Voting and Electoral Competition

Citation

Callander, Steven Joseph (2002) Voting and Electoral Competition. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/67R6-E540. https://resolver.caltech.edu/CaltechETD:etd-01252008-133155

Abstract

The behavior of individuals and groups in the political realm is subject to many and varied incentives. These incentives impact significantly not only the candidates who win elections, but also the policies that they implement. This thesis analyzes several aspects of this problem that have until now gone unexplained.

Part 1 contains two models of candidate competition. Chapter 1 details a model of competition under the plurality rule that simultaneously explains two well-documented empirical regularities: that typically only two parties compete in each election (Duverger's Law), and that these parties choose non-centrist policy platforms. I show that if, and only if, competition is for multiple districts does an equilibrium consistent with these phenomena exist. I characterize bounds on district heterogeneity for this to be true, which can be interpreted as describing the domain for Duverger's Law. In Chapter 2, I turn attention to the run-off rule and study a similar model to that of Chapter 1. I find that this subtle change to the counting rule has a significant impact on the incentives and equilibria of the model. In the traditional single district environment there now exists a continuum of two-party non-centrist equilibria, which are robust to simultaneous competition for multiple districts.

In Part 2, I investigate the behavior of voters, and particularly the effect of vote timing on voter behavior and election outcomes. In Chapter 3, I study a model of sequential voting and explain when and why the commonly observed phenomena of bandwagons and momentum arise. I show that only if voters have a desire to vote for the winning candidate, in addition to their desire to select the better candidate, is momentum observed and bandwagons begun. In Chapter 4, I compare these results with analogous results for when voting is simultaneous and characterize when each process is superior. The conclusions confirm commonly held views about the front-loading of U.S. presidential primaries: that in tight races a simultaneous vote is preferred, but in lopsided races a sequential vote is better. Strangely, the superior performance of sequential voting in lopsided races is precisely because bandwagons occur.

Thesis Files