<![CDATA[Purpose –This study examines the effect of sectoral public expenditure comprising health, education, social protection, village fund (Dana Desa), and capital infrastructure expenditure on poverty dynamics across 13 districts/cities in Central Sulawesi Province, Indonesia, over the period 2015–2024. Design/methodology/approach – Using balanced panel data (N=13, T=10, NT=130), this study estimates a Fixed Effect Model (FEM) with Driscoll-Kraay standard errors robust to heteroscedasticity, serial autocorrelation, and cross-sectional dependence. The Chow Test (F=80.014; p<0.001) confirms the relevance of individual effects, and the Hausman Test (Chi²=4.469; p=0.484) does not reject the Random Effect specification; nevertheless, FEM is retained on theoretical grounds because district-specific unobserved heterogeneity is expected to correlate with fiscal allocations. Diagnostic tests confirm serial autocorrelation (Breusch-Godfrey: χ²=53.775; p<0.001) and cross-sectional dependence (Pesaran CD: z=5.750; p<0.001), justifying Driscoll-Kraay inference. Two nested model specifications are estimated: Model 1 (health, education, social protection) and Model 2 (all five fiscal variables including Dana Desa and capital expenditure). All expenditure variables are deflated using the regional CPI (base year 2018=100), and the Dana Desa variable is transformed as ln(Dana Desa + 1) to accommodate the structural zero values for Kota Palu. Finding/Results – All fiscal expenditure variables are measured as total district-level real expenditure (deflated by CPI, base year 2018=100); population size is controlled for implicitly through district fixed effects. In Model 1, health expenditure is the only variable that consistently and significantly reduces poverty (coefficient = −2.178; p<0.001). In the full Model 2, health expenditure retains its negative direction but with only marginal significance (coefficient = −1.552; p=0.076), indicating weaker evidence in the full specification. Capital expenditure (+1.499; p<0.001) and social protection (+0.822; p=0.003) are significant in Model 2. Education expenditure shows no significant short-to-medium-term effect across both models. Village Fund (Dana Desa) loses significance after CPI deflation, indicating that earlier nominal-data findings likely reflect spurious inflation-driven correlations rather than genuine real welfare effects. Originality/Value – This study provides the first CPI-deflated panel analysis of multi-sectoral public expenditure and poverty in Central Sulawesi, addressing methodological gaps in the existing literature. The findings have important implications for regional fiscal policy reform in Eastern Indonesia, particularly regarding the spatial reallocation of capital expenditure and governance strengthening for the Village Fund (Dana Desa).]]>
Copyrights © 2026
