Another perspective on Milberg Weiss

Michael Perino has written a paper, The Milberg Weiss Prosecution: No Harm, No Foul?, which has gathered significant attention, including a presentation at AEI, and substantial notice in Saturday's NYT by Joe Nocera . The paper addresses whether the actions of Milberg Weiss and its now disgraced principals, including Melvyn Weiss and Bill Lerach, injured the very shareholders they were supposedly championing by paying kickbacks to named plaintiffs.

One might ask at the outset why this matters. There is now no legal doubt that the Milberg defendants broke the law by paying the kickbacks. Whether or not their kickbacks hurt shareholders, the defendants’ behavior was disgraceful and ought to be punished severely.

And yet the question does matter. To begin with, it has salience in the court of public opinion. Those who sympathize with Milberg’s efforts over the years can still ask if the means justified the ends. This becomes harder to do if Milberg was ripping off the people they were supposed to be defending.

More importantly, there is a larger policy debate about the role of class action lawyers. If kickbacks undermine the positive effects of class actions, we ought to be very concerned about them. On the other hand, if kickbacks are not harmful or even improve class actions, then perhaps the laws they violate should be changed.

Either way Lerach and Weiss go to jail, but the long term public policy outcome may differ. For example, there is a debate underway about whether plaintiffs’ contingent-fee lawyers should get favorable tax treatment for litigation expenses. We might like to know whether these lawyers are doing society any good before deciding how they should be taxed. And there are even broader issues at stake about the role of empirical evidence and criminal prosecution of business.

So Perino’s article is important. What does it show? In a nutshell, Perino’s findings respond to an important contention of the Milberg defendants, that the kickbacks didn’t hurt the shareholders. Defendants claim the kickbacks came out of their pockets and motivated the name plaintiffs to monitor counsel’s efforts and their fee requests. Class members therefore actually may have gained from the kickbacks in the form of higher settlements and lower fees. But Perino purports to show that the fees were actually higher and varied less in kickback cases for which the Milberg defendants were indicted, while the settlements were no higher. This suggests that the fees didn’t motivate higher settlements and lower fees, and therefore that class members were harmed as compared to a world in which there were no kickbacks. In effect, rather than reducing defendants’ benefits from the litigation, they were a sort of quid pro quo through which the defendants bought acquiescent lead plaintiffs.

But there are two main problems with the data and conclusions. The first major point, and perhaps the more important from a broader perspective, is that even if the evidence shows what it purports to show, this does not establish that the class members were harmed. That is because the paper looks at only one side of the theoretical story -- whether the fees motivated better monitoring by named plaintiffs. In fact, this story is not particularly interesting because we should already have assumed that even non-bribed plaintiffs would have little incentive or ability to affect class action outcomes. And, indeed, there’s evidence to support our skepticism: Theodore Eisenberg & Geoffrey P. Miller, Attorney Fees in Class Action Settlements: An Empirical Study, 1 J. Empirical Legal Stud. 27, 28 (2004) found that the size of the award determines fees, that objectors to fee awards have no discernable effect on the size of the award, and that even the enactment of the PSLRA, which gave a significant role to institutional plaintiffs, did not decrease securities class action fees.

In focusing on how the kickbacks affected fees and settlements as compared to non-kickback cases, Perino misses another important question: whether class action fees and payments to plaintiffs helped class members by motivating the filing of high-quality class action complaints.

The apparent reason why Perino does not focus on this theory is simply that the defendants did not rely on it. We can readily understand why the defendants made nothing of this argument: it would not have helped their litigation position to concede that the kickbacks reduced net recoveries and argue the class members were still better off because of the kickbacks’ effects in motivating lawsuits. But regardless of defendants’ litigation position, society should care a lot about this incentive effect.

To understand why the kickbacks might have motivated higher quality class actions, it is important to understand the incentives under which class action lawyers operate. They are paid on a contingency basis – they earn nothing unless they win. Moreover, at the time of filing a class action complaint the lawyer does not know if the client will be the lead plaintiff or whether the lawyer will end up representing the lead plaintiff because the court decides these questions. Under the rules prevailing prior to the Private Securities Litigation Reform Act (PSLRA), the lead plaintiff was often the first to file, provided that plaintiff met all the standing rules. All of this means that lawyers may invest a lot of effort in a complaint for which they may get no reward.

Weak incentives to invest in writing the complaint are especially problematic because of the importance of the complaint. Given the dynamics of litigation and settlement, the key moment in the case is the resolution of a motion to dismiss or for summary judgment. These cases almost never go to trial – trial is just too expensive for everybody for reasons that go beyond this short discussion. When the plaintiff survives the preliminary motion stage, the defendants have sharply greater incentive to settle. Thus, securities class action complaints are the foundation of the case law of securities regulation. Lower quality complaints may reduce the deterrent value of securities litigation and lead to lower-quality securities law.

It follows that if class actions are to play a positive role, it is important that plaintiffs’ lawyers be motivated to put a lot of effort into complaints that clearly identify what defendants did wrong. See Kobayashi and Ribstein, Class Action Lawyers as Lawmakers, 46 Ariz. L. Rev. 733 (2004).

