Reevaluating the Evidence on Trade and Productivity

Researchers have estimated productivity changes surrounding trade liberalization for different countries, using different techniques, and have generally reached the conclusion that gains are positive. In this paper we study how different techniques influence the quantitative results by calculating productivity gains for Chile following the liberalization of the 1970s and 1980s using several methods used in the literature and incorporating new cost and market structures. To do this, we specify and estimate production functions for firms using methods suggested by Olley and Pakes (1996), Levinsohn and Petrin (2003), and Cooper, Haltiwanger, and Willis (2006). We modify these techniques to include decisions on exporting and exit and to allow for convex and nonconvex forms of labor and capital adjustment costs. Additionally, we allow for nontraded goods sectors – which are frequently ignored in this literature – to asses the differential impact of opening to trade on traded and nontraded good sectors. In agreement with other studies, we find aggregate productivity gains are driven mostly by changes in the composition of plants. In addition, we find that nonconvex adjustment costs are crucial to replicating the firm level data on exporting and factor demands and that there are strong interactions between the traded and nontraded good sectors. In our preferred specification, which allows for this richer structure, we find that plant level productivity grows more than that found in other studies.