermediaries like auditors, analysts, lawyers, institutional investors as well the extent of domestic enforcement by capital market regulators. In fact, the cost of enforcement by the SEC and private litigation also depends on the availability of a local infrastructure (e.g. lawyers and auditors) to support enquiries and action. We use the measure rule of law in the home country as discussed in La Porta et al.,(2006) as a summary proxy measure to capture the variation across countries on all of the dimensions discussed above. 9 We believe this parsimony is desirable and necessary as many of the local institutional development measures are highly endogenous and develop together. Our argument above suggests that restatement frequency will decrease with this rule of law measure, conditional on the extent of existing earnings management of the firm. We distinguish between two explanations of restatements (signal of frequent accounting irregularities vs. prudent enforcement of restatement reporting) by conditioning our analysis on existing level of earnings management. If firms correctly report their accounting irregularities, the frequency of restatements will be positively associated with proxies for the extent of earnings management. In contrast, if restatements are likely to be concealed, we expect to find less significant relationship between restatement frequency and earnings management. In our empirical tests, we examine whether the relationship between level of earnings management and tendency to restate financials differs systematically with the strength of home country enforcement. Our prediction is that the association between restatement probability and level of earnings management will increase with home country enforcement. 9 Rule of law measures the extent to which agents have confidence in and abide by the rules of society (as measured in in year These include perceptions of the incidence of both violent and non-violent crime, the effectiveness and predictability of the judiciary, and the enforceability of contracts. See LaPorta et al., (2006) for details. 10

12 H2: The relation between the probability to restate and earnings management will be less positive when the cross-listed firm is from a country with weak rule of law. 2.2 Restatement disclosure method choices of foreign firms listed in the US. Prior studies show that, conditional on reporting a restatement, firms tend to make disclosure choices that minimize the cost of the restatement (Myers et al. [2010]; Files et al. [2009]). Disclosure choices from restatements arise from announcement medium and timing of the restatements. Restatements are generally announced in one (or more) of four different ways: through 1) a separate filing (e.g., form 8-K or 6-K), 2) a press release, 3) an amended financial statement (e.g., 10-K/A, 10-Q/A, 20-F/A, or 40-F/A) or non-timely filing (e.g., 10-NTK or equivalent), or 4) a regularly scheduled financial statements (10-K, 10-Q, 20-F, or 40-F). The first two types, i.e., separate filings or press releases, are the more visible methods which clearly indicate the existence of the restatement and tend to occur on a timely basis. The other two types (i.e., amended statements or regularly scheduled filings) sometimes known as stealth restatements provide information with lesser publicity at the time of the regular filings. If home country characteristics affects foreign firms disclosure methods, we predict that foreign firms are more likely to disclose restatements using stealth medium, and more so if the firm is from a weak rule of law country. In August 2004, the SEC introduced the Final Rule: Additional Form 8-K Disclosure Requirements and Acceleration of Filing Date (SEC 2004) which provided specific rules underlying the use of different disclosure methods of restatements. The 2004 Final Rule limited firms from using stealth restatements (and instead required firms to file an 8-K form) for restatements deemed to be material. However, confusion regarding which restatements constituted as material resulted in ambiguity with respect to the disclosure method required for 11

