Market risk and counterparty risk are the two risks that a bank’s trading activities create. Market risk is the risk of losing money because prices move: rates, credit spreads, equity prices, currencies, commodities, volatilities. Counterparty credit risk is the risk that a client or a bank on the other side of a derivative defaults when the derivative is worth something to us. A third family, the valuation adjustments known collectively as XVA, puts a price on counterparty risk, funding and capital and charges it to the trades that cause them.
At Meridian Bank these activities are split between the first and the second line of defence. The XVA desk sits in the front office of the CIB, next to the rates and commodities traders. It prices the credit valuation adjustment (CVA) and the funding adjustment (FVA) into every uncollateralised client trade, books them as a reserve and hedges the CVA with credit default swaps where it can. It is a trading desk with a P&L, a VaR limit and a budget. Market Risk Management (MRM) and Counterparty Credit Risk (CCR) sit in the Risk function under the Chief Risk Officer. They do not trade. They measure, set limits, challenge and escalate. Risk Methodology builds the models (VaR, expected shortfall, exposure simulation, wrong-way risk) and Model Validation, independent of Methodology, approves them. Product Control, in Finance, produces the daily P&L, performs independent price verification (IPV) and books the valuation reserves. Regulatory Reporting turns all this into own funds requirements, COREP templates and Pillar 3.
The function earns nothing but decides how much capital the trading floor can use. Case assumptions for Meridian: a group one-day 99% VaR of about EUR 25m in normal markets, market-risk RWA of about EUR 15bn, CVA reserves of about EUR 350m on the uncollateralised client book, and a Global Markets division that consumes about a fifth of the group’s capital. Each euro of VaR limit that MRM grants to a desk is, through the capital model, a slice of the 13% of RWA that Meridian allocates to its businesses (see E02 and E03). A 10% error in a VaR model therefore shifts hundreds of millions of capital, and a missed scenario can wipe out a year of a desk’s revenue in a week.
MRM covering the CIB has about 40 people (case assumption): desk-aligned risk managers, a stress-testing team, VaR production and a small climate-risk cell being folded into the desk teams. CCR has about 25: credit officers approving trading lines, an exposure analytics team running the potential future exposure (PFE) engine, and collateral risk. Adjacent functions: Treasury (C15), banking-book Credit Risk (E01), Compliance (E10) and the Climate and Environmental Risk unit, which owns the bank’s climate scenarios.
A realistic week for the commodities and credit market-risk manager in November 2026:
WHY IT MATTERS ON THE DESK — On a trading floor, every desk has a limit, and every limit is a number that MRM computes and defends. When a climate shock arrives, the traders will ask one question: “how much room do I have?” The answer comes from the models taught in this module.
IN ONE SENTENCE — Market risk is measured in three layers: sensitivities say how much a position loses per unit of move, VaR and expected shortfall say how large a bad day is likely to be, and stress tests say what happens on a day nobody has seen.
Sensitivities are the trader’s language. The DV01 is the change in value for a one basis point rise in rates; the CS01 for a one basis point rise in the credit spread; the delta of a commodity position is its change in value for a 1% (or one unit) rise in the commodity price; the vega for a one-point rise in implied volatility. They are additive within a risk factor and are the basis of limits (a desk may hold at most EUR 25,000 of CS01 on a single issuer, for instance) and of the standardised capital rules.
Value-at-risk (VaR) answers: what loss will be exceeded only with a small probability over a given horizon? A one-day 99% VaR of EUR 1m means that on one day in a hundred the bank expects to lose more than EUR 1m. Expected shortfall (ES) answers the next question: when that bad day comes, how much do we lose on average? ES looks inside the tail, where VaR only looks at its edge.
INTUITION — VaR is the height of the dyke: the sea will overtop it one day in a hundred. Expected shortfall is the average depth of water in the town on the days it does. Two dykes of the same height can protect very differently if one sea brings ripples and the other tsunamis.
Historical simulation revalues today’s positions under each of the last 250 or 500 days’ market moves; it assumes no distribution but cannot imagine what did not happen. Parametric methods assume normal returns and combine sensitivities, volatilities and correlations. Monte Carlo simulates paths from a model and handles options. Most European banks use historical simulation for regulatory VaR; we use the parametric version because it shows every moving part.
FORMULA
is the vector of one-day P&L sensitivities to one-standard-deviation moves of each risk factor (in euros); the correlation matrix of the risk factors; the one-day standard deviation of the book’s P&L; the standard normal quantile at confidence (2.326 at 99%, 1.960 at 97.5%); the horizon in days, scaled by the square root of time; the standard normal density. With a normal distribution the 97.5% ES equals , almost exactly the 99% VaR (), which is why the Basel Committee chose 97.5% for ES when it replaced the 99% VaR.
The square root of time assumes independent daily returns and an unchanged position. Neither holds in a crisis. The FRTB’s liquidity horizon assigns each risk factor the number of days needed to exit or hedge it in stress.
INTERACTIVE LAB — From daily volatility to VaR and ES over a liquidity horizon
See how VaR and ES scale with confidence and horizon for a single linear position, and why the choice of horizon matters more than the choice of measure.
| Parameter you can move | Range | Default |
|---|---|---|
| Position value | 1 to 100 EUR m | 10 EUR m |
| Daily volatility | 0.2 to 6 % | 2.4 % |
| Horizon | 1 to 120 days | 10 days |
| Result at default settings | Value |
|---|---|
| VaR 99% over the horizon | 1.766 EUR m |
| ES 97.5% over the horizon | 1.774 EUR m |
| ES 97.5% / VaR 99% | 1.0049 |
| Annualised volatility (250 days) | 37.9 % |
Try it. Move the horizon from 10 to 40 days: the risk doubles. Moving from VaR 99% to ES 97.5% changes it by half a percent. Arguments about the measure matter less than arguments about the horizon.
A EUR 10m position with a daily volatility of 2.4% (38% a year, the order of magnitude of EU allowances in calm years) has a ten-day 99% VaR and ES of EUR 1.77m; at 40 days the ES doubles to EUR 3.55m. Stressed VaR (Basel 2.5, in force in the EU until end-2026) recomputes VaR on a stressed year; backtesting compares each day’s VaR with realised P&L, and more than four exceptions in 250 days raises the capital multiplier.
Until 31 December 2026, EU banks use the original CRR rules: standardised charges by risk type (for commodities, a simplified approach of 15% of the net plus 3% of the gross position), or internal models at three times VaR plus stressed VaR. From 1 January 2027 the Fundamental Review of the Trading Book (FRTB), transposed by CRR2 and CRR3, replaces them, with temporary adjustments for three years (section 3.4).
The FRTB has two approaches. The standardised approach applies to every desk as fallback. It adds the sensitivities-based method (SBM) (delta, vega and curvature sensitivities times regulatory risk weights, aggregated with regulatory correlations under three scenarios), the default risk charge (DRC) and the residual risk add-on (RRAO). C08 computes it on a bond book; this module adds a commodity position.
The internal model approach (IMA) is granted desk by desk. It replaces VaR by a 97.5% ES computed over a stressed period, with liquidity horizons of 10 to 120 days by risk-factor category. Under the Basel standard (MAR33), the horizon is 10 days for interest rates in major currencies and large-cap equities, 20 days for energy and carbon emissions trading prices and for investment-grade sovereign spreads, 40 days for high-yield sovereign and investment-grade corporate spreads, 60 days for high-yield corporate spreads, and 120 days for credit spread volatilities. Risk factors without enough real price observations are non-modellable and capitalised by a separate stress-scenario charge. Each desk must pass backtesting and the profit-and-loss attribution (PLA) test, which checks that the risk model’s P&L matches the front office’s. The capital is a multiplier of at least 1.5 times the ES plus the non-modellable charge plus a default risk charge computed with a one-year 99.9% model.
FORMULA
is the ten-day ES of the portfolio shocked on all risk factors; the ten-day ES when only the risk factors with a liquidity horizon of at least are shocked; the ordered horizons 10, 20, 40, 60 and 120 days. A factor with a 40-day horizon is counted in every term up to 40 days, so its contribution grows as the square root of 40 over 10, that is twice the ten-day figure.
IN ONE SENTENCE — Counterparty exposure is the replacement cost of a derivative in the future, which we model as a distribution because we do not know which way the market will move.
The expected exposure is the average positive value at date across simulated scenarios; the potential future exposure a high quantile of the same distribution, used for credit lines; the expected positive exposure (EPE) the time-average of EE. For a swap the profile is a hump: uncertainty grows, then remaining cash flows shrink (section 5.6).
Netting under an ISDA Master Agreement limits the claim to the net value of all trades (C07 shows a 29% saving). Collateral under a Credit Support Annex (CSA) does the rest: variation margin (VM) settles value changes daily, leaving what can happen during the margin period of risk (MPOR), ten business days for most bilateral sets; initial margin (IM) covers that residual. Financial counterparties above EMIR thresholds must exchange both; most corporates, below the clearing threshold, sign no CSA, and the bank carries the exposure.
