The Effect of Treatment on the Untreated - Free Primary Education in Kenya


I investigate the effects of expanding access to education on students that would have been students in the absence of expansion to access. Any policy that expands access to education and causes children previously unable or unwilling to attend school to attend will change the composition of the average classroom, and will change school staffing decisions. Previous research has established that the composition of a student?s peers has significant effects on his performance, that the size and composition of a classroom has significant effects on teaching strategies, and of course that teacher quality significantly affects student performance. To explore this question, I analyze the implementation of a free primary education program in Kenya using data from the Demographic and Health Surveys. As a first pass, I compare the outcomes of students of ethnicities that were highly likely to be students prior to the implementation of the policy to those of ethnicities that were unlikely to be students prior to the implementation of the policy. Using a regression kink design, I find that the literacy rate of likely affected students exposed to the policy significantly increase due to the policy, and that the literacy rate of likely unaffected students exposed to the policy is not significantly affected by the policy. However, using a difference-in-differences method, I find that younger cohorts of unaffected students are increasingly negatively affected by exposure to the policy. I am currently implementing a sibling analysis, using students whose siblings attended school prior to the policy implementation as an additional measure of likelihood of being affected by the policy.