Early warning models for systemic banking crises: can political indicators improve prediction?

This study provides the first attempt to evaluate whether a logit early warning system (EWS) for systemic banking crises can produce better predictions when political indicators are used alongside traditional macro-financial indicators. Based on a dataset covering 32 advanced economies for the period 1975-2017, we show that the inclusion of political indicators helps improve the predictive performance of the model. While the improvement is small, it is statistically significant and consistent for several different performance measures and robustness tests. Among the newly employed political variables, variables indicating the political ideology of the ruling party and the time in office of the incumbent chief executive show significant correlations with the likelihood of systemic banking crises. The results suggest that a systemic banking crisis is less likely when the government is left-wing and when the chief executive officer has been in office longer.