Do private schools really produce more learning?
It is undisputed that private school students perform better on a range of tests than government school students. It is also undisputed that probably a substantial portion of this difference is accounted for not by the schools but by the type of households these children come from and their greater socio-economic advantage – in short, if the private schools had to teach the students currently in government schools, they might not be able to do better either!
This has been the key question on which quantitative research on private schools has focused on in India – and it seemed that, to the extent possible in data on students at one point of time, similar children in private schools scored better than their peers in government schools even accounting for those of the background characteristics that were observed in the data. And yet consensus was difficult – who could say whether the data observed everything that differed between private and government school students?
This impasse changed this year with two studies which significantly advance the robustness of our knowledge in this area. In a paper published in the Journal of Development Economics this year, ungated version here, I used panel data on two cohorts of children in Andhra Pradesh and Telangana to study whether the amount children learnt in private schools (in technical terms, the “value-added” of private schools) differed from that in government schools. The basic idea is simple: to the extent that unobserved factors are already reflected not only in a child’s test scores now but also his test scores from earlier years, we can get much more robust answers to this question by studying the amount children learn over time in government vs. private schools rather than what they know at any one point – these models, called “value-added models”, have been shown to replicate the same results as experiments and quasi-experiments in several recent projects [click each link!].
In a nutshell, I found that in the core academic subjects – Mathematics and Telugu – private school students in primary grades were not learning any more than government school students. But they were learning a lot more English – and that is why parents said they were sending children to private schools in the first place. In Telugu medium-schools, these extra English skills were gained without any loss in core math or Telugu skills. At secondary school ages, there’s somewhat more evidence of greater learning in math and Telugu – but the gains are modest, and certainly a fair distance from where any absolute expectation of what we expect children to know at the end of secondary schooling. I also don’t find any evidence that private schools increase students’ “psychosocial skills”, specifically the extent to which they believe they can shape their future outcomes through their own efforts (agency) or their belief in their own abilities (self-efficacy). These results agree very closely with results from a paper by Karthik Muralidharan and Venkatesh Sundararaman, also coincidentally in AP at the same time and also published this year, thus boosting the confidence in the results from both studies even more; see here for Karthik’s summary of his work.
So the basic message seems quite clear: private schools do seem to produce more learning than government schools but, with the exception of English, these differences are not very large.
What about the effect of private schooling on the inequalities in learning?
Concerns about whether private schools cause higher achievement is only a part of the story. To the detractors of private schools, the bigger concern is about whether they entrench socio-economic advantage by providing greater chances to those from more advantaged backgrounds to begin with. It is clear that students in private schools tend to be from richer backgrounds, urban and with a lower proportion of children from disadvantaged castes. The differences are, though, perhaps less muted than most observers assume: while 30% of students in government schools are first generation learners, so are 20% of private school students(see p.15). In thinking of Indian private schools, just as in thinking of other aspects of the education system such as the syllabus, we seem to think disproportionately of only the top end. In some dimensions, most notably gender, the gap in private school attendance seems to have widened over time while in others, such as the rural-urban gap, it has narrowed in recent years. Still, it remains very much the case that access to private schools is far from equitable.
Does this matter for the outcomes of children?
A leading argument against segregation of students by socio-economic status or other markers such as race and caste has been that being seated in the same classrooms, studying together, has independent learning value and teaches the students valuable social skills. In a particularly important contribution, much less well known than it deserves, Gautam Rao at Harvard University looked at the effect of bringing in students from poorer backgrounds (“Economically Weaker Sections”) in Delhi to elite private schools on both the rich and the poor students. He finds that having poor classmates makes wealthy students more prosocial and generous, more likely to volunteer for charities in schools, and less likely to discriminate against poor children. For any educational system, these are important outcomes. And this comes without an increase in classroom disruption, or lower test scores in math or Hindi: the only sign of a trade-off at all is a modest (and only marginally significant) decline in English scores.
So, in sum, the empirical evidence thus far on stratification does seem to indicate that there are good reasons to be concerned: the social and economic inequalities in access exist, in some dimensions they may be increasing, and they have important effects on the non-academic (but crucially important) outcomes of children.
The unexplored (and unevaluated) potential of the private sector
The existing evidence on the private sector, as it exists above, is not too optimistic: there are gains in absolute achievement, although often none in some core academic domains, and there are serious concerns about the effect of the inequality in access to private schools which is undesirable in itself but also, as Rao’s study points out, with potentially large negative effects on non-academic outcomes that we hope an education system will deliver. This does not, however, mean that the private sector has little to offer by way of improving the dire learning crisis in Indian schools. What are the ways in which it might do so?
