Principal investigator: Adriano Valladão Pires Ribeiro
Article title: INPUTS, INCENTIVES, AND COMPLEMENTARITIES IN EDUCATION: EXPERIMENTAL EVIDENCE FROM TANZANIA
Article authors: Isaac Mbiti, Karthik Mutalidharan, Mauricio Romero, Youdi Schipper, Constantine Manda and Rakesh Rajani
Location of the intervention: Tanzania
Sample size: 350 schools
Sector: Education
Type of intervention: Transfer of resources to schools and teachers
Primary variable of interest: Passing the math, Swahili, and English assessments.
Evaluation method: Experimental Evaluation (RCT)
Policy Problem
Improving the education system in developing countries is a global challenge, with hundreds of millions of dollars spent annually by governments and donors. Despite this, increased spending on education does not always translate into improvements. This suggests that increased resources have a limited impact, and other constraints also play a significant role in determining educational outcomes. A complementary constraint to spending would be the effort expended by teachers; that is, the provision of materials resulting from increased spending would only improve the quality of teaching if accompanied by high teacher commitment, and vice versa.
Evaluation Context
The complementarity highlighted above can be tested in an experiment in Tanzania, a place where both resources and motivation and effort are lacking among teachers. In 2001, tuition fees for public primary education in the country were abolished, an important step towards universal primary education with enrollment growth from 52% in 2000 to 94% in 2008. Nevertheless, the level of education remained low, with only one-third of third-grade students having the expected proficiency in mathematics and literacy of a second-grade student.
The rapid growth in enrollment numbers has not been matched by increased spending on primary education, resulting in a lack of school infrastructure, basic teaching materials, and an average of nearly 50 students per teacher. Furthermore, a World Bank study The study revealed that one in four teachers was absent from work, and half of the teachers who were at school were absent from the classroom. The curriculum was taught for about two hours, less than half of what was scheduled, and 47% of teachers reported that they would choose another career if possible.
Intervention Details
Having identified the problems of both lack of resources and demotivated teachers, a program to mitigate each of these obstacles individually, and another that addressed both simultaneously, were launched. The interventions were implemented by Twaweza, a civil society organization in East Africa focused on citizens and the delivery of public services, in partnership with the government to facilitate their execution and evaluation. It is also noteworthy that the program was applied to a nationally representative sample of 350 schools in 10 districts in Tanzania throughout 2013 and 2014.
Some schools were selected to receive additional per-student funding in addition to the resources initially provided by the government. The funds were announced by Twaweza to participating schools and some community members, including students' parents, at the beginning of each school year. The funds would be transferred to the school's account in two installments, one in April and the other in August or September, and the way they were spent was decided by the school administration. Despite this, schools were required to be accountable, recording and publicly sharing information with the community about the income and expenditure of the funds.
The teacher incentive program provided a monetary payment to teachers based on their students' performance in an independent assessment conducted by Twaweza. Given the emphasis on early primary education, the program was limited to teachers in grades one through three and focused on mathematics and literacy in English and Swahili (the official language of Tanzania). The rule adopted was simple and depended only on the number of students who passed these assessments. Finally, the schools selected for both programs would follow the same protocols described above. It is important to highlight the difference between the transfer of resources to schools, which was unconditional, and the teacher incentive, which depended on student test results.
The selection of participating schools in each program was done randomly. First, the 10 districts were randomly selected. Then, the 35 participating schools were drawn, with 7 receiving the resource transfer, 7 receiving only the teacher incentive, 7 participating in both programs, and the remaining 14 forming the control group receiving no resources.
Methodology Details
The variables used to evaluate the effects of the programs were the results on math, English, and Swahili tests. The random selection of schools for the control or treatment group allows for comparison of the impacts of resource transfer, teacher incentives, and both, since differences in average exam results could be attributed to the adopted program. The idea is that, by considering the individual characteristics of the schools, teachers, and students, the inequality observed in the assessments would be fully explained by the implementation of the project. That is, the only element that would explain the difference in the products of interest between the schools in the control group and the treatment group would be participation in one of the programs.
Results
An assessment of the transfer of resources to schools in isolation indicates that the funds were primarily spent on books and classroom materials, such as blackboards, chalk, and maps. Other expenses were administrative and for student support, such as security and food. It is also noted that the expenditure per student was double the amount compared to the control group, and there was no increase in teacher salaries as per the program's rules. Regarding the assessments, following the methodology described above, no difference was found in the exam results between the schools in the control group and the schools that received only the cash transfer. In other words, although the expenditure per student did increase, no effect on learning was observed.
Schools that only provided incentives to teachers showed an increase in exam pass rates. By the end of the second year, students were 37%, 17%, and 70% more likely to pass the math, Swahili, and English exams, respectively. Finally, schools that received both financial aid and teacher incentives also recorded an increase in per-student spending similar to schools that received only the extra funds. However, this time, the impacts on assessment results were positive; by the end of the second year, the probability of passing the math, Swahili, and English tests increased by 49%, 31%, and 116%, respectively. It is noteworthy that the effects of combining the programs were even greater than those indicated by teacher incentives alone.
Lessons in Public Policy
The results above demonstrate that several factors are restricting the advancement of educational quality in developing countries and that simply increasing spending per student is not enough. Furthermore, mitigating only some factors and not others can also limit improvements. Another lesson to highlight is that more funding for education can increase learning when accompanied by better incentives for the effective use of resources. In this case, there were clear rules for schools to follow regarding accountability, information was made available to the general public and, especially, to the affected community, and a simple model for incentivizing teachers was in place.
Reference
MBITI, Isaac; MURALIDHARAN, Karthik; ROMERO, Mauricio; SCHIPPER, Youdi; MANDA, Constantine; RAJANI, Rakesh. “Inputs, Incentives, and Complementarities in Education: Experimental Evidence from Tanzania”. The Quarterly Journal of Economics, Volume 134, Issue 3, August 2019, Pages 1627–1673, 2019.
