My attempt at describing a paper in words. The paper is by Anirban Mitra and Debraj Ray and I am going to offer a simplified version of it.
Suppose an aggressor faces a victim and decides whether to attack the victim or not. Each has wealth which can be stolen at some cost. Each belongs to some group, e.g Hindu or Muslim. The aggressor’s group can decide how to much to invest in a “conflict infrastructure” that reduces the cost of an attack to a member of the aggressor group. How do the number of attacks change as a function of aggressor and victim incomes?
First, suppose victim incomes increase keeping aggressor incomes fixed. There are two effects. Keeping investment in conflict infrastructure fixed, certainly the aggressor group has a greater incentive to attack and this effect increases attacks. However, in principle, if investment in conflict infrastructure goes down significantly, this might reduce the number of attacks. As the benefits of attacking have gone up and aggressor incomes are kept fixed, this effect is never going to be large enough to cancel out the first one. The first conclusion is that as victim incomes go up so do the number of attacks.
Second, suppose aggressor incomes go down keeping victim incomes fixed. There are two effects. The first effect is the same: a fall in aggressor incomes increases their incentive to attack the victims, in the same way that a rise in victim incomes increases the incentive to attack, keeping investment in conflict infrastructure fixed. But as aggressor income has gone down, this can increase the costs of investing in conflict infrastructure. If this second effect is large enough, it can cancel out the first effect. The second conclusion is that the impact of a change in aggressor incomes on the number of attacks is ambiguous.
Then the authors use the model to offer an interpretation of Hindu-Muslim violence in India. They have data on violence and also data on household expenditures by ethnic group. They show that a change in Hindu expenditures has an insignificant effect on ethnic violence but that an increase in Muslim incomes has a large, positive and significant effect on violence.
Using the model to interpret the data, we would conclude that Hindus are the aggressor group and Muslims are the victim group. The combination of the theory and the data is necessary to offer this interpretation. I like this aspect of the paper as well as the quite surprising elicitation of the identity of the aggressor group vs. the victim group. Why is there this asymmetry? The authors offer some interesting speculations based on Indian Partition. I highly recommend the paper: the theory and the empirical analysis and simple enough that a theorist or an empiricist can understand the whole paper. The conclusions are surprising and the methodological approach is attractive.
Cheap Talk 
2 comments
Comments feed for this article
January 14, 2010 at 1:58 pm
Anshuman
Muslims in India are much more poor, uneducated and backward compared to the hindus, especially compared to upper caste hindus. There have only been a few exceptions, such as the Gujrat riots, where the “hindus” (it is questionable if the tribals involved in the attack are hindus in any meaningful way) attacking the muslims have been poorer than the muslims. For the most part, the muslims, who have suffered at the hands of hindu/political violence, have been very poor and have not necessarily been displaced.
While property theft is always a motive for attacking and displacing a population, I find it hard to believe that any epsilon gain in the poor muslim population’s wealth is a driver at all of this violence.
It is commonly understood in India, or at least in the circles that I have lived in, that most of the hindu-muslim violence is either organized/initiated/directed/ended by the state or by politicians out of power or clamoring for power. Given the deep seated hatred hindus and muslims in India have for each other, the private armies/goon-squads of the politicians readily find young unemployed hateful men to participate once the violence has gained sufficient momentum. These riots are usually the result of political power moves of ambitious politicians. While an increase in muslim wealth, accompanied by hindu unemployment, might lead to some envy and resentment and might make it easier for the politicians to involve non-professional rioters, I find it hard to believe that this is the case in all but a few of the hindu-muslim riots in the last century in India.
January 14, 2010 at 4:31 pm
Lyle_s
Is it possible, then, that the muslim population is generally so poor that they sit below the line that makes violence affordable? Maybe the increase in Muslim income gives them just enough money to justify a spend on violence. I didn’t read the paper but the summary suggests that only the theory and the economic evidence led them to decide that the Hindus were the attackers, not actual crime evidence. Maybe the Muslims have been suffering from a severe case of the second effect.