See this post on the CFPB's blog.
https://www.consumerfinance.gov/about-us...ing-guidelines/

Apparently a CFPB/OCC/NCUA joint effort. Our institution fall in the 501 - 100k entry line. Currently under FDIC examination the resubmission threshold is 8 out of 79 files (or 4 errors in same key data field)... this is reducing it to 4 out of 79... Has anyone heard if the FDIC will follow suit? (or FRB for that matter).

Good Grief.. more than double the number of fields per file and then reduce the margin for error by half. They might as well just send out hand grenades and make you pull the pin out and pop it back in 100 times without it blowing up.... probably a higher chance for success.. (and less stress)

From the ABA this morning:

Agency Guidelines on HMDA Data Testing Fall Harder on Smaller Lenders
The federal banking agencies yesterday issued guidelines for how examiners will test the accuracy of data collected and reported by financial institutions under the Home Mortgage Disclosure Act.

Despite coming after concerns expressed by ABA and others about the burdens imposed by unreasonable error tolerances that require a bank to resubmit its HMDA data -- in light of the vastly expanded data fields that must be reported beginning in March 2019 -- the new guidelines are expected to have the opposite result, creating disproportionate expectations for smaller volume lenders. For example, an examiner will review 30 loan files of a bank that only makes 50 mortgage loans but only 159 files for a bank that originates 100,000 loans. Moreover, a small number of errors in any given data field will trigger review and resubmission.

In comments filed to the proposed guidelines, ABA called for more reasonable tolerances. The association intends to ask the congressional banking committees to review the new guidelines and consider whether the level of perfection required will undermine the ability of banks to serve their customers. Read more. For more information, contact ABA’s Rob Rowe.

From the article linked above:

...hand me one of those grenades....

seriously... banks are going to need HMDA ARMIES just to keep up...

Id like to see a study on increased rates of alcoholism in compliance professionals over the last three years hahaha

With these guidelines being released does anyone believe that the HMDA rules will get delayed like some are hoping for? Or is this the final nail in the coffin to that actually occuring?

I didn't think it would happen to begin with, but now I am even more certain that it won't. (Doesn't mean I can't still hope for it though.)

This was the CFPB's doing.
Delaying implementation will be a POTUS and/or congressional move.
I'm still hoping....
So keep those cards and letters and sacrifices to the gods going!

That's got to be the resolve of the industry to squash this rogue agency. The current admin needs to rid itself of the holdovers from the prior admin, especially this bureau director that continues to thumb his nose at "elected officials".

The ABA's letter on these guidelines was dated March 2016. Yet, this agency continues on its path to push an agenda where there
appears to be no just cause other than to burden the industry with expanded data collection. Why? What value added does this give the public? In 37 years of reporting HMDA Data I have yet to hear from a single person or group that they benefited from any such information. This is nothing short of regulatory "gotcha" just another form of "choke point" by a rogue agency.

Our small CB reported less 200 HMDA Lars for the past 3 years. Has the bureau considered an exemption for small creditors?

Just got the FIL from the FDIC. Oh joy.

I used to like HMDA but it's turning in to nearly a life and death situation to get things right.

^^^this.
Now, ccman, let's be fair. CFPB is Dodd/Frank not the former Potus'.
I don't think that benefitting the public,was ever the driving force once you got past stopping redlining.
As for 2018 HMDA, that is simply a government guise for mining private information that there is absolutely no valid reason to collect.
yea...I probably am very close to joining the tin foil hat league.

So the only good thing, I think (?), is that it took out FILE error rate, and just left the FIELD error rate (for the resubmission threshold). Is that what you all understand as well?

I agree that we (falling into the 501-100K LAR count) will have a hard time if our initial sample threshold is only 2 of one type of field errors.

I am going to go ahead and update our Analysis review spreadsheet (we present this to management once a month with our thresholds showing whether we would have had to have resubmit or not).

Banker K, are you saying that it's your understanding that if 9% of the sample size has errors, but no one field has an error rate over say 5%, then no resubmission would be required?

I've only scanned over it...but that would be better than a 6.4 or 5.4 percent related to any field error.

Does anyone know why the Resubmission Threshold column in the table contains both a specific number and a percentage? For example, take the line for a LAR count of 191-500. 5.1% of that size would be 10-26 loans, which is a lot more than the "3" shown in Column D. So is the resubmission threshold the specific number or the percentage?

The % is the resubmission threshold divided by the sample size.

Ah, thank you!

Thank you Sinatra Fan and ahou! Thought I was the only one trying to figure those percentages out! Read through all of the examples and they only used the numeric value--no mention of percentages. Could have cut that portion of the column out and avoided confusion. My greatest concern is the examiner discretion afforded by item #4 (page 2). The procedures make it much easier on the examiners, but much tougher on filers.

They do give a little tolerance in some minor areas, but also give examiners discretion that is troubling. “Examiners may direct a financial institution to correct the Action Taken data field and resubmit the HMDA LAR even if the number of Action Taken errors found in the Total Sample does not equal or exceed the Resubmission Threshold in Column D in the HMDA Table.” [emphasis mine]

No matter how many times you scrub the data, I don't know if you'll ever feel confident the data is correct.

They could have saved a bunch of paper and simply said, "We demand absolute perfection!"

Just wanted to follow up with everyone now that this has had time to sink in.

Based on the exam guidelines, the resubmission and additional sampling thresholds are errors in the same (presumably) key data field. For instance if in the initial sample, you had two errors on Action Date it would trigger the additional sample, and if an additional error in an Action Date field (based on LAR size) was identified, you would resubmit.

Based on what I read, you could technically according to the "guidance" have 10 errors all in different fields in the initial 35 file sample. The examiner is not required to pull an additional sample at that point, but may pull an additional sample or require refiling at their discretion if the data is considered too inaccurate to rely on.

Is this how everyone else is reading it? Essentially, resubmission thresholds are based on errors in same key data field only (with examiner discretion otherwise)?