
























Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Community
Ask the community for help and clear up your study doubts
Discover the best universities in your country according to Docsity users
Free resources
Download our free guides on studying techniques, anxiety management strategies, and thesis advice from Docsity tutors
The article discusses the motivation behind going-private transactions and the desire to escape the burdens and costs of public ownership, including litigation risk. The author considers whether alternative steps, such as governance reforms, may be sufficient against litigation exposure. The article analyzes various measurable attributes of governance to predict subsequent litigation exposure. The findings have implications for the private-equity market and for corporate/securities law more generally.
Typology: Essays (university)
1 / 32
This page cannot be seen from the preview
Don't miss anything!
Eric L. Talleyt
Many going-private transactionsare motivated-at least^ ostensibly-by^ the^ desire to escape the burdens and costs of public ownership.Although these burdens have^ many purported manifestations,one commonly cited is the risk of litigation,which may be borne both directly by the firm and/or its fiduciaries or reflected in director and officer insurancepremia funded at company expense. An important issue for the "litigation risk" justification of privatizationis whether alternative (and less expensive) steps falling short of going private-such as governance reforms-may augursufficiently against litigation exposure. In this Article, I consider whether, controllingfor other^ variables^ related^ to^ firm- specific attributes,various measurable attributesof governance help to predict subsequent litigation exposure. Although there are some governance features (such as multiple board service, the presence of a staggered/classifiedboard, institutional investing, and the proactive adoption^ of^ a^ governance^ policy)^ that^ predict^ subsequent^ liability^ exposure, most governance indicia appearto^ be^ of^ negligiblepredictive^ value,^ both^ statistically^ and economically. In light of these findings, this Article discusses implicationsfor both the private-equity market and for corporate/securitieslaw more generally.
INTRODUCTION Prior to its quasi-hibernation in late 2007, the private-equity (PE) market rose to historically unprecedented levels. From 2002 through the third quarter of 2007, the total annual number of PE deals nearly doubled, and the associated annual dollar value of PE deals approx- imately quadrupled.' Many privatizations -particularly during and^ after the promulgation of the Sarbanes-Oxley Act of 20022 (SOX) and its regulatory progeny-were^ concentrated^ among^ micro-cap^ and^ small-
t Professor of Law and Co-director, Berkeley Center in Law, Business and the Economy, UC Berkeley School of Law; Robert B. and Candice J. Haas Visiting Professor of Law, Harvard Law School; Senior Economist (Adjunct), RAND Corporation. Thanks to Robert Bartlett, Alicia Davis-Evans, Todd^ Henderson,^ Kim^ Hogrief,^ Arthur^ Levitt, Patrick McGurn, Nell Minow, Maureen Mulligan, Adam Pritchard, Robert Revile, two anonymous referees, and conference participants at The University of Chicago Law School, University of Michi- gan, and the American Bar Association Tort Trial & Insurance^ Practice^ Section^ for^ helpful^ comments and discussions, and to the RAND Corporation Institution for Civil Justice for generous support. Many thanks as well to the Corporate Library for granting me access to their data. All errors are mine. I See Per Stromberg, The New Demography of Private Equity, in Anuradha Gurung and Josh Lerner, eds, The Global Economic Impact^ of^ Private^ Equity Report^2008 3,^13 figures^ 1A^ and^ 1B (World Economic Forum 2008), online at httpJ/www.wefonun org/pdflcgi/pe/FulReport.pdf (visited Jan 11, 2009). 2 Sarbanes-Oxley Act of 2002, Pub L No 107-204,116 Stat 745, codified in relevant (^) part at 15 USC § 7201 et seq.
