On December 21, 2018, the Consumer Financial Protection Bureau (CFPB) announced its final policy guidance regarding the public disclosure of loan-level HMDA data collected under the 2015 HMDA rule. The guidance is effective for HMDA data that will be published by the CFPB in 2019 for 2018 reportable mortgage activity.
The CFPB will disclose most of the data fields collected in accordance with the 2015 HMDA rule, without modification. This includes – with one exception – all HMDA fields that were publicly reported prior to the effective date of the 2015 HMDA rule, as well as several new data fields, including loan term, combined loan-to-value ratio, interest rate, total loan costs, total discount points, lender credits and rate spread.
To address privacy concerns, the CFPB will exclude or otherwise modify several data fields. Modified data fields will include loan amount, property value, applicant age, debt-to-income ratio and the number of units in the dwelling. For these fields, the data will be disclosed based on ranges of data instead of actual values. The modification of loan amount – which will be disclosed in $10,000 bins – represents a departure from how these data were previously published. This modification was driven by concerns for privacy based on research showing that loan amounts can be used to re-identify borrowers by matching to public records.
Excluded data fields will consist of application date, action taken date, property address, credit score and all free-form text fields. Perhaps the most controversial of the excluded data fields is credit score. The Bureau received comments from community groups promoting the publication of these data in a modified form due to the critical importance of this factor in Fair Lending decisions. Ultimately, the CFPB determined that the risks to privacy outweighed the benefits of transparency in disclosing these data, even with modification.
Based on our experience in Fair Lending analysis, the additional data that will be disclosed to the public will enhance the level of HMDA data analysis that can be accomplished by researchers and other mortgage market stakeholders. However, the exclusion of credit score and the modification of variables like loan amount and debt-to-income ratio will reduce the level of precision that we would expect to accomplish with the complete, non-modified data that we typically receive from our clients for Fair Lending analyses.