Industrial structure and employment growth in the 1990s in Appalachian counties

Employment growth in the 1990s and its relationship with the initial industrial structure in 1990 are examined in the case of Appalachian counties, after controlling for labor-market conditions and other factors, such as labor mobility, natural amenities, and market size. Spatial exploratory data analysis of the competitive employment growth (CEG) in Appalachian region shows that strong spillover effect of CEG exists among 410 counties. Counties with higher employment growth rates are concentrated in the north side of Atlanta Metropolitan area around Interstate highway 1-75. Counties with lower growth rates concentrated in Central Appalachia, along the convergent border of three states, Kentucky, Virginia, and West Virginia. Another low growth rate concentration is in Northeast Pennsylvania. The existence of spatial autocorrelation affects my empirical model's explanatory power, the significance levels, and the values of coefficients of independent variables. There is no specific theoretical base for the county-interacting mechanism of this empirical model, whereas, the magnitude of each independent variable's impact on employment growth depends on the spatial-weight matrixes. To find a better match, I compare the parameters of the model by using two different weight matrixes, i.e., weights based on physical neighbor interaction and weights based on commuting ties.