Agglomerative Forces and Cluster Shapes

We model spatial clusters of similar …rms. Our model highlights how agglomerative forces lead to localized, individual connections among …rms, while interaction costs generate a de…ned distance over which attraction forces operate. Overlapping …rm interactions yield agglomeration clusters that are much larger than the underlying agglomerative forces themselves. Empirically, we demonstrate that our model’s assumptions are present in the structure of technology and labor ‡ows within Silicon Valley and its surrounding areas. Our model further identi…es how the lengths over which agglomerative forces operate in‡uence the shapes and sizes of industrial clusters; we con…rm these predictions using variations across both technology clusters and industry agglomeration. JEL Classi…cation: J2, J6, L1, L2, L6, O3, R1, R3. Key Words: Agglomeration, Clusters, Industrial Organization, Silicon Valley, Entrepreneurship, Labor Markets, Technology Flows, Patents, Natural Advantages. Comments are appreciated and can be sent to [email protected] and [email protected]. An earlier version of this paper was titled “Tipping Points and Agglomeration Bubbles.” This research is supported by Harvard Business School and the Kau¤man Foundation. Alexis Brownell and Kristen Garner provided excellent research assistance. We thank Jim Davis, Gilles Duranton, Ed Glaeser, Vernon Henderson, Guido Imbens, Yannis Ioannides, Sonia Ja¤e, Ramana Nanda, Steve Ross, Scott Stern, and seminar participants at Aalto University School of Economics, American Economic Association, Bank of Finland, Boston University, Boston Urban and Real Estate Seminar, European Regional Science Association, Georgia Institute of Technology, Harvard University, NBER Productivity, Philadelphia Federal Reserve Board, University of California San Diego, University of Pennsylvania Wharton School, and University of Virginia for their insights. Kerr thanks the Bank of Finland for hosting him during a portion of this project. The research in this paper was conducted while Kerr was a Special Sworn Status researcher of the US Census Bureau at the Boston Census Research Data Center (BRDC). Support for this research from NSF grant ITR-0427889 [BRDC] and an NSF Graduate Research Fellowship [Kominers] is gratefully acknowledged. Research results and conclusions expressed are the authors’and do not necessarily re‡ect the views of the Census Bureau or NSF. This paper has been screened to ensure that no con…dential data are revealed.