13 each restatement (Acito et al. [2009]). 10 More importantly for our purposes, this rule was not applicable for foreign firms because foreign issuers as exempt from using 8-Ks for current reports in the first place. Foreign firms instead furnish (not file) current reports on form 6-K which are generally applicable to any material event (Latham and Watkins [2010]). 11 This provided more room for interpretations for foreign firms when it comes to the exact disclosure method required for different restatements. We thus predict that the tendency to disclose material restatements using visible disclosure methods (i.e., non-stealth restatements) will vary with home country enforcement. That is, we hypothesize that link between restatement materiality and restatement publicity will weaken when firms are from a weak rule of law country. H3: Firms from weak rule of law countries are more likely to disclose material restatements with less publicity than those from strong rule of law countries. 3. Sample and Research Design 3.1 Sample construction and descriptive statistics Our sample of cross-listed firms includes all foreign firms listed on major US exchanges (NYSE and NASDAQ) from 2000 to Consistent with prior literature, we include both American Depository Receipts (ADRs) and foreign firms directly listed on the U.S. stock exchanges (Karolyi [2006]). We exclude both OTC traded firms and private equity issuers 10 The official position of the SEC on materiality is The omission or misstatement of an item in a financial report is material if, in the light of surrounding circumstances, the magnitude of the item is such that it is probable that the judgment of a reasonable person relying upon the report would have been changed or influenced by the inclusion or correction of the item. (SAB 99, 11 Filed information is subject to liability provisions of Section 18 of the Exchange Act of 1934 and is automatically incorporated by reference into issues registration statement. Furnished information, on the other hand, is not subject to the same liability section and is not automatically incorporated into the registration statement, unless the issuer specifically requests its incorporation. Unlike filed 8-Ks which hold the preparer liable for any false or misleading information, 6-Ks are only furnished thereby hold the preparer liable only when the preparer is proven to have intentionally provided information that is false or misleading. (Morrison and Foerster, LLP Frequently Asked Questions about Foreign Issuers ) 12

14 because such firms are not required to register under the Securities Act of 1933 and therefore do not need to follow US disclosure practices (Doidge [2004]). We use the country of the headquarter information from Compustat (using variable LOC) to identify our sample of foreign firms. We classify firms as foreign if they are headquartered in a foreign country regardless of the place of incorporation. 12 We drop firm-years with no financial data to compute the control variables in our regression model from Compustat, CRSP, and Worldscope. These selection criterions provide us a sample of 1,364 unique foreign firms and 7,890 firm-years. The restatement sample is obtained from Audit Analytics. Audit Analytics data have been used in prior restatement studies (for example Myers et al. [2011], Badertscher and Burks [2011]) and Scholz (2008) documents that this database provides the broadest coverage of companies and sample period for restatements among other competing sources. 13 Next, we partition the cross-listed firm sample by the level of legal enforcement in the home country. Following prior literature (La Porta et al. [2006), we use the rule of law index from the World Bank (Worldwide Governance Indicators) as the measure of legal enforcement. We classify firms into strong and weak rule of law country-firms using the sample country median of the rule of law index (=1.64). Table 1 (presents the distribution of firm-year observations and restatements for all countries in each of the two groups. The table shows that firms from weak rule of law countries are, on average, less likely to restate that those from strong rule of law countries. Of the 3,061 firm-years in the weak group, 234 (7.64%) were 12 Firms with foreign headquarters yet incorporated in the US frequently represent foreign firm that acquired a subsidiary domiciled in the US (i.e., reverse merger firms). Since such firms are better characterized as non-us firms, we include the reverse-merger firms in our foreign firm sample. 13 For example, another widely used restatement database compiled by the Government Accountability Office (GAO), identified 294 restatements announced by listed firms in 2004 while Audit Analytics identified 418 restatement announcements during the same period. 13

15 eventually restated, and of the 4,829 firm-years in the strong group, 550 (11.39%) were later restated. Overall about 10% of firm years are later restated to correct material errors. Table 2 panel A presents the descriptive statistics across our sample firms from strong and weak rule of law countries and their respective matched samples of US companies. The matched sample is obtained by performing an exact match on year and industry, and propensity score match on size, growth opportunities, leverage and performance (ROA) for each firm year. Compared to firms from countries with strong rule of law, firms from weak rule of law countries are similar in size, have lower total sales, fewer growth opportunities, lower market capitalization, and higher profitability, as measured by return on assets (ROA). These firms are audited by a big five audit firm less frequently (65.9%) compared to firms from strong rule of law countries (69.1%) and US matched sample (79.4%), they have less ownership by US institutions (18.1% vs. 20.1% and 60.7%), and less analyst coverage (4.03 vs. 6.5 and 9.2). Table 2 also shows descriptive statistics of the earnings management measures. With the exception of % of firm-years with small positive income measure (5.8% for weak rule of law countries and 6.7% for US matched sample), firms from countries with weak rule of law, on average, have higher level of earnings management on all the measures relative to the matched sample of the US firms. Consequently, firms from weak ROL countries have significantly higher level of earnings management on the index variable as well (p-value < 0.001). On the other hand, difference in earnings management between firms from strong ROL countries and their matched US sample is not statistically significant (p-value of 0.652). 3.2 Measures of earnings management We use four commonly used measures of earnings management (EM) all estimated at the firm-year level. We use un-restated financials to compute all measures of earnings management. 14

