The American Accounting Association is the largest community of accountants in academia.

Founded in 1916, we have a rich and reputable history built on leading-edge research and publications. The diversity of our membership creates a fertile environment for collaboration and innovation. Collectively, we shape the future of accounting through teaching, research and a powerful network, ensuring our position as thought leaders in accounting.

Annual Meeting Presentation

August 2012

Does the Revolving Door Affect the SECs Enforcement Outcomes?

American Accounting Association | Publications

Does the Revolving Door Affect the SECs Enforcement Outcomes?

Ed deHaan Doctoral Student Foster School of Business University of Washington Seattle, WA 98195 Email: edehaan@uw.edu Simi Kedia Professor, Finance and Economics Rutgers Business School 94 Rockefeller Road New Brunswick, NJ Email: skedia@rbsmail.rutgers.edu Kevin Koh Nanyang Business School Nanyang Technological University Singapore Email: akohwl@ntu.edu.sg

Shivaram Rajgopal* Schaefer Chaired Professor of Accounting Goizueta Business School Emory University 1300 Clifton Road NE Atlanta, GA 30322 Email: shivaram.rajgopal@emory.edu July 2012

Abstract:
We provide empirical evidence on the consequences of the revolving door phenomenon at the SEC. If future job opportunities make SEC lawyers exert more enforcement effort to showcase their expertise, then the revolving door phenomenon will promote more aggressive regulatory activity (the human capital hypothesis). In contrast, SEC lawyers can relax enforcement efforts in order to curry favor with prospective employers in the private sector (the rent seeking hypothesis). We collect data on the career paths of 336 SEC lawyers that span 284 SEC enforcement actions against fraudulent financial reporting over the period 1990-2007. We find evidence consistent with the human capital hypothesis. Specifically, the intensity of enforcement efforts, proxied by the fraction of losses collected as damages, the likelihood of criminal proceedings and the likelihood of naming the CEO as a defendant, are higher when the SEC lawyer leaves to join law firms that defend clients charged by the SEC. Our evidence is thus inconsistent with popular concerns that revolving doors undermine the SECs enforcement efforts.

*Corresponding author. We would like to thank Jonathan Karpoff, Scott Lee and Gerald Martin for graciously sharing
their SEC enforcement data. We thank Scott Lee, Zoe-Vonna Palmrose, Srini Sankaraguruswamy, Michael Smallberg of POGO, Jun Yang and seminar participants at ALEA Meetings held at Stanford University, Fordham University, Miami University, National University of Singapore, University of Florida, Texas A&M University and Delaware University for their comments. We are grateful for financial support from the Foster School of Business, Nanyang Business School, Goizueta Business School and the Whitcomb Center at Rutgers Business School.

Does the Revolving Door Affect the SECs Enforcement Outcomes?

At a minimum, the revolving door has undermined the integrity of the SECs oversight on numerous occasions, and the SEC isnt policing as aggressively as it should, said Nick Schwellenbach, POGOs director of investigations, quoted in Hilzenrath (2011).

1.0 Introduction
In this paper, we examine whether revolving doors are associated with compromised regulatory oversight by the SEC. In particular, we investigate whether regulatory enforcement against financial reporting fraud is influenced by the future job prospects of prosecuting SEC lawyers. Revolving doors lead to both the SEC hiring officials from firms that they regulate as well as SEC officials leaving to work for firms that are regulated. 1 The revolving door exists because (i) the SEC needs industry specific expertise to regulate its constituents effectively, and (ii) regulated firms value experience and knowledge of complex regulations to minimize their cost of compliance. However, revolving doors can compromise enforcement outcomes if: (i) prior experience with industry makes SEC personnel unduly sympathetic to industrys interests; or (ii) the prospect of future employment makes SEC personnel go easier on violations by regulated firms. The media, members of Congress, academics, former employees of the SEC and investors have all raised questions about the impact of the revolving door on the SECs efficacy and independence. Indeed, the current SEC chairwoman, Mary Schapiro (2009, page 28) testified that the SEC must seek to avoid conflicts created by employees walking out the door and going to a firm and leaving everybody to wonder whether they showed some favor to that firm during their time at the SEC. A GAO report (2011) contends that even the mere appearance of a conflict of interest could undermine confidence in the enforcement process at the SEC. Despite the inherent importance of the SECs revolving door phenomenon, there is surprisingly little systematic evidence on the

For instance, Peter H. Bresnan, a former Deputy Director in the SECs Division of Enforcement, resigned in December 2007 and joined the law firm of Simpson Thacher & Bartlett LLP. In November 2009, Mr. Bresnan filed a statement advising the SEC that he had been retained to represent a client [name redacted] in connection with SEC v. Bank of America Corp. (09-Civ-6892 (JSR)) (S.D.N.Y.). A reverse example relates to the hiring of Robert S. Khuzami, the SECs director of enforcement, who was previously the general counsel of Deutsche Bank.

topic. Our paper attempts to provide such evidence by examining whether SEC lawyers future career prospects influence their enforcement efforts while at the SEC. Crucial to whether revolving doors enhance or compromise regulatory effort is the reason why the regulator is being hired by industry. If the SEC official is being hired for his knowledge of the complex regulatory environment and technical expertise, he will have an incentive to invest in his human capital at the regulatory agency to increase his future prospects in industry, which, in turn, will make him enforce regulations more aggressively (Che 1995). Moreover, as Salant (1995) suggests, SEC personnel might follow aggressive enforcement practices to signal their competence to their prospective employers. We label these arguments as the human capital hypothesis. In contrast, if the SEC official is being hired for his ability to lobby and influence decision makers at the agency, he is likely to under-emphasize or even compromise enforcement outcomes to curry favor with prospective employers (the rent-seeking hypothesis). In this paper, we discriminate between these hypotheses by investigating whether and how future job opportunities influence SEC lawyers regulatory enforcement outcomes. For each lawyer that prosecuted financial reporting fraud on behalf of the SEC, we hand-collect data on whether he left the SEC and, if so, the name of his future employer.2 The rent-seeking hypothesis implies that lawyers that leave the SEC to work for a private law firm, hereafter referred to as revolvers, will be associated with lenient or lax enforcement while at the SEC. In contrast, the human-capital hypothesis implies that revolvers will be associated with aggressive enforcement while at the SEC. We use four proxies of aggressive enforcement effort. The first is the monetary value of the damages collected by the SEC, as a percentage of the shareholder losses from the misconduct. The second is whether the SEC wins versus settles the case. However, our win versus settle results should be interpreted with caution as 93% of cases are settled and it is unclear whether settling is driven by lax or aggressive enforcement. The third outcome is whether, in addition to prosecuting his own administrative and civil

The revolving door concern covers several professionals such as lawyers, accountants and consultants. We focus on lawyers because case dockets allow us to match lawyers with specific SEC enforcement. Moreover, detailed resumes, necessary to trace career paths, are relatively easy to obtain for lawyers as compared to other professionals.

charges, the SEC lawyer refers the case to the Department of Justice (DOJ) for simultaneous criminal proceedings. The fourth outcome is whether the SEC lawyer pursues individual charges against the CEO of the firm, an outcome considered aggressive because : (i) naming individual officers antagonizes influential people who might hinder the SEC lawyers future employment prospects; and (ii) individuals are likely to defend more rigorously than when only the company is named (Eagelsham 2011). We obtain case docket information for all available SEC civil litigation enforcement actions against financial reporting fraud filed between the years 1990-2007. We then search through Bloomberg Law databases, supplemented with LexisNexis Court Link, to collect data on the names of the SEC lawyers prosecuting each case, the defendant law firms, the parties charged, the monetary damages and the outcome of the case. We rely on the LexisNexis Academic database, the Marindale Companys database, Freedom of Information Act requests, and general web searches to gather data on the age, education, and the identity of the post-SEC employer of each SEC lawyer identified above. Our final sample includes 336 unique lawyers that worked on 284 SEC enforcement cases over the sample period. About 58% (or 196) of the 336 lawyers continue to work for the SEC by the end of the sample period. About 11%, or 37 lawyers, leave the SEC to join employers other than law firms, and the remaining 31% of the lawyers quit to join private law firms (referred to as revolvers). The revolver lawyers potentially work for law firms that represent clients before the SEC and are hence most likely to face conflicts of interest. However, our initial tests provide no evidence that the revolver lawyers enforcement outcomes differ from those of non-revolver SEC lawyers. Note that some revolver lawyers leave the SEC to join firms that frequently represent clients before the SEC, while other revolvers join law firms that specialize in areas other than SEC regulation. The revolver lawyers SEC experience, whether it be expertise in SEC regulation or in lobbying decision-makers, should be more relevant for law firms that specialize in SEC matters and actively practice before the SEC (labeled as SEC specialist firms). We identify each firms level of SEC specialization based on the count of the number of cases the firm defends against the SEC in our sample. Further examination of the data reveals that enforcement outcomes are significantly more aggressive for revolver lawyers that join SEC specialist firms.
3

This effect is also economically significant. Revolver lawyers that join SEC specialist law firms that defend twice against the SEC relative to law firms that defend only once, equivalent to a move from the median to the 75th percentile, are associated with an increase of 15.9% in damages, an increase of 4.2% in the likelihood of criminal proceedings, and an increase of 4.4% in the likelihood of naming the CEO as a defendant. However, we find that revolver lawyers joining SEC specialist firms are less likely to win or more likely to settle the case. As mentioned earlier, winning a case is a somewhat ambiguous measure of aggressive enforcement. Overall, the evidence is consistent with the human capital hypothesis. We next investigate two potential sources of cross-sectional variation in the data. First, the lawyers last year of employment at the SEC is likely to be associated with magnified conflicts of interests. If the lawyer has already developed a reputation for tough enforcement, then such a reputation is potentially less likely to be affected by his behavior in the last year. Hence, we would expect the revolver lawyers last year at the SEC to be associated with weaker enforcement. We find some support for this conjecture. Second, younger SEC lawyers that are investing actively in their careers are potentially more responsive to external job market pressures relative to mature SEC lawyers with established reputations. We do not find significant evidence that younger revolver lawyers who join SEC specialist firms are associated with tougher enforcement outcomes relative to older lawyers. Finally, we investigate whether defendant law firms, that have hired several SEC lawyers in the past and hence boast of large SEC alumni, are able to extract milder enforcement outcomes for the clients they represent against the SEC. We find no evidence of such influence. This result further supports the earlier inference that revolving doors are not associated with lax enforcement outcomes. Our results are robust to three alternate explanations and concerns. First, aggressive enforcement outcomes obtained by revolver lawyers, while at the SEC, might be attributable to higher ability as opposed to their career aspirations. We attempt to control for lawyer ability and experience in our tests by including variables to for whether the lawyer attended an Ivy League law school and the number of enforcement cases he worked at the SEC. Moreover, if aggressive enforcement is attributable to latent ability, then enforcement outcomes should be aggressive for all the best SEC employees that leave for lucrative opportunities outside
4

the SEC. However, as discussed, aggressive enforcement outcomes increase only for lawyers who join defense law firms that specialize in SEC matters, as opposed to any private law firm. However, even if latent ability explains our results, the policy implications of our study remain.3 The existence of revolving doors that allow capable lawyers to work at the SEC before they move to other opportunities is not detrimental to the SECs enforcement efforts. Second, our result that aggressiveness of enforcement outcomes is increasing in the SEC specialization could reflect the strong recruiting networks established by these frequent hirers of SEC personnel. Such recruiting networks, however, do not materially affect our results. The hypotheses still speak to whether these recruiting networks, with potentially greater information, value competence (human capital) or lobbying potential (rent seeking). Third, a skeptic can question whether an SEC lawyer has significant discretion over the penalty structure imposed on the culpable firm. However, if one were to argue that SEC lawyers have little or no influence over enforcement outcomes, then the debate over revolving doors compromising regulatory efforts is moot. This seems unlikely as SEC Chairman Mary Schapiro explicitly expressed concerns about revolving door lawyers during her confirmation hearing in January 2009 (Schapiro 2009). Hence, a maintained assumption in the paper is that SEC lawyers can affect enforcement outcomes. A few caveats are in order. First, our results do not imply that the current implementation of the SECs revolving door policies dominates other alternate policies designed to achieve higher enforcement levels such as, for instance, closing the revolving door and simultaneously increasing compensation to attract the very best lawyers. The results simply suggest that the current implementation of revolving doors is not associated with observable compromise in enforcement effort. Second, our results can only speak to the average enforcement outcomes of revolver lawyers there may indeed exist idiosyncratic cases of revolver lawyers favoring potential future employers. Finally, due to data limitations we are not able to study the reverse revolving door phenomenon i.e., the impact of SEC hiring from industry on its enforcement efforts.
3

