Political Information Cycles: When Do Voters Sanction Incumbent Parties for High Homicide Rates?

Do voters sanction incumbent parties for their performance in office? I argue that how governments are held to account depends upon when voters consume informa-tion about the relevant incumbent party. If news consumption follows electoral cycles, short-term performance indicators in the news prior to elections may powerfully shape voting behavior. In the context of local homicides and Mexican municipal elections, I test this theory’s central implications using three distinct sources of plausibly ex-ogenous variation. First, I show that voters indeed consume more news before local elections, and that homicides before such elections increase the salience of public se-curity and reduce confidence in the mayor. Second, electoral returns confirm that pre-election homicide shocks substantially decrease the incumbent party’s vote share and re-election probability. However, such sanctioning is limited to mayoral elections, and is barely impacted by longer-term homicide rates. Finally, the punishment of homicide shocks relies on, and increases with, access to local broadcast media stations. These effects are only pronounced among less-informed voters, who principally engage with politics around elections. The findings demonstrate the importance of when voters consume news, and contribute to explaining the mixed electoral accountability often observed outside consolidated democracies and in federal systems.