HMDA and CRA Data: What Community Bankers and Analysts Need to Know
A bank with a 1.10% ROA, a CET1 ratio north of 13%, and a clean NPL book can still have a "Needs to Improve" CRA rating quietly sitting in its …
A bank with a 1.10% ROA, a CET1 ratio north of 13%, and a clean NPL book can still have a “Needs to Improve” CRA rating quietly sitting in its regulatory file — waiting to surface in the next merger application review. That combination is more common than most compliance teams want to admit, and it illustrates exactly why HMDA data and CRA ratings belong alongside the financials in any serious bank analysis.
The call report tells you how much a bank lends and how those loans are performing. HMDA data and CRA ratings answer the question regulators actually care about when a bank files an expansion application: to whom does this institution lend, and how well does it serve the communities where it collects deposits?
What HMDA Data Actually Contains
The Home Mortgage Disclosure Act requires covered institutions to collect and publicly disclose application-level detail on their mortgage lending. Not aggregates — individual records. For each covered application, the filing includes loan amount, loan type and purpose, property location down to the census tract, action taken (originated, approved but not accepted, denied, withdrawn, or incomplete), and applicant demographic characteristics.
The census-tract granularity is the core value of the dataset. The call report (FFIEC 031/041/051, Schedule RC-C) gives you aggregate 1–4 family residential balances and closed-end mortgage breakdowns — but it tells you nothing about the geography of those loans or whose applications got denied. HMDA fills that gap. It is the foundational dataset for fair-lending analysis: mapping where a bank lends, where it does not, and whether denial rates vary across applicant groups or neighborhoods in ways that warrant scrutiny.
The CFPB administers HMDA and publishes the data annually. The raw files are large and require some care — the column schema shifted across filing years, action-taken codes changed with the 2018 rule overhaul, and census tract vintages do not always align with the geographic lookup tables most researchers expect. That friction is worth knowing before you start.
That chart shape — origination mix skewing toward upper-income tracts relative to peers operating in the same market — is the first thing a fair-lending examiner pulls. If your bank looks like the left bar in that chart, the exam conversation starts there.
The CRA: What It Requires and Who Examines It
The Community Reinvestment Act was passed in 1977, specifically to counter redlining. The law requires a bank’s federal regulator to assess how well it meets the credit needs of its entire community, including low- and moderate-income neighborhoods, and to make that assessment public.
Which regulator runs the exam depends on charter type: the OCC for national banks, the Federal Reserve for state member banks, the FDIC for state nonmember banks. Credit unions are not subject to CRA — they operate under a separate framework through the NCUA. Exam frequency is risk-based; smaller banks with clean records may go several years between exams, while institutions with prior rating issues or significant growth see more frequent reviews.
The exam evaluates lending activity, investment activity (qualified investments in LMI areas), and service (branch access and community development services). The weight of each component depends on bank size. Institutions under $600M in assets face a streamlined lending test rather than the full three-part evaluation that large banks face. The 2023 CRA modernization rule introduced material changes to assessment area definitions and performance metrics, though implementation has been subject to legal challenge and the timeline remains unsettled.
CRA Ratings: Four Levels, Uneven Consequences
The rating scale is four levels. The consequences are not evenly distributed.
| CRA Rating | Regulatory Meaning |
|---|---|
| Outstanding | Exceeds community credit need expectations |
| Satisfactory | Adequately meets community credit needs |
| Needs to Improve | Falls short of expectations |
| Substantial Noncompliance | Largely fails to meet community credit needs |
Roughly 95% of rated banks land at Satisfactory or Outstanding — around 88% Satisfactory, 9% Outstanding, based on published FFIEC distribution data. The bottom two ratings are rare. They also hit disproportionately hard.
Under the Bank Merger Act and the Bank Holding Company Act, regulators explicitly weigh CRA performance when evaluating applications for mergers, acquisitions, new branches, and other expansions. A “Needs to Improve” rating does not automatically kill a deal, but it creates a documented obstacle that acquirers and their counsel track through the entire approval process. “Substantial Noncompliance” has derailed transactions outright.
That is the piece analysts outside compliance departments frequently miss: the CRA rating is a strategic variable, not a compliance checkbox. A bank that plans to grow through acquisition — or that expects to be acquired — carries its CRA rating into every regulatory negotiation that follows.
How These Datasets Differ from the Call Report
These are complementary datasets, not substitutes for each other.
| Call Report / UBPR | HMDA | CRA | |
|---|---|---|---|
| What it measures | Financial condition and performance | Mortgage lending decisions | Community reinvestment record |
| Granularity | Institution-level aggregates | Application-level records | Rating plus narrative evaluation |
| Primary analytical use | Safety, soundness, profitability | Fair-lending, geographic coverage | Regulatory risk, M&A due diligence |
| Filing form | FFIEC 031/041/051 | Loan Application Register (LAR) | Examination |
| Public? | Yes | Yes | Yes |
| Frequency | Quarterly | Annual | Periodic (risk-based) |
The UBPR gives you peer-adjusted ratio analysis built from call report data — NIM (community bank average ~3.3–3.5% as of Q1 2025), efficiency ratio (60–65% typical, >70% concerning), asset quality metrics. None of that answers whether the bank is making mortgage loans in the LMI census tracts within its assessment area. For that, you need HMDA data. Both questions matter; neither dataset alone answers both.
