IN ONE SENTENCE — The accounting function turns contracts into numbers that a board signs, an auditor certifies and a supervisor uses, and for sustainable instruments it decides whether a green or ESG-linked feature changes the measurement of a deal or only its label.
Where it sits. Accounting is part of Group Finance, under the Chief Financial Officer. At Meridian Bank four teams touch every sustainable instrument. The accounting policy team (often called technical accounting, a dozen people at group level) writes the bank’s IFRS manual, issues opinions on new products and talks to the auditors about judgements. Product control, attached to the CIB, produces the daily profit and loss of the trading desks, checks valuations independently of the front office and books reserves. Financial reporting and consolidation produces the quarterly and annual financial statements and the FINREP regulatory reports to the ECB. The IFRS 9 impairment team, often shared between Finance and Risk, runs the expected credit loss (ECL) models, the scenarios and the overlays. Around them sit the finance business partners (the “CFOs” of Corporate Banking, Global Markets and Treasury), the tax department and the external auditors, who spend several weeks a year inside the bank.
Who the clients are. Accounting has no external clients, but it has three demanding internal ones: the business lines, which want to know before signing what a new product will do to their revenue and volatility; Risk and Capital Management, who take accounting values as the starting point of regulatory ratios (CET1 begins with accounting equity); and the Sustainability Reporting team (see E06), whose Green Asset Ratio is computed on the gross carrying amounts that accounting produces.
How it “earns money”. It does not, but it moves money in two ways. First, a classification decision changes the timing of profit: the same loan measured at amortised cost shows smooth interest income, while at fair value through profit or loss (FVTPL) it shows every move in credit spreads. For a bank whose shares trade on a multiple of earnings, a few tens of millions of unexplained volatility is expensive. Second, impairment is the largest judgement in a bank’s accounts: in Meridian’s case (an assumption of this course) the ECL allowance runs to a little over 1% of the loan book, and a 10-point change in the weight of a downside scenario can move it by several hundred million euros. The quality of the accounting function is visible in the cost of equity, not in a revenue line.
The team on a sustainable product. When Corporate Banking signs its first sustainability-linked loan (SLL) with a new feature, the deal team sends a new product approval (NPA) request. The accounting policy team writes a short memo: classification, measurement, fee recognition, disclosure. The IFRS 9 impairment team confirms that the loan system can carry a flag and the margin grid. Product control checks whether any part of the deal needs fair value. The auditors review the memo at the next quarterly close. The same chain applies when Treasury buys a green bond for the liquidity portfolio, when a client asks Meridian how its own sustainability-linked bond (SLB) will look in its accounts, or when the commodities desk starts carrying an inventory of allowances.
WHO DOES WHAT — A realistic week in the accounting policy team in November 2026. Monday: the Q3 close is over; the team drafts the note on the first application of the May 2024 amendments to IFRS 9 and IFRS 7, which Meridian applies from 1 January 2026 and must describe in its 2026 annual report. Tuesday: Corporate Banking asks whether a new SLL with a 15 bp ratchet and a penalty paid to a charity still passes SPPI; memo due Friday. Wednesday: the impairment team presents its climate overlay for the year-end ECL to the Provisions Committee; the auditors ask for the evidence behind the 20% weight of the transition scenario. Thursday: product control flags that the carbon inventory of the commodities desk is measured at fair value but a new sub-portfolio was booked at cost; the team reconciles. Friday: a client of the DCM desk (see C02) asks for an informal view on how its SLB step-up should be accounted for; the team answers that it cannot advise the client, but explains the published practice.
Adjacent functions. Accounting depends on data it does not produce. The sustainability-linked margin grid sits in the loan system maintained by Loan Agency (see B17); the KPI results come from the borrower’s compliance certificate, checked by the sustainability coordinator (see B03). Scenario weights come from Economic Research and Risk (see E01 and E08). Fair value inputs for power and carbon come from the trading desks and are checked by product control (see C10). The auditors test all of it. A recurrent finding of auditors and supervisors is that the “ESG flag” on loans is set manually, inconsistently between systems, and without the grid that the IFRS 7 disclosure needs.
Why a sustainable-finance course needs this module. Three reasons. Every product of this course ends in a set of accounts, and a structurer who proposes a feature that forces fair value measurement will see the deal blocked by the client’s CFO. The accounting of sustainable instruments changed materially in 2026, with two sets of IFRS 9 amendments entering into force on 1 January 2026. And the bank’s sustainability metrics are built on accounting values: the Green Asset Ratio uses gross carrying amounts, and the Pillar 3 ESG templates reconcile to FINREP.
The product of an accounting function is a measurement. Before adding any sustainability feature, a student must master four pieces of IFRS 9: classification, the effective interest rate, impairment, and the treatment of derivatives and hedges.
IN ONE SENTENCE — IFRS 9 puts every financial asset in one of three measurement boxes, amortised cost, fair value through other comprehensive income (FVOCI) or fair value through profit or loss, according to why the entity holds it (the business model) and what its cash flows look like (the SPPI test).
INTUITION — Think of a landlord. If the flat is bought to rent out, its value is the rent (amortised cost). If it is bought to rent out but sold when a good offer comes, the owner watches both rent and price (FVOCI). If it is bought to resell quickly, only the price matters (FVTPL). The SPPI test then checks that what the tenant pays is really rent, not a share of the tenant’s business.
Business model. It is assessed at portfolio level, on how the portfolio is managed and reported to key management, how managers are paid and the history of sales. “Hold to collect” (HTC) portfolios, such as Corporate Banking’s loan book, collect contractual cash flows; occasional sales because of credit deterioration are compatible. “Hold to collect and sell” (HTC&S) portfolios, such as Treasury’s liquidity portfolio, collect cash flows and sell to manage liquidity or yield. Everything else, including trading, is “other” and goes to FVTPL.
SPPI. The contractual cash flows must be solely payments of principal and interest on the principal outstanding, where interest is consideration for the time value of money, credit risk, other basic lending risks (liquidity), costs and a profit margin. A loan whose return varies with the borrower’s revenue or with an equity index fails. So does a loan with leverage on interest. A floating rate with a margin passes; a prepayment at par plus reasonable compensation passes.
Liabilities. Financial liabilities are measured at amortised cost, except those held for trading or designated at FVTPL. A hybrid liability that contains an embedded derivative not closely related to the host must be split: the derivative goes to FVTPL. For assets, there is no splitting: a hybrid asset is assessed as a whole by the SPPI test. This asymmetry is central to section 5.5.
FORMULA
is the amount initially recognised (price paid net of fees received, or proceeds net of fees paid for a liability); the estimated contractual cash flows of period ; the effective interest rate; the amortised cost at the end of period ; interest income (or expense) of period is .
The EIR spreads every fee that is an integral part of the yield over the expected life. An arrangement fee received by a lender reduces the initial carrying amount and therefore increases the EIR above the coupon; interest income is higher than the cash coupon each year, and the carrying amount accretes back to par. Fees that are not part of the yield (a fee for a service performed, or a commitment fee when drawing is not probable) fall under IFRS 15 and are recognised when the service is delivered or over the commitment period.
INTERACTIVE LAB — Effective interest rate and amortised cost of a loan with an upfront fee
See how an upfront fee raises the effective interest rate above the coupon and how the carrying amount accretes back to par.
| Parameter you can move | Range | Default |
|---|---|---|
| Loan nominal | 1 to 100 EUR m | 10 EUR m |
| Fixed coupon | 0 to 10 % | 4 % |
| Upfront fee received | 0 to 5 % of nominal | 1 % of nominal |
| Tenor | 1 to 10 years | 5 years |
| Result at default settings | Value |
|---|---|
| Effective interest rate | 4.226 % |
| Interest income in year 1 | 0.4184 EUR m |
| Cash coupon each year | 0.4000 EUR m |
| Amortised cost at end of year 1 | 9.9184 EUR m |
Try it. Set the fee to 0: the EIR equals the coupon and the line is flat. Then raise the tenor to 10 years with a 1% fee: the EIR falls, because the fee is spread over more years.
On a EUR 10m five-year loan at 4% with a 1% fee, the lender recognises EUR 9.9m, an EIR of 4.226%, and EUR 0.418m of interest income in year 1 against a coupon of EUR 0.400m. The EUR 0.1m fee is not revenue on the signing day; it is interest, spread over five years.
Changes in estimated cash flows. When the lender or the issuer revises its estimate of future contractual cash flows (other than for credit losses), IFRS 9 (paragraph B5.4.6) requires it to recompute the carrying amount as the present value of the revised cash flows at the original EIR, and to book the difference immediately in profit or loss: the catch-up adjustment. For floating-rate instruments, the periodic re-estimation of cash flows to reflect movements in market rates changes the EIR instead (paragraph B5.4.5). Which of the two paragraphs governs a margin that moves with a KPI is exactly the question sustainable instruments raise.
FORMULA
is the probability weight of macroeconomic scenario ; the horizon (one year in Stage 1, the remaining life in Stage 2 and Stage 3); the marginal probability of default in year ; the loss given default; the exposure at default, including expected drawings on commitments; discounting is at the EIR.
