Charles Silver has another post on the DMI blog on his studies; he even takes time to respond to one of the deranged anonymous commenters, though he has yet to address the critique of his previous post, even though it's been more than a month since I've made it. I'll be happy to post such a response on Point of Law if his hosts there are forbidding him from responding on that site.
Once again, Silver's post exhibits confirmation bias, stretching ambiguous data to reach conclusions that do not necessarily follow from the data.
Malpractice payments stack up at the policy limits, suggesting that insurance policies cap recoveries even when patients deserve much more.
Well, that's one possible suggestion. Another possible suggestion is that malpractice payments stack up at the policy limits regardless of the merits of the case: in the game theory of litigation, the plaintiff always proposes a settlement at policy limits, and the defendant insurance company has to decide whether to accept the settlement offer or accept the risk of being liable for more than policy limits if the results go against the defendant. Insurance policy limits would thus provide a focal point for settlement even when patients deserve much less because they are suing without actionable malpractice, but there is some small percentage chance of a jackpot verdict that forces insurers to settle at policy limits in response to an offer to do so or risk bad-faith litigation. We simply do not know because the "quality of the case" variable is entirely absent from the Texas Department of Insurance dataset used by Silver, yet Silver draws conclusions by assuming, without evidence, the very premise that is challenged by reformers. The $1 million settlement could reflect a defendant getting away with $3 million in damages and 100% chance of losing, or it could be a lottery case where $20 million in damages is possible 4% of the time, with the other 96% of juries getting it right and awarding zero.
What we do know from Silver's dataset is that insurers do ignore insurance limits and settle for above policy limits a significant percentage of the time. Silver's hypothesis works only if one assumes the policy limits are firm; the alternative hypothesis works even if the policy limits are not firm—and Silver's dataset shows that, indeed, the policy limits are not absolute. Silver provides no explanation why insurers would settle some meritorious cases at policy limits and others above policy limits; it's not clear to me that any such explanation is possible. If, however, policy limits provide a focal point regardless of the merits of the case where there is a risk of a jackpot verdict, then Silver's results say nothing about the merits of reform in one direction or the other.
Silver also writes:
Doctors almost never use personal assets to resolve malpractice claims. The claim that �every physician is one lawsuit away from financial ruin� is a myth.
This is, once again, a non sequitur—"almost never" is hardly "never," especially to a risk-averse physician. (
It's also unclear where Silver draws the premise from; the TDI dataset he uses does not collect data over the use of defendants' assets. Update: my mistake, it does.)
Perhaps the Silver paper does do analysis of the merits of the underlying cases and has a secret stash of data outside the TDI dataset such that he can draw such confident conclusions. (Update: here's the paper on SSRN.)
Update, May 26: Justinian Lane takes it upon himself to respond, but his only point is that "only" 2.4% of settlements are above policy limits. There are three reasons why this fails to rebut my argument. First, as I mentioned briefly in my previous post (though not this one), and as Silver fails to acknowledge in his DMI posts, though he indirectly acknowledges in the paper itself, the TDI dataset is incomplete, and is not capturing a number of settlements above policy limits. Second, the relevant number is not 2.4% because that denominator includes the 70% or so of settlements where the insurance limit is not a constraint on settlement because the final settlement is well below the policy limit. The relevant number is that at least one in fourteen settlements at or above policy limits is above policy limits. (Forty percent of verdicts above policy limits are settled above policy limits.) That is a significant number, because it is consistent with my model, but not Silver's model. Indeed, this leads to the third point, which is that the oversimplified Silver argument is falsified by any payments above insurance limits. The fact that plaintiffs do have recourse beyond policy limits means that the policy limits for a single doctor are not a cap—and don't forget that in any sizable case, the plaintiff is going to be seeking to fulfill damages from multiple insurance policies, a fact not captured by the way Silver measures the data. (Lane proposes an explanation that insurers are accepting settlement offers slightly above policy limits to avoid legal expenses. This is nowhere to be found in the Black/Silver papers, and for good reason: it contradicts the game theory by which insurers face Stowers risk for failure to settle. See American Physicians Ins. Exch. v. Garcia, 876 S.W.2d 842 (Tex. 1994) (no Stowers liability for denying settlement offer larger than policy limits).)
Perhaps Lane can attempt to defend Silver's exaggerations in his earlier post, too, since Silver appears not to want to.