The Political Economy of Resource Allocation Decisions in Campaigns

研究美国政治的学者最关心的一个问题是，在竞选活动中花费的资源如何影响选举结果。这个问题引起人们的兴趣至少有两个原因。首先，从逻辑上讲，它先于民主理论中的几个重要问题：拥有大量资金的政党和候选人能否通过金钱进入有权势的政治职位？竞选人对金钱的需求是否允许那些控制大笔资金并为竞选活动捐款的个人和利益集团对政策结果施加“过大”的影响？其次，有必要了解金钱和其他资源在竞选活动中的作用，以便预测各种竞选财务法规（如竞选支出限制、捐款限制和公共补贴选举）可能产生的影响。这个项目的重点是参与若干同时但不同的选举竞争的政党所面临的资源分配问题。具体来说，一个政党应该如何在不同的竞选中分配其竞选资源，以最大限度地实现其目标？在加拿大、英国、印度、新西兰、美国等实行单一制的议员选举和实行“单一制”的选举人团制度的美国总统选举中，难免会出现这样的问题。首席调查员将竞选支出视为一种广告形式，这增加了候选人赢得他或她正在寻求的职位的可能性。由于竞选公职的候选人中只有一人能够获胜，候选人对手的支出也会影响他或她获胜的可能性。研究人员采用博弈论框架来捕捉这种相互作用。该项目包括理论和实证两个部分。在理论工作中，研究者将分析几种不同的战役资源分配模型。这些模型在以下方面有所不同：1)竞争政党的数量，2)各政党的目标函数，3)“选举生产函数”的某些细节（即，竞选资源如何转化为选票），例如现任者和挑战者之间的差异，以及4)世界是否以一种相当静态的方式被对待，只有一次选举（或者，政党非常“短视”）。只关注当前选举)，或者以一种更动态的方式关注一系列选举，其中一次选举的结果可能影响未来选举的结果。在实证工作中，研究人员对1979年、1980年和1984年加拿大大选的数据进行了各种模型的测试。然后，他利用这些模型中最好的模型，使用适当的公式，估计竞选支出、在任等因素对选举结果的影响。除了增加了关于选举和竞选资金具体问题的知识储备之外，这项研究的新颖之处在于，它使用博弈论来非常密切地指导实证测试的规范。事实上，很少有人尝试从经验上检验博弈论模型，也很少有人认真对待识别和估计这些模型所涉及的计量经济学问题。

A fundamental concern to scholars who study American politics centers on how resources spent in election campaigns influence electoral outcomes. This question is of interest for at least two reasons. First, it logically precedes several important issues in democratic theory: Can parties and candidates with large amounts of money buy their way into powerful political offices? Does the need for money by those who seek office allow individuals and interest groups who control large sums of money, and who contribute to election campaigns, to wield "too much" influence on policy outcomes? Second, it is necessary to understand the role of money and other resources in election campaigns in order to predict the likely effects of various campaign finance regulations, such as campaign spending limits, contribution limits, and publicly subsidized elections. This project focuses on the resource allocation problems faced by political parties that are involved in a number of simultaneous but distinct electoral contests. Specifically, how should a party allocate its campaign resources across the various contests in order to maximize its objectives? Such problems arise naturally in legislative elections with single-member districts and first-past-the-post electoral rules, such as Canada, Great Britain, India, New Zealand and the U.S., and in U.S. presidential elections under the electoral college with the "unit- rule." The Principal Investigator treats campaign spending as a form of advertising, which increases the probability that a candidate wins the office he or she is seeking. Since only one of the candidates competing for an office can win, spending by a candidates' opponents also affects his or her probability of winning. The researcher employs a game-theoretic framework to capture this interaction. The project involves theoretical and empirical components. In the theoretical work, the researcher will analyze several different models of campaign resource allocation. The models vary in 1) the number of competing parties, 2) the objective functions of the various parties, 3) certain details of the "electoral production function" (i.e., how campaign resources are translated into votes), such as the differences between incumbents and challengers, and 4) whether the world is treated in a rather static fashion, with a single election (or, alternatively, parties are extremely "myopic," concerned only with the current election), or in a more dynamic fashion, with a sequence of elections in which the outcome of one election may affect the outcomes of future elections. In the empirical work, the researcher tests the various models on data for the Canadian general elections of 1979, 1980, and 1984. Then, using the best of these models, he estimates the effects of campaign spending, incumbency, and so on, on the election outcomes, using the appropriate formulas. Besides adding to the stock of knowledge about specific issues in elections and campaign finance, this research is novel in that it uses game-theory to guide very closely the specification of empirical tests. In fact, there have been very few attempts to test game-theoretic models empirically, and to treat carefully the econometric problems involved in identifying and estimating these models.