We present results from a large-scale randomized experiment across 350 schools in Tanzania that studied the impact of providing schools with (a) unconditional grants, (b) teacher incentives based on student performance, and (c) both of the above. After two years, we find (a) no impact on student test scores from providing school grants, (b) some evidence of positive effects from teacher incentives, and (c) significant positive effects from providing both programs. Most importantly, we find strong evidence of complementarities between the two programs, with the effect of joint provision being significantly greater than the sum of the individual effects. Our results suggest that combining spending on school inputs (which is the default policy) with improved teacher incentives could substantially increase the cost-effectiveness of public spending on education.A randomized controlled trials registry entry is available at https://www.socialscienceregistry.org/trials/291 1 Some of these include poor student health and nutrition, low student attendance, mismatch between curriculum/pedagogy and student learning levels, and low levels of teacher effort and accountability. Each of these challenges has been extensively documented in multiple developing country settings. See Glewwe and Muralidharan (2016) and Mbiti (2016) for reviews and references to primary sources. 2 The Government's capitation grant policy aimed to provide schools with TZS 10,000/student. The program we study provided another TZS 10,000/student over and above this grant, effectively doubling the school grant. Since teacher salaries were paid directly by the government and did not pass through the schools, these grants (from the government and the program) were the main source of discretionary funding available to schools. Schools were not permitted to use grant funds to augment teacher salaries or hire new teachers (consistent with government rules for capitation grant expenditure).
1We conducted the experiment in a large nationally-representative sample of 350 public schools (and over 120,000 students) across 10 districts in mainland Tanzania. We randomly allocated schools to four groups (stratified by district): 70 received unconditional school grants, 70 received the teacher performance pay program, 70 received both programs, and 140 were assigned to a control group. The study was powered adequately to test for complementarities, and we gave the same importance to testing for complementarities as testing for the main effects of the two programs. 3 All programs were implemented by Twaweza, a leading Tanzanian non-profit organization.We report four sets of results. First, the school grant program significantly increased per-student discretionary expenditure in treated schools. We find evidence of reduction in school and household spending in the Grant schools. Even after this reduction, there was a significant increase in net discretionary school-level spending per student in treated schools (excluding teacher salaries). However, this increase in spend...