Is the Price Right? An Empirical Study of Fee-Setting in Securities Class Actions

Is the Price Right? An Empirical Study of Fee-Setting in Securities Class Actions

Every year, fee awards enable millions of people to obtain access to justice and strengthen the deterrent effect of the law by motivating lawyers to handle class actions. But little research exists on why judges award the amounts they do or whether they size fee awards correctly. The process remains a black box. Through a detailed study of 431 securities class actions that settled in federal district courts from 2007 through 2012, this Article presents the first empirical study to peer inside that black box. In contrast to prior analyses, this study relies on the actual court filings in each case to create an original, comprehensive data set on all points at which federal judges are likely to consider issues relating to fees. These data enable us to paint a picture of the fee-setting process that is unusually detailed and nuanced—and that falsifies many common beliefs.

Among this Article’s major findings are that: (1) federal judges often deviate from the path Congress laid out in the Private Securities Litigation Reform Act of 1995 (PSLRA), which requires lead plaintiffs to set the terms of class counsel’s retention and federal judges to serve as backstops against abuses; (2) fees are generally lower in federal districts that see a high volume of securities class actions than in districts that handle lower volumes; (3) judges in high-volume districts are significantly more likely to cut fees than low-volume judges; (4) the “decrease–increase” rule, according to which fee percentages decline as settlements become larger, operates mainly in high-volume districts; and (5) judges appear to cut fees randomly, typically with very little explanation for why they did so. Finally, this Article finds that so-called “lodestar cross-checks,” which are supposed to help judges moderate fee awards, have unintended effects. All else equal, fee awards are significantly higher when fee requests include cross-checks than when lawyers use only the percentage method. A plausible explanation is that lawyers are anticipating judges’ reactions to fee requests and acting strategically. They include lodestar information when their requests may appear excessive and they omit it either when they expect judges to grant their requests or they think that the lodestar data will not help their cause.

In sum, there is little evidence that courts’ current actions in securities class actions move class counsel’s fees closer to the “right price.” This Article therefore proposes a set of procedural reforms, which courts could easily adopt, to make fee-setting in securities class actions more transparent, more compatible with the PSLRA’s normative goals, and more predictable. The reforms would encourage lawyers to invest in class actions at more appropriate levels, with salutary effects for plaintiffs and the integrity of the financial markets.


  1. The PSLRA’s Mechanisms for Selecting the Lead Plaintiff and Class Counsel and for Setting Attorneys’ Fees
  2. The Data on Fee-Setting in Securities Class Actions
    1. The Sample
    2. Historical Findings
      1. Descriptive Statistics
      2. Regression Analysis
  3. Should Courts Set Fees Ex Ante? A Proposal
    1. Prior Reform Proposals
    2. The Proposal
      1. The Proposal Respects the PSLRA’s Preference for
        Private Ordering
      2. The Proposal Creates Superior Incentives to Invest in Litigation



Every year, class action settlements bring $10–$20 billion into federal courts. 1 See Brian T. Fitzpatrick, An Empirical Study of Class Action Settlements and Their Fee Awards, 7 J. Empirical Legal Stud. 811, 825 (2010) [hereinafter Fitzpatrick, Settlements] (studying all class actions settled in federal courts in 2006 and 2007 and reporting total recoveries of $22 billion and $11 billion, respectively). And every year, federal judges award billions of these dollars to plaintiffs’ attorneys in payment of fees and reimbursement of expenses. 2 Id. at 831 (reporting $2.9 billion in fee and expense awards in 2006 and $2.1 billion in 2007). The payments are essential. But for these awards, the incentive to wage class actions—which entail enormous commitments of time and financial resources 3 The average case in our sample lasted four years from initial filing until court approval of the award of attorneys’ fees and expenses. During that time period, a firm’s attorneys and other professionals would have worked hundreds or thousands of hours on the case without compensation. The law firm would also typically have paid the substantial expenses associated with litigating a class action lawsuit. In the average case in our sample, courts reimbursed firms for $727,598 in expenses. The firm must also bear all of the costs associated with the cases in which it was unable to obtain any recovery for the class. See C.S. Agnes Cheng et al., Institutional Monitoring Through Shareholder Litigation, 95 J. Fin. Econ. 356, 371 (2010) (finding courts dismissed 36.9% of sampled cases without any damages award). —would disappear. One can say, without exaggeration, that federal judges enable millions of people to obtain access to justice each year by rewarding lawyers for litigating class actions successfully.

Yet the process by which judges set fee and cost awards remains a black box. Settlements go in; awards come out. Little is known about the mechanism that earmarks dollars for attorneys. 4 The empirical literature on fee awards in class actions is large, but only one study to date has looked at fee-award procedures in any detail. See Lynn A. Baker, Michael A. Perino & Charles Silver, Setting Attorneys’ Fees in Securities Class Actions: An Empirical Assessment, 66 Vand. L. Rev. 1677 (2013) (examining fee-setting practices in three district courts). All of the other studies focus solely on inputs (settlements), outputs (fee and cost awards), and case characteristics (subject matter of the litigation, named plaintiff type, duration, etc.). See, e.g., Theodore Eisenberg & Geoffrey P. Miller, Attorney Fees and Expenses in Class Action Settlements: 1993–2008, 7 J. Empirical Legal Stud. 248, 250, 259–60 (2010) [hereinafter Eisenberg & Miller, Fees and Expenses] (finding attorneys’ fees vary with size of client recovery, fee method, and riskiness of case, and noting inter-circuit variations in fee awards); Theodore Eisenberg & Geoffrey P. Miller, Attorney Fees in Class Action Settlements: An Empirical Study, 1 J. Empirical Legal Stud. 27, 52 (2004) [hereinafter Eisenberg & Miller, Empirical Study] (documenting “strong correlation” between fees and settlement size); Fitzpatrick, Settlements, supra note 1, at 830–45 (finding fees vary with settlement size, case age, and where case litigated); Eric Helland & Jonathan Klick, The Effect of Judicial Expedience on Attorney Fees in Class Actions, 36 J. Legal Stud. 171, 187 (2007) (finding positive correlation between class-action fee awards and measures of court congestion); Robert H. Lande & Joshua P. Davis, Benefits from Private Antitrust Enforcement: An Analysis of Forty Cases, 42 U.S.F. L. Rev. 879, 891–93 (2008) (reporting data on size of fee awards in antitrust cases); William J. Lynk, The Courts and the Plaintiff’s Bar: Awarding the Attorney’s Fee in Class-Action Litigation, 23 J. Legal Stud. 185, 195–209 (1994) (empirically comparing lodestar and percentage-of-recovery approaches and concluding neither could be conclusively rejected “as an explanation of common-law class-action fee award determination”); Michael Perino, Institutional Activism Through Litigation: An Empirical Analysis of Public Pension Fund Participation in Securities Class Actions, 9 J. Empirical Legal Stud. 368, 369–70 (2012) [hereinafter Perino, Activism Through Litigation] (finding positive correlation between attorneys’ fees and investor recovery, negative correlation between fees and public-pension cases, and positive externality related to institutional activism). Consequently, it is difficult to know why fee awards are sized as they are. Studies have shown that more dollars flow to lawyers when settlements are larger; 5 See Michael A. Perino, Markets and Monitors: The Impact of Competition and Experience on Attorneys’ Fees in Securities Class Actions 21, (St. John’s Univ. Sch. of Law Legal Studies Research Paper Series, Paper No. 06-0034, 2006), (on file with the Columbia Law Review) (noting “near linear relationship” between attorneys’ fees and settlement sizes). that, in percentage terms, awards tend to decline as recoveries rise; 6 Eisenberg & Miller, Empirical Study, supra note 4, at 54 (providing data showing “[a]s client recovery increases, the fee percent decreases”). and that investors get more “bang for the buck” in securities-fraud class actions when public pension funds serve as lead plaintiffs. 7 Perino, Activism Through Litigation, supra note 4, at 383 (presenting data suggesting “public pensions do a better job of maximizing recoveries than other types of lead plaintiffs”). But the causes of these phenomena remain hidden from view because researchers have not peered inside the fee-setting mechanism.

The reason for this is simple: It takes work—a lot of it—to study fee-setting processes in action. In securities-fraud cases, which account for about 40% of all federal class actions and for over 70% of all federal class action recoveries, 8 See, e.g., Fitzpatrick, Settlements, supra note 1, at 818, 825 (noting “[s]ecurities” cases were 40% of total settlements approved by federal judges in 2006 and 35% in 2007, and composed 76% and 73% of recoveries by dollar value in 2006 and 2007, respectively). the mechanism first cranks up when the court appoints the lead plaintiff and lead counsel at the start of litigation. To study just this part of the process, one must review the motions, memoranda, and declarations filed by the parties and lawyers seeking control of the case, along with the orders entered by the court. With seventy to eighty settlements per year and two or more parties per case seeking appointment as the lead plaintiff, it takes substantial effort just to determine which parties and lawyers sought control, who won, and whether judges considered compensation terms when selecting the winners. 9 Most existing studies of fee awards in securities class actions control for the type of investor that served as lead plaintiff, see, e.g., Perino, Activism Through Litigation, supra note 4, at 384. However, no study quantifies the frequency with which lead plaintiffs and the lawyers they hire bargain over fees, describes the fee terms that such arm’s-length bargaining produces, or documents the frequency with which privately negotiated fee agreements are offered into the record for judges to review and consider when ruling on class counsel’s motion for attorneys’ fees at the end of the case. Any belief that the documented reduction in agency costs under the Private Securities Litigation Reform Act of 1995 (PSLRA) is due to lead plaintiffs who aggressively negotiate ex ante fee agree­ments with the attorneys they hire is only a surmise. To make matters more challenging, the relevant documents must often be obtained directly from courts because they are not available electronically. The process must then be repeated for each case at the settlement stage, where the items to be collected and reviewed include motions for settlement approval and fee awards, supporting affidavits, objections, responses to objections, and the court’s various orders.

Do the stakes justify the effort needed to peer inside the black box? Yes. More than billions of dollars in compensation and access to justice for millions of people are at issue. The law’s ability to influence conduct outside the courthouse also hangs in the balance. By setting fees too high or too low, judges would incentivize lawyers to bring too many class actions or too few. Excessive litigation would over-deter primary conduct that is desirable; insufficient litigation would under-deter primary conduct that is unwanted. 10 For a standard account of the relationship between deterrence and litigation rates, see, e.g., Steven Shavell, Foundations of Economic Analysis of Law 389–418 (2004) (discussing basic theory of litigation). These consequences will occur wherever we rely on private enforcement through class litigation: antitrust, 11 See Clayton Act, 15 U.S.C. § 15(a) (2012) (“[A]ny person who shall be injured in his business or property by reason of anything forbidden in the antitrust laws . . . shall recover . . . a reasonable attorney’s fee.”); Int’l Travel Arrangers, Inc. v. W. Airlines, Inc., 623 F.2d 1255, 1262 n.10 (8th Cir. 1980) (noting other courts’ “‘recognition that the purposes of the antitrust laws are best served by insuring that the private action will be an ever-present threat to deter anyone contemplating business behavior in violation of the antitrust laws’” (quoting Perma Life Mufflers, Inc. v. Int’l Parts Corp., 392 U.S. 134, 139 (1968))); see also In re Ins. Brokerage Antitrust Litig., 579 F.3d 241, 285 (3d Cir. 2009) (granting award of attorneys’ fees in Sherman Act litigation). civil rights, 12 See Civil Rights Attorney’s Fees Awards Act of 1976, 42 U.S.C. § 1988(b) (2012) (“In any action or proceeding to enforce [certain provisions of federal civil-rights statutes] the court, in its discretion, may allow the prevailing party, other than the United States, a reasonable attorney’s fee as part of the costs . . . .”); McDaniel v. County of Schenectady, 595 F.3d 411, 414–15 (2d Cir. 2010) (discussing district court’s broad discretion to fashion award of attorneys’ fees). consumer protection, 13 See, e.g., Truth in Lending Act, 15 U.S.C. § 1640(a)(3) (2012) (imposing upon creditors who violate consumer-protection statute liability for “reasonable attorney’s fee as determined by the court”); McGowan v. King, Inc., 616 F.2d 745, 746 (5th Cir. 1980) (noting “determination of a reasonable attorney’s fee is a matter for the sound discretion of the trial judge” in Truth in Lending Act litigation). and a host of other areas, 14 See generally 3 Alba Conte, Attorney Fee Awards 415–16, 483 (3d ed. 2004) (discussing litigation fee incentives in environmental and consumer-credit claims). including financial fraud. 15 See 2 id. at 455 (discussing attempts to prevent excessive litigation over desirable financial conduct). The deep reason for opening the black box is to learn whether the procedural system operates in ways that are reasonably thought to advance the goals of substantive laws.

The need is especially acute in the securities area, both because securities-fraud cases comprise a large percentage of the federal class action docket 16 See, e.g., Fitzpatrick, Settlements, supra note 1, at 818 (noting securities class-action settlements “comprised a large percentage” of all federal class-action settlements in 2006 and 2007). and because Congress instructed judges legislatively how to handle fee awards in these cases. 17 See infra text accompanying notes 24–27 (describing lead plaintiffs’ rights and responsibilities, including choosing counsel for the class and setting compensation terms). In 1995, Congress adopted the Private Securities Litigation Reform Act (PSLRA) because it viewed the securities class action system as dysfunctional and abusive. 18 Some have disputed this description. See Myriam Gilles & Gary B. Friedman, Exploding the Class Action Agency Cost Myth: The Social Utility of Entrepreneurial Lawyers, 155 U. Pa. L. Rev. 103, 105–06 (2006) (defending deterrent effects of securities class-action system and noting irrelevance of class-compensation reforms); Joel Seligman, The Merits Do Matter: A Comment on Professor Grundfest’s “Disimplying Private Rights of Action Under the Federal Securities Laws: The Commission’s Authority,” 108 Harv. L. Rev. 438, 439–41 (1994) (disputing argument that securities class-action claims destroy capital formation and emphasizing societal benefits of such litigation). One concern for Congress was that the litigants who served as representative plaintiffs were often small investors who would not or could not monitor class counsel effectively. 19 See H.R. Rep. No. 104-369, at 32 (1995) (Conf. Rep.), reprinted in 1995 U.S.C.C.A.N. 730, 731 (“[The PSLRA’s] provisions are intended to increase the likelihood that parties with significant holdings in issuers, whose interests are more strongly aligned with the class of shareholders, will participate in the litigation and exercise control over the selection and actions of plaintiff’s counsel.”). Among other abuses, lawyers paid kickbacks to individuals who held a few shares of likely litigation targets who were willing to serve as the needed representative plaintiff. Those individuals obviously could not be expected to vigorously monitor class counsel, particularly with respect to attorneys’ fees requests. Kickbacks were often a percentage of the lawyers’ fee and, therefore, the representative plaintiff would have an incentive to maximize rather than minimize fees. See Michael A. Perino, The Milberg Weiss Prosecution: No Harm, No Foul?, Briefly . . . Perspectives on Legislation, Regulation and Litigation, May 2008, at 1, 19–23, [] (showing how rational plaintiffs attempting to maximize their return might pursue strategy allowing lawyers to charge excessive fees). They were figureheads for lawyers who financed the cases, ran them as they wished, and, for all important purposes, were the real parties in interest. 20 See John C. Coffee, Jr., Understanding the Plaintiff’s Attorney: The Implications of Economic Theory for Private Enforcement of Law Through Class and Derivative Actions, 86 Colum. L. Rev. 669, 677–80 (1986) [hereinafter Coffee, Understanding] (“[I]n the context of class and derivative actions, it is well understood that the actual client generally has only a nominal stake in the outcome of the litigation.”); Jonathan R. Macey & Geoffrey P. Miller, The Plaintiffs’ Attorney’s Role in Class Action and Derivative Litigation: Economic Analysis and Recommendations for Reform, 58 U. Chi. L. Rev. 1, 7–8 (1991) (noting entrepreneurial attorneys in derivative and class litigation are “not subject to monitoring by their putative clients” and “operate largely according to their own self-interest”).

Another concern was that the traditional solution to this agency cost problem—ex post monitoring by federal judges—had proven inadequate. Although charged by the rules to act as guardians for absent class members, judges were at least as interested in getting cases off their dockets as they were in rooting out abuses. 21 See Janet C. Alexander, Do the Merits Matter? A Study of Settlements in Securities Class Actions, 43 Stan. L. Rev. 497, 566 (1991) (“Judges . . . typically display a strong interest in seeing large, complex cases such as securities class actions settle rather than go to trial and often take an active role in promoting settlement.”). Judges lacked the means to police abuses too. They could not rely on the parties who proposed settlements to bring problems to their attention, 22 Many commentators have made the point that neither defendants nor class counsel have incentives to bring defects in settlements to judges’ attention. For the canonical source, see Coffee, Understanding, supra note 20, at 714 (“[T]he plaintiff’s attorneys and the defendants can settle on a basis that is adverse to the interests of the plaintiffs. At its worst, the settlement process may amount to a covert exchange of a cheap settlement for a high award of attorney’s fees.”); id. at 717 (explaining lodestar method of regulating attorneys’ fees, which breaks connection between recoveries and fees, creates environment in which plaintiffs’ attorneys and defendants can collude, to detriment of class members); see also Edward Brunet, Class Action Objectors: Extortionist Free Riders or Fairness Guarantors, 2003 U. Chi. Legal Forum 403, 403, 404 & n.2 (showing Professor Coffee’s critique of settlement bargaining in class actions has gained widespread acceptance). and they could not (always) rely on objectors to do so either. Objectors only appeared in some cases, and their agendas were often suspect. Many objectors were hold-up artists bent on extorting payments. 23 On the frequency with which objectors appear in class actions of different types, including securities cases, see Theodore Eisenberg & Geoffrey Miller, The Role of Opt-Outs and Objectors in Class Action Litigation: Theoretical and Empirical Issues, 57 Vand. L. Rev. 1529, 1549 (2004) [hereinafter Eisenberg & Miller, Role of Opt-Outs] (“Across all case types . . . the median objection rate is zero and the mean is 1.1 percent of class members.”). On the impact of objectors on judicial fee determinations, see id. at 1563 (“Overwhelmingly, fees are determined by a single factor: the size of the recovery for the class as a whole. We found no significant association between the number of dissenters and either the gross fee or the fee as a percentage of class recovery.”). On objectors’ motives, see Bruce D. Greenberg, Keeping the Flies Out of the Ointment: Restricting Objectors to Class Action Settlements, 84 St. John’s L. Rev. 949, 950 (2010) (describing tactics used by objectors to extract money from class counsel and derail settlement process); see also Brian T. Fitzpatrick, The End of Objector Blackmail?, 62 Vand. L. Rev. 1623, 1624 (2009) [hereinafter Fitzpatrick, Objector Blackmail] (describing “objector blackmail” in which objectors demand out-of-pocket payment from class counsel in exchange for dropping bad-faith appeals to fee awards, and common responses of class counsel).

By enacting the PSLRA, Congress gave class-action procedure a substantial overhaul. 24 Compare Private Securities Litigation Reform Act of 1995, Pub. L. No. 104-67, 109 Stat. 737, 737–49 (codified as amended in scattered sections of 15 U.S.C. and 18 U.S.C.) (specifying detailed requirements for procedural issues including notice to class members and appointment of lead plaintiff), with Fed. R. Civ. P. 23 (setting forth class-action procedures in more general terms and leaving more details to district judges’ discretion); see also 15 U.S.C. §§ 77z-1(a)(3), 78u-4(a)(3) (2012) (establishing procedural guidelines for determining lead plaintiffs). Seeking to rely less on judges and objectors and more on incentives, it sought to put class actions under the control of sophisticated investors with large financial stakes. 25 See, e.g., H.R. Rep. No. 104-369, at 32 (1995) (Conf. Rep.), reprinted in 1995 U.S.C.C.A.N. 730, 731 (noting Congress’s intent to “increase the likelihood that parties with significant holdings in issuers, whose interests are more strongly aligned with the class of shareholders, will participate in the litigation and exercise control over the selection and actions of plaintiff’s counsel”). The hope was that these investors would seek to maximize their own net recoveries by maximizing the net recoveries for everyone. 26 See Elliott J. Weiss & John S. Beckerman, Let the Money Do the Monitoring: How Institutional Investors Can Reduce Agency Costs in Securities Class Actions, 104 Yale L.J. 2053, 2056 (1995) (noting institutional-investor plaintiffs “might be well situated to monitor the conduct of plaintiffs’ attorneys as proxies for all members of plaintiff classes”); see also Elliot J. Weiss, The Lead Plaintiff Provisions of the PSLRA After a Decade, or “Look What’s Happened to My Baby,” 61 Vand. L. Rev. 543, 547 (2008) (noting institutional-investor plaintiffs have economic incentive to litigate efficiently). They would use contingent-fee arrangements to incentivize excellent attorneys to obtain good results, while using competition among lawyers to obtain bargain rates. Sophisticated investors would also evaluate settlements, reducing the burden on judges, who are poorly placed to figure out whether proposed deals are good or bad. 27 Weiss & Beckerman, supra note 26, at 2066–67 (describing limitations on trial judges’ ability to evaluate parties’ proposed settlement deals).

Empirical studies of inputs and outputs suggest that the PSLRA has been a reasonable success. Although many of the most sophisticated private investment funds still refuse to serve as lead plaintiffs, 28 See Michael A. Perino, Have Institutional Fiduciaries Improved Securities Class Actions? A Review of the Empirical Literature on the PSLRA’s Lead Plaintiff Provision, in Handbook of Institutional Investment and Fiduciary Duty 146, 150 (James P. Hawley et al. eds., 2014) [hereinafter Perino, Institutional Fiduciaries] (referencing 2005 PricewaterhouseCoopers study finding “union and public pension funds served as lead plaintiffs in an average of 4.8% of the cases filed in the first three years . . . after passage of” PSLRA). public pension funds have volunteered in numbers, 29 See, e.g., Stephen J. Choi, A.C. Pritchard & Jill E. Fisch, Do Institutions Matter? The Impact of the Lead Plaintiff Provision of the Private Securities Litigation Reform Act, 83 Wash. U. L.Q. 869, 900–01 (2005) (noting pronounced “emergence of public pension funds as class representatives” following PSLRA’s adoption, but finding “no statistically significant correlation between public pension fund participation and reduced fee awards” even post–PSLRA); James D. Cox & Randall S. Thomas with the Assistance of Dana Kiku, Does the Lead Plaintiff Matter? An Empirical Analysis of Lead Plaintiffs in Securities Class Actions, 106 Colum. L. Rev. 1587, 1622 (2006) (noting “some support for the idea that public pension fund lead plaintiffs have the most potential to improve client monitoring of class counsel”). and their use of the mechanisms created by the PSLRA appears to be reducing agency costs. 30 See, e.g., Cheng et al., supra note 3, at 358 (arguing public pension funds “achieve better litigation outcomes and governance reform than other institutional investors”); James D. Cox, Randall S. Thomas & Lynn Bai, There Are Plaintiffs and . . . There Are Plaintiffs: An Empirical Analysis of Securities Class Action Settlements, 61 Vand. L. Rev. 355, 379 (2008) (finding “positive and significant impact on settlement size from the presence of a public pension fund”); Perino, Activism Through Litigation, supra note 4, at 369–70 (finding presence of public pension funds correlated with higher recoveries and lower fees). Cases led by public pension funds tend to have higher recoveries and lower attorneys’ fees than other securities class actions, controlling for important case characteristics. 31 See Perino, Activism Through Litigation, supra note 4, at 382–89 (presenting empirical results suggesting public pension fund participation “lead[s] to higher settlement amounts, all else equal” and “[o]n average, fees in public fund cases were 3.3 percent lower than in other cases during the same time period”). Even in cases led by other kinds of plaintiffs, attorneys’ fees have declined significantly over time. 32 Id. at 370 (“[I]n cases without public pension lead plaintiffs that were settled after 2002, fee awards dropped on average 2.2 percent.”). The overall reduction in fees appears to be a positive externality associated with public pension fund participation in these cases. Fee awards in cases with these kinds of lead plaintiffs are visible to other judges, creating downward pressure on fees in cases with other lead plaintiff types. Id. at 373, 387 (“[B]ecause institutions tend to become involved in the larger, higher-profile cases, . . . with increases in institutional involvement, even judges in districts that do not see substantial amounts of securities class actions may be able to observe more cases with lower fees and may adjust their own fee awards accordingly.”).

Without looking inside the black box, however, it is impossible to know whether the credit for these successes belongs to the PSLRA or to other aspects of the fee-setting mechanism. The documented reduction in attorneys’ fees and other agency costs could be the result of hard bargaining between institutional lead plaintiffs and class counsel. 33 This has been suggested by Professor Elliot Weiss, for example, who has con­tended that the mechanism he co-conceived is working as it should. In an article reflecting on developments in the post–PSLRA era, he wrote:

“Consistent with our expectations, institutional investors that have sought appointment as lead plaintiff generally have negotiated fee arrangements with the law firms they have retained that provide for percentage fees far lower than had been the norm prior to passage of the PSLRA. Many institutional lead plaintiffs also have actively monitored class actions in which they have served as lead plaintiff and have pushed for larger settlements, recoveries from individual defendants responsible for corporate frauds, and governance reforms directed at preventing corporate wrongdoing from recurring.”

Weiss, supra note 26, at 552–53. Professor Weiss could be right, but the studies on which he relies shed no light on how the observed reduction in attorneys’ fees and other agency costs has been obtained.
But the credit could, at least in part, also belong to federal judges, who may use the fees awarded in cases with public pension lead plaintiffs as the model for fees in other cases. The fact that fees have fallen in securities class actions of all types since the enactment of the PSLRA, including those led by other investors, 34 See Perino, Activism Through Litigation, supra note 4, at 385–89 (presenting empirical findings that fee requests “nearly 3 percent lower [post–2002] than in the reference period (pre–1999) when public pension fund participation was either nascent or nonexistent”). suggests that judges are playing an important role.

To learn how fee awards in class actions are set and whether the PSLRA is working as Congress intended, we created an original database with information on all of the 431 securities class action settlements that were announced between January 1, 2007 and December 31, 2012 in the federal district courts in the United States. 35 See infra section II.A (describing study’s sample of cases). This database is uniquely rich because it contains details gathered from the actual court filings in each of the cases, in addition to the settlement amounts and fee awards on which other studies rely. The data include the number and type of plaintiffs who sought control of class litigation, the terms of any agree­ments regarding fees and costs that were disclosed to the courts, the amounts requested as fees, the formulas for calculating fees that judges were asked to employ, the presence and number of objectors, and many other details. This information enables a deeper look inside the fee-setting black box.

This granular approach to analyzing the fee-setting process yields a picture that is far more nuanced than—and significantly different from—that suggested by the conventional wisdom, which is that institutional investors are solely responsible for recent improvements in the operation of securities class actions. 36 See Perino, Institutional Fiduciaries, supra note 28, at 150–56 (surveying post–PSLRA empirical studies focused predominantly on connection between institutional plaintiffs and recent improvements, with limited attention to other factors). Many of this study’s findings show that the PSLRA has not worked as hoped. For example, although the statute was supposed to encourage lead plaintiffs to bargain over fees with class counsel at the start of litigation, we find that cases with ex ante fee agreements are the exception rather than the rule. 37 See infra section II.B.1.b (showing average fee requests are higher where evi­dence of ex ante fee agreement is absent). And although courts are making smaller fee awards, in the vast bulk of cases that does not appear to be because they are explicitly choosing to abide by fee agreements negotiated between lead plaintiffs and their chosen counsel. 38 See infra notes 92–94 and accompanying text (finding little evidence judges actually consider ex ante fee agreements when awarding lead counsel position). The truth appears to be that, instead of enforcing (and requiring) such ex ante fee agreements, the courts in most cases set fees in precisely the same manner they did before passage of the PSLRA—ex post, after a settlement has already been reached. 39 See infra section II.B.1.a (reporting lead plaintiff candidate or court discussed ex ante fee agreement in only 11.29% of cases).

