As published in ABA Bank Compliance (now ABA Risk and Compliance), July/August 2021
There is no better time to perfect your Home Mortgage Disclosure Act (HMDA) compliance program. The Biden administration, along with acting Consumer Financial Protection Bureau (CFPB) Director Uejio and CFPB Director nominee, Rohit Chopra, have all indicated the priorities of the CFPB will include regulation and enforcement of fair lending and racial equity issues. Since HMDA data is critical to analyzing fair lending, lenders should be prepared for additional scrutiny and emphasis on the accuracy of HMDA data.
Implementing a strong, comprehensive HMDA compliance program will help ensure the accuracy of your HMDA data, and in turn, should ensure the information gleaned from the data portrays an accurate picture of your fair lending performance. Building a process that is sustainable and ensures data integrity continuously throughout the year requires a strong foundation. The foundation, or core pillars, of an effective HMDA compliance program include:
- Board and management oversight;
- Policies and procedures;
- Training; and
- Monitoring and reporting.
The First Pillar – Board and Management Oversight
Similar to an institution’s overall compliance management system, the HMDA compliance program requires diligent board and management oversight. The board is ultimately responsible for administering an effective program that ensures accurate HMDA data and must be kept apprised of the institution’s efforts to comply with HMDA and fair lending requirements. In addition to being confident that HMDA data is accurate, the board should understand the institution’s story—what the HMDA data reveals about the institution’s lending patterns and fair lending performance.
The Second Pillar – Policies and Procedures
Robust policies and procedures are the second pillar of a strong HMDA compliance program and should provide the guiding principles and detailed guidance for the institution. Procedures should provide direction on how to perform day-to-day HMDA related activities, starting with defining a HMDA reportable application all the way through the Loan Application Register (LAR) submission process. Procedures should be tailored to individual roles within the institution including loan originators, processors, and compliance personnel. They should also address the different loan products offered, application channels, and application outcomes. While HMDA-related procedures should be comprehensive and much broader than what can be covered here, below are a few basic topics that, if addressed in HMDA procedures, can help avoid common HMDA reporting errors.
HMDA source document. Develop a document that maps the source of the information used to populate each HMDA data point (e.g., application date, loan type, loan purpose) reported on the LAR. Tailor the document for different loan products, different application channels and different decision outcomes.
For example, one data point is the application date. The source document would list where the application date is found in the institution’s system and/or loan file for the various types of applications—online, in-person, and applications received by commercial lenders, if applicable. The source document can be a listing of the data points in one column and the corresponding source of the data in a second column. A simple way to get started creating a source document is to use the CFPB’s Regulatory Reporting and Overview Reference Chart that describes each of the HMDA data points and expanding it to include the source of each data point specific to the institution’s operations.
A comprehensive HMDA source document can also help to demonstrate consistency, which is very important in HMDA reporting. It is also a useful training tool to ensure consistency and understanding across the institution.
Preapproval program: If the institution offers a qualified preapproval program as defined by HMDA, procedures should document (and personnel should understand) what distinguishes a qualified preapproval request from other requests. With the limited supply of houses on the market, homebuyers, home sellers, and realtors are expecting a prequalification or preapproval letter to help a potential buyer’s chances of getting an accepted offer. Understanding whether the institution offers a qualified preapproval program is essential.
A preapproval program for purposes of HMDA reporting requires the financial institution to conduct a comprehensive analysis of the applicant’s creditworthiness, including income verification, and then issue a written commitment for a home purchase loan. The commitment must be valid for a designated time and up to a specified amount, and can be subject only to limited conditions including finding a suitable property and that no material changes in credit worthiness or financial condition occur. There are a few other limited conditions allowed that are not related to financial condition or creditworthiness. (Refer to 12 CFR 1003.2(b)(2); Comment 2(b)-3, for more information.)
If the institution’s preapproval program (or equivalent program by any other name) does not meet these specific requirements, then it is not a qualified preapproval program for HMDA purposes and requests should not be reported as such on the LAR. Instead, the institution must determine if the request is a HMDA reportable application, or a prequalification request or inquiry that is not reportable, as further discussed below.
Application or Inquiry. HMDA defines an application as “an oral or written request for a covered loan that is made in accordance with procedures used by the financial institution for the type of credit requested.” Unlike the Real Estate Settlement Procedures Act (RESPA), which specifies six pieces of information that, once obtained, automatically constitutes an application for RESPA purposes, HMDA allows each financial institution discretion in defining an application based on its products and procedures. Institutions should be cautious about defining an application too narrowly, as that could result in inappropriately omitting applications from the LAR.
