Vote Buying and Accountability in Democratic Africa
[Thesis]
Kramon, Eric
Posner, Daniel N.Geddes, Barbara
2013
Posner, Daniel N.Geddes, Barbara
2013
The distribution of cash to voters during elections, vote buying, is extremely widespread in many democracies. That vote buying is so widespread raises concerns about the quality of emerging democratic institutions and the potential for elections to deliver better and more accountable government. I develop a new theory to explain why politicians in new democracies distribute money to potential voters. I argue that cash handouts are effective because they convey information to voters about the extent to which a candidate will protect and serve their interests in the future, especially with respect to the provision of patronage resources. I test this informational theory with observational and experimental data collected in Kenya, as well with existing data from a larger set of African countries. As vote buying is secretive and sensitive, and so survey and interview responses are subject to response bias, I use several survey and experimental methods to improve descriptive and causal inferences about vote buying. In a variety of empirical tests, I provide evidence directly consistent with the informational theory; I show that patterns in the prevalence and geographic allocation of vote buying across and within African countries are best explained by the informational theory; and I provide evidence that helps to rule out existing explanations. Chapter 2 analyzes data from a nationally representative survey and survey list experiment, a method that reduces response bias in survey questions, to show that cash handouts influenced the vote choice of about 20 percent of Kenyans during the country's 2007 elections. Chapter 3 presents the informational theory and provides preliminary evidence from existing ethnographic studies and survey data from 18 African countries. Chapter 4 shows that existing explanations for vote buying, which focus on the role of political machines or on the mobilization of voter turnout, are insufficient to explain widespread cash handouts in Kenya and other African settings. Chapter 5 analyzes survey experimental data to show that, when Kenyan voters hear that a political candidate has distributed cash, they prefer that candidate to an otherwise identical candidate who has not done so. This effect is especially strong among poorer voters. Additionally, vote buying increases voters' expectations that a candidate will provide them with patronage and private benefits in the future, direct evidence consistent with the informational argument. Chapter 6 shows that, in conveying this information about patronage, vote buying reinforces and perpetuates patterns of ethnic voting---that is, the propensity of voters to support members of their own ethnic group at the polls. Experimental results show, in contrast to psychological or expressive theories of ethnic voting, that participants only prefer coethnic candidates and only expect to benefit from their patronage when they are engaged in vote buying. I demonstrate external validity by showing that vote buying has the most influence on vote choice when voters are targeted by members of their own ethnic group. Chapter 7 uses data about the geographic allocation of local public goods projects in Kenya to show that vote buying is associated with more patronage allocations after an election. This result is consistent with the idea that handouts can be an informative signal about future performance. Chapter 8 shows that cash handouts not only help to convince voters that they should support a particular candidate, they also mobilize them to turn out to vote in order to gain access to the patronage resources that the patron will allocate in the future. The last empirical chapter, Chapter 9, steps back from the question of why vote buying is effective to ask why vote-getting strategies takes the form of direct vote buying in some places but not others. With data from across Africa, I show that direct vote buying is most prevalent where local intermediaries, in this case traditional rulers, are not present or powerful enough to deliver large numbers of voters in a block. That is, politicians in Africa directly hand out cash in settings where they must win votes without the assistance of strong local brokers, a pattern that the informational theory is best suited to explain. I conclude the dissertation by discussing the implications of the results, which complicate normative interpretations of vote buying.