Recent papers, for e.g. Baranchuk, MacDonald and Yang (2011) and Edmans, Gaibix and Landier (2009) point out that innate ability and effort are complimentary. It is worth stressing that the point of the paper is to study whether rent seeking compromises enforcement or not. The issue of whether, in the absence of rent seeking, higher enforcement is due to ability or effort has little relevance for the policy debate on revolving doors at the SEC. 5

In sum, our paper is among the first to empirically examine the impact of revolving doors on the SECs regulatory effort. Our evidence that SEC regulatory efforts are not, on average, compromised as a result of lawyers leaving the Commission alleviates widely expressed concerns about the detrimental effect of revolving doors. The remainder of the paper is organized as follows. Section 2 discusses the background and hypotheses. Section 3 describes the data and the empirical specification. Section 4 presents the empirical results. Section 5 reports additional tests and section 6 provides some concluding remarks. 2. Literature Review and Hypotheses 2.1 Literature review and institutional background The revolving door phenomenon is not unique to the SEC. 4 Prior literature, however, primarily examines revolving doors in the context of regulating utilities, broadcasters and the insurance industry. Che (1995) presents a model where regulators, during their tenure with the regulatory agency, can invest either in technical expertise or in lobbying capital via social connections with officials at the regulator. The model predicts that if firms hire former regulators for their technical expertise, then regulators will want to acquire such expertise, which, in turn, will lower the cost of monitoring the firm. Salant (1995) also models the revolving door and shows that regulators may monitor constituents aggressively to signal their superior ability to potential future employers in the industry. Both Che (1995) and Salant (1995) motivate our human capital hypothesis. However, if regulators are hired for their lobbying capital (i.e., influencing regulators both during and after working at the SEC), the revolving door will discourage the acquisition of useful expertise, and will result in laxer regulation (our rent seeking hypothesis). An important empirical question is what drives industry to hire regulators their technical and regulatory expertise, or their lobbying potential? We attempt to answer this question in the context of SEC enforcement against financial reporting fraud. The empirical evidence on revolving doors is sparse. Gormley (1979) studies the impact of prior industry experience and finds that commissioners of the Federal Communication Commission (FCC), who

The poster child for these conflicts of interest, albeit not at the SEC, is Darleen Druyun who was responsible for the Air Forces weapons acquisition program and left the government in January 2003 to join Boeing. It was subsequently revealed that she was negotiating the terms of her Boeing employment while she was handling a proposal to lease refueling aircraft from Boeing a proposal that was more costly than purchasing the aircraft outright (See Revolving Door Working Group 2005). 6

were former broadcasters, vote for rules that favor the broadcasting industry. Cohen (1986) documents that prior industry-experience makes FCC commissioners more supportive of industrys interests. However, Dal Bo (2006) raises concerns that, with both these studies, it is hard to disentangle the effect of the revolving door from the political affiliation of the commissioners. Spiller (1990) posits and finds that regulators who preside over more lenient regulatory periods are more likely to get jobs in industry. In contrast, Glaeser et al. (2000) argues that the career prospects of an enforcement official are strengthened by cultivating a reputation for aggressive enforcement and not by pandering to potential target-employers. Grace and Phillips (2008) consider the pre and post-employment history of state insurance commissioners but find no evidence to suggest that consumers in states with regulated prices paid more for insurance relative to consumers in states where the competitive insurance markets. Finally, Vidal, Draca and Fons-Rosen (2011) find evidence consistent with revolving door lobbyists selling access to powerful politicians. Overall, there is little and/or mixed empirical evidence on the impact of revolving doors on regulatory outcomes. The recent 2011 Project on Government Oversight (POGO) report identifies several high profile cases, including Bear Stearns and the Allen Stanfords Ponzi scheme, in which the revolving door appears to have been responsible for staving off SEC oversight. Members of Congress, academics, former employees of the SEC and investors have raised questions about the impact of the revolving door on the SECs efficacy and independence (e.g., Coates 2000, Freeman 2004, Langevoort 2006, Gadinis 2009, Lewis and Einhorn 2009, Grassley 2011, Perino 2004). Our study is the first to: (i) obtain data on the career choices of SEC enforcement lawyers; and (ii) directly test for the existence and the consequences of the SECs revolving door phenomenon. 2.2 SEC enforcement process Before we discuss the details of our research design and construction of relevant variables, we outline the SEC enforcement process in brief. The SECs inquiry into a culpable firms reporting practices consists of multiple steps. Fraudulent firms may be brought to the SECs notice in various ways, including news reports, a routine review of the SEC filings, or tips from whistle blowers. The SEC conducts informal investigations for a subset of these firms. The informal investigation can develop into a formal investigation
7

if questionable activity is suspected. The SEC does not publicly disclose the names of firms that are under informal or formal investigation. After the investigation, the SEC may drop the case or proceed to the regulation period. If the SEC initiates a regulation against the firm, it can choose to bring an administrative proceeding or a civil litigation, or both. Administrative proceedings are heard by an administrative law judge, who is independent of the SEC and issues a decision that includes recommended sanctions. In contrast, in a civil action, the SEC files a complaint with a U.S. District Court and asks the court for a sanction. The choice of administrative proceedings or civil litigation depends on the type of sanction being sought.5 When the misconduct warrants it, the SEC might bring both types of proceedings. The SEC can also refer the case to the Department of Justice (DOJ) for criminal proceedings, which is usually reserved for cases of severe misconduct. We can observe the names of the lawyers involved only for civil litigation and not for administrative proceedings. Hence, we only examine civil litigation by the SEC. Administrative proceedings, criminal proceedings or both also accompany many of these civil actions. 2.3 Hypothesis development Talented SEC lawyers can potentially earn substantially higher wages in the private sector, especially at a law firm. The SEC seeks to control potential conflict of interests via post-employment restrictions. These restrictions bar former employees from: (i) appearing before the SEC; or (ii) assisting others in appearing before the SEC on matters in which they participated personally and substantially while they were at the Commission. The SEC also requires former employees to file statements (known as CFR Title 17 letters) when they expect to appear before the agency on behalf of outside parties for up to two years after leaving. Since SEC regulations facilitate disclosure but do not discourage moving to a law firm, future labor market prospects are important to many SEC lawyers. This is evidenced by the fact that 42% of lawyers in our sample leave the SEC by the end of our data collection period. Lawyers who intend to leave

For example, the SEC may bar someone from the brokerage industry in an administrative proceeding, but an order barring someone from acting as a corporate officer or director must be obtained in a federal court. It is our understanding that administrative proceedings involve milder sanctions relative to civil litigation. 8

the SEC and maximize their future job opportunities are likely to invest in skills that are valued by the external job market. If future employers value the knowledge of SEC regulations and practices, the SEC lawyer will bolster his human capital in these areas. The human capital hypothesis implies that future job prospects, or the existence of revolving doors, are likely to lead to aggressive enforcement outcomes that are consistent with the SECs objectives (Che 1985). In contrast, if the lawyer is being hired to lobby the SEC he is unlikely to focus on enforcement effort or, worse, will compromise enforcement to curry favor with defendant law firms (i.e., potential future employers). The rent seeking hypothesis implies that the prospect of future job opportunities is likely to be associated with laxer enforcement outcomes. These alternate outcomes are summarized in our first hypothesis: H1: Under the human capital hypothesis, revolving doors are associated with tougher enforcement by SEC lawyers that eventually leave the SEC. In contrast, the rent seeking hypothesis implies that revolving doors are associated with laxer enforcement outcomes. Both the human capital and rent seeking hypotheses imply that the SEC lawyer will choose to invest in activities either expertise in SEC regulation or SEC lobbying potential - that are potentially valued by external employers. If the SEC lawyer joins a law firm that does not practice before the SEC, his expertise in SEC regulation or lobbying is likely to be less relevant to enforcement outcomes for the empirical tests of H1. Therefore, the SEC lawyers effort in building his human capital or lobbying potential should be increasing in the relevance of his SEC experience to his potential employer. This is summarized in our second hypothesis: H2: Tougher or laxer enforcement outcomes under the human capital and the rent seeking hypotheses are increasing in the relevance of the lawyers SEC experience to the future employer. To test H2, we construct a variable SEC_SPECIALIST to capture the number of times the revolvers post-SEC employer appears as a defendant law firm in our sample of SEC civil trials. The revolver lawyers

SEC experience is likely to be more valuable to his future employer if the employer actively participates before the SEC, i.e., if SEC_SPECIALIST is high. H2 predicts that under the human capital (rent seeking) hypothesis, the intensity of enforcement efforts should be increasing (decreasing) in SEC_SPECIALIST. We further develop and test two cross sectional implications of our hypotheses. First, the last year of employment of the lawyer at the SEC is likely to be associated with magnified conflict of interests. Even under the human capital hypothesis, a lawyer, having already acquired a reputation for aggressive enforcement, may in his last year reduce enforcement effort due to lower involvement or explicit quid pro quos. As both hypotheses point towards laxer enforcement prior to leaving the SEC, we study whether the last year at the SEC is associated with reduced enforcement effort. Second, though all SEC lawyers are affected by the external labor market, they are likely to differ in the degree to which they invest in acquiring the skills that maximize their future job opportunities. Specifically, young (mature) SEC lawyers investing actively in their careers will be more (less) responsive to external job markets pressures. This implies that both the hypotheses are likely to be associated with a stronger effect for younger lawyers. 3.0 Data and Empirical Specification 3.1 Data collection process We begin with data on SEC enforcement actions against fraudulent financial reporting graciously provided to us by KLM. The KLM dataset covers 865 cases of regulatory enforcement actions by the SEC over the period 1979 to 2007. We exclude all enforcement actions prior to 1990 as case dockets for these actions are generally unavailable. As discussed earlier, the SEC enforcement actions can involve administrative proceedings, civil litigation or both. As administrative cases do not have case dockets which identify the lawyers involved, we restrict our sample to enforcement actions that were subject to civil litigation. 6 After other required data screens, detailed in Table 1, the final sample consists of 284 cases

As we only examine SEC enforcement actions that involve civil litigation, the sample consists of the more severe violations. This is reflected in the fact that the mean cumulative abnormal return around the revelation of the accounting fraud is -21% for enforcement actions with civil litigation relative to -16% for those that have only administrative proceedings (un-tabulated). Moreover, targets of litigated cases are more likely to delist (36%) relative to those with administrative proceedings (21%). 10

involving 336 lawyers.7 As one case can be worked on by multiple lawyer, the unit of observation for our analysis is the lawyer-case level. We have 666 such lawyer-case observations in our sample. 8 For each enforcement action, we identify the corresponding SEC litigation releases that are available after 1995 from the SECs website at http://www.sec.gov/litigation/litreleases.shtml. For SEC enforcement actions prior to 1995, we rely on the Lexis.com database. We extract the following data items from each litigation release: (i) the case docket identifying information; (ii) the charged parties (e.g., company, CEO, CFO); (iii) the outcome of the case (e.g., win versus settle); and (iv) the monetary damages. We then use the Bloomberg Law database and LexisNexis Court Link to obtain civil case dockets. We collect the following data from the case dockets: (i) the SEC lawyers names and office locations; (ii) the name of the defendant; and (iii) the defense law firm. We obtain data on each SEC lawyers age, education, work history, and post-SEC employer from the following sources: (i) LexisNexis Academic database (source: Martindale-Hubbell(R) Law Directory); (ii) Martindale Company database (http://www.martindale.com/Find-Lawyers-and-Law-Firms.aspx); (iii) a general web search including professional network sites such as LinkedIn; and (iv) CFR Title 17 Letters that we obtained from the SEC invoking the Freedom of Information Act for the years 2004 2010.9 We construct the SEC_SPECIALIST variable by counting the number of times each law firm appears as a defense firm in our sample of SEC litigation. For example, the law firm of Baker and Hostetler appeared before the SEC four times. For the two lawyers that leave the SEC and join Baker and Hostetler, the variable SEC_SPECIALIST takes the value 4. Appendix A illustrates the data collection process for one litigation release against Oliver Transportation Inc and its employees, filed on 17th of December, 1998. The case docket from Bloomberg

