Retrospective Voting in the Premodern World: The Case of Natural Disasters in the Roman Republic

Political scientists frequently debate how “blind†retrospective voting is in the modern world, and the extent to which voters punish or reward officials for things beyond their control, such as disastrous flooding. Results vary based on the time and location of the study, as well as methodology. We test these arguments in a radically different time and place: the Roman Republic between 218 and 166 B.C.E. We articulate a clear theoretical expectation: that voters observe and reward response to local crises. We then leverage the exogenous nature of river flooding as well as the random assignment to office location in the Roman Republic to estimate the causal effect of flooding on those randomly assigned to be in the city when floods occurred. Analyzing 272 second-tier executive officials in the Republic, we find that significant flooding of the Tiber River substantially increased the ability of those randomly placed in the city to achieve the top executive office in the Republic in the near future. Among those randomly assigned to be outside of the city, we find no evidence of any effect on short-term career advancement. These results indicate that floods were opportunities to perform for city voters. This further illustrates that theories of voting and elections inductively formed based on scholars' familiarity with modern American elections may be limited in explanatory power to those contexts.