This paper develops a model of optimal urban structure which combines data from hedonic regressions with a version of the assignment model. It is proposed that the data be updated in such a way that the model can be run over many time periods. The optimization is structured in such a way that external effects can be considered. The data for these are also obtained from the hedonic regressions. All of the methodology (hedonic regressions, demand estimations, linear assignment optimizations) is currently operational. The usefulness of the approach is further underscored by the ability to track welfare effects over space and over time; surplusses accruing to each renter and owner group (maximum willingness to pay less actual price paid) are computed from the dual of the optimization.