Overall, the lack of data availability within cases with civil litigation does not appear to be systematically related to any observable firm characteristics. Un-tabulated analyses suggest no statistically significant differences between civil litigation cases with and without all available data for the following variables: return on assets, book-to-market value, stock beta, trigger date cumulative abnormal returns, and firms failure rates.
8

Discussions with academic fellows who have served at the SEC reveal that though the lead lawyer of the SEC team is usually the most senior lawyer, he does not necessarily have the most decision rights in assessing penalties and enforcement outcomes. Hence, we do not restrict our analysis to lead lawyers alone. All data collection was independently completed by two research assistants. Discrepancies between the coding of outcomes between these two assistants were investigated and reconciled by a third research assistant. 11
9

Law identifies Leonatti and Baker PC as one of the defense firms in the civil case. We also identify four lawyers representing the SEC. One of the SEC lawyers is Mr. Willam R. Baker III, based in Washington DC. The Marindale-Hubble Law directory states that Mr. Baker now works for the law firm of Lathan & Watkins LLP. The directory also identifies that Mr. Baker obtained his J.D. in 1983 from Georgetown University. Additional information from his website profile reveals that Mr. Baker has worked in SEC for 15 years prior to joining Latham & Watkins. Linking back to our database of cases and the defense law firm, we could identify that the Mr. Bakers current employer, Latham & Watkins LLP, has been involved in six cases defending a client in a SEC-related case. However, Mr. Baker is not identified as the lawyer on those cases. 3.2 Measuring enforcement outcomes The first enforcement outcome is the amount of monetary damages collected by the SEC. Larger damages represent tougher enforcement and also enhance the publicity value of enforcement cases and direct public attention to a lawyers prosecutorial efforts. However, Karpoff, Lee and Martin (KLM) (2008a) suggest that higher damages could signal more severe violations rather than aggressive enforcement. To account for this confound, we construct a variable labeled DAMAGES that normalizes the damages collected by an estimate of the loss to shareholders from the violation. The loss to shareholders, a proxy for egregiousness of the violation, is measured as the decline in shareholder value implied by the three-day cumulative abnormal return around the trigger date (TRIGGER_CAR), where the trigger date, provided to us by KLM, is the date when the violation becomes public. Under the human capital (rent seeking) hypothesis, we expect revolver SEC lawyers to be associated with higher (lower) DAMAGES in their cases prosecuted while at the SEC. As seen in Table 2, the average DAMAGES ratio is 0.4% suggesting that the damages collected by the SEC are on average 0.4% of the estimated loss in shareholder value from the violation. There are no monetary damages imposed by the SEC in 199 cases, or 30% of the sample, resulting in a large standard deviation (187.4%).10 The low value of the damages relative to the loss in shareholder value is consistent

Note that the number of available observations for Damages is 624 lawyer-case. Monetary damages are unknown or cannot be found for the remaining 42 cases (666-624 cases). 12

10

with Jackson (2007) who documents that monetary sanctions imposed by the SEC are small relative to that extracted by private litigation. Our second measure of enforcement outcomes is an indicator variable that takes the value of one when the SEC wins the case (WIN). Winning the case is a natural and intuitive proxy for the energy behind enforcement efforts. However, this variable is subject to multiple interpretations. For instance, most cases appear to be settled out of court (93% of cases settled and 7% won). Moreover, even if the case is settled, the SEC lawyers could extract large monetary damages or other tough enforcement actions from the culpable firm. Indeed, the correlation of WIN with damages collected is low at 0.05. Finally, the defendants propensity to settle is potentially an increasing function of the intensity of the prosecuting lawyers efforts, in which case settling could be indicative of more aggressive enforcement. Hence, it is unclear whether settlements capture lax or aggressive enforcement. The third measure is an indicator variable that takes the value of one when the case involves a criminal proceeding (CRIM_CASE). As discussed before, the SEC has discretion in referring a case to the DOJ for initiating criminal proceedings against the firm. 11 We predict that there is a higher likelihood of observing a CRIM_CASE under the human capital hypothesis than under the rent seeking hypothesis. In our data, about 44.7% of the cases are accompanied by criminal proceedings. Finally, we code an indicator variable that is set to one if the case named the CEO as a defendant (CEO_CHARGE). Many of the CEOs and other individuals that are named as defendants are barred from working in corporate America either temporarily or indefinitely. Eviction from industry is a harsh penalty leading to the increased likelihood that the CEO, and therefore his firm, will fight back harder (Eagelsham 2011). Further, Gadinis (2009) argues that corporate liability helps deflect sanctions away from managers and employees, which, in turn, provides judges, juries and regulators with the opportunity to castigate misconduct without sending a real human to jail. Therefore, naming individual officers, especially the

11

It is our understanding that the federal securities laws give the SEC the right to bring civil enforcement actions based on scienter or the intent and or knowledge of wrongdoing. The Department of Justice has the right to file criminal charges based on violations of those same sections if the conduct is also willful. This implies a higher burden of proof in criminal cases. 13

CEO, likely requires a greater burden of proof and risks antagonizing influential people that might impact the revolver lawyers future job opportunities. In our data, 54.2% of the cases name the CEO as a defendant. It is worth stressing that, conceptually, the four enforcement outcomes are not independent of one another. Hence the four measures, when considered together, are likely to present a more complete picture of the enforcement outcomes. Indeed, Table 2, panel B shows that the correlation between the four enforcement outcomes is small and even negative in some cases. 3.3 Univariate Evidence There is significant variation in the nature of post-SEC employment among those lawyers that leave the SEC.12 Of the 188 observations involving revolver lawyers, the median SEC lawyer joins a firm that has defended against the SEC once in our sample (See Table 3, Panel A). The top 25% of the revolver lawyercase observations relate to law firms that have defended at least twice against the SEC over the sample period. We begin by comparing the enforcement outcome variables of revolvers versus non-revolvers. As displayed in Table 3, Panel B, CRIM_CASE and CEO_CHARGE are both higher for the revolver cases. There is no statistical difference between the two groups for DAMAGES and WIN. These univariate differences are amplified when we consider revolver lawyers who join law firms that are more likely to defend against the SEC, i.e., when SEC_SPECIALIST is 2 or more (comparison group 2) and when SEC_SPECIALIST is 4 or more (comparison group 3). In both these subgroups, we find that CRIM_CASE and CEO_CHARGE are higher for the lawyers that join high SEC_SPECIALIST firms relative to lawyers that are still at the SEC or left to join a non-law firm. The differences for DAMAGES and WIN are not significant at conventional levels. Panel B of Table 3 also indicates that the cases involving revolvers look different from those related to career SEC lawyers on dimensions other than the enforcement outcomes. Specifically, revolvers tend to be involved with firms that are smaller, are more likely to delist, and suffer more negative stock price reactions upon the initial fraud announcement. Moreover, revolver lawyers are much more likely to have Ivy League

12

We do not have the data to ascertain whether the separation of the lawyer from the SEC was forced or voluntary. 14

degrees relative to career SEC lawyers. We control for these characteristics in our multivariate analyses to follow. 3.4 Estimated models and associated variables The models we estimate generally contain the same explanatory variables. We use four dependent variables based on the four enforcement outcomes. We use two key treatment variables in our analysis: (i) REVOLVER, an indicator variable that is set to one if the SEC lawyer on the case left the SEC to work at any law firm; and (ii) SEC_SPECIALIST, which is a count variable that equals the number of times the revolver lawyers post-SEC law firm shows up as a defense firm (against the SEC) in our sample. Control variables fall into three categories: (i) the characteristics of the company charged by the SEC; (ii) case characteristics, including those that capture the severity of the violation; and (iii) lawyer characteristics, including those related to his competence. All variables discussed below are further detailed in Appendix B. We control for the following characteristics of the firm targeted by the SEC: (i) a measure of firm size, proxied by the natural log of total assets (ASSETS), (ii) book-to-market (BTM); (iii) operating performance (ROA); (iv) stock beta (BETA); and (v) FAILED_FIRM, which is an indicator variable set to one if the firm delists before the end of regulation period. All estimations include fixed year effects to control for time trends.13 In some specifications, we perform sensitivity tests replacing the year fixed effects with a SOX indicator variable that takes the value one after 2002 and controls for potential changes in the enforcement efforts after the passage of Sarbanes Oxley Act. Because the enforcement outcomes are a function of the case and the severity of the violation, it is also important to control for case characteristics. To control for loss in shareholder value arising from the violation we include the three-day abnormal equity return around the initial revelation of the accounting misconduct (TRIGGER_CAR). We also include the buy and hold market adjusted return for the 11 months

13

In some logit specifications (e.g., the WIN models) there is no variation in the dependent variable in some years. In such cases, we only include dummies for years with variation to prevent losing data under the standard year fixed effects model.

15

ending one month prior to the violation end date (PRE_VIO_END_CAR) to capture any anticipation of the fraud and its severity by the equity market. Next, we include the natural log of the length of the violation period (VIO_LENGTH) as Agrawal and Chadha (2005) have argued that the longer the violation period, the more material the violation. We also control for the natural log of the length of the regulation period (REG_LENGTH). The more egregious and complex the case, the greater is the regulation period. Lastly, we control for the press coverage of the case. Increased media attention captures both the severity of the case as well as political pressure and is likely to be associated with greater enforcement effort. MEDIA_HITS is the number of press articles that mention the SECs litigation release from one month before to one month after the litigation release announcement date.14 We control for lawyer competence in two ways. First, we include an indicator variable if the lawyers went to an Ivy-league Law School, IVY. Our use of IVY is motivated by prior research that has used education background to proxy for ability (see for instance Zawel 2005 and Chidambaram, Kedia and Prabhala 2011).15 Second, we proxy and control for lawyer experience with a count variable of total number of cases worked by the SEC lawyer up to and including the current case (EXPERIENCE). 4.0 Empirical Results: Hypotheses 1 and 2 4.1 Damages We begin with an analysis of DAMAGES, which is monetary damages scaled by shareholder loss from the violation. We cluster all standard errors by case since each observation is measured at the lawyercase level, and there is more than one SEC lawyer per case. We begin by estimating whether REVOLVER lawyers, i.e., those that leave to join another law firm, are associated with higher damages. As seen in column (1) of Panel A of Table 4, the coefficient on REVOLVER is insignificant, which implies that there is no difference in damages collected between revolver lawyers and those that stay at the SEC. In column (2), we refine the specification by replacing REVOLVER with the variable SEC_SPECIALIST. The coefficient

14

The search involved variations of 1) company name 2)SEC and 3) law in LexisNexis Academic database (Source: US Newspaper and wires).
15

Consistent with Zawel (2005) we use an extended list of schools for our Ivy-League dummy. In particular, schools regarded as Ivy League are Harvard, Cornell, Yale, Princeton, Columbia, Brown, Dartmouth, MIT, Stanford, University of Chicago, and University of Pennsylvania. 16

of SEC_SPECIALIST is positive and statistically significant (t-statistic = 2.01). Thus, damages collected are increasing in the likelihood that the lawyers post-SEC employer defends culpable firms against the SEC. This evidence is consistent with aggressive enforcement and the human capital hypothesis. The effect of SEC_SPECIALIST is economically meaningful as well: an increase from 1 to 2, a move from the median to the 75th percentile of the variable, increases DAMAGES by 15.9%.16 Finally, in column (3) of Panel A, we replace year fixed effects with a SOX indicator variable. Though the SOX coefficient is not significant by itself, its introduction reduces the significance on the SEC_SPECIALIST variable. This outcome is attributable to our inability to control for year fixed-effects, given that SOX imperfectly controls for broad time-trends in damages. Not controlling for year fixed effects also substantially reduces the estimated R-squared of the model. The results in Panel A of Table 4 also indicate that damages are higher for larger firms, riskier (higher beta) firms, and firms with longer violation periods. Most of the other variables are insignificant, with the exception of MEDIA_HITS, which is negative and significant in the specification involving SOX and excluding year fixed effects (column 3). This result is counterintuitive as it implies that damages collected as a fraction of losses are lower when there is higher media coverage. However, this finding is likely attributable to the correlation between media coverage and shareholder losses (Pearson coefficient = 0.37, untabulated). Recall that we normalize damages collected by shareholder losses. A positive correlation between media coverage and shareholder losses can explain the negative coefficient on the MEDIA_HITS variable. 4.2 Win or settle? We go on to examine the second enforcement outcome, whether the SEC won the case or settled. The dependent variable, WIN, is an indicator variable that takes the value one when the SEC wins an

The coefficient of SEC SPECIALIST in Model 2 in Table 4 is 0.086. The fitted value of the DAMAGES evaluated at the sample mean of the control variables, a baseline year of 2000, and SEC SPECIALIST of 1 is 0.537. Therefore the increase of 0.086 represents 15.9% of the conditional mean of DAMAGES. Related to the same specification, note that about 30% of the cases report no damages assessed. In un-tabulated work, we also estimated a Tobit model to control for truncation and the results are materially unchanged.

