Identifying Sub-centres Using Contiguity Matrices

This paper presents a simple algorithm for identifying sub-centres using standard properties of contiguity matrices to simplify the identification procedure. Sub-centres are identified for Atlanta, Baltimore-Washington, Boston, New York and Philadelphia. A sub-centre is defined to be a set of contiguous tracts that each have significantly positive residuals in a first-stage non-parametric regression and that together have at least 10 000 employees. The algorithm is also applicable when potential sub-centre sites are identified by other methods, such as minimum employment density cut-offs.