محافظه کاری، تحقیقات SEC و کلاهبرداری

We present evidence on the relationship between firms that have engaged in fraudulent financial reporting and accounting conservatism. We empirically investigate the extent to which US firms identified by the SEC in their Enforcement Releases demonstrate higher levels of conditional conservatism in order to mitigate information asymmetry and agency problems. Specifically, by assessing the timing of changes in the litigation risk environment for fraud firms, we document how differences in heightened legal liability guide changes in conservative accounting behavior. Compared to a matched non-fraud control sample, we document that fraud firms have significantly lower levels of accounting conservatism in the pre-fraud period. Consistent with changes in potential legal liability, we find an increase in accounting conservatism for fraud firms during the SEC investigation period. Subsequently, during the public discovery of fraud, any increases in accounting conservatism are marginal and appear to converge back to lower levels compared to the SEC investigation period. Overall, our findings suggest more temporary changes in conservative reporting in the short-term for fraud firms. We also document that increased levels of accounting conservatism for fraud firms are not due solely to the passage of the SOX Act. Our findings aid in explaining fraud firms’ incentives and opportunities for accounting conservatism and lend support for why standard setters, regulators and auditors should continue to monitor and re-evaluate conservatism’s short-term effects that are conditioned on changes in a firm’s risk environment.

A large stream of research has evolved on accounting conservatism. However, little is known about whether or how conservatism manifests itself in fraud firms. The motivation for this study is to shed light on the current debate which centers on whether conservative financial accounting represents a universal desirable attribute of qualitative accounting information. Prior research has theoretically linked increases in accounting conservatism to unexpected changes in litigation risk (Watts, 2003b). Yet, it is not entirely clear whether firms accused of fraud, which are characterized by high levels of litigation risk, have higher levels of accounting conservatism or when changes in conservatism are likely to occur. Since Givoly et al. (2010, p. 221) state that it is important to examine the “episodic” nature of earnings management, we examine whether accounting conservatism, based on our sample, is sustainable. US GAAP affords managers considerable reporting discretion, with accounting conservatism being one of those reporting practices (Bagnoli and Watts, 2005 and Guay and Verrecchia, 2007). Studies reveal managers use accounting conservatism as a means to address information asymmetry and agency costs in their private information and financial reporting strategies (Chaney, 2007 and LaFond and Watts, 2008). In this current study, we document how agency conflicts and the likelihood of legal liability affect fraud firms’ corporate disclosure behaviors. Specifically, by assessing the timing of changes in the litigation risk environment surrounding the period of manipulation in the financial statements for fraud firms, we provide new insight on how fraud firms temporarily alter their conservative accounting practices in order to reduce information asymmetry and potentially regain investor confidence. The timing of changes is relevant because it is often assumed that fraud firms’ aggressive accounting behaviors and accounting conservatism practices, in general, are not likely to persist, but rather vary in response to operating incentives and the likelihood of litigation. Understanding how accounting conservatism manifests itself in fraud firms also indirectly provides new evidence on the prediction of and remediation effects to accounting failures. Although increases in accounting conservatism are suggested to aid in facilitating the effective monitoring of managers and contracts (Ball et al., 2000, Watts, 2003a and Watts, 2003b), we are motivated to study the interrelationship between fraud firms and conservatism because of the concern for how accounting conservatism, characterized as a monitoring-disclosure mechanism, is reflective of providing an economic benefit to market participants. Thus, labeling accounting conservatism as a quality metric under all circumstances appears problematic. Moreover, in a recent study, Jackson and Liu (2010) provide evidence that managers use the flexibility of conservatism as a strategic earnings management process which contradicts the notion that accounting conservatism provides an economic benefit. By focusing on individual accrual accounts, they show that the allowance for doubtful accounts has become more conservative over time and firms subsequently manage bad debt expense downward to meet or beat analysts’ forecasts of earnings. Prior research has generally regarded accounting conservatism as a mechanism that provides an economic benefit in response to a firm’s incentives. Chen et al. (2007) state that contract efficiency can be improved upon by using accounting conservatism. Ahmed et al. (2002) report that accounting conservatism mitigates