Fair Lending Analysis: 3 Methods for Determining Borrower Race & Ethnicity
When examiners look at your loan portfolio for fair lending, they want to see how your financial institution performed across different races and ethnicities. Was one group given favorable treatment for the credit decision, pricing, or terms of the loan?
The problem when answering this question is that an FI doesn’t always know the race or ethnicity of its borrowers. Mortgage borrowers are given the option to share that information, but it’s not required. For consumer lending such as auto loans, credit cards, and personal loans FIs aren’t even allowed to ask or collect race or ethnicity in most cases.
As a result, lenders have to determine borrowers’ ethnicity and race using “proxy” data, or data based on the FI’s best estimates using the information it does have, including publicly available demographic data.
There are three main methods for determining an applicant’s race and ethnicity when conducting a fair lending analysis.
3 Methods for Determining Borrower Race & Ethnicity
- Ethnicity from Fed Tables *
- Bayesian Improved Surname Geocoding (BISG) **
- Ethnicity and Race based on Census tables ***
These methods are all relatively similar, combining last name with location to determine the likelihood that an individual is of a specific race and ethnicity. The methods all draw from U.S. Census data and differ only in their calculations.
How accurate are these methods?
When, it gets returned with a probability of accuracy. For instance, it might say that someone named Page Simmons in Fort Lauderdale, Fla., has an 85 percent chance of being African American.
It’s up to the financial institution to decide what level of accuracy to require. For instance, it may decide that a result must have an 80 percent probability of being correct to be included in the analysis.
Can I use all the methods at once to get the most accurate answer?
No. That’s not a good idea for two reasons.
- That means your methods won’t match.
- It opens you up to accusations of cherry picking, a particularly bad idea when talking about fair lending.
I have a strong fair lending record. Do I really need to do this?
Even if your FI has an immaculate record, it’s always a good idea to do your own internal reviews and look for any weaknesses. Most institutions come across at least a few red flags when analyzing their data. By analyzing your data yourself, it gives you an opportunity to research any issues that come up and determine if it’s a fair lending issue or an anomaly.
Start by determining the race and ethnicity of borrowers to see if the FI is compliant with fair lending when it comes to:
It’s much better to discover these issues yourself and figure out what happened on your own timeline vs. the pressure of figuring it out with the examiner there. This way you’ll have a prompt response to the examiner’s concern. It looks bad when examiners uncover an issue and you don’t know what happened.
This sounds complicated. How do I do it?
This isn’t something most FIs are equipped to do on their own. Rather than hire in-house expertise, many FIs outsource the actual analysis to a third-party provider and products that can scrub the data and run the analysis. The FIs then set thresholds for acceptable levels of probability and look for reasons behind any results that suggest potential fair lending issues.
Don’t be intimidated by proxy data. Make sure your FI has a way to gather the race and ethnicity information it needs to thoroughly analyze its fair lending performance.