Testing Nonlinear Dynamics in Terms of Trade with Aggregated Data: Implications for Economic Growth Models

Abstract

For many decades, world trade has grown on average nearly twice as fast as total world output. International trade flows have exploded since the 1980s, however high and middle-income countries continue to make up the main players in international trade. Favorable movements in global export prices lead to similar movements in terms of trade in developed and developing countries, but it still did not stop the latent deterioration of terms of trade of undeveloped countries. Though we can detect general convergence in the dynamics and lower volatility of the terms of trade, this still does not explain us the evolution of the terms of trade. We still have to asses if the movements in this variable were random. This paper examines long time aggregated series data (1960 - 2015) of the terms of trade for the variety of (grouped) countries to find out if there are any signs of nonlinearity in these series. Finding evidence of nonlinearity suggests that economic models that include terms of trade can improve by switching from linear to nonlinear modeling strategy. For this purpose, we use BDS nonparametric test as it is one of the most popular tests for nonlinearity. We can use it as a portmanteau test or miss-specification test when applied to the residuals from a fitting model. Results reject the nonlinearity presumption going in favor of linear behavior of the terms of trade.