What influence do emotions that originate outside politics have on German federal elections? In this paper, we estimate the causal effects of football (soccer) results from the first German Bundesliga on party-list district data during the election weekend in 2013. The results from a difference-in-differences design analysis relative to the 2009 elections reveal that football results are a quasi-random treatment and thus function as a natural experiment with regional variation. Incumbent parties benefited from a positive outcome for the local football team, as measured by the difference in goals in a statistically significant result. In contrast, opposition parties did not lose clearly from these positive outcomes. The results for the oppoistion parties were sometimes significant, but always had the expected points estimates. All effects were moderated by the timing of the matches relative to the closing of the election: the greater the time lag, the smaller the effects, although these effects were not always statistically significant, either. There was no evidence that betting odds, as the best information available before the match about the potential result, moderated the effects. Overall, the results only partially support the aggregated implications of the micro-logic of "blind retrospection" due to imprecise and instable estimates.