The University of Chicago Law Review
cap issuers, (^) an observation that has now been documented numerous times elsewhere. The going-private wave (^) of the last half-decade undoubtedly had many causal drivers (not the (^) least of which was relatively cheap access to debt available during the period). According to many commentators, researchers, (^) and the privatizing companies themselves, however, the pri- vate-equity wave of the (^) last half-decade was at least partially inspired by an organizational desire to escape the (espoused) burdens (^) of public ownership, including (^) litigation risk.' While difficult to test directly, this claim is at least a plausible one. Indeed, SOX (and its (^) regulatory progeny) substantially enhanced the power of both government (^) and self-regulating organizations (SROs) to commence enforcement (^) actions against public issuers and their fiduciaries.^5 Moreover, the post-SOX (^) regulatory land- scape also gave private plaintiffs greater leverage (^) in bringing suit against public companies. To be sure, most of the (^) provisions of SOX specifically disclaim the creation of a private right of action, but at the (^) same time the legislation included key features (^) that almost certainly enhanced the attractiveness of securities litigation. (^) Most directly, it increased (pros- pectively) the limitations period for filing a securities fraud (^) action (both from occurrence and discovery),' (^) thereby enhancing the value of the real option for shareholders to seek (^) redress of their complaints through litiga- tion. Somewhat less directly, SOX liberalized the (^) utilization of "fair-fund" escrows in which to park moneys collected by the SEC from statutory
3 See, for example, Ehud Kamar, Pinar Karaca-Mandic, and Eric Talley, Going-private (^) Deci- sions and the Sarbanes-Oxley (^) Act of 2002: A Cross-countryAnalysis, 25 J L, Econ, & Org *3 (forth- coming 2009), online at http'//ssm.com/abstract=901769 (^) (visited Jan 11, 2009); Robert R Bartlett III, Going Private (^) but Staying Public: Reexaming the Effect of Sarbanes-Oxley on Firms' Going-private Decisions, 76 U Chi (^) L Rev 7,33-38 (2009) (examining data on companies' going-private decisions and concluding that only small-cap (^) and medium-cap companies have done so to avoid SOX requirements). 4 See, (^) for example, Ehud Kamar, Pinar Karaca-Mandic, and Eric (^) L. Talley, Sarbanes-Oxley's Effect on Small Firms:What Is the Evidence?, in Susan M. Gates and Kristin J. Leuschner, eds, In (^) the Name of Entrepreneurship?The (^) Logic and Effects of Special Regulatory Treatment for Small Busi- ness 143, (^165) table 5.3 (RAND 2007), online at httpJ/www.rand.org/pubs/monographst2007/RAND_ MG663.pdf (visited Jan 11, 2009) (offering a concise summary of the literature on the effects (^) of SOX on small firms and large firms); Maurice R. Greenberg, Regulation, Yes; Strangulation,No, Wall St J A10 (Aug 21,2006) (highlighting (^) the negative reaction to SOX compliance among public corporation (^) executives); Alan Murray, For Sarbanes-Oxley Bashers; Some Perspective, Wall St J A (Nov 16, 2005) (acknowledging the costs of SOX for smaller (^) firms and the complaints from busi- nesses of all sizes, but questioning whether the costs of regulation (^) can really explain larger firms going private). 5 See Kamar, Karaca-Mandic, and Talley, (^) Sarbanes-Oxley's Effects on Small Firms at 145- (cited in note 4). (^6) See Robert Serio and Matthew Kahn, Private Rights (^) of Action and the Sarbanes-OxleyAct of 2002, 38 Sec Reg (^) & L Rep 668,669-71 (Apr 2006) (highlighting possible implied causes of action arising out of SOX). 7 See 28 USC § (^) 1658.
[76:
The University of Chicago Law Review
combination of current and prospective fiduciaries of the issuers (who may demand more compensation in exchange for the added risk), and the issuers themselves (who also face liability exposure and in any event generally pay director and officer (D&O) insurance premiums asso- ciated with litigation risk). Anecdotally, it is notable that during the last few years numerous D&O insurers have begun to "grade" issuers' go- vernance based on perceived litigation risk, and in addition numerous private vendors have moved into the market of predicting such litiga- tion susceptibility." Moreover, beyond the availability of low-cost debt capital, many ascribe the private-equity wave of the last seven years to a combination of (1) a desire to remove the regulatory and litigious over- hang that drags down returns of public companies; and (2) a desire to put a public issuer in a type (^) of "quarantine" away from the oversight of corporate law, activist (^) shareholders, and securities regulation/litigation, where the public focus on "good governance" may run wide of the mark of what is appropriate (^) for that firm.'^3 To the extent that the various go- vernance reforms (^) implemented under SOX were efficacious in reducing agency (^) costs and fraud (and associated litigation), then the above two arguments would not be very convincing. On the other hand, if the go- vernance reforms championed in the (^) post-SOX environment did not have much of an effect in reducing litigation costs associated with secur- ities litigation, then it would lend some support to these possible de- fenses of the going private. My empirical findings, while qualitatively mixed, appear (^) to be more consistent with the (^) latter argument above. That is, the predictive rela- tionship I am able to uncover between governance choices and prospec- tive litigation risk is relatively (and in some ways surprisingly) modest. While there (^) are particular governance features (for example, multiple board service, the proactive adoption of a corporate governance poli- cy, (^) and to some extent the existence of a classified/staggered board) that bear relatively strongly and robustly on prospective litigation (^) risks, most factors-and indeed most of those promulgated by SOX that the data studied here can measure -appear to have little predictive effect on the incidence of litigation and a firm's exposure once sued. It is important to note from the outset that this Article does not at- tempt to analyze (^) executive compensation either as a component of go-
12 See, for example, The Corporate Library, Securities Litigation Risk Analyst, online at http:// www.thecorporatelibrary.com/info.php?id=49 (visited Jan 11, 2009) (advertising software that pre- dicts the likelihood of a securities class action against a company using factors like the company's "governance (^) risk"). 13 Consider Stanley B. Block, The Latest Movement (^) to Going Private:An Empirical Study, 14 J Applied Fin 36,37 (Spring/Summer 2004) (reporting that firms that had recently gone private most often cited as their primary reason the cost of being public, in both dollars and time).
