A passer-by reads a newspaper on display at a news stand in Sao Paulo on July 9, 2014. NELSON ALMEIDA/AFP/Getty Images
Since Brazil’s humiliating 7-1 loss to Germany in the World Cup semifinals, the Internet has been filled with speculation about what this might do to Brazilian President Dilma Rousseff’s re-election hopes this fall (see here, here, here, and here for example). With this in mind, I reached out to Stanford University political scientist Neil Malhotra, who has conducted and published research on the effect of the outcome of sporting events in the United States on election results.
JT: Why are political scientists interested in whether results from sporting events affect election outcomes?
Malhotra: A canonical model of voting developed by political scientists is the “retrospective voting” model, which posits that voters use the past performance of incumbents to guide their decision making. There seems to be a lot of evidence for this model. U.S. presidential election results, for example, can be remarkably and accurately predicted by economic growth preceding the election. The harder question is how to interpret this behavior. Some scholars believe that such behavior reflects people rationally evaluating government performance and that it allows for the selection of good leaders. Others, such as Larry Bartels and Chris Achen, believe that people engage in “blind retrospection”: what looks like measured evaluation of performance is actually just emotional responses to the prevailing mood of the day. However, it’s hard to test whether voters are actually engaging in retrospective voting by looking at how the economy influences elections, because both mechanisms could be operating. Similarly, even looking at how voters respond to events such as floods and shark attacks has issues because voters could expect the government to respond to these events. However, sports outcomes represent a great empirical test because they can affect people’s mood and satisfaction with the status quo, but are totally irrelevant to government activity. In sum, we should be interested in whether sport results affect election outcomes because it tells us about how democracy works.
JT: What did you find in your research on this topic? Did incumbents do worse when their teams lost?
Malhotra: We collected data on college football outcomes around elections, along with county-level election returns for presidential, gubernatorial, and Senate elections from 1964-2008. We found that losses immediately preceding Election Day (compared to wins) caused incumbent to lose about 1 percentage point of vote share. The effects were about double for major college football teams, defined as those with high attendance or those that had won national championships. This result is quite robust. We conducted “placebo” tests by predicting election results with games that occurred *after* Election Day and found no results. We created a natural experiment where we controlled for which team was expected to win (using betting spreads) and found that surprising wins and losses (like the Brazil-Germany match) have a more powerful effect.
JT: In your research, you examined elections that took place very shortly after the sporting events in question. With Brazil’s election not scheduled until October, is that too far away to expect a “World Cup” effect?
Malhotra: I don’t want to extrapolate too much from our results, but my intuition would suggest that the effects probably wear off over time. We did not observe effects of games played three weeks or more before Election Day. However, one thing we didn’t consider in our paper was candidate strategy. Challengers may create campaign themes around Brazil’s devastating loss and therefore make the effects more long-lasting. Also, the games we studied were regular season games. This particular loss may be especially painful to a soccer-obsessed country like Brazil, given that it was a loss on home soil in the most important soccer competition.
JT: Given the work you’ve done in this area, what’s your take away point for everyone speculating about the effect of this game on Rousseff’s re-election chances?
Malhotra: I am fairly confident that at least part of what we call “retrospective voting” is not a rational assessment of performance but rather mood-driven satisfaction or dissatisfaction with the status quo. To the extent that feelings from the World Cup loss (which is a big deal in a country like Brazil) can be strategically used by politicians to make people unsatisfied with the status quo (despite actual economic data to the contrary), the game could have an effect.