Every asset at amortised cost or FVOCI, and every loan commitment, carries an allowance from day one. Stage 1: 12-month ECL. Stage 2: lifetime ECL when credit risk has increased significantly since initial recognition (SICR), assessed by comparing the lifetime PD now with the lifetime PD expected at origination for the same remaining life; 30 days past due is a backstop. Stage 3: credit-impaired, lifetime ECL and interest on the net carrying amount. The ECL must be unbiased and probability-weighted, and must use “reasonable and supportable” forward-looking information. Unlike the regulatory expected loss of the IRB approach (a through-the-cycle PD and a downturn LGD over one year), IFRS 9 uses point-in-time parameters, scenario weighting and lifetime horizons. The two numbers do not match, and the difference flows into capital (section 6.1).
A derivative under IFRS 9 has three features: its value changes with an underlying variable (a rate, a price, an index, a credit rating “or other variable, provided in the case of a non-financial variable that the variable is not specific to a party to the contract”); it requires little or no initial investment; it is settled at a future date. Derivatives are measured at FVTPL. A contract to buy or sell a non-financial item that can be settled net is treated as a derivative unless it was entered into and continues to be held for the entity’s expected purchase, sale or usage requirements: the own-use exemption. An own-use contract is an executory contract, off balance sheet until delivery, unless it is onerous.
Hedge accounting is optional. In a cash-flow hedge, the effective portion of the derivative’s fair value change goes to OCI and is recycled to profit or loss when the hedged forecast transaction affects it; the ineffective portion stays in profit or loss. The hedged forecast transaction must be “highly probable” and the volume must be identifiable, which is where renewable PPAs, whose volumes depend on the weather, used to fail (section 5.6).
IFRS 13 classifies fair values in three levels: quoted prices in active markets (Level 1), observable inputs (Level 2), significant unobservable inputs (Level 3). Under IFRS 9 (paragraph B5.1.2A), when the fair value at initial recognition is not evidenced by a Level 1 price or a valuation using only observable data, the difference between the model value and the transaction price is deferred and recognised only as the inputs become observable or the instrument is settled. Product control maintains a day-one profit and loss reserve for this. A ten-year Spanish solar PPA is the textbook Level 3 instrument.
No IFRS standard deals specifically with emission allowances. IFRIC 3 Emission Rights was issued in December 2004 and withdrawn in June 2005, after EFRAG advised against endorsement because it created an accounting mismatch. Since then, entities choose a policy under the IAS 8 hierarchy. Allowances are not financial instruments under IFRS (they are not a contractual right to receive cash), so IFRS 9 does not apply to the allowances themselves, only to net-settled contracts on them that are not own-use. Section 3.5 describes the policies in use.
Sustainable instruments raise five accounting questions, and each of the case’s five instruments illustrates one.
| Instrument | Question | Standard |
|---|---|---|
| Sustainability-linked loan held | Does the KPI feature fail SPPI? | IFRS 9 B4.1, IFRS 7 |
| Green bond held | Does the label change anything? | IFRS 9 |
| SLB issued | Is the step-up an embedded derivative? | IFRS 9 4.3 |
| Renewable PPA | Own-use or derivative; hedge? | IFRS 9 2.4, 6 |
| Allowances, credits | Asset, inventory or nothing? | IAS 38, IAS 2, IAS 37 |
A use-of-proceeds label (green loan, green bond) changes nothing in the contractual cash flows: if the issuer misallocates the proceeds, the investor has no contractual remedy beyond reputation. The instrument is classified like its conventional twin. A sustainability-linked feature changes the cash flows, so it changes the analysis. Climate risk, finally, enters through estimates: ECL, impairment of non-financial assets, useful lives, provisions.
On 30 May 2024 the IASB issued Amendments to the Classification and Measurement of Financial Instruments (Amendments to IFRS 9 and IFRS 7), the result of its post-implementation review of IFRS 9 classification. They are effective for annual reporting periods beginning on or after 1 January 2026, with early application permitted. The EU endorsed them by Commission Regulation (EU) 2025/1047 of 27 May 2025, published in the Official Journal on 28 May 2025 (dates verified on the IFRS Foundation news page and EUR-Lex record in November 2026). They contain four elements:
Transition requires no restatement of comparatives. The SPPI and disclosure elements could be early applied on their own.
PITFALL — The amendments did not create a safe harbour for every SLL. “Not significantly different” is not defined by a percentage, the assessment must cover all contractually possible scenarios (not only the probable ones), and a feature that pays the lender on an event that is not specific to the debtor (a national emissions index, a share price) remains outside. A ratchet of 3 bp passes easily; a ratchet of 50 bp on a 70 bp margin needs a documented analysis, and may fail.
On 18 December 2024 the IASB issued Contracts Referencing Nature-dependent Electricity (Amendments to IFRS 9 and IFRS 7), effective for annual reporting periods beginning on or after 1 January 2026, with early application permitted. The EU endorsed them by Commission Regulation (EU) 2025/1266 of 30 June 2025, published on 1 July 2025 (verified on the IFRS Foundation news page, the EUR-Lex record and the German standard-setter’s endorsement note in November 2026). The scope is narrow: contracts that expose the entity to variability in the underlying amount of electricity because generation depends on uncontrollable natural conditions (wind, sun, water), whether physical PPAs or financial ones such as virtual PPAs. The IASB stated that the amendments may not be applied by analogy to other items.
There is no IFRS standard on climate. The IASB’s position is that existing standards already require entities to reflect climate-related risks where material. On 28 November 2025 the IASB published Disclosures about Uncertainties in the Financial Statements, a set of illustrative examples (accompanying material, not part of the standards and therefore not subject to EU endorsement) that use climate-related uncertainties to show how existing requirements apply: materiality judgements, disaggregation, IAS 36 impairment with a sensitive emission-cost assumption, estimation uncertainty, IFRS 7 and IAS 37. A near-final version circulated in July 2025. Because they have no effective date, regulators expect preparers to reflect them in the next reporting cycle.
In April 2024 the IFRS Interpretations Committee finalised an agenda decision on climate-related commitments under IAS 37: a public net-zero pledge can create a constructive obligation, but no provision is recognised when the pledge is made, because there is no past event; costs of reducing future emissions are future operating costs; if the entity has committed to offset its emissions by buying carbon credits, it recognises a provision as it emits, to the extent it can measure it reliably.
For banks, the supervisory channel is stronger than the accounting one. The ECB’s 2020 Guide on climate-related and environmental risks expects banks to reflect these risks in credit risk management, including provisioning, and its July 2024 report on IFRS 9 overlays and model improvements for “novel risks” lists climate and environmental risks among the risks that models capture poorly and that post-model adjustments must address with documented, scenario-based methods. ESMA, in its common enforcement priorities for 2025 annual reports (October 2025), stressed connectivity between financial statements and sustainability statements; its priorities for 2026 reports, published on 13 October 2026, focus on geopolitical risks, restructuring provisions and debt restructuring, not on climate, but the connectivity expectation remains.
Three families of policy coexist for EU ETS allowances held for compliance.
| Policy | Allowances | Obligation |
|---|---|---|
| Net liability (cost) | Free at nil; bought at cost (IAS 38 or IAS 2) | Carrying amount of allowances held, market for shortfall |
| Market-value liability | As above | Whole obligation at market price |
| Gross (IFRIC 3 style) | All at fair value at grant, deferred grant income | Market price |
The net liability approach is the most common among European industrials. Entities that trade allowances in the ordinary course of business hold them as inventory under IAS 2; commodity broker-traders may measure inventories at fair value less costs to sell, with changes in profit or loss (IAS 2 paragraph 3(b)), which is Meridian’s policy for the trading desk. Voluntary carbon credits have no specific guidance either: they are usually intangible assets under IAS 38 (at cost) when held for retirement against the entity’s own commitments, or inventory when held for sale. The IASB researched “pollutant pricing mechanisms” and, on 29 January 2025, decided to defer the decision on whether to add a standard-setting project until its next agenda consultation: in November 2026 there is still no project on the work plan.
The accounting life of a sustainable instrument starts before signing and ends in the annual report. The flow below follows a new type of SLL at Meridian; the other instruments of the case follow the same chain with different owners.
INTERACTIVE LAB — From term sheet to annual report: the accounting chain for a sustainability-linked loan
Follow who decides what, with which document, from the new product approval to the published IFRS 7 note.
| # | Who | Does what | Deliverable |
|---|---|---|---|
| 1 | Loan structuring (Corporate Banking) | Submits the term sheet with the margin grid and KPI definitions to new product approval | NPA request |
| 2 | Accounting policy team | Assesses business model, SPPI including the significance of the ESG-linked feature, fee treatment and disclosure | Accounting memo |
| 3 | External auditor | Reviews the memo for a new feature before first booking | Audit clearance or comments |
| 4 | Loan operations and agency | Books the loan with the ESG flag, the grid and the maximum range of margin change in the loan system | Static data in loan system |
| 5 | Finance business partner | Sets the fee split between EIR and IFRS 15 and the revenue plan | Revenue recognition sheet |
| 6 | IFRS 9 impairment team | Computes Stage 1 ECL with scenario weights including the transition scenario | ECL run and overlay proposal |
| 7 | Provisions committee | Approves scenarios, weights, overlays and stage transfers | Committee minutes |
| 8 | Sustainability coordinator and agent | Receive the annual compliance certificate and verifier report; notify the new margin | Margin adjustment notice |
| 9 | Product control or loan accounting | Applies the new margin prospectively or books a catch-up per policy | Accounting entry |
| 10 | Financial reporting | Aggregates gross carrying amounts and ranges of cash-flow changes for IFRS 7 20B to 20D | Disclosure note draft |
| 11 | Sustainability reporting team | Uses the same gross carrying amounts for GAR and Pillar 3 ESG | Reconciled templates |
| 12 | Audit committee | Reviews key judgements and approves the financial statements | Signed annual report |
Step 1, the NPA. Any instrument with a feature new to the bank goes through the new product approval process, chaired by the business line’s COO, with Risk, Compliance, Legal, Tax, Operations and Finance signing off. Accounting’s question is narrow and concrete: can the bank’s systems and policies handle the instrument without manual workarounds? For an SLL, this means a field for the KPI-linked margin component, a field for the maximum range of change and a flag that feeds the disclosure.
