Guide 3’s uninsured-deposit line tied the 10-K to the call report
The SEC’s September 2019 rewrite of bank statistical disclosure added a total for uninsured deposits, later tied to the regulatory reporting method. One estimate then served investors, supervisors and the FDIC, so its calculation became a data governance problem.
Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.
Two-minute summaryThe dek and the first sentence of each section, quoted from the article
The SEC’s September 2019 rewrite of bank statistical disclosure added a total for uninsured deposits, later tied to the regulatory reporting method. One estimate then served investors, supervisors and the FDIC, so its calculation became a data governance problem.
- The week On September 17, 2019, the Securities and Exchange Commission proposed rules to update the statistical disclosures that bank and savings and loan registrants give investors.
- What the proposal changed Most of the proposal was codification.
- Defining the uninsured deposit figure The deposits item carried the proposal’s most consequential change.
- One figure, three users Once the methods were aligned, a bank’s estimate of uninsured deposits served three audiences with different purposes.
- Where the calculation goes wrong An uninsured deposit estimate looks like a single number.
- Where the argument is weakest The argument here is that a one-line addition to bank disclosure tied the investor figure to the regulatory figure, and so made the calculation behind it a matter for data governance as much as for finance.
- With hindsight The SEC adopted the final rules in September 2020, in a release published in the Federal Register on October 16, 2020, at 85 FR 66108.
The week
On September 17, 2019, the Securities and Exchange Commission proposed rules to update the statistical disclosures that bank and savings and loan registrants give investors. The proposal, Release No. 33-10688, would replace Industry Guide 3, Statistical Disclosure by Bank Holding Companies, with a new Subpart 1400 of Regulation S-K. It would also eliminate disclosures that overlapped with Commission rules, U.S. generally accepted accounting principles or International Financial Reporting Standards. The SEC set a 60-day comment period following publication in the Federal Register. The proposal appeared there on October 3, 2019, at 84 FR 52936.
Guide 3 was old and informal. It was first published in 1976 as a convenient reference to the statistical disclosures the Division of Corporation Finance’s staff sought from bank holding companies. When the Commission published it, it said the guides were not Commission rules and did not bear its official approval. They represented the staff’s policies and practices. Guide 3 called for disclosure in seven areas: the distribution of assets, liabilities and equity with interest rates and differentials; investment portfolios; loan portfolios; loan loss experience; deposits; return on equity and assets; and short-term borrowings. Guide 3 was last amended in 1986, before the accounting standards that now govern how banks classify investment securities. On March 1, 2017, the Commission published a request for comment on possible changes to it, asking among other things whether disclosures that overlapped with U.S. GAAP, IFRS or other regulatory disclosure regimes should be dropped.
The proposal would write the disclosure requirements into regulation and apply them by name to bank holding companies, banks, savings and loan holding companies and savings and loan associations. According to the SEC’s fact sheet, the new rules would require disclosure about average balances with related interest income, expense and rates; the weighted average yield of debt securities by maturity; a maturity analysis of loans, split between predetermined and floating or adjustable rates; an allocation of the allowance for credit losses with certain credit ratios; and information about deposits, including amounts that are uninsured.
That last item was new in an important way. Guide 3 had asked for the amount of domestic time deposits of $100,000 or more by maturity. The proposal would instead require the amount of time deposits in uninsured accounts by maturity, with no fixed dollar threshold. And it would add a new line: the amount of uninsured deposits as of the end of each reported period. For U.S. federally insured institutions, the proposal defined uninsured deposits as individual deposits in U.S. offices above the FDIC insurance limit, plus investment products such as mutual funds, annuities and life insurance policies.
The proposal also changed reporting periods. Guide 3 had called for five years of loan portfolio and loan loss data and three years of other information. The proposal would align the periods with the financial statements, keeping five years of credit ratios only in initial registrations. It would keep Guide 3’s preference for daily averages, allowing weekly or month-end averages where daily data would involve undue burden or expense, provided they were representative and the basis was disclosed.
