Citizens living in the developing world are frequently exposed to adverse conditions generated by negative events, such as economic crises, natural disasters, and crime waves. These circumstances tend to reduce citizens’ incomes and diminish their living conditions. Do voters change their political behavior after negative events, and if so, how and why? The extant literature tends to focus on how citizens punish or reward the incumbent candidate or party based on a model of (mis)attribution of responsibilities. This approach overlooks the forward-looking dimension of electoral decisions. When the incumbent has handled a negative event poorly, we can expect that voters will punish her or him and select a candidate from among the pool of challengers, but we do not know which other candidate would be more likely to be preferred. In my dissertation, Political Preferences in Adverse Conditions, I argue that affected citizens, besides blaming or rewarding incumbents for how they responded to the negative shock, are also selecting the political candidate who they believe can best enhance their welfare. I hold that victims focus on improving their standard of living, which will affect the policy issues they most care about. As a consequence, victims are more likely to prefer candidates who can better address these new policy preferences.

In my job market paper, I study the impact of natural disasters on victims’ political preferences. Climate change has increased our concerns about the frequency and intensity of disasters. According to my theory, citizens affected by catastrophes seek to reduce the gap between their living conditions before and after the disaster. This leads them to focus on welfare and social policies, for example, new housing. However, it is difficult to study the consequences of disaster damage because it can be correlated with several unobserved characteristics. I focus on flash floods that occurred in Chile in 2015. I use the natural experiment created by the flood damage to identify exposed and unexposed citizens, and I implement a conjoint survey experiment to measure the impact of the disaster on voters’ political preferences. The findings show that material damage due to the flood increased the probability of preferring left-wing candidates by 12 percentage points. These candidates could be associated with welfare and social policies that could improve victims’ living conditions. Qualitative evidence from interviews illustrates how disaster victims emphasize the importance of welfare policies that could improve their living conditions. In addition, I compare these findings to the actual electoral response to the flood and provide survey evidence from a different natural disaster to increase external validity. Besides punishing or rewarding incumbents, victims select the candidate who can provide the social policies they need. As a consequence, left-wing parties and candidates should have a natural electoral advantage after disasters.

In another paper derived from my dissertation, I study how crime victims change their policy preferences. I show that crime victims are more likely to support strong-handed measures to reduce crime, such as allowing state repression. These results reveal that crime can change what people think the state should be allowed to do, which can have important political implications. I use panel data from Brazil and implement strategies for reducing sensitivity to hidden biases. For example, I focus on individuals who were not crime victims during a previous wave, which allows me to mitigate some methodological concerns such as serial victimization and reverse causality. In addition, I use survey data from 18 Latin American countries to improve the external validity of the findings.

In a third paper drawn from my dissertation research, I implement an original survey experiment in Chile to compare how voters react to disasters and crime victimization. The survey randomly presents different hypothetical scenarios to participants, and then evaluates their policy and electoral preferences using a conjoint survey experiment. The three scenarios are: being a disaster victim, being a crime victim, and being neither as a control condition. I argue that the disaster treatment makes respondents more likely to support government welfare distribution, and to vote for left-wing candidates. The crime treatment, in contrast, makes participants more likely to support strong-handed measures to reduce crime, and to vote for right-wing candidates. The survey also shows that citizens are able to connect simple policy outcomes (e.g., welfare and iron-fist policies) with particular ideological labels (e.g., left and right) in Chile. In countries where ideology is less salient, affected citizens should be able to connect policy outcomes with party labels or other candidate characteristics.

These three dissertation papers teach us that policy preferences are not fixed: they can change according to the context. In addition, voters can exploit simple shortcuts to connect candidates with their new policy concerns. As a consequence, disaster and crime victims are more prone to vote for candidates whom they would not vote for in other contexts.