In their paper "Predicting Exporters with Machine Learning," Francesca Micocci and Armando Rungi explore the use of machine learning techniques to predict the likelihood of firms becoming successful exporters. The authors train and test various algorithms using financial information on both exporters and non-exporters in France from 2010-2018, ultimately demonstrating their ability to predict the distance of non-exporters from export status. They 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, and note that predictions are robust to changes in definitions of exporters and in the presence of discontinuous exporting activity. The authors also discuss how their exporting scores can be useful for trade promotion, trade credit, and assessing aggregate trade potential. For instance, they estimate 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 enter foreign markets. This highlights the potential for machine learning techniques to aid in predicting export success for firms by providing valuable insights for policymakers and businesses alike. Overall, this study provides evidence for the effectiveness of these methods in helping companies make informed decisions about entering international markets.
Francesca Micocci and Armando Rungi used computers to guess if a company will be good at selling things in other countries. They looked at information about companies in France from 2010-2018. One computer program called BART-MIA was the best at guessing, with up to 90% accuracy. The guesses were still good even if they changed how they defined what it means to be an exporter. Knowing how good a company is at exporting can help people who want to promote trade or lend money to them. Companies that are not very good at exporting need more money to start selling things in other countries. Using computers can help us figure out which companies will be successful at selling things in other countries, which is helpful for people who make decisions about trade and businesses.
Definitions
- Machine learning techniques: using computers to learn patterns and make predictions
- Exporters: companies that sell goods or services in other countries
- Bayesian Additive Regression Tree with Missingness In Attributes (BART-MIA): a specific machine learning algorithm used by the authors
- Accuracy: how often the computer's predictions are correct
- Discontinuous exporting activity: when a company starts and stops selling things in other countries
- Trade promotion: activities that encourage companies to sell more things in other countries
- Trade credit: lending money to companies so they can buy materials or products for export
- Aggregate trade potential: how much all the companies in a country could sell if they all tried
Predicting Exporters with Machine Learning: A Study by Francesca Micocci and Armando Rungi
In their paper, "Predicting Exporters with Machine Learning," Francesca Micocci and Armando Rungi explore the use of machine learning techniques to predict the likelihood of firms becoming successful exporters. The authors train and test various algorithms using financial information on both exporters and non-exporters in France from 2010-2018, ultimately demonstrating their ability to predict the distance of non-exporters from export status.
The Research
The authors tested several different machine learning algorithms, including a Bayesian Additive Regression Tree with Missingness In Attributes (BART-MIA), Support Vector Machines (SVMs) and Random Forests (RFs). They found that BART-MIA performed better than other techniques, achieving an accuracy of up to 0.90. Furthermore, they note that predictions were robust to changes in definitions of exporters and in the presence of discontinuous exporting activity.
Implications for Businesses & Policymakers
The research conducted by Micocci and Rungi has important implications for businesses looking to enter international markets as well as policymakers interested in promoting trade. For instance, they estimate 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 enter foreign markets – highlighting the potential for machine learning techniques to aid in predicting export success for firms by providing valuable insights into what is required for success abroad.
Additionally, these findings can be used by policymakers when assessing aggregate trade potential or when making decisions about trade promotion or credit policies. By providing useful information about which companies are likely to succeed at exporting goods or services abroad, governments can make informed decisions about how best allocate resources towards encouraging exports within their countries’ economies.
Conclusion
Overall, this study provides evidence for the effectiveness of these methods in helping companies make informed decisions about entering international markets while also aiding policymakers when making decisions related to trade promotion or credit policies. With its ability to accurately predict export success based on financial data alone, machine learning offers a powerful tool that can help businesses navigate global markets while also allowing governments assess aggregate trade potential within their own economies – potentially leading towards greater economic growth overall through increased exports worldwide
In this contribution, we exploit machine learning techniques to evaluate whether and how close firms are to becoming successful exporters. First, we train and test various algorithms using financial information on both exporters and non-exporters in France in 2010-2018. Thus, we show that we are able to predict the distance of non-exporters 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.