The proposal explained why fewer years of history would be enough. When Guide 3 was issued, only two years of financial statements were required, and commenters at the time had said five years of history would be extremely difficult to obtain where detailed breakdowns were requested. In 2019 the SEC pointed instead to EDGAR, its electronic filing system: the history a registrant no longer had to repeat would generally be accessible through its prior filings. The change was small in itself. It showed how the Commission now assumed past disclosures were stored, searchable and comparable.
The September 2019 proposal looked like housekeeping: an old staff guide turned into a regulation, with overlaps trimmed. One new line in it mattered more than the rest. The proposed disclosure of total uninsured deposits, and the final rule’s instruction that the figure follow the methodology banks use for regulatory reports, joined a bank’s annual report to its call report. After the bank failures of March 2023, the FDIC made that estimate the base for a special assessment, and found some banks had reported it in a way that was inconsistent with the call report instructions. This article takes the argument in five parts: what the proposal changed, how the uninsured deposit figure was defined, the three users of that one figure, where the calculation goes wrong, and where the argument is weakest.
What the proposal changed
Most of the proposal was codification. Guide 3 had shaped bank disclosure since 1976 without being a rule. Banks provided it because the staff expected it, and other registrants with lending and deposit activities, such as savings and loan holding companies, provided it to the extent it applied. The proposal made the scope explicit and the requirements binding.
From staff practice to regulation
A staff guide and a regulation differ in how they are read. Under a guide, a registrant and the staff could discuss whether a given table was needed. Under a rule, the requirement is the text. The proposal kept flexibility where it mattered to preparers. Registrants could choose other deposit categories if those described their deposits better, and the disclosures did not have to sit in a particular part of the filing. But the core items became obligations with defined terms.
For the people who produce these numbers, that shift changes the standard of evidence. A disclosure required by rule needs a documented method, a repeatable process and an owner who can explain any change from one period to the next. A table produced because the staff had always expected it can survive on institutional memory. A table required by regulation should not.
Overlaps removed
Commenters on the 2017 request had said that Guide 3 disclosures overlapping with Commission rules, GAAP and IFRS should go. The proposal removed several. It did not carry forward return on equity and assets or the short-term borrowings disclosures into the new subpart. It also proposed changes to Article 9 of Regulation S-X, which governs bank holding company financial statements. Some commenters had pointed to overlap with the banking agencies’ public regulatory reports. Several noted that the primary purpose of that reporting differs from the Commission’s disclosure objectives, and one said it may be of limited use to investors given its volume and level of detail. The proposal did not replace SEC disclosure with a reference to the call report.
Periods and averages
Two technical choices affected data systems directly. First, aligning reporting periods with the financial statements meant fewer years of history for most items. Commenters on the 1976 Guide 3 Release had said five years of history would be extremely difficult to obtain where detailed breakdowns were requested. Second, the proposal kept daily averages as the default for average balances, with weekly or month-end averages as a fallback where daily data would be unduly burdensome.
The fallback had an interesting history. Guide 3’s general instruction on foreign registrants said that, in evaluating a claim that compiling information such as daily averages would be an undue burden, the staff would consider among other factors the registrant’s electronic data processing capacity. In 1976 that was a real constraint. By 2019, a bank that could not produce daily average balances from its core systems had a data problem, not a computing one.
The new items
The proposal added or reshaped several disclosures. It asked for the weighted average yield of each category of debt securities not carried at fair value through earnings, by maturity, and an explanation of how the yield was calculated. It asked for loans by maturity and, for loans due after one year, the split between predetermined and floating or adjustable rates. It defined four credit ratios: the allowance for credit losses to total loans, nonaccrual loans to total loans, the allowance to nonaccrual loans, and net charge-offs to average loans. Each ratio would be disclosed with each component of its calculation, and with a discussion of the factors behind material changes.
Requiring each component alongside each ratio is a small instruction with a large effect. It forces the numerator and denominator into the filing, so a reader can check the arithmetic and compare definitions across banks. It also means the components must agree with other figures elsewhere in the filing. That is a reconciliation problem, and reconciliation is where data quality becomes visible.
History moved to the filing archive
The shorter reporting periods rested on an assumption about data, stated in the proposal. Historical information dropped from new filings would generally remain accessible in a registrant’s prior filings on EDGAR. For an analyst, that is true. For the bank, it means each year’s published table becomes a permanent record that later tables will be compared against. A restated average balance or a changed loan category is visible to anyone who lines up two filings.