Regulatory exposure uses SA-CCR (C07, C10): 1.4 times replacement cost plus an add-on from supervisory factors (0.5% rates, 4% FX, 18% commodities other than electricity, 40% electricity), then risk-weighted like a loan.
FORMULA
is the market loss given default (60% by convention for senior unsecured, from the 40% recovery used in CDS); the expected exposure at date ; the survival probability of the counterparty, so that is the probability of default between two dates; the discount factor; the default intensity implied by the credit spread of the counterparty (the “credit triangle”).
The CVA is charged to the client in the price and booked by the XVA desk as a reserve against the derivative’s fair value. The DVA, its mirror for the bank’s own default, is deducted from CET1. The FVA pays for funding the uncollateralised exposure: the client pays nothing until settlement while the hedge is margined daily. The MVA funds initial margin posted to a CCP; the KVA, the cost of capital, Meridian handles through RAROC.
Since 1 January 2025, CRR3 also requires own funds for CVA risk, the risk that the CVA itself moves with the counterparties’ spreads: the standardised SA-CVA, which needs supervisory approval, or the basic BA-CVA. The EU kept an exemption, in Article 382(4) CRR, for trades with non-financial counterparties established in the Union; Vantoria and Castellan therefore attract no CVA-risk capital, although their CVA is still priced and reserved.
Wrong-way risk (WWR) is the case in which exposure tends to rise when the counterparty’s credit quality falls. General WWR comes from common macro factors; specific WWR from a legal or economic link (a client selling the bank protection on itself, or posting its own shares as collateral). Independence between exposure and default, assumed by the formula above, then understates CVA. The opposite case, right-way risk, is common in corporate hedging: a company that hedges a cost it bears pays the bank when the cost falls, that is, when it is doing well.
Meridian’s hierarchy: a board-approved VaR and stress-loss appetite; desk VaR limits; sensitivity limits (CS01 per issuer and sector, commodity delta by tenor, vega); stop-loss triggers; and PFE lines per counterparty approved by a credit officer. A hard limit needs prior approval to exceed; a soft limit forces a discussion. Every breach is logged, explained and remedied or approved one level up, through the daily risk report, the weekly stress report and the monthly Market Risk Committee pack.
PITFALL — A limit that is never close to being used is not a control, it is decoration. Limits should be calibrated so that a desk uses 50% to 80% of them in normal conditions; otherwise a new risk, such as a climate scenario, can grow inside the slack without anyone noticing.
Climate change does not create a new risk type in a trading book. It creates new drivers of the existing risk factors. The Basel Committee’s 2021 report on transmission channels and its 2022 principles distinguish physical and transition drivers; for a trading book the transition drivers dominate because the horizon is short and policy news arrives in a day.
| Driver | Risk factor hit | Example in the case |
|---|---|---|
| Carbon policy news | EUA price, power prices | ETS reform, Market Stability Reserve, price-control debate |
| Sector repricing | Credit spreads by sector | Chemicals and cement widen, renewables tighten |
| Commodity demand shifts | Gas, oil, power curves | Industrial gas demand falls with plant closures |
| Volatility regime | Implied and realised vols | Carbon vol doubles around policy dates |
| Correlation regime | Cross-asset correlations | Carbon and emitters’ spreads move together |
| Physical events | Power, gas, agricultural prices | Heatwave and low rivers raise power prices |
| Liquidity | Bid-offer, exit time | Brown bonds become hard to sell |
Three features defeat standard models: jumps (a policy decision can move carbon 10% or more in a day), regime change (correlations flip sign when carbon becomes the driver) and horizon mismatch (scenarios span years, risk is measured over ten days). As Roncalli’s handbook argues, a historical VaR cannot contain a shock that has not happened; a climate scenario must be translated into an instantaneous shock plus stressed risk parameters, the method of section 5.4.
There is no Pillar 1 rule that makes market-risk capital depend on climate. The standards are supervisory expectations, enforced through the SREP and Pillar 2.
Climate is absent from Pillar 1, but carbon and energy are not. Four rules matter on the desk.
FRTB standardised approach. Commodities are split into eleven buckets. Bucket 3, “energy: electricity and carbon trading”, carries a 60% delta risk weight, the highest of the energy buckets and well above the 18% to 45% of most others. ISDA argued in a July 2021 paper with Copenhagen Economics that the stressed volatility of EU allowances (about 56% a year) justified roughly 37%, and that the 99% correlation between spot and forward positions was too low for EUAs. The Basel standard has not changed.
FRTB internal model. Energy and carbon emissions trading prices have a 20-day liquidity horizon, investment-grade corporate credit spreads 40 days, high-yield corporate spreads 60 days.
SA-CCR. EU allowances use the 18% supervisory factor of “other commodities”; electricity has 40%. A two-year uncollateralised EUA forward therefore carries an add-on of 18% of its notional before the 1.4 alpha.
CVA risk. Article 382(4) CRR exempts transactions with EU non-financial counterparties. Most carbon and energy hedges for industrial clients fall in the exemption.
FRTB go-live and the Delegated Act. The FRTB applies in the EU from 1 January 2027, the last postponement the CRR allows. On 4 June 2026 the Commission adopted a delegated act of temporary adjustments (Delegated Regulation (EU) 2026/1221), applying for three years from 1 January 2027 after scrutiny by Parliament and Council. According to public summaries of the text, it contains an optional institution-specific multiplier that offsets the capital increase relative to the previous rules for banks with internal models; a general 0.9 multiplier on the standardised own funds; relief on the PLA test, which becomes a monitoring tool during the transition; and, for EU ETS instruments, an adjustment that raises the tenor correlation used in the commodity delta charge. On 3 August 2026 the EBA issued a no-action letter on the boundary and internal risk transfer provisions so that banks using the multiplier do not have to run two boundary frameworks.
ECB climate factor in the collateral framework. Since 15 June 2026 the Eurosystem applies a climate factor to marketable assets issued by non-financial corporations used as collateral in its credit operations: an additional reduction in collateral value, larger for issuers more exposed to transition risk, of at most 5%. On 24 July 2026 the ECB announced its extension to non-financial corporate credit claims, at the earliest by the end of 2027. This is the first time a central bank’s haircut schedule depends on climate, and private CSAs and repo schedules are being asked whether they should follow (see C11 and D08).
Carbon market policy risk. The carbon market itself became a source of policy-event risk in 2025 and 2026, with debates on price-control mechanisms, the use of the Market Stability Reserve and the free allocation path under CBAM (C10 covers the Commission’s July 2026 proposal). For a market-risk manager this means a calendar of event dates on which the carbon price can jump, in either direction.
Three places in the pricing of a derivative can carry climate risk.
PITFALL — Calling a CVA “climate-adjusted” because the counterparty spread already reflects climate risk is circular. The spread prices the client’s default risk on average; it says nothing about whether default is more likely precisely when the trade is worth most to the bank. That dependence is what WWR modelling must add.
The “deal” of a risk function is a risk decision. This section follows two linked processes in November 2026: the onboarding of a client commodity hedge with wrong-way risk, and the weekly climate stress of the trading book that the hedge joins.
INTERACTIVE LAB — From a client hedge request to the climate stress report
Follow a carbon-intensive client’s commodity hedge through credit approval, XVA pricing, booking, daily risk, the climate stress run and escalation.
