Are Shareholder Lawsuits Useful or Frivolous?

March 18, 2004

Unedited transcript prepared from a tape recording

8:45 a.m.	Registration
9:00	Panelists:	Eric Helland, President’s Council of Economic Advisers and Claremont McKenna College Adam Pritchard, University of Michigan Law School
	Discussant:	Michael Perino, St. John’s University School of Law
	Moderator:	Jonathan Klick, AEI and Florida State University College of Law
11:00	Adjournment

Proceedings:
MR. KLICK:  I'm Jon Klick.  I'm the Associate Director of the Liability Project here that I run with Michael Greve at AEI and also Richard Epstein at the University of Chicago Law School.  For folks who aren't familiar with our project, basically what we started out doing and what we hope to do is basically set up more or less a tort reform project that differs from all the tort reform projects that have failed in the past, specifically by looking at empirical research on the litigation crisis or supposed litigation crisis and political economy issues involved in the foundations of the political economy--in the torts crisis.

And so today's event fits nicely in that general umbrella in that we are looking rigorous empirical research that tries to determine whether or not reforms have worked in a particular area, specifically securities litigation.  One of the probably primary failings of the tort reform movement is that immediately and without any sort of investigation, tort reform people call for reform.  And they're not terribly specific on what reform is, but they know that there must be some kind of reform, without really examining what reforms have done in the past in individual areas of the torts crisis.

And so what we're hoping to do today is shed a little bit of light on the specific issue of whether the Private Securities Litigation Reform Act of 1995 has done anything to discourage frivolous lawsuits in the securities area, and then also from a more general perspective, try to determine whether securities cases do tend to be frivolous or whether they have got merit.

And so we're going to start off with Adam Pritchard from the University of Michigan Law School who's got a new working paper entitled "Do the Merits Matter More?  Class Actions under the Private Securities Litigation Reform Act of 1995," which he co-authored with Marilyn Johnson an Karen Nelson.

Professor Pritchard?

[Pause.]

MR. PRITCHARD:  You need me on record as fumbling around to try and find my presentation.  Okay.  I'm glad we've got that for posterity.

Jon described this as new.  That's true and false.  This paper has been kicking around for a little while, but I have a new set of regressions; we've collected some additional data.  So it's certainly a new twist, and I think the new data actually tells a more interesting story about how class actions work and a little bit about the effect of the Private Securities Litigation Reform Act.  And as Jon said, this is co-authored work.  Marilyn and Karen are the accounting professors part of this team; I'm the law professor, which is to say they do the heavy lifting on the regressions, and I try to translate it into English.

All right.  The question about whether the merits mattered at all in securities class actions was a lively one in the early 1990s, and I have to give credit to Janet Cooper Alexander as being one of the first people to prompt this debate, because I am nakedly stealing from the title of her paper.  So she concluded that, no, the merits didn't matter because she found a lack of variation in settlements.

Shockingly, corporate issuers say the merits are entirely irrelevant.  If the stock price drops, they get sued.  And the accountants and underwriters, quite predictably, also say no, they're just getting sued because they have the ability to pay.

If you actually look at the data, though, it's not nearly as clear as the accountants and issuers would have you think.  The settlements tend to be higher in cases of flagrant fraud.  And it's not true that simply having a large stock price stop leads to a lawsuit.  Some do and some don't.  The claim was that all 10-percent stock price drops would get you sued, and that's just clearly not the case.

And one other point which has got to be right, the plaintiffs' lawyers are working on a contingency fee.  So as between strong cases and weak cases, you are more likely to get paid if you bring a strong case.  So basic economics would tell you that the plaintiffs' lawyers ought to be looking for the cases where they are most likely to get a settlement.  And you would think the ones that had fraud would be those cases.

All right.  So that's kind of the framework of the debate in the early 1990s.  Congress, in its infinite wisdom, decided that the plaintiffs' bar was hurting companies and investors, and we got the PSLRA out of that.  And I think the three most important provisions in the PSLRA--there are some other provisions that matter, but for purposes of discouraging meritless or thin suits, the three pieces I've put up here are the most important:  the pleading standards which require some particularity as to what the misstatement is, and the biggest obstacle in the pleading standard, the requirement that you allege facts which would give rise to a strong inference that the defendant acted with the required state of mind.  So you have to plead facts in your complaint before you have had discovery that would suggest that the defendants not only made a misstatement but knew or were reckless in making the misstatement.

The other big innovation, the safe harbor for forward-looking statements, it raises the level of scienter for forward-looking statements to actual knowledge.  Even if you have made a misstatement with actual knowledge, if you have surrounded it with meaningful cautionary language as to why your projections for the future may not come to pass, you're going to be immunized from liability.  And then the other thing that makes lawsuits a lot cheaper for issuers to defend is a discovery stay, so plaintiffs can't have--they can't depose the executives.  They can't get the documents from the company that they might use to use the strong inference of scienter required by the tougher pleading standard.  So the combination of the first two with this discovery stay makes it a lot easier to win on a motion to dismiss if you are an issuer--at least in theory.

All right.  So what happened after Congress passed the PSLRA?  Well, the first thing that happened was that Bill Clinton vetoed the PSLRA, which led to the first congressional override of a Clinton veto.  The market thought that was a good thing, according to a study by my co-authors, Johnson and Nelson.  And then going forward, the number of lawsuits has not declined in any measurable way.  It's probably up slightly now, I would say.  There was a blip with the laddering lawsuits in 2000 and 2001, but now it's returned to a steadier pace, but still at least as many lawsuits as before the act and maybe just a little bit more.  And then among those lawsuits people have tried to keep track of what's being alleged, and it's pretty clear that there are more accounting and insider trading allegations, presumably in an effort to try and satisfy that tougher pleading standard.  And there is some evidence that at least the insider trading allegations may be easier to bring then to win, and suits with the insider trading allegations tend to be dismissed after the PSLRA.

So we have a variety of hypotheses that we want to test relating to the effect of the act.  And I have my nice slide here with all six of them.

The first is that measures, objective measures--we like to call them objective because we're doing the coding, so measures of accounting, insider trading, and governance factors are going to be more important in explaining lawsuit filings after the PSLRA.  The governance factors, the notion there is that companies with stronger monitoring environments ought to be less likely to be sued.  So if we see a company with a weak monitoring environment, that ought to correlate more highly with lawsuits after the act.

The safe harbor ought to be giving some protection to earnings projections by companies.  If they are properly counseled, they ought to be able to effectively immunize those forecasts from lawsuits.  So that ought to be a less important predictor.

And going down from the level of the complaint to what is alleged in the complaint, do the allegations--is there a correlation between the allegations made in the complaint and our objective measures of aggressive accounting and insider trading?  You'd like to think there would be a closer connection after the act.

All right.  The complaint of the issuers was that they were getting sued because of stock price drops.  They failed to meet earnings, the stock price dropped, they automatically get sued.  Then if the act is working as these issuers had hoped, an earnings shortfall will be less important in explaining an allegation of a false forward-looking statement.  The notion here is that earnings shortfall happens and then you go back in time as the plaintiff's lawyer and try and find a forward-looking statement where you said that earnings were going to be great.

And going to the last stage of the litigation process, what actually happens to these lawsuits, we expect our same factors to be more important in explaining the outcomes, and we predict that forecasts in earnings shortfalls should be less important.  So we're trying to capture both the effect of the pleading standard, requiring more substance in the complaints, as well as the safe harbor, which is intended to give some protection to forward-looking statements.

All right.  To test our hypotheses, we collect a sample of companies in the computer hardware and software industry.  The reason for choosing this sample rather than some other samples is that firms in this sector got sued before the act and they get sued after the act, and in order to do our comparison of what's happening before and after, we need a fairly stable industry in terms of exposure to litigation.  And other industries tend to have varied over time a little bit on that.

In order to do our analysis, we have to have proxy statements, CRSP data and Compustat data, so that necessarily excludes firms doing their IPOs.  This is a study of secondary market lawsuits, so firms that were sued on the basis of an alleged misrepresentation that affected the trading in their securities in the after-market.  And once we satisfy all those data requirements, we end up with 50 suits from before the act and 64 afterward.  And then we construct a matched sample from the same industry code, and then we have done our very best to try and match on the stock price drop that--the minimum one-day return for the sued firms.  We tried to find the firm that didn't get sued that had the closest drop minimum one-day return for the non-sued company, which I'm not going to put up the descriptive variables.  That was only partially successful.  It was hard to find companies that had price drops that were the same.

