Theoretical and Experimental Research on Voting and Bargaining Games

California Institute Of Technology, Pasadena CA

Investigators

Abstract

Intellectual merits: This research project studies strategic voting and committee bargaining mechanisms under conditions of incomplete information from both theoretical and experimental perspectives. Many questions remain unresolved about the general theoretical problem of designing optimal voting mechanisms to efficiently aggregate preferences and information, without explicit side payments, and also about actual behavior under such mechanisms. Theoretically, there are questions of the relative performance of different voting mechanisms, where performance can be evaluated relative to several different objectives, including economic efficiency, informational efficiency, and equity. The research approaches voting behavior from a mechanism design perspective, and compares the performance of different voting mechanisms in different kinds of environments. The research encompasses five projects in political economy. Three of these projects will explore promising extensions of the research initiated under the current grant. These are: (1) The Dynamic Political Economy of Public Infrastructure; (2) Strategic Voting and Information Acquisition in Committees; and (3) Vote Trading in Committees. The first follows naturally from the dynamic political bargaining model studied under the current grant. Bargaining agreements evolve over time, with past bargaining outcomes setting the table for current and future rounds of bargaining. We extend that pure private good model to incorporate a dimension of durable public goods, or "public infrastructure." We propose to compare the effects of different voting rules and government organization on the bargaining process, especially with respect to the effects on efficient investment in public infrastructure. The second project evolved from experimental studies conducted by the PI and collaborators on the "swing voter's curse," a phenomenon whereby less informed voters have incentives to either abstain or even vote strategically against the outcome they would choose if the decision they would choose on their own. The extension endogenizes the acquisition of information by voters. The third part of the proposed research involves studying new voting mechanisms that allow voters to express preference intensities over a multiple decisions across different issue dimensions. We apply competitive equilibrium analysis to this kind of environment and propose a new equilibrium concept with vote trading, called "ex ante exchange equilibrium." We then study the predictive power of this model by setting up controlled laboratory "vote markets." The emphasis is on the comparative statics predictions and the efficiency properties. The two additional projects are: (4) The Emergence of Efficiency in Dynamic Coordination Problems; and (5) Dynamic Cursed Equilibrium with an Application to Sophisticated Voting in Dynamic Agendas with Incomplete Information. The analysis of the data will explore these extended models of limited rationality to help explain where and why the theory seems to be adequate and where and why it misses. Project (4) characterizes efficient dynamic equilibria subject to constraints imposed by the symmetry structure of the coordination game. We will design and conduct experiments that look at the effects of the number of players in the group and also the effects of the symmetry structure of the game. Project (5) explores a new specification of limited strategic sophistication that extends the cursed equilibrium to extensive form games. It extends the cursed equilibrium logic from traditional "mixed strategies" (strategic form) to behavioral strategies. The original formulation was done in strategic form and so the idea of cursed equilibrium did not have bite in many sequential contexts. We apply this to a variety of sequential games, in particular sequential voting with incomplete information. Broader impacts: In addition to basic research, the proposed research has an education component, by training graduate students in experimental economics and economic theory. The ultimate goal of the research is to better understand how procedures in commitees affect decision making and with and eye to understanding how these procedures may be modified to improve decision making and overcome obstacles such as conflicting preferences and beliefs, and asymmetric information. This better understanding in the long run can improve performance of organizations and policy-making institutions. The performance of voting procedures are evaluated according to traditional economic welfare criteria, informational efficiency, and equity. There is extensive software development proposed under the grant. This software may be used freely by other researchers in experimental economics and will be publicly available as open source code.