Analyzing of Multivariate Time Series with Vector Autoregressive (VAR) Models

This study applied a Vector Autoregressive (VAR) model to examine the active relationship between global oil and gold prices using this data from January 2015 to June 2019. Stationarity was tested through the Augmented Dickey-Fuller (ADF) test, with necessary transformations applied. Lag order selection criteria (RMSE, AIC, BIC, HQC, and FPE) identified VAR (1) as the optimal model. Findings revealed that gold prices are highly persistent and mostly influenced by their individual past values, while oil prices exert a negative effect on gold price changes. Diagnostic tests confirmed model adequacy and stability. Forecasting results suggested a gradual decline in gold prices and a slight increase and stabilization in oil prices over time. The study concludes that VAR models provide a reliable and effective framework for analyzing inter-market relationships and generating short- to medium-term financial forecasts. All analyses were conducted using Python programming.