What Do We Know about Contingency Fees?

September 22, 2004

Unedited transcript prepared from a tape recording

Proceedings:
MR. GREVE:  We're here to discuss attorneys fees and class actions.  I don't have to explain to this expert audience that it is a very intensely controversial issue, and one of enormous practical relevance.  I'll just mention two things that have concerned people.

One is that at least at the higher end of class action compensation, asbestos, tobacco, a lot of that money, after all the yachts and the jets have been purchased, finds its way back into the litigation system itself, and another big chunk of these fees finds its way into the political and into the legal process.  That's altogether fitting, I think, that one of the former chairmen, I suppose, of the finance committee of one of the presidential companies was, or is the leading asbestos lawyer in the country.

Given that political and economic salience in the role of these fees and what they do in the larger political and legal environment, it's actually amazing how little we know about contingency fees, and on the theory--and about the likely effects of various reform proposals that have been advanced on the theory that there's nothing that economists can't solve or at least have firm views on.

We've assembled, really, the leading authorities in the country on attorney's fees and the economic aspects.  We'll have two paper presentations and a set of comments.

We'll start with Alex Tabarrok, who together with Eric Talley [?], has produced a really very interesting paper on the effects of attorneys fees, caps, and the experience with those caps in a particular state.  He'll present that paper first, and to comment on that paper, we have Lester Brickman from Cardozo Law School who himself is an authority on attorneys fees and has written extensively on attorneys fees, especially in asbestos litigation, and I should say that the CVs of the speakers are in your materials.

And then we will hear from Jon Klick, whose paper is half comment, I suppose, on a preceding paper by Ted Eisenberg, the fourth speaker here that has caused a major stir and have an explanation of why attorneys fees and contingency fees in class actions, in particular, might be higher in some jurisdictions than others.

We'll first have the papers and the comments and then move to a wider discussion.  I have to say that we have to end ten minutes before the appointed hour, so that would make it ten of four.  Without further delay, Alex.

MR. TABARROK: I want to begin by talking a little bit about some of the benefits of our contingency fees and the most obvious one is that it improves the alignment of lawyer and client interests and this is important because of the difficulty that clients have in monitoring what their lawyers are doing.

It's difficult not to know how many hours a lawyer is putting into a case, more difficult to know how much effort they're putting into a case.

This has two problems.  The first problem is that because the clients don't know what the lawyers are doing, the lawyers may not put enough effort and hours into the case.

The second problem is that if, let's say if the case loses, or if the client gets less than the expected, which is almost always going to be the case I think, the client is likely to blame the lawyers.  You know, "you must not have worked hard enough, you must not have fought hard enough" about this case.

And at least when you have contingency fees, not only does the lawyer then have an incentive to work in order to have the case win and earn more money, but the lawyer can then say look, we're in the same boat.

So if it goes wrong, it goes wrong for me just as well as it goes wrong for you, and in this way disputes between the client and the lawyer can be reduced.

 We also see contingency fees in other areas of the economy, it's worth pointing out, sales people, for example, on commission, is essentially a contingency fee.  The managers may not be able to monitor the sales people, and so they put them on commission, which increases their incentive to do what's appropriate in order to earn sales.

If I have a critique of contingency fees, I might actually argue that we don't see them enough, and I would say wouldn't it be nice to have contingency fees in other areas of the economy?

I'd like to see contingency fees in medicine, for example.  Here's another case where the patient has no idea what the--how much effort and time the doctor is going to put into the case.

You know, Is the doctor reading the journals?  Is the doctor going to the conferences and keeping up with the latest techniques, and so forth?

And contingency fees could actually help in this area.  I'd like to see, for example, 20/20 vision or your money back.  That would be a good deal.

Another benefit of contingency fees is greater access to the legal system.  Now of course is the legal system is crummy, then this is a cost; okay.  And we certainly want to take that into account.

But I think it's better to--if the legal system is crummy, we ought to solve that, fix that problem, okay, and keep these two problems separate.

I think it's fairly clear, but let me just give you an analogy here.  If you have a great product, you have a great idea for a new product, but in order to get that idea out in the market you have to build a prototype and you've got to invest some research and development funds, and so forth.

And in order to do this, let's suppose you go to a bank and you say, Could you loan me some money so that I can develop my product.  The bank is going to say "forget it."  The bank is going to say our highest possible return, say 10 percent, you know, a year, or something like that, that does not justify, cannot justify the possibility that you're going to go completely bankrupt.  Okay.

So the bank won't give you money.  So what do you do?  Well, what you do is you go to a venture capitalist, and the venture capitalist will give you up-front funds in return for a share of your idea.

The capitalist is willing to do that because their upside potential is unlimited in exactly the same way that your upside potential is unlimited.

So just to upset people, I like to say that the contingency fee lawyer is the venture capitalist of the legal system; some people would say the vulture capitalist of the legal system.

But I won't steal all Lester's lines.

Another benefit of contingency fees is that the contingent fee lawyer screens cases, and I can perhaps best explain what this is all about by quoting Elihu Root who once said about half of the practice of the decent lawyer consists in telling would-be clients that they're damn fools and they should stop.

Well, if you think about this, the contingent fee lawyers will tell their clients they're damn fools more often than hourly fee lawyers.

And the reason is that the contingent fee lawyer is going to look at the case, is going to weigh the probability of winning, is going to assess whether it has merits or not, and if it doesn't, they're going to say "forget it," it's not worthwhile.  They'll tell their clients it's not worthwhile pursuing this case cause you're not likely to win.

On the other hand, the hourly fee lawyer may very well say sure, I'll take the case, why not?  They're going to be paid by the hour.

This has a testable implication, suggests that drops should be higher under hourly fees, and the intuition is that as time passes and cases move through the legal systems, clients are going to gain information.

With better information, some clients with weak cases will drop their cases, decide not to pursue them any further.

In other words, some of the hourly-fee clients will learn what a contingent fee lawyer would have told them on day one and they'll drop their cases, wiser but poorer.

Now some states restrict contingency fees, especially in medical malpractice cases.  So we can expect that in these states and types of cases, a greater proportion of lawyer compensation will be hourly in nature, will be noncontingent in nature.

So thus we predict that drops will be higher in states that restrict contingency fees.  That's the prediction.  And we're going to have two tests, okay, two tests of the relationship between contingency fees and drops.

The first test is different states at a single point in time.  So we're going to compare drops in states that limit contingency fees in medical malpractice cases but not in other types of cases.  We're going to compare that with drops in states that do not limit contingency fees.

And we're going to use auto cases in which contingency fees are not limited in either type of state to control for other factors, and if this is not clear now, I think it will become clear shortly.  That's the first test.

The second test, we're going to look at the same state before and after restrictions on contingency fees.

And here, we're going to use a short treatment window to control for other factors.  So we have two different tests, two different types of controls.

Let's look at the first type of test.  The result is that drops are significantly higher in states that restrict contingency fees.  Drops are significantly higher.  And just focus for the moment on states with limits on contingency fees in medical malpractice cases and look at medical malpractice cases.

The proportion of drops there, here, is 18.3 percent.  Now look at states without limits on contingency fees.  Again in medical malpractice cases, the proportion of drops is 5 percent.

So many more cases are dropped in states which limit contingency fees, suggesting that there's something to the screening function of contingent fee lawyers.

Now you might say okay, but there's a lot else going on between these two different types of states.  Okay.  Maybe it's other factors, maybe it's not the contingency fee, maybe it's other factors which are driving this result.

So to try and control for some of these possible other factors, we look at auto cases, drops in auto cases, in states with limits on contingency fees in medical malpractice cases and states without those limits.

So if you look at the drops for auto cases in states with limits on contingency fees in medical malpractice cases, you see it's 9.7 percent.

If you look at in states without limits on contingency fees, it's 10.2 percent, in other words, the same.  Okay.  So what we see here is that drops are much higher in medical malpractice cases in states that limit contingency fees and this seems to coincide with the types of cases on which contingency fees are limited and not with other factors to do with differences in the states.  Okay.

So we see the effect where the contingency fees are expected to have bite and we don't see the effect where contingency fees are not expected to have any effect.

And I should point out that in the longer paper, for those of you who are interested, we also do this in a regression context with a bunch of other variables, like differences in the laws between states, differences in the type of defendant, the type of plaintiff.

There's a bunch of other things we can troll for, the story stays the same.  Okay.  So the simple method of looking at auto cases, difference in difference method, the simple method tells us the whole story.

I have copies of the longer paper, you can get it on my Web site, if you're interested in those further results; but the story is the same.

Okay; so that's test number one.  Test number two.  Florida limited contingency fees in November of 1985, November 1st, '85, and we're going to look at drops in a 300 day window before and after that limitation, before and after November 1st, and this narrow window helps us to control for other factors.

