The Automated Rules behind HMDA Reporting

2 0

When it comes to HMDA files reviews, leveraging rules-based, data-driven digital technology can ease the burden of HMDA reporting by helping lenders shift from a largely cumbersome to a highly automated process. Automation helps replace manual processes in three specific areas related to HMDA reporting, which we will cover in detail in this post:

1. Data validation, an existence check on reportable data
2. Audit testing, a data integrity review
3. Data transformation, a preparation of data for submission of the final report

1) Data validation – Often the first set of checks done on a lender’s HMDA data, automation helps with data validation or edit checks, as they are often called in the industry.  This is where automated rules determine whether the lender has all the LAR data needed for reporting. This existence check determines if there is missing or incongruent data for the lender’s reportable HMDA fields. Confirming the presence of Denial Reasons on denied loans or checking for the right amount of digits in a credit score are good examples of what data validation rules can determine. 

2) Audit Testing– HMDA automation that includes automated rules to perform a quality audit or “logical checks” for data defects prior to submission is also beneficial to keep lenders out of hot water with regulators. Many industry HMDA technology providers without this option advise clients that these “data scrubs” as they are often called are already being done on a regular basis via the QC process. As a strategy, that has a lot of holes, however. Pre- and Post-Close QC generally does not cover loan applications that never make it through the origination process. If a lender is doing some type of discretionary reviews of loans with adverse actions, they could pick up on some of these. However, all of these reviews are typically done via sampling and therefore not on every loan file. A more prudent approach would be to perform some of the basic audit tests across all applications for data fields required for HMDA submission.

Automated audit rules perform tests for things like the accuracy of the Rate Spread or geocode data from the LOS. Auditors can then manually review this type of discrepancy within seconds by quickly looking at the loan file in more detail to determine the true values. Another discrepancy an audit test might find is if there is a duplicate denial reason on the file.  While you can have up to five, you cannot have two of the same.

When audit testing incorporates both loan file data and documents, additional discrepancies might be found. For example, if the LOS data contains a value (such as credit score) that differs from the source documents, auditors could then look back and validate the truth using other data and document sources being leveraged in the audit. These types of data discrepancies are more common than we’d all like to think, given our reliance on Loan Origination Systems.

3) Data transformation– In its simplest definition, data transformation is a set of tasks required to change data fields and/or formats in preparation for outbound file generation. This exported information then becomes consumable by another application that expects these values and formats.  

When it comes to HMDA reporting, automated systems can much more efficiently transform the reportable fields of loan file data for the Loan Application Register (LAR) submission. The very specific formatting requirements laid out by the CFPB can be codified into data transformation rules and standardized for each loan file. 

One straightforward example of a HMDA data transformation requirement is the need to mark all conventional loans with a number “1” for LAR submission.  A more complex example relates to credit score.  If the Action Taken on a particular loan was “Withdrawn,” the LAR requires the credit score value be transformed to reflect “8888” for “Not Applicable.” In both instances, the loan record in the system remains intact, but the output file reflects the values needed for proper submission.  

While this type of task seems simple, the automation of it is extremely helpful for originators with a large amount of reportable applications / loans. In this case it becomes impractical to export rows and rows of data into an Excel file and create macros to do the transformation manually ‘offline” for this final important step in the reporting process.

HMDA automation offers originators a lot of beneficial data validation, auditing and analysis tools to keep regulators at bay.  Given the outcome of HMDA Transaction Testing Sample Sizes and Thresholds done by regulators, resubmission of LAR data may be required with just 3 or 4 errors on the sample loans tested, whether your LAR counts are large or small, it is important for lenders to ensure these fields are reported correctly.^[1]  

As well, if a financial institution’s HMDA data are collected through multiple data collection and reporting systems,” The Federal Financial Institutions Examination Council (FFIEC) reserves the right to “test a single sample from the financial institution’s entire HMDA LAR, test separate samples from each system, or test samples from selected systems chosen based on risk.”^[2]

The FFIEC is also tasked with verifying the accuracy of the data in HMDA LAR sample(s) against the corresponding loan files which they request from the financial institution. FFIEC examiners will document any difference between the LAR data and the loan file and seek explanation from the financial institution, which could lead to penalty.^[3]  Therefore, it is a worthwhile exercise to perform loan quality audit tests on data and documents for at least a portion of loans prior to submission. 

LoanLogics’ portfolio of solutions for HMDA provide flexibility for data only reviews, as well as data and documents reviews prior to submission of the HMDA LAR. Our LoanHD® HMDA DirectCheck and LoanLogics® HMDA Complete™ solutions, were recently enhanced with over 100 new rules earlier this year to further automate both the data validation and data transformation workflows.  Leveraging the robust set of audit rules for data integrity checks helps further review the file for quality control prior to submission to the CFPB. Taking it one step further, comprehensive discrepancy reports enable lenders to see what the rules identified as errors and provide an export to update the system of record, often the loan origination system, to prevent ongoing systemic data issues.

[1] https://www.federalreserve.gov/supervisionreg/caletters/CA%2017-2%20Attachment%20HMDA%20Resubmission%20Guidelines%20Final%2008-10-17.pdf

[2] https://www.federalreserve.gov/supervisionreg/caletters/CA%2017-2%20Attachment%20HMDA%20Resubmission%20Guidelines%20Final%2008-10-17.pdf

[3] https://www.federalreserve.gov/supervisionreg/caletters/CA%2017-2%20Attachment%20HMDA%20Resubmission%20Guidelines%20Final%2008-10-17.pdf

About the Author

Natalie Henderson

For over 11 years, Natalie Henderson has been an integral leader for LoanLogics’ data and document solution, LoanLogics® IDEA (Intelligent Data Extraction and Automation). As a member of the product management team, her expertise centered around managing and building the solution’s extensive rules-based automation engine capabilities. Recently, Natalie has transitioned into a key position where she manages go to market strategy for the company, administering the successful roll out of all product launches and significant releases.