HMDA reporting trips up small banks more than almost any other consumer compliance obligation. The data points are numerous, the definitions are technical, and errors in geocoding or applicant data can trigger examination findings, or worse, feed into fair lending analyses that paint an inaccurate picture of your lending. This guide covers who must report, what data is required, where small banks qualify for exemptions, and the errors examiners catch most frequently.
Key Takeaways:
- HMDA reporting under Regulation C (12 CFR Part 1003) applies to banks and thrifts meeting specific asset and origination thresholds
- The EGRRCPA partial exemption reduces required data points for smaller institutions, but core fields still must be reported
- Common errors include incorrect geocoding, missing applicant demographic data, and misidentification of covered loans
- LAR submission to the CFPB is due by March 1 for the prior calendar year's data
Who Must Report HMDA Data
HMDA reporting obligations are governed by Regulation C (12 CFR Part 1003). Not every bank is a HMDA reporter. The institutional coverage test has several components:
For depository institutions (banks, thrifts, credit unions), you must report if you meet ALL of these:
- You have a home or branch office in a metropolitan statistical area (MSA) on the preceding December 31
- You meet the asset threshold, for banks and thrifts, the threshold is adjusted annually. For 2026, institutions with assets above $56 million (check the CFPB's annual announcement for the current year's figure) must evaluate whether they meet the remaining criteria
- You originated at least 25 closed-end mortgage loans OR at least 100 open-end lines of credit in each of the two preceding calendar years
If you fall below the loan origination threshold in either of the two preceding years, you are not required to report, even if you exceed the asset threshold.
Voluntary reporting: Banks below the reporting thresholds may voluntarily report HMDA data. Some community banks choose voluntary reporting to demonstrate CRA commitment or to have their lending data included in market analyses.
What Loans Are Covered
HMDA covers dwelling-secured loans and applications, which includes a broader set of transactions than many banks realize.
Covered transactions:
- Purchase-money mortgages (first and subordinate liens)
- Home improvement loans secured by a dwelling
- Refinancings of dwelling-secured loans
- Home equity lines of credit (HELOCs)
- Reverse mortgages
- Construction-only and construction-to-permanent loans
- Multifamily dwelling loans (5+ units)
Not covered:
- Loans secured by vacant land (unless the loan purpose is home purchase or improvement)
- Agricultural purpose loans
- Commercial purpose loans secured by multifamily properties where the primary purpose is commercial
- Temporary financing (bridge loans with a term of less than 12 months)
- Purchases of partial interests in loans
The "dwelling-secured" definition extends to manufactured homes, multifamily properties, and mixed-use properties where the property includes a dwelling. Community banks with agricultural lending portfolios must carefully distinguish between farm dwelling loans (covered) and agricultural purpose loans (not covered).
Required Data Points
HMDA reporters must collect and report a substantial number of data points for each covered application and origination. The full list under Regulation C includes:
Application and loan data:
- Universal Loan Identifier (ULI), a unique identifier for each application
- Application date and action taken date
- Action taken (originated, approved but not accepted, denied, withdrawn, incomplete, purchased)
- Loan type (conventional, FHA, VA, USDA/RHS)
- Loan purpose (home purchase, refinancing, cash-out refinancing, home improvement, other)
- Loan amount
- Interest rate
- Rate spread (APR minus average prime offer rate)
- Lien status
- Loan term
Property data:
- Property address
- Census tract (geocoded from the property address)
- Property type (single family, manufactured home, multifamily)
- Construction method
- Total units
- Property value
Applicant data:
- Ethnicity
- Race
- Sex
- Age
- Income
- Credit score and scoring model
- Debt-to-income ratio
Denial and pricing data:
- Denial reasons (up to four per application)
- Total loan costs
- Origination charges
- Discount points
- Lender credits
- Prepayment penalty term
- Introductory rate period
This is not the complete list; Regulation C specifies additional fields. But these are the fields where errors most commonly occur.
The EGRRCPA Partial Exemption
The Economic Growth, Regulatory Relief, and Consumer Protection Act (EGRRCPA) of 2018 created a partial exemption that reduces HMDA data collection for qualifying institutions.
Who qualifies: Banks and credit unions that originated fewer than 500 closed-end mortgage loans in each of the two preceding calendar years AND fewer than 500 open-end lines of credit in each of the two preceding calendar years.
What the partial exemption does: Exempt institutions do not need to report approximately 26 data points that the 2015 HMDA rule added. The exempted fields include:
- Universal Loan Identifier (can use a non-universal identifier)
- Property address (census tract is still required)
- Rate spread
- Credit score
- Automated underwriting system information
- Debt-to-income ratio
- Combined loan-to-value ratio
- Loan term
- Introductory rate period
- Non-amortizing features
- Property value
- Manufactured home information
- Total loan costs and origination charges
- Discount points and lender credits
- Prepayment penalty term
- Application channel
What the partial exemption does NOT exempt: Even with the partial exemption, banks must still report the core HMDA data points: applicant demographic data (race, ethnicity, sex, age, income), loan amount, action taken, loan type, loan purpose, lien status, and census tract. The partial exemption reduces burden but does not eliminate the reporting obligation.
Geocoding and Census Tract Accuracy
Geocoding, converting a property address into a census tract number, is one of the highest-error-rate fields in HMDA data. Incorrect geocoding affects fair lending analyses, CRA evaluations, and the bank's own internal analytics.
