Modeling volatility of precious metals markets by using regime-switching GARCH models

This paper aims to test the existence of regime changes by using MSGARCH models for modeling volatility of the four most famous precious metals, i.e. Gold, Silver, Palladium, and Platinum. We fitted around 72 MSGARCH models with different regimes (k = 1,2,3) to the log-returns of each precious metal to test in-sample analysis of volatility. We compare 72 models for in-sample analysis by using the Akaike information criterion (AIC) and choose the best models for all precious metals' series. Further, one-day ahead Value-at-Risk forecasting was conducted by the best MSGARCH. Our finding suggests the existence of regime changes in the GARCH process in most cases analyzed. We also find that regime-switching GARCH models outperform single-regime GARCH specifications when predicting the Value-at-Risk. The results indicate that using MSGARCH models may provide accurate Value-at-Risk predictions, and hence effective in portfolio optimization, pricing of derivatives and risk management etc.