The Deterrence Effects of SEC Enforcement and Class Action Litigation
Jared JenningsDoctoral Student Foster School of Business University of Washington Simi Kedia
Associate Professor, Finance and Economics Rutgers Business School
94 Rockefeller Road New Brunswick, NJ Email: [email protected]
Shivaram Rajgopal*
Schaefer Chaired Professor of Accounting Goizueta Business School
Emory University 1300 Clifton Road NE
Atlanta, GA 30322
Email: [email protected]
June 2011
Preliminary and Incomplete Please do not cite without permission Abstract:
The United States’ (U.S.) Congress specifically requires the SEC to deter potential miscreants via its enforcement actions against firms that engage in fraudulent financial reporting. The U.S. is also unique in allowing private enforcement via class action lawsuits. In this paper, we investigate whether SEC
enforcement actions and class action lawsuits, over the years 1996-2006, deter aggressive financial reporting behavior among the peers of fraudulent firms. We find significant deterrence associated with both SEC enforcement actions and class action lawsuits. The average peer firm, subject to SEC action and/ or litigation, reduces discretionary accruals equivalent to 14% to 22% of the median return on assets (ROA) in the aftermath of such enforcement. The results also inform target selection criteria associated with greater deterrence. Moreover, repeated and sustained enforcement in an industry, as opposed to isolated investigations, provides more effective deterrence.
*Corresponding author. We acknowledge helpful comments from Gautam Gausaumi, David Erkens, and workshop participants at Drexel University, Rutgers University, University of Washington, 3rd Annual Conference at Baruch College, AAA Western Meetings - 2011. We would like to thank Jonathan Karpoff, Scott Lee and Gerald Martin for graciously providing us with the SEC enforcement data. We are grateful for financial support from the Foster School of Business, Goizueta Business School and the Whitcomb Center at Rutgers Business School.
1
The Deterrence Effect of SEC Enforcement and Class Action Litigation
1.0
IntroductionThe SEC has been heavily criticized for its handling of the $7 billion Stanford fraud and has been sued for its “gross negligence” in not discovering the long running fraud by Bernard Madoff, among others.1 Is the SEC’s failure to discover these large frauds in a timely manner a testimony to its
ineffectiveness? Should the SEC even try to achieve complete compliance? As Stigler (1970) points out, complete enforcement is very costly to be a feasible enforcement strategy. In the absence of complete enforcement, a regulator has to focus on effective deterrence such that fewer of the non-targeted firms adopt questionable practices. Indeed, the United States’ (U.S.) Congress specifically requires the SEC “to achieve an appropriate level of deterrence in each case and thereby maximize the remedial effects of its enforcement actions” (H.R. Rep. No. 101-616, at 13, 1990).
Despite the obvious importance of understanding whether SEC enforcement deters fraud or aggressive accounting, empirical evidence on the issue is sparse. In this paper, we assess whether peer firms become less aggressive in their financial reporting decisions when a target firm in their industry has been subject to SEC enforcement for violating GAAP (generally accepted accounting practices).
As a non-trivial fraction of the SEC’s investigations are confidential, peer firms are likely to be uninformed about (i) the probability of being subject to an SEC investigation; or (ii) the practices the SEC is investigating. Consequently, observing a public SEC enforcement action in its industry against a target firm is likely to increase a peer firm’s knowledge about SEC activity and cause it to revise upward its subjective probability of attracting such an action against itself. Such an increase in the probability of getting “caught” will decrease a peer firm’s propensity to engage in aggressive earnings management.
Any potential deterrence effect of SEC enforcement actions is subject to counter arguments. The SEC publicly targets a very small fraction of firms – in our sample only 0.74% of firms were subject to
1
See “Foundation Blames Madoff Loss on SEC Negligence in Suit” by Chad Bray in the Wall Street Journal, 24
September 2010 and “SEC Under Fire over Stanford Fraud Inquiry” by Jean Eaglesham in Financial Times on September 23, 2010.2
SEC enforcement. At these low levels of enforcement, a substantial fraction of misreporting is likely to go undetected. Therefore, if potential miscreants consider the probability of detection to be too low, they are unlikely to change their behavior. As the deterrence effect of SEC enforcement is ultimately an empirical issue, we attempt to estimate it by examining whether peer firms in the industry of the target significantly reduce their discretionary accruals in the year after the SEC enforcement action.
Though the SEC is responsible for public enforcement, the U.S. is unique in also allowing private enforcement in the form of securities class action litigation. Securities class action litigation for alleged reporting irregularities is more likely against an average firm - in our sample 1.28% of firms are subject to class action litigation. Moreover, Coffee (2007) claims that plaintiff attorneys extract more funds from the corporate pocketbooks than all federal and state regulators combined. The greater likelihood of class action litigation, combined with higher monetary sanctions, likely renders lawsuits as a potentially effective way to deter reporting irregularities at peer firms. However, class action lawyers are likely to only target firms with an ability to pay damages and are hence likely to miss high deterrence targets that lack financial resources. Moreover, senior managers rarely contribute to class action settlements from their own funds and therefore may have no incentive to change their behavior when others in their industry are subject to class action litigation. The possibility of frivolous lawsuits casts further doubt on the potential deterrence effect of lawsuits. In this paper, we also examine whether class action litigation is associated with a move towards more conservative accounting by peer firms.
During our sample period 1996-2006, the SEC initiated 474 investigations against several firms accused of financial statement misrepresentation, based on the data graciously provided to us by Karpoff, Lee and Martin (2008, hereafter KLM). Focus on the 1996-2006 period is driven by the absence of class lawsuit data before 1996. Over 1996-2006, there were 1,111 securities class action lawsuits filed as per the Stanford Securities Class Action Clearinghouse database against firms for GAAP violations. In our sample, 283 cases or 60% of all SEC enforcement actions are also accompanied by litigation. The overlapped sample of SEC enforcement actions and class action lawsuits, however, accounts for a small fraction (25%) of all class action litigation.
Firm characteristics likely determine whether a firm will be subject to SEC enforcement, litigation, or both. Specifically, firms subject to class action litigation have higher stock price declines during the fiscal year prior to the filing of the lawsuit and have higher cash holdings than those firms subject to SEC enforcement actions. This is not surprising as “provable damages” and ability to pay are both considerations in the lawyers’ decision to litigate. Firms that are subject to SEC enforcement and not litigation have better prior stock return performance, lower levels of cash holdings, and the lowest growth rate in sales. Target firms subject to only SEC enforcement also belong to industries that are the least populated, i.e., have the smallest number of firms in the industry.
In order to tease out the potential deterrence effect of SEC enforcement from that of securities litigation, we estimate the deterrence effects associated with targets for three separate groups: (i) targets subject to both SEC enforcement action and lawsuits; (ii) targets that are subject to only SEC
enforcement; and (iii) targets that are subject to only class action litigation. We find significant evidence of deterrence in all three categories -- peer firms reduce their discretionary accruals after target firms in their industries are subject to SEC enforcement, class action lawsuits or both. This deterrence effect of SEC enforcement and class action litigation is qualitatively similar across different definitions of industry and peer firms (four digit SIC or five digit NAICS). Deterrence is both statistically and economically significant. We find that the reversal of income-increasing accruals for the average peer firm, when targets are subject to both SEC and litigation, is 0.35% of total assets. Considering that the median ROA (return on assets) in the sample is 2.5%, the reversal per peer firm represents a 14% reversal of an average firm’s annual ROA. The estimated deterrence when targets are subject to only SEC enforcement (0.45% of assets) or only litigation (0.56% of assets) is higher, although the differences between these three sets of reversal estimates are not statistically significant.