Step 2, the memo. The accounting memo is a four- to ten-page document: facts, issues, references to the standards, analysis of alternatives, conclusion, and the disclosure consequence. For a new ESG-linked feature the core of the memo is the SPPI analysis of section 5.1: is the event specific to the debtor, and are the cash flows “significantly different” in any possible scenario? The memo is a precedent: later deals with the same features cite it rather than repeating the work, which is why the policy team insists on a template term sheet.
Step 3, the auditors. The auditors do not approve the accounting, but they will test the judgement at the next close. Raising a new product early avoids a restatement later. In the first years of SLLs, auditors often asked for the SPPI analysis to be documented loan by loan; since the 2024 amendments clarified the test, banks rely more on a template analysis reused for every deal with standard features, with a deal-level check only when the grid departs from the template.
Steps 4 and 5, booking and fees. Loan operations capture the static data. The finance business partner decides how fees are recognised: an upfront fee on a revolving facility is split between the part related to expected drawings (deferred and included in the EIR of the drawn loans) and the part related to the undrawn commitment (recognised over the commitment period under IFRS 15). The sustainability coordination fee that Meridian earns as SLL coordinator (see B03) is a fee for a service under IFRS 15, recognised when the service is delivered.
Steps 6 and 7, impairment. The ECL is computed monthly or quarterly. The Provisions Committee, chaired by the CFO and the CRO, approves the scenarios and weights produced by Economic Research, the model outputs and any overlay. Climate enters as a scenario, as a sector overlay, or as a qualitative SICR trigger (section 5.3).
Steps 8 and 9, the annual KPI test. The borrower’s compliance certificate and the verifier’s report arrive within 120 to 180 days after the financial year end. The agent notifies the new margin (see B17). Accounting applies the policy chosen in step 2: for revolving drawings priced off a floating rate, a prospective change; for a fixed-rate term loan or bond, a catch-up adjustment under B5.4.6 (section 5.2).
Steps 10 to 12, reporting. At year end, financial reporting collects, for all instruments flagged as having ESG-linked features, the gross carrying amount and the range of possible changes in cash flows. The same gross carrying amounts feed the Sustainability Reporting team’s GAR and Pillar 3 templates (see E06). The audit committee reviews the key judgements: ECL overlays, Level 3 valuations, the SPPI conclusions of new products. A bank that discloses EUR 8bn of SLLs in its sustainability statement and EUR 5bn in its IFRS 7 note will receive a question from its enforcer, because ESMA’s connectivity expectation applies to exactly this kind of figure.
WHY IT MATTERS ON THE DESK — Every hand-off in this chain is a control point. The most frequent weakness is not a wrong policy but a broken data chain: a margin grid not captured, a KPI result not fed back, a flag set differently in the loan system and in the sustainability data warehouse.
The case follows five instruments already modelled elsewhere in the course, now seen through the eyes of the accountants. Three sit in Meridian’s books: its EUR 75m share of the Borealis Packaging sustainability-linked RCF (see B03), a position in the Helios Renewables green bond held by Treasury (see C01), and the carbon inventory of the commodities desk (see C10). Two sit in the books of Meridian’s client Castellan Cement: its EUR 600m SLB (see C02) and its Spanish solar virtual PPA (see C10). Meridian’s accounting policy team does not account for Castellan, but the DCM, structuring and commodities desks must understand what their products do to a client’s accounts, and the credit analysts must read those accounts.
Common assumptions for the case: Meridian’s financial year is the calendar year; EURIBOR and the swap curve are flat at 2.50%; all amounts are case assumptions unless the text cites another module.
Question. Meridian holds EUR 75m of the Borealis RCF in its Corporate Banking loan book (a hold-to-collect portfolio). Are the drawn loans measured at amortised cost, or does the KPI ratchet force fair value through profit or loss?
Data (from B03). Drawn margin 70 bp over EURIBOR; commitment fee 35% of the margin (24.5 bp); a symmetrical per-KPI ratchet of 1 bp per KPI, three KPIs, so a maximum of plus or minus 3 bp on the margin, which flows into the commitment fee through the 35% ratio; tenor five years. Meridian’s average share drawn: EUR 15m; undrawn EUR 60m. The KPIs are Borealis’s own scope 1 and 2 emissions, recycled content and a third operational KPI; the results are verified by an external party.
Step 1: the nature of the event. The trigger is the borrower meeting or missing targets on its own emissions and operations: an event specific to the debtor. The change in interest does not introduce exposure to an unrelated risk such as an equity price or a commodity index. The feature is “not directly related to changes in basic lending risks or costs”, so the quantitative condition of paragraph B4.1.10A applies: the cash flows must not be significantly different from those of the same loan without the ratchet, in any contractually possible scenario.
Step 2: the quantitative test. The worst contractually possible scenario is the maximum adjustment applied every year. Meridian’s accounting manual compares two measures with the benchmark instrument: the change in the annual interest rate, and the change in the present value of all contractual cash flows.
FORMULA
is the maximum margin change in either direction (3 bp); the reference rate (2.50%); the base margin (70 bp); the annuity factor at rate over years, so that is the present value of the maximum change as a share of the principal of a loan priced at par.
INTERACTIVE LAB — SPPI significance test for an ESG-linked margin ratchet
Measure how far the cash flows of the Borealis SLL can deviate from those of the same loan without the ratchet, and how the answer changes for larger ratchets.
| Parameter you can move | Range | Default |
|---|---|---|
| Meridian’s average drawn amount | 0 to 75 EUR m | 15 EUR m |
| Meridian’s undrawn commitment | 0 to 75 EUR m | 60 EUR m |
| EURIBOR | 0 to 6 % | 2.5 % |
| Base margin | 20 to 300 bp | 70 bp |
| Commitment fee as share of margin | 20 to 50 % | 35 % |
| Maximum ratchet in either direction | 0 to 100 bp | 3 bp |
| Tenor | 1 to 10 years | 5 years |
| Result at default settings | Value |
|---|---|
| Maximum change in annual interest rate | 0.94 % of interest |
| Maximum change in PV of cash flows | 0.137 % of principal |
| Maximum annual change on drawn amount | 4,500 EUR |
| Maximum annual change, drawn and undrawn | 10,800 EUR |
Try it. Raise the ratchet to 16 bp: the change in annual interest reaches the 5% review trigger of Meridian’s manual. Then try 50 bp: a sixth of the interest depends on the KPIs, and the memo would have to argue that this is still not ‘significantly different’.
Result. The ratchet can change the annual interest rate on drawn amounts by 0.94% (3 bp out of 320 bp) and the present value of the contractual cash flows by 0.137% of the principal. In euros, Meridian’s share of interest and fees can move by at most EUR 4,500 a year on drawn amounts and EUR 10,800 including the commitment fee. The loans are measured at amortised cost, and the undrawn commitment stays off balance sheet with an ECL provision.
Reading. Three points make the conclusion robust. The measure covers all possible scenarios (every target missed every year), not the expected path. The test is made against the same loan without the feature, not against market rates. And the result is far from any reasonable threshold: the IASB deliberately gave no percentage, so Meridian’s manual sets a review trigger, not a pass mark, at 5% of interest: below it, a standard memo; above it, a detailed analysis signed by the head of accounting policy. A 16 bp ratchet on this loan would reach the trigger. Ratchets of 2.5 to 10 bp, which is where the market is (see B03), stay well below it.
PITFALL — A ratchet that pays the lender more when the borrower misses a target is not a problem for SPPI; a feature where the borrower pays a fixed penalty to a charity, or where the margin depends on the bank’s own portfolio emissions, needs a separate analysis. And an SLL that passes SPPI is still disclosed: the amendments add IFRS 7 information precisely for the instruments that stay at amortised cost.
Question. How does Meridian recognise the upfront fee, and how does it account for the margin change when Borealis misses or meets its targets?
Data. Upfront (participation) fee: 25 bp on EUR 75m, EUR 187,500 (B03). Meridian’s expectation of drawing: 20% of the commitment on average (EUR 15m of EUR 75m). Accounting policy: the share of a commitment fee or upfront fee corresponding to expected drawings is deferred and included in the EIR of the drawn loans; the rest is recognised on a straight-line basis over the commitment period under IFRS 15. For the ratchet, the case also studies a variant: a EUR 15m five-year term loan to Borealis with the same grid, whose margin rises by 1 bp after the first annual test, for the remaining four years.
