I dont' know if this question has been asked already (did a search and couldn't find much). But my question is if the Federal Reserve has STATED tolerance levels when they do a review of a bank's HMDA-LAR. In other words, when we conduct our internal audit of HMDA data, we expect our data to be within a 3% error rate for all of the HMDA reported variables. I know this has to be a very low tolerance level (this is just the way we've done it for the longest time so I'm not sure if there is a Fed mandated benchmark to compare ourselves to). If the Fed sees an error rate of 5%..10%...15%...at what point do they say, this data needs to be redone? Also, what does the Fed say when an error rate is between 5-10%. Good, bad, needs improvement...anyone willing to share their thoughts on this???

From one Anon. to another....

I think this first year after the new HMDA data elements the examiners will cut banks a little slack and the 5-7% range should be ok. But I believe each year going forward will provide more drama. Banks will be pounded (i.e.e, scrub all files) to get this data right, however difficult and $$$ that may be.
In future years I would be happy if our bank could get to the 3% error level. Let's face it, nobody and I mean nobody will ever be a 0%. Any bank that says they are, I'd like to test. Give me the original loan file and your LAR and I'll get you to 5% within a day. The rules are just too convoluted........................

Thanks for the info....but where does this 5%-7% number come from? If they are willing to accept a range of 5-7% on the new data, what has been the tolerance levels in the past? 5%? Are these written in guidance anywhere? Even for past HMDA data (not including the new reporting requirements) Or is this just your experience?

The OCC published information on this before the changes and looked at 5%. But that doesn't mean you won't be cited. It typically triggers a refiling and complete scrub since more errors means the data is not reliable. It won't be of sufficient quality to use for fair lending or CRA, as examples.

I think you'd be hard pressed to get an official tolerance level stated by any agency today. If the rest of your compliance program is poor or if HMDA has been poor, they may be less tolerant. So I think there are external factors, and add in different agencies and regions and many banks could see differences with the same error rates. This is just an opinion.

The best advise is, if you think you have a problem, check it out. If you find a problem, correct it. It is to late to refile for this year, but you'd still want clean data if it will be used elsewhere.

We just included a summary of the HMDA tolerances in our latest E-newsletter. You can get it at our website (scroll down to the bottom to see the Newsletters). Click on the "June 2005" link.

What a great newsletter David! That information was perfect of course Do you mind if I link...or would you, that newsletter to a similar question on resubmission in the CRA threads? It's spot on, and saves me hours of typing LOL

7% is frightening...careful anon...you could be gunning for trouble. I set a limit of 3% at the MOST overall. And, you could go through my files and you wouldn't be able to make legit claims I was over about 1 1/2 to 2 % at the highest. Differences of opinion in intrepretation do not lead to error reports, only "wrong" (meaning you didn't justify the difference of opinion when asked about it or you're just flat wrong) reporting leads to an error ratio.

Thanks. You're a prince of a guy and pretty HMDA smart too

Wow, thanks David and everyone else. This is great information. Exactly what I was looking for. Thanks again for all your help. I realy appreciate it!!!

It's been 4 years since I was a Fed examiner, so I don't know if tolerances have changed. But it was 5% for the overall LAR and 10% for critical fields before we required the data to be resubmitted.

Anon examiner...That level must fall under the "if they're in a bad/good mood" section of the tollerances. OCC quotes the same levels as David posted in his newsletter.