If the census tabulates caste data with education, occupation and the place of residence, etc., it could be meaningful
Too many of India’s policies, especially in recent times, are made in a data vacuum. The Right to Education seeks to rid the country of unrecognised private schools, but there is no official data on just how many children study in these schools and how good or bad this education is. The National Rural Employment Guarantee Act seeks to provide jobs to the unemployed, but the data on this is collected only once in five years by the NSSO. The government is happy to do caste-based affirmative action even while it has no data on how real the discrimination is.
It is in this context that the proposal to ask for caste details in the census has to be seen. Those in favour of it argue that if you must have reservations along caste lines, you must know just how many people there are in each caste group. Those opposed to it argue that since the Supreme Court has put a 50 per cent ceiling on reservations, and there are more SC/ST/OBCs, the exact number needn’t be collected. Two, since there are benefits in being an SC/ST/OBC, people will flock to declare they’re SC/ST/OBC — the OBC share of the population rose from 36 per cent in the 1999-2000 NSSO survey to 41 per cent in the 2004-05 one for this very reason.
So, it is likely the same thing will happen if caste details are canvassed in the census. There will then be pressure on the courts to raise the levels of reservation, leading possibly to another anti-Mandal type of bloody agitation. There is, however, an even more fundamental issue when it comes to canvassing caste details: in the absence of ways to put them in perspective, the data is just incendiary.
Let’s say you know that while SC/STs are 25 per cent of the population, they comprise just 12 per cent of those with “administrative, executive and managerial” jobs. Before you shout discrimination, keep in mind SC/STs also comprise just 14 per cent of all graduates — and since you have to have passed out of school to become a graduate, the solution is clearly not affirmative action, it is ensuring low dropout rates in school.
Or let’s say you get to know that while 20 per cent of OBCs are graduates, just 7 per cent of them have professional/administrative/managerial jobs. It’s natural to shout discrimination, but keep in mind the respective figures for upper castes are 34 and 11 per cent.
All of this, and a lot more, form the crux of a book just written by Rajesh Shukla and me (out on the stands at the end of the month) based on the NCAER’s annual income survey across the country. While it would be foolish to deny caste discrimination, the book tries to analyse the reasons for the difference in incomes — how much is due to education, how much is due to where you live (villages, small towns, metros, prosperous states, poor states), how much is due to your occupation, and so on.
The results are illuminating. To cite just one type of example, while an illiterate ST household earns Rs 22,456 per year, this rises to Rs 85,023 for a graduate household; an illiterate upper class household earns Rs 31,511 versus Rs 135,086 for a graduate household. At all levels of education categories, upper castes earn more than STs. But the caveats are critical. Here are a few:
# The difference in income between an illiterate and a graduate ST is far higher than the difference between an ST and any other caste group — that is, education is critical (and keep in mind the point about the high school dropout rates).
# As salaries are the highest (and the relative differences between caste groups the least) in the modern services sector, a big reason for low ST incomes (regardless of their education level) is that very few of them are in this sector.
# Since incomes are the lowest (and relative income differences between caste groups the highest) in low-income states, a very major reason for low ST incomes is that they are mostly located in these states.
# As incomes are the lowest in rural areas (and relative income differences among the highest), and this is where the greater share of ST households are, this is another reason for lower ST incomes (again, this applies to all education groups).
While the census does not ask for information on incomes, it does ask for information on some surrogates — pucca house, own kitchen/bathroom, TV, car, mobile etc. If this information is married with information on education, occupation and location (rural, urban, small town, metro etc.) in meaningful cross-tabulations of the type discussed above, the exercise could yield powerful results on caste discrimination. The raw numbers, it must be underscored, are not relevant, more so since, as in any such exercise, very large numbers will either not report their caste or will declare it incorrectly. Contrary to the generally held view, the census does collect information on caste and religious groups even today — the SC/ST data comes from the census today. So, asking respondents to declare whether they are OBCs isn’t such a major issue.
It needs, of course, to be said that since census is used to validate most sample surveys, getting the enumeration wrong will have huge negative implications. Getting this right is critical, and the analytics have to be top class — so make the unit-level records public and then let top academics/institutions have a crack at interpreting them. The census will throw up as many interpretations of caste discrimination as we have over poverty using the same NSSO data. Let’s not shy away from the debate.