Policy choice for the margin change. Two readings of IFRS 9 compete. One treats the ESG adjustment like a floating-rate reset: the new margin applies prospectively and changes the EIR (B5.4.5). The other treats it as a revision of estimated cash flows: the carrying amount is recalculated at the original EIR and the difference is booked immediately (B5.4.6). Most large banks apply the second reading to fixed-rate instruments and to term loans, and treat short revolving drawings, which are repriced at each rollover, prospectively in practice because the catch-up is negligible.
INTERACTIVE LAB — Upfront fee split and the catch-up adjustment when an ESG margin changes
Split the upfront fee between the EIR and IFRS 15, and compare the catch-up adjustment under B5.4.6 with a prospective treatment.
| Parameter you can move | Range | Default |
|---|---|---|
| Meridian’s commitment | 10 to 200 EUR m | 75 EUR m |
| Upfront fee | 0 to 100 bp | 25 bp |
| Expected average drawing | 0 to 100 % | 20 % |
| Commitment period | 1 to 7 years | 5 years |
| Term loan variant: nominal | 1 to 100 EUR m | 15 EUR m |
| EIR of the term loan | 0 to 8 % | 3.25 % |
| Margin change after the test | -5 to 5 bp | 1 bp |
| Remaining years after the test | 1 to 6 years | 4 years |
| Result at default settings | Value |
|---|---|
| Upfront fee | 187,500 EUR |
| Part deferred into the EIR | 37,500 EUR |
| IFRS 15 part recognised each year | 30,000 EUR |
| Approximate EIR uplift on drawn amounts | 5.00 bp |
| Catch-up booked on the test date (B5.4.6) | 5,542 EUR |
| Prospective change in interest each year | 1,500 EUR |
Try it. Set the change to -3 bp (all targets met): the catch-up is a loss of about EUR 16,600 for the lender on the test date. Change the remaining years to 1: the catch-up and the prospective change converge.
Result. Of the EUR 187,500 upfront fee, EUR 37,500 (the 20% related to expected drawings) is deferred and raises the EIR of the drawn loans by about 5 bp a year; EUR 30,000 a year of the rest is recognised over the five-year commitment under IFRS 15. In total the income statement shows the same EUR 37,500 a year as the RAROC of B03, but in two different lines: interest income and fee income. On the term-loan variant, a 1 bp increase for four years changes interest by EUR 1,500 a year; under the catch-up method the lender books EUR 5,542 on the test date and then earns the original EIR.
Reading. The amounts are immaterial for Meridian, which is the honest conclusion: the ESG feature of an SLL is a disclosure and data problem far more than a measurement problem. The catch-up matters in two situations: large instruments with long remaining lives (a EUR 1bn ten-year SLB held by an insurer), and the timing of the estimate. Under B5.4.6 the revision is booked when the estimate changes, which may be before the formal test: a lender that knows in December that a borrower will miss its target should, in principle, revise then.
Question. What ECL does Meridian hold on its Borealis exposure, how does a climate transition scenario enter it, and what would push the exposure to Stage 2?
Data (case assumptions). Exposure at default for IFRS 9 purposes: drawn EUR 15m plus 40% of the undrawn EUR 60m, EUR 39m (Meridian’s behavioural drawdown assumption happens to equal the regulatory CCF). LGD 40%. EIR used for discounting 3.25%. Remaining life four years. One-year PD at origination: 0.20% (BBB). Economic Research’s scenarios at year end 2026, with one-year point-in-time PDs for Borealis produced by the sector model:
| Scenario | Weight | PD Borealis |
|---|---|---|
| Baseline | 50% | 0.18% |
| Adverse macro | 30% | 0.35% |
| Disorderly transition | 20% | 0.60% |
The disorderly transition scenario is built from the NGFS “delayed transition” narrative (see E08): a sharp rise in carbon prices after 2028, the EU packaging regulation tightened, and a fall in demand for rigid plastics. It is the channel by which climate enters the ECL. Meridian’s SICR rule: Stage 2 if the probability-weighted PD has risen to at least 2.5 times the PD at origination (the bank’s quantitative criterion compares lifetime PDs; the case uses one-year PDs as a proxy).
INTERACTIVE LAB — Probability-weighted ECL on the Borealis hold with a transition scenario
Compute the weighted PD, the 12-month and lifetime ECL, compare with the regulatory expected loss and test for a significant increase in credit risk.
| Parameter you can move | Range | Default |
|---|---|---|
| Weight of adverse macro scenario | 0 to 60 % | 30 % |
| Weight of disorderly transition scenario | 0 to 70 % | 20 % |
| PD in baseline | 0.03 to 5 % | 0.18 % |
| PD in adverse macro | 0.03 to 10 % | 0.35 % |
| PD in disorderly transition | 0.03 to 10 % | 0.6 % |
| PD at origination | 0.03 to 5 % | 0.2 % |
| LGD | 10 to 80 % | 40 % |
| Exposure at default | 1 to 100 EUR m | 39 EUR m |
| Discount rate (EIR) | 0 to 8 % | 3.25 % |
| Remaining life | 1 to 7 years | 4 years |
| Result at default settings | Value |
|---|---|
| Weight of baseline | 50 % |
| Probability-weighted PD | 0.315 % |
| 12-month ECL (Stage 1) | 49.14 EUR k |
| Lifetime ECL (Stage 2) | 180.74 EUR k |
| Regulatory expected loss (PD at origination) | 31.20 EUR k |
| Weighted PD / PD at origination | 1.575 x |
| Stage under Meridian’s SICR rule | 1 |
| ECL added by the transition scenario versus giving its weight to the baseline | 13.10 EUR k |
Try it. Double the weight of the transition scenario to 40%: the Stage 1 ECL rises by EUR 13,000 but Borealis stays in Stage 1. Then raise all three PDs as in a downgrade after a debt-funded acquisition (0.45%, 0.80%, 1.60%): the ratio crosses 2.5 and the lifetime ECL is nine times the original Stage 1 figure.
Calculation. Weighted PD: . Stage 1 ECL: $0.315% % = $ EUR 49,140. The ratio to the origination PD is 1.575, below the 2.5 trigger: Stage 1. Had the exposure been in Stage 2, the lifetime ECL over four years, discounted at 3.25%, would have been EUR 180,740.
Result. Meridian holds EUR 49,140 against Borealis, of which the part related to the undrawn commitment is presented as a provision on the liability side. That is 57.5% more than the regulatory expected loss of EUR 31,200 used in the RAROC (see B03 and section 6), the same 1.575 ratio as the PDs because LGD and EAD are common. Without the transition scenario (its 20% weight given to the baseline), the ECL would be EUR 36,040: the climate scenario adds about EUR 13,100 on this name.
A stage transfer. In the spring of 2028 Borealis announces a debt-funded acquisition (the event of B03 section 5.7), and its sector model PDs rise to 0.45%, 0.80% and 1.60%. The weighted PD becomes 0.785%, the ratio 3.9: the exposure moves to Stage 2 and the allowance jumps to EUR 447,330, an ECL charge of almost EUR 400,000 in one quarter, nearly twice the annual revenue of the hold.
Reading. Three lessons. First, climate enters ECL mostly through scenario weights and sector PD paths, and for a single investment-grade name the effect is small; it becomes material at portfolio level, in carbon-intensive sectors, and through stage transfers. Second, the ECL and the SLL ratchet are independent: missing a sustainability target is not in itself a significant increase in credit risk, but the reasons for the miss (a failing transition plan, an acquisition of a carbon-intensive business) may well be. Third, the transfer cliff, not the Stage 1 level, is what hurts profit and loss. This is why supervisors ask banks to document how transition risk feeds the SICR assessment, not only the scenario weights.
WHO DOES WHAT — When the impairment team proposes a climate overlay, the auditors’ first question is always the same: “Is this already in the model?” A sector PD model calibrated on past defaults contains no carbon price shock, so an overlay can be justified; but if the scenario already contains the shock, the overlay double counts. The overlay memo must show the reconciliation.
Question. Meridian’s Treasury bought Helios green bonds at issue for its liquidity portfolio. How are they measured, and what happens when yields rise?
Data (case assumptions, bond terms from C01). Nominal bought: EUR 25m at the reoffer price of 99.864 (cost EUR 24.966m); coupon 3.95% annual; eight years. The liquidity portfolio is managed to collect coupons and to sell when Treasury needs liquidity or rebalances (HTC&S, see C15). Helios is rated BBB; Treasury’s ECL parameters: 12-month PD 0.20%, LGD 40%. Scenario for the first year end: the bond’s yield rises by 40 bp (a widening of the issuer’s spread and of the OAT curve), with seven years remaining.
Classification. The green label imposes an allocation of proceeds and reporting, but the bond’s cash flows are a fixed coupon and principal, with no contractual link to the use of proceeds: SPPI is met without any reference to the 2024 amendments. With an HTC&S business model the bond is at FVOCI. Interest at the EIR and ECL go to profit or loss; the rest of the fair value change goes to OCI and is recycled on sale.
