Accounting the US-China Trade War and COVID-19 Effects in Forecasting Gold Price using ARIMAX-GARCH Model

Investor demand for gold as a safe investment tool tends to increase during the Covid-19 pandemic. In addition, global uncertainty due to the trade war between the US and China also affects the movement of gold prices which makes gold prices tend to fluctuate. Modeling and prediction of gold prices are very important to minimize risk factors. The 10 years range of data (2001-2021) was analyzed. It was found that the conditions of the trade war between the US and China, as well as the Covid-19 pandemic had a significant impact on the movement of gold prices. By using these two variables as exogenous variables in the Autoregressive Integrated Moving Average with Exogenous Variable (ARIMAX) approach, we get a model that does not meet the classical assumptions of parametric time series analysis, which is the serial correlation in the squared residual indicating heteroskedasticity. To overcome this problem, the Autoregressive Conditional Heteroscedasticity – Generalized Autoregressive Conditional Heteroscedasticity (ARCH-GARCH) approach is used to capture the effect of volatility. The final model is ARIMAX(4,2,1)-GARCH(0,2) with MAPE value of 3.53% and 5.76% for training and testing data. Thus, the model has been able to predict gold prices with high accuracy.