Condorcet efficiency with adaptive parties in a spatial model

Abstract

Several recent articles analyze the decision rules that politicians use to adjust their policy strategies in response to past election results and to rival candidates/parties’ strategies. Findings presented in these articles shed light on the expediency of using alternative voting systems from the viewpoint of their potential to produce an election winner preferable in a particular sense or according to a particular criterion. We consider one such criterion, the Condorcet criterion, which evaluates the ability of the candidates to defeat all others in a series of pair-wise votes. We assume that voters choose the candidate (party) based on the candidates’ announced policies, and we report results of a computer simulation of adaptive politicians competing in a multi-dimensional issue space. Our simulation findings suggest that voting systems’ Condorcet efficiencies vary significantly with the decision rules that the competing politicians employ to adjust their policy strategies. Our findings also suggest that, consistent with conclusions reported in earlier research by Merrill, voting systems’ Condorcet efficiencies depend on the degree of spatial dispersion between the parties or candidates contesting the election.