How patenting and self-employment have affected US metropolitan growth

The bidirectional relationship between population and employment is analyzed across 377 US metropolitan areas during 4 overlapping decades in the period 1990-2015. Applying 2SLS regression to the regional adjustment model, current population depends on lagged population, current employment, and various natural and human-created amenities. Likewise, current employment depends on current population, lagged employment, and a series of economic attributes (wages, professional workers, etc.), including self-employment (entrepreneurship) and patenting (inventiveness). Pooled estimation shows that “people followed jobs†before 2000 but then “jobs followed people†afterward. Amenities (+) invariably affected population numbers while wages (-) invariably affected job numbers. Once endogenized, self-employment had a strong impact on job numbers during the entire 25-year period, while patenting had a much weaker and uneven impact on those numbers. Special consideration is given to spatial lag effects, which were found to be mostly negative (indicating competition) in both instances.