Spatial heterogeneity of unemployment in Turkiye: Evidence from OLS, WLS and GWR

Unemployment, as a multidimensional phenomenon, remains a persistent challenge for both policymakers and researchers, particularly in regions characterized by structural disparities.This study investigates the determinants of provincial unemployment rates in Türkiye with an emphasis on spatial heterogeneity.Using data from 81 provinces for 2024, the analysis employs Ordinary Least Squares, Weighted Least Squares, and Geographically Weighted Regression methods.While Ordinary Least Squares and Weighted Least Squares models reveal limitations due to heteroskedasticity and spatial autocorrelation, the Geographically Weighted Regression approach successfully addresses these issues by incorporating spatial variation into parameter estimation.The findings demonstrate that employment exerts an unemploymentreducing effect in all provinces, statistically significant in most of them, whereas utilized agricultural land significantly decreases unemployment particularly in the south-eastern border provinces around Lake Van and in north-central Anatolia.Migration exhibits heterogeneous impacts, increasing unemployment in the Western Black Sea and Central Anatolia while reducing it in the east, and exports reveal regionally divergent effects, with positive associations in some provinces and negative in others.Model comparisons confirm the superiority of Geographically Weighted Regression, which explains nearly 80% of the variation in unemployment and provides the best overall model fit.The results highlight that spatially sensitive approaches are essential for producing reliable estimates and for designing effective, region-specific labor market policies in Türkiye.