Basel IRRBB Pillar 2 Guidance: Comments Due 9 February 2027

The Basel Committee on Banking Supervision published proposed additional Pillar 2 guidance on interest rate risk in the banking book (IRRBB) on 9 November 2026, with comments due by 9 February 2027. The proposed IRRBB Pillar 2 guidance keeps the 15% of Tier 1 outlier test and the six prescribed shock scenarios as they are. It adds a defined category called “Supervisory Information”: data that banks would report to supervisors for ongoing monitoring and examinations, covering modelling practices, stress testing, hedging controls, embedded losses and earnings indicators.

The Committee agreed to consult at its meeting of 28 and 29 September 2026, after an empirical assessment of the 2016 IRRBB standard identified specific shortcomings in banks’ management of the risk. The publication page ties those shortcomings to periods of elevated inflation, sharp interest rate increases and heightened market volatility. The proposals sit in two places: amendments to SRP31, the IRRBB chapter of the Basel Framework, and a rewrite of SRP98, its application guidance.

Nothing in the document binds a bank on its own. The BCBS Charter states that the Committee has no formal supranational authority and that its decisions have no legal force, so the guidance reaches a reporting desk only through a jurisdiction’s own rules. The useful work before February is a data inventory: which of the proposed Supervisory Information items a bank already produces, which it could produce, and which definitions are too loose to report consistently.

Related reading: Interest Rate Risk in the Banking Book (IRRBB): EBA Guidelines and CSSF Expectations for Luxembourg Banks

IRRBB Pillar 2 guidance: the dates that matter

• 28 and 29 September 2026: the Committee, meeting in Indonesia, agreed to consult on additional Pillar 2 guidance for banks and supervisors (media release of 1 October 2026).
• 5 November 2026: separate consultation on consequential amendments to the Basel consolidated guidelines, the vehicle for moving application guidance chapters out of the Basel Framework. Comments are due by 21 December 2026.
• 9 November 2026: IRRBB consultative document published, status “Consultative”.
• 9 February 2027: deadline for comments on the IRRBB proposals, submitted through the BIS comment upload page. The Committee publishes comments on the BIS website unless a respondent asks for confidential treatment.
• No implementation date: the consultation proposes none. In the annex, the “Version effective from” field for the revised SRP98 is blank, and the SRP31 header still carries the 1 January 2026 version date.

Two separate BCBS consultations carry different deadlines. The IRRBB proposals close on 9 February 2027, but whether the rewritten SRP98 ends up in chapter MIR20 of the consolidated guidelines depends on the consequential-amendments consultation, which closes seven weeks earlier on 21 December 2026.

Inside the consultative document and its marked-up annex

The document has a short narrative part and a long annex. Section 3.1 lists additional guidance for banks under five headings: model risk management and model governance, stress-testing approaches, IRRBB control approaches, treatment of embedded losses, and earnings indicators. Section 3.2 adds a peer comparison framework for supervisors. Section 3.3 introduces Supervisory Information. The annex then shows every proposed edit to SRP31 and SRP98, with additions in red and deletions struck through.

The scope section describes the result as standalone guidelines that complement the current IRRBB framework and replace the existing SRP98 application guidance. SRP98 would also be renamed “Elaboration and application guidance on interest rate risk in the banking book”. Footnote 3 explains why the Committee wants the material in MIR20: to reduce the proliferation of guidance across its website and to separate standards from guidelines more clearly. Our note on the BCBS consolidated guidelines and the chapters that may move covers that migration.

The split matters for anyone citing the text in a policy. The scope section says certain foundational elements have also been incorporated into SRP31, so part of the new material would sit in the Framework chapter and the rest in the guidance chapter. The drafting register also shifts. Many of the new SRP31 sentences are framed as “Sound risk management includes”, and the new SRP98 paragraphs lean on “may”. I read that as expectation-setting language, and a team mapping obligations should record each item with the modality it actually carries.

What the proposals leave untouched

The outlier test text is unchanged apart from its cross-references. The supervisor should implement at least one outlier/materiality test comparing the bank’s maximum change in economic value of equity (EVE) under the six prescribed shock scenarios with 15% of Tier 1 capital. Supervisors may add further tests, provided they apply them throughout their jurisdiction in the same form; these may use a different capital measure or capture risk relative to earnings, with a threshold at least as stringent as 15% of Tier 1.