| # | Who | Does what | Deliverable |
|---|---|---|---|
| 1 | Climate and Environmental Risk unit with Risk Methodology | Translates the NGFS short-term scenarios and the ETS policy calendar into instantaneous shocks to carbon, gas, power and sector spreads, and stressed volatilities and correlations | Climate scenario specification, approved annually |
| 2 | Market Risk Committee | Approves the scenarios and sets a climate stress-loss limit for the commodities and credit desks | Committee minutes and limit letter |
| 3 | Commodities sales | Receives Vantoria Chemicals’ request to fix the price of its gas purchases for 2027 and 2028 | Trade request with volumes, tenor and indicative price |
| 4 | Counterparty Credit Risk officer | Checks the PFE against Vantoria’s trading line, flags wrong-way risk under the WWR policy and asks for a scenario analysis | Credit approval with WWR flag and conditions |
| 5 | XVA desk | Prices CVA with and without wrong-way risk, FVA and MVA, and decides how much of the WWR charge to pass to the client | XVA quote attached to the trade |
| 6 | Commodities sales and trader | Execute the swap with Vantoria and the offsetting TTF futures on the exchange | Trade confirmation under the ISDA Master Agreement; futures fills |
| 7 | Middle office and Product Control | Book the trade, the day-one sales credit and the XVA reserve; run IPV on the forward curve | Day-one P&L and reserve booking |
| 8 | Market Risk Management | Computes overnight VaR, ES, sensitivities and limit usage of the desk | Daily risk report |
| 9 | Market Risk Management, stress-testing team | Runs the weekly historical and climate scenarios, including the stressed-covariance add-on | Weekly stress report with limit utilisation |
| 10 | Head of commodities desk and MRM | Handle a breach of the climate stress limit: reduce the carbon inventory or obtain a temporary increase | Breach note and remediation plan signed by the CRO’s delegate |
| 11 | Model Validation | Reviews the WWR model and the climate overlay of the PFE engine | Validation report with findings and deadlines |
| 12 | Regulatory Reporting and Finance | Compute market-risk own funds, SA-CCR exposure and the CVA reserve movement; prepare COREP and Pillar 3 | Quarterly own funds report and disclosures |
Step 1, scenario design. The Climate and Environmental Risk unit owns the bank’s climate narratives; Risk Methodology converts them into numbers the risk engine can use. The specification lists, for each scenario, the shocks to each risk factor (prices, spreads, curves), the stressed volatilities and correlations, the liquidity horizon assumed and the source of each calibration. Sources are named: NGFS short-term scenarios for the macro narrative, the 2022 ECB stress test for sector spread multipliers, historical policy events for carbon jumps. The document is approved by the Market Risk Committee and reviewed by Model Validation.
Step 2, limits. The committee sets a stress-loss limit per scenario family, not only a VaR limit, because VaR cannot see the scenario. Meridian’s climate stress limit for the combined commodities and credit flow books is EUR 3m (case assumption), against a ten-day VaR limit of EUR 2.5m.
Steps 3 to 6, client trade. Vantoria Chemicals, the BBB-rated chemicals producer of C08, buys gas as fuel and feedstock. It asks to fix the price of 3 TWh over 2027 and 2028. The CCR officer applies Meridian’s WWR policy, which requires a scenario analysis for any client whose credit and whose trade both depend on the same transition driver. The XVA desk prices the CVA with and without WWR. The trade is documented under the ISDA Master Agreement already in place with Vantoria; there is no CSA. The trader hedges on the exchange with TTF futures, which are cleared and margined daily.
Steps 7 to 9, booking and daily risk. Product Control books the day-one sales credit and the XVA reserve, and the IPV team checks the gas forward curve against broker quotes. The trade has no market risk for the desk once hedged, but it adds counterparty exposure and a funding need: if gas prices fall, Meridian pays variation margin on its futures while Vantoria pays nothing until each monthly settlement.
Steps 10 to 12, escalation, validation and reporting. A breach note states the size, the cause, the action and the deadline. The validation report on WWR models typically challenges the calibration of the dependence parameter, which rests on few data points. Regulatory reporting brings everything into the quarterly COREP market risk and counterparty templates and the Pillar 3 disclosures.
The case follows Meridian’s market-risk manager for commodities and credit, and her counterparty-risk colleague, through the week of 9 to 13 November 2026. The book combines positions that earlier modules built: the credit flow book of C08, the carbon hedges for Castellan Cement of C10 and the swaps of Borealis Packaging of C07. One new trade is added: a gas price hedge for Vantoria Chemicals. All positions, volatilities, correlations and scenario shocks are case assumptions chosen to be realistic for November 2026; they are not market data.
Question. Which positions carry market risk after hedges, which carry only counterparty risk, and how large are their sensitivities?
Data. The credit flow desk holds the positions of C08’s Monday book, rebuilt after the episode studied there (case assumption): long Helios Renewables bonds with a net CS01 of EUR 29,594 per bp, long Vantoria Chemicals bonds with a net CS01 of EUR 20,071 per bp, and EUR 65m of iTraxx Main five-year protection with a CS01 of EUR 466.2 per million (C08, section 5.1). Rate risk is hedged with futures. The carbon desk has sold Castellan Cement the three forward tranches of C10 (3.409 Mt for December 2027, 2028 and 2029) and bought the same quantities in ICE EUA futures of the same December expiries, so its delta on these trades is zero; it also carries an inventory of 100,000 allowances bought from smaller clients and auctions to serve flow, at EUR 85 (case assumption, consistent with C10). The rates desk has hedged Borealis’s swaps (C07) with other banks at mid. The new trade: on 12 November Vantoria fixed the price of 3 TWh of gas (1.5 TWh a year in 2027 and 2028) at EUR 32.00 per MWh against a TTF forward of EUR 31.60; the commodities desk bought the same volume in TTF futures on the exchange.
| Position | Desk | Size | Market risk after hedge | Source |
|---|---|---|---|---|
| Helios bonds, net | Credit flow | EUR 50m long | CS01 EUR 29,594 | C08 |
| Vantoria bonds | Credit flow | EUR 40m long | CS01 EUR 20,071 | C08 |
| iTraxx Main protection | Credit flow | EUR 65m bought | Short credit | C08 |
| EUA inventory | Carbon | 100,000 t | Delta EUR 85,000 per 1% | New |
| Castellan EUA forwards | Carbon | 3.409 Mt | Hedged by futures | C10 |
| Borealis IRS and CCS | Rates and FX | EUR 600m | Hedged at mid | C07 |
| Vantoria gas swap | Commodities | 3 TWh | Hedged by futures | New |
Daily risk-factor volatilities and correlations estimated by the VaR team on the last twelve months (case assumptions): Helios spread 2.5 bp, Vantoria spread 4.0 bp, iTraxx Main 2.0 bp, EUA price 2.4% (38% a year). Correlation 0.5 between the two issuers’ spreads, 0.6 between each issuer and the index, and minus 0.2 between the EUA price and each spread: in the last year carbon rallied on risk-on days, when spreads tightened.
Reading. The table separates two worlds. Four lines carry market risk: the bonds, the index hedge and the carbon inventory. The client trades with Castellan, Borealis and Vantoria have almost no market risk once hedged, but they carry counterparty risk, because the clients pay no margin while Meridian’s hedges are margined or are with banks under CSAs. Climate enters both worlds: carbon and sector spreads in the first, the credit of carbon-intensive clients in the second.
WHO DOES WHAT — The market-risk manager covers the credit flow and carbon desks; the gas trade sits with the commodities desk, which reports to the same head of trading. Her daily report shows desk by desk. The question she asks this week, because the Market Risk Committee asked it in October, is whether the desks’ risks add up or offset when carbon is the driver. The answer cannot come from the desk reports, only from the combined book.
Question. What are the ten-day 99% VaR and 97.5% ES of the combined book in normal markets, which position dominates, and what does the FRTB liquidity horizon add?
Calculation. Each position’s one-day P&L for a one-standard-deviation move: minus CS01 times spread volatility for the bonds, plus the index CS01 times the index volatility for the protection, the inventory value times the EUA volatility for the allowances. The book’s standard deviation combines them with the correlation matrix (section 2.1). The liquidity-adjusted ES applies the cascade formula of section 2.2 with a 20-day horizon for the EUA price and a 40-day horizon for the three credit spreads (investment-grade corporate under MAR33).
INTERACTIVE LAB — Baseline VaR and ES of the combined book, and the FRTB liquidity-horizon adjustment
Combine the credit, index and carbon positions into one parametric VaR and ES, see each position’s stand-alone VaR and the diversification benefit, and apply the liquidity horizons.
| Parameter you can move | Range | Default |
|---|---|---|
| CS01 Helios bonds | 0 to 60000 EUR per bp | 29594 EUR per bp |
| CS01 Vantoria bonds | 0 to 60000 EUR per bp | 20071 EUR per bp |
| iTraxx Main protection | 0 to 150 EUR m | 65 EUR m |
| Index CS01 per EUR 1m | 300 to 600 EUR per bp | 466.2 EUR per bp |
| EUA inventory value | 0 to 50 EUR m | 8.5 EUR m |
| Daily vol Helios spread | 0.5 to 10 bp | 2.5 bp |
| Daily vol Vantoria spread | 0.5 to 15 bp | 4 bp |
| Daily vol iTraxx Main | 0.5 to 10 bp | 2 bp |
| Daily vol EUA price | 0.5 to 8 % | 2.4 % |
| Correlation EUA price / spreads | -0.9 to 0.9 | -0.2 |
| Ten-day VaR limit | 0.5 to 10 EUR m | 2.5 EUR m |
| Result at default settings | Value |
|---|---|
| CS01 of the index protection | 30,303 EUR per bp |
| One-day P&L standard deviation | 244.1 EUR k |
| Ten-day VaR 99% | 1.796 EUR m |
| Ten-day ES 97.5% | 1.804 EUR m |
| Ten-day ES 97.5%, credit factors only | 0.751 EUR m |
| ES with liquidity horizons (20 days EUA, 40 days spreads) | 2.764 EUR m |
| Sum of stand-alone ten-day VaRs | 3.081 EUR m |
| Diversification benefit | 1.286 EUR m |
| EUA share of the sum of stand-alone VaRs | 48.7 % |
| Stand-alone ten-day VaR of the EUA inventory | 1.501 EUR m |
| Ten-day VaR 99% of the credit positions alone | 0.747 EUR m |
| Utilisation of the VaR limit | 71.8 % |
Try it. Set the EUA inventory to zero: the VaR of EUR 110m of bonds and their hedge is smaller than the VaR of EUR 8.5m of allowances. Then set the correlation to +0.4 and watch the carbon long start to hedge the credit long.
