The exacerbating effect of police presence: A multivariate point process analysis of the Naxal conflict

The effect of police presence is an important consideration when designing counter-insurgency activities. While there has been much empirical research investigating the effects of counter-insurgency in a military setting, the effect of police presence in countering political dissidents has seen more limited attention. We argue that police presence at violent events is likely to exacerbate subsequent levels of violence rather than reduce them. To assess the validity of this claim we analyze a unique police dataset from the Indian state of Andhra Pradesh, which details the timing and location of police-recorded politically motivated violent offenses attributed to Naxals over a ten year period. The data also includes an indicator to determine whether or not police are present at each event, which we use to measure the exacerbating effect of police presence. Our analytical strategy is inspired by recent literature concerning crime and political violence that examine event interdependence at disaggregated levels of analysis. We calibrate a series of novel multivariate point process models, which are designed to test a series of hypotheses. Our results show that: 1) the Naxal conflict displays patterns of spatio-temporal clustering; 2) police presence at violent events exacerbated the conflict, leading to more events than would have otherwise occurred, at least in the short term; and 3) the space-time regularities can be usefully employed to predict the likely locations of future events.