bondholder–shareholder conflicts over dividend policy. In this respect, accounting conservatism serves as a protective action that benefits lenders by providing them with a barrier to reduce downside risk. In particular, accounting conservatism leads to contracting benefits for borrowers in the form of lower interest rates (Zhang, 2008) and firms that institute higher standards of corporate governance have exhibited higher levels of accounting conservatism (García Lara et al., 2009). Moreover, Kim and Pevzner (2010) find a reduction in future bad earnings news for firms with conservative accounting practices. Chung and Wynn (2008) find that directors’ and officers’ liability insurance coverage and cash indemnification for Canadian firms reduces expected legal liability, causing a decrease in conservative earnings. García Lara et al. (2011) provide evidence that conservatism imposes greater verification requirements relating to increasing earnings, thus reducing information uncertainty and leading to a lower cost of capital. Conversely, prior research has also regarded accounting conservatism as a distortion of accounting information. While accounting conservatism is not a direct qualitative characteristic of the FASB conceptual framework, the FASB has been considering revising conservative disclosure practices in order to achieve more “neutrality of information” (FASB, 2005).2 They state that in order for accounting information to be relevant, it has to be verifiable, neutral, and contain representational faithfulness.3Givoly and Hayn (2002) contend that accounting conservatism may create unwanted bias and noise. Bias and noise in the financial statements generate “soft” accounting numbers4 (Watts, 2006 and Mackintosh, 2006). In one of the first studies to provide evidence on the adverse effects of the conservatism doctrine, Mensah et al. (1994) document that, based on the regulatory environment, HMOs intentionally understated profits in order to manage industry norms. Similarly, Givoly and Hayn (2000, p.292) maintain that accounting conservatism does not mitigate but rather “induces [information] asymmetry in the timeliness of incorporating economic events in reported earnings.” An understatement of earnings in the current period can lead to an overstatement of earnings in future periods which can alter earnings quality (Penman and Zhang, 2002). We also note that the use of fraudulent accounting tactics, in regards to conservative financial reporting, is evident in SEC Accounting and Auditing Enforcement Releases (AAERs) where firms have been accused of manipulating reserves.5 In its 2010 study, the Deloitte Forensic Center revealed that one of the 12 main categories of alleged fraud schemes included the manipulation of reserves. In fact, the manipulation of reserves accounted for 7% of total fraud schemes for the period 2000–2008.6 Further, COSO (2010) reports that some SEC AAER cases denote there is an involvement by senior management (e.g., tone at the top) in the fraud manipulations, thus also supporting the exploitation of management discretion of financial reporting. The notion that fraud firms can use accounting conservatism at their discretion, and are incentivized due to changes in their litigation landscape, questions the usefulness of conservatism in evaluating a firm’s reporting quality. As a result, our study has implications for other select groups of firms who may articulate the use of accounting conservatism based on discretionary motives. Our sample consists of US public firms that have been identified by the US Securities and Exchange Commission (SEC) in their Enforcement Releases during the period 1985–2007. Contemporaneous research studies have used these civil litigation proceedings as proxies for fraud and litigation risk (Erickson et al., 2006; Carcello and Nagy, 2004; Palmrose and Scholz, 2004, Bonner et al., 1998, Dechow et al., 1996 and Feroz et al., 1991).7 The selection of this ex post sample of fraud firms provides a natural setting to study changes in accounting conservatism that are expected to vary with changes in litigation risk.8 We hand collect data based on reading each SEC Enforcement Release to determine the information content (i.e., fraud manipulation dates, dates of enforcement releases, details of the fraud, etc.). Our final sample is composed of firms based on a variety of types of fraud.9 Our empirical results show that the timing of changes in the litigation risk environment guides changes in conservative accounting behavior and this discretionary behavior is shown to significantly differ between fraud and non-fraud firms. Specifically, we document significant differences in conservatism levels between these two groups of firms over time. These differences are evident whether we utilize an aggregate asymmetric timeliness of earnings measure or a firm-specific C-Score measure of conservative accounting. Based on these two conservatism measures, we find that fraud firms, compared to non-fraud firms, have significantly lower levels of conservatism in the pre-fraud period than during the period they are manipulating their financial statements. Next, as time elapses, we find increasing differences in accounting conservatism for fraud firms during the SEC investigation