All right.  So we have our damages, accounting, insider trading, governance variables.  We didn't think these up on our own.  They have been found in prior work to correlate with litigation or SEC enforcement actions.  The premise of the work in this area is that if the SEC is bringing an action for fraud, they have scarce resources, plenty of companies that they could sue, so they are going to focus their energies on the clearest cases of fraud, so the bringing of an enforcement action is a decent proxy for fraud.  And for our forecasting variables, we are trying to capture the effect of the safe harbor.

All right.  So our first category is the damages variables.  Bigger companies tend to get sued.  If you've got the big stock price drop, you are likely to be sued as well.  We include this variable in the regressions despite having matched our sample on that basis because we were only partially successful in doing the minimum return matching.  And share turnover, if there are more shares being traded, it is more like--the damages calculation is going to be greater.  So we assume that plaintiffs' lawyers, part of their calculation is what sort of settlement they can expect to get.  The greater the damages, the greater the settlement.

For our accounting variables, a firm that has to restate its earnings anytime during the class period, and that's a very conspicuous signal for plaintiffs' lawyers, that there has been a misstatement.  There's still the obstacle of proving scienter and loss causation and damages.  But the restatement is pretty close to an admission that you have made the misstatement; that would be a requirement for the lawsuit.  And then a couple variables that you might think of as governance variables or you may think of them as relating to accounting, firms vary in how frequently their audit committee meets.  The assumption here is that firms that meet more often, that audit committees that meet more often are doing a more vigorous job of monitoring.  And more independent audit committees we would think would be more likely to take a close look at the company's accounting.

All right.  Insider trading variables, we have two.  One is just a raw measure of what are the net sales during the class period by the insiders.  They are ranked by decile so we have got a 1 through 10 categorization, and every firm is put into one of those ranks.  And then an abnormal insider trading measure, which is closer to the standard stated by the courts, which is that insider trading is not evidence of scienter standing alone, it has to be compared with how much trading did you do in prior periods.  So it's only a spike in the sale of stock by the insiders that suggest they know something is wrong at the company.

Then three forecasting variables.  The earnings shortfall is if the company's made a disclosure that they're not going to meet their earnings expectations, earnings are going to be lower, either on the minimum return date or the class end date.  And then a positive forecast is if you're making any sort of positive prediction about earnings during the class period.  And a negative forecast, maybe firms try to manage expectations as a way of avoiding litigation, so they're trying to talk down the market because they are concerned that they won't meet their earnings projections, and our prediction is that if they are successful in talking down the market, they're more likely to avoid a lawsuit.

Governance variables.  If we have directors who have been around for a while on the board, we think they are more familiar with the company's operations, better able to process the information.  Directors who are serving on a large number of boards may have difficulty monitoring this board that they are sitting on.  And, again, as with the audit committee, more outsiders on the board, we think more likely to have greater--likely to have better monitoring in that circumstance.

All right.  So now we put those independent variables into a regression with our dependent variable being the filing.  So the match firms didn't get sued, the sued firms did get sued, so that's our dependent variable.  And when we look at the independent variables for the pre-PSLRA period, big firms get sued.  Firms that have made a positive forecast and firms that didn't meet their earnings expectations.  So none of the variables that you would think would correlate with the likelihood of fraud are significant in the pre-PSLRA period.

Post-PSLRA, market caps still important.  If you're a big firm, you're more likely to get sued.  The stock price drop is an important predictor of suits.  So those damages variables have not become unimportant after the PSLRA.  But some of the variables that suggest problems, either with the company's accounting, the restatement, or the governance, insider trading is significant, busy and independent are significant, all in the predicted direction.  So that's encouraging.

Earnings shortfalls still an important basis for  bringing suit, just like the damages variables are.  So this is not unambiguously an improvement.  And looking at the effect of the safe harbor, note that the positive forecast is significant before but not afterward.  That suggests that the safe harbor is giving some protection to forward-looking statements by companies, as Congress intended.

All right.  Now we move on to the second set of regressions that involve the allegations made in the complaints, and comparing the--these are the descriptives on this.

Accounting, somewhere around, let's see, 32 percent of the cases before the PSLRA had accounting allegations, and for our sample that goes to 48 percent afterward, which--that just doesn't even look right on the graph.  I don't know.  I typed it in a 48 and it's showing up at 60.  It's really 48.

The insider trading allegations, 38 percent before, 38 percent of the suits had insider trading allegations before, and 61 percent afterward.  Once again, a gross inflation by PowerPoint.  It's not me.  It's PowerPoint.  It's a big spike.  It's not that big.

And then forecasting allegations, allegations of false forward-looking statements, a big factor before, 86 percent had allegations of that sort before the PSLRA; and then a number that surprised me a little bit is 74 percent here were understating the amount in my bar graph, 74 percent had forecasting allegations after the act, which is a little bit higher than I expected.  I thought it would be lower because of the safe harbor.

MR.           :  [inaudible].

MR. PRITCHARD:  For the forward-looking statements, it was 86 percent.  So here I think my bar is kind of close.  It seems to do better in the pre- period than the post- period for some reason.

All right.  So what do we find when we do the regressions.  We take out the damages variables.  Presumably, the damages variables are relevant to the decision to bring suit, but once you have made that decision, they should not have an effect on what you include in the complaint.  And our dependent variables are just whether there was an accounting allegation, an insider trading allegation, or a false forward-looking statement allegation.  So that's the dependent variable, and obviously our sample here is just the firms that were sued.  We're not looking at the match firms in this point.

And for accounting allegations, we don't have any significant variables before the PSLRA, so restatements do not correlate with accounting allegations before the PSLRA.  Afterward, restatements do.  This is kind of interesting to me in terms of lawyer strategy, I guess.  We include the other allegations as independent variables in these models, and forecasting allegations are inversely correlated with accounting allegations.  So there's a substitution.  If you have a forecasting allegation, you're less likely to have an accounting allegation.  We're looking for some basis to file suit, and we find it, and we're done, apparently.

Audit independence is significant but contrary to the predicted direction.  So this is not an unambiguously good story here.

The insider trading allegations, before the Private Securities Litigation Reform Act, they correlate with the busyness of the directors.  There's no correlation with either our raw measure of insider trading or the abnormal measure.  And after, it correlates with the raw measure of insider trading allegations, so that's better, I suppose, than the pre-PSLRA period.  But they don't correlate with abnormal insider trading, which is the standard that the courts say is the relevant one.  And as with accounting allegations, the insider trading allegations negatively correlate with forecasting allegations.

All right.  Then the last of our allegations variables, earnings shortfalls are correlated with having a forecasting allegation.  Note here that the positive forecasts, which you would have thought would be the basis for the statement of an allegation that you've made false forward-looking statements, are not correlated.  And afterward, earnings shortfalls still correlated there and the substitution effect that we saw with other variables, but positive forecasts still not correlating with the allegation of a false forward-looking statement.

So the last set of regressions is whether--I was trying to capture whether the merits drive the outcome of the suits, and we've excluded the damages variables.  Those are obviously relevant to settlement amounts, but our question is more do the suits where there is some reason to think that fraud has occurred result in a settlement, and the ones that fraud is unlikely to have occurred do not.  So we're looking at a more binary dependent variable which is attempting to capture that.  We don't distinguish simply between settlement and dismissal, which would be the obvious way of sorting out the cases.  There are settlements out there that are for such trivial amounts, we think they roughly correlate with defense costs, and we have a very conservative estimate of defense costs here as being less than $2 million.

So, on the one side, we have lawsuits that result in a settlement of more than $2 million, and on the other side, we have suits that were dismissed or resulted in this nuisance value settlement, as we describe it.

All right.  Pre-PSLRA, abnormal insider trading does correlate with settlement.  The negative forecast is inversely correlated, so that's actually a decent--you can't manage down the earnings expectations in the pre-PSLRA period, is what that suggests.  And average tenure is once again significant as a governance variable.