[76:
2009] Public Ownership,Firm Governance, and Litigation Risk 339
vernance structure or as a predictor of litigation risk. This omission is deliberate and has multiple justifications. Primarily, the relationship be- tween compensation and litigation risk is already one that has been ex- plored extensively in the literature. Previous research has found that the structure of an executive's compensation package (and in particular, the fraction of one's compensation that comes through incentive payments, bonuses, and/or stock and options compensation) is relatively predictive of later accounting restatements, SEC investigations, and private securi- ties litigation." A contemporaneous paper to this one considers whether a company's voluntary disclosure of a Rule 10b5-1 compensation plan is a marker of litigation risk." There the authors find, somewhat surprising- ly, and using controls similar to those used (^) here, that disclosure of a 10b5-1 plan is strongly associated with future litigation risk.^16 Moreover, the exclusion of compensation from my analysis has some rationale in the data. It turns out (^) that most of the available data on executive com- pensation comes from Compustat's ExecuComp Database, which focuses solely on relatively large issuers in the S&P 1500. Given that many sued firms come from a much smaller capitalization stratum -and it is these firms that appear to have been the most likely to utilize going-private strategies -using compensation data (even as a control) would tend to eliminate from view the set of firms that are among the most interesting for this study. Finally, including executive compensation as a "right-hand side" variable can create problems from an econometric perspective, since compensation and (^) litigation susceptibility are endogenously part of an overall system. Thus, if one were to include executive compensation as
14 See, for example, Shane A. Johnson, Harley E. Ryan, Jr, and Yisong S. Tian, Managerial Incentives and CorporateFraud:The Sources of Incentives (^) Matter *5 (European Finance Associa- tion 2006 Zurich Meetings, Feb 2008), online at http://ssrn.com/abstract=395960 (visited Jan (^) 11, 2009); Eric L. Talley and Gudrun Johnsen, Corporate Governance, Executive Compensation and Securities Litigation *4 (University of Southern California Law School Olin Law & Economics Working Paper No 04-7, May 2004), online at http://ssrn.com/abstract=536963 (visited Jan 11, (^) 2009) ("[W]e estimate that each 1% increase in the fraction of a CEO's contract devoted to medium- to long-term incentives (rather than short-term compensation) predicts a 0.3% increase in expected litigation and a $3.4 million dollar increase in expected settlement costs."); Bin Ke, Do Equity-based Incentives Induce CEOs to Manage Earnings to Report Strings of Consecutive EarningsIncreases? *2 (14th (^) Annual Conference on Financial Economics and Accounting, Feb 2004), online at http:// ssm.com/abstract=446540 (visited Jan 11, 2009) (concluding that (^) executives with high equity-based compensation are more likely to manage earnings reports to ensure there is a string of consecu- tive earnings increases for their own personal gain). 15 See (^) M. Todd Henderson, Alan D. Jagolinzer, and Karl A. Muller, (^) Scienter Disclosure (University of Chicago Law School Olin Law & Economics Working Paper No 411, July 2008), online at http://ssrn.com/abstract=1137928 (visited Jan 11, 2009). 16 Id at *2-3 (finding that insiders may voluntarily (^) disclose information prior to strategic trades in order to mitigate future litigation risks).
