Document Type



This paper extends applications of unconditional and conditional β-convergence and σ-convergence analysis to poverty rates in a panel data sample of Pennsylvania counties during the period 1990-2019. Spatial structural breaks between rural and urban counties in Pennsylvania plus the possibility that Philadelphia County is an outlier are acknowledged to avoid spurious inferences. The findings support the existence of unconditional β-convergence in the pooled, urban, and rural samples with non-metropolitan areas exhibiting the greatest convergence. However, the largest conditional β-convergence is observed for urban counties, and this outcome is robust to the exclusion of Philadelphia County. Graphical evidence evinces a greater degree of σ-divergence in rural areas relative to the pooled and urban samples with metropolitan areas exhibiting neither convergence nor divergence in the absence of Philadelphia County. Statistical evidence based on ADF and DF-GLS tests reveals the presence of σ-divergence in the pooled and rural samples but weaker findings for the urban counties. Panel data tests for unit roots indicate σ-convergence for the full and rural samples but mixed results for the urban sample depending upon the test employed and whether Philadelphia County is included or not. The findings indicate that further investigation of tailored policy responses to poverty in different geographic areas within the same state is warranted.