The six scenarios (parallel up, parallel down, steepener, flattener, short rates up, short rates down) stay the same, and so does the currency shock table. That table already reflects the recalibration the Committee approved in July 2024, which had an implementation date of 1 January 2026. The standardised framework, including the 19 time buckets and the caps on core non-maturity deposits (NMDs), is printed in black in the annex, meaning no proposed edit.

Paragraph numbers are a different matter. New paragraphs are inserted throughout SRP31, so references move even where the text does not.

National texts quote these numbers. Canada’s OSFI Guideline B-12 (2027 version, published 10 September 2026) tags its outlier test sentence “[Basel Framework, SRP 31.82]”. Internal IRRBB policies, model documentation and ICAAP chapters that cite SRP paragraph numbers would need the same update once a final text is published.

Supervisory Information: what the table lists and what it leaves open

The proposed SRP98.77 defines Supervisory Information as information that banks report to supervisors as part of ongoing monitoring or examinations related to IRRBB. Its stated purposes are to close supervisory data gaps, improve supervisors’ understanding of the basis for senior management’s decisions, strengthen independent validation and address any lack of transparency in internal reporting. SRP31 is amended to match: the Principle 10 paragraph on collecting information from banks would add “including supervisory information” and “including through ongoing monitoring or examinations”.

The table is a list of examples of what banks “may report”. It sets no template, frequency, reference date, format or materiality threshold, and Section 3.3 of the consultative document calls it “a table with examples”. Each supervisor would decide what to collect and how, much as the existing SRP31 already allows the precise information collected to differ among supervisors.

One governance step would become explicit. A new SRP31 paragraph in Principle 7 says sound risk management includes governance for senior management to receive and examine relevant Supervisory Information before it is reported to supervisors. For a reporting team, that would translate into a sign-off gate in the production timetable, with the package reaching senior management early enough for review before the submission date.

Three earnings ratios beside the EVE test

The consultation’s case for earnings indicators is stated directly: given the long-term nature of the balance sheet run-off horizon, the Supervisory Outlier Test alone might conceal short- and medium-term impacts of rate movements on earnings. Section 3.1(e) names three firm-specific measures of net interest income (NII) sensitivity, repeated in the proposed SRP98.30:

• (NII shocked minus NII baseline) divided by Tier 1 capital;
• (NII shocked minus NII realised) divided by NII realised;
• change in NII divided by the three-year average of core operating profit.

Banks are “encouraged” to address both NII and EVE in risk management and to take both into account when assessing risk capacity. SRP98.30 adds detailed descriptions of earnings risk by balance sheet item with and without hedging, pass-through rates (defined as the proportion of the offered rate on an instrument compared with a reference rate), repricing gaps, the effect of fair value changes on NII, and reverse stress tests that identify the conditions under which a bank’s earnings are severely threatened, for example when distributable profits become depleted.

The list is framed as measures that “may” be used “according to jurisdictional and firm-specific considerations”, with no threshold attached to any ratio. Under Principle 12, outlier tests come with published criteria and, for additional tests, a threshold at least as stringent as 15% of Tier 1; the earnings ratios carry neither.

Several inputs are undefined. The document does not say which shock scenario feeds “NII shocked” in the ratios, over what horizon, which period “NII realised” covers, or what counts as “core” operating profit. The existing guidance describes earnings measures as typically covering one to three years and at most five, which leaves a lot of room. Two banks could report the same ratio on different bases, and peer comparison is exactly what the Supervisory Information is meant to support.

The first ratio has a familiar shape for EU banks. Article 5(1) of Delegated Regulation (EU) 2024/856 already defines a large decline in NII for the supervisory outlier test as a decline of an institution’s one-year net interest income by more than 5% of its Tier 1 capital under either of the two parallel supervisory shock scenarios in Article 1(2). The EBA’s final report on revised SREP Guidelines, published in June 2026 and covered in our note on the EBA revised SREP guidelines, also lists a comparison of worst NII projections against current realised NII among the additional dimensions competent authorities should consider analysing, when deemed appropriate, where an institution is identified as an outlier under the supervisory outlier test on net interest income (paragraph 258). The Basel document mentions neither, and how an EU process would treat the overlap is an open question.