16

17

outcome that happens only in 7% of the cases. The results of estimating a logistic regression are reported in Table 4, Panel B. As before, we find no significant effect of REVOLVER on the propensity to win the case (see column 1). The coefficient of SEC_SPECIALIST in column (2), however, is negative and significant (zstatistic = -1.8). Thus, SEC lawyers that leave to join SEC specialist law firms are more likely to settle rather than win the case. The estimated marginal effect of SEC_SPECIALIST is -0.014 (evaluated at the sample means, a baseline year of 2000 and SEC_SPECIALIST of 1, not tabulated) implying that an increase in SEC_SPECIALIST from 1 to 2 decreases the likelihood of winning by 1.4%. The negative coefficient on SEC_SPECIALIST could be interpreted as prima facie evidence of lenient enforcement and the rent seeking hypothesis. However, four considerations temper that interpretation. First, the economic impact of SEC_SPECIALIST on WIN, being just 1.4%, is fairly small. Second, as the SEC wins only 7.4% of the cases and settlements are the norm, we cannot over-interpret results related to settlements or wins as a proxy for enforcement effort. Third, target firms might have agreed to settle the case in exchange for higher damages or other tougher enforcement outcomes - the correlation of WIN with DAMAGES and other outcomes is low (see Panel B of Table 2). Fourth, quickly settling a case may sometimes be a more efficient outcome for the SEC than a protracted trial. The results in Table 4, Panel B also indicate SEC is more likely to settle with firms that report higher ROA. The only case characteristic that is significant in explaining WIN is length of the regulation period. This is not surprising as settlements likely require less time than taking a case to trial. Neither lawyer EXPERIENCE nor IVY is significant. Lastly, the significantly negative SOX coefficient indicates that more cases are being settled after SOX (column 3). One interpretation is that case workload at the SEC increased manifold post SOX which, in turn, led lawyers to press for more settlements to conserve the scarce enforcement resources of the SEC. As the regulatory environment became tougher after SOX, an increased propensity to settle after SOX also implies that this is not a straight forward measure of aggressive enforcement efforts. 4.3 Criminal proceedings

18

In this section, we examine aggressive enforcement by SEC lawyers that leads to civil litigation accompanied by criminal proceedings. The dependent variable, CRIM_CASE, takes the value of one if the SEC has also pursued criminal proceedings against the target firm. Table 5, Panel A presents the results of the logistic estimation for this case outcome. Consistent with prior results, the coefficient on REVOLVER lawyers is insignificant (see column 1). However, as before, the coefficient on SEC_SPECIALIST is positive and significant (z-statistic = 1.98 in column 2). Thus, the likelihood of criminal charges is increasing when the revolver lawyers join SEC specialist law firms. This effect is also economically important; an increase in SEC_SPECIALIST from 1 to 2 increases the likelihood of a criminal proceeding by 4.2%.17 Including a dummy for SOX (column 3) produce substantially unchanged results. Table 5 also indicates that the likelihood of criminal charges is increasing in the length of the regulation period, MEDIA_HITS, EXPERIENCE and TRIGGER_CAR. The positive coefficient on TRIGGER_CAR implies that a less severe fraud is associated with higher likelihood of criminal proceeding. This counter-intuitive finding related to returns around the fraud announcement arises because TRIGGER_CAR along with MEDIA_HITS and length of the regulation period all capture the severity of the case. As all these variables are highly correlated their effect is jointly determined. 18 As longer regulation period and greater press coverage are both associated with a higher likelihood of criminal proceedings, our overall results suggest that severe cases are more likely to be associated with criminal proceedings. 4.4 CEO named as a defendant Finally, we evaluate whether revolver lawyers are more likely to file charges against the CEO of the target firm (CEO_CHARGE). Consistent with prior results, the coefficient on REVOLVER is not significant and that of SEC_SPECIALIST is positive and significant (z-statistic = 2.36) in column (2). Thus, the data are again consistent with the hypothesis that SEC lawyers who join SEC specialist law firms are more aggressive in filing criminal charges against the CEO. An increase in SEC_SPECIALIST from 1 to 2 increases the

17

The estimated marginal effect for SEC SPECIALIST is 0.042 when evaluated at control averages, a base year of 2000, and for SEC SPECIALIST of 1. These have not been tabulated for brevity.
18

In unreported results, the coefficient of TRIGGER_CAR is not significant when MEDIA and REG_LENGTH are excluded. 19

likelihood of naming the CEO as a defendant by 4.4%. In line with results in the earlier section, the inclusion of SOX does not impact the reported results. Firm characteristics are important in the likelihood of CEO being named. CEOs of (i) smaller firms, (ii) firms with higher valuations (coefficient on book to market is negative); and (iii) more profitable firms (coefficient on ROA is positive) are more likely to be charged. Further tests, not tabulated for brevity, show that the positive relation between propensity to charge the CEO and ROA is found only when ROA is negative, thus shedding some light on this potentially counter-intuitive finding. Moreover, CEOs are more likely to be charged when: (i) the firm attracts higher press coverage; and (ii) the firm has a longer violation period. Once again, there is no evidence to suggest that lawyer education or experience explains whether the CEO is charged or not. Taken together, the collection of results in Tables 4 and 5 provides no evidence that enforcement outcomes of revolver lawyers differ in general from those that stay at the SEC or join employers that are not law firms. However, this evidence changes systematically when we examine the nature of the law firm that the revolvers join. In particular, three of the four enforcement outcomes point towards aggressive enforcement by SEC lawyers that leave and join law firms that frequently defend against the SEC. The exception is that revolver lawyers that join SEC_SPECIALIST firms are more likely to settle (or less likely to win). As previously discussed, the relation between WIN and lawyer effort is somewhat ambiguous. Overall, the results are consistent with the human capital hypothesis. 5.0 Additional Tests 5.1 Last year in office As discussed in Section 2.3, the lawyers last year of employment at the SEC is likely to be associated with magnified conflicts of interest and, therefore, is potentially associated with reduced enforcement efforts under both the human capital and rent seeking hypotheses. To study such horizon effects, a binary variable LAST_YEAR is set to one for cases that conclude within one year of a revolver lawyer leaving the SEC. Sample sizes are reduced as we dont have data on the last year of office for all lawyers. As shown in Table 6, the coefficient of LAST_YEAR is significant and negative for three of our enforcement
20

variables. Revolvers in their last year at the SEC: (i) collect significantly lower damages; (ii) have a lower probability of charging a CEO; and (iii) are associated with a lower likelihood of winning. The coefficient on LAST_YEAR is insignificant when the dependent variable is the likelihood of criminal charges (column 3). In this case, the coefficient of SEC_SPECIALIST is also not significant, implying potential power issues arising from the smaller sample. To better understand the horizon problem, we would ideally like to investigate how the revolver lawyers incentives for enforcement effort changes over his tenure at the SEC. In particular, it would be ideal to compare a given lawyers enforcement outcomes in his early cases versus those immediately prior to leaving the SEC. As seen in Table 7, 60.2% of the revolver lawyers and 57.9% of the non-revolver lawyers participated in only one case in our sample. Only 15 revolver lawyers, related to 54 cases, are involved with at least one case before and one case during their last year with the SEC. As seen in Table 8, such revolver lawyers are less likely to win the case in the last year at the SEC. However, there is no statistically discernible difference in the other enforcement outcomes between the last year and the rest of their tenure at the SEC. In sum, there is weak evidence that the last year at the SEC tends to be associated with more lenient enforcement outcomes. 5.2 Younger lawyers Though all SEC lawyers are likely to be aware of whether the external labor market values competence or lobbying potential, lawyers are likely to invest in these skills to varying degrees depending on how actively they seek to advance their external job opportunities. In particular, young (mature) SEC lawyers are more (less) likely to be responsive to external job market pressures. To examine this issue, we would ideally like to use data on lawyer age; however we were able to locate date of birth for only a small fraction of the lawyers. Consequently, we rely on a proxy for age based on the year the lawyer took the Bar exam. In particular, for revolvers we calculate the difference between the year the lawyer left the SEC and the year he cleared the Bar exam. We create binary variables YOUNG (MATURE) that takes the value 1 if the period between the lawyers Bar exam date and the SEC departure date is less (greater) than the sample median of 17 years. We decompose the SEC_SPECIALIST term into: (i) SEC_SPECIALIST*YOUNG; and (ii)
21

SEC_SPECIALIST*MATURE to examine whether young lawyers are more responsive to the pressures of the external market. Therefore, we do not estimate the main effect related to the SEC_SPECIALIST term. Partial results, i.e., only the coefficients for relevant variables, are displayed in Table 9, Panel A. Sample sizes are again reduced by limited data availability. When compared across the four enforcement outcomes, we find that the coefficient on SEC_SPECIALIST* YOUNG LAWYER is positive and significant for CRIM_CHARGE (z-statistic = 2.09) and CEO_CHARGE (z-statistic = 2.95) while the coefficient on SEC_SPECIALIST*MATURE is not significant, which suggests greater responsiveness from younger lawyers. However, the estimated coefficients on the two interaction terms are not statistically different from each. Nor are the interaction coefficients significantly different from each other for the remaining two outcomes of WIN and DAMAGES. Overall, the evidence is suggestive that younger lawyers are more aggressive though this difference is not statistically significant. 5.3. SEC alumni Lastly, we examine the possibility that the lawyers efforts to build competence (human capital) and influence (rent seeking) are not mutually exclusive. An underlying assumption in our tests thus far is that efforts to boost human capital and rent seeking are distinct from each other. However, it is possible that these efforts are related, in that a competent lawyer who invests in human capital is also influential. If rent seeking is also increasing in competence then, under both hypotheses, we may observe increased enforcement effort by revolvers while at the SEC. The detrimental effect, if any, of revolving doors may only be seen in the exercise of this influence by revolvers after they leave the SEC. Thus, SEC alumni, with influence over friends and colleagues still at the SEC, may impact the agencys ongoing enforcement decisions to benefit clients of their current employer. If so, defendant law firms that employ more SEC alumni should be associated with milder enforcement outcomes. To investigate this conjecture, we construct a variable labeled DEFFIRM_HIRES_AVG to measure the average number of SEC lawyers hired in our sample by all the defendant law firms involved in the case (most cases have

22

multiple defending law firms).19 To ensure that DEFFFIRM_HIRES_AVG captures the effect of SEC alumni and not the number of defense firms we include the log of the number of defense firms as a control. Results of regression models including DEFFIRM_HIRES_AVG are presented in Table 9, panel B. Sample size is again reduced due to limited data availability. The SEC_SPECIALIST coefficient is largely unchanged although the WIN coefficient is no longer significant. DEFFIRM_HIRES_AVG is insignificant in all four specifications, which is inconsistent with enforcement outcomes differing for defendant law firms with a greater presence of SEC alumni.20 These findings are largely unchanged when SEC_SPECIALIST is removed from the models (un-tabulated). Overall, a greater number of SEC alumni do not appear to have observable influence on ongoing enforcement efforts. 5.4 Robustness tests In this section, we perform several robustness tests to check the sensitivity of our results. Several of these tests require data that is unavailable for all lawyers and cases, thereby reducing our sample sizes. Although we incorporate a host of controls for firm characteristics, case characteristics and lawyer characteristics, our previous models do not control for the quality and skill of the defending law firm. We expect the aggressiveness of SEC enforcement outcomes to decrease with the skill and competence of the defending law firm. Following Rider (2011), we proxy for the quality of the defense firms using prestige scores assigned by Vault, with higher scores representing highly regarded law firms.21 Vault prestige scores are based on an annual survey of 16,000 U.S. attorneys. When a case has more than one defense law firm, we compute the average prestige score for the defense firms to capture the overall quality of the defense team (SCORE_DEFENDANT) and also include a control for the natural log of the number of defending firms.

