I'm not a big fan of Thomas Sowell's in general (here is a link to his weekly columns; here is a column where he summarizes the argument of his book), but the research he has done here is helpful. He does succeed in showing that the reservations systems in places like India, Sri Lanka, and Malaysia have many problems. But he fails to synthesize five different affirmative action experiments into a global critique of the practice. Each country is too complicated, and the comparisons that can be drawn are at best limited. However, the research he has done is still useful, albeit in a cautionary way.
Here are some ideas and facts about reservations in India from Sowell's book that I found interesting. The information I'm passing on isn't slanted either pro-reservations or against them. I do provide my own interpretations in parentheses, but other interpretations of the facts might be possible:
1) Caste violence still occurs. Sowell cites a 1991 India Today article on the public lynching of cross-caste lovers in a village 100 miles from Delhi. But more importantly, Sowell cites government statistics: "Government statisticts on atrocities against untouchables never fell below 13,000 per year during the decade of the 1980s and reached well over 16,000 in 1984. Far from abating with time, these officially recorded atrocities escalated to more than 20,000 a year in the 1990s. "
That's 20,000 a year. Most atrocities occur in a small number of states (Sowell cites Bihar and Uttar Pradesh especially). They are primarily rural occurences; Sowell acknowledges that strong caste feeling -- the kind that can lead to violence -- is rare in India's big cities.
(This evidence supports the idea of government intervention, at least as a matter of basic civil rights. It doesn't directly support the reservation system)
2) Unfilled seats. Reservations for Scheduled Castes (SC) in schools and government posts remain largely unfilled, whereas reservations for Other Backward Classes (OBCs) are generally filled to capacity. Sowell cites a 1997 study that indicates that nationally preferential policies only benefit 6 percent of Dalit families. Moreover, the same study reported that "none of India's elite universities and engineering institutes had filled its quota for members of scheduled castes."
(This could be read in many ways -- but at the very least, it proves there are problems and imbalances in the reservations system. OBCs are not necessarily 'backward')
3) Continued underrepresentation. People from the scheduled castes and scheduled tribes continue to be absent from white collar positions. "For the country as a whole, members of the scheduled castes and scheduled tribes -- combined -- did not receive as much as 3 percent of the degrees in engineering or medicine, though together they add up to nearly one-fourth of the population of India."
(This suggests that reservations have not been wholly successful, though perhaps even 3 percent is a dramatic improvement over what one might have seen 50 years ago)
4) Other economic and practical factors. The government provides scholarship to SC students to attend school, but sometimes that is not enough: "Even when the government provides primary schooling free of charge, the costs of books and supplies may not be affordable by very poor people. For secondary education, rural students especially may not always find a school nearby, so that those whose parents cannot afford the costs of commuting or relocating -- and paying for housing and boarding -- have little realistic prospect of attending, regardless of preferential admissions policies." (32)
(This evidence suggests that reservations may not be systematic enough to really create equality of opportunity)
5) Dominance of some SCs. Some Scheduled Castes do better than others with the system, raising the demand in some quarters for 'quotas within the quota'. A particular case in point are the Chamars, historically a leather-working (and therefore untouchable) caste. There are lots of statistics here, so I'll quote at length:
In the state of Maharashtra, the Chamars are among the most propserous of the scheduled castes. A study found that they were 17 percent of the state's population and 35 percent of its medical students. In the state of Haryana, the Chamars received 65 percent of the scholarships for the scheduled castes at the graduate level and 80 percent at the undergraduate level. Meanwhile 18 of the 37 untouchable groups in Haryana failed to get any of the preferential scholarships. In the state of Madhya Pradesh, Chamars were 53 percent of all the scheduled caste students in the schools of that state. In Bihar, just two of the 12 scheduled castes in that state--one being the Chamars-- supplied 61 percent of the scheduled class students in school and 74 percent of those in college.
(This suggests that caste status should really be indexed with a family's economic background where there is evidence that certain communities are no longer impoverished. Wealthy families from low castes should not be given preferential treatment.)
There's more. Sowell also has sections where he talks about local reservations issues in Assam, Andhra Pradesh, and Maharashtra (he is compelling on Assam and Andhra, but not on Maharashtra -- the Shiv Sena is a separate can of worms). He also refers to some of the bizarre ways people work the system (i.e., higher caste students who are 'adopted' into lower caste families in order to benefit from reservations). Most of these examples are, however, anecdotal.
What is lacking in Sowell's chapter on India is any treatment of the Mandal Commission report, which recommended the expansion of the reservation system in the 1980s. When the V.P. Singh government implemented the recommendations of the Commission in 1990, it led to chaos, and eventually the collapse of the government.
There are certainly other limitations and flaws too, but at least he provides some helpful data and a starting point for discussion.