G-SIB Window Dressing: Basel Moves Indicators to Annual Averages

On 16 October 2026 the Basel Committee on Banking Supervision published final amendments to SCO40, the chapter of the Basel Framework that scores global systemically important banks (G-SIBs). The G-SIB window dressing amendments change what a bank reports for the ten stock indicators behind its score: an average across the reporting year, where the framework currently takes the year-end position. The Committee agreed to implement them by 1 January 2030, with a best-efforts parallel run from 1 January 2028. A separate change applies much sooner. From the end-2026 assessment exercise, conducted in 2027, the threshold that triggers G-SIB data submission and public disclosure rises from EUR 200 billion to EUR 350 billion of leverage ratio exposure measure.

The score carries a direct capital consequence. It drives whether a bank is on the G-SIB list and, if so, which bucket it sits in (supervisory judgment can also add banks to the sample and adjust bucket allocations), and the bucket sets the higher loss absorbency (HLA) requirement in Common Equity Tier 1: 1.0% for scores of 130 to 229 basis points, rising to 3.5% for 530 to 629. The Committee’s stated problem is that some banks temporarily lower certain indicators at year-end, so the snapshot understates their systemic importance. Averaging removes the single date that made that worthwhile, and it hands reporting teams a data build: month-end and, for some items, daily values that a year-end process never had to produce.

Related reading: FR Y-15 Reporting: The Fed’s Systemic Risk Report, Schedule by Schedule

G-SIB window dressing amendments: the dates in order

The amendments arrive in two annexes, each a revised version of SCO40. Annex 1 applies from 1 January 2027 and carries the threshold change. Annex 2 applies from 1 January 2030 and carries the averaging. The calendar below is taken from the implementation section of the final document and from the operational timetable paragraph in each annex.

Two of those rows are easy to misread. The 2030 date is when the revised chapter applies, but the data that feeds it starts accruing on 1 January 2029, because an average over 2029 needs every month-end and every business day of 2029. And the transitional year begins a full year earlier again.

Why the Committee stopped trusting the year-end snapshot

The Committee consulted on this in March 2024. Its press release of 7 March 2024 proposed that banks in the G-SIB assessment exercise report and disclose most indicators as averages over the reporting year, with comments due by 7 June 2024, and it published a working paper with the supporting analysis alongside. The final document says the 2024 proposal was based on evidence that some banks temporarily lower the values of certain indicators at year-end, and that the final text reflects consultation feedback plus additional evidence collected through the G-SIB assessment exercise. The Committee’s meeting release of 1 October 2026 had already signalled the outcome, stating that window-dressing by banks undermines the intended objectives of its standards.

The concern is older than the consultation. In October 2018 the Committee issued a statement on leverage ratio window-dressing after observing heightened volatility in money and derivatives markets around reference dates such as quarter-ends. It called the behaviour unacceptable and said banks should avoid transactions whose only purpose is to report higher ratios on reporting days. The leverage ratio disclosure standard was revised in 2019 to add ratios based on daily average values of securities financing transactions (SFTs), and the Committee’s October 2026 Regulatory Consistency Assessment Programme (RCAP) report on Canada describes that revision as a response to temporary reductions of transaction volumes around reference dates.

The G-SIB amendment works on the measurement side. Once a stock indicator is the average of twelve month-ends, a balance sheet that shrinks for the last week of December moves the average very little, and the daily columns for derivatives and on-balance sheet SFT items leave even less weight on any single date. The Committee also says it will continue to monitor the prevalence of window-dressing as part of its periodic review process.

Which indicators get averaged, and how often

Thirteen indicators feed the G-SIB score. Ten are stock measures and three are flows. The flow indicators (payments activity, underwritten transactions in debt and equity markets, and trading volume) already accumulate over the year, so the averaging table marks them “n/a”. The other ten get an averaging frequency that depends on which group a bank falls into. The final document defines the terms: a daily, monthly or quarterly frequency means the average of daily values, month-end values or quarter-end values over the reporting year.

* Extended scope of consolidation that includes insurance activities. Source: Table 1 of the final document and Table 2 inserted into SCO40 by Annex 2. Data items that the table does not list take the same averaging frequency as the indicator they belong to.