Result. The index protection has a CS01 of EUR 30,303 per bp. The book’s one-day P&L standard deviation is EUR 244,100; the ten-day 99% VaR is EUR 1.796m and the ten-day 97.5% ES EUR 1.804m. The stand-alone VaRs add up to EUR 3.081m, so diversification saves EUR 1.286m. The EUA inventory alone has a stand-alone VaR of EUR 1.501m, 48.7% of the stand-alone total, while the credit positions together, EUR 90m of bonds and their hedge, have a VaR of EUR 0.747m. With liquidity horizons the ES rises to EUR 2.764m: the ten-day ES of the credit factors alone (EUR 0.751m) counts again with a weight of the square root of two, because spreads are deemed to need 40 days to exit.
Reading. Two facts frame the rest of the case. First, a EUR 8.5m carbon inventory carries twice the risk of EUR 90m of investment-grade bonds hedged with an index. Carbon is a volatile commodity, not a quiet regulatory instrument. Second, the measured correlation of minus 0.2 between carbon and spreads means that, in the last twelve months, carbon and credit losses tended to arrive together: both are long “risk-on”. Whether that remains true when carbon policy, rather than the business cycle, drives the market is the question of the next two stages.
PITFALL — The correlation of minus 0.2 is a twelve-month average of a relationship that changes sign. In 2022, when gas prices drove both carbon and industrial credit, carbon and emitters’ spreads moved together; in calm years carbon trades with equities. A VaR model calibrated on one regime gives the wrong diversification in the other, and climate shocks are precisely regime changes.
WHO DOES WHAT — In the Friday morning call the market-risk manager does not read the VaR aloud first; she reads the composition. “Book VaR 1.8 million against 2.5, but half of the stand-alone risk is a hundred thousand allowances. The credit book is quiet.” The carbon trader answers that the inventory is there to serve industrial clients before the December expiry. Both statements are true, and the question for the stress run is whether the inventory still hedges anything when policy, not the business cycle, moves the market.
Question. What does the book lose under an instantaneous disorderly-transition shock, and under the opposite policy shock? Is the climate stress limit of EUR 3m respected?
Data. Risk Methodology has translated two narratives into ten-day shocks (case assumptions). Sudden wake-up, inspired by the NGFS scenario of the same name: governments signal a faster path, the carbon price jumps, carbon-intensive sectors reprice, renewables are barely affected, and industrial gas demand falls with announced plant closures. Policy rollback: political pressure leads to a price-control mechanism and an extended free allocation, the carbon price falls, emitters’ spreads tighten, and renewable issuers widen on lower expected power prices and subsidy doubts.
| Risk factor | Sudden wake-up | Policy rollback |
|---|---|---|
| EUA price | +35% | −40% |
| Vantoria spread | +90 bp | −15 bp |
| Helios spread | +5 bp | +15 bp |
| iTraxx Main | +20 bp | +5 bp |
| TTF gas forward | −30% | +10% |
The magnitudes are calibrated on observed episodes: EUAs fell by more than a third between August 2022 and March 2023 and rose by similar amounts within months in 2021, and chemicals spreads widened by several tens of basis points more than the index in the 2022 energy crisis. The gas shock matters for the Vantoria swap in section 5.6, not for the market risk of the hedged book.
The scenario specification records more than the numbers. For each shock it names the narrative variable that drives it (the carbon price path, sector output, gas demand), the historical analogue used for the size, the horizon over which positions are assumed frozen (ten days, the time to reduce the inventory in a stressed market), and the people who signed it: the Climate and Environmental Risk unit for the narrative, Risk Methodology for the translation, the Market Risk Committee for approval. Model Validation’s main challenge in 2026 was the Helios shock: renewable issuers are often assumed to gain in a transition scenario, but in the rollback narrative they lose, and in the wake-up narrative their spreads still widen a little with the index. The sign of a “green” asset’s move is a scenario choice, not a law.
Calculation. P&L of each line equals minus CS01 times the spread move for bonds, plus the index CS01 times the index move for the protection, and inventory value times the price move for allowances. Convexity is ignored; for a 90 bp move on a five-year bond it would reduce the loss by less than 3%.
INTERACTIVE LAB — Repricing the book under two climate scenarios, and the stress-limit utilisation
Apply the two climate scenarios, scaled up or down, to the positions and compare the result with the EUR 3m climate stress-loss limit.
| Parameter you can move | Range | Default |
|---|---|---|
| Scenario | rollback / wakeup | rollback |
| Severity (share of the calibrated shocks) | 0 to 200 % | 100 % |
| CS01 Helios bonds | 0 to 60000 EUR per bp | 29594 EUR per bp |
| CS01 Vantoria bonds | 0 to 60000 EUR per bp | 20071 EUR per bp |
| iTraxx Main protection | 0 to 150 EUR m | 65 EUR m |
| Index CS01 per EUR 1m | 300 to 600 EUR per bp | 466.2 EUR per bp |
| EUA inventory | 0 to 500 thousand t | 100 thousand t |
| EUA price | 30 to 150 EUR/t | 85 EUR/t |
| Climate stress-loss limit | 1 to 10 EUR m | 3 EUR m |
| Result at default settings | Value |
|---|---|
| P&L Helios bonds | -443.9 EUR k |
| P&L Vantoria bonds | 301.1 EUR k |
| P&L iTraxx protection | 151.5 EUR k |
| P&L EUA inventory | -3,400.0 EUR k |
| P&L of the book | -3,391.3 EUR k |
| Utilisation of the stress-loss limit | 113.0 % |
Try it. Switch to the wake-up scenario and set the inventory to zero: the book loses EUR 1.35m, because the long Vantoria bonds are no longer offset by carbon. The inventory that breaches one limit protects against the other scenario.
Result. Under policy rollback the book loses EUR 3.391m: EUR 3.400m on the allowances, EUR 443,900 on Helios, partly offset by EUR 301,100 of gain on Vantoria and EUR 151,500 on the index protection. The limit is used at 113.0%: a breach. Under sudden wake-up the book gains EUR 1.627m: the allowances earn EUR 2.975m, which more than offsets the EUR 1.806m lost on Vantoria. Halving the inventory to 50,000 tonnes cuts the rollback loss to EUR 1.691m, a utilisation of 56.4%.
Reading. The carbon inventory is a hedge in one climate scenario and the main source of loss in the other. A climate stress programme that only ran the “transition” direction, the natural reflex of a sustainability team, would have reported a gain and missed the breach. For a trading book, a climate scenario must be run in both directions, because the desk is long or short policy and not “green” or “brown”. The head of the carbon desk reduces the inventory to 60,000 tonnes over the following week, selling into client demand; until then, the CRO’s delegate grants a temporary limit of EUR 3.5m, recorded in the breach note.
Question. Beyond the instantaneous repricing, how does a carbon-driven regime change the VaR of the book, and how large is a climate add-on to the risk measure?
Data. In the transition regime (case assumptions, consistent with the scenario narratives): the daily volatility of Vantoria’s spread and of the EUA price rises by 50%, to 6.0 bp and 3.6%; the correlation between the EUA price and Vantoria’s spread turns positive at +0.4 (carbon up, Vantoria wider, and vice versa); other parameters are unchanged.
FORMULA
is the instantaneous repricing of stage 5.3; and are the baseline one-day sensitivities to one-standard-deviation moves and the correlation matrix; and their stressed versions, with higher volatilities and the new correlation. The decomposition follows the logic of Roncalli’s handbook (section 13.2), which separates repricing, risk-premium and covariance effects; over ten days the risk-premium term is negligible and we drop it.
