This dissertation studies politicians' and voters' behavior. The first chapter studies whether politicians strategically respond when confronted with potential electoral backlash from increased transparency. I answer this question using exogenous variation generated by randomized audits to local governments in Brazil. I show that the execution of an audit leads to an increase in the number of public employees hired by the mayor. This effect is greater in municipalities where auditors uncovered higher levels of corruption. I find evidence consistent with mayors hiring more employees as a form of patronage to compensate for the loss of electoral support resulting from the audit. I closely examine the education sector using additional detailed data and find that hiring more school employees does not improve student outcomes, revealing limited direct consequences of audits on public goods production. Moreover, I show that an audit increases the share of payroll expenditures but decreases capital investment and that this substitution translated into a deterioration in the quality of school assets. These results suggest that patronage enables politicians to offset the potential electoral penalty of an audit by hiring employees who do not contribute much to public goods production. In the second chapter, we study voters' response to marginal changes to the fine for electoral abstention in Peru, leveraging variation from a nationwide reform. A smaller fine has a robust, negative effect on voter turnout, partly through irregular changes in voter registration. However, representation is largely unaffected, as most of the lost votes are blank or invalid. We also show that the effect of an exemption from compulsory voting is substantially larger than that of a full fine reduction, suggesting that non-monetary incentives are the main drivers behind the effectiveness of compulsory voting. In the third chapter, I examine changes in the Peru's two largest export products: copper and gold. Using difference-in-differences estimators, I show that higher levels of gold production increases (decreases) the vote share for anti-mining (pro-mining) parties. Conversely, an increase in copper production leads to a close to zero—and not statistically significant—change in the vote shares for anti- and pro-mining parties. Moreover, I find evidence consistent with voters caring about the environmental damage caused by gold production when casting their vote and prioritizing this effect over improvements in socioeconomic indicators. Overall, these findings highlight that voters heavily weigh social considerations compared to self-interest when deciding for whom to vote.