Jointly forecasting gold and silver prices via threshold regime-switching bivariate HAR model

초록

The gold and silver markets have constantly attracted the intense interest of investors throughout history. The prices of gold and silver often move in tandem, reflecting their shared characteristics as precious metals and safe-haven assets. In order to analyze the correlated data of gold and silver prices, this work proposes a Markov regime-switching bivariate heterogeneous autoregression (HAR) model with threshold. In the model, two regimes are distinguished by time-dependent correlation coefficient series. The empirical study finds that the threshold regime-switching HAR (TRS-HAR) model of gold and silver prices performs better in both fitting and forecasting compared to existing models. By adopting the bivariate normal approximation for the residuals of the model, joint elliptic prediction regions are constructed, which reveals that the TRS-HAR model with the joint bivariate normal approximation is effective in predicting the co-movement of gold and silver prices.

키워드