18 August 2008

NYT on Rejected Offers

Earlier this month, the New York Times published an article on trial verdicts in civil litigation where there are rejected settlement offers. The results were as follows:

Defendants made the wrong decision by proceeding to trial far less often, in 24 percent of cases, according to the study; plaintiffs were wrong in 61 percent of cases. In just 15 percent of cases, both sides were right to go to trial — meaning that the defendant paid less than the plaintiff had wanted but the plaintiff got more than the defendant had offered.

The vast majority of cases do settle — from 80 to 92 percent by some estimates . . . and there is no way to know whether either side in those cases could have done better at trial. . . the findings [are] based on a study of 2,054 cases that went to trial from 2002 to 2005 . . . .

On average, getting it wrong cost plaintiffs at about $43,000; the total could be more because information on legal costs was not available in every case. For defendants, who were less often wrong about going to trial, the cost was much greater: $1.1 million.

“Most of the time, one of the parties has made some kind of miscalculation or mistake,” said Jeffrey J. Rachlinski, a law professor at Cornell who has studied how lawyers and clients decide to go to trial . . . .

The study’s authors have analyzed some data from New York and, after a review of 554 state court trials in 2005, have found parties to lawsuits making the wrong decision at comparable rates. . . .

As part of the study, which is the biggest of its kind to date, the authors surveyed trial outcomes over 40 years until 2004. They found that over time, poor decisions to go to trial have actually become more frequent. . . .

[P]oor decisions by plaintiffs to go to trial “are associated with cases in which contingency fee arrangements are common,” according to the report. “On the defense side, high error rates are noted in cases where insurance coverage is generally unavailable.”


The facts seem plausible. The analysis, however, isn't rich enough and comes to the wrong conclusions as a result. As an attorney who has done plaintiffs' personal injury work on a contingent basis, insurance defense work on an hourly basis, and commercial litigation in which neither party is insured, I will suggest some more plausible and nuanced explanations for these facts.

It is a relatively trivial matter to show mathematically that in a world where litigation costs were irrelevant, plaintiffs and defendants had equally accurate access to information and expertise, all judgments were paid in full, and no one was unduly motivated to settle by financial necessity, we would expect that plaintiffs and defendants would do better than final settlement offers equally often and in equal amounts. But, of course, we don't live in such a world. Why is the real world different?

Lack of expertise and communications skills on the part of lawyers, the explanation suggested in the New York Times article is one possibility, and may explain trendlines over the long term towards somewhat less accuracy in predicting outcomes generally. But, as economists are eager to illustrate with a host of examples, betting on irrationality to explain a large sample of regular phenomea is usually a losing bet. The study's findings themselves, discount the importance of expertise in this situation.

[T]he study . . . found that factors like the years of experience, rank of a lawyer’s law school and the size of a law firm were less helpful in predicting the decision to go to trial. More significant was the type of case.


Notably article states without attribution that: "Law schools do not teach how to handicap trials, nor do they help develop the important skill of telling a client that a case is not a winner." This is simply not the case. In fact, while these points don't appear in any single course, and instead appear across the curriculum, both were actually hammered into students repeatedly when I was in law school.

Betting that experienced attorneys are consistently acting in a manner that is grossly irrational is an even weaker bet.

The hard lesson for practicing lawyers to learn is not how to handicap a lawsuit, or to try to control client expectations, as we were taught in law school, but to learn that what makes a particular settlement acceptable to a client often depends on facts other than the objective merits of their case at trial.

A better explanation is that plaintiffs and defendants acted rationally, but that they had different incentives.

The Impact of Litigation Costs

One important fact to consider in a tort case, such as one involving an accident or medical malpractice, is that the parties make litigation decisions, not the lawyers.

A plaintiff in a personal injury case often pays for legal representation with a contingency fee. In the classic case, where the lawyer gets a third of the payout, a plaintiff who gets a judgment for exactly the amount of the last settlement offer is no worse off financially. The legal fees are the same.

In contrast, because insurance companies, on behalf of their clients, typically pay defense lawyers who litigate a case to trial on an hourly basis, the insurance company (who pays both the judgment and legal fees, and is the real party in interest in most tort litigation) must come out many thousands of dollars better than the last offer in case in order to come out ahead.

As a result, even in a world with perfect information, so long as the verdict is somewhat uncertain (as it always is in personal injury cases where non-economic damages are always determined on an ad hoc basis by a jury), we would expect plaintiffs to do better than the final offer less often than defendants at trial, even if all other things were equal. Indeed, this result affirms the fact that Plaintiffs' attorneys are acting ethically by deferring to client judgment, even when their own interests differ from those of the client.

In fact, probably about a third or more of the gain defendants typically make at trial by not settling in the cases where they make the right call is erased by additional legal fees.

Similarly, in cases where the defendants don't have insurance (where defendants tend to have much higher "error rates"), the plaintiffs lawyers will typically be paid by the hour, so plaintiffs have an incentive to settle close cases to avoid the litigation costs associated with going to trial to far greater extent in these cases than in cases where they are represented on a contingency basis.

Bad Faith Liability

There is also another reason why defendants settle generously when the real party in interest is the insurance company, which is called the doctrine of bad faith.

An insurance company has a duty to act in good faith in handling an insureds claim, including how it litigates cases on behalf of a defendant. The insured can't settle the case without the insurance company's permission (without losing coverage). And, the defendants' insurance policy limits are disclosed early in litigation.

If a plaintiff offers to settle for the insurance policy limits (or less), and the insurance company doesn't agree, and then the judgment at trial is for more than the policy limits, the insurance company is likely to be sued by the insured who was sued for the amount of the judgment in excess of policy limits that was incurred on the theory that the insurance company acted in bad faith by rolling the dice with the insureds money, rather than holding the insured harmless with a possibly generous settlement.

