The Promise and Pitfalls of Conflict Prediction: Evidence from Colombia and Indonesia

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 one fine- grained 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.

The impact of forced displacement on host communities: A review of the empirical literature in economics

“…In a regression framework, we test these implications by estimating the relationships between iand j-specific natural resource endowments with whether or not regions i and j were involved in the same conflict. In keeping with recent work that finds time-invariant characteristics are often better at predicting conflict than shocks (Bazzi et al, 2018), we focus on the longer-run "endowment" of resources in the cross-section. To address concerns regarding identification that arise from the use of the cross-sectional dimension of the data rather than time-varying shocks, we show that our results are robust to controlling for local geographic, agricultural, and climatological characteristics, as well as spatial fixed e↵ects of varying size.…”

“…Recent work shows that cross-sectional variation is a better predictor of conflict than time-varying shocks(Bazzi et al, 2018).…”

Conflict and the Composition of Economic Activity in Afghanistan

“…The irregular nature of Afghanistan's warfare gives support to this assumption (Ciarli et al 2015). While a deterioration of the overall security situation could be anticipated as the international presence was scaled down, where exactly conflict would unfold is much more difficult to predict (Bazzi et al 2019). This uncertainty over security risks makes local surges in conflict plausibly exogenous from the point of view of the empirical analysis.…”

Resources, Conflict, and Economic Development in Africa

“…In a regression framework, we test these implications by estimating the relationships between iand j-specific natural resource endowments with whether or not regions i and j were involved in the same conflict. In keeping with recent work that finds time-invariant characteristics are often better at predicting conflict than shocks (Bazzi et al, 2017), we focus on the longer-run "endowment" of resources in the cross-section. To address concerns regarding identification that arise from the use of the cross-sectional dimension of the data rather than time-varying shocks, we show that our results are robust to controlling for local geographic, agricultural, and climatological characteristics, as well as spatial fixed effects of varying size.…”

“…Recent work shows that cross-sectional variation is a better predictor of conflict than time-varying shocks(Bazzi et al, 2017).…”