Let the Rain Settle It: Estimating the Effect of Dissent on Repression

Abstract

Although many factors have been found to influence government willingness to violate human rights, there is a scholarly consensus that state repression is an attempt to control popular dissent. However, dissent and repression are endogenous, in that the state and citizens decide on their actions in expectation of the other’s behavior. To date, empirical models have not accounted well for this endogeneity, leaving scholars either unable to estimate the effects of an independent variable on rights violations (at best) or drawing incorrect conclusions as to these effects (at worst). To solve this problem, we follow recent US voting scholarship and propose an new instrument to model the endogeneity of dissent to repression: rainfall. We argue that while rainfall will deter citizens from active protest, it is unlikely to affect repressive capacity. We couple annual data on rainfall deviations with data from the Social Conflict in Africa Database (SCAD), which reports information on protest and repression in all African states from 1990 to 2009. Using an instrumental variable (IV) approach, we find that dissent fails to have either a statistically or substantively significant effect on repression when we take endogeneity seriously.

Keywords: state repression, mobilized dissent, instrumental variable, strategic interactions