We also study target firms’ characteristics that are likely to be associated with greater deterrence. We find that visible targets, proxied as targets with large market capitalization, or greater market share by sales, or high sales growth, are associated with incremental deterrence only when targets are subject to both SEC action and litigation. Characteristics of the target industry, as opposed to the target
characteristics, have a greater impact on deterrence. In particular, SEC enforcement in competitive industries and litigation in populated industries is associated with significant incremental deterrence.
We also examine whether the intensity of SEC enforcement or litigation within an industry is associated with differing deterrence. In other words, does the SEC need sustained activity in an industry to make firms change their behavior? Or should the SEC spread its enforcement dollars widely -
investigate all industries even though there are few cases in each industry? We examine this issue by identifying new targets in their industry and studying whether these new targets are associated with incrementally greater deterrence. A newly targeted industry is operationalized as one that has not experienced a SEC investigation or lawsuit in the last three years. We find no evidence that newly targeted industries are associated with any deterrence – from SEC enforcement or from litigation. Hence, effective deterrence requires sustained SEC activity and litigation in the industry.
Finally, we exploit the long time series of SEC enforcements from 1976 to 2006 to study
deterrence over the past three decades.2 As we do not have litigation data over this time period we are not able to study possible deterrence from litigation. However, we are able to control for which among the SEC enforcement actions were also subject to litigation. We find significant evidence over this long time period that SEC enforcement actions are associated with change in behavior, in particular reduction in accruals, of both target as well as peer firms.
The long time series of SEC enforcements actions also allows us to examine the potential influence of the SEC commissioner and his/ her political affiliation. The general belief that Republican presidents are good for business or Democratic presidents favor more regulation – would suggest
potentially greater deterrence in Democratic regimes. Consistent with this intuition, we find significantly higher reversals in accruals for target firms under Democratic regimes. However, there is no difference in the reaction of peer firms in one regime relative to the other.
2
The KLM dataset contains SEC enforcement actions over the 1976-2006 time period. KLM have also collected data on which of these enforcement actions are also subject to litigation.
We contribute to the literature in three ways. First, our paper is among the first to systematically evaluate the potential deterrence effects of SEC enforcement actions and class action lawsuits. Our primary finding is that both SEC enforcement actions, as well as, securities class action litigation are associated with significant deterrence. Such evidence of significant deterrence associated with SEC enforcement actions is not entirely consistent with the criticism directed at the SEC in the recent past. Though it is hard to say what level of deterrence the SEC should have achieved, the results clearly suggest that SEC enforcement policy has made peer firms more conservative in their reporting strategy.
Second, we contribute by documenting that class action lawsuits, although often maligned as frivolous and socially wasteful (e.g., Easterbrook and Fischel 1985; Thakor, Nielsen and Gulley 2005; and Langevoort 1996), can have positive externalities by curbing aggressive reporting behavior of peer firms. Though class action litigation does not explicitly seek to deter, this byproduct of litigation has a significant impact on the reporting policies of peer firms.
Finally, our results also provide guidance as to the choice of target firms that are likely to maximize potential deterrence. Large firms, fast growing firms and those with a high market share in their industry are associated with somewhat greater deterrence. Targeting competitive industries, more populated industries and sustained enforcement activity in the industry is also associated with higher deterrence.
The remainder of the paper is organized as follows. Section 2 discusses the background and hypotheses. Section 3 describes the data and empirical specification used to evaluate the deterrence effects of SEC enforcement actions and lawsuits. Section 4 presents the base empirical results; section 5 presents cross sectional results; section 6 discusses the effect of newly targeted industries and finally section 7 summarizes and concludes.
Becker (1968)’s pioneering work on the economic model of crime and optimal penalties suggests that firms will commit fraud when the expected benefits from these activities exceed the expected costs. The underlying mechanism in this economic model of crime is deterrence, i.e., firms faced with higher probability of getting caught are less likely to commit fraud.
2.1 Deterrence through SEC enforcement
There are two potential channels through which the SEC’s enforcement policy can reduce firm’s propensity to commit fraud: (i) specific deterrence or the potential for the lower likelihood of
misreporting by firms that have been investigated by the SEC; and (ii) general deterrence or when peer firms increase their subjective probability of getting caught subsequent to the SEC targeting a culpable firm in their industry. Of these, the role of specific deterrence is likely to be limited because the SEC investigates only a small fraction of firms and the likelihood of getting investigated twice by the SEC is low.3 Further, such knowledge of the SEC’s enforcement activity fades with time and management turnover within the firm.
The potentially more important channel for deterrence is the impact on other firms in the industry that observe SEC enforcement actions. Firms that are poorly informed about the likelihood of an SEC investigation are likely to become more educated about the SEC’s policing activity when they observe the SEC launch an enforcement action on another firm in their industry. An optimal enforcement policy followed by the SEC would involve selecting target firms to maximize the deterrence among these peer firms. Such general deterrence is the primary focus of our study.
Because a non-trivial fraction of the SEC’s investigations are confidential, peer firms are likely to be uninformed about (i) the probability of being subject to an SEC investigation; or (ii) the practices the SEC is investigating.4 Consequently, observing a public SEC enforcement action in its industry against a
3
Of course, the market is likely to increase its scrutiny of a firm that has been investigated once by the SEC. Hence, a follow up investigation by the SEC might not be necessary.
4
Levitt (1998) points out that firms engaging in aggressive accounting may be poorly informed about the likelihood of detection. As there are no published statistics on how many complaints the SEC gets, which among these are informally investigated by the SEC, and finally what fraction are formally investigated, it is quite feasible that firms are uninformed about the probability of an SEC investigation. Consistent with this conjecture, Kedia and Rajgopal
target firm is likely to increase a peer firm’s knowledge about SEC activity and cause it to revise upward its subjective probability of attracting such an action against itself. Such an increase in the probability of getting “caught” will decrease a peer firm’s propensity to engage in aggressive earnings management.
The SEC can potentially also use a penalty structure to generate deterrence. A large enough penalty, even if associated with a low likelihood of detection, can generate significant deterrence. However, there are limits to how large the penalties can be. As pointed by Stigler (1970), the marginal deterrence associated with heavy penalties is small if minor crimes carry large penalties. In other words, if a thief loses his hand for stealing $5, then he might as well steal $5,000. The SEC’s penalty structure has been examined by KLM (2008) who find that SEC penalties are increasing in the misreporting, and though non-trivial are not noticeably large. Coffee (2007) claims that plaintiff attorneys extract more funds from the corporate pocketbooks than all federal and state regulators combined further highlighting the fact that SEC penalties are not high. Moreover, penalties are not known at the time the SEC
investigation first becomes public and are hence unlikely to be associated with change in peer firm behavior around the time the target firm is investigated. Consequently, we do not focus on penalties in this study.
