Data suggests a steady increase in the share of students attending private schools in rural India. This column analyses data from the largest and longest evaluation of a school choice programme in India, conducted over four years in the state of Andhra Pradesh, to assess whether private schools are indeed more effective than government schools.
The recently released Annual Status of Education Report (ASER)1 on rural education in India contains two main findings. First, learning levels among primary school age children in rural India continue to be shockingly low despite a steady increase in education spending under Sarva Shiksha Abhiyan (SSA) and the Right to Education (RTE)2. Second, there has been a steady increase in the fraction of parents abandoning free government schools in favour of fee-charging private schools, with the share of private school enrolment in rural India increasing from 19% in 2006 to 29% in 2013. While reliable annual data does not exist for urban India, the private school share in urban India was estimated at 58% in 2005 (using the Indian Human Development Survey (IHDS) data), and is likely to be considerably higher in 2013.
The ASER report shows, as do other data sources like IHDS (Desai et al 2009) and Young Lives (Singh 2013), that students of comparable age and standard in private schools score significantly higher than their counterparts in government schools. However, this doesn’t necessarily mean that private schools caused the better performance of their students. First, students in private schools are more likely to come from socio-economically and educationally advantaged backgrounds. Second, they also typically have 1-2 years of extra schooling (Lower Kindergarten (LKG)/ Upper Kindergarten (UKG)) compared to students in government schools. So, the better performance in private schools could simply reflect these other factors and not the actual effectiveness of the schools. Thus, a critical open question for education in India is this: “Are private schools more or less effective than government schools – holding all other factors constant?”
The Andhra Pradesh ‘School Choice’ Study
Answering this question is crucial for policy. Clause 12 of the RTE requires private schools to reserve 25% of their seats for students from Economically Weaker Sections (EWS), with the government reimbursing private schools for their fees (up to a maximum of per-child spending in government schools). If public money is going to be used to fund private schools, we need to understand whether private schools are more (or less) effective than government schools after holding all other factors constant.
The Andhra Pradesh School Choice Project, a long term research study covering over 6,000 students in 180 villages for four years (2008-2012), was designed to precisely and credibly answer this question3. Under the project, we invited parents of children in rural government schools to apply for vouchers (scholarships) which would cover all the costs (fees, books, uniforms) for their children to go to any private school of their choice in the village. The project offered the voucher to two cohorts of students starting in class 1 and class 2, and committed to providing the voucher till the completion of primary school (class 5).
The key design feature that enabled a statistically-valid comparison between government and private schools was that the scholarship was offered by lottery to a randomly-selected subset of applicants. This lottery-based selection thus created a treatment group (those who got the vouchers) and a control group (those who did not) – who were identical, on average, on all other socioeconomic characteristics and previous school experience except for winning the lottery. Thus, any differences in education outcomes between the treatment and control groups over time can be attributed purely to the change in schooling environment (made possible by the voucher) and will not be confounded by other factors.
What do private schools do differently?
We find that private school teachers have lower levels of education, training, and experience; and are paid much lower salaries (on average, less than one sixth of government teacher salaries). However, they have much better measures of effort and time-on-task (lower rates of absence, more likely to be actively teaching and to be in control of the class, when measured during surprise visits to schools). Private schools also have a longer school day and year, significantly lower pupil-teacher ratios, and much lower rates of multi-grade teaching (the lower teacher salaries allow them to hire significantly more teachers per student). Private schools are also more likely to have functioning toilets (for boys and girls) and scored better on measures of school sanitation and hygiene. Overall, we find that private schools are worse than government schools on input-based measures of teacher quality, but that they do much better on measures of teacher effort and active teaching.
Are private schools more effective?
In Telugu and Maths, we find that lottery winners who went to private schools don’t perform any better than lottery losers. However, we also found that private schools spend much less instructional time on Telugu (40% less time) and Maths (32% less time), and use this extra time to teach more English, Science/ Social Studies (EVS), and also Hindi as a third language (which is not taught in government schools). We find positive effects of vouchers on test scores on all of these subjects (large and significant for Hindi). Thus, adjusting for instructional time, we see that private schools are more productive because they are able to deliver equivalent outcomes as government schools on Telugu/ Maths even with substantially less instructional time, and used the extra time to deliver better outcomes on other subjects (especially Hindi).
We also find suggestive evidence of important differences in impact by medium of instruction of private school attended. In particular, going to an English medium private school led to worse scores on Telugu, Maths and EVS, but much better scores on English/ Hindi (relative to staying in a government school). However, going to a Telugu medium private school led to better outcomes on all subjects relative to staying in a government school (but less than English medium schools in English/ Hindi). These results suggest that switching medium of instruction may hurt accumulation of content knowledge (Maths/ EVS) for EWS students. This is consistent with evidence from cognitive neuroscience that first-generation learners are best off being taught in their native language (which allows for reinforcement at home), with English being taught as a subject. The results also suggest that private schools may be even more effective when the medium of instruction is not disrupted. These results though are only suggestive and a lot more research is needed in this area.
