Forecasting political developments with the help of financial markets

Abstract

Using standard econometric techniques, I test the relative predictive accuracy of naïve autoregressive models in predicting political events with competing models that contain information from financial markets. The evidence assembled shows that information from stock markets sensibly improves the predictive accuracy of models that largely rely on the auto-regression of the series under examination to produce forecasts. The lagged development of the Tel Aviv Stock Exchange allows me to predict the level of cooperation in the conflict between Israel and the Palestinians the following day more precisely than with an alternative model that just takes the lagged political interactions into account. Because violent events come much more often as a surprise than cooperative developments, the inclusion of financial market information does, in return, not improve the accuracy in the prediction of the daily level of conflict in the Middle East.