Abstrakt
The generalized autoregressive conditional heteroscedasticity (GARCH) process is a particular modelling scheme, which is capable of forecasting the current level of volatility of financial time series. Recently, recursive estimation methods suitable for this class of stochastic processes have been introduced in the literature.
They undoubtedly represent attractive alternatives to the standard non-recursive estimation procedures with many practical applications. It is truly advantageous to adopt numerically effective estimation techniques that can estimate and control such models in real time.
However, abnormal observations (outliers) may occur in data. They may be caused by many reasons, e.g. by additive errors, measurement failures or management actions.
Exceptional data points will influence the model estimation considerably if no specific action is taken. The aim of this contribution is to propose and examine a robust recursive estimation algorithm suitable for GARCH models.
It seems to be useful for various financial time series, in particular for (high-frequency) financial returns contaminated by additive outliers. The introduced algorithm can be effective in the risk control and regulation when the prediction of volatility is the main concern since it distinguishes and corrects outlaid bursts of volatility.
Real data examples are presented.