19

The number of SEC lawyers hired by a defense firm is constructed from within our dataset i.e., it is the number of SEC lawyers in our dataset that have joined a particular defense firm. This measure is not able to account for SEC lawyers with no civil enforcement and hence not captured by our data, or those that left the SEC prior to 1990 (the beginning of our dataset) and, hence, likely underestimates the number of lawyers who join private law firms who appear before the SEC.
20

In robustness tests we also create another variable, DEFFIRM_HIRES_MAX which is the maximum number of SEC lawyers hired by a defendant law firm among all the defendant law firms that are involved in the enforcement. The results are qualitatively similar with this variable and have not been reported for brevity. Available at http://www.vault.com/wps/portal/usa/rankings/individual?rankingId1=2&rankingId2=1&rankings=1&regionId=0&rankingYear=2007. 23
21

Sample sizes are reduced as defense firm data is not available for all cases. However, our results for SEC_SPECIALIST are substantially unchanged even in the smaller sample and in the presence of these additional control variables (See Panel C of Table 9). The only result that is impacted is the likelihood of winning, which is now insignificant. We find that the quality of the defense team (SCORE_DEFENDANT) is, by itself, not significant. The number of defense law firms is positively related to winning the case and to the CEO being charged. It is not surprising that the number of defending firms is higher when CEO charges are involved as individual defendants typically engage their own law firms. In our second set of robustness tests, we try to better control for the experience and competence of the SEC lawyers. In our base regressions, we have accounted for the number of cases that lawyers worked on while at the SEC and whether they attended an extended Ivy League. Here, we add an additional control for prior experience based on the number of years elapsed between the current case and when the lawyer passed his bar exam, YEARS_AS_LAWYER. We are able to obtain this data for 637 of the 666 lawyer-case observations. As seen in panel D of Table 9, introduction of this new variable does not materially impact our results. The coefficient of SEC_SPECIALIST continues to be significant in all columns. YEARS_AS_LAWYER is itself significant only in the CRIM_CASE regression (column 3). Finally, we create an alternate proxy for the relevance of the revolver lawyers SEC experience to the hiring law firm. A potential limitation of the SEC_SPECIALIST measure is that it only captures the extent to which the post-SEC employing law firm is involved with civil litigation in our sample of cases. A lawyers SEC experience is also likely to be valuable in other dealings with the SEC, especially in SECs informal investigations and inquiries that are not publicly disclosed. Thus, SEC_SPECIALIST may not adequately capture all potential law firms that practice before the SEC. Our alternate proxy is based on the expectation that companies that are being investigated by the SEC are likely to hire prestigious law firms. Thus, a law firms prestige rank (again from Vault) is a potential indicator of the extent to which the firm values a revolver lawyers SEC experience. We create a variable HIRINGFIRM_RANK that takes a value of zero for unranked law firms, 1 for law firms with a rank from 100 (being the lowest ranked firm) to 21, and a value of 2 for top-20 law firms. The rank of the law
24

firm not only likely captures the market share of the law firm in providing legal services against the SEC, but also the desirability of the law firm as a future employer for the revolver lawyer. As shown in Panel E of Table 9, our results are qualitatively similar when SEC_SPECIALIST is replaced with HIRINGFIRM_RANK, with the exception that we observe no significance in the damages regression. Thus, damages collected by lawyers that leave for the most prestigious law firms are no different from the damages collected by those that stay at the SEC. However, those that leave to join prestigious law firms are still associated with a higher likelihood of: (i) bringing criminal charges against the firm; and (ii) charging the CEO. The finding that SEC lawyers that join more prestigious law firms are associated with tougher (or at least not laxer) enforcement outcomes while at the SEC, again implies that sought after employers value the lawyers SEC regulation expertise (i.e., human capital) over their lobbying potential (i.e., rent seeking). 6. Conclusions Several critics, including members of Congress, have argued that the SECs integrity has been damaged by the revolving door phenomenon whereby former SEC employees seek employment at law firms that defend firms regulated by the SEC. The concern is that a conflict of interest could bias SEC oversight and adversely affect the credibility of the SECs enforcement initiatives. In an effort to bring empirical evidence to bear on these allegations, we investigate the association, if any, between enforcement outcomes and career opportunities of SEC lawyers in cases involving fraudulent financial reporting. We collect data on the career choices of SEC lawyers and outcomes of civil litigation while these lawyers are at the SEC. We find no significant differences in the enforcement outcomes for SEC lawyers that eventually leave the SEC to join law firms relative to those who stay back at the SEC. However, the results change when we consider lawyers that leave to join law firms that specialize in defending clients against the SEC. In particular, lawyers that join SEC specialist law firms are associated with stronger enforcement efforts, as proxied by: (i) higher damages collected; (ii) a higher likelihood of criminal proceedings; and (iii) a higher likelihood of naming the CEO as a defendant. The likelihood of winning the case is lower for lawyers that join such SEC specialist firms, but it is unclear whether settling a case is a result of lax or
25

aggressive enforcement. There is some indication that conflicts of interest arise primarily in the last year at the SEC. Finally, we find no evidence that SEC alumni in defense firms practicing before the SEC are able to exercise influence over ongoing enforcement efforts. These results alleviate concerns expressed by the press, policy makers and Congress that revolving doors are detrimental to the SEC regulatory efforts, at least, on average. If anything, future job prospects make SEC lawyers increase their enforcement efforts while they are at the SEC. The results can also potentially inform discussions on the effect of revolving doors for other regulatory agencies. However, much remains to be done. Further research can fruitfully explore other characteristics of the labor market for regulators and how this impacts regulatory effort.

26

References Agrawal, A., and S. Chadha. 2005. Corporate governance and accounting scandals. Journal of Law & Economics 48 (2):371-406. Baranchuk, N., G. MacDonald and J. Yang, 2011, The economics of super managers. Review of Financial Studies, 24(10), 33213368. Che, Y. K. 1995. Revolving doors and the optimal tolerance for agency collusion. The RAND Journal of Economics:378-397. Chidambaran, N. K., S. Kedia, and N. R. Prabhala. 2011. CEO director connections and corporate fraud. SSRN eLibrary. Coates IV, J. C. 2000. Private vs. political choice of securities regulation: A political cost/benefit analysis. Virginia Journal of International Law 41:531. Cohen, J. E. 1986. The dynamics of the "revolving door" on the FCC. American Journal of Political Science:689-708. Dal B, E. 2006. Regulatory capture: a review. Oxford Review of Economic Policy 22 (2):203. Eagelsham, J. 2011. At SEC, strategy changes course. September 11. Wall Street Journal. Edmans, A., Gabaix,X., and A.Landier. 2009. A calibratable model of optimal CEO incentives in market equilibrium. Review of Financial Studies 22, 4881-4917. Files, R. 2011. SEC enforcement: Does forthright disclosure and cooperation really matter? Journal of Accounting and Economics. Freeman, J. P. 2004. Statement of John P. Freeman Professor of Law, University of South Carolina Law School before the Senate Governmental Affairs Subcommittee on Financial Management, the Budget, and International Security. Gadinis, S. 2009. The SEC and the financial industry: Evidence from enforcement against broker-dealers. SSRN eLibrary. Glaeser, E. L., D. P. Kessler, and A. Morrison Piehl. 2000. What do prosecutors maximize? An analysis of the federalization of drug crimes. American Law and Economics Review 2 (2):259. Gormley Jr, W. T. 1979. A test of the revolving door hypothesis at the FCC. American Journal of Political Science:665-683. Government Accountability Office. 2001. Securities and Exchange Comission: Human capital challenges require management attention (GAO-01-947). Government Accountability Office. 2011. Securities and Exchange Commission: Existing post-employment controls could be further strengthened (GAO-11-654). Grace, M. F., and R. D. Phillips. 2008. Regulator performance, regulatory environment and outcomes: An examination of insurance regulator career incentives on state insurance markets. Journal of Banking & Finance 32 (1):116-133. Grassley, C. 2011. Grassley: Report shows SEC's revolving door must be slowed 2011 [cited October 7 2011]. Available from http://www.grassley.senate.gov/news/Article.cfm?customel_dataPageID_1502=34491. Hilzenrath, D. (2011). SEC Staffs revolving door prompts concerns about agencys independence May 13, 2011. Available at http://www.washingtonpost.com/business/economy/sec-staffs-revolving-door-promptsconcerns-about-agencys-independence/2011/05/12/AF9F0f1G_story.html Jackson, H. E. 2007. Variation in the intensity of financial regulation: Preliminary evidence and potential implications. Yale Journal on Regulation 24:253. Karpoff, J. M., D. S. Lee, and G. S. Martin. 2008a. The cost to firms of cooking the books. Journal of Financial and Quantiative Analysis 43 (4):32. Karpoff, J. M., D. S. Lee, and G. S. Martin. 2008b. The consequences to managers for financial misrepresentation. Journal of Financial Economics 88 (2):193-215. Langevoort, D. C. 2006. The SEC as a lawmaker: Choices about investor protection in the face of uncertainty. Washington University Law Review 84:1591. Lewis, M. E., David. 2009. The end of the financial world as we know it. The New York Times, 1/3/2009. Perino, M. A. 2004. SEC enforcement of attorney up-the-ladder reporting rules: An analysis of institutional constraints, norms and biases. Villanova Law Review 49:851.
27

Project on Government Oversight. 2011. Revolving regulators: SEC faces ethics challenges with revolving door 2011 [cited May 13 2011]. Available from http://www.pogo.org/pogo-files/reports/financialoversight/revolving-regulators/fo-fra-20110513.html. Revolving Door Working Group (2005), A Matter of Trust: How the Revolving Door Undermines Public Confidence in Government - And What To Do About It", available at http://www.cleanupwashington.org/documents/RevovDoor.pdf. Rider, C. 2011. Networks, Hiring and Attainment: Evidence from Law Firm Dissolutions, Working paper, Emory. Salant, D. J. 1995. Behind the revolving door: a new view of public utility regulation. The RAND Journal of Economics:362-377. Schapiro, M. 2009. Hearing before the Senate Subcommittee on Banking, Housing, and Urban Affairs, January 15, 2009. Available from http://www.gpo.gov/fdsys/pkg/CHRG-111shrg50221/pdf/CHRG111shrg50221.pdf Spiller, P. T. 1990. Politicians, interest groups, and regulators: A multiple-principals agency theory of regulation, or "let them be bribed". Journal of Law and Economics 33 (1):65-101. Toonkel, J. 2011. New SEC executive's background could bode well for fund industry 2011 [cited May 11 2011]. Available from http://www.investmentnews.com/article/20110123/REG/301239978. US Senate. 2009. Hearing Before the Committee on Banking, Housing, and Urban Affairs: Nominations of Mary Schapiro, Christinga D. Romer, Austan D Goolsbee, Cecilia E. Rouse, and Daniel K. Tarullo. Vidal, Jordi Blanes, Mirko Draca and Christian Fons-Rosen 2011, Revolving Door Lobbyists, Working paper, London School of Economics. Zawel, M. 2005. Defining the ivy league. Untangling the Ivy League (College Prowler).

