We propose to view social conflicts in Africa as having strong similarities with earthquake occurrences and hence to consider the spatial-temporal Epidemic Type Aftershock Sequence (ETAS) model. The parameters of this highly parameterized model are estimated through simulated annealing. We consider data for 2012 to 2016 to calibrate the model for four African regions separately, and we consider the data for 2017 to evaluate the forecasts. These forecasts concern the amount of future large events as well as their locations. Examples of our findings are that the model predicts a cluster of large events in the Central Africa region, which was not expected based on past events, and that in particular for East Africa it apparently holds that small conflicts can trigger a larger number of conflicts.

, , ,
, ,
hdl.handle.net/1765/129572
Journal of Forecasting
Department of Econometrics

van Hengel, G., & Franses, P. H. (2018). Forecasting Social Conflicts in Africa Using an Epidemic Type Aftershock Sequence Model. Journal of Forecasting, 2, 284–308. Retrieved from http://hdl.handle.net/1765/129572