In a democracy, when and why are some politicians electorally punished for corrupt acts, while others get off scot-free? Some answers are commonsense: major scandals generally draw more ire than minor malfeasance; media coverage (and hence voter knowledge) matters; and citizens consider a variety of performance indicators—not just corruption or lack thereof—in selecting politicians. But the details are hazy. Some studies suggest politicians who get caught are more likely face electoral loss, but others find little to no such correlation. Likewise, we know anticorruption candidates often flounder for political reasons, but sometimes they succeed against the odds. So what drives, or contributes to, voter backlash against corrupt politicians?
A recent paper by Harvard scholars Horacio Larreguy, John Marshall, and James Snyder addresses this question in the context of mayoral elections in Mexico. Its conclusions should give pause to anticorruption activists looking for broad-brush solutions. In brief, the paper finds that the devil is in the details: local media coverage can reduce a corrupt incumbent’s vote share, but regional or national media doesn’t seem to matter much; voters do punish corrupt politicians on average, but certain political parties are punished much more than others for the same misconduct; and guaranteeing an audit of public programs reduces malfeasance, but merely threatening a possible audit has little if any effect.
These nuanced findings provide insight into voters’ habits, but they also reinforce the notion that corruption is deeply political—and therefore anticorruption interventions must be context-specific. To unpack this all a bit more, consider the study’s main findings:
- First, the authors found a statistically significant drop in voter support for mayors whose corrupt acts were reported by local media, particularly local radio and television stations. But regional and national media outlets did not have the same effect. (The authors note that their study was designed to err on the side of not finding significant effects, so it is possible that the regional or national stations had some impact. Nevertheless, the difference is striking.) Why is local media more effective? The data do not clearly answer that question. Local stations might be more in tune with voters concerns, and thus they might report more salient facts. Citizens might have greater access to local TV and radio stations, or may simply pay greater attention to local rather than national reporting. Regardless of the root cause, the discrepancy in effect suggests that the medium for transparency matters as much as the message.
- Second, the study also found a difference in electoral treatment between mayors from Mexico’s two largest national political parties, PRI and PAN. PRI falls to the political left of PAN, and the authors note it has “a stronger ‘pro-poor’ reputation” than PAN. After finding that PRI mayors were punished more severely for diverting funds from the poor than their PAN counterparts, the authors posited that this could be a “hypocrisy” effect—voters penalizing politicians for failing to live up to their promises. If so, that’s good news for anticorruption activists because it means that anticorruption rhetoric creates expectations of transparency that will influence citizens’ voting decisions. Get politicians to talk the talk, and voters will insist they walk the walk. But the trend could also reflect differing voter preferences. Voters more concerned with concrete performance than transparent process, for example, might broadly gravitate toward PAN, while the latter generally prefer PRI. And in any case, this finding reinforces the point above—political dynamics (here, differences between political parties) are a critical influence on how voters perceive corruption and what they are willing to tolerate. These dynamics probably also vary by region and constituency within a pluralistic democracy, meaning that a targeted transparency campaign could succeed in one district only to flop in the next.
- Third, only actual, reported audits, which the mayors knew would occur, stemmed bad behavior. In other words, threatening audits, or randomizing audits across regions, did not appear to dissuade officials from acting corruptly. This is, at first blush, very depressing. Mexico’s mayors either do not take the threat of auditing seriously or don’t believe they’ll be criminally convicted if they get caught. Changing the political dynamic—say, through a series of high-profile prosecutions and convictions—could help ratchet up the stakes of corruption and create a stronger incentive against malfeasance.
Overall, the paper reinforces the insight that voters tolerance for corruption is driven largely by political factors–including perception of likely punishment, party affiliation and messaging. Because local context is critical, these results suggest that caution is in order when prescribing broad-brush anticorruption interventions.