28

Appendix A: Example of the Data Collection Process Enforcement Action: Against Oliver Transportation and employees beginning 12/17/1998. Step 1: SEC Litigation Release from SEC Website
SECURITIES AND EXCHANGE COMMISSION LITIGATION Release No. 16003 1998 SEC LEXIS 2724 December 17, 1998 The Securities and Exchange Commission yesterday filed an enforcement action in the Eastern District of Missouri charging eight individuals with perpetrating a financial fraud at Oliver Transportation, Inc. (OTI), a now defunct trucking company formerly headquartered in Mexico, Missouri. The complaint alleges that: From the time OTI went public in 1993 until it ceased operations in August 1995, OTI's senior management and other employees unlawfully inflated OTI's financial results by fabricating phony customer orders, and in turn accounts receivable, for trucking services. As a result of the recording of phony receivables, OTI's financial statements and other disclosures in its June 1993 registration statement and subsequent periodic reports filed with the Commission contained materially false and misleading information. By the time the fraud was uncovered in August 1995, nearly half of OTI's reported receivables were based on phony customer orders. As to the roles played by each of the defendants, the complaint alleges, among other things, that. . John F. "Pete" Oliver (OTI's founder and chairman) devised and initiated the fraud to obtain funds under a bank loan secured by the company's [*2] accounts receivable. In addition, Oliver sold 32,000 shares of OTI common stock, receiving proceeds of $ 129,869, when he knew that OTI's financial statements and periodic reports contained materially false and misleading information. . Willard A. "Tony" Meador (OTI's president until October 1994) directed the entry of the phony customer orders into OTI's books and records. While aware that OTI's financial statements and other disclosures were materially false and misleading, Meador also sold 100,500 OTI shares, receiving proceeds of $ 288,637. . Wayne M. Sampson (the company's initial chief financial officer and later Meador's successor as president), James R. Gehringer (the director of operations), and Julie McNabb-Meador and Patrick M. Jacobi (both billing supervisors), each assisted in the entry and tracking of the phony orders in OTI's books and records. . Steven M. Gross (Sampson's successor as chief financial officer) and Michael W. Roberts (controller), both certified public accountants, knowingly reported the phony accounts receivable in OTI's financial statements included in periodic filings with the Commission. In addition to civil money penalties, the complaint seeks to [*3] permanently enjoin the defendants from violating the antifraud, books and records, and internal accounting control provisions of the federal securities laws. The Commission also requested that the court order Oliver and Meador to disgorge their ill-gotten gains from their insider trading and permanently bar each from serving as an officer or director of any public company. Simultaneously with the filing of the complaint, without admitting or denying the complaint's allegations, Oliver, Sampson, Gehringer, Roberts, and Gross each agreed to settle the charges against them by consenting to final judgments. The final judgments against Oliver and Sampson prohibit each from violating Section 17(a) of the Securities Act of 1933, and Sections 10(b) and 13(b)(5) of the Securities Exchange Act of 1934 and Rules 10b-5, 13b2-1, and 13b2-2 thereunder. The judgment against Oliver also bars him from serving as an officer or director of a public company under Section 21(d)(2) of the Exchange Act. The judgment against Gehringer prohibits him from violating Section 17(a) of the Securities Act and Sections 10(b) and 13(b)(5) of the Exchange Act and Rules 10b-5 and 13b2-1 thereunder. Gross and Roberts [*4] consented to judgments enjoining them from violating Sections 10(b) and 13(b)(5) of the Exchange Act and Rules 10b-5, 13b2-1, and 13b2-2 thereunder. As part of their settlements, Gross and Roberts have agreed to the entry of Commission orders barring each from appearing or 29

practicing before the Commission as accountants. The Commission agreed not to seek imposition of civil money penalties against the settling defendants based on their demonstrated inability to pay. For the same reason, the Commission also agreed to waive the payment of disgorgement by Oliver. The charges filed against Meador, McNabb-Meador, and Jacobi are pending before the court.

Step 2: Docket Information regarding the case from Bloomberg Law

U.S. District Court Eastern District of Missouri (LIVE) (Hannibal) CIVIL DOCKET FOR CASE #: 2:98-cv-00075-DJS SEC v. Oliver, et al Assigned to: Honorable Donald J. Stohr Demand: $0 Cause: 15:77 Securities Fraud Plaintiff Securities and Exchange Commission represented by Carleasa A. Coates SECURITIES AND EXCHANGE COMMISSION 450 Fifth Street, N.W. Washington, DC 20549 202-272-2550 Fax: 202-942-9569 LEAD ATTORNEY ATTORNEY TO BE NOTICED Kathleen M. Hamm SECURITIES AND EXCHANGE COMMISSION 450 Fifth Street, N.W. Washington, DC 20549 202-272-2550 Fax: 202-628-1471 LEAD ATTORNEY ATTORNEY TO BE NOTICED Keith A. O'Donnell SECURITIES AND EXCHANGE COMMISSION 450 Fifth Street, N.W. Washington, DC 20549 202-272-2550 Fax: 202-628-1471 LEAD ATTORNEY ATTORNEY TO BE NOTICED William R. Baker, III SECURITIES AND EXCHANGE COMMISSION 450 Fifth Street, N.W. Washington, DC 20549 202-272-2550 Fax: 202-628-1471 LEAD ATTORNEY ATTORNEY TO BE NOTICED Date Filed: 12/16/1998 Jury Demand: Defendant Nature of Suit: 850 Securities/Commodities Jurisdiction: U.S. Government Plaintiff

30

Defendant John F. Oliver TERMINATED: 12/23/1998 Defendant Willard A. Meador TERMINATED: 07/18/2000 represented by Willard A. Meador 25 Shadescrest Road Birmingham, AL 35226 205-989-1565 PRO SE

Defendant Wayne M. Sampson TERMINATED: 12/23/1998 Defendant James R. Gehringer TERMINATED: 12/23/1998 Defendant Julie McNabb-Meador TERMINATED: 07/18/2000 represented by Louis J. Leonatti LEONATTI AND BAKER P.C. 123 E. Jackson Street P.O. Box 758 Mexico, MO 65265-0758 573-581-2211 Fax: 573-581-6577 Email: lou@leonatti-baker.com TERMINATED: 07/18/2000 LEAD ATTORNEY ATTORNEY TO BE NOTICED Randall P. Baker LEONATTI AND BAKER P.C. 123 E. Jackson Street P.O. Box 758 Mexico, MO 65265-0758 573-581-2211 Fax: 573-581-6577 Email: randy@leonatti-baker.com TERMINATED: 07/18/2000 LEAD ATTORNEY ATTORNEY TO BE NOTICED Defendant Patrick M. Jacobi represented by Ronald E. Jenkins JENKINS AND KLING, P.C. 10 S. Brentwood Boulevard Suite 200 Clayton, MO 63105 314-721-2525 Fax: 314-721-5525 Email: rjenkins@jenkinskling.com LEAD ATTORNEY ATTORNEY TO BE NOTICED

31

Step 3: Information on former SEC Lawyer William Baker from Marindale-Hubble Law Directory
MARTINDALE-HUBBELL (R) LAW DIRECTORY Practice Profiles Section WILLIAM R. BAKER, III Latham & Watkins LLP 555 11th Street, N.W., Suite 1000 Washington, District of Columbia 20004-1304 Telephone: 202-637-2200 Telecopier: 202-637-2201 Email: webmaster@lw.com Url: http://www.lw.com POSITION: Partner PRACTICE-AREAS: SEC Enforcement; Securities Disclosure; Corporate Investigations; Securities Litigation; White Collar Criminal Defense; Corporate Governance. ASSOCIATIONS: American Bar Association ADMITTED: 1984, District of Columbia LAW-SCHOOL: Georgetown University (J.D., 1983) COLLEGE: University of Notre Dame (A.B., 1979) ISLN: 909291912

Step 4: Additional Information on former SEC Lawyer William R Baker from Internet Search:
http://www.lw.com/people/WilliamRBakerIII William R. Baker III is a partner in the Washington, D.C. office of Latham & Watkins. Mr. Baker's practice includes a broad range of business regulatory and corporate governance matters, including representing corporations, auditing and other professional firms, investment banks and other financial institutions in US Securities and Exchange Commission (SEC) and other regulatory enforcement proceedings. In addition, Mr. Baker conducts internal investigations on behalf of management and boards of directors. He regularly counsels clients on SEC reporting, disclosure, compliance and corporate governance requirements. Prior to joining Latham, Mr. Baker was Associate Director of the Division of Enforcement at the SEC, where he worked for 15 years. In that capacity, he was responsible for supervising all types of SEC enforcement activities, including Investigations involving issuer accounting fraud and other disclosure violations, insider trading, market manipulation and broker-dealer misconduct. During his tenure as Associate Director, Mr. Baker lead numerous high-profile investigations that resulted in several landmark enforcement actions, including the global settlement in 2000 involving the Commission, Department of Treasury, Department of Justice, Internal Revenue Service, National Association of Securities Dealers and 21 securities firms that resulted in those firms paying a total of US$195 million to resolve claims that the firms charged excessive markups on government securities, and the Commission's action against WorldCom Inc., involving one of the largest financial frauds in history. While at the Commission, he was a recipient of the SEC's Stanley Sporkin Award, awarded by the Chairman of the SEC in recognition of outstanding contributions to the Enforcement program, and of the Commission's Law and Policy Award. Mr. Baker has been recognized as a leading securities litigation lawyer by Chambers USA and The Legal 500 US. 32

He is the co-author of "Corporate Internal Investigations after Sarbanes-Oxley" published in Volume II of The Practitioner's Guide to the Sarbanes-Oxley Act (American Bar Association 2005) and is a contributor to Securities Law Techniques (Matthew Bender). He is Co-chair of the Sub-Committee on SEC Enforcement and Civil Litigation of the American Bar Association Business Law Section and serves on the Advisory Council of the SEC Historical Society. From 2001-2004, Mr. Baker was an adjunct professor at George Washington University Law School, where he taught Securities Regulation. He is a frequent speaker and panelist on securities law issues at programs organized by a wide variety of groups, including the American Bar Association, the District of Columbia Bar Association, the Association of the Bar of the City of New York, the Securities Industry Association, The Bond Market Association, the Justice Department's National Advocacy Center, the Practicing Law Institute, Georgetown University Law Center and Stanford Law School. After graduating from law school, Mr. Baker clerked for Judge Douglas W. Hillman, United States District Judge for the Western District of Michigan.

33

Appendix B Variable Definitions

The first panel lists variables derived from data collected by Gerald Martin and used in Karpoff et al. (2008a, 2008b). The second panel lists variables based on data manually collected as described in Appendix A. The third panel lists variables based on additional data available via Compustat and CRSP. All accounting data is as of the year-end preceding the violation end date. Continuous accounting and returns-based variables are winsorized at 1% and 99%.

Variables Based on the Karpoff, Lee, and Martin Dataset Crim_Case Binary variable equal to one if a criminal case is filed Failed_Firm Binary variable equal to one if the firm delists before the end of the regulation period Vio_Length Natural logarithm of the length of the violation period, in days Reg_Length Natural logarithm of the length of the regulation period, in days Reg_End_Date Filing date of the concluding regulatory proceeding Trigger_Date The date that the public learned of a potential SEC violation Vio_End_Date End of the SEC infraction violation period

Manually-Collected Data CEO_Charge Damages Experience Hiringfirm_Rank Ivy

Last_Year Media_Hits

Revolver Score_Defendant SEC_Specialist Win Years_as Lawyer Deffirm_Hires_Avg

Binary variable equal to one if the firm's CEO is personally charged by the SEC Monetary damages scaled by Trigger_Loss (defined below) Total number of cases worked by the SEC lawyer up to and including the current case Vault prestige score of the revolvers post-SEC law firm Binary variable equal to one for lawyers who graduated from one of the following ivy law schools: Harvard, Cornell, Yale, Princeton, Columbia, Brown, Dartmouth, MIT, Stanford, University of Chicago, or University of Pennsylvania. Binary variable equal to one if the case concludes within one year of the lawyer leaving the SEC Count variable for the number of press articles identified in the LexisNexis Academic database for the period from one month before to one month after the SECs litigation release. Binary variable equal to one for lawyers who leave the SEC to work at law firm Average of the defendant law firms Vault prestige scores Count variable for the number of times the lawyer's post-SEC employer law firm appears in our defense firm sample Binary variable equal to one if the case is won. Equal to zero if the case is settled Difference between the year of the case and the year that SEC lawyer passed the Bar Exam. Number of SEC lawyers hired by the defense law firms in our sample. If there are more than one defense firms, this is the average across all defense firms involved.

Variables Based on Compustat / CRSP Assets Natural log of total assets Beta Market model beta. Calculated using value weighted market returns over the 11- month period ending one month prior to the violation end date BTM book value / market value Pre_Vio_End_CAR Buy-and-hold, market-adjusted returns for the 11-month period ending one month prior to the violation end date ROA income before extraordinary items / total assets SOX binary variable equal to one if the Reg_End_Date is after 2002 Trigger_CAR Three-day buy-and-hold market-adjusted returns around the trigger date Trigger_Loss Loss of market value in the three days around the trigger date. Calculated as the market value of equity two days prior to the trigger date multiplied by Trigger_CAR

34

Table 1 Sample Refinement

The original sample is based on the population of SEC enforcement actions used by Karpoff, Lee and Martin (2008). The complete sample include 865 regulatory enforcement actions by the SEC from 1979 2007. The enforcement actions arise from violations of the Securities Exchange Act of 1934, as amended by the Foreign Corrupt Practices Act of 1977. See Appendix A for a detailed description of the data collection process.

