DRAFT ONLY- Do Not Cite Without Permission of the Authors The Topography of Metropolitan Employment: Identifying Centers of Employment in a Polycentric Urban Area∗ by

Increasingly, U.S. metropolitan areas are polycentric. While this is well recognized, there is little consensus as to the appropriate method for identifying centers of employment and their extent. Discussions of sprawl and decentralization, agglomeration and productivity, and the impacts of transportation or land-use regulation on urban structure depend crucially on the spatial accounting of employment within a metropolitan area. Existing methods for subcenter identification suffer from either strong assumptions about parametric form, misspecification, or reliance on local knowledge to calibrate model parameters. Using data from the greater Los Angeles metropolitan area, this paper introduces a non-parametric method for mapping the topography of employment. The method is benchmarked against several representative alter-natives to subcenter identification. The importance of the difference in approaches is made clear by comparing their measured concentration of the greater Los Angeles metropolitan area. Results indicate that the more flexible non-parametric approach yields both greater accuracy in defining subcenter boundaries and better resolution identifying a wide range of subcenters. These attributes should help inform research that employs density as an independent or depen-dent variable