Post-PSLRA, restatements correlate with the likelihood of settlement.  Average tenure is still significant.  It looks like having your directors around for a longer period of time means that you are having better monitoring from those directors, according to this set of outcomes.

So taking you back to my hypotheses, it does look like the lawsuit filings are better explained by some of our variables that are attempting to proxy for likelihood of fraud.  It does seem that the safe harbor is giving some protection in terms of discouraging lawsuit filings.  And there is some evidence that accounting and insider trading allegations are more closely correlated to accounting and insider trading problems, which you would like to see that.

The last three, though, we don't find any evidence that earnings shortfalls have become less important, either in the filing of lawsuits or alleging false forward-looking statements.  And the outcome regressions I don't think provide strong evidence that the merits matter more to the outcomes.  And we don't find evidence that the forward-looking statements in the earnings shortfalls are less important.

So the conclusion, the merits matter somewhat more post-PSLRA, at least in the filing of lawsuits and the allegations that are made in those lawsuits.  And Congress I think can conclude that the safe harbor is providing some protection for forward-looking statements, which was their goal.

On the other hand, we don't really see a clear shift when we look at the outcomes of the lawsuits.  And going to what our paper can show or not show, we can't tell you how many good lawsuits didn't get filed because of the barriers erected by the PSLRA.  So that is a cost of the reform act that our methodology doesn't tell you anything about that, but it's obviously relevant to whether the act worked.

The fact that plaintiffs' lawyers are still focusing on raw measures of insider trading, despite the frequent instruction from the courts that they should be looking at abnormal insider trading is worrisome, I suppose.  And obviously there may be factors that we can't capture in our models.

MR. KLICK:  Okay.  Thank you, Professor Pritchard.

Our discussant today, Professor Michael Perino, has asked if he could give all of his comments after both papers since there's significant overlap.  I, on the other hand, am going to focus on the econometric issues and the methodological issues of each of the two papers, and they are sort of significantly different between the two papers.  So I'll quickly make my comments in that regard on Professor Pritchard's papers and give him a second or two to respond before we move on to Professor Helland's paper.

First of all, I'm not really sure what you're achieving through your matching methodology.  In principle, we like to use matching to control for unobservable variation, and you happen to be matching on something that we do observe, which is price drops.

Now, presumably you're making the assumption that the unobservables are roughly the same for the sued firms and for the firms that aren't sued conditional on this similar price drop.  I'm wondering if that's a plausible assumption, and it may be descriptive in that in the finance journals and the econometric journals matching methodologies were sort of all the rage in the early 1990s until, you know, folks like Heckman(ph) and some other econometricians came along and said, really, the unobservables are really quite important here and we really don't know much about heterogeneity in unobservability.  So I'm wondering if you really are achieving anything with that matching.

Now, that said, if you are going to go with a matching methodology, I wonder if it might not be more profitable to match on the basis of firms that the PLSRA was meant to affect versus firms that it wasn't meant to affect.  Okay, so the idea is presumably there was some crisis in securities litigation that prompted the PSLRA, but as you noted, there are some industries or there are some kinds of firms that basically there was no crisis.  Right?  And so why not match on the basis of those things, say match with firms in different industries or match with smaller firms that clearly weren't meant to be affected by the PSLRA, and that might give you a better hold on controlling the unobservability that might or might not be sort of leaking into your PSLRA variables.

Secondly, I wonder if there's something a bit misleading about the way you present your results, the idea being you look at this methodology or this regression reporting where you present results for pre-PSLRA and post-PSLRA, and you say, well, if something wasn't significant beforehand but is significant after-hand, that is maybe some evidence that there has been an effect.

But it's interesting if you look through your tables and you actually check to see if the differences in the coefficients are significant pre- and post-, in quite a few of them it's not.  For example, one of your sort of important results is that accounting allegations matter more post-PSLRA than pre-PSLRA because post- there's a key statistic of negative 0.197 versus pre- 1.24.  But if you look at the significance of the difference, the difference is insignificant.  So I wonder if the significance in your second set of results is really being driven by the fact that you've got larger sample sizes, 64 cases as opposed to 50 cases.  And I'm wondering if a different way of running these regressions might not yield sort of more germane results.

PROFESSOR PRITCHARD:  As far as the matching, the matching is not so much intended to control for the unobservables but to make data collection a tractable proposition, because I think if we're not matching, we should just be doing all of the firms in this sector as the sample and then--but life is short, and it would take a long time to collect the proxy data and the insider trading data, which all has to basically be done either by hand or through a very laborious computer process to do that.

So the matching procedure provides us with the advantage of having a relevant sample.  It does have some effect on the size of the coefficients because we're not matching the population.  But it shouldn't affect the significance of the coefficients.  So it's basic--it's not intended to capture the unobservables.  We're just trying to make the study possible to do.

In terms of trying to compare pre- and post- more directly, we are working on a set of pooled regressions with an interaction variable for before--for after the PSLRA is enacted, and it may be possible that we'll--to do those, but we're still working out the kinks with them.  We just got the forecasting data put into the regressions.  So that's something we're planning to explore.

MR. KLICK:  Okay.  Thank you, Professor Pritchard.

Our second paper is going to be presented by Eric Helland, who is currently on leave from the Economics Department at Claremont McKenna College.  He's on leave at the Council of Economic Advisers as a senior economist.  His paper is entitled "A Secondary Market Test of the Merits of Class Action Securities Litigation:  Evidence from the Reputation of Corporate Officers."

Eric?

PROFESSOR HELLAND:  Well, let me start by saying thank you for coming.  I apologize for not wearing the jacket, but after the Red Line exploded, I walked in from American University and figured it could stay in the office and I would just come straight here.

The second is, as Jon mentioned, I am with the Council of Economic Advisers, so the usual disclaimer applies.  This is not the opinion of the White House or the Council, but just me as a professor on leave.

Let me start, then, with kind of a general description.  As you pointed out, there is a lot of overlap.

In recent years, corporations have come under a great deal of scrutiny for how they report their finances.  My favorite estimate comes from the Brookings Institution, which is that basically $37 to $42 billion in lost GDP has resulted sort of from uncertainty about corporate governance.  If you don't like sort of macro forecasts, you could also go with the GAO's estimate that about 10 percent of the companies between '97 and 2002 restated their earnings.

As everyone knows who is here, this has raised considerable concerns about the U.S. system of enforcing financial regulations, and a large chunk of that enforcement effort actually depends on the actions of sort of private attorneys general.  And that actually requires that these cases be meritorious.

So this is a graph that actually comes from a consultancy, and the reason it looks so smooth prior to '86 is that I interpolated--they only gave every two years, so don't draw anything out of the smoothness before or after.

What I wanted to sort of raise is there's an important series of cases--or case, I guess it is, that occurs in the early '80s that's kind of alleged to have increased the frequency of sort of class actions, the idea basically being this expanded the role of private attorneys general.  And then as we've talked about a little bit, PSLRA in '95 sort of restricts that.

So the lesson I want to sort of draw from this, this is the trend in directors' insurance premiums with '74 kind of as a base.  And what it says is that these private attorneys general are driving this market for directors' insurance.  That may not be a huge surprise, but--so the question then basically is:  Are these cases meritorious?  And by meritorious, what I mean is that they serve to deter future fraud.  And what that essentially means is if you're filing cases that don't actually identify any fraud, then it's going to be hard for them to deter future fraud.  That's the basic idea.

Now, the problem is that private parties have an incentive to bring cases that the government wouldn't file even if the SEC had an unlimited budget, and the reason is that a defendant will often agree to settle a case in which he would prevail at trial to avoid the considerable cost of litigation in particular discovery.

And, actually, the literature has been summarized, so I'll sort of skip of this slide, just to mention that the conclusions range from sort of these cases all being frivolous, the Alexander paper, to a fair number being frivolous and sort of everywhere in between.  And there is actually a small literature looking at the link between corporate governance and private securities litigation, and I list a number there.

