PEMODELAN REGRESI DATA PANEL PADA POLA KEMISKINAN: STUDI KASUS KABUPATEN/KOTA DI PROVINSI BALI

Abstract

Poverty remains one of the key indicators of development success and continues to challenge in Bali Province. This study combines panel data regression with comprehensive diagnostic testing and spatial analysis using Local Indicators of Spatial Association (LISA) to examine the determinants of the Poverty Rate (PR) across nine regencies/cities in Bali Province, using 72 observations from 2017–2024. Independent variables include economic growth (EG), unemployment rate (UR), district/city minimum wage in natural logarithm form ( ), and average years of schooling (AYS). Three panel data approaches were estimated, namely the Common Effect Model (CEM), Fixed Effect Model (FEM), and Random Effect Model (REM). Based on the Chow and Hausman tests, FEM was selected as the best model. Diagnostic tests revealed autocorrelation and cross-sectional dependence, so inference relied on Arellano robust standard errors. The results show that minimum wage and average years of schooling significantly reduce poverty, while economic growth and unemployment rate are not significant. LISA analysis identified a significant Low-Low cluster in Denpasar, reflecting spatial spillover in southern Bali. These findings highlight minimum wage policy, education access, and place-based regional strategies as key instruments for poverty reduction.