A Non Parametric Study of the Volatility of the Economy as a Country Risk Predictor

This paper intends to explain Venezuela’s country spread behavior through the use of Neural Networks, incorporating as input data the IGAEM’s (monthly economic activity general index, an index of economic indicators constructed by the BCV) VaR (value at risk), a measure of the shocks affecting country risk (country spreads) of emerging markets (the EMBI +Global) and EEUU’s short term interest rates (6 month T-bills). The non parametric method used allows approaching the problem of finding the non linear relationship between these inputs and the country risk. The relative performance of the networks that were found was subsequently evaluated using the method of excess predictability (Anatolyev & Gerko, 2005). The network’s performance in terms of predicting Venezuela’s country risk behavior was satisfactory according to the evaluation method employed.