Embedded losses: from a capital factor to a monitored number

Embedded losses are already in today’s SRP31. The capital adequacy factors in SRP31.69 include “the impact of embedded losses”, and SRP31.77 asks supervisors to consider the level of IRRBB “including embedded losses” relative to capital and earnings. The consultation turns that reference into a measured, reported quantity.

A new footnote defines embedded gains and losses as the difference between book value and market value of all instruments in the banking book. A new paragraph (31.39 in the annex) says sound risk management includes monitoring the impact on capital headroom, particularly for securities held at amortised cost. The Principle 9 paragraph would define capital headroom as the surplus of a bank’s capital resources above all minimum regulatory requirements and buffers.

The proposed SRP98.34 says holistic monitoring of capital headroom may include four elements:

• for banks with significant holdings of investment securities, the amount of embedded gains and losses on securities held at amortised cost, with the impact on capital headroom reported to senior management and supervisors;
• transparency on the share of EVE sensitivity attributable to amortised-cost securities under different rate shocks;
• stratification of the amortised-cost portfolio by asset class, maturity and fixed versus floating rate, with a summary of hedging provided as Supervisory Information;
• internal stress tests under high levels of embedded losses, with the effect on EVE and capital headroom.

The definition and the monitoring focus cover different populations. The footnote measures book-to-market differences on every banking book instrument, loans included, while the headroom work centres on amortised-cost securities. The existing SRP98 text gives the background to that distinction: a loss in economic value on amortised-cost instruments does not automatically become an accounting loss, whereas for fair-valued items rate changes hit accounting values, P&L and available capital directly. “Significant holdings” carries no threshold in the text.

Supervisors get matching tools. The annex extends the SRP31 list of supervisory considerations: where embedded losses in the securities portfolio may rapidly escalate under stress, supervisors should evaluate the effect of rate changes on embedded losses and gains, on capital and on liquidity positions, and take appropriate actions. SRP98.78 adds that supervisors may ask for additional stress tests or closer monitoring, and may discuss the matter with management of a bank identified as particularly vulnerable to further valuation declines.

Validation and behavioural assumptions get firmer wording

A new SRP31 paragraph says sound risk management includes testing key IRRBB models and material behavioural assumptions through backtesting and/or other independent model validations. The vendor-model paragraph is reworded so that validation covers the justification of any change in modelling assumptions and prompts recalibration for poor model performance. A new “Validation activities” block in SRP98 (98.74 to 98.76) lists what validation may assess:

• the justification for floors applied to rate shocks, since floors can limit the impact of falling and negative rate scenarios;
• the choice of discounting yield curves;
• scenarios adverse enough to capture the realisation of option risk;
• changes in modelling assumptions and the use of management decisions and overlays;
• recalibration or redevelopment when backtesting shows material degradation of a model’s predictive ability, with overrides, overlays or other compensating controls in the meantime where poor performance can be attributed to exceptional circumstances.

Challenger models appear as an optional tool, “readily available to supervisory authorities upon request when established thresholds are exceeded”. Banks using a vendor model may assess whether its assumptions reflect their own risk profile, customer base and business model.

For NMDs the expectations tighten. The modelling-assumptions paragraph (SRP31.42 in force, 31.43 in the annex) replaces its Table 1 with a longer list of segmentation factors and macroeconomic variables for fixed rate loans, loan commitments, term deposits and NMDs, including deposit guarantee coverage, digital propensity and pass-through magnitude. Among the factors banks may consider for behavioural model time series, the proposed SRP98.41 lists historical coverage of at least one full interest cycle or a sufficiently long and representative period. SRP98.49 names insured versus uninsured status among the segmentation criteria.

The standardised caps on core deposits in Table 4 stay as they are: 90% and five years for retail transactional, 70% and 4.5 years for retail non-transactional, 50% and four years for wholesale. The proposed SRP98.49(1) adds that banks should not rely exclusively on those caps, because they may be insufficiently conservative under certain market conditions and firm-specific risk profiles.

