SEC filings / tags
The comparability tax
XBRL was meant to make filings machine-comparable. In FY2024, across the 358 bank and thrift holding-company 10-Ks, 18 us-gaap elements are reported by 95% or more of filers. 438 us-gaap elements are reported by exactly one filer. Alongside them sit 5,079 distinct company-extension names, of which 4,713 appear in exactly one filing. A screen built on the shared vocabulary is a screen with about 18 columns.
This is not a style complaint. It is the mechanism behind the finding on the index page: SVB Financial Group tagged the carrying amount of its held-to-maturity book with an element it minted itself, so a screen filtered to us-gaap elements sees the fair value of that book and has nothing to compare it against. The bank drops out of the screen while both numbers stay in the filing.
How many filers share a tag
Each bar counts distinct tags, grouped by the share of the 358 FY2024 bank 10-K filings that report them at least once on the consolidated face.
Of the 1,590 us-gaap elements that appear at all in FY2024 bank 10-Ks, 75 reach half the filers. Extensions cannot reach far by construction, because an extension is scoped to the filer that minted it: 5,069 of the 5,079 extension names in these filings are used by under 2% of filers, and where two filers happen to mint the same name they are still two elements in two namespaces.
Source: SEC Financial Statement Data Sets, US Securities and Exchange Commission, Division of Economic and Risk Analysis. Facts as filed: uncleaned, unrestated, not audited by FinObservatory. Selection: consolidated facts (segments and coreg empty) in FY2024 10-K submissions with SIC 6020-6036 and prevrpt = 0, one filing per company-year. Bars count DISTINCT TAG NAMES, not facts and not filings. Methodology
The tags nearly everyone reports
The 25 most widely reported us-gaap elements in FY2024 bank 10-Ks. Only 18 of the 1,590 us-gaap elements in these filings reach 95% of filers, so a screen written on shared vocabulary alone is working with about that many columns.
| us-gaap element | Filings | % of filers |
|---|---|---|
| LiabilitiesAndStockholdersEquity | 358 | 100.0% |
| NetCashProvidedByUsedInOperatingActivities | 358 | 100.0% |
| Assets | 357 | 99.7% |
| Liabilities | 357 | 99.7% |
| NetCashProvidedByUsedInFinancingActivities | 357 | 99.7% |
| NetCashProvidedByUsedInInvestingActivities | 357 | 99.7% |
| IncomeTaxExpenseBenefit | 354 | 98.9% |
| NoninterestExpense | 353 | 98.6% |
| RetainedEarningsAccumulatedDeficit | 353 | 98.6% |
| NoninterestIncome | 351 | 98.0% |
| AccumulatedOtherComprehensiveIncomeLossNetOfTax | 349 | 97.5% |
| EarningsPerShareBasic | 349 | 97.5% |
| InterestIncomeExpenseNet | 348 | 97.2% |
| StockholdersEquity | 346 | 96.6% |
| InterestIncomeExpenseAfterProvisionForLoanLoss | 344 | 96.1% |
| NetIncomeLoss | 342 | 95.5% |
| CommonStockSharesAuthorized | 341 | 95.3% |
| Deposits | 341 | 95.3% |
| ComprehensiveIncomeNetOfTax | 339 | 94.7% |
| EarningsPerShareDiluted | 339 | 94.7% |
| InterestAndDividendIncomeOperating | 334 | 93.3% |
| CommonStockSharesIssued | 330 | 92.2% |
| LaborAndRelatedExpense | 329 | 91.9% |
| IncomeLossFromContinuingOperationsBeforeIncomeTaxesExtraordinaryItemsNoncontrollingInterest | 325 | 90.8% |
| InterestExpenseOperating | 325 | 90.8% |
The same line, invented over and over
The 25 company-extension names minted by the largest number of separate FY2024 bank 10-K filers. Each name below appears in 6 to 34 different filings, and in each of them it is a different element in a different namespace.
