<![CDATA[Labor force participation is a crucial indicator of regional economic development, providing essential evidence for effective employment policies. However, empirical studies analyzing labor force determinants frequently rely on Poisson regression, which assumes equidispersion and produces inefficient estimates when socioeconomic count data exhibit overdispersion. Despite this widespread issue, the application of Negative Binomial Regression (NBR) to model labor force participation remains limited, particularly in South Sulawesi, Indonesia. Therefore, this study aims to examine the effects of Gross Regional Domestic Product (GRDP), average years of schooling, poverty rate, and the population aged 15 and older on the labor force in South Sulawesi, while identifying the most appropriate regression model for overdispersed count data. Using a Generalized Linear Model (GLM) approach, the study compares Poisson regression and NBR using 2023 secondary data from Statistics Indonesia (BPS). Model performance was evaluated through overdispersion testing, Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). Results demonstrate significant overdispersion in the data, rendering the Poisson model inadequate. Conversely, NBR demonstrates superior performance with lower AIC and BIC values. Furthermore, the findings reveal that average years of schooling and the population aged 15 and above have statistically significant positive effects on the labor force, whereas GRDP and the poverty rate do not. These findings imply that educational attainment and demographic structures are stronger determinants of labor force participation than macroeconomic conditions in South Sulawesi. Ultimately, NBR provides a more robust framework for modeling such data, offering reliable empirical evidence for regional workforce planning.]]>
Copyrights © 2026
