孔多塞悖论再探

The Condorcet paradox revisited

Social Choice and Welfare · 2016
被引 16
人大 A-ABS 3

中文导读

在一个三人三选项的多数投票讨价还价模型中,研究了孔多塞悖论,发现当风险系数的几何均值不超过1时存在平稳子博弈完美均衡,且均衡中不会出现孔多塞循环。

Abstract

We analyze the Condorcet paradox within a strategic bargaining model with majority voting, exogenous recognition probabilities, and no discounting for the case with three players and three alternatives. Stationary subgame perfect equilibria (SSPE) exist whenever the geometric mean of the players’ risk coefficients, ratios of utility differences between alternatives, is at most one. SSPEs ensure agreement within finite expected time. For generic parameter values, SSPEs are unique and exclude Condorcet cycles. In an SSPE, at least two players propose their best alternative and at most one player proposes his middle alternative with positive probability. Players never reject best alternatives, may reject middle alternatives with positive probability, and reject worst alternatives. Recognition probabilities represent bargaining power and drive expected delay. Irrespective of utilities, no delay occurs for suitable distributions of bargaining power, whereas expected delay goes to infinity in the limit where one player holds all bargaining power. An increase in the recognition probability of a player may weaken his bargaining position. A player weakly improves his bargaining position when his risk coefficient decreases.

Condorcet悖论策略性议价模型平稳子博弈完美均衡议价权力