So the argument is is that within this timeframe we wouldn't expect much else to have changed, other than the fact that contingency fees were limited.

So it's a different type of test, different type of control.  What do we find?  Well, we find again that drops in Florida increase after contingency fees are limited.

Before limits, we have a drop proportion of 22.7 percent.  After limits, 26.2 percent.  So not as big a increase, about 3.5 percentage points, and you wouldn't expect as big an increase with a short timeframe.

And that's about 15 percent.  3.5 percentage points are about a 15 percent increase, and again it's statistically significant, all that kind of stuff.

So in both cases we're finding that drops increase when you see limits on contingency fees, suggesting that the hourly lawyers take on more cases, that clients realize eventually that the cases that they're paying for aren't going to win and they drop them.

All right.  What about some of the costs and criticisms of contingency fees?  Well, one which is fairly--it's talked about a lot in the literature, is that a lawyer has a greater incentive to settle than a client, because the lawyer bears all the costs of a trial but receives only a third of the benefits.  Okay.

This may be true.  You know, in theory, this is correct.  But if it's a problem, there's actually some simple solutions.

One, you can raise the contingency fee conditional on going to trial.  So you can have 25 percent, if settled, 33 percent if the case goes to trial, and that's not uncommon.

Two, if this is a "big deal," you can split the costs and benefits between the client and the lawyer equally.  So you can say the lawyer gets a third of the settlement but only bears a third of the costs.  And I haven't seen contracts explicitly set up like this but it's not that uncommon either for clients to bear some of the costs.  So something approximating like this is certainly possible.

So it seems to the extent that this is a problem, there are some solutions.  Empirical result, both our own, and in other papers, don't find a big settlement effect.

Now you might argue because the papers which have looked at this have not been well done, that's quite possible, I'm not going to argue for this in a big way, I'm just saying that of the tests of this settlement critique which have been made, nobody finds that it's a "big deal."  What else?

Time to settlement.  Here we have an interesting debate because Wally Olson says most litigants tire of their fights.

MR.          :  [inaudible.]

MR. TABARROK:  Okay.  Let me tell you what he says and you can fix the computer while--yeah, I'm still looking at it.

MR.          :  Wally Olson often has that effect.

[Laughter.]

MR. TABARROK:  Right.  So Olson says that the litigants tire of their fights, if not at first, then after a while, and, quote, this is a quote: "At some point the clients would rather get on with their lives and hold out for a little more and the lawyer with a big war chest has an incentive to make you wait in order to go for the extra money."

Said Bernstein, David Bernstein--

MS.          :  Sometimes this works--

MR. TABARROK:  Bernstein, who's also against contingency fees, he says it's not that the lawyers make you wait, he says time is money in the legal business so it pays the contingent fee attorney to settle as quickly as possible.

So Olson says the contingent fee lawyer makes you wait, Bernstein says he makes you go quickly, okay, so the contingent fee lawyer can't win.

What do we think?  We think Bernstein is correct, that cases will be quicker under contingency fees, but we put this in a more positive light.  The reason we think this is that the hourly fee lawyer has an incentive to rack up the hours.  Okay; it's pretty straightforward.

So we tested this.

MR.        :  [Inaudible.]

[Laughter.]

MR.        :  You're not being paid on a contingency fees--

MR. TABARROK:  That's right.

[Computer problem continues.]

MR. TABARROK:  Lester was trying.  That explains it.  I must have been doing well, then, if he was trying to--sabotage, yeah.

MR.        :  Well, there is a disconnect.

[Pause.]

MR. TABARROK:  All right.  So we haven't lost too many times.  Using the same data I talked about earlier, we look for this, who's right? Olson or Bernstein and why?

Well, states that have restrictions on contingency fees have settlement times which are longer by 22 percent, and again this is relative to auto cases as a control factor as well as other factors.  So even after taking into account differences between the states, the ones which restrict contingency fees on medical malpractice cases, those cases take 22 percent longer to settle.

Same thing in Florida.  After you had restrictions on contingency fees, the time to settle increased by 13 percent.

So under our explanation, the guy was earning money by the hour, so the hours are racked up.

Now what about this excessive returns to contingent fee lawyers?  And here I'll just quote Lester.  Says, "Contingency fee lawyers have not only flouted ethical rules and fiducial protections but have also imposed substantial rents on tort claimants as the price for tort claiming.  Contingent fee lawyers generate substantial rents and obtain inordinately high rates of return, not infrequently amounting to thousands and even tens of thousands of dollars an hour."

Well, I think it's easy to see how one could conclude this.  Suppose a lawyer puts 30 hours of time into a case, okay, and let's suppose that damages are uncertain, a half probability, 50 percent change of 25,000 and a 50 percent chance of five thousand.

Then when you look at some of the winning cases, some of the cases that win, the equivalent hourly fee is going to be $833 an hour.  But if you look on the ones that lose, it's going to be just 166, are the ones that earn less.  Just 166.  So the same time, depending upon whether the case wins or loses, because there's uncertainty, you can generate, quite easily come up with anecdotes where the hourly rate is extremely high.

Remember the earlier analogy I gave to venture capitalists?  It's common to hear about venture capitalists who multiply their investment ten or even a hundred times.  But those huge returns are paralleled by many investments that return little or nothing.

We also have to remember that legal disputes are tournaments with big first prizes and lousy second prizes.  What I mean by that, it's good to win, it's bad to lose.  So the market for lawyers is similar to a winner-take-all market, and the big-name lawyers, the Joe Jamals and Dickie Scruggs, and so forth, they're the super stars of their profession, and they're paid exorbitantly for the same reasons that Julia Roberts and Derick Jeter are paid exorbitantly.

So I don't think, I'm not necessarily--in fact I'm not necessarily--I'm not in favor of this.  I think it's a sad comment on, you know, our legal system and society, when some of these people can earn these extremely high fees for impeding good law; okay.  But that's not my point.

My point is is that the explanation for those high fees is not to do with the contingency fees, it's to do with the fact that they're super stars.  It's to do with the fact that the very top guy is paid a lot more than the next very best guy because winning is everything.  If you don't win, you lose, and so you want to hire the best guy.

Data collected by Herbert Kritzer indicated hourly and contingency fee lawyers make about the same.  Lester has a number of criticisms of that data.  Since I'm an economist, I say, you know, theory rules, so even though I do some data work.  And look, economic theory says look--it says you can explain why lawyers, as a group, are earning a lot of money because of restrictions on entry into the profession and various things like that.

But what is utterly implausible is to suggest that the hour--is to suggest that the contingent fee lawyers are making much, much more than the hourly fee lawyers, because any lawyer can switch.  It's just implausible to think that the plaintiff lawyers are making a lot more than the defense lawyers.  You know, what? Because the defense lawyers are willing to give up income in order to protect their clients, they would do a good job, you know, they love it so much?  I don't think so.

If the plaintiff lawyers were making a lot more money, the defense lawyers would switch over.  You have to have it that these types of people are earning the same amount.  All right.

If fees harm the interests of clients, on the contrary, we found evidence that restrictions on contingency fees contribute to wasted plaintiff effort and longer times to settlement, which is not too surprising.  Clients watch out for their own interests, and the market for lawyers is pretty competitive.

But then we have Olson's comment.  He says look, the case against the contingency fee has always rested on the danger it poses, not to the one who pays it but to the opponent, and more widely to justice itself.  As other nations recognize, it can yoke together lawyer and client in a perfectly harmonious and efficient--I just want to pause on that because I think while--I'd like to see Wally and Lester debate this, because Lester argues that it's a rip-off for the client as well as a rip-off for society.  Wally says perfectly harmonious between the lawyer and the client.  Efficient.  All right.

But that it is still a rip-off for society.  Olson continues: There are things lawyers will do when a fortune for themselves is on the line, that they won't do when it's just a fortune for a client.

The raw data, however, does not appear to be consistent with this argument, and I'm going to show you some of the raw data.  I want to presage these comments by just saying that this is just the most basic sort of data.  There's no big research design like we had earlier.  So there's lots of criticisms that one could make of the following sorts of data, so it's sort of preliminary work.  Nevertheless, it's more than anybody else has done, including opponents, so I think the data's worthwhile to take a look at.

First of all, awards in medical malpractice cases are actually higher, higher in states that limit contingency fees.  Here's the data.  So in medical malpractice cases in limit states, the mean award is 500,000.  In no limit states it's 225.  That's the mean; yeah.  Now you might say--

MR.          :  [Inaudible.]

MR. TABARROK:  No; no.  This is state court.  '92.  '92.

Now you might say, okay, but there's difference between the states, and indeed there are.  Auto cases have a higher mean in states which limit as well as compared to states which don't limit.  Nevertheless, it seems about double in both cases.  Nevertheless--and you don't see any strong evidence, for sure, that restrictions on contingency fees are holding awards down.

