Abstract

In this study, we demonstrate that a common approach in using the Autoregressive Integrated Moving Average model is not efficient to forecast all types of time series data and most specially, the out-of-sample forecasting of the time series that exhibits clustering volatility. This gap leads to introduce a competing model to catch up with the clustering volatility and conditional variance for which, we empirically document the efficient and lower error use of the Generalized Autoregressive Conditional Heteroscedasticity model instead.

This work is licensed under a Creative Commons Attribution 4.0 License.