How much does international trade affect business cycle synchronization ?

In a recent article, Jeffrey Frankel and Andrew Rose (1998) examine the hypothesis that greater trade flows between two countries cause greater synchronicity between their business cycles. The increase in business cycle synchronicity may be seen as rationalizing a common monetary policy and, so, a shared currency. Arguing that product specialization would lower the synchronicity of business cycles, Frankel and Rose posit that a regression of output correlation on overall trade will indicate whether (positive) common demand shocks and productivity spillovers dominate or (negative) specialization effects do. The authors apply instrumental variables to confirm a causal relationship. In this paper, we refine the estimation in two ways. First, we test for instrument validity and find that the confirming null hypothesis is rejected in most cases. We find evidence to suggest that the instrumental variables method applied is inappropriate and results in inflated coefficients. We develop and apply an alternative OLS-based estimation procedure. Second, we add structure-of-trade variables to the model to separate the effects of intra- and inter-industry trade flows. Although our results suggest that the Frankel and Rose model overestimates the effect of trade on business cycle correlation, the overall results of our model are consistent with theirs. With our own model estimation, we find that specialization generally does not significantly asynchronize business cycles between two countries.