SMU SOE Seminar (Nov 10, 2017, 2-3:30pm): Learning to Import from Neighbors

This paper studies how learning from neighboring firms affects the behaviors of new importers. We first develop a learning model in which firms update their beliefs about the import price in foreign markets based on several factors, including the number of neighboring firms that import from that market, the level and heterogeneity of their import prices. The updating proceeds according to the Bayesian rule. The model predicts that a positive signal about import prices revealed by neighboring importers encourages entry and increases initial imports from the same country. The signal plays a stronger role when it is revealed by more neighbors. Using a transaction-level dataset of Chinese importers over the 2000-2006 period, we find supporting evidence for the model's predictions. Furthermore, importer learning displays heterogeneous effects on different firms and exhibits a spatial decay structure. Our results are robust to controlling for various fixed effects, an alternative entry definition, and subsamples consisting of ordinary trade firms and direct importers, respectively.

Keywords: Learning to Import, Bayesian Update, Spillover, Uncertainty, Signals

JEL Classification: F1, F2, D8