Sequential Voting Models

项目摘要该提案包括两个部分。第一部分也是主要部分是与米歇尔·皮乔内的合作。在最近政治经济学研究的基础--投票模型中，投票可以扮演两个角色：它可以聚合私人信息和/或偏好。虽然在大多数投票模型中，选民同时对每个问题进行投票，但我们分析了顺序投票博弈。这使得有可能解决许多关于投票的两个角色的有趣问题。例如，关于汇总信息，哪种机制更好：同时投票还是顺序投票？当选民可以选择何时投票时，他们会选择一种有效地汇总信息的方法吗？我们还研究了在提前或晚些时候投票时两个相反考虑因素的影响：早期选民更有可能决定一组“相关的”替代方案，而后来的选民将知道哪些替代方案是相关的，因此不太可能“浪费”他们的选票。那么，当选民可以选择何时投票时，会出现什么样的顺序结构？那些预计会有相似偏好的选民会选择在时间上距离较近还是相距较远的投票？该项目解决了因考虑顺序投票环境而产生的这些问题和其他问题。这揭示了各州在何时举行初选方面的“博弈”，或者在一些地区投票完成之前宣布投票结果的含义。第二部分包含两个关于博弈论基础的项目(但在其他方面是无关的)。第一个，与巴特·利普曼和阿尔多·鲁斯蒂奇尼联合，为克雷普斯提出的不可预见的偶然事件模型*推导出独特的主观概率，并在早期的工作中被我们三人进一步推广。第二，与德鲁·福登堡和大卫·莱文合作，为不完全信息的游戏开发了基于学习的解决方案概念，如自我确认均衡的概念，以及纳入理性的概念。*D.M.Kreps(1979)：‘弹性偏好的表示定理’，‘计量经济学’，47 565-576。

Project Abstract The proposal contains two parts. The first and main part is joint work with Michele Piccione. In voting models, which underlie much recent research in political economy, the voting can have two roles: it can aggregate private information and / or preferences. Whereas in most voting models the electorate votes simultaneously on each issue, we analyze sequential voting games. This makes it possible to address many interesting questions on both roles of voting. For example, concerning aggregating information, which mechanism is better: simultaneous or sequential voting? When the electorate can choose when to vote, will they choose a method that is efficient for aggregating information? We also study the implications of two opposing considerations in voting earlier or later: early voters are more likely to determine the set of `relevant` alternatives and later voters will know which alternatives are relevant and hence are less likely to `waste` their vote. What sequential structures will then occur when voters can chose when to vote? Will voters who expect to have similar preferences choose to vote close together or far apart in time? The project addresses these and other questions that arise from considering sequential voting environments. This sheds light on the implications of states `gaming` on when to have their primaries, or of announcing the outcomes of votes in some districts before voting is completed in others. The second part contains two projects on the foundations of game theory (but that are otherwise unrelated). The first, joint with Bart Lipman and Aldo Rustichini, derives unique subjective probabilities for the model of unforeseen contingencies proposed by Kreps* and further extended by the three of us in earlier work. The second, joint with Drew Fudenberg and David Levine, develops, for games of incomplete information, solution concepts that are based on learning, such as the notion of self-confirming equilibria, and that incorporate rationality as well. *D. M. Kreps (1979): `A Representation Theorem for 'Preference for Flexibility',` Econometrica 47 565 - 576.