Equally troubling is the finding that the market for attorneys’ fees may be even more imperfect than previously suspected. 40 See infra section II.B.1.b (presenting descriptive findings on fee requests and awards). It is certainly true that some judges in securities class actions have complained that they find it difficult to set fees ex post because they lack data on what an actual client would agree to pay an attorney ex ante to litigate a securities class action. See Taubenfeld v. AON Corp., 415 F.3d 597, 599 (7th Cir. 2005) (“Although is it [sic] impossible to know ex post exactly what terms would have resulted from arm’s-length bargaining ex ante, courts must do their best to recreate the market by considering factors such as actual fee contracts that were privately negotiated for similar litigation, information from other cases, and data from class-counsel auctions.”); Goldberger v. Integrated Res., Inc., 209 F.3d 43, 52 (2d Cir. 2000) (“[W]e cannot know precisely what fees common fund plaintiffs in an efficient market for legal services would agree to, given an understanding of the particular case and the ability to engage in collective arm’s-length negotiation with counsel.”). That problem is less severe in the post–PSLRA period because judges can now look to the fees that public pension funds have negotiated ex ante. Indeed, it seems possible that the overall decline in fees since passage of the Act is attributable to the influence the public-pension cases may have had on fee-setting in other cases. See Perino, Activism Through Litigation, supra note 4, at 385–89 (presenting empirical data on impact of public-pension-fund plaintiffs on fee requests). Data reveal that the market for attorneys’ fees is geographically segmented. Judges from high-volume districts (those that see securities class actions more frequently) set fees that are significantly different from those set by judges from low-volume districts. 41 See infra section II.B.1.b (discussing fee awards granted by judges from low- and high-volume districts); infra section II.B.2.b (detailing variables’ effect on fee awards). The data also suggest that plaintiffs’ attorneys may be aware of and may seek to exploit these market imperfections by asking courts for significantly higher fees in low-volume districts than in high-volume ones. 42 See infra sections II.B.1.b–c (outlining discrepancies in fee-award requests in high- and low-volume districts); infra section II.B.2.a (analyzing High Volume variable’s effect on fee requests).

Perhaps most contrary to the goals of the PSLRA are our findings regarding courts’ current handling of fee requests in the minority of cases (about 15%) in which courts cut the requested fee. 43 See infra section II.B.1.c (presenting findings). Regression analyses show that these judicial fee cuts are effectively random events, driven more by judges’ predilections and biases than by the merits of the fee requests. 44 See infra section II.B.2.c (noting variables that correlate with increased proba­bility of fee cut). The unpredictability of fee cuts necessarily creates uncertainty on the part of class counsel, who cannot reliably estimate the returns on effort. This likely discourages lawyers from investing time and resources optimally in class action cases, to the detriment of investors who will lose compensation in particular cases and be victimized by frauds too often. Fee-related uncertainty weakens the law’s power to discourage wrongdoing.

Taken as a whole, these findings suggest that the current system of ex post fee-setting in securities class actions is deeply flawed. This Article therefore proposes a mechanism that would both work better and be more consonant with Congress’s intent, as expressed in the PSLRA. 45 See infra section III.B (proposing “set of arrangements to improve the PSLRA’s effectiveness”). The cornerstone for this proposal is the belief that attorneys’ fees in securities class actions should be set ex ante. The lead plaintiff should negotiate a fee when retaining counsel to handle the case. The lead plaintiff should be required to disclose the terms of the negotiated fee agreement to the district court when offering a law firm for appointment as class counsel. The district court should review the negotiated fee terms before appointing class counsel and should approve those terms unless they are clearly unreasonable or not the products of arm’s-length negotiations. And when reviewing class counsel’s request for a fee award at the end of litigation, the district court should apply the agreed and previously approved terms unless unforeseen developments have rendered those terms clearly excessive or unfair. 46 See infra section III.B (proposing “set of arrangements to improve the PSLRA’s effectiveness”). Among the advantages of this Article’s proposal are that it respects the PSLRA’s preference for private ordering, and it creates superior incentives for attorneys to invest the optimal amount of time and resources in the litigation, thereby maximizing the recovery to the class. 47 See infra sections III.B.1–2 (discussing proposal’s advantages over existing procedures).

The Article begins in Part I by briefly discussing the lead plaintiff and lead counsel selection mechanisms contained in the PSLRA. Part II sets out the design of the six-year nationwide study and discusses the empirical findings in greater depth. Part III discusses the normative implications that flow from these findings. It also considers several procedural reforms that would make fee-setting in securities class actions more transparent and compatible with the PSLRA by enhancing the flow of information to judges.

I. The PSLRA’s Mechanisms for Selecting the Lead Plaintiff and Class Counsel and for Setting Attorneys’ Fees

Under the PSLRA, the court chooses the lead plaintiff through a straightforward procedure. The party that is first to file a securities class action provides notice of the lawsuit to all investors by publication in a “widely circulated national business-oriented publication or wire service.” 48 15 U.S.C. §§ 77z-1(a)(3)(A)(i), 78u-4(a)(3)(A)(i) (2012). Interested investors then have sixty days to file a motion nominating themselves for the position of lead plaintiff. 49 Id. §§ 77z-1(a)(3)(A)(i)(II), 78u-4(a)(3)(A)(i)(II). Upon review of the timely submitted applications, the court is to appoint the “most adequate plaintiff” as the lead plaintiff. 50 Id. §§ 77z-1(a)(3)(B)(i), 78u-4(a)(3)(B)(i). The PSLRA imposes on the court a presumption that the most adequate plaintiff is the applicant that has “the largest financial interest in the relief sought by the class” and that “otherwise satisfies the requirements of Rule 23 of the Federal Rules of Civil Procedure.” 51 Id. §§ 77z-1(a)(3)(B)(iii)(I), 78u-4(a)(3)(B)(iii)(I). The PSLRA further specifies that the lead plaintiff is obligated, “subject to the approval of the court, [to] select and retain counsel to represent the class.” 52 Id. §§ 77z-1(a)(3)(B)(v), 78u-4(a)(3)(B)(v). The statute also explicitly limits the total attorneys’ fees and expenses that may be awarded to “a reasonable percentage of the amount of any damages and prejudgment interest actually paid to the class.” Id. §§ 77z-1(a)(6), 78u-4(a)(6).

In practice, when an investor files its application to be appointed lead plaintiff it typically has already chosen the law firm that it will recommend that the court appoint as class counsel. 53 The candidate for lead plaintiff will typically use its chosen legal counsel to file its motion for appointment as lead plaintiff. In addition, the candidate having selected and retained competent class counsel is “one of the best ways for a court to ensure that [an investor seeking appointment as lead plaintiff] will fairly and adequately represent the interests of the class.” In re Cendant Corp. Litig., 264 F.3d 201, 265 (3d Cir. 2001). One court explains the process of appointing lead plaintiff and class counsel as follows:

“Once adequate notice is given, PSLRA sets out a two-step process for appointing the lead plaintiff of which the first step is to decide whether to consolidate the actions . . . . The second step is the selection of adequate plaintiff whose counsel will be lead counsel for litigation subject to Court approval. 15 U.S.C. § 77z-1(a)(3)(B)(iii)(I)(bb) . . . . As one Court observed, ‘while it is clear that determination of lead plaintiff and lead counsel are separate questions . . . in most cases, these issues are decided concurrently.’”

Katt v. Titan Acquisitions, Ltd., 133 F. Supp. 2d 632, 646–47 (M.D. Tenn. 2000) (emphasis added) (quoting Fischler v. AmSouth Bancorporation, No. 96-1567-Civ-Y-17A, 1997 WL 118429, at *1 (M.D. Fla. 1997)).
Indeed, that chosen law firm usually serves as counsel-of-record for the investor in the process of applying for lead plaintiff status and in requesting that the investor’s chosen counsel be appointed counsel for the class. 54 This is not surprising since the alternative would be for a lead plaintiff candidate to proceed pro se in seeking appointment as the lead plaintiff. All of the lead plaintiff applications in the data set of 431 cases were filed on behalf of the lead plaintiff candidate by the lead plaintiff candidate’s chosen counsel.

The PSLRA’s mandate that the lead plaintiff “select and retain” counsel to represent the class suggests that a lead plaintiff candidate will hire its chosen law firm, including negotiating the terms of its compensation (subject to judicial review). 55 See Baker et al., supra note 4, at 1684 (supporting this allocation of responsibility on grounds “sophisticated lead plaintiff . . . with a large financial stake in the case and substantial experience in securities litigation” has requisite information, skills, and incentives to secure appropriate counsel for appropriate fee). Professors Elliott Weiss and John Beckerman, who conceived of the lead plaintiff mechanism, 56 See generally Weiss & Beckerman, supra note 26, at 2096, 2105 (“Courts would benefit were institutional investors with large stakes in class actions to serve as lead plaintiffs.”). hoped and expected that institutional investors would apply to be appointed lead plaintiffs and, when given control of class actions, “would act as reasonably diligent litigation monitors, negotiating arm’s length fee arrangements with plaintiffs’ attorneys and overseeing the prosecution and settlement of the actions in which they were involved.” 57 Weiss, supra note 26, at 551; see also Weiss & Beckerman, supra note 26, at 2105 (arguing institutional investors are best situated to secure appropriate representation for the class and negotiate fair terms of settlement). In other words, unlike passive small investors, institutional investors were supposed to be active participants in the litigation. They were expected to drive hard bargains with lawyers competing for the opportunity to earn fees by representing investor classes and to demand zealous representation from the lawyers they retained.

Consistent with the above analysis, the Third Circuit requires district court judges to consider fee agreements when appointing lead plaintiffs. 58 See In re Cendant Corp. Litig., 264 F.3d 201, 265 (3d. Cir. 2001) (holding “courts should . . . consider . . . whether the movant has demonstrated a willingness and ability to select competent class counsel and to negotiate a reasonable retainer agreement with that counsel” in assessing whether particular lead plaintiff applicant satisfies Rule 23 adequacy requirement). The trial judge’s job, according to the Third Circuit, is to ensure that the lead plaintiff “fairly and adequately represent[s] the interest of the class.” 59 Id. at 266. And “one of the best ways” a trial judge can do this is by “inquir[ing] whether [a lead plaintiff candidate] has demon­strated a willingness and ability to select competent class counsel and to negotiate a reasonable retainer agreement with that counsel.” 60 Id. at 265. There are a handful of cases outside the Third Circuit that take a similar approach. See, e.g., In re Quintus, 201 F.R.D. 475, 482 (N.D. Cal. 2001) (“[T]he best way for the court to assess a potential lead plaintiff’s adequacy is to consider the manner in which he has retained counsel and negotiated an attorney’s fee for the class.”).

The reported cases, however, suggest that this ex ante approach is infrequently employed. Judges outside the Third Circuit rarely examine fee agreements when appointing lead plaintiffs, 61 See, e.g., In re Cavanaugh, 306 F.3d 726, 732–33 (9th Cir. 2002) (“[T]he district court has no authority to select for the class what it considers to be the best possible lawyer or the lawyer offering the best possible fee schedule.”). and even within the Third Circuit, many district courts do not seem to do so. 62 See Steamfitters Local 449 Pension Fund v. Cent. European Distrib. Corp., No. 11-6247, 2012 WL 3638629, at *13 (D.N.J. Aug. 22, 2012) (accepting magistrate judge’s recommendation to approve lead plaintiff’s choice of counsel without discussion of fee arrangements when motion to appoint counsel was unopposed); Blake Partners, Inc. v. Orbcomm, Inc., No. 07-4517, 2008 WL 2277117, at *7–8 (D.N.J. June 2, 2008) (noting Third Circuit requires courts to consider “whether the movant has demonstrated a willingness and ability to . . . negotiate a reasonable retainer agreement with that counsel” but neglecting to discuss movant’s fee arrangement (internal quotation marks omitted) (quoting In re Cendant Corp. Litig., 264 F.3d at 265–66)); In re Sterling Fin. Corp. Sec. Class Action, No. 07-2171, 2007 WL 4570729, at *6–8 (E.D. Pa. Dec. 21, 2007) (determining, without describing fee arrangement, that proposed lead plaintiff “demonstrated a willingness and ability to select competent class counsel and to negotiate a reasonable retainer agreement with that counsel” (internal quotation marks omitted) (quoting In re Cendant Corp. Litig., 264 F.3d at 263)); Lowrey v. Toll Bros., No. 07-1513, 2007 U.S. Dist. LEXIS 99501, at *10 (E.D. Pa. June 29, 2007) (appointing lead plaintiff and expressing confidence it “will endeavor to negotiate a reasonable fee arrangement with counsel”). Prior to this Article’s study, however, there was no systematic, comprehensive information about the extent to which district courts’ existing practices comport with Congress’s intent in enacting the PSLRA. 63 See H.R. Rep. No. 104-369, at 34–35 (1995) (Conf. Rep.), reprinted in 1995 U.S.C.C.A.N. 730 (noting “increasing the role of institutional investors in class actions will ultimately benefit shareholders and assist courts by improving the quality of representa­tion” and “courts could be more confident settlements negotiated under the supervision of institutional plaintiffs were ‘fair and reasonable’ than [with those] negotiated by unsu­pervised plaintiffs’ attorneys” (quoting Weiss & Beckerman, supra note 26, at 2105)). In brief, no systematic information was available about the frequency with which lead plaintiffs negotiate ex ante fee agreements with class counsel; the frequency with which courts consider such agreements when appointing lead plaintiffs; the frequency with which class counsel invokes an ex ante fee agreement when seeking an award of fees from the court at the end of the litigation; or the frequency with which the court invokes (and enforces the terms of) class counsel’s ex ante fee agreement with the lead plaintiff when awarding attorneys’ fees at the end of the litigation. Nor was systematic information available about how often or to what extent judges overrule lead plaintiffs when awarding fees. Finally, it was not known whether, on average, judges or institutional investors with large financial stakes tended to be more parsimonious with regard to attorneys’ fees or whether there are systemic differences across courts with respect to fee awards.

Before turning to the details of this Article’s study, it is worth pausing briefly to describe how fees are typically set in class actions. 64 Many class actions unrelated to the securities laws employ the process described in this paragraph. See, e.g., Sullivan v. DB Invs., Inc., 667 F.3d 273, 330 (3d Cir. 2011) (employing this process in antitrust case); McDaniel v. County of Schenectady, 595 F.3d 411, 421–22 (2d Cir. 2010) (applying process to civil rights claim); Vizcaino v. Microsoft Corp., 290 F.3d 1043, 1047 (9th Cir. 2002) (applying process to employee benefits claim). When submitting a proposed settlement for judicial review, class counsel also submits an application for an award to cover attorneys’ fees and litigation costs. More often than not, class counsel requests a percentage of the common fund as fees, and judges calculate fees on this basis. 65 See Boeing Co. v. Van Gemert, 444 U.S. 472, 476 (1980) (“Each individual recovery was to carry its proportionate share of the total amount allowed for attorney’s fees, expenses, and disbursements.”). When setting the fee percentage or determining fees on another basis, judges enjoy broad discretion. The PSLRA and Federal Rule of Civil Procedure 23(h) specify only that fee awards be “reasonable.” 66 15 U.S.C. § 78u-4(a)(6) (2012) (“Total attorneys’ fees and expenses awarded by the court to counsel for the plaintiff class shall not exceed a reasonable percentage of the amount of any damages and prejudgment interest actually paid to the class.”); Fed. R. Civ. P. 23(h) (“In a certified class action, the court may award reasonable attorney’s fees and nontaxable costs that are authorized by law or by the parties’ agreement.”).

In the vast majority of cases, plaintiffs’ law firms, in order to establish the reasonableness of their fee request, provide information on the number of hours their attorneys worked on the case multiplied by the hourly rates common for attorneys in that district. Indeed, some courts mandate that attorneys provide detailed information on this figure—known as the “lodestar.” 67 See Gunter v. Ridgewood Energy Corp., 223 F.3d 190, 195 n.1 (3d Cir. 2000) (describing suggested practice of district courts cross-checking percentage fee award with amount calculated under lodestar methodology). For general descriptions of the lodestar methodology and the problems it creates, see Third Circuit Task Force, Court Awarded Attorney Fees, 108 F.R.D. 237, 243 (1986) [hereinafter 1986 Task Force Report] (defining “lodestar” as benchmark fee figure court uses to asses fee requests based on “multiplica­tion of the number of compensable hours by the reasonable hourly rate”); Charles Silver, Unloading the Lodestar: Toward a New Fee Award Procedure, 70 Tex. L. Rev. 865, 869 (1992) [hereinafter Silver, Unloading the Lodestar] (proposing “new set of standards and procedures for use in place of the lodestar method in all fee-shifting cases, regardless of the kind of relief sought”). Lodestar amounts are usually below the fee requested. 68 In a non-trivial number of cases, however, the lodestar exceeds the fee request. See, e.g., In re Heartland Payment Sys., Inc. Customer Data Sec. Breach Litig., 851 F. Supp. 2d 1040, 1088 (S.D. Tex. 2012) (“The record and law support reducing the lodestar by a negative multiplier to avoid a windfall to class counsel, given the value of the settlement obtained.”). The multiplier applied to adjust the lodestar amount to the percentage requested is supposed to compensate attorneys for the risks associated with handling the case on a contingency basis. In cases where the multiplier is unusually high (or, theoretically, unusually low), the court has the ability to adjust the percentage award. 69 In no instance has a court declared a class counsel’s multiplier too small and adjusted it upwards, either in this study’s data set or in the published cases with which the authors are familiar. The literature on the lodestar methodology is vast, 70 See, e.g., 1986 Task Force Report, supra note 67, at 238 (presenting findings of task force convened to “study the [lodestar method] and to make recommendations on the criteria to be utilized in determining attorneys’ fee awards,” in response to difficulties encountered in first ten years of experience with lodestar method). For academic critiques of and commentary on the lodestar approach, see Charles Alan Wright et. al., Federal Practice and Procedure § 2675.2 (3d ed. 2014); Lynk, supra note 4, at 209 (surveying controversy around lodestar method and concluding empirical results “reject both a strong form of the lodestar proposition . . . and a strong form of the percentage-of-recovery proposition”); Charles Silver, Due Process and the Lodestar Method: You Can’t Get There from Here, 74 Tulane L. Rev. 1809, 1810 (2000) (arguing judges should use percentage-based compensation formulas rather than lodestar method in common-fund cases because of due process imperative to minimize conflicts in class actions); Silver, Unloading the Lodestar, supra note 67, at 869 (proposing “new set of standards and procedures” to replace lodestar method). and although it has been subject to withering judicial and academic criticism, 71 See Goldberger v. Integrated Res., Inc., 209 F.3d 43, 48 (2d Cir. 2000) (noting lodestar methodology creates “temptation for lawyers to run up the number of hours for which they could be paid”); 1986 Task Force Report, supra note 67, at 248 (concluding lodestar methodology “creates a disincentive for the early settlement of cases” because it places premium on hours worked). the approach is still commonly used by courts 72 See, e.g., In re AT&T Corp. Sec. Litig., 455 F.3d 160, 164 (3d Cir. 2006) (“Attorneys’ fees are typically assessed through the percentage-of-recovery method or through the lodestar method.”). In this study’s sample, 43.36% of the judicial fee orders explicitly reference the percentage method with a lodestar cross-check. Virtually all of the remaining cases used the percentage method alone. (and therefore also by class counsel in their fee requests 73 In this study’s data set, 92.16% of the fee applications used the percentage method with a lodestar cross-check. Virtually all of the remaining applications used the percentage method alone. ) as a “cross-check” on the reasonableness of the percent­age fee requested.

In most cases, this process occurs in a largely non-adversarial setting. Defendants are typically indifferent to fee awards because in a common fund case, the fee is deducted from the settlement and therefore simply involves the allocation of an already fixed sum between class counsel and class members. 74 A more adversarial process is likely where the statutory cause of action provides that defendants are liable to compensate plaintiffs for the attorneys’ fees they have incurred in successfully pursuing the action in addition to any damages. See, e.g., 15 U.S.C. § 15c(a)(2) (2012) (permitting successful plaintiffs suing under antitrust laws to recover “reasonable attorney’s fee” from defendants); Civil Rights Attorney’s Fees Awards Act of 1976, Pub. L. No. 94-559, 90 Stat. 2641 (codified at 42 U.S.C. § 1988(b) (2012)) (permitting court, in its discretion, to allow prevailing party, other than the United States, “reasonable attorney’s fee as part of the costs”). A larger fee costs a defendant no more and a smaller fee saves a defendant no money. The primary challenges to excessive fees are supposed to come from class members, but prior studies show that objections to fee requests are rare. 75 See Eisenberg & Miller, Role of Opt-Outs, supra note 23, at 1550 tbl.3 (finding median number of objectors to securities class-action settlement was zero).

II. The Data on Fee-Setting in Securities Class Actions

This Part contains the study’s analyses of fee-setting in securities class actions. Section A provides information on how we created the study’s sample. Section B reports the empirical findings.

A. The Sample

Using lists of case names and docket numbers provided by the Stanford Securities Class Action Clearinghouse, which supplied a comprehensive list of securities class actions filed in federal court since passage of the PSLRA, 76 See About Us, Stanford Law Sch. Securities Class Action Clearinghouse, [] (last visited Aug. 1, 2015) (“[O]ur population of records consist [sic] of securities class action lawsuits filed after 1995.”). we identified every securities class action filed in every federal district court in which the parties announced a settlement from January 1, 2007 through December 31, 2012. 77 These cases were filed between October 1998 and August 2011. In some cases, the court’s order awarding attorneys’ fees and/or granting final approval for a settlement announced during our study period was not filed until 2013. We included these cases in the sample. Some cases with multiple partial settlements had announced settlements that occurred prior to our study period. In order to make the data analysis as complete as possible, we included these earlier partial settlements in the sample as well.

The sample omits all cases which were part of In re Initial Public Offering Securities Litigation. 671 F. Supp. 2d 467 (S.D.N.Y. 2009). That consolidated action involved separate suits filed against issuers and underwriters of 309 initial public offerings made during the late 1990s. Id. at 470. Each individual case had its own lead plaintiff, but settlement negotiations were largely handled by a six-member executive committee of law firms. Id. at 473 n.35. The executive committee negotiated a global settlement on behalf of plaintiffs in all of the consolidated cases. See id. at 473–74 (summarizing executive committee’s fee request). The attorneys’ fee request was based on the committee’s work as well as on the work of the firms in the individual actions. Id. The unique governance structure of these cases is inconsistent with the lead plaintiff mechanism, which is the focus of this study.

The only other cases omitted from the list of cases the Stanford Clearinghouse generated were cases without monetary settlements, cases that were not subject to the PSLRA, cases in which the court did not certify a class, and cases which were outside the study period.
Hand-collected, comprehensive data for each of these 431 cases was then obtained by examining the following documents in the court record for each case: (1) all motions for appointment of lead plaintiff and lead counsel, along with all supporting memoranda, affidavits, declarations, and other documents; (2) court orders appointing lead plaintiffs and approving lead counsel; (3) all motions and supporting memoranda, affidavits, declarations, and other documents requesting an award of attorneys’ fees; (4) any filed objections to the settlement, including objections to the proposed attorneys’ fees; (5) any filed responses to the objections, and any related court orders; (6) court orders granting final approval to the settlement; and (7) court orders awarding attorneys’ fees.

In these documents for each case, we looked for evidence that the lead plaintiff candidates had negotiated fee agreements with their chosen counsel at the outset of the litigation. We also examined whether judges considered the existence or terms of an ex ante fee agreement when appointing the lead plaintiff, approving a lead plaintiff’s proposed class counsel, or awarding fees to class counsel at the end of the litigation. We then coded the characteristics of the parties that were ultimately chosen as lead plaintiffs. Finally, we looked at the arguments lawyers offered to justify their fee requests, the bases for any objections to the fee request, and the explanations judges gave for their decisions.

B. Empirical Findings

Section II.B presents this Article’s empirical findings below in two parts. Section II.B.1 first sets out various descriptive statistics summa­rizing the examination of the actual court filings in all of the 431 cases in the six-year data set. Section II.B.2 then presents and discusses a series of regression analyses aimed at providing a better understanding of how ex ante fee agreements, the participation of public pension funds, and judicial experience with securities class actions are correlated with fee requests and awards.

1. Descriptive Statistics. — This Article’s nationwide data set contains a wealth of new information about the operation “on the ground” of the fee-award process in securities class actions. Section II.B.1 summarizes and discusses that information below (and in Tables 1 and 2) with a focus on three areas: the lead plaintiff selection process, fee requests and awards, and judicial fee reductions.

Table 1: Summary Statistics (Continuous Variables) 78 Settlement data are in constant 2012 dollars.

Obs. Mean Std. Dev. Median Min Max
Fee Request 431 0.246 0.072 0.250 0.000 0.340
Fee Award 431 0.238 0.073 0.250 0.000 0.340
Settlement (Millions) 431 52.20 165.00 10.70 0.352 2,420.0
Ratio Award to Request 431 0.968 0.094 1.000 0.253 1.018
Judge Frequency 431 2.807 1.981 2.000 1.000 8.000
Case Age (Years) 431 4.008 2.120 3.514 1.006 13.778

Table 2: Summary Statistics (Categorical Variables) 79 The frequencies for lead plaintiff types exceed 100% because cases often feature more than one type of lead plaintiff.

Obs. Yes % No %
LP Competition 431 305 70.77 126 29.23
Ex Ante Agreement Cited in LP Motion 425 48 11.29 377 88.71
Ex Ante Agreement Cited in LP Order 430 21 4.88 409 95.12
Ex Ante Agreement Cited in Fee Request 431 78 18.10 353 81.90
Ex Ante Agreement Cited in Fee Award 431 30 6.96 401 93.04
Public Pension LP 431 148 34.34 283 65.66
Other Institution LP 431 155 35.96 276 64.04
Individual LP 431 196 45.48 235 54.52
Fee Request Uses Lodestar Cross-Check 430 397 92.33 33 7.67
High Volume District 431 209 48.49 222 51.51
High Volume Judge 431 136 31.55 295 68.45
Objection to Fee 431 99 22.97 332 77.03
Lead Plaintiff Supports Requested Fee 430 149 34.65 281 65.35
Fee Award Uses Lodestar Cross-Check 431 194 45.01 237 54.99
Fee Cut 431 62 14.39 369 85.61

a.  The Lead Plaintiff Selection Process. — The study’s analysis began by looking for cases in which proposed lead plaintiffs offered the court proof of the ex ante fee agreements they had negotiated. Although Congress and the drafters of the lead plaintiff mechanism seemed to anticipate that such agreements would be the norm, 80 See Weiss, supra note 26, at 551 (noting Weiss and Beckerman hoped institution­al investors would negotiate “arm’s length fee arrangements with plaintiffs’ attorneys”). there is little evidence that they play a significant role in a court’s selection of the lead plaintiff. There were very few cases—just 11.29% 81 In the Global Crossing litigation, one lead plaintiff candidate, Staro Asset Management (“Staro”), mentioned in its moving papers that it had negotiated favorable fee terms with its chosen counsel. See Memorandum of Law in Support of Motion of Staro Asset Management, LLC for Appointment as Lead Plaintiff on Behalf of the Cumulative Convertible Preferred Stock Purchaser Class and for Approval of Selection of Lead Counsel at 14, Manson v. Winnick, No. 1:02-cv-00910 (S.D.N.Y. Apr. 26, 2002) (on file with the Columbia Law Review) (“Staro carefully negotiated a highly favorable, below market rate, fee agreement with its chosen Lead Counsel, for the benefit of the preferred stock securities class. Staro will present this fee agreement to the Court in camera should the Court wish to see it.”). The Global Crossing litigation produced five separate settlements, each of which yielded a fee award. In the data set, the unit of analysis is the individual settlement. This structure means that if ex ante agreements are more common in cases with multiple, partial settlements, the paper’s methodology may overestimate the frequency of these agreements.

