Monday, February 09, 2004

WTC Insurance Case

A federal court will soon decide whether the 9/11 attack on the World Trade Center constituted one event or two. The answer matters because the insurance policy carried by the WTC’s owners specified payouts on a per-event basis. If 9/11 attack was one event, they will receive $3.5 million from the insurance companies; if it was two, they will receive $7 million.

There is no obvious right answer here. The biggest lesson, one that I emphasize to my law & economics students, is that no contract is ever complete. There will always be contingencies unforeseen by the parties and left unspecified in the contract, because it is costly to predict unusual events and insert clauses to deal with them, and there are diminishing returns to doing so. The parties to a contract rationally choose to leave some things for the courts to decide ex post, and that’s why the courts have to establish default rules.

Still, I’ll take a stab at finding the most correct answer. While the 9/11 attack was highly unusual, non-independent events are hardly unknown to contract law. Consider the damage done to your car by a hailstorm – would it be considered one event (a hailstorm) or many events (hailstone 1, hailstone 2, … , hailstone 100…)? I’m not an insurance lawyer, but I strongly suspect it’s the former. Similarly, if you have cancer, and the cancer leads to the need for two different procedures (an operation to remove a tumor, and a series of chemotherapy treatments), your insurance policy’s “per condition” cap presumably applies to the total from both, not each one individually. In each case, the probability of a second event, conditional on the other having occurred, is greater than zero. Given that my car got hit by one hailstone, there’s a greater chance than usual chance it will get hit by another. If I need to get a tumor removed, the chance of my also needing chemo is greater than if I hadn’t needed to get a tumor removed.

The general rule, then, is that when two events are not statistically independent, they are treated as one for insurance purposes. (I’m assuming positively related events; I choose to ignore cases in which one event reduces the likelihood of another.) In the present case, the two airplanes that hit the WTC were pretty clearly part of a coordinated attack, and thus not statistically independent. So I would probably side with the insurance companies, unless their contracts defined “event” in some novel way. The precedent established would be consistent with the general pattern I noted above.

Naturally, people writing insurance contracts would have the right to contract around the precedent by defining “event” in some more specific way (and adjusting premiums upward accordingly, since the liability created for the insurance company would be greater). The same would be true of the opposite precedent as well (with a downward adjustment of premiums). This is one of the great virtues of contract law – if you don’t like the court’s rules, you can typically write your own, with the consent of the other party, of course. But since the rule I advocate here seems congruent with current practice in less dramatic circumstances, I think it would not require as much rewriting of existing contract forms as would the alternative.

UPDATE: Just to be clear, I realize there’s some ambiguity in the rule I suggest here. Among other, we would have to ask what other factors should be taken as given before doing the calculation of statistical independence. For example, an elderly person might be more susceptible to a variety of illnesses; if we didn’t control for age, then a person with condition X might have a greater chance (than a randomly selected person) of also having condition Y, even though X and Y are not related. To deal with this and similar cases, my rule might have to be adjusted to some threshold higher than zero correlation. The implication for events like hailstorms and WTC attacks is still pretty clear, though.

No comments: