Indirect Exports and Wholesalers: Evidence from Interfirm Transaction Network Data

Abstract

A substantial fraction of international trade is facilitated by wholesalers, who enable manufacturers to indirectly export their products to foreign markets. Using large-scale Japanese interfirm transaction network data, this paper add empirical findings on the difference in the features between indirect and direct exporters. We base our analyses on a simple Melitz-type trade model with indirect export alternative as presented in Ahnet al. (2011), which predicts sorting of firms to direct, indirect, and non-exporters along the size dimension. This pattern is also confirmed in our data, and the distributions of sales, in-degree (the number of suppliers), out-degree (the number of customers), and labor productivity are ordered for direct, indirect, and non-exporters in terms of first order stochastic dominance, which is consistent with findings in Ahn et al. (2011) and Bernard et al. (2014). We then perform multinomial logit analyses in which a manufacturer chooses one export mode out of direct export, indirect export through wholesalers, and no export. Consistent with the model, the estimated intercept is lower and slope of sales is steeper for direct exporting. We also find that in-degree raises the probability of direct exporting implying a cost sharing mechanism of firms with more suppliers. Out-degree raises the probability of exporting in general (both indirect and direct). This implies a higher product appeal and broader demand base for firms who have more customers in the domestic market. Industry heterogeneity in the propensity of indirect and direct exporting is also analyzed.