So one could argue with this, but the prima facie case is that, if anything, you know, awards are higher, and maybe, as I say, there may be other explanations for that, but we don't see, on the first pass, we don't see any evidence that these restrictions are helping things.

You recall Olson's comment that a lawyer on contingency fee will do things which a lawyer on the hourly fee won't.  If this is true, if the contingency fee increases the incentive to behave unethically, as a theory I think it sounds good.  How big is the effect?  How big is this effect?

So we try and measure this.  We look at complaints per lawyer and what we find is that the complaints per lawyer are virtually identical in states that limit and do not limit contingency fees in medical malpractice cases.

It's 10.98 percent in states with limits; it's 10.26 percent in states without limits.

So we don't see, if the lawyers are behaving unethically, the clients or their counterparties or other parties don't seem to be complaining about it.

All right.  Let me just, some brief discussion and conclusions.  I'm not saying that this market is a perfectly normal market, there are some peculiarities.  Competition seems to be primarily on non-price grounds, leading to an oversupply of lawyers.

This is similar to what we see in the real estate market with commissions.  So it's a peculiar market and Lester has done a lot of fascinating work bringing out some of those peculiarities.  I also would agree that contingent fee reformers have some good ideas, such as basing fees on additions to settlement.

So if you go and the defense offers 100,000 to settle, then the deal you should make with your lawyer is we'll have a contingent fee on anything above 100,000 that you get.

I would point out, however, it's a good idea, it's going to increase incentives in an appropriate way, but I would point out that this is going to make contingency fees higher, it's actually going to increase contingency fees.  It's going to be on a smaller base but the number's going to be higher.

But perhaps this will convince people to settle cases more often than they otherwise would, once they see how much of the additional take a lawyer would want.  So I think it's not a bad idea.

My main point is the restrictions on contingency fees are unlikely to be a method of tort reform, good method of tort reform, and remember here, basically you're talking about impeding freedom of contract and interfering with freedom of contract often has unintended effects.

In particular, what we have found is that restrictions on contingency fees increase drops and lengthen the time to settlement.

There are lots of problems with the tort system.  You know, I've written papers about the problems of elected judges, written papers about the problems with poor and minority juries, problems with the jury system, controversial stuff, and all kinds of things that we ought to look at.  Particularly, the law is bad.  Okay.

My point--we've paid lawyers using contingent fee for over a 100 years.  Tort crisis dates to 1980's.  Tort reformers really cannot shy away, there's no easy substitute here, they can't shy away from advocating substantive changes in the law, so that's the direction I think we have to go and not focus so much on how we pay the lawyers.

Thanks very much.

MR.          :  Thanks a lot, Alex.

Lester Brickman.  I'm sure you'll have a lot to say.

MR. BRICKMAN:  When I first heard the introduction, I thought I was in the wrong room.  I'm not here to talk about attorney fees and class actions, for sure, though I think it's an interesting subject, but that's another day, and I'm not here to discuss the merits of contingency fees or the abuses.  I've written about these extensively and I can refer you to my writings.  I don't know if there's a bio included.

But I'm here to talk about Alex's paper and I am sorely tempted to respond to a couple of things said but if I do, I really won't get to the paper.  Well, I'll make one exception.

One of the last things Alex said was that one of the measurements he uses was to look at complaints per lawyer and compare them, you know, to the various states on the basis of their legal regimes.

This is somewhat characteristic of the entire paper.  There just are an enormous number of misstatements, because disciplinary agencies, lawyer disciplinary agencies do not take cognizance of fee disputes.  Fee disputes are just something they will not take on.  They tell you "Go sue."  There's virtually no discipline.  I don't say "no," but you know, as a percentage, it's under a tenth of a percent.

The statement that some of the proposals that I'm associated with, like the early offer proposal, will actually increase lawyer contingency fees is true.  However, it will decrease total lawyer take by 10 or $15 billion, which will have a very significant effect on contingency fee practice.  The idea that it would interfere with freedom of contract, the basic premise of lawyer self-regulation is that there is no freedom of contract, that contracts with a lawyer are fiduciary in nature.  I would be happy to take away all lawyer self-regulation and substitute freedom of contract.  That would be a splendid addition.

Let me start with some basic premises.

I guess I plead guilty to disconnecting Alex earlier, but I think it's only fitting since I think there's an enormous disconnect between the model he is using for his paper and the reality of contingency fee practice.

Let me start by saying that all states limit contingency fees in tort cases, and I'm only talking about tort cases.  They only differ in the degree of regulation.

Let me start by saying that all states limit contingency fees to 50 percent maximum.  That was not true a 100 years ago but it is certainly true for the last 50 years.

The reasons given are sort of frivolous and I won't go there.  I have suggested, in a recent article, that but for state limitations on contingency fees to 50 percent, if it were regulated, quote, by the market, the standard contingency fee today would be somewhere in the range of 60 to 70 percent.

Now a number of states more severely limit contingency fees and that is the point of the paper.  There are some generally applicable contingency fee limitations, for example, in New York state.  By court rule, tort case contingency fees are limited to a third, and then there are more specialized limits, particularly for medical malpractice.  Again New York has an even more stringent limit.  California, a number of states, Illinois, have limits on contingency fees in medical malpractice.

So all states limit contingency fees.  It's just a matter of degree.

It strikes me or it seems to me that the premise of the article is that contingency fee caps are a surrogate for hourly rates, and hourly rates induce low, or lower quality litigation, that is, lawyers bring even lower quality cases, they take longer to resolve, and so on.

It is a critical, it is a fundamental premise of this paper that contingency fee caps are a surrogate for hourly rates.

I disagree entirely, and I'll mention that momentarily.  let me just state my own thesis in very broad and very didactic terms.

My research and understanding of contingency fees leads me to very different conclusions about the form and nature of the enterprise.

My overall conclusion is that the volume of contingency fee litigation is a function of effective hourly rates of return of contingency fees lawyers, which I have tried to measure in a variety of ways using surrogates in some recent publications.

That is to say, the higher the effective hourly rate, the greater the quantum of litigation.

The litigation explosion which has been pooh-poohed by a number of people, including some at this table, which I have suggested is real, is a function of the increase in effective hourly rates over the last 40 years which I have determined to be in the range of a thousand to 1400 percent after adjustment for inflation.

Now some other conclusions that I've reached.  Damage caps reduce contingency fee income.  Why do they do that?  Because damage caps, contingency fees are a percentage of damages.  If you have damage caps, they're lower than they would be in the absence of caps, you get less contingency fee income, and therefore, under my conclusion, you get less contingency fee-funded litigation.

And I've described some empirical demonstrations of this.  There's a good deal of literature about the California MIKRA, that's their medical malpractice limitation, and how this results in lower insurance rates.  I have demonstrated that insurance rates and contingency fees are related as functions.

I'm tempted to talk about contingency fees in medicine which would be an even more disastrous public policy than some other proposals; but I can't go there now.

Do contingency fees promote excessive litigation?  Well, I haven't expressed that view.  It strikes me that excessive litigation is simply a political question, which doesn't lessen its importance, it's simply that whether we have too little, too much, or just right Goldie Locks amounts of litigation, are purely a political question.

If effective hourly rates go up, then there will be more litigation.  Is that excessive?  It depends.  Do you like vanilla?  Do you like chocolate?  It's a political issue.

You can make some empirical arguments I think about the economic effects of greater amounts of contingency fee tort litigation and these appear in the literature.  If you want to go there, I recommend the literature.

Now there is a part of the paper that deals with the quality of a case in an attempt to measure the effect of contingency fees on case quality.

Quality is not really defined here, but if by quality, the usual synonym is meritorious and the flip side is frivolous, I would suggest that meritorious is a frivolous term, and of course frivolous, the flip side is--I mean these are terms that are vacuous, they have no empirical reference.  They are terms I generally do not use in my scholarly literature; of course I use them in my political literature because they obviously have a great deal of political content.

At pages 519 and to 520 of the paper, the authors say, "As contingency fees are restricted to a greater degree, the portion of a lawyer's compensation that is due to the contingency fee will fall and the portion that is tied to ours will increase."  This is simply an empirically invalid model.

This is not--contingency fees lawyers, and again I'm only talking about tort cases, do not combine contingency fee and hourly rates.  They do in commercial litigation but not in 99 percent of tort litigation.

So the surrogate that is the fundamental core of the paper, that contingency fees caps are an equivalent of hourly rates, and then you make the comparison between hourly rates and unrestricted contingency fees, falls, because the model is simply invalid.

Page 520 dealing with the unrestricted contingency, lawyers waiving deposition fees, photocopying filing fees, et cetera, et cetera.  Well, maybe somewhere but not on Planet Earth.