16 Our first measure of earnings management (EM1) is the portion of small positive income (Burgstahler and Dichev [1997]). For each firm-year, we calculate the percentage of years with small positive income using a three-year rolling window, where small positive income is defined as net income that falls between 0 and 1% of the firm s total asset. This measure was used in international setting by Lang et al (2003) and shown to be appropriate in a setting where model estimation-based earnings management measures are likely to be subject to large measurement error. Our second measure of earnings management (EM2) is the magnitude of total accruals measured as the ratio of absolute value of total accruals to absolute value of operating cash flows from Leuz et al. (2003). Magnitude of the total accruals is used as a proxy for the use of managerial discretion, and scaling by operating cash flows adjusts for the differences in firm economics. We use approach in Dechow et al (1995) to measure total accruals. Our third measure of earnings management is the accruals quality (EM3), which captures the estimation errors in the accounting process by measuring how well accruals map into cash flow realizations (Dechow and Dichev [2002] and modified in Wysocki [2009]). We operationalize this measure as the standard deviation of the residual from a firm-level regression of prior and future operating cash flow. Finally, we use a smoothing measure to capture the level of management discretion in the reported earnings. The accounting literature has traditionally used a strong negative correlation between changes in accruals and operating cash flows to proxy for management intervention over and beyond the natural level of accruals accounting (e.g., Francis et al. [2005]). We operationalize this by calculating the three-year rolling spearman correlation between changes 15

17 in accruals and changes in operating cash flows. This measure is our final EM measure (EM4). 14 We aggregate the four measures into an EM index for each firm-year by first ranking the individual measures, approach similar to that used in Leuz et al. (2003). We use the average percentile rank of all four EM proxies. Since firm-level EM measures are known to have large measurement errors, we use the quintile ranks of the aggregate EM index in our empirical analysis. We sign the measures so that higher values reflects higher EM. Table 2 panel B provides the level of earnings management by level of home country enforcement. Consistent with prior literature (Lang et al. [2006]; Leuz et al. [2003]), our sample shows that firms from weak rule of law countries have higher levels of earnings management compared to firms from strong rule of law countries, measured using any of the four variables of earnings management and the index variable, and compared to the US matched sample, on three of the four earnings management variables used, and the index variable. It is worthy to note that despite having higher level of earnings management, on average, firms from weak rule of law countries show less propensity to restate their earnings. In our empirical tests, we statistically test whether the extent of earnings management and restatement probability systematically varies by the level of country enforcement. 3.3 Restatement characteristics Table 3 presents the descriptive statistics relating to restatement characteristics. Out of the 360 restatements by foreign firms, 116 are made by firms from weak rule of law countries and 244 by firms from strong rule of law countries. On all restatement characteristics, the mean values for restatements from weak rule of law countries are not different compared to those 14 Appendix A also defines each measure and provides their sources in prior literature. 16