INTERACTIVE LAB — A green bond at FVOCI: interest, ECL and the OCI reserve after a rise in yields
Record one year of the Helios green bond held at FVOCI, and compare profit and loss with amortised cost and FVTPL.
| Parameter you can move | Range | Default |
|---|---|---|
| Nominal held | 5 to 100 EUR m | 25 EUR m |
| Purchase price | 95 to 105 % of par | 99.864 % of par |
| Coupon | 0 to 8 % | 3.95 % |
| Maturity at purchase | 2 to 15 years | 8 years |
| Change in yield after one year | -150 to 150 bp | 40 bp |
| 12-month PD | 0.03 to 5 % | 0.2 % |
| LGD | 10 to 80 % | 40 % |
| Result at default settings | Value |
|---|---|
| EIR at purchase | 3.970 % |
| Interest income year 1 | 0.9912 EUR m |
| Amortised cost at end of year 1 | 24.9697 EUR m |
| Fair value at end of year 1 | 24.3781 EUR m |
| Stage 1 ECL | 0.0200 EUR m |
| OCI reserve at end of year 1 | -0.5717 EUR m |
| Profit and loss, FVOCI or amortised cost | 0.9712 EUR m |
| Profit and loss if FVTPL | 0.3996 EUR m |
Try it. Set the yield change to 0: the OCI reserve equals the ECL, because at FVOCI the allowance is booked in profit or loss against OCI rather than against the carrying amount. Then try -100 bp: FVTPL would show a profit of more than EUR 2.5m.
Calculation. The EIR is the purchase yield, 3.970%. Interest income in year 1: $24.966 % = $ EUR 0.9912m, against a coupon of EUR 0.9875m; amortised cost at the year end EUR 24.9697m. Stage 1 ECL: $0.20% % = $ EUR 0.0200m. After a 40 bp rise in yield, the fair value with seven years left is EUR 24.3781m.
Result. The balance sheet shows the bond at EUR 24.378m. Profit and loss shows EUR 0.9712m (interest less ECL), exactly as at amortised cost. The OCI reserve is minus EUR 0.5717m: the fair value loss of EUR 0.5916m below amortised cost, offset by the EUR 0.0200m ECL which, at FVOCI, is charged to profit or loss against OCI and does not reduce the carrying amount. Had the bond been at FVTPL, profit and loss would have shown only EUR 0.3996m.
Reading. The green label changes nothing in this accounting. It matters elsewhere: the bond enters the numerator of Meridian’s Green Asset Ratio to the extent of the alignment of Helios’s allocated proceeds (see E06), measured on its gross carrying amount; and, as a corporate bond rated BBB, it can at best qualify as a Level 2B liquid asset with a 50% haircut under the LCR rules, independently of the label (see C15). The OCI reserve is not a free lunch for capital: unrealised losses on FVOCI securities reduce CET1, because the temporary prudential filter that the EU allowed for some sovereign exposures expired at the end of 2022. A Treasury that buys long green bonds for its GAR without thinking about rate risk buys CET1 volatility.
PITFALL — “We will hold them, so we can put them at amortised cost” is a classic error. The business model is a fact about how the portfolio is managed, not an intention about one bond. A liquidity portfolio that is sold to raise cash in stress, by definition, is not hold to collect.
Question. How does Castellan account for its EUR 600m SLB, and what happens in its accounts when it misses the FY2029 target?
Data (from C02). Nominal EUR 600m; coupon 3.625% annual; seven years from 17 November 2026; net proceeds EUR 597.524m; step-up of 25 bp from the interest period starting 17 November 2030, so three higher coupons paid in 2031, 2032 and 2033 (EUR 1.5m each); probability of a miss estimated by Castellan at issue: 11.5%. The FY2029 intensity misses the target by 6 kg (C02 section 5.7). For simplicity the revision is computed at the anniversary of 17 November 2029, close to Castellan’s year end, when management already knows that the target will be missed.
Is the step-up an embedded derivative? The SLB is a financial liability at amortised cost. For a liability, an embedded derivative must be separated if its economic characteristics are not closely related to the host and a separate instrument with the same terms would be a derivative. The definition of a derivative excludes contracts whose underlying is a non-financial variable specific to a party to the contract. Castellan’s scope 1 intensity is a non-financial variable, and it is specific to Castellan. The prevailing view, followed by most European issuers and their auditors, is therefore that the step-up is not a derivative and stays in the host, measured at amortised cost. A minority view argues that emissions are influenced by market variables (output, carbon prices) or that the variable is not “specific” enough when it is a group-wide KPI that third parties also observe; under that view the step-up would be measured at FVTPL. The May 2024 amendments addressed the asset side (SPPI) and disclosures for both sides, but not the embedded derivative question for liabilities, which the IASB left to existing requirements.
Measurement. Castellan’s initial EIR is computed on its estimate of cash flows. Policy A uses the most likely outcome (target met, no step-up). Policy B includes the step-up weighted by the estimated probability of a miss. When the estimate changes, B5.4.6 applies: the carrying amount is recomputed at the original EIR and the difference is booked in finance costs.
INTERACTIVE LAB — Castellan’s SLB at amortised cost: EIR, carrying amount and the catch-up on a missed target
Compute the issuer’s EIR, the carrying amount when the miss becomes known, and the catch-up loss under two estimation policies.
| Parameter you can move | Range | Default |
|---|---|---|
| Nominal | 100 to 1500 EUR m | 600 EUR m |
| Coupon | 0 to 8 % | 3.625 % |
| Net proceeds | 90 to 1500 EUR m | 597.524 EUR m |
| Step-up | 0 to 100 bp | 25 bp |
| Number of stepped-up coupons | 1 to 6 | 3 |
| Probability of a miss at issue (policy B) | 0 to 100 % | 11.5 % |
| Year in which the miss becomes known | 1 to 4 year | 3 year |
| Result at default settings | Value |
|---|---|
| EIR, policy A (no step-up expected) | 3.693 % |
| EIR, policy B (probability-weighted) | 3.704 % |
| Carrying amount before revision, policy A | 598.509 EUR m |
| Catch-up loss, policy A | 4.038 EUR m |
| Carrying amount after revision, policy A | 602.547 EUR m |
| Catch-up loss, policy B | 3.572 EUR m |
| Next-year interest expense without the miss | 22.103 EUR m |
| Next-year interest expense after the miss | 22.252 EUR m |
Try it. Set the probability at issue to 100%: policy B then books no catch-up at all, and spreads the step-up over the whole life through a higher EIR. Set the step-up to 75 bp, a level seen in some early SLBs: the catch-up triples.
Calculation. Policy A: the EIR on the cash flows without step-up is 3.693%, the all-in cost of C02. Three years after issue the carrying amount has accreted from EUR 597.524m to EUR 598.509m. When the miss becomes certain, the carrying amount is recomputed with the three extra coupons of EUR 1.5m, discounted at 3.693%: EUR 602.547m. The difference, EUR 4.038m, is a finance cost booked at once. Interest expense in the following year rises from EUR 22.103m to EUR 22.252m. Policy B: the probability-weighted EIR is 3.704%; the catch-up is smaller, EUR 3.572m, because 11.5% of the step-up was already being accrued.
Result. Missing the target costs Castellan EUR 4.5m of extra coupons (C02), but in the accounts the cost appears as a EUR 4.04m charge in FY2029, the year of the miss, not as EUR 1.5m a year in 2031 to 2033. The present value is front-loaded.
Reading. Two consequences for bankers. For the credit analyst reading Castellan’s FY2029 accounts, a one-off finance cost labelled “remeasurement of sustainability-linked bond” is the accounting trace of a missed target, and should be read with the sustainability statement. For the DCM desk, an issuer that fears this visibility may prefer policy B, or a step-up structure with smaller amounts; but the accounting should follow the economics, not drive them. Had the step-up been separated as a derivative at FVTPL, Castellan would have recorded fair value changes every quarter as the market’s probability of a miss moved (C02 values the step-up at EUR 0.42m to EUR 1.12m at issue depending on the probability used), a volatility no treasurer wants.
NOTE — The issuer’s catch-up is the mirror of the holder’s. An insurer holding the SLB at amortised cost books a gain of the same order on its share, and an investor holding it at FVTPL simply sees the price move, as in C02 section 5.6.
Question. How does Castellan account for its ten-year virtual PPA signed at EUR 32 per MWh, what do the December 2024 amendments change, and would a physical PPA have been own-use?
Data (from C10). Pay-as-produced vPPA on a 150 MW Spanish solar park, 262,800 MWh in year 1, degradation 0.4% a year, ten years from 2027; strike EUR 32; capture rate 60% falling by 1.5 points a year; baseload forwards EUR 60 (2027 to 2029) and EUR 55 (from 2030); discount rate 3.70%. Value at signing on Meridian’s model: minus EUR 3.05m. Case assumptions for the first year end: the liquid 2027 to 2029 forwards rise by EUR 10 per MWh, the long-term view is unchanged; hedge ineffectiveness 5% of the fair value change (negative-price clause and timing differences). For the physical alternative: consumption of the Iberian plants 550 GWh a year; 20% of the solar output arrives in hours when it exceeds the plants’ load and must be sold to the grid (Meridian’s power desk’s hourly simulation).
The vPPA is a derivative. It is net settled and Castellan never receives the electricity from the park, so own-use is not available: the contract is at FVTPL. It is a Level 3 instrument (capture rates and prices after 2029 are unobservable), so at inception the fair value is the transaction price, zero, and the model value of minus EUR 3.05m is a day-one difference that is deferred and recognised over the life of the contract.
