Does exporting cause productivity growth? Evidence from Chilean firms

Does exporting increase firm productivity, or are increased export sales caused by the firm's ability to increase its productivity? This paper provides new empirical evidence on the causal relation between trade and productivity, adopting a structural vector autoregressive analysis combined with identification algorithms from the machine learning literature. We focus on a well-studied country (Chile) and on already-exporting firms (intensive margin). We identify the contemporaneous and lagged causal structure between firm productivity and export growth using two different machine learning algorithms based on independent components analysis (ICA), which exploit the non-Gaussian distribution of the data to recover the independent structural shocks that drive the observed variables. Our findings show that, for Chilean firms, productivity growth causes export growth in the same year, but that the reverse does not apply. Export growth also has no causal effect on TFP growth in subsequent years. To increase sales in the foreign market, firms should first also increase productivity. The increased presence in the foreign market does not contribute to such productivity growth.