Accountable or Untouchable? Electoral accountability in Romanian local elections

While retrospective models of voting posit that voters should "vote the rascals out", a wave of recent research has found that this is rarely the case. We investigate this question in a context in which many sitting politicians have recently been indicted on corruption charges - the municipal level in Romania, a surprisingly under-researched case in this sub-field. Romania provides a good case for electoral accountability. Not only do Romanians deeply detest corruption, the party system also contains many parties that would make it easy for voters to switch from a corrupt to a cleaner alternative. We collected an original data register of electoral and socio-political data on roughly 3200 localities together with all cases of corruption charges published by the Romanian anti-corruption agency, the Directia Nationala Anticoruptie (DNA), accounting for magnitude and timing of the scandal as well as the judicial outcome for the indicted mayor. In all, we find that 81 sitting mayors elected in 2012 were charged with corruption prior to the 2016 election. We test the electoral impact of corruption on the incumbent mayors on four outcomes indicating electoral accountability commonly used in the literature - retirement, vote share compared to the previous election, voter turnout, and reelection using difference and difference and a pairwise matching designs, inter alia. The results show that Romanians do punish their corrupt incumbent mayors to a quite high extent compared to the clean mayors. However, due to the large vote margins, the punishment is not severe enough to make them lose more often than similar "dead' mayors, although they tend to not run for reelection at much higher rates. Turnout is unaffected by corruption at the municipal level. In line with previous results, we thus find a certain amount of electoral accountability, but not to the extent that the 'rascals are thrown out'.