UC Berkeley

This dissertation investigates how political parties can undermine the representation of citizen interests in new democracies. Conventional wisdom has emphasized the centrality of parties in mediating the relationship between voters and politicians, and has often attributed the representational deficit observed across the developing world to the lack of stable partisan attachments or the ephemeral nature of political parties. I show that this may not be the case. To the contrary, under a political geography that enables political parties to repeatedly monopolize electoral support from voters in subnational elections, parties and the internal processes that govern their selection of candidates can function sever, rather than strengthen, the connection between voters and their representatives.

My theory focuses on how conditions typical of many new democracies in Sub-Saharan Africa---local one party dominance and centralized control over candidate selection---shape the incentives of politicians to serve the interests of their constituents. Despite the institutionalization of competitive multiparty elections at the national level in new democracies, political parties are often able to consistently dominate their competitors in subnational elections. Under conditions of local one party dominance, politicians who contest local office become beholden to the selectorate which decides who the party's nominee will be, without much regard to the electorate. Yet for the party leader, who controls the selection of candidates within their parties, local politicians who amass an independent support base by serving the interests of their constituents pose a significant threat towards maintaining her position in the party hierarchy. I argue that the party leader selects candidates in a way that minimizes the risk of politicians building such an independent support base, encouraging responsiveness to constituents only in select locales where they are electorally vulnerable. As a result, politicians are incentivized to divert their effort and resources away from serving their constituents towards other activities that benefit the party leader.

I support these claims using a multi-method research strategy that pieces together qualitative, quantitative, and experimental evidence based on 18 months of fieldwork in Kenya. I first combine insights from more than 70 politician interviews and analyses of nationally representative surveys and constituency-level electoral returns across six African democracies to establish that African parties often hold a monopoly on local power. Moreover, using detailed inquiry into the organization of political parties in Kenya and a series of experiments conducted among Kenyan primary voters, I also show that party leaders possess both the institutional tools and the persuasive influence over partisans that enable them to command control over the candidate selection process. Finally, I use supervised machine-learning methods on a large data set collected through web-crawling to document the existence of a nomination tournament, in which party leaders select candidates that invest significantly in "party-oriented" rather than "constituency-oriented" behavior over their terms in office.

Substantively, the findings contribute to the emerging consensus that democratic elections are necessary but insufficient to foster better representation and responsiveness for the people. However, while the dominant narrative in the comparative politics has focused on structural-institutional factors such as ethnicity, clientelism, or electoral systems to understand this deficit, I shift the attention rightfully back to political parties. In fact, the conclusions of the dissertation suggest that ideal of "representative democracy" is likely to remain elusive unless democracy within political parties is realized. When power and authority over party institutions and decision-making processes accumulate to a single individual or a small group of elites, and without systematic checks to constrain their power, party leaders have the potential to effectively become autocrats within their domain; manipulating elected representatives who should primarily be interested in tending to their constituents to serve their political ambitions, thereby derailing democratic process that they should protect.