Three-factor profile analysis with GARCH innovations

The technique of ANOVA has been widely used in Economics and Finance where the observations are usually time-dependent but the model itself is treated as independent in time. In this paper, we extend an ANOVA model by releasing the assumption of independence in time. We further release the assumption of homoskedasticity in the traditional profile analysis by introducing GARCH innovations in our proposed profile analysis that allows for both autoregressive and moving average components in the heteroskedastic variance to display a high degree of persistence. We reprise the model with regards to the issue of American Depository Receipts by releasing the time dependence assumption that has been ignored in the literature. Applying our model, we find that the returns from the stocks and the American Depository Receipts are time dependent and hence the traditional ANOVA cannot fully explore the time effect from the data.

• Type: Journal
• Year: 2008
• Where: MCS
• Authors: Pui-Lam Leung, Wing-Keung Wong
• Keywords: American Depository Receipts, ANOVA, MCS 2008, Model, Pattern Recognition