One mechanism, previously written about by Karthik Muralidharan, looks at how Clause 12 of the Right to Education Act may be much better designed to maximize the chances of improving both test scores and equity within the current system and the current legal framework. This is, of course, an important (and feasible) reform that deserves close attention.
But the private sector in education, including all non-state providers of education, might well be able to offer much more. The most unrecognized role of the private sector, but perhaps the most important, is as an engine of innovation. Thinking of the sources of knowledge about learning levels in India in the past decade, and how to improve them at scale, two of the most significant advances have been (a) the documenting, year-after-year, of the immense and stubbornly persistent learning deficits by the ASER reports and (b) the potential for unqualified volunteer teachers to increase learning significantly for children lagging behind the curriculum. The initial impetus on both these fronts came from civil society and non-state actors, especially the PROBE Report and the extensive work done by Pratham. Nor are NGOs and civil society groups the only possible source for such innovation. Various outlets have written about the intriguing example of the Bridge academies in Kenya which with a heavily-scripted curriculum aims to standardize pedagogy and provide the teaching methods support that most teachers have not received – whether because they never did get teaching qualifications or, as likely, the teaching qualifications just did not prepare them for the practical aspects of classroom management, lesson planning and day-to-day pedagogical practice. There are also outfits in India which are engaged in curricular reform, even within the current curriculum: one which I came into contact with in the last year is the XSEED program and there are probably many others I am unaware of.
How can these innovations from the private sector help transform Indian education, given that non-state organizations themselves are unlikely to reach the scale of the public education system and, in the case of the commercial private sector, suffer from the same selection on ability to pay as the private school sector itself?
At least three models spring to mind.
The first is the model of Private Public Partnerships, akin to Charter Schools in the US, where the government pays for the students at the rate in government schools and ensures admissions are non-selective but private operators decide how the money is spent and how teaching and classrooms are organized. Although often controversial, this model is already being considered or experimented upon by various state governments (see e.g. this school in Delhi operated by Ark India or the draft policy in Rajasthan). Experimenting with this model can help answer two key questions that we do not yet have the answers to: first, we know that the private sector can produce similar gains in learning to private schools at a third the total cost but can it translate higher funding into much better results? And second, what is the effect of inserting a high-quality education provider into the market for education — does it lead to an improvement also in other schools through competition?
The second is the model where innovations are developed in the private sector, enabled by much greater decentralization, but then adopted into the public sector. An example here is Pratham’s “teaching at the right level” approach developed through their extensive work on remedial education but now also extended to government schools. Whether all such knowledge transfers will work in government schools remains, of course, an empirical (and evaluable!) question since it may be that the binding constraint of weak governance is too strong.
The third is a model where private providers supplement existing government schools. Across the country, a very large number of students go to after-school private tuition – so there clearly is a demand for this sort of instruction which is currently met by a large number of unorganized small-scale tutors with little causal evidence of impact. It is in this space of supplementary and remedial instruction that larger-scale private providers could significantly raise quality by promoting suitable pedagogical approaches – by e.g. the Balsakhi program promoted by Pratham or computer adaptive learning tools (such as the Mindspark centres in Delhi run by Educational Initiatives).
Where does all of this leave us?
The existence and the rapid proliferation of the low-cost private schooling sector provides a stark indictment of the current state schooling sector. As Karthik Muralidharan memorably summarized: “What does it say about the quality of your product that you can’t even give it away for free”?
That said, at least on the basis of the evidence we now have, we certainly should not expect the private sector alone to provide a substantive solution to the ‘learning crisis’ in our schools in terms of the basic abilities to read, write and do simple mathematics: while private sector schools are certainly more productive, by producing the same learning gains at a fraction of the total cost spent per student in the state sector, the absolute increment in these fundamental skills is too small compared to the enormous gap between students’ actual achievement and any objective standards of quality we expect a functioning school system to deliver. And it is prudent, even when documenting the greater productivity or the modest absolute gains in the private sector, to worry about the effects of a stratified schooling system. Vouchers and other means to enable poorer students to access private schools may help keep inequality of access in check but a lot depends on the precise design of policies (and whether they can be manipulated): past experience on voucher schemes haven’t always been successful in this regard (as the Chilean example shows).
The private sector, through greater innovation and greater nimbleness than the government schooling system, might well provide the pedagogical innovations needed to address the incredibly low productivity of Indian schools. Perhaps through PPP models and voucher schemes, it can also demonstrate the true potential of learning levels that current per-student spending in the state sector could produce. And through supplementary and remedial education, it might mitigate the effects of a failing school system.
Eventually, however, solving the learning crisis comprehensively will require substantive reform in the government schooling system – much as we might wish to ignore the elephant in the room of weak governance in government schools, that probably still is where the most promising reforms lie.
An edited version of this blog was first published on Ideas 4 India on 28 October 2015.