period, which significantly contrasts to the pre-fraud and fraud manipulation periods. During the public discovery/announcement of fraud, compared to the SEC investigation period, we find that any increases in accounting conservatism are marginal and appear to converge back to lower levels for fraud versus non-fraud firms. Using the C-Score measure of accounting conservatism, we find that during the public discovery of fraud, fraud firms’ conservatism levels are still higher compared to pre-fraud levels. Additionally, because of concerns that changes in conservative reporting may be due to regulation, we control for the post-SOX period. We find that, in spite of increased regulation, significant differences in conservatism levels remain between fraud and non-fraud firms as the litigation landscape changes. In summary, our findings show that conservative accounting practices vary across time, a finding consistent with more temporary, less sustainable levels of conservatism for fraud firms. Therefore, we provide evidence that since fraud firms can exercise strategic discretion over conservative financial reporting that selectively fits their incentives, the timing differences of conservative accounting have implications for public policy. These results also indirectly provide evidence that differences in accounting conservatism across time can provide clues regarding the internal management practices of firms. We contribute to the literature by examining accounting conservatism based on fraud firms’ motivations and their changing litigation landscape. We note that conservatism levels, in general, are not static for all firms. A limitation of prior studies is that they have modeled conservative practices over long-term horizons. Over the course of time, conservatism levels change and the environment which prompts the onset of conservatism changes also. Kwon et al. (2001, p. 29) show that in an unlimited liability setting, any bias may be random and accounting conservatism would not be expected to develop as a “pervasive and enduring phenomenon.” Therefore, fraud firms would not be able to use higher levels of accounting conservatism over a long time horizon and continue in perpetuity. We argue that it is important to understand the motivations for these changes since conditional accounting conservatism is based on managerial discretion.10 Managerial discretion often increases during periods of uncertainty and information asymmetry, as is often the case with fraud. As such, it is important to gain a better understanding of this relationship for firms accused of fraud and to document the timing of changes in their litigation risk environment. We provide evidence that, without the proper disclosure of conservative accounting policies, information asymmetry may be exacerbated. The uniqueness of our study is based on several aspects. Our first objective is to determine whether or not US firms accused of fraud are more conservative in their financial accounting practices. Moreover, we examine when shifts in these incentives and subsequent increases in accounting conservatism are most likely to occur. Next, we measure accounting conservatism from the perspective of conditional conservatism by examining the aggregate asymmetric timeliness of earnings (Basu, 1997). Still, due to studies that reveal potential limitations with the aggregate Basu (1997) model (Givoly et al., 2007 and Dietrich et al., 2007), we also take into consideration the C_Score (Khan and Watts, 2009), a hybrid measure that represents a firm-specific measure of accounting conservatism derived from an extension to the aggregate Basu (1997) model, but which takes into consideration the cross-sectional and intertemporal variation in the market-to-book ratio, leverage, and size. Lastly, our study contributes to the literature by following the suggestion by Watts (2003b, p.294) that “finer tests of variations in accounting conservatism” during expanding regulatory periods be performed in order to determine the effect of regulation on conservative financial reporting. As such, we also determine whether the period of regulation surrounding the Sarbanes–Oxley (SOX) Act of 2002 has led accounting conservatism to vary between fraud and non-fraud firms. The SOX Act corresponds to a shift in the information environment representing more litigiousness for non-compliance. Therefore, we document whether the reaction to the future threat of litigation for fraud firms is due to external SOX regulations that impose stricter requirements or due to firms’ internal managerial incentives. As such, by controlling for the SOX Act, we determine whether the Act can be credited to the increased levels of accounting conservatism for our sample of fraud firms or whether accounting conservatism can be attributed to the onset or remediation of fraudulent financial reporting. The remainder of this study is organized as follows: in the next section, we review the relevant prior literature and present the hypotheses. Next, we discuss the sample selection process and data sources and then we define the research methodology. Thereafter, we present the empirical results and additional sensitivity tests. Lastly, we summarize the major findings and discuss limitations and directions for future research.

