Discussion of Private Schooling in India Abhijit V. Banerjee
An important policy question • This paper takes on a key policy question—do private schools work better than public schools? • Very important both for determining the regulatory stance of the government and for deciding on voucher’s etc. • And not obvious. • Public school teachers are much better paid (often by a factor of 4) and, as a result, might be better educated and trained. • On the other hand there is the now very well-known problem of incentives in the public sector
A hard problem • We cannot use judgment of parents: parents have a hard time estimating the benefit of private schooling for their child • Zhang (2008) shows that parents in China queue up to try to get their children into more expensive schools, but except for the highest performing children everyone actually does worse as a result. • But then there is an identification problem • Are children who are sent to private school more motivated/gifted on average? • Selection could operate within a family, within a neighborhood, within a district
The eternal search for an instrument • What we want is something that influences private school participation but not performance • Not easy to find--they use: • Presence of a private school: but don’t private schools where education is a priority /where public schools are bad: • could lead to upward bias because of prioritization • Downward bias because bad public school quality. • Early English in Govt. school/English language instruction: Does it vary at the State level? If so what determines where you set up these special schools? • Probably leads to upward bias.
More instruments • Social networks: Even they are skeptical • Mysterious “missing schedule” • Cook in government school: • Very intriguing reduced form: villages where there is a cook in the local government school systematically under-perform other villages • Worth investigating---is the school meal program actually hurting childrenbecause parents value the wrong things • We do not know the direction of bias: upward because more privileged village or downward because more backward village? • Worth investigating government policy for allocating cook
Another approach: Family fixed effects • Problem: who would you send to school if you had to send one child to school? • The good news is that the two approaches and indeed the OLS, give very similar results. • At least the family fixed effect and the IV approaches use very different types of variation • The effects are not small: .3-.4 standard deviations is substantial by 1st world standards though dwarfed by state effects. • Probably will not convince the skeptics
What else can you do? • Randomized trials: randomly allocate vouchers • Angrist et al. • However how do you deal with equilibrium effects: what happens to school quality as demand goes up? • Kremer and Muralidharan try to deal with this by randomly choosing voucher villages in AP • They also randomly allocate vouchers, so will get an estimate of the supply elasticity. • But villages are also small….
Another approach • In a number of papers Jishnu Das and his co-authors have studied public versus private education in Pakistan • As in India private education is much cheaper • Their results make a strong case for private education based on the following observation: • In English, the difference between children in private and government schools is twelve times as large as the difference between children from poor and non-poor households after controlling for observed differences between children. • Put differently, a huge part of the variation in performance is explained by school differences within the same village. • Can un-observables do so much heavy lifting? • Why do people with the same observables choose differently? • What would be the effect of an information campaign?
Still it may be worth trying • To see whether we see the same patterns in India • How does the school effect within a village compare with the effect of wealth? • They do not present school effect regressions with village fixed effect • Wealth effects are hard to interpret because the scale is not shown (and why are there 25% in the top quintile) • However the OLS effect of private school is the same size as the OLS effect of college educated parents! • (why are control effects in the IV identical to the OLS?)
On balance • All the evidence looks favorable to private education but there is no smoking gun.. • But even if we find the smoking gun we will need to think about • Whether this model (educated unemployed set up schools for want of anything better to do) will remain sustainable if growth continues? • Whether this can be a means of delivering post primary education?