We exploit a large-scale natural experiment to estimate the causal impact of a switch in payment modality on demand for a popular life insurance product. Requiring upfront lump-sum payments reduces demand by about 30 percent relative to when paying by installment is possible. The context makes it unlikely that common neoclassical explanations have power, including the roles of price, income, liquidity constraints, information, convenience, and discount rates. We describe three behavioral explanations consistent with the finding. The sample includes 207,000 poor women involved in microfinance borrowing, and the result helps illuminate the low demand for merit goods in similar contexts.