What the amendments change. Before 2026, designating a pay-as-produced vPPA as a cash-flow hedge was hard: the hedged item had to be a specified, highly probable volume of forecast purchases, while the vPPA’s volume depends on the sun. The December 2024 amendments allow Castellan to designate as the hedged item a variable volume of its forecast electricity purchases equal to the volume the park actually produces, and presume that volume highly probable. Changes in the vPPA’s fair value then go to OCI, to be recycled when the hedged purchases affect profit and loss.
INTERACTIVE LAB — The vPPA in Castellan’s accounts: day-one difference, hedge accounting and the own-use test of a physical PPA
Measure the vPPA at signing and after a move in forward prices, compare profit and loss with and without cash-flow hedge accounting, and test the net-purchaser criterion of a physical PPA.
| Parameter you can move | Range | Default |
|---|---|---|
| Strike | 20 to 60 EUR/MWh | 32 EUR/MWh |
| Baseload forward 2027-2029 | 30 to 120 EUR/MWh | 60 EUR/MWh |
| Baseload view from 2030 | 30 to 120 EUR/MWh | 55 EUR/MWh |
| Capture rate in 2027 | 40 to 100 % | 60 % |
| Annual fall in capture rate | 0 to 4 points | 1.5 points |
| Discount rate | 1 to 8 % | 3.7 % |
| Move in 2027-2029 forwards at year end | -30 to 30 EUR/MWh | 10 EUR/MWh |
| Ineffective share of the change | 0 to 50 % | 5 % |
| Physical PPA: share of output sold to the grid | 0 to 80 % | 20 % |
| Physical PPA: annual consumption | 100 to 1000 GWh | 550 GWh |
| Result at default settings | Value |
|---|---|
| Model value at signing | -3.05 EUR m |
| Day-one difference released each year (straight line) | 0.305 EUR m |
| Change in fair value at year end | 4.277 EUR m |
| Profit and loss without hedge accounting | 4.277 EUR m |
| OCI with hedge accounting | 4.063 EUR m |
| Profit and loss with hedge accounting | 0.214 EUR m |
| Physical PPA: electricity sold to the grid | 52.56 GWh |
| Physical PPA: electricity bought from the grid | 339.76 GWh |
| Purchases / sales | 6.46 x |
Try it. Move the forwards by minus EUR 20: without hedge accounting Castellan reports a loss of more than EUR 8m in a year in which nothing has happened to its plants. Then raise the share sold to the grid to 80%: the purchases-to-sales ratio falls but stays above 1, because the plants still buy at night.
Calculation. At signing the model value is minus EUR 3.05m: Castellan recognises the vPPA at zero and defers the difference, released at about EUR 0.305m a year if the policy is straight-line. At the year end, a EUR 10 rise in the 2027 to 2029 forwards raises the value of the vPPA by EUR 4.277m (the sum over three years of output times capture rate times EUR 10, discounted). Without hedge accounting the whole gain is in profit and loss. With a cash-flow hedge designated under the amended rules, EUR 4.063m goes to OCI and only the ineffective EUR 0.214m reaches profit and loss.
The physical alternative. Had Castellan signed a physical PPA with delivery to its Iberian plants, it would sell 52.56 GWh a year back to the grid at midday, when production exceeds load, and buy 339.76 GWh from the grid at other hours: purchases are 6.5 times sales, the surplus is bought back in the same market within the day, and Castellan has been and expects to remain a net purchaser. Under the amended own-use rules the physical PPA is an executory contract, off balance sheet, with IFRS 7 disclosures of the volume variability and the estimated future purchase cash flows. Before the amendments, regular sales of surplus would have put own-use at risk and forced the contract to FVTPL.
Reading. The amendments remove a genuine accounting obstacle to renewable PPAs, which the IASB took up in early 2024 at the request of preparers and finalised within the same year, a rare speed for standard-setting. For Meridian’s structuring and commodities desks, the practical consequence is that a corporate client can now hedge-account a pay-as-produced vPPA, so the accounting argument against PPAs is weaker. What remains is the economic argument of C10: cannibalisation, basis risk and a Level 3 valuation that the client’s auditors will challenge.
Question. How does Castellan account for its 2027 EU ETS obligation, how does Meridian account for the allowances its desk holds, and how does Meridian account for its own carbon-credit commitment?
Data. Castellan 2027 (from C10): emissions 10.92 Mt; free allocation 9.223 Mt; 1.357 Mt bought forward from Meridian at EUR 87.84 (futures EUR 87.34 plus EUR 0.50 margin) and delivered in December 2027. The forwards were own-use (physically settled, for surrender), so they stayed off balance sheet until delivery. EUA price at 31 December 2027: EUR 90 (case assumption). Castellan’s policy: net liability approach. Meridian’s commodities desk (case assumptions): an inventory of 0.40 Mt bought at EUR 85, hedged with short futures; costs to sell EUR 0.05 per tonne. Meridian’s own operations: 18 kt of residual operational emissions in 2027 that the bank has publicly committed to compensate with carbon-removal credits, expected cost EUR 20 per tonne.
INTERACTIVE LAB — Allowances and credits: Castellan’s emission liability under two policies, Meridian’s inventory and its own offset commitment
Compare the emission liability under the net and market-value policies, measure the trading inventory at fair value less costs to sell, and size the IAS 37 provision for a voluntary offset commitment.
| Parameter you can move | Range | Default |
|---|---|---|
| Castellan emissions 2027 | 5 to 15 Mt | 10.92 Mt |
| Free allocation | 0 to 15 Mt | 9.223 Mt |
| Allowances bought and delivered | 0 to 5 Mt | 1.357 Mt |
| Price paid | 40 to 160 EUR/t | 87.84 EUR/t |
| EUA price at year end | 40 to 160 EUR/t | 90 EUR/t |
| Meridian inventory | 0 to 2 Mt | 0.4 Mt |
| Inventory purchase price | 40 to 160 EUR/t | 85 EUR/t |
| Costs to sell | 0 to 1 EUR/t | 0.05 EUR/t |
| Meridian residual operational emissions | 0 to 100 kt | 18 kt |
| Removal credit price | 5 to 200 EUR/t | 20 EUR/t |
| Result at default settings | Value |
|---|---|
| Castellan shortfall at year end | 0.340 Mt |
| Liability, net approach | 149.80 EUR m |
| Liability, market-value approach | 152.73 EUR m |
| Difference between the two | 2.93 EUR m |
| Meridian inventory at FV less costs to sell | 35.98 EUR m |
| Gain on inventory in profit and loss | 1.98 EUR m |
| Loss on short futures hedge | -2.00 EUR m |
| Meridian provision for its offset commitment | 0.36 EUR m |
Try it. Lower the year-end price to EUR 70: under the market-value approach the liability falls below the net approach, because allowances bought at EUR 87.84 stay at cost while the obligation they will settle is marked down.
Calculation. Castellan’s shortfall at the year end is Mt. Under the net approach the liability is the carrying amount of allowances held (free ones at nil, bought ones at EUR 87.84) plus the shortfall at EUR 90: EUR 149.80m. Under a market-value approach the whole obligation in excess of free allowances is at EUR 90, EUR 152.73m, while the bought allowances stay at cost: a EUR 2.93m mismatch that reverses when they are surrendered. At Meridian, the 0.40 Mt inventory is measured at EUR 89.95, EUR 35.98m, a gain of EUR 1.98m, offset by the EUR 2.0m loss on the short futures (the difference is the costs to sell). Meridian’s own commitment generates a provision of EUR 0.36m, recognised as it emits during 2027, in line with the April 2024 IFRIC agenda decision.
Reading. Three policies for the same allowances produce liabilities that differ by a few million euros at EUR 90 and by over EUR 24m at EUR 70. None is wrong; all must be disclosed, and analysts comparing two cement groups must restate. For Meridian the essential point is the policy symmetry on the desk: inventory at fair value less costs to sell matches futures at FVTPL. Had a new sub-portfolio been booked at cost under IAS 38, the futures hedge would have created EUR 2.0m of artificial volatility, which is exactly the reconciliation the accounting policy team performed on the Thursday of section 1. Voluntary credits bought for the bank’s own commitment are intangible assets at cost, derecognised when retired; their quality (removal versus avoidance, see C10) is a sustainability-reporting question, not an accounting one.
Question. What will Meridian disclose at 31 December 2026 under IFRS 7 paragraphs 20B to 20D for the instruments of the case?