That raises the value of keeping the logic behind each table versioned. When a category definition changes, the bank should be able to show which periods used which definition, and why. Without that record, a reader who finds a break in a series has no way to tell a change in the business from a change in the method.
Defining the uninsured deposit figure
The deposits item carried the proposal’s most consequential change. The proposal explained why. Disclosures about significant deposits from a small number of depositors, or about certain types of deposits such as uninsured deposits, could give investors insight into a registrant’s reliance on particular funding sources and the risks related to them. Insured retail deposits can be a reliable funding source and help mitigate liquidity risk. Uninsured deposits are a different kind of funding.
From a threshold to a definition
Guide 3 used a fixed dollar threshold: time deposits of $100,000 or more. A fixed threshold drifts out of date as the insurance limit changes, and it measures account size rather than insurance status. The proposal replaced it with insurance status. Time deposits would be reported in two groups, U.S. time deposits above the FDIC limit and time deposits otherwise uninsured, each by remaining maturity: three months or less, over three through six months, over six through 12 months, and over 12 months. The SEC noted that having no defined dollar threshold would let the requirement accommodate changes in the insurance limit.
The total
The proposal then asked for something Guide 3 never had: the total amount of uninsured deposits at the end of each reported period. That figure covers all deposit types, time deposits among them. It speaks directly to the question investors would later ask in a crisis: how much of this bank’s funding is not covered by deposit insurance.
What the total does not say
The total is a single period-end amount. It does not show how concentrated the uninsured balances are, which industries or counterparties hold them, how long they have stayed, or how quickly they could leave. Item 1406 asks for time deposits that are uninsured by remaining maturity, which gives some sense of contractual term for that slice. It asks nothing comparable for demand or savings balances, which can leave at once.
Those gaps are not flaws in the rule. A disclosure item has to be comparable across hundreds of registrants, and concentration or behavior data would be hard to standardize. But the gaps shape what a bank should track internally. The reported total is a floor for analysis, not a measure of run risk. A bank that manages only to the reported number knows how much is uninsured. It does not know who would move first.
The proposed definition included investment products such as mutual funds, annuities and life insurance policies. Those products are not deposits in the ordinary sense, and including them would have mixed two different things. The final rule, adopted in September 2020, dropped the express reference. The SEC said banks already evaluate whether a product is subject to deposit insurance, so more guidance was unnecessary.
Estimates and the regulatory methodology
The final rule made one more change that is central to this article. In response to commenters’ concerns about overlapping regimes, it said registrants should determine uninsured deposits using the same methodologies and assumptions they use for regulatory reporting, to the extent applicable. Where a precise measure is not reasonably practicable, a registrant may disclose an estimate, saying that it is an estimate, and the estimate must use the same methodologies and assumptions as its regulatory reporting. The SEC said this would reduce complexity, better align the rules with U.S. bank regulatory reporting and reduce cost.
That instruction is what tied the two reports together. The 10-K figure was no longer an independent calculation that might differ from what the bank told its regulators. It was meant to be the same calculation. Any error in the regulatory estimate would flow into the investor disclosure, and the other way round.
Foreign registrants and branches
Foreign banking registrants were handled differently. The proposal and the final rule both let them disclose the definition of uninsured deposits appropriate to their country of domicile, since insurance regimes differ. The final rule’s pointer to regulatory reporting methods gave them a parameter: use the method your own regulators expect, to the extent applicable. The SEC said that should reduce some of the comparability concerns for registrants operating in different jurisdictions.
On the U.S. regulatory side, the FDIC’s 2023 rule drew on two reports. Insured banks report estimated uninsured deposits on the call report. U.S. branches and agencies of foreign banks report on the FFIEC 002. A foreign group with U.S. operations can therefore face three definitions at once: its home regime for its own disclosure, the U.S. call report for an insured subsidiary, and the FFIEC 002 for a branch. Keeping the three apart, and labeling which one appears where, is part of the same governance job.
