Developing Exp-FIGARCH Hybrid Models for Time Series Modelling

Abstract

In this paper, we introduced a new hybrid model namely Exponential Autoregressive-Fractional Integrated Generalized Autoregressive  Conditional Heteroscedasticity (ExpAR-FIGARCH) model and study financial data. The Daily Nigeria All Share Stock Index that exhibit  nonlinear, volatility and long memory effect were analyzed in the study. The existing ExpAR-Generalized Autoregressive Conditional  Heteroscedasticity (ExpAR-GARCH) model were estimated and compared with the proposed ExpAR-FIGARCHmodel. Results showed that  the new hybrid model is better based on efficient parameters, serial correlation analysis and forecast measures of accuracy. Therefore, as  a conclusion, the current study indicates that the ExpAR-FIGARCHmodel performed better compared to the ExpAR-GARCHhybrid  model. Therefore, the ExpAR-FIGARCHmodel is a better option for modeling nonlinear, volatility and long memory characteristics of time  series. Future study should focus on the application of the developed hybrid ExpAR-FIGARCHmodel using health, meteorological and  economic data.