The potential deterrence effects of SEC enforcement actions are subject to several counter-arguments. First, SEC enforcement actions are not numerous enough to constitute a significant threat to most firms. Indeed, as discussed in detail later, only 788 SEC enforcement actions have been filed against firms over the period 1976-2006 (KLM 2008). At these levels of enforcement, many cases of fraud will potentially go undetected given that 17,000 registrants file four quarterly financial statements every year. Thus, if potential miscreants consider the probability of detection to be too low, SEC enforcement actions will not provide effective deterrence.
Second, the SEC suffers from severe resource constraints. For instance, the SEC receives in excess of 20,000 complaints a year (SEC 2001). Hence, any attempt to prioritize its enforcement efforts
(2011) find that firms vary in their awareness of the SEC’s enforcement activity and such differential awareness impacts their propensity to engage in earnings management that results in a restatement.
will leave a lot of potential violators off the hook. Third, the SEC is more interested in investigating ongoing violations rather than those that are not current (see Cox, Thomas and Kiku (2003)). Such preferences and other political pressures may lead the SEC to target firms that have limited deterrence effect. Ultimately, the deterrence effects, if any, of SEC investigations is an empirical question.
2.2 Deterrence through class action lawsuits
In contrast to the SEC enforcement actions, class action lawsuits are relatively more frequent. Coffee (2006) reports that between 2.1% to 2.8% of firms were subject to securities class action lawsuits every year between 1995 and 2005.5 Thus, the likelihood of detection is far higher under private
enforcement via class action lawsuits relative to public enforcement by the SEC. Much like firms may be uninformed about the likelihood of SEC enforcement, they might be uninformed about the likelihood of class action litigation. Firms targeted by class action litigation, as well as peer firms that observe others in their industry being sued are likely to get better informed about the likelihood of being subject to litigation. The resultant increase in the subjective probability of being litigated against will result in a lower likelihood of earnings management.
Moreover, Jackson (2007) finds that the majority of the total monetary sanctions imposed on firms stem from private class action lawsuits and not via public enforcement actions of the SEC. He finds that over the years 2002-2005, the average payments made by culpable firms on account of SEC
monetary sanctions total $801 million relative to $1.923 billion attributable to class action rewards and settlements. The much higher level of monetary sanctions associated with class action litigation also increases the expected costs of engaging in questionable behavior and therefore is likely effective at deterrence.
However, one cannot automatically infer that class action lawsuits are more effective at deterring potentially fraudulent behavior relative to public enforcement by the SEC. Securities class action
5
In our sample, we find that 1.28% of firms are subject to litigation. Note that that lower fraction of firms subject to litigation in our sample relative to Coffee (2006) is due to the fact that we focus only on GAAP related violations
lawsuits are often criticized for being frivolous (see Easterbrook and Fischel 1985; Thakor, Nielsen and Gulley 2005; and Langevoort 1996). Targeting of innocent firms is likely to encourage fraudulent behavior as it diminishes the marginal cost of engaging in such behavior. Further, class actions lawyers are known to target large firms with deep pockets. This creates the possibility that a whole set of firms, i.e. those that are small and cash strapped, are not pursued by class action lawyers and potential
deterrence arising from these firms is missed by securities class action litigation.
Thirdly, senior managers and directors rarely, if ever, contribute to class action settlements (Alexander 1996 and Black et al. 2006) as their liability is covered by D&O insurance. Legal scholars have argued that the absence of personal monetary liability to managers in securities litigation is likely to limit the potential deterrence effect of securities litigation (See Coffee 2007, Klausner 2009). However, managers in many large public firms hold equity in their firms and see significant reduction in their wealth due to the stock price decline that follows in the aftermath of securities litigation. Such stock price declines cause many managers to suffer significant personal monetary losses when their firms get sued.
Lastly, Cornerstone Research (2007) reports that auditors and underwriters are less likely to be named in a securities class action lawsuits. This lack of liability on these important gatekeepers reduces their incentives to keep errant managers in check. However, this appears to be changing in recent times as 25% of lawsuits in 2009 named an auditor or an underwriter as liable (Cornerstone 2009). Therefore, the efficacy of class action lawsuits in deterring potential reporting fraud is again an empirical issue.
2.3 Interaction of SEC enforcement and securities litigation
Many cases of corporate wrongdoing are subject to both SEC enforcement and securities litigation. Klausner (2009) suggests that cases where the SEC and litigation overlap are likely to be the most egregious cases of corporate wrongdoing. Those that are less egregious are likely to be targeted by securities litigation but not by a resource constrained SEC. If the overlap of SEC enforcement and securities litigation proxies for the most egregious cases, then we would expect these to be associated with the greatest deterrence.
Alternatively, the presence of SEC enforcement might make the lawsuit against the firm more credible. Consequently, such cases are likely to settle for larger amounts than private suits without SEC proceedings (see Cox, Thomas and Kiku 2003). Block, Nold and Sidak (1981) find that increasing antitrust enforcement, in the presence of a credible threat of large damage awards in lawsuits, has the deterrent effect of reducing collusive pricing. Thus, a combination of SEC action and litigation is likely to be the most costly to the target firm in terms of damages and hence potentially have a greater deterrent effect than either SEC action or litigation by itself.
Although these overlapped cases of SEC enforcement and litigation are likely to be the most egregious cases, they may not be associated with the highest level of deterrence. To the extent these egregious cases involve blatant one-of a-kind fraud cases, such as seen in Enron or Worldcom, peer firms may perceive such cases to be idiosyncratic with few implications for their own behavior. It is quite plausible that milder misreporting activities are possibly more widely practiced and action against such activities may be associated with greater deterrence. By way of an analogy, while we are likely to reduce our driving speed when we see another car flagged by the police for speeding, we are unlikely to change our behavior following a murder trial where a motorist ran over a pedestrian.
3.0 Data and Empirical Specification
The data on the SEC enforcement actions is obtained from Karpoff, Lee and Martin and extends from 1976 to 2006. To be consistent with the post PLSRA litigation data that are available from 1996 onwards, we also examine SEC enforcement over the 1996 to 2006 time period.