The Global Crossing litigation was also exceptional in that, although the judge did not appoint Staro to the lead plaintiff position, he did appoint Staro’s counsel to an executive committee with which class counsel was instructed to consult throughout the liti­gation. See Order at 5, In re Glob. Crossing, Ltd. Sec. and ERISA Litig., No. 02-MD-1472 (S.D.N.Y. Dec. 18, 2002) (on file with the Columbia Law Review) (“Counsel for Staro . . . shall serve as members of an Executive Committee of the Plaintiffs in the GC Securities Actions . . . . Lead Counsel shall consult with members of the Executive Committee, and shall keep members of the Executive Committee informed of significant developments in the litigation.”). This action was of questionable propriety. See In re Cendant Corp. Litig., 264 F.3d 201, 273–77 (3d Cir. 2001) (emphasizing lead plaintiff is to select lawyers to represent class and set their compensation terms, and trial judge must respect lead plaintiff’s decisions, when reasonable).
—in which the lead plaintiff candidate or the court discussed an ex ante agreement during the appointment process. 82 Of the 425 cases for which lead plaintiff moving papers were available, only 48 discussed the presence of an ex ante fee agreement. In the remaining 377 settlements, the lead plaintiff moving papers were not available either electronically or directly from the court. See infra Table 2. The study then looked at whether lawyers vary their approach to obtaining the lead plaintiff position for their clients, such as by referencing an ex ante fee agreement in the motion to be appointed lead plaintiff, depending on the frequency with which courts handle securities class actions. The study analyzes that question in several ways. First, the sample was segmented using an indicator variable (High Volume) with a value of 1 for cases litigated in one of the three districts with the largest number of securities class actions—the Central and Northern Districts of California, and the Southern District of New York—and a value of 0 otherwise. 83 The study utilizes these three districts because in the nationwide sample of settlements, these courts collectively accounted for 42.46% of the settlements. The Northern District of California heard the fewest (42 cases) of these three districts, but that was more than twice the number of cases as the next nearest court (the Northern District of Illinois, with 18 cases). Overall, lead plaintiff candidates discuss ex ante fee negotiations in 11.29% of cases—13.59% of the cases litigated in high-volume districts and 9.13% in other districts. 84 See infra Table 2 (reporting ex ante fee agreements present in 11.29% of cases). Although this difference suggests the possibility that counsel for lead plaintiff candidates use court-specific strategies in the lead plaintiff application process, the difference is not statistically significant. 85 The Pearson chi-square is 2.1074 (probability = 0.147).

Next, the study looks at judges’ experience with securities class actions. Because the study comprises data on all settlements during the six-year study period, the frequency with which any individual judge appears in the sample provides a rough measure of judicial experience with these cases. 86 To be sure, this measure is not perfect. It does not account for cases that were dismissed or otherwise terminated prior to a settlement. Nor does it include cases that a judge may have handled prior to the study period. The sample contains 246 individual judges; 160 of those judges contributed a single fee decision, but the number of decisions per judge ranges from 1 to 8. On average, each judge contri­buted 2.81 decisions to the data set, with a standard deviation of 1.98. Based on these data, we created a variable (High Volume Judge) that takes a value of 1 if a judge had four or more fee decisions in our sample and 0 otherwise. 87 One standard deviation above the mean number of opinions is 4.79. Setting the value at 4 rather than 5 was necessary to ensure that the high-volume judges were not just those sitting in high-volume districts. Discussion of fee arrangements in the lead plaintiff applica­tion process is significantly more frequent in cases handled by high-volume judges (16.42%) than in cases handled by other judges (8.93%), 88 The Pearson chi-square is 5.1284 (probability = 0.024). suggesting that lawyers may modify their strategy for obtaining lead counsel positions depending on the judge who hears the case. That is, lawyers may be more willing to discuss, or at least mention, fee arrange­ments in their filings when a high-volume judge is handling the case, presumably because they expect that information to strengthen the case for appointment of the lead plaintiff candidate that they represent (and therefore also their own appointment as lead counsel).

The study also analyzed the relationship between competition for the lead plaintiff position and the frequency with which lead plaintiff candidates discuss ex ante fee agreements. Overall, there is a good deal of competition to capture the lead plaintiff position. Overall, 71% of the cases had more than one motion to be appointed lead plaintiff, with a mean (median) of 3.22 (3) motions per case. One case had seventeen different applications for appointment as lead plaintiff. 89 In re Am. Home Mortg. Inv. Corp. Sec. Litig., No. 07-MD-01898 (E.D.N.Y. July 31, 2007) (on file with authors). Not surprisingly, the cases with competition turn out to yield significantly larger settlements, suggesting that prospective lead counsel may have the ability to identify the more lucrative or otherwise higher quality cases at the earliest stages of the litigation. Cases with multiple lead plaintiff motions had mean (median) settlements of $65.0 million ($12.0 million), compared to only $21.2 million ($7.5 million) for the cases with only one motion. 90 The difference in means is significant at less than 5%: t = -2.5151; probability = 0.0123. Competition for the lead plaintiff position is also correlated with a higher frequency of discussion of an ex ante fee agreement in the lead plaintiff moving papers. Where there was no competition, fee arrangements were discussed in only 4.00% of the cases, compared to 14.33% of the cases with competition. 91 While this difference is statistically significant, it is worth emphasizing that this finding does not imply that increased competition causes lawyers to compete on price. The Pearson chi-square is 9.4041 (probability = 0.002). After all, the presence of multiple lead plaintiff applications increases the likelihood that at least one of those applications will mention fees. The study found no evidence that competition for the role of lead plaintiff is more frequent in high-volume districts.

While evidence of ex ante fee agreements in the moving papers was sparse, there was even less evidence that judges actually considered ex ante fee agreements when awarding the lead counsel position. In only 4.88% of the cases did the court’s order appointing the lead plaintiff and class counsel make any mention of such an agreement. There were no significant differences between high- and low-volume districts, but high-volume judges were twice as likely (7.35% to 3.74%) as other judges to mention ex ante agreements in their selection decisions. 92 This difference in frequencies is not significant. The Pearson chi-square is 2.6108 (probability = 0.106). Judges are much more likely to focus on fee arrangements when the lawyers have raised them, but discussion in the moving papers is by no means a guarantee that courts will address fees when appointing the lead plaintiff. Just 25% of the orders appointing the lead plaintiff mentioned fee arrangements in cases where fees were discussed in the lead plaintiff motion papers, compared to only 2.39% of fee orders in the remaining sample. 93 This difference is significant. The Pearson chi-square is 46.3529 (probability < 0.001). Thus, in the vast majority of cases, courts seemingly did not consider the presence of a negotiated fee arrangement to be pertinent to the lead plaintiff and lead counsel appointment decisions. We speculate that this result may be the product of the PSLRA, which explicitly makes the size of the lead plaintiff’s stake in the case the overwhelming consideration for the court when appointing the lead plaintiff. 94 See 15 U.S.C. § 77z-1(a)(3)(B)(iii)(I) (2012) (“[T]he court shall adopt a presumption that the most adequate plaintiff . . . is the person or group of persons that . . . in the determination of the court, has the largest financial interest in the relief sought by the class . . . .”); id. § 78u-4(a)(3)(B)(iii)(I) (“[T]he court shall adopt a presumption that the most adequate plaintiff in any private action arising under this chapter is the person or group of persons that . . . has the largest financial interest in the relief sought by the class . . . .”).

While the courts discussed such ex ante fee agreements in only a few cases, those that did emphasized that the presence of such an agreement is relevant, either with respect to whether the lead plaintiff satisfies the adequacy prong of Rule 23 or to whether the court should approve lead plaintiff’s proposed lead counsel. For example, the judge in Taubenfeld v. Career Education Corp. appointed the candidate with the largest financial interest as lead plaintiff, but noted that the plaintiff’s moving papers did not discuss its chosen counsel’s proposed fee structure. 95 See Taubenfeld v. Career Educ. Corp., No. 03 C 8884, 2004 WL 554810, at *5 (N.D. Ill. Mar. 19, 2004) (appointing as lead plaintiff movant with “largest financial interest in the relief sought by the class . . . [who also] satisfies the requirements of Rule 23,” but reserving ruling on appointment of lead counsel until lead plaintiff’s chosen firm “submit[s] to the court information concerning its proposed fee structure”); see also In re Motorola Sec. Litig., No. 03 C 287, 2003 WL 21673928, at *7 (N.D. Ill. July 16, 2003) (appointing lead counsel where proposed fees did “not appear unreasonable”); Johnson v. Tellabs, Inc., 214 F.R.D. 225, 229 (N.D. Ill. 2002) (approving proposed lead counsel where proposed fees did not appear unreasonable). As this was “significant information the court would like to review prior to making a determination on lead counsel,” the court reserved its decision appointing lead counsel until the law firm submitted the requested fee information. 96 Taubenfeld, 2004 WL 554810, at *5; see also Mayo v. Apropos Tech., Inc., No. 01 C 8406, 2002 WL 193393, at *5–6 (N.D. Ill. Feb. 7, 2002) (requesting potential lead plaintiff to submit information on, among other things, “agreed fees”). Unfortunately, there are too few instances of this kind of judicial demand for additional information in the data set to determine whether they are correlated with lower fee requests or awards. The simple fact is that most courts, perhaps because of time or other constraints, appear to be indifferent to such evidence when they make their lead plaintiff selections.

b. Fee Requests and Awards. — The data show that ex ante fee agree­ments are more frequently invoked when the lead counsel applies for a fee award than when they seek appointment as lead counsel. Overall, in 18.10% of the cases, lead counsel argued that the presence of a nego­tiated ex ante agreement with the lead plaintiff justified its requested fee. 97 That is, 78 of 431 cases. See infra Table 2. Although this argument was more frequent in high- versus low-volume districts (19.14% versus 17.12%), the difference was not statisti­cally significant. 98 The Pearson chi-square is 0.2968 (probability = 0.586). Lead counsel was significantly more likely to make this argument in front of a high-volume judge (23.53%) than a low-volume judge (15.59%). 99 The Pearson chi-square is 3.9554 (probability = 0.047). Consistent with the view that sophisticated institutional investors make the best monitors of class counsel, evidence in the record of an ex ante fee agreement is most prevalent when a public pension fund is the lead plaintiff. Evidence of such an agreement was present in 27.70% of the cases with public pension lead plaintiffs compared to just 13.07% of cases with other lead plaintiff types. This difference was significant. 100 The Pearson chi-square is 14.0299 (probability < 0.001).

Somewhat more frequently (34.65% of the cases), attorneys noted in their fee-request filings that the lead plaintiff supported the requested fee, often without any discussion of whether that support was premised on an ex ante fee agreement. In many cases, the lead plaintiff’s support came solely as a result of an ex post evaluation of the fairness of the fee. In such a case, the lead plaintiff or its representative would typically file a supporting declaration indicating that it thought the requested fee was fair based on the work performed and the result achieved. It is possible that the variables in our regressions (discussed below) for ex ante fee agreements and public pension plaintiffs might not be measuring separate effects. Plaintiffs’ attorneys said in informal interviews that, in their experience, some form of ex ante fee agreement is present in virtually all cases with public pension funds. 101 See Baker et al., supra note 4, at 1694, 1701 (examining fee-setting practices). For that reason, we have speculated that these variables may be “measuring some variation of the same thing—the impact that a sophisticated and engaged lead plaintiff has on fee requests.” 102 Id. at 1701.

Some evidence from this Article’s current, nationwide study, however, is at odds with that anecdotal evidence. In cases with both public pension lead plaintiffs and evidence of an ex ante fee agreement, the average fee request is 13.31%. By contrast, in cases with a public pension lead plaintiff but no evidence of an ex ante fee agreement, requests are significantly higher, averaging 22.42%. 103 The t-statistic is 7.4212 (probability < 0.0001). See infra Tables 2 & 4. These data suggest that fee agreements negotiated at the beginning of cases have a substantial moderating effect on fee requests, even when cases are under the control of sophisticated institutional investors. To put the matter another way, public pension funds may come in two types—those that act aggressively to reduce agency costs and those that do not—and the presence of an ex ante fee agreement may help distinguish one from the other. Relatedly, the absence of an ex ante agreement may also correlate with instances of pay-to-play, a practice known to weaken the enthusiasm of public pension funds for bargaining hard with attorneys over fees. 104 See Stephen J. Choi, Drew T. Johnson-Skinner & A.C. Pritchard, The Price of Pay to Play in Securities Class Actions, 8 J. Empirical Legal Stud. 650, 651 (2011) (examining influence of law-firm campaign contributions on class-counsel selection in cases where lead plaintiff was public pension fund headed by elected official).

Overall, the mean (median) fee request in the sample of cases is 24.60% (25.00%). In cases without ex ante agreements, overall fee requests are 26.07%, compared to 17.93% in cases with evidence of ex ante agreements. These differences are significant. 105 The t-statistic is 10.0697 (probability < 0.0001). The same pattern exists with respect to fee awards. In cases without an ex ante agreement, fee awards averaged 25.12%, compared to 17.62% in cases with evidence of an ex ante agreement. 106 Here, too, the differences are significant. The t-statistic is 8.9813 (probability < 0.0001). Of course, these simple comparisons cannot tell us whether it is the presence of an ex ante fee agreement that is driving this result. Inflation-adjusted settlements, for example, are significantly larger in the cases with ex ante agreements ($136 million) than in the cases without such agreements ($33.8 million). 107 The t-statistic is -5.0576 (probability < 0.0001). Fee requests and awards measured as a percentage of settlement tend to decline as settlement size increases, so it is possible that the difference in the size of the settlements explains the difference in fees.

To get a better sense of how settlement size and the presence of an ex ante fee agreement might be correlated with fee requests and awards, Figure 1 depicts fee requests and awards as a percentage of the settle­ment. Figure 1 shows that the majority of fee requests are clustered into one of three values—25%, 30%, or 33.33%. A 25% fee request is the mode for the sample (occurring in 26.22% of the cases), an unsurprising result given the number of courts, particularly in the Ninth Circuit, that explicitly use 25% as a benchmark for awarding fees. 108 See, e.g., Hanlon v. Chrysler Corp., 150 F.3d 1011, 1029 (9th Cir. 1998) (“This circuit has established 25% of the common fund as a benchmark award for attorney fees.”); Camden I Condo. Ass’n v. Dunkle, 946 F.2d 768, 775 (11th Cir. 1991) (“[D]istrict courts are beginning to view . . . 25% as a ‘bench mark’ percentage fee award which may be adjusted in accordance with the individual circumstances of each case . . . .”). The vast majority (74.36%) of fee requests in cases with ex ante agreements, however, are below 25%. The slopes for the predicted values in cases with and without agreements are close to parallel, which suggests that the effect of an ex ante fee agreement does not vary with the settlement size.

Figure 1: Fee Requests and Awards in Cases with and without Ex Ante Agreements 109 Settlements are log-transformed and in constant 2012 dollars.


There is substantial evidence that the court’s familiarity with securities class actions makes a difference not only with respect to fee awards, but also with respect to fee requests. Experience, in other words, matters both in terms of what lawyers ask for and what courts give them. These differences become particularly pronounced if one looks jointly at both judicial experience and settlement size. As settlement size increases, lawyers ask for and are awarded significantly larger fees in those districts that see fewer securities class actions. Fee requests averaged 26.30% in the low-volume districts compared to just 22.79% in the high-volume districts. 110 The t-statistic is 5.2377 (probability < .0001). Overall, fee awards were significantly higher (25.73%) in the low-volume districts than in the high-volume ones (21.67%). 111 The t-statistic is 6.0344 (probability < .0001). In the top quartile of settlements in our nationwide study, average fee requests in the low-volume districts were 23.10% compared to 18.50% in the high-volume districts. 112 The t-statistic is 3.4859 (probability = 0.0007). Fee awards for those cases average 22.41% in the low-volume districts versus 17.46% in the high-volume districts. These differences are also significant. 113 The t-statistic is 3.7891 (probability = 0.0003).

Figure 2 provides a starker illustration of how fee-request practices vary between high- and low-volume districts. To understand the signifi­cance of these results, it is important to emphasize that one of the most robust findings of existing empirical research is that awards of attorneys’ fees exhibit scale effects—as settlement size increases, fee percentages decline. 114 See, e.g., Eisenberg & Miller, Empirical Study, supra note 4, at 54 (finding “[a]s client recovery increases, the fee percent decreases”); Fitzpatrick, Settlements, supra note 1, at 837 (“Regression analysis . . . confirms that after controlling for other variables, fee percentage is strongly and inversely associated with settlement size among all cases, among securities cases, and among all nonsecurities cases.”); Perino, Activism Through Litigation, supra note 4, at 388 tbl.4 (presenting results of regression analysis showing inverse correlation between fee award percentage and settlement size). This proposition is also generally accepted by courts. See Silverman v. Motorola Sols., Inc., 739 F.3d 956, 959 (7th Cir. 2013) (“Negotiated fee agreements regularly provide for a recovery that increases at a decreasing rate.”). Until now, most commentators simply assumed that these scale effects were uniform across districts and judges. But as Figure 2 illustrates, that is simply not the case. Fee requests and awards are again shown as a percentage of the settlement. In low-volume districts, Figure 2 shows that, on average, fee requests are less sensitive to settlement size, with requests declining only slightly as settlements increase. There are, to be sure, some low fee requests (defined as requests below 20%) in the low-volume districts, but those cases overwhelmingly have either a public pension lead plaintiff (71.22% of the cases) or evidence of an ex ante fee agreement (53.57% of the cases). In low-volume districts, there were 132 cases that had neither a public pension lead plaintiff nor evidence of an ex ante fee agreement. Only five of those cases (3.79%) had fee requests below 20%.

Figure 2: Fee Requests and Awards in Cases in Low and High Volume Districts 115 Settlements are log-transformed and in constant 2012 dollars.


By contrast, in high-volume districts, the downward slope of the fitted values line is steeper. 116 In high-volume districts, the correlation between log settlement values and fee requests is -0.4689, compared to -0.3077 in low-volume districts. In these districts, lawyers ask for substantially lower percentage fees as settlement size increases than in the low-volume districts. Again, however, these simple comparisons of means cannot tell one whether this effect is really associated with the volume of securities cases settled in the high-volume districts or is instead the result of the kinds of securities cases that are brought in those districts. For example, it remains possible that sophisticated lead plaintiffs and ex ante fee agreements are driving this result. Over 75% of the cases in high-volume districts with fee requests below 20% involved public pension lead plaintiffs. Just under 50% of such cases had evidence of an ex ante fee agreement.

A similar pattern emerges when we compare high- to low-volume judges (Figure 3). In both sub-samples, there are scale effects, but the rate of decline is much sharper for both requests and awards when a high-volume judge hears the case. Here, too, caution is prudent because other case characteristics could be driving these results. Nonetheless, the picture that is beginning to emerge is one that is far more nuanced than indicated in previous studies. 117 See supra note 114 (citing previous studies). Not surprisingly, it appears that lawyers behave strategically based on the court in which they are litigating and the particular judge before whom they are appearing. It has certainly been known for years that courts and judges are not monolithic when it comes to their fee awards. 118 See Theodore Eisenberg, Geoffrey Miller & Michael A. Perino, A New Look at Judicial Impact: Attorneys’ Fees in Securities Class Actions After Goldberger v. Integrated Resources, Inc., 29 Wash. U. J.L. & Pol’y 5, 27–31 (2009) (identifying significant inter-circuit variation in fee award practices); Helland & Klick, supra note 4, at 181 (attributing differ­ences in fees to differences in court congestion). Some judges have earned reputations for vigorously scrutinizing fee requests and for awarding relatively low fees. 119 See In re HPL Techs. Inc. Sec. Litig., 366 F. Supp. 2d 912, 918 (N.D. Cal. 2005) (Walker, J.) (rejecting 15% fee after concluding “lodestar cross-check warrant[ed] a fee lower than that sought by lead counsel”). What is new, however, is that those differences are not attributable solely to the idiosyncrasies of individual judges. 120 But see infra section II.B.2.c (noting such personal characteristics likely remain important in understanding fee-setting process). The more experience courts or judges have with securities class actions, the more parsimonious they appear to become. But these phenomena are not simply the product of experienced courts or judges slashing fee requests more than their less experienced counterparts (a point we address in more detail below). Lawyers appear to be anticipating what the court is likely to do, asking for lower fees (particularly as cases get larger) when they appear in courts with large numbers of class actions or before high-frequency judges, and asking for higher fees in other situations.

Figure 3: Fee Requests and Awards in Cases with Low and High Volume Judges 121 Settlements are log-transformed and in constant 2012 dollars.


The fee request–award dynamic thus appears to be a complex version of the Ultimatum Game, a classic of behavioral economics 122 See Martin A. Nowak, Karen M. Page & Karl Sigmund, Fairness Versus Reason in the Ultimatum Game, 289 Science 1773, 1773 (2000) (“The Ultimatum Game is quickly catching up with the Prisoner’s Dilemma as a prime showpiece of apparently irrational behavior.”). in which two players decide how to divide a fixed sum of money. In the standard form of the game, the Proposer is asked to allocate a sum of money between herself and the other player, the Responder. The Responder may either accept the offer (in which case the money is split as proposed) or reject it. In the latter case, neither player receives any­thing. 123 See Christine Jolls, Cass R. Sunstein & Richard Thaler, A Behavioral Approach to Law and Economics, 50 Stan. L. Rev. 1471, 1489–90 (1998) (describing operation of Ultimatum Game); Richard H. Thaler, Anomalies: The Ultimatum Game, 2 J. Econ. Persp. 195, 195–96 (1988) (outlining hypothetical Ultimatum Game scenario). Under the traditional assumptions of classical economics, the Proposer should offer the Responder the smallest sum of money possible and the Responder should accept it because she will be better off than with the alternative (which is nothing). But repeated experi­ments have shown that Responders tend to reject offers that are too far outside what might be considered the objectively fair outcome, an even split. In the majority of studies, offers tend to average between 40% to 50% of the fixed total sum, with responders usually rejecting offers below 30%. 124 Ernst Fehr & Simon Gächter, Fairness and Retaliation: The Economics of Reciprocity, in Advances in Behavioral Economics 510, 512 (Colin F. Camerer et al. eds., 2004); Ernst Fehr & Klaus M. Schmidt, A Theory of Fairness, Competition and Cooperation, in Advances in Behavioral Economics, supra, at 271, 277. Proposers, in other words, appear to anticipate these reactions and so tend to offer divisions that fall within the likely acceptable range. 125 See Jolls, Sunstein & Thaler, supra note 123, at 1490–97 (“Responders are thus willing to punish unfair behavior, even at a financial cost to themselves . . . . [T]his response seems to be expected and anticipated by Proposers; they typically offer a substantial portion of the sum to be divided—ordinarily forty to fifty percent.”). The reason why Proposers make offers larger than what would be predicted under the classic economic model remains something of a mystery. They might “have a taste for fairness,” Thaler, supra note 123, at 197, or they might simply be acting strategically to avoid the possibility that low offers will be rejected. See Richard A. Posner, Rational Choice, Behavioral Economics, and the Law, 50 Stan. L. Rev. 1551, 1564–65 (1998) (“To gain anything from playing the game, the proposer has to make an offer generous enough to induce the respondent to accept.”). The latter explanation is likely the better one.

An attorney’s decision about what fee to request in a class action shares some basic characteristics with the Ultimatum Game. 126 For a discussion of the application of the Ultimatum Game to contingent fees more generally, see Eyal Zamir & Ilana Ritov, Notions of Fairness and Contingent Fees, 74 Law & Contemp. Probs. 1, 9–10 (2011) (discussing key similarities and differences between Ultimatum Game and contingent-fee situations). As in the game, the attorney must propose a division of a fixed sum of money. There are two key differences. The first is that in the absence of an ex ante fee agreement, the court rather than the lead plaintiff will play the role of the Responder, although in doing so it is supposed to be acting in the best interests of the class. 127 In the Ultimatum Game, the Responder bears a cost for rejecting the proposed allocation because she gets nothing under those circumstances. The court can also be thought to bear a cost when it rejects a fee request because the judge will now have to write an opinion explaining why the proposed fee was unreasonable. The willingness to accept such costs is important for judges in high-volume districts because it will allow them to develop a reputation for accepting only fair fee requests. See Nowak et al., supra note 122, at 1774 (noting “individuals who accept low offers run the risk of receiving reduced offers in the future” whereas “the costly act of rejecting a low offer buys the reputation that one accepts only fair offers”). The second is that the court has the final decision about how the pot will be split. But the existence of an accepted range of fair values—in this case the existing precedents regarding fee awards—will tend to constrain the attorney’s decision about what fee to propose. Indeed, in variations on the Ultimatum Game in which Proposers have information about which offers have been accepted in the past, offers tend to converge toward those demonstrably acceptable values. 128 See id. (explaining where information on past offers is available and game is repeated, “evolutionary dynamics tend to favor strategies that demand and offer a fair share of the prize”). Rather than run the risk that a court will respond to a high demand by awarding an especially low fee, lawyers appear to make proposals that are within a range of values deemed to be fair, given the precedents in the relevant jurisdiction. The difference between high- and low-volume districts is that in the former there are likely to be more precedents and a tighter range of “fair values,” which constrain class counsel’s ability to request as much as the lawyers might want. In low-volume districts and with judges who handle these cases infrequently, there are fewer precedents, allowing attorneys to pitch their requests at the higher end of the perceived “fair value” range. Simply put, lawyers ask for as much as they think they can reasonably get, but what they think they can get will vary depending on where and before whom a case is pending.

This dynamic changes when a sophisticated institutional investor serves as the lead plaintiff. Particularly where that lead plaintiff negotiates fee terms ex ante with retained counsel, the lead plaintiff has the ability to engage in real arm’s-length negotiations, rather than simply responding to the lawyer’s fee proposal. 129 See Jill E. Fisch, Aggregation, Auctions, and Other Developments in the Selection of Lead Counsel Under the PSLRA, 64 Law & Contemp. Probs. 53, 62 (2001) (“Institutions are developing ongoing relationships with plaintiffs’ firms and increasing sophistication in evaluating and negotiating fee arrangements.”); Jill E. Fisch, Lawyers on the Auction Block: Evaluating the Selection of Class Counsel by Auction, 102 Colum. L. Rev. 650, 705–06 (2002) [hereinafter Fisch, Auction Block] (discussing how institutional lead plaintiffs bargain over fees when selecting prospective lead counsel). In the former case, the parties are likely to take into account the anticipated risks associated with the case and will have incentives to design fee structures that maximize the plaintiffs’ net recovery. The latter involves a zero-sum game—every dollar the judge gives the lawyers is a dollar out of the class members’ pockets. As we discuss in Part III, there are substantial reasons to believe that, given the dynamics and uncertainties inherent in class-action litigation, sophisticated lead plaintiffs who negotiate fees with their counsel at the start of litigation are better suited to setting fees appropriately—that is, to paying the “market rate”—than are judges who simply respond to fee requests after the substantive aspects of the case are complete.

