Call Us: US - +1 845 478 5244 | UK - +44 20 7193 7850 | AUS - +61 2 8005 4826

“Deprivation and Race Riots

Our main result ðproposition 1Þ states that if a group is relatively poor
to begin with, an increase in the average incomes of the group—controlling for changes in inequality—must raise violence perpetrated against
that group. In contrast, the effect on violence perpetrated by that group
on members of the other group is generally negative.2
We use a unique data set on Hindu-Muslim violence between 1950 and
1995, compiled by Ashutosh Varshney and Steve Wilkinson, and extended by us to 2000. It summarizes reports from The Times of India on
Hindu-Muslim conflicts in India in the second half of the twentieth
century. We use counts of the number of people killed or injured or the
number of riot outbreaks.
We match the data to the large-scale household surveys that are conducted quinquennially as part of the National Sample Surveys ðNSSÞ. We
use data from three consecutive “thick rounds”: the thirty-eighth in
1983, the forty-third in 1987–88, and the fiftieth in 1993–94. We compute average per capita monthly expenditures in each round for Hindu
2 These nuanced connections between economic growth and conflict suggest that the
overall relationship between the two could be nonmonotonic. Dube and Vargas ð2013Þ
make a parallel observation in the context of resource shocks and violence in Colombia.
economic theory of conflict 721
This content downloaded from on Tue, 9 Sep 2014 10:18:07 AM
All use subject to JSTOR Terms and Conditions
and Muslim households in 55 regions and so work with a three-period
In several different panel specifications with different sets of controls,
Hindu per capita expenditures have a negative effect on conflict, while
the coefficient on Muslim per capita expenditures is significant and positive. The coefficients are also large. Depending on the exact specification ðsee table 3 below for baseline resultsÞ, a 1 percent increase in Hindu
per capita expenditure is predicted to decrease casualties by anywhere
between 3 percent and 7 percent, while the same increase in Muslim per
capita expenditure increases casualties by 3–5 percent. We conclude that
an increase in Hindu prosperity is negatively associated with greater religious fatalities in the near future, while the opposite is true of Muslim