3.1. The year of SEC enforcement
The SEC’s inquiry into the firm consists of multiple steps. Miscreant firms may be brought to the SEC’s notice in various ways; news reports, a routine review of the SEC filings, or tips from whistle blowers. Public trigger dates, marked as trigger events in Figure 1 (reproduced from KLM (2008)), are days on which the questionable practice is made public through restatement announcements, auditor departures, delayed SEC filings, management departures, etc. From among these firms, the SEC chooses
some firms for which it conducts an informal investigation. If questionable activity is suspected, this informal investigation may develop into a formal investigation. The SEC does not publicly disclose the list of firms that are under informal or formal investigation. Investigated firms however, may issue a press release announcing that they are under informal or formal investigation by the SEC. When
available, KLM collect those dates of voluntary disclosure. After the investigation, the SEC may drop the case or continue on to the regulation period. A civil, administrative, or criminal proceeding is pursued by the SEC in the regulation period. The beginning and the end of the regulatory period is available for all firms in the KLM dataset.6
To examine how SEC enforcement action impacts peer firms in the industry, we are interested in ascertaining the first date on which the SEC involvement with the target firm becomes public. To this effect, we consider the earliest of the informal investigation date, formal investigation date and beginning of the regulatory period as the beginning date of the SEC’s involvement. Note that the actual SEC’s involvement is likely to predate the date we capture for SEC involvement because (i) the SEC does not disclose the identity of firms that it is informally investigating; (ii) these informal investigation dates are frequently not publicly available; and (iii) even when available, the informal investigation dates are voluntarily disclosed by targeted firms. This data limitation is an issue when studying how the target firms responds to the SEC since its exposure to the SEC starts before this date. However, it is not a concern for the study of how peer firms respond to the SEC’s action against a target firm as peers can only respond to a public disclosure of the SEC involvement. The timeline of an SEC enforcement action, as laid out in KLM (2008), is reproduced in Figure 1 below.
6
If the SEC continues with a civil proceeding, the regulation period is generally initialized by a Wells Notice which documents the intent of the SEC to file charges against the firm. Also note that the dataset consists of all
enforcement actions and not just Accounting, Auditing Enforcement Releases (AAERs). The SEC gives a secondary designation of an AAER to all proceedings that involve an accountant or an auditor.
To reiterate, the date of the SEC investigation used in the analysis is the first known date on which the SEC’s involvement with the target firm is made public. However, it could be argued that the SEC’s investigation may have been anticipated (and hence caused peer firms to change behavior) when the problems at the target firm first surfaced i.e., at the trigger date. Therefore, for robustness, we also use the trigger date as the date of the SEC investigation and rerun our analyses.
3.2 Securities class action data
We use the Stanford Securities Class Action Clearinghouse to identify class action lawsuits filed between 1996 and 2006. We extract the company name, exchange ticker symbol, and lawsuit filing date and match these to firm level data obtained from Compustat. As we are interested in when the peer firm learns about the lawsuit, we identify the start of the lawsuit as the filing date of the lawsuit. Stanford Securities Class Action Clearinghouse also identifies which lawsuits are GAAP and non-GAAP related. Since our main research question addresses the deterrence effects of aggressive accounting behavior, we isolate our class action lawsuit sample to include only those lawsuits that are brought because of GAAP related issues in the financial statements.
We have retained in our sample class action lawsuits related to GAAP violations even if some of these cases were subsequently dismissed. We study deterrence or change in behavior of peer firms at the time the lawsuit is filed. The outcome of litigation (whether it will be subsequently dismissed or not) is not known at the time of the filing. Therefore peer firms’ response is not likely to vary by whether the lawsuit is finally dismissed or not. Nevertheless, to make sure that this does not impact results we have
estimated our model after excluding dismissed cases and continue to find qualitatively similar results. These results are not reported for brevity.
3.3 Timing and construction of our variables
If the filing date of litigation or the SEC investigation date occurs between the first quarter of fiscal year t and the first quarter of fiscal year t+1, we identify year t as the year of the lawsuit and/or SEC enforcement. We include the first quarter of year t+1 because discretionary accruals can still be
manipulated until earnings are released to the public. See Figure 2 for a graphical depiction of how year t is identified.
To identify the effect of SEC enforcement (litigation) on the target firm, we estimate the difference in accruals from before (Pre) to after (Post) the announcement of the enforcement and/or litigation. Similarly, the effect on peer firms is captured by change in accruals from the “Pre” to the “Post” period. The Pre period is defined as t-2, and t-1 where year t is the date of the SEC enforcement and/or litigation. The Post period is defined as year t and year t+1.
Our initial sample consists of all non-financial and non-regulated firms for whom we can compute accruals over the period 1996-2006. Financial firms, defined as firms in the SIC codes 6000-6999, are deleted because interpreting accrual numbers for these firms is difficult. Utilities, falling in SIC codes 4900-4999, are deleted because regulations are likely the first order driver of their accrual
behavior.7 Consistent with Kothari et al. (2005), we delete all firm-years that have fewer than 10 observations for any industry-year because we estimate abnormal accruals for each industry and year. See Table 1 for further details on the sample selection process. After applying the stated data filters, we are left with 64,462 firm-year observations with 5,860 firms per year.
Table 2 reports the distribution of SEC actions and class action lawsuits for every year in the sample. Several observations are worth noting. The SEC investigates an average of only 43 firms every year over the 1996-2006 time period. SEC activity peaks in 2002 (67 actions) presumably in response to the increase in accounting scandals such as Enron and Worldcom and following the bursting of the tech bubble in 2000. On average, the SEC investigates only 0.74% of firms in the sample (43 targets/5,860 firms in the sample). This limited number of enforcement actions points to resource constraints and increases the need to target the miscreants carefully so as to maximize the deterrence value of such investigations on peer firms.
Given how closely public and private enforcement are intertwined in the U.S., it would be useful to understand the proportion of these SEC investigations that are accompanied by a lawsuit. Columns (2) and (3) report this decomposition. On average, 60% of the SEC investigations attract class action
lawsuits (column 4). Column (5) reports the number of lawsuits not accompanied by SEC investigations during our sample period. As is obvious, class action lawyers pursue many more miscreant firms for GAAP related misconduct relative to the SEC (an average of 75 firms per year). For the average firm in the sample that was subject to both SEC investigations and a class action lawsuit, lawsuits precede SEC enforcements by 297 days.
7
We delete 107 lawsuits and 49 SEC investigations that belong to SIC codes 6000-6999. We delete 32 lawsuits and 15 SEC investigations that belong to SIC codes 4900-4999.
There are several firms that are subject to securities litigation and not subject to SEC enforcement. This is because the SEC is resource constrained and cannot investigate all firms.
Somewhat surprising is the finding that about 40% of all firms subject to SEC enforcement actions are not sued. To understand better why these firms are not subject to private litigation, Table 3 displays
descriptive statistics for these firms in the year prior to the lawsuit or SEC investigation and how they differ from other firms subject to both SEC enforcement and litigation. There are some significant differences in firm characteristics.
Firms that are subject to only SEC enforcement do not experience a large drop in stock price on the public announcement of the SEC investigation relative to those that are also sued. In particular, the average stock price decline for firms that were subject to only SEC enforcement is 2.9% relative to -11.5% for those that are subject to both. This difference is significant at the 1% level. Interestingly, firms subject to SEC investigations only have the lowest Altman Z-scores relative to firms (i) subject to both SEC and class actions; and (ii) subject only to litigation, and these differences are statistically significant. Hence, it appears that lawyers avoid firms heading into insolvency although the SEC does not. This is not surprising as the lawyers are interested in collecting damages from the target firm but the SEC’s objective is not solely the collection of fines.
On a related point, firms that are subject to litigation hold somewhat higher holdings of cash. Because stock price declines are related to the damages claimed under litigation and firm’s cash holdings are indicative of their ability to pay damages, this data suggest that firms subject to only SEC actions are unattractive targets for opportunistic lawyers that initiate private litigation. Firms that are not sued also have significantly lower sales growth and a higher fraction of tangible assets when compared to target firms only subject to litigation and target firms subject to both an SEC investigation and litigation.