Three cells deserve a second look. Size and the two intra-financial system indicators carry two frequencies at once: the indicator is averaged over month-ends, but its on-balance sheet SFT and derivative component is averaged daily. A team that builds one month-end feed for total exposures will have covered the indicator and missed the sub-item. OTC derivatives notional is the only full indicator that goes daily. And Level 3 assets stay on quarter-end values for every bank, G-SIBs included, which makes it the only asterisked indicator without a monthly or daily column.

The footnote on unlisted data items has a long reach. The G-SIB reporting template collects far more lines than the thirteen indicators, and under that footnote every data item that belongs to one of the ten stock indicators takes the frequency of that indicator.

The 100 basis point line between two averaging regimes

Every bank in the exercise averages. The final standard sets quarter-end averaging as the minimum for all banks, and the monthly and daily columns apply to two groups: banks designated as G-SIBs in the previous year, and near-G-SIBs. SCO40.11, as inserted by Annex 2, defines a near-G-SIB as another bank with a systemic score equal to or above 100 basis points in the previous year. That line sits 30 basis points below the 130 basis point cut-off for G-SIB designation. SCO40.12 then puts banks that scored below 100 basis points, or that had no score calculated, on quarter-end averages.

Which column applies is therefore a lookup against last year’s result. The final document notes that the previous year’s results for all major banks, G-SIBs and near-G-SIBs included, are published on the Committee’s G-SIB dashboard. A bank that sits near the line cannot plan on the lighter column until that year’s score is out.

A new near-G-SIB gets lead time. Footnote 5 to SCO40.11 says a bank that reaches 100 basis points, and was not already subject to SCO40.11 in the previous year, reports on the higher-frequency basis from the second year following the year in which its score is published. The Committee’s own example: a score of 100 basis points in the end-2030 assessment means higher-frequency averages from 1 January 2033, reflected in the end-2033 assessment conducted in 2034. The stated purpose is to give the bank time to build the capability.

The lead time runs one way only. SCO40.13 gives national authorities discretion to require higher-frequency averages than either column sets out, and the final document gives two examples of when it might be used: specific concerns about window-dressing by a particular bank, group of banks or indicator, and a bank that is expected to fall temporarily below the 100 basis point threshold. A one-year dip below the line can still leave a bank on monthly and daily averages if its supervisor uses that discretion.

EUR 350 billion: the threshold change that applies first

Annex 1 rewrites SCO40.32 from 1 January 2027. Banks whose leverage ratio exposure measure, including exposures arising from insurance subsidiaries, exceeded EUR 350 billion at the previous year-end, using the exchange rate applicable at that financial year-end, are required to submit the full set of data used in the assessment methodology to their supervisors. National authorities should require the same banks to publish the 13 indicators, with the trading volume indicator split into its two sub-indicators: fixed income instruments, and equities and other securities. The figure replaces EUR 200 billion.

The timing is set out precisely. The new threshold applies from the end-2026 assessment exercise, and the data that decides it is end-2025 data. A bank’s end-2025 position (31 December 2025 for a bank whose financial year ends on that date) therefore already determines whether it submits and discloses for the end-2026 exercise. The test itself is unchanged in kind: the final document states that the criterion remains the leverage ratio exposure measure, insurance subsidiaries included, as of the previous year-end.

The rewrite also moves the data-submission duty. In the current text, the paragraph on the sample of banks (SCO40.19) says that sample banks submit the full data set. Annex 1 strikes those words from SCO40.19 and attaches the submission duty to the threshold in SCO40.32 instead. SCO40.33 then keeps two groups in scope even below EUR 350 billion: banks added to the sample by supervisory judgment, and banks that were classified as G-SIBs in the previous year. Both must meet the data submission and disclosure requirements in SCO40.32.

For a bank between EUR 200 billion and EUR 350 billion that was not a G-SIB last year and was not added by supervisory judgment, SCO40.32 and SCO40.33 no longer set a submission or disclosure duty. Whether that change reaches the bank depends on its national rules, which is where the EU’s own EUR 200 billion reporting threshold comes in (see the section on national rules below). The sample paragraph keeps its three membership criteria, including the 75 largest banks by leverage ratio exposure measure. The exposure measure itself is defined in the Basel leverage ratio standard, and our leverage ratio RCAP findings for six jurisdictions show how far local versions of that measure can drift from it.