| Extension name | Filings | % of filers |
|---|---|---|
| ServiceChargesOnDepositAccounts | 34 | 9.5% |
| IncreaseDecreaseInAccruedInterestReceivableAndOtherAssets | 12 | 3.4% |
| AccruedInterestPayableAndOtherLiabilities | 11 | 3.1% |
| CreditLossExpenseReversal | 11 | 3.1% |
| MortgageBankingIncome | 11 | 3.1% |
| OccupancyAndEquipmentExpense | 11 | 3.1% |
| FeesAndCommissionsDepositorAccounts1 | 10 | 2.8% |
| RegulatoryAssessments | 9 | 2.5% |
| InterestAndFeeIncomeLoansAndLeasesTaxable | 8 | 2.2% |
| OtherServiceChargesAndFees | 8 | 2.2% |
| DebitCardExpense | 7 | 2.0% |
| InterchangeIncome | 7 | 2.0% |
| InterestAndFeeIncomeLoansAndLeasesTaxExempt | 7 | 2.0% |
| InterestBearingTimeDeposits | 7 | 2.0% |
| ProvisionForCreditLosses | 7 | 2.0% |
| DataProcessing | 6 | 1.7% |
| FederalHomeLoanBankStockDividends | 6 | 1.7% |
| FinancingReceivableAndNetInvestmentInLeaseExcludingAccruedInterestBeforeAllowanceForCreditLoss | 6 | 1.7% |
| IncreaseDecreaseInAccruedInterestPayableAndOtherLiabilities | 6 | 1.7% |
| InterestPayableAndOtherLiabilities | 6 | 1.7% |
| NoninterestIncomeServiceChargesOnDepositAccounts | 6 | 1.7% |
| OtherRealEstateOwnedExpense | 6 | 1.7% |
| PaymentsToAcquireEquitySecurities | 6 | 1.7% |
| ServiceCharges | 6 | 1.7% |
| TimeDepositsLessThan250000 | 6 | 1.7% |
Source: SEC Financial Statement Data Sets, US Securities and Exchange Commission, Division of Economic and Risk Analysis. Facts as filed: uncleaned, unrestated, not audited by FinObservatory. Selection: distinct tag names on the consolidated face of FY2024 bank and thrift 10-K filings, ranked by the number of filings reporting each name. Methodology
The extension share is rising
The share of all consolidated facts in bank and thrift 10-Ks that carry a company-extension tag rather than a taxonomy element, by fiscal year.
In FY2011, 9.5% of the 195,715 consolidated facts in bank 10-Ks were company extensions. In FY2024 it is 11.3% of 130,034. The path between them is not a straight climb: the series bottoms out at 7.6% in FY2017 and then rises in each of the 7 fiscal years that follow, to FY2024. FY2011 carried 329 consolidated facts per filing and FY2024 carries 363.
Source: SEC Financial Statement Data Sets, US Securities and Exchange Commission, Division of Economic and Risk Analysis. Facts as filed: uncleaned, unrestated, not audited by FinObservatory. Selection: all consolidated facts (any unit, any duration) in bank and thrift 10-K submissions with prevrpt = 0, one filing per company-year, counted per fiscal year. Methodology
The rest of the financial sector, at sector level only
The estate holds only financial filers: every submission carries an SIC code in 6000-6999. The rest of this module measures depository holding companies and nothing else, but the tagging mechanics can be profiled across every major group in the estate. This table makes no financial claim about any of them. It counts tags.
| SIC major group | 10-K filings | Companies | Consolidated facts | Extension share of facts | Distinct extension names |
|---|---|---|---|---|---|
| 60xx Depository institutions | 6,751 | 905 | 2,581,741 | 9.0% | 23,121 |
| 61xx Nondepository credit institutions | 1,413 | 275 | 398,135 | 12.0% | 7,932 |
| 62xx Security and commodity brokers and exchanges | 2,531 | 374 | 566,507 | 15.7% | 8,470 |
| 63xx Insurance carriers | 2,092 | 230 | 709,241 | 9.5% | 6,254 |
| 64xx Insurance agents and brokers | 289 | 44 | 80,716 | 11.1% | 1,198 |
| 65xx Real estate | 1,757 | 323 | 363,883 | 12.7% | 7,717 |
| 67xx Holding and other investment offices | 7,745 | 2,120 | 1,743,506 | 12.8% | 31,222 |
Pooled across every fiscal year in the estate, the lowest extension share of the 7 major groups is Depository institutions at 9.0% of consolidated facts, and the highest is Security and commodity brokers and exchanges at 15.7%. Depository institutions sit at 9.0%. Even at that rate, the carrying amount of SVB’s held-to-maturity book was a company extension.
Source: SEC Financial Statement Data Sets, US Securities and Exchange Commission, Division of Economic and Risk Analysis. Facts as filed: uncleaned, unrestated, not audited by FinObservatory. Selection: all 10-K submissions with prevrpt = 0 in the estate, grouped by the two-digit SIC major group of the submission, consolidated facts only, every fiscal year pooled. Methodology
What this cannot tell you
- An extension is not by itself a red flag. Filers legitimately extend the taxonomy for lines it does not model, and a filer with a high extension share may simply have an unusual balance sheet. This page measures how comparable the filings are, not how honest they are.
- Two filers using the identical extension name are still two separate elements, in two separate namespaces, and the SEC does not certify that they mean the same thing. The right-hand table counts name collisions, and nothing stronger should be read into it.
- The coverage ladder counts a tag as reported if it appears at least once on the consolidated face of the filing. It does not measure whether the filer used the tag correctly, and FinObservatory has not audited any filer’s tagging.
- The counts here cannot be compared to SEC filers outside finance. The estate contains only SIC 6000-6999 submissions, so a claim of the form “banks tag worse than manufacturers” is not computable from this data and is not made.