Contingency fees lawyers never waive these costs.  Indeed, contingency fee rates have actually increased marginally over the last 30 or 40 years because there has been a significant shift in lawyers' practices.

The majority of lawyers used to apply the contingency fee percentage to the net after expenses were deducted, to the net recovery.

Now for the most part, with exceptions, they apply it to the gross.  So they take their one-third or 40 percent from the top.

When they take it from the net, they share proportionately the expenses with clients.  When they take it from the gross, the client ends up paying a 100 percent of the expenses.

So in New York, by court rule, incidentally, it must be on net.  They still limit it that way.  But in parts of the country, in most of the country where it's unlimited, there's been a shift, and contingency fee lawyers never, you know, waive these expenses.  They always itemize them.  contingency fees lawyers say, oh, gosh, we don't keep hours, you know, we could not possibly charge hourly rates because we do not have recordkeeping facilities, and we don't go there, except for expenses.

Now with expenses, they have a record of every pencil and every piece of paper, because that is deducted, and there are profit margins involved there.

Page 520.  "There is little reason to believe that restrictions on contingency fees will reduce the expected compensation of lawyers."

Well, my thesis would just add one word; not.  That is to say, there is little reason to believe that restrictions on contingency fees will not reduce the expected compensation of lawyers.

The idea, page 520 again, that as contingency fees are restricted as in California and New York for med mal, the lawyer's non-contingency hourly fees increase--again this is simply not an empirically valid model.

What happens when you have fee caps or damage caps, which are just one step removed from fee caps, they're functionally the same, is that lawyers will--contingency fee lawyers screen all cases.  They generally reject at least 50 percent of cases that come into them.  Med mal lawyers, where risks are higher, will reject 70 to 80 percent, on average, of people seeking their services.

Where you have either damage caps or fee caps, what you do is you move the marker, the dividing line or the lawyer between I will take this case and I will not take this case, because of the risk elements.  You simply move it over to lower risk.  If there's lower potential return, the lawyer will take on cases with lower risk.

That is, he will askew cases with higher risk, that he would have taken had the higher returns justified that higher risk.

There is an interesting footnote in the paper, footnote five at page 520.  "We are implicitly assuming that the market for legal services is competitive..." and then it goes on from there.

I have just written an article in which I have said that the market for contingency fee financing of tort claiming services is not competitive.  A summary of that article is contained in the current issue of Regulation magazine published by Cato.  If you have an interest, you can pick up a copy.

I'm not going to go into any details but I have put forth a fairly substantial case.  You can agree or disagree, obviously.  But I have put forth what I consider to be, and I think reasonably can be considered to be the most effective presentation of the subject, at least on the side in which I come out on.

Page 522.  "Plaintiffs who hire lawyers have weaker cases and that will lead to higher settlement rates and lower settlement amounts."

Again, that does not reflect the reality of contingency fee practice.  I've just written an article dealing with effective hourly rates of return and with a focus on automobile accident practice.

The outcomes that are effected by whether or not a client hires a lawyer can be measured.  One of the measures, which is really a measure of litigiousness, because you can take two populations of people involved in automobile accidents and compare them on the basis of whether or not they do or do not hire a lawyer.

If they hire a lawyer, their medical expenses are going to be four times higher than if they don't hire a lawyer.

This is called medical cost buildup because under contingency fee math, for every dollar you spend on a doctor, a dollar goes in the lawyer's pocket, ergo, when you have a contingency fee you have far higher medical expenses.

This has nothing to do with medical needs.  It's purely a function of the contingency fee system.

MR.          :  [Inaudible.].

MR. BRICKMAN:  To somebody who handles their own case.  I'm comparing the two populations.  I'm comparing people with the identical injury in an automobile accident, let's say a weight-bearing bone break.  If you hire a lawyer, your medical expenses will be 4X.  If you don't hire a lawyer, your medical expenses will be X.

A measure of the litigiousness represented by this choice is the BIPD ratio.  I won't go into this in much detail.  If you have an interest, you can follow up in an article I published recently in Washington of St. Louis.  BI, bodily injury, P-D is property damage.  Every automobile accident gives rise to a property damage claim but not every one gives rise to a bodily injury claim.

The ratio of bodily injury to property damage claims is in fact a surrogate as a measure of litigiousness and it varies significant, from as little as, let's say, ten BIs per PD, or per hundred PDs, to 105 BIs to a 100 PDs in various locations in the country, and the principal differences are not geographical as they are the concentration of lawyers.

The higher the concentration of lawyers, the more lawyers as a percentage are hired to represent claimants, the higher the BIPD ratio.

And so the weaker/stronger case model is not the appropriate model here.  Actually, the data that I looked at indicates that if you hire a lawyer--[start side 1B.]

[in progress] lower net return than if you don't hire a lawyer, in all cases except one, that is one type, and that would be soft tissue injuries.  In soft tissue injuries, it is profitable for the claimant to hire a lawyer, but that's a whole 'nother story.

The effect of fee caps at page 524.  Less litigation and therefore lower med mal insurance costs.  That's my proposition that I'm advancing.

Fee caps, reduced litigation.  If you want more litigation as a political matter, you simply raise the standard contingency fee, or you enact fee shifting laws as Congress has done, which has that same effect.

If you want lower med mal insurance costs, you either cap damages or you cap fees because those are surrogates, or their equivalent functions, that reduces the amount of med mal litigation, it will raise the average med mal or the median, probably the average as well, because you are eliminating at the margins the weaker cases that would be taken in the absence of the fee cap.

When you have med mal fee limits, instead of lawyers screening and rejecting 60 to 80 percent, they will reject 70 to 90 percent, depending upon the severity of the limit.

The more they screen the higher the value of the cases they accept.

Now a statement at 537.  Capping non-economic damages reduces settlement time by 35.7 percent.

I have not attempted to look at the underlying data and I'm perfectly willing to accept it, but I would offer a potentially, an alternative explanation.  I don't necessarily say that what I'm about to say is correct.  I'm simply saying it's an alternative that might be explored.

Since fee caps result in lawyers taking less risky cases, in other words, the high-risk cases at the margin are not taken, then relatively weaker cases are weeded out so that cases taken on the average are stronger, and therefore you could expect quicker settlements if they're stronger cases, if you're comparing that to a body of cases in which you include weaker cases as well.

There's also a statement on page 539: When contingency fees are restricted, that's strongly suggestive of a decline in lawyer monitoring, resulting in an increase in low-quality cases.

But here, this depends upon the validity of the model, that when you cap contingency fees, you have an hourly rate model which I simply reject.  That simply is not reflective of the reality of contingency fee tort practice.

In other words, my thesis is, and my writings are to the effect that you have the exact opposite approach.  So not a lot to agree on.

MR.          :  Thank you very much.  Alex, we'll give you a chance to respond but I want to get around to the second half, if that's okay.

MR. TABARROK:  I do have [inaudible.].

MR.          :  Go ahead.

MR. TABARROK:  I will be quick.  Let me just look at the issue of this complaints per lawyer.  The idea of looking at that was not to look about complaints per fees.  It was precisely to test Lester's theory of what he had brought out, that you have all this medical cost build-up, that is, under contingency fees there's an incentive for the lawyer to get the doctor, you know, to boost up the damages, you know, to increase, to say that there are injuries which aren't real injuries.

That's the sort of thing, very precisely Lester's theory, which we would expect to see, to show up in complaints per lawyer.

MR. BRICKMAN:  Why?

MR. TABARROK:  Because it's unethical, it's unethical to have your--for the lawyer to get the doctor to lie about claims or to boost them up.

MR. BRICKMAN:  The doctor--he's not lying--

MR. TABARROK:  Or to boost them up.

MR. BRICKMAN:  He treats you more often because when you have contingency fees you also have lawyers selecting the doctors, statistically.  So under contingency fee math, I didn't also point out that for every extra dollar that goes in the lawyer's pocket you get one in the plaintiff's pocket.  So they have financial incentive.

MR. TABARROK:  I agree with you, I agree with you, so--

MR. BRICKMAN:  So there are not going to be complaints about getting more money.

MR. TABARROK:  No, no, no, no.  As Olson, and as you are saying, there are things lawyers will do when a fortune for themselves is on the line, that they won't do when it's just a fortune for a client.  You'd expect that to show up in complaints, not just from clients, but from the opposing lawyer, from judges, from other people.

MR.          :  The client will never complaint about [inaudible.]

MR. TABARROK:  The client's not going to complain but you expect it from other people.

MR.          :  Why can't it be [Inaudible.] dissatisfaction with the legal system in the United States--

MR.          :  [inaudible] and the Harvard study tells us [inaudible].

MR.          :  But nobody brings them according to that [inaudible].

