Short answer: CAMELS is the six-part rating an examination assigns to every US bank: Capital adequacy, Asset quality, Management, Earnings, Liquidity, and Sensitivity to market risk, each scored 1 to 5, plus a composite. The rating is confidential supervisory information and is never published. What is public is the call report data that moves alongside five of the six components. In the quarter ending June 30, 2026 the median US bank ran an 10.96% tier 1 leverage ratio, a 0.47% nonperforming loan ratio, a 1.23% ROA, and an 80.16% loans to deposits ratio. Management is the component with no public number, and it is the one most likely to cap your composite.
What the rating is, and what it costs you
CAMELS comes out of the Uniform Financial Institutions Rating System, used in common by the federal banking agencies. Six components, each rated 1 through 5 with 1 the strongest, plus a composite on the same scale. The composite is not an average. It is an examiner judgment informed by the six, and one weak component routinely carries it: a bank with five 2s and a 4 in asset quality is not a composite 2.
The line that matters is between 2 and 3. Composite 1 and 2 banks are considered fundamentally sound and get expedited processing on applications. A 3 means the agency has found problems it expects management to fix, and it changes the relationship: more frequent contact, targeted reviews between exams, scrutiny on growth and capital distributions. A 4 or 5 means unsafe or unsound condition and almost always comes with a formal enforcement action.
Two consequences land harder than bankers expect. Assessment pricing for small institutions runs partly off a weighted average of the component ratings, so a downgrade becomes a recurring expense line. And the agencies can reclassify a bank into a lower prompt corrective action capital category on an examination finding, which is how a bank with perfectly adequate published capital ratios loses the ability to accept brokered deposits.
Why the rating is confidential and stays that way
The rating is confidential supervisory information. It belongs to the agency, not to the bank that received it, and the agencies restrict its disclosure by regulation. A bank may share the report of examination with its board, its officers, and in defined circumstances its attorneys and outside accountants. It may not publish the rating, put it in an offering document, or give it to a counterparty.
The rationale is that supervision depends on candor. An examination works because management hands over the problem credits, the validation with the ugly findings, and the minutes where the limit breach was argued about. Make the grade public and every exam becomes a disclosure event, the incentive flips toward managing the record instead of fixing the problem, and a composite 3 at a liquid bank becomes self-fulfilling the week it leaks.
One practical consequence gets ignored. Do not confirm your rating to anyone outside the permitted circle, including when it is good, because confirming a 1 is still a disclosure and it builds a pattern in which your silence in a later year carries the answer. Meanwhile everything around the rating is public: consent orders, written agreements, penalties, capital categories, and the full quarterly financials. A bank can be downgraded to a 3 and hold it for eighteen months with nothing visible changing, which is why counterparties reconstruct what they can from the numbers below.
The public number behind each component
None of these ratios produces a rating. They tell an examiner where to look first, the same job the off-site surveillance systems do between examinations. Sitting in the tail of your peer distribution reliably predicts examiner attention; a bad answer to the question that follows is what moves a component.
| Component | Public call report proxy | Banks | Bottom quartile | Median | Top quartile |
|---|---|---|---|---|---|
| Capital | Tier 1 leverage ratio | 4,238 | 9.70% | 10.96% | 13.00% |
| Capital | Equity to assets | 4,238 | 9.06% | 10.56% | 12.73% |
| Asset quality | Nonperforming loans to loans | 4,203 | 0.12% | 0.47% | 1.17% |
| Asset quality | Allowance for credit losses to loans | 4,205 | 0.98% | 1.19% | 1.44% |
| Management | No public proxy exists | n/a | n/a | n/a | n/a |
| Earnings | Return on assets | 4,230 | 0.83% | 1.23% | 1.65% |
| Earnings | Efficiency ratio | 4,234 | 53.08% | 61.81% | 72.18% |
| Liquidity | Loans to deposits | 4,238 | 65.73% | 80.16% | 90.94% |
| Liquidity | Brokered share of deposits | 4,238 | 0.00% | 0.00% | 4.32% |
| Liquidity | Uninsured share of deposits | 998 | 23.76% | 32.08% | 42.10% |
| Sensitivity | AOCI as a share of equity | 4,237 | -13.05% | -4.95% | -1.05% |
Source: BankingLens, computed from FFIEC call reports for the quarter ending June 30, 2026. Quartiles are calculated across the full population of filing banks in each size band, not sampled. Non-insured non-deposit trust companies are left out of every statistic: they take no deposits and make no loans, so a margin or a funding cost computed for them has no meaning. Bottom quartile is the 25th percentile and top quartile the 75th; for ratios where lower is better, such as nonperforming loans and the efficiency ratio, the bottom quartile column is the stronger end. Only 998 banks report an estimate of uninsured deposits, so that row describes larger institutions rather than the whole industry.
