Irrational Economist_ Making Decisions in a Dangerous World - Erwann Michel-Kerjan [68]
Here I will give some examples of insurance buyer behaviors that involve such inconsistencies, indicate the kinds of analysis associated with approaches (a) and (b), and use these to discuss more generally the methodological strategy that should be used in each of these circumstances.
AVERSION TO DEDUCTIBLES
Insurance firms never plan on paying back all of the premiums they collect in the form of policyholder benefits, because they have administrative costs and capital costs that they have to cover. These costs are termed loading and are usually described as a proportion of benefits paid out. For example, a premium of $130 might lead to payouts of $100 per insured person, in which case the loading proportion would be $30 or 30 percent. What kind of insurance does it make sense to buy when you know that some of your premium will go to cover costs in this way?
A well-known proposition in insurance theory established in 1953 by Nobel Laureate economist Ken Arrow (who has also contributed to this book) maintains that the optimal policy will be full coverage above a deductible, so above a predetermined deductible of, say, the first $1,000 of the claim, the insurance will pay 100 percent of the losses incurred.1
In many circumstances, however, consumers do not seem to be rational in this sense; they seem to prefer policies with low or no deductibles even when the higher premium to cover them is very large relative to the amount of additional money one would expect to collect: For auto insurance, individual health insurance, or homeowners insurance, even comfortably middle-class households tend to choose plans with deductibles of only a few hundred dollars, and they do so even in the face of high proportional loadings. This preference for a low-deductible policy by so many buyers seems too strong to square with reasonable assumptions about risk aversion: People generally seem to prefer low deductibles even in policies that cover risks that are small relative to their wealth.
Indeed, when I show my students at Wharton a numerical example where the increase in the deductible is actually smaller than the reduction in premium (such that their guaranteed premium saving is more than the worst that could happen to them under the higher deductible), some still say they prefer the low-deductible policy.
How to explain this? One strategy is to posit a different (or at least an enriched) utility function in which “peace of mind” and “freedom from regret” have positive values (rather than only gain or loss, as predicted by the expected utility [EU] model), and that “peace of mind” extends on down to losses that are actually inconsequential relative to the person’s income or wealth (but presumably large enough to be upsetting). A related view is this: Especially if your wealth is ample, why worry yourself to death to save a few dollars on insurance?
One modification of the EU model that would predict low deductibles has been proposed by Schlesinger and Doherty (1985): It may be that there are “losses” associated with the uncertain event that are not directly insurable. In a second-best sense, it may then be preferable to overinsure the insurable event (e.g., monetary damage or cost) to compensate for the inability to insure the other event. The substitution cannot be perfect, of course, but paying for coverage in this indirect way may be better than no coverage at all.
These two different explanations imply two potentially different roles for public policy. If the EU model were accepted, the discovery of a rational explanation for deductible aversion would obviate the need for government to take corrective action regarding overinsuring. In contrast, if the