Here's the potential difficulty.  The potential difficulty is that these things may be endogenous.  Imagine the following story:  I name--I'll pick on you for a moment.  I name you as a defendant in this suit, and we happen to see that your board has a large number of insiders on it.  Well, that may make you a more attractive company for a lawsuit because the attorney can mention this in court.  Or option two is that it may actually be that that symbolizes weaker governance.

So one of the problems with this is--and the critics have sort of alleged this.  This is nothing that I'm raising here--that a lot of our tests suffer from this problem that unobserved characteristics may cut both ways.  Maybe we're identifying things that make it easier to win a lawsuit, and maybe we're identifying things that actually indicate fraud.

I'm going to try something very different in this paper, which is I want to assess the merits of private securities cases using the reputational penalties paid by officers and directors who serve on a board accused of fraud.  The idea is this:  Given the amount of discretion and hidden information implicit in corporate management, outside directors have a reputation--I'm sorry, have an interest in maintaining a reputation for trustworthiness.

So, again, to continue to pick on you, the idea is if you're a director of a company--right?--you have a considerable interest in indicating to people in the directors market that you're an honest person.  The reason is that you are going--basically, the shareholders' proxy on the board and you're looking out for their interests.

So if private securities class actions are meritorious, we would expect directors to pay a reputational penalty when they're accused of fraud.  If it comes out--pick on your for a moment--that you have been implicated in a fraud, other companies may look at this and say, "Not a director for us."

I also want to point out--and it's something that I'll sort of get back to in a bit--not all boards are equally prestigious or lucrative, so I'm going to examine different types of directorships, and I'll come back to that statement in a minute.

Now, you might ask, Well, is it actually the case that shareholder interests drive the market for outside directors?  And I'll come back to this point directly.  But the argument might be maybe, in fact, being a fraudulent director, to continue to pick on you, makes you a really attractive candidate for certain firms.  You know, we really would prefer people that didn't do a lot of monitoring, and the fact that you didn't before means that you're really great for our firm.

Let me cite a little bit of evidence that that's not true, and then I'll also show you some evidence that I actually estimate.

There's kind of a conventional wisdom, if you will, starting sort of with  (?)  Jensen and Klein and Leffler that additional board appointments are sort of a signal of quality, that directors bond their service on a board with their reputation.  They don't usually own a lot of equity in the company.  And what that means is that directors who serve on more boards, up to a point, have greater reputational capital.  And then there's some evidence from some studies I cite here, as well as some others, that show basically doing things that are not in shareholders' interests mean that you are not as likely to be appointed to other outside boards.

My favorite one of these is a test actually that looked at voting on certain anti-takeover provisions.  Basically what they find is that if you approved an anti-takeover provision that the Pennsylvania Legislature allowed in a certain period, those people were less likely to serve on outside boards in the future.  So having nothing to do with the one they're currently on, but they simply are less attractive in the directors market because they've done something that shareholders perceive of as not in their interest.

I guess one other point to make before we go too much further is:  Why would a director care?  And the answer is these directorships are quite lucrative.  Yermak (ph) estimates that the present value of another outside directorship is almost a million dollars, and that's without considering equity options.  And Yermak's actually looking at the Fortune 500.  I'm actually looking at all firms.  So this estimate may be a little high--or actually is likely high for what I'm sort of looking at.  But basically the point is that serving on a board actually involves some significant financial compensation, and so you want to protect that reputation.

What's the sample?  One of the things you can do when you're visiting--and I was visiting at the University of Chicago last year--is that you can actually engage in insane data collection exercises.  So what I did basically was take Compact Disclosure, which maintains an electronic database of all proxy statements, and it's basically every company that files with the SEC, available electronically.  It contains the name and age of all directors, and it also identifies what company you serve on.

So I took all of these back as far as I could get them, which is the early '90s, downloaded this information, dragged it kicking and screaming in this data, and constructed a unique identifier for each company--which basically means looking at the list and figure out which one's which--and then for each individual director.

The individual directors, since there's actually something like 300,000 of director-years in this data set, you can't actually do that by hand.  So what I do is identify each director using last name, first initial, middle initial, and age, and then going back and seeing if certain directors show up, say, only once on a board--that's unlikely--or if John Q. Smith, for example, age 42, ends up having 5,000 outside directorships.  And then I go back to the proxy statements and sort of sort these out.

So after about six months, you get something that looks like a panel of all directors serving between 1994 and 2002 on all publicly traded companies in the U.S.

The measure of reputation is the change in one of several classifications of outside directorships.  Here's what I basically do:  If your sample starts, say--if the company starts in '93 and I observe the board then, I don't know when you were added, but let's suppose that at some point, say '96, Jon Klick is added to the board.  I have the board data before.  He appears on the board in '96.  He gets a new directorship.  Right?  Let's suppose he serves for a while on that board and exits in 2000.  That's an exit.

So what I'm basically going to do is look at the net directorships, new boards minus exits, every year with Jon as the unit of observation.  So for every year that he's in the sample, I'm going to basically watch and see what the turnover is in his portfolio.

The reason I'm doing it that way is, one, occasionally companies don't file with the SEC.  That is -- [tape ends].

[In progress] --and approximately, sometimes the filings simply end.  So, I don't want to count as an exit on the board if in 2000, the ABC Company simply vanishes.  I don't know if Jon has actually left that board, if the company has been delisted or if it simply stopped filing.  So, I simply treat those as not being exits.

So, all I'm looking at is existing boards, companies that are not going in the sample and what happens to Jon.

The datasets for this, for the securities cases, comes from the Security Class Action Alert, which is a litigation reporter.  Other crazy things you can do if you are visiting in Chicago is, take a stack of that high of litigation reporters off the floor.  It's a monthly publication.  I basically coded them all in from 1985 to the present.

It actually, at various points, lists a fairly comprehensive list of securities cases.  I also went through Lexis-Nexis, but I didn't find any that were not in the Securities Class Action Alert.

What I did find is that, there is a delay for the Securities Class Action Alert of getting cases that are pending, that have had no action.  So, I went to the Stanford Class Action Clearinghouse and basically added those in.  I have a fairly comprehensive list of all of the private securities litigations.

Generally, the SEC part of this, I went to the SEC's Accounting and Auditing Enforcement, which contains the data on cases filed by the SEC and essentially constructed a parallel sample of SEC cases.

I apologize for the ocular trauma.  Basically, I pay for my own slides at the council.  This is the first time I have presented this electronically, so I cram a lot on every piece of plastic.

Basically, what I want to do is I want to vary this, because directorships vary in their prestige, how lucrative they are.  So, what I am going to do is take that sample I had constructed of net directorships and then restrict net directorships to certain classifications.

So, for example, some directorships have a pension system.  For a subset of firms that I know, whether you have that, have information on whether you have a pension system or not, so I look at the net changes in those firms, board positions with pensions, directorships with stock options, top quartile[ph] option value, that is, how much the option at issue is worth, top quartile of directors fees.  That comes from ExecuComp [ph].

I also want to measure the change in prestige of board positions.  So, I look at top quartile in industry adjusted return, top quartile at sales, top quartile of employment and directorships in the Fortune 500.

One concern I mentioned is, suppose that a lawsuit actually makes you more attractive to poorly governed companies.  That is, perhaps it is the case that if you have a company where the CEO has served for a long time, the company has by a lot of our measures looked as if the governance measures aren't strong [tape skip] directors who don't know [unintelligible] fraud is more attractive.

So, what I do then is look at a couple of different measures of governance, namely, the top quartile of block ownership, the top quartile of insiders serving on the board, board size, CEO tenure and then also I have some information on directorships that I indemnify their director.  It basically says that, if you are a director on this and directors' and [unintelligible] insurance does not cover you, we will pick up the rest.  There are a bunch of different measures.

A couple of independent variables, what is corporate performance--I will talk about this for just a few seconds.  There are a lot of studies indicating that corporate performance--the board you serve on as an outside director, how well those companies do actually determine how attractive you are in director labor market.  So, what I do is include a couple of measures.  The problem is--I'll point out--is the directory is highly correlated.  So, you can use any one of these and the results don't change.  I have an obvious preference for which one.

I use the average return on equity for your firm, that is, the average return for your board.  Now, in any given year, most of the sample only serve on one board.  So, it is not actually clear that average is the right way to describe this.

