International DevelopmentHuman capitalPositive

The Balsakhi Tutoring Program

MIT Poverty Action Lab (Banerjee, Cole, Duflo, Linden) · Vadodara and Mumbai, India · 2002

Summary

The Balsakhi program addressed a structural failure in Indian primary education: students who fall behind early tend to stay behind because instruction is pitched at grade level regardless of actual learning. By pulling struggling students out for targeted instruction from a local community member — rather than a credentialed teacher — the program produced significant learning gains at very low cost (approximately $2 per student per year in some implementations). The two-year follow-up was particularly important: gains persisted even after children moved to schools without the program, suggesting the learning was genuine rather than test-day coaching. The program became a template for what is now called Teaching at the Right Level (TaRL), a globally replicated approach developed by Pratham and evaluated in multiple countries with consistently positive results.

Research question

"Can hiring young community women to tutor struggling students improve learning outcomes at low cost?"

Methodology

Intervention

Schools were randomly assigned to receive a 'balsakhi' — a young woman from the local community hired to provide targeted instruction to students in grades 3 and 4 who were falling behind in basic literacy and numeracy. Balsakhis worked with small groups of 15–20 struggling students during regular school hours.

Assignment

School-level randomized controlled trial; 122 schools in Vadodara and 77 schools in Mumbai randomized to treatment or control

Sample size

Approximately 14,000 students across the two cities

Primary outcome

Math and verbal test scores at end of program year and one year after program ended

Effect estimate

0.14 SD improvement in Vadodara after one year; 0.28 SD in Mumbai; effects persisted two years after the program ended even for students who had moved to other schools

Decision

Results influenced the design of remedial education programs across India and contributed to Pratham's Teaching at the Right Level (TaRL) methodology, now implemented in multiple countries; program model adopted in Kenya, Ghana, and elsewhere with similar positive results

Result

Positive

0.14 SD improvement in Vadodara after one year; 0.28 SD in Mumbai; effects persisted two years after the program ended even for students who had moved to other schools

Evidence strength

Strong

Randomized trial, replicated across multiple sites or studies.

Replication status

Replicated

Institution

MIT Poverty Action Lab (Banerjee, Cole, Duflo, Linden)

Location

Vadodara and Mumbai, India

Year

2002

Policy area

International Development

Mechanism

Human capital

Related experiments

Tennessee, USA · 1989

Project STAR — Small Class Size

Positive

Wales, UK · 2007

Universal Free School Breakfast Program

Positive

Ypsilanti, Michigan, USA · 1962

Perry Preschool Project

Positive

Mentioned in the Dispatch

Issue 13 · June 9, 2026

The land question.

What happens when farmers get formal title to land they already farm — and why the answer is more complicated than property rights economists expected.

← Back to registryWhat the evidence shows →