C. Judicial Fee Reductions. — If we anticipate that lawyers will moderate their fee requests when they appear in high-volume districts or before high-volume judges, then it is difficult to predict the frequency with which different courts will cut requested fees. On one level, fee cuts might in general be quite infrequent. This would be so if lawyers accurately anticipate the amounts judges are willing to award and tailor their fee requests accordingly. One might also expect fee cuts to be less frequent among experienced judges and in experienced courts because lawyers’ predictions will be better informed. Alternatively, because a low-volume court has comparatively little basis for comparison when presented with a given fee request, it might be inclined to find reasonable a fee request that a high-volume court would consider “high” or “too high.” As we noted previously, fee requests are typically non-adversarial. 130 See supra Part I (discussing non-adversarial nature of fee requests). Defendants are indifferent to fee awards because they are paid out of the common fund and objectors tend to be infrequent. As a result, lawyers requesting fees may be inclined to skew their briefs toward the higher end of the range of perceived “fair values.” Under these circumstances, a court that sees securities class actions infrequently might cut fees only rarely.

As it turns out, fee-cut decisions do present something of a mixed bag. As shown in Table 2, in 14.39% of cases (that is, in 62 of the 431 cases in the data set), the court awarded a smaller fee than lead counsel requested. 131 See infra Table 2 (reporting courts cut fees in 62 of 431 cases for which docu­ments were available online). Put another way, in 6 out of 7 cases, plaintiffs’ counsel received precisely the fee requested. From this perspective, courts seem to take a light touch, but that conclusion needs to be placed in context. Concerns about judicial fee cuts have less to do with their frequency than with the circumstances under which they occur.

Moreover, judicial aggressiveness when reviewing fee requests often comes from within. In only 19 of the 62 cases in which fees were reduced (30.65%) did an objector formally challenge the size of the requested fee. 132 See infra Table 3. The objection rate was, to be sure, higher in the cases where the court cut the requested fee. Overall, objections occurred in only 22.97% of the cases in the study. See infra Table 2. But the difference in the frequency of objections between the cases where the court cut requested fees (30.65%) and the cases where it did not (21.68%) was insignificant. The Pearson chi-square is 2.411 (probability = 0.120). See infra Table 2. Thus, in 7 out of 10 cases in which the court cut the requested attorneys’ fees, the court did so sua sponte, without the lead plaintiff or any other class member questioning the size of the fee request.

Indeed, the willingness of judges to cut fees appears to be unrelated to actual objections from class members to proposed fees. If we measure the ratio of award to request, we find that for the overall sample the mean is 0.9683. 133 The median is 1.0, see infra Table 1. That is, on average, lawyers could expect to get 97 cents for every dollar they requested in fees. 134 If one relies on the median of 1.0, lawyers could expect to get 100 cents on the dollar in fee awards. In the subset of cases in which the court cut fees, the average fee award is actually higher when there is an objection to the fee request (80 cents on the dollar) than when there is no objection (77 cents on the dollar), although the difference is not statistically significant. 135 See infra Table 1 & Appendix. This is a ratio of award to request of 0.8035 when there is an objection to the fee request, compared to 0.7737 when there is no objection. The t-statistic is -0.7413 (probability = 0.4614). The requested fees in the 62 fee-cut cases ranged from 6.54% to 33.33% of the relevant recovery, with the fee awards ranging from 6.29% to 30.00%. The mean (median) ratio of award to request in these cases was 0.7828 (0.800), with a range from 0.2533 to 0.9908. The data also provide mixed evidence about the deference that courts show to large institutional lead plaintiffs. Based simply on broad descriptive statistics, we found that fee cuts were neither more prevalent when an individual was the lead plaintiff in the case nor less prevalent when a public pension fund was the lead plaintiff, two relationships that might be expected given that the PSLRA encourages courts to defer to institutional investors. Individuals were lead plaintiffs in 40.32% of the cases with fee cuts. 136 The other lead plaintiffs in these fee-cut cases were public pension funds (27.42%) and other institutions, including union funds and private institutions (32.26%). See infra Appendix. These differences between individuals and other lead plaintiff types were not significant. 137 The Pearson chi-square is 1.7160 (probability = 0.424). See infra Table 2. There does seem to be, however, some relationship between the type of lead plaintiff and the size of the fee cut. The mean ratio of award to request is significantly higher in cases with a public pension lead plaintiff (0.9821) than in cases with other lead plaintiff types (0.9611), 138 The t-statistic is -2.1997 (probability = 0.0284). See infra Table 1 & Appendix. indicating that courts cut fees less in cases with public pension funds. This deference is consistent with the general view among academics that public pension funds, which have large portfolios, few conflicts of interest, and a high degree of sophistication, are best suited to serve as monitors in securities class actions. 139 See Joseph A. Grundfest & Michael A. Perino, The Pentium Papers: A Case Study of Collective Institutional Investor Activism in Litigation, 38 Ariz. L. Rev. 559, 578–79 (1996) (arguing large portfolios enable major mutual and pension funds to take “broader view” of systemic concerns and avoid collective-action problems that might threaten individual investors); Perino, Activism Through Litigation, supra note 4, at 373–74 (noting features of public pension funds suitable to effective monitoring); Weiss and Beckerman, supra note 26, at 2111 (discussing public pension funds’ ability to bear cost of activism on corporate governance issues).

We also found limited evidence, again based on simple frequency comparisons, that courts show greater deference to fee requests that were the product of ex ante fee agreements. Judges did not award the request­ed fee in 8.97% of the settlements with ex ante fee agreements compared to 15.58% of the cases without such agreements. Although fee cuts were twice as likely in cases without ex ante agreements, the difference in frequencies was not statistically significant. 140 The Pearson chi-square is 2.2639 (probability = 0.132). Within the fee-cut cases, there was no significant difference in the ratio of award to request for cases with ex ante fee agreements (0.8210) and those without (0.7779). The same was largely true across the entire sample. Overall, lawyers were awarded 98.42% of their requested fees in cases with ex ante fee agreements compared to 96.48% in cases without such agreements, a difference that was significant at about the 10% level. 141 The Pearson chi-square is -1.6454 (probability = 0.1006). Nor is there evidence that the lead plaintiff’s expressing support of the requested fee made a difference in the likelihood of a judicial fee cut. Cuts were about as likely in the cases with such support (12.75%) as in those without it (15.30%). 142 These slight differences were not statistically significant. The Pearson chi-square is 0.5134 (probability = 0.474).

As in the other aspects the study examined, however, there is some evidence of inter-court and inter-judge variation. High-volume districts appear to scrutinize fee requests somewhat more vigorously than low-volume districts. To be sure, the overall level of judicial scrutiny remains quite low. The average ratio of award to request in high-volume districts (0.9571) is significantly lower than in low-volume districts (0.9789). 143 The t-statistic is 2.4252 (probability = 0.0157). Most striking is the fact that fee cuts are almost twice as likely in the high-volume districts (18.66% of cases) as in the low-volume districts (10.36% of cases). 144 The Pearson chi-square is 6.0215 (probability = 0.014). See infra Table 4. A similar pattern emerges for the judges who most frequently rule on fee requests. Fee cuts are more likely among high-volume judges (19.12%) than among low-volume judges (12.20%), although the difference is significant at only the 10% level. 145 The Pearson chi-square is 3.6134 (probability = 0.057). High-volume judges also have lower ratios of award to request (0.9580) than low-volume judges (0.9731), but the difference is not significant. Indeed, in the sub-sample of cases with fee cuts, there are no significant differences in the size of cuts among high-volume and low-volume districts or among judges who see these cases more or less frequently.

Table 3: Rationales for Judicial Fee Reductions

Rationale Number (Frequency)
The requested fee is “too large” 25 (40.32%)
The requested fee is “too large given the work performed by the attorneys” 22 (35.48%)
The requested fee is “too large given lead counsel’s actual risk of non-recovery” 19 (30.65%)
Requested fee is “out of line with fees in similar cases” 19 (30.65%)
Requested fee fails a lodestar cross-check 20 (32.26%)
The court cannot rely on the market for setting attorneys’ fees 3 (4.84%)
Requested fee not the result of arm’s-length bargaining 1 (1.61%)

There does appear to be some difference, however, between the deference that high-volume judges give to ex ante fee agreements compared to their low-volume counterparts. In only 2 of 32 cases (6.25%) in which a high-volume judge was evaluating a fee request that was the product of an ex ante agreement did that judge reduce the requested fee. By comparison, low-volume judges reduced requested fees in 5 of 46 cases with ex ante fee agreements (10.87%). Although the numbers are too small to demonstrate statistical significance, they suggest that judges who see more securities class actions may defer more readily to negotiated fee agreements.

Finally, we examined the rationales courts offered for their decisions to reduce the requested fees. Given that fee awards raise significant policy concerns regarding the structure of the civil litigation system and its deterrent effects, and given the importance of fee awards to lawyers and class members, one might hope that judges would give some explanation for their decisions to cut the requested fees. That is simply not the case. In 40.3% (25) of the 62 fee-reduction cases, the court gave no reason or justification at all for its decision to cut the fee; it simply awarded a lower percentage or dollar amount than that requested by lead counsel. 146 See infra Table 3. In the remaining cases, the court gave one or more reasons for the fee reduction, which are summarized in Table 3. For the most part, judges offered only the most cursory explanations for their decisions. When courts offered other explanations, such as the low risk the case entailed, or the low quality or excessive quantity of the work performed by the attorneys, these discussions were usually cursory at best. The typical decision was an unpublished order only a few pages in length. While hardly adequate, these decisions were often an improvement over the cases in which the court awarded the requested fee without any reduction. In many of those cases, courts simply signed a conclusory proposed order that class counsel had submitted with its fee application. In those cases, it is difficult to avoid the inference that the court simply rubber-stamped the requested fee with little if any independent evaluation or analysis.

Courts’ unwillingness or inability to explain why they award the fees they do leaves important questions unanswered. Are fee awards random or is there a pattern to the data? What factors do courts really consider in determining whether a fee request is reasonable? Do courts actually apply the factors they claim to consider? Are there systematic differences among courts or among judges in awarding fees? Do lawyers know about these differences and exploit them in their fee requests? What, if any, role do ex ante fee agreements play in fee awards? What factors explain judicial decisions to cut requested fees? The next section examines these questions in greater depth.

2. Regression Analyses. — To better understand how ex ante fee agreements, the participation of public pension funds, and judicial experience with securities class actions are correlated with fee requests and awards, we constructed a series of linear regressions with either the fee request or the fee award as the dependent variable. We included indicator variables for: (1) evidence of an ex ante fee agreement (Ex Ante); (2) whether there was competition for the lead plaintiff position (Competition); (3) the presence of a public pension fund (Public Pension) or other institutional investor (Other Institution) as lead plaintiff; (4) the age of the case in years from filing until the date the court awarded attorneys’ fees (Case Age); (5) whether the case was litigated in a high-volume district for securities class actions (High Volume); and (6) whether the case was litigated in front of a high-volume judge (High Volume Judge). As an alternative way of measuring judicial experience with securities class actions, we created an additional variable (Judge Frequency) which measures the actual number of fee decisions a judge has in our data set. In our regressions for fee awards, we also included indicator variables for: (1) whether an objection to the fee was filed (Fee Objection), and (2) whether the court cut the requested fee (Cut). All models include as an independent variable the log-transformed, inflation-adjusted settlement in the case. To give a clear sense of the magnitude of the effects these variables have, we centered the regressions at the mean settlement size in the database ($52.2 million). 147 To account for potential variation among circuits, the regressions reported here use robust standard errors, clustered by circuit. See Perino, Activism Through Litigation, supra note 4, at 381, 389 n.28 (using similar methodology).

To analyze the circumstances in which courts reduce the requested fee, we also constructed a logit model which takes Cut as a dependent variable. This regression employs independent variables that are largely observable by case participants when the fee request is made. Using these variables enables us to determine whether class counsel can make reliable predictions as to when fee cuts are likely. To a large extent, the variables are the same ones used in the regressions for fee awards and requests.

a. Fee Requests. — Although infrequent, ex ante fee agreements appear to have a powerful influence over fee requests. The effect of an ex ante fee agreement on a fee request is negative and significant, both statistically and economically, even controlling for other relevant variables. 148 See infra Table 4 (reporting negative and statistically significant effect of ex ante agreement and public pension fund on fee requests). At the mean settlement value, the average fee request in a case without an ex ante agreement is 29.8% of the settlement amount (or $15.56 million). By contrast, the mean fee request in cases with an ex ante agreement was 23.8% (or $12.42 million), a 20% reduction in the fee requested. The presence of a public pension fund has similar effects. Cases with public pension lead plaintiffs have fee requests that average 24.8%, a reduction of about $2.6 million at the mean settlement amount. Other types of lead plaintiffs, including union-affiliated pension funds and other kinds of private institutional investors, have no statistically significant correlation with fee requests, a result that is consistent with prior studies. 149 See, e.g., Perino, Activism Through Litigation, supra note 4, at 387 (“[N]either UNION nor OTHER INSTITUTION is significant so there is no evidence that the participation of these institutions has any influence on fee requests.”).

Table 4: Regressions for Fee Requests 150 Settlement data are in constant 2012 dollars, log-transformed and centered at the mean settlement value. Robust standard errors are in parentheses. The probability values are indicated as follows: *** p<0.01; ** p<0.05; and * p<0.1.

(1) (2) (3) (4) (5) (6) (7)
Settlement -0.006* -0.001 -0.001 -0.001 -0.001 -0.001 -0.001
(0.003) (0.004) (0.005) (0.004) (0.005) (0.005) (0.005)
Competition -0.012*** -0.012*** -0.012*** -0.012*** -0.013*** -0.013*** -0.014***
(0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002)
Public Pension -0.050*** -0.043*** -0.050*** -0.049*** -0.050*** -0.049*** -0.049***
(0.009) (0.012) (0.009) (0.009) (0.009) (0.009) (0.009)
Other Institution -0.003 -0.004 -0.003 -0.004 -0.003 -0.003 -0.003
(0.004) (0.003) (0.003) (0.003) (0.003) (0.004) (0.003)
Ex Ante Agreement -0.060*** -0.043*** -0.057*** -0.057*** -0.058*** -0.058*** -0.059***
(0.015) (0.010) (0.013) (0.012) (0.014) (0.014) (0.014)
High Volume District -0.028*** -0.041*** -0.035*** -0.023*** -0.038*** -0.032*** -0.044***
(0.005) (0.004) (0.005) (0.005) (0.006) (0.007) (0.008)
Case Age (Years) -0.003* -0.003* -0.003** -0.003** -0.003** -0.003** -0.003**
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Settlement x High Volume -0.010** -0.009** -0.009* -0.010** -0.009* -0.009**
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004)
Public Pension x Ex Ante -0.035
Judge Frequency -0.004*** -0.001
(0.001) (0.002)
High Volume x Judge Frequency -0.005**
High Volume Judge -0.014*** -0.003
(0.004) (0.006)
High Volume Judge x. High Volume -0.017*
One Opinion 0.003
One Opinion x High Volume 0.019**
Constant 0.298*** 0.305*** 0.316*** 0.310*** 0.311*** 0.309*** 0.309***
(0.005) (0.009) (0.010) (0.009) (0.010) (0.009) (0.012)
Observations 431 431 431 431 431 431 431
R-squared 0.430 0.449 0.453 0.456 0.448 0.450 0.447

Competition for the lead plaintiff position is also correlated with lower fee requests, although the effect is smaller than for either ex ante fee agreements or the presence of a public pension fund. On average, the fee requests in cases with such competition decline from 29.8% to 28.6%. Competition seems not to influence fee award size directly, however, because compensation terms do not influence judges’ decisions to appoint lead plaintiffs. Instead, competition may be a proxy for case quality. Higher quality encourages lawyers to seek control of class actions because it implies lower risk, and it induces them to compete on price for the same reason. 151 Model 1 also suggests that the age of the case has a minor impact on the fee request, but not in the predicted direction. See infra Table 4. Although one might expect the requested fee to be higher in cases that take longer, Case Age is actually negatively correlated with fee requests. Infra Table 4. For every additional year the case is litigated, the fee request declines by 0.3%. This association, however, is only significant at the 10% level.

Perhaps the most significant finding in Model 1 is that plaintiffs’ attorneys appear to adjust their fee requests based on the volume of securities class actions settled in the district court where a case is pending. The variable High Volume is negative and significant. All else being equal, fee requests in the Northern and Central Districts of California and in the Southern District of New York (which collectively see roughly half of all securities class actions) are, on average, 2.8% lower than requests in other districts. At the mean settlement value, the lead counsel in a case in a high-volume district asks for about $1.46 million less in fees than in a case filed in a low-volume district. 152 In an unreported model, the High Volume variable was replaced with indicator variables for each circuit, with the Second Circuit as the reference circuit. The results are consistent with the analysis in Model 1. In eight of the remaining eleven circuits, fee requests were either higher than in the Second Circuit or statistically indistinguishable. In five of those circuits (the Third, Fourth, Eighth, Tenth, and D.C. Circuits) those differences were significant. For the most part, these are the circuits that see the fewest securities class actions. See Stanford Law Sch. Securities Class Action Clearinghouse, [] (last visited Aug. 1, 2015) (reporting Fourth, Eighth, Tenth, and D.C. Circuits have had lowest levels of securities class-action filings since 1996).

The only circuits with significantly lower fee requests were the Seventh (1.7% lower fee requests on average) and Ninth (2.0% lower) Circuits. Id. Again, these results are largely consistent with the study’s High Volume variable. Two of the three districts defined as high-volume districts were within the Ninth Circuit. While the Seventh Circuit does not see a large number of securities class actions, it has developed a reputation for closely scrutinizing fee requests in other types of class actions. See In re Synthroid Mktg. Litig., 325 F.3d 974, 975–76, 979–80 (7th Cir. 2003) (reviewing district court’s fee award after remand and overturning portion of fee award that was lower than rate agreed in certain class members’ ex ante fee agreements); In re Synthroid Mktg. Litig., 264 F.3d 712, 718 (7th Cir. 2001) (overturning district court order capping class counsel’s fees at 10% and remanding with instruction to district court to “estimate the terms of the contract that private plaintiffs would have negotiated with their lawyers, had bargaining occurred at the outset of the case . . . when the risk of loss still existed”); In re Cont’l Ill. Sec. Litig., 962 F.2d 566, 568, 572 (7th Cir. 1992) (overturning district court’s fee award of only 50% of class counsel’s requested fees, noting aim “is to give the lawyer what he would have gotten in the way of a fee in an arm’s length negotiation, had one been feasible”). The Seventh Circuit also contains the district court (the Northern District of Illinois) that saw the fourth largest number of securities class-action settlements in this study’s sample.

To further examine this relationship, Model 2 includes two interactions—(1) between Mean Settlement and High Volume and (2) between Ex Ante and Public Pension—in order to test the combined effects of those variables on fee requests. With these interactions, the constant in Model 2 (0.305) represents the average fee request in a case litigated in a low-volume district without either an ex ante fee agreement or a public pension lead plaintiff. 153 The 95% confidence interval for the constant is 0.2864 to 0.3100. As noted previously, one of the most consistent results in studies of attorneys’ fees in class actions generally is that settlement size is strongly correlated with fee requests and awards. Researchers have found that as settlements increase, fee requests (measured as a percentage of the settlement) decrease. 154 See Baker et al., supra note 4, at 1695 (“Measured as a percentage of recovery, fee requests decline as settlements increase.”); Eisenberg & Miller, Empirical Study, supra note 4, at 250 (“Fees and costs both exhibit scale effects, with the percent of each decreasing as the class recovery amount increased.”); Fitzpatrick, Settlements, supra note 1, at 837 (“[A]fter controlling for other variables, fee percentage is strongly and inversely associated with settlement size among all cases . . . .”). The overwhelming empirical support for this proposition makes one result in Model 2 particularly notable. With an interaction term between settlement size and high-volume districts, the variable Settlement is now negative but insignificant. What does this result mean? In districts that see few securities class actions, fee requests do not appear to vary significantly with the size of the settlement. In other words, regardless of settlement size, when a lead counsel in a low-volume district is not constrained by either a negotiated ex ante fee agreement or an active public pension lead plaintiff, the counsel requests fees of around 30% of the settlement.

By contrast, in high-volume districts fee requests are significantly lower, averaging 26.4% at the mean settlement value, about a $2.1 million reduction in the requested fee. It is important to emphasize that this reduction in fee requests occurs even in the absence of a public pension lead plaintiff or an ex ante fee agreement, suggesting that it is the court with relatively greater experience handling securities class actions that is providing this moderating influence. This finding substantially modifies the existing understanding of the relationship between fee requests and settlement size in class action settlements. Previously, most scholars have concluded that class action fee requests declined uniformly across courts as settlement size increased. 155 See supra note 154 and accompanying text (surveying relevant scholarship). But the results here suggest both imperfect information (by courts and plaintiffs) in the market for fees and plaintiffs’ attorneys who seemingly seek to exploit that information asymmetry. 156 One could construe this finding as evidence of a bifurcated or stratified market for class-action attorneys, rather than as imperfect information by courts, if there were market-based reasons for the same law firm to charge a different percentage of the recovery when prosecuting a class action in one district court rather than another. But it is unclear what those justifications would be. The decline in fee requests associated with increasingly large settlements is driven almost entirely by the requests made to the subset of courts that see these cases most frequently. Outside of that handful of districts, lawyers appear to ask for uniform fees, regardless of settlement size.

How big is the difference in fee requests in high- versus low-volume districts when the settlements involved are substantially larger than average? This relationship can be seen in the interaction between Settlement and High Volume, which was negative and significant. 157 See infra Table 4. This result indicates that as settlement values increase, the distance between fee requests in low- versus high-volume districts continues to widen. To understand the magnitude of this effect, consider a one-standard-deviation increase in the size of the settlement, which would mean a settlement of approximately $217.2 million. In the absence of either a public pension lead plaintiff or an ex ante fee agreement, the average fee request in a low-volume district is 30.3%, virtually unchanged from the fee request for an average settlement. In the same case in a high-volume district, the average fee request is only 23.8%. At these averages, class counsel could be expected to ask for $14.1 million more in fees in the case litigated in the low-volume district. These differences suggest substantial, previously undocumented inefficiencies in the market for attorneys’ fees in securities class actions. Lawyers, who naturally act in accordance with their own financial best interests, appear to be aware of these inefficiencies and strategically adjust their fee requests to take advantage of them.

The question that naturally arises concerns the mechanism though which high-volume courts can have this moderating influence on fee requests. What prevents an attorney (at least in the absence of an ex ante fee agreement or a sophisticated public pension fund) from asking for the same fees in both high- and low-volume districts? One obvious way courts might influence fee requests is through their fee award prece­dents. For example, the Ninth Circuit explicitly uses a benchmark fee of 25% in securities class actions, 158 See, e.g., Hanlon v. Chrysler Corp., 150 F.3d 1011, 1029 (9th Cir. 1998) (noting Ninth Circuit has “established 25% of the common fund as a benchmark award for attorney fees”). a percentage that is similar to the nationwide average fee request we report. 159 The average for all cases in our sample was 24.6%. See infra Table 1. A benchmark, of course, works uniformly, regardless of settlement size, but precedents need not be so precise to have a moderating influence on fee requests. The Second Circuit, for example, does not explicitly use such a benchmark, 160 See Goldberger v. Integrated Res., Inc., 209 F.3d 43, 52 (2d Cir. 2000) (“We are . . . disturbed by the essential notion of a benchmark.”). but in numerous cases district courts there have awarded fees substantially below 25%, 161 In re Bristol-Myers Squibb Sec. Litig., 361 F. Supp. 2d 229, 231 (S.D.N.Y. 2005) (awarding attorneys’ fee of 4% of settlement). and the Court of Appeals has suggested that fees in securities cases are too high. 162 See Goldberger, 209 F.3d at 52 (questioning whether “fully informed group of plaintiffs able to negotiate collectively would routinely agree to pay their lawyers a fee of 25% of a multi-million dollar settlement”). As indicated earlier, a judge in a high-volume district is also more likely to have ready access to, or feel obliged to seek out and consider, the body of unpublished fee decisions from the same district. Those decisions may affect published decisions within the district, ultimately making it more difficult for class counsel to request fees that are substantially out of line with the fees other judges from the same court have awarded in larger cases.

The final question that Model 2 addresses is whether the effects that public pension lead plaintiffs and ex ante agreements have on fee requests are conditional, or are simply two different ways of measuring the same thing—the influence that an active, sophisticated lead plaintiff has on fee requests. 163 See Baker et al., supra note 4, at 1701 (observing from informal interviews with plaintiffs’ attorneys that “some form of ex ante agreement is present in virtually all cases with public pension lead plaintiffs”). With the interaction between Public Pension and Ex Ante included in the regression, the two separate individual variables now represent partial effects—the effect of a public pension lead plaintiff without an ex ante agreement and vice versa at mean settlement values. Both individual variables remain negative and significant in Model 2, with magnitudes that are identical. In the presence of either a public pension lead plaintiff or an ex ante agreement, fee requests average just under 26.2%, all else being equal. The interaction term in Model 2, while negative, is insignificant. In other words, there is no evidence of any greater effect on fee requests when the class action has both a negotiated ex ante fee agreement and a public pension lead plaintiff. Consequently, it appears that each of these two variables is indeed measuring the influence of an active, sophisticated lead plaintiff on fee requests. Together they suggest that this aspect of the PSLRA is working largely as Congress intended. In cases where a real plaintiff plays a meaningful role in selecting and retaining counsel, some of the agency costs typically associated with securities class actions are reduced.

Models 3 through 6 supplement the analysis of high-volume districts by looking at whether attorneys moderate their fee requests when they appear before judges with significant experience handling securities class actions. The descriptive statistics for the sample suggested precisely this kind of strategic behavior, and these regressions confirm those results. Whether judicial experience is measured as the total number of decisions the judge has in the data set (Models 3 and 4) or if instead by pooling high-volume judges (Models 5 and 6), the same dynamic emerges. The more experience an individual judge has with securities class actions, the lower the fees that class counsel requests. This phenomenon, however, appears to be limited to experienced judges in high-volume districts. In models with an interaction for high-volume districts and judges, the interaction is negative and significant while the variable that measures judicial experience is insignificant. This result suggests that lawyers put a great deal of care into structuring their fee requests. For the most experienced judges in high-volume districts, fee requests average 4.9% less than requests made to less experienced judges in low-volume districts. Experienced judges in low-volume districts (now represented by the Judge Frequency or High Volume Judge variables) see fee requests that are statistically indistinguishable from their colleagues who have seen fewer such cases.

Plaintiffs’ lawyers in securities class actions take seriously the deci­sion to lower their fee requests. They certainly do so in the presence of a public pension lead plaintiff or an ex ante fee agreement. But in situations where the lead plaintiff does not appear to be monitoring their behavior, attorneys ask for uniformly high fees regardless of settlement size. As section II.B.1.b shows, experienced judges from districts that see these cases frequently award lower fees. Attorneys appear to anticipate these awards and moderate their requests when they appear before these judges.

