Association for Public Policy Analysis & Management

How does regional entrepreneurship evolve? This research focuses on the drivers of regional sectoral change through new business activity, and productivity growth through higher labor market reallocation, as a means to more accurately identify industry clusters for greatest returns to targeted entrepreneurship policy making. This research applies a matrix of occupational skill-weighted co-specialization probabilities of industries in 366 U.S. metropolitan areas to estimate industry “relatedness”, and develops a clustering technique that is highly correlated with metropolitan entrepreneurship rates and labor market reallocation rates. These two processes, entrepreneurship and labor market reallocation, are critical drivers of economic development and represent incentive targets for regional policy makers. While orthodox clustering techniques face some limitations in identifying cross-sector interactions that underscore these processes, the technique proposed here directly emphasizes the co-evolution of regional cross-sector interactions, and thus is centered on innovative growth and structural transformation. Accordingly, the clustering technique has four features that bare direct relevance to policy studies of entrepreneurship and innovation. 1. The clusters are metropolitan specific and reflect the economic ‘histories’ of regions, thus are dynamic and sensitive to time trends. 2. The clusters are identified at the intersection of industries and their composite occupations, thereby engaging a nuanced degree of human capital specificity. This translates to better identification of skill-gaps given a region’s unique industry-occupation composition. 3. With its focus on entrepreneurship and skills, this method provides a better identification of the sources of new jobs in metropolitan areas. 4. The relatedness matrix provides a similarity measure that reveals a network of interdependence across metropolitan economic activity. This yields to innovative new techniques for studying the network properties of regional clusters, for instance, in terms of regional resilience to exogenous shocks, including technological shocks.