MR.          :  Well, but in terms of the example that you just described, the client who would bring a disciplinary complaint is never going to complain.  It's the doctor who's saying come back, come back, come back.  It's an understanding of the system and it does create that [inaudible].

MR. TABARROK:  So I agree with the whole thing, I'm just arguing that that's the sort of thing you would expect to show up in complaints for a lawyer.  But if you want to propose another test, I'm willing to do that.  But I don't want to get on this too much.

The second point is about hourly versus contingent fee.  I could just be very brief here.

Look.  Suppose you have a sales person who is on commission.  Okay.  And then there's a law which says no more commission, sales people cannot be on commission.  What is going to happen to their wages?  Their wages are going to go up.

Suppose you have--look about waiters, waitresses.  Suppose you say no tips.  We're going to have a legal rule which says no tips.  A tip, right, is a percentage of the meal; it's like a contingency fee.  Okay.  You get a tip, it's based upon a percentage of the meal, and you only get it if you do good service.  It's exactly a contingent fee.  A tip is a contingent fee.

Suppose you were to say for waiters, no contingent fees, no tips.  What is going to happen to their wages?  It's going to go up; it's going to go up.  Exactly the same thing with lawyers.  You can't control the total compensation of lawyers by controlling how they get paid.  All you're going to do when you control how they get paid is have them get paid in some other way.

MR.          :  I do want to pursue that because it did strike me as a significant point.

When you say the drop rate goes up, right? is that in fact, do you think--I mean, is that contingent on--is that because you think they switch to hourly compensation forms?  I mean, my knowledge of the legal profession is these people who work for contingency fees will not go to hourly compensation; end of debate.

MR.          :  That's not true.

MR.          :  Do we know anything about that?

MR.          :  Yes.

MR. TABARROK:  Go ahead.  I was going to call on you.

MR.          :  [Off-mike].  [Static on line.]

MR. TABARROK:  And if you restrict them going on contingent fee, then of course they're going to look for other ways to earn their money, to earn an income.

MR.          :  Right. But presumably there is an anger to that; right?  Because I mean, for example, you could--what are the actual--could you find out what the actual arrangements in the states that you studied were and what the actual arrangements--I mean, in the short time window that you have--

MR. TABARROK:  We don't have that data.  I mean, that would be a good thing to look at.  But I don't think I'm saying anything controversial here.   I mean, we know, for example, AEI's done lots of work here, why do we have health insurance bundled with jobs; right?  And this goes back to World War II.  You had a freeze on wages, but firms wanted to increase compensation so they did it another way.  This is just exactly the same idea.

MR.          :  Contingency fee caps are not a freeze on wages; they're a freeze on percentages that you can apply to damages [inaudible].  The aggregate effect of contingency fee caps or damage caps is to reduce aggregate lawyer income, unless they can enlarge the envelope, which lawyers are very good at, in terms of more, inculcating more areas for litigation.

MR.          :  Jon Klick.

MR. KLICK:  I too am presenting a paper that's joint with Eric Talley.  If you weren't sort of out in LA, it would have been more efficient to have him come and fight Lester and Ted on his own.

MR.          :  Oh, it wouldn't have been half as good.

MR. KLICK:  That's true.  He's not as exciting a guy as we are.  But our paper, as Michael said, is in effect two different papers.

One is sort of a substantive addition, hopefully, to what we know, and the other is maybe not quite a critique but a disagreement with some analysis that Ted did, because we use his same data.  But in order to get to where we wanted to go, we had to make some different choices than he did.

So what's the main question that we're looking at?  How do judges and how do the incentives that judges face affect their willingness to sign off on attorney's fees in settlements, in class action settlements specifically?

So in order to start this question, we try to figure out what are judges' incentives, and we go back to an article in '93 by Judge Posner, who as best we can tell, is the first one to actually systematically investigate this question.

And so he asked the question, What do judges maximize? and his sort of Posnerian answer was, "The same thing everyone else does."  They maximize their welfare.

The only difference for judges is that most of what they do, and most of their opportunities, don't have to do with increasing their financial incomes.  They're pretty much restricted in increasing their financial incomes and their income opportunities in the future.  So they will try to maximize along other margins; okay.  And so Posner's basic point is that judges will choose a relatively high level of leisure, conditioned on maintaining their level of prestige and their reputation, and conditional on not getting reversed too often.

And so the question is, How do judges increase their leisure?  Well, Posner doesn't sort of answer this completely.  He suggests one possibility.  He says, well, judges might just not put all that much effort into their cases and they'll let their caseload grow, the case queue grow.

The problem is, and he admits this, this isn't a perfect solution because they can't simply let their case queue grow indefinitely because if a given court basically has a humongous backlog, there may be incentives for the legislatures to add more judgeships, which will diminish the prestige of existing judges.

And so what Eric and I thought of, and there were some people who have implied this in the literature already, but we thought of this alternate strategy.  What judges can do is they can seek to resolve cases quickly with little effort.  Expediency, in many cases, however, will lower their reputation or increase their reversal rates, but class action settlements are attractive along this dimension since the judge can clear a complex case from his docket and he's unlikely to face sort of vocal challenges.

And so the implication of Posner's argument might be, is that judges have this incentive to facilitate settlements in class action cases.

Now what does that have to do with attorney's fees?  Well, given that one of the main things judges are supposed to do in sort of policing class action settlements is they're supposed to police the attorney fee.  However, if the reject the proposed attorney's fee, it's going to slow down the settlement process; right?  It's going to cut against their incentive to gain more leisure.

Now if they sign off on particularly egregious fee arrangements, it's not quite clear that this is going to substantially hurt their reputation, because presumably the defendant in these cases doesn't care; right?  They could care less whether they're paying money to the lawyers or they're paying money to the plaintiffs, the plaintiff class, and presumably the plaintiff class is too defuse to monitor this.

I mean, there are objectors in some cases but, by and large, people don't object.  I mean, one of the reasons that we ostensibly have the judge police settlements is because the class is too defuse itself to monitor what the lawyers are doing.

And so an implication of this, an empirical implication of this, attorney's fees, we might expect to rise as court congestion rises; right?  And so it shouldn't come as a surprise when the conventional wisdom tells us that there's an upward trend to both court condition and attorney's fees.

In fact, sort of an implication of Posner's argument is that there's a causal relationship between this, between these two things.  But wait, here's where Ted comes in; right?  Earlier this year, Professor Eisenberg and Professor Miller had an article that got quite a bit of attention, in fact it was an 1800-word article on the front page of the business section of the New York Times, with his picture, before the paper had even been published, under the headline, Study Disputes View of Costly Surge in Class Action Suits, and in Ted and Professor Miller's paper, they basically, among lots of other things, they basically say, well, the hypothesis that attorney's fees are increasing, over time, finds very little support in our data.

Now to be fair, we had Ted on one of these panels a few months ago, and he made the point, and I believe him, I don't think he's lying--he claims that the trend result really wasn't central to their findings.  They hung their hat more on other findings.  This was just something interesting that came about.  Unfortunately, it's the result that sort of got all the attention in the popular press.

And so the first step in what we wanted to do to get to our ultimate goal of testing this congestion effect is we wanted to sort of use Ted's data but sort out some things that we saw as maybe problems, or if not quite problems, differences in judgments.

And our thought is that in many ways Ted and Jeff's result of no trend is questionable, for a couple of reasons.  One is they examine both state court and federal court cases but their process of building the data set was to look at Westlaw and Lexus-Nexus, and unfortunately, there's quite incomplete coverage of state trial court cases in both of these data sets.

And so if you look into it a bit, it appears as though the states that we think of as the judicial "hell holes" actually show up less often in these data sets than do some other states.

And so with the data set, you might be examining, you know, the states that have relatively few problems, at least the tort reformer's perspective.

Another thing that we find a bit problematic in the way Ted and Jeff analyze their data is that they're starting off with a relatively small data set, only 370 cases.  Now there's nothing you can do about that; right?  I mean, the data sets are small and in these litigation studies they almost are always small.

But a number of choices that Ted and Jeff make effectively make the data sets smaller.  So instead of analyzing the whole 370 cases at once, they partition these things.  So they look at fee-shifting cases separately from non-fee-shifting cases.  They look at cases with Lodestar, reported Lodestar amounts separately from cases that don't have reported Lodestar amounts.

So basically they're not even dealing with a data set of 370 cases.  They're dealing with substantially smaller data sets.  Ideally, what you would like to do is analyze all the data together and then sort of control econometrically for the differences that you're worried about between fee-shifting cases and non-fee-shifting cases, and so on.

Another thing that's particularly problematic with respect to the trend result is that the data set is unbalanced.  So some years in the data set have relatively few cases, say thirty, whereas other years have 70 or 80 cases.  It turns out that for some technical reasons, this is going to make it a bit problematic to get a good idea of what's going on in sort of temporal time trends.

