Abstract
Temporary incentives are offered in anticipation of persistent effects that are seldom estimated. We use a nationwide randomized experiment in the Philippines to estimate effects of two incentives for health insurance three years after their withdrawal. We find that both temporary incentives had persistent effects on enrollment. A premium subsidy had a small but highly persistent effect. Application assistance offered to those initially unresponsive to the subsidy had a much larger but less persistent effect. The subsidy persuaded those with higher initial stated willingness to pay to enroll and keep enrolling. The offer of application assistance to initial non-compliers with the subsidy achieved a larger immediate effect by drawing in those who stated they valued insurance less and were less likely to re-enroll when the incentives were withdrawn.