We analyzed 80,000+ CRA exam records across more than 35 years.Here's what we found.
We took the public CRA record back to 1990 and joined it to component scores and bank financials. That's 80,000+ exam records. Across the findings, one pattern recurs: CRA outcomes are shaped not only by what a bank does, but by its size, its market, and how deliberately it directs and documents qualifying activity.
Finding 1
The largest banks are ten times as likely to be rated Outstanding.
Roughly 7% of exams at the smallest banks earn an Outstanding rating. At banks over $100 billion, roughly 77% of exams do. A CRA rating is therefore meaningful only when compared with exams at banks of similar size and exam type.
% of exams rated Outstanding, by bank size (Jan 2017 – Mar 2026)
Source: FFIEC CRA Ratings, 11,090 exams Jan 2017 – Mar 2026, assets at exam
Small and Intermediate Small banks classified by exam method; large banks split by assets at exam. Strategic-Plan and Wholesale/Limited-Purpose banks (a distinct model) are drawn apart on the chart. Small = under $412M; Intermediate Small = $412M–$1.6B.
Why this probably happens
1. Large banks have more ways to source and execute CRA-qualifying activity.
Large banks can source and underwrite investments that are harder for smaller institutions to execute efficiently. They are also more likely to have dedicated tax-credit, municipal-finance, CDFI, foundation, and CRA teams. A $200 billion bank may have an entire Community Development division. A $2 billion bank may have a single CRA Officer.
2. An Outstanding rating carries more strategic value for a large bank.
For a large bank, an Outstanding rating can smooth regulatory approval for mergers, branch expansion, and other strategic initiatives. It can also serve as a hedge against fair lending scrutiny and other discrimination compliance risks. Smaller banks may rationally conclude that the incremental benefit of pursuing Outstanding does not justify the additional investment.
Finding 2
The bigger the bank, the more the top rating depends on community development.
We examined how often banks rated Outstanding overall also received the highest rating on the Investment Test, which measures community-development investments.
Earned Outstanding overall AND an Outstanding community-development score
share of banks rated Outstanding, by asset size
Earned Outstanding overall WITHOUT an Outstanding community-development score
the same banks, inverted
Source: FFIEC CRA performance evaluations and component scores, verified against the underlying performance-evaluation documents. Counts are banks rated Outstanding overall, grouped by asset size.
Why this probably happens
Community Development may be the easiest place for large banks to distinguish themselves.
Retail lending is constrained by factors largely outside a bank's control:
- Local demand
- Housing supply
- Interest rates
- Borrower credit quality
Banks operating in the same assessment area often serve very similar applicant pools. That naturally compresses performance on the lending test.
Community Development is different. Banks choose which affordable housing projects to finance, which CDFIs to support, which tax-credit investments to make, and which community partnerships to build.
In other words, the lending test measures a market opportunity, but community development activities measure a management choice.
If lending performance is clustered around Satisfactory, Community Development becomes the differentiator.
Finding 3
Needs-to-Improve banks hold more capital, make fewer loans as a percentage of assets, and keep more cash. Satisfactory and Outstanding banks look nearly identical on the same measures.
We matched 8,230 exams conducted since 2017 with each bank’s year-end financial data, then compared capital, lending, and cash ratios across the rating scale—from Needs to Improve to Outstanding.
Needs to Improve vs Satisfactory: capital, loans, and cash
133 Needs-to-Improve exams · 7,127 Satisfactory exams · group averages · one scale for all bars; note the two denominators
Outstanding vs Satisfactory: the same measures
967 Outstanding exams · 7,127 Satisfactory exams · group averages · same scale as above
Within each size band, Outstanding and Satisfactory banks differ in capital levels by about a point.
The data show association, not causation. Banks with weaker CRA performance may hold more capital partly because supervisors require it. The figures are group averages, and the Needs-to-Improve sample includes only 133 exams, so those results should be interpreted with caution.
Source: FFIEC CRA ratings joined to bank Call Reports, 8,230 exams published 2017 to 2026.
Why this probably happens
The most surprising finding in our research is that Outstanding banks do not appear to deploy materially more capital than Satisfactory banks. Their balance sheets look remarkably similar.
The difference may lie in how they deploy that capital—and how effectively they receive CRA credit for it.
Outstanding banks may be better at:
Choosing CRA-creditable investments.
