Incentives for impact: Evidence from an RCT of a remedial education program funded by a development impact bond

This paper presents the results of a clustered randomized controlled trial of a remedial education program funded by a development impact bond (DIB). Remedial education programs can be effective at improving student learning, but the magnitude of impact varies. Experimental evidence from India has shown effects of remedial education programs on test scores ranging from zero to +0.7 standard deviations, with implementation details – who delivers the instruction, how often, whether in school or elsewhere – substantially mediating impact. These programs are thus an ideal candidate for DIBs, which have been proposed as a funding structure that can incentivize implementers to design the most effective version of their program for their local context. In the Educate Girls DIB, volunteers provided remedial instruction to students in government primary schools in Rajasthan, India. While the terms of the DIB itself were not exogenously varied, the RCT results suggest that the DIB created an environment conducive for program innovation. After one year of programming, treatment students gained a modest +0.07 SD in test scores relative to the control group. After extensive changes to program implementation, by the end of Year 3 treatment students gained +0.44 SD relative to control students. Heterogeneity analysis suggests that Educate Girls revised program activities in response to early-year evaluation results, leading to further impact. While these results suggest that DIBs may create an enabling environment for implementers to innovate, questions remain around which aspects of DIBs spur program impact, and whether DIBs are the optimal financing instrument for sustaining impact.