Collaborative Research on Spending and Incumbency in Congressional Elections: A Game-Theoretic Approach

本项目研究了美国国会选举中竞选支出、任期和选举结果之间的关系。该方法结合了正式的政治/经济方法、统计模型和1970-1990年期间的经验数据分析。这些理论模型旨在理解挑战者和在位者之间支出竞赛背后的战略因素。所采用的博弈论方法强调了在投票支出联系中观察两个因果方向的重要性。一方面，竞选资金/支出活动影响胜利的幅度，可能是谁赢得选举的决定性因素；在另一个方向上，候选人和捐助者对选举可能有多接近的预期会影响竞选财务/支出决策。这两个方向都很重要，支出决策和投票结果的联合性或同时性导致了测量和估计的困难统计问题，以及提出了一些具有挑战性的理论问题。大多数先前的研究主要集中在一个方向（上面提到的第一个方向），并试图拟合一个“投票方程”来衡量支出对选举结果的影响。当前的方法明确地将此与“支出方程”的估计相结合，其中预期亲密度的相互影响与支出对最终结果的影响同时进行估计。这使研究人员能够理清主要利益变量之间的一些复杂的相互作用：现任者的支出、挑战者的支出和现任者的投票份额。这项研究的初步结果对先前几项研究的中心发现提出了实质性的修订，并指出了在这种博弈论方法框架内有趣的新研究方向。这些新的方向，无论是理论的还是实证的，构成了本研究项目的议程。

This project investigates the relationships between campaign spending, incumbency and electoral outcomes in U.S. Congressional elections. The approach combines formal political/economic methodologies, statistical modelling, and analysis of empirical data during the period 1970-1990. The theoretical models are aimed at understanding the strategic factors underlying spending races between challengers and incumbents. The game-theoretic approach employed emphasizes the importance of looking at both causal directions in the vote-spending connection. In one direction, campaign finance/spending activities affect margins of victory and can be a deciding factor in who wins an election; in the other direction, candidate and contributor expectations of how close the election is likely to be affect campaign finance/spending decisions. Both directons are important, and the jointness or simultaneity of spending decisions and voting outcomes leads to difficult statistical problems of measurement and estimation, as well as raising a number of challenging theoretical issues. Most previous research focuses primarily on one direction (the first one mentioned above), and attempts to fit a "vote equation" to measure the effect of spending on electoral outcomes. The current approach explicitly combines this with an estimation of "spending equations," where the reciprocal effect of anticipated closeness is simultaneously estimated with the effect of spending on the final outcome. This enables the investigators to disentangle some complex interactions between the variables of primary interest: incumbent spending, challenger spending, and the vote share of the incumbent. Preliminary results of this investigation suggest substantial revisions to the central findings of several prior studies, and point to interesting new directions for investigation within the framework of this game-theoretic approach. Those new directions, both theoretical and empirical, form the agenda of this research project.