Another problem potentially is when Ted and Jeff look at fees as a percentage of total client recovery.  Instead of just looking at the percentage directly, they take the square root of the percentage, which is fine for forecasting purposes, which I think is implicit in what Ted and Jeff were hoping to do.

But for drawing inferences about what's really doing on causally, perhaps, it's problematic because it dampens variation in your data.

And then lastly they include some subjective risk variables.  Basically if the judge mentioned that this case seems to be high risk, they controlled through that.  If he mentioned that it's low risk, they controlled through that.  Unfortunately, if you want to draw sort of casual inferences, this is problematic.

If I'm a judge and I want to provide cover for giving an awfully high fee amount, I simply call it high risk; right?  There's no sort of objective standard here for controlling for these things, and it turns out that the results are at least somewhat sensitive to the inclusion of these variables.

So what happens when we look at Ted and Jeff's data, not looking at the congestion effect yet, but when you just look at their data and try to redo more or less what they were doing, at least with respect to the time trend, making different judgments here.

Well, we find that there is a positive trend coefficient in all the specifications and in the regressions that look at the fee as a percent of total recovery, it's actually statistically significant and it's, you know, practically significant.  You know, through the course of their data set, it turns out to be something like a 5 percent increase over time.

Now, again, as I said, ideally, we would like to do some fancier things with the econometrics and so if we do these sort of interaction models that control for everything we would like to, under some choices we still find statistical significance.  Other choices, it disappears.

And so the question is, well, what is correct?  Well, if we had 10,000 data points, okay, if we had a huge data set, we could do, you know, we could control for everything that we want to, but given that we have a relatively small data set, you've got to make some tradeoffs and our basic position is that in the tradeoffs that Ted and Jeff made, they were basically, you know, "rigging the game" so as not to find a trend.

But that may not be problematic given that that wasn't what they were particularly looking to examine.  The problem comes in what third parties have sort of taken the paper and said about it, beyond that.

So now anyway, to the sort of substantive part and sort of the more interesting part.  After we sort of remedy or make different judgments on those econometric issues, we merge in Ted and Jeff's data with some data from the Federal Judicial Center on caseload data for each district, basically congestion data.

And so the raw data basically suggests a quite small but at least visible and upward positive relationship between congestion and attorney's fees.  But if we try to do things a bit more rigorously with some regressions, what do we find?

Well, we find that in the regressions where we look at the level of the attorney fee, we find that an increase in congestion of one percent basically leads to an increase in attorney's fees by point one percent, and this coefficient is statistically significant.

For the percentage fee regressions, we find the coefficient that basically implies that if a given court has an increase in one standard deviation of the congestion measure, we find that attorney's fees rise by basically point eight, and this has a relative effect of basically three percent.

So if a given court has more congestion than another court, on the order of one standard deviation or congestion, we can basically see the relative of change in attorney fees go up by 3 percent, and this is statistically significant as well.

Again, if we're interested in what the trend coefficient does, it drops slightly but it remains statistically significant in percent fee regressions, but again that part might change depending on what choices you make on your specifications.  The congestion effect doesn't change.  The congestion effect remains, almost regardless of what choices you make econometrically.

Now there is something peculiar with this data, at least with respect to this question.  Again, since we're dealing with a small data set, it turns out that outliers matter and, in particular, there's a case in Ted's data from the northern district of Alabama in 1999, and this case isn't all that interesting per se, but it turns out that the northern district of Alabama terminated about 19,000 breast implant cases in 1999.  So this is really doing screwy things with the congestion variable for this court.

Basically in that year, for the northern district of Alabama, terminations per judge were over three thousand.  So this is "off the charts" compared to other districts in the data set.  It's even off the charts for this district.

In the year before, this district only saw 843 terminations.  The year after, it only saw 700 terminations.  So what happens if we throw out this case basically?  Again, if we had a huge data set, we wouldn't want to do that.  We wouldn't think that one case would matter, particularly.  But in a small data set, it may matter.

So what happens?  Well, we still get a positive congestion effect but all the coefficients jump up significantly.  We basically see coefficients tripling.  So what does that mean?   Well, if we throw out this case, we find that a one percent increase in congestion leads to a .15 percent increase in fees.

In the percentage fee regressions, we find that an increase in congestion of one standard deviation leads to a relative increase in attorney's fees of 6 percent, and this is statistically significant again.

Again, the trend, the time trend again is sensitive to specification, but it's always positive, at least in the coefficient.

So what are our conclusions?  Well, there is good evidence, or at least from this data set it is consistent, the analysis is consistent with this Posnerian theory of judicial expediency; right?  Judges do seem to sign off on higher attorney's fees when they're busier, when their court is busier, when it's more congested.

We're dealing with incomplete data here, so we can't do everything that we'd like to do to sort of nail down a causal effect here but it's at least consistent with this theory.  Now the question might be is this really what's driving any apparent trend?

Well, it appears from the data that these two phenomena are independent.  There is a congestion effect but there's also sort of an independent time trend effect.

So where do we go from here?  Well, absent better data sets, we're just hoping to sort of go forth and see if we can replicate these basic results in existing sort of incomplete data sets, to find out if it's really just an artifact of Ted's data or some of our choices.  Thanks.

MR.     :  Thank you, Jonathan.

Ted.

MR. EISENBERG:  Thank you.  It's a pleasure to be back here, and two larger lessons to come out of Jon's somewhat disagreement with us, that I think have an actual hopeful sign.  One is he asked for the data and we gave it to him, and he could do his own independent analysis of it, however incorrect, and, you know, we can have an honest--you know, you get better analysis if more people have access to the data.

The other thing to his credit is he sent me the paper before he published it, and so that we could react in a way, you know, that might actually help him and lead to a better product.  I think those two lessons are good and people should do more of that.

There's really two papers going on here.  One is a critique of us and one is the congestion point as an explanation.  So let me start with the substantive part of your paper that doesn't threaten me; just some suggestions.

One is, and I honestly don't know what you've done exactly, so part of this is question and part of this is recommendation.  It's critical to your paper to measure congestion, to have a decent measure of congestion.

As I quickly read it, it's the number of cases terminated divided by the number of judgeships, which is certainly a plausible measure, but a question would be, Does that include the criminal docket?  And it does; just add them up?

MR. KLICK:  The main result is basically robust, too, if we exclude or include, but these results are including--

MR. EISENBERG:  Okay.  So that was important; that's good.  The other question is, Does the administrative office of the Federal Judicial Center periodically publish, deals with weighted caseloads?, that is, not all cases are the same, especially if you're a federal court in an area with a state prison, you will get a huge number of state habeas corpus cases which will substantially increase your caseload, but they're sort of low case, low work cases, and precisely because if things like--or you might get one antitrust case that is a huge case.  Obviously the simple division of number of cases terminated by number of judgeships is respectable but I think the state of the art could improve it by somehow taking into account weighted caseloads.

That could be done in one of two ways.  Actually, either look at the caseload factors I think the FJC or AO makes available, and so weight your data in accordance with that system.

The other would be to put in the model some sort of variables for case categories.  After all, the weighting system is really a function of case category.  So that if you put in the model, you know, case categories, and I forget what case categories are in here, securities or tort or whatever, you might get a more precise measure of congestion that accounts for caseload rather than just the pure number of cases terminated.

A third thing I think to look to would be stage of disposition.  Now in our data they're all settled, so maybe that doesn't matter.  You're just working our data on this?   Yeah.  So that may not be so good.  But, in general, if you're looking at congestion, you know, trials are different than, early dismissals are different than summary judgments, and it would be great to know what stage we're talking about.

So that sort of measures the congestion.  I think we can improve it slightly.

The second question is a modeling issue, and I haven't thought it through, but it seems to me what you've got basically is fee as a function of, in your case, two critical variables and a bunch of others, congestion, and year.  I'm not sure the year one's so important to you actually.  But fee as a function of congestion.

I don't think congestion is exogenous, that is, congestion should be modeled as a function of, say, case type or fee or year.  That is, it's always nasty, but if you can solve them simultaneously, I think it's a more persuasive model, because certainly congestion's not exogenous to all [inaudible-static on line].  It's got to be tied up with it.

We know, for example, an article I did with Claremont years ago, that people respond to the queue for trials.  That is, if you look at the choice between judge trial and jury trial, people actually know how long it'll take to get one and they might actually choose one over the other based on, you know, the length of the queue.

And so that type of analysis I think would be useful to you to at least thing about.  The horrible thing about it is you have to solve simultaneous equations and they don't behave, and all that stuff, but it's at least worth exploring.

In the slightly bigger picture, what you're doing, the importance of congestion, workloads on courts, resonates with a huge project the ABA is behind this year and that's the vanishing trial.