Putting all this together, lawyers who want their investments in their complaints to pay off must be able to quickly find clients who had standing to sue (e.g., were shareholders at the time of the alleged fraud), were willing to sue, and would not dump the lawyer later in the litigation. The kickbacks helped accomplish these objectives in three ways. First, they encouraged people to be plaintiffs. This in itself is no small matter, since being a plaintiff means, among other things, having to show up for depositions and putting up with other hassles during the lawsuit. At the same time, the plaintiff gets little reward from the lawsuit because he has to share recovery with the other class members. Plaintiffs, in other words, need some incentive to come forward. See Theodore Eisenberg & Geoffrey P. Miller, Incentive Awards to Class Action Plaintiffs: An Empirical Study, 53 UCLA L. Rev. 1303, 1307 (2006).

Second, the kickbacks created a stable of plaintiffs that the lawyers could draw from quickly who had invested in the securities that ultimately became the subject of litigation.

Third, the payments encourage the plaintiffs to be loyal to particular lawyers. Plaintiffs otherwise might have an incentive, once a lawyer has gotten them to the ball in style, to pursue the case with a different lawyer who may have better litigation skills but who didn’t have either the skill or the incentive to invest in the complaint.

It follows that, even if the kickbacks reduced net recoveries in given cases – that is, assuming the same suit would have been filed without the kickback – they still may have made class members better off because they motivated better lawsuits. Indeed, Perino himself notes in his article that “Milberg Weiss is clearly good at what it does.” Moreover, eleven years ago, even after the PSLRA changed the system of appointing lead plaintiffs to one in which lead plaintiffs were institutional investors, with presumably better incentives regarding appointment and monitoring class counsel, Perino told the Senate Banking Committee:

Since passage of the Act, Milberg Weiss appears to have become an even more dominant presence. Milberg Weiss' appearance ratio in 1996 stood at about 59% nationwide and 83% in California.Milberg Weiss' increased significance can be explained by the fact that: (1) it is likely the best capitalized plaintiffs' firm and therefore best able to finance the delays associated with slower procedures under the Reform Act; (2) it has the most diversified portfolio of plaintiffs' claims and is therefore better able to absorb the risk associated with litigation under the new regime; and (3) it is best situated to internalize the externalities associated with the need to invest to create new precedent interpreting the Act's novel provisions.

We should emphasize that even if the kickbacks did help class members by improving the quality of securities litigation, this does not make them right. They were still against the law, and flagrant violations of the lawyers’ ethical obligations. But we already knew that without Perino’s article. The whole point of Perino’s paper is that even if the kickbacks were illegal, it matters whether the kickbacks hurt shareholders. Our analysis bears on that point.

Bruce Kobayashi and I have discussed in a pair of articles the importance of these hypotheses regarding incentives in class actions that were not discussed in Perino’s paper. In Class Action Lawyers as Lawmakers, we proposed mechanisms for protecting lawyers’ investments in complaint-drafting post-PSLRA so as to promote incentives to produce high quality complaints. In The Hypocrisy of the Milberg Indictment: The Need for a Coherent Framework on Paying for Cooperation in Litigation, we argued that plaintiffs’ incentives to come forward are important, and that indeed these incentives have been recognized in the criminal context via prosecutors’ ability to offer plea bargains to cooperating witnesses – including in the Milberg cases themselves. We contended that an appropriate system would recognize the incentive issue and allow it to play out under court supervision in the civil context as now happens in criminal cases. Instead, illegalizing witness and plaintiff payments in civil cases drives these payments underground, eliminating the possibility of judicial supervision.

To repeat, the empirical implication of our theory is that even if class members bore the cost of the kickbacks, and even if the kickbacks did not improve settlements or reduce fees compared to Milberg’s non-kickback cases, the kickbacks may have helped the class members by motivating Milberg to file better cases.

Indeed, Perino himself shows that “there is a positive and significant correlation between the dependent variables and the presence of Milberg Weiss as a lead counsel in the case” (Perino at 36). This is consistent with our hypothesis that the kickbacks improved the quality of Milberg’s cases by giving them an incentive to invest more in the cases. Nevertheless, Perino does not take that correlation into account in reaching his bottom-line conclusion that the kickbacks harmed class members because, he says, “neither this study nor any of the previous studies draw a causal link between Milberg Weiss’ participation and higher settlement amounts” (id). Rather, according to Perino, Milberg may just have been able to get a “leg up. . . in the best cases” (37). That’s true. But the difference between a “good” and “bad” case has a lot to do with the lawyer’s skill and investment of effort in the case. This seems to be borne out by Milberg’s continued success even after the PSLRA reduced the potential role of kickbacks. In any event, exactly the same thing about correlation and causation can be said about all of the paper’s empirical conclusions, not just the correlation between the presence of Milberg and higher settlements.

The bottom line is that, while Perino’s data (assuming it shows anything) responds to Milberg’s argument about kickbacks, and therefore is potentially useful on that score, it does not show what it purports to show, and what has earned it significant publicity – that the kickbacks hurt the plaintiffs.