These differences in firm characteristics, suggesting that some firms are more attractive targets for private litigation, are reinforced when we examine firms that are subject to only litigation. These target firms, the largest group, have significantly higher cash holdings, and higher sales growth than those that are not litigated against. The announcement period stock price decline for firms that are only
litigated is less negative than that for the group that is subject to both SEC and litigation (11.5% vs. -4.8%). This data imply that firms subject to both SEC actions and litigation constitute the most egregious cases of fraudulent activity. The less egregious cases, especially those with higher cash, including possible frivolous ones, are subject to only litigation.
The finding that target firms in the three categories analyzed (subject to only SEC enforcement, subject to both SEC and litigation, and subject to only litigation) are significantly different from one another is not surprising. As mentioned earlier, one downside of private litigation is that it is focused on firms with “provable” damages and ability to pay. Firms pursued by the SEC and not by lawyers either (i) do not have the financial resources to pay damages or are closer to bankruptcy; or (ii) are chosen by the SEC not for their egregious misconduct but rather for political or other reasons.
3.5 Discretionary accruals estimation
We measure abnormal accruals using the modified Jones model controlling for performance and is estimated by two-digit SIC code and year, as specified in Kothari, Leone, and Wasley (2005). In particular, we estimate normal accruals as a function of the change in cash sales (control for operating activities assuming cash sales are non-discretionary), level of property, plant, and equipment (control for nondiscretionary depreciation expense), and ROA (control for operating performance). That is, we first estimate equation (1) specified as:
Tot_Acc i,t = α + β1 (Chg_Sales i,t - Chg_A/R i,t) + β2Gross_PPE i,t + β3ROA i,t + εi,t (1)
where Tot_Acc is accruals for firm i at year t using the statement of cash flows approach as detailed in Hribar and Collins (2002), Chg_Sales is change in sales, Chg_A/R is the change in accounts receivable, Gross_PPE is gross property, plant and equipment and ROA is return on assets. Each of the variables included in (1) is scaled by total assets at the end of year t-1. One appealing feature of specifying accruals as a proportion of total assets at the end of the previous year is that we can identify the extent to which ROA for a firm (which is the sum of CFO and Tot_Acc) was inflated via aggressive accrual
activity. The portion of total accruals unexplained by normal operating activities and operating performance is discretionary accruals i.e., the error term from the above regression.
3.6 Empirical specification
As discussed earlier, we examine the change in discretionary accruals after the SEC / litigation for target and peer firms. Consequently, we estimate the level of discretionary accruals before (denoted by “Pre”) and after (denoted by “Post”) enforcement and litigation. This is done separately for three different groups: 1) firms subject to only SEC action (Only_SEC); 2) firms subject to both SEC action and litigation (SEC_with_Lit); and 3) firms subject to only litigation (Only_Lit). Estimating changes in accruals separately for these three groups will allow us to shed some light on the role of the SEC vs. private litigation in deterring peer firms’ aggressive reporting practices. The estimated regression is as follows:
εi,t = α + β1 Pre_Only_SECi + β2 Post_Only_SECi + β3 Pre_ SEC_with_Liti + β4 Post_ SEC_with_Liti + β5 Pre_Only_Liti + β6 Post_Only_Liti + β7 Pre_Peer_Only_SECi + β8 Post_Peer_Only_SECi + β9 Pre_Peer_SEC_with_Liti + β10 Post_Peer_SEC_with_Liti +
β11 Pre_Peer_Only_Liti + β12 Post_Peer_Only_Liti +
δ
Controls i,t + year dummies + u i,t (2)In equation (2), εi,t is the estimated discretionary accrual, from Equation 1, for firm i during year t and the indicator variables prefixed “Pre_” (“Post_”) are set to one in year t-1 and t-2 (t and t+1) for the target firm relative to the date of the SEC investigation or lawsuit filing date. As the objective of the paper is to evaluate the impact of enforcement activity on peer firms, we set indicator variables prefixed with “Peer_” to one in year t for all peer firms where the target firm is either subject to an SEC
investigation, lawsuit, or both. Peer firms are defined as firms not directly subject to the SEC investigation or lawsuit that share the same four-digit SIC code and year as the target firm. The
“Pre_Peer_” (“Post_Peer”) variables are set equal to one in year t-1 and t-2 (t and t+1) for all peer firms relative to the SEC investigation date or the lawsuit filing date.
We also include several control variables to explain cross-sectional variation in discretionary accruals. These control variables are (i) Log Size , the natural log of the firm’s market value, is included to control for political visibility (Watts and Zimmerman 1986); (ii) Log Sales Growth, the log of sales in current year divided by sales in the prior year, and Bk/Mkt, the firm’s book value of equity divided by the firm’s market value of equity, both of which are included to control for cross-sectional variation in growth (Dechow, Kothari and Watts 1998, Desai et al. 2004); (iii) Lt Debt, long-term debt scaled by lagged total assets, is included to control for differences in leverage; (iv) EP Ratio, earnings per share divided by share price, is included as a control for profitability and growth prospects; and (v) CFO, defined as operating cash flow scaled by lagged total assets, is included as an additional control for operating performance (Kasznik 1999). We also include two industry level variables, the average industry sales growth (Ind Sales Growth) and the average industry book to market (Ind Bk/Mkt), to control for potential industry trends in accruals. Standard errors are clustered by firm to address serial correlation in error terms. Year dummies are added to control for any systematic yearly variation in discretionary accruals and correct for cross-sectional correlation. Table 4 reports summary statistics on each of the variables listed above.
4: Results: Base Model of Deterrence
The base results are reported in Table 5, Model 1. The control variables are mostly statistically significant with the appropriate sign. For instance, discretionary accruals are higher for larger firms (p-value = <.01), firms with greater sales growth (p-(p-value = <.01), low book-to-market firms (p-(p-value = <.01), higher earnings-to-price firms (p-value = <.01) and more levered firms (p-value = <.01). As expected, contemporaneous cash flows are strongly negatively related to discretionary accruals (p-value = <.01). The average industry sales growth is negatively associated with discretionary accruals (p-value = <.01). The average industry book-to-market ratio is positively associated with discretionary accruals (p-value = <.01). The adjusted r-squared from the regression is a reasonable 45%.