Only the instruments with contractual cash flows that can change on an event not directly related to basic lending risks are in scope. The Helios green bond is not (no such clause). The carbon inventory is not a financial instrument. The vPPA and the SLB sit in Castellan’s books, not Meridian’s. The Borealis RCF is in scope, and it is representative of the bank’s whole SLL book.
| Disclosure element | Borealis RCF, Meridian share |
|---|---|
| Nature of contingent event | Borealis meeting annual targets on three KPIs, verified |
| Range of change in cash flows | Margin plus or minus 3 bp; commitment fee plus or minus 1.05 bp |
| Effect, maximum per year | EUR 10,800 (EUR 4,500 on drawn amounts) |
| Gross carrying amount of drawn loans | EUR 15m (average; year-end figure used) |
| Measurement | Amortised cost; undrawn commitment with ECL provision |
In the published note, Meridian aggregates by product (SLL, sustainability-linked bonds held, other ESG-linked instruments) and by range of possible change (for example below 5 bp, 5 to 15 bp, above 15 bp), and explains in a sentence that no instrument failed the SPPI test because of an ESG-linked feature. The same gross carrying amounts must reconcile with the SLL figures in the sustainability statement.
| Figure | Value | Stage |
|---|---|---|
| Borealis ratchet: change in interest / in PV, worst case | 0.94% / 0.137% | 5.1 |
| Maximum annual effect on Meridian’s share | EUR 10,800 (EUR 4,500 drawn) | 5.1 |
| Ratchet reaching the 5% review trigger | 16 bp | 5.1 |
| Upfront fee: deferred in EIR / IFRS 15 per year | EUR 37,500 / EUR 30,000 | 5.2 |
| EIR uplift on drawn amounts | about 5 bp | 5.2 |
| Term-loan variant: catch-up for +1 bp over 4 years | EUR 5,542 | 5.2 |
| Weighted PD / Stage 1 ECL / lifetime ECL | 0.315% / EUR 49,140 / EUR 180,740 | 5.3 |
| Regulatory EL for comparison | EUR 31,200 | 5.3 |
| After the 2028 acquisition: weighted PD / stage / ECL | 0.785% / Stage 2 / EUR 447,330 | 5.3 |
| Helios bond: EIR / interest year 1 / amortised cost | 3.970% / EUR 0.9912m / EUR 24.9697m | 5.4 |
| Helios bond after +40 bp: FV / OCI reserve | EUR 24.3781m / minus EUR 0.5717m | 5.4 |
| Helios bond: P&L at FVOCI / if FVTPL | EUR 0.9712m / EUR 0.3996m | 5.4 |
| Castellan SLB: EIR policy A / B | 3.693% / 3.704% | 5.5 |
| Carrying amount before / after revision | EUR 598.509m / EUR 602.547m | 5.5 |
| Catch-up loss policy A / B | EUR 4.038m / EUR 3.572m | 5.5 |
| vPPA: day-one difference deferred / annual release | EUR 3.05m / EUR 0.305m | 5.6 |
| vPPA after +EUR 10 forwards: FV change / OCI / P&L | EUR 4.277m / 4.063m / 0.214m | 5.6 |
| Physical PPA: sold / bought / ratio | 52.56 / 339.76 GWh / 6.46x | 5.6 |
| Castellan emission liability: net / market value | EUR 149.80m / EUR 152.73m | 5.7 |
| Meridian inventory: FV / gain / futures loss | EUR 35.98m / 1.98m / minus 2.0m | 5.7 |
| Meridian offset provision | EUR 0.36m | 5.7 |
Accounting values are the raw material of capital. Three links matter for the instruments of the case.
ECL against regulatory expected loss. For exposures under the IRB approach, the CRR compares the bank’s credit risk adjustments with its regulatory expected loss. A shortfall (provisions below EL) is deducted from CET1; an excess can be added to Tier 2 up to 0.6% of the IRB credit RWA. On Borealis, the Stage 1 ECL of EUR 49,140 exceeds the EL of EUR 31,200: the EUR 17,940 excess is eligible Tier 2, so the ECL hits CET1 through profit and loss but partly returns as Tier 2. A Stage 2 transfer widens the excess. Under the standardised approach there is no such comparison; general credit risk adjustments count in Tier 2 up to 1.25% of credit RWA.
OCI. Unrealised losses on the Helios bond at FVOCI reduce CET1 immediately, through accumulated OCI. The minus EUR 0.57m of section 5.4 is a CET1 deduction of the same amount, with no RWA relief (the RWA of the bond depend on its exposure value and the issuer’s risk weight, not on the price move in this simple view).
Fair value and prudent valuation. For instruments at fair value, CRR Article 105 and the EBA’s regulatory technical standards require additional valuation adjustments (AVAs) deducted from CET1, for model risk and uncertainty in particular. A bank holding Level 3 PPA derivatives (Meridian chose not to, see C10) or a Level 3 carbon book would deduct AVAs on top of its accounting reserves.
The RAROC convention of the course uses regulatory expected loss and a smoothed revenue line. The accounts show something else in a bad year. The block below reproduces the B03 RAROC, then shows the effect of the ratchet and of an ECL stage transfer on the accounting return on allocated capital.
INTERACTIVE LAB — RAROC of the Borealis hold versus the accounting return in a year of stage transfer
Reproduce the RAROC of B03 line by line, apply the ESG ratchet, and compare with the accounting return when the ECL moves from Stage 1 to Stage 2.
| Parameter you can move | Range | Default |
|---|---|---|
| Revenues before ratchet | 50 to 500 EUR k | 225.9 EUR k |
| Cost-income ratio on revenues before ratchet | 20 to 70 % | 40 % |
| Ratchet applied to the margin | -3 to 3 bp | 0 bp |
| Drawn amount | 0 to 75 EUR m | 15 EUR m |
| Undrawn amount | 0 to 75 EUR m | 60 EUR m |
| Regulatory PD | 0.05 to 5 % | 0.2 % |
| ECL at start of year | 0 to 1000 EUR k | 49.14 EUR k |
| ECL at end of year | 0 to 1000 EUR k | 447.33 EUR k |
| Result at default settings | Value |
|---|---|
| EAD (CCF 40%) | 39.0 EUR m |
| F-IRB risk weight | 39.02 % |
| RWA | 15.22 EUR m |
| Allocated capital | 1,978.2 EUR k |
| Revenues after ratchet | 225.9 EUR k |
| Operating costs | 90.4 EUR k |
| Regulatory expected loss | 31.2 EUR k |
| RAROC | 3.96 % |
| Accounting profit after tax in the year | -197.0 EUR k |
| Accounting return on allocated capital | -9.96 % |
Try it. Set the closing ECL equal to the opening one: the accounting return is 5.14%, above the RAROC, because the accounts charge no expected loss in a year without ECL movement. Then restore EUR 447.33k: one stage transfer wipes out more than two years of profit on the hold.
Line by line. EAD is $15 + 40% = $ EUR 39m; the F-IRB risk weight for a PD of 0.20%, an LGD of 40% and a maturity of 2.5 years is 39.02%; RWA EUR 15.22m; allocated capital at 13%, EUR 1,978.2k (EUR 1.978m). Revenues EUR 225.9k; operating costs at 40%, EUR 90.36k; expected loss $0.20% % = $ EUR 31.2k. Result after 25% tax over capital: RAROC 3.96%, the figure of B03. With the full 3 bp discount Meridian loses EUR 10.8k of revenue and RAROC falls to 3.55%.
The accounting view. The accounts do not charge EL; they charge the change in ECL. In a quiet year (ECL unchanged), the accounting profit after tax is EUR 101.66k, a 5.14% return on the allocated capital, higher than the RAROC. In the year of the 2028 stage transfer the ECL rises from EUR 49.14k to EUR 447.33k and the hold loses EUR 196.99k after tax, minus 9.96% on allocated capital. Over a full cycle the two views should converge; in any single year they can diverge widely. The sustainable feature itself (plus or minus 3 bp) moves RAROC by 0.4 point; the climate scenario moves the ECL, and through it the accounting result, by much more.
Accounting classification and funding interact in Treasury’s liquidity portfolio. The Helios bond held at FVOCI is funded at the FTP rate of an eight-year asset, and its interest margin (the yield over the swap rate) is Treasury’s carry. Its value in the LCR depends on its HQLA eligibility, not on the label. On the commitment side, the Borealis undrawn amount requires an HQLA buffer under the LCR (10% outflow for committed credit facilities to non-financial corporates), carried at 40 bp in Meridian’s convention; the IFRS 9 drawdown assumption used for ECL and the LCR outflow rate are different numbers with different purposes, and the policy team checks that they are not confused in the EAD of the ECL model.
| Instrument | Holder | Measurement | Sustainable feature |
|---|---|---|---|
| Borealis SLL | Meridian | Amortised cost | Passes SPPI; IFRS 7 20B-20D |
| Helios green bond | Meridian Treasury | FVOCI | None; label ignored |
| Castellan SLB | Castellan | Amortised cost | Step-up in host; catch-up |
| Solar vPPA | Castellan | FVTPL or cash-flow hedge | Variable-volume hedged item |
| Allowances, credits | Both | Cost, or FV less costs to sell | Policy choice |
Green Asset Ratio and BTAR. The GAR numerator and denominator are built on gross carrying amounts, the IFRS figures before ECL allowances. The Helios bond counts in the numerator in proportion to the Taxonomy alignment of its allocated proceeds; the Borealis drawn amount counts in the denominator and in the numerator only to the extent of Borealis’s turnover-based alignment (an SLL is general purpose, not use of proceeds). The Castellan SLB, if Meridian held it, would follow the same general-purpose logic. Trading-book allowances are excluded. Since Commission Delegated Regulation (EU) 2026/73, applicable from 1 January 2026 (financial year 2026), the trading-book and fees KPIs are postponed to 2028 (see C01 and E06).