One figure, three users
Once the methods were aligned, a bank’s estimate of uninsured deposits served three audiences with different purposes. After 2023, the FDIC also used it to bill banks.
Investors
For investors, the figure is a measure of funding risk. A bank with a high share of uninsured deposits depends on depositors who have more reason to move their money quickly under stress. The proposal’s reasoning anticipated this use. Under Subpart 1400, the figure appears in the annual report alongside deposit averages and rates. The final rule did not require the disclosures to be presented in a machine-readable format unless the registrant chose to include them in the financial statements, so unless a bank placed the disclosures in its financial statements, the figure would appear as untagged text and tables in the filing.
Read by people, not machines
That choice deserves more attention than it received. Several commenters had argued against requiring the new disclosures in a structured format. Two noted that registrants may provide the items in Management’s Discussion and Analysis, so tagging them would leave some of that section in XBRL and some not. A few commenters wanted all Subpart 1400 data tagged in Inline XBRL regardless of location, so the figures would be consistently machine-readable across registrants. The Commission left the items outside the financial statements unless a registrant chose otherwise. Outside the statements, they were neither audited nor subject to the XBRL requirement.
The consequence is that the uninsured deposit figure in a 10-K reaches users through whatever table design the bank chooses. An analyst comparing banks has to find and transcribe it. Where the figure sits outside the financial statements, no audit opinion covers it and no tagged value can be checked automatically against the call report, so a bank’s own controls are the main defense against a transcription or version error between the two.
Supervisors
For the banking agencies, the call report’s estimate of uninsured deposits is a safety and soundness indicator. The FDIC said so in 2023: the amount of estimated uninsured deposits reported on the call report is monitored as one of many indicators of safety and soundness, and its accuracy is of the utmost importance. Supervisors use it to judge how stable a bank’s funding is.
The deposit insurance fund
For the FDIC, the same estimate also feeds the deposit insurance fund. The FDIC explained that reported estimated uninsured deposits are used to determine estimated insured deposits when it calculates the fund’s reserve ratio, the ratio of the fund balance to insured deposits. In 2023 it became the base for a special assessment, discussed in the hindsight section below.
What that means for the data
A number with three users needs one definition, one calculation and one owner. If the finance team that prepares the 10-K computes uninsured deposits separately from the regulatory reporting team that prepares the call report, the two figures can drift. The SEC’s final rule made drift a compliance issue, because the 10-K estimate must use the regulatory methodology. A bank that wants the figure to be right in every place it appears would build it once, from the deposit system of record, with documented logic, and publish the same result to every consumer.
The practical design follows. A governed calculation takes account-level data from the core deposit systems, applies the insurance-status logic, aggregates, and produces the period-end figure with an audit trail. The regulatory report and the SEC filing both draw from that output. A reconciliation confirms they match, and a variance analysis explains any period-to-period change. The model, its assumptions and its changes are documented and approved. Neither rule requires this design in that form. It is the architecture that makes the two rules’ alignment hold.
Lineage and evidence
Lineage matters because the figure is an estimate. An auditor, examiner or investor asking how it was produced needs to see which source data fed it, which rules were applied and which manual adjustments, if any, were made. Each adjustment is a point of risk. The FDIC’s 2023 findings, described below, concerned exactly the kind of judgment that can enter as an adjustment: deciding to reduce uninsured deposits for collateral or to exclude certain balances. A lineage record shows whether such a step happened, who approved it and on what basis.
Who owns the number
The figure crosses at least five teams. Deposit operations owns the account data and the ownership coding. Regulatory reporting owns the call report schedule and its instructions. The controller’s or SEC reporting team owns the 10-K. Treasury, through asset-liability management, uses the figure to manage liquidity. A data office, where one exists, owns the shared definitions and lineage. Internal audit tests the result.
A workable split gives one team the calculation and makes the others consumers. In our view the natural owner is regulatory reporting, because the SEC rule now points to the regulatory method. The SEC reporting team then certifies that the 10-K figure equals the regulatory figure or explains why it differs. Treasury receives the same output at a higher frequency for its own monitoring. The disclosure committee, which already reviews the 10-K, adds one standing question: does the uninsured deposit total in this draft match the call report for the same date?
