What Supervisors See—and Change
This chapter focuses on the information side of U.S. banking supervision: how an examiner’s scrutiny can change the reported capital signal that later supports enforcement or closure. The reader’s practical task is to follow one potential loss from an illiquid loan, through a provision and a revised financial statement, to book equity and a supervisory rating—without confusing a loss being recognized with a loss being created.
A bank’s capital is not a pile of cash sitting in a separate room. In simplified terms, it is the value left for owners after the bank’s assets are measured and its obligations are subtracted. That measurement becomes difficult when the assets are loans whose repayment depends on uncertain future events. A borrower may still be making payments today while the loan’s likely recovery value is deteriorating. The bank must decide when that expected loss belongs in its accounts. Supervisors examine that decision.
Why capital and CAMELS become decisive in distress
Two capital measures help explain the supervisory signal. The equity-to-assets ratio compares book equity with the bank’s total assets. The Common Equity Tier 1, or CET1, ratio compares a high-quality form of capital with risk-weighted assets, which give more weight to riskier exposures. Neither measure is a complete statement of economic value. Both are, however, observable ways to classify a bank and compare its condition over time.
These ratios answer slightly different questions. Equity-to-assets asks how large the accounting cushion is relative to everything the bank owns. CET1 asks how that cushion compares with a risk-adjusted measure of the assets that could generate losses. A bank can therefore have the same total-assets ratio as another bank while presenting a different CET1 ratio if their portfolios carry different regulatory weights. The study uses both measures for different purposes, but its dynamic analysis often uses equity-to-assets as a parsimonious measure of book capitalization. That choice keeps the mechanism visible: when recognized losses reduce the asset-side value or retained earnings, the book cushion becomes thinner.
The distinction between “book” and “economic” value matters. A loan can be worth less than its recorded amount even before the loss is fully reflected in book equity. Conversely, a bank can look thinly capitalized under a conservative assumption while still having a chance of recovery if borrowers repay more than expected. Supervisory systems need a rule-bound measure to act on, but that measure is an estimate with consequences, not a direct reading of an invisible physical quantity.
The study of failing banks finds that this measurable signal becomes especially important as distress deepens. In the paper’s all-bank data from 2000 through 2023, institutions with CET1 ratios above 7.5% were typically rated CAMELS 1 or 2. Institutions below 4% were almost always rated 4 or 5. Banks with a CET1 ratio above 5% essentially never failed in the plotted sample, while the probability of closure over the following year approached 70% when equity-to-assets fell below 2%.
These are empirical relationships in the study, not universal legal cutoffs. They show how sharply the supervisory process reacts when a bank’s reported capital becomes very low. Failing banks experienced an average decline of about eight percentage points in equity-to-assets during the five years before failure and ended with an average ratio near 3% immediately before closure. Their average CAMELS rating rose by 2.8 points over the same period, moving from the range associated with ordinary institutions toward ratings of 4 or 5.
CAMELS is an acronym for six dimensions of safety and soundness: capital adequacy, asset quality, management, earnings, liquidity, and sensitivity to market risk. The composite rating runs from 1, the strongest rating, to 5, the weakest. It is therefore tempting to treat CAMELS as an independent expert opinion that confirms or contradicts the capital ratios. The evidence suggests a more conditional relationship.
In troubled banks, the composite rating closely follows the capital signal. The capital component is necessarily connected to observed capitalization, but the relationship also appears in asset quality and earnings, which deteriorate when loan losses are recognized. Management and other components add information, yet they do not erase the importance of hard capital evidence once a bank is near the boundary of failure. When a bank is well capitalized, supervisors have more room to exercise judgment about management and other qualitative conditions. When equity is very low, the room for discretion narrows because a weak capital position is easier to verify and harder to explain away.
This does not make the rating redundant. A capital ratio can reveal the size of the accounting cushion without explaining why it is shrinking or whether management is responding responsibly. Asset quality can show the source of the deterioration; earnings can show whether the bank is generating enough income to rebuild capital; liquidity can reveal whether it can meet near-term demands. The composite rating brings these dimensions together. What changes in distress is the relative weight of judgment: once the capital cushion is nearly gone, a favorable view of management cannot by itself restore solvency.
That asymmetry is central to the broader argument. Supervision is not a contest between a supposedly objective ratio and a wholly subjective rating. It is an interaction. Supervisors interpret several forms of evidence, but in a distressed bank the reported capital measure becomes the anchor around which the composite judgment and formal action organize themselves. A rating downgrade then brings more attention, and more attention can change the financial information that produces the next rating.