We first begin with target firms that are subject to litigation. These firms, irrespective of whether litigation is accompanied with SEC enforcement, have significant negative accruals in the years after being targeted. Somewhat different results are seen for targets that are subject to only SEC enforcement. The estimated coefficient for these firms after being targeted (Post_Only_SEC) is not significantly different from zero.8
Though a significantly negative coefficient in the “Post” period suggests that discretionary accruals were negative after enforcement/ litigation, it does not necessarily capture the change in
discretionary accruals. Consider a firm that manages earnings upward in the pre-enforcement period but either ceases to manage earnings upward or reduces its upward earnings management in the
post-enforcement period. In such a scenario, we would not observe a significant negative coefficient in the Post period but the change from the Pre to Post period will be negative, consistent with our hypothesis. Therefore, we test whether the coefficient on the “Pre_” variable is greater than the coefficient on the “Post_” variable.9
Tests reporting the statistical significance of the change in coefficients on “Pre_” and “Post_” are reported at the bottom of the table. We find that though the discretionary accruals for target firms subject to SEC enforcement alone in the post period were not statistically significant, they are
significantly lower than the discretionary accruals in the pre period. In particular, the difference in accruals between the pre and post period (Pre_Only_SEC and Post_Only_SEC), interpreted as an accrual reversal, for target firms is 1.42% of total assets (p-value of the difference in the coefficients = 0.0749). Similar significant reversals in accruals for target firms are also seen in the other two categories, i.e.,
8
One potential reason for the insignificant coefficient could relate to the confidential nature of the SEC
investigation. We can only observe the publicly known date of the SEC action, which is likely to be much later than the beginning of the SEC’s involvement with the target firm. Thus, the absence of change in the target firm’s behavior after such a pubic announcement is therefore not surprising. However, peer firms’ behavior should only change around the publicly known date of the SEC’s involvement with the target. Our inability to time the correct date of SEC involvement should not impact the results for peer firms.
9
This approach evaluates whether the average discretionary accrual level in the pre-enforcement period is higher than the average discretionary accrual level in the post-enforcement period and does not assume that the level of discretionary accruals, absent manipulation, is zero. Hence, if discretionary accruals are biased even after the inclusion of the control variables, as long as the bias does not differentially change between the pre-enforcement and the post-enforcement period for the firm, we can conclude with some assurance that the SEC enforcement affected the target and the peer firms’ accrual behavior.
when targets are subject to SEC enforcement and litigation or to litigation alone. In particular, the reduction ranges from -1.86% of total assets for targets hit by both the SEC and the lawyers (difference between Pre_SEC_with_Lit and Post_SEC_with_Lit, p-value of the difference in the coefficients = 0.023) to -1.5% of total assets for targets affected by litigation alone (difference between Pre_Only_Lit and Post_Only_Lit, p-value of the difference in the coefficients = 0.0009). These are large magnitudes considering that the median ROA for a firm in the sample is 2.54% as per Table 4. However, the size of the reversal of accruals should not be surprising considering that the analysis thus far speaks only to the target firms. These magnitudes are however reassuring as they imply that our discretionary accruals model to capture earnings management has the ability to pick up the expected changes in reporting practices at target firms. We now turn to the impact of enforcement against targets on peer firms.
Peer firms in the industry of targets firms for all the three categories of enforcement experience declines in accruals. In particular, a comparison of the coefficients on (i) Pre_Peer_Only_SEC with Post_Peer_Only_SEC suggests an average accrual reversal per peer firm of 0.45% of total assets (p-value = 0.0052); (ii) Pre_Peer_SEC_with_Lit relative to Post_Peer_SEC_with_Lit reveals an average accrual reversal per peer firm of 0.35% of total assets (p-value = 0.0246); and (iii) Pre_Peer_Only_Lit with Post_Peer_Only_Lit points to an average accrual reversal per peer firm of 0.56% of total assets (p-value = <.01). The reduction in discretionary accruals for each category is not statistically different from each other category. As the median ROA in our sample is 2.54%, the documented reversals represent 14% to 22% of return on assets per peer firm. In summary, we find significant reduction in discretionary accruals for target, as well as, peer firms after SEC enforcement and/or litigation.
As discussed earlier in the paper, we have used the earliest of the informal SEC investigation, formal investigation and the beginning of the regulatory period as the date of the SEC investigation in model 1. Though this date is the first known instance of when the SEC action is made public, it is plausible that peer firms anticipate the SEC enforcement action before its public disclosure. In other words, it is possible that on the trigger date – the date when the accounting problems at the target firm became public knowledge – there was anticipation among the peers that the SEC will investigate the
target. In such an event, the peer firms may change their behavior around the trigger date and not wait till the SEC publicly announces an investigation against the target. Therefore, model 2 uses the trigger date as the date of SEC enforcement action. As noted earlier, there are many firms that announce accounting problems but only a very small fraction of these are subject to SEC enforcement. Therefore, using the trigger date to time the SEC action may correctly capture the investigation date for a very small fraction of targets where SEC involvement was a certainty but is likely to create noise for many other targets where potential SEC action is a small possibility.
The results, reported in model 2 of Table 5, point to significant deterrence for all categories of targets; although the effect is much weaker for the targets that are subject to only SEC enforcement. The estimate for the reduction in peer accruals is -0.29%, instead of -0.45% in model 1. The p-values are 0.083 instead of being less than 0.01 in model 1. This fact pattern suggests that it is difficult to predict SEC enforcement actions and use of trigger date likely creates noise for the majority of targets subject only to SEC enforcement. Note that this use of the trigger date to time SEC enforcement activity does not impact the overlapped category significantly, i.e., when targets subject to both an SEC investigation and lawsuit. This is because we use the early of the lawsuit filing data or SEC enforcement date to time these events and the lawsuit is usually filed within days of the trigger date. Therefore it is not surprising that use of trigger date to time SEC enforcement has little material impact for this category.
Finally, we evaluate the robustness of our results by using another definition of industry to identify peer firms. In particular, we use the 5 digit NAICS codes instead of the four digit SIC codes to identify the peer firms. As seen in model 3, we continue to find significant evidence of decrease in accruals for peer firms in all three categories of enforcement. In fact, the estimates of the drop in accruals for peer firms are higher – ranging from 0.85% to 1.09% of total assets with accompanying p-values of less than 1%. These results clearly suggest that both class action litigation, as well as, SEC enforcement are associated with significant deterrence in peer firms and that this result is robust to a different
5. Cross Sectional Variations in Deterrence Effects
General deterrence is achieved when peer firms take notice of target firms in their industry that are subject to enforcement and litigation. An attendant increase in information about enforcement and the associated increase in their probability of being subject to similar action might lead peer firms to reduce earnings management. It seems intuitive to conjecture that target firms that are more visible, more likely to be known by others in the industry and/or more likely to be mimicked by others are likely to be associated with increased deterrence. In this section, we develop several proxies for greater visibility of target firms and examine whether these target characteristics result in higher deterrence. In particular, we use the target size, the target firm’s rate of growth, and the target firm’s market share in the industry to proxy for the impact of visible targets on their peer firms.
Along with target characteristics, characteristics of the target industry may also influence deterrence. We discuss these in Section 5.4. For parsimony, we only tabulate the impact of target
characteristics using the four digit SIC as the definition of industry. We have replicated these results with five digit NAICS codes. Moreover, in this section, we designate the earliest of the three dates (informal investigation, formal investigation and the beginning of the regulation period) as the date of the SEC enforcement action.
5.1 Deterrence associated with larger firms
As larger target firms are more likely to be visible, we expect these larger targets to be associated with greater deterrence. We use the market value of equity to proxy for target size. To capture greater deterrence, we create a High Deterrence dummy that takes the value one when the target firm is in the top market value quintile in its four digit SIC for that year. To estimate the greater deterrence associated with this group, we include an interaction term of the High Deterrence dummy with Pre and Post dummies in model (2). The results of estimating this revised model are reported under Model 1 in Table 6. As before, the difference in accruals for the High Deterrence firms before and after the enforcement activity and the corresponding F-statistics for such differences can be found at the bottom of Table 6.
