What happens if Colleges declares a quota for a particular identity group under the following conditions?
1) the identity group has relatively few people who can easily study hard and come up to the general level whereas most of its members find it very costly to achieve even a basic level.
In this case, the few will put in just enough work to beat the many so as to secure a place at the College with the lowest possible outlay. However, since they don't have experience of studying hard they may continue to lag behind the others in the College.
2) the few who can easily study hard come from a 'creamy layer'- i.e. have better connections than the vast majority. (In India, the 'creamy layer' of the Backward Castes, but not the Dalits, is excluded from affirmative action)
In this case, such students would find it rational to put in very little work both at School and at College because they will try to use their connections to get a job rather than rely upon their own merit. If no one can get a job through connections, then students will soon learn that there is no point not studying hard at School, even if one has a guaranteed place thanks to the quota, because if you don't study hard at School you won't have acquired the habit of industriousness and thus will fail at College and end up unemployed and thus worse off than if you never went to College in the first place.
However, if some people can get jobs through connections then this lesson is never learned. On the contrary, what happens is that the lazy fellows who got jobs create a poor impression and so those without connections, even if they were industrious students, find their qualification does not get them employment.
Notice that the 'creamy layer' can remain 'creamy' from generation to generation but only if they can keep up their connections. They can never advance on the basis of merit. However, since a person who both studied hard and who has connections can make more of his life than one who only has connections, it follows that the 'creamy layer' won't get as much out of education as the general category student who, even if he has connections, still has to compete with people like himself in order to get College entry in the first place. Thus the 'creamy layer', over time, will find itself falling behind what would otherwise have obtained as a result of the very affirmative action from which they benefited in the short run. As for the vast mass of non-creamy layer students, the quota system helped few of them and even those few faced prejudice in getting employment because of the laziness of the 'creamy layer' graduates.
One solution, favored by the creamy layer, is to ensure promotions are subject to quota in all industries. However, this only postpones the problem. Cosmetically, the creamy layer is getting creamier, but they fall behind nevertheless because they don't have the skills to match their position. Thus they can't manage their own wealth as cleverly as the hard working general category student who went on to be a hard working employee having to cover for the incompetence of his 'creamy layer' colleague. This hard working employee is acquiring skills relevant to his own wealth management and valuable insights relating to the upbringing of his own children.
The Creamy layer may demand a quota of Directorships in the Enterprises set up by the hard working general category people but this will only be granted in a cosmetic manner. All that has happened is that the Creamy layer are extracting a rent based on their Social Identity. But this rent comes at huge cost to the class they claim to represent. They themselves would have been much better off if no quotas had been put in from the outset.
There is one caveat. Suppose some creamy layer types who do the minimum work to get in on the quota are randomly disqualified in favor of others from the same identity class, then they have an incentive to study hard (since this is relatively cheap for them to do) so as to guarantee their place. Similarly, if quotas in jobs and promotions and so on have this random element added on- such that some one who 'just' qualifies is randomly rejected- then Creamy layer candidates who have low cost of skill acquisition have an incentive to work hard so as to guarantee their success.
It might be argued that a better solution is simply to disqualify 'creamy layer' candidates ab ovo. However, there would still be the problem that non-creamy layer candidates with low cost of skill acquisition have an incentive to put in the bare minimum of effort. In other words, there has to be randomization in selection even for first generation 'creamy layer' aspirants, otherwise all that happens is that the second generation 'creamy layer' is skimmed off and behaves like the general category. In this case, there is no incentive to continue to identify with the disadvantaged group for them. They have been co-opted by the general category. Thus they can't act as a mimetic locomotive pulling their entire class upwards. By contrast, with randomized disqualification, the creamy layer has an incentive to work extra hard because the over all quality of their identity class would be lower than otherwise and this would make it more expensive for them to secure their own future through their contacts.
This may well have a 'mimetic effect' within their identity class.
Perhaps, Dr. Ambedkar, who well understood Gabriel Tarde's mimetic theory, favored a sunset clause for Quotas for this purely economic reason.