C: capital adequacy
The component is not about the level of the ratio, it is about whether capital is adequate for the risk in this balance sheet, which is why two banks at identical tier 1 leverage get different capital ratings. Examiners work the composition of capital, asset growth against retained earnings, the dividend the holding company needs to service its own debt, the concentrations that consume capital disproportionately, and whether there is credible access to new capital if the plan fails.
Publicly this tracks tier 1 leverage and equity to assets. The median bank ran 10.96% tier 1 leverage, middle half 9.70% to 13.00%, against a 5% prompt corrective action threshold for well capitalized and a community bank leverage ratio election above 9%. Almost nobody operates near the statutory line, and a bank that does has said something about its capital plan before an examiner opens a file.
| Bank size (total assets) | Banks | Bottom quartile | Median | Top quartile |
|---|---|---|---|---|
| All banks | 4,238 | 9.70% | 10.96% | 13.00% |
| Under $100M | 543 | 10.58% | 12.76% | 15.96% |
| $100M - $300M | 1,214 | 9.86% | 11.30% | 13.63% |
| $300M - $1B | 1,433 | 9.61% | 10.81% | 12.58% |
| $1B - $3B | 626 | 9.61% | 10.54% | 11.97% |
| $3B - $10B | 264 | 9.63% | 10.45% | 11.66% |
| $10B - $100B | 126 | 9.43% | 10.25% | 11.08% |
| Over $100B | 32 | 7.98% | 9.44% | 10.05% |
Source: BankingLens, computed from FFIEC call reports for the quarter ending June 30, 2026. Quartiles are calculated across the full population of filing banks in each size band, not sampled. Non-insured non-deposit trust companies are left out of every statistic: they take no deposits and make no loans, so a margin or a funding cost computed for them has no meaning.
Capital falls with size, monotonically, from a 12.76% median under $100 million to 9.44% above $100 billion. The smallest banks are not better capitalized because they are better run. They hold it because they have no other loss absorption, no debt market to tap in a hurry, and often one concentration that could take a double-digit bite. Benchmark against your band.
Worth noticing: median tier 1 leverage of 10.96% sits above median equity to assets of 10.56%, the reverse of what people expect from a regulatory ratio that deducts intangibles. Most of that is the AOCI opt-out, which keeps unrealized securities losses out of regulatory capital while GAAP equity carries them in full; the two also use different denominators, average assets against period-end assets. The size of your own gap is a rough read on how much of your capital is an accounting election. Full distribution in tier 1 leverage benchmarks.
A: asset quality
This is where examiners spend the most hours, and almost none of them on your nonperforming loan ratio. They pull a sample of credits, re-grade them independently, and count how many they move; that downgrade rate is the finding. From there it runs through classified and special mention assets against capital, policy exceptions and who approved them, appraisal age on commercial real estate, charge-off timeliness, and the CECL methodology down to the qualitative factors.
Publicly, watch nonperforming loans to loans alongside the allowance to loans. The median bank reported 0.47% nonperforming against a 1.19% allowance, and the cycle is benign: the bottom quartile of the nonperforming distribution is 0.12%, and among banks under $100 million a full quarter report none at all. The two distributions cover slightly different populations, 4,203 and 4,205 banks, so implied coverage at the median is an approximation, not a bank-level statistic.
The pairing that starts a provisioning conversation is a nonperforming ratio at or above the 1.17% top quartile with an allowance at or below the 0.98% bottom quartile. Separately neither is remarkable. Together they say the allowance was not rebuilt as the book deteriorated, which is an earnings finding as much as a credit one. The largest banks sit in the opposite corner, 0.85% median nonperforming against a 1.44% median allowance, which is consumer exposure carrying a lifetime loss estimate, not distress. See also nonperforming loan benchmarks and the Texas ratio.
M: management
There is no public proxy for this one. That is not a gap in the data, it is the structure of the rating, and it is why every attempt to reverse engineer a composite from call report ratios fails.