The other thing I do is, compute for those who serve on more than one board [tape skip] how many boards were in the bottom quartile of industry-adjusted performance and how many were in the top [unintelligible] percentile of industry-adjusted performance.  That says you have two boards that are doing really badly but, oh, there is one doing really well.

I tried a couple of different measures of this, adjusting for size and so on.  It doesn't seem to make much difference.

I also include whether the director reaches retirement age, whether the CEO of the company--and if there is a company in the sample that there is an insider on.  These are all outside positions, but it may be because you actually are the chief financial officer for a company in the sample that has been less active.

Include an indicated variable for CEO turnover.  If the CEO of the company leaves, that is usually a big event in this literature.  Then current board positions--well, what I should say is actually lags by one year.  How many boards did you serve on last year and then relating to your business hypothesis, the idea that you serve on more than six boards.  Maybe that makes you less attractive, because you are on so many.  Then industry effects for reason [unintelligible].

Let me skip that for when it is time.

Basically, that says of the--all the director years, how many people served on a board--well, so much for my skipping it--but how many people served on a board that had a class action allegation between '85 and 2002.  A way to think about that number basically is, I have listed you as having a fraud allegation.  After the fraud allegation is revealed, that is, either basically filed or actually notated--is it revealed sort of before that, usually a filing.

What this says is, out of the 300,000 director years that have been available, about 66,000 people have served on a board that, during the period they were serving on, was accused of committing fraud, that is, essentially during the class period.  That is a distribution of outside directorships--it actually looks a lot like Charles Murray's Human Achievement Graph.

Basically, most people served any given year on one board and it tapers off to very, very few serving on 15.  One reason for this is, I'm only looking at boards, serving on boards within my sample, for one thing and, I'm not looking at mutual funds.  If you looked at mutual funds you would basically have another spike somewhere out there, sort of in the mid-30s.  Mutual funds, serving on fund boards, you usually serve on a whole bunch.  So, I'm excluding that.

What do I find?

Well, a fraud accusation in a private securities case has a statistically significant and positive impact on the reputation of directors.  That is, a fraud allegation increases the number of net outside directorships by almost a hundred percent.

To put this in context, what this says is if Jon here, my erstwhile victim, served on a board during a class period where he is accused of having committed fraud in some capacity, he actually does better in the market after that is revealed.  Now, Jon made me promise that I would not use sort of deep econometric discussions, but this is a fixed effect estimate.  Every director has an individual interstep, which means that we are really looking at the change before and after.

So, what this says is that, before he was accused of fraud, Jon didn't do as well as after he was accused of fraud.  Now, that seems particularly puzzling, right, if you think about it, but let me give you a couple of stories that might suggest why being accused of fraud doesn't harm a director.

One is that, the critics of [unintelligible] litigation are correct.  That is, the average case does not identify an actual fraud, but in fact, is designed to elicit settlement.  If this were the case, which you might expect, is that good directors, active directors, directors who have whatever it is that people want in the directors' market, directors like Jon, are good targets.

Why?  They don't want to be tied up in discovery.  He has four hundred board meetings to go to or, at least, eight, right.  So, the fact of the matter is, he makes a better target.

An alternative--[unintelligible] targeting is--but suppose most directors are utterly inert.  They do nothing while they serve on a board.  Jon, by contrast, is an active director.  He approves projects; he speaks in public; he does things that we might want him to do and that actually makes him a target simply because he has opened his mouth.  It also means that he is an active director and he is sought out.

An alternative possibility is that, it is simply direct.  That is, Jon learned something from being sued.  He learned how to sort of be a better director in the process of actually being dragged into court and deposed and having to defend what he says.  It causes him to learn something.

I don't know which one of those--they are all consistent with it.  It is, unfortunately, also consistent with the other hypothesis that I will come back to in a second.  It is just that he is more attractive, because he is totally unobservant.  So, let me come back to that for one second [tape skip] quick discussion of control variables.

I have no idea of how much time I have left.  Thank you, I'll continue on.

A few discussions.  One is, as I mentioned, these are highly correlated, but having a board in the top quartile of performance actually reduces the number of boards you have.  It turns out that, that changes sign if you leave out the average return on your board.  The reason is that often people only serve on one board in my sample.  So, these are picking up much the same thing.

One standard deviation in the number of boards in the bottom quartile actually reduces the number of outside directorships and the average return going up increases the number.  Becoming a CEO of a company in the sample reduces boards, the number of boards you hold.  The exit of the CEO of one of the companies on which you serve as an outside director actually causes a sizable reduction in the number of boards.

Skipping over the rest of these, which are largely intuitive, the one I want to point out is, becoming the CEO of a Fortune 500 company, which is a very, very rare event, has a 373 percent increase in the number of boards.  That is, you are a very attractive outside director if you become the CEO of a Fortune 500 firm, but beyond that, nothing is quite as helpful as being accused of fraud.

[Laughter]

PROFESSOR HELLAND:  Now, what does that mean?

Well, if you look at this, what this says is--I don't think we should take that point too literally, by the way.  Let me look at a couple of different measures.  Maybe what is going on is, Jon is accused of fraud.  He was serving on the board of GM but now, in fact, he is actually only able to add the Rockville Concrete Company in three other small Maryland companies.  So, I'm picking this up as a positive effect, but in reality, he is really on hard times.

That actually doesn't seem to be what is going on.  If you look at directorships with pensions in the top quartile of compensation, in ones that have options, the top quartile of values for those options, sales, employment and the Fortune 500, they are all positive.  So, what that says, for example, is Jon's net outside directorships in the Fortune 500 go up after he is accused of fraud.

What about my other hypothesis?

These are my governance measures.  Is it the case that, in fact, he becomes more attractive to firms that are poorly governed?  Do long-serving CEOs or CEOs that pack the board with lots of insiders of the corporation like him better?  The answer is no, there doesn't seem to be any difference in these coefficients than the ones I showed you before, which suggest that the behavior of what I'm calling weakly governed firms looks pretty much like the behavior of all firms.

Let me now do something different.  As you correctly pointed out, kind of the conventional wisdom or, at least, the way I read it is, that the SEC, when it actually identifies and files a case, which is a fairly rare event, actually does go after cases of pretty flagrant fraud.  So, what that would suggest is, we should see a difference.

Now, when you look at public allegations of fraud--and this is a little bit unfair, because in my sample, basically every public allegation of fraud by the SEC has a shareholder suit that goes along with it.  But if you look at that, what you find--and again, this is a rare event--is that it is consistently negative.

If you are accused by the SEC of having committed fraud, your net outside directorships decrease.  This is true even when you don't look at the firm--if you were still on the board of a firm accused by the SEC, that is, in the outside directors' market.  This isn't just you are on a board and the SEC accuses the board of fraud and you exit.  It is actually driven by other firms sort of finding you a less attractive director and not adding you or not reappointing you.

Again, I apologize for ocular trauma.

A couple of different ways of hashing private securities cases.  Maybe dropped and dismissed cases shouldn't be in here at all.  Maybe the court has ruled that there is not a legitimate fraud allegation and we don't want to consider those.  It doesn't seem to matter.  That is, if you look at these, they are still positive, directors accused in cases that actually have a settlement, rather than being dropped or dismissed.

What I'm doing, by the way, is consolidating cases.  If there are multiple cases here and they eventually are consolidated--say, five are dropped and one continues, I'm not counting that as being dropped or dismissed.  It is only if all the allegations covering a class period go away.

What about dropped, dismissed and pending?  Maybe the problem here is that, pending cases shouldn't be counted.  Again, if you drop out, it is significant, but in general, positive.

What about ones that are in the top quartile of awards?  So, this is the argument.  If strike suits are ones that you file basically to get a settlement--I hadn't heard the $2 million number, but I like that one.  I have to go back and check.

What I did--I'm actually very hopeful that the top quartile of settlement awards is over $2 million.  I don't actually know.

If you take the top quartile--I should know what that number is off the top of my head, but I [tape skip].  If you take the top quartile and only count those cases in the top quartile of settlements or that go to trial--I should say not award, but the top quartile that go to trial, these are cases that the plaintiff clearly got a large settlement or intended to go all the way with.  Now, the problem is that, since they basically all settle, I'm not getting much off the trial part.  That doesn't matter.  What drives this is the top quartile of settlements.

