To better assess whether financial institutions meet the lending needs of communities, the Consumer Financial Protection Bureau (CFPB) changed rules governing data collected under the Home Mortgage Disclosure Act (HMDA) in 2015. For the first time in September 2019, the modified and expanded data were made public.
The new HMDA Data Browser hosts the 48 data points collected under HMDA for 2018 and beyond. Due to a change to the identification process, all HMDA data collected before the regulatory change (data preceding 2017) are hosted separately.
Woodstock Institute built a file to allow our team to easily compare respondent data across the two different identification systems to conduct a longitudinal study of mortgage lending in the Chicagoland area. Now, we are sharing that code file with you.
Congress passed the Home Mortgage Disclosure Act (HMDA) in 1975 to provide the public with critical information on mortgage lending trends by financial institutions. HMDA data gave the public the ability to determine whether banks were serving the credit needs of their communities, and to flag discriminatory lending known as “redlining.” Insights into lending patterns can be determined based on variables such as loan type, the income of the borrower, the loan action taken, or the race of the borrower.
Until 2010, when Congress enacted the Dodd-Frank Act, the Consumer Financial Protection Bureau (CFPB) and the Federal Financial Institutions Examination Council (FFIEC) maintained joint control over the publication of HMDA data. While the data was primarily hosted on the FFIEC website, the CFPB laudably developed a tool called HMDA Explorer, which allowed users to navigate and analyze national mortgage loan data in-browser and to create easy-to-understand tables summarizing specific data points so that non-academic users could obtain loan information without the use of sophisticated or high-performing systems to download and process such large datasets.
Changes to HMDA Data Points
The new rules (passed in 2015, made available in 2019 for all data collected on or after January 1, 2018) broaden the scale of information financial institutions must report under HMDA. Some of the data points were modified, others were new. All-in-all, 25 new HMDA data points are now collected, 14 were modified, and 9 remain mostly unchanged.
New data points include:
- Age of borrower
- Credit score
- Interest rate
- Loan term
- Property value
- Debt-to-income ratio
Modified data points include:
- Ethnicity of borrower
- Race of borrower
- Loan amount
- Reason for denial
- Loan purpose
- Sex of borrower
What’s New in the HMDA Data Browser
Here’s what the Data Browser looks like.
Woodstock Institute is concerned about the removal of the functionality to do year-by-year comparisons of data in the transition from Data Explorer to the Data Browser. Why? Users looking to analyze HMDA data across the years must download datasets for each individual year. In addition to the issue of computer processing capacity for the non-academic user, this is problematic on another level: the 2018 modifications to the HMDA variables also came a change in the way that financial institutions were identified within each row of data.
Pre-2018 HMDA data used a variable called “Respondent ID,” a numerical identifier, to indicate which financial institution was associated with a mortgage loan application. The 2018 HMDA data replaced Respondent IDs with Legal Entity Identifiers (LEIs), throwing a wrench into the process of comparing the 2018 dataset with pre-2018 HMDA data. For non-technical users, it has therefore become quite difficult to determine trends by financial institution across the years.
Commendably, the CFPB has provided publicly accessible resources for mapping LEIs to Respondent IDs and the names of their respective financial institutions. These resources can be accessed via a link on the home page of the HMDA data under Snapshot National Loan-Level Dataset. Under 2018, users can download a csv file under the category of “Reporter Panel.”
However, the CFPB’s resources are unfortunately buried under technical terminology. This language can confuse even the most astute academic users.
Woodstock Institute’s Crosswalk File
To make this translation process as painless as possible for all users of HMDA data, Woodstock Institute has produced a simplified translation file (aka, a crosswalk file) that allows users to map an LEI to its matching Respondent ID and to the name of the financial institution for IL data.
To compare lending data across years by financial institution, users need to download the HMDA dataset for each year that the user would like to analyze. The Respondent Name column (Column B) will provide the official name of the financial institution for each Respondent ID in all pre-2018 HMDA datasets. For the 2018 HMDA data and all HMDA data going forward, the Respondent Name column provides the financial institution names for all LEIs in the dataset. This crosswalk file, of course, also allows users to match LEI (Column A) to the corresponding Respondent ID (Column C).
For any additional questions about how to use this crosswalk file, or about any of our data analysis, contact Janice Guzon at (312)368-0310 or email@example.com.