Forecasting Through a Structural Break: Provincial Revenue Targets Under Fiscal Reform

Abstract

Revenue forecasting in subnational budgets is known to be optimistically biased, but the literature has concentrated on political and institutional drivers of that bias within a stable tax regime. This study examines a different and less studied case: forecasting across a legislated change in the tax base itself. Indonesia's Law 1 of 2022 introduced a district surcharge on provincial vehicle taxes effective January 2025, mechanically transferring a large share of provincial revenue to districts. Using published budget and accountability documents for Central Java province, the study compares monthly realised vehicle tax revenue for 2018 to 2022 with the targets set for 2024 and 2025. Two findings emerge. First, realised revenue in the pre-reform period was highly volatile, with vehicle transfer tax contracting 28.44 per cent in 2020 and recovering 24.57 per cent in 2021, while motor vehicle tax varied far less, indicating that the two bases carry different forecasting risk and should not be projected by a common method. Second, the 2025 targets embody a structural break of a magnitude that no time-series method could anticipate: the motor vehicle tax target falls 35.6 per cent and the transfer tax target 31.8 per cent against 2024, with total provincial revenue down 16.3 per cent. The study argues that budget credibility assessment must distinguish forecast error from legislated reallocation, since conventional accuracy measures would record the latter as a forecasting failure.