INTERACTIVE LAB — Climate add-on: repricing plus the effect of stressed volatilities and correlations
Compute the stressed VaR of the book in a carbon-driven regime and decompose the climate add-on into repricing and covariance components.
| Parameter you can move | Range | Default |
|---|---|---|
| Repricing scenario | rollback / wakeup | rollback |
| Volatility multiplier, Vantoria spread | 1 to 3 | 1.5 |
| Volatility multiplier, EUA price | 1 to 3 | 1.5 |
| Correlation EUA price / Vantoria spread in the regime | -0.9 to 0.9 | 0.4 |
| EUA inventory value | 0 to 50 EUR m | 8.5 EUR m |
| Result at default settings | Value |
|---|---|
| Baseline ten-day VaR 99% | 1.796 EUR m |
| VaR with the new correlation only | 1.470 EUR m |
| Stressed ten-day VaR 99% | 2.135 EUR m |
| Covariance add-on | 0.339 EUR m |
| Repricing add-on | 3.391 EUR m |
| Total climate add-on | 3.731 EUR m |
| Baseline VaR plus climate add-on | 5.526 EUR m |
Try it. Set the regime correlation back to -0.2: the covariance add-on rises from EUR 0.34m to EUR 0.84m. The new correlation is good news for this book, because carbon then hedges the Vantoria bonds.
Result. The new correlation alone lowers the VaR from EUR 1.796m to EUR 1.470m: when carbon and Vantoria’s spread move together, the long allowances offset the long Vantoria bonds. Higher volatilities push it back up to EUR 2.135m, a covariance add-on of EUR 0.339m. Adding the rollback repricing loss of EUR 3.391m, the total climate add-on is EUR 3.731m and the stressed figure EUR 5.526m, three times the baseline VaR. In the wake-up direction the repricing term is zero (the book gains) and the add-on is the covariance term only.
Reading. The decomposition shows where the danger lies. The covariance effect is modest and even partly favourable, because the regime change turns carbon into a hedge of carbon-intensive credit. The danger is the jump, which no VaR model calibrated on the past year can produce. This is why supervisors ask for stress tests rather than “climate VaR”: the add-on is dominated by a scenario judgment, not by statistics. The pitfall is to present the EUR 5.5m as a probability-weighted measure; it is a conditional loss with no probability attached.
Question. From 1 January 2027, what own funds does the EUA inventory require under the FRTB, and how does that compare with the rules in force until 31 December 2026?
Data. Inventory EUR 8.5m (100,000 tonnes at EUR 85), linear, no options, so curvature and the residual risk add-on are zero. FRTB standardised: bucket 3 risk weight 60%; the Delegated Act’s 0.9 multiplier on standardised requirements during 2027 to 2029. FRTB internal model, approximated: ES 97.5% on a stressed daily volatility of 3.5% (about 56% a year, the stressed figure of the ISDA study), liquidity horizon 20 days, multiplier 1.5. Current rules: the simplified approach for commodities (15% of the net position plus 3% of the gross position) for a standardised bank; for an internal-model bank, three times the ten-day 99% VaR plus three times the stressed VaR, with the 2.4% and 3.5% daily volatilities. Meridian converts own funds into RWA at 12.5 and allocates capital at 13% of RWA.
INTERACTIVE LAB — Own funds for EUR 8.5m of allowances: current rules against the FRTB
Compare the market-risk own funds and allocated capital of a carbon inventory under the simplified approach, the current internal model, the FRTB standardised approach (with and without the 0.9 multiplier) and an approximate FRTB internal model.
| Parameter you can move | Range | Default |
|---|---|---|
| Inventory value | 1 to 50 EUR m | 8.5 EUR m |
| FRTB risk weight, bucket 3 | 20 to 80 % | 60 % |
| Delegated Act multiplier on standardised | 0.5 to 1 | 0.9 |
| Daily volatility, current | 0.5 to 6 % | 2.4 % |
| Daily volatility, stressed | 0.5 to 8 % | 3.5 % |
| Liquidity horizon | 10 / 20 / 40 / 60 | 20 days |
| Result at default settings | Value |
|---|---|
| Current simplified approach | 1.530 EUR m |
| Current internal model (3 x VaR + 3 x sVaR) | 11.068 EUR m |
| FRTB standardised | 5.100 EUR m |
| FRTB standardised after 0.9 multiplier | 4.590 EUR m |
| FRTB internal model (1.5 x stressed ES) | 4.666 EUR m |
| Allocated capital, FRTB standardised after multiplier | 7.459 EUR m |
| Allocated capital, FRTB internal model | 7.581 EUR m |
| Risk weight implied by the internal-model estimate | 54.9 % |
Try it. Set the risk weight to 37%, the figure ISDA argued for in 2021: the standardised charge falls below the internal-model estimate. Set the liquidity horizon to 60 days, the horizon of ‘other commodities’: the internal model would cost 73% more.
Result. A standardised bank today holds EUR 1.53m of own funds against the inventory. From January 2027 the FRTB standardised approach requires EUR 5.10m, or EUR 4.59m after the 0.9 multiplier: three times more. A bank with an internal model today holds EUR 11.07m, because the multiplier of 3 applies to both VaR and stressed VaR; under the FRTB internal model the approximate figure is EUR 4.67m, a risk weight of 54.9%, close to the regulatory 60%. Allocated capital at Meridian is EUR 7.46m under the standardised approach after the multiplier and EUR 7.58m under the internal model.
Reading. Three lessons. First, the 60% risk weight is not absurd when set against stressed carbon volatility over a 20-day horizon; ISDA’s argument for 37% rests on a different reading of the same data, and the debate is about calibration, not principle. Second, an inventory worth EUR 8.5m consumes about EUR 7.5m of Meridian’s capital: holding allowances is almost a full-equity activity, which is why carbon desks hold small inventories and prefer to hedge client forwards with exchange futures of the same expiry, which net to zero in the standardised approach. Third, the climate stress of stage 5.3 is not in Pillar 1 at all: the regulatory capital sees the volatility of the carbon price, not the direction of policy. Climate stress losses enter the bank’s capital only through the ICAAP and Pillar 2.
WHO DOES WHAT — The head of carbon trading takes two decisions from this stage before January. She keeps the client inventory but caps it, because the flow business with industrial clients needs allowances on the shelf in November and December, when compliance buyers hedge. And she instructs the desk to hedge every client forward with the future of the same December expiry, even when the later contracts are less liquid and cost a few cents more in bid-offer. Product Control is asked to report the residual tenor mismatch weekly, in tonnes and in FRTB own funds.
PITFALL — “Hedged” means different things to the trader and to the regulator. The Castellan forwards and the offsetting futures net to zero only if they have the same expiry. Hedging the 2029 forward with the more liquid December 2027 future, a common practice, leaves two positions in different tenors that the standardised approach correlates at 99%: the residual charge is the square root of 2% of the weighted position, several million euros of own funds on the 2029 tranche alone. The Delegated Act’s higher tenor correlation for EU ETS instruments reduces this, which is exactly the problem it targets.
Question. What is the CVA of the Vantoria gas swap, how much does wrong-way risk add, and why is the Castellan carbon hedge the opposite case?
Data. Vantoria pays a fixed EUR 32.00 per MWh and receives the TTF price on 1.5 TWh a year over two years, settled quarterly in this simplified model (eight quarters, equal volumes). Meridian receives fixed and pays floating; it is owed money when gas prices fall below EUR 32. TTF forward EUR 31.60; lognormal volatility 45% a year; discount rate 2.50%; Vantoria’s five-year credit spread 150 bp (consistent with its 2031 bond in C08); market LGD 60%, so the default intensity is 2.5% a year; Meridian’s funding spread for FVA 12 bp (the two-year FTP liquidity premium of the course). For wrong-way risk the XVA desk uses a two-state model (case assumption): with a probability of 20%, the economy is in the sudden wake-up state, where the gas curve is 30% lower (industrial demand destruction) and Vantoria’s default intensity is four times that of the normal state. The two states are calibrated so that their average gas forward equals the market forward and their average default intensity equals the market intensity: only the co-movement changes.
FORMULA
is the market default intensity; the probability of the stress state; the multiplier of the default intensity in that state; the gas price shock in the stress state (); and the gas forwards in the normal and stress states; the CVA formula of section 2.4 with expected exposures computed as the value of a put on gas struck at the fixed price, times the remaining volume.
