Estimating and Forecasting GARCH Volatility in the Presence of Outiers

The main goal when fitting GARCH models to conditionally heteroscedastic time series is to estimate the underlying volatilities. It is well known that outliers affect the estimation of the GARCH parameters. However, little is known about their effects when estimating volatilities. In this paper, we show that when estimating the volatility by using Maximum Likelihood estimates of the parameters, the biases incurred can be very large even if estimated parameters have small biases. Consequently, we propose to use robust procedures. In particular, a simple robust estimator of the parameters is proposed and shown that its properties are comparable with other more complicated ones available in the literature. The properties of the estimated and predicted volatilities obtained by using robust filters based on robust parameter estimates are analyzed. All the results are illustrated using daily S&P500 and IBEX35 returns.