Q&A with Bob Avery, Creator of the HMDA Lender File

The Philadelphia Fed’s Consumer Finance Institute (CFI) recently began hosting the HMDA Lender File, a data set that can be used in conjunction with publicly available HMDA filing data to better understand the characteristics of firms receiving mortgage applications and originating loans. The data set enables users to connect institutions across subsidiaries and over time, going back to 1990.

Bob Avery, CFI visiting scholar and creator of the file, shared background on the history of the file, how he creates it, and what value it can provide to HMDA data users.

Q: When did you start creating the HMDA lender file, and what research or policy need drove you to make it?

BA: The lender file was initially created in the mid-1990s, when I was part of a group of economists at the Federal Reserve Board of Governors tasked with facilitating HMDA data processing and producing a public report accompanying each year’s data release. We produced the file for our own purposes and to help correct errors in the transmittal sheet (TS). We started releasing it publicly when we realized that other users would have some of the same needs we had. Because it was our best guess as to the identity of the HMDA reporters, this was not an “official” file but rather a “research” file, with the typical caveats.

Q: Tell us, in a nutshell, how you create the HMDA Lender File and what it contains.

BA: Each year, the 5,000 or so lenders that file loan-level mortgage data in compliance with the annual HMDA statute also file a TS, which provides information about who they are. The information supplied is “"bare-bones" — institution name, tax ID, legal entity identifier (LEI), city, state, and primary regulator (e.g., NCUA, OCC, nonbank). Often this information is not sufficient to fully identify the reporter and is, in some cases, wrong or missing. The TS is a regulatory filing and thus cannot be edited or changed by the regulators.

Our lender file is an attempt to expand on the TS. We match the information in the TS to external sources to fully identify who the reporter is and what organizations they are members of and to provide a limited amount of financial information. We use multiple sources to do this, starting with regulatory data collected as part of the National Information Center (NIC), Credit Union and Bank Call reports, Federal Home Loan Bank membership lists, and the lender’s public website. The specific fields that we report and their sources are detailed in our documentation file.

The file contains information on each lender for each filing year from 1990 through 2024. Because the data collection process changed significantly in 2018, we have two similar but slightly different files, 1990–2017 and 2018–2024.

Q: How has the HMDA Lender File changed over time?

BA: There were major revisions to the HMDA data collection in 2004 and 2018 that impacted the reporting process. For example, prior to 2004, manufactured housing, second-lien, and subprime loans were reported but not explicitly identified. Consequently, prior to that, lists of lenders specializing in these loans were created by regulators and flagged on our lender file. After 2004, that was no longer necessary. Starting in 2018, LEI and tax ID were included on the TS file (and on our file as well). That year also represented a significant expansion of the HMDA reporters to include lenders specializing in business, reverse annuity, and open-ended loans.

The most important change in the file, though, is its accuracy. Prior to the mid-2000s, the NIC system did not have comprehensive coverage of mortgage bank subsidiaries of banking institutions, savings and loan holding companies, or independent mortgage companies. Coverage was vastly improved after the Great Recession, such that almost every HMDA reporter was in the NIC system, making creation of our file much easier.

Q: What do you see as some of the key insights HMDA data users can glean from the HMDA Lender File?

BA: Perhaps the most important value of the file is that it allows users to identify connections among the filers. For example, all members of a common organization, or affiliates in a joint venture. It also allows users to correctly identify the legal status of a filer. For example, is it a depository institution? Is it covered by the Community Reinvestment Act? Is it an international institution? Is it a bank or nonbank? Does it have a parent?

These linkages will allow users to examine behavioral differences among broad groups of lenders and to potentially test the efficacy of legal constraints and laws.

Q: Tell us about your choice to make the HMDA Lender File broadly available to the public.

BA: The purpose of the HMDA legislation was to provide the public with information to evaluate mortgage loan providers, but in some cases, it is hard for the public to easily identify who the lender is. This is particularly the case when a lender is part of a larger organization but the name alone doesn’t identify it as such. This file is an attempt to make the identification easier in alignment with the spirit of the original HMDA law. HMDA data are available to anyone. It is important that our lender file be as well.

The lender file only contains information that is already in the public domain, although not necessarily linked to the lender. This is important because it is a public, not a regulatory, release.