18 from their matched sample in a statistically significant way except that companies in weak ROL countries are more likely to issue a stealth restatement than the matched sample. For the strong ROL countries restatements are more material as measured their impact on income or equity compared to the matched sample of US firms. Market reaction to the restatement announcements, while less negative on average, is not significantly different in statistical terms for weak rule of law firms compared to the US matched sample. Buy and hold 3-day, 5-day and 30-day returns centered around the earliest mention of the restatement are -1.1%, -1%, and -3.1% for weak ROL firms, respectively, and - 4.1%, -3.3%, and -6.1% for the matched US sample. For these event windows firms from strong ROL countries experience -1.6%, -1.7% and -3.5% event period returns. The univariate comparisons indicate that restatements by firms from weak rule of law countries, though less frequent, are qualitatively and quantitatively similar to the restatements of comparable US firms and foreign firms from better enforcement regimes. The restatement announcement returns shown for US firms is comparable to that reported in other studies (Myers et al. [2011], Scholz [2008]) focused on the time period following Sarbanes-Oxley. In terms of other consequences, univariate comparisons seem to suggest that once a firm makes a restatement, it is likely to face similar regulatory or legal action, regardless of the firm s origin. More precisely, 7.8% of restatements made by firms from weak ROL countries are followed by SEC investigation, while the figure is around 10.2% for the matched sample of US firms. Of the restatements, 7.8% is followed by litigation (including SEC-initiated litigation) for weak ROL firms and 10.7% for the matched sample of US firms. As a comparison, 8.6% of restatements made by strong-rol firms are SEC-investigated, and 9.5% are litigated. We observe significant differences in CEO turnover following restatements CEO turnover for the 17

19 weak and strong ROL firms are significantly lower than the CEO turnover rates for the corresponding matched samples of US firms. 3.4 Research design Frequency of restatements and home country enforcement We first examine the effect of home-county enforcement on restatement probability of foreign cross-listed firms. Using the restatement prediction model in equation (1), we examine the likelihood of foreign firms to restate financial statements as a function of firm-level characteristics and (i.e. size, leverage, growth, profitability) and country-level characteristics. Restatement it = β 0 + β 1 Weak_ROL c + β 2 EM_Index it + β 3-13 Firm_Controls it + β Country_Controls ct + Industry FE + Year FE+ ε it. (1) The dependent variable Restatement it equals 1 if firm i restated financial statements for year t, and 0 otherwise. Weak_ROL c is the primary variable of interest and equals 1 for firms from countries with Rule of law score below our sample country median, and zero otherwise (La Porta et al[2000]). EM_Index it is the earnings management index variable, as described in section 3.3. The same model is also estimated for the US matched sample for comparison purposes. In addition to the earnings management measures, we use number of control variables hypothesized to affect likelihood of a restatement. Financial characteristics of firms found previously to affect propensity to restate include size, leverage, profitability, and growth (Kinney and McDaniel [1989]; DeFond and Jimbavlo [1991]). We also include complexity, measured as number of business segments, and firm s monitoring environment: auditor, analyst following and ownership. Furthermore, to mitigate the possibility that weak rule of law partition is capturing difference in local accounting versus US GAAP, capital market development, 18

20 economic growth or differences in auditor legal liability, all of which may be associated with propensity to restate, we include these as control variables in our specifications. Finally, we include both year and industry fixed effects to control for unobservable time and industry factors that may affect restatement probability. The industry fixed effect captures differences in industry characteristics that could affect restatements and year fixed effect captures unobserved economy wide changes (e.g., enforcement) that affect restatements in a given year. Note that the likelihood of restatement for foreign firms can vary by level of home country enforcement, because home country enforcement can affect 1) the likelihood of accounting irregularities in the reported financials (Lang et al. [2006]; Leuz et al. [2003]) and/or 2) the likelihood of firms to detect and report existing accounting irregularities. In our next empirical tests, we attempt to distinguish the two explanations by examining the effect of home country enforcement by the level of earnings management in the reported earnings Home country enforcement and restatements conditional on level of earnings management Our main prediction is that the association between restatement likelihood and earnings management will vary with home country enforcement. The intuition is that the companies engaging in earnings management should be more likely to restate and the restatement outcome is more likely in the presence of a stronger private and public enforcement culture in that economy. Gong et al. (2009) follow a similar approach in assessing the relevance of internal control reports using measures of earnings quality. We test this prediction using the following model. Restatement it = β 0 + β 1 EM_Index it + β 2-12 Firm_Controls it + β Country_Controls ct + ε it. (2) The dependent variable Restatement it, equals 1 if firm i restated financial statements for year t, and 0 otherwise. In order to compare estimates on our main variable of interest, the β 1 19

