Irrational Economist_ Making Decisions in a Dangerous World - Erwann Michel-Kerjan [72]
Suppose you are considering buying insurance coverage that you know to be priced close to cost against the possibility of an event that imposes a loss of $L, where L = dollar amount of loss. You do not have to worry about how likely the event is or try to collect some data or ask others about it. Instead, you need only plug in your estimate of how much of the premium goes for loading, interpret the rest as a measure of expected loss (and expected benefit from insurance), and decide whether or not your risk premium (which you can calculate when you know the loss and its probability) is or is not greater than the loading. This information will also tell you how large a deductible to choose.
This means that the first question a rational person should ask when confronted with an offer of insurance coverage in an unfamiliar setting (say, long-term care insurance) is not how much the loss (or loss probability) might amount to but, rather, what the loading (including profit) on the offered coverage is. Unless you have some inside information that your loss probability is much different from what the insurer would think is average, knowing the loading and knowing the premium will give you almost all the information you need—if not for a perfect choice, then for one that is pretty good.
Unless there is a conscious attempt at secrecy, finding out the loading embodied in an insurance offer is usually easier than trying to find the data on losses and their frequencies. If the insurance market has multiple sellers, if there are past data on premiums paid in and benefits paid out, and if you expect the future to be much like the past, the needed information is in principle available from the insurer’s own accounting data. Often these data must be submitted to a state insurance regulator; alternatively, a prospective buyer may ask the insurer for this information. If the insurer refuses to furnish it, that alone speaks volumes about the likely low value of the insurance relative to its cost.
The key point here is that rational insurance purchasing can be based on a small amount of fairly objective data (or assumptions about data) and need not depend on the feelings and facts currently known by insurance purchasers.
CONCLUSION: WHEN RATIONAL AND BEHAVIORAL DECISION MAKING COMPLEMENT EACH OTHER
My conclusion from these applications is that the rational and the behavioral need not be in serious conflict if there is reason to believe that insurance markets are working reasonably well. Given that entry into insurance is relatively unrestricted, “working reasonably well” should be the rule rather than the exception, although spirited discussion about how well is “reasonably well” will be commonplace. There are some exceptions to rational behavior for relatively small and linked risks, such as warranties and rental car insurance. But for those risks that might (if uninsured) make a big difference to a person’s expected utility, we can expect markets to work reasonably well, and usually they do. Of course, a large number of little risks can mount up to a big risk, but insurance firms seem adept at bundling risks to a given asset in a given use (as in the form of all-perils coverage). Things are not going to work that well for new, large, unknown losses—but you have probably guessed that. This is precisely the focus in the next part of the book.
RECOMMENDED READING
Arrow, K. J. (1963). “Uncertainty and the Welfare Economics of Medical Care.” American Economic Review 53, no. 5: 941-973.
Bundorf, M. K., and M. V. Pauly (2006). “Is Health Insurance Affordable for the Uninsured?” Journal of Health Economics 25, no. 4: 650-673.
Gruber, J. (2008). “Massachusetts Health Care Reform: