SARIMA Modeling with a State Space Approach for Tax Revenue Forecasting

This document is a research article that evaluates the effectiveness of classical SARIMA and SARIMA State Space models in forecasting monthly tax revenue for the Sumedang Regency Primary Tax Service Office. The study emphasizes the importance of accurate forecasting for fiscal planning and public service provision, particularly in the context of regional tax revenue, which is a significant source of local income. The chosen model, SARIMA(0,1,1)(0,1,1)12, was selected based on AIC values and residual analysis. The article presents a comparison of model accuracy, revealing that while both models yield similar accuracy, the SARIMA State Space model demonstrates lower RMSE and Theil’s U values, indicating its robustness. The findings suggest that state space modeling can effectively address the complexities of fiscal time series data, providing a theoretically grounded approach for enhancing revenue forecasting strategies at the regional level. The study contributes to both practical applications in regional tax management and the broader literature on forecasting complex time series data.