Abstract
This study examines the conditional associations between institutional integrity, inflation, unemployment, and regional economic growth across ten provinces in Sumatra, Indonesia, during the 2020–2024 pandemic-to-recovery period. Annual secondary data from Statistics Indonesia (BPS) and the Corruption Eradication Commission (KPK) form a balanced panel of 50 province-year observations. The analysis considers pooled, fixed-effects, and random-effects specifications and uses the Chow, Hausman, and Breusch–Pagan Lagrange Multiplier tests for model selection. The selection evidence is mixed: the Chow cross-section F test does not reject pooled estimation at the 5% level (p = 0.0539), the Chow chi-square result supports province-specific effects (p = 0.0140), the Hausman test favors fixed effects (χ²(3) = 18.005400, p = 0.0004), and the LM test does not favor random effects over pooled estimation (p = 0.2400). The Fixed Effects Model is retained as the primary specification because the Hausman result and the research context indicate that unobserved provincial characteristics may be correlated with the regressors. Institutional integrity is negatively but not significantly associated with growth (β = −0.018254, p = 0.8662), inflation is positively associated with growth (β = 0.428665, p = 0.0070), and unemployment is negatively associated with growth (β = −2.097523, p = 0.0001). The model is statistically significant overall (F = 2.754790, p = 0.008951), with R² = 0.471862 and adjusted R² = 0.300574. Because the design is observational and does not resolve endogeneity or common year shocks, the coefficients are interpreted as conditional associations rather than causal effects. The findings suggest that regional recovery strategies may benefit from greater attention to employment conditions, the sources of inflationary pressure, and institutional strengthening linked to measurable administrative outcomes, while these implications should be interpreted cautiously given the observational design.