2009] Public Ownership,Firm Governance,and Litigation Risk 341
the components of the index to improve its predictive power; 2 still oth- ers have attempted to combine their index with other data to improve upon it. 21 Other ways that governance may matter have been more elu- sive but are still the topic of significant collective research. For example, recent research on governance and executive compensation appears to confirm the argument (albeit weakly) that "well governed" firms also tend to structure executive compensation in a manner that more closely ties pay to performance.T Nevertheless, the enterprise of empirical corporate governance has not been free from controversy. A recent study, for example, finds that even though popular corporate governance scores do well in explaining past performance, they fare much more poorly in predicting future per- formance or litigation risk.2 It would, of course, seem unlikely on a priori grounds that such scores should perform well in predicting litiga- tion risk given that they were not crafted with that in mind; rather, their strength (or at least purported strength) is in predicting other elements of shareholder value.T My enterprise in this Article, in (^) contrast, is much more fundamental-to determine whether the primitive building blocks of a corporate governance ranking (^) themselves have predictive pow- er-even if the prevailing (^) indices that aggregate those scores are less reliable. The fact that even these primitives have a predictive ability that is at best modest provides yet another insight into why their aggrega- tion into an off-the-rack governance "score" might similarly fare poorly. Another important literature that is related to this Article is the large body of work on the determinants of securities litigation, and in particular how the relative incidence of frivolous and meritorious suits
Paper No 491, Sept 2004), online at http://ssrn.com/abstract=593423 (visited Jan 11, 2009) (analyzing a subset of the GIM index consisting of six factors-four that concern shareholder voting power and two measures taken in preparation to hostile takeovers-and concluding that these (^) six factors are largely responsible for the relation between performance and corporate governance). 26 See, for example, Robert M. Bowen, Shivaram Rajgopal, and Mohan Venkatachalam, Accounting Discretion,Corporate Governanceand Firm Performance *20 n 11 (14th Annual (^) Confe- rence on imancial Economics and Accounting, Jan 2003), online at http://ssrn.com/abstract= (visited Jan 11, 2009) (considering several board characteristics and the GIM index as separate measures of governance). 27 See, for example, Lawrence D. Brown and (^) Marcus L. Caylor, Corporate Governance and Firm Performance *3-4 (Working Paper, Dec 2004), online at http://ssrn.com/abstract= (visited Jan 11, 2009) (considering the GIM in conjunction with proxies for board monitoring, institutional ownership, managerial ownership, incentive compensation by bonus or stock op- tions, and auditor expertise). 28 See Jay C. Hartzell and Laura T. Starks, InstitutionalInvestors and Executive Compensa- tion, 58 J Fin 2351, 2352 (2003). 29 See Robert Dairies, Ian Gow, and (^) David Larcker, Rating the Ratings: How Good Are Com- mercial Corporate Governance Ratings? *29 (Stanford Law School Olin Law & Economics Working Paper No 360, June 2008), online at http://ssrn.com/abstract=1152093 (visited Jan 11, 2009). 30 Significantly, Daines, Gow and Larcker report that most commercial ratings do not perform well as predictors even of standard shareholder value measures. Id at *21-26.
The University of Chicago Law Review
has changed over the years.^3 Although I have little to say about the is- sue of whether meritorious lawsuits have increased over the last decade, my findings may have at least some tangential relevance. Intuitively, one might conjecture that as the incidence of frivolous litigation de- creases, the connection between governance and litigation risk should grow stronger. The moderate to underwhelming results I find here, then, might^ also^ be^ consistent^ with^ a^ claim^ that^ regardless^ of^ their^ tra- jectory over the last decade, securities class actions remain a relatively noisy and unpredictable function of governance choices. One significant caveat that deserves^ explicit^ mention^ before^ pro- ceeding concerns the nature of most (if not all) attempts to understand the empirical relationship between governance and^ observable^ outcomes: it is difficult to overemphasize the caution one should exercise about in- terpreting the results of the sort of empirical exercise conducted here as a test of causal theories relating governance and other outcomes. In a manner similar to (though less extreme than) the executive compensa- tion discussion above, governance attributes within a firm are frequent- ly endogenous, making it difficult to find reliable, independent statistical instruments for predicting those choices. While there are many things one can try to do to confront this problem (such as lagging the inde- pendent variables of interest, adding additional controls, and using oth- er measures), the problem of endogeneity bias is unlikely to crumble away. This criticism is, of course, true both for those who purport to find a relationship between governance and a variable of interest and those who purport to find little or no relationship. Consequently, should there be refinements that would better address these issues than those uti- lized here (and there undoubtedly are), then my results must be inter- preted in light of those possible refinements. My analysis proceeds as follows. Part I describes the overall ar- chitecture and structure of the dataset, providing summary statistics of the variables that are available within it. Part II then presents an anal- ysis of the incremental effects of numerous^ governance^ characteris- tics - controlling for a number of other market characteristics - on the prospective incidence of securities class actions. Part III conducts a
31 See generally, for example, Stephen J. Choi, Do the Merits Matter Less after the Private Securities Litigation Reform Act?, 23 J L, Econ, & Org 598 (2007) (looking at the impact of the Private Securities Litigation Reform Act of 1995 (PSLRA) on meritorious securities litigation and concluding that although the PSLRA has reduced the incidence of "nuisance suit litigation," it has also worked to reduce more meritorious litigation aimed at smaller companies and com- panies engaged in fraud whose existence is not evinced in pre-filing "hard evidence"); Marilyn E Johnson, Karen K. Nelson, and A.C. Pritchard, Do the Merits Matter? The Impact of the Private Securities Litigation Reform Act, 23 J L, Econ, & Org 627 (2007) (looking at the same question and finding evidence that, post-PSLRA, plaintiffs' attorneys are more precisely targeting firms likely to have committed fraud).
