Parameter changes in GARCH model

Abstract

A new method for detecting the parameter changes in generalized autoregressive heteroskedasticity GARCH (1,1) model is proposed. In the proposed method, time series observations are divided into several segments and a GARCH (1,1) model is fitted to each segment. The goodness-of-fit of the global model composed of these local GARCH (1,1) models is evaluated using the corresponding information criterion (IC). The division that minimizes IC defines the best model. Furthermore, since the simultaneous estimation of all possible models requires huge computational time, a new time-saving algorithm is proposed. Simulation results and empirical results both indicate that the proposed method is useful in analysing financial data.

Keywords:

• GARCH(1,1)
• information criterion
• model selection
• parameter change

Acknowledgements

I am grateful to two anonymous referees for their very useful comments and suggestions. Needless to say, any remaining errors are mine.