Forecasting Volatility Dynamics of Libyan GDP Using EMD-GARCH and EMD-NGARCH Models

Abstract

Accurate volatility forecasting is crucial for financial decision-making, risk management, and economic policy. This study investigates the performance of hybrid models for forecasting volatility, with an empirical application to Libyan GDP. We develop two hybrid models that integrate the Empirical Mode Decomposition (EMD) method with traditional volatility models: Generalized Autoregressive Conditional Heteroskedasticity (GARCH) and Nonlinear Autoregressive Conditional Heteroskedasticity (NGARCH). The EMD technique decomposes the financial time series into intrinsic mode functions (IMFs) to isolate underlying patterns. These components are then modeled using the GARCH and NGARCH frameworks, creating the EMD-GARCH and EMD-NGARCH models. Their forecasting performance is compared against a standard GARCH model using out-of-sample data for Libyan GDP, evaluated with Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Directional Accuracy (DA). The results indicate that both hybrid models significantly outperform the standard GARCH model. The EMD-NGARCH model demonstrates superior performance, achieving the lowest forecast errors and proving most effective at capturing the asymmetric and nonlinear features inherent in the data. The integration of EMD with nonlinear volatility models provides a superior framework for forecasting volatility in complex economic environments. The EMD-NGARCH model, in particular, offers a powerful tool for enhancing risk management and strategic financial planning for economies like Libya's.