STOCHASTIC VOLATILITY MODELS AND STYLIZED FACTS

This paper highlights the ability of the discrete stochastic volatility models to predict some important properties of the data, i.e. leptokurtic distribution of the returns, slowly decaying autocorrelation function of squared returns and the Taylor effect. Although, there are many methods proposed for stochastic volatility model estimation, in this paper Markov Chain Monte Carlo techniques were considered. It was found that the existent specifications in the stochastic volatility literature are consistent with the empirical properties of the data. Thus, from this point of view the discrete stochastic volatility models are reliable tools for volatility estimation.