What gets assessed is the control environment: whether directors challenge management in the minutes or merely ratify, whether the board package reports against limits or only against results, what happened the last time a limit was breached, internal audit and loan review and their reporting lines, key person depth, and the response to prior matters requiring attention. That last item is disproportionate. An unresolved prior finding reads as a governance failure rather than a technical one, and it moves management before it moves the component the finding came from.
One part of this is visible to you in advance: the accuracy of your own regulatory reporting. An amended call report, or internal reports that do not tie to what was filed, is a management finding that costs nothing to prevent and is expensive to explain. To see your numbers the way an off-site system sees them, start with how to read a call report and the call report versus UBPR distinction.
E: earnings
Earnings are the first defense against loss and the main source of capital, so the component is graded on sufficiency, quality, and sustainability rather than on the headline number. Examiners test how much of the quarter is nonrecurring, whether earnings were propped up by an under-provisioned allowance, and whether the margin depends on funding that reprices faster than the assets.
Publicly this is return on assets and the efficiency ratio. The median bank posted a 1.23% ROA, middle half 0.83% to 1.65%. The old rule of thumb that 1% ROA is satisfactory has quietly gone stale: 1.00% is below today's median and above the bottom quartile, a mediocre result rather than a passing grade. The efficiency ratio median was 61.81%, quartiles 53.08% and 72.18%.
The band detail is what a small bank needs walking in. Under $100 million, bottom quartile ROA is 0.46% and the median efficiency ratio 70.24%, against 1.07% and 56.56% in the $3 billion to $10 billion band. That is fixed cost, not management quality. The comparison that belongs in the room is your band, and within it the trend and the composition of revenue: 0.60% on a granular recurring base is a different conversation from 1.10% on securities gains. Distributions in ROA benchmarks and efficiency ratio benchmarks.
L: liquidity
Examiner expectations have moved further here than anywhere else since 2023, and the loans to deposits ratio is the least interesting thing in the file. What gets tested is funding structure and whether the contingency plan is real: unpledged securities as a share of the portfolio, collateral already posted at the Federal Home Loan Bank, whether discount window access has actually been tested rather than merely documented, whether the funding plan has named triggers with named owners, and deposit concentration by depositor and by category.
| Bank size (total assets) | Banks | Bottom quartile | Median | Top quartile |
|---|---|---|---|---|
| All banks | 4,238 | 65.73% | 80.16% | 90.94% |
| Under $100M | 543 | 45.12% | 65.63% | 82.79% |
| $100M - $300M | 1,214 | 61.86% | 76.41% | 88.46% |
| $300M - $1B | 1,433 | 69.41% | 81.66% | 91.49% |
| $1B - $3B | 626 | 74.35% | 85.35% | 94.92% |
| $3B - $10B | 264 | 78.70% | 88.31% | 94.63% |
| $10B - $100B | 126 | 75.50% | 84.80% | 90.78% |
| Over $100B | 32 | 59.68% | 74.86% | 82.62% |
Source: BankingLens, computed from FFIEC call reports for the quarter ending June 30, 2026. Quartiles are calculated across the full population of filing banks in each size band, not sampled. Non-insured non-deposit trust companies are left out of every statistic: they take no deposits and make no loans, so a margin or a funding cost computed for them has no meaning.
The brokered distribution surprises people. Across all banks the median brokered share is 0.00%: more than half of US banks carry none at all, and the 75th percentile is only 4.32%. Under $100 million, even the top quartile is 0.00%. Usage concentrates in the larger bands, where the $10 billion to $100 billion group runs a 3.69% median on a 0.06% bottom quartile. The point is not that brokered funding is bad. It is that a $400 million bank carrying 12% brokered is outside its peers, whose bottom quartile is 0.00% and top quartile 4.15%, and it will be asked to justify that against its contingency plan rather than its cost of funds.
Uninsured deposits are reported by only 998 banks, effectively the larger ones, so there is no industry-wide figure to quote. Among reporters the median uninsured share is 32.08%, quartiles 23.76% and 42.10%, rising to a 44.87% median above $100 billion. The $300 million to $1 billion row holds just 11 banks and is not representative of that band. If your uninsured share is top quartile, expect the exam to move quickly from the ratio to the depositor concentration list. More on the denominator in loan to deposit benchmarks.
S: sensitivity to market risk
For nearly every community bank this means interest rate risk, and it is graded on the modeling as much as the exposure. Examiners want economic value of equity and net interest income simulations across a range of shocks, documented assumptions, particularly non-maturity deposit decay rates and betas, validation by someone who does not run the model, and evidence of what the bank did the last time a board limit was approached. Real exposure with a well-governed model rates better than modest exposure with a spreadsheet nobody has validated since 2019.