[76:
The University of Chicago Law Review
interested in examining how such securities market measures predict litigation risk, it is relatively well established that such measures can play a substantial predictive role in securities litigation risk, and I therefore included key variables as controls. I therefore utilize a num- ber of CRSP variables to serve as controls for my analysis. In merging the principal datasets, it was important to remain mind- ful of the fact that only a small minority (ranging from 2 to 5 percent) of publicly traded firms are sued in any given year. Thus, one cannot ex- pect to have a one-to-one match between the governance and litigation databases. Consequently, the only firm-year observations that were dropped were those involving firms that appeared in the litigation da- tabase but for which I was unable to find a match within the CL and CRSP database. These dropped observations almost certainly bias the sample in favor of larger issuers that the Corporate Library tends to track. (As with most ratings and proxy advisory firms, the CL tracks slightly more than half of all publicly traded firms, skewed toward the larger capitalization issuers.) The merge was performed according to the six-digit (historical) CUSIP code of the issuer or-if that was not available -the issuer's exchange ticker code. The merged dataset consists of an unbalanced panel spanning the five-year period of the panel, comprised of 9,455 firm-years and 377 securities class actions. The summary statistics for the merged dataset are reflected in Tables la-c below. Note that in many cases, the Corpo- rate Library governance data was missing values for its key (^) variables for a number of sued firms. Thus, most of the later regressions that fol- low will utilize only a portion of CL data (around 50 percent of the raw data, sometimes less). As can be seen from Table la, this creates some bias in the direction of larger US issuers (for example, 93 percent of the sample is comprised of US corporations, the mean number of employees is over 85,000, and fully 39 percent of the firms in the merged dataset appear in the Russell 1000 index for that year).
[76:
20091 Public Ownership,Firm Governance,and LitigationRisk 345
TABLE 1A SUMMARY STATISTICS- MARKET CONTROL VARIABLES Standard Variable Observations Mean Deviation Minimum Maximum
Company Age (years) 5370 42.35102 47.38025 0 230
US Corporation (1 if yes) 9455 0.9353781 0.2458705 0 1
Delaware Corporation (1 if yes) 9455 0.5359069 0.4987354 0 1
Employees (Log) 8918 8.487771 1.685237 (^0) 19.
In Russell 1000 Index 9455 0.3922792 0.4882842 0 1 In S&P Midcap Index 9455 0.2070862 0.4052393 0 1 In S&P Smallcap Index 9455 0.2956108 0.456341 0 1 Mean Monthly Price (Log) 9142 3.126659 0.808775 -2.996 7. Mean Monthly Volume (Log) 9159 11.39143 1.594765 3.004 16. Mean Gross Abnormal Return (Log) 9121 0.0059051 0.1352223 -1.243 1. Mean Return Volatility 9114 0.3816908 0.2580139 0.028 4.
Table lb reports on the attributes of the 377 firms subject to suit. Note that the incidence of litigation against the issuers in the dataset is higher than it is among all publicly traded firms. 35 Between 4 and 5 per- cent of the sample (slightly less for all firm-years) is named in a securi- ties class action in the reporting years. This is more than twice the size of the historical litigation incidence rate, a fact that is not surprising given the larger capitalization of the sample relative to the entire popu- lation of public companies. The distribution of settlement values re- ported for the sued firms is right-skewed, with a mean value of $50 mil- lion, but a median of approximately $10 million and a 75th percentile of just under $30 million.
35 By most estimates, there are around 12,000 to 15,000 publicly traded companies in the United States, and historically there have been around 200 to 220 class actions filed per year, which would yield a prediction of around a 1.3 to 1.8 percent litigation rate. See Walkers Re- search, online at http://www.walkersresearch.com (visited Jan 11, 2009) (subscription required).
