Expected income and labor market choices of US married couples: A locally weighted regression approach

• We define the geographic boundaries of US labor markets using Census metro areas.
• We use a loess regression to estimate the wage offer from each labor markets.
• We find the local measures affect US couples’ migration decisions significantly.
• The state average measures are not significant in the location choice process.

This paper applies a locally weighted scatterplot smoothing (loess) method to estimate the spatially heterogeneous wages of demographic groups of workers across precisely defined US labor markets. We estimate a location choice model using data from the National Longitudinal Survey of Youth (NLSY79) using these estimates of labor market specific wages for men and women as determinants of their place of residence. We compare estimates of this model to a model using more aggregated measures of wages and locations from CPS. We show that potential wages based on these more refined definitions of labor markets and demographic groups provide more explanatory power in a simple migration model than do those based upon less detailed definitions of labor markets and demographic groups.