Economic determinants of regional trade agreements revisited using machine learning

While traditional empirical models using determinants like size and trade costs can predict RTA formation reasonably well, we demonstrate that allowing for machine-detected nonlinear patterns helps to improve the predictive power of RTA formation substantially. We find that the fitted tree-based methods and neural networks deliver sharper and more accurate predictions than the probit model. For the majority of models, the allowance of fixed effects increases the predictive performance considerably. We apply our models to predict the likelihood of RTA formation of the EU and the USA with their trading partners, respectively.