The sequencing matters at quarter end. The call report and the 10-K or 10-Q follow separate calendars and can be prepared by different people under different deadlines. If the SEC team drafts its tables before regulatory reporting finalizes its schedule, the two can diverge by a late adjustment. A simple rule closes the gap: the SEC disclosure pulls the figure from the locked regulatory output, never from a separate extract.
Where the calculation goes wrong
An uninsured deposit estimate looks like a single number. It rests on several layers of data and judgment, and each can fail.
Rules applied to the wrong data
The FDIC’s November 2023 final rule recorded two errors it had observed in how some institutions reported or amended estimated uninsured deposits. Some incorrectly reduced uninsured deposits to the extent the deposits were collateralized by pledged assets. The FDIC said that was wrong, because the existence of collateral has no bearing on the portion of a deposit covered by federal deposit insurance. Others incorrectly reduced the amount by excluding certain intercompany deposit balances. Both errors lowered the reported uninsured figure.
Public funds and pledged collateral
The collateral error has a predictable source. Banks that hold deposits from states and political subdivisions may be required by state law to secure them with pledged assets. The FDIC noted that the call report already collects one such category, preferred deposits, defined as uninsured deposits of states and political subdivisions that are secured or collateralized as required under state law. The definition itself says they are uninsured. A treasury or public-funds team that thinks of those balances as safe, because they are collateralized, may expect them to drop out of the uninsured total. They do not.
The FDIC also explained why it did not treat collateralized deposits differently for the special assessment. In its view, collateral does not fully mitigate run risk, and collateral may not always be sufficient to cover the full amount. For a bank with a large public-funds book, that is a reason to document the treatment of each public deposit category explicitly, with a citation to the call report instruction, and to have the public-funds team review it.
Those are rule errors, not arithmetic errors. Someone decided that collateral or affiliation changed insurance status, and the calculation applied that decision. A governed calculation with documented, reviewed rules would surface such a decision as a change to logic that needs approval, not as an unreviewed adjustment inside a spreadsheet.
Inconsistent interpretation across banks
The FDIC also received comment letters saying banks may be reporting uninsured deposits differently or inconsistently, and one saying some banks were confused about whether to include collateralized deposits. On July 24, 2023, it issued a Financial Institution Letter on estimated uninsured deposits reporting expectations. The letter reiterated longstanding instructions, said each institution is responsible for the accuracy of its call report, and said institutions that had reported incorrectly should amend their call reports.
Inconsistency across banks matters for the SEC disclosure too. Investors compare banks. If two banks with similar deposit bases compute the figure differently, the comparison misleads. The SEC’s decision to anchor the figure to the regulatory methodology was meant to reduce that risk. It works only as well as banks apply the regulatory instructions.
Affiliates and intercompany balances
The intercompany error was a judgment about affiliation. Commenters on the FDIC’s 2023 proposal argued that deposits of affiliates, subsidiaries or a parent holding company are stable and present little run risk, because entities within one organization are unlikely to withdraw funds in a crisis. The FDIC disagreed. It said there was no clear evidence that intercompany deposits are more stable than other deposits, that governance and creditor behavior can differ between entities in the same group, and that an affiliated entity’s deposits are insured to the same extent as an unaffiliated entity’s.
For a holding company with several subsidiaries, the practical point is that an organizational chart is not a deposit insurance rule. Each affiliate balance is a depositor relationship like any other, and the calculation should treat it that way unless the instructions say otherwise. Where a bank believes a different treatment is justified, the place to argue it is with the regulator, in writing, before the figure is filed.
Data quality at the account level
Below the rules sit the data. An estimate of insurance status depends on what the deposit systems record about each account and depositor. Gaps in customer identification, inconsistent ownership coding, accounts held in several systems after a merger, and balances booked in subsidiaries or foreign offices all affect the result. A bank with one modern core system and a simple deposit base has a short list of these issues. A bank that has grown through acquisition, with several cores and sweep or affiliate arrangements, has a long one.
Those are familiar data management problems. What the 2019 proposal and its final rule added was a public consequence. An error that might once have affected only an internal liquidity report now appeared in an SEC filing, a call report and, from 2023, an assessment invoice.