One way to see how limited the situations are in which attorneys will independently reduce their fee requests is to flip the analysis around to focus on the least experienced judges. Model 7 includes a variable (One Opinion) that takes a value of 1 if a judge has only one fee decision in the data set and 0 otherwise. Since that opinion must come from the same case, the lawyers effectively have no information about how that particular judge might award fees. 164 As noted previously, such information may be available to the extent that a judge issued fee decisions before our study period. As in previous models, Model 7 interacts this variable with the High Volume district variable. The fee requests made to judges from low-volume districts with only one decision are indistinguishable from the average fee request of about 31% made in a low-volume district. For experienced judges in high-volume districts, fee requests are on average 4.40% lower than fee requests in low-volume districts. For judges without an established track record in high-volume districts, however, fee requests are 1.90% higher than more experienced judges in those districts. Fee requests remain lower for inexperienced judges in high-volume districts than for judges in low-volume districts (presumably because of the abundance of precedents from other judges in the district supporting lower fees), but not as low as for judges from the high-volume district with a more established record. These findings support the hypothesis that plaintiffs’ attorneys adjust their fee requests according to both the relevant judge’s individual experience and the collective experience of the relevant court in litigating securities class actions.

b. Fee Awards. — The models in Table 5 examine how the same variables we analyzed with respect to fee requests affect fee awards. The descriptive statistics suggest that in a typical case the judge displays a relatively light touch when it comes to reducing the requested fees. Fee cuts occur in a minority of cases. While such cuts can be large when they occur, on average lawyers get the vast majority of the fees they request—as illustrated earlier, about 97 cents on the dollar. 165 See supra section II.B.1.c (finding ratio of award to request for overall data set is 0.968 (mean) and that court awarded smaller fee than lead counsel requested in 14.39% of cases in data set). The higher fee requests observed in low-volume districts suggest that attorneys might anticipate higher fee awards when their cases are litigated in those districts.

Table 5: Regressions for Fee Awards 166 Settlement data are in constant 2012 dollars, log-transformed and centered at the mean settlement value. Robust standard errors are in parentheses. The probability values are indicated as follows: *** p<0.01; ** p<0.05; and * p<0.1.

(1) (2) (3) (4) (5) (6) (7)
Settlement -0.007* -0.002 -0.002 -0.002 -0.001 -0.002 -0.002
(0.003) (0.004) (0.005) (0.005) (0.005) (0.005) (0.005)
Competition -0.010*** -0.009*** -0.009*** -0.010*** -0.010*** -0.010*** -0.011***
(0.003) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002)
Public Pension -0.045*** -0.037*** -0.045*** -0.044*** -0.045*** -0.045*** -0.044***
(0.008) (0.011) (0.008) (0.008) (0.009) (0.009) (0.008)
Other Institution -0.003 -0.004 -0.004 -0.004 -0.004 -0.004 -0.004
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004)
Ex Ante Agreement -0.057*** -0.038*** -0.054*** -0.054*** -0.055*** -0.055*** -0.056***
(0.016) (0.011) (0.014) (0.013) (0.015) (0.015) (0.015)
Objection -0.003 -0.003 -0.005 -0.005 -0.005 -0.005 -0.004
(0.007) (0.008) (0.008) (0.008) (0.007) (0.007) (0.008)
Fee Cut -0.038*** -0.037*** -0.039*** -0.039*** -0.038*** -0.039*** -0.038***
(0.004) (0.004) (0.006) (0.006) (0.005) (0.006) (0.004)
High Volume District -0.031*** -0.042*** -0.036*** -0.023*** -0.039*** -0.033*** -0.046***
(0.005) (0.005) (0.006) (0.004) (0.006) (0.007) (0.008)
Case Age (Years) -0.002 -0.002 -0.002* -0.002* -0.002* -0.002* -0.003*
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Settlement x High Volume -0.009* -0.008* -0.008* -0.009* -0.008* -0.008*
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004)
Public Pension x Ex Ante -0.040
Judge Frequency -0.004*** -0.001
(0.001) (0.002)
High Volume x Judge Frequency -0.005**
High Volume Judge -0.016*** -0.004
(0.004) (0.006)
High Vol. Judge x High Volume -0.018*
One Opinion 0.002
One Opinion x High Volume 0.019**
Constant 0.291*** 0.295*** 0.309*** 0.302*** 0.304*** 0.302*** 0.301***
(0.007) (0.008) (0.011) (0.010) (0.010) (0.010) (0.012)
Observations 431 431 431 431 431 431 431
R-squared 0.419 0.438 0.439 0.443 0.436 0.439 0.434

As in the models for fee requests, 167 See supra section II.B.2.a (detailing models for fee requests). Model 1 in Table 5 begins without interaction terms. At mean settlement values, fee awards in low-volume districts average 29.1%. In high-volume districts, fee awards average 26.0%. In either kind of district, ex ante fee agreements and public pension lead plaintiffs have a significant moderating influence on fees. Average fees in all cases with an ex ante fee agreement are 23.4% at mean settlement values. For cases with public pension lead plaintiffs, fee awards average 24.6%. When judges cut fees, we observe reductions of similar magnitude: Fee cuts average about 3.8%.

It is noteworthy that when courts award a fee that is lower than the requested amount they apparently pay little heed to objectors. We found no evidence that objections to fee requests have any significant relationship to fee awards. There are several potential explanations for this result. One is that objections are infrequent. Objections to fee requests were filed in only about one-fifth of the cases (22.97%), and one-third of the cases with objections (34.34%) had only a single objection. Of the 125 cases with any kind of objection, only 20 (16.00%) had 5 or more objections. Most courts consider a low rate of objections to be support for the proposed settlement and fee, 168 See, e.g., In re AT&T Corp. Sec. Litig., 455 F.3d 160, 165 (3d Cir. 2006) (holding courts should consider “presence or absence of substantial objections” in evaluating fee request); In re Tyco Int’l, Ltd. Multidistrict Sec. Litig., 535 F. Supp. 2d 249, 269 (D.N.H. 2007) (concluding class supported proposed fee because there were only 11 objections to proposed fee, out of more than 2.4 million class members who received notice of settlement). although the strength of such an inference seems dubious given the collective-action problems facing class members. Many plaintiffs’ lawyers complain about the presence of “professional objectors,” who file pro forma objections to proposed class settlements simply in an attempt to extract a greater share of the settlement or to garner attorneys’ fees for providing what they contend are independent benefits to the class. 169 See, e.g., In re Initial Pub. Offering Sec. Litig., 721 F. Supp. 2d 210, 215–16 (S.D.N.Y. 2010) (“[P]rofessional objectors undermine the administration of justice by disrupting settlement in the hopes of extorting a greater share of the settlement for themselves and their clients.”); Barnes v. FleetBoston Fin. Corp., No. 01 Civ. 10395, 2006 WL 6916834, at *1 (D. Mass. Aug. 22, 2006) (“Repeat objectors to class action settlements can make a living simply by filing frivolous appeals . . . . The larger the settlement, the more cost-effective it is to pay the objectors rather than suffer the delay of waiting for an appeal to be resolved (even an expedited appeal).”); Greenberg, supra note 22, at 963–68 (providing plaintiffs’ class action attorney’s perspective on and analysis of problems wrought by professional objectors). For a discussion of this phenomenon, see Fitzpatrick, Objector Blackmail, supra note 23, at 1633–37 (noting class counsel often “willing to dip into their own pockets to pay objectors to drop their appeals” in order to avoid delaying receipt of fee awards). If courts believe that these professional objectors predominate (and there is some evidence that they do), 170 See O’Keefe v. Mercedes-Benz United States, LLC, 214 F.R.D. 266, 295 n.26 (E.D. Pa. 2003) (“Federal courts are increasingly weary of professional objectors: [S]ome of the objections were obviously canned objections filed by professional objectors who seek out class actions to simply extract a fee by lodging generic, unhelpful protests . . . .” (citation omitted)). it is not surprising that courts pay little attention to objectors when awarding fees.

Competition for the lead plaintiff position at the start of the case is also correlated with lower fee awards at the end of the case. On average, fee awards are about 1% less in cases with that kind of competition. While it is possible that competition signals to the court that the case is more lucrative and perhaps less risky, leading to a reduction in fees, this does not appear to be the case. Instead, the magnitude of the decline in fee awards is about the same as the decline in fee requests in these cases. In other words, courts appear to be awarding lower fees in cases with competitors for the lead plaintiff position because attorneys are asking for lower fees in those cases. This result, along with the general ineffectiveness of objections to fee requests, underscores the importance of focusing on ex ante versus ex post strategies for addressing fees, a matter treated in more detail in Part III.

The models for fee requests, showed a weak (negative) correlation with Case Age. 171 See infra Table 4. Model 1 shows no statistically significant correlation between the number of years the case was litigated and the size of fee awards, although some of the other models demonstrate a weak (negative) correlation. The mixed evidence for this variable is interesting, given the standards courts typically employ in awarding fees. For example, Johnson v. Georgia Highway Express, Inc., one of the leading class action fee precedents, adopts a twelve-part test for evaluating fee requests. 172 See 488 F.2d 714, 717–19 (5th Cir. 1975) (presenting twelve-part test). Two elements of that test—“the time and labor required” to reach the settlement and the opportunity costs associated with taking a particular case—relate to the length of time a case has been litigated. 173 Id. at 717–18 (including as two of twelve factors “time and labor required” of attorneys and “preclusion of other employment by the attorney due to acceptance of the case”). The duration of the case is also correlated, albeit imperfectly, to the hours an attorney has devoted to a case. Indeed, consistent with appellate court precedent in the Ninth, 174 See Vizcaino v. Microsoft Corp., 290 F.3d 1043, 1047 (9th Cir. 2002) (“[T]he lodestar may provide a useful perspective on the reasonableness of a given percentage award.”). Second, 175 See Goldberger v. Integrated Res., 209 F.3d 43, 50 (2d Cir. 2000) (“The district court’s use of the lodestar method . . . was a permissible exercise of its discretion.”). and Third 176 See In re Rite Aid Corp. Sec. Litig., 396 F.3d 294, 305 (3d Cir. 2005) (“[W]e have suggested it is ‘sensible’ for district courts to ‘cross-check’ the percentage fee award against the ‘lodestar’ method.” (citation omitted)). Circuits, a sub­stantial minority of courts in our nationwide data set (45.01%) used a lodestar cross-check to evaluate the requested fee. 177 See supra notes 67–73 and accompanying text (providing background on lodestar method). The prevalence of these time-based factors in legal doctrine and the insignificance of Case Age in our findings suggest a substantial disconnect between what these courts say they do and what they actually do when awarding fees. Instead, courts in practice seem to follow something closer to a pure percentage approach to fee awards. 178 Several circuit court decisions have adopted just such an approach. See Swedish Hosp. Corp. v. Shalala, 1 F.3d 1261, 1267–71 (D.C. Cir. 1993) (“We adopt a percentage-of-the-fund methodology . . . primarily because it is more efficient, easier to administer, and more closely reflects the marketplace.”); Camden I Condo. Ass’n v. Dunkle, 946 F.2d 768, 774 (11th Cir. 1991) (“[W]e believe that the percentage of the fund approach is the better reasoned in a common fund case.”). The pure-percentage approach was used in 43.72% of the fee awards examined.

Given the prevalence among courts of using a lodestar cross-check to justify fee awards, 179 Indeed, some judges have argued that conducting such a cross-check is an ethical imperative. See Vaughn R. Walker & Ben Horwich, The Ethical Imperative of a Lodestar Cross-Check: Judicial Misgivings About “Reasonable Percentage” Fees in Common Fund Cases, 18 Geo. J. Legal Ethics 1453, 1454 (2005) (“[W]e argue that courts making common fund fee awards are ethically bound to perform a lodestar cross-check.”). we investigated whether there is any evidence that this procedure has any meaningful impact on fees. Although the results are not reported separately in Table 5, a regression similar to Model 2 was run that included as an explanatory variable whether the court stated in its fee order that it had used a lodestar cross-check in arriving at its fee award. If this approach by courts was useful, the regression should show systematic differences between those courts that use a cross-check versus those courts that award fees on a simple percentage basis. That is not the case. There was no statistically significant difference between fee awards using a lodestar cross-check and fee awards under the percentage of the recovery approach. 180 The coefficient for the cross-check variable was actually positive (0.004), but insignificant. The t-statistic is 1.25 (probability = 0.237). This result supports previous academic criticisms of the lodestar methodology. 181 See supra note 70 (presenting academic criticisms of lodestar approach). It also suggests that the time and resources lawyers devote to tracking lodestar amounts and the enormous number of hours that some judges spend reviewing firm billing records is largely a waste of time.

Even though cross-checked fee awards do not differ statistically from those based on the percentage method alone, cross-checks may influence the size of awards in ways that are hard to detect. For example, expecting judges to cut fee requests that seem excessive on a lodestar basis, lawyers may moderate their requests before submitting them. Because the vast majority of fee requests (92.33%) included lodestar comparisons, lodestar-related moderation may occur even in circuits that permit judges to award fees on the basis of reasonable percentages alone. That is, lawyers may mask the lodestar method’s fee-constraining effect by anticipating the concerns with hours worked that judges may have.

To gain more insight into the possible impact of lodestar cross-checks, a regression was run on fee awards in which a fee application’s inclusion or exclusion of a cross-check was an independent variable. Awards were actually 2.4% higher when fee requests included cross-checks than when lawyers urged judges to apply solely the percentage approach. 182 The t-statistic is 2.85 (probability = 0.016). Because the regressions included Case Age as an independent variable (which remained insignificant), the difference cannot be explained by the possibility that cases in which fee requests included lodestar cross-checks warranted higher fees because they took longer to resolve.

An alternative possibility is that lawyers use lodestar-related information strategically. They may include it when it tends to shore up fee requests that might strike judges as excessive if sought on the basis of percentages alone; and they omit it when they expect judges to grant their percentage-based requests or think that it may tend to make percentage-based requests seem excessive (e.g., because the number of hours expended was small). Although these speculations do not prove that lawyers strategize in the manner described, they have several obvious strengths. They comport with lawyers’ self-interest, which should lead them to couch fee applications as attractively as possible. They take seriously the belief that judges place weight on time expended when sizing fee awards. And they plausibly assume that lawyers base their arguments on prior beliefs about judges’ attitudes.

As with the analysis of fee requests, Model 2 adds interaction terms between settlement size and high-volume districts and between ex ante agreements and public pension lead plaintiffs. The results continue to be very similar. On average in a low-volume district, courts award attorneys’ fees of about 29.5% in cases without either a public pension lead plaintiff or an ex ante fee agreement. There is no evidence that those fees vary with settlement size. In a high-volume district, by comparison, courts in similar cases award, on average, only 24.4% in attorneys’ fees. The interaction term Settlement x High Volume remains significant in Model 2. On average, fee awards in districts that frequently see securities class actions decline as settlement size increases. For a settlement one standard deviation above the mean ($217.2 million), fee awards in low-volume districts average about 29.2% compared to 22.8% in high-volume districts. This is an average difference of about $13.9 million per fee award, an amount that is both statistically and economically significant. 183 See infra Table 5.

Put another way, the significance of the interaction between settlement values and high-volume districts with regard to fee awards, as opposed to fee requests, suggests that plaintiffs’ attorneys are able to exploit imperfections in the market for attorneys’ fees. That is, class counsel request larger fees in districts that are more likely to award them. If that were not the case, then fee awards in low-volume districts would match fee awards in high-volume districts, regardless of the size of the fee requests.

If the PSLRA were working as intended, one would not expect to find inter-court variations like these. Both the plaintiffs’ securities bar and the investment markets operate nationwide; any plaintiffs’ law firm can appear in a securities class action in any federal district court. 184 See Stephen J. Choi and Robert B. Thompson, Securities Litigation and Its Lawyers: Changes During the First Decade After the PSLRA, 106 Colum. L. Rev. 1489, 1514–29 (2006) (showing how small number of firms dominate national market for securities class actions). If lead plaintiffs consistently shopped for lawyers and drove hard bargains when negotiating ex ante fee agreements, one would expect to find fairly uniform fee terms and fee awards across courts after controlling for relevant case characteristics.

Plausibly, judges in low-volume districts are operating in a comparatively lower information environment than their counterparts in high-volume districts. Because they see these cases less frequently and because so many fee awards are unpublished, 185 Fees are typically awarded via a court order which is filed and is public, but which is not a published opinion of the court. Of the 431 cases in the data set, for example, a Westlaw search showed that only 113 (26.2%) had fee opinions which were published in the Federal Reporter. they have a smaller pool of precedents on which to base their decisions. As noted above, judges seem to generally disregard objectors, perhaps because they deem them an unreliable signal of excessive fee requests. 186 See supra notes 132–134 and accompanying text (reporting empirical findings that judges’ willingness to cut fees appears unrelated to class members’ objections). Defendants are indifferent to fee requests because the fees are paid out of the common fund. 187 In this study’s nationwide data set, there was not one instance in which a defendant objected to a proposed fee. Furthermore, it is common practice for settlement agreements to state that the defendant “agrees not to oppose a fee request by class counsel of X dollars/percent or less.” See also supra Part I. In such a non-adversarial setting, it is hardly surprising that a time-constrained federal judge would be inclined simply to sign the proposed fee order that class counsel has drafted.

The effects on fee awards associated with ex ante fee agreements and public pension funds remain the same in Model 2. Individually, both variables are negative and significant, with magnitudes that are roughly similar. The interaction term remains negative, but insignificant. At this stage, there is no evidence of any greater effect on fee awards when there is both an ex ante agreement and a public pension lead plaintiff. The variables Cut and Fee Objection are similar to the results under the previous model. Judicial fee cuts average about 3.7%. There is no evidence that objections are correlated with fee awards.

Models 3 through 7 in Table 5 examine the impact on fee awards of individualized judicial experience with securities class actions. The results mirror those found in the regressions on fee requests. Overall, experienced judges award lower fees than less experienced judges, but this appears to be true only for experienced judges in districts that see substantial securities class action filings. To illustrate this point, consider Models 5 and 6, which included an indicator variable for judges who contributed four or more decisions to the data set. Viewed in isolation (Model 5), all such judges appear to award fees that are, on average, 1.6% lower than those awarded by less experienced judges. But interacting that indicator variable with high-volume districts shows that all the reduction comes from the most experienced judges in the highest volume districts. Contrast that result with judges from high-volume districts with the least amount of experience (Model 7). The fees that the less experienced judges award are lower than those awarded by their counterparts in low-volume districts, but not nearly as low as the awards by more experienced judges in their own districts. These results suggest that the more judges see securities class actions, the less willing they are to award fees to the attorneys who bring them.

c. Judicial Fee Reductions. — In order to evaluate the relative impact of ex ante and ex post approaches to attorneys’ fees, it is useful to consider the variables correlated with judicial fee reductions. To do that, the study contains a logit model with Cut as the dependent variable. The model uses the same variables employed in this Article’s fee regressions because we wanted to see whether these variables could also predict the cases in which fee cuts occurred. What is noteworthy about these variables is that, with the exception of Objection, they are readily observable to case participants prior to any fee application. The coefficients for the logit model reported in Table 6 are the marginal effects of each independent variable when the other independent variables are set to their means.

Table 6: Logit Regression for Judicial Fee Reductions 188 This Table reports average marginal effects for the independent variables with delta-method standard errors. Robust standard errors are in parentheses. The probability values are indicated as follows: *** p<0.01; ** p<0.05; and * p<0.1.

Average Marginal Effect Standard Error
Settlement 0.071*** 0.021
Competition -0.007 0.037
Public Pension -0.121*** 0.044
Other Institution -0.019 0.037
Ex Ante Agreement -0.116** 0.053
High Volume District 0.064 0.049
Case Age (Years) -0.011 0.009
Objection 0.014 0.041
Settlement x High Volume -0.037* 0.022
High Volume Judge 0.130** 0.058
High Volume Judge x High Volume -0.136* 0.071
Observations 431
Pseudo R-squared 0.0997

Table 6 shows that there are some variables that increase or decrease the probability of a fee cut. For example, there is evidence that courts do in fact defer to the fee arrangements negotiated by public pension funds and to ex ante agreements. Indeed, the effect on the probability of these two variables is nearly identical. In the presence of either one, the probability of a fee cut declines by about 12%.

Table 6 also shows that, all else being equal, the probability that the judge will reduce the requested fee is related to the settlement size. A closer look at those probabilities, however, shows a seemingly anomalous result. The likelihood of a fee cut increases in low-volume districts (represented here by the Settlement variable). By contrast, in high-volume districts (Settlement x High Volume), the probability of a fee cut decreases with settlement size. The explanation for this counter-intuitive result most likely lies in the findings with respect to fee requests. On average, as settlement size increased, plaintiffs sought lower fees in high-volume versus low-volume districts. With lower fee requests, there was likely less need for high-volume courts to cut fees. A similar dynamic occurs with high-volume judges. Outside the high-volume districts, judges with more experience with securities class actions are more likely to cut fees. In high-volume districts, because the most experienced judges see lower fee requests, they are less likely to cut the requested fees. In other words, ex ante steps to reduce fee requests reduce the likelihood that judges will make ex post adjustments in their fee awards.

While these results are consistent with the linear regressions for requests and awards, it is important to recognize the limitations in this analysis. Perhaps the best way to evaluate the utility of a logit model is to see how well it classifies cases. In other words, do the model’s predictions for which cases will have fee cuts match up with what actually happened in the cases in the sample? Overall, the model correctly predicted 85.61% of the cases. In the abstract, that sounds like a powerful predictive tool, until one realizes that there were no fee cuts in 85.61% of the cases in the database. Identical results would be obtained by simply guessing that there were no fee reductions in any of the sampled cases in the study. A closer examination reveals just how poorly the model performs. It correctly predicted all but 2 of the 369 cases without fee cuts, but identified only 2 of the 62 cases in which reductions actually occurred. In other words, when it came to identifying the situations in which judges did cut fees, the model was wrong more than 96% of the time.

Notwithstanding the model’s low predictive power, it does illustrate that ex ante fee agreements, public pension lead plaintiffs, and judicial experience are all associated with the probability of a fee cut. The real lesson of the analysis, however, is that these variables tell only a small part of the story. Logically, that realization suggests two possible alternatives. The first is that judicial fee reductions are, for all intents and purposes, random events. The second is that other unobserved (and perhaps unobservable) case or judicial characteristics play the dominant role in determining whether the court will cut fees.

Either scenario raises serious questions about how fees are set in securities class actions (and in class actions of other types). Some judges may believe they have the ability to identify and accurately reduce “excessive” fee requests, but if they do they are relying on criteria other than those tested here. The standards that courts explicitly employ also contemplate that judges have the ability to assess ex post a vast array of factors, including how hard the attorneys worked, how much risk they faced, the quality of the results they obtained, and the relationship of the fee request to similar cases. 189 See Johnson v. Ga. Highway Express, Inc., 488 F.2d 714, 717–19 (5th Cir. 1975) (articulating twelve-part test); Detroit v. Grinnell Corp., 495 F.2d 448, 470 (2d Cir. 1974) (“There are many parameters that affect the value of legal services and which, therefore, must be considered by a court in evaluating a fee request.”). Perhaps long experience with any given case or with securities class actions more generally provides judges information that is more nuanced than can be captured here. Indeed, that might partially explain why fees in high-volume districts differ so dramatically from fees in low-volume districts.

But something more is likely going on. Perhaps the unpredictability of fee cuts derives from the fact that fee reductions are primarily the product of subjective assessments by judges based on their own idiosyncrasies, biases, and heuristics rather than the objective facts of any given case. Temperament, judicial philosophy, experience with class action litigation, personal wealth, pre–judicial work history, personal earnings history, political ideology, and other individual and largely unobservable judicial characteristics may well—consciously or unconsciously—drive judges’ decisions about whether and how much to reduce fees. 190 This theory will be the subject of further exploration in future work. For now, it is sufficient to note that many of those writing in the field of judicial decisionmaking believe that judicial decisions are partially, and perhaps even primarily, the product of political predilections, personal experiences, and deeply ingrained worldviews. See Gregory C. Sisk, Michael Heise & Andrew P. Morriss, Charting the Influences on the Judicial Mind: An Empirical Study of Judicial Reasoning, 73 N.Y.U. L. Rev. 1377, 1383 (1998) (citing Joel B. Grossman, Social Backgrounds and Judicial Decision-Making, 79 Harv. L. Rev. 1551, 1552 (1966)). Unfortunately, this literature tends to focus on the Supreme Court or on intermediate appellate courts. Comparatively few studies examine the decisionmaking of trial court judges. Those that do have arrived at mixed results. Some studies find that Democratic judges tend to render more “liberal” decisions than their Republican counter­parts. See Robert A. Carp & C.K. Rowland, Policymaking and Politics in the Federal District Courts 32–34 (1983) (compiling data suggesting Democratic-appointed district judges render more politically left-liberal decisions than Republican-appointed judges). Others find that various ideological measures are significant in predicting the outcomes of Establishment Clause cases and Voting Rights Act cases. See Adam B. Cox & Thomas J. Miles, Judging the Voting Rights Act, 108 Colum. L. Rev. 1, 3 (2008) (finding judicial ideology significantly influences judicial decisionmaking in Voting Rights Act cases); Gregory C. Sisk & Michael Heise, Ideology “All The Way Down”?: An Empirical Study of Establishment Clause Decisions in the Federal Courts, 110 Mich. L. Rev. 1201, 1204 (2012) (discovering political factors played major role in judges’ Establishment Clause decisions). These results, however, are far from uniform. Other empirical work finds that ideology or other judicial characteristics play a very small role in district court decisionmaking. See Orley Ashenfelter, Theodore Eisenberg & Steven J. Schwab, Politics and the Judiciary: The Influence of Judicial Background on Case Outcomes, 24 J. Legal Stud. 257, 257 (1995) (suggesting judges’ political ideologies are not significant predictors of judicial decisions); Theodore Eisenberg & Sheri Lynn Johnson, The Effects of Intent: Do We Know How Legal Standards Work?, 76 Cornell L. Rev. 1151, 1190 (1991) (finding identity of appointing President did not have statistically significant impact on outcomes in inten­tional discrimination cases); Max Schanzenbach, Racial and Sex Disparities in Prison Sentences: The Effect of District-Level Judicial Demographics, 34 J. Legal Stud. 57, 57 (2005) (suggesting there is little evidence of political composition of district affecting sentencing disparities). Indeed, the importance of such variables might partially explain the finding that in one-third of the fee-reduction cases the court offered no justification or explanation at all for its decision. 191 See infra Table 3. When courts do offer an explanation, the most common one is simply that the requested fee is “too large.” 192 See infra Table 3 (reporting “too large” explanation offered by court in 40.32% of fee-cut cases).

This study’s data show that class counsel can expect some kind of fee reduction in about one out of every seven cases, and that some judges may be more likely than others to cut the requested fees. 193 For example, 27 judges in the sample authored 4 or more fee decisions. 12 of those judges did not have a single decision that cut the fees lead counsel had requested. By contrast, 5 judges cut fees in 50% to 75% of their decisions. But whether fees will be cut in any particular case will likely be impossible to determine at the time the fee request is made. And little more is likely to be known after a given court’s fee ruling because the courts’ fee opinions offer no reliable explanations for those decisions.

The inability to account for fee cuts in securities class actions on the basis of objective factors raises a broader concern for class actions of all types. The unpredictability of fee cuts implies that, to a significant extent, fee awards are unpredictable too. This is likely true for all class actions, because the doctrines and procedures that govern fee awards are largely the same across different substantive areas of the law. If anything, fee awards in other types of federal lawsuits may be more uncertain than those in securities fraud class actions because the PSLRA applies only to the latter. In cases of other types, fee awards are governed by federal common law, which places only the weakest of constraints on the discretion of district court judges. Uncertainty regarding fees can be expected to discourage lawyers from investing optimally in class actions by creating downside risks that lawyers will avoid. This uncertainty ultimately harms both class members (by impairing the quality of the representation they receive) and the general public (by weakening the deterrent effect of the law).

