A Two-Stage Approach to Predicting Genocide and Politicide Onset in a Global Dataset
Benjamin E. Goldsmith
University of Sydney
Charles Robert Butcher
National Centre for Peace and Conflict Studies, University of Otago
University of New South Wales (UNSW)
affiliation not provided to SSRN
March 20, 2012
We present what is, to the best of our knowledge, the first published set of out-of-sample forecasts of genocide and politicide based on a global dataset. Our goal is to produce a prototype for a real-time model capable of forecasting one year into the future. While building on the current literature, we take several important steps towards better forecasting. We implement a two-stage modelling approach that considers both the likelihood of instability and the likelihood of genocide in a single estimate. Our sample is restricted only by the available data, rather than through selection of controlled cases, since such an approach would not be available for forecasting into the future. We explore factors exhibiting variance over time to improve year-on-year forecasting performance. And we produce annual lists of at-risk states in a format that should be of use to policy makers seeking to prevent such mass atrocities. Overall our out-of-sample forecasts for 1988-2003 predict 81.8% of genocide onsets correctly while also predicting 78.7% of non-onset years correctly, an improvement over a previous study using a case-control in-sample approach. We produce 16 annual forecasts based only on previous years’ data, which identify seven of eleven cases of genocide/politicide onset within the handful of the top 5% of at-risk countries per year. This represents a considerable step toward useful real-time forecasting of such rare events. We conclude by suggesting ways to further enhance predictive performance.
Number of Pages in PDF File: 31working papers series
Date posted: March 22, 2012 ; Last revised: March 1, 2014
© 2014 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo4 in 0.359 seconds