Aggregation and insurance-mortality estimation

Abstract

One goal of government health insurance programs is to improve health, yet little is known empirically about how important such government interventions can be in explaining health transitions. We analyze the child mortality effects of a major health insurance expansion in Costa Rica. In contrast to previous work in this area that has used aggregated ecological designs, we exploit census data to estimate individual-level models. Theoretical and empirical econometric results indicate that aggregation can introduce substantial upward biases in the insurance effects. Overall we find a statistically significant but quite small effect of health insurance on child mortality in Costa Rica.