Timing
The figure is reported as of a period end. Deposits move daily. A period-end estimate that is correct on its date can still mislead if balances swing sharply around quarter end. The proposal’s preference for daily averages applied to average balance disclosures, not to the uninsured total. A bank that wants to understand its own exposure, as opposed to reporting it, would track uninsured balances more often than the reports require.
Monthly or weekly internal estimates also make the quarter-end figure easier to defend. If the reported number moves sharply from one quarter to the next, the bank can show the path between them and separate deposit flows from method changes. A figure that is calculated only four times a year has no such context.
Controls an examiner or auditor would expect
The documentation follows from the risks. A written methodology tied to the call report instructions. A list of every rule and assumption, with an owner and approval date. A reconciliation between the call report figure and the SEC disclosure each period. Variance analysis with explanations. Evidence of review for any manual adjustment. A record of amendments filed and why. Those are the artifacts that would let a bank answer an examiner’s or auditor’s question quickly, and they are the same artifacts that would have caught the errors the FDIC described.
Testing the estimate before it is filed
Controls on paper do not prove the figure is right. A few tests do. The first is re-performance: an independent analyst recalculates the total from the account-level extract using the documented rules and gets the same answer. The second is a sample: pick accounts near the insurance limit, accounts with several owners and accounts held through trusts or agents, and check that the coding matches the documents on file. The third is a movement test: compare the change in the uninsured total with the change in large balances over the quarter, and require an explanation when they diverge.
Three metrics tell a chief financial officer or chief data officer whether the process is under control. One is the number and size of manual adjustments each quarter. Another is the share of deposit balances whose ownership category could not be determined automatically. The last is the gap, if any, between the call report figure and the SEC figure before reconciliation. None should be a surprise at quarter end, and all three should trend down as the data improves.
The effort scales with the bank. A regional holding company with several charters and a public-funds book might run these tests every quarter with a dedicated team. A single-charter community bank might run them once a year and rely on simpler rules the rest of the time. Either way, the test results belong in the evidence file the disclosure committee and the auditors see.
Where the argument is weakest
The argument here is that a one-line addition to bank disclosure tied the investor figure to the regulatory figure, and so made the calculation behind it a matter for data governance as much as for finance. Some objections deserve a direct answer.
The proposal did not cause the link
Banks report estimated uninsured deposits to the banking agencies on the call report, and the SEC did not invent the measure. That is true. What the proposal and final rule did was move the figure into the SEC filing as a required item and specify that it use the regulatory methodology. Before, an investor might find the figure elsewhere or not at all. After, it was part of every covered annual report, with the same basis as the regulatory number.
Hindsight makes one line look prophetic
Reading the 2019 proposal through 2023 risks overstating what anyone foresaw. The SEC’s stated reason was transparency about funding and liquidity risk, not a prediction of bank runs. The argument here does not depend on foresight. It depends on the mechanics: a figure defined in 2019 and 2020 with a single methodology became, after 2023, the basis for supervisory attention and an assessment. The governance burden followed from the uses, whatever anyone expected at the time.
The SEC expected little effort
The Commission’s own burden estimate cuts against the argument. For the final rules, it estimated that the burden of an affected annual report would rise by three burden hours per year, and an affected registration or offering statement by one hour. If the figure is a governed data product, why did the regulator expect so little work?
The answer is that the SEC measured the cost of disclosure, not the cost of getting the number right. A bank that already produced a sound regulatory estimate could copy it into the 10-K in minutes. The hours that matter sit upstream, in the deposit data, the insurance logic and the review of judgment calls. Those costs were already owed to the banking regulators. The SEC rule raised the consequence of getting them wrong.
Why not cite the call report instead
Some commenters on the 2017 request had pointed to overlap with the banking agencies’ public regulatory reports. Other commenters said the banking agencies’ reporting serves a different primary purpose, and one said it may be of limited use to investors given its volume and level of detail. The Commission kept a separate SEC disclosure. That decision is a second reason the figure needs one owner. Had the SEC simply cross-referenced the call report, there would be one number in one place. Instead, there are two presentations of the same estimate, prepared on different calendars for different readers, and a bank that wants them to agree has to make them agree.
