There is a friendly review in Commentary of a new book by Thomas Sowell, called Affirmative Action Around the World: An Empirical Study. To my knowledge, it's the first book to look at assistance programs for underprivileged minorities in a trans-national perspective.
A couple of quick points going from the text of the review. One of Sowell's examples is India -- no surprise there. The statistics are well known: reservations for scheduled and backward castes are at 50% in Indian universities, and it is highly unlikely that there will be a change anytime soon. Virtually everyone I know who has gone through the Indian system complains about it (though I should say that most of those people were not born into low-caste households). It is clear that the quota system needs to be reformed, but certainly there are more ways of doing this than simply abolishing it. For instance, I wonder if it might be possible to implement "affirmative action" in India along the Michigan model (consider race as one among many factors for what the candidate). The courts will have to do this if it ever happens, since it seems to me that reforming the reservation system will be a political impossibility for the next few years. At any rate, to draw from this the Indian case the idea that the problems in the Indian system means that "proliferation is the rule," seems to me highly questionable.
Another place where Sowell oversimplifies is Malaysia, where there are reservations for native Malays (who are the majority) against the ethnic Chinese. Here, Sewell again uses his slanted rhetoric of "competence," rather than addressing the real issue. Here, it is clear that closed business networks operating within ethnic groups tend to be the biggest reason why groups such as the Chinese in Malaysia continue their economic dominance. It is not "competence."
Finally (for now), the assertion that the anti-Tamil laws enacted in Sri Lanka in the 1950s are a form of "affirmative action for a less competent group," and that this is a direct cause of the Civil War, is a gross oversimplification. The de-recognition of the Tamil language by the Sri Lankan government cannot be called "affirmative action" for Sinhalas! It was quite something else entirely.
A non-polemical study organized along the same lines, and even employing the same case studies, could be helpful. But instead of trying to lump all of the different national cases into a single vocabulary of "affirmative action for" in the interest of making alarmist racial/political points, the writer would need to be sensitive to the particular histories and ethnic formations of each nation, including the role played by colonialism in the formation of the current system. One would also need to study in depth how reservations have improved life for underprivileged minorities -- something that Sowell apparently takes no interest in. One would further need to use work from anthropologists who have studied how closed ethnic business practices create an environment of domination that is almost impossible to alter without help from the state. It may not be organized discrimination (i.e. there is no "Jim Crow" in Malaysia), but it is a problem.
(More on this once I've read the complete book.)