The 2028 parallel run and the first averaged score

The Committee built in a transitional period that starts on 1 January 2028. For the end-2028 exercise, published in 2029, banks are asked to report higher-frequency data on a best-efforts basis. The final document says reporting banks will be required to report both financial year-end values and, on a best-efforts basis, their averaged values. Full application follows: higher-frequency data from 1 January 2029, for the exercise published in 2030 and based on 2029 data.

The best-efforts label describes the quality expected of the averaged figures. The same passage of the final document attaches an expectation to supervisors: during the transitional period, they are expected to apply supervisory judgment where material differences between year-end and averaged values are observed. The document leaves the outcome of that judgment open, and the arrangement puts a bank’s year-end figures and its averages side by side in the same submission a year before the averages become binding.

The capital consequence lands later than the data. Under the Annex 2 operational timetable (SCO40.38), the methodology applies from 2030 based on end-2029 data, and the corresponding HLA requirement applies from 1 January 2032. Annex 1 sets the same spacing for the current cycle: end-2026 data, assessment from 2027, HLA from 1 January 2029. A daily OTC derivatives notional captured on a business day in early 2029 therefore feeds a capital requirement that takes effect about three years later.

What the data build involves

The size indicator is total exposures as defined for the Basel III leverage ratio. The leverage ratio standard computes the exposure measure on a quarter-end basis (LEV20.6, as the Committee’s October 2026 RCAP report on Switzerland restates). For a G-SIB or near-G-SIB, the G-SIB size indicator becomes the average of month-end values, which implies producing the exposure measure at every month-end, eight more runs a year than the quarterly leverage ratio calculation requires, with the on-balance sheet SFT and derivative component captured daily on top.

Banks that already disclose leverage ratios based on daily average SFT values have a related data set. Whether it matches the G-SIB sub-item definition is a question for the reporting instructions, and the most recent instructions cited in the Committee’s October 2026 consultation are those for the end-2025 exercise, dated January 2026, which predate the amendments.

Insurance subsidiaries are the second scope trap. Six indicators carry the extended scope of consolidation that includes insurance activities: total exposures, both intra-financial system indicators, securities outstanding, OTC derivatives notional and Level 3 assets. For a G-SIB or near-G-SIB, five of those six move to month-end or daily averaging, and the insurance entities’ data has to arrive at the same frequency as the banking group’s. A group that today collects insurance figures once a year for the year-end G-SIB template needs monthly feeds at minimum, plus daily feeds for OTC derivatives notional and for the on-balance sheet SFT and derivative items.

The leverage ratio precedent shows how a daily-average rule can be diluted in implementation. The Committee’s October 2026 RCAP report on Australia records that APRA allows the daily SFT average to be calculated as a “best estimate” on a consistent and prudent basis, that some Australian banks use average month-end values accordingly, and that the Assessment Team rated the deviation potentially material because it can hinder market participants’ ability to assess whether disclosed leverage ratios are affected by window-dressing. That finding shows how the Committee graded an approximated daily average in the leverage ratio context.

The standard does not say how to treat weekends and public holidays in a daily average, how to handle a business day missing from a source system, or how the daily SFT and derivative component combines with the month-end remainder of the same indicator. SCO40.11 leaves that detail to the reporting instructions published on the Committee’s website.

Template GSIB1 after the change

Public disclosure runs through Template GSIB1 in chapter DIS75. Annex 1 updates its scope from 2027: the template is mandatory for banks that in the previous year were classified as G-SIBs, had a leverage ratio exposure measure above EUR 350 billion, or were included in the assessment sample by supervisory judgment. The content stays at least 13 indicators, with trading volume split into rows 10a (fixed income instruments) and 10b (equities and other securities). The template is annual, must also appear in the bank’s financial year-end Pillar 3 report, and must be fully consistent with the data submitted to supervisors for onward transmission to the Committee.

That consistency requirement is what carries averaging into the public numbers. The Annex 2 version of the disclosure paragraph (SCO40.35) adds that the ten stock indicators must be calculated as described in SCO40.11 and SCO40.12. From the exercise based on 2029 data, a bank’s GSIB1 total exposures figure is an average, and it will generally differ from the year-end leverage ratio exposure measure reported elsewhere in the same Pillar 3 report.