Now, it switches to being negative.  Now, directors who serve on boards where there is a settlement in the top quartile, now they experience damage to their reputations.  Their net directorships decrease.

The last one--and I think this is actually why I'm on the panel--is if you look post-1995, post-PLSRA, these are cases that are filed after 1995--what do you find?  Remember that we saw a drop in that trend in directorships.  What happens is, it is now negative.

Now, the total effect is probably still positive.  What this says is, after 1995, directors' reputations take a hit.  That is, being filed in this, relative to what happened before, now the average director is sort of more likely to have a negative effect.

So, what are the key findings?

Well, the effect of serving on the board of a company charged with fraud in a private securities class action increased the net number of outside directorships.  The result is robust to a mind-numbing number of [unintelligible] cases.  This is consistent with the average case being a strike suit, one that doesn't convey actual fraud and that the market is essentially discounting this.  There are a couple of reasons why that I mentioned.

One is that strike suits may be more likely to elicit settlement from a director who essentially got a high opportunity [unintelligible] or it may be that they developed useful human capital.

It also suggests that PLSRA seems to have shifted the distribution of strike suits towards more meritorious--sorry--shifted distribution of suits away from strike suits and towards more meritorious suits.  That is one way to read these results.

On that note, I will stop.

MR. KLICK:  Thanks, Eric.

Again, I will go over my quick, sort of boring comments on the methodology and then I will hand it over to Professor Perino to give us the real upshot.

Eric, one thing I wondered--and this may be sort of a computational constraint.  Have you run these models with account[?] data models, because this is all account data?

Presumably, it is not going to change, but under some situations it might give you different results that would likely be more methodologically accurate.  Again, you are dealing with large sample sizes, so you might have some trouble running these.  It is really something to check for [unintelligible], because by your nodding, I suspect you didn't do it  yet.

[Laughter]

MR. KLICK:  This is sort of half methodological and sort of half application.  As economists, we always like to look for a silver bullet, right.  One cause is generating all the variation in the data, really or all the important variations in the data.  Really, I suspect there are pooled effects here.

Presumably, all your hypotheses are correct or all the tested hypotheses are correct on some dimension in that, yes, we look for board directors who have been tested under fire.  We look for board directors who have a handful of other characteristics and presumably any treatment effect that you find is some pooling of this.

I was wondering if there was a way to tease out these different effects.  Specifically, what happens if you look at the treatment effect differing by industry?  So, some industries seem to be, according to the conventional wisdom, the targets of more of these strike suits than others.  If their coefficient for fraud allegations differs significantly from industries where strike suits are sort of known to be common.  I wonder if you can use that as some way to tease out these different effects.

Also, along these lines, this reimposed PLSRA result, I wonder if this is consistent with trial by fire.  So, if pre-PLSRA we get lots of these crazy suits and it is really valuable to have directors who have sort of put up with this stuff in the past.  Then post-PLSRA, we don't really have so many of these suits and so testing by fire isn't all that important.

I wonder if the fact that you are sign flips there is really telling us something.  I don't know if you can get anything more through the empirics on this, but you may consider thinking about that result quite a bit more.

One last sort of question is, this assumption that SEC filings or SEC suits or actions or what have you seem to impose a negative penalty on outside directorships, I wonder what the causal mechanism is here.  Is it really that these are directors who aren't very good or is there evidence that the SEC sort of targets high profile people?

I'm thinking of my former employer and good buddy Wendy Graham, right.  Does she have a big X on her forehead?  They might think Wendy is a great person and sort of a sharp person and would like, all things being equal, to have her on the board, but recognize that the SEC is going to target these high profile people.  So, regardless of what we do, we are going to have a  higher likelihood of SEC investigations and actions.

Thanks.

Do you want to answer those first and then I'll introduce Michael.

PROFESSOR HELLAND:  Thank you.  Actually, these are really good comments.

The count data one--let me ask one pointy-headed comment, which is can you run those with negative numbers?  Then the answer is--I've got to try that.  The treatment effect, that's a good idea and I'll sort of look into that.  Actually, I like that a lot.

The targeting one about high profile people, I'm not actually sure.  The one thing I have done is exclude people who are banned basically.  That seemed sort of an unfair reputational test.  If the SEC says, you know, you aren't ever allowed to serve on a board, you probably are going to go down in my sample.  So, I sort of took those people out.

The other thing I did was, people who are actually named in a suit, I have actually run this without them and list only the other people who served on the board.

The contrast here is that, in the private cases, it is actually very hard, because they all settled, to identify which director it was they hit.  Often, it is a large number of officers and directors or something.  When you go and look at the case, it is often very hard to sort of parse out who specifically was named.  That seems to be somewhat fluid.  So, we took the whole board as a first pass.

In the case of the SEC, it is actually very easy.  They identify people.  It actually makes it difficult to find which company it was, because they often will say, this person is sort of implicated and then you have to figure out what company they are talking about.  So, I had actually run it leaving out that person.

That doesn't seem to be driving it.  It actually seems to be the rest of the board.  So, I will make amends for that.

Thank you.

MR. KLICK:  Now, for more policy relevant comments.  We are going to here from Professor Michael Perino, who teaches securities regulation and litigation at St. John University School of Law.  He has written quite extensively in this literature.

Professor Perino.

PROFESSOR PERINO:  Thank you.

I don't know if it is more relevant, but now that you guys have done all the hard work, I can do the easy questions, the policy questions.

A couple of points on your point and just one small point on yours before I forget it.

In terms of targeting high profile people, let's think Martha Stewart, for example, just to pick a name at random.  My sense from the SEC--and maybe Adam will address this, since he used to be there--is not that they go looking for high profile people, but having found a potential violation, all things being equal, they are more likely to bring the case against a high profile person, which makes some sense.  From the deterrence perspective, you get the free publicity, more bang for your deterrence buck.  Point number one.

Point number two, on your DNO [ph] graph, you show a big spike up in DNO rates in the mid-1980s.  DNO is going to cover two kinds of cases, cases under state law, alleging the directors' fiduciary duty to the shareholders and cases under federal law, essentially alleging these kinds of securities fraud.

The big spike in the mid-1980s is as a result of those [inaudible] cases.  So, you are not picking up the securities cases in that spike.

Okay, now the policy questions.

What I wanted to focus on was, essentially, what do the findings in these papers tell us about how we should structure our system of securities law enforcement and the relative weight we should accord public versus private enforcement.  This is what we are really talking about here.  You can imagine three different models of securities enforcement.

We can have a pure public enforcement system, a pure private enforcement system, which I don't think anybody other than Bill Rackey [ph] probably wants or the mixed model we have now, some other version of the mixed model we have now.  Then there is the question, of course, as always, what is going to give us the greatest benefit of securities enforcement with the least amount of cost.  That is really the question we are talking about here.

Now, the reforms that have been put in place so far essentially have been reforms that have contemplated that we are going to maintained a mixed model of securities litigation.  What we want to do is to limit the cost, the main cost being the non-meritorious suits that are out there.  The way we are going to do that is essentially to put in place a series of procedural reforms.

We are going to make the initial hurdle that the plaintiffs have to get over in the beginning of the cases higher.  We are not going to give them the benefit of discovery in order to get over that initial hurdle.  We are going to make--and you can say that this is now leaking over to substantive reforms.  We are going to make some kinds of allegations much harder to satisfy in terms of false forecasting allegations.

So, what you can see, obviously, is essentially what we have done is, we have created a screening device.  We're giving the courts the ability--at least, we hope we are.  We are giving courts the ability to weed out more of the bad cases at the filing stage and let the good cases proceed or, at least, that is the theory anyway.

What these two papers both present I think is, in some sense, good news.  The screening device seems to work.  On average, the cases seem to be better than they were prior to reform act.  Of course, the screening device that we are using is, by its very nature and obviously so, a very crude one.  It is only going to be partially effective in weeding out bad cases.

Why?  Well, there are a number of reasons why.  This goes to the ultimate empirical question here we are talking about here, the relative [unintelligible] of the average security case.  Well, who is to say what is a good securities case and what is a bad securities case?  In many ways, whether we have a strong inference that the defendants acted with fraudulent intent is really going to be in the eyes of the individual judge.  That is likely to be relatively resistant to appellate court review.