When determining when an inquiry becomes an application, institutions should consider the customer’s perspective and expectations. Does the customer believe they applied for a loan and are they expecting to receive a Loan Estimate (LE) to use for comparison shopping? Does the online application process clearly indicate whether the customer is submitting “an application” or merely an inquiry about the amount of a loan for which they may qualify? Also consider how to address inquiries that result in adverse action and whether those represent HMDA reportable applications. Procedures should provide clear guidance to ensure inquiries and applications are defined, identified, and reported consistently and accurately.
Non-originated loans: A frequent source of HMDA reporting errors are applications that do not result in an originated loan. Nuances between denied, withdrawn, approved but not accepted, or incomplete applications are often reported incorrectly, and detailed procedures addressing these nuances can minimize these errors. Procedures should clarify that applications are only reported as withdrawn when they are expressly withdrawn by the applicant before the financial institution makes a credit decision or before the applicant satisfies underwriting or creditworthiness conditions if a conditional approval is provided. The treatment of counteroffers, which generally should be reported as denied applications unless accepted by the applicant, should also be detailed.
The Third Pillar – Training
Personnel involved in the HMDA process should receive periodic training on the technical requirements of collecting HMDA data as well as the institution’s specific procedures. Training should be specific to key roles in the HMDA process including loan originators, processors, and compliance personnel. Be sure to train commercial loan originators who make HMDA reportable loans secured by single and multi-family residential dwellings. While senior management and the board may not require detailed HMDA training, they should understand the importance of accurate HMDA data and what the data reveals about the institution’s lending. Various sources of training are available including through industry groups, online training services, conferences, and webinars. Training can also include internal staff meetings and team training sessions.
The Fourth Pillar – Monitoring and Reporting
The fourth pillar of an effective HMDA compliance program is periodic monitoring of the accuracy of the HMDA data, and reporting of the findings. Regulators expect the process to be ongoing and sustainable throughout the year, not just a lastminute scrub of the data prior to submitting it. The monitoring process should be documented and include sufficient transactional testing to ensure compliance with regulatory and procedural requirements. When establishing your monitoring process, consider the following:
Lending Summary. A good starting point for monitoring HMDA data is creating and assessing a lending summary. A lending summary can provide insights into the accuracy of your data and includes categorizing the data on the LAR for reasonableness, and variances from what was expected. Some items to assess could include:
- Does the total number of loans for the period reviewed make sense compared to the prior month, (or prior quarter or prior year) based on volumes at the institution and in the market as a whole?
- How many preapprovals, approvals, withdrawn, denied, incomplete applications are reported, and do they align with expectations?
- When sorted by monitoring information, such as race and sex, does it reflect the diversity of your market area?
There are numerous ways to categorize and assess your data to determine whether the data aligns with expected results. Investigate and understand unexpected outcomes.
Reconciliation/Omissions Testing: To avoid overlooking HMDA applications from the LAR, periodically reconcile the LAR data to loan origination system (LOS) data for the same time period. This is much easier if HMDA reportable applications are clearly defined and distinguished from inquiries or leads as discussed above. Some differences between the LOS and LAR are expected, including customer inquiries that are recorded on the LOS but are not HMDA reportable, and applications that are still in process. Other differences may represent HMDA errors. Records on the LOS that are not on the LAR may be applications that were incorrectly omitted from the LAR. Records included on the LAR but not on the LOS for the same time-period could be applications that were decisioned in a prior year that should have been reported on the prior year’s LAR. If you maintain HMDA reportable applications on more than one LOS (such as home equity lines of credit or commercial loans) be sure to reconcile the data from each applicable system.
Transaction Testing. Transaction testing should be performed to identify potential weaknesses in HMDA related procedures and errors in the HMDA data. The scope and type of transaction testing should be based on loan volumes, and ideally should include both manual file reviews and systemic data integrity testing.
Manual File Review: Manual file reviews are completed by comparing the information on the LAR to the documentation in the LOS or loan file, such as the application form, transmittal summary, adverse action notice, underwriter or processor notes or logs, etc. The source document described above can be very helpful when completing manual reviews as it should outline the source of information to be used to consistently validate each item on the LAR.
Sample sizes to be tested can be based on a statistical sampling methodology, and easily determined using a statistical sample calculator found on the internet. Alternatively, sample sizes can be based on the Federal Financial Institutions Examination Council (FFIEC) HMDA Examiner Transaction Testing Guidelines which outline the number of transactions the regulatory examiners will test when validating HMDA data, along with the corresponding error rates that, if exceeded, may require correction and resubmission of the data. According to the FFIEC’s guidelines, the sample of items tested is based on the number of items on the LAR. For example, based on the FFIEC’s table above, the total sample size for a LAR with 1,000 applications is 79.
Although sample testing can never ensure 100 percent accuracy, sufficient transaction testing should provide good insights into the accuracy of your data and identify errors that need to be corrected. Manual reviews may also identify potential concerns that may not be detected during a systemic review.
