Irrational Economist_ Making Decisions in a Dangerous World - Erwann Michel-Kerjan [112]
Stage 1 risk is semi-mutualized in binding long-term contracts with separate treatment of stage 1 and stage 2 risk. Let us assume that stage 2 risk is reasonably well behaved in the sense that it is not highly contemporaneously correlated across regions and insurance markets can secure geographical diversification. These are strong assumptions, but they establish the minimal conditions under which ongoing short-term insurance would function. Now let us assume that stage 1 risk is subject to serial and contemporaneous correlation. To the extent that it is, diversification does not help. However, we know much about insurance design under such circumstances. The mutual structure allows all idiosyncratic risk to be effectively insured. Yet the systematic risk is shared by all policyholders through equity participation in the overall risk pool.7 This can be achieved through organizational structure (mutual or reciprocal insurance companies) or by contract design (participating policies issued by a joint stock insurance firm).
One might conjecture that an optimal long-term insurance contract would have the following structure. First, idiosyncratic and diversifiable stage 1 premium risk would be mutualized. Each policyholder would have his or her premiums indexed to changes in the total risk in the pool. This would both ensure overall premium adequacy and provide some degree of risk protection for those in regions where risk increased more rapidly. Stage 2 risk would then be insured under normal conditions. In short, the stage 2 risk would be borne by short-term insurance contracts secured by insurer capital in the usual way, whereas the stage 1 risk would be borne communally by means of indexed premiums. However, if it did transpire that stage 2 risk exhibited higher correlation, then one could have a secondary level of mutualization under which all policyholders funded actual deficits (or shared surpluses) in the annual results by means of additional premiums or refunds.
Both types of contracts—the semi-mutualized insurance and the constant premium contract—raise concerns about the disincentives for homeowners to take measures to mitigate their own risk. In both cases, policyholders are protected from, at least, idiosyncratic changes in risk; but would this erode homeowners’ incentive to mitigate? Certainly, either form would compromise incentives for short-term mitigation. For long-term mitigation, the issue is more complex. Perhaps the central issue is that incentives for mitigation rest largely on the prospect that home values do indeed reflect risk. If long-term insurance were bundled with home ownership (such that rights to the policy transferred with ownership of the house), then insurance would provide a hedge against falling house prices due to increased hazard. However, it is apparent that decoupling house prices from hazard risk would simply encourage people to continue to move to high-risk areas, as we have seen in the state of Florida. Under normal insurance practice, the policy attaches to the policyholder rather than to the home, so this may not be an issue.
Stage 1 risk is hedged through non-insurance mechanisms. The benefits of a mutual approach to stage 1 risk also could be achieved by non-insurance mechanisms. To hedge against local changes in insurance premiums caused by changes in local risk, one could take derivative positions in house prices. In fact, such derivatives are already traded in some metropolitan areas. A derivative trade can easily be constructed in which individuals are protected from the fall in their house price relative to the national average change in house prices. A market for such contracts would protect people from the idiosyncratic stage 1 risk. Of course, the local changes in house prices would reflect not only changes in hazard risk but also changes in regional economic factors.