The blog of the Urban Institute
June 6, 2014

Weaker credit or racial discrimination: the data are unclear

June 6, 2014

Some readers of our June 5 blog post on mortgage denial rates opined that the discrepancy in mortgage denial rates between white and minority applicants with weaker credit indicates racial discrimination.

It is important to note, however, that we do not know whether the clear racial discrepancy among those with weaker credit is due to variations in treatment of applicants with the same credit profile (i.e., racial discrimination) or variations in the level of credit weakness among these groups (i.e., minority applicants have weaker credit profiles than white applicants). We were not able to fully test either hypothesis, since the credit profile distribution of the mortgage applicants is unknown.

The only information available to researchers is the credit profile distribution of mortgage borrowers, which by itself is not enough to test the above hypotheses.

Nevertheless, we thought it might be helpful to provide further information about the credit profile distribution of the mortgage borrowers:

We analyzed the credit profile distribution of originated GSE loans by matching Home Mortgage Disclosure Act and CoreLogic data. We considered borrowers to have a weak credit profile if their loan fit into either of these two categories:

  1. A combined loan-to-value ratio (CLTV) above 90, or
  2. A back-end debt-to-income ratio (DTI) above 45, or
  3. A CLTV below 90 and DTI below 45 with a FICO credit score below 680.

Note that an individual borrower could have a relatively strong credit profile if one only considers a single factor and not all three factors: CLTV, DTI, and FICO score.

When we further subdivide the borrowers, an interesting pattern emerges within the group of borrowers with the weakest credit profile: African American and Hispanic borrowers with weaker credit—the lowest 25th percentile—tend to have lower FICO scores than their white and Asian counterparts with weaker credit. The difference within the 25th percentile between African American and white borrowers is 19 units.

If the pattern for applicants who were denied a loan is consistent with that found in the bottom quartile of borrowers, which would be intuitive, African American mortgage applicants may simply have weaker credit profiles than their white counterparts. This could account for the racial discrepancy in the denial rates of applicants with weak credit profiles.


As an organization, the Urban Institute does not take positions on issues. Experts are independent and empowered to share their evidence-based views and recommendations shaped by research.


Would it be worthwhile estimating a selection model and predicting a "shadow credit profile" for denied loans? I don't know if you have the data to do this in a credible way, but it might be an interesting approach.