21 coefficients, across the foreign firms and its matched US firm sample, we estimate equation (2) as a seemingly-unrelated regression model. If restatements by firms from weak ROL countries are as informative as restatements of equivalent US firms, we expect to find no significant difference in β 1 coefficients between the two samples. To the extent that restatements by firms from weak enforcement countries are less informative, we predict the coefficient to be less positive for these firms relative to the US matched sample Home country enforcement and restatements conditional on level of internal control weaknesses A common criticism with the earnings management measures is that it can subject of considerable measurement error hence be noisy proxy for managerial discretion (Dechow et al. 2010). We thus repeat the analyses with internal control weaknesses as an alternative proxy for earnings quality following Doyle et al (2007b) who find that material control weaknesses are associated with low accruals quality. We collect internal control weaknesses (ICW) disclosure from Audit Analytics. Doyle et al. (2007a) note that the reporting requirement of ICW disclosure under section 302 of Sarbanes- Oxley may be interpreted as a voluntary reporting. Prior literature shows that good internal control systems increase the reliability in the financial reporting (Doyle et al. [2007], Ashbaugh- Skaife et al. [2008]). Although companies had the option to voluntarily report on the state of their internal controls since the passage of Sarbanes-Oxley Act, they were not required to do so until much later. Furthermore, deadline to comply with the reporting requirement and include auditor s assessment of management s assertions regarding control effectiveness was phased in gradually, starting first with US accelerated filers in 2004, followed by foreign-private issuer 20

22 accelerated filers in 2006, and finally all other firms in In order to capture mainly the mandatory reporting of ICW, we focus our analyses on the period starting with fiscal year 2007 and ending in Table 2 shows the descriptive statistics of the ICW disclosure across our country subsamples. For the sample of firm from weak rule of law countries over the period which do provide auditor s attestation on internal controls, 8% of firm-years are originally reported as having internal control weakness(es). During the same period, 3.9% of US matched sample firm years had similar assessments. For the firms from strong rule of law countries, the number of firm years with weak internal control assessment is 5.1%. We test whether the association between internal control weaknesses and earnings quality varies systematically by home country investor protection using the following equation. Restatement it = β 0 + β 1 IC_Weakness it + β 2-12 Firm_Controls it + β Country_Controls ct + ε it. (3) The dependent variable Restatement it, equals 1 if firm i restated financial statements for year t, and 0 otherwise. As before, we expect that the association with internal control weakness and restatement probability to systematically vary by level of home country ROL. We compare estimates on the IC_Weakness term across the foreign firms and its match US firm sample using a seemingly-unrelated regression model. To the extent that restatements by firms from weak rule of law countries are less informative, we predict the coefficient to be less positive for these firms relative to the US matched sample Disclosure choice and home country enforcement 15 See White & Case LLP Guide for Foreign Private Issuers: Preparing your Upcoming Annual Report on Form 20-F 16 It is worthy to note that we measure the presence of internal control weaknesses and restatement concurrently. Since restatements almost always result in an ICW disclosure, one can be concerned that there is a mechanical relationship between ICW disclosure and restatements. To address such concerns, we use the assessment of internal controls for the year prior to the year firms realize the need to change the internal control assessment due to restatement of their financials. 21