2009] Public Ownership,Firm Governance, and LitigationRisk 347
TABLE 1C SUMMARY STATISTICS (^) - CORPORATE GOVERNANCE VARIABLES
Standard Variable Observations Mean Deviation Min (^) Max
Insiders Control Institutional Investor Majority Institutional Investor Ownership Stake Classified/Staggered Board Majority Outsider Board Outside Board Members Meet Business Ethics Code Governance Policy Directors' Base Pay ($000, 2005 dollars) Outside Board Members (percent) Other CEO Board Members (percent) Board Members with > 15 Years Experience (percent) Board Members Serving on _ 4 Boards (percent) Board Members > 70 Years Old (percent) Women Board Members (percent) Dominant (^) Shareholder Audit Committee Independent Compensation Committee Independent Nominating/Governance Committee Independent CEO Is Chair
6124 6125 8525 9430 9421 6108 5852 9418 8877 9421 9421
9421 14.63755 17.3344 0 100
4029 0.8880616 0.3153298 0 3329 0.3331331 0.4714045 0
One should also note from the three tables above the fact that some of the governance variables have greater breadth across the dataset than do others. (For example, the "CEO Is Chair" indicator variable has significantly more missing data than, say, whether the board is classi- fied/staggered.) This will cause the effective sample size of the estima- tions below to vary (depending (^) on data coverage) across different forms of governance variables.
II. INCREMENTAL GOVERNANCE EFFECTS ON LITIGATION INCIDENCE Perhaps the most transparent (and least confusing) way to think about how governance metrics might predict litigation risk is to study a number of them sequentially in isolation. I turn to this task below, analyz- ing the incremental likelihood of securities litigation, controlling for in-
1 1
1 1 1 1 1 400 100 100
The University of Chicago Law Review
dustry and capital market effects, and for various measures of gover- nance performance. In the interests of conserving time and space, I limit my attention to the governance factors that are of greatest interest.
A. Litigation Exposure: Component-wise Effects
Consider first the likelihood that an issuer (^) is subject to a private suit as a function of its governance characteristics and a set of market controls. Specifically, consider the following functional relationship, in which the dependent variable is the probability that an issuer is subject to suit:
In the above expression, the dependent variable y, l is the event that a class action is filed against the company in the year following the reporting year; f(.) denotes an increasing function bounded between 0
securities market controls for each issuer i at each reporting year t. In all the specifications that (^) follow, the controls include logged price, logged monthly volume, the logged gross abnormal returns, return volatility, logged number of employees, Delaware incorporation, exchange dum- mies, industry dummies, capitalization dummies, and foreign issuer sta- tus; ., is an error term for firm i in year t. Finally, x,, denotes a single go- vernance characteristic of the issuer, considered sequentially. (In Part lI.B, I will report on combinations of and interactions between governance characteristics in fuller specifications.) In all specifications below, I use a random-effects logit likelihood function (though I obtained similar results with random- and fixed-effect linear probability estimates).^37 Although I will not delve deeply in what follows into the coefficient estimates of the baseline model - treating them strictly as controls - it is perhaps worth reporting on the estimates of these control variable coef- ficients as an initial matter. Table ld presents these results; keep in mind that the estimation procedure takes advantage of the panel structure of the data to estimate a random-effects logit specification, in which the dependent variable is the probability of the filing of a securities class action within the succeeding reporting year. Because of the presence of numerous binary variables, the coefficient for each variable in Table ld (and many of those that follow in this Part) is expressed in terms of an odds ratio: thus, values falling (^) below one reflect a smaller predicted like-
37 It would, in principle, also be possible to use a conditional logit (fixed-effects) approach here. However, because there is not a considerable amount of variation among the governance variables over time, I report on a random-effects estimator throughout.
