![]() | Only 14 pages are availabe for public view |
Abstract One of the main assumptions of the stationary time series models is that the variance of the errors terms is constant. In many practical applications, this assump- tion may not be realistic. The rst model that provides a systematic framework for volatility modeling is the autoregressive conditional heteroskedasticity model of Engle (1982), and it is often denoted by ARCH models. Although the ARCH model is sim- ple, it in often requires many parameters to adequately describe the volatility process. Therefore, Bollerslev (1986) and Taylor (1986) proposed to gain greater parsimony by extending the model in a similar manner as the AR model when moving to mixed ARMA models. They suggested the generalized ARCH (GARCH) model. Bilinear (BL) time series models have been studied by Gabr and Subba Rao (1981, 1984). This class of time series has been found to provide a better t as well useful in many areas including biological sciences, ecology and engineering. Bilinear class may be regarded as a non-linear extension of ARMA. |