Diversity of human capital and regional growth

In this paper we study the impact of diversity on regional growth by extending the existing literature in such that we differentiate between industry diversity and human capital diversity. In order to measure human capital diversity we construct a regional measure based on individual occupational data. In fact, based on panel data for German regions we find empirical evidence that regions with higher degrees of human capital diversity exhibit higher GDP-per-capita and employment growth, as well as higher patent-output per R&D-worker. So far the empirical literature whether regional specialization or regional diversification encourage knowledge spillover and therefore promote regional growth mainly focused on the diversity of the regional industry structure and thus disregards that knowledge transmission merely occurs between individuals. This was already described by Jacobs (1969), and is also recognized by Glaeser et al. (1992) who emphasize the importance of interaction between people in close geographical distance for innovation. Nevertheless, the existing literature typically relies on the regional industry diversity as an indicator for the breadth of the local knowledge base. However, we argue that the diversity of skills and knowledge at the individual level rather than the diversity of industries reflects the scope of the local knowledge base and the potential for spillover. A more fruitful approach should therefore take a more disaggregated view on this topic by looking at the diversity of skills and abilities at the level of individuals. We apply two strategies to assure that human capital diversity is not just a proxy of the regional industrial structure. First we calculate a variable for the regional industry diversity equivalent to the occupational diversity using regional industry employment shares at the three digit industry level. Second we incorporate regional employment shares of 27 out of 28 aggregated industries as additional explanatory variables in our regression. Furthermore, skill ...