This paper decomposes the black-white gap in homeownership propensity into an endowment component and a residual component across its distribution. We document evidence that supports the view that the racial gap differs across its distribution and that studies that only examine the gap at the conditional means are misleading. We find that although household characteristics could explain the racial gap for households that are more likely to own homes, there is still a substantial portion of the gap that remains unexplained particularly for households with a low propensity to own homes. We estimate the counterfactual distribution using a semiparametric procedure which has the advantage of looking at disaggregated data and does not restrict the sample data to segregated neighborhoods. We illustrate this flexibility by decomposing the racial gap at the city-level.