Finally, it is important to highlight that the average cost per student in the private schools was only one-third of the per-child spending in government schools. Thus, even though private schools were not more ´effective´ in improving learning outcomes in the main subjects of Maths and Telugu, they were clearly more ´productive´ than government schools, delivering similar outcomes at a much lower cost per student.
Implications for policy
These results suggest that both sides of the public versus private debate need to exercise caution in their recommended policy approaches to improve education quality in India. Advocates of private schools need to confront the fact that lottery-winners did not learn more in Maths and Telugu compared to the lottery-losers (suggesting that most of the observed differences between public and private schools reflect socio-economic factors). Given the abysmally low levels of overall literacy and numeracy, they also need to recognise that increasing the share of private-schooling, in its current form, is unlikely to solve India´s education quality problem. While private school teachers are clearly more accountable and work harder than their public school counterparts, it is possible that the binding constraint to education quality lies elsewhere - including a mismatch between the curriculum and student learning levels, and an education system that disproportionately values top performing students and does not care about helping low-performing students to achieve functional literacy and numeracy and realise their full potential. These systemic pathologies afflict both public and private schools and an excessive focus on private schooling as a panacea may divert attention from these fundamental issues of pedagogy and learning.
At the same time, the verdict on government schools is even worse. Empirical research on public education in India has clearly shown that increasing inputs (including teacher qualifications, training and salary) has had no impact on learning outcomes. So advocates of pouring more resources into government schools need to confront the fact that private schools are able to achieve equal or superior outcomes using teachers who are less qualified, and paid much less, suggesting that better management, and greater teacher accountability and effort can more than compensate for lower qualifications and salaries. In other words, private schools may have a pedagogy problem, but public schools have both a pedagogy problem and a governance problem.
Since private schools achieved equal or better outcomes at one-third the cost, the fundamental question that needs to be asked is "How much better could private management do if they had three times their current level of per-child spending?" Thus, in addition to focusing on improving the effectiveness of government schools at the current level of spending, the results suggest that policymakers should be open to experimenting with models of education provision with public funding (to ensure universal access) and private provision (for better school management).
Overall, policy discussions need to move away from debates of ´public´ versus ´private´ provision of education, which are (a) too simplistic because averages hide enormous variation within both public and private schools, and (b) not very useful because both systems are unlikely in their current form to deliver significant improvements in outcomes. Rather, the focus should be on the design of better education ´systems´ that aim to deliver superior outcomes by leveraging the strengths of both the public and the private sector while mitigating the weaknesses of the other. Clause 12 of the RTE provides the ideal context in which to have this discussion of education systems.
In a follow-up column next week, the author will outline a suggested implementation protocol for Clause 12 of the RTE.
- Sarva Shiksha Abhiyan (SSA) is Government of India´s flagship programme for achievement of Universalisation of Elementary Education in a time bound manner, as mandated by the Right to Education (RTE) – the 86th amendment to the Constitution of India making free and compulsory education to the children of 6-14 years age group, a Fundamental Right. SSA is being implemented in partnership with state governments to cover the entire country and address the needs of 192 million children in 1.1 million habitations.
- The study was carried out under my technical leadership by the Azim Premji Foundation in partnership with the Government of Andhra Pradesh, with financial support from the Legatum Institute, and the World Bank. All findings presented here represent the analysis and synthesis of results by Prof. Muralidharan, and incorporate several additional analysis and robustness checks based on comments from several global experts in the economics of education. But they do not represent the official views of any of the organisations named above, and have not yet been formally peer-reviewed. The discussions of policy implications are all the personal opinions of Prof. Muralidharan based on synthesising a decade of field work and research on education in India.
- Desai, Sonalde et al. (2009), ´Private Schooling in India: A New Educational Landscape´, in Suman Bery, Barry Bosworth, and Arvind Panagariya (eds.), India Policy Forum, 5, Sage, New Delhi.
- Muralidharan, K and M Kremer (2008), ´Public and Private Schools in Rural India´, in Paul Peterson and Rajashri Chakrabarti (eds.), School Choice International, MIT, Cambridge.
- Muralidharan, K and V Sundararaman (2013), ‘The aggregate effect of school choice: Evidence from a two-stage experiment in India’, NBER Working Paper Series, Working Paper 19441.
- Singh, Abhijeet (2013), ´The Private School Premium: Size and sources of the private school advantage in test scores in India´, Oxford.