Stylized Facts of Return Series, Robust Estimates, and Three Popular Models of Volatility

It is well-established that the …nancial time series display some stylized fatcs such as volatility clustering, high kurtosis, low starting and slow-decaying autocorrelation function and the Talyor e¤ect as well. In order to evaluate volatility models’capacity in capturing such facts, we apply both standard and robust measures of kurtosis and autocorrelation of squares to GARCH, EGARCH and ARSV models. Such robust measures provide a fresh viewing angle in model evaluations which may be useful because the …nancial time series are often contaminated with some outliers. Keywords: GARCH, EGARCH, ARSV, extreme observations, autocorrelation function, kurtosis, robust measure, con…dence region. JEL codes: C22, C50. Acknowledgement: We thank Changli He for useful suggestions. Responsibility for any errors and shortcomings in this work remains ours. Zhenfang Zhao: [email protected], Timo Teräsvirta: [email protected]