Universal Journal of Accounting and Finance Vol. 12(1), pp. 13 - 23
DOI: 10.13189/ujaf.2024.120102
Reprint (PDF) (760Kb)

Assessing the Forecasting Performance of GARCH Models in the Presence of Instabilities

Azwifaneli I. Nemushungwa *
Faculty of Management Sciences, Commerce and Law, University of Venda, South Africa


Proper modeling and anticipation of business volatility play an important role in risk management, derivatives allocations, and the price of assets in a finance research field. Hence, it means the requirement for forecasting models that deliver accurate forecasts during those periods. An accurate forecasting model is critical for interpreting financial data, and confirming the time series is stationary is a vital step in model development. Moreover, one must select a model that does well in a particular asset class. There is no single best model across all asset classes and periods. As a result, identifying a model that performs well in specific asset classes is critical. Several forecasting models are available to decision-makers, and no single model emerges as the best all-around. This is due to instability in predicting performance, which varies by state and is based on time-varying economic factors. As a result, this study compares the forecasting abilities of some GARCH models for exchange rate data on the South African market by using traditional and fluctuation tests within normal students, and general error distribution assumptions. Forecast accuracy was assessed using four model accuracy measures: root mean square error, mean absolute error, mean absolute percentage error, and the Theil inequality coefficient. The Giacomini and Rossi (2010) fluctuations test and the Diebold and Mariano test were used to evaluate relative predicting skills in the face of instabilities. In contrast, the Rossi-Sekhposyan (2016) test was utilized to determine if absolute predicting performance is robust to instabilities. It is revealed that symmetric GARCH models do not overperform those models with asymmetry under several assessment indicators and error distributions. Giacomini and Rossi's (2010) test confirms the efficiency of all models undertaking the t-distribution approach. However, the Sekhpoysan (2016) test suggested that despite all models generating good forecasts, an individual model might be making weak predictions compared to the others. The practical implication is that all models can make accurate predictions. It may perform poorly solely when compared to other models.

Forecasting Abilities, Traditional Tests, Fluctuation Tests, Symmetric GARCH Models, Asymmetric GARCH Models, Exchange Rate Data, South Africa

Cite This Paper in IEEE or APA Citation Styles
(a). IEEE Format:
[1] Azwifaneli I. Nemushungwa , "Assessing the Forecasting Performance of GARCH Models in the Presence of Instabilities," Universal Journal of Accounting and Finance, Vol. 12, No. 1, pp. 13 - 23, 2024. DOI: 10.13189/ujaf.2024.120102.

(b). APA Format:
Azwifaneli I. Nemushungwa (2024). Assessing the Forecasting Performance of GARCH Models in the Presence of Instabilities. Universal Journal of Accounting and Finance, 12(1), 13 - 23. DOI: 10.13189/ujaf.2024.120102.