Strategic agents in voting games

• Main Author: Hortala-Vallve, Rafael
• Published: London School of Economics and Political Science (University of London) 2005
• Subjects: 324.6
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.416950

The first part of this Thesis asks whether we can devise voting rules that allow strategic voters to express the intensity of their preferences. As opposed to the classical voting system (one person - one decision - one vote), we first propose a new voting system where agents are endowed with a fixed number of votes that can be distributed freely between a predetermined number of issues that have to be approved or dismissed. Its novelty, and appeal, relies on allowing voters to express the intensity of their preferences in a simple manner. This voting system is optimal in a well-defined sense: in a setting with two voters, two issues and uniform independent priors, Qualitative Voting Pareto dominates Majority Rule and, moreover, achieves the only ex-ante (incentive compatible) optimal allocation. The result also holds true with three voters as long as the valuations towards the issues differ sufficiently. Experimental evidence is provided supporting equilibrium predictions and showing that Qualitative Voting is better able to replicate the efficient outcome than Majority Rule. More generally in a setting with an arbitrary number of voters and issues, we show: (1) that a mechanism is implementable only if it does not undertake interpersonal comparisons of utility; (2) the impossibility of implementing strategy-proof mechanisms that are sensitive to the voters' intensities of preferences and satisfy the unanimity property. The second part of the Thesis studies the interaction between politicians' strategic behaveiour and voters' turnout decision: politicians diverge to motivate citizens to vote and they adapt their policies to the most sensitive voters -thus less sensitive voters abstain on the grounds of perceiving politicians being too similar. Moreover, citizens in central/moderate positions abstain. We find support for our predictions using NES data: (1) a perceived low difference between the Democratic and Republican parties tends to decrease a citizen's probability to vote and (2) moderate citizens vote less.