Multivariate Garch Models Comparison in Terms of the Symmetric and Asymmetric Models

Abstract

Modelling the Multivariate Generalised Autoregressive Conditional Heteroscedasticity (GARCH) model included both symmetric and asymmetric processes. The information includes monthly data for the Consumer Price Index, Crude Oil Price, Exchange Rate, and Inflation Rate from January 2005 to December 2021. For the analysis, E-view Statistical software was employed. The aforementioned four macroeconomic variables show a tendency for volatility to cluster across time. Both symmetrical and asymmetrical processes had the volatility condition. The residual recursive plot was used to analyse the structural fluctuation in the series. The plot showed continual movement in the inflation rate as well as a downward and upward movement in the consumer price index, exchange rate, and crude oil price. The Akaike Information Criterion (AIC), Hannan-Quinn Information Criterion (HQIC), and Schwarz Information Criterion were chosen as the best models based on information criteria (SIC). Symmetric and asymmetric modelling techniques were used to create the Economic Variables Multivariate GARCH (M-GARCH) models. To calculate the covariance and correlation between the four variables, M-GARCH models were utilised. The main finding of the estimation of all M-GARCH models is that the symmetric models (Diagonal BEKK and Constant Conditional Correlation, or "CCC") have the lowest values of the model information criteria compared to the asymmetric models (Diagonal BEKK and CCC), while the asymmetric models (Diagonal VECH) have the lowest values of the model information criteria compared to the symmetric models (Diagonal VECH). Based on the findings, it was discovered that while analysing the interaction between the four economic variables returns series, the Symmetric Diagonal BEKK and CCC model outperformed the Asymmetric Diagonal VECH model.