Modeling Migration Effects on Agricultural Lands: A Growth Equilibrium Model

Growth in population and employment can result in increased demands on agricultural land for non-agricultural uses. This study develops a growth equilibrium model at the county-level for the state of West Virginia. Each growth model is a structural equation model that addresses the endogeneity of population densities and employment densities. Three single-decade models are specified – 1970s, 1980s, and 1990s using a two-stage least squares regression technique. Each model is unique, reflecting changing population and economic structures over time. For West Virginia, our results suggest that jobs follow people. Population density in 1990 was positively related to areas with higher proportions of farmland in cropland in 1980, but was negatively related to those areas where larger proportions of cropland were in pasture. End of decade employment was not statistically associated with quantity of farmland in any of the models. There are at least two explanations for this result. First, the location of cropland in 1980 may be serving as a proxy for an omitted variable in the model (e.g., cropland may be spatially correlated with other factors contributing to population growth patterns in West Virginia). Second, the implied direction of causation maybe wrong. Growth may be a push factor in the decline of agricultural land, but agricultural land is not a pull factor for growth. We will expand this analysis to look at recursive models and other systems of equations models that better capture the relationships between agricultural land and growth