Neural Networks applied for Macroeconomic Analysis

In this work, we review the dependency between exports of traditional products from Ecuador and certain macroeconomic indicators using four Neural Networks models and a Walk Forward Validation approach. We study the effect of the Country Risk (EMBI+) and Foreign Direct Investment (FDI) indicators on the exportation of Banana, Shrimp, Cocoa and Coffee. The information source covers Time Series of macroeconomic indicators and historical export data considering a period of 22 years. The outcome show us correlation coefficients around 0.94 when we analyze cocoa with ARMAX structures. Consequently, a dependence is determined between the Cocoa exportations and the macroeconomic indicators studied.