Comparative Analysis of Exchange Rate Volatility Forecasting and Value at Risk Measurement Using GARCH-Type Models: An Empirical Study Based on Major Reserve Currency Pairs

This study compares the performance of GARCH(1,1), EGARCH(1,1), and GJR-GARCH(1,1) models in forecasting exchange rate volatility and estimating Value at Risk (VaR) for the major reserve currency pairs USD/EUR and USD/CNY. Using weekly return data from 2015 to 2025, the analysis shows that both series display fat tails, volatility clustering, and clear conditional heteroskedasticity. USD/CNY exhibits heavier tails and more persistent volatility. Based on maximum likelihood estimation and rolling-window forecasts, the asymmetric EGARCH and GJR-GARCH outperform the symmetric GARCH model. EGARCH delivers the most stable short-term volatility forecasts across both markets. Regarding distributional assumptions, the t-distribution improves the modelling of tail risks, especially for USD/CNY, although it introduces trade-offs for VaR stability during extreme events. Overall, the results highlight important differences between developed and emerging currency markets. They also provide practical guidance for selecting appropriate volatility models and distributional assumptions in exchange rate risk management.