Essays on Party Politics

Abstract

This thesis comprises three essays on party politics. The first chapter discusses the strategic behavior of the central leadership and the grassroots organizations of a political party. The focus is on the degree of centralization of fundraising, which determines the locus of control over the nature of the party's platform. In a model of central and local fundraising, I show that (i) public and private funding at the central level crowd out the fundraising efforts of the national party's local organizations, leading to the centralization of power; and (ii) that there exists a collective action problem among local organizations as they attempt to free-ride on transfers from the center. The equilibrium degree of centralization under various conditions is simulated to develop testable hypotheses. Canadian data from 2005 to 2013 are then used to study the impact of central party funding on net transfers from the center to the party’s local organizations. The empirical results show that an increase in central party funding (e.g., public funding) increases net transfers, while higher funding by the national party negatively impacts local fundraising efforts.

In the second chapter, I develop a methodology to deal with continuous treatment levels within a regression discontinuity design, and I apply this to study the effect of public subsidies on fundraising. Introducing continuous treatment allows the estimation of a dose-response function for outcomes conditional on the level of treatment. Using this framework, I study empirically how public subsidies in Canada allocated to grassroots political organizations impact their private fundraising: in the Canadian case I explore, a local candidate who wins over 10% of the vote was eligible for reimbursement of 60% of his electoral expenses. The results uncover a crowding-out effect. Wealthier grassroots organizations exhibit a decrease in fundraising after receiving the reimbursement. A survival effect also appears as the poorer grassroots organizations are found to increase their fundraising activities following reimbursement.

Finally, the third chapter considers issues of randomness and unobserved heterogeneity that typically arise when using a regression discontinuity design (RDD) to study the incumbency effect, or indeed in any controlled experiment. In studies of incumbency, randomness depends on the selection of 'close' elections as typically measured by a 'small' winning vote margin. However, the voters' willingness to change their vote is a critical component of a better, volatility-adjusted, measure of winning margins. We study empirically how the use of volatility-adjusted instead of simple margins alters the outcome of an RDD analysis of incumbency for Canadian federal elections. Unobserved heterogeneity in the selection of constituencies used in such an analysis arises when the electoral readiness of a political organization - a potentially important determinant of its candidate's success - is not considered. We investigate empirically how allowance for the quality of the organizations contesting an election affects the outcome of an RDD study of incumbency over the history of federal elections. We find that adjusting for volatility alters conclusions about the exact magnitude of the incumbency effect while taking organizational quality into account eliminates the incumbency effect in the Canadian case.