The CFPB provides a geocoding tool that maps addresses to census tracts using the most current census data. Banks should use this tool or a validated vendor system, not manual lookups.
Common geocoding errors include:
- Using outdated census tract boundaries: Census tracts change with each decennial census. The 2020 census tract boundaries are current, and banks that haven't updated their geocoding systems from 2010 boundaries will report incorrect tracts.
- P.O. box addresses: HMDA requires the property address, not the borrower's mailing address. If the file shows a P.O. box, geocoding fails.
- Rural properties without standard addresses: Properties without a standard street address require alternative geocoding methods (coordinates, parcel data, or county assessor records).
- Manufactured homes on leased land: The property location may differ from the mailing address or dealer address.
Examiners verify geocoding accuracy by pulling a sample of loans and re-geocoding the property addresses. Systematic errors, especially errors that cluster in specific census tracts, trigger additional fair lending review.
LAR Submission Process and Timeline
HMDA data is submitted to the CFPB through the Loan/Application Register (LAR) via the HMDA Platform at ffiec.cfpb.gov.
Key deadlines:
- March 1: Annual LAR submission deadline for the prior calendar year
- Quarterly reporting: Institutions with 60,000+ covered originations and applications must file quarterly (within 60 days of quarter-end). Small banks are generally not subject to quarterly reporting.
Submission process:
- Prepare the LAR in the specified pipe-delimited format
- Upload to the HMDA Platform
- Run the platform's built-in edit checks (syntactical, validity, quality, and macro edits)
- Review and verify all flagged edits
- Sign and submit the LAR
The HMDA Platform runs extensive edit checks that flag potential errors. Syntactical edits check formatting. Validity edits check that reported values are within acceptable ranges. Quality edits flag unusual but potentially valid entries. Macro edits flag aggregate patterns that seem unusual (e.g., an unusually high denial rate for a particular loan type).
Quality and macro edits require the bank to verify the data is correct, not necessarily change it. If the data is accurate but unusual, the bank can verify and submit.
Common Errors and Examination Findings
Based on CFPB and FFIEC examination data, the most frequent HMDA errors at small banks include:
- Incorrect action-taken codes: Misclassifying a withdrawn application as denied, or failing to report preapproval requests that progressed to a decision
- Missing or incorrect applicant demographic data: Failing to report race, ethnicity, or sex when the applicant declines to provide it (the bank must report based on visual observation or surname for in-person applications)
- Wrong loan purpose codes: Misclassifying cash-out refinancings as rate-and-term refinancings
- Geocoding errors: Using the wrong census tract, especially after the 2020 census boundary changes
- Excluding covered loans: Failing to report home improvement loans or HELOCs that meet the reporting criteria
- Incorrect rate spread: Calculating rate spread against the wrong benchmark index or failing to report rate spread for above-threshold loans
These errors matter because HMDA data feeds directly into fair lending analyses. An inaccurate LAR can create the appearance of fair lending issues where none exist, or mask real issues that go unaddressed.
How Canarie Helps Banks Manage HMDA Compliance
HMDA compliance involves a year-round cycle, data collection during origination, mid-year scrubs, year-end compilation, edit resolution, and the March 1 submission deadline. Each step requires coordination between loan officers, operations staff, and compliance personnel.
Canarie tracks each stage of the HMDA cycle as a compliance task with assigned owners and deadlines. Mid-year data quality reviews are scheduled automatically, edit resolution tasks are assigned to the right team members, and submission milestones are tracked with evidence of completion. Your compliance team has a documented record of every step from data collection through final submission.
See how Canarie helps banks stay on top of reporting deadlines →
Frequently Asked Questions
Can a bank lose its HMDA partial exemption?
Yes. The partial exemption is evaluated annually based on origination volume in the two preceding calendar years. If your bank originates 500 or more closed-end mortgage loans in both of the two preceding years, you lose the partial exemption and must report the full set of data points for the following calendar year. Banks near the 500-loan threshold should monitor their origination volume and prepare for potential full reporting.
Does HMDA require reporting applications that are withdrawn before a decision?
Yes. Applications withdrawn by the applicant must be reported with an action-taken code of "4" (application withdrawn). Applications that the bank determines are incomplete after requesting additional information must be reported with an action-taken code of "5" (file closed for incompleteness). Only inquiries and prequalification requests that do not result in a formal application are excluded from reporting.
How does HMDA define "home improvement" for reporting purposes?
A home improvement loan is reported as HMDA-covered if it is secured by a dwelling, regardless of the stated purpose. If the loan is not secured by a dwelling, it is reported only if the borrower states the purpose is home improvement. The dwelling-secured test is the primary determinant. A personal loan used for home improvement but not secured by the dwelling is not covered unless the borrower specifically states the home improvement purpose at origination.
What happens if a bank submits HMDA data with errors?
The CFPB may require resubmission if errors exceed a certain threshold. More practically, HMDA data quality is reviewed during consumer compliance examinations by the bank's primary federal regulator. Significant error rates can result in a MRA, a requirement to resubmit corrected data, and increased examiner scrutiny of the bank's data collection processes. Errors in applicant demographic data are particularly sensitive because they affect fair lending analyses.