from EVIDENCE to POLICY Learning what works for better programs and policies June 2019 PAKISTAN: What’s the best way to invest in private schools? Private schools that cater to low-income students are popular to determine how best to structure cash grants to support ru- with parents seeking alternatives to government schools. But ral private schools serving low-income students.* The team these private schools, which are often owned by local entre- tested two models: In one set of villages, all private school preneurs, may lack the resources and incentives to expand en- owners were given a grant; in another set, only one private rollment or improve quality. They generally operate in markets school was targeted in each village. In both groups of villages, where access to credit is limited and where there aren’t loan schools increased spending on school infrastructure relative products tailored to their to the control group. However, test scores increased only needs. This means that among students who were in private schools in villages where any improvements have all the private schools received the grants. In these schools, EDUCATION to be financed through the owners increased teachers’ wages too, which may have school fees or their own helped them attract better teachers. They also raised school funds. When donors and fees, which helped increase their revenues, while schools that investors step in to pro- were the only ones in the village to receive the money gener- vide support to private ally relied on raising enrollment to increase their revenues. schools, they tend to fo- The results indicate that improving access to finance may cus on larger operators help schools make improvements that benefit students, while with a chain of schools, which typically implies selective fund- also raising their own profits through increases in either en- ing to a limited number of schools rather than broad support rollment or fees, or a combination of the two. The research to the schooling market. Is this the best way to support private team has now partnered with a major microfinance institute schools that cater to poor families? Could supporting the entire in Pakistan, Telenor Bank, and others to develop and evaluate market, rather than select schools—or chains—lead to more the impact of new loan products for private schools that cater competitive pressure to invest in quality improvements that to low-income students in Punjab. Among other things, the promote students’ learning? And is a market for loans for pri- research team will measure whether making loans available vate schools sustainable? leads schools to make changes that improve student learning In Pakistan, a research team supported by funders that and also raise revenues so they can repay the loans. included the Strategic Impact Evaluation Fund (SIEF) sought Context About one-third of all children of primary school age in Paki- across all Pakistani schools lags behind what is in the curricu- stan attend private school, and most of these schools keep their lum, students in these schools tend to outperform their peers fees low and cater to poor students. Schools hold down costs iw government schools. By grade three, the test scores of stu- largely because they hire as teachers local, unmarried young dents in these private schools put them ahead by 1.5 to 2.5 women with a secondary school education.** While learning years of learning. *This study is registered in the AEA RCT Registry with the unique identifying number **Tahir Andrabi, Jishnu Das, Asim Ijaz Khwaja, “Students today, teachers tomorrow: AEARCTR-0003019. The research paper was funded also through grants from the Identifying constraints on the provision of education,” Journal of Public Economics, Aman Foundation, Templeton Foundation, National Science Foundation, and Research 100 (2013) 1–14 on Improving Systems of Education (RISE) with support from UK Aid and Australian Aid. In Punjab Province, where this study took place, private produce higher test scores at lower costs than government schools enroll close to 40 percent of children between the ages schools but they don’t show continual quality improvements of six and 10 years. Median school fees are under $2 a month, nor do they expand enrollment, based on data from the which is less than half the daily minimum wage in the prov- Learning and Educational Achievement in Pakistan Schools ince. These schools operate in competitive environments. In (LEAPS), an ongoing study of education in Punjab. Two- the area where the study took place, 64 percent of villages thirds of school owners surveyed through LEAPS reported had at least one private school, and many had a mix of up to that they would like to borrow to improve their schools, but five private and government schools. These private schools only 2 percent had school-related loans. Intervention and Evaluation The researchers tested the impact of two different approaches two government primary schools. Researchers randomly for making funding available: providing all private schools in a selected 266 of the villages for the experiment. The villages village with a one-time cash grant of Rs. 50,000 (about $500 were divided into two treatment groups and one control. In at the time), versus giving the grant to only one of the private 114 villages, researchers randomly assigned one private school EDUCATION schools in a village. The money was distributed in a minimum in each village to receive the cash grant; in 75 villages, all of two installments over a one-year period. The amount was 228 private schools were offered the grant; and in the control equal to about five months of profit and was enough to pay group of 77 villages, none of the 249 private schools received for two teachers a year or to buy an additional 25 to 100 desks a grant. The baseline survey was carried out in July 2012 and and chairs. There were no conditions attached to receiving the distribution of the money started later the same year, halfway money, although school owners had to submit a nonbinding through the 2012-2013 school year. There were five rounds business plan and they had to open a one-time use bank of follow-up surveys starting in May 2013, and then every account to receive the money. three to four months through November 2014 focusing The intervention took place in Faisalabad district in on enrollment, fees, and revenues. Children were tested at Punjab and it covered all villages that had at least two non- baseline and then again 14 months later. The typical school governmental schools, which was 42 percent of the district’s year runs from April to March, with a three-month break in 786 villages. On average, each village had three private and the summer. Findings School owners used the money on schools: All The first year of the program, all of the school owners school