The visible piece is accumulated other comprehensive income as a share of equity, the mark already taken on the available-for-sale book. The median bank sits at -4.95% of equity, the bottom quartile at -13.05%. Read that as a floor on rate exposure, not a measure of it: AOCI says nothing about held-to-maturity, where the same losses sit unrecognized, and nothing about long-duration fixed-rate loans funded with deposits that reprice immediately. The banks that failed in 2023 did not have unusual AOCI. They had unusual HTM books and unusual depositors. Distribution in bank unrealized losses and AOCI.
How to use this before an exam
Run every ratio above against your own asset band rather than the industry, and write down the ones where you sit in the tail. Those are the questions you will be asked, and there are only four per metric: why is it off peer, what changed since the last examination, what is your board-approved limit and did you breach it, and what did you do about the prior findings here. A quantified answer to an off-peer ratio is not defensive. It is the clearest available evidence that management understands its own balance sheet, the one component with no public number attached to it.
Frequently asked questions
What do the six CAMELS components stand for?
Capital adequacy, Asset quality, Management, Earnings, Liquidity, and Sensitivity to market risk. Each is rated 1 through 5, with 1 the strongest, and the examination also assigns a composite rating on the same scale. The composite is not an average of the six; it is an examiner judgment informed by them, and a single weak component can drive it.
Is a bank's CAMELS rating public?
No. CAMELS component and composite ratings are confidential supervisory information. They belong to the supervisory agency rather than to the bank, and the bank is not permitted to disclose them publicly. Enforcement actions such as consent orders are public, and capital ratios are public through the call report, but the rating itself never is.
Can you figure out a bank's CAMELS rating from its call report?
No, and attempts to do it fail for a structural reason: the Management component has no public proxy at all, and it is the component most likely to cap a composite. Public data tells you where a bank sits in its peer distribution on capital, credit, earnings, liquidity, and rate risk, which is where examiners look first. It does not tell you what they found in the loan files or the board minutes.
What is a good CAMELS composite rating?
A composite 1 or 2 is satisfactory and is what the large majority of banks carry. A 3 signals supervisory concern and brings more frequent contact, restrictions on expedited application processing, and pressure on growth. A 4 or 5 means the bank is in unsafe or unsound condition and is generally accompanied by a formal enforcement action.
Which call report numbers line up with each CAMELS component?
Capital tracks the tier 1 leverage ratio, median 10.96% in the quarter ending June 30, 2026, and equity to assets, median 10.56%. Asset quality tracks nonperforming loans to loans, median 0.47%, and the allowance to loans, median 1.19%. Earnings track return on assets, median 1.23%, and the efficiency ratio, median 61.81%. Liquidity tracks loans to deposits, median 80.16%, the brokered share of deposits, median 0.00%, and the uninsured share among the 998 banks that report it, median 32.08%. Management has no public proxy.
What tier 1 leverage ratio do examiners expect?
Prompt corrective action sets 5% tier 1 leverage as the well capitalized threshold, and the community bank leverage ratio framework sets its election level above 9%. Neither is where banks actually operate: the median US bank ran 10.96% in the quarter ending June 30, 2026, with the middle half between 9.70% and 13.00%. Examiners judge the level against the risk in the balance sheet, not against the statutory minimum.
Where these numbers come from
Every figure on this page is computed from the FFIEC call reports that all 4,296 FDIC-insured banks file each quarter, for the quarter ending June 30, 2026. The quartiles above are computed across 4,238 of those 4,296 filings: the other 58 are non-insured non-deposit trust companies, which take no deposits and make no loans, so a margin or a funding cost computed for one has no meaning, and they are left out of every statistic on this site. Nothing here is modeled, estimated, or sampled, and nothing on this page is drawn from any supervisory source. You can run every ratio above for a specific bank against its own asset band and FFIEC peer group on the BankingLens dashboard, or pull the underlying institutions on Bank Peer Intel. Plans start at $29 a month (pricing).
Related reading
- What is a good tier 1 leverage ratio for a bank?
- What is a normal nonperforming loan ratio for a bank?
- What is a good ROA for a bank?
- What is a good efficiency ratio for a bank?
- What is a good loan to deposit ratio for a bank?
- Bank unrealized losses and AOCI
- What CRE concentration ratio triggers regulatory scrutiny?
- How to build a bank peer group