Is it really possible to tell if insurance markets are functioning effectively without too many high- or low-risk buyers of policies? That is the challenge of new research by London School of Economics professors David de Meza and David Webb.
Their study, published in the November 2017 issue of the Economic Journal, suggests that standard tests for the twin phenomena of adverse and advantageous selection – too many or too few buyers who are likely to make a claim – lead to false diagnoses of the nature of market failure. They conclude that tests should be based on what influences the marginal buyer of an insurance policy, not the average buyer.
The researchers note that it has long been recognised that insurance markets are vulnerable to the curse of ''adverse selection''. Those with the most to gain from buying insurance fall into a particular category. They know that they are likely to claim, but they get a good deal because the company is not able to detect that they are bad risks.
Nevertheless, companies anticipate – or learn – that they will disproportionately attract unobservable bad risks, and so set their premiums accordingly. Good risks are therefore priced out of the market, exacerbating the problem.
In the worst cases, a ''death spiral'' may be triggered and the market completely disappears. Advocates of Obamacare often attribute the failure of private markets for health insurance to adverse selection.
The opposite problem of ''advantageous selection'' is also possible. For example, insurance appeals to the risk-averse who also avoid exposing themselves to hazardous situations. Rather than bad risks driving out good, the reverse occurs, leading to market over-expansion.
Empirical work finds evidence of both adverse and advantageous selection. The new research explores whether these tests are well founded.
The standard test asks whether buyers of high-cover policies are more likely to claim than those with low-cover policies. If there is no difference, it is concluded that neither type of selection is present so there is no inefficiency, as an early, influential study found.
There is though a paradox. If asymmetric information is not a problem, the theory on which the test is built implies that no low-cover policies will be sold, in which case the test cannot be applied!
The authors of the new study argue that administrative and related costs, which are often substantial in practice, are one reason why low-cover policies are chosen even when information is asymmetric. The test now loses most of its power.
Buyers of low-cover policies may be a mixture of high-risk types who are taking advantage of relatively low premiums made possible by low-risk buyers put off taking more cover by the loading on high-cover policies. Claim rates may be independent of coverage, but the coverage choice would be very different if pricing reflected individual risk.
There is a second problem with the standard test. Some buyers of a policy would not change their choice even if premiums were very different. Mispricing due to asymmetric information is not an efficiency concern for this group.
What matters for efficiency are the characteristics of those who would switch their choices if premiums were to change. The risk profile of the two groups need not be the same. Only the second group matters for efficiency, but the standard test averages across the inert and the alert.
This leads to false diagnoses of the nature of market failure. Tests should be based on marginal not average selection effects.
This research illustrates how neglect of these considerations leads to policy recommendations that are likely to be wrong. Similar issues arise in credit markets and other settings.
''False Diagnoses: Pitfalls of Testing For Asymmetric Information In Insurance Markets'' by David de Meza and David Webb is published in the November 2017 issue of the Economic Journal. David de Meza and David Webb are at the London School of Economics.