Neighborhood effects on productivity : a knowledge spillover analysis using geocoded data

This paper exploits a geocoded, publicly audited, full population dataset on employment and wages in Sweden's city areas, to analyze the relationship between density of economic activity and individual wages. The analysis is based on 250-by-250 meter (about 0.15 miles), 1 km2, 9 km2 and 100 km2 squares. Fixed effects individual-level wage regressions confirm that workers are positively affected by proximity to economic activity, but that the results are dependent on spatial resolution. A doubling of density in a 250 meter square is associated with a 3.7 percent increase in wages, holding individual characteristics and industry structure constant. Basing the analysis on any of the larger squares decreases the elasticity by about one third, confirming the modifiable areal unit problem, and lending support to arguments in favor of the attenuation of human capital spillover effects.