Using a proprietary data set of credit bureau records, Cohen-Cole (2011) finds evidence that lenders are using the racial composition of a borrower's neighborhood to set credit limits on revolving accounts. Using the same credit bureau data, I revisit this work and reach two main findings. First, an undocumented decision in constructing the variables appears to have introduced a distortion that is highly correlated with neighborhood racial composition and appears to increase the size of the reported disparity. Second, when neighborhood income is controlled for, the results presented as evidence of redlining disappear.

Some have suggested that housing policy, embodied by the Community Reinvestment Act (CRA) and affordable housing goals of the government-sponsored enterprises (GSEs), caused the subprime crisis. We examine if these programs led to worse mortgage outcomes using two approaches. The first examines whether more activity by CRA-covered lenders, or more loan sales to the GSEs, was associated with worse outcomes. The second uses regression discontinuity to determine if outcomes were worse at the geographic thresholds used by each program. Our results suggest that neither program played a significant role in the subprime crisis.