[76:
The University of Chicago Law Review
large-cap firms and small-cap (^) firms attracting the greatest attention. 9 In particular, variables that are related to the likely (^) stakes associated with a securities settlement (such as price, volume, (^) abnormal returns, and (^) volatility) have tremendous explanatory value. Consistent with other studies on securities litigation, the most predictive single element appears to be the volatility in the return of the issuer's common stock." In addition, however, the higher the price and volume of the stock (holding constant the volatility), the greater the likely loss claim, and thus the more lucrative the case. With (^) the underlying structural model described above operating as a baseline, Table 2a now proceeds to consider the added (^) predictive effect of (^) governance variables. Each line from the Table reports on the marginal (^) coefficient estimates of the above specification with the hig- hlighted governance attribute added as the x,,, variable. (^) (That is, each row (^) from the Table represents a coefficient estimate from a different regression.) (^) As with the baseline estimation above, the coefficient for each variable is reported as an odds ratio, so as to (^) facilitate the interpre- tation of the coefficient for dummy variables. Before discussing (^) the factors in the Table that are significant, it is perhaps worth noting just how many of them (^) are not. Specifically, there does not appear to be much predictive power to any of the factors relating to corporate control by insiders, classified/staggered (^) boards, outside majorities on either the board itself or on significant commit- tees (audit, compensation, or nominating/governance), (^) and the shared role of CEO and chairperson;' in many respects, both (^) the reforms im- plemented by SOX and the recent (^) publicity around shareholder rights have concentrated (^) on addressing and altering each of these perceived
39 The (^) omitted category in the capitalization rankings contains (^) firms that appear neither in the Russell (^1000) nor the S&P small- and medium-cap indices. Because this omitted category constitutes a mixture of middling to small-cap firms (^) that are not in an index, the mid-cap order statistic coefficients have an interpretation that is more (^) challenging. Alternatively controlling for (logged) capitalization (^) does not significantly improve the predictiveness of the baseline model beyond these indicator measures of capitalization. 40 Throughout the analysis, I measure volatility (^) in any year as the standard deviation of the logged monthly gross returns during the year. This is consistent with standard asset-pricing ap- proaches in option-pricing theory. See Mark Grinblatt and Sheridan Titman, (^) FinancialMarkets and CorporateStrategy 280-83 (McGraw-Hill 2d ed 2002). (^) Although there is nothing sacrosanct about using stock volatility (for example, public debt holders, (^) preferred shareholders, and even option holders can also be securities fraud plaintiffs), this measure is (^) directly tied to derivatives prices, and the presence of common shareholders (^) in such suits is nearly ubiquitous. 41 One word of caution: (^) for the "CEO Is Chairman" regression, the CL data is more li- mited, and I was forced to drop some of the capitalization variables from (^) the baseline regression. However, even if one (^) drops all control variables, the "CEO Is Chairman" factor never appears to play any appreciable predictive role.
[76:
2009] Public Ownership,Firm Governance,and LitigationRisk 351
"problems." (^2) It may well be (^) that they are problems, (^) but the data (^) do
not appear to suggest that shareholders were reacting to these prob- lems by ratcheting up their own securities litigation efforts. Adding to the intrigue is the effect of a large sample size: in large datasets, statis- tical significance is not terribly hard to come by in its own right. Moreover, the approach used here-adding in each governance varia- ble individually in sequence rather than including them all simulta- neously-is even more likely to render results that are statistically significant. The fact, therefore, that only seven of the twenty gover- nance variables explored yield statistically significant predictions of litigation exposure is telling- not definitive, of course, but telling.
42 See, for example, Gretchen Morgenson, Soviet-style Proxies Made in the USA., NY Times Cl (June 25,2006).
2009] Public Ownership,Firm Governance, and Litigation Risk 353
odds of a shareholder suit declines. This finding is intuitive, and is consis- tent with greater willingness by institutional shareholders to use direct monitoring and governance pressure (rather than threatened litigation) as a core managerial discipline device. Similarly, the existence of a domi- nant shareholder also predicts a lower susceptibility to later class action litigation and is mildly statistically significant. One potential interpreta- tion of this effect is that both institutional-ownership-dominant share- holders may be both good monitors and well positioned to exercise more control over the company, thereby forestalling shareholder litiga- tion. Significantly, this effect does not appear to carry over to companies controlled by insiders, which face significant agency-cost problems not- withstanding the insiders' enhanced stake in company value. Next, consider the effects of board structure. As noted above,^ many of the reforms brought about by SOX, including outside majorities on the board and on significant board committees, appear not to matter much at all in predicting susceptibility to^ litigation exposure.^ However, there are some factors that do appear to play a role. Most notably, the firm's possession of a business ethics code^ significantly^ increases^ the predicted susceptibility to suit, with an order statistic^ of^ almost twenty. Conversations with practitioners suggest that this may^ make^ some^ sense for a number of reasons. First, the existence of a business ethics code is voluntary and^ may^ be^ endogenously^ determined^ by^ a^ "bad^ apples"^ ef- fect-that is, bad managers are forced^ to^ adopt^ business ethics^ codes^ in addition to being sued more. 3 In addition, the existence of a^ business ethics code provides a benchmark against which to measure subsequent behavior. Divergence from a stated ethics code can provide particula- rized evidence of intent to defraud, making things somewhat easier on prospective plaintiffs." Conversely, the possession of a formal corporate governance policy^ predicts^ a^ substantially lower^ litigation-risk^ threshold. One possible interpretive story here is that a^ governance^ code^ clarifies the processes that^ shareholders^ may expect^ to^ be^ accorded^ should^ they attempt to challenge management^ through^ nonlitigious^ means,^ thereby either providing valuable procedural information to shareholders or sending a signal that the corporation will be receptive to such endeavors.'
