Game theory has proven to be a useful tool for understanding the logic of bargaining when actors have conflicting interests. Bargaining models have been applied by political scientists to analyze everything from the effects of open and closed rules on the distributive politics of legislative appropriation to the study of war initiation and termination. In contrast to the theoretical advances, the quantitative analysis of bargaining has progressed at a much slower rate. Typically, scholars analyzing bargaining data employ "off-the-the-shelf" OLS, FGLS, or Tobit models. However, Signorino's previous research on the empirical analysis of strategic behavior showed that for inferences to be correct, the statistical estimator must be structurally consistent with the hypothesized theoretical process. In fact, the Principal Investigator demonstrates in the current project that the above statistical models are structurally inconsistent with respect to bargaining behavior. Signorino's previous research dealt with strategic discrete choice models. Yet, in many important contexts - e.g., bargaining involving territory, policy, or money - the choice set available to actors is better represented as continuous, rather than discrete. Therefore, new statistical models and estimation techniques must be developed in order to analyze bargaining data. The investigator has already made substantial progress in this area. However, much remains to be done. The goals of this research project are the following: o Create statistical techniques for analyzing ultimatum bargaining games, such as the divide the-dollar game and the Romer-Rosenthal agenda setting models; and, create statistical techniques for analyzing axiomatic bargaining models. o Use these models to analyze existing data (which they already have) on 1. Bargaining in territorial disputes and in international political economy. 2. Bargaining in Congressional politics. 3. Bargaining in society, including (a) Among Indians of the Amazon. (b) Among U.S. citizens and Russian citizens. o Incorporate these new estimators into STRAT, a program for estimating statistical strategic models, and produce modules for the techniques in STATA and in R.
Broader Impact. This research has the potential to influence multiple subfields of Political Science, as well as Economics and Anthropology. The research effort will also assist in graduate training, since graduate students will provide research assistance. The research will be presented in class lectures, seminars, and at conferences. Finally, the results of this research will be made publicly available not only through research publications, but also as part of statistical software.