For example, a manual review would include underwriter or processor logs that would not be assessed in a systemic review, and these logs sometimes include information that contradicts the system data. For example, a note in an underwriter’s log may reveal that an application reported as withdrawn was actually denied and should be reported as a denied application for HMDA purposes.
No matter what method is used for selecting the sample for manual HMDA monitoring, ensure the sample is representative of your LAR and application activity. Consider items such as products offered, application channels, and decision outcomes. Further consideration can be given to locations, branches, or even loan originators if appropriate.
For example, if 80 percent of the applications on the LAR are refinances, then about 80 percent of the monitoring sample should be refinances. If 10 percent of the applications on the LAR are withdrawn, then about 10 percent of the monitoring sample should be withdrawn applications. Selecting a sample that is representative of the application activity will improve the effectiveness of monitoring and yield better results.
Systemic Data Integrity Testing: Systemic data integrity testing assesses 100 percent of the LAR data and typically consists of a series of tests including validity and logic testing, and omissions testing. Depending on the size of the institution’s operations and LAR, third-party vendors can be used to perform systemic data integrity reviews using sophisticated software tools, or procedures can be developed internally using common spreadsheet software to do some basic analysis. Logic testing consists of a series of tests developed to compare HMDA data fields to other data on the LAR and LOS for accuracy and reasonableness. For example, the application date reported on the LAR could be compared to the LOS fields for system file started date, the intent to proceed date, and the LE disclosure date to determine if the dates occur in a logical order. Logically, the application date would be on or after the system file started date, on or before the LE disclosure date, and on or before the intent to proceed date. If any of these dates are not in logical order, additional file review would be warranted. There are numerous creative logic comparisons that can be developed to systemically validate LAR data.
Outliers and Red Flags. Reviewing the HMDA data for outliers or red flags may also identify data integrity issues when comparing your data to other internal lending information, the national aggregate or peer data. Outliers could include items like debt-to-income ratios that are too high or too low compared to your standard underwriting criteria, or income amounts or loan amounts that are too high or too low based on the nature of your business. Significant differences in the institution’s application withdrawal rates or denial rates compared to peer data could also be indicative of HMDA data integrity issues.
Pulling it All Together: Correction and Remediation Efforts
An important part of any monitoring process is remediating any identified issues or errors. Individual errors should be corrected, but more importantly, the root cause(s) of errors should be identified and remediated to ensure errors do not reoccur. Root cause analysis may identify other issues with the HMDA compliance program that should be addressed including a need for additional training, stronger procedures, or operational changes. (For more information on root cause analysis, see Root Cause Analysis and Remediation: Putting Findings into Practice in the March–April issue of ABA Bank Compliance on page 28.) It is also important to report results and issues to the Board, in order to support their role in oversight, the first pillar.
Your Fair Lending Performance: What Story Does Your HMDA Data Tell?
While HMDA is a data collection regulation and focuses on accurate reporting of HMDA data, the data tells a story about your fair lending efforts. One of the main purposes of HMDA is to provide the public with loan data that can assist in identifying possible discriminatory lending patterns and enforcing anti-discrimination statutes. (For more information, see Leaning Against Racial Injustice: Your Fair and Responsible Banking Program and Beyond in the January–February issue of ABA Bank Compliance, on page 10.) While it is important to ensure HMDA data is accurate and the compliance program is effective, it is equally important (or perhaps even more important) to understand what the data says about the financial institution’s fair lending performance. With the Biden administration’s stated focus on racial equities, boards and key executives should waste no time in understanding their HMDA story.
HMDA data is used by the public and regulators to identify potential differences in lending outcomes based on a prohibited basis such as race, age, and ethnicity. Lending performance areas that should be analyzed to provide insight into potential disparities, include, but are not limited to, the following:
- Processing Times: Assess how long it takes to process or disposition applications to identify potential differences in service levels provided to applicants.
- Underwriting Disparities: Analyze underwriting or decision outcomes to understand what applicants are receiving loans and what applicants are denied.
- Pricing Differences: Determine whether differences exist between what control group applicants (generally white, male) and prohibited basis group applicants are charged.
- Product Placement: Determine whether certain applicants receive less advantageous products, i.e., higher-priced mortgage loans, or adjustable-rate mortgage loans, on a prohibited basis.
- Redlining: Assess whether potential geographic or demographic gaps in lending patterns exist in higher minority geographies.
Complete and accurate HMDA data is essential to understanding where and to whom loans are being made. Regulators have signaled that continued and enhanced scrutiny should be expected. Today, every board member, key executive, legal and compliance team member should know the institution’s story—where and to whom you lend. Establishing a comprehensive HMDA compliance program will ensure the accuracy of your HMDA data, and when analyzed effectively, it will in turn help manage your fair lending risk.