owners who received the cash grants who received the grants spent between 39 percent and 44 increased spending on their schools, with more than percent more than the control group on fixed costs like room half the money spent on desks, chairs, and other upgrades, new furniture, school infrastructure, and education infrastructure improvement. materials. This increase in investments amounted to 70 This policy note is based on “Upping the ante: the equilibrium effects of unconditional grants to private schools,” Tahir Andrabi, Jishnu Das, Asim Ijaz Khwaja, Selcuk Ozyurt, Niharika Singh, World Bank Policy Research Working Paper, 2018; and forthcoming at AEJ: Economic Policy. percent of the total grant amount in the villages where all while in villages where all schools received the grant private schools received the cash grants and to 61 percent in the rise in revenues was driven by an increase in the villages where only one private school received the money. school fees. How schools spent the money differed depending on Revenues increased in all schools that received grants, indicat- if they were the only school in the village to get the ing they either expanded the number of students they en- money or one of many schools: The real difference rolled or raised fees, or did a combination of the two. After was in spending on teachers—schools in villages two years, schools that were the only ones in the villages to where all private schools received grants increased receive the grants had an average of 22 more students than teachers’ salaries. the control group, an increase of around 13 percent. These schools were also nine percentage points less likely to close Spending on teachers’ salaries and other variable expenses down during the study period. When all schools in a village rose by 12 percent when all private schools in a village re- received the grant, they were not any less likely to close when ceived grants. The increase was driven by higher salaries compared with the control group and enrollment in each for teachers and not an increase in the number of teachers. school rose by an average of nine students. These schools were more likely to have hired new teachers, and they paid higher salaries for their teachers. This suggests that schools were trying to attract better teachers to raise the quality of the schools, consistent with prior research in Pakistan linking students’ test scores and teachers’ wages.*** In schools that were the only ones in their villages to receive the grant, variable monthly spending didn’t increase signifi- cantly. Student learning improved only when all private schools in a village received cash grants. Researchers tested the students in math, English, and the local language, Urdu, at baseline and 14 months after the start of the intervention. In villages where all private schools received grants, the test scores of children in the private schools rose by 0.21 standard deviations, which corresponds to an additional six months of learning over two years. Test score gains remained when researchers restricted their sample to those students who had been in the schools before the grants, ruling out gains because the composition of the students in these The returns on these grants were remarkably high. In villages schools had changed. However, students in the schools that where only one private school received the grants, the authors were the only ones in the villages to receive a grant showed no compute returns on investment between 61 percent and 83 statistically significant test score improvements. percent and in villages where all private schools received the grants, the returns were lower at 12 percent to 32 percent, but All schools that received cash grants showed an still exceptional compared to typical programs in this sector. increase in revenues. In schools that were the only ones in the village to receive a grant, revenue School fees, however, only went up when all private increases were driven by increases in enrollment, schools in the village received grants. ***Natalie Bau and Jishnu Das, “The Misallocation of Pay and Productivity in the Public Sector: Evidence from the Labor Market for Teachers,” World Bank Work- ing Paper, WPS8050 (2016) The investments in learning—­ particularly, the increased when only one school in a village received money, revenues spending on teacher salaries—allowed schools in villages rose because enrollment increased; in villages where all pri- where all schools received the grant to increase their fees. Av- vate schools received the money, revenues rose primarily erage monthly tuition fees were about Rs. 19 higher in these from charging higher fees. The fact that schools increased schools than in control schools, or an eight percent hike. their overall expenditures despite the grant being essentially There was no increase in school fees in schools that were the unconditional suggests that school investments offer better only ones in the villages to receive the grant. In other words, returns relative to other investment options. Conclusion In some low- and middle-income countries, private schools schools increased their investments relative to a control that target poor families are used to fill gaps in government group, focusing mainly on infrastructure, and their revenues services or are attracting families that believe the quality is rose. In schools in villages where every school was offered better than in government schools. While government-run a grant, teachers’ salaries also increased, as did student education systems can rely on public financing to improve learning. In villages where only one school received a grant, services, private providers’ access to financing is more schools increased enrollment but not student learning. The EDUCATION limited. This evaluation sought to test the effectiveness of results indicate that opening up credit could be an effective offering unconditional cash grants to small private schools route for improving small-scale private schools. Currently, in rural Pakistan and measured the effects on enrollment, the SIEF-supported research team is testing the impact of school investments, school revenues, and student learning. offering loans to private schools in Pakistan that cater to The evaluation also explored whether it mattered if one low-income students, to see whether they generate similar school in the village received the money or if all schools impacts as the unconditional cash grants. did. The evaluation found that with both types of grants, 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 low and middle income 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 and the London-based Children’s Investment Fund Foundation (CIFF). THE WORLD BANK, STRATEGIC IMPACT EVALUATION FUND 1818 H STREET, NW, WASHINGTON, DC 20433 Series editor: Aliza Marcus