What determines exports of U.S. states : distance or location factors? /

Abstract

This study is an empirical investigation into the export performance of the United States using state level data. Only a few studies have tried to disaggregate national level data. While spatial issues play an increasing role in international trade models, most applications make use of the traditional gravity model, which is based on physical distances. However, the significant reduction in transportation and communication costs experienced over the past few decades makes it questionable that physical distance matters any more.

The study is arguing on two fronts. First, it examines to what extent it may be useful to consider proximity measures other than traditional physical distance variables. To examine this possibility, this study employs an augmented gravity equation with a number of alternative physical and nonphysical/psychic distance measures. Second, the study explicitly allows for the possibility that distance no longer plays any role in explaining trade patterns, regardless of how distance is defined. For this purpose, the study includes a large number of alternative variables that can potentially explain trade patterns. Many of these measures are related at least indirectly to the location decisions of firms that are engaged in the exporting business.

The results of the study indicate that physical distance continues to play a role in determining state manufacturing exports. However, nonphysical/psychic distance, trade agreements, state location characteristics, and conditions in export markets also affect a state's manufacturing exports. These latter effects appear to vary for high-tech and low-tech products.

Tests for the presence of spatial autocorrelation in state manufacturing export equations indicate that physical and cultural proximity does not matter for spatial autocorrelation. Rather, it is the human development proximity of states that explains their similarity in export performance. The results of the spatial autocorrelation analysis are robust only for total and high-tech exports.