Adaptive learning with a unit root: An application to the current account

Abstract

This paper develops a simple two-country, two-good model of international trade and borrowing that suppresses all previous sources of current account dynamics. Under rational expectations, international debt follows a random walk. Under adaptive learning, however, the model's unit root is eliminated and international debt is either a stationary or an explosive process, depending on agents’ specific learning algorithm. Some stationary learning algorithms result in debt following an AR(1) process with an autoregressive coefficient less than 0.8. Because unit roots are a common and problematic feature of many international business cycle models, our results offer a new approach for generating stationarity.