Modeling Time Varying Volatility and Skewness

Abstract

Financial returns have been traditionally assumed to be normally distributed. However returns have been shown to possess thick tails as well as being leptokurtic and asymmetric. Accurate modeling of these features is of the upmost importance for financial investors. This study makes use of the skewed generalized error distribution for capturing asymmetries and thicker tails in the returns distribution. In addition, both volatility and skewness are modeled as time varying. Using daily returns data, we find evidence of time varying skewness in both S&P 500 and oil spot price. Also we find that variance is greater in days after a weekend/holiday for both sets of data, but there is no such holiday effect in skewness.