Pages

Saturday 17 August 2024

Soumitra Shukla's junk Social Science

 Soumitra Shukla has a silly paper here which proposes paying MNCs in India a subsidy equal to about 5 percent of annual starting salary if they hire a candidate from a disadvantaged caste. Apparently 200 dollars is enough to neutralize elite advantage accordingly to this silly lad's Junk Econometrics.

Interestingly he does not take into account the higher incentive for 'disadvantaged' caste members, particularly those who have been to an elite college, to go in for a Government job where they will be able to accumulate much more power and money. This means you will have higher attrition for 'disadvantaged' hires because it is worth their while to sit the UPSC year after year. Since the cost of losing a guy you trained for a year or two is very high compared to the 200 dollars on offer, Shukla's proposal is foolish. 

On the other hand, it sounds plausible that 'elite advantage' arises most substantially at the 'non-technical interview stage'. However, this may merely track proficiency in English. 

Shukla writes- Unstructured interviews that screen on conversations related to shared experiences, aspects of cultural fit, and personal hobbies are not unique to India. Such practices are also common in the screening practices of elite U.S. colleges and corporate America.

All Americans who attended an elite US college speak good English. This is not necessarily the case in India. A person with very poor English may be able to score well in Exams precisely because the Exams don't test comprehension. On paper, there is nothing to distinguish, Ms. Bannerjee, whose great grand father was knighted by the British and who might struggle to understand any Indian language and Ms. Murmu who is the first person from her village to attend College. She may be brilliant. She may have worked harder. But she has seldom had the opportunity to converse in English to the required standard. This means, in a work environment, she may not initially be able to keep pace with her batchmates. True, a really good HR manager would go ahead and make the hire and arrange for the girl to get a bit of extra help. This will pay off massively in years to come. Ms. Murmu will end up being promoted over the head of Ms Bannerjee. The problem is that the graduate recruiters don't get the benefit for their long sighted choice. By contrast, nobody complains if you send them more Bannerjees with posh accents. 

It is one thing to end up being forced to hire Bannerjee and Chatterjee because their English is good and quite another to pay for the privilege of having to associate with arrogant, Anglicized, Brahmins. Nobody wants to do that- least of all Brahmins. 

The employer willingness to pay for caste

there is no such willingness. If you really lurve Brahmins so much just hire one as your cook. Warning- most of them are Vegetarians.

is calculated through a reduced-form caste coefficient in the employer’s utility function.

Where is the 'conversational English' coefficient? Surely that matters to MNCs?  

The “caste penalty” is identified under the plausible assumption that—conditional on resume screening and performance in both technical tests and non-technical group debates—parental background, neighborhood, and subjective impressions of cultural fit are plausibly weakly correlated with productivity.

No. Speaking good English is highly correlated with productivity in MNCs. So is being the sort of guy the client likes and trusts. There is a premium on language, not caste. True, a small minority of people from traditionally 'learned' castes switched to English generations ago (they also Anglicized their names- e.g. Bandopahdyaya became Bannerjee) and this had a hysteresis effect. It may be that in competition with 'creamy layer' disadvantaged castes, there is still some generational advantage but there are two other factors which need to be mentioned.

1) Graduates from elite colleges whose parents have a bit of money may do better by going abroad as 'students' (this generally also permits them to work or operate a business) and this may put them on a much better trajectory. This is particularly the case if you didn't bother too speak much English at Skool or Collidge. Once you have got MA and green card from Yukay Amrika, no one will dare make fun of your accent. They will bow down and worship you.

2) At the 'group debate/ HR interview' stage, selectors screen on the basis of a notion of Corporate Culture. They may say to each other 'that fellow is too cocky. Let him go to our rival.' Equally, some candidates get the message that they are too aggressive or too passive and thus won't fit.  In other words, there is mutual screening and signaling. Since we are talking about a 'matching problem' which is asymmetric- the employers wants minimal attrition whereas the graduate doesn't mind using one firm as a stepping stone to another- a more complicated game is being played than Shukla will admit. Now, it may be that certain supposed caste based traits play into this because even bad information reduces uncertainty, and so it may be that people of backgrounds for which there is no stereotype are disadvantaged. My point is, if you know what stereotype people have about your sort, you can play against that stereotype and thus make your mark. 

One final remark. To speak of people as being disadvantaged suggests that they will perform less well than those who are spoken off as privileged. Other things being equal, it is the latter who will be hired even if the princely sum of 200 dollars is given to you if you hire a smelly retard. 


No comments:

Post a Comment