So, we have a very crude screen right there.

We also have a situation where different courts apply different standards.  Now, in another paper Adam wrote with [unintelligible], they seem to find that the standards the court applies has a big difference at the end of the day in terms of the likelihood of getting dismissed.

The one interesting thing to see in terms of the reputational, secondary market effects might be does it matter what standard we are using to evaluate these cases.  Do these cases with the more rigorous review at that early pleadings stage have a greater reputational effect?  Why else are we going to see this as a relatively crude screening device?

Obviously, courts are going to make mistakes.  You can't tell often in that initial stage whether something is a good or a bad case.  Plaintiffs' attorneys have some ability to disguise the non-meritorious case as a meritorious case.  If you look at plaintiffs' attorneys behavior, it is very interesting.  You can explain a lot of it as being an attempt to cause courts to make mistakes, mistakes, obviously, letting cases go through at that initial pleadings stage.

So, courts consistently, across the board criticize plaintiffs' attorneys for essentially constructing convoluted and lengthy complaints that are hard for the courts to evaluate.  Well, the reason they might want to do that, obviously, is to make the court's screening efforts much more difficult to do and to increase, essentially, the number of false positives that are out there.

I think the imprecision of that standard often tends to explain Adam's finding about high levels of insider sales being a significant explanatory variable, while abnormal sales aren't.  There is a difference between what courts say they do and what, in fact, they actually do.  So, insider sales must be unusual.  That's what the courts say, but they are not entirely consistent in how they apply that standard.

Because of that inconsistency and if, as I suspect the case is, it is relatively cheap for the plaintiffs' attorneys to investigate and to file the complaint, it is going to make a lot of sense to them to bring all cases in which they find high levels of insider sales.  They know that they are going to lose some of them, but enough will get by that initial pleadings stage for it to make sense then to pursue all of those cases.

So, we have a relatively crude screening device.

In terms of forward-looking statements, it may be that they are still a significant percentage of the cases.  My sense, when I looked at this initially was that, the forward-looking forecast with no other allegations had fallen by the wayside.  It was forward-looking cases that were combined with something else.  It might as well--we have it; let's throw it in, kind of thing.

So, given the reforms we put in place, what we have is pretty expected.  We see more meritorious cases, but still a large number of non-meritorious cases are out there.

Can we construct a better screen?  I doubt it.  We are talking about something that is inherently imprecise.  Was there a fraud or wasn't there a fraud?  I don't think we can do a better screen in that area.

So, what should our regulatory response be?  That is really the question.  Before I talk about that, I want to point out what I think is an interesting paradox about these reforms.  You can call it the paradox of pleading standards, if you want.

Go back to initial first principles that Eric talked about.  Why do we have securities class actions in the first place?

Well, the whole idea behind securities class actions was this private attorneys general system.  The SEC can't do it all.  They don't have the budget.  We want to maximize enforcement resources.  We want to complement the SEC's efforts in this area and criminal enforcement efforts as well.  Plans are actual economic actions [unintelligible].

Now, what Congress had in mind was, well, let's see.  We will put in place a higher pleadings standard.  What is that going to do?  Well, it is going to force attorneys to do better research.  Before they file cases, to do a better job of themselves weeding out the weak cases, because they know they are not going to get by.

Actually, that's not what happened.  In a paper I did, I looked at the effect of the Ninth Circuit adopting the highest pleadings standard.  I looked at how long it took to file the case from the end of the class period until the time the action was filed.

If you look at that, what you find is, before adoption of a higher pleadings standard, the highest form that is out there in the courts, the cases actually were filed more slowly, significantly more slowly than after they adopted the higher pleadings standard, exactly the opposite effect of what you expect.

What is the explanation for that?  Well, I think the explanation for that is that, plaintiffs' attorneys as rationale economic actors will naturally gravitate to cases with the lowest search costs.  The effect of the higher pleadings standard means increased risk is going to force them to focus on more obvious cases of fraud.

So, it might be--I don't know if you find this, but is there more overlap with SEC actions in post or pre.  There is an obvious problem there.  We found a way that reduces some of the cost of securities class action lawsuits.  We put these higher procedural burdens in place, but the cost eliminates the original benefit for having securities class actions in the first place, because the were supposed to complement SEC actions.  Maybe now we are just duplicating SEC actions.

Now, that is the empirical question.  We don't know the answer to that question.  We don't know if that, in fact, is actually happening.  It is something that is worth looking at.  Then that leads us to another question, which is, well, if that is the effect of these kinds of reforms, well, what should our response be.  Go back to our three potential models for securities litigation.  What should we do?  Does it mean we should just jettison private securities lawsuits?

I suppose you could come to that conclusion.  I don't think it will work down the street, on Capitol Hill, but that is one possible conclusion.

I think there is another set of empirical question we have to ask out there.  We also have to understand what--and this really hasn't been, in my mind, addressed in the literature at all at this point.  What is the dynamic interaction between the SEC's enforcement efforts and the enforcement efforts of private plaintiffs' attorneys?  Does the SEC in some sense structure what they are going to do in enforcement based on what the private actors are already doing?  We can think about this in a number of ways.

For example, where are we most likely to have under-enforcement in private securities litigation?  Well clearly, as Adam's variables show, it has to be sufficiently large damages to create a sufficient attorneys fee at the end of the day in order to get the attorney involved.  So, where we are going to have under-enforcement in cases are where the damages--it might be a real fraud, but we are not going to get a big enough payoff at the end of the day to make it worthwhile for the plaintiff's attorney.

Well, what has the SEC done?  Well, what we see is the SEC traditionally has focused a lot of its enforcement efforts on cases that, in fact, aren't going to  yield that kind of high-based payday for plaintiffs' attorneys.  Stock manipulation cases, cases involving penny stock fraud, insider trading kinds of cases, all the kinds of cases where it might make sense and that might be a perfectly rational response to the best way to allocate enforcement resources.

We know the plaintiffs' attorneys are going to deal with these public fraud cases.  Let them do that and we will do something else, stuff that we know they are not going to do.

Now, I think it is unreasonable to believe that the SEC is ever not going to bring high profile cases involving publicly traded companies.  As a matter of political pressure, either because Congress is going to ask lots of questions like, why haven't you brought an action against Jeff [inaudible] or because Elliot [inaudible] is going to say, well, you see, the SEC is not doing anything.  So, I have to bring this case.

The SEC is going to get involved in these cases and, maybe that makes some sense.  There does seem to be some reputational impact on the most serious cases.  It also is true that, if you look at the data on who pays at the end of the day, most of the settlements still come out of either the company or out of insurance.

So, we may not have a big enough deterrent impact on just those reputational effects, at least, in the most serious cases.  So, it might make sense for the SEC to pursue a strategy that basically says, let's leverage off the private plaintiffs' efforts.  Private plaintiffs bring cases.  We watch what is going on.  We see if it really seems like a serious fraud case, in which case, we can come in and then seek maybe an officer-director bar and prevent this person from being a director in the future, that sort of thing.

So, I just want to conclude by saying, before we take that next step of, okay, here are the empirical results of what the reform act has done, we need to think a lot harder about the question I think hasn't been addressed so far.  That is, how these various enforcement actors work together in this system that we have.

MR. KLICK:  Thank you, Professor Perino.

I just want to second that last point.  It is interesting in sort of multiple--of the events that we have had as part of this project, this issue keeps arising.  We have some optimistic assumption that private and public efforts are going to sort of balance each other's shortcomings out, but we really don't have any empirical evidence of that.  We don't have any empirical evidence of what direction these counter-acting or multiplicative [sic] actions are doing.  So, that is an institutional question that really deserves much more research.

Unless either of our two presenters have civic points they want to reply to in Professor Perino's comments, I will move to the floor.

QUESTION:  [Inaudible].

MR. KLICK:  Questions, please wait for the microphone and state your name and affiliation, please.

[Pause]

QUESTION:  Brad Shingleton[ph], Deutch Telecom.  Do any of the panel members have any feelings about how the pending Class Action Reform Bill in Congress might affect your conclusions or the issues as you see them?