We first discuss the results for targets that are subject to both SEC enforcement and litigation. In this overlapped category, potentially representing the egregious or obvious cases of misreporting, size of the target is important for deterrence. There is a significant reduction in accruals for peer firms only when the target is large, as evidenced by incremental accrual reversals of 0.84% of total assets for the average peer firm (p-value = <.01). When the target is not large there is no significant change in behavior of peer firms. This is because the difference in the coefficients on Post_Peer_SEC_with_Lit and
Pre_Peer_SEC_with_Lit is not statistically significant (p-value = 0.859).
However, the importance of size changes for other categories. When we consider targets that were sued but not subject to SEC action (potentially those with less egregious problems), we find that size of the target has no incremental effect. In other words, all targets, irrespective of size, are associated with a reduction in accruals and there is no incremental reduction in accruals when the targets are large.
Finally, we obtain somewhat counter intuitive results for targets that are subject to SEC
enforcement alone. We find that the incremental effect of large targets is positive. In other words, peer firms of large targets are associated with an increase (rather than decrease) of accruals of 1.14% of total assets (p-value = 0.0002). One potential reason for this anomalous result could be that the SEC’s objective in pursuing these firms may be political. This inference is reinforced by the observation that these visible targets investigated by the SEC somehow did not attract private litigation. Consequently, peers of these targets are likely to perceive these enforcements as idiosyncratic rather than as indicative of the SEC’s interest in their industry. Such perceptions are likely to cause little deterrence. In short, targeting larger firms does not appear to be associated with higher deterrence across all categories of targets.
5.2 Growing firms
We consider the impact of targeting growing firms on the reporting behavior of peer firms. Firms with high growth rates, even if they are not large, likely garner attention from other firms in their
is in the bottom book-to-market quintile in its 4-digit SIC code for the year. The results are reported in Table 6, Model 2.
The results are similar to those for firm size. In particular, growth matters most when targets are subject to both SEC enforcement and litigation. For these targets, there is evidence of deterrence only if the target is a high growth firm, as evidenced by accrual reversals of -0.55% per peer firm (p-value = 0.06). For the other two categories of targets, i.e., those that are subject to only SEC actions or only litigation, although there is significant reduction of accruals for peer firms of all targets, there is no incremental deterrence from high growth targets. This is because the difference in coefficients (i) High * Post_Only_SEC and High * Pre_Only_SEC; and (ii) High * Post_Only_Lit and High * Pre_Only_Lit is not statistically significant.
5.3 Firms with greater market share
Next, we examine whether targets with greater market share are more likely to get noticed by peer firms and thus create incentives for peers to become less aggressive in their financial reporting. To capture this effect, the high deterrence dummy takes the value one when the target firm is in the top sales ratio quintile at the 4-digit SIC level for the year. We define sales ratio to be the firm’s sales scaled by the industry’s total sales. The results are displayed as Model 3 of Table 6.
Targeting firms with large market share appears to have stronger incremental effects. Not only for targets that are subject to both SEC enforcement and litigation, but also for targets that are subject to only litigation there is incrementally significant deterrence from targets with high market share.
However, like firm size, market share is not associated with significant deterrence for targets that are subject to only SEC enforcement.
In summary, proxies for target visibility appear to be associated with greater deterrence though the evidence is mixed. In particular, target visibility appears to play a role only when targets are subject to both SEC and litigation. If being subject to both SEC and litigation proxies for more egregious problems or at least cases that are widely perceived to be problematic, then the visibility of the target
creates deterrence. For targets subject to only SEC (potentially due to political pressures) or for targets subject to only litigation (potentially frivolous), there is overall deterrence but there is no evidence that the visibility of the target has any incremental deterrence effect.
5.4 Target industry characteristics
We begin by studying the impact of targeting firms in competitive industries. Competitive industries have smaller profit margins, and peer firms may be more aggressive in the pre-enforcement period to compete with the target firm’s inflated or fraudulent financial statements. Competitive pressures may also make peer firms more keenly aware of practices and fortunes of other firms in the industry. Such awareness is likely to increase the deterrent effect of SEC enforcement and litigation in competitive industries. We proxy for industry competition using the Herfindahl index, calculated by summing the squared sales ratio for each firm in the 4-digit SIC code for the year. A lower Herfindahl index suggests greater competition and therefore the High deterrence dummy takes the value one when the target belongs to an industry in the lowest quintile of Herfindahl index for the year.
The results displayed under Model 4 of Table 6 suggest a significant impact of competition in SEC enforcement actions. All SEC enforcement actions, whether or not accompanied by litigation, are associated with significant incremental deterrent effects when the target firm is in a competitive industry. In particular, the average peer firm in a competitive industry where the target firm is subject to SEC action alone reverses incrementally more accruals equivalent to 1.49% of total assets (p-value = <.01). The analogous incremental reversal of accruals for the average peer firm is 1.15% of total assets (p-value = <.01) when targets (i) are subject to both SEC action and litigation; and (ii) belong to a competitive industry. For targets that are subject to only litigation, there is overall evidence of significant deterrence, like before, but no incremental deterrence from competitive industries.
We also examine industries that are highly populated, i.e., have a large number of firms by two digit SIC code and year. We expect that a higher number of firms in each industry increases the probability of further enforcement action in the industry, resulting in greater deterrence effects. To
capture this, the High deterrence dummy takes the value one if the number of firms in the four digit SIC is in the top quintile of all firms for that year. The results reported in Table 6, Model 5 show that
incremental effect of populated industry is not uniform across all categories. In particular, the
incremental effect is statistically significant for litigation (both with and without SEC) but not for targets in the SEC only category. The average peer firm experiences an accrual reversal of 1.44% (3.34%) of assets when the target is (i) subject to both SEC action and a lawsuit (just a lawsuit); and (ii) if the target is in a crowded industry.
In summary, industry characteristics of the target have an incrementally larger impact on deterrence than target characteristics. More specifically, targets of SEC enforcements in competitive industries have incrementally higher deterrence and targets of litigation in populated industries are associated with incrementally higher deterrence.
6. Additional analyses
6.1 How much enforcement is enough?
In this section we examine whether the frequency with which the SEC targets an industry is associated with differential levels of deterrence. The objective is to complement the analysis in the previous sections that attempts to identify whether the characteristics of the target or its industry is associated with effective deterrence. Because deterrence is achieved when peer firms become aware of the SEC enforcement, one potential policing strategy is to target every industry and hence spread enforcement efforts widely to achieve maximum deterrence. However, if peer firms are aware that the SEC will try to widen its enforcement across all industries, then an enforcement action in their industry may signal that no one else will be investigated and therefore completely eliminate deterrence. Moreover, repeated SEC enforcements may be required for firms to take notice and to change their behavior.