Stress testing frequency and the four control dimensions

New SRP31 text says sound risk management includes stress tests on a regular basis and more frequently when warranted by market conditions, changes in the portfolio or increased exposure levels. Another addition says it includes identifying the risk scenarios that could significantly impact a bank’s IRRBB. Sensitivity testing should consider different possible levels and scenarios, and the proposed SRP98.32 adds multiple shock levels, non-parallel curves and the historical and implied volatility of rates.

One deletion changes modality. The SRP31.35 sentences on measuring risk under a reasonable range of scenarios, including negative rate scenarios in low-rate environments, are struck from the Framework chapter. Similar content reappears in the proposed SRP98.31, where banks “may also consider” negative rate scenarios. In the current SRP31 text, banks “should also consider” them.

The control approaches in Section 3.1(c) come in four dimensions, which reappear as the four items under “Banks’ IRRBB control approaches” in the SRP98.77 table: a qualitative summary of controls including limits and authorised hedging instruments; an analysis of the impact of hedging instruments on IRRBB metrics, with an assessment of hedge effectiveness when the supervisor requests it; delegated powers and lines of responsibility; and exposures both with and without hedging. A footnote to the SRP98.77 table adds that limits and the corresponding management actions should be clearly defined, with the rationale for each documented.

The “with and without hedging” view is the operational catch. SRP98.77 lists IRRBB exposures with and without the authorised hedging instruments, at the relevant level of aggregation, under stress testing, and exposures both with and without hedging under control approaches, and the earnings row lists baseline and shocked NII by balance sheet item with and without hedging. A bank whose IRRBB engine stores only post-hedge results would first need to establish whether it can produce a pre-hedge view; the consultation prescribes no method.

Peer groups and trend analysis on the supervisor’s side

SRP31 already contains peer comparison. The current text asks supervisors to compare key behavioural and strategic assumptions across banks (SRP31.79) and asks jurisdictions running off-site reviews to put in place reporting schemes that enable peer comparison (SRP31.73). The new element is a structured method built on Supervisory Information.

A new paragraph (31.81 in the annex) says supervisors evaluate a bank’s quantitative and qualitative information against an identified peer group and use the results as dynamic indicators of potential outlier banks. The proposed SRP98.79 gives three example dimensions for building reference groups:

• static, such as size or business model;
• operational, such as cross-jurisdictional presence and assets or liabilities in different currencies;
• idiosyncratic, such as drivers of EVE and NII sensitivity, duration gap, behavioural optionalities, market power or embedded losses relative to capital ratios.

Jurisdictions may choose other dimensions. The text’s own example is supervisors comparing banks’ assumptions on deposit volumes and rate sensitivities with peers in the same competitive market. SRP98.80 adds trend analysis across rate cycles to spot banks whose metrics are approaching regulatory or internal limits or deteriorating steadily, and to check modelled duration or portfolio valuations with supervisory tools.

Where the proposals would land in an EU reporting stack

EU banks already file a quarterly IRRBB package. Article 21 of Implementing Regulation (EU) 2024/3117 requires template J 01.00 from all institutions each quarter, with the breakdown, repricing cash flow and parameter templates differentiated for large, other and small and non-complex institutions, and the qualitative templates J 10.01 and J 10.02 (large institutions) or J 11.01 and J 11.02 (others) annually. The outlier tests come from Article 98(5) CRD and Delegated Regulation (EU) 2024/856.

My reading is that the overlap is partial. J 01.00 already reports changes in EVE and NII as ratios to Tier 1 capital under the supervisory scenarios, which resembles the first earnings ratio. Sensitivity results by balance sheet component resemble what the J 02.00 to J 04.00 breakdowns capture, and the qualitative templates cover methodology. Backtesting results, management overlays reported to the board, embedded-loss projections for amortised-cost securities and the two ratios built on realised NII and core operating profit have no obvious J-template home.

Anything an EU supervisor added would also face Article 104(2) CRD, in the wording introduced by Directive (EU) 2019/878. Additional or more frequent reporting may be imposed only where it is appropriate and proportionate and the information requested is not duplicative. Information is deemed duplicative where the same or substantially the same information has already been otherwise reported to the competent authority or may be produced by it, and the authority may not require additional information it previously received in a different format or level of granularity where that does not prevent it from producing information of the same quality and reliability. The Basel document says nothing about EU implementation, the J templates or national reporting schemes, so the EU material here is context for sizing data gaps, with no prediction attached.