Settling a case typically protects an insurance company from paying more than the policy limits. Going to trial creates a risk for the insurance company of paying more than the insurance contract says it should, in an undetermined, but almost unlimited amount in a state like New York (where the study was conducted) which has weak limits on non-economic and punitive damages.

The average $1,100,000 penalty paid by defendants in the 24% of cases where they come out worse at trial than they do had they accepted a final settlement offer illustrates the high stakes risks to defendants of failing to settle in the face of a good offer.

Once again, defendants backed by insurance companies have an incentive to settle unless they are very sure that they will come out considerably better at trial, to avoid the downside risk.

Strategic Litigation Strategies

One suspects that another reason that defendants represented by insurance companies usually do better than a final offer, even considering litigation costs, is that insurance litigants are repeat players who have strategic interests in winning at trial.

Settlement occurs in the shadow of what is likely to happen at trial. If you litigate repeatly, you have a long term interest in developing a good track record so that you can be credible in settlement discussions. In theory, an ethical lawyer for an injured plaintiff, who will usually be a one time litigant, doesn't have this incentive.

As a result, a rational insurance company or other repeat litigant will offer to settle close cases for considerably more than they are worth, so that the cases that actually go to trial will be winners. This incentive is due in part to the fact that settlements are not a matter of public record, while trial verdicts are a matter of public record. Notably, in the case of medical malpractice suits, both settlement amounts and verdict amounts are maintained publicly, and settlement is much less common in these suits than in other personal injury suits.

The flip side of plaintiffs coming out behind significantly in the 8%-20% of cases that actually do go to trial, is that plaintiffs are presumably doing better on average than they could have at trial in the other 80%-92% of cases, which is to say that in a narrow sense, defendants lose by settling cases.

Judgment Proof Defendants

The particularly high "error rate" of defendants in uninsured cases may also be easy to explain. Defendants will typically reject an offer that they cannot pay or that will put them out of business, regardless of the likelihood that they will lose and lose big in a case. At trial, an asset poor defendant either isn't wiped out by the judgment, or loses everything the defendant has, while a settlement for everything the defendant has makes that loss a certainty. A rational defendant in a situation like that will often reject a settlement knowing that it is better than the most likely result at trial.

The large gap between the judgment and the last offer before trial in such cases also suggests that in these kinds of cases, litigation costs are almost irrelevant relevative to the amount at stake on the merits, so plaintiffs have little incentive to accept big discounts in order to avoid the cost of litigating the case to trial in these kinds of cases.

Declining Predictive Accuracy

One should expect that attorneys are less accurate in predicting trial outcomes than they used to be. At least one reason is simple. Far fewer cases, on a percentage basis, go to trial than was the case forty years ago, and the number of trials has steadily declined. The less data there is out there, the harder it is to predict the outcomes.

The decline in the number of trials isn't random either. Pre-trial motions to dismiss and motions for summary judgment resolve a much larger percentage of cases (predominantly easy cases) before trial. So, the cases that do go to trial tend to be closer. The increasing availability of compulsory pre-trial factual investigation, called discovery, has greatly increased the cost of litigation, and hence the down side of going to trial, so the margin of error that must be present for an insurance company to litigate a case is greater. And, mandatory mediation, in which a third party prods the parties to reach any possible settlement, has likely become the norm and encouraged an ethic of settlement.

The most experienced trial attorneys typically either got their trial experience in criminal law, where trials are far more common but the litigation environment differs a great deal from civil trials, or got their experience as junior civil litigation attorneys in an environment when fewer and fewer cases are going to trial. While lawyers aren't necessarily avoiding trial because they aren't good trial litigators, the rarity of civil jury trials to provide examples in future cases makes it hard for anyone to predict their outcome. The smaller the sample size a person has, the harder it is for them to accurately predict the outcome on a new event.

It is also important to note that defendants are ten times as likely, as measured by the arithmetic mean result, to make the wrong decision about going to trial as plaintiffs.

As noted above, some of this outcome may have to do with defendants in non-insurance cases that are rational despite carrying a high risk of loss. There is every reason to believe, given the large downside risks of going to trial, that insurance company lawyers would be more conservative than the overall numbers indicate. This may also have something to do with the fact that the insurance company employees making decisions for defendants' lawyers are often playing with someone else's money, while plaintiffs making decisions are usually making decisions for themselves.

This may also have something to do with the pressure on insurance company lawyers to keep settlements in cases that don't go to trial reasonable. It may be worth a few big losses in order to secure a lot of little wins.

And, finally, it could be simply a statistical fluke. Settlement offers rarely get any lower than zero, and generally will rarely be more than the ability of a defendant or the defendant's insurance company to pay, which means settlement offers from plaintiffs are rarely more than $300,000 in cases against individuals (which make up the numerical bulk of the cases). In contrast, there is no real limit to how high a jury verdict can get. Overall, jury verdict numbers are highly skewed by a very small number of very high verdicts. It could be that 95%+ of insurance defense defendants that make the "wrong choice" come quite close in magnitude to the actual jury verdict, and that a tiny 5% or fewer of verdict, some in uninsured cases and some in insured cases, are driving a very high average error.

Bottom Line

In the case of social phenemena as complex of lawsuits, simple explanations drawn from data sets that are insufficiently rich in detail, are likely to be deceptive.

This data set, for example, rather than impuning the judgment of plaintiffs' lawyers, instead suggests that plaintiffs lawyers are, in the highest ethical tradition, putting their clients' interests above their own, while insurance defense lawyers may roll the dice with somebody else's money more often than they should, although perhaps with reasons that make sense in the institutional culture of insurance companies.

I look forward to seeing the actual published journal article, where breakdowns by type of case may either enhance or dispute the analysis I've offered above, this September.

No comments: