Research

Working Papers*

International Migration Amid Insurgent Violence (Job Market Paper)

This paper develops a method for inferring international migration events from anonymous cell phone metadata and estimates the effects of political violence on human mobility and displacement. Using call detail records from more than 15,000 unique mobile phone users over a 17-month period, I analyze patterns of individual movement during a violent period of the Taliban insurgency in Afghanistan (2011-2012). This project illustrates how digital trace data can provide behavioral and economic context in otherwise data-scarce settings. I leverage the spatial and temporal richness of call data to develop a series of classifying criteria to infer which attrition events are plausibly due to leaving the country. Upon identifying potential cross-border migration episodes, I estimate an increasing likelihood of migration with additional exposures to insurgent violence, consistent with the literature on the determinants of forced displacement.

Internet Access and Immigration

This paper describes the effects of the expansion of internet infrastructure on immigration between Mexico and the United States. Specifically, it asks: does the arrival of mobile internet (3G) in a Mexican municipio fundamentally change individuals' migration choice? There are several margins on which one might expect an effect--the binary migration choice itself, as well as several related decisions like optimal routes and crossing points, US destination choice, and uptake of travel aid. The project seeks to measure the relative strength of two different mechanisms within this migration choice decision. On one hand, 3G access introduces (or strengthens) an important information channel for shaping expectations of life in the destination country, and potentially avoiding risks en route. On the other hand, 3G is a valuable amenity whose addition to the home community could increase the utility of staying. Additionally, the paper examines a relevant side effect of the aforementioned information channel, namely the possibility of price convergence for the illicit services of border-crossing guides, or coyotes.

Cell Phone Coverage and Violence: Evidence from a Radio Propagation Model (with Robert Gonzalez and Lin Yang)

This paper examines the impact of access to mobile phone coverage on insurgent violence. Access to coverage can lower violence by facilitating information sharing by civilians and shielding informers from retaliation. On the other hand, coverage can increase violence by reducing the cost of producing violence (e.g., facilitating coordination among insurgents, remote detonation of IEDs). Empirically, this paper finds evidence that the net effect of access to mobile phone coverage is to lower insurgent violence. We estimate a radio-wave propagation model that uses variations in terrain topography and the spatial distribution of mobile phone towers to predict signal strength on the ground for each cell of a 1X1 kilometer grid of Afghanistan. We then employ a regression discontinuity design that compares cells at the margin of the signal strength threshold required for coverage. Cells with just enough coverage experience a two percentage point drop in the likelihood of any attack and a 0.8 percentage point drop in the likelihood of an IED. Further analysis suggests that information sharing is a key mechanism. The impact of coverage is larger in cells where detection of insurgent activities by civilians is more likely: near populated areas, primary roads, and during morning hours.