The second major point about Perino’s paper is that, on a close look, it is far from clear that the data shows even what it purports to show – that the kickbacks hurt plaintiffs’ recoveries in given cases. This calls for some extended statistical analysis, which we give here.

In general, Perino’s data bears on important issues of public policy. The kickbacks likely affected the incentives of the Milberg defendants. If the effect benefited class members, then perhaps we should rethink whether the kickbacks should be illegal.

It is also important to stress broader lessons of this analysis. First, empirical work is becoming very important in the legal literature and in litigation. There is no doubt that this work can be very useful. On the other hand, there is a danger of over-persuasion. Thus, as continuing controversies concerning “junk science” have taught us, it is extremely important to be cognizant of the limitations of data, and careful about what it does and does not show. This warning applies not only to statistical methods, but also to the need to have sound theories motivating the development of the hypotheses being tested. We understand that we (and Perino for that matter) may have left behind some numerically challenged readers, who may be tempted to skip over the technically daunting analysis to the clearly framed conclusion. But that is part of our point: the analysis is at least as important as the conclusion.

Second, it is particularly important to get the data and theories right with respect to Milberg. Lerach, in particular, is for some people the personification of a particular brand of evil – the Jeff Skilling or Ken Lay of the trial bar. Lerach’s flamboyance can divert attention from the real issues. In particular, is the problem with class actions in general or with the Milberg defendants in particular? Could it be that it was the class actions themselves that made Lerach & Co. evil, with the kickbacks only the mechanism used for putting them out of business? Was it the very fact that the Milberg defendants were so good at class actions that made them bad?

Third, given the notoriety of Milberg’s business and Lerach in particular, there is a particular danger of punishing business people because they are in an unpopular (though legal) business, and not for what they’ve actually done wrong. This is no less true of Lerach and Weiss than it is of Lay and Skilling. Indeed, Nocera, in the article cited at the beginning of this piece, is candid about this when he concludes that “if putting [Lerach] in prison for his little kickback scheme — rather than his far more venal form of economic extortion [of class actions] — is a little like putting Al Capone in jail for tax evasion, well, so be it.”

Given this potential for bias we need to be especially careful about our data and theories. Notably, Nocera cited Perino’s data as an important basis for his conclusion. For all these reasons, it is very important to be sure that Perino’s data actually shows what it purports to show. As we have demonstrated, that is not clear.

Comments

[Ralph Adamo]

The fact that the New York Times gave publicity to the Perino "study" shows that they haven't learned a thing from the Jayson Blair scandal. In that infamouse case, Blair, a writer for the NYT simply made up stuff in his articles. Now Nocera and the NYT are hanging their hats on more made-up stuff.

If there was any merit to Perino's article, wouldn't the Wall Street Journal have at least looked at it and given it to a real statistician for a critique?

Perhaps the WSJ did, and that's why the WSJ isn't saying anything about Perino's article.

In any case, Perino uses flawed statisitics and tries to artificially boost everything in his favor to try to claim that his results are statistically significant. In spite of his flawed and biased approach, he STILL ended up with results that even he admits were statistically insignificant. I guess the NYT never read that part.

So how did Perino get around this little problem? He relied on a technique known as bootstrapping. Bootstrapping might normally be considered when estimating and testing population parameters in situations where the sampling distribution is unknown. In Perino’s study, the sampling distribution is not unknown, and, in fact, Perino worked with an extensive database in which all the case details were known. Neither the sampling distribution was unknown, nor were any assumptions violated. So bootstrapping shouldn't even have been used in the first place.

So what is bootstrapping?Bootstrapping is a method that involves repeatedly taking random samples of size n (with replacement) from the original sample and then calculating the value of the point estimate. To illustrate how bootstrapping would work, suppose we had a sample of 4 data points: 1, 3, 5, and 9, and we estimate the median from this sample as 4. Applying bootstrapping, we repeatedly take samples with replacement from the 4 data points and get the following, using 10 cycles:

1. Median (3,5,1,1) = 2
2. Median (3,9,9,1) = 6
3. Median (5,9,9,9) = 9
4. Median (5,1,9,5) = 5
5. Median (4,9,9,5) = 7
6. Median (1,3,9,5) = 4
7. Median (3,3,9,3) = 3
8. Median (9,5,3,3) = 4
9. Median (1,5,3,1) = 2
10. Median (9,3,3,9) = 6

Using bootstrapping, we have a pseudo-sample of 10 re-estimated medians (i.e., 2, 6, 9, 5, 7, 4, 3, 4, 2, and 6). The behavior of this pseudo-sample mimics the behavior of the original median of 4, as the median of the pseudo-sample is also 4. Then, the standard deviation of these 10 values is computed, 2.3, which is an estimate of the variability of the median.

This illustration uses 10 pseudo-samples drawn from the original sample of 4 values. In practice, bootstrapping normally entails a very large number of samples, such as 10,000 and is performed using a computer program.

I'm not saying that bootstrapping doesn't have its place in statistics. But I am saying that Perino hasn't established any possible foundation for using it. I would defy the NYT and Perino to find any decent statistician who would support his use of bootstrapping in his paper.

So, given these facts, the "study" is merely unsubstantiated propaganda against the Milberg firm.