On this page
Can a small bonus, paid only when children learn more, make government teachers teach better? Karthik Muralidharan and Venkatesh Sundararaman tested that in rural Andhra Pradesh, publishing "Teacher Performance Pay: Experimental Evidence from India" in the Journal of Political Economy in 2011.pdf) (volume 119, issue 1, pages 39-77).
How was the Andhra Pradesh study designed?
The study was a randomised controlled trial in government-run rural primary schools (Classes 1 to 5), part of the Andhra Pradesh Randomized Evaluation Study run with the state government, the Azim Premji Foundation and the World Bank (published paper, author copy.pdf)).
Sample: 300 schools drawn to represent rural Andhra Pradesh, with 100 offered group bonuses, 100 offered individual bonuses and 100 left as the comparison group.
Bonus rule: Rs. 500 for every percentage point of average test-score gain above 5%, in maths and language.
Size of the reward: the average bonus was set at around 3 percent of a typical teacher's annual salary.
Testing: independent tests designed by Education Initiatives, with "mechanical" questions (rote) and "conceptual" questions (understanding), plus science and social studies tests that carried no bonus.
Two more sets of 100 schools got an extra contract teacher or a cash grant instead, to compare incentives with inputs of similar cost.
What did teacher performance pay do to test scores?
After two years, students in incentive schools scored 0.27 standard deviations higher in maths and 0.17 standard deviations higher in language than students in comparison schools (Muralidharan and Sundararaman, 2011.pdf)). The average effect across both subjects was 0.22 SD.
Other findings from the same paper (author copy.pdf)):
Gains showed up on both mechanical and conceptual questions, which suggests real learning rather than drilling.
Students also scored 0.11 SD higher in science and 0.18 SD higher in social studies, subjects with no bonus attached.
Individual bonuses beat group bonuses by year two: 0.28 SD against 0.15 SD.
Input schools gained 0.09 SD against 0.22 SD for incentive schools, although the bonuses cost around 25 percent less than the inputs.
What changed inside the classroom?
Surprisingly, observers sitting in classrooms saw no difference in teacher or student attendance, or in visible teaching activity. The difference showed up in teacher interviews. Asked unprompted how they prepared for the end-of-year tests, 42% of incentive-school teachers mentioned extra homework against 20% in comparison schools, and 30% mentioned practice tests against 14% (paper, table 9.pdf)). The authors read this as more effective effort by teachers already present, some of it (after-school classes, practice tests near exams) outside what observers could see.
What does the study not show?
It is about primary school. Classes 1 to 5 in rural government schools, not Class 11 physics.
Some evidence is self-reported. The homework and practice-test differences come from interviews, not observation.
Results elsewhere are mixed. The paper itself notes a Kenyan trial by Glewwe, Ilias and Kremer where scores rose during the bonus programme but students did not keep the gains afterwards.
Design matters. The bonus depended on all students' average gain, tests were run by outside teams and graded centrally, and the working paper version says the programme was framed as "recognition" of good teaching rather than accountability.
What happened when the programme kept running? In a five-year follow-up (a 2012 working paper), Muralidharan reported that students who entered in Class 1 and spent all five primary years in schools where teachers were paid these bonuses scored 0.54 SD higher in maths and 0.35 SD higher in language than control students (NBER Reporter, 2018). That measures continued bonuses, not gains surviving after bonuses stop, and independent replication in other Indian states would strengthen the case.
What can Indian students and families take from it?
This part is our inference, not a finding of the study. The trial suggests clear, measured targets moved learning, and the extra effort looked ordinary: homework, practice tests, attention to weaker students. You can borrow the structure for yourself:
Measure a baseline. Take one timed chapter test now and note your score.
Set a gain target, not a final mark. "Five points better than last month" mirrors the study's improvement-based bonus.
Make it individual. In the study, individual teacher bonuses beat group ones by year two; for your own practice, track your own number even in a study group.
Test for understanding. Include "why" questions, not only formula plugging.
Keep reward as recognition. Parents, praise the effort and progress rather than paying per mark; our explainer on Deci and Ryan's self-determination theory covers why control-style rewards can backfire.
For weekly practice tests, the Fermi Problem Library has worked problems with a step-by-step solver that explains the reasoning rather than printing an answer.
For more Indian school trials, read what the Teaching at the Right Level trials found and what the Mindspark study in Delhi found about adaptive software.



