Adverse Selection Does Not Explain Why Utilization Rises with Premiums: Evidence from a Health Insurance Experiment in India
Abstract: Information asymmetries or community-rating can lead to adverse selection into health insurance. We use a multi-armed RCT that varies insurance premiums to study selection into a health insurance program in India, Rashtriya Swasthya Bima Yojana (RSBY). Limited fiscal capacity in low and middle income countries (LMICs) may necessitate charging premiums, which may exacerbate adverse selection. Moreover, the degree of selection may differ in LMICs due to limited healthcare supply and knowledge, and constraints that interact with selection in a priori ambiguous ways, e.g., liquidity. We find mixed evidence on selection into the program. While those who purchase insurance when premiums are high use insurance more, they are no higher risk than those who purchase at lower prices. To interpret these findings we appeal to a literature in development economics that studies why price sometimes increases utilization of a product. That literature suggests three explanations: selection, price as a signal of quality, and sunk costs. Given this framing, our positive correlation test is not dispositive because it is consistent with all three theories. However, the evidence that enrollment does not vary with risk is inconsistent with selection. We are also able to rule out that price signals quality. Our study employs a two-stage randomization that varies the premium that neighbors pay, holding constant what a household pays. We find that people do not utilize insurance more when their neighbors are charged a higher price. These results suggest that sunk costs effects, rather than adverse selection, explain the effect of price on utilization in our context.
