Testing for Adverse Selection with “Unused Observables”

This paper proposes a new test for adverse selection in insurance markets based on observable characteristics of insurance buyers that are not used in setting insurance prices. The test rejects the null hypothesis of symmetric information when it is possible to find one or more such “unused observables” that are correlated both with the claims experience of the insured and with the quantity of insurance purchased. Unlike previous tests for asymmetric information, this test is not confounded by heterogeneity in individual preference parameters, such as risk aversion, that affect insurance demand. Moreover, it can potentially identify the presence of adverse selection, while most alternative tests cannot distinguish adverse selection from moral hazard. We apply this test to a new data set on annuity purchases in the United Kingdom, focusing on the annuitant’s place of residence as an “unused observable.” We show that the socio-economic status of the annuitant’s place of residence is correlated both with annuity purchases and with the annuitant’s prospective mortality. Annuity buyers in different communities therefore face different effective insurance prices, and they make different choices accordingly. This is consistent with the presence of adverse selection. Our findings also raise questions about how insurance companies select the set of buyer attributes that they use in setting policy prices. We suggest that political economy concerns may figure prominently in decisions to forego the use of some information that could improve the risk classification of insurance buyers. JEL Classification Codes: D82, G22