Randomized Evaluations of Educational Programs in Developing Countries: Some Lessons

This paper reviews recent randomized evaluations of educational programs in developing countries, including programs to increase school participation, to provide educational inputs, and to reform education. It then extracts some lessons for education policy and for the practice and political economy of randomized evaluations. Education is widely considered to be critical for development. The internationally-agreed Millennium Development Goals call for universal primary school enrollment by 2015. However, until recently there were no good assessments of how best to achieve this goal or how much it would cost. Some argue that it will be difficult to attract additional children to school, since most children who are not in school are earning income their families need. Others argue that children of primary-school age are not that productive, and modest incentives or improvements in school quality would be sufficient. Some see school fees as essential for ensuring accountability in schools and as a minor barrier to participation, while others argue that eliminating fees would greatly increase school participation. The simplest way to increase school participation is to reduce the cost of school, or even pay for attendance. Schultz (forthcoming) takes advantage of randomized order of program phase-in to examine the impact of the PROGRESA program in Mexico, which provided cash grants to families conditional on their sending their children to school. He finds an average increase in enrollment of 3.4% for all students in grades 1 through 8; the increase was largest