INTERACTIVE LAB — CVA of a client commodity swap with and without wrong-way risk
Build the expected exposure profile of an uncollateralised swap in which the client buys a commodity forward, compute CVA under independence and under a two-state wrong-way model, and see the sign flip when the stress state moves the price the other way.
| Parameter you can move | Range | Default |
|---|---|---|
| Forward price | 10 to 150 EUR per unit | 31.6 EUR per unit |
| Fixed price paid by the client | 10 to 150 EUR per unit | 32 EUR per unit |
| Total volume | 0.5 to 10 million MWh or million t | 3 million MWh or million t |
| Tenor | 1 to 5 years | 2 years |
| Price volatility | 10 to 100 % a year | 45 % a year |
| Discount rate | 0 to 5 % | 2.5 % |
| Client credit spread | 20 to 1000 bp | 150 bp |
| Market LGD | 20 to 90 % | 60 % |
| Probability of the stress state | 0 to 50 % | 20 % |
| Default intensity multiplier in the stress state | 1 to 10 | 4 |
| Price shock in the stress state | -60 to 60 % | -30 % |
| Funding spread for FVA | 0 to 100 bp | 12 bp |
| Result at default settings | Value |
|---|---|
| Default intensity, normal state | 1.5625 % |
| Default intensity, stress state | 6.2500 % |
| Forward, normal state | 33.62 EUR |
| Forward, stress state | 23.53 EUR |
| Mid-market value of the trade to Meridian | 1,166.8 EUR k |
| Peak expected exposure (independence) | 11.51 EUR m |
| Expected positive exposure (independence) | 8.66 EUR m |
| CVA under independence | 248.3 EUR k |
| CVA with wrong-way (or right-way) risk | 338.8 EUR k |
| CVA with dependence / CVA under independence | 1.3641 |
| Extra CVA per unit of volume | 0.0301 EUR |
| FVA | 20.3 EUR k |
| Wrong-way surcharge (difference of the two CVAs) | 90.4 EUR k |
Try it. Set the shock to +35% with the Castellan inputs of the second test (forward 87.34, fixed 87.84, 1.357 Mt, one year, 40% volatility, 120 bp spread): the dependence lowers the CVA by 19%. Then set the multiplier to 1: part of the wrong-way effect is only the fatter distribution of prices.
The chart shows the three exposure profiles. Under the market forward the expected exposure rises for three quarters, as uncertainty about gas prices grows, then falls as quarterly settlements reduce the remaining volume. In the stress state the profile starts high and only declines: the swap is already deep in the money for Meridian when gas is 30% lower, and every settlement simply pays down part of a large claim. The dashed average profile is above the market profile throughout, because a mixture of two price levels is a wider distribution than a single forward, and an exposure is an option on that distribution.
Result, Vantoria. At mid-market the trade is worth EUR 1.167m to Meridian (the EUR 0.40 per MWh margin on 3 TWh, discounted). The expected exposure peaks at EUR 11.51m after nine months and the expected positive exposure over the life is EUR 8.66m. Under independence the CVA is EUR 248,300 and the FVA EUR 20,300. In the two-state model the default intensity is 1.5625% in the normal state and 6.25% in the stress state, and the gas forward EUR 33.62 and EUR 23.53; the CVA rises to EUR 338,800, a ratio of 1.364: wrong-way risk adds 36%. Of that, a ratio of 1.059 comes from the fatter price distribution alone (multiplier set to 1), the rest from the dependence. Per MWh the extra charge is EUR 0.030.
Result, Castellan. For the 2027 tranche of the Castellan carbon forward (C10: 1.357 Mt, forward EUR 87.34, client price EUR 87.84, one year, volatility 40%, Castellan spread 120 bp), the stress state is a carbon price up by 35%. Castellan buys allowances forward, so Meridian is exposed when carbon falls, that is, when Castellan is doing well. The CVA falls from EUR 97,500 to EUR 79,100, a ratio of 0.812: right-way risk.
Reading. The same climate scenario creates wrong-way risk with one carbon-intensive client and right-way risk with another. The difference is not the client’s carbon intensity, it is the direction of the hedge relative to the driver. Vantoria hedges an input whose price falls in the transition scenario that hurts it; Castellan hedges an input whose price rises in the scenario that hurts it. A WWR policy that flags clients by sector (“all cement and chemicals companies are wrong-way”) would be wrong half the time. Proportion matters too: wrong-way risk adds EUR 90,400 to a trade with a EUR 1.17m margin. The CVA is not where the danger lies; the exposure jump is (stage 5.7).
WHO DOES WHAT — The XVA trader charges Vantoria the independence CVA in the price, as for any BBB client, and books the wrong-way surcharge of EUR 90,400 as a reserve funded from the desk’s sales credit, after a discussion with the CCR officer. Passing it to the client would add EUR 0.03 per MWh to the fixed price, small enough to accept but hard to explain to a treasurer who compares three bank quotes on the screen. Product Control records the WWR reserve as a model reserve, reviewed quarterly.
Question. If Vantoria signed a CSA, would collateral protect Meridian against the gas price jump of the stress state?
Data. The jump is a 30% fall in the gas forward, from EUR 31.60, on the full 3 TWh (the stress state arriving at the start of the trade). Three set-ups: no CSA (today); a CSA with daily variation margin and a zero threshold; the same CSA plus an initial margin covering a ten-day 99% move at the 45% volatility, the standard calibration of a margin model. A margin period of risk of ten business days. The initial margin may be posted in cash or in securities; in the second case we allow a fall in their value in the same scenario.
INTERACTIVE LAB — Exposure after a climate price jump: no CSA, variation margin, initial margin
Measure the exposure left after a sudden price jump under different collateral set-ups, and the shortfall of an initial margin calibrated on normal volatility.
| Parameter you can move | Range | Default |
|---|---|---|
| Remaining volume | 0.5 to 10 TWh | 3 TWh |
| Gas forward before the jump | 10 to 100 EUR/MWh | 31.6 EUR/MWh |
| Fixed price paid by the client | 10 to 100 EUR/MWh | 32 EUR/MWh |
| Price jump | -60 to 0 % | -30 % |
| Volatility used by the margin model | 10 to 120 % a year | 45 % a year |
| Margin period of risk | 5 to 20 business days | 10 business days |
| CSA threshold | 0 to 30 EUR m | 0 EUR m |
| Fall in value of securities posted as initial margin | 0 to 30 % | 0 % |
| Result at default settings | Value |
|---|---|
| Value to Meridian before the jump | 1.20 EUR m |
| Value change from the jump | 28.44 EUR m |
| Exposure, no CSA | 29.64 EUR m |
| Exposure, variation margin only | 28.44 EUR m |
| Initial margin held | 19.85 EUR m |
| Exposure, variation and initial margin | 8.59 EUR m |
| Share of the jump covered by initial margin | 69.8 % |
Try it. Raise the margin-model volatility until the initial margin covers the jump: about 65% a year, which is roughly the stressed volatility of European gas, and far above what a client will accept to post.
Result. Before the jump the swap is worth EUR 1.20m to Meridian (undiscounted, hence slightly above the EUR 1.167m of stage 5.6); the jump adds EUR 28.44m. Without a CSA the exposure becomes EUR 29.64m. Daily variation margin does not help against a jump: the move happens before the next call, and if Vantoria defaults in the same scenario the exposure is the full EUR 28.44m. An initial margin of EUR 19.85m, calibrated on 45% volatility over ten days, covers 69.8% of the jump and leaves EUR 8.59m uncovered; if the margin is posted in securities that fall 5% in the same scenario, EUR 9.58m.
Reading. Collateral turns a diffusion risk into a jump risk; it does not remove the jump. This matters for climate because policy shocks arrive as jumps. Two consequences on the desk. First, the collateral schedule matters: initial margin in Vantoria’s own bonds would be specific wrong-way risk and is excluded by Meridian’s collateral policy, and bonds of the chemicals sector would carry general wrong-way risk; cash or high-quality government bonds are the clean answer, and the ECB’s climate factor is a reminder that collateral values can depend on the issuer’s transition risk. Second, the client is not obliged to sign a CSA: as a non-financial counterparty below the EMIR clearing threshold, Vantoria is exempt from margin requirements, and most industrial treasurers refuse to tie up cash.
WHO DOES WHAT — The CCR officer proposes a middle way that Vantoria’s treasurer may accept: no CSA while Vantoria stays at BBB, and a rating trigger in the ISDA Schedule that requires collateral above a threshold if it is downgraded to BB+. The XVA trader warns that a rating trigger is itself wrong-way: it bites exactly when Vantoria is short of cash, and the cliff can accelerate the default it is meant to protect against. The committee accepts the trigger with a EUR 10m threshold, and asks for the stressed exposure to be reported monthly.
Question. What does the market-risk manager report to the Market Risk Committee, and which limits change?
Data and result. The table collects the figures of the case as they appear in the risk report of Friday 13 November 2026.
| Metric | Value | Limit or trigger | Status |
|---|---|---|---|
| Ten-day VaR 99%, combined book | EUR 1.796m | EUR 2.5m | 72% used |
| Climate stress, policy rollback | EUR −3.391m | EUR 3.0m | Breach, 113% |
| Climate stress, sudden wake-up | EUR +1.627m | EUR 3.0m | No loss |
| EUA inventory | 100,000 t | 150,000 t | Within |
| Vantoria exposure in stress (no CSA) | EUR 29.64m | PFE line EUR 25m | Above line in stress |
| WWR CVA surcharge | EUR 90,400 | Reserve | Booked |
The WWR surcharge is the difference between the two CVAs of stage 5.6. The committee takes three decisions (case narrative). It confirms the temporary limit of EUR 3.5m for two weeks while the inventory is reduced. It adds a two-sided carbon policy limit: the carbon desk’s net delta to the EUA price may not exceed EUR 60,000 per 1% move, which at today’s price forces the inventory well below 100,000 tonnes and caps the loss in both scenarios at once. And it asks CCR to replace the sector-based WWR flag by a direction-based flag: each client hedge is tagged by whether its value to Meridian rises or falls in the bank’s climate scenarios, and the counterparty line is checked on the stressed exposure, not only on the PFE.