The transitional year is quieter. The transition paragraph speaks of reporting both year-end and best-efforts averaged values; it does not say the best-efforts averages are disclosed, and the requirement to calculate the disclosed stock indicators as averages appears only in the 2030 text.

Two existing features carry over. GSIB1’s accompanying narrative may explain the use of estimates and the method used, mergers or modifications of legal structure, and the bucket allocation, which gives banks a place to explain an averaged figure. And the in-force SCO40.34 already requires banks to disclose when the data used to calculate their scores differs from what they previously published, with accurate figures due in the financial quarter immediately following the finalisation of the Committee’s score calculation. For how EU supervisors oversee Pillar 3 disclosures generally, see our article on the EBA’s Pillar 3 peer review.

Where national rules still say year-end

A Basel standard binds no bank directly. The Committee describes itself as having no formal supranational authority; its decisions have no legal force and rely on members’ commitment to implement them, and our Basel III implementation tracker follows the gap between agreed and adopted rules. For G-SIB indicators, the EU and US rules reviewed here still use a year-end reference date.

European Union

Article 131 of the Capital Requirements Directive (Directive 2013/36/EU) governs identification of global systemically important institutions (G-SIIs), with at least five buffer sub-categories under Article 131(9), as replaced by Directive (EU) 2019/878: the lowest carries a G-SII buffer of 1% of total risk exposure amount, and the buffer for each higher sub-category rises in steps of at least 0.5%. Article 131(18) mandates regulatory technical standards on the identification methodology, “taking into account any internationally agreed standards”. Those standards are in Commission Delegated Regulation (EU) No 1222/2014. In its consolidated version of 1 December 2021, Article 5(1) bases the indicator values on reported data of the preceding financial year-end, and Article 3(2) requires relevant authorities to report to the EBA, by 31 July each year, the indicator values of entities with a total exposure measure above EUR 200 billion.

The EU also runs a quarterly feed. Article 20 of Commission Implementing Regulation (EU) 2024/3117, applicable since 28 June 2025, requires supplementary G-SII reporting each quarter from EU parent groups, and from standalone institutions on an individual basis, with a total exposure measure of at least EUR 125 billion (insurance subsidiaries included for groups) and a presence in a Banking Union participating Member State. The remittance dates are 1 July, 1 October, 2 January and 1 April. Those are quarter-end observations of the kind the Basel “other banks” column averages, but nothing in the Basel text or the EU regulation says one can stand in for the other. None of the sources reviewed for this article shows an EU proposal implementing the averaging or the EUR 350 billion threshold. For the EU’s domestic systemic buffers, see our EBA 2026 O-SII list.

United States

The Federal Reserve’s GSIB surcharge rule in 12 CFR Part 217, subpart H, identifies a bank holding company as a GSIB when its method 1 score equals or exceeds 130 basis points (section 217.402). Section 217.404 computes each systemic indicator score from the amount reported as of 31 December of the previous calendar year, with the indicators defined by reference to the FR Y-15. For the size indicator, the FR Y-15 instructions for Schedule A already require domestic Category I, II and III banking organisations to report quarter averages, using daily data for on-balance sheet items and the average of the three month-end balances for off-balance sheet items, so for those filers the total exposures amount reported as of 31 December is a fourth-quarter average. In the regulation text itself, averaging appears in one place: the method 2 short-term wholesale funding score in section 217.406 uses amounts calculated for each business day of the previous calendar year. The 2026 edition of the CFR shows no change to the 31 December reference in section 217.404. The Federal Reserve has a separate proposal pending: its notice of proposed rulemaking published in the Federal Register on 27 March 2026 (comments closed 18 June 2026) would require GSIBs to report certain FR Y-15 indicators as averages of daily or monthly values over the reporting quarter, and would calculate method 1 and method 2 scores from average values over a calendar year. It is a proposal, not a binding rule. It would take effect two calendar quarters after a final rule is adopted, so its timetable runs separately from the Basel Committee’s 2030 date.

