A prominent feature of economic geography in America is the positive correlation amongst local incomes, housing costs and city population. This paper embeds a “black box” agglomeration economy within a more neoclassical general equilibrium model of local wages, rents and population to assess the ability of various conceptual models to predict this cross-sectional variation. I use exogenous changes in housing supply to induce changes in population and examine whether the changes in rents and wages move in the same direction under neo-classical assumptions, agglomeration economies in production, congestion in production, or urbanization economies in consumption. On their own, none of these urban scale effects generate the observed pattern. All urban scale effects generate a negative correlation between rents and population. Combining natural advantage with the urban scale effects improves the models’ output. It generally predicts positive correlations amongst the three variables, although some of these effects are ambiguous in the production agglomeration model. If natural advantage and housing supply constraints vary more-or-less independently, the results suggest a better fit of the data is provided by either the congestion in production or the agglomeration in consumption models. The micro-economics of such consumption-oriented agglomeration economies have received less attention than production-oriented agglomeration economies. The results of this model thus suggest that consumption-oriented agglomeration and congestion should receive more attention in the future.