from EVIDENCE to POLICY Learning what works for better programs and policies March 2016 INDIA: Do Kids in Private Schools Learn More? Primary school education is a basic building block for children’s dents—with perhaps more motivated parents who push them— development, preparing them for success later in life. But in many who are choosing these schools. countries, poor children often don’t finish school even if it’s avail- In India, where such fee-charging private schools are growing able to them. Those who in number and popularity, the Legatum Institute, the World Bank, do stay in school may not the British government’s Department for International Develop- learn much. The qual- ment, the educational non-profit Azim Premji Foundation, and ity of education can be the government of the Indian state of Andhra Pradesh worked to- so low that children end gether to evaluate alternatives for improving education and giv- up completing primary ing children more choices. One initiative, known as the Andhra school without learning Pradesh School Choice Project, sought to make private schools to read or do basic math. available to poor families by setting aside a number of spaces and One response to the subsidizing the full tuition costs. The main finding of the evalua- EDUCATION perceived low-quality of tion was that students who attended low-cost private schools did public schools has been as well as students in government-run schools on most subjects. the rapid growth of pri- But they also learned additional subjects, despite the fact that the vate schools that cater largely to the poor through low monthly cost of educating a child was 60 percent lower than the cost in fees, responding to demand from parents seeking what they be- government-run schools. The results are relevant for policy- lieve will be better opportunities for their children. The main makers in India and for educational experts everywhere who public policy concern is whether these schools really do deliver are facing questions and sometimes controversies about the better educational opportunities or whether it’s the better stu- value of supporting low-cost private schools. Context The public school system in India incorporates two extremes— reimbursing tuition up to the cost of a public school education. excellent access and very poor learning. More than 96 percent of This provision is only now taking effect, after the Supreme Court primary-school aged children are enrolled, putting India close upheld its constitutionality in 2013. to universal education. Yet less than 40 percent of children aged There has been a lot of debate in India about the implica- 6-14 can read at or above a second-grade level. Parents looking tions of government steps to increase access for disadvantaged for better educational opportunities for their children are increas- students to private schools. Opponents argued that, among ingly turning to private schools that promise higher quality. In other things, students in private schools would be negatively rural India, more than a quarter of children aged 6-14 are en- affected by an influx of poorly prepared students, and that rolled in private schools, while in urban India more than half of these students wouldn’t be able to keep up with the curric- the children are in private schools. ulum. Prior to the law’s being enacted, the Indian state of For the most disadvantaged families, even these private Andhra Pradesh, working with the Legatum Institute, the schools can be out of reach. The Government of India, concerned Azim Premji Foundation, and under the technical leadership that the poorest children don’t have access to these schools, in- of the World Bank, implemented and evaluated a four-year cluded a provision to boost access in the sweeping Right to Edu- research program that aimed to answer key questions regard- cation Act passed in 2009. Private schools were required to set ing the likely impact of the law’s provision for opening up aside 25 percent of their seats for the poor, with the government private schools to more poor students. Evaluation The project was launched in the 2008-2009 school year in were still in private schools. The voucher covered tuition and five districts in Andhra Pradesh, covering a total of 180 vil- school uniforms, books and other supplies. It didn’t cover the lages. The project, which focused on families with a child in cost of transportation if the school picked was outside the kindergarten or the first year of school, offered vouchers to village, nor did it include money in lieu of the free midday cover the child’s private schooling costs through the end of meal students in government schools gets. The value of the primary school (fifth grade). A two-stage randomization was voucher was set at roughly 40 percent of what government established in order to better measure the impact of the pro- schools spend to educate a child. The average tuition in these gram on voucher and non-voucher students by taking into private schools in the study was about a third of the cost to account possible spillover effects (for instance, students who educate a child in a government school. were in private schools to begin with might suffer from an Learning outcomes were measured through independent influx of less prepared classmates from public schools). student tests after two years and then after four years in math, The first lottery assigned entire villages into treatment English, and Telugu, the state’s native language and the lan- and control groups. A second lottery, carried out only in guage of instruction in the state’s public schools. At the end treatment villages, assigned families into groups that either of four years, tests were also given in science, social studies received or didn’t receive the scholarships. Fifty-nine percent and Hindi, which is the national language of India. Tests were of around 11,000 eligible households in the 180 villages given to all the students in treatment and control villages who EDUCATION applied. At that point, the villages were randomly divided had applied for the scholarship voucher, and to a represen- equally between control and treatment groups. In order to tative sample of students who either hadn’t applied or were distribute the vouchers, a second lottery was then held just for already in private school when the program started. In ad- households in the treatment group. Of the 3,097 households dition, household and school surveys were carried out each in the treatment villages that applied for vouchers, 1,980 were year. Data collectors also made unannounced visits to schools accepted and 1,210 took the scholarship and enrolled in a pri- to collect data on school processes and teacher effort, and also vate school. By the end of the four-year study, 1,005 students surveyed teachers and households. Results Students who received a voucher to go to a private don’t teach at all). Students also did much better than school didn’t do any better than students in government school students in Hindi, which private government schools in the main subjects of math schools teach as a third language. Private schools also give and Telugu language, but the private schools also students more time for computer use, arts, crafts, sports spent less time teaching these subjects and gave and study hall. Overall, private schools were not more students more time in other subjects. effective at improving test scores in math and Telugu, but they were more productive with their time, since they Students who attended private school with vouchers achieve similar outcomes with less time and use the extra didn’t do any better in math and Telugu than students time to teach more subjects. in the government schools. However, private schools spent less