An Integrated Spatial Model of Population Change in South Carolina Counties

Existing studies of population change have grown to consider spatial dependence. The current literature highlights the importance of location when modeling population change. The evidence clearly suggests that population change in one area is dependent on population change in neighboring areas. What is missing in the existing literature is an analysis of which specific features in surrounding areas impact the local area. We employ a model using spatially lagged explanatory variables (SLX) to model population change in South Carolina. In addition to the normal impacts of the explanatory variables in a standard OLS regression, the SLX model measures the impact of the independent variables in contiguous areas. It also allows for separation of spillovers from rural counties versus those from urban counties. We find that the impact of neighboring counties is distinctly different for rural versus urban counties. Local urban population growth is influenced by neighboring counties, both urban and rural. However, local rural population growth is influenced only by neighboring rural counties. For rural counties, the share of retirees and recreational activities in surrounding rural counties are significant in explaining population change. For urban population growth, birth rates, income, and single female pregnancy rates in neighboring urban counties are significant. Retiree share and single female pregnancy rates in neighboring rural counties are also significant in explaining urban population change.