23 Finally, we examine whether restatement disclosure practices differ between firms from different legal enforcement regimes. Restatements may be announced in various ways: a separate filing (e.g., form 8-K or 6-K), a press release, amended financial statements (e.g., 10- K/A, 10-Q/A, 20-F/A, or 40-F/A), regularly scheduled financial statements (10-K, 10-Q, 20-F, or 40-F) and non-timely filings (e.g., 10-NTK or equivalent). Restatements that are announced using a separate filing form, press release or non-timely filing clearly indicate the existence of the restatement thus provide the most visibility in the restatement announcement. The other two types of restatements (i.e., those announced in the amended statements or in regularly scheduled financial statements), also known as stealth restatements, manage to provide information with minimum publicity in their regular filings. Prior studies show that the manner in which a restatement is disclosed affects investor reaction. For example, Files (2011) and Myers et al., (2010) show that stealth restatements are associated with less negative stock performance around the announcement date. Thus, if firms from countries with weak legal enforcement use disclosure choice opportunistically, we expect that the decision to announce restatement using more visible disclosure methods (i.e., nonstealth restatements) will be less associated with restatement magnitude. To test the hypothesis, we estimate the following model: Non-Stealth Restatement it = β 0 + β 1 Magnitude it + β 2-12 Firm_Controls it + β Country_Controls ct + ε it. (4) The dependent variable, Non-Stealth Restatementit, is equal to 1 if firm i restates in year t using a separate filing (form 8-K or 6-K) or a press release, and 0 otherwise. Magnitude is an indicator variable equal to 1 if the restatement amount, as a percentage of total assets, is in the top quartile of all restatements, and 0 otherwise. 22

24 We estimate a seemingly unrelated regression model to compare the coefficient estimate for the firms from weak legal enforcement countries and their matched sample, and firms from strong legal enforcement countries and their corresponding matched sample. The matched sample allows us to control for other observable characteristics that differ across firms listed from weak and strong legal enforcement countries. 4. Empirical Results We discuss the results of the empirical analysis in this section. 4.1 The role of home country enforcement on restatement frequency We first examine the effect of home-county enforcement on restatement likelihood. Table 4 Panel A shows the percentage of firm-years restated in the weak and strong ROL countries and each group s US matched sample. Foreign firms from weak ROL countries restate their financial less frequently (7.6%) than firms from strong ROL countries (11.4%). This is in contrast to the matched US firm samples, which show a significantly higher frequency (14.6% and 15.8% respectively) of restatements. Table 4 Panel B presents the results from estimating the logistic regression in equation (1). Model (1) shows the estimated coefficient using the cross-listed firm sample. The coefficient estimate on Weak ROL is negative and statistically significant (coefficient = , p- value<0.01), indicating that firms from weak ROL countries are less likely to restate their financials. It is worthy to note that the level of earnings management has no significant effect on restatement probability (coefficient =0.045, p-value=0.275) for the cross-listed firms. This is in contrast to the matched sample of US firms (Model (2)) which show a greater probability of restatements when firms have higher EM (coefficient =0.092, p=0.004). Consistent with prior 23

25 studies (Kinney and McDaniel [1989]; DeFond and Jimbavlo [1991]), we find that firm size, profitability (ROA), and sales are all significant determinants of restatements. Overall, our findings suggest that for cross-listed firms, home country enforcement is a stronger predictor of restatement probability than the reporting quality of the firm. In Model (3) and (4), we use an alternative measure of reporting quality, the internal control effectiveness, as a predictor of restatements. As before, we estimate equation (1) for the cross-listed and the matched US sample. However, since internal control weakness disclosure reporting became mandatory only after 2007 (SOX section 404), our analysis is limited to the recent years after Model (3) shows the estimated coefficients from the cross-listed firm sample. As before, we find that firms with weak home country enforcement are less likely to restate (-1.446, p- value 0.003). The coefficient estimate on internal control weakness is positive and significant (=0.790, p-value 0.022) suggesting that IC weaknesses are strong predictors of restatements. We find similar effect on IC weakness for the matched US sample (Model (4)). Overall, the results imply that firms from weak rule of law countries are less likely to restate than firms from strong rule of law countries, and such findings are not driven by weak rule of law firms having better earnings quality, as proxied by earnings management and internal control effectiveness. 4.2 Association between earnings management and restatement likelihood We next test if the association between restatement likelihood and earnings management will vary with home country enforcement. Panel A of Table 5 presents the univariate result of the difference in restatement probability for firms in the highest and lowest earnings management quintile. For every country subsample, firms in the high earnings management group restate much more frequently compared to firms in the lower earnings management 24