Thanks to the expanded underwriting and pricing fields that are part of the 2018 HMDA dataset, users of the popular LendingPatterns™ software can identify file records based on qualification criteria and outcomes that are candidates for further review. Specifically, users can identify applicants that appear similarly situated based on some qualifications but had different underwriting outcomes. Using a user-uploaded Private Data file, you can compare overlap approvals to overlap denials. The analysis output is a list of file records to investigate.


With a list of potential matches in hand, it’s up to you to manually review the files, looking for justification to explain the underwriting and pricing. Simply stated, are the reasons for different outcomes supported by the information in the application file?


Identify top ranked loan programs. The “Distribution by Product” LendingPatterns(TM) report ranks loan products.

Choose pairs of interest to set the target and control groups. The prohibited bases that you can analyze are race, gender and age.

Create applicant profile using qualification factors (e.g., credit score)

Select “Raw HMDA Export”. From our new pop-up box, select the HMDA fields for export. Choose a sample size of all records or a random sample that allows you to specify the number of records.


From the Excel spreadsheet output report (below), use filters to refine the target matches. You can include or exclude records.   For example, if you’re looking at underwriting, consult the denial reason column for more information regarding the decision outcome.


  1. Working with a large dataset to find strong matches will be a major chore, if not impossible, in Excel.
  2. The HMDA data are lacking in a number of fields that could be critical to your analysis. For example, you will not have information about conventional bond programs, pricing adjustments, underwriting and pricing exceptions, etc.

Reach out to for help with this or if you have any other questions.