Uneven accountability in the wake of political violence

The government faces a principal–agent problem with lower-level state officers. Officers are often expected to use the state coercive capacity endowed to them to politically benefit the government. But officers can shirk from the government’s demands. An officer’s actions during bouts of large-scale and highly visible electoral violence reveal the officer’s type, thereby providing the government with the information necessary to solve its principal–agent problem for the future. The government holds officers who used their authority to perpetuate incumbent-instigated violence accountable through positive rewards, while holding officers who used their authority to perpetuate opposition-instigated violence accountable through negative sanctions. We find evidence in support of the theory using micro-level archival data on 2,500 local officer appointments and fine-grained satellite data on the locations of violence in the aftermath of Kenya’s 2007 election. The Kenyan government was more likely to fire officials whose jurisdictions saw opposition-instigated violence that targeted government supporters. But we find the opposite result where violence was instigated by incumbent supporters: there, officers were less likely to be fired if violence occurred in their jurisdiction. Our results indicate that leaders can manipulate accountability processes after political violence to further politicize the state.