Input-based versus Output-based Incentive Contracts in Health Care: Experimental Evidence from India

Performance pay for developing country health providers ubiquitously rewards health input use (i.e., service delivery indicators).  In principal-agent relationships in which principals have complete information about input productivity and input use can be verified, this is optimal. However, if input use is difficult to verify and principals lack contextual information about health production functions, rewarding health outputs (good health status) may be superior.  In this paper, we directly compare input- and output-based incentive contracts for maternity care providers in rural India.  We find that input contracts reduce rates of post-partum hemorrhage, the leading cause of maternal mortality in India, by 28% – but no health improvement is associated with output incentives.  Part of the explanation appears to be that providers respond less to incentive contracts with greater risk that provider effort will not be rewarded. We also find evidence suggesting that provider overconfidence explains the heterogeneity in responses to incentive contracts.