Sign up to receive our monthly newsletter!

6 HMDA Data Integrity Issues You Should Know About

Feb 10, 2020 by Loretta Kirkwood

For those that are scrubbing their HMDA data, we wanted to share a few data integrity issues clients have identified or that have been the focus of recent regulatory examinations. Of course, we also want to offer some quick tips for addressing these key “red flags”.


The most important thing to remember is that data integrity issues often arise from inconsistences between HMDA data and the data used to process and underwrite a mortgage loan application. It is critical to understand the nuances of how data flows through your loan origination system (“LOS”) and review any business rules designed to translate that data into reportable HMDA fields.

Application Date

Testing the accuracy of application date or action date is most often a file-by-file review. We suggest that calculating the number of days from application to final action can quickly identify areas of potential risk. For example, any loan origination that shows only one (1) day from application to origination, is likely to have an error in either the application date or action date. It generally takes much longer than one day to decision and close a mortgage / HMDA application. Similarly, testing for extended length of time in originations such as 180 days or more, may identify an error unless the loan is a single closing construction/permanent loan.

Action Type

There is significant confusion in how to report the final disposition on an application where the customer chose not to proceed. The HMDA Guide contains an Action Taken Chart in Appendix B which provides various scenarios. However, there may be challenges in coding final action using business rules. The following are examples of application outcomes that have created the most confusion:

  • Approved Not Accepted
    • Application was approved – ONLY customary commitment or closing conditions are outstanding:
      • Borrower did not respond
      • Borrower expressly withdrew the application
  • Denied
    • Application was denied BEFORE:
      • Borrower withdrew
      • File was closed for incompleteness
    • Application was approved subject to underwriting or credit conditions, which were not met.
    • Counteroffer:
      • Borrower did not accept
      • Borrower did not respond
  • Withdrawn
    • PRIOR to a credit decision – borrower expressly withdraws application
    • Application was approved subject to underwriting or credit conditions:
      • Borrower expressly withdraws prior to satisfying conditions
    • Counteroffer accepted by borrower:
      • Conditionally approved subject to underwriting or credit conditions
      • Borrower expressly withdraws prior to satisfying conditions

Visual Observation

Another area of concern is ensuring the accurate reporting of Visual Observation, specifically when an application was taken face-to-face. If a lender captures how an application was taken (face-to-face, internet, mail, telephone), testing for consistency in reporting Visual Observation assists in identifying potential issues. Examples of testing include:

  • Application was taken face-to-face, and race, ethnicity, and/or gender are coded as Information Not Provided
  • Application was taken via internet, mail, or telephone and Visual Observation is coded as 1-Collected on the basis of visual observation or surname


We have heard from several clients that recent regulatory exams have definitely focused on the accuracy of age. This isn’t really a surprise, since age is a new demographic field and is being used in fair lending reviews. The verification process involves calculating the age using date-of-birth and application date.

Denial Reasons

It is important to understand the potential risk in reporting of credit information such as credit score, debt-to-income, and combined loan-to-value allows. It is now easy to identify conflicts between denial reasons and actual credit information. For example, for applications denied for debt-to-income, originated and approved not accepted applications should be reviewed to identify debt-to-income ratios that are higher than denials. Of course, it’s important to ensure that testing focuses on similar products and terms.

S.2155 / Partial Exemption

Equally important is the issue of data collection when an institution qualifies for the partial exemption under S.2155. While the partial exemption rule is intended to limit the burden to smaller institutions, it is important to remember that data required during a regulatory fair lending examination may include many of the data fields that are not required to be reported for HMDA.

It shouldn’t really be a big surprise that additional data is being required during an exam. Before the expanded HMDA data requirements, lenders were often required to provide what was called HMDA “plus” data for fair lending reviews and fields needed to verify “derived” fields such as geocode and rate spread. Just remember, examiners “CAN” request data that was not required to be reported. We highly recommend that lenders that qualify for the partial exemption, continue to capture data fields that are needed for HMDA data integrity reviews and fair lending analysis.

As always, we value your feedback and will continue to share customer experiences and challenges with the QuestSoft community!