Predicting Exporters with Machine Learning

Abstract In this contribution, we exploit machine learning techniques to evaluate whether and how close firms are to become successful exporters. First, we train various algorithms using financial information on both exporters and non-exporters in France in 2010–2018. Thus, we show that it is possible to predict the distance non-exporters are from export status. In particular, we find that a Bayesian Additive Regression Tree with Missingness In Attributes (BART-MIA) performs better than other techniques with an accuracy of up to 0.90. Predictions are robust to changes in definitions of exporters and in the presence of discontinuous exporting activity. Eventually, we discuss how our exporting scores can be helpful for trade promotion, trade credit, and assessing aggregate trade potential. For example, back-of-the-envelope estimates show that a representative firm with just below-average exporting scores needs up to 44% more cash resources and up to 2.5 times more capital to get to foreign markets.