A municipal bond and a LIHTC investment can occupy the same place on the balance sheet. Only one may meaningfully advance a CRA rating.
Defining assessment areas strategically.
Banks are evaluated where they choose to operate. Well-designed assessment areas increase the likelihood that qualifying activity receives credit.
Capturing credit for work already being done.
A loan to a nonprofit health clinic, affordable housing developer, or community organization may qualify for CRA consideration, but only if it is properly identified and documented. The activity may already exist. The exam credit may not.
The lesson is simple: Outstanding may be less about deploying more capital than deploying the same capital more intentionally.
Finding 4
Lower-income borrowers face the largest lending disparities in expensive urban markets, along the southern border, and in the farm valleys. The gaps are smallest across much of the Midwest.
Nationally, 39% of American families are lower-income, but only about 25% of 2024 mortgage originations went to a lower-income borrower—a 14-point gap. The map shows how lending disparities vary by county. In the median county, the gap is about 12 percentage points; the colors indicate whether each county’s gap is wider or narrower than that benchmark.
Lending to lower-income borrowers versus the share of lower-income families, county by county
every US county · 2024 mortgage originations
In the median county, lending to lower-income borrowers runs about 12 percentage points below their share of local families.
Geography matters in CRA exams because examiners assess each bank against the demographics and market conditions of the communities it serves. The same lending gap can therefore mean different things in different places.
Read the map through the lens of affordability. In high-cost markets, fewer lower-income families can afford today’s home prices, regardless of which bank they walk into.
For each county, the borrower-distribution test compares two figures: the share of mortgage originations made to lower-income borrowers and the share of families that are lower-income. In the median county, the lending share trails the family share by 11.8 percentage points.
Nationally, the gap is about 14 percentage points. That figure is calculated separately and is not the benchmark used in the map. The map is centered on the median county gap, so its colors show whether a county’s gap is wider or narrower than the median—not whether the gap has disappeared. A mint-colored county may still have a meaningful gap.
Family shares are interpolated from FFIEC income brackets and are accurate to approximately one percentage point. The color bands correspond to absolute gaps of roughly 2, 9, 15, and 22 percentage points. Counties with fewer than 25 originations are shown without a value. Puerto Rico is included in the data but is not displayed in this projection; it has the widest gap of any jurisdiction on this measure, a median of about 33 percentage points across its municipios.
Sources: 2024 HMDA originations and 2025 FFIEC census data.
Why this probably happens
The gap appears to reflect housing economics as much as banking.
The shortfall is smallest across much of the Midwest and largest in expensive urban markets, along the southern border, and in the farm valleys.
Several factors likely contribute:
- Lower home prices
- Smaller required down payments
- Greater availability of entry-level housing
- Smaller appraisal gaps
- Higher rates of homeownership among lower-income households
The implication is important for CRA officers.
A Los Angeles bank should not benchmark itself against a bank in Des Moines.
The appropriate comparison is not the national average, but peer institutions serving markets with similar affordability challenges.
Explore
Explore more than 35 years of the Outstanding rate.
The share of banks rated Outstanding peaked at 26% in 1996 and has shifted with each major change to the exam since. The dashed line marks the 1995 exam reforms.
% of banks rated Outstanding, 1990 to 2025
Until 1995, exams graded effort: outreach, marketing, and paperwork showing a bank tried to serve local credit needs. The 1995 reform graded results instead: the loans, investments, and services a bank actually delivered (60 Fed. Reg. 22156). The intermediate-small tier followed in 2005. A 2023 update was blocked in court before it took effect and is now set to be withdrawn, so exams still run on the 1995 rules.
Source: FFIEC CRA Ratings, all public exams 1990–2025
Why ratings matter
FairPlay CRAi
Three ways to put these findings to work.
Sign up for continued insights.
Volume Two is next: new findings from the record as we publish them.
CRAi Insights is FairPlay’s research series for the people who work with CRA exams every day: CRA and community development officers.
Each volume examines the public CRA record from a new angle. Future volumes will take on one question at a time—how outcomes vary by bank size, exam type, and local market.
Sources: FFIEC CRA performance evaluations, with component scores and Call Report financials joined by FairPlay. The full public record spans 80,000+ exams from 1990 to 2025; each chart states the subset it draws on. Finding 4: 2024 HMDA originations and FFIEC 2025 census data. Counties with fewer than 25 originations are shown without a value.