The 2025 Home Mortgage Disclosure Act (HMDA) National Snapshot data was made publicly available by the Consumer Financial Protection Bureau (CFPB) and the Federal Financial Institutions Examination Council (FFIEC) on June 23, 2026. Institutions using RiskExec now have access to the dataset for peer analysis, benchmarking, and year-over-year redlining review.
As with any public HMDA data release, the Snapshot is not just a new file to load. It is a new analytical baseline. Before relying on it for peer comparisons, institutions should understand how it differs from the earlier 2025 Modified LAR (mLAR) release.
The RiskExec team reviewed the Snapshot against the prior mLAR data and identified differences in agency codes, record counts, filer populations, and Tax ID values. Some appear to reflect institution resubmissions after the initial filing. Others relate to how institution-level information is sourced between the National Snapshot and prior mLAR data.
The differences are not all the same type, and they do not all carry the same analytical weight. The most important ones are outlined below.
As part of our standard pre-release review, RiskExec compared the 2025 HMDA National Snapshot against the prior mLAR data. Four differences stood out.
For 81 institutions, the agency code in the National Snapshot differs from the prior mLAR data. Together, these institutions account for just under 500,000 records in the peer dataset.
Agency code matters because it identifies an institution’s supervising agency. It can also affect how institutions are grouped for peer and supervisory analysis. When agency codes change, peer sets and year-over-year comparisons can shift in ways that are easy to miss.
The differences appear to stem from the underlying source of the agency code. The National Snapshot reflects the transmittal sheet submitted by the institution to the CFPB. The prior mLAR data relied on the CFPB Institutions API, which may no longer reflect the updated institution information captured in the National Snapshot.
The National Snapshot also includes 44 institutions with record counts that differ from the earlier mLAR release. These differences account for more than 215,000 records in the dataset.
In most cases, this appears to reflect institution resubmissions after their initial filing. This finding does not include institutions where the total record count stayed the same, but the underlying loan-level data may have changed.
The National Snapshot includes 12 institutions that were not present in the initial mLAR data release.
While that number is relatively small, new entrants can change the composition of a market, particularly in narrower geographies or product segments.
We also found 15 institutions where the Tax ID in the National Snapshot data differs from the Tax ID reflected in the prior mLAR data. While Tax ID is generally not a central field for HMDA or fair lending analysis, these differences indicate another reconciliation gap between the datasets.
Public HMDA data reflects the source files, APIs, and resubmissions available at the time it is published or updated. RiskExec relies on the data and APIs made available by the CFPB. When those sources differ, or when institutions resubmit data after an initial filing, the public dataset can change in ways that are not always obvious to downstream users.
The point for institutions is straightforward: public HMDA data can include gaps, corrections, or reconciliation differences that are not always documented in detail. If no one is checking for them, those differences can carry into peer analysis, market share calculations, and year-over-year comparisons.
Before treating the latest data as final for analysis, institutions should review whether the changes affect the specific peer groups, geographies, and product lines they rely on.
We recommend five checks before relying on the latest data for peer analysis, benchmarking, redlining review, or reporting:
Before RiskExec makes CFPB-published data available in the platform, it goes through automated validation and manual review.
Automated validation confirms the data can be processed consistently. Manual review helps identify changes that may affect how the data should be interpreted. The findings in this article came from that manual review.
That additional review step is important because the cost of an unnoticed data shift is usually borne by the institution using the data in an analysis.
The 2025 HMDA National Snapshot is ready to use, but it should be read carefully. The differences RiskExec identified are not a reason to disregard the dataset. They are a reason to review peer groups, filer populations, and year-over-year comparisons before drawing conclusions from the data.
RiskExec is sharing these findings so institutions can account for the differences now rather than discover them later in a benchmarking, redlining, or board reporting context. If you would like help reviewing how these findings affect your specific peer groups or assessment areas, our team is ready to assist.
RiskExec provides compliance data, analytics, and reporting tools for HMDA, CRA, fair lending, and related programs. This article is for informational purposes and does not constitute legal or regulatory advice.
The 2025 HMDA National Snapshot is the public national HMDA dataset made available by the CFPB and FFIEC for analysis of mortgage lending activity. Institutions often use it for peer analysis, benchmarking, market share review, and fair lending analysis.
The National Snapshot can differ from earlier Modified LAR data because institutions may resubmit HMDA data after the initial filing, and because certain institution-level fields may be sourced or reconciled differently across public data sources.
No. The differences are not a reason to disregard the dataset. They are a reason to review peer groups, filer populations, record counts, and year-over-year comparisons before relying on the data for conclusions.
Institutions should confirm peer group composition, review agency code changes, identify newly added filers, evaluate record count differences, and document any treatment decisions that affect the analysis.
