Cash on delivery: Results of a randomized experiment to promote maternal health care in Kenya

• Full vouchers and conditional cash transfers were highly effective at increasing institutional deliveries among poor, rural women in Kenya. • Co-payments reduced the effectiveness of vouchers and unconditional transport subsidies were not effective. • A free care policy instituted by the government also demonstrated no effect on institutional deliveries. • SMS text reminders did not have any effect on institutional deliveries. We conducted a randomized controlled experiment to test whether vouchers, cash transfers, and SMS messages were effective in boosting facility delivery rates among poor, pregnant women in rural Kenya. We find a strong effect of the full vouchers and the conditional cash transfers: 48% of women with access to both interventions delivered in a health facility, while only 36% of those with neither did. Amongst women who did not receive a cash transfer, we find that a small copayment dramatically reduced voucher effectiveness, suggesting a discontinuous impact of cost-sharing on the demand for health services. Both the unconditional cash transfer and the text messages had limited effect on the use of health services. Finally, we also find no evidence that a government policy to eliminate user fees increased demand for maternal health services.