Threshold Approaches to Asymmetric Models for Heteroscedastic Time Series

초록

Asymmetric volatility or the so called leverage effect has recently attracted considerable attention owing to empirical evidences from various time series. This review paper discusses two threshold approaches to account for asymmetries in the Generalized AutoRegressive Conditional Heteroskedasticity (GARCH) context. The first one involves directly modeling the volatility via threshold functions such as in threshold GARCH models (see, Hwang and Kim [1]). The second approach is indirect for which errors are asymmetric, for instance, skewed t-distributed. A threshold concept is introduced for errors (Hwang [2]) as an indirect asymmetry in volatility. Quasi-likelihood and Godambe optimum scores are discussed to estimate parameters. As an illustration for cell lineage studies, a random coefficient model in bifurcating process is discussed.

키워드