Title Pending 2267

This is an accepted article with a DOI pre-assigned that is not yet published.Childhood stunting remains one of Indonesia’s most persistent development challenges, and its reduction is increasingly understood as a problem of public financial management rather than clinical intervention alone. This study examines whether the fiscal architecture of decentralised government — how sub-national revenue is raised and how it is spent — is associated with the prevalence of under-five stunting. Using a panel data from 34 Indonesian provinces observed over three years (2021–2023, N = 102), we estimate pooled ordinary least squares, random-effects and fixed-effects models relating stunting prevalence to five fiscal-decentralisation ratios: the own-source revenue ratio (OSRR), the mining revenue ratio (MRR), the village fund ratio (VFR), the employee expenditure ratio (EER) and the capital expenditure ratio (CER). Formal specification testing (an F-test for individual effects, a Breusch–Pagan Lagrange multiplier test and a Hausman test) guides model selection. The own-source revenue ratio is the most robust correlate of stunting, carrying a negative and highly significant coefficient across all specifications (p < 0.01). The mining revenue ratio displays a near-zero bivariate association that becomes significantly negative once provincial heterogeneity is controlled, indicating a suppression effect that nuances the “resource-curse” interpretation prevalent in the Indonesian literature. Village-fund, personnel-expenditure and capital-expenditure ratios show no robust within-province association. The findings suggest that fiscal autonomy, more than the broad composition of expenditure, is the fiscal feature most closely linked to better child-health outcomes, with implications for intergovernmental transfer design and budget accountability.