43 Recall that SOX did not require companies to adopt business ethics codes; rather, § 406 requires them to disclose whether they have adopted one consistent with the criteria laid out in the section.^ See^ SOX^ §^ 406(a),^116 Stat^ at^ 789,^ codified^ at^15 USC^ §^ 7264(a).^ Listing^ require- ments at the SRO level subsequently required all issuers to adopt one. See, for example, NYSE, Listed Company Manual at § 303A.10 (cited in note 10). 44 See Harvey L. Pitt and Karl A. Groskaufmanis, Minimizing CorporateCivil and Criminal Liability:A Second Look at CorporateCodes of Conduct, 78 Georgetown L J 1559,1560 (1990). 45 Although it is often plaintiffs'^ attorneys^ (and^ not^ shareholders)^ who^ make^ an^ initial^ deci- sion to bring class^ actions,^ an availing^ governance^ code might^ still^ augur^ against litigation^ risk^ by
The University of Chicago Law Review
Next, consider the attributes of the directors themselves. Although the existence of an outside majority on the board does not itself appear to have much predictive (^) power (see Table 2a), the overall fractional re- presentation of outside directors on the board (that is, outside board membership measured on a continuous scale) does appear to have a nontrivial (^) dampening effect.4 Delving deeper into the data, it turns out that most of this effect is identified (^) from inframarginal boards possess- ing (^) a minority of outside directors: increasing outside composition (even if (^) not to a majority) appears to substantially reduce litigation risk. 7 This may be because minority outside directors, while formally (^) powerless to effect change on the board, may act as effective whistleblowers." Finally, the data pick up two other predictive component-specific effects. First, the percentage representation on a board of members who serve on multiple (four or more) boards appears to predict significantly dampened litigation risk. This effect remains robust (^) (and even a bit stronger) with other baseline models or permutations of controls. (^) In some respects, multiple board service would (^) not automatically seem to predict a lower susceptibility to litigation. (^) However, there is one sense in which board overlap can serve as an important information dissemi- nation device. Many practitioners with (^) whom I have spoken about this effect agree that maintaining a degree (^) of multiple board experience provides useful avenues for shared knowledge among members (^) of the board. (^) This interboard learning effect may prove significant in coming years given the decline in multiple (^) board service witnessed in the post- SOX era (a point that I shall return to in Part IV).' Finally, the data (^) suggest that the percentage composition of women on the board tends to predict lower litigation exposure. Intriguing as it
inducing greater information revelation to the market sooner, which in turn can undermine the viability of a later securities fraud action. 46 Even though the value of the (^) odds ratio seems modest, remember that this variable is measured continuously from 0 percent to 100 percent, and thus the coefficient measures the odds ratio difference of (^) a move of 1 percent in outside board representation. 47 Estimating this same model for companies that do not have an outside majority (^) yields a nearly identical coefficient estimate and standard error. 48 See, for example, Julie Creswell, A Board in Need of an Emily Post, NY Times C1 (Sept 7, 2006) (discussing the controversy over (^) Hewlett-Packard obtaining phone records to determine the source of board (^) leaks and the role that Tom Perkins, an outside director, played in publicizing it). 49 My results on multiple board service are a bit distinct from those found in a recent paper that concentrates more directly on board attributes per se. See Stephen P. Ferris, Murali (^) Jaganna- than, and A.C. Pritchard, Too Busy to Mind (^) the Business? Monitoring by Directors with Multiple Board Appointments, 58 J Fin 1087, 1107-08 table VIII (^) (2003) (using a matched sample approach to consider the effect of multiple board service for firms sued between 1996 and 1998). Ferris, Jagannathan, and Pritchard find essentially (^) no significant difference. See id at 1109. The longer panel and later time (^) period studied here, along with somewhat more controls, may be playing part of the role in explaining (^) the difference in results. In addition, my results consider multiple board service in terms of a dummy variable (that is, service on four or more boards), which (^) may pick up the especially experienced board members.
[76: