Adverse Selection in Low-Income Health Insurance Markets: Evidence from a large-scale RCT in Pakistan

Around the world, there is a growing interest to provide insurance policies to lowincome households. Selection of high-risk individuals into the insurance pool is an often cited impediment for the sustainability of such schemes, though. We provide robust evidence on the presence of adverse selection from a large randomized control trial on health insurance in rural Pakistan. Our experimental setup allows us to separate adverse selection from moral hazard, to estimate how selection changes at different points of the demand curve and to test measures against adverse selection. The results suggest that there is substantial adverse selection if health insurance coverage can be individually assigned. In particular, adverse selection tends to become worse with higher premium prices, creating a trade-off between cost recovery and the quality of the insurance pool. In contrast, adverse selection is mitigated when bundling insurance policies at the household or higher levels. Further analyses suggest that adverse selection in individual products has non-negligible welfare consequences and that these are less pronounced in relative terms when bundling policies.