Hello all,

I've been working on fair lending for a few months, have completed lots of work, and I'm stuck on the transactional comparison portion. My problem is my bank does not have large enough loan volume for me to do a true and accurate transactional comparison for fair lending. Under exam guidance when this occurs, it simply says to expand your time period for the review, to include more loans and obtain your sample that way. I've gone back a year and a half and still do not have a large enough controlled group sample (trying to use surrogate names as well) to complete this.

Has anyone else encountered this? If so, how far back did you go? It seems a bit much to go back years and years to get a large enough sample, especially since then products may change. I've documented my review well, and was thinking that some language stating we don't have a large enough universe to complete an accurate fair lending transactional comparison would be sufficient. Is that okay? And if so, does anyone possibly have some language from a fair lending audit, review, or even exam that they could provide to that effect? Something that states I've done this review and determined that because of the lack of application data the analysis could not produce a meaningful result, and therefore, not presented.

Thanks in advance for any assistance.

I simply exclude groups that don't allow for meaningful comparison/analysis. For example, our lending area has a very small population of Native Hawaiian or Other Pacific Islander. I simply would say that we recieved 1 application, which was approved, but that there was not enough applications for meaningful analysis. Is this what you mean?

Yes. My bank is primarily a commercial bank. We had about 75 total applications on our HMDA LAR for 2011. I can't get a sample of 5 on any protected class. I've even looked at commercial and attempted to use surrogate names (as described in the examiner guidance) to gather a sample for comparison of commercial loan products, and again, I'm coming up short. I took a sample of all loans from January 2011, through the 1st Quarter of 2012. If I go back any further, I think the results will be even more skewed. I guess an examiner could look at the available data and come to the same conclusion, but I want to be sure I'm covering everything, and be able to show that we went through the process of attempting a true fair lending review.

Is this proportion (<5/75) consistent with the demographics of your defined lending area? What I do for all protected classes is look at the census data. If 5% of my applications are from Black or African American applicants and the population of my lending area is about 5% Black or African American, I'm comfortable having too few for analysis. However, if my lending area is 50% Black or African American, then I'm concerned. Simply put, some banks simply operate in areas with very low prohibited basis group populations and as long as the demographic data support your lending activity, I wouldn't be concerned.

In terms of timeframe, I conduct a review on an annual basis and consider a year of data.

Are you able to at least make a male vs. female comparison.

J5, the following is from the Interagency Fair Lending Examination procedures, which by your post you are familiar with. See especially step "c" under "set the sample size.

C. Transactional Underwriting Analysis - Residential and Consumer Loans.
Step 1: Set Sample Size
a. For each focal point selected for this analysis, two samples will be utilized: (i) prohibited basis group denials and (ii) control group approvals, both identified either directly from monitoring information in the case of residential loan applications or through the use of application data or surrogates in the case of consumer applications.

b. Refer to Fair Lending Sample Size Tables, Table A in the Appendix and determine the size of the initial sample for each focal point, based on the number of prohibited basis group denials and the number of control group approvals by the institution during the twelve month (or calendar year) period of lending activity preceding the examination. In the event that the number of denials and/or approvals acted on during the preceding 12 month period substantially exceeds the maximum sample size shown in Table A, reduce the time period from which that sample is selected to a shorter period. (In doing so, make every effort to select a period in which the institution’s underwriting standards are most representative of those in effect during the full 12 month period preceding the examination.)

c. If the number of prohibited basis group denials or control group approvals for a given focal point that were acted upon during the 12 month period referenced in 1.b., above, do not meet the minimum standards set forth in the Sample Size Table, examiners need not attempt a transactional analysis for that focal point. Where other risk factors favor analyzing such a focal point, consult with agency supervisory staff on possible alternative methods of judgmental comparative analysis.

d. If agency policy calls for a different approach to sampling (e.g., a form of statistical analysis, a mathematical formula, or an automated tool) for a limited class of institutions, examiners should follow that approach.


You're right - when you are matching, it is for a single prohibited basis group. You could not have a black female compared to a white male. If there is an issue, it could not be determined if it was because of race or gender.

Since you do not have enough for comparative file reviews, I'm sure the examiners will look at the individual files to ensure there are no violations of the ECOA or FHA. If they do see a violation, they will test the loans for similarities, and you would have a violation of the reg, but not because of a comparison. There are minimum numbers because of statistical accuracy issues. They could not extrapolate the information from a very low sample - however they could cite individual exceptions.

Good luck - it seems that you and the bank are covering all the bases.

KPOC, while very important for any fair lending review, that would probably fall under the application tests/marketing/outreach rather than comparative file review.

Southern Banker, you are absolutely correct. However, what I typically do is use the marketing piece to help justify why I have so few applications for comparative file review, rather than just moving on without looking at why sample sizes are so small.

Maybe I shouldn't mix my fair lending review sections.

Thank you everyone for taking the time to reply. It's greatly appreciated.

As for the demographics, we are assessing that in our marketing outreach. As I said we're primarily commercial, we do very few consumer loans, and most of what we do make is through referrals and such. Although we've recently begun getting more involved in consumer residential, and while we don't advertise, we're cognizant that marketing to the areas we serve is still a factor.

I've been doing the individual reviews for ECOA and FHA. I'm not as concerned about an examiner spotting an individual exception as I am with them stating we're being discriminatory. All these news articles lately hype up everything, and it makes it a little nerve racking, so I want to be as thorough as I can, even if the available data is limited. I also wanted to show some kind of end result, even if it was mostly verbiage, stating that a review was conducted but a true and accurate fair lending comparison is simply not possible. Maybe I'm just overthinking everything too much.

Thanks again for the input and references.

Rock . . . . .Johnny5 . . . . Hard Place

With examiners, you'll usually find yourself between a rock and a hard place. [pun intended]

Like KPOC indicated, you may want to concentrate on marketing. In those rare instances where there is communication between the bank and someone outside the bank (potential customer), make sure it is covered somehow in policy, practice, procedures and validation.

For example, if the LO's talk to your companies for internal referrals, make sure it covers all of your customers - bigger ones to the minority owned and those in LMI areas. If they discretionary solicit from brokers, ensure that they do not just go to the ones in the suburbs, but also call on the ones in LMI areas. Document their outreach (if any) especially to protected class sources.

Marketing appears to be the only areas [inferred] that might have some discretion, especially for a "commercial" bank.

Hahaha thanks Southern Banker. I'll be sure to give that a thorough review. I want to have everything outlined for the examiners so their questions are very minimal, if any. We already have a huge binder dedicated specifically to fair lending that is somewhat of a roadmap to our processes.