Insurance versus Savings for the Poor: Why One Should Offer Either Both or None

This paper analyzes data from a novel field experiment designed to test the impact of two different insurance products and a secret saving device on solidarity in risk-sharing groups among rural villagers in the Philippines. Risk is simulated by a lottery. Risk-sharing is possible in solidarity groups of three and insurance is introduced via less risky lotteries. Our main hypothesis is that formal market-based products lead to lower voluntary transfers among network members. We also test for the persistence of this crowding-out of solidarity. We find evidence for a reduction of solidarity by insurance if shocks are observable. Depending on insurance design, there is also evidence for persistence of this effect even if insurance is removed. Simulations using our regression results show that the benefits of insurance are completely offset by the reduction in transfers. However, if secret saving is possible solidarity is very low in general and there is no crowding out effect of insurance. This suggests that introducing formal insurance is not as effective as it is hoped for when the monetary situation can be closely monitored, but that it might be a very important complement when savings inhibit observing financial resources. The implication for policy is that microsavings should be offered simultaneously with microinsurance.