Low-cost private schooling in the developing world context: new evidence

In the past decades, there has been a significant increase in the share of pupils attending private schools throughout the developing world. Many of these schools are profit-making, in stark contrast to the private schools that exist in most developed countries. Indeed, recently, Laura McInerney highlighted in the Guardian the government’s inconsistent approach to for-profit schools: whereas it maintains that such schools should not be allowed in England, it gladly sponsors them in developing countries.

And the private school revolution in these countries is indeed controversial, having been met with enthusiasm by some, who believe that it offers a solution to the failures of the state school systems in these countries, and with apprehension by others, who believe it weakens the state school system and exacerbates inequality.

Up until now, however, there has been little rigorous research on the impact of low-cost private schools in a developing world context. While non-experimental research often has suggested strong positive effects of private school attendance, there haven’t been any experimental evaluations. In a sophisticated paper - 'The Aggregate Effect of School Choice: Evidence from a Two-stage Experiment in India' - economists Karthik Muralidharan and Venkatesh Sundararaman report the findings from a large-scale randomised voucher experiment in the Indian state of Andhra Pradesh. The programme provided a voucher to finance pupils to attend low-cost private primary schools. Applicants were encouraged to apply, and receipt of a voucher was determined by lotteries. This ensures that, on average, covariates between lottery winners and lottery losers are balanced.

But the paper went farther than any voucher study has ever done. The experiment not only randomised receipt of vouchers among households; it also randomised the villages in which the programme would be conducted. This allowed the authors to analyse potential spill-over effects of the voucher programme on pupils in the state sector. This is an important methodological contribution to existing lottery studies, which cannot separate school productivity effects from peer effects – or take into account potential competition effects on state schools.

The authors’ findings display the importance of separating causation from correlation. Indeed, while the cross-sectional correlation suggests that private school pupils perform 0.65 standard deviations (SD) better than state school pupil in Telugu and mathematics, this correlation entirely displays differences in pupil ability. There is no impact at all of private school attendance in these subjects (or in social science and science) after four years in the programme.

However, pupils winning a voucher performed 0.13 SD better in English and a whopping 0.55 SD higher in Hindi (a subject not taught by state schools), with the average effect across all subjects also being 0.13 SD after four years. Among the pupils who actually opted for a private school once they had won a voucher, the average impact is 0.26 SD after four years. Overall, therefore there was a positive, albeit relatively small, effect of private school attendance.

However, these positive effects are also obtained with a significantly lower school budget: state schools spend on average more than three times as much per pupil as the private schools. Additionally, private schools have less instruction in Telugu and mathematics, but their pupils still do not perform worse. This strongly indicates that the low-cost private schools are considerably more productive than the state schools.

Given the two-stage randomisation design, the authors are also able to study the spill-over effects of the voucher programme on pupils in state schools. They find no effects at all, suggesting that these pupils neither benefit nor are hurt by vouchers. Similarly, the authors find no impact at all on pupils who already attended private schools. This backs up recent evidence that displays that the existence of peer effects has been grossly exaggerated in general.

On the other hand, it calls into question whether state school pupils benefit from competition from the private sector. Yet it’s not clear whether we should expect competition effects here given the strong regulations and little autonomy present in the studied school sector, where it’s also unlikely that money follow pupils. One of the key conditions for school competition to even have a chance to work is that schools have incentives and sufficient autonomy to react to competition. Furthermore, the area under study is rural, which naturally limits competition.

Nevertheless, the authors do find some evidence that the effect of winning a voucher is higher in areas that have more choice and competition between private schools. In these areas, the effect of attending private schools is strongly significant across all subjects and the effect size after four years amounts to 0.2 SD in English to 0.5 SD in Telugu and Science/Social Sciences. While more research is needed, this is indicative evidence that voucher programmes are more likely to be successful in areas where there is more choice.

Overall, the paper provides evidence that warrants a more nuanced attitude towards private schools in the developing world, both among critics and supporters. Private schools do slightly outperform state schools on average, albeit only in Hindi and English, despite the fact that they spend considerably less on average per pupil. They are therefore clearly part of the solution rather than part of the problem. At the same time, it’s also clear that these schools do not provide the vastly superior outcomes for which they sometimes are heralded – it turns out that pupil ability and background can explain much (even most) of the private school “effect”.

Furthermore, it’s important to note that an important reason why these primary private schools can be so cheap in the first place is because they hire young female, untrained teachers who are paid considerably less – and these women have often been educated in government secondary schools. This “invisible” public-private partnership appears to be important for the ability of the private schools to be so cheap in the first place.

Finally, the external validity of the experiment appears limited to similarly poor rural regions in developing countries. In other words, it has limited relevance for policymakers in the developed world, apart from suggesting the importance of rigorously evaluating policies before adopting them nationwide. Indeed, allowing a randomised two-stage trial with for-profit schools and a pure voucher programme in England would advance our understanding of the impact of choice and competition also in a developed world context.

Gabriel H. Sahlgren

This comment piece is also the Editor's Pick in the CMRE Monthly Research Digest_03_15. The piece reviews a paper by Karthik Muralidharan and Venkatesh Sundararaman, ‘The Aggregate Effect of School Choice: Evidence from a Two-stage Experiment in India', The Quarterly Journal of Economics, February 2015, which may be downloaded here. A free copy of the working paper may be downloaded here.

You can download free copies of back issues of the CMRE Monthly Research Digest here.

to subscribe.