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

Some recent analyses challenge previous reports which show that economically important trade significantly reduces the probability of militarized disputes between countries. Beck et al. (1998) address the effect of temporal dependence in the time-series data on empirical support for the liberal peace, 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 challengers. When analyzing all dyads, we find no relationship between interdependence and peace, but the pacific benefits of trade become evident among the politically relevant dyads (those including a major power, or two contiguous states), among whom the great majority of disputes occur. Subsequently, we introduce, in stages, an alternative method of controlling for temporal dependence, our theoretically preferred measures of interdependence and proximity, 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 benefits. This support is largely robust to the methods of controlling for temporal dependence and to whether an attempt is made to explain involvement in disputes or merely their onset. We find no evidence that asymmetric trade increases conflict.