Initially published in the Sept/Oct issue of ABA Bank Compliance magazine.
Day in and day out, you gather HMDA data from multiple sources, review edits, correct errors, double check transactions, deal with system limitations and challenges, and monitor your data quality. As your Loan Application Register grows, it sometimes gets away from you, which usually results in sleepless nights or nightmares about data and edit codes. As the submission deadline creeps up, you scrub your data and second guess your efforts, usually at 3:00 a.m. You close your eyes, cross your fingers, and push the button. HMDA is submitted! Whew! After you strongly consider running away, you decompress and ease back in. While the challenges are fresh in your mind, here are a few things you can do to gain a few nights’ sleep while improving data quality and monitoring for fair lending risk.
Establish a first line of defense data validation process. At times, compliance professionals hear that HMDA is “Compliance’s job.” Well, compliance professionals, especially those who wear a CRA or fair lending hat, use the data. They are not collecting or reporting the data. Your business lines gather, enter, use, and report various data throughout the life cycle of an application. In reality, they own the data that forms the basis for HMDA reporting.
Ensure that your business lines understand this approach, because if HMDA data is inaccurate, other non-HMDA data points may also be inaccurate. Additionally, the business lines possess the documentation that supports the HMDA data reporting. This documentation is key to ensuring that all reported data elements are accurate, but how can a financial institution be sure? The business lines should ensure the accuracy of their data and establish a validation process that entails the following elements and processes:
- Ensure your institution has data collection and transaction coverage procedures for each loan operating system (LOS). These procedures should assist the business lines with how to collect and report the data based on the action taken and specific product features.
- Establish a workflow that identifies covered transactions, collects the HMDA data throughout the application process, and triggers the application data for reporting. HMDA collection should be seamless throughout the application process.
- Periodically import your HMDA data from the LOS to your reporting system. Proactively distribute edit reports to the business lines with guidance for how to clean up validity, syntactical, or quality issues noted and correct for any missing or inaccurate data.Request that they update the LOS for the next import.
- Require business lines with reporting responsibilities to perform their own transaction coverage review. Ensure all covered transactions are reporting. For any dwelling-secured application or loan that isn’t deemed a covered transaction, track it for future reference.This would include commercial transactions that do not meet the purpose test, temporary financing, or transactions in which your institution did not make the credit decision.
- Require business lines to perform a validation review on a representative sample, comparing each reporting element to the relevant source documentation. Request that they update the LOS for the next import.
Take time to learn the HMDA pain points. Meet with the business lines and ask what they struggle with the most. Is it transaction coverage or LOS challenges? Do your business lines need more training? Which data fields have the highest error rate? Is there a specific business line or product that has the most errors? Review your compliance monitoring reports for inaccurate data fields. Perform a root cause analysis and create an action plan. Consider the following:
- Create alerts or require certain fields to be populated upon action taken. Consider stopping the transaction from going to the next milestone until all HMDA fields are entered.
- Create defaults in your LOS for fields that are always the same. If you do not offer reverse mortgages, default to code “2” and remove user edit capability.
- Work closely with your LOS vendor to ensure they are accurately calculating and collecting the data.
- Review the FFIEC’s Filing Instructions Guide for HMDA Data Collected (FIG) to understand the reporting expectations for the current reporting year. Get ahead of potential issues before your next submission. The FIG provides file, data, and edit specifications for the reporting year.
- Update your procedures, checklists, and job aids and continue to train the business lines.HMDA submissions are never a “one and done” matter.
- Consider leveraging other data fields in your LOS to review HMDA data fields for accuracy. For example, compare the Closing Disclosure’s issue date with the HMDA action taken date for originated transactions. If the Closing Disclosures issue date is after the action taken date, you need to ensure any updated Total Closing Costs, Origination Fees, Discount Points and Lender Credits are accurately reporting.
- If you use a technology like CRA Wiz, leverage User Defined Edits. These edits are invaluable by isolating data that may not have a validity or quality edit but is still inaccurate.For example, if your institution offers a fixed rate HELOC product that also has an interest-only feature and a three-year pre-payment penalty, you can create a user defined edit to isolate any transaction that doesn’t have any one of those fields accurately reporting.
Are you eligible for a partial exemption? Whew! Well, not so fast. While you may not be required to report certain data fields, consider collecting and validating the data. Should you no longer qualify for the partial exemption, you would be ready for the increased collection and reporting. Take advantage of the partial exemption by leveraging that additional data for your own internal fair lending analysis and address any risk as well.
Monitor the Not Provided Demographic Information Data Collection. Establish a monitoring threshold for race, ethnicity and gender that report as “not provided.” Determine the level of “not provided” activity globally and by every lender. Also consider which application method results in the highest “not provided” response rate. Assess why the threshold has been exceeded or which application method results in higher “not provided” responses. It is possible your lenders are not asking for the information as required, or that your application collection process needs some work.
If telephone applications have a higher percentage of not provided responses, it is possible that your lender is not providing the disclosure and asking for the demographic information. Lenders have reported that this is the most uncomfortable part of interviewing an applicant for a loan. Enhance the training and monitoring regarding demographic information collection.
Evaluate applications received via USPS mail, email, or dropped off for completeness of information provided on the disclosure and collection form. Was a collection form provided to the applicant? If not, the applicant wasn’t given the opportunity to provide the government monitoring information as required. Review the application forms in use to ensure that the disclosure and collection form is evident as explained in Regulation B, 12 CFR 1002.13, and illustrated in Appendix B to the regulation.
If your application has multiple pages, incorporate the disclosure and collection form as part of the main page and not as a separate page or attachment. Ensure your online data collection process is presented in a way that encourages the applicant to provide any information they choose. If it is buried in a link or is too cumbersome to complete, the applicant may not bother. Also, consider programming the system to require a response from the applicant before they can submit the application. The applicant can still choose to not provide the information, but at least they have been given the opportunity.
Trend your fair lending risks and respond timely. Now that you have worked on some solutions for improving your data quality, consider trending the denial rates and pricing disparities over a rolling period. Create quarterly or monthly focal point reports with denial rates and pricing disparities by each protected class. Do this for all of your primary products and ensure you have a historical look back (say 2018 and 2019). Create a trending report that tracks the denial ratios and pricing disparities that are statistically significant and/or over the benchmark.
Update the trending report periodically to watch for increased risk by a protected class. Further, trend by NMLS ID, branch, region, MSA, and so on. Do what you can to isolate the areas that may be increasing the fair lending risk to your institution. If you have a protected class that “pops,” do further analysis. Remember, HMDA data is available to the public. Your examiner, competitors and the public can gather and analyze your data at any time. You need to know the story behind the data and if you have an issue, you must respond.