Estimating the Productivity Selection and Spillover E¤ects of Imports

Economists emphasize two channels through which import liberalization a¤ects productivity, one operating between and the other within \u85rms. According to the for-mer, import competition triggers market share reallocations between domestic \u85rms with di¤erent technological capabilities (selection). At the same time, imports can also improve \u85rmstechnologies through learning externalities (spillovers). We present evidence for a sample of industrialized countries over the period 1973 to 2002. First, in the long run, import liberalization lowers productivity in domestic industries through selection. This \u85nding con\u85rms the prediction of models with \u85rm heterogeneity, in-cluding Melitz and Ottaviano (2008), in which unilateral liberalization lowers the pro\u85ts of domestic relative to foreign exporters. Second, if imports involve advanced foreign technologies, liberalization also generates technological learning that can on net raise domestic productivity. Third, for short time horizons of up to three years, a surge in imports typically raises domestic productivity. Because the number of \u85rms at home and abroad does not change much in the short-run, new competition from foreign rms has a pro-competitive e¤ect. We also \u85nd that high entry barriers, especially regulation, slow down the process of market share reallocation between \u85rms. Over-all, the results support models in which trade triggers both substantial selection and technological learning. We are grateful to Jan DeLoecker, Ann Harrison, Margaret McMillan, Gianmarco Ottaviano, Carol Shiue, Catherine Wolfram, and Stephen Yeaple, as well as seminar audiences at the European Central Bank and UC Berkeley (Haas) for comments. Thanks also to Ray Fisman, Mark Roberts, and Virginia Sarria-Allende who helped with the data. Views expressed in this paper are those of the authors and do not necessarily reect those of Industry Canada