PITFALL — A stressed exposure above the line is not a breach in Meridian’s framework, because lines are set on PFE. Many banks are in the same position: climate stress exposures are reported, discussed and filed. Until a stressed-exposure limit exists, the analysis has no teeth.
The committee pack also carries the figures into the bank’s other processes. The rollback loss enters the ICAAP’s climate stress, where it is added to the banking-book losses of E01 under the same narrative, so that the board sees one climate number per scenario rather than one per department. The Vantoria and Castellan exposures in the stress state enter the large-exposure and concentration review. And the WWR reserve enters the quarterly valuation committee, where Product Control explains its movement together with the other model reserves.
| Figure | Value | Stage |
|---|---|---|
| Index CS01 / one-day P&L standard deviation | EUR 30,303 / EUR 244,100 | 5.2 |
| Ten-day VaR 99% / ES 97.5% / ES with liquidity horizons | EUR 1.796m / 1.804m / 2.764m | 5.2 |
| Stand-alone VaR: EUA inventory / credit positions | EUR 1.501m / 0.747m | 5.2 |
| Diversification benefit | EUR 1.286m | 5.2 |
| Rollback P&L / limit use / with 50,000 t | EUR −3.391m / 113.0% / 56.4% | 5.3 |
| Wake-up P&L / of which Vantoria / allowances | EUR +1.627m / −1.806m / +2.975m | 5.3 |
| VaR with regime correlation / stressed VaR | EUR 1.470m / 2.135m | 5.4 |
| Covariance add-on / total climate add-on | EUR 0.339m / 3.731m | 5.4 |
| Own funds on EUR 8.5m of EUAs: simplified / internal model today | EUR 1.53m / 11.07m | 5.5 |
| FRTB SA / after 0.9 / FRTB IMA approx. | EUR 5.10m / 4.59m / 4.67m | 5.5 |
| Vantoria swap: mid value / peak EE / EPE | EUR 1.167m / 11.51m / 8.66m | 5.6 |
| CVA independent / with WWR / ratio | EUR 248,300 / 338,800 / 1.364 | 5.6 |
| Castellan 2027 tranche: CVA independent / right-way / ratio | EUR 97,500 / 79,100 / 0.812 | 5.6 |
| Exposure after jump: no CSA / VM / VM and IM | EUR 29.64m / 28.44m / 8.59m | 5.7 |
IN ONE SENTENCE — A client hedge consumes capital three times over (counterparty credit risk, CVA risk, and market risk on whatever the trader leaves unhedged), and climate changes none of the formulas but several of the inputs.
Market risk. From 1 January 2027 Meridian’s trading desks compute own funds under the FRTB. The credit flow desk’s bonds require about EUR 8.7m of own funds under the standardised approach (C08, section 5.5); the carbon inventory EUR 4.59m after the 0.9 multiplier (section 5.5); the hedged client trades almost nothing, provided the hedges have matching tenors. Meridian has applied for the internal model on its rates and FX desks and keeps commodities and credit on the standardised approach for 2027, a common choice for desks whose PLA results are uncertain; the Delegated Act’s relief on the PLA test and the optional multiplier reduce the cost of the decision for three years. Climate appears nowhere in these formulas. It appears in Pillar 2: the ECB expects the ICAAP to cover material climate risks, and Meridian holds internal capital for the climate stress losses of section 5.3 in its economic capital model.
Counterparty credit risk. The Vantoria swap is measured with SA-CCR: replacement cost equal to the mid-market value in Meridian’s favour, plus an add-on of 18% of the notional for gas (maturity factor 1 for an unmargined trade of more than a year), times 1.4. The exposure is risk-weighted with Vantoria’s F-IRB parameters (PD 0.20%, LGD 40%) and an effective maturity of one year, the floor of the formula and close to the notional-weighted life of a two-year swap settled quarterly. The offsetting TTF futures are cleared: trade exposures to a qualifying CCP carry a 2% risk weight, plus a small charge for the default fund contribution.
CVA risk. Exempt: Vantoria is an EU non-financial counterparty (Article 382(4) CRR). If the same trade were with a commodity trading house classified as a financial counterparty, the BA-CVA or SA-CVA would add RWA of the same order as the counterparty RWA (C07 shows a ratio above two for Borealis’s swaps under the BA-CVA).
Leverage and large exposures. The SA-CCR exposure enters the leverage ratio measure without risk weights, and the exposure to Vantoria adds to the group’s large-exposure count together with the bonds held by the credit desk. In the wake-up scenario, both rise at once: a concentration that only a cross-desk view reveals.
Data. Day-one revenue: the mid-market value of EUR 1.167m less the CVA and the FVA transferred to the XVA desk, spread over the two years of the trade. Funding of the initial margin on the TTF futures (20% of notional, case assumption) at the two-year liquidity premium of 12 bp is deducted as a cost of funds (the MVA). Operating costs: 60% of revenues. Expected loss: PD times F-IRB LGD of 40% times EAD. Operational-risk RWA: EUR 2.5 per euro of revenue. The stressed column applies the sudden wake-up state to the same trade: the gas forward 30% lower, which raises the replacement cost, and Vantoria downgraded to BB (PD 1.00%).
INTERACTIVE LAB — RAROC of the Vantoria gas hedge: as priced, with the wrong-way reserve, and in the stress state
Compute SA-CCR exposure, RWA, expected loss and RAROC of the client gas swap, and show how the wrong-way reserve, the price jump and a downgrade change it.
| Parameter you can move | Range | Default |
|---|---|---|
| Volume | 0.5 to 10 TWh | 3 TWh |
| Gas forward | 10 to 100 EUR/MWh | 31.6 EUR/MWh |
| Client fixed price | 10 to 100 EUR/MWh | 32 EUR/MWh |
| Tenor | 1 to 5 years | 2 years |
| CVA under independence (from 5.6) | 0 to 2000 EUR k | 248.335 EUR k |
| CVA with wrong-way risk (from 5.6) | 0 to 2000 EUR k | 338.763 EUR k |
| FVA (from 5.6) | 0 to 500 EUR k | 20.29 EUR k |
| Charge the wrong-way reserve to the trade | no / yes | no |
| Vantoria PD | 0.03 to 5 % | 0.2 % |
| Effective maturity | 1 to 5 years | 1 years |
| Initial margin on the futures hedge | 5 to 40 % of notional | 20 % of notional |
| Liquidity premium on margin funding | 0 to 50 bp | 12 bp |
| Gas price move in the stress state | -60 to 30 % | -30 % |
| Vantoria PD in the stress state | 0.03 to 10 % | 1 % |
| Result at default settings | Value |
|---|---|
| Mid-market value at inception | 1,166.8 EUR k |
| Adjusted notional | 94.80 EUR m |
| SA-CCR add-on | 17.064 EUR m |
| Exposure at default | 25.523 EUR m |
| F-IRB risk weight | 26.69 % |
| Day-one revenue after CVA and FVA | 898.2 EUR k |
| Margin funding cost (MVA) per year | 22.8 EUR k |
| Net revenue per year | 426.4 EUR k |
| Operating costs (60%) | 255.8 EUR k |
| Expected loss | 20.4 EUR k |
| Counterparty credit RWA | 6.812 EUR m |
| Operational-risk RWA | 1.066 EUR m |
| Allocated capital (13% of RWA) | 1,024.1 EUR k |
| RAROC after tax | 10.99 % |
| Net revenue per year for a 12% RAROC | 465.8 EUR k |
| Exposure at default in the stress state | 58.22 EUR m |
| Risk weight in the stress state | 65.14 % |
| RAROC in the stress state | -0.92 % |
Try it. Raise the client price to EUR 32.03 (the extra CVA per MWh passed on): RAROC barely moves. Then set the stress PD to 0.20%: most of the stressed loss of return comes from the downgrade, not from the exposure jump.
Result.
| Line (EUR k per year unless stated) | As priced | WWR reserve charged |
|---|---|---|
| Day-one revenue after CVA and FVA (EUR k, once) | 898.2 | 807.8 |
| Net revenue per year, after MVA of 22.8 | 426.4 | 381.1 |
| Operating costs (60%) | −255.8 | −228.7 |
| Expected loss (0.20% x 40% x EUR 25.52m) | −20.4 | −20.4 |
| RWA: counterparty / operational (EUR m) | 6.812 / 1.066 | 6.812 / 0.953 |
| Allocated capital | 1,024.1 | 1,009.4 |
| RAROC after tax | 10.99% | 9.81% |
The SA-CCR add-on is EUR 17.06m on an adjusted notional of EUR 94.80m; with the replacement cost the EAD is EUR 25.52m, at a risk weight of 26.69%. Priced as a normal BBB trade, the hedge earns a RAROC of 10.99%, below the 12% hurdle; it would need EUR 465,800 of net revenue a year. Charging the wrong-way reserve lowers RAROC to 9.81%. In the stress state the EAD more than doubles to EUR 58.22m, the risk weight at a 1% PD is 65.14%, and RAROC becomes −0.92% (−1.19% with the reserve).