*     *     *

The picture that emerges from these nationwide data is both nuanced and significantly different from the one Congress had in mind when it gave lead plaintiffs responsibility for handling attorneys’ fees in securities class actions. 194 On Congress’s expectations, see John F. Olson et al., Pleading Reform, Plaintiff Qualification and Discovery Stays Under the Reform Act, 51 Bus. Law. 1101, 1144–45 (1996) (“Congress’s goal, in short, was to ‘protect[] investors who join class actions against lawyer-driven lawsuits by giving control of the litigation to lead plaintiffs with substantial holdings of the securities of the issuer.’” (quoting H.R. Rep. No. 104-369, at 32 (1995), reprinted in 1995 U.S.C.C.A.N. 730)). Our data suggest that some lead plaintiffs do take seriously the responsibility Congress bestowed on them, carefully choosing their lawyers, aggressively negotiating fees with them upfront, and actively monitoring them throughout. But such plaintiffs are in the minority. In most cases, lead plaintiffs appear passive, seemingly agreeing at the time of settlement to whatever fees their attorneys deem appropriate.

Even more troubling are our findings regarding the role of the courts in the fee-setting process. We found no evidence that the actions taken by the courts move class counsel’s fees closer to the “right price.” Instead, the data showed that the courts facilitate, rather than prevent, the exploitation of market imperfections by class counsel, enabling them systematically to obtain higher fees from courts and judges that see securities class actions less frequently than from more experienced courts. And although judges do sometimes cut class counsel’s fees, those decisions were unpredictable. That is, judicial fee cuts are as likely to result in fees that are further from the “right price” as they are to move them closer to that ideal.

In sum, there is little to celebrate in the current state of affairs, and reason to think that even small improvements in the fee-setting process might yield significantly better results.

III. Should Courts Set Fees Ex Ante? A Proposal

This Part considers some possible improvements to the current process by which class counsel’s fees are determined under the PSLRA. Section III.A begins by examining previous reform proposals. Section III.B then offers an original reform proposal, which we believe improves upon those prior efforts.

A. Prior Reform Proposals

In 2002, a Task Force convened by the Third Circuit issued a report on the manner of selecting and compensating lawyers who handle class actions. 195 Third Circuit Task Force, Selection of Class Counsel, 208 F.R.D. 340 (2002) [hereinafter 2002 Task Force Report]. It encouraged judges presiding over securities cases to base fee awards on negotiated agreements between lead plaintiffs and class counsel. After observing that “[t]he PSLRA establishes a model of client control that extends . . . to . . . negotiation of the fee,” the Task Force “conclude[d] . . . that strict scrutiny of the fee agreement [would be] inconsistent with the client-driven litigation model established in the PSLRA.” 196 Id. at 425. Instead, it urged district court judges to “presume that the fee is reasonable when it is the result of an agreement between the ‘most adequate’ plaintiff and chosen counsel.” 197 Id. Judges should override this presumption only when the agreed compensation is “clearly excessive,” “has been rendered unfair by unforeseen developments,” or “was not reached by arm’s-length negotiation between the lead plaintiff and counsel.” 198 Id. at 425–26; see also Restatement (Third) of Law Governing Lawyers § 125 (2000) (stating “more appropriate arrangement” for reducing settlement-related conflicts in class actions is for client and lawyer to “negotiate [lawyer’s fee] initially . . . at the outset of the relationship,” while “disclos[ing] to the client that the ultimate award may be scrutinized by the opposing party and approved by the court”).

The Task Force issued its proposal on the heels of the Third Circuit’s decision in Cendant. 199 In re Cendant Corp. Litig., 264 F.3d 201 (3d Cir. 2001). The Task Force issued its proposal the year after Cendant was decided. See 2002 Task Force Report, supra note 195. Its recommendation hews closely to the approach endorsed in that case, which also encouraged district court judges to defer to negotiated fee agreements. 200 See 2002 Task Force Report, supra note 195, at 425 & n.323 (citing Cendant, 264 F.3d at 220) (“[A] court should presume that the fee is reasonable when it is the result of an agreement between the ‘most adequate’ plaintiff and chosen counsel.”). This mirrors Cendant’s recommendation that:

“[C]ourts should accord a presumption of reasonableness to any fee request submitted pursuant to a retainer agreement that was entered into between a properly-selected lead plaintiff and a properly-selected lead counsel. This presumption will ensure that the lead plaintiff, not the court, functions as the class’s primary agent vis-à-vis its lawyers. Further, by rendering ex ante fee agreements more reliable, it will assist those agreements in aligning the interests of the class and its lawyers . . . .”

Cendant, 264 F.3d at 282.

In support of this recommendation, the Third Circuit cited Weiss & Beckerman, supra note 26, at 2105, which observed courts “might well feel confident in assuming that a fee arrangement an institutional investor had negotiated with its lawyers before initiating a class action maximized those lawyers’ incentives to represent diligently the class’s interests, reflected the deal a fully informed client would negotiate, and thus presumptively was reasonable.” 2002 Task Force Report, supra note 195, at 400 n.227.
The Third Circuit also agreed that judges could properly override the presumption when “unusual and unforeseeable changes, i.e., those that could not have been adequately taken into account in the negotiations,” arose and when “the (properly submitted) retained [sic] agreement fee is clearly excessive.” 201 Cendant, 264 F.3d at 283. The Court thought it unlikely that “candidates for [the] lead plaintiff designation [would] be deterred by the understanding that their retainer fee arrangement [sic] with Lead Counsel will be subject to judicial review for clear excessiveness.” 202 Id. at 282–83.

Clearly, the Task Force’s recommendation is compatible with the holding of Cendant. Both place great weight on ex ante fee agreements between lead plaintiffs and class counsel, and both set out similar grounds for overriding the terms of ex ante agreements on ex post judicial review. But the Task Force did not simply follow the Third Circuit’s jurisprudence down the line. Its recommendation differs from Cendant’s holding in a key respect: The Task Force, but not Cendant, encouraged judges to set preliminary fee terms early in the litigation. 203 See 2002 Task Force Report, supra note 195, at 420–21 (“[T]he Task Force recommends that the topic of attorney fees should be addressed at the early stages of the case as well as throughout the prosecution of the case.”). The Task Force followed Cendant in recommending deference to fee agreements negotiated ex ante between attorneys and lead plaintiffs but added that judges should set fee terms when appointing class counsel instead of addressing fees for the first time when litigation concludes, as they usually do. 204 See id. at 425–26 (“If there are concerns about aspects of the agreement that ought to be aired earlier rather than later, early review would provide guidance to both the lead plaintiff and the lead counsel.”); see also Charles M. Silver, Dissent from Recommendation to Set Fees Ex Post, 25 Rev. Litig. 497, 497 (2006) (encouraging judges to set fees at or near start of class action litigation and contending “tradition of setting fees ex post is responsible for much that is wrong with the modern class action”). Although the Task Force agreed that judges had to review fee awards for reasonableness at the end of litigation, 205 See 2002 Task Force Report, supra note 195, at 363 (“Even if one concludes that an ex ante bidding process is advantageous, it is difficult to believe in the class action context that one would abolish a final look at the attorneys’ work at the conclusion of the case.”). it emphasized that the terms of class counsel’s compensation should be addressed at the appointment stage as well. 206 See id. at 420–21. Specifically, the Task Force recommended that:

“[T]he topic of attorney fees should be addressed at the early stages of the case as well as throughout the prosecution of the case. At the outset of the case, the court may be well-advised to direct counsel to propose the terms for a potential award of fees; the potential fees might be established within ranges, with the court making it clear to the parties that the fee remains open for further review for reasonableness. A preliminary fee arrangement may provide a helpful structure for the court when it conducts its reasonableness review at the end of the case.”

Id. at 420–21 (footnotes omitted).

The Task Force is far from the only authority to encourage judges to set preliminary fee terms when appointing class counsel. Successive editions of the Manual for Complex Litigation advise judges to establish ground rules for fee awards and cost reimbursements at this juncture. 207 See Manual for Complex Litigation (Fourth) § 14.211 (2004) (“At an early conference or in an early pretrial order after consultation with counsel, it is helpful to establish guidelines and procedures that will lighten the burdens on the participants, clarify expectations, and reduce the opportunities for disputes.”); Manual for Complex Litigation (Third) § 24.21 (1995) (“Disputes will be reduced if the court advises the parties at the outset of the litigation what method will be used for calculating fees and . . . the range of likely percentages.”). A series of guides for judges published by the Federal Judicial Center since the 1980s also endorses this approach. 208 See Alan Hirsch & Diane Sheehey, Awarding Attorneys’ Fees and Managing Fee Litigation 113–14 (2d ed. 2005) (“Many judges stress the importance of informing attorneys, early in the case, what is expected of them in regard to attorneys’ fees.”); Alan Hirsch & Diane Sheehey, Awarding Attorneys’ Fees and Managing Fee Litigation 109 (1994) (quoting judges stressing importance of setting ground rules for fees at outset of litigation); Barbara J. Rothstein & Thomas E. Willging, Managing Class Action Litigation: A Pocket Guide for Judges 33 (3d ed. 2010) (“When appointing counsel, consider entering an order with express provisions about the standards and procedures you expect to use in reviewing requests for attorney fees and costs.”); Barbara J. Rothstein & Catherine R. Borden, Managing Multidistrict Litigation in Products Liability Cases: A Pocket Guide for Transferee Judges 14 (2011) (“Although fees will not be awarded unless and until there is a settlement, early judicial involvement can have a major impact on the fairness and reasonableness of fee requests.”); Thomas E. Willging & Nancy A. Weeks, Attorney Fee Petitions: Suggestions for Administration and Management 5–8 (1985) (“[A] court’s guidelines on attorneys’ fees will be fairer and easier to enforce if announced in advance so that counsel have an opportunity to alter any nonconforming practices.”). Other appellate courts have expressed a preference for ex ante fee agreements in class litigation. 209 See Silverman v. Motorola Sols., Inc., 739 F.3d 956, 958 (7th Cir. 2013) (“[E]stablishing a fee structure at the outset of a suit is desirable; unlike auctions, which private markets in legal services do not use, ex ante fee structures are common and beneficial to clients.”); In re Synthroid Mktg. Litig., 264 F.3d 712, 719 (7th Cir. 2001) (“Only ex ante can bargaining occur in the shadow of the litigation’s uncertainty; only ex ante can the costs and benefits of particular systems and risk multipliers be assessed intelligently.”); Swedish Hosp. Corp. v. Shalala, 1 F.3d 1261, 1269 (D.C. Cir. 1993) (“Mechanisms which may facilitate a judge in more closely approximating the market include . . . encouraging class counsel to enter into preliminary non-binding fee agree­ments with class members . . . .”). Setting attorneys’ fees ex ante was one of the primary motivations for the now-abandoned experiments in some courts to auction the role of class counsel. 210 See In re Auction Houses Antitrust Litig., 197 F.R.D. 71, 85 (S.D.N.Y. 2000) (stating fee structures and auction process were intended “to align attorney-client interests more closely, reduce agency costs, and help ensure that the class action mechanism acts as an effective mechanism of justice”); In re Bank One S’holders Class Actions, 96 F. Supp. 2d 780, 784–85 (N.D. Ill. 2000) (discussing how ex ante fee calculations in auction bidding replicate free legal market better than does lodestar approach); In re Lucent Techs., Inc., Sec. Litig., 194 F.R.D. 137, 155–58 (D.N.J. 2000) (requiring bidding firms to provide “statement of the dollar amount, as well as percentage, of any recovery the firm will charge in the event of a recovery”); In re Oracle Sec. Litig., 131 F.R.D. 688, 696–97 (N.D. Cal. 1990) (“It is far better to make the selection of counsel based on the benefits to the class implicit in a firm’s price quote rather than on qualitative grounds that provide a judge little in the way of legitimate assistance.”). For a critique of auctions, see generally Fisch, Auction Block, supra note 129, at 652 (arguing auctions fail to solve agency problems and alter judge’s role). One can also find a strong hint in this direction in the report a prior Third Circuit Task Force issued in 1986. The earlier Task Force recommended “that in the traditional common-fund situation . . . the district court, on motion or its own initiative and at the earliest practicable moment, should attempt to establish a percentage fee arrangement agreeable to the Bench and to plaintiff’s counsel.” 211 1986 Task Force Report, supra note 67 (footnote omitted). Although members of the 1986 Task Force disagreed as to when the “earliest practicable moment” was likely to arrive, the report noted that “high management judges will want to settle the fee question at the outset of the case” and that “[a]ll lawyer members of the Task Force . . . desired early clarification of the fee issue.” 212 Id. at 255 n.62. Finally, some judges have set basic fee terms up front, 213 Most famously, Judge John F. Grady did so in In re Cont’l Ill. Sec. Litig., 572 F. Supp. 931, 933–35 (N.D. Ill. 1983) (specifying “guidelines” under which court would evaluate “[a]ny fees and expenses for which court approval” subsequently sought by plaintiffs’ counsel). Judge Grady’s order is discussed in detail in Thomas E. Willging, Fed Judicial Ctr., Judicial Regulation of Attorneys’ Fees: Beginning the Process at Pretrial 2 (1984) (using Judge Grady’s Continental Illinois guidelines as “major focus” of study of fee-setting issues). See also Memorandum Opinion and Order at 6, In re HealthSouth Corp. Sec. Litig., No. CV-03-BE-1500-S (N.D. Ala. July 14, 2005) (on file with the Columbia Law Review) (ordering co-lead plaintiffs to report to court “fee arrangement for lead counsel and other counsel participating in the stockholder class action”); Stockholder Plaintiffs’ Submission in Response to the Court’s Memorandum Opinion and Order, Dated July 8, 2005 at 1, In re HealthSouth Corp. Sec. Litig., No. CV-03-BE-1500-S (N.D. Ala. July 14, 2005) (on file with the Columbia Law Review) (noting co-lead plaintiffs agreed to fee structure to be shared with court in camera at later date). and others have encouraged lawyers to bring the matter of fees to their attention early on. 214 See In re First Fid. Bancorporation Sec. Litig., 750 F. Supp. 160, 163 (D.N.J. 1990) (“Unfortunately, counsel did not avail themselves of the procedure outlined in the [1986] Task Force Report, wherein the contingent fee is to be negotiated and approved at the outset of the litigation.” (citing 1986 Task Force Report, supra note 67, at 255–58)).

It is not clear whether Plaintiffs’ lawyers who bring cases under the PSLRA prefer to disclose the fee terms they negotiate ex ante (if any) to the court (perhaps in camera) as part of the lead plaintiff and lead counsel application process or to keep them secret. The data indicate that they only infrequently volunteer in their lead plaintiff and lead counsel appointment filings to provide this information to the court. 215 See supra section II.B.1.a (reporting lead plaintiff candidate or court discussed ex ante fee agreement in only 11.29% of studied cases). The data set includes only one formal exchange on this topic by lead plaintiff competitors, who put forth strikingly different views. In In Re: Accredo Health, Inc., Securities Litigation, counsel for two public pension funds, which were joint applicants for lead plaintiff, contended that a court’s Rule 23 “adequacy” inquiry should require evidence that the lead plaintiff applicant has both selected competent class counsel and negotiated “a reasonable retainer agreement” with them. 216 See Memorandum of Law in Support of the Motion of the Teachers’ Retirement System of Louisiana and Louisiana School Employees’ Retirement System for (1) Appointment as Lead Plaintiffs; (2) Approval of Their Selection of Lead Counsel; and (3) Consolidation of All Related Actions at 7, Ferrari v. Accredo Health, Inc., No. 03-cv-2216 (W.D. Tenn. June 11, 2003) [hereinafter Louisiana Funds Memorandum] (on file with the Columbia Law Review). Chosen counsel for the two Louisiana public pension funds was Bernstein Litowitz Berger & Grossmann LLP. See Louisiana Retirement Funds’ Memorandum of Law in Further Support of Their Motion for Lead Plaintiff Further to the Court’s July 2, 2003 Order at 2–3, In re Accredo Health, Inc. Sec. Litig., No. 03-cv-2216 (W.D. Tenn. Apr. 19, 2005) [hereinafter Louisiana Funds’ Supplemental Memorandum] (on file with the Columbia Law Review); Memorandum in Support of the Motion of the Teachers’ Retirement System of Louisiana and the Louisiana School Employees’ Retirement System for Reconsideration of the Court’s June 20, 2003 Lead Plaintiff Order at 5–6, In re Accredo Health, Inc. Sec. Litig., No. 03-cv-2216 (W.D. Tenn. Apr. 19, 2005) [hereinafter Louisiana Funds’ Memorandum in Support of Reconsideration] (on file with the Columbia Law Review) (contending because competing applicants for lead plaintiff had not provided court with information about “how they selected their respective counsel” and “whether they negotiated a reasonable fee agreement with counsel” that “they have not shown that they will be able to adequately handle the task of representing the Class”); see also In re Conseco, Inc. Sec. Litig., 120 F. Supp. 2d 729, 733 (S.D. Ind. 2000) (contending court should consider “quality of information” regarding “adequacy” offered in support of lead plaintiff’s application). This counsel (Bernstein Litowitz) described their ex ante fee agreement in general terms and offered “to disclose the agreement to the Court for an in camera review should the Court so request.” 217 Joint Declaration of William T. Reeves, General Counsel for Louisiana Teachers, and Joe Seymour, Chairman of the Board of Trustees of LSERS ¶¶ 11–12, Ferrari v. Accredo Health, Inc., No. 03-cv-2216 (W.D. Tenn. June 11, 2003) (on file with the Columbia Law Review). The plaintiffs described their retainer agreement by saying:

“11. We have each been solicited over time by representatives from various firms that handle these types of securities class actions. After due consideration, the Louisiana Retirement Funds propose the Law Firm of Bernstein Litowitz Berger & Grossman LLP to serve as Lead Counsel in this case. We have entered into a retainer agreement with proposed Lead Counsel based on a percentage of the recovery. The percentages vary depending on the stage of the proceedings when the case is resolved, whether through settlement or judgment. The percentage figures are all substantially below the fees typically awarded in these types of cases. We are prepared to disclose the agreement to the Court for an in camera review should the Court so request. We understand, however, that the Court has the ultimate responsibility for setting the fees and expenses to be paid to counsel.
12. As Lead Plaintiffs in this Action, we will work diligently to maximize the recovery for the Class.”

Equally prominent counsel (Milberg Weiss) for a competing, individual lead plaintiff applicant argued in opposition that a retainer agreement should not be part of a court’s “adequacy” inquiry because “the Court, not the agreement, will determine the proper fee, if any, to award in this case” and thus “a retainer agreement at this stage of the litigation may well be illusory.” 218 See Debra Swiman’s Consolidated Opposition to All Other Motions for Appointment as Lead Plaintiff at 14–16, Ferrari v. Accredo Health, Inc., No. 03-cv-2216 (W.D. Tenn. June 11, 2003) [hereinafter Swiman’s Consolidated Opposition] (on file with the Columbia Law Review) (discussing irrelevance of retainer agreement in adequacy inquiry). Swiman further contended that although “the Louisiana Group dedicates four full pages in its memorandum of points and authorities to herald its adequacy” that the “Louisiana Group’s efforts in this regard are window-dressing that are simply not required by the PSLRA.” Id. at 14 (internal citations omitted). Swiman’s chosen counsel was Milberg Weiss Bershad Hynes & Lerach LLP, id. at 10. Counsel for the Louisiana Funds, however, replied that:

“Swiman’s argument that the retainer agreement is irrelevant because the Court will ultimately determine counsels’ fees misses the point . . . . A reasonable fee agreement is significant at the lead plaintiff stage because it evidences that the lead plaintiff movant controls counsel and that the litigation is not lawyer-driven. It is an indicator of lead plaintiff movants’ adequacy which Swiman has failed to fulfill.”

Louisiana Funds’ Supplemental Memorandum, supra note 216, at 3 n.3 (citing In re Cendant Sec. Litig., 264 F.3d 201, 266 (3d Cir. 2001)).
That law firm neither acknowledged in its filings whether it had negotiated a fee agreement with its client nor volunteered any information about terms. 219 Although Swiman’s counsel never stated whether it had negotiated an ex ante fee agreement with her, and never offered to provide the court any such retainer agreement, such an agreement apparently did exist at the outset of the litigation. Swiman’s sworn “Certification and Authorization of Named Plaintiff Pursuant to Federal Securities Laws” was Exhibit A to the “Declaration of B.J. Wade in Support of Debra Swiman’s Motion for Appointment as Lead Plaintiff and Approval of Her Selection of Lead Counsel.” Declaration of B.J. Wade in Support of Debra Swiman’s Motion for Appointment as Lead Plaintiff & Approval of Her Selection of Lead Counsel, Exhibit A (Certification & Authorization of Named Plaintiff Pursuant to Fed. Secs. Laws) at 1, Ferrari v. Accredo Health, Inc., No. 03-cv-2216 (W.D. Tenn. June 11, 2003) (on file with the Columbia Law Review). This earlier filing stated that:

“[Swiman] authorizes and, upon execution of the accompanying retainer agreement by Milberg Weiss, retains Milberg Weiss Bershad Hynes & Lerach LLP (‘Milberg Weiss’) to file an action under the federal securities laws to recover damages and to seek other relief against Accredo Health, Inc. (‘Accredo’). Milberg Weiss will prosecute the action on a contingent fee basis and will advance all costs and expenses. The Accredo Health, Inc. Retention Agreement provided to the Plaintiff is incorporated by reference, upon execution by Milberg Weiss.”


While there is considerable, if not universal, 220 Compare 2002 Task Force Report, supra note 195, at 425–26 (endorsing ex ante fee-setting accompanied by limited ex post review), with Task Force on Contingent Fees, Tort Trial & Ins. Practice Section of the Am. Bar Ass’n, Report on Contingent Fees in Class Action Litigation, 25 Rev. Litig. 459, 485 (2006) (“[F]ees should be set ex post except in cases for which an ‘empowered plaintiff’ or an auction is deemed appropriate.”). support for negotiating fees ex ante, substantial problems remain in the way that courts have handled these issues. For example, judges have adopted the ex ante approach explicitly in only a minority of cases. 221 See, e.g., In re Cardinal Health Inc. Sec. Litig., 528 F. Supp. 2d 752, 758–59 (S.D. Ohio 2007) (applying “presumption of reasonableness to ex-ante fee agreements”); In re Glob. Crossing Sec. & ERISA Litig., 225 F.R.D. 436, 466 (S.D.N.Y. 2004) (“[C]ourts pre­sume fee requests submitted pursuant to a retainer agreement negotiated at arm’s length between lead plaintiff and lead counsel are reasonable.”). They sometimes state in general terms that they will respect retainer agreements, but do not specify the precise process they will follow in reviewing those agreements. 222 Even the Third Circuit has vacillated. For example, in In re AT&T Corp. Securities Litigation, the district court approved a fee award of 21.25%, noting that the amount “resulted from a sliding scale formula negotiated by lead counsel and lead plaintiff New Hampshire Retirement Systems at the beginning of the case, and had been subsequently approved by each court-appointed lead plaintiff.” 455 F.3d 160, 167–68 (3d Cir. 2006). On appeal, the Third Circuit affirmed the award, but it noted significant concerns about the fee negotiation process when doing so:

“[T]he PSLRA has not eliminated all difficulties with establishing fair and reasonable attorneys’ fees in class action suits. As objectors note, the adversarial process is often ‘diluted’ or entirely ‘suspended’ during fee proceedings, and fee requests often go unchallenged. Lead plaintiffs, having previously negotiated a fee arrangement with lead counsel, will rarely oppose a fee request.”

Id. at 168 (citation omitted).

The Third Circuit did not explain why a lead plaintiff should oppose a fee request ex post after agreeing to the terms on which the request is based ex ante. More to the point, requiring lead plaintiffs to second-guess ex post the fee terms they contract for ex ante would weaken class counsel’s incentive to expend resources by rendering the agreed terms unreliable. Such a requirement would also deviate from the usual understanding of contracts, which is that they commit parties in advance to terms that cannot be altered unilaterally after one side has substantially performed.
In many cases, review of the retainer agreement seems to come only at the end of the case when the court is awarding fees rather than when it selects the lead plaintiff and lead counsel. 223 See, e.g., Union Asset Mgmt. Holding A.G. v. Dell, Inc., 669 F.3d 632, 644–45 (5th Cir. 2012) (discussing district court’s consideration of fee agreement between lead plaintiff and class counsel when awarding fees from the common fund recovery); In re Enron Corp. Sec., Derivative & ERISA Litig., 586 F. Supp. 2d 732, 749 (S.D. Tex. 2008) (reviewing fee agreement ex post and applying favorable presumption to its terms). Courts also tend to be frustratingly vague when it comes to specifying the circumstances under which they will choose not to defer to such agreements. For example, in In re AT&T Corp. Securities Litigation, the Third Circuit emphasized that:

[T]he presumption of reasonableness set forth in Cendant does not diminish a court’s responsibility to closely scrutinize all fee arrangements to ensure fees do not exceed a reasonable amount. We caution against affording the presumption too much weight at the expense of the court’s duty to act as “a fiduciary guarding the rights of absent class members.” 224 455 F.3d at 168–69 (citing In re Cendant Corp. Litig., 264 F.3d 201, 231 (3d Cir. 2001)); see also In re Bristol-Myers Squibb Sec. Litig., 361 F. Supp. 2d 229, 236 n.8 (S.D.N.Y. 2005) (“[T]he fact that the 7.5% fee was negotiated with institutional Lead Plaintiffs . . . should not and, here, does not, lead to the conclusion that this percentage is presumptively fair.”).

Vague statements such as these create too much uncertainty over whether and to what extent the court will actually defer to a vigorously negotiated ex ante fee agreement.

The following proposal addresses all of these concerns.

B. The Proposal

As a general matter, the authorities discussed in the preceding section are on the right track: All support a shift from the traditional end-of-litigation approach currently taken by courts to set fees in most class actions to a start-of-litigation regime in class actions governed by the PSLRA. Having considered the details of those prior recommendations, which vary, we tentatively propose the following set of arrangements to improve the PSLRA’s effectiveness:

  1. The lead plaintiff should negotiate a fee when retaining counsel to handle the case;
  2. The lead plaintiff should disclose the terms of the negotiated fee to the district court when offering a law firm for appointment as class counsel; 225 Plaintiffs have legitimate reasons for wanting to prevent defendants from acquiring information about their fee terms, especially at the outset of litigation. Therefore, fee agreements should be submitted under seal for in camera review by the district court. The 2002 Task Force made a similar proposal for dealing with bids submitted in connection with fee auctions. See 2002 Task Force Report, supra note 195, at 393–94 (“As a policy, sealed bids make sense. If the defendant knows about the terms of the winning bid, it will be aware of class counsel’s financial incentives.”).

    Fee terms have been publicly disclosed in class actions that produced partial settlements. This happened, for example, in the Global Crossing litigation, which produced five settlements. When seeking approval of the fee request in the first settlement, class counsel revealed that it had entered into an ex ante agreement providing for a declining scale of fee percentages. Lead Counsel’s Memorandum of Law in Support of Petition for an Award of Attorneys’ Fees and Reimbursement of Expenses at 5–6, In re Glob. Sec. Litig., No. 1:02-cv-00910 (S.D.N.Y. July 29, 2004) (on file with the Columbia Law Review). The disclosure enabled the non-settling defendants to identify break points in the fee schedule where, in future settlements, class counsel’s marginal rate of compensation would decline. This problem can be avoided by filing fee agreements under seal for in camera review by the court, which may refer generally to the terms when awarding fees without discussing the terms in detail.
  3. The district court should review the negotiated fee terms before appointing class counsel and should uphold them unless they are clearly unreasonable or not the products of arm’s-length negotiations; and
  4. When reviewing class counsel’s request for a fee award at the end of litigation, the district court should apply the agreed terms unless unforeseen developments have rendered those terms clearly excessive or unfair. In the rare instance in which a court determines that the agreed terms merit modification, the court should provide an opinion that articulates its reasons for deviating from the agreed terms.