Pillar 3 ESG. The EBA templates on transition risk report gross carrying amounts by sector, with accumulated impairment and accumulated negative changes in fair value, and Stage 2 and Stage 3 amounts. The ECL figures of section 5.3 flow directly into them: a Stage 2 transfer of Borealis would appear in the packaging line of template 1.
Financed emissions. PCAF attributes Borealis’s emissions to Meridian using the outstanding amount (drawn) over EVIC, and the Helios bond using the outstanding amount at book value or market value according to the bank’s chosen PCAF option. Both come from accounting data. The undrawn commitment carries no financed emissions under PCAF, which is one reason a mostly undrawn RCF book looks cleaner than its risk (see E07).
| Risk | Control | Owner |
|---|---|---|
| ESG feature fails SPPI unnoticed | Term-sheet template, NPA memo, review trigger | Accounting policy |
| Margin grid or KPI result not in system | Agent notice reconciled to loan system | Loan operations |
| IFRS 7 and sustainability figures diverge | Single ESG flag, joint reconciliation | Financial reporting |
| Climate overlay double counts or is arbitrary | Overlay memo, back-testing, committee | IFRS 9 impairment |
| Stage transfer missed after a KPI miss | Watch-list trigger on transition-plan failure | Credit risk |
| Level 3 valuation of PPA or carbon book | Independent price verification, AVAs | Product control |
| Policy inconsistency on allowances | Group policy, sub-portfolio mapping | Accounting policy |
| Business model misdescribed for green bonds | Portfolio mandate, sales history review | Treasury and Finance |
Red-flag checklist.
Open debates. Where “significantly different” begins: no number, and auditors’ firm-wide views differ. Whether a KPI margin change is a B5.4.5 reset or a B5.4.6 catch-up. Whether an SLB step-up is ever an embedded derivative. Whether climate belongs in scenarios, in overlays or in PD models (supervisors push towards models). And whether allowances should be standardised at all, after twenty years of policy choice.
INTERACTIVE LAB — Self-assessment: IFRS 9 and sustainable instruments
Check the key concepts of classification, measurement, impairment, PPAs and carbon accounting.
Q1. Since when do the May 2024 amendments on ESG-linked features apply?
Answer: B. Effective for annual periods beginning on or after 1 January 2026; endorsed in the EU by Regulation 2025/1047.
Q2. A green bond held in a hold-to-collect-and-sell portfolio is measured at:
Answer: C. The label does not change cash flows; SPPI is met and the business model gives FVOCI.
Q3. Which condition applies to an ESG feature not directly related to basic lending risks?
Answer: A. Paragraph B4.1.10A requires the comparison in all contractually possible scenarios.
Q4. Borealis’s worst-case ratchet changes annual interest by:
Answer: D. 3 bp out of 320 bp (2.50% plus 70 bp) is 0.94%.
Q5. Under B5.4.6, a revision of estimated cash flows is booked:
Answer: B. The carrying amount is recomputed at the original EIR and the difference goes to profit or loss.
Q6. Why is an SLB step-up usually not separated as an embedded derivative by the issuer?
Answer: C. The derivative definition excludes non-financial variables specific to a party.
Q7. At FVOCI, the Stage 1 ECL is recorded:
Answer: A. The carrying amount stays at fair value; the allowance sits in OCI.
Q8. A virtual PPA is generally:
Answer: D. It is net settled with no delivery, so own-use cannot apply.
Q9. The December 2024 amendments allow a buyer of a pay-as-produced PPA to designate as hedged item:
Answer: B. The hedged item can be a variable nominal amount aligned with the nature-dependent volume.
Q10. The own-use clarification for physical PPAs relies on the entity being:
Answer: C. Sales of surplus do not taint own-use if the entity is and expects to be a net purchaser.
Q11. A commodity broker-trader may measure allowance inventory at:
Answer: A. IAS 2 paragraph 3(b) allows fair value less costs to sell for broker-traders.
Q12. Under the April 2024 IFRIC decision, a net-zero pledge to buy carbon credits leads to a provision:
Answer: D. The past event is the emission, not the announcement.
Q13. On the IRB approach, ECL above regulatory EL is:
Answer: B. The excess counts in Tier 2 within the cap; a shortfall is deducted from CET1.
Q14. Which IASB publication of November 2025 uses climate as the illustrative uncertainty?
Answer: C. Illustrative examples published on 28 November 2025.
Exercise 1: SPPI significance. A EUR 40m four-year SLL pays EURIBOR (2.00%) plus 150 bp, with a two-way ratchet of plus or minus 12.5 bp. Compute the worst-case change in annual interest and in PV of cash flows, and the maximum annual euro effect.
Solution. Base rate . Change in annual interest: . Annuity factor at 3.50% over four years: . Change in PV: of principal. Annual effect: $40 % = $ EUR 50,000. Below Meridian’s 5% review trigger; a standard memo concludes amortised cost, with an IFRS 7 disclosure of the range.
Exercise 2: issuer catch-up. An issuer’s EUR 400m five-year SLB has an EIR of 4.00% (target-met estimate). At the end of year 2 it becomes certain that a 50 bp step-up will apply to the coupons of years 4 and 5. Compute the catch-up.
Solution. Step-up per coupon: $400 % = $ EUR 2.0m. Discounted at 4.00% from the end of year 2: year 4 is two periods away, ; year 5 is three periods away, . Catch-up: EUR 3.627m, booked in finance costs in year 2. The total extra cash paid is EUR 4.0m.
Exercise 3: FVOCI. A bank buys EUR 10m of a five-year 3.00% annual-coupon green bond at par. After one year, yields rise by 50 bp. Stage 1 ECL parameters: PD 0.10%, LGD 45%. Compute profit and loss and the OCI reserve.
Solution. EIR 3.00%; interest income EUR 0.300m; amortised cost stays EUR 10m. Fair value with four years left at 3.50%: price , so EUR 9.8163m. ECL: $0.10% % = $ EUR 4,500. Profit and loss: $300{,}000 - 4{,}500 = $ EUR 295,500. OCI reserve: $9.8163 - 10 + 0.0045 = $ minus EUR 0.1792m.
Interview questions.
Does a green bond’s label affect its accounting? No. The label creates obligations on the use of proceeds and reporting, but the investor’s contractual cash flows are those of a conventional bond. The bond is classified by business model and SPPI like any other. The label matters for the GAR, for the investor’s sustainability reporting and possibly for its portfolio mandate. A bank that held green bonds at a different measurement from conventional ones would have a business model problem, not a green one.
Why did the IASB amend IFRS 9 for ESG-linked loans? Because the SPPI test defined interest by reference to basic lending risks, and a margin moving with emissions did not obviously fit. Practice diverged, and some lenders feared FVTPL for large SLL books. The May 2024 amendments clarified that a contingent feature specific to the debtor can be compatible with SPPI if the cash flows are not significantly different from those without it in all scenarios. They also added disclosures, so investors can see the size and range of these features.
How would you test a 25 bp ratchet on a 60 bp margin? I would first check the nature of the event: specific to the borrower, not a market variable. Then I would compute the worst-case deviation in interest (25 over EURIBOR plus 60) and in PV, compared with the identical loan without the ratchet. With EURIBOR at 2.50% the interest change is about 8%, which is above a typical internal review trigger. I would document why it is or is not significantly different, involve the auditors early, and disclose the range under IFRS 7.
An issuer misses its SLB target. What appears in its accounts? If the issuer had estimated that it would meet the target, it revises its cash flows and books a catch-up at the original EIR: a one-off finance cost equal to the present value of the extra coupons. Interest expense is then slightly higher for the remaining life. If it had separated the step-up as an embedded derivative, the effect would have appeared progressively through fair value changes. Either way the analyst should link the charge to the sustainability statement.
How does climate enter IFRS 9 ECL? Through forward-looking scenarios and their weights, through sector-specific PD and LGD paths, through overlays when models cannot capture the risk, and through the SICR assessment. Supervisors expect documented, scenario-based methods and a check against double counting. For investment-grade names the Stage 1 effect is small; the material effect comes from stage transfers in exposed sectors and from collateral values for physical risk.
Why do the December 2024 PPA amendments matter for a bank? They make own-use available for physical PPAs when the buyer is a net purchaser, and they make cash-flow hedge accounting workable for pay-as-produced virtual PPAs. Corporate clients therefore face less profit and loss volatility from PPAs, which removes an obstacle to signing them. For the bank’s structuring and commodities desks, that widens the client base, while the valuation and credit questions remain.
How should a trading desk account for EU allowances? Allowances are not financial instruments under IFRS, so IFRS 9 does not apply to them. A broker-trader can measure them as inventory at fair value less costs to sell through profit or loss, which matches the futures used to hedge them. If they were held at cost as intangibles, the hedge would create artificial volatility. The bank must apply one policy consistently and disclose it.
Why can the RAROC and the accounting result of a loan diverge? RAROC charges a regulatory expected loss each year and uses smoothed revenues; the accounts charge the change in ECL, which is point-in-time, scenario-weighted and lifetime after a stage transfer. In a quiet year accounting profit exceeds the RAROC numerator; in a stage-transfer year it can be negative. Over a cycle they converge. Business lines should be steered on RAROC but should understand the accounting volatility they create.
Standards and amendments (IFRS Foundation).
EU law.
Supervisors and enforcers.
Professional summaries consulted (public).
Further reading.