Are High-Tech Employment and Natural Amenities Linked?: Answers from a Smoothed Bayesian Spatial Model

Abstract

We investigate the recently advanced theory that high-technology workers are drawn to high amenity locations and then the high-technology jobs follow the workers. Using a novel data set that tracks high-technology job growth by U.S. county, we estimate spatial parameters of the response of job growth to the level of local natural amenities. We achieve this estimation with a reasonably new class of models, smooth coefficient models. The model is employed in a spatial setting to allow for smooth, but nonparametric response functions to key variables in an otherwise standard regression model. With spatial data this allows for flexible modeling such as a unique place-specific effects to be estimated for each location, and also for the responses to key variables to vary by location. This flexibility is achieved through the non-parametric smoothing rather than by nearest-neighbor type estimators such as in geographically weighted regressions. The resulting model can be estimated in a straightforward application of analytical Bayesian techniques. Our results show that amenities can definitely have a significant effect on high-technology employment growth; however, the effect varies over space and by amenity level.