Gaps the consultation leaves open

The Committee welcomes comments on “all aspects” and lists no numbered questions, so respondents set their own agenda. The text leaves these points without an answer:

• an implementation date or transition period;
• the frequency, format and reference date of Supervisory Information;
• scenario, horizon and threshold for the three earnings ratios;
• definitions of “NII realised” and “core operating profit”;
• a size test for “significant holdings of investment securities”;
• how the new expectations scale for smaller banks, beyond the existing SRP31.5 principle that implementation should be commensurate with a bank’s nature, size and complexity, with supervisors focusing on systemic risks inherent in large, complex or internationally active banks.

Each of those gaps would need a bank-specific definition before any figure could be compared across banks.

Building a comment and a data inventory before 9 February 2027

A bank preparing for this consultation has two outputs to produce: a gap map against the Supervisory Information table, and, if it chooses to respond, comments on the definitions that drive cost. A workable sequence:

1. Map each SRP98.77 row to an existing source: ALCO and board packs, the ICAAP, model validation reports, and in the EU the J templates. Mark each item as produced, producible from existing data, or not available.
2. Compute the three earnings ratios for recent quarter-ends on the bank’s own definitions, and document the choices for scenario, horizon, “NII realised” and “core operating profit”. The documented choices become comment material.
3. Test whether the securities book can be stratified by asset class, maturity and fixed or floating rate, with embedded gains and losses projected under different rate shock scenarios and the effect on capital headroom calculated.
4. List the backtesting evidence for NMD, prepayment and deposit pricing models, including the thresholds that would trigger reporting.
5. Check whether exposures can be produced both with and without hedging, at the relevant level of aggregation for the stress testing row.
6. Add a senior management review step to the timetable for any IRRBB Supervisory Information submission, matching the new Principle 7 paragraph on relevant Supervisory Information.
7. List every policy, model document and ICAAP chapter that cites SRP31 or SRP98 paragraph numbers.

Our guide to ICAAP and ILAAP reporting covers how IRRBB capital already flows into the internal assessment, which is where the capital headroom definition would land first.

Frequently Asked Questions

Which banks does the consultation target?

The document has no scope paragraph by bank type. The BCBS Charter expects full implementation of Basel standards by members and their internationally active banks, while SRP31 itself describes IRRBB as a risk encountered by all banks and asks supervisors to gauge their responses for banks with low IRRBB profiles. Whether a smaller domestic bank is captured depends on how its jurisdiction implements the final text.

Does Supervisory Information become part of our Pillar 3 disclosure?

Supervisory Information is defined as information reported to supervisors. Public disclosure stays under Principle 8, which points to the DIS70 disclosure chapter, and the consultation does not amend that principle. Nothing in this text would make the embedded-loss projections or the earnings ratios public.

Our securities are held at fair value through other comprehensive income. Does the embedded-loss work still apply?

The footnote definition covers book-to-market differences on all banking book instruments, so the measure itself reaches beyond amortised cost. The capital headroom monitoring and the stratification element focus on securities held at amortised cost. A portfolio at fair value still contributes to EVE sensitivity and to the stress results that the Supervisory Information table lists.

Our bank floors post-shock rates at zero. Does the consultation change that?

The national discretion for supervisors to set floors on post-shock rates under the six scenarios, provided the floors are not greater than zero, carries over unchanged (31.97 in the annex). The new validation guidance lists the justification of floors that banks apply to rate shocks among the items validation may assess, because such floors can limit the impact of declining and negative rate scenarios. The first is a supervisory parameter; the second would need its own justification in the validation file.

Will our comment letter be published?

Yes, by default. The consultation states that all comments will be published on the BIS website unless a respondent specifically requests confidential treatment.

If SRP98 moves to MIR20, does the guidance gain or lose force?

The Committee presents the move as a way to reduce the proliferation of guidance and separate standards from guidelines more clearly. Since BCBS decisions have no legal force, neither chapter binds a bank directly, and a policy that quotes SRP98 would need to point to MIR20 once the consolidated-guidelines amendments are final.

