Explaining and Predicting Double Tax Treaty Formation with Machine Learning Algorithms

The paper confronts country pairs that have tax treaties with country pairs that do not and investigates determinants for this distinction based on their gravity characteristics using machine learning techniques. It trains machine learning algorithms to distinguish such country pairs and selects random forest as the algorithm with the highest accuracy (94.3%) to use it for predictive purposes. The paper identifies 59 country pairs likely to have tax treaties based on their gravity characteristics. Countries/regions with the highest number of predicted new tax treaties are Germany (9), Saudi Arabia (8), Brazil (7), Myanmar (7), and Hong Kong (6). The paper investigates the machine learning prediction from the point of the current tax treaty status of the identified country pairs. Out of these country pairs, 31 are known to lead tax treaty negotiations, to have initialed a tax treaty, or to have already signed a tax treaty, 6 country pairs have signed or are negotiating an exchange of information agreement or a transport tax treaty, 3 country pairs used to have tax treaties, which were terminated. There is no public information available about ongoing negotiations for only 19 countries, less than a third of all countries where the machine learning algorithm would predict a treaty. This supports the validity of the machine learning techniques for prediction purposes. These results present important insights for policymakers when deciding over which treaty to pursue and which treaties may present a threat to a country’s international tax policy.