To examine this issue, we identify New Enforcement actions in the industry as targets that are the first in their 4 digit SIC to be subject to SEC actions in the past three years. Similarly, we define a new litigation event as a target that belongs to an industry that has not experienced a lawsuit or investigation in
the prior three years.10 As we need three years of enforcement and litigation data to construct this New variable, we can only use data from years 1999 to 2006 for this analysis.
Similar to the prior cross-sectional tests in the previous section, we interact the New variable with each peer enforcement variable to ascertain the incremental effect of enforcement activities in industries that are newly targeted by either the SEC or by litigation. Finding results suggesting that new SEC enforcement actions or lawsuits are associated with increased deterrence would imply that the SEC may be more successful deterring aggressive accounting behavior if it were to spread its efforts more widely across industries.
In Model 1 of Table 7, we begin our analysis by estimating our base model, similar to that of Table 5, in this period from 1999 to 2006. Consistent with Tables 5, we continue to find a significant reduction in accruals for peer firms of targets in all categories. This is reassuring as it implies that our main results are not specific to the time period from 1996 to 2006.
Model 2 of Table 7 reports the results with the New variable interacted with each of the peer enforcement variables. The evidence is quite clear - these new targets are not associated with incremental deterrence for either litigation and/or SEC enforcement. To assess the total impact of New targets – we add the incremental effect of New to the main effect – and find that there is no significant change in accruals of peer firms of New targets of SEC enforcements and/or litigation. The p-value for the decline in total (as opposed to incremental) accruals for peer of New targets of SEC enforcement alone is 0.2427. The p-value for the change in accruals for new targets of SEC enforcement and litigation is 0.6412 and for that of peer of new targets of only litigation is 0.8878. This fact pattern suggests that the first targets in the industry are not associated with any significant deterrence. The results imply that repeated and sustained presence of the SEC and private litigation in the industry is more likely to make peer firms take notice and change their behavior.
10
6.2. SEC enforcement actions from 1976 to 2006
In the previous sections, we study SEC enforcements over the period 1996 to 2006 due the limitation that class action litigation data is available over this time period. However, KLM have collected data on SEC enforcement actions from 1976 to 2006. In this section, we examine the entire time series of SEC enforcements to ascertain whether our results hold in the prior two decades. Though this extended time period sheds considerable light on the role and effectiveness of the SEC, it suffers from two limitations. First, there is lack of data on securities class action litigation over this entire period. Therefore we cannot study the role of litigation on deterrence. Though we do not have data on firms that were subject to litigation, KLM have collected data on which of their SEC enforcement actions were also subject to litigation. Consequently, we can control for the differences between SEC enforcement by itself and that accompanied by litigation.
Secondly, operating cash flow data as per FAS 95 is available only since 1988. This matters because Hribar and Collins (2002) show that inferring accruals by subtracting operating cash flows as reported under FAS 95 from net income is associated with less measurement error as opposed to inferring accruals from changes in the balances of working capital accounts as per the balance sheet. With the longer data we are forced to use balance sheet data to estimate our model of discretionary accruals for periods before 1988.
The results for this extended estimation are displayed in Table 8. As can be seen in Model 1, SEC enforcement action is associated with a significant reduction of discretionary accruals for both the target firm, as well as, peer firms. In Model 2, we separately estimate the effect of SEC enforcement by itself and when accompanied by litigation. Consistent with results presented earlier, there is significant evidence of a reduction in accruals for peer firms in both categories. The reduction in accruals for peer firms is to the tune of -0.67% of total assets for the average peer firm when targets are subject to both SEC and litigation and is -0.44% of total assets for the average peer firm when targets are subject to only SEC enforcement both with p-values of less than 1%. In short, over this extended time period that spans
three decades, there continues to be significant evidence that SEC enforcement actions are associated with significant reduction in the discretionary accruals of peer firms.
The long time series of SEC enforcement actions facilitates a more robust study of SEC
deterrence and also allows us to study the role of SEC commissioner and whether the political affiliation of the commissioner affects target and peer firm behavior. If the general belief is that Republican presidents are “good for business,” then it might be the case that SEC enforcements in Republican
regimes are associated with lower deterrence. Hong and Kostovetsky (2010) find that Democratic mutual fund managers hold less of their portfolios in industries deemed socially irresponsible. This also implies that if Democrats are more “socially responsible,” Democratic regimes may be associated with tougher regulation and hence more deterrence.
As the US president designates one of the SEC commissioners as the Chairman, we use the political party of the US president to proxy for the political leaning of the SEC.11 We include a
Republican dummy that takes the value one if the US president at the time of the SEC enforcement action is a Republican. Analogous to the structure of the tests before, we interact the Republican indicator variable with the various deterrence variables to capture any incremental effects from political affiliation.
The results for this estimation are displayed in Table 8, Models 3 and 4. The overall results are similar to Model 1 and 2. Hence, we concentrate only on the incremental effects that arise from the SEC commissioner being Republican. We find no significant incremental effect on peer firms in a Republican presidential regime. In other words, the reduction in accruals of peer firms in not influenced by the political party in control of the White House. In contrast, there is a significant difference in the reporting behavior of the targeted firm. There is a significantly lower reversal of accruals for target firms when the SEC commissioner is Republican (-1.86%) relative to the average reversal when the SEC commissioner is
11
The SEC consists of five commissioners appointed by the US president with the advice and consent of the senate. Their terms last five years and are staggered to one commissioners term end every year. To ensure that the SEC remains non-partisan, no more than three commissioners may belong to the same political party. The president designates one of the commissioners as the chairman.
Democratic (-5.38%). This is consistent with the belief that target firms perceive the regulatory action to be less severe under a Republican regime.
7. Conclusions
In this paper, we study whether SEC enforcement actions are associated with significant change in behavior of peer firms towards greater compliance. As complete compliance is not feasible, a rational enforcement policy implies enforcement efforts that maximize deterrence. Maximum deterrence is also explicitly mandated in directives from the US Congress as one of the main objectives of the SEC’s enforcement policy. Private securities class action litigation, though it does not aim to explicitly deter others, also has the potential to generate deterrence as such enforcement is more frequent and imposes higher monetary sanctions than the SEC.
We study accrual based earnings management in peers, operationalized as firms in the same industry, as the targeted firm in the aftermath of SEC enforcement and litigation to ascertain the existence and magnitude of deterrence. The results suggest significant reduction in accruals for peer firms of targets that are subject to SEC enforcement and/or litigation. Such reversal of accruals is not only highly statistically significant but also economically important. On average, every peer firm reduces
discretionary accruals to the tune of 14% to 22% of its average ROA. This evidence of significant deterrence is robust to different definition of industry. It is also not isolated to events in a few years. Significant evidence of deterrence is seen in sub-samples and also over the extended time period from 1976 to 2006.
The results also inform target selection criteria that are associated with greater deterrence. We find that proxies for target visibility are associated with incrementally higher deterrence only when targets are subject to both SEC enforcement and litigation. When targets are subject to either SEC enforcement or litigation, proxies for target visibility are not associated with incremental deterrence. In contrast to target characteristics, target industry characteristics have greater incremental impact on deterrence for peer firms. SEC actions in competitive industries and litigation in populated industries are associated