fishin wrote: You are up one level to high. What you've described would be an "MA" (Metropolitan Area") in HMDA-Speak. Each MA is broken down into MSAs and MDs to allow for more granularity.
Metro NYC would be an MA. The Bronx and Manhattan would be MSAs.
Thanks for clarifying. Nevertheless, the study's model would still classify a Bronx in the same category as a Manhattan borrower. They both live inside an MSA, and the model doesn't distinguish between MSAs. It only distinguishes between "Inside an inner city", "inside
some MSA, but not an inner city", and "rural".
fishin wrote:Within the MSAs things are further broken down into MDs. There is no size limit there but they tend to use the postal codes which gives you the individual neighborhoods in major cities (Boston has ~40 postal code for example.)
Again, thanks. So I guess the variable I'd like to see controlled or is "bankruptcy rate within borrower's MD" or something similar. This would help distinguish between raw racism and proxies of financial risk that may not be reflected in the credit record.
Another point that comes to mind: Does the study say anywhere what the probability of a black person is to receive a lower rate than a comparable white? Hypothetically, if the distribution was higher: 30%, same: 40%, lower: 30%, there would be no reason to suspect discrimination even on the face of the data. Do I really expect that the likelihood of a lower rate are really 30%? No, but it would still be nice to know what the actual chances are.