In reviewing the recently published HMDA aggregate data, I noticed that there are several records in each of our assessment area MSAs that have state codes that do not correspond to the applicable MSA. For example, for the Fort Collins MSA 22660 (see below), under the State Code GA, FL, and IN are listed when CO is the correct code for this MSA. The county and census tracts correspond correctly to CO and the MSA. I looked up several of the LEI numbers and they appear to be non-depository lenders (mortgage lending companies) and credit unions. I haven't been able to find information regarding this practice of reporting a different state code not within the MSA due to these entity types. I'm working on comparing our loan data with aggregate lenders within our MSAs in CO. I'm unsure if it would be accurate to consider these lenders when we compare our bank to our peers within the applicable MSAs. Thank you.

The example below is the correct Fort Collins, CO MSA and other than state code the other fields are correct.
derived_msa-md state_code county_code census_tract
22660 GA 8069 8069001603
22660 FL 8069 8069002600
22660 IN 8069 8069001305

I think everyone is hoping that the data gets better. My understanding is that it is currently a mess.

It's important to note that CFPB has said that the data will be updated as needed after resubmissions, so the data you're looking at now won't necessarily be the same as the data your examiners will be using in the future.

I just took a look at the Fort Collins MSA data, and ... oh boy. It does look like the biggest issue is with one mtg company, and ALL of the apps they actually took themselves have the problem. The large majority of their LAR entries in that MSA are for purchased loans, and those correctly list CO. I would not include them.

I don't think anyone really knows what to do with the data right now.

I received an email from the CFPB in response to my earlier question. See below:

"The mismatch in data comes from a quality edit, Q604: The reported State and County are not a valid combination. If neither State nor County were reported NA, then the County must be located within the State. Because this was a quality edit, filers were able to verify that their data was accurate (meaning a mismatch in geography data could pass). For future years, this edit has become a validity edit, meaning the data has to match in order to continue the submission process. In the case stated, please assume the state in question is CO based off the other data fields'.

I hope this helps everyone when reviewing aggregate data.