Wednesday, October 31, 2007

Caste and religious discrimination in India's private sector companies

Here's the abstract of this paper by Sukhadeo Thorat and Paul Attewell:

This article examines the prevalence of discrimination in the job application process of private sector enterprises in India. The study is based on a field experiment where authors replied to job advertisements in major English dailies sending three applications to each call – as an upper caste Hindu applicant, as a dalit and as a Muslim. Using statistical analysis they assess the data and find that discriminatory processes operate even at the first stage of the application process.

The authors used a neat method: Look for recruitment ads for entry-level or near entry-level jobs in private sector companies. Send three identical CVs with different names that would easily identify the (male) candidate's background: upper caste, Dalit or Muslim. And wait for who gets the call for the next step in the process (typically, an interview).

What did they find?

... [T]here are statistically significant effects of both caste and religion on job outcome. Appropriately-qualified applicants with a dalit name had odds of a positive outcome that were 0.67 of the odds of an equivalently qualified applicant with a high caste Hindu name. Similarly-qualified applicants with a Muslim name had odds of 0.33 of an otherwise equivalent applicant with a high caste name.

I would say this conclusion is pretty devastating. We will have to wait and see how the industry people respond to this study.

* * *

Thanks to Madhukar Shukla (who also has an extended commentary) for the pointer. I was keen to read the paper ever since it was mentioned in Ram Mohan's blog. I'm glad it's now available in an open-access journal like EPW.

* * *

Before closing, I also want to highlight something else from the paper:

The most common answer to an application was no response whatsoever. Rejection letters were rare: only 17 applications (one-third of 1 per cent) resulted in rejection letters. In other cases, those we classified as positive outcomes, employers either phoned or wrote to certain applicants asking to interview the person (or in some cases requesting the applicant appear for a written test). There were 450 positive outcomes of this type (9.4 per cent of all experimental applications).

The response rate itself is less than 10 percent! Unbelievable, isn't it? What kind of professionalism is this?


  1. Anonymous said...

    Don't mean to be glib here, but the authors proved quite indisputably that caste plays a role in hiring in the India pvt sector. Agreed.

    However, do they give any reason why this is so? I have gone through their entire paper(it's very good piece of work , India is crying out for good empirical work). Nowhere, do I see that they have found a causal relationship.

    My explanation:firms know a dalit name when they see one, figure out that the dude got his university seat through a quota(which is almost always a correct conclusion) and then decide that merit-wise he's not up to snuff. My theory will however fail to explain the phenonmenon for muslims who don't have any quota in most educational instis but one has to go through their dataset in more detail to know that(or employers might be incorrectly assuming that).

    I passed out from a NIT in '05. I know from personal observation that lower SC get worse jobs and ST even worse. Why? Because companies aren't dumb. When they are paying lakhs of rupees to freshers they want to be sure.

    I have a huble suggstion for the researchers. Do this same study but this time use a pvt univ like SIBM, BITS-P, DA-IICT,RVCE PESIT, BMS,etc (which don't have reservations) and see if the results repeat. If yes, then the Indian pvt sector is seriously racist. If not, the Indian pvt sector might be check-mating the politicians.

    In a twisted way this study shows why reservations don't always work the way they are supposed to.

    Btw, I'm really all for a united, vibrant India in which everyone has a shot at making something of themselves. So all you haters, think about this before you flame me.

  2. Madhat said...

    abi, fix the link for the paper!

  3. Anonymous said...


    The results were kind of expected, given the amount of casteism, racism prevalent in the general population.

    Just if I tell you that such a study was done and you wont wait for the results, you would have guessed it already. The best part of the paper is in the fact that now we have statistics with us to cry about it.

    The comment given by "dal...." above gives a different angle to the picture which could be partially true.

  4. Just Simple said...

    Caste is a big reality and its disheartening to know its existence in private sector too.

    In IIT's as far as I know, companies went for higher CPI's and not just who got admitted on a reserved seat. The companies need what one learnt in four years in college, not what he studied for getting admission. The caste difference came only when government companies came for recruitment.

    I think it is mentioned in this case 'Identical CV's'

  5. Anonymous said...

    Is the Economic and Political Weekly Journal peer-reviewed?

  6. Anonymous said...

    Leave the private sector alone; if I had a company I would hire only people with crooked noses and no one can challenge me. But does the public sector hire really impartially? I hear that people are asked to pay cash in return for govt jobs.

  7. Anonymous said...

    The problem of reservation esp in higher ed and value of education are a irony of sorts.

    On one hand, we say reservations will improve the lot of people by giving them equal educational opportunties. On the other hand, almost all agree that the bulk of the education esp in a good univ comes from the selection effect. That is the mistake Anurag makes. Anurag will likely take two contradicatory stands on both these issues.

    Anurag points to an observable but can he deny that SC/ST students at his insti have worse labour market outcomes on an aggregate basis. There are exceptions to be sure, but we are all pretty smart here at nanopolitan to ignore outliers.

    At my insti also, companies ask for a lot of things CPI, extra-currics,etc but to say that companies aren't using caste just because tehy don't ask for it is kind of premature. This very blog has poitned out studies showing higher earnigns for taller people, more beautiful people, people with outgoing personalities,etc. All high paying jobs have some level of client interaction which demand a ceratin thereshold of iq + mannerisms.

    The major flaw of the study in my opinion is that it uses a predictor of productivity(univ education) as an instrument to document discrimination. But labour market outcomes are more than jsut univ degrees and the Indian pvt sector(no matter how screwed and it's pretty screwed up) has figured after a ton of trials which type of people tend to work out in their firms. Can you blame them for taking into account all this non-CV information to make hiring decisions? This paper establishes a good reference point for debate,mainly that discrimation exists at least to teh naked eye and I totally agree with that. What is more puzzling is why this exists in a competitive labour market esp for the type of firms the authors study? The answer might be a lot more harder to digest than crying discrimination hoarsely.

  8. Ludwig said...

    Thanks for the link, Abi. Haven't read the paper (yet), but at first glance, they appear to have tried what Bertrand and Mullainathan attempted in Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination. That paper is one of the references in the EPW one, and of course was about black vs white hiring.

  9. Ludwig said...

    dal mein kuch kaala hain...

    While I hear your argument that it's possible that the companies are using caste as a proxy for whether the candidate got lower marks in the entrance/qualifying exam (that's what you mean essentially, right?) I find that there is one piece of information missing in the paper that might've shed some light. The authors say

    "All the résumés and cover letters presented strong applicants for the job opening: they had suitable degrees from reputable universities, and (where indicated in the advert) appropriate job experience and skills."

    They don't say if the resume included a CGPA/CPI/% age type of data, and based on what we know we can only assume it didn't.

    Suppose that the resumes did include CGPA/marks, and everyone had the same marks, and then the outcome was the same i.e. apparently disciminatory, then would you say that hirers are simply discriminating, rather than using the name as a proxy for the candidate's entrance/qualifying exam marks?

    I think that would make a much stronger case for arguing that there is some blatant discrimination based purely on prejudice and nothing else. Of course, you could still argue that the hirer is doing some thinking on these lines:

    "Candidate A has a CGPA of 8.5 and has a name that sounds upper caste. Candidate B has a CGPA of 8.5 and has a name that sounds dalit. Even though both have the same CGPA in college, it is likely that B got in through a quota, so if I have to pick between A & B, I should pick A."

    Maybe we should ask the authors if they did include marks in the resumes...