We link the theory of accounting conservatism to fraudulent financial reporting by focusing on firms’ incentives and opportunities by examining the role of accounting conservatism in the context of (1) a monitoring-disclosure tool to deter litigation and (2) an earnings management tool that could provide flexibility in altering financial statements in order to maintain and restore financial reporting credibility. Specifically, we assess the timing of changes in the litigation risk environment for fraud firms, compared to a matched non-fraud control sample, and document how heightened legal liability guides changes in conservative accounting behavior. Our measurement approach is based on conditional conservatism represented by the aggregate asymmetric timeliness of earnings (Basu, 1997) and the C_Score, a hybrid, firm-specific measure of conservatism that incorporates the market-to-book ratio, leverage, and firm-size (Khan and Watts, 2009). The findings of our paper are similar in spirit to Farber (2005) who show that fraud firms do take actions in an attempt to restore trust and curry favor with investors, regulators, and auditors. Similar to the theory developed by Kothari et al. (1988) and Basu (1997), we empirically document that accounting conservatism arises during high litigation risk periods for fraud firms. Compared to a matched non-fraud control sample, we document that fraud firms have significantly lower levels of accounting conservatism at least 3 years prior to the first occurrence of fraud manipulation in the financial statements and they do not exhibit higher levels of accounting conservatism during the fraud manipulation period. A low level of accounting conservatism during the pre-fraud period could be used to signal the likelihood or predictability of fraud. In this sense, determining when changes in conservative financial reporting occur could be used to combat the discovery of potential fraudulent practices. This finding complements Farber (2005) who shows that fraud firms have weaker governance systems in place prior to committing fraud and suggests a potential weakness in corporate governance for these firms. During the SEC investigation period, we document that fraud firms anticipate heightened litigation risk and increase their use of accounting conservatism to counterbalance the future threat of litigation and changes in legal liability. By observing changes that occur along the litigation landscape beyond a mere pre- versus post-analysis, we show that while fraud firms certainly have the right and predisposition to take steps to repair their reputation, repairing it before the public has become aware of the manipulation (during the SEC investigation period) implies that managers and auditors use accounting conservatism as an incentive to time the reduction of their litigation risk. In this sense, accounting conservatism is used as an implied communication mechanism to reduce information asymmetry and to provide investors with an indication that the firm is sound. However, given that we utilize an ex post fraud sample, we show that in this specific setting, accounting conservatism does not facilitate the effective monitoring of managers and does not reduce shareholder conflicts. Moreover, accounting conservatism is significantly higher during the SEC investigation period than during the pre-fraud manipulation period. During the public discovery of fraud, any increases in accounting conservatism are marginal and appear to converge back to lower levels. Further, we also document that increased levels of accounting conservatism for fraud firms are not due solely to the passage of the SOX Act. Our findings aid in explaining fraud firms’ incentives and opportunities for accounting conservatism and lend support for why standard setters, analysts, regulators, and auditors should continue to monitor and re-evaluate its increasing effects. We show that conservative accounting serves as a self-protection mechanism used by managers of fraud firms to convey private information. Our findings suggest more temporary changes in conservatism in the short-term for fraud firms and provide support for why it is important to consider the litigation landscape as firms increase their use of accounting conservatism. Our results have implications for how accounting measures can vary based on short-term incentives, creating non-sustainable accounting metrics. Additionally, our findings have implications for financial statement analysis as the US moves towards convergence of International Financial Reporting Standards (IFRS), creating modifications related to current disclosure practices. Similar to other studies that have used SEC Enforcement Releases, there is a potential for a type I error due to the SEC misclassifying a firm as having committed fraud. However, the SEC generally only pursues firms whose evidential matter indicates a high probability of violation after a thorough investigation and substantial evidence has been attained (Caskey and Hanlon, 2011 and Feroz et al., 1991). There is also the probability of a type II error because some firms that may have committed fraud have not been pursued by the SEC (Erickson et al., 2006). Further, we note that, although there is the probability that the SEC has correctly identified fraud firms, the first (initial) year of fraud manipulation in the financial statements may have been misspecified. Future research could benefit by gaining an understanding at the firm-level of how accounting conservatism reacts to changes in agency problems in an international context by taking into account litigation risk beyond the scope of legal traditions and origins. Research could also investigate how fair value accounting is associated with accounting conservatism and changes in litigation risk.