The gains from insurance arise from the transfer of income across states. Yet, by requiring that the premium be paid upfront, standard insurance products also transfer income across time. This project considers the implications of this transfer across time for insurance demand, with an application to crop insurance. Crop insurance offers large potential welfare gains in developing countries, yet demand has remained persistently low. We show that the intertemporal transfer can help explain such low demand, especially among the poor, and in a randomized control trial in Kenya we test a crop insurance product which removes it. The product is interlinked with a contract farming scheme: as with other inputs, the buyer of the crop offers the insurance and deducts the premium from farmer revenues at harvest time. The take-up rate for pay-at-harvest insurance is 72%, compared to 5% for the standard pay-upfront contract, and the difference is largest among poorer farmers. Additional experiments and outcomes provide evidence on the role of liquidity constraints, present bias, and counterparty risk, and find that enabling farmers to commit to pay the premium just one month later increases demand by 21 percentage points.