time teaching these two subjects and more Private schools weren’t only doing more, they time on other subjects including English, social studies delivered this education at a fraction of the cost of and science, and Hindi (which the government schools educating a student in a government school. This note is based on The Aggregate Effect of School Choice: Evidence from a Two-stage Experiment in India,” by Karthik Muralidharan and Venkatesh Sundararaman. Quarterly Journal of Economics, Vol. 130, August 2015, Issue 3. Generous financial support for this study was provided by the Legatum Foundation. The per student private school cost in the Andhra Pradesh school at the four-year mark. One difference was they didn’t de- sample—based on what the schools charged for monthly tu- vote as much time on homework as the typical private school ition—was less than a third of what the government spent to students. And while typical private school students spent 20 educate a student in its schools. Even after adding in the extra minutes less a day playing with friends, students who received money for a voucher winner’s payments for books, uniforms vouchers to go to private school played with their friends as and other supplies, it was still only about 40 percent of what much as before. it cost to educate a student in a public school. Private schools managed to deliver these savings mainly because they paid their Critics have also said that private school students teachers much less than government teacher salaries, on average might be negatively affected by an influx of pre- less than a sixth of what government teachers received. sumably poorer and less academically prepared voucher students. This didn’t happen. Teachers hired by the private schools had less experience, less education and less training, but The evaluation didn’t find any negative impact on students they were more likely to show up in the classroom, who were already in private school. This was the case regard- actively teach and keep their class under control. less of the number of voucher holders—up to the 25 percent limit—who enrolled. During unannounced visits by the survey team over the four-year period, private schools outperformed government schools when it came to observed quality of teaching. In private schools, teachers were more likely to be in the class- room and more likely to be actively teaching. Teachers also were more in control of their classes and more effective at maintaining discipline. Private schools were also less likely to group different grades together into one classroom and have one teacher teach all of them. In government schools, 24 percent of teachers were likely to be absent on any given day, compared with nine percent of teachers in private schools. Thirty-five percent of government teachers were likely to be actively teaching at any given point, compared with 50 percent of private school teachers. Private schools also offered a cleaner environment, which The evaluation was also able to measure whether the can reduce the transmission of disease. Private schools were departure of voucher holders from government schools for less likely to have garbage dumped on their grounds, they private schools had a negative impact on the remaining stu- had fewer swarms of flies and fewer pools of stagnant water. dents in government schools. Test results didn’t show any They also were more likely to have separate, functioning toi- negative impact. lets for boys and girls. Overall, private schools weren’t more effective Concerns that voucher students wouldn’t be able to in terms of improving the main subjects of math adjust to the new setting proved unfounded. and Telugu, but they were much more productive because they achieved the same results as Private school students spent more time than government stu- government schools but at much lower costs. dents in school and more time doing their homework—43 min- utes more a day in school, and 23 minutes extra on homework. A free education in private school was clearly worth it. Students Students who received vouchers were able to adjust to longer who used vouchers to go to a private school learned as much as school days and the overwhelming majority were still in private their public school counterparts when it came to the core sub- jects. They also learned more Hindi and had more time on hand, these low-cost private schools in their current form are other subjects. When it came to reducing social segregation, unlikely to significantly improve learning for disadvantaged vouchers helped do so without any negative consequences. students. On the other hand, they are much more cost effec- But private schools didn’t actually offer a better education tive, since they deliver the same learning at a much lower cost. in core subjects. Although the teachers in these schools ex- A next step for research and policy would be to experiment erted more effort—by showing up, teaching and keeping the with voucher and school-choice models where private schools classes under control—they were less experienced and hadn’t (or charter schools) are reimbursed at the same rate as the cost received the same training as government teachers. This may per child in government schools and aren’t allowed to selec- have affected their ability to boost learning. So on the one tively admit students. Conclusion Concerns that the program would have a negative effect on high-quality education to all children in an inclusive and eq- fee-paying students proved unfounded. Similarly, concerns uitable manner. A key open question for education policy that voucher students wouldn’t be able to keep up with the in low-income setting is to study the extent to which pri- work also proved unfounded. It turned out that the low- vate schools that have the same level of spending per child cost private schools were more productive than the govern- as government schools can improve learning outcomes ment schools in terms of offering more classes and teaching without selectively admitting students. EDUCATION core subjects in shorter periods of time, despite hiring less experienced teachers and paying them less than government school teachers. As the evaluation shows, vouchers don’t hurt students—neither those who receive them, nor those who are Private vs Public their new or old classmates—which means they can be an ef- • Private schools in the sample generally had longer school years, fective tool for expanding access across socio-economic lines lower pupil-teacher ratio and better facilities, such as working toilets (and separate toilets for girls), drinking water and electricity. and giving poor children the opportunity to be exposed to • Government schools were more likely to have a cupboard with the variety of classes the private schools offer. Nevertheless, books that students could borrow and a radio. such programs require careful attention to design to deliver The Strategic Impact Evaluation Fund, part of the World Bank Group, supports and disseminates research evaluating the impact of development projects to help alleviate poverty. The goal is to collect and build empirical evidence that can help governments and development organizations design and implement the most appropriate and effective policies for better educational, health and job opportunities for people in developing countries. For more information about who we are and what we do, go to: http://www.worldbank.org/sief. The Evidence to Policy note series is produced by SIEF with generous support from the British government’s Department for International Development. THE WORLD BANK, STRATEGIC IMPACT EVALUATION FUND 1818 H STREET, NW WASHINGTON, DC 20433 Produced by the Strategic Impact Evaluation Fund Series editor and writer: Aliza Marcus