Total Karpoff, Lee, and Martin Sample Less: Cases beginning before 1990 Cases post-1990 Less: Firms not on CRSP & Compustat

865 -160 705 -121 584 -121 463 -79 384 -60 -9 -14 301

Less: Non-Litigated Cases Litigated Cases Less: Docket Unavailable Litigated Cases with Docket Less: Firms without an identified trigger date Less: firms without 3-day returns surrounding the trigger date Less: Firms without sufficient CRSP & Compustat data for the basic models Final Sample For Data Collection

**

Less: Cases for which no lawyer information can be obtained Final Sample of Cases for Analysis

-17 284

***

Number of Individual Lawyers Identified Final Sample of Lawyer-Cases

336 666

* Untabulated analysis shows that the litigated cases tend to have larger trigger date losses than the unlitigated cases (-21% versus 16%) and have a higher likelihood of failing before the end of the regulation period (36% versus 21%). Differences in size, return on assets, book-to-market, stock beta, and length of the violation period are insignificant. ** Untabulated analysis shows that the log of the violation period length is slightly longer for firms for which dockets are available (6.38 versus 6.60). *** Untabulated analysis shows no significant differences in the aforementioned variables between the cases for which we have and dont have lawyer data.

35

Table 2 Sample Summary Information

The sample contains information for 284 SEC enforcement cases and 336 individual SEC lawyers, for a total of 666 lawyer-cases. All variables are defined in Appendix B. Panel A contains summary statistic. Panel B presents correlation coefficients.

Panel A: Summary Statistics Mean Firm Characteristics Assets BTM ROA Beta Failed_firm SP500 Lawyer Characteristics Experience Ivy Last_Year Revolver SEC_Specialist Case Characteristics Pre_Vio_End_CAR Trigger_CAR Vio_Length Reg_Length Media_Hits Coop Enforcement Outcomes Damages Win Crim_Case CEO_Charge 5.703 0.532 -0.130 1.044 0.330 0.146 p25 3.935 0.206 -0.151 0.479 0.000 0.000 Median 5.612 0.395 0.000 0.934 0.000 0.000 p75 7.238 0.685 0.042 1.561 1.000 0.000 Std. Dev. 2.446 0.496 0.367 0.764 0.471 0.353 N 666 666 666 666 666 666

2.452 0.228 0.138 0.282 0.435

1.000 0.000 0.000 0.000 0.000

1.000 0.000 0.000 0.000 0.000

3.000 0.000 0.000 1.000 0.000

2.411 0.420 0.345 0.450 1.204

666 666 616 666 666

-0.050 -0.204 6.576 5.072 6.119 0.219

-0.598 -0.389 6.118 3.584 1.000 0.000

-0.200 -0.126 6.596 6.263 4.000 0.000

0.163 -0.022 7.079 7.086 9.000 1.000

0.810 0.223 0.835 2.749 6.899 0.512

666 666 666 666 666 666

0.004 0.074 0.447 0.542

0.000 0.000 0.000 0.000

0.000 0.000 0.000 1.000

0.009 0.000 1.000 1.000

1.874 0.261 0.498 0.499

624 666 666 666

36

Panel B: Correlation coefficients of select variables Pearson (Spearman) above (below) the diagonal

Damages Damages Win Crim_Case CEO_Charge Revolver SEC_Specialist 1.00 -0.02 -0.14 0.00 0.01 0.00

Win 0.05 1.00 -0.05 -0.04 0.00 -0.02

Crim_Case -0.08 -0.05 1.00 0.15 0.08 0.06

CEO_Charge 0.01 -0.04 0.15 1.00 0.07 0.09

Revolver 0.06 0.00 0.08 0.07 1.00 0.75

SEC_Specialist 0.05 -0.04 0.06 0.12 0.58 1.00

37

Table 3 Lawyer Employment Information

The sample contains information for 284 SEC enforcement cases and 336 individual SEC lawyers, for a total of 666 lawyercases. Panel A provides detail on whether SEC lawyers are still with the SEC, left for a non-law firm, or left to a law firm. Panel A also provides additional detail of SEC_Specialist. SEC_Specialist is a count variable that increases by one each time a postSEC employer appears as a defense firm in our cases sample. For instance, an SEC lawyer that leaves the SEC for a firm that is listed on the defense team for three cases will have SEC_Specialist = 3.

Panel A: SEC Lawyer Employment Information & SEC_Specialist Summary Information Lawyer-cases for lawyers still with the SEC Lawyer-cases for lawyers that quit to join a non-law firm Lawyer-cases for lawyers that quit to a law firm ("Revolvers") Total 389 89 188 666

SEC_Specialist Distribution Info.

n 188

mean 1.5

min 0

p25 0

median 1

p75 2

Max 8

38

Panel B: Differences between revolvers and non-revolvers

Non-Revolver lawyers are those that are either still with the SEC or have left the SEC for a non-law firm. Revolver lawyers are those that left the SEC to work at a law firm. SEC_Specialist 2+ lawyers are those that left the SEC to work at a law firm that shows up at least twice as a defense firm in our sample. SEC_Specialist 4+ lawyers are those that left the SEC to work at a law firm that shows up at least four times as a defense firm in our sample. The t-stat of difference columns show the results of a t-test that the comparison group mean differs from the Non-Revolver group. All other variables are as described in Appendix B. Standard errors for the t-statistics of differences in means are clustered by case. *** Indicates the difference is significant at 1%; ** at 5%; and * at 10%, two-tailed. Comparison Group 1 Comparison Group 2 Comparison Group 3 A B C D Average: Average: t-stat of Average: Average: t-stat of Non-Revolver Revolver difference SEC_Specialist 2+ t-stat of difference SEC_Specialist 4+ difference (n=478) (n=188) (B A) (n=59) (C A) (n=33) (D A) Firm Characteristics Assets 5.817 5.41 [1.85]* 4.957 [2.58]** 5.06 [1.92]* BTM 0.547 0.492 [0.90] 0.423 [1.85]* 0.42 [1.59] ROA -0.132 -0.127 [0.15] -0.185 [0.98] -0.216 [1.08] Stock Beta 1.063 0.998 [0.94] 0.945 [1.02] 1.176 [0.74] Failed_firm 0.303 0.399 [2.07]** 0.44 [1.81]* 0.454 [1.67]* SP500 0.149 0.139 [0.31] 0.102 [1.01] 0.091 [1.04] Lawyer Characteristics Experience Ivy Last_Year Case Characteristics Pre_Vio_End_CAR Trigger_CAR Vio_Length Reg_Length Media_Hits Coop Enforcement Outcomes Damages22 Win Crim_Case CEO_Charge 2.529 0.174 -0.058 2.255 0.367 -0.029 [1.32] [4.51]*** [0.31] 2.034 0.272 -0.093 [2.20]** [1.50] [0.31] 1.606 0.425 -0.103 [5.20]*** [2.76]*** [0.33]

-0.058 -0.19 6.582 4.942 6.096 0.236

-0.029 -0.24 6.56 5.403 6.176 0.175

[0.31] [2.38]** [0.31] [1.74]* [0.14] [1.27]

-0.093 -0.212 6.466 5.219 5.559 0.067

[0.31] [0.67] [0.87] [0.69] [0.59] [2.73]***

-0.103 -0.229 6.396 5.194 6.757 0.091

[0.33] [0.96] [1.09] [0.51] [0.54] [1.63]

-0.067 0.073 0.423 0.519

0.181 0.074 0.511 0.601

[1.04] [0.06] [1.88]* [1.78]*

-0.012 0.034 0.56 0.729

[0.30] [1.56] [1.83]* [3.13]***

0.247 0.03 0.637 0.818

[1.51] [1.33] [2.52]** [4.25]***

22

The Damages variable is negative in some cases because we scale Damages, a negative number, by TRIGGER_LOSS, which is abnormal return around the trigger date x MVE. A positive abnormal return around the trigger date will lead to a negative damages_loss variable. This is the case for 11.56% of our sample. Note that that the mean of -0.067 in column A and of -0.012 in column C are insignificantly different from zero (t = 0.96 and 0.07 respectively).

39

Table 4 Empirical Model for Damages and Winning

The sample contains 666 lawyer-cases. Panel A reports OLS estimation with the dependent variable being DAMAGES i.e., monetary damages scaled by Trigger_Loss. Panel B reports logit estimation with the dependent variable being an indicator variables that takes the value one if the SEC won the case. All variables are as listed in Appendix B. T-statistics in brackets are based on heteroskedasticity-robust standard errors that are clustered by case. *** indicates significance at 1%; ** at 5%; * at 10%, based on two-tailed tests.

Panel A: Damages
Model 1 Firm Characteristics Assets BTM ROA Beta Failed_Firm Case Characteristics Pre_Vio_End_CAR Trigger_CAR Vio_Length Reg_Length Media_Hits Sox Lawyer Characteristics Experience Ivy Revolver SEC_Specialist Constant -3.099 [-2.12]** 624 0.076 Yes 0.231 [2.00]** -0.255 [-0.77] -0.375 [-0.98] 0.237 [2.01]** -0.066 [-0.24] -0.192 [-0.78] -0.382 [-0.57] 0.284 [1.87]* -0.047 [-0.66] -0.021 [-1.45] Model 2 0.233 [2.01]** -0.243 [-0.74] -0.371 [-0.96] 0.232 [2.04]** -0.060 [-0.22] -0.184 [-0.74] -0.414 [-0.62] 0.298 [1.89]* -0.043 [-0.61] -0.020 [-1.43] Model 3 0.162 [1.76]* -0.200 [-0.69] -0.282 [-0.72] 0.147 [1.99]** -0.098 [-0.34] -0.198 [-0.72] -0.417 [-0.69] -0.006 [-0.05] -0.019 [-0.37] -0.033 [-1.88]* -0.285 [-1.09] -0.009 [-0.34] 0.095 [0.31] Model 1 -0.259 [-1.56] 0.730 [1.58] -1.446 [-2.70]*** -0.608 [-1.34] -1.040 [-1.63] -0.327 [-0.82] -0.518 [-0.56] -0.182 [-0.64] 0.312 [2.08]** 0.017 [0.42]

Panel B: Win
Model 2 -0.263 [-1.57] 0.696 [1.50] -1.452 [-2.73]*** -0.618 [-1.38] -1.020 [-1.60] -0.336 [-0.85] -0.505 [-0.53] -0.175 [-0.62] 0.296 [2.00]** 0.017 [0.41] Model 3 -0.210 [-1.40] 0.738 [1.28] -1.511 [-3.17]*** -0.717 [-1.64] -0.524 [-0.95] -0.181 [-0.50] -0.556 [-0.58] -0.242 [-0.74] 0.234 [1.67]* 0.010 [0.17] -1.042 [-2.04]** -0.123 [-1.02] 0.329 [0.79]

-0.000 [-0.01] 0.034 [0.12] 0.230 [1.17]

0.001 [0.03] 0.049 [0.16]

-0.112 [-0.93] 0.030 [0.08] 0.083 [0.23]

-0.108 [-0.92] 0.108 [0.28]

0.086 [2.01]** -3.216 [-2.15]** 624 0.076 Yes

0.072 [1.51] -0.554 [-0.59] 624 0.030 No

-2.550 [-1.18] 666 0.322 Yes

-0.227 [-1.80]* -2.375 [-1.09] 666 0.326 Yes

-0.299 [-2.09]** -0.532 [-0.22] 666 0.228 No

Observations Adjusted R2 Year Fixed Effects

40

Table 5 Empirical Estimation for Likelihood of Criminal Cases and Charging the CEO
The sample contains 666 lawyer-cases. Panel A reports logit estimation where the dependent variable is CRIM_CASE a binary variable equal to one if the SEC files criminal charges in the case. Panel B reports logit estimation where the dependent variable, CEO_CHARGE is a binary variable equal to one if the SEC also charges the CEO of the targeted firm. All variables are as listed in Appendix B. Z-statistics in brackets are based on heteroskedasticity-robust standard errors that are clustered by case. *** Indicates significance at 1%; ** at 5%; * at 10%, based on two-tailed tests.