From examination to a revised balance sheet
Banks have meaningful discretion in estimating provisions for potential loan losses and deciding when to charge off loans that have already been provisioned. That discretion is not necessarily misconduct. It reflects the difficulty of valuing assets whose cash flows arrive over time. It does mean that two banks with similar loan books can report different timing or amounts of loss recognition, especially while the quality of their borrowers is uncertain.
A provision is an expense recognized against an expected loss; an allowance is the accumulated balance intended to absorb that loss; a charge-off removes an amount judged uncollectible from the books. The precise accounting treatment is more detailed than this, but the direction of the mechanism is important. More conservative estimates generally raise provisions or charge-offs, reduce current income, and leave less equity retained inside the bank. A supervisor does not need to discover a new bad loan on the examination date for the examination to have a financial effect. The examination can change when an existing uncertainty is recorded as a loss.
An on-site examination tests those judgments. An examiner can review loan files, challenge assumptions about repayment, assess the quality of management, and require the bank to record a more conservative allowance or provision. The immediate accounting path is straightforward. A higher provision reduces current income. Lower income reduces retained earnings. Lower retained earnings reduce book equity, unless another source of capital offsets the change. The reported capital ratio can therefore fall even though the underlying loan loss existed before the examination.
The Call Report is important in this process because it is the public financial statement through which banks report their condition to regulators and the market. The study finds that revisions are common enough to matter: from 2007 through 2023, nearly 30% of bank-quarter observations contained a revision to at least one line item. Net income was revised in 4.6% of cases and equity in about 5% of cases. Revisions to a report can affect assets, income, equity, loan-loss provisions, and other indicators that feed into the capital picture.
Timing makes the supervisory connection visible. The probability that a bank revised a previously submitted Call Report rose slightly in the two months before an on-site examination, then increased sharply in the examination month and the two months afterward. Among banks on the rotating schedule of federal and state-led examinations, revisions were more pronounced after federal-led exams, which the paper treats as the stricter side of the comparison. Revisions were also much more likely for banks that were downgraded or eventually failed.
The direction of the revision is more revealing than the fact of revision alone. For banks within one year of failure, the probability of a downward equity revision after an examination was about 30%, compared with 3% for an upward revision. The corresponding figures for net income were 28% downward and 6% upward. Loan-loss provisions were revised upward in about 21% of these cases and downward in only 3%. The pattern is consistent with troubled banks being required to recognize losses that had previously been delayed, understated, or estimated less conservatively.
Consider two banks that have each made loans to a weakening property market. In the first bank, managers and examiners agree that the expected recovery has fallen enough to increase provisions in the current report. In the second, the same deterioration is acknowledged only after a later review. If the loans ultimately produce the same recovery, the first bank will look weaker sooner even though the examination did not cause the market decline. Its earlier recognition may also constrain dividends, asset growth, or new lending sooner. The second bank may appear stronger for longer, but that appearance does not necessarily represent more economic capital.
The revisions also accumulate. In the five years before failure, the equity-to-assets ratio reported in the original Call Reports was higher than the ratio in the final revised submissions. The cumulative difference from revisions amounted to nearly three percentage points of the roughly eight-point decline in equity-to-assets. A small adjustment in any one quarter can therefore become consequential when the same bank is repeatedly remeasured as its condition worsens.
The sequence has an important limit. A supervisor’s examination does not necessarily cause the borrower to default or make the asset lose value. The loss may have been present but uncertain, and the examination may bring it into the accounts sooner. The study’s examination-window pattern alone cannot always distinguish a revision caused by supervisory pressure from a revision that a bank had already planned and merely accelerated after the examination. The comparison of federal- and state-led examinations, and the paper’s later causal design, strengthen the case that at least some of the difference is produced by supervisory strictness.
The revised July 2026 version of the New York Fed staff report estimates this effect directly. Among banks exposed to a federal-led examination early in the Global Financial Crisis, the probability of a downward revision to equity was 1.2 percentage points higher, the probability of a downward revision to net income was 1.3 points higher, and the probability of an upward revision to loan-loss provisions was 0.74 points higher. The estimates are causal only under the paper’s rotating-schedule assumptions, and they describe a restricted crisis sample. Their significance is the mechanism they reveal: stricter supervision changes what the bank must acknowledge, not merely how an unchanged balance sheet is labeled.
This is why “supervision creates losses” is the wrong shorthand. A better formulation is that supervision can create more conservative hard information about losses. That information can lower reported book equity, worsen a CAMELS rating, and move a bank into a capital category with legal consequences. The bank’s economic condition and its reported condition are connected, but they are not the same object.
Once the reported capital signal changes, it is no longer just an accounting detail. It can determine which forms of monitoring, enforcement, or restriction are available under the supervisory system. The next institutional question is how that system responds to the new category—and how much time remains before the bank fails.