Morbidity risk and health-insurance coverage

There is evidence from different countries suggesting that agents covered by health-insurance policies overuse healthcare services and are more likely to present a poor health condition. We access data from Brazil to test whether individuals covered by health-insurance plans are more likely to report suffering from a list of twelve different health problems, namely: arthritis or rheumatism, bronchitis or asthma, cancer, chronic renal disease, cirrhosis, depression, diabetes, heart disease, hypertension, lumbar pain, tendinitis, and tuberculosis. Our analysis is possible thanks to a complementary health schedule included in the 2003 National Sample Survey of Households (PNAD) survey, conducted by the Brazilian Institute of Geography and Statistics (IBGE). This complementary schedule contains self-reported information on individual health-insurance coverage and incidence of these medical conditions.

The Brazilian institutional context is relatively standard. A regulatory reform approved in 1998 restricted risk-based premiums such that they depend exclusively on seven pre-determined age categories. Our data allows us to divide health-insurance contracts in two groups: (i) the individual policies, which are paid by the insuree directly to the carrier; and (ii) the employer-contracted policies, which are contracted through an employer. In principle, adverse selection should act more intensively in individual policies than in employer-contracted plans, since the choice of the latter type of policy depends on many other factors besides ex-ante individual risk. On the other hand, morbidity risk caused by moral hazard should affect both types of policies equally, given that they present the same coverage level and, then, the same incentive structure. We estimate the risk-coverage relationship for each of those pre-determined age categories. Contrary to previous results from other countries, we do not find a positive correlation between self-reported morbidity and insurance coverage.