Short Term Forecasting of Indonesia’s Policy Interest Rate Using ARIMA

Abstract

This study examines the short-term dynamics of Indonesia’s policy interest rate and develops forecasts using an autoregressive integrated moving average (ARIMA) approach. The analysis uses 95 monthly observations from January 2018 to November 2025 to generate forecasts for December 2025–February 2026. Model selection combines stationarity testing, information criteria, and residual diagnostics. The results indicate that the series becomes stationary after first differencing. Among the three specifications considered, ARIMA(1,1,1) provides the lowest Akaike and Bayesian information criteria, while residual tests reveal no significant autocorrelation. The model projects an approximately stable policy rate of 4.74% throughout the forecast horizon. A retrospective comparison shows that Bank Indonesia maintained the policy rate at 4.75% during all three months. Although the ARIMA forecasts closely approximate the observed rates, a no-change benchmark achieves lower forecast errors over this period. These findings highlight the persistence of Indonesia’s policy rate and the importance of benchmark comparisons in evaluating forecasting performance. The model provides a statistical baseline for short-term monetary analysis that can complement assessments of broader macroeconomic developments