Alex,

First of all, thank you for your gracious response to my last posting. This type of give-and-take is exactly what we hope for in our Point of Law featured discussions. And I must say I'm flattered by your reaction to my thoughts -- not only your words of praise but also you actions, i.e., that an economist of your stature would actually run a new regression at 2:30 in the morning to generate an immediate response.

Since we do have two competing theories, the answer is ultimately empirical. And I don't know that we'll be able to come to a clear answer this week; I expect that this discussion could have some useful offshoots going forward. I'm pulling together this response quickly myself at 6:30 in the morning -- before heading to the airport, and I'm not generally a morning person -- so don't hold me to this snapshot reaction as if I'd given the matter extended thought; but here's my immediate take on your new regression results.

I agree with you that your theory does match the assumption that after a contingency fee cap is imposed, expected awards would decrease -- given that you postulate that plaintiffs' lawyers switch to hourly fees and exploit their clients by milking them through more bad cases being filed and more billing, i.e., time spent, per case.

I'm not sure though that in my view of the world, I'd expect average awards over all cases to increase. For any given case picked up, I'd expect that to be true after the drop decision, holding attorney behavior and other legal rules constant. The first clause there -- "for any given case picked up" -- is important: because although the screening mechanisms for lawyers pre-discovery are crude (the very foundation of my theory vis-a-vis the drop results), they're not non-existent. Lawyers, or at least good lawyers, certainly know on average how likely various types of cases are to win or lose, and how big an award they're likely to generate, based on certain characteristics. Although you pick up a lot of these characteristics as independent variables in your Florida regression (i.e., the basic severity of injury, the type of practice, and the type of defendant), there are undoubtedly factors missing. And in any event, there are significant factors to consider for the Florida data that could be affecting those results.

Crucially, as you and Eric notice, there was a "rush to file" in Florida prior to the November 1985 filing deadline, before the contingency fee caps went into place. Florida lawyers certainly seemed to think that the fee caps made a difference. Given the structure of the fee caps in Florida, which only kick in at $1 million (more on that later), it's very likely that these cases rushed to the courthouse were of the really-high-dollar variety. Today, those would be infant birth defect cases and the like; I'm not sure what they be in Florida in 1985, but it wouldn't surprise me at all that attorneys' rush to file such cases before the fee caps kicked in would overwhelm the discovery screening effects.

Perhaps even more important than this rush to the courthouse effect, it's important to consider the other "confounding problem" you and Eric note about the Florida data:

Florida enacted another series of reforms in 1986, the first of which took effect in 1987. These reforms included limiting contingency fees in nonmedical malpractice cases, eliminating joint and several liability, capping non-economic damages, instituting a period payment schedule for large awards, implementing a collateral sources offset rule which reduces the amount of the award by payments from other sources, and capping punitive damages.

19 J.L. Econ. & Org. at 531-32. Given that damage caps certainly could effect award outcomes, these changes might be a significant explanatory factor for the fall in awards, assuming that the damage caps kicking in starting in 1987 would affect a larger subset of the cases filed in the 10 months after the November 1985 deadline than those filed in the 10 months before. Without your full dataset and without a lot of time to consider the issue -- I've only just reread your section with the Florida data specifics -- I'm going to stress again that this reaction is tentative. But I do think that there could be other reasons for a fall in average awards in the Florida dataset other than the contingency fee cap itself.

Time to Settlement: Theoretical Issues

Now I want to look at the second major empirical finding in your contingency fee paper, namely that in states with contingency fee caps, we see an increase in the time it takes to settle a case. You and Eric conclude that this phenomenon is explainable through your hourly-fee theory, i.e., that in states with contingency fee caps, plaintiffs' lawyers switch to the same types of cases on an hourly basis -- and stretch out the time to settle so that they can charge more billable hours to their poor unknowing clients.

Having observed hourly-fee attorneys in action, I'd be the first to concede that each marginal hour of work billed isn't necessarily in the client's self-interest in terms of the value added to the case. I'd also admit, as I did at the outset, that contingency fees help mitigate this incentive misalignment; the incentives for contingency fees on average tend to run in the opposite direction -- the attorney wants to get as much as possible, yes, but for as little work as he can.

But just because I agree with you that hourly-fee lawyers have an incentive to string out cases when they shouldn't, and contingency fee lawyers have an incentive to settle on the cheap and move on to the next case, doesn't mean I buy the theory you and Eric advance. Again, I know of no evidence that lawyers in New York, Illinois, or California are doing significant amounts of medical malpractice litigation on an hourly basis. In your most recent post, you acknowledge that rather than switching to a pure hourly fee, lawyers might shift from a pure contingency fee to other forms of cost-shifting. That's slightly more persuasive to me, but I'd still want to see some real evidence that we see this in practice -- for the med-mal cases you're analyzing -- before I bought into it very much.

Before I get into the specifics of your empirical analysis, let me delve a bit into what we'd expect to see if we accept my alternative view of the world. I don't think that there's a clear theoretical prediction about the effects of contingency fee caps on settlement time. On the one hand, by giving lawyers an incentive to weed out bad cases, contingency fee caps increase the number of dropped claims, as you observe. That trend would directly reduce the time to settlement. Moreover, lawyers working on an unlimited contingency fee, the expected return is higher all else being equal. Thus, it's worth it to invest more time in any given case -- and we'd expect contingency fee caps to lead to quicker settlements.