That is, Mark Ellender [ph] has a paper, it's coming out, we have a big issue of the Journal of Empirical Legal Studies in November dedicated completely to that topic, and one of the things people worry about there is why is the trial disappearing, and one of the basic worries is, you know, court congestion and the like.

And so Sheri Diamond in that issue has a paper modeling factors that might be leading to the [audio static] disappearance of trial, and you might get some benefit out of that because she's thinking through some of the things you are.  Okay.  Can I do this?

[No audio for approximately a minute.]

If combining them fundamentally changes the result from insignificant to significant, you've got to worry.  How much have I accounted for the heterogeneity of the two data sets I've just combined?  And I don't think it's a weakness on our part that we refuse to combine them.

In fact it seemed like the right thing to do, because they are so different in terms of the law that governs, the percent of fees that winds up being awarded and the like.

The other thing I might say is our results, we analyzed not one but two data sets, one was ours, all published opinions we can get our hands on.  The other is the class action reports data, which is actually larger, it's over 600 cases.  I just peeked at our results.  And using the class action data, we get exactly the same nonsignificant time effect.

So to make ours go away, I think--I can send you that data too, but you've got to make it significant in both data sets to at least do the equivalent of what we've done, because we worried about our data and we tried to get ahold of, I think the next--well, maybe it's better in some ways, weaker in others, but we found no effect in either data set, and my bottom line recommendations is, before I believe your result I need a simple picture or a simple table, and if you can't give one, you need to explain to the reader why, and if you can't, I probably won't believe you.  Thank you.

MR. KLICK:  Just a couple of comments.  It's not exactly clear that your CAR regressions and analysis are exactly equivalent to what you had before, because in the CAR you can't control for a number of the things that you did.  So that's still a bit of an empirical question.  But you're right--the challenge is for us now to replicate what we've done in other data sets.  So no doubt you're right.

Now in terms of refusing to combine sort of disparate case types, that may well be a reasonable principal to stand with.  The problem then is, though, you start running regressions on samples of 60 cases, right? and you get--

MR. EISENBERG:  [inaudible] cases only add a very small amount to the total.

MR. KLICK:  No; no.  But you end up--you differ--you separate out not only fee shifting versus non fee shifting.  You also effectively separate out cases that have a reported Lodestar amount with cases that don't.  And again we're already starting out with what's a small n.  It's a small n, sort of regardless, even if you don't parse.  Parsing only sort of makes it a bit worse.

And so the question is, well, do you do regressions and recognize that it's small n's and so it's not entirely clear that not being able to reject the null really does say that there's nothing here, or do you say we just don't have enough data to run regressions, and sort of make any causal inferences here or any sort of correlation inferences here, and have very much confidence in them.

That's no doubt a question, you know, it's a judgment call.  In terms of whether a result can show up in a graph or not, you know, the reason that we use regressions in economic is because life isn't two-dimensional; right?  It's just not.  And there plenty of instances--I can give you plenty of instances where a result looks clear on a graph and once you start controlling for things, that result disappears, it reverses.  Sometimes it stays the same; no doubt.

And you can do it, I mean, you can do it with real data, you can do it with synthetic data.  I'm not so sure that sort of hanging a hat of trying to explain the world, hanging your hat on a two-dimensional picture really creates this line in the sand that you're trying to make it create.

But beyond that, again, it basically does come down to judgment calls on our differences with respect to you.

MR. EISENBERG:  I would say if you can't reduce it to the simple story, you need to be able to explain why, because you can go variable by variable and split up the data into the--you know, we spilt it up, you combined it.  I say when you disaggregate it your effects go away, and so I mean, one, I agree that there are some results that are plausible, that only show up in regression.  But there are not that many.

You'll find most of the good articles give a nice simple story, up front, and then confirm it with regression analysis.

MR. KLICK:  That may be true.  I can give you a good story on why it doesn't show up.  Again, you can do this at home if you want to try it with a spreadsheet or something like that.  Try to set up a trend where in early years you've got five, you know, five observations, and in later years you've got a 100 observations; right?

MR. EISENBERG:  Fine.  That may be the explanation, and do it with the data.

MR. KLICK:  Fair enough; fair enough.

MR.          :  I just have one question which is prompted by a comment that Ted made, and I saw you nod, and it was a question I had, which is I just want to sharpen this, which is the possible endogeneity there with congestion; right?  Just on a gut level, put the econometrics to one side--is it more likely that congestion causes the fees to go up?  Or is the other inference more likely, with respect to a certain kind of case, that is to say, whether lawyers have a lot of control over where to file it; right?  Which is--

MR.          :  [inaudible].

MR.          :  Until there's a queue effect, and the effect that Ted mentioned kicks in and you want the case resolved.

MR. KLICK:  Yeah.  Clearly, in any sort just sort of straight, cross-sectional correlation here, you're always going to worry about endogeneity and this is a very plausible story of endogeneity, so maybe the congestion effect is the sort of reverse causality, and without sort of probably bigger data sets, we're kind of limited in how well we can control for that through systems of equations, through natural experiments, and things like that.

One sort of thing that might give us pause in immediately jumping to the endogeneity problem is that even in the worst courts, these class action cases make up sort of a vanishingly small percentage of total terminations; right.  Most of these courts are swamped with sort of criminal cases, right? and sort of non class action type cases.

Now does that defeat the endogeneity argument?  No.  But it sort of attenutates its effects and plausibility.

MR.          :  [inaudible--off microphone source].

MR. KLICK:  Sure.  We basically take the FJC's definitions, you know, so basically, you know, the case sort of leaves the docket of the district.

MR.          :  [inaudible].

MR. KLICK:  That's right.

MR.          :  [inaudible].

[Audio fades out for approximately a minute or two.]

MR.          :  Where do plaintiffs learn that tactic?

MR.          :  Pardon me?

MR.          :  Where do plaintiffs learn that tactic of imposing costs on the other side?

MR.          :  [Off-mike.]

[Speaking off-mike continues.]

MR.          :  If you'll yield for one quick comment.  The chicken and egg question that you raised, you know, is congestion the reason for the higher fees or is it that the higher fees attracts congestion.  The term used in the literature is the "magnet court," and I just note that the leading magnet courts, such as that in Madison County, Illinois, state courts, Jefferson County, Texas, Palm Beach County, Florida, are unrepresented in Ted's data.  There are no cases, no state court cases in that data base.  So on a per capita basis, you know, Madison County has more class actions than the rest of the world combined.  They have the highest per capita--

MR.          :  How many certified?

MR.          :  Certify as opposed to settlement class actions?  I think they're all certified.  They're certified in one form or another.

MR.          :  I don't think you know what you're talking about.

MR.          :  All right.  Well, I think all of the class actions filed in Madison County are certified in the form that either certified in settlements--

MR.          :  Even Beisner [ph] hasn't claimed that.

MR.          :  Well, if they're dismissed--

MR.          :  They just lost one, I think.  Go ahead.

MR.          :  Do you acknowledge that your data base does not have these cases?

MR.          :  Of course.  But I also don't acknowledge that anyone has solid evidence about the behavior of Madison County.

MR.          :  I think American corporations that have paid out billions of dollars have some pretty solid evidence.

MR.          :  I think they should have paid a few million to study the outcome of cases in the place.  Do you know the medical malpractice win rate in Madison County?

MR.          :  Is it different than the rest of the country?  I would assume it's the same, 20, 30 percent.

MR.          :  Then why is it a "hell hole"?

MR.          :  Because of class actions and because of asbestos litigation.

MR.          :  Well, asbestos I can understand.

MR.          :  [inaudible].

MR.          :  Okay.  Yeah.  Asbestos maybe.  I have one comment, just going back to medical.  Harvard, the malpractice study says most victims don't sue.

MR.          :  90 percent.

MR.          :  You say 70-80 percent, or 60 or 80 percent of the lawyers won't take a medical--I mean 60 [inaudible].  Don't we need a system to have more medical law suits?

MR.          :  Well, you need different kinds of cases [off-mike].

MR.          :  [Off-mike--inaudible] is recoverable.

MR.          :  Yeah; okay.

MR.          :  [off-mike comment.] not just the Harvard study.  I think the Harvard study--

MR.          :  I think it's consistent.  People don't sue.  Injured people don't sue.

MR.          :  [Off-mike.]  Yeah.  Old people with low recovery, low dollar recovery potential are not going to get their suit brought.  But the point that it makes [inaudible] is that the client has very little to do with contingent fee litigation, and I think you have to start understanding that to see that key decisions are made by lawyers who are financing the cases, who are controlling the caseload and who, in a sense, are at great risk in one way, but they're not at the level of risk that they ought to be because their opponent's costs and transaction costs are never recoverable.