Panel A: Criminal Case

Model 1 Firm Characteristics Assets BTM ROA Beta Failed_Firm Case Characteristics Pre_Vio_End_CAR Trigger_CAR Vio_Length Reg_Length Media_Hits Sox Lawyer Characteristics Experience Ivy Revolver SEC_Specialist Constant -6.331 [-3.06]*** 666 Yes 0.326 0.014 [0.17] -0.252 [-0.71] 0.330 [0.79] 0.322 [1.24] 0.385 [0.98] 0.158 [0.69] 1.305 [1.68]* 0.192 [0.79] 0.578 [5.58]*** 0.080 [2.05]** Model 2 0.018 [0.22] -0.236 [-0.66] 0.354 [0.85] 0.329 [1.26] 0.387 [0.99] 0.167 [0.74] 1.284 [1.65]* 0.207 [0.86] 0.587 [5.75]*** 0.081 [2.06]** Model 3 -0.001 [-0.01] -0.205 [-0.60] 0.568 [1.38] 0.434 [1.64] 0.400 [1.12] 0.153 [0.75] 1.298 [1.78]* 0.339 [1.38] 0.552 [5.47]*** 0.071 [2.09]** 0.214 [0.52] 0.105 [2.07]** -0.235 [-1.00]

Panel B: Charging the CEO

Model 1 -0.347 [-3.60]*** -1.302 [-3.17]*** 0.909 [1.90]* 0.134 [0.49] 0.297 [0.79] -0.324 [-1.46] -0.535 [-0.62] 0.524 [2.16]** -0.006 [-0.09] 0.057 [2.10]** Model 2 -0.345 [-3.57]*** -1.278 [-3.17]*** 0.929 [1.91]* 0.143 [0.52] 0.292 [0.77] -0.321 [-1.43] -0.550 [-0.64] 0.544 [2.22]** -0.001 [-0.01] 0.057 [2.09]** Model 3 -0.295 [-3.20]*** -1.236 [-2.97]*** 0.978 [2.17]** 0.210 [0.84] 0.273 [0.77] -0.251 [-1.21] -0.746 [-0.93] 0.426 [2.10]** 0.035 [0.59] 0.046 [1.74]* 0.196 [0.48] 0.003 [0.08] 0.217 [0.97]

0.117 [2.26]** -0.236 [-0.99] 0.253 [1.12]

0.119 [2.30]** -0.251 [-1.04]

-0.011 [-0.29] 0.192 [0.87] 0.054 [0.27]

-0.008 [-0.22] 0.128 [0.58]

0.181 [1.98]** -6.435 [-3.13]*** 666 Yes 0.329

0.192 [2.04]** -6.484 [-3.20]*** 666 No 0.300

-3.076 [-1.66]* 666 Yes 0.199

0.187 [2.36]** -3.246 [-1.74]* 666 Yes 0.204

0.205 [2.59]*** -1.369 [-0.95] 666 No 0.152

Observations Year Fixed Effects Pseudo R2

41

Table 6: Impact of the Last Year in the SEC on Enforcement

The sample contains 666 lawyer-cases. The dependent variable is as listed in each column header. All variables are as listed in Appendix A. Z-statistics in brackets are based on heteroskedasticity-robust standard errors that are clustered by case. *** Indicates significance at 1%; ** at 5%; * at 10%, based on two-tailed tests.

Damages / Loss Firm Characteristics Assets BTM ROA Beta Failed_Firm Case Characteristics Pre_Vio_End_CAR Trigger_CAR Vio_Length Reg_Length Media_Hits Lawyer Characteristics Experience Ivy SEC_Specialist Last_Year Constant 0.183 [1.73]* -0.258 [-0.75] -0.340 [-0.88] 0.161 [1.68]* -0.046 [-0.17] -0.286 [-1.20] -0.700 [-1.02] 0.199 [1.65] -0.069 [-0.92] -0.007 [-0.83] 0.017 [0.78] -0.165 [-0.77] 0.119 [2.15]** -0.433 [-1.93]* -2.210 [-2.05]** 578 Yes 0.111

Win -0.148 [-0.86] 0.567 [1.20] -1.742 [-3.15]*** -0.778 [-1.49] -1.015 [-1.53] -0.619 [-1.14] -0.167 [-0.17] -0.335 [-1.08] 0.282 [1.93]* -0.019 [-0.40] -0.067 [-0.69] -0.018 [-0.05] -0.133 [-1.40] -1.453 [-1.99]** -1.673 [-0.77] 616 Yes 0.343

Criminal Case 0.065 [0.74] -0.407 [-1.10] 0.253 [0.60] 0.294 [1.10] 0.391 [1.01] 0.110 [0.48] 1.188 [1.53] 0.218 [0.89] 0.590 [5.95]*** 0.073 [1.81]* 0.124 [2.40]** -0.268 [-1.10] 0.139 [1.20] -0.059 [-0.18] -7.174 [-3.46]*** 616 Yes 0.335

CEO Charged -0.337 [-3.45]*** -1.313 [-3.18]*** 0.954 [2.03]** 0.177 [0.62] 0.174 [0.46] -0.266 [-1.19] -0.675 [-0.76] 0.599 [2.35]** 0.017 [0.26] 0.063 [2.26]** -0.011 [-0.28] 0.195 [0.87] 0.353 [3.31]*** -0.834 [-2.60]*** -3.466 [-1.80]* 616 Yes 0.211

Observations Year Fixed Effects Adjusted / Pseudo R2

42

Table 7: Distribution of Number of Cases Litigated

This table details the total number of cases litigated by each of the 336 individual SEC lawyers. Revolver lawyers are those that left the SEC to work at a law firm. Non-Revolver lawyers are those that are still with the SEC or left the SEC for employment other than with a law firm. ^^ Indicates that the difference in means between revolvers and non-revolvers is insignificant at 10%.

Number of Cases Litigated 1 2 3 4 5 6 7 8 10 12 14 15 Total Lawyers Mean Cases Per Lawyer Median Cases Per Lawyer

Non-Revolvers Number of Percentage of Lawyers Total 135 57.9% 49 20 9 8 3 1 3 2 1 2 0 233 2.05 1 21.0% 8.6% 3.9% 3.4% 1.3% 0.4% 1.3% 0.9% 0.4% 0.9% 0.0%

Revolvers Number of Percentage of Lawyers Total 62 60.2% 24 10 1 2 2 1 0 0 0 0 1 103 1.83 1 ^^ 23.3% 9.7% 1.0% 1.9% 1.9% 1.0% 0.0% 0.0% 0.0% 0.0% 1.0%

Total 197 73 30 10 10 5 2 3 2 1 2 1 336

43

Table 8: Enforcement by Last Year at the SEC

The sample include 15 SEC revolver lawyers that left to join law firms and 54 cases that they worked at while at the SEC. Last_Year is coded as one if they were working on the case in their last year at the SEC. A t-test (Wilcoxon) of the differences in means (medians) is used to assess whether there is a difference between enforcement outcomes by last year at the SEC. ** significant at 5% based on two-tailed test

Last_Year = 0 N 26 28 28 28 Mean 0.59 0.18 0.39 0.75 Median 0.00 0.00 0.00 1.00 N 26 26 26 26

Last_Year = 1 Mean 0.01 0.00 0.62 0.58 Median 0.00 0.00 1.00 1.00 Diff Means -0.58 -0.18 ** 0.22 -0.17 Diff Medians 0.00 0.00 ** 1.00 0.00

Damages Win Crim_Case CEO_Charge

44

Table 9: Additional Results

The sample contains 666 lawyer-cases. The dependent variable is as listed in each column header. Only partial results, i.e. the variables of interest have been reported. Variables included in the estimation but not displayed are Assets, BTM, ROA, Beta, Failed_Firm, Pre_Vio_End_Car, Trigger_Car, Vio_Length, Reg_Length, Media_Hits, year fixed effects, and a constant. Number_Deffirms is the log of the number of law firms that comprise the defense team. Score_Defendant is the average vault score for defense firms for 2007. Years_as_Lawyer is the difference between the year of the case and the year that SEC lawyer cleared the bar exam. Hiringfirm_Rank is based on prestige scores for law firms obtained from Vault.com. It takes the value of 2 when revolver joins a law firms with a top 20 rank, a value of 1 if the law firm joined has a rank from 21 to 100 and 0 if the law firm joined is not ranked. Panel A estimates the effect of lawyer age. A young (mature) lawyer dummy takes the value 1 if the time from the entering the bar to leaving the SEC is below(above) the median (17 years). Panel B estimates the effect of SEC alumni. Deffirm_Hires_Avg is the average number of SEC lawyers hired by the defendant firms in the case. All variables are as listed in Appendix A. Z-statistics in brackets are based on heteroskedasticity-robust standard errors that are clustered by case. *** Indicates significance at 1%; ** at 5%; * at 10%, based on two-tailed tests.

Damages / Loss Panel A: Role of Lawyer Age Lawyer Characteristics Experience Ivy SEC_Specialist* Young Lawyer SEC_Specialist*Mature Lawyer Constant

Win

Criminal Case

CEO Charged

0.004 [0.15] -0.149 [-0.67] 0.012 [0.37] 0.138 [1.49] -1.863 [-1.84]* 569 Yes 0.071

-0.074 [-0.78] -0.108 [-0.27] -0.290 [-2.12]** -0.285 [-1.26] -1.597 [-0.73] 607 Yes 0.331

0.129 [2.53]** -0.281 [-1.14] 0.236 [2.09]** -0.003 [-0.02] -7.247 [-3.52]*** 607 Yes 0.333

-0.014 [-0.35] 0.131 [0.58] 0.315 [2.95]*** 0.115 [1.04] -3.505 [-1.80]* 607 Yes 0.205

Observations Year Fixed Effects Adjusted / Pseudo R2 Panel B: Influence of SEC Alumni Lawyer Characteristics Experience Ivy SEC_Specialist Defense Firm Characteristics Log Number of Defense Firms Deffirm_Hires_Avg Constant

0.003 [0.06] 0.173 [0.41] 0.130 [2.12]** -0.250 [-0.77] -0.092 [-0.29] -5.517 [-1.96]* 403 Yes 0.119

0.035 [0.43] -0.209 [-0.48] -0.177 [-1.21] 1.258 [1.86]* 0.543 [0.82] -1.466 [-0.33] 434 Yes 0.354

0.030 [0.52] -0.627 [-2.55]** 0.226 [2.55]** -0.017 [-0.05] -0.341 [-0.89] -6.154 [-2.54]** 434 Yes 0.340

-0.066 [-1.08] -0.378 [-1.21] 0.261 [2.37]** 1.046 [2.48]** 0.196 [0.54] -1.134 [-0.43] 434 Yes 0.226

Observations Year Fixed Effects Adjusted / Pseudo R2

45

Table 9 Continued
Damages / Loss Panel C: Controlling for Quality of Defendant Law Firms Lawyer_Characteristics SEC_Specialist 0.125 [2.12]** Defense_Characteristics Log Number of Defense Firms -0.212 [-0.71] Score_Defendant 0.101 [0.97] Observations 403 Adjusted / Pseudo R2 0.124 Panel D: Additional Controls for Lawyer Experience Lawyer_Characteristics Experience -0.013 [-0.46] Ivy 0.091 [0.32] Years_as_Lawyer 0.005 [0.49] SEC_Specialist 0.083 [1.92]* Observations 597 Adjusted / Pseudo R2 0.055 Panel E: Alternate Proxy for Hiring Law firm Lawyer_Characteristics Experience Ivy Defense_Characteristics Hiringfirm_Rank Observations Adjusted / Pseudo R2 Win Criminal Case CEO Charged

-0.181 [-1.25] 1.200 [1.85]* 0.040 [0.18] 434 0.350

0.231 [2.66]*** -0.023 [-0.07] -0.165 [-1.19] 434 0.344

0.250 [2.22]** 1.023 [2.54]** 0.093 [0.82] 434 0.227

-0.054 [-0.40] 0.353 [0.87] -0.033 [-1.10] -0.207 [-1.87]* 637 0.296

0.154 [2.97]*** -0.137 [-0.57] -0.031 [-1.99]** 0.158 [1.76]* 637 0.324

0.003 [0.07] 0.159 [0.71] -0.011 [-0.73] 0.213 [2.61]*** 637 0.204

-0.000 [-0.02] 0.121 [0.37] -0.086 [-0.72] 624 0.074

-0.082 [-0.75] 0.212 [0.57] -0.979 [-2.26]** 666 0.339

0.115 [2.20]** -0.308 [-1.31] 0.395 [2.14]** 666 0.330

-0.013 [-0.33] 0.098 [0.46] 0.321 [1.90]* 666 0.203

46