Brokers and the Insurance of Non-Verifiable Losses

Consider how the insurance industry responded to the events of September 11, 2001. It is unclear whether many of the losses at the World Trade Center were really covered under insurance policies. While many policies anticipated some level of terrorist activity and this was covered (or not excluded), most policies excluded acts of war. The events of September 11 and after seem to span terrorism and war. Indeed the U.S. president has continued to refer to the post-September 11 environment as a war situation, and the response has engaged the country in actual wars. Despite some ambiguity in whether the September 11 events were covered, leaders in the insurance industry quickly announced that they would not fight these claims. No doubt reputation and patriotism fed into this decision.

Now, compare this anecdote with the following. Several observers have noticed a recent and, supposedly, disturbing trend in insurance markets. Apparently, insurers are now more likely to dispute large claims, to offer less than 100 cents on the dollar, or to try to get away without paying. Richard and Barbara Stewart have labeled this the "loss of certainty effect," and Kenneth Abraham has talked of the "de facto big claims exclusion."¹ One reason for such disputes is that large claims threaten the solvency of the insurer, and such offers may be seen to resemble workouts in which distressed noninsurance firms negotiate with creditors. But the issue here is with the willingness, not the ability, to pay. These writers see the "big claims exclusion" as degradation of the insurance market because risk-averse consumers will place a lower value on such uncertain insurance. Indeed, they see a potential downward spiral of the insurance market if this practice continues.

The loss of certainty may be characterized as ex post bargaining over a settlement rather than a straightforward appeal to the policy conditions. Yet such bargaining should not be a surprise when claims are unusual and it is unclear whether they are really covered. For example, it is a matter of real dispute whether many environmental losses (for example, for cleanup of Superfund sites) are really covered and, if so, how the many policies in force over the long gestation period of such losses should contribute. Indeed, losses of this nature and duration were probably not anticipated when the policies were written, and therefore the policy wording is simply unclear.

Incomplete contract theory provides a very different view of these trends. In a world with rapidly evolving technology and shifting sociopolitical institutions, we might expect to be exposed to new types of losses. As with more traditional losses, there may be a comparative advantage in the transfer of such risk from individuals and firms to insurers and reinsurers whose capital and portfolio structure enables them to absorb such unknown losses at lower cost. But the novelty of these losses presents a problem. If the nature of losses cannot be anticipated with any precision (or if the variety of such potential losses is wide), then it may simply be infeasible to write enforceable contracts to share risk. Can we, then, find a way of arranging the affairs of individuals and potential insurers such that there is sharing of risk, despite the absence of enforceable insurance?

Our model works as follows. Many losses can be anticipated, and enforceable insurance policies can be written against these losses. Let us call such losses verifiable. Insurers establish a relationship to cover the verifiable losses, and a contract is written. However, the parties supplement this contract by creating a "forfeit," should the relationship break down. The idea of the forfeit is that because the parties both have something to lose, this will encourage bargaining over the non-verifiable loss even though it is not formally covered in the policy. This is the familiar "holdup" problem. The nature and size of the forfeit are set in place...