MR.          :  I mean, I think in class actions that's almost certainly right, that the lawyers run it.  I think in individual medical malpractice cases, a lot of the lawyers put on a "dog and pony" show to try and win the case, and the client actually chooses.

MR.          :  [Off-mike.]  I think that that's right, but, again, I don't want to steal your thunder--but again there's a kind of fungibility about that, the fungibility of the lawyers who take those cases [inaudible] they're well-equipped enough, experienced [inaudible] enough to make the same evaluations about a given case, and so the competition is to get the best guy you can get [inaudible] take that kind of case.

MR.          :  Yeah.  But there's competition.

MR.          :  I agree that there's competition with that.

[Simultaneous conversation.]

MR.          :  It just begs the question of--that's right--because there is no competition [inaudible] the contingent fee is [inaudible] had nothing to do with risk.

MR.          :  Probably true.  It is an indictment, I think, of the tort system in this country, that if somewhere a surgeon makes an egregious error and removes the wrong limb or operates on the wrong side of the brain, somewhere, a lawyer becomes a millionaire as a consequence.

MR.          :  I mean, don't you have to explore the alternative?  What if the lawyer doesn't become a millionaire and the person gets no relief?

MR.          :  [inaudible] not the alternative.

MR.          :  Oh.  What system's in place to fix that?

MR.          :  Well, early offer is a system that will fix that right away, and that's exactly what's wrong with orally offer, that it will fix it and it will reduce compensation to lawyers by tens of billions of dollars.

MR.          :  Can I just ask a question, I mean, that seems to me to hang over this but I didn't fully understand it.  This is about Jon's data.  To what extent do you think that congestion should produce higher fees?  Or should it produce higher fees only in cases where there's some real asymmetry between the lawyers and the plaintiffs? because why else should we think that just the fee should make settlement more likely?

MR. KLICK:  I think that's right.  I think, you know, we're only looking at these class action settlement cases which are, you know, in sort of the Posner model, are particularly attractive; right?

You can clear the case, nobody's going to bitch about it, you're done; case is gone; right?  Sort of beyond that, I don't know, you'd have to sort of analyze different constraints and find out which cases are more or less likely to create, you know, people being angered or something like that.

MR.          :  We don't want to monopolize--

[Start tape number two.]

[in progress].

MR. KLICK:  Quickly, we find, almost
regardless of specification, this positive correlation between court congestion and attorney's fees in these class action settlements, and it actually turns out to be, they're not only statistically significant but sort of practically and importantly significant.

MR.          :  [Off-mike.]

MR.          :  Yeah, and it's an approximation.  I mean, the whole basis of lawyer self-regulation is that the lawyer-client contract is not an arm's length contract.  It's a contract, it's a fiduciary contract, and, you know, the concept of reasonable fees, for example, is a diametric opposite, in some sense, of arm's length market bargaining.

It is this self-regulatory nature that lawyers resort to in order to avoid the kind of societal regulations that they impose on others.

I would be very happy--and then of course they use that self-regulatory control to insulate themselves from Sherman Act and otherwise applicable doctrinal regimes.

I would be very happy to strip the self-regulatory power of the bar in exchange for an arm's length relationship.  Let lawyers bargain for fees in the absence of the ethical rules that they impose, that preclude price competition in contingency fees.

I mean, they prohibit the very market developments that would occur but for these regulations in order to preclude or at least impede, but actually preclude.

MR.          :  Are you saying a lawyer can't charge a 20 percent fee?

MR.          :  He can't?  No, I'm not saying that.  I'm saying he can't prefund litigation.  He cannot pay the client in advance for expenses.  He can't compete with other lawyers on the basis of I will advance you litigation costs.  There are a variety of ways in which he--you cannot have for-profit referral services, to that precludes lawyer brokerages in the same sense that you have mortgage brokerages.  I mean, there are--

MR.          :  [Off-mike].

MR.          :  Well, I'd be much happier to have a regime in which we had a market economy for lawyers, yes, because right now, with lawyer control, you have ethical provisions enacted by the bar, and enforced, which preclude price competition in contingency fees.

The reason you don't have a 20 percent contingency fee is because of these ethical rules, not that they prohibit it, but they prohibit the kind of bargaining would take place if the market were unimpeded by these regulations.

MR.          :  I think it's just descriptively inaccurate.  We have variation in the percent of fees.

MR.          :  I disagree with you on the facts.

MR.          :  All right.

MR.          :  And I think the empirical data is that the standard one-third fee applies.

MR.          :  It certainly doesn't apply here.

MR.          :  No; it doesn't.  I'm not talking--the idea that one-third should apply, that the standard contingency fee in a tort case should apply to a class action is a straw man which I think you effectively destroyed.

MR.          :  Can I just ask something.  One of the things that struck me in reading Lester's article is--

MR.          :  Which one?

MR.          :  The asbestos study.  The theoretical approach strikes me as very persuasive, that is, to look at law firms as firms, that stitch together portfolios.  There is, to my knowledge, very, very little attention to that kind of a question in the literature.  There is an endless literature in class actions on the alignment of incentives and how one should do this, and what to do with [inaudible] and all the rest of it.

There's very little on the industry itself and its industry structure, and how these places behave, and it's a very, very different world from what it was 20 years ago, and the structure itself, my just impression is that this market has, consistent with what Alex said, become bifurcated in a very strange way, there are a lot of big players now that are really ongoing legal enterprises.  There are a lot of--

MR.          :  [Off-mike].

MR.          :  Right.

MR.          :  [Off-mike].

MR.          :  Right.  And then there are a lot of, you know, little players whose portfolios and risk profile, and so forth, looks very, very different, who don't have access to venture capital which the big firms already have.

All of this is extremely poorly understood, it seems to me, and once you have this model of the legal system as sort of an arena of entrepreneurship, you had better damn well make sure that the firms are understood.

So my question is, Is there anything in that, I mean, in that sort of venue [inaudible]?

MR.          :  It's a completely correct topic, we should be pursuing.  You know, I've dealt with some of the plaintiff firms, and my wife's a writer, and I keep thinking that she should go write a book about how they do this cause it's fascinating.

I don't think there are--I mean, I think it's an evolving thing.  They get together in to a quote, firm, for a case, and they disaggregate after, and it's very personal, they sort of know--you know, the big players know each other, the same guy's showing up in the--

MR.          :  [Off-mike].

MR.          :  There's not a whole lot of--you know--and we're talking about a very small piece of the legal system in some sense, the very top end of the plaintiffs bar, which is what you read about.  But it's absolutely fascinating.  They fight like hell.

MR.          :  [Off-mike].

MR.          :  In the tobacco ca--yeah, you'll see--

MR.          :  [Off-mike] nothing I think you can say that scientifically, is the case.  There's a tremendous amount [inaudible] how those firms [inaudible] how they got the cases, how [inaudible], how they dealt with [inaudible], how they dealt with [inaudible] like the fact that Dickie Scruggs, a major contributor to Democrats [inaudible], you know, and how those sorts of things play out, and why the Congress won't make significant reforms in that area [inaudible].  There are a lot of [inaudible] one thing that you might look at in terms of [inaudible] lost time.

[inaudible] 30 years ago, when any federal judge had [inaudible] go in on a Friday [inaudible] dozens of cases [inaudible]to the point where people could assess ultimate probability [inaudible] transaction costs.  That's when a case settles.  That doesn't exist [inaudible].  Judges who are looking [inaudible].  I think that's worth your study and I don't know [inaudible].  Of course that's a federal court phenomenon although there are [inaudible].

MR. KLICK:  Yeah, that's actually a great idea, and yeah, this research is sort of at a fairly early stage, that we can certainly do things like that.  Regarding your earlier comment about terminations being sort of maybe not the best congestion measure, the results are basically robust to basically any of the congestion measures the FJC provides now.

Again, they're probably all, admittedly, imperfect but it's at least enheartening that they sort of all go in the same direction, but this last suggestion's really quite good.  Thank you.

MR.          :  One anecdote on the tobacco that plugs into the fee arrangements, at least in some states, when it was laughable to sue the tobacco industry, and it was laughable for many years.  It was a hopeless case.  Some of the plaintiffs' lawyers went door to door to the defense firms and said, Will you do this work? and they all said no.

The risk was solely on the people who were willing to take it, and the sort of hourly firms wouldn't touch it.

MR.          :  [Off-mike].

MR.          :  Oh, absolutely.

MR.          :  [Off-mike].

MR.          :  Someone should write a book about how these guys get together, separate.  I mean, it's really--you know, I mean, since it's helping shape public policy--

MR.          :  [Off-mike].

MR. GREVE:  We'll end on that cheerful and harmonious note.  I thank all of you for attending and I thank our speakers and please join me in thanking all of them.