Predicción de la volatilidad en los mercados del petróleo mexicano a través de modelos CGARCH asimétricos bajo dos supuestos distribucionales

Abstract

In this paper, three symmetric and asymmetric CGARCH models are estimated to evaluate and improve volatility forecasts in Mexican crude oil markets under different distributional assumptions (Normal and Laplace). Empirical evidence shows that CGARCH and
CGARCH-A2 models yield the most accurate one-five-and twenty-day out-of-sample volatility forecasts for Istmo and Maya crude oil returns in comparison to the traditional GARCH models, including the CGARCH-A1 based model. These results are supported using symmetric and asymmetric forecast error measures and the Hansen´s (2005) superior predictive ability test. The improvement in volatility forecasting has important economics and financial implications for participants in Mexican crude oil markets, in particular the Mexican government.