Citation:
Abstract:
Performance-based financing (PBF) programs have been introduced in numerous developing countries to increase the provision and quality of health services through financial incentives. Despite growing evidence about short-term impacts of PBF, less is known about medium-run impacts and scale-up effects, and about how PBF compares to other financing approaches. In this paper, we extend the initial evaluation of Rwanda’s PBF program to identify medium-run and scale-up effects of incentives and unconditional financing relative to a new “business as usual” counterfactual. We use data from the Demographic and Health Surveys from Rwanda and several Sub-Saharan African countries from 2001 to 2010, using two control group strategies: all available control regions, and a subset of regions that are similar to Rwanda based on pre-intervention trends in covariates and outcomes. We then use difference-in-differences regressions to measure the Rwandan program’s impacts on four indicators: institutional deliveries, antenatal tetanus prophylaxis, completion of any antenatal visits, and completion of four antenatal visits. The results are similar using the various control groups and in additional robustness checks. In the short-run and relative to no intervention, both performance-based and unconditional financing raised institutional delivery rates and completion of four antenatal visits. In the medium-run, relative to no intervention and in addition to the initial short-run impacts, performance-based incentives resulted in further improvements for institutional deliveries. Program scale-up was effective, with few differences between intervention arms after all areas received performance-based incentives. There were few effects on antenatal tetanus prophylaxis or on completion of any antenatal visits. Together, the results suggest that PBF can have persistent effects for some indicators, but unconditional financing can also be effective. Moreover, the analysis demonstrates how observational research methods and secondary data can generate new insights on completed trials.
Also available at CGD Working Paper 497.