Analisis Hubungan Dinamis dan Peramalan Inflasi di Indonesia: Pendekatan Vector Autoregression (VAR) dan Bayesian Vector Autoregression (BVAR)

Abstract

This study analyzes the dynamic relationships between inflation and key macroeconomic variables and compares the forecasting accuracy of the Vector Autoregression (VAR) and Bayesian Vector Autoregression (BVAR) models in Indonesia over the 1989–2024 period. The study employs annual data on inflation, exchange rates, money supply (M2), real Gross Domestic Product (GDP), Producer Price Index (PPI), world oil prices, and global inflation. VAR and BVAR models are used to examine the dynamic interactions among the variables, while Impulse Response Function (IRF) and Forecast Error Variance Decomposition (FEVD) are employed to assess the responses of inflation to macroeconomic shocks and the contribution of each variable to inflation forecast error variance. The forecasting performance of the models is evaluated by comparing their forecast errors. The results show that inflation exhibits dynamic responses to shocks in all selected macroeconomic variables and gradually returns toward equilibrium following the shocks. The FEVD results indicate that, after its own shocks, the exchange rate provides the largest contribution to the forecast error variance of inflation, followed by world oil prices, real GDP, money supply (M2), global inflation, and PPI. Furthermore, the forecasting evaluation shows that the BVAR model with the Minnesota Prior produces lower forecast errors than the VAR model. These findings indicate that incorporating Bayesian shrinkage through the Minnesota Prior can improve the forecasting performance of inflation. The BVAR model therefore provides a more accurate approach for forecasting inflation in Indonesia.