“Preference Heterogeneity and Insurance Markets: Explaining a Puzzle of Insurance,” D. Cutler, A. Finkelstein & K. McGarry (2008)

I mentioned a few days ago that, in the absence of irrationality (or “switching costs” or “bounded rationality” or whichever nomenclature you prefer), it is often very difficult to generate functioning insurance markets. The standard models assume that people with more risk desire more insurance, all things equal: holding risk preferences equal, this just means people try to maximize expected utility. But what of the case where people with lower risk are the ones demanding more insurance, or “advantageous selection”?

Consider long-term care insurance. The type of people who buy a lot of such coverage are also likely people who are thoughtful and careful in other areas of their life, and hence people who will have lower long-term care costs otherwise. In such a case of advantageous selection, we avoid the usual informational problems. How true is this empirically? Cutler and his coauthors use great data from the Health and Retirement Study that includes measures of a number of “risky” activities (smoking, heavy drinking, lack of seatbelt use, etc.) as well as contemporaneous measures of use for five types of insurance (annuity, long-term care, Medigap, life and hospital care). Across a wide scope of risk measures, those who do risky activities are less likely to have insurance; those who wear seatbelts also buy annuities. Examining actual use of the insurance products, those who undertake risky activities are also more likely to utilize the insurance. In some markets, these two empirical facts combined can tell us why adverse selection does not destroy the market: in the standard adverse selection story, more life insurance is demanded by those who know they will die sooner, but because of heterogeneity is preferences for risk, those who live longer also turn out to be those who demand more life insurance. This is an interesting effect which seems important enough not to neglect in theoretical models of insurance markets.

One last thing, not totally related to this paper, but which I wanted to jot down in general: Chiappori and his coauthors (in a 2006 RAND) provide a very useful clarification to the whole asymmetric information story. Unfortunately – and probably because the name chosen was “asymmetric information” – there is a somewhat common belief that hidden information, in general, destroys efficiency in markets. This is not true. The world is full of hidden information: buyers don’t know which firm makes goods at least cost, sellers don’t know what buyer valuations are, etc. The RAND paper gives the example of car salesman not knowing which buyers want which colors for their cars. In competitive markets, such asymmetry is totally unimportant: we still reach first-best. Even if sellers don’t know what buyer valuations are, it doesn’t matter because in competition they have no power to change their price anyway. The problematic asymmetric information is that where one party has information which affects the value the other party places on the good. This is true in the used car lemons example of Akerlof, or in the example of individuals “selling” their health risk to an insurance company. This distinction is often elided, but I think it’s the right one to keep in mind when considering problem of information asymmetry.

http://dash.harvard.edu/bitstream/handle/1/2640581/cutler_preference.pdf?sequence=2 (AER P&P – there is a longer WP version, of course)

%d bloggers like this: