Assessing the liberal peace with alternative specifications: Trade still reduces conflict

Some recent analyses challenge previous reports which show that economicall y important trade significantly reduces the probability of militarized disp utes between countries. Beck et al. (1998) address the effect of temporal d ependence in the time-series data on empirical support for the liberal peac e, while Barbieri (1998) makes a number of important changes in theoretical specification and measurement. Using data for nearly the entire post-World War II era (1950-92), we first replicate the specifications of the challen gers. When analyzing all dyads, we find no relationship between interdepend ence and peace, but the pacific benefits of trade become evident among the politically relevant dyads (those including a major power, or two contiguou s states), among whom the great majority of disputes occur. Subsequently, w e introduce, in stages, an alternative method of controlling for temporal d ependence, our theoretically preferred measures of interdependence and prox imity, and new dyadic estimates for unreported trade. With these sequential modifications we find increasingly strong support for the liberals' belief that economic interdependence and democracy have important pacific benefit s. This support is largely robust to the methods of controlling for tempora l dependence and to whether an attempt is made to explain involvement in di sputes or merely their onset. We find no evidence that asymmetric trade inc reases conflict.