Can Performance Pay for Teachers Improve Student Learning in India?
Educating India’s Children
With around 200 million children aged six to 14, which is nearly two-thirds the size of the entire United States population, India has the largest primary school system in the world. While India’s primary-school enrollment rates have significantly improved — more than 95 percent of appropriately aged children are enrolled in school — the actual learning level of these students is still dismal. More than 60 percent of enrolled students do not even read at a second grade level (Pratham 2012).
Figuratively speaking, throwing money at the problem is a common response to low levels of education and learning in developing countries, and the government of India has substantially increased spending on primary education in the past decade. Unfortunately this increase did not translate into better-educated children. In some of my previous collaborative research, we show that a central challenge in translating increased spending into better-educated children is limited teacher motivation and accountability. On any given day, 25 percent of teachers in Indian public schools are absent and that less than half were actually teaching when observed during unannounced visits.
Motivating teachers is clearly a key education-policy imperative. Following the dissemination of the results on teacher absence, the education secretary of the Indian state of Andhra Pradesh (AP) struck out to implement a performance-based pay structure that would recognize and reward highly effective teachers. Evidence on the impact of this program was presented in a recent paper co-authored with Venkatesh Sundararaman of the World Bank.
Experimental Design and Research Questions
The pay-for-performance program is distinct because it was evaluated using a rigorous randomized evaluation. Schools were selected by lottery to participate in the program; schools selected for the program (treatment) and those not selected (control) are identical on average on all characteristics. This structure means that any difference in outcomes from these schools over time can be attributed to the pay-for-performance program. The experiment was carried out in a large representative sample of 300 rural public schools in AP. The 300 schools were divided evenly among a group teacher-incentive program, an individual teacher-incentive program, and a control group.
This experimental design allowed us to answer a comprehensive set of questions, including 1) Can performance pay for teachers based on students’ test scores improve student achievement? 2) What, if any, are the negative consequences of teacher incentives based on student test scores? 3) How do school-level group incentives compare with teacher-level individual incentives? 4) How does teacher behavior change in response to performance pay? and 5) How cost effective are teacher incentives relative to other uses for the same money?
Pay-for-performance creates better-educated students who out performed students in control schools in all five grades, all districts and all levels of test question difficulty.
Our main finding was that the teacher pay-for-performance program was quite effective in improving student learning. At the end of two years of the program, students who attended the incentive schools performed significantly better than those in control schools by 0.27 and 0.17 standard deviations in math and language tests respectively (these correspond to improvements of 11 and 7 percentile points for a median student in math and language). We also found that the gains are very broad based, with students in incentive schools out-scoring the students in control schools in all five grades, all districts and all levels of test question difficulty.
There are no adverse consequences as a result of the pay-for-performance programs and no evidence of teachers trying to game the system.
We found no evidence of any adverse consequences resulting from incentive programs. In fact, incentive schools did significantly better on mechanical components of the test that were designed to reflect rote learning, and conceptual components of the test that were designed to capture deeper understanding of the material. This suggests that the gains in test scores represent an actual increase in learning outcomes as opposed to merely reflecting teaching to the test. A positive spillover effect was also suggested because students in incentive schools did significantly better not only in math and language, for which there were incentives, but also in science and social studies, where there were no incentives. We also found no evidence of any adverse gaming of the incentive program by teachers.
Teachers start teaching with financial incentives.
We found that school-level group incentives and teacher-level individual incentives performed equally well in the first year, but the individual-incentive schools outperformed the group-incentive schools after two years of the program. We measure changes in teacher behavior in response to the program with teacher interviews as well as direct physical observation of teacher activity. Our results suggest that the main impact of the incentive program was not increased teacher attendance but greater (and more effective) teaching effort conditional on showing up for work.
Performance-based bonuses effectively increase test scores and are more cost effective when compared to similar amounts of spending on education without performance conditions.
We found that performance-based bonus payments to teachers were a significantly more cost effective way of increasing student test scores compared to spending a similar amount of money unconditionally on additional schooling inputs. For example, in a parallel set of studies, two other sets of 100 randomly chosen schools were provided with an extra teacher as well as a cash grant for school materials. We find that while students in these schools also did better, the students in the schools with the teacher-incentive programs did markedly better than those in input schools (with the incentive program being three times more cost effective).
This work — along with the experimental evaluations of other interventions to improve education outcomes in the same setting — has had considerable impact on policy over the past couple of years. The results have been featured in op-ed columns in leading newspapers in India and international news magazines such as The Economist. The World Bank has written policy notes for governments on the basis of this work. I have given several presentations to senior policymakers in the government of India and senior staff at international agencies like the World Bank and the British aid agency (DFID), as well as public lectures in India and the United States. We also made a documentary that interviewed teachers about the program to showcase their experiences with it. Overall, this research is seen as both academically rigorous as well as relevant to policy in developing countries, which is a feature I try to maintain in most of my ongoing research.
Karthik Muralidharan is an assistant professor of economics at the UC San Diego, where he has been on the faculty since 2008. Born and raised in India, he earned an A.B. in economics (summa cum laude) from Harvard University, an M.Phil. in economics from Cambridge University and a Ph.D. in economics from Harvard. He is a faculty research fellow at the National Bureau of Economic Research (NBER), a junior affiliate at the Bureau for Research and Economic Analysis of Development (BREAD), a member of the Jameel Poverty Action Lab (J-PAL) network, an affiliate at the Center of Evaluation for Global Action (CEGA), and a research affiliate with Innovations for Poverty Action .