The importance of asymmetric autoregressive stochastic volatility models in financial markets

Abstract

The leverage effect is one of the most relevant stylized facts to modelling time-varying financial volatility because the effect that negative returns has on volatility is different to that produced by positive returns. For this reason, this paper focuses on two goals: (i) to analyse the main stylized facts of volatility using 20 indexes from the most important stock markets in Asia, Europe, North America and South America in the sample period, namely between January 3, 2000 and February 4, 2014; and, (ii) to estimate conditional heteroskedasticity using the classic econometric models typically used for this purpose (AGARCH, GJR-GARCH, EGARCH, APARCH) along with the symmetric (ARSV) and asymmetric (TA-ARSV) stochastic volatility models, and then to compare their results to determine which model best explains the main stylized facts of financial returns. The results show, on the one hand, that financial returns have the same characteristic features, regardless of the particular stock market and, on the other hand, that the TA-ARSV is the best model to explain the stylized facts.