The Banking Union consultation running alongside

The same 16 October 2026 release opened a consultation on writing the Committee’s 2022 agreement on the European Banking Union (EBU) into SCO40. Under that agreement a parallel set of scores is calculated for EBU-headquartered G-SIBs, recognising 66% of the score reduction that would result from treating intra-EBU exposures as domestic under the cross-jurisdictional activity indicator. When the relevant authority proposes it, the bucket allocation the Committee recommends to the Financial Stability Board is based on the parallel score, any downward adjustment is limited to one bucket, and G-SIB classification is unaffected.

This part is still a proposal. The consultation would amend SCO40 and the reporting instructions to describe the parallel-score calculation, add the relevant data items to the GSIB1 content for EBU banks as part of the cross-jurisdictional activity indicator, and apply with immediate effect once finalised because it formalises existing practice. The consultation text does not address the averaging years. Read with the averaging table’s rule that unlisted data items follow their parent indicator, EBU parallel-score items would sit under cross-jurisdictional activity, which is averaged over quarter-ends for every bank. That is my reading, and a point respondents could ask the Committee to confirm.

Frequently Asked Questions

Our score in last year’s assessment was exactly 100 basis points. Are we a near-G-SIB?

The revised SCO40.11 says “equal to or above 100 basis points”, which puts a bank at exactly 100 in the G-SIB and near-G-SIB column. The explanatory section of the final document describes near-G-SIBs as banks with a score above 100 basis points. In my reading the paragraph inserted into the Basel Framework is the operative text, and a bank sitting exactly on the line has good reason to get its supervisor’s reading before building to the lighter column.

Does the averaged exposure measure decide whether we cross EUR 350 billion?

Under the 2030 text it does not. The disclosure paragraph (SCO40.35 in Annex 2) still tests the leverage ratio exposure measure at the previous year-end. The averaged measure is used elsewhere: the sample criterion for the 75 largest banks (SCO40.22(1) in Annex 2) switches from the financial year-end exposure measure to the measure calculated as defined in SCO40.11 and SCO40.12. A bank can therefore be ranked on one figure and tested against the threshold on another.

Which exchange rate converts an averaged stock indicator into euros?

The amendments add no averaged rate for stock indicators. The currency paragraph, renumbered SCO40.15 in Annex 2, still says the 31 December spot rate is used to convert each of the 13 indicator values into euros, and keeps average rates over the fiscal year for converting individual payments data into the bank’s reporting currency. My reading is that a stock average is built in the reporting currency and translated once at the year-end spot rate; the reporting instructions should confirm it.

Our financial year ends on 30 June. How does averaging work for us?

DIS75 still allows national authorities to let such banks report indicator values based on their position at 31 December. The amendments describe averages over the reporting year (SCO40.11 and SCO40.12) and over the financial year (the explanatory text), and they do not say how either aligns with a 31 December interim position. National instructions will have to settle it.

We complete an acquisition in the middle of the reporting year. How is it averaged?

Neither annex addresses it. The GSIB1 narrative already lets a bank explain mergers or modifications of its legal structure, and the reporting instructions are where the Committee places calculation detail, so the treatment of pre- and post-acquisition months is a question to raise with the supervisor before the transitional year.

Do these amendments change leverage ratio disclosure or TLAC reporting?

Annex 1 and Annex 2 revise SCO40 and the GSIB1 template in DIS75 only. Leverage ratio disclosure keeps its own safeguard, the daily-average SFT figures added in 2019, and nothing in the two annexes touches total loss-absorbing capacity (TLAC) reporting.

Related Articles

• FR Y-15 Reporting: The Fed’s Systemic Risk Report, Schedule by Schedule: how US holding companies report the systemic indicators behind the GSIB surcharge.
• Leverage Ratio RCAP 2026: Where Six Jurisdictions Depart from Basel: the leverage ratio exposure measure and daily-average SFT disclosure as implemented in six jurisdictions.
• Basel III Implementation by 2027: What the BCBS Tally Leaves Out: where adopted rules lag the Basel standards they are meant to implement.
• EBA 2026 O-SII List: Which EU Banks Carry an O-SII Buffer: the EU’s domestic systemic buffers alongside the G-SII framework.
• HKMA D-SIB Designation 2025: Five Banks, HLA Buckets Unchanged: how a domestic systemic framework maps scores to higher loss absorbency buckets.