The figure is still an estimate
A skeptic could say that no amount of governance makes an estimate precise, so the effort has limits. The final rule accepted estimates explicitly. Governance does not remove estimation. It makes the estimate consistent, explainable and correctable, which is what all three users need.
Differences by registrant
Subpart 1400 applies to banks, bank holding companies, savings and loan associations and savings and loan holding companies that file with the SEC. A large bank holding company with several insured subsidiaries and foreign offices faces the full set of issues: multiple systems, intercompany balances and foreign deposits, for which foreign registrants must disclose their own definition of uninsured deposits. A small savings and loan holding company with one subsidiary bank has a simpler calculation but may have fewer people to review it. Credit unions are outside the list. Insured banks that do not file with the SEC may still report the regulatory estimate, so the call report half of the argument can apply to them even though the 10-K half does not. Size matters here too. According to the FDIC, institutions with less than $1 billion in total assets as of June 30, 2021, were not required to report estimated uninsured deposits on the call report for December 31, 2022. A small SEC registrant below that line still owes the 10-K figure, and the rule’s reference to regulatory methodology applies only to the extent applicable. Such a bank has to document its own method with less outside structure to lean on.
Across all of them, the useful test is the same: could the bank show, for any period, how its uninsured deposit figure was produced, that it matches across every report where it appears, and why it changed?
With hindsight
The SEC adopted the final rules in September 2020, in a release published in the Federal Register on October 16, 2020, at 85 FR 66108. They took effect November 16, 2020, and registrants had to apply them for the first fiscal year ending on or after December 15, 2021. The final text of Item 1406(e), as it stands in the Code of Federal Regulations today, requires the amount of uninsured deposits at the end of each reported period and says any estimate must use the same methodologies and assumptions as the registrant’s regulatory reporting.
In March 2023, Silicon Valley Bank and Signature Bank failed. The Federal Reserve’s review of its supervision of Silicon Valley Bank, released with a cover letter dated April 28, 2023, said the bank was an outlier because of its highly concentrated business model, interest rate risk and high level of reliance on uninsured deposits. It said deposit outflows were over $40 billion on March 9 and that management expected $100 billion more the next day. The letter said the Federal Reserve should re-evaluate the stability of uninsured deposits in its standardized liquidity rules and in firms’ internal liquidity stress tests.
The FDIC then turned the regulatory estimate into a bill. Its final rule, published November 29, 2023, imposed a $16.3 billion special assessment to recover the deposit insurance fund’s loss from protecting uninsured depositors at the two banks. The base was each institution’s estimated uninsured deposits reported for the quarter ended December 31, 2022, excluding the first $5 billion, at a quarterly rate of 3.36 basis points over eight quarters. The FDIC said the amounts collected could change because of corrective amendments to reported uninsured deposits, described the reporting errors it had found, and said it was conducting an Assessment Reporting Review of how institutions report estimated uninsured deposits. Of the $18.7 billion in estimated losses at the two banks, it attributed $16.3 billion to the protection of uninsured depositors. Because of the $5 billion deduction, the special assessment did not apply to banking organizations with total assets under $5 billion.
The assessment ran into 2026. An interim final rule published December 19, 2025, said the loss estimate attributable to uninsured depositors stood at $16.7 billion as of September 30, 2025, and that six quarterly collections had brought in $12.7 billion. It reduced the rate for the eighth collection quarter, with an invoice payment date of March 30, 2026, to 2.97 basis points to avoid overcollecting. It also provided for offsets to regular assessments if collections exceeded losses after the litigation between the FDIC and SVB Financial Trust is resolved and again when the receiverships end.
What held up was the 2019 logic: uninsured deposits told investors something real about funding risk. What the proposal could not show was how much weight the figure would carry. By 2024, the same estimate appeared in SEC filings, fed supervisory monitoring, set the deposit insurance fund’s reserve ratio inputs and determined assessment invoices. A calculation that had once been a reporting detail had become a governed data product, whether or not each bank treated it as one.
Spotted an error? Tell us and we will check it and correct the article where needed.