| In sub-Saharan Africa, low levels of taxation are correlated with low levels of accountability and high levels of corruption. This work argues that existing theories of why taxation increases accountability overlook a key micro-level mechanism. In particular, I argue that taxation fundamentally decreases citizens' tolerance for corruption and increases their willingness to hold government officials accountable. I develop this theory formally and then test it using a set of novel laboratory experiments in Uganda, demonstrating support for the hypothesized effect and the proposed mechanism. I then expand these findings in two key areas. First, I use a survey experiment of politically-active Ugandans to show how the framework of accountability developed in prior chapters can also explain other types of variation in citizens' demand for punishment. Then, I develop a game-theoretic model of how and when rent-seeking governments will tax citizens, using the evolution of taxation in Uganda since 1996 to support the model's predictions.;I argue that loss aversion can explain why taxed citizens are more likely to engage in costly actions to punish wrongdoing by government officials. By pushing citizens below their reference point, taxation increases the economic utility citizens lose from corruption. If those economic losses are also tied to the psychological, expressive benefits citizens receive from taking part in collective action such as protesting or voting, taxed citizens will be more likely than untaxed citizens to punish government officials for corruption. Laboratory experiments in Uganda show that taxation increases citizens' willingness to punish a non-accountable "Leader" by 13% overall, and by 30% among adult, wage-earning men, who are the group who has the most exposure to taxation in Uganda. Further experiments show that the results are due to the proposed loss-aversion mechanism, rather than alternative mechanisms such as fairness norms. I then use a conjoint survey experiment of politically-active Ugandans to show that, even outside of a laboratory setting, Ugandan citizens are more likely to want to prosecute an official for corruption when he has stolen tax funds, rather than central transfer or aid money; the experiment also provides broader support for the theory of accountability developed in the dissertation, showing that citizens are more willing to punish corruption when it involves officials of whom they have higher expectations, when it violates societal norms, or when it directly impacts goods and services that citizens value.;Given that taxing citizens increases accountability demands, how should we expect a rent-seeking government to respond? The final third of the dissertation develops a game-theoretic model to examine when governments will tax citizens, and when they will seek to engage in tax bargaining as opposed to taxing coercively. I demonstrate that democratization can actually lead to a decrease in taxation, especially when tax bargaining is difficult to sustain; I also provide predictions for when tax bargaining is most likely. I then use the evolution of taxation in Uganda since 1996 to support the model's predictions; where tax bargaining is not feasible, increased political competition in Uganda is associated with a decrease in taxation and public goods provision, while when tax bargaining is feasible an increase in competition leads to an increase in accountability. |