But other factors would seem to cut the other way. Specifically, the very incentives that cause plaintiffs' lawyers to be more cautious about taking weak cases in states with contingency fee caps, due to the reduced upside, are also likely to encourage lawyers to be more careful about discovery in general -- thus spending more time on the average case. Also, while contingency fee lawyers should be more risk neutral than their clients, they aren't generally going to approach pure risk neutrality except for megafirms with broad portfolios of cases, like the Milberg Weiss securities firm. So if I'm right that contingency fee caps increase the average probability of case success, then we'd expect to see an increase in settlement time; with uncapped contingency fees, risk averse attorneys with more long-shot claims are more likely to cut and run for the cheap settlement.

Those aren't the only arguments out there, but it's obvious that it's not wholly clear to me in which direction we'd expect contingency fee caps to influence settlement time in my view of the world.

Time to Settlement: The Florida Data

You and Eric observe that the time it took to resolve a case increased in Florida after contingency fee caps went into place in November 1985. In your regression analysis, you control for the stock of cases -- thus holding constant court "congestion effects" that may have occurred due to the "rush to file" before the caps took effect. With that control, the increase in time to settlement remains statistically significant.

Rather than reflecting attorneys' switching to hourly fees starting in November 1985, and bilking their clients, I'd guess that this effect has much more to do with the way the Florida reform was actually structured. The main Florida caps only kick in at $1 million. That's around the median jury verdict today, but well above it in 1985. Indeed, Lester Brickman observes that in 1984 the mean med-mal jury verdict nationally was just over $900,000 in 2001 dollars (at p. 710). The median is certainly a good bit lower than that (means are driven upward by large outliers in tort cases), the inflationary effects from 1984 to 2001 are substantial (but not worth my while to calculate since it's not central to my argument), and Florida may well have had lower verdicts on most cases than the rest of the country (since its cases are disproportionately brought by seniors, who have lower economic damages than cases brought by earners in their prime). The point here is that Florida's capping of contingency fees above $1 million is not likely a good explanatory factor for the increase in time to settlement you observe.

What may well be a good explanatory factor is the way Florida structured its fee limitations below $1 million. Fees were capped at one-third for recoveries through the time of filing an answer or a demand for arbitration; but fees were only capped at 40 percent from that point through the trial of the case. Assuming lawyers in Florida took advantage of this statutory cap and structured it into their fee arrangements after November 1985 when the law went into effect, contingency fee lawyers in Florida faced a direct inducement to string out cases beyond the filing of an answer or arbitration demand because they'd take home a bigger piece of the pie. That perverse incentive probably goes a long way toward explaining the results you observe.

Time to Settlement: The Cross-State Data

Like your time-series analysis in Florida, your cross-state analysis also finds that the time to settlement increases in states that have contingency fee caps. Again, though, I'm not convinced that this observation results from attorneys in those states bilking their clients using hourly fees, rather than other factors.

Specifically, I think your observed results are a function of anomalies in your data set. You draw from the 1992 sample of data from 75 large metropolitan counties in a Bureau of Justice Statistics report. When looking at Appendix table 2 (p. 8) of the report, I immediately notice the very long processing time in 2 of the 3 largest counties in your "contingency fee cap" states, namely New York, NY, and Cook, IL (Chicago). Cook County is a notorious "judicial hellhole," and a prominent left-wing litigator has called the Bronx civil jury "the greatest tool of wealth redistribution since the Red Army."

On a hunch, I looked at the summary data for civil jury results in the 1992 survey, presented in this BJS report. I looked at the 4 counties in your "cap states" with 200 or more civil jury trials in the sample -- the two aforementioned MSAs plus Los Angeles and Orange counties in California. I then examined the 7 counties in the "non-cap states" with 200 or more civil jury trials: Hennepin, MN; St. Louis, MO; Cuyahoga, OH; Philadelphia, PA; Bexar, TX; Dallas, TX; and Houston, TX. What I found confirmed my suspicions. The large counties in the "cap" sample had juries more likely to award for the plaintiff (54 percent to 50 percent), thus reducing risk to the attorneys (and reducing the incentive for attorneys to "cut and run" through early settlement). The "cap" counties had higher average verdicts ($834,000 to $697,000), thus increasing the expected return to attorneys (and increasing the amount of time a reasonable attorney would spend working on a case). And, significantly, the "cap" counties had a higher percentage of plaintiff winners getting a verdict of $1 million or more (14 percent to 8 percent) -- again, increasing the percentage of cases for which contingency fee attorneys would be willing to work more hours on a case, caps notwithstanding.

Of course, these data differences also show up in your summary data showing that the "cap" counties have higher average awards than "non cap" counties -- and indeed, that the differential is greater in med-mal cases than in auto cases (unsurprisingly, since the high-end med-mal verdicts bumping up the mean will be relatively larger than the biggest auto cases). But all that tells us is that contingency fee caps -- or more precisely, the caps imposed in these states -- are not sufficient to ameliorate the tort system's problems entirely. I don't think any reasonable person would claim otherwise.

But what your analysis tells us is that even these caps can have a significant effect in screening undesirable cases, to the tune of inducing a higher drop rate of around 15 percent. That's nothing to sneeze at.