HMDA Partial Exemption: Pros/Cons
by: Greggory B. Oberg, Esq.
Why Aren't You Using the HMDA Partial Exemption?
No seriously, this isn't a "click bait" headline, I'm asking a legitimate question. Why aren't you taking advantage of the EGRRCPA "partial exemption" from mid-2018? Eighteen months after the exemption passed, we're still seeing a fair percentage of clients who qualify for the partial exemption not taking advantage of the rollback. There are more than a few completely reasonable answers to this question, but I'm genuinely curious on pros and cons. Down the rabbit hole we go.
(if you're a HMDA reported eligible for the partial exemption, email me at firstname.lastname@example.org or call me, would love to hear thoughts!)
Little Bit of Background
Like many lenders I spoke to during the 2018 summer, I was a little sick and tired of HMDA. It just kept changing, and the CFPB had already signaled 2018 was warmer and fuzzier than before on the enforcement front via their December 2017 Christmas gift to industry. So when another change came along--one which undid the work we'd spent well over a year implementing, training, and testing our technology--it just didn't seem worth re-tooling the system again. The general consensus seemed to be "eh, my LOS already does it; and I'm not even sure I’ve got everything I must get right in order."
For a little while, I was firmly in that camp, and hadn't thought twice about it for quite a while….but last week something caught my attention I hadn't seen before. While I don't see nearly as many LARs (I always think of Lars Ulrich when I say that) as some of my colleagues--h/t Stephen Michael--I do see a fair amount in the course of advising on fair lending compliance management systems.
Understandably, I hadn't seen an exempt code on a LAR for 2018 data; or at least hadn't taken note of the fact. Basically the same in 2019. I felt this was anecdotally a decent sample to conclude many of our clients are not taking advantage of the partial exemption.
Finally, I saw one in the context of an independent fair lending risk assessment I was working on. I hit a wall on comparative evidence of disparate treatment in pricing, ALL THE GOOD STUFF WAS "1111."
This lead to a few logical questions that I couldn't help but address:
Does my hypothesis that New England community lenders in general tend not to use the partial exemption?
If so, what benefit to using the partial exemption? and
If so, what costs are associated with a shift to reporting Exempt for 26 data points.
Question #1: Usage of Partial Exemption - Nationwide
Simplest way to test seemed to be looking at aggregate data. Or at least that's how I'm writing it up, because I'm not admitting how many different avenues I ran down before I hit the logical one.
Per the CFPB Interpretive Rule implementing EGRRCPA, (creating the partial exemption, published September 7, 2018, 83 FR 45325) it was anticipated that approximately 3,300 institutions would be eligible for the partial exemption for data collected in 2018.
After a bit more digging, an August 30, 2019 news release by the CFPB accompanied the FFIEC's release of the 2018 data. In that news release (linked in resources at bottom of article, if I missed one you want email me), the FFIEC indicated that 2,251 institutions utilized the partial exemption for data collected in 2018.
In the interest of fairness--we're proving my hypothesis, after all--the Rule and FFIEC data differed on the number of reporters as well. While the FFIEC is clearly correct in fact, the CFPB's margin of error in estimating 2018 reporting institutions is relevant to understanding whether the 3,300 institution estimate is reasonable
CFPB estimated 5,852 reporting institutions in the 2018 Rule, compared to an actual number of 5,683; a difference of 169, or a little under 3% applying that error rate +/- to the estimated institutions eligible for the partial exemption, we expect 99 institutions up or down, so anywhere between 3,200 and 3,400.
Bottom line, best guess is between 66% and 70% of institutions who were eligible for the partial exemption took advantage of it. Region-specific data is a LOT of work, and I'm not sure that a) I can get it; and/or b) it would get us much more accurate estimates.
Benefits of Partial Exemption
I want to run through a few logical benefits for which I have not fully evaluated, and rely on my readers to correct misconceptions. But, as I see it, here are a few of the benefits:
Quicker Scrubs: you simply set XLS to filter data, go to the exempt rows, and if anything is NOT 1111, you messed up. That eliminates 26 fields, ~15 are critical if I remember my recent blog correctly
Remove "Fat Fingering", etc.: HMDA is hard for some of us without having to make calculations and track ups and downs in loan amount, qualified income, etc. Fields like DTI and CLTV--even when properly calculated in the first instance--are subject to typographical errors when transposing LAR data. I've seen it, it happens…shouldn't, but does. Talk to Paul Bates to improve your lending tech.
No More (or less, at least) Interpretation: many of these new fields have the dual flaw of being both (a) new, and thus little to no "industry knowledge" on regulator interpretation; and (b) are kinda complex and result from poorly worded regulations. If you're partially exempt, forget em.
But the real benefit here goes back to the pricing analysis I mentioned above. I couldn't do any data analysis on 1111 for interest rate, lender credits, etc. This got me thinking…if I can't, can the regulators?
First thing we have to figure out is if the data is even available. I think any regulator arguing that you are required to produce that data when partially exempt AND your systems are not configured to do so, should be asked to present his support for this theory.
But assuming it is, what happens? I've seen some LOS that collect all the data, but then a single click converts them to exempt. Stands to reason that the institution in that instance has the data. So the question becomes "can the regulator ask for it?"
The answer is obviously yes, good luck hiding a non-privileged document on a theory of "but I don't wanna." If you have the record, it's reasonable to assume they'll ask. And any institution seeking to stay on the regulators' good side best comply with all requests. Better, get out ahead of it, as discussed last week.
BUT-the possibility that they will not ask is the upside here. If that happens, one of the seven risk factors on the Interagency Guidelines Risk Assessment tool is potentially wiped out, or at least made much harder to address.
Just my opinion, but I'd have the full LAR ready (if you have it) for review, and discuss the findings of your internal fair lending analysis performed on the complete data. Get out ahead, cooperate, and self-identify; things will go better for everyone.
So We Should All Use it, Right?
Not so fast. I would say that, all else equal, go ahead. But in never is. Many of the same concerns that arguably caused ~30% of eligible lenders to eschew the partial exemption still exist today.
Regulatory Amendment Fatigue (RAT? I'm trying this out) is still a think, with multiple HMDA related rulemaking set to occur in the near future. Just as in 2018, the simple fact that you've got other priorities exists. The benefit here is arguably nil if you are in an excellent position for fair lending data…or if you aren't but you flag it when they walk in the door.
The cost is at least something. Retraining costs would be negligible for most, your vendors (I've probably trained half of your institutions on HMDA) will adapt their tailored training to the new reality. If you're training internally, the trainer shouldn't require too much time and effort to amend old materials.
On the policy front, it very well may be negligible as well. This depends on tech. If you have to take efforts to "scrub" the LAR of exempt data, maybe you just report it. If you called Paul Bates and got everything streamlined, it's more like a click of a button.
Biggest downside I see would only occur if the data visibility internally decreases. Those who need to understand pricing and fair lending trends should not lose any access to this data. Arguably, vendors should be provided the data as well; TruPoint or SCA Statistical Analysis aren't very useful when ~15% of the good data is non-existent.
On balance, economics seem to weigh against any investment in shifting...but this is based on what I"m seeing, and I'm not a banker.
…AND the Answer Is?
This is where I need you. What considerations am I missing, and what are you doing? Email me or call me, love to discuss.
With the info I know currently, it's pretty much a coin toss for me. My aspirational approach says "bring it." If my data isn't up to snuff, then my department isn't doing it's job. The realist in me says HMDA data re-submission is not something we want to do, and how can I not take a relatively low-cost control to eliminate the potential for error in 26 fields?