It doesn't matter

There’s little evidence that reservations help

Using someone else’s (Surjit Bhalla) column name for the heading of yet anther person’s (TCA Srinivasa-Raghavan is off for two weeks sunning himself in Gurgaon, where he lives) regular Okonomos slot is, perhaps, not the best thing to do, but nothing describes the government’s policy on reservations for the disadvantaged in jobs and in educational institutes better than this. Since I haven’t been able to find any robust papers using relevant data on India, here’s a summary of some papers from the National Bureau of Economic Research (NBER) on the subject.

The first paper, by David Card and Alan B Krueger* begins by saying that when, between 1996 and 1998, California and Texas eliminated the use of affirmative action in college and university admissions, admission rates for black and Hispanic students fell by 30-50 per cent. Case shut and closed, you’d be entitled to say. But hold on a minute. What the duo have done after this is to examine the behaviour of these minority groups after affirmative action ended — did they continue to send their SAT scores to these elite universities, or not? This is particularly important in the case of India since it has often been argued — when Murli Manohar Joshi was trying to get the IITs and the IIMs to reduce fees — that lack of affirmative action turns away the underprivileged from better quality educational institutes that, by and large, tend to be privately owned and, often enough, take no aid from the government.

Card and Krueger use confidential micro data from the population of SAT-takers in these two states between 1994 and 2001 (to get a good feeling of the “before” and “after”). First, they found that while there was a sharp drop in the admissions in the elite campuses like Berkeley and the UCLA, the overall drop in the state was hardly significant — that is, the minorities got admissions to other colleges in the state. So, if there were good quality government-run or aided colleges, the issue of reservations in unaided colleges/schools in India wouldn’t even be an issue.

Second, they found that the average mean differences in SAT scores for blacks and Hispanics versus Asians and whites remained more or less the same before and after 1997 — in other words, the impact of affirmative action in raising overall education standards was low, hardly surprising since educational differences are not so easily fixed. The analysis also showed that “there is little evidence of a systematic fall off in interest of minority students in the elite or other selective public institutions in California or Texas”. That is, the numbers of those applying to these schools among the minorities hardly showed any change. And, when you break this up further into those with good SAT scores and those with mediocre ones, the falling off in applications due to the removal of affirmative action is barely noticeable, even in the immediate years of the changed policy.

Another paper, by Roland G Fryer Jr and Glenn C Loury** is more broad brush and reviews a lot of the literature on the impact of affirmative action. One paper it reviews, for instance, finds evidence to show that when two groups of similar whites were asked for their views on blacks, the group to whom affirmative action was mentioned tended to affirm negative racial stereotypes like blacks are mostly lazy, the other group to whom affirmative action was not mentioned tended to have less extreme views.

Another study reviewed by the duo is even more interesting. It shows that when there is a modest goal of affirmative action (like, say, a 5 per cent reservation) there is a tendency for minorities to invest more in education since only the most qualified of them will get jobs. If, however, the goal is a lot more aggressive, the minorities know their chances of getting a job are higher and so invest less in bettering their skills. An important policy conclusion for those in government here. The authors call this outcome a patronising equilibrium.

There is also an interesting discussion on whether affirmative action should take place earlier on or later — at primary and secondary school or in college, or in jobs. The conclusions are mixed, and depend upon the control variables (for instance, assisting students in a particular neighbourhood is more beneficial than when the same effort is spread across a wider area since the peer group externalities are strong in the first case). But one important result is that affirmative action is more effective earlier on only if the attrition rates are low. In India, the dropout rates for SC/STs are higher than those for others, and so the impact of reservations at the school level obviously gets dissipated. Another paper reviewed shows, using data on legal education, that affirmative action has actually hurt black students as they enroll in top schools because of this, but fare much worse than the class and so are less likely to graduate from law school — the authors admit the study is controversial! * Would the elimination of affirmative action affect highly qualified minority applicants?; NBER Working paper 10366