How Practitioners Use HMDA and CRA Data
Fair-lending and compliance teams run HMDA to compute denial-rate disparities across protected classes, map origination density against the bank’s deposit footprint, and compare LMI lending penetration against peer institutions in the same geographies. When the numbers look off relative to peers, internal file reviews and potential remediation begin well before an examiner schedules a visit.
A specific pattern worth watching: a bank whose denial rate for minority applicants runs 8–12 percentage points above its denial rate for non-minority applicants with similar credit profiles has a fair-lending problem that HMDA data will surface, even if the CRA rating is Satisfactory. The examiner sees the same data you do.
M&A analysts and strategic advisers check CRA ratings as standard due diligence on any bank acquisition target. A target carrying “Needs to Improve” gets scrutiny on deal certainty, not just deal value. The acquirer’s own CRA standing matters equally — regulators have used public comment periods on merger applications to extract specific LMI lending commitments from acquirers with mixed community reinvestment records.
Board directors and senior management should understand that the LMI origination share is the single most scrutinized HMDA metric in CRA exams. A bank running five percentage points below its peer group in LMI originations — while posting strong ROA and clean asset quality — is exactly the institution that ends up in a difficult exam conversation. Financial strength does not offset a weak lending record in the exam framework.
One additional dataset worth noting: Section 1071 of Dodd-Frank requires banks to collect and report small-business loan data in a manner parallel to HMDA. The CFPB finalized the collection rules, though implementation is still rolling out. When that data becomes broadly available, it will extend the geographic credit-access analysis that HMDA now enables for mortgages into commercial and small-business lending — a significant expansion of the analytical picture.
Pulling HMDA and CRA Data Programmatically
Analysts who want to automate peer comparisons can pull HMDA summary statistics and CRA ratings from the BankRegAPI alongside call report financials:
from bankregreports import BankReg
brr = BankReg("brr_xxx")
# HMDA summary for a specific institution
hmda = brr.hmda.summary(rssd_id=611240, year=2023)
# Returns denial_rate_pct, lmi_share_originated_pct, peer_lmi_share_pct, census_tracts_served
# CRA rating history
cra = brr.cra.rating(rssd_id=611240)
# Returns exam_date, rating, regulator, prior_rating, evaluation_url
The response includes the bank’s LMI origination share alongside the peer average for the same assessment area — the comparison that matters most in exam prep.
What Examiners Actually Focus On
The exam conversation for most community banks centers on three questions derived from HMDA data:
- What share of originations went to LMI census tracts, and how does that compare to peers in the same market?
- Does the bank’s application volume in LMI tracts reflect meaningful outreach, or is it incidental?
- Are denial rates consistent across applicant groups after controlling for creditworthiness factors?
Examiners also look at geographic gaps — LMI tracts within the assessment area where the bank has essentially no origination activity. A pattern of lending in middle- and upper-income tracts while systematically absent from LMI tracts in the same market is harder to explain than a below-average LMI share on its own.
The CRA performance evaluation narrative, which is a public document available through the FFIEC, often contains more specific and actionable information than the rating itself. A Satisfactory rating with a narrative that highlights weak investment activity or geographic gaps tells you something the letter grade alone does not. Read the narrative.
Frequently Asked Questions
What is HMDA data? HMDA (Home Mortgage Disclosure Act) data is publicly available, application-level information about mortgage lending. It records loan amount, type and purpose, property location by census tract, action taken on each application, and applicant demographics. The CFPB publishes it annually.
What is a CRA rating and why does it matter? A CRA rating is the public grade a bank’s federal regulator assigns after examining how well the institution meets community credit needs, including in low- and moderate-income areas. The four levels — Outstanding, Satisfactory, Needs to Improve, and Substantial Noncompliance — are published alongside a full narrative evaluation. Ratings of Needs to Improve or worse can impede regulatory approval for mergers and expansion, making the rating a material factor in any institution’s strategic planning.
How is HMDA data different from call report data? The call report reports aggregate financial condition quarterly. HMDA is annual, application-level mortgage data showing who receives credit and where. They address different questions and are most useful together.
Are HMDA and CRA data public? Yes. HMDA data is published annually by the CFPB. CRA ratings and performance evaluation narratives are public records available through the FFIEC website. Both are accessible without any subscription or account.
Which banks are subject to CRA? FDIC-insured commercial banks and savings institutions. Credit unions are not subject to CRA. Coverage thresholds for specific exam components vary by asset size.
What does “assessment area” mean in a CRA exam? An assessment area is the geographic footprint where a bank operates and takes deposits — typically defined by the counties where it has branches or a significant volume of deposit activity. CRA performance is evaluated relative to credit needs and peer lending within that footprint.
The data in this post is available through the BankRegReports platform. Pull peer benchmarks, Call Report metrics, UBPR trends, and enforcement history for any FDIC-insured bank — no data engineering required. Explore the platform →