Our nationwide data 226 See supra section II.B (discussing nationwide data). suggest that these proposed procedures will require most courts, lead plaintiffs, and law firms that represent plaintiffs in securities class actions to change their current practices. 227 See supra section II.B.1.a (concluding courts do not consider fee agreements when choosing lead plaintiff, but suggesting judges could demand information regarding fee agreements ex ante). But these changes should be relatively simple to make. And, as the following sections explain, the result should be fee awards that more closely approximate the “right price” as envisioned by the PSLRA.

1. The Proposal Respects the PSLRA’s Preference for Private Ordering. — In keeping with the purpose of the PSLRA, our proposal treats the lead plaintiff as the bargaining agent for the class, with the district court judge serving as a backstop in case the lead plaintiff fails to do its job. 228 The Conference Report states the intent of the PSLRA in general terms—the expectation being that “the lead plaintiff provision [would] encourage institutional investors to take a more active role in securities class action lawsuits.” H.R. Rep. No. 104-369, at 34 (1995), reprinted in 1995 U.S.C.C.A.N. 730, 733. In support, the Conference Report quotes Weiss & Beckerman, supra note 26, at 2105, as follows: “Institutions with large stakes in class actions have much the same interests as the plaintiff class generally; thus, courts could be more confident settlements negotiated under the supervision of institutional plaintiffs were ‘fair and reasonable’ than is the case with settlements negotiated by unsupervised plaintiffs’ attorneys.’’ H.R. Rep. No. 104-369, at 35 (1995), reprinted in 1995 U.S.C.C.A.N. 730, 734 n.6. This is one example of a lead plaintiff serving as a reliable bargaining agent for a class. The Conference Report then adds a second example, observing that the “lead plaintiff provision solves the dilemma of who will serve as class counsel. Subject to court approval, the most adequate plaintiff retains class counsel. As a result, the Conference Committee expects that the plaintiff will choose counsel rather than, as is true today, counsel choosing the plaintiff.” Id. Plainly, Congress hoped that lead plaintiffs in securities class actions would bargain effectively on behalf of all investors when prosecuting class actions. See Charles Silver, Reasonable Attorneys’ Fees in Securities Class Actions: A Reply to Mr. Schneider, NAPPA Report, Aug. 2006, at 7, 7 (“[J]udges should intercede only when there is good reason to believe that lead plaintiffs failed to set fees at market rates.”). This arrangement makes sense because many lead plaintiffs, especially sophisticated institutional investors with large financial stakes, can be relied upon to demand appropriate terms when hiring attorneys. 229 See Fisch, Auction Block, supra note 129, at 718–19 (“More troubling is the court’s rejection of selection decisions made through a competitive process by investors with substantial stakes . . . [which] either reflect a rejection of the lead plaintiff’s ability to perform its role or an effort to try to whipsaw counsel into accepting a lower price.”). They have good information about the market for fees because lawyers compete for opportunities to represent them. They have considerable bargaining leverage over attorneys, for the same reason. They know that price and quality matter, not just price alone, so they will offer higher fees when, in their judgment, higher fees are likely to generate larger net recoveries. Lastly, they have incentives to save money. By promising supra-market rates for legal services, they would needlessly diminish their own recoveries. These effects are not just theoretical. Our empirical analyses show that cases with public pension lead plaintiffs and ex ante fee agreements have fees that are materially different from those in other cases. 230 See supra section II.B.1.b (discussing fee requests and awards).

Not all lead plaintiffs are likely to bargain hard over fees, however. Even today, many class actions are led by individual investors with fairly small stakes in the litigation. 231 See supra section II.B.1.c (discussing judicial fee reductions); see also Cox, Thomas & Bai, supra note 30, at 366 (finding empirically “greatest number of securities class action settlements have as their plaintiff either an individual or a group of individuals, but not a financial institution”). Our data showed that the fees in cases led by union-affiliated institutional investors are statistically indistinguishable from the fees in cases led by individual investors. These plaintiffs may have little incentive to bargain, limited access to counsel, and little information. Other lead plaintiffs are pension funds whose managers may favor particular lawyers as a result of past campaign contributions or other relationships. 232 See 2002 Task Force Report, supra note 195, at 402–03, 411–12 (noting negative consequences of “pay-to-play” dynamic); see also supra note 104 and accompanying text (discussing “pay-to-play” dynamic). They may agree to pay higher fees than they should because their agents have been corrupted. Ex ante review of fee agreements enables judges to distinguish lead plaintiffs who are doing their jobs from those who are not, before litigation proceeds very far.

To make this assessment, judges will need evidence. Therefore, lead plaintiff candidates would be expected to submit materials showing that they acted as zealous agents for the absent investors they seek to represent. For example, a pension fund’s general counsel might prepare an affidavit describing the number of law firms that monitor the fund’s portfolio for signs of fraud, the number of firms that bid for the opportunity to represent the fund in the current litigation, the range of the competing bids, and the reason(s) supporting the choice of the law firm offered to the court as counsel for the class. The affidavit might also describe the relationship between the fund and the law firm, including any political contributions that lawyers associated with the firm may have made to anyone associated with the fund, along with the fees the fund agreed to pay outside counsel when suing on its own in other matters.

Over time, the body of evidence bearing on the reasonableness of ex ante fee contracts would grow, and judges’ initial fee assessments would become better informed and more reliable. In cases with competing lead plaintiff candidates, judges could require all contenders to submit their fee agreements. This would increase the amount of information available and pressure all lawyers involved to offer market rates. Judges would also have a firm basis for disciplining lead plaintiffs who failed to bargain hard by referencing fees negotiated in comparable cases. If this evidence were readily available, it is reasonable to expect that it would have the greatest impact on, and utility for, judges in low-volume districts or judges with little or no prior experience with securities class actions. As we have shown, it is in cases presided over by those judges that lawyers have the greatest ability to exploit the existing imperfections in the market for attorneys’ fees. 233 See supra notes 116–120 and accompanying text (summarizing empirical data on fee-award differences between high- and low-volume districts and judges). Making these data readily available to all judges may help to reduce the current information asymmetries that we believe underlie our empirical findings. 234 See supra section II.B.2.a (discussing effect of ex ante fee agreements on fee requests).

When reviewing negotiated fee terms, district court judges would either uphold them or find them clearly unreasonable. Both findings would require an evidentiary basis. A visceral reaction by the judge that a proposed fee arrangement promises too much money to lawyers would not suffice; nor would other unsubstantiated concerns. Rather, when striking down a fee negotiated ex ante, a judge would have to show that the lead plaintiff failed to act as a zealous agent for the class or that the proposed fee deviates from the market rate so greatly as to be unacceptable.

Even so, the fee-setting process would be simpler than it is today. For the initial review, a judge would need only the lead plaintiff’s evidence and information about prevailing market rates, which could be obtained from an expert report. For the final review, a judge would require evidence of unforeseen developments, if any, that rendered the previously approved, negotiated fee terms clearly excessive or unfair. For example, a defendant may have offered to settle immediately after the lead plaintiff was appointed, so that the result obtained could not possibly be attributed to class counsel’s work. 235 The category of unforeseen developments should include only occurrences that normally prevent lawyers from collecting contingent fees agreed to in advance. Ordinarily, neither a quick recovery, a large recovery, nor a high effective hourly rate has this effect. Consequently, that the enforcement of fee terms agreed to ex ante and judicially approved at the start of class litigation would yield a high effective hourly rate would not ordinarily count as a basis for revising those terms downward ex post. Ordinarily, such evidence, when it exists, will be within the court’s judicial cognizance or could readily be obtained by questioning the parties.

By comparison, the existing fee-setting process is inefficient and unprincipled. When petitioning for fees, lawyers may feel pressure to inflate their hours and billing rates. 236 The history of lodestar-based fee awards in class actions is replete with allegations of over-billing. A famous discussion of the charges leveled in In re Fine Paper Antitrust Litigation appears in John C. Coffee, Jr., Rescuing the Private Attorney General: Why the Model of the Lawyer As Bounty Hunter Is Not Working, 42 Md. L. Rev. 215, 252–61 (1983) [hereinafter Coffee, Rescuing the Private Attorney General]; see also In re Fine Paper Antitrust Litig., 751 F.2d 562, 570–71 (3d Cir. 1984) (discussing allegations). Objectors scour fee filings for errors and improper charges, accuse class counsel of being greedy, and attempt to extract payments by filing appeals. 237 In many instances, activities like these are designed to set up appeals of settlement approvals that are accompanied by demands for payoffs. For a discussion, see Fitzpatrick, Objector Blackmail, supra note 23, at 1633–40 (describing dynamics of “objector blackmail”). Judges approve fees without knowing how the requested fees compare with those prevailing in the market for legal services. 238 Judges sometimes contend that reliable data on prevailing market rates are unavailable. See Vizcaino v. Microsoft Corp., 290 F.3d 1043, 1050 (9th Cir. 2002) (asserting efforts to mimic market when setting fees for class counsel are “entirely illusory”); Goldberger v. Integrated Res., Inc., 209 F.3d 43, 52 (2d Cir. 2000) (observing “‘hard data’ on analogous situations—such as attorneys’ fees sophisticated corporate plaintiffs typically agree to pay—are ‘sketchy’” (quoting William J. Lynk, The Courts and the Plaintiff’s Bar: Awarding the Attorney’s Fee in Class-Action Litigation, 23 J. Legal Stud. 185, 187 (1994))). And they cut fees randomly, for reasons that are often unspecified or unclear. 239 See supra section II.B.1.c (discussing difficulty of predicting judicial fee reductions). By tying fees to the market and reducing the need for lodestar cross-checks on the back end, our proposal would diminish the burden on judges and all but eliminate opportunities for extortionate objections.

The existing federal class action rule empowers judges to demand information about fees and to set fee terms when appointing class counsel. Rule 23 permits a judge to order candidates for the role of class counsel “to provide information on any subject pertinent to the appointment and to propose terms for attorney’s fees and nontaxable costs.” 240 Fed. R. Civ. P. 23(g)(1)(C). It also allows judges to include “provisions about the award of attorney’s fees or nontaxable costs” in the order appointing class counsel. 241 Fed. R. Civ. P. 23(g)(1)(D). One can plausibly argue that judges should consider ex ante fee agreements when assessing lead plaintiffs’ adequacy as well.

When reviewing fees upfront, judges would have to receive materials in camera and possibly meet with competing lead plaintiff candidates and their attorneys on an ex parte basis. Although one-sided procedures are disfavored in litigation, 242 See Steven Lubet, Ex Parte Communications: An Issue in Judicial Conduct, 74 Judicature 96, 96 (1990) (“The Code of Judicial Conduct provides that, except as authorized by law, judges may ‘neither initiate nor consider ex parte or other communications concerning a pending or impending proceeding.’” (quoting Model Code of Judicial Conduct Canon 3(A)(4) (Am. Bar Ass’n 1990))). they are required in this instance because defendants must be denied information about plaintiffs’ fee arrangements. Otherwise, defendants might use this information to class members’ disadvantage, such as by proposing settlements that target break-points in class counsel’s fee schedule. 243 For example, suppose that a defendant knew that class counsel’s fee percentage would decline from 25% of marginal dollars to 10% when the settlement exceeded $50 million. The defendant could then offer exactly $50 million, knowing that class counsel would have little incentive to hold out for more. If the reasonable settlement value was, say, $75 million, the defendant might save the additional $25 million, on which class counsel stood to earn only an extra $2.5 million. Why would class counsel risk $12.5 million in fees (25% of $50 million) by rejecting the offer on the table in order to earn only $2.5 million more? This is why most judges insisted on secrecy when experimenting with auctions as a means of setting class counsel’s fees. 244 See 2002 Task Force Report, supra note 195, at 366–67 (noting judges keep bids sealed “to prevent defendants from obtaining an unfair advantage by knowing the winning bidder’s fee arrangement” because defendants “might be able to pressure counsel into a settlement that maximizes the attorney’s recovery but is unfair to the class”). Judges have also preserved confidentiality by discussing fees with class counsel in chambers on an ex parte basis. This occurred in In re HealthSouth Corp. Securities Litigation, where, after appointing the lead plaintiffs, Judge Karon O. Bowdre ordered their “counsel to appear for [an] ex parte conference in [the judge’s chambers] . . . to discuss fee arrangements.” 245 Memorandum Opinion and Order at 1, In re HealthSouth Corp. Sec. Litig., CV-03-BE-1500-S (N.D. Ala. July 14, 2005) (on file with the Columbia Law Review).

2. The Proposal Creates Superior Incentives to Invest in Litigation. — By setting class counsel’s fees at the appointment stage, courts would strengthen class counsel’s incentives to devote resources to the litigation, while also compensating them appropriately for incurring risks. When fees are first addressed at the end of litigation, lawyers’ rights are under-specified. 246 See Coffee, Rescuing the Private Attorney General, supra note 236, at 233 (noting incentives and risk for attorneys investing in private action). They know what they will receive if they lose—nothing—but they can only guess at what they will be paid if they win. Because class actions require lawyers to bear large, undiversified risks in terms of both time and cash outlay, uncertainty surrounding their compensation terms likely inclines them toward parsimony. This predictably harms investors by reducing their recoveries. 247 See id. at 230–35 (arguing risk aversion, inadequately specified fee interests, and other defects in class action regulation, lead plaintiffs’ attorneys to engage in low-intensity litigation that produces cheap settlements).

a. Lawyers Currently Have Difficulty Predicting Fees. — Within broad limits, fee awards in securities class actions are predictable. Indeed, the evidence compiled here suggests that plaintiffs’ lawyers, particularly when they are dealing with an experienced judge from a high-volume district, adjust their requests based on their assessments of what the judge is likely to award in fees. Still, the limits on judicial discretion remain broad, the outliers among fee awards are not trivial, and the variables that ultimately will be most relevant to the court’s fee determination are unknowable at the commencement of the case. 248 See Samuel R. Berger, Court Awarded Attorneys’ Fees: What is “Reasonable”?, 126 U. Pa. L. Rev. 281, 284 (1977) (recounting leading class-action lawyer’s argument that fee request outcomes largely depend “upon ‘the roll of the dice’—from court to court and from case to case”).

At the most general level, in the 148 securities fraud cases led by public pension funds in this study’s nationwide data set, mean fee awards were 19.53% with a standard deviation of 7.80%. This implies that in about two-thirds of the cases the fee award fell within the 11.73%–27.32% range, while awards in the remaining cases were above or below these extremes. Settlement size is by far the best predictor of fee awards (explaining more than 90% of the variation in fee awards 249 See Eisenberg & Miller, Fees and Expenses, supra note 4, at 278–79 (presenting empirical findings that “over 90 percent of the variance in the fee is explained by the size of the recovery”). ), but at the outset of a litigation lawyers cannot predict settlement size with any degree of precision. And even if they could, the lawyers would still have difficulty predicting their eventual fees because courts often deviate from prevailing norms—usually by awarding less than settlement size alone would lead one to expect. 250 See supra section II.B.2.c (analyzing variables correlated with judicial fee reductions). With little ability to predict the cases in which such reductions will occur and given a predictable downward bias from courts, a rational plaintiffs’ attorney might well choose to under-invest in the average case in order to guard against a substantial reduction in fees in any given case.

Fee awards also vary in response to other factors whose impact, in the absence of reliable ex ante fee terms, can only be guessed. For example, in percentage terms, fees and recoveries vary inversely. A larger recovery may imply a smaller percentage fee, although our data have shown that relationship to prevail only in the high-volume districts. 251 See Jordan Milev et al., Recent Trends in Securities Class Action Litigation: 2011 Year-End Review 21 (2011), [] (studying post–PSLRA cases and reporting “median plaintiffs’ attorney[s’] fees [were] a third of the settlement amount [for settlements below $5 million], while for settlements of over $500 million, fees [fell] to below 10%”). But in any given case, the relationship is poorly specified. Judges exercise broad discretion and they vary in their generosity toward plaintiffs’ lawyers, with substantial differences, in courts that see securities class actions more or less frequently. 252 See supra section II.B.2 (analyzing differences between high- and low-volume districts). Class counsel in a case in a high-volume district can be confident that greater success will cause their fee percentages to decline, but they cannot predict the magnitude of the effect. Long-term trends also matter. Perino found that “[m]ean fees in cases settled after 2002 were 24.9 percent, significantly lower than the period prior to 1999 (28.1 percent) or the period 1999–2002 (27.1 percent).” 253 Perino, Activism Through Litigation, supra note 4, at 389. Lawyers who file cases today can only guess whether the trend toward lower fees will stop, continue, or accelerate.

b. Uncertainty over Fees Harms Class Members. — Notwithstanding the lack of an empirical study showing that investors do better when class counsel’s fees are set ex ante rather than ex post, it seems safe to infer that the failure to specify lawyers’ rights in advance discourages them from investing optimally in cases. 254 Empirical studies conducted prior to this one did find that the presence of institutional investors, chiefly public pension funds, as lead plaintiffs reduced agency costs in securities class actions. See Perino, Institutional Fiduciaries, supra note 28, at 146–47 (reviewing studies). But the authors of these studies had no data on the time or manner in which fee awards were set, so they could not distinguish between cases in which fees were set ex ante and ex post. When sophisticated clients hire lawyers in private market transactions, which are presumably efficient, they routinely set fee terms up front. 255 See David L. Schwartz, The Rise of Contingent Fee Representation in Patent Litigation, 64 Ala. L. Rev. 335, 360 (2012) (reporting lawyers and clients generally fix fee percentages and responsibility for expenses upfront in patent representations); see also In re Synthroid Mktg. Litig., 264 F.3d 712, 718 (7th Cir. 2001) (“This is what happens in actual markets. Individual clients and their lawyers never wait until after recovery is secured to contract for fees. They strike their bargains before work begins.”). This is “[t]he best time” to negotiate because, as Judge Frank Easterbrook observed, “[o]nly ex ante can bargaining occur in the shadow of litigation’s uncertainty; only ex ante can the costs and benefits of particular [compensation] systems and risk multipliers be assessed intelligently.” 256 Synthroid, 264 F.3d at 718–19.

An empirical study of patent engagements supports Judge Easterbrook’s claim that sophisticated clients set lawyers’ fee terms ex ante. When Professor David L. Schwartz studied contingent-fee patent representations, he found highly structured working arrangements and consistent ex ante bargaining over fee percentages and responsibility for costs. 257 Schwartz, supra note 255, at 360 (reporting contingent-fee patent-litigation industry developed “graduated rate” and “flat rate” as “two main ways of setting the fees” and two most often negotiated provisions of contingent-fee agreements were “contingent percent and whether the law firm advanced the costs of the litigation”); see also Stephen D. Susman, Fee Agreements in Technology Litigation: Harmonizing Client and Attorney Interests, Presented at 20th Annual Technology and Computer Law Conference 1–2 (May 23–25, 2007) (on file with the Columbia Law Review) (providing sample terms used in fee agreements in patent representations, including terms governing contingent percentages and responsibility for litigation expenses). Ex ante bargaining also occurs in securities lawsuits when sophisticated clients opt out of class actions and hire lawyers to pursue tag-along suits. 258 See Synthroid, 264 F.3d at 720–22 (looking to securities “benchmark[s] for determining legal fees”); see also Kevin LaCroix, Ohio Joins the Time Warner Opt-Out Settlement Parade, D&O Diary (Mar. 7, 2007),
JG33-UGEY] (reporting State of Ohio agreed to pay “17.5% contingency fee to Lerach Coughlin” for representation in opt-out securities case).
To the best of our knowledge, class actions comprise the only area of contingent-fee practice in which compensation terms are currently fixed when litigation ends. 259 State bar disciplinary rules require contingent-fee arrangements to be set out in writings that expressly address compensation percentages, responsibility for litigation costs, and other matters. But these rules do not require the terms to be written down at the outset of litigation. See, e.g., Model Rules of Prof’l Conduct r. 1.5 cmt. (Am. Bar Ass’n 2010) (suggesting attorneys provide clients with customary fee-arrangement information in new client-lawyer relationships).

Why do clients and lawyers normally set contingent percentage fees when representations begin? Many reasons, including some identified by Judge Easterbrook, 260 Synthroid, 264 F.3d at 718–19 (discussing ethical and practical reasons for setting fees ex ante). are apparent. First, it is usually easier to negotiate fees ex ante than ex post. At the start of litigation, there is no money to divide. There is only the prospect of forming a joint venture between a client and a lawyer that seeks to maximize the parties’ joint wealth by offering the lawyer compensation terms that will motivate the lawyer to work hard on behalf of the client.

When fees are set at the end of litigation, by contrast, the amount to be recovered is already known. This heightens the conflict between the client and the attorney because every additional dollar for one means a dollar less for the other. This is why judges, who sit as absent class members’ guardians, may think fee-setting is a zero-sum game, and also may think they are protecting class members when they set fees as close to zero as possible. They may see no upside to class members from paying higher fees because they are setting fees at a time when lawyers’ services are no longer required.

Second, lawyers and clients should want to set compensation terms early on because this encourages lawyers to bear beneficial risks. Principals hire agents because agents can perform certain tasks more efficiently than principals can themselves. 261 See generally Principles of the Law of Aggregate Litig. § 1.05 cmt. a (Am. Law Inst. 2010) (citing Adolf A. Berle & Gardiner C. Means, The Modern Corporation and Private Property (1932)); Eugene F. Fama & Michael C. Jensen, Separation of Ownership and Control, 26 J.L. & Econ. 301 (1983); Michael C. Jensen & William H. Meckling, Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure, 3 J. Fin. Econ. 305 (1976) (“A foundational insight of the economic literature on agency relation­ships is that ownership of assets and control of their disposition must often be separated to achieve economies of scale, to take advantage of the division and specialization of labor, to bear risks efficiently, and to realize other advantages.”). (Otherwise, principals would be better off acting directly.) In contexts where lawyers work on contingency, efficient working arrangements saddle lawyers with risks they can bear more readily than clients. This includes risks associated with the deployment of legal services. Lawyers pay for these services initially by lending their time and bearing costs. Compensation and reimbursement come, when they do, from recovered funds. Under common arrangements, lawyers are reimbursed for expenses dollar for dollar. 262 The reimbursed expenses do not usually include the law firm’s general “overhead.” The overhead expenses are ultimately borne by the lawyers (out of the “fees” portion of their award), consistent with standard practice in other contingent-fee arrangements. See, e.g., ABA Comm’n on Ethics & Prof’l Responsibility, Formal Op. 93-379 (1993) (“A lawyer may not charge a client for overhead expenses generally associated with properly maintaining, staffing and equipping an office; however, the lawyer may recoup expenses reasonably incurred in connection with the client’s matter . . . so long as the charge reasonably reflects the lawyer’s actual cost for the services rendered.”). For a compelling argument that class action attorneys should be able to receive a greater than dollar-for-dollar return on their investments of money (in addition to time) in contingent-fee litigation, see Morris A. Ratner & William B. Rubenstein, Profit for Costs, 63 DePaul L. Rev. 587, 587–90, 613 (2014) (explaining that allowing profits for costs would increase judicial access). Their profit comes entirely from the percentage fee. 263 Of course, attorneys may have incentives to inflate their costs as an alternative way to profit from a case. Indeed, in the minority of cases in which there are objections to the settlement or fee request, allegations that attorneys are inflating their costs are a common theme. This compensation structure creates a strong, though still imperfect, alignment of interests because it encourages lawyers to bear the costs and risks associated with litigation activities that are likely to benefit clients. It also reduces the monitoring burden on clients, who can usually trust lawyers to exercise good judgment when delivering services, incurring expenses, and evaluating settlement offers.

In contingent-fee relationships, then, the fee percentage does the heavy lifting. It motivates lawyers to help clients by making good decisions that maximize joint gains. This is why clients and lawyers both benefit from fixing lawyers’ compensation in advance. Leaving the core fee term unsettled creates uncertainty about compensation, causing lawyers to decline some risks that clients would rationally want them to take. The predictable result will be suboptimal recoveries that leave clients and lawyers both poorer than they might have been.

Third, when contingent-fee percentages are set ex ante, the risks associated with litigation are palpable because no one knows how they will play out. Will the class recover? If so, how large will the settlement be? The answers can only be estimated at the start of litigation. By contrast, when judges set fees ex post, everything is known. This creates significant potential for the hindsight bias to poison judges’ assessments of litigation risks. The predictable result is that fee percentages will be set too low.

Hindsight bias is a defect in human reasoning that causes people who know actual outcomes to misestimate ex ante odds. 264 For discussions of the hindsight bias and its effect on judgments about legal ma­tters, see Christine Jolls et al., A Behavioral Approach to Law and Economics, 50 Stan. L. Rev. 1471, 1523 (1998) (explaining jurors’ hindsight bias); Jeffrey J. Rachlinski, A Positive Psychological Theory of Judging in Hindsight, 65 U. Chi. L. Rev. 571, 600–01 (1998) (discussing legal effect of hindsight bias); Jeffrey J. Rachlinski, Heuristics and Biases in the Courts: Ignorance or Adaptation?, 79 Or. L. Rev. 61, 72 (2000) (highlighting hindsight bias’ effects on court determinations regarding burdens of proof). It specifically leads people to overestimate the ex ante probability of the outcomes they observe. To document the impact of hindsight bias on federal judges, researchers gave more than 150 federal magistrates a statement describing a case in which a prisoner appealed after being sanctioned by a trial judge for filing a frivolous complaint. 265 Chris Guthrie et al., Inside the Judicial Mind, 86 Cornell L. Rev. 777, 801–02 (2001) (explaining judicial study methodology). One-third of the statements indicated that the appellate court affirmed the sanction; another third indicated that the appellate court imposed a lesser sanction; and the last third indicated that the appellate court vacated the sanction entirely. 266 See id. (“We randomly assigned the judges to one of three conditions: the ‘Lesser Sanction’ condition; the ‘Affirmed’ condition; or the ‘Vacated’ condition.”). The judges were then asked to make a prediction of the result that was most likely to occur on appeal. As expected and in keeping with the hindsight bias, their estimates reflected the information they received about the actual outcomes. The judges were “much more likely” to choose the outcomes they were told had actually occurred. 267 Id. at 803.

The parties to real client–lawyer relationships would fare poorly if they allowed hindsight bias to influence their assessments of lawyers’ compensation. Knowing that litigation produced a recovery, they would overestimate the ex ante odds of winning and set lawyers’ fee percentages too low. This would discourage lawyers from taking risks when they should. Setting fees up front and enforcing ex ante fee agreements avoids this problem. 268 See Kerr v. Quinn, 692 F.2d 875, 878 (2d Cir. 1982) (“[T]he likely response of the bar [to a request for representation] is to be determined in light of the posture of the case at the time counsel is sought, not in the light of hindsight.”); see also id. (“The statutory purpose of encouraging the bringing of meritorious claims is not furthered by denying attorney’s fees in cases which happen to turn out well for plaintiffs without regard to the initial risk.”).

