Predicting Exporters with Machine Learning (Online event)

04  March 2021    3:00 pm

Armando Rungi, IMT School for Advanced Studies Lucca

In cooperation with:

Research Centre International Economics (FIW) 


This is an online event via Zoom. Please register for the dial-in link, reminders and e-mail updates. The dial-in link will be sent to you shortly before the event.



The presentation is based on a paper with the same title co-authored with Francesca Micocci.

In this contribution, we exploit machine learning techniques to predict a firm’s ability to export. In a pure prediction framework, we train a Bayesian Additive Regression Tree (BART) on the financial accounts of 57,021 manufacturing firms in France in the period 2010-2018. We obtain a relatively high accuracy of prediction, with a precision-recall at 0.91. Then, we show how predictions could be of help in assessing a firm-level exporting score, i.e., the distance of a firm from export status. We argue our exporting score has the potential to inform target-specific and evidence-based policies of internationalization. 

The presentation, when available, will be posted online after the event.

Armando Rungi is an Assistant Professor of Industrial Organization and International Trade at IMT Lucca, working for the research unit AXES (Analysis of Complex Economic Systems). He obtained his PhD at the Bocconi University, where he also graduated in international economics.