Key Takeaways

• EU G-SII identification (Delegated Regulation 1222/2014) still runs on preceding financial year-end values and the US method 1 score (12 CFR 217.404) on amounts reported as of 31 December, although the FR Y-15 size indicator behind that score is already a fourth-quarter average; a bank reporting under both regimes and Basel may need year-end and averaged figures side by side until local rules change.
• Read last year’s score off the Committee’s G-SIB dashboard before scoping the build: a score of 100 basis points or more puts the bank in the G-SIB and near-G-SIB column (month-end or daily averaging for most stock indicators, quarter-end for cross-jurisdictional activity and Level 3 assets), and below it means quarter-end averaging for every indicator.
• For a G-SIB or near-G-SIB, extend month-end and daily feeds to insurance subsidiaries for the five asterisked indicators averaged at those frequencies (Level 3 assets stays on quarter-end values); the extended scope of consolidation applies at the averaged frequency.
• Reconcile year-end and averaged values through the 2028 parallel run and document any material gap, since supervisors are expected to apply judgment where one appears.
• Ask the national supervisor early for the reporting instructions that will govern the averaged years; the most recent set the Committee cites covers the end-2025 exercise and predates the amendments.
• Where a daily series is approximated, record the method; the GSIB1 narrative already allows banks to explain estimates and how they were made.

Sources and References

• Basel Committee on Banking Supervision, Global systemically important banks: revised assessment framework to address window dressing (16 October 2026), publication page: bis.org; full text: bis.org (PDF)
• BCBS press release, Basel Committee finalises measures to address window-dressing in the G-SIB framework and consults on G-SIB methodology review (16 October 2026): bis.org
• BCBS consultative document, Global systemically important banks: methodology review (16 October 2026, comments by 14 January 2027): bis.org; full text: bis.org (PDF)
• BCBS press release of 1 October 2026 on the Committee’s meeting: bis.org
• BCBS press release, Basel Committee consults on measures to address window-dressing in the G-SIB framework (7 March 2024): bis.org; consultative document page: bis.org
• BCBS, Statement on leverage ratio window-dressing behaviour (18 October 2018): bis.org
• Basel Framework, SCO40 Global systemically important banks, version in force from 9 November 2021: bis.org
• BCBS, G-SIB cut-off score and bucket thresholds: bis.org; G-SIB assessment page and dashboard: bis.org
• BCBS, RCAP assessments of the Basel leverage ratio framework (October 2026): Canada bis.org (PDF); Australia bis.org (PDF); Switzerland bis.org (PDF)
• Directive 2013/36/EU (Capital Requirements Directive), Article 131: eur-lex.europa.eu; Article 131(9) and (18) as replaced by Directive (EU) 2019/878: eur-lex.europa.eu
• Commission Delegated Regulation (EU) No 1222/2014 on the identification of G-SIIs, consolidated version of 1 December 2021, Articles 3 and 5: eur-lex.europa.eu
• Commission Implementing Regulation (EU) 2024/3117 on supervisory reporting, Article 20: eur-lex.europa.eu
• 12 CFR Part 217, subpart H (sections 217.402 to 217.406), 2026 edition: govinfo.gov (PDF)
• Federal Reserve Board, Instructions for the Preparation of the Systemic Risk Report (FR Y-15), Schedule A (Size Indicator) general instructions: federalreserve.gov (PDF)
• Federal Reserve Board, notice of proposed rulemaking amending the GSIB surcharge rule in Regulation Q and the FR Y-15 (Docket No. 1889, RIN 7100-AH22; Federal Register, 27 March 2026), proposal only: govinfo.gov

First checkpoints: the end-2026 exercise and 1 January 2028

The threshold test comes first and needs one number: the end-2025 leverage ratio exposure measure, insurance subsidiaries included, converted at the applicable year-end rate and set against EUR 350 billion, together with a check on whether the bank was a G-SIB last year or was added to the sample by supervisory judgment. The averaging build has a fixed start after that. The artifact to have in place before 1 January 2028 is an indicator-by-indicator capture plan that maps each of the ten stock indicators, and each of their data items, to the frequency in its column, with the source system, the insurance entities in scope and the daily cut-off time named for every line.

Disclaimer: The information on RegReportingDesk.com is for educational and informational purposes only. It does not constitute legal, regulatory, tax, or compliance advice. Always consult your compliance officer, legal counsel, or the relevant supervisory authority for guidance specific to your institution.