Friday, June 02, 2006

ET digs economic demographics

Even as the imbroglio over reservation for OBCs in higher education institutes continues, with daggers drawn between proponents of the pro- and anti-quota stance, ET in partnership with National Council of Applied Economic Research’s senior fellow Rajesh Shukla did some quick analysis of the government’s own National Sample Survey (NSS 55th round 1999-2000) data on per capita consumption to come out with some startling figures.

From the front page lead story in today's Economic Times. So, what did this quick analysis show?

If you needed proof that social policy in this country operates quite in isolation of hard economic data, you need to look no further than the NSS data. OBCs — 35.8 % of India’s population and not 52% as the government would make us believe — compare favourably with the General category (36.5% population) on overall consumption figures as well as for individual products and services. [...]

At the top end, an average OBC spends Rs 15,436 compared to Rs 16,923 for General, and bottom 20% is the same for everyone, with OBC spends of Rs 3,293 hardly any different from the General’s spend of Rs 3,336.

We know why the ET gave this story such prominence. Still, this is something that's going to be important in the on-going tussle over the quota policy. The Supreme Court is going to be interested in data of this kind. I can already imagine Karan Thapar barking at dazzling some hapless minister with this data...

This paragraph, however, is intriguing (and revealing, too):

What also emerges from this NSS data is the fact that on consumption spends there is hardly any difference in the bottom 20% of all categories, SC/ST (43.3% of its population in this segment), OBC (37.7%) and others (18.9 %) with APCCE of Rs 3,172, Rs 3,293 and Rs 3,336 respectively.

It's interesting that while ET publishes the relative fractions of the three categories in the bottom 20% in the NSS data, it doesn't give this break-up for the top 20%! I wonder why ...

Prof. Shukla has a great series of posts on reservation. Do check them out:

Quota/Reservation: More 'reserved' than others

Who are the OBCs?

The merit of reservations.

A further nuance, which I can't claim to understand very well, is the following from this post by Prof. Madhukar Shukla of XLRI:

While it is true that that the last caste-based census was held in 1931, the Mandal Commission (and before that, Kaka Kakelkar Commission, 1961), had based their recommendations on their own sample surveys - one may have disagrements about the sample survey methodology, but it is not true that these reocommendations were based on 1931 data!

The differences in the estimates of OBCs among different surveys (52% by Mandal Commission, 32% by NSSO, 29% by National Family & Health Survey, etc.) is not because one is more accurate than the other - but because each uses difference criteria to define OBCs.

Can someone help me with figuring out the origin of this difference? Preferably with some online resources?


  1. Anonymous said...

    the "nuance" about different estimates of OBC population is like this: Mandal Commission partly used the 1931 Census data, partly the '11 indicators' and a sample survey to identify the OBCs. NSSO, on the other hand, simply asked people if they are OBCs or not. To the best of my knowledge, the National Family & Health Survey, added some extra parameters (e.g., type of house, availablity of water, toilet, electricity, etc., possession of watch, radio, cycle, etc) in identifying OBCs. The differences in the estimates are due to these different methodologies.

  2. Anonymous said...

    I can help you with this much, no group will want to leave the OBC categorization. So, all this talk about surveys etc. does not really work to our advantage, you can rest easy about that. The only undisputed fact is the percentages of so called privileged caste (12%), religious minorities (12%) and SC/ST groups (24%). Hence the 52% figure stands. Happy now?

  3. Anonymous said...

    Profs Abi/Madhukar:

    Which would you say is the more accurate estimate? Also why the differences (self-presentation concerns on the self-report?).



  4. Abi said...

    Madhukar: Thanks a lot. That clarifies things. While the Mandal Commission's estimates were 'derived' using the 1931 census data (together with a bunch of other things), do you happen to know what the 1931 caste-wise break-up is?

    Barbarindian: If you can't provide hard data (or pointers to hard data), could you at least refrain from posting your rant here? We all know your 'moral clarity' disallows you to take any nuanced view about quotas; except, of course, for the AIIMS quota. So, you don't have to repeat this message in every post here. Thank you.

    Neela: Since I have very little knowledge of the methodologies, I am unable to answer your question. My understanding is that the NSSO 1999 data is not all that unimpeachable; because it used a small sample (and so, its estimates come with larger error-bars). However, I do know that a massive NSSO survey was done in 2005, and the government is sitting on the data without releasing them. Several right wing economists read some dark motives behind the government's unwillingness to disclose the survey's results (for example, they insinuate that perhaps they show a big drop in poverty rates ...) In any event, we will just have to wait for NSSO-2005 results.

  5. Anonymous said...

    I have a complete list of all backward class commissions constituted in India since 1918. Next step is to find out how to get hold of these reports.

    Hard data is key and any policy must be on the basis of truth and openness.

    That is the position of reality check from day one. That is why I support the students one-point demand for a judicial commission before implementing any drastic quota (that will cost us 12,000 crores).

