Health system changes under pay-for-performance: the effects of Rwanda's national programme on facility inputs.

Pay-for-performance (P4P) programmes have been introduced in numerous developing countries with the goal of increasing the provision and quality of health services through financial incentives. Despite the popularity of P4P, there is limited evidence on how providers achieve performance gains and how P4P affects health system quality by changing structural inputs. We explore these two questions in the context of Rwanda's 2006 national P4P programme by examining the programme's impact on structural quality measures drawn from international and national guidelines. Given the programme's previously documented success at increasing institutional delivery rates, we focus on a set of delivery-specific and more general structural inputs. Using the programme's quasi-randomized roll-out, we apply multivariate regression analysis to short-run facility data from the 2007 Service Provision Assessment. We find positive programme effects on the presence of maternity-related staff, the presence of covered waiting areas and a management indicator and a negative programme effect on delivery statistics monitoring. We find no effects on a set of other delivery-specific physical resources, delivery-specific human resources, delivery-specific operations, general physical resources and general human resources. Using mediation analysis, we find that the positive input differences explain a small and insignificant fraction of P4P's impact on institutional delivery rates. The results suggest that P4P increases provider availability and facility operations but is only weakly linked with short-run structural health system improvements overall.