A little bit louder now: Signaling, interests, and the liberal peace

This dissertation examines competing explanations for the liberal peace (the idea that trade between states creates pacific relations between those states) and uses statistical analysis to argue that traditional "interest-based" explanations (Russett and Oneal, 2001) do not explain aspects of interstate conflicts as well as more recent "signaling-based" explanations (Fearon, 1994; Gartzke et. al., 2001). Specifically, the dissertation examines the probabilities for interstate conflict between states when a measure that examines whether one state is economically dependent on the other is included. The results suggest that more dependent states are more likely to both initiate interstate conflicts and use higher levels of hostility in ongoing conflicts than non-dependent states. These results, I argue, are consistent with signaling-based explanations for the liberal peace, and wholly inconsistent with interest-based explanations. Additionally, by examining the actions of dependent states, the dissertation looks at one set of boundary conditions (Mansfield and Pollins, 2001) in establishing the limits for the liberal peace. Next, the dissertation argues that the two dominant data sets used to study the liberal peace (Gleditsch, 2002; Barbieri, 2002) are both flawed: the Gleditsch data by biased GDP measures and the Barbieri by missing values. Finally, the existence of the liberal peace itself is called into question: results of empirical tests indicate that higher amounts of trade with specific trading partners lead states to greater probability of initiating interstate conflict and using higher levels of hostility in ensuing interstate conflict with that trading partner.