    If interested stop by this blog

  6. Anonymous said...


    Great job man. So, you are saying about 26% OBC and we already have reports of about 23% OBCs in schools. So no quota? Yippieeee!


    I will add you as a co-blogger. Let's protest against the quota together.

  7. Anonymous said...

    Abi, I discussed the 1931 issue briefly here.

  8. Anonymous said...

    Thanks Abi - I am not familiar either but the difference struck me as quite large to be ignored.

    I had one more question based on what Madhukar wrote about the Mandal criteria. Do you know if any of the classes identified as OBCs were forward castes? There has been a lot of discussion that the Mandal criteria does not exclude poor forward castes and I was wondering if mandal himself had classified any forward castes as backward (yeah, kind of confusing I know!) based on the multiple criteria.



  9. Anonymous said...


    One of the biggest confusion in this debate (unfortunately amplified by media) is that the 'C' is OBC stands for "Class/Community", and not for "Caste" (though caste is one of the criteria for determining whether one belongs to OBC or not)

    Since Mandal Commission criteria defines the OBCs based on not just the social background/caste, but also on educational and income background, OBC is not a static category across regions. A particular "caste" - whether 'forward' or 'backward - can be classified as belonging to a backward "community" depending other criteria (i.e., educational and economic deprivation). This would also vary across states and regions. Thus, it is likely that a particular caste in one region may be classified as part of backward community, but not in another region.

    Having said that, it is also true that in our society caste does correlate with "opportunity structure", and therefore, correlates with economic and educational backwardness.

  10. Abi said...

    RC: I look forward to the data when you get hold of the BC Commission reports. I too support the demand for a commission with powers to regulate the quota system.

    Barbarindian: Not so fast! The demand for quotas/affirmative action is for elite institutions too. In its absense, what we will have is 'separate but equal', and that will just not do. Moreover, in our current situation, it's actually 'separate *and* unequal'

    Dilip, Neela, Madhukar: Thanks for your comments, taking this discussion forward.

    My understanding is that the Mandal Commission's classification is not family specific. It's still at the caste level: within each state (and Dr. Bruno tells me that it's at the district level in TN), it assesses all the castes against its criteria, and classifies some of them as OBCs.

    Theoretically, then, an economically and educationally backward 'upper caste' can be classified as an OBC. Following what Madhukar said, since our caste "does correlate with opportunity structure", an upper caste being classified as an OBC would be a rarity indeed.

  11. Anonymous said...

    It's still at the caste level: within each state (and Dr. Bruno tells me that it's at the district level in TN), it assesses all the castes against its criteria, and classifies some of them as OBCs.

    Does anyone have any evidence that this happened in atleast one instance ?

    From what I have gathered so far and talking to one sociologist, no community was actually evaluated against those criteria. That set of criteria is just a guideline for communities that request to be included in the OBC list after 1991(ie, which seek reclassification from OC to OBC).

    If anyone knows of any community that was actually evaluated against those 11 criteria, please point me there. It is important.

  12. Anonymous said...

    The Mandal Comission report is based on a sample survey of only 810 villages out of 5 lakh!

    More has been discussed by Karan Thapar here.
    Can anyone get hold of the speech by Rajiv Gandhi, the Lok Sabha page has archives that date back only up to 1991.

  13. Anonymous said...


    Thanks for answering my question so patiently. I think I now understand a bit more. Pls correct me if I am wrong:

    1. A certain caste/community can be classified as OBC in one state but not in another. So, for example, X community may classify as OBC in A state but not in B state, right?

    2. The overall classification is a group-wise classification not an individual-level classification. So it is highly likely that a poor brahmin for example will get excluded from being classified as an OBC because his/her community itself might not be OBC in that particular state/area.


    Another question as iwork through this: Do the backward classes in the South correspond to these Mandal classifications? I think there is a lot of angst that backward classes in the South for example, are actually the ruling political party or richer landowners. As per Mandal, this should nto really happen, should it? Is Mandal for the entire country or only for the North? In that case, would we see a re-classifcation of backward classes in the South?



  14. Abi said...

    Neela: Hmmm..., the questions are getting a little too deep into the nitty gritty of how a caste gets classified as OBC. While we wait for someone to clarify it, here's my guess: it's done at the state level by something like the Backward Classes Commission. However, if this is correct, the role of NCBC (the National Commission on BC) will have to be clarified!

    Ankan: there is nothing wrong with sampling if it is done carefully and properly (like, following the guidelines laid down by statisticians). There probably are other problems with Mandal Commission's methodology (including improper sampling); but number of villages sampled, by itself, is not a valid or conclusive argument against the MC; you will have to show why it's not valid/conclusive.

  15. Abi said...

    Ankan: Rajiv Gandhi's speech does cover some of the things that I mentioned about the methodological problems with Mandal.

    And, I forgot to add: thanks for that link!