ANSWER:  I don't think it covers securities fraud class actions, but I haven't studied it carefully.  I think it is for state law, cause of action on other [inaudible].

ANSWER:  I think that is probably right.  I suspect that any reform that is going to essentially do the same kind of procedure but more in this area.  So, for example, one thing the reform act purported to do was to impose sanctions risk on plaintiffs' attorneys in cases that get dismissed.  It turns out, it doesn't really do that, because all it requires is that the court review whether sanctions are appropriate.  Since most of these cases settle, there is really not much of an incentive to impose sanctions on anybody in one of these cases.

If you were to make a more meaningful sanctions risk for plaintiffs' attorneys, it is likely to only reinforce the paradox that I'm talking about here.  It is likely to only reinforce that we are just going to focus on the most obvious cases of fraud.

ANSWER:  I haven't seen a lot of evidence that the judges actually do the mandatory sanctions inquiry after a dismissal.

ANSWER:  A, they tend not to do it.  If they do it, they do it in a sentence.  There is no basis for sanctions here.  As far as I know, they are now almost ten years of reform act litigation.  There has been one case that imposed any sort of meaningful sanctions, a couple of hundred thousand dollars.

ANSWER:  The only thing I would add to that is that, most of these cases, from what I have found so far, end up at the federal level, most of the securities cases.  That is, I found relatively few at the state level.  That act, I think, specifically is addressing issues largely at the state level.

It gets at the larger question that--I don't know that I would want to make this case, but you potentially could.  This is kind of the best case scenario for the private attorney-general model.  A small court in the middle of Nowhere, Alabama looking at cable rates is probably less likely to be a good example for the private attorney-general model.

So, in that sense, I don't know how that bears on the Class Action Bill, but I do think we are looking at, in my experience, probably the best case.  If these cases are meritorious, we don't necessarily want to conclude that the cable, you know, overcharging cable rates is meritorious.

QUESTION:  Michael Greve, American Enterprise Institute.  I have one question for Eric and then maybe one for the panel.

My question about the directors is simply this.  Isn't there some direct way of getting at this in the following sense?  If there is such a boost to the reputation, right, they should advertise it presumably, right?

[Laughter]

QUESTION:  No, I'm serious.  I mean, you know, Bumford Motor Company has been sued 16 times.  I don't know how that market works, the search market, but maybe headhunters and stuff have some opinion about this.

It could be true that everyone in the industry holds false beliefs about it and thinks it is a terrible damage to one's reputation.

ANSWER:  I had that thought, which was I wanted to actually--so when I asked people, the general statement was, it doesn't matter.  Now, my other thought was, perhaps, I should open a mutual fund, which said that I should look and find companies run by people who had been sued a lot.

I think what this says, although I would not want to conclude too much from this--is it probably has more to do with targeting than a direct attack.  Just in my discussions with people who do this, most of them on the directors' side--I haven't actually talked to any headhunters.  Their statement is, no, it is not helpful at all to be sued.  I didn't learn anything, period, but, no, it hasn't hurt me at all in the market.

So, they tend to confirm--I think that's what the [unintelligible] said--people who are active tend to end up being targeted and the market generally views them as not having been damaged by that.

QUESTION:  My other question for the panel and maybe for Adam personally is this.  You mentioned at the beginning that the Private Securities Litigation Reform Act has not affected all industries--no, sorry.  That the composition of suits pre-reform and post-reform isn't the same for all industries, which is why you picked the computer industry, right.

Now, this could be true as--I mean, this could be the case for any number of reasons.  One is some industry changes.  So, it was a high-flying industry before.  Afterwards, it behaves like utilities.  So, of course, the suit composition changes.

It could also be the case, could it not, that the screening, as you described it, of the Private Securities Litigation Reform Act affects different industries in different ways.  I wonder whether there is any research on that or if I am totally off base?

ANSWER:  No, it is an interesting question.  The general assumption is that, it is more economic effects within the industry.  So, savings and loans got sued right and left at the end of the '80s and the early part of the '90s.  Now, that industry is very different after Congress did a bunch of things to the industry.

The telecommunications industry, they did not get sued in the early 1990s, much higher incidence [unintelligible] have sued.  It has more to do with the froth of the tech bubble sweeping in telecommunications firms, I think.

It is an interesting question as to whether the act had some effect on the industry choice of suits.  I'm not aware of research on that question.

ANSWER:  I looked certainly at the makeup of high technology firms.  You find that, essentially, pre-imposed reform act, the numbers are about the same.  Firms in high technology are about twice as likely to be sued as firms in other industries.

It may be a number of factors working together.  For example, high technology firms tend to have relatively--at least, they did for a while--relatively large market capitalizations, a high turnover in the stock price, damage-related factors that are going to make them more attractive targets for law suits.  Combine that with a higher degree of equity compensation, that is, more trading by insiders, high levels of trading during the time when the fraud is allegedly out in the marketplace and you have the factors built in that might suggest that these are more profitable targets for plaintiffs' attorneys to--or it could also be that this sort of--from specific human capital built up.

You learn the high technology industry.  We know there are a lot of targets out there.  We get good at suing high technology firms.  We know what to look for.  We know the kinds of allegations to make.  So, it is profitable for us to keep going back to those cases.

MR. KLICK:  Are there any other questions?

QUESTION:  Hi, I'm Mark Salzburg [ph].  I am an attorney here in Washington.  I have a question for Adam Pritchard.

I wondered if one of those variables you looked at or if you saw any correlation--is by the expansion in coverage of capital markets in the business, perhaps and, how that relates to shareholder losses.  I think what the reform act--plaintiffs often get put in the position where they have some kind of indicia of fraud, but they don't really have enough to meet the big guys [unintelligible] requirements.  So, they are in this Catch-22 where they need to rely on outside sources, such as, investigative journalism or somehow a fortuitously leaked document or a whistleblower.

Because of this really high quality business journalism that we are seeing, I think that really informs the substance of a lot of law suits.  The one I can think of most famously is the Daimler-Chrysler litigation where I think a lot of people were aware of the reporting, that this wasn't really a merger of equals.  It was only when this was sort of splashed across the front page of the "Financial Times" that, that case really became strong enough for the lawsuit to jell.

So, I wondered if you could talk to that.

ANSWER:  I thought the prompting in the Daimler-Chrysler suit was not rigorous investigative journalism, but a rather stupid CEO--

[Laughter]

ANSWER:  --who said something he shouldn't.

The question of whether newspapers are having an effect is not something we have looked at.  I'm trying to think how you might try and code for that.

The first thing I would think is, can I code for it.  You could code for the number of stories, I suppose, about particular companies and see if there was a correlation between press coverage of that company.  It would be harder to code for press coverage that was negative.  I have seen studies that do that.  I'm a little bit skeptical that the coding is done in a way that is not subjective.

I have seen studies that--well, there is a story that says this company is engaged in fraud and if you have a story that says the company is engaged in fraud, you are more likely to get sued.

The company didn't say it had engaged in fraud and there is no adjudication of that.  So, people can read newspaper stories very differently.  So, that would be my concern about trying to capture that.

A more indirect but maybe a more objective way of getting at the amount of attention a company receives might be the [unintelligible] analyst coverage, to see if companies with more analysts following them--which would tend to correlate with more turnover, if that makes an incremental difference above and beyond share turnover in terms of who gets sued.

MR. KLICK:  Anymore questions?

[Tape ends]

MR.        :  [In progress] believe it is on litigation in general and particularly the reforms of litigation.  The empirical research probably isn't going to answer questions, but it is going to suggest sort of more interesting research, institutional research.  It is the question of how do public and private actors interact and does that improve welfare or worsen welfare.

These are all questions that we need to have answer before we are going to be able to craft useful policy.

That is really quite important, because a lot of people in this area think that they know the answers.  They think they know that things are bad in certain areas and that reform, however defined, is the answer or more public scrutiny is the answer or more reliance on market mechanisms is the answer.  We really don't know.  These aren't the sort of questions that can be answered on an a priori basis.

So, that is why empirical research is really what is needed in this area.

Thank you very much for coming and I hope the redline is no longer on fire.