Staff Report, Federal Reserve Bank of New York 127
Borders have a sizable negative impact on trade flows. Given the vast number of individual goods potentially traded, this border effect could have two possible explanations: (1) less international than domestic trade in the goods that are actually traded between countries (flow), or (2) differences between the sets of goods traded internationally and domestically - that is, fewer goods are available as exports than are sold in the home market (availability). Most of the previous literature on border effects has ignored the possible role of this second factor, instead reporting a single border effect that contains the embedded assumption that identical sets of goods are available in the domestic and export markets. In contrast to this assumption, evidence on the activities of firms shows that only a fraction of domestic products areactually exported. This paper provides theoretical and empirical work that incorporates the distinction between the flow and availability explanations of border effects. A model that includes heterogeneous fixed costs of trade illustrates how either of these two factors could underlie a given border effect. The empirical work incorporates the fact that not all firms export by examining only the fraction of total domestic production attributable to those firms that actually do sell abroad. The results suggest that a portion of the border effect is indeed due to differences between the sets of goods available domestically and internationally. I find that, on average across industries, about one-half of the border effect is due to the flow explanation, while the remaining half may be attributed to availability. Given that the policy and welfare implications of border effects depend on the relative importance of these two explanations, future work should take care to specify clearly which aspect of the border effect is being measured.