Related Articles

• Interest Rate Risk in the Banking Book (IRRBB): EBA Guidelines and CSSF Expectations for Luxembourg Banks: the EU legal basis, outlier tests, behavioural modelling and J-template reporting for Luxembourg institutions.
• BCBS Consolidated Guidelines Final: Seven Basel Chapters May Move: the 5 November 2026 consolidated guidelines and the consequential-amendments consultation that governs where SRP98 ends up.
• FINMA Circular 2019/2 Consultation: Earnings Risk Gets an Outlier Test: our note on FINMA’s draft revision of the Swiss interest rate risk circular for banks.
• ESPREP-STT Reporting in Luxembourg: The Retired IRRBB Stress Test: how a national IRRBB return was replaced by the harmonised EBA templates.
• EBA Revised SREP Guidelines: What EU Banks Must Review in ICAAP, ILAAP and Pillar 2 Capital: the EU supervisory review changes that sit alongside IRRBB assessment.

Key Takeaways

• Teams that want to comment on SRP98 moving into MIR20 file under the consolidated-guidelines consultation, which closes on 21 December 2026, ahead of the IRRBB deadline.
• The proposals leave the outlier test, the shock table and the Table 4 core deposit caps unchanged; plan 2027 on the currently effective values and recheck them against any final text.
• Once a final text appears, refresh every internal cross-reference to SRP31 and SRP98 paragraph numbers.
• Size the five SRP98.77 rows now; format, frequency and thresholds would come from each national supervisor.
• The undefined inputs to the NII ratios (scenario, horizon, “NII realised”, “core operating profit”) are the strongest comment points.
• Check the ICAAP’s headroom definition against “surplus above all minimum regulatory requirements and buffers” before modelling embedded-loss impacts.
• The SRP98.77 table lists IRRBB exposures with and without hedging in the stress testing and control approaches rows, and baseline and shocked NII by balance sheet item with and without hedging in the earnings indicators row, so check whether the bank can produce both views; only the stress testing row adds “at the relevant level of aggregation”.

Sources and References

• Basel Committee on Banking Supervision, Interest rate risk in the banking book: additional Pillar 2 guidance, publication page (9 November 2026): bis.org
• Basel Committee on Banking Supervision, Consultative Document: Interest rate risk in the banking book, issued for comment by 9 February 2027 (PDF): bis.org
• BIS comment upload page for BCBS consultations: bis.org
• BIS media release, Basel Committee meets to advance supervisory and regulatory initiatives and discuss risks and vulnerabilities to the global banking system (1 October 2026): bis.org
• Basel Committee on Banking Supervision, Consolidated guidelines: consequential amendments, consultation (5 November 2026): bis.org
• Basel Framework, SRP31 Interest rate risk in the banking book (version effective 1 January 2026): bis.org
• Basel Framework, SRP Supervisory review process, chapter PDF including SRP31 and SRP98: bis.org
• Basel Committee on Banking Supervision, Interest rate risk in the banking book standard (April 2016): bis.org
• BIS press release on amendments to the IRRBB standard, including recalibrated shocks with a 1 January 2026 implementation date (3 July 2024): bis.org
• Basel Committee on Banking Supervision, Charter: bis.org
• Commission Delegated Regulation (EU) 2024/856 on supervisory shock scenarios, common modelling and parametric assumptions and what constitutes a large decline: EUR-Lex
• Commission Implementing Regulation (EU) 2024/3117 on supervisory reporting, Article 21 (IRRBB reporting): EUR-Lex
• Directive (EU) 2019/878 (CRD V), amending Article 104 of Directive 2013/36/EU: EUR-Lex
• Directive 2013/36/EU (CRD), consolidated version of 11 July 2026, Article 98(5) supervisory outlier tests: EUR-Lex
• European Banking Authority, Final Report on revised SREP and supervisory stress testing Guidelines (June 2026): eba.europa.eu
• Office of the Superintendent of Financial Institutions, Guideline B-12 Interest Rate Risk Management (2027): osfi-bsif.gc.ca

Before 9 February 2027: the IRRBB data inventory

The Committee has set no implementation date, and any reporting change would arrive through national rules. The consultation does fix the vocabulary: Supervisory Information, capital headroom, embedded gains and losses, and three earnings ratios. A bank that builds its inventory against the SRP98.77 table now, documents its own definitions for the ratios and embedded losses, and decides by January whether to comment will have its material ready for the 9 February 2027 deadline.

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.