When fees are set ex post, the outcomes of all litigation risks are known and the hindsight bias has many opportunities to distort judges’ estimates of ex ante odds. Judges know, for example, whether the complaint survived the motion to dismiss (instead of failing it as complaints often do), 269 On rates at which motions to dismiss securities fraud class action complaints are granted, see Cornerstone Research, Securities Class Action Filings: 2012 Year in Review 9 fig.7 (2013),‌Securities-Class-Action-Filings‌%E2%80%942012-Year-in-Revie.pdf [] (reporting 38% of securities class actions lasting longer than 60 days were dismissed on first motion to dismiss). whether a class was certified (which can be a significant hurdle), 270 On certification rates in securities class actions where certification is contested, see Cornerstone Research, Securities Class Action Filings: 2013 Year in Review 10 (2014),‌GetAttachment/‌d88bd527-25b5-4c54-8d40-2b13da0d0779/Securities-Class-Action-Filings-2013-Year-in-Review.pdf [] (“For cases filed between 2002 and 2010, 48 percent were granted class certification in part or in full.”). whether the class obtained a recovery (which not all class actions do), and whether the recovery was large (which is a rarer accomplishment still). 271 On the rates at which securities class actions obtain recoveries, see Cornerstone Research, Securities Class Action Filings: 2014 Year in Review 12 (2015), [] (“For filings from 1996 to 2013, 49 percent have settled, 41 percent have been dismissed, and 9 percent are ongoing.”). On the size of recoveries in securities class actions, see Cornerstone Research, Securities Class Action Settlements: 2014 Review and Analysis 1 (2015), [] (reporting, in 2014, “total settlement dollars in securities class actions hit their lowest mark in 16 years,” median settlement was $6 million, and largest settlement was $265 million). Predictably, this information will cause judges to overestimate the lawyers’ likelihood of success and underestimate the risk the lawyers’ bore when they began the litigation, thereby exerting downward pressure on the judges’ assessments of appropriate fees. Case law expressly ties fees percentages to the risks lawyers incur. 272 See, e.g., Singleton v. Domino’s Pizza, LLC, 976 F. Supp. 2d 665, 683 (D. Md. 2013) (“In determining the reasonableness of an attorneys’ fee award, courts consider the relative risk involved in litigating the specific matter compared to the general risks incurred by attorneys taking on class actions on a contingency basis.” (quoting Jones v. Dominion Res. Servs., 601 F. Supp. 2d 756, 762 (S.D. W. Va. 2009))); see also In re Schering-Plough Corp. Enhance Sec. Litig., No. CIV.A. 08-2177 DMC, 2013 WL 5505744, at *3 (D.N.J. Oct. 1, 2013) (“In reviewing an attorneys’ fees award in a class action settlement, the Third Circuit looks at a number of factors . . . [including] the risk of nonpayment . . . .”). Judges who are convinced that litigation risks were low cannot be expected to award high fees.

If courts lower fees based on these ex post assessments of risk, they also run the risk of awarding fees that are too low for another reason illustrated in the empirical data. Plaintiffs’ lawyers are likely to have the greatest expertise in evaluating ex ante the potential risks and rewards of any potential case. The data suggest that attorneys consider these risks in their fee requests. Case competition (a proxy for case quality) is negatively correlated with fee requests, i.e., lawyers already ask for lower fees in what appear to be higher quality cases. 273 See supra section II.B.2.a (summarizing regression analysis of relationship between case competition and fee awards). If judges acting ex post were to reduce fees even further, they run a substantial risk of setting fees that are too low given the risks the case entails.

The proposal prevents hindsight bias from distorting judges’ risk assessments by mimicking what normally occurs outside the class action context when clients hire lawyers on contingency. Fees are set at the outset of litigation, before any actual results are known. 274 See In re First Fidelity Bancorporation Sec. Litig., 750 F. Supp. 160, 163 (D.N.J. 1990) (comparing counsel seeking contingent fees ex post “to placing a wager after the outcome of the event is known or playing poker with everyone’s cards face up,” and contending “percentage of recovery should be negotiated and fixed while the risks and amount of recovery are still unknown”).

The theoretical case in favor of setting contingent fees ex ante is thus straightforward. Fee arrangements that motivate lawyers to maximize recoveries serve clients best. The market pressure that would likely exist if the proposal is adopted should therefore encourage lawyers to offer fee arrangements that have this effect. These arrangements will reduce uncertainty about lawyers’ compensation to efficient levels, because higher levels of uncertainty would harm clients by discouraging lawyers from investing optimally. Actual transactions in all market sectors, including transactions involving sophisticated clients, show that efficient arrangements set lawyers’ compensation terms at or near the start of litigation. Consequently, the current judicial practice of setting fees ex post in securities class actions is presumably inefficient and makes investors worse off.

c. The Proposal Restores Objectivity and Reduces Arbitrariness. — Finally, under this proposal, an ex ante fee contract would give class counsel considerable protection from a court’s subsequent, arbitrary fee reduction by establishing a presumption of reasonableness following initial review. By accepting a lead plaintiff’s recommendation of class counsel without complaint, a judge would establish that the lead plaintiff was a zealous bargaining agent and that the proposed fee terms set out in the ex ante fee agreement were reasonable given everything that was then known about the litigation and the prevailing market for attorneys’ services. Because few grounds would exist for a court to revise this assessment ex post, the negotiated fee terms would be reliable and lawyers’ incentives would be secure. A lawyer who challenged an ex post fee cut on appeal would also look like any other lawyer who sues to enforce a contractual right to payment. 275 In addition, a lawyer appealing a fee cut will be less likely to be accused of acting adversely to his/her clients in making such an appeal. By comparison, when appealing a low fee award today, class counsel must establish that a trial judge abused his or her discretion. 276 See In re Nortel Networks Corp. Sec. Litig., 539 F.3d 129, 134 (2d Cir. 2008) (“We will not overturn a district court’s award of attorneys’ fees ‘absent an abuse of discretion.’” (quoting Goldberger v. Integrated Res., Inc., 209 F.3d 43, 47 (2d. Cir. 2000))). Because some evidence providing a basis for a low fee award can likely be found in every record, this is a very hard showing for class counsel to make.

Wal-Mart Stores, Inc. v. Visa U.S.A., Inc., 277 396 F.3d 96 (2d Cir. 2005). an antitrust class action, provides an excellent example of arbitrary fee-setting. There, the Second Circuit affirmed a $220 million fee award on a settlement that provided for about $3.4 billion in compensatory relief. The Second Circuit agreed with the district court judge that the low fee percentage—6.511%—was proper because the recovery was enormous. 278 Id. at 122. When litigation started nearly a decade earlier, of course, no one knew that even $1 would be recovered. 279 Id. at 101 (reporting litigation commenced in 1996). It was therefore a question of fact whether the parties would have given the possibility of a mega-fund recovery much weight when negotiating the lawyers’ compensation ex ante. Because “five of the nation’s largest merchants” agreed to pay more than 18% of their recoveries in fees when engaging counsel for the case, a negative answer enjoyed considerable support. In other words, there was reason to doubt that a sophisticated representative of the entire class would have used the possibility of a mega-fund recovery to bargain for a 6.5% fee. 280 Id. at 123. (One might similarly doubt that experienced attorneys would ever have agreed to so low a contingent fee ex ante.)

The appellate court could have suggested, with greater plausibility, that a sophisticated bargaining agent would have insisted on a sliding scale of fees. This happened in the Enron litigation, in which the Regents of the University of California agreed ex ante to a scale that entitled class counsel to 10% of the first billion dollars recovered, 9% of the second billion, and 8% of any larger amount. 281 In re Enron Corp. Sec., Derivative & ERISA Litig., 586 F. Supp. 2d 732, 778 (S.D. Tex. 2008) (holding 9.52% of settlement “fair and reasonable” attorney-fees award). Given the eventual $7.2 billion recovery, the contractual fee scale entitled class counsel to $688 million. Judge Easterbrook has noted that sliding scales like these make sense in securities class actions:

“Awarding counsel a decreasing percentage of the higher tiers of recovery enables them to recover the principal costs of litigation from the first bands of the award, while allowing the clients to reap more of the benefit at the margin (yet still preserving some incentive for lawyers to strive for these higher awards).”

Silverman v. Motorola Sols., Inc., 739 F.3d 956, 959 (7th Cir. 2013).
The plausibility of a fee scale being readily admitted, it is nonetheless true that we know of no instance in which a real client and a real lawyer agreed that the latter’s compensation would be based on a scale of percentages that would have reduced the Wal-Mart fee into the $220 million range. The Enron scale is the lowest percentage observed during this Article’s study. Had it been applied to the $3.4 billion recovery in Wal-Mart, the fee would have been $302 million. 282 This $302 million figure is calculated based on the sum of $100 million (derived from the product of $1 billion * .10), $90 million (derived from the product of $1 billion * .09), and $112 million (derived from the product of $1.4 billion * .08). The actual fee award was almost $80 million less.

Neither the Second Circuit nor the district court connected the Wal-Mart fee award to the contingent percentage fee paid by any real client in any actual case. Both plucked the 6.511% figure out of thin air. Worse, the Second Circuit downplayed the importance of getting lawyers’ incentives right. It remarked that “[i]n this case, the district court’s decision in favor of protecting the instant class from an excessive fee award militates against awarding attorneys’ fees based purely on economic incentives.” 283 Wal-Mart Stores, 396 F.3d at 123. Plainly, both the Second Circuit and the district court missed the seemingly obvious point that fee awards based on terms that motivate lawyers to maximize recoveries cannot be excessive insofar as class members are concerned. Rather, such fees are precisely the ones that class members would willingly and rationally agree to pay ex ante when hiring lawyers directly. 284 A source of arbitrariness that is hard to document but unquestionably important is the vast difference between federal judges’ meager salaries and the enormous amounts lawyers earn when class actions yield mega-fund recoveries. How can federal district court judges, who earn less than $200,000 per year, not be envious when lawyers whose credentials may be no better than theirs, and who may work no harder than they do, pocket tens or hundreds of millions in fees in a single case?

The analysis just presented appears to have altered the thinking of Judge Gleeson, who was the district court judge in the Wal-Mart case that settled in 2003. 285 See In re Visa Check/Mastermoney Antitrust Litig., 297 F. Supp. 2d 503, 522 (E.D.N.Y. 2003) (holding requested attorney fees to be excessive). When awarding fees ten years later in the Payment Card Interchange Fee litigation, which settled for $5.7 billion in 2013, he repeatedly emphasized the need for “concrete guideposts” that would secure lawyers’ financial incentives, and he explicitly took market rates into account. 286 See In re Payment Card Interchange Fee and Merch. Disc. Antitrust Litig., 991 F. Supp. 2d 437 (E.D.N.Y. 2014) (awarding fees in connection with settlement given final approval in In re Payment Card Interchange Fee and Merch. Disc. Antitrust Litig., 986 F. Supp. 2d 207 (E.D.N.Y. 2013)). The $5.7 billion is after reductions for opt outs. See id. at 439. Judge Gleeson mentioned “concrete guideposts” in In re Payment Card Interchange Fee, 991 F. Supp. 2d at 442 (“The law sets only minimal constraints on fee awards. Within the boundaries of those constraints, it offers no concrete guideposts.”); id. at 444 (“Despite the absence of concrete guideposts . . . .”); see also id. at 446 (“In my view, a guidepost is sorely needed.”). In this case, one of us, Professor Silver, submitted an expert report on fee awards to which Judge Gleeson referred in his opinion. See id. at 440 (citing Declaration of Professor Charles Silver Concerning the Reasonableness of Class Counsel’s Request for an Award of Attorneys’ Fees, In re Payment Card Interchange Fee, 991 F. Supp. 2d 437 (No. 1:05-md-01720-JG-JO), 2013 WL 1562407). Thus, he adopted a declining scale of marginal percentages because “sophisticated clients often require counsel to accept a smaller percentage of a recovery as the size of the recovery increases.” 287 Id. at 444. He also set the percentages at levels that, in his assessment, resembled those sophisticated clients employ, starting with 33% of the first $10 million recovered and declining steadily across identified ranges until reaching a low of 6% on the last $1.7 billion. 288 Id. at 445. Summing the marginal increments, Judge Gleeson concluded that a 9.56% fee was warranted, a percentage very close to the agreement-based fee approved in Enron. 289 See id. (“The result is a total fee of $544.8 million, or 9.56% of the total fund.”); see also In re Enron Corp. Sec., Derivative & ERISA Litig., 586 F. Supp. 2d 732, 766 (S.D. Tex. 2008) (holding 9.52% fee was “fair and reasonable”). Judge Gleeson in his 2014 opinion also twice endorsed the principle of setting fees in class actions in advance. In the first passage on this subject, he noted that “[i]n PSLRA litigation, a lead plaintiff may bargain with lead counsel over fees.” 290 In re Payment Card Interchange Fee, 991 F. Supp. 2d at 443. Then, after adding that in the Second Circuit “a district court need not defer to such a bargain when actually making the award at the end of litigation,” he observed that “[i]t would be helpful to have a negotiated benchmark from which to work” and said that “in a future case, [he would] consider employing [his] authority under Rule 23(d) to require such negotiation—perhaps with court-appointed counsel to represent a cross-section of the plaintiffs for the purposes of the fee negotiation.” 291 Id.

Obviously, we agree that a negotiated benchmark set by the lead plaintiff and its chosen counsel at or near the start of a securities class action would both make the court’s eventual fee award more objective and give class counsel better incentives during the litigation. We dissent, however, from Judge Gleeson’s proposal to appoint counsel to negotiate on behalf of the class, partly because the PSLRA vests lead plaintiffs with this responsibility and partly because appointed, temporary counsel would have no obvious interest in obtaining fee terms that are optimal for a class. 292 See 15 U.S.C. §§ 77z-1(a)(3)(B)(v), 78u-4(a)(3)(B)(v) (2012) (empowering most adequate plaintiff to “select and retain” class counsel). For a proposal that would use a monitor incentivized to maximize a class’s recovery to set class counsel’s fee, see Alon Klement, Who Should Guard the Guardians? A New Approach for Monitoring Class Action Lawyers, 21 Rev. Litig. 25, 69–70 (2002) (“By awarding the monitor a percentage of the recovery, his interests are effectively aligned with the class; he would choose the lawyer who would maximize the class’s net recovery and offer the best possible fee arrangement for the class.”).

The second time Judge Gleeson discussed ex ante fee-setting in his 2014 opinion, he zeroed in on lawyers’ incentives:

A final reason to employ the schedule methodology advanced here is for the benefit of counsel in future cases. If plaintiffs’ lawyers know in advance (that is, at the start of a case) that such a schedule will be used, it will alter their thinking for the better. A graduated schedule ensures that the greater the settlement, the greater the fee, and it therefore avoids certain incentive problems that come from simply scaling an overall percentage down as the size of the fund increases . . . . Using such a schedule as a guideline for future cases—from which departures based on case-specific circumstances may of course be warranted—will permit counsel to make reasonable decisions ex ante in those future cases. 293 In re Payment Card Interchange Fee, 991 F. Supp. 2d at 446 (internal citation omitted).

This is both exactly right and a significant improvement on the lack of concern about attorney incentives expressed in the Wal-Mart opinions. 294 In re Visa Check/Mastermoney Antitrust Litig., 297 F. Supp. 2d 503 (E.D.N.Y. 2003), aff’d sub nom. Wal-Mart Stores, Inc. v. Visa U.S.A. Inc., 396 F. 3d 96 (2d Cir. 2005). When setting fees, judges should keep uppermost in their minds that they are creating incentives for attorneys. Realizing this, their only object should be to select fee terms that motivate lawyers to maximize net recoveries for claimants. Choosing a fee arrangement for any other reason would disserve class members by discouraging their lawyers from representing them zealously, thereby creating a serious risk that class members would be denied due process of law.

In our nationwide study, we found that judges usually grant fee requests in full and are somewhat more deferential when requests are based on ex ante agreements. 295 See supra section II.B.2.c (presenting regression analyses findings on judicial fee reductions). Readers may therefore wonder why this Article argues the current fee-setting process lacks objectivity and is rife with arbitrariness. Don’t judges mostly do what this Article contends they should do, by regularly approving fee requests supported ex post by lead plaintiffs? No—we are concerned that judges currently act improperly both when they grant fee requests and when they cut them. In most of the 85% of cases where judges awarded the fees that class counsel requested, they neither set fees up front, made any effort to determine whether lead plaintiffs bargained zealously on behalf of the investors they represented, nor compared requested fee percentages to prevailing market rates. 296 See supra section II.B.1.c (discussing empirical findings on judicial fee reductions). Consequently, in these cases it is hard to know whether fee awards were too large, too small, or about right. The only thing one can be confident of is that, while the lawsuits were ongoing, the lawyers’ incentives were deficient because their compensation terms were vague. In the remaining 15% of the cases where judges cut fee requests, the same three problems were present. Fee reductions only made the fact of arbitrariness clearer by highlighting defects in the way judges think about fee regulation. And, as the data show, those fee reductions were random or the product of unobserved variables. 297 See supra section II.B.2.c (discussing judicial fee reductions).

In nearly all cases in this Article’s nationwide six-year study, judges appear to have misunderstood the role that ex ante fee agreements between lead plaintiffs and lawyers should play in the fee-setting process under the PSLRA. Usually, judges displayed no interest in the content or existence of such ex ante agreements. Judges did not require lead plaintiff candidates to submit them as part of the application process, and the candidates rarely volunteered them. Similarly, only a handful of judges expressed interest in these agreements when evaluating class counsel’s request for fees at the time of settlement. We are left to surmise that most federal district court judges may have failed to grasp that the core implication of the PSLRA is that a lead plaintiff is supposed to act as a zealous bargaining agent for a class, and that Congress therefore assigned lead plaintiffs weightier responsibilities than named plaintiffs in other class actions are normally required to bear. 298 Even so, one could argue (1) that judges should also consider ex ante fee agreements when deciding whether a lead plaintiff candidate and proposed class counsel are adequate under Rule 23(a) of the Federal Rules of Civil Procedure, and (2) that judges should make more frequent use of the power to set fees up front recognized in Rule 23(g). To ensure that lead plaintiffs do the job they sign up for, judges should obtain and evaluate ex ante fee agreements routinely at the outset of the case.


Every year, federal judges presiding over securities class actions award billions of dollars in attorneys’ fees. Until now, little has been known about how these awards are actually made. The nationwide, six-year empirical study presented in this Article is the first to peer carefully, comprehensively, and systematically inside the black box of the fee award process. The findings suggest that the current practice of ex post fee-setting is both deeply flawed and not consistent with Congress’s goals when adopting the PSLRA twenty years ago. This Article therefore proposes a set of procedural reforms that courts can easily adopt and that we believe will make fee-setting in securities class actions more transparent, more compatible with the normative goals of the PSLRA, and ultimately more beneficial for class members, class counsel, and the general public.

In the meantime, this Article aspires to encourage potential lead plaintiffs and the attorneys who represent them to privately resolve fee-related matters when litigation begins, and to present their ex ante fee agreements to courts for in camera review when requesting control of class litigation. Doing so should result in fee awards at the time of settlement that are more predictable and more likely to be grounded in prevailing market rates than in the predilections and biases of individual judges.

Finally, we hope that our findings will cause judges who undertake to award attorneys’ fees (or to review such fee awards on appeal) in securities cases and class actions of other types to think more deeply and carefully about how best to perform this task. Academic researchers can contribute to this important judicial work by undertaking empirical studies along the lines of the one presented here. 299 The authors are currently at work on two additional studies, one on the impact of federal district court judges’ backgrounds on their treatment of fees and one on ex ante fee agreements that may have been entered into by lead plaintiffs in securities fraud class actions but not disclosed to courts. Other topics that could usefully be studied empirically include: portfolio monitoring, a service that many plaintiffs’ firms offer to investors in hope of alerting them to instances of fraud and obtaining legal engagements; comparison shopping by institutional investors seeking to hire counsel for securities fraud cases; and fees paid and recoveries obtained by investors who opt out of class actions and pursue individual cases.


Appendix: Fee-Cut Cases 300 The columns represent the following data: (1) Lead Plaintiff Type; (2) Gross Settlement (Millions); (3) Fee Request; (4) Fee Award; (5) Objection to Fee Request?; (6) Award-to-Request Ratio; and (7) Reasons Given for Fee Cut.
For column (1), the Lead Plaintiff Types are indicated as follows: I = Individual; PPF = Public Pension Fund; UF = Union Fund; and PI = Private Institution.
For column (7), the Reasons Given for Fee Cut are indicated as follows: (A) requested fee too large; (B) requested fee too large given work performed; (C) requested fee too large given actual risk of nonrecovery; (D) Court cannot rely on market for attorneys’ fees; (E) requested fee is out of line with fees in similar cases; and (F)requested fee fails lodestar or other cross-check.

Case (Court) (Judge) (1) (2) (3) (4) (5) (6) (7)
Am. Bus. Fin. (E.D. Pa.) (O’Neill, J.) I $16.77 0.292 0.25 Yes 0.86 A
Am. Intl. Gp. (S.D.N.Y.) (Batts, J.) PPF $72 0.1325 0.0909 Yes 0.69
Am. Tower (D. Mass.) (Wolf, J.) UF $14 0.25 0.2269 Yes 0.91
Ambassadors (E.D. Wash.) (Quackenbush, J.) UF $7.5 0.25 0.217 0.87 ABEF
Arotech (E.D.N.Y.) (Dearie, J.) I/PI $2.9 0.3325 0.25 0.75 ABC
ArthroCare (W.D. Tex.) (Sparks, J.) PPF $7.4 0.3 0.2297 0.77 ACEF
Bank of America (S.D.N.Y.) (Castel, J.) PPF $2,425 0.0654 0.0629 Yes 0.96
Barrick Gold (S.D.N.Y.) (Berman, J.) I $24 0.3 0.2 0.67 CEF
Bayer (S.D.N.Y.) (Pauley, J.) I $18.5 0.12 0.1189 0.99 F
Cadence (N.D. Cal.) (Conti, J.) UF $38 0.25 0.189 0.76 AB
Canadian Superior (S.D.N.Y.) (Scheindlin, J.) I $3.12 0.3 0.25 0.83 ABCF
Cell Therapeutics (W.D. Wash.) (Pechman, J.) I $19 0.3 0.23 Yes 0.77 ABCE
Children’s Place (S.D.N.Y.) (Scheindlin, J.) UF $12 0.27 0.15 0.56 ABEF
Chiron (N.D. Cal.) (Walker, J.) UF $30 0.25 0.153 Yes 0.61 ABCDF
Citigroup (S.D.N.Y.) (Stein, J.) I $59 0.165 0.12 Yes 0.73 F
Coca-Cola (N.D. Ga.) (Hunt, J.) UF $13.75 0.2604 0.21 Yes 0.81 AE
Coinstar (W.D. Wash.) (Pechman, J.) PPF $60 0.25 0.23 0.92 BF
Collins & Aikman (1) (E.D. Mich.) (Rosen, J.) UF/
$10.8 0.27 0.231 0.86
Collins & Aikman (2) (E.D. Mich.) (Rosen, J.) I $12.26 0.307 0.2304 0.75
E*Trade (S.D.N.Y.) (Oetken, J.) PI/I $79 0.3333 0.2780 Yes 0.83
EnergySolutions (S.D.N.Y.) (Koeltl, J.) UF/
$26 0.27 0.25 0.93
Escala (S.D.N.Y.) (Hellerstein, J.) PPF $18 0.22 0.1805 0.82 ABC
Evergreen (D. Mass.) (Gorton, J.) UF $25 0.3 0.24 0.80 F
Forest (S.D.N.Y.) (Wood, J.) UF $65 0.245 0.14 0.57 ABC
Globalstar (S.D.N.Y.) (Preska, J.) PPF $1.5 0.2 0.183 0.92
Harmonic (N.D. Cal.) (Hamilton, J.) PI/I $15 0.27 0.25 0.93
Hexion (S.D.N.Y.) (Berman, J.) I $18 0.2 0.07 0.35 ABCEF
Home Solutions (1) (N.D. Tex.) (Godbey, J.) I $3.5 0.3 0.287 0.96
Home Solutions (2) (N.D. Tex.) (Godbey, J.) I $5.1 0.3 0.22 0.73
IMAX (S.D.N.Y.) (Buchwald, J.) PI $12 0.25 0.19 0.76 ABCE
IndyMac (C.D. Cal.) (Wu, J.) I $6.5 0.3 0.25 0.83
JA Solar (S.D.N.Y.) (Koeltl, J.) I $4.5 0.3 0.23 0.77 EF
JP Jeanneret (S.D.N.Y.) (McMahon, J.) UF $207.9 0.196 0.187 Yes 0.95 B
Lehman (S.D.N.Y.) (Kaplan, J.) UF $40 0.1925 0.1289 0.67 ABCF
Magma Design (N.D. Cal.) (Breyer, J.) I $13.5 0.3 0.1855 0.62 ACE
National Lampoon (C.D. Cal.) (Anderson, J.) I $1 0.333 0.25 0.75 ABCE
NexCen (S.D.N.Y.) (Cedarbaum, J.) I $4 0.3 0.2125 0.71 ABCE
NovaGold (S.D.N.Y.) (Cote, J.) PPF $26.78 0.213 0.163 0.77 ABC
Nuvelo (N.D. Cal.) (Breyer, J.) I $8.92 0.3 0.28 Yes 0.93
OCA (E.D. La.) (Vance, J.) I $6.5 0.3 0.285 Yes 0.95 ABEF
ORBCOMM (D. N.J.) (Walls, J.) I $2.45 0.33 0.22 0.67
Pall (S.D.N.Y.) (Seybert, J.) PPF $22.5 0.275 0.222 Yes 0.81
Pro Net Link (S.D.N.Y.) (Preska, J.) I $1.22 0.3 0.2857 0.95
Qwest (D. Colo.) (Blackburn, J.) UF/I $400 0.24 0.15 Yes 0.63 AE
RHI (S.D.N.Y.) (Hellerstein, J.) UF $2.5 0.3 0.076 0.25
Sadia (S.D.N.Y.) (Scheindlin, J.) UF/I $27 0.324 0.3 0.93 AE
Salomon (S.D.N.Y.) (Lynch, J.) PI $35 0.31 0.27 0.87
ShoreTel (N.D. Cal.) (Breyer, J.) I $3 0.3333 0.26 0.78
SLM (S.D.N.Y.) (Pauley, J.) PI $35 0.25 0.225 0.90
Sterling (E.D. Pa.) (Stengel, J.) PPF $10.25 0.3 0.2847 Yes 0.95
TECO (M.D. Fla.) (Whittenmore, J.) UF/
$17.35 0.285 0.25 0.88 B
Terayon (N.D. Cal.) (Wilken, J.) I $2.73 0.333 0.25 0.75
Tongxin (C.D. Cal.) (Kronstadt, J.) I $3 0.25 0.2383 0.95
TVIA (N.D. Cal.) (Whyte, J.) I $2.85 0.272 0.228 0.84
UnitedHealth (D. Minn.) (Rosenbaum, J.) PPF $925.5 0.1192 0.07 Yes 0.59 AD
US Unwired (E.D. La.) (Africk, J.) I $9.7 0.3 0.173 0.58
UTStarcom (N.D. Cal.) (Illston, J.) I $9.5 0.3333 0.25 0.75 BCF
Wachovia (S.D.N.Y.) (Sullivan, J.) PPF $627 0.175 0.12 Yes 0.69 ABDEF
Warner Chilcott (S.D.N.Y.) (Pauley, J.) UF/PI $16.5 0.275 0.18 0.65 C
Wash. Mutual (W.D. Wash.) (Pechman, J.) PPF $208.5 0.225 0.21 0.93 F
Wyeth (S.D.N.Y.) (Sullivan, J.) UF $67.5 0.245 0.2 Yes 0.82 AEF
Xerox (D. Conn.) (Thompson, J.) PPF/I $750 0.2 0.16 Yes 0.80 CEF