Working Paper: CEPR ID: DP13829
Authors: Christopher Blattman; Oeindrila Dube; Samuel Bazzi; Matthew Gudgeon; Richard Peck; Robert Blair
Abstract: Policymakers can take actions to prevent local conflict before it begins, if such violence can be accurately predicted. We examine the two countries with the richest available sub-national data: Colombia and Indonesia. We assemble two decades of finegrained violence data by type, alongside hundreds of annual risk factors. We predict violence one year ahead with a range of machine learning techniques. Models reliably identify persistent, high-violence hot spots. Violence is not simply autoregressive, as detailed histories of disaggregated violence perform best. Rich socio-economic data also substitute well for these histories. Even with such unusually rich data, however, the models poorly predict new outbreaks or escalations of violence. “Best case” scenarios with panel data fall short of workable early-warning systems.
Keywords: conflict prediction; Indonesia; Colombia; civil war; machine learning; forecasting
JEL Codes: D74; C52; C53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| socioeconomic factors (P23) | violence prediction (D74) |
| historical violence data (Y10) | violence prediction (D74) |
| detailed histories of disaggregated violence (D74) | violence prediction (D74) |
| rich socioeconomic data (D31) | violence risk (D74) |
| terrain ruggedness (R14) | violence (D74) |
| ethnic diversity (J15) | violence (D74) |
| historical violence data (Y10) | socioeconomic factors (P23) |
| socioeconomic factors (P23) | historical violence data (Y10) |