Abstract:
|
After decades of violence and conflict, and following several natural disasters, there are over a million internally displaced people (IDPs) in Afghanistan. However, statistics on internal displacement are notoriously unreliable, due to the lack of formal reporting procedures and the difficulty of tracking people over time. In this paper, we develop a method to measure internal displacement from passively collected 'digital trace' data, and to estimate the causal effect of discrete events on human displacement. We use these methods to study the effect of violence on internal displacement in Afghanistan, using several years of pseudonymized mobile phone data obtained from Afghanistan's largest mobile phone operator. Analyzing the response to thousands of conflict-related events catalogued by the Uppsala Conflict Data Program, we find evidence of substantial short- and long-term displacement. These effects are heterogeneous, and depend on factors such as the location of the violent event and pre-event peace time.
|