Teacher performance pay in Rwanda

Coauthors: Clare Leaver; Owen Ozier; Pieter Serneels.

Further details:

• STARS trial in the American Economic Review
• IPA project page
• AEA trial registry
• Development Impact blog post on blinded pre-analysis strategy.

The developing world faces a crisis of learning. While there have been notable gains in student enrollments, school quality has in fact worsened in many developing countries. Literacy rates among those completing five years of schooling have actually fallen by 24 percent in Sub-Saharan Africa and by more than 35 percent in South Asia since the 1950s.¹ Because investments in schooling translate into economic opportunities only when they generate learning,² this deterioration in school quality translates into an enormous lost opportunity – the improvements in living conditions that did not materialize because of school systems’ failure to deliver learning.

A growing body of evidence suggests that the human resources in developing-country education systems are a key piece of the puzzle to unlock their learning potential. Surveys indicate that, across seven Sub-Saharan African countries, only two thirds of primary school teachers possess mastery of fourth-grade curricula in English and Math; these surveys also reveal teachers to be absent from schools 23 percent of the time.³ And pervasive teacher churn places further strain on school resources: in Rwanda, 20 percent of primary school teachers separate from their jobs annually, resulting in a patchwork of teaching arrangements within schools.⁴ But in spite of these diagnoses and scattershot evidence on policies that may improve teacher effort levels, little is known about the scope for improving the recruitment, effort, and retention of effective teachers at scale.

The STARS randomized, controlled trial

Work led by gui2de and recently published in the American Economic Review,⁵ has shown how researchers and policymakers can design contracts for teachers that strengthen these margins. Researchers worked with government counterparts to design a randomized, controlled trial conducted in public primary schools in areas of Rwanda struggling with poor learning outcomes and low teacher retention rates. The STARS (Supporting Teacher Achievement in Rwandan Schools) trial demonstrated that a contract that rewarded teachers not only for the learning gains that they delivered in the classroom, but also directly for their inputs of presence, preparation, and pedagogy, could raise student learning by 0.21 standard deviations annually. These gains are equivalent to an extra year of status-quo learning for each year spent in school, effectively doubling the rate of learning growth.

An adaptive, experimental scale-up

gui2de is now leading the development of the next phase of this work. Grants awarded by USAID’s Development Innovation Ventures (DIV) and the French Development Agency’s Fund for Innovation in Development (FID) will support a three-year program to learn how to scale the lessons generated by earlier experimental evidence. Leveraging recent investments in a system of comprehensive student assessments and Rwanda’s imihigo framework for civil-servant performance contracts, the project will provide lessons on adaptation to government systems, while building a model to be operated on national scale. By its third year, this project aims to reach scale in 10 of Rwanda’s 30 districts, strengthening the learning environment for an estimated 370,000 pupils annually.

Consensus estimates of the economic returns to education⁶ suggest that the potential benefits of a successful scale-up represent an extraordinary return on the investment of public funds. A gain equivalent to one year of schooling, even for a population as poor as that of Rwanda, is estimated to yield an impact of between USD 21 and 66 million in increased earnings for each year spent in the work force by the students in our proposed study districts.

Beyond these development outcomes, the STARS scale-up will continue to address key social-scientific questions. A deep partnership with the Rwanda Education Board will allow us to learn how the design of teacher placement mechanisms can harness teacher preferences to strengthen recruitment, effort, and retention. Randomized trials run within each of the first two years of the project will generate insights into teachers’ responses to alternative contract designs. And we will develop methodological tools for “adaptive experimentation”– learning quickly from the lessons in one year to inform refinements of designs across years – a promising tool for the evaluation of social policies across a range of contexts.