Reading. The wrong-way reserve costs 1.2 points of RAROC; the stress state costs twelve. Climate risk in a client hedge does not show in the price; it shows in the capital and the expected loss after the scenario. Coverage accepts a sub-hurdle RAROC on the hedge because Vantoria is a relationship client with lending and DCM business (B03 and C08 show the same logic). The CCR officer adds a condition: if Vantoria’s rating falls below BBB−, any new hedge requires a CSA.
The FVA of section 5.6 (EUR 20,300) is the desk’s funding charge on the expected positive exposure. It understates the liquidity problem in a jump. In the stress state the gas forward falls 30%; Meridian’s long TTF futures lose about EUR 28.4m, which the clearing house calls in cash the same day, while Vantoria owes the matching amount only as quarterly settlements fall due over two years. Treasury funds the gap, and the LCR captures part of it: the historical look-back approach (Delegated Regulation (EU) 2015/61, Article 30(3), with the methodology of Delegated Regulation (EU) 2017/208) adds an outflow equal to the largest net 30-day collateral flow of the past 24 months. The CCP’s initial margin also rises when volatility jumps, which C10 shows for carbon futures. Meridian’s Treasury charges the commodities desk a contingent liquidity cost on the stressed margin call, not only on the expected exposure; the charge is small in calm markets and is the reason the desk limits the tenor of uncollateralised client hedges to three years.
For Meridian the swap and the futures are held for trading and measured at fair value through profit or loss under IFRS 9. Fair value under IFRS 13 includes the CVA and the FVA, booked as valuation adjustments; the wrong-way reserve is a model reserve that Product Control reviews quarterly. TTF swaps of up to two years are Level 2 (observable forward curve); long-dated carbon forwards beyond the liquid December contracts may be Level 3, with day-one profit deferred. Under the prudent valuation rules (Article 105 CRR and the EBA RTS), additional valuation adjustments for close-out costs, model risk and concentration are deducted from CET1; a WWR model with uncertain calibration is a natural candidate for a model-risk adjustment. On the client side, Vantoria can designate the swap as a cash flow hedge of its forecast gas purchases under IFRS 9, provided the purchases remain highly probable; a plant closure in the stress state would make part of the hedged volume no longer expected, and the corresponding gains or losses would be recycled to profit or loss, an accounting echo of the over-hedging problem of C10.
The trading book is largely outside the bank’s sustainability reporting today. The Green Asset Ratio covers banking-book exposures; the trading book KPI and the fees and commissions KPI of the Taxonomy Article 8 Delegated Act were postponed by the Commission’s simplification act of 4 July 2025 to 1 January 2028. The Pillar 3 ESG templates of Article 449a CRR focus on the banking book (exposures to carbon-intensive sectors, real estate collateral, alignment metrics; the GAR and BTAR templates are stopped for large institutions under the EBA’s final draft of 22 June 2026, not yet adopted). PCAF’s financed emissions standard covers loans, bonds and listed equity held on balance sheet, and its facilitated emissions standard covers capital markets underwriting; neither attributes emissions to commodity derivatives or to a carbon allowance inventory. The bonds held by the credit desk are a grey zone: some banks include trading inventory in financed emissions, most exclude it.
This does not make the trading book irrelevant to the bank’s sustainability profile. Meridian’s transition plan (see E12) commits to supporting clients’ decarbonisation, and its carbon desk, which helps Castellan buy allowances efficiently, and its gas hedges for Vantoria are part of how carbon-intensive clients manage the transition. The honest reporting answer is qualitative: disclose the trading activities linked to carbon and fossil fuels, the climate stress results of the trading book in the ICAAP summary, and the policy on wrong-way risk with carbon-intensive clients. Overstating the carbon desk as “green” would be a red flag; so would ignoring it.
| Risk | Where it shows | Control | Owner |
|---|---|---|---|
| Carbon policy jump | EUA inventory, carbon options | Two-sided stress limit, delta cap | MRM, carbon desk head |
| Regime change in correlations | VaR understates or overstates hedges | Regime-specific stress, VaR backtesting | Risk Methodology |
| Tenor mismatch in hedges | FRTB SA charge, basis P&L | Matched-expiry hedging rule | Trader, MRM |
| Wrong-way risk with clients | CVA, stressed exposure | Direction-based WWR flag, stressed-exposure line | CCR officer, XVA desk |
| Jump not covered by collateral | Exposure after margin | Collateral eligibility, IM calibration on stress | Collateral risk |
| Liquidity of margin calls | Treasury, LCR | Contingent liquidity charge, tenor cap | Treasury, desk |
| Model risk in climate overlays | Reserves, capital | Independent validation, model-risk AVA | Model Validation, Product Control |
| Misleading “green” labels on trading activity | Disclosures, marketing | Product approval, disclosure review | Compliance, Sustainability reporting |
Red-flag checklist for a reviewer of a trading-book climate framework:
What changed recently: the FRTB is finally a date, not a debate; collateral frameworks have started to price climate; and supervisors have moved from asking for climate risk “integration” to fining delays. Open debates: whether a short-horizon measure can say anything about climate at all (many risk managers, and Roncalli’s handbook, argue that only scenario analysis can); how to calibrate WWR when defaults linked to transition shocks are almost absent from data; whether the EU’s CVA exemption for corporates hides a growing climate-linked exposure; and whether trading books should enter financed-emissions metrics.
INTERACTIVE LAB — Market risk, counterparty risk and climate: check yourself
Test the key concepts and numbers of the module.
Q1. Why did the Basel Committee choose 97.5% for expected shortfall?
Answer: B. Under a normal distribution the 97.5% ES (2.338 standard deviations) is almost the 99% VaR (2.326), so the level of capital is broadly kept while the measure captures the tail.
Q2. What is the FRTB minimum liquidity horizon for energy and carbon emissions trading prices?
Answer: C. MAR33 sets 20 days for energy and carbon emissions trading prices and 60 days for their volatilities.
Q3. Under MAR33, what is the liquidity horizon for investment-grade corporate credit spreads?
Answer: A. Investment-grade corporate spreads have 40 days, high-yield corporate 60 days; 20 days applies to investment-grade sovereign spreads.
Q4. In the case, which position dominates the stand-alone VaR?
Answer: D. The inventory has a stand-alone ten-day VaR of EUR 1.501m, 48.7% of the sum of stand-alone VaRs.
Q5. Why must a trading-book climate stress be run in both directions?
Answer: B. In the case the transition scenario produces a gain and the rollback scenario breaches the limit.
Q6. What does the FRTB standardised approach charge on a linear EUA position?
Answer: B. Bucket 3, electricity and carbon trading, has a 60% delta risk weight; 18% is the SA-CCR factor, 15% plus 3% the current simplified approach.
Q7. Vantoria buys gas forward from Meridian; in the stress scenario gas falls and Vantoria’s credit worsens. This is:
Answer: C. Exposure rises when credit quality falls through a common macro driver, without a legal link: general wrong-way risk.
Q8. Why is the Castellan carbon forward right-way risk in the wake-up scenario?
Answer: B. Meridian’s exposure to Castellan arises when carbon falls, that is when Castellan is better off.
Q9. Why does daily variation margin not protect against a policy jump?
Answer: A. Margin covers diffusion between calls; a jump within the margin period of risk is uncovered.
Q10. Which EU rule removes CVA-risk capital on a hedge with an EU industrial client?
Answer: A. Article 382(4) exempts transactions with EU non-financial counterparties; the CVA is still priced and reserved.
Q11. In the RAROC of the Vantoria hedge, what reduces return most?
Answer: C. The reserve costs about 1.2 points; the stress state takes RAROC from 10.99% to -0.92%.
Q12. What does the ECB climate factor do?
Answer: B. Since 15 June 2026 it applies to marketable assets of non-financial corporations, with a maximum additional reduction of 5%.
Q13. Why is a ‘climate VaR’ over ten days hard to interpret?
Answer: B. Historical data cannot contain the scenario; the add-on is driven by an assumed jump and stressed parameters.
Q14. Which tenor-mismatch effect does the 2026 Delegated Act target for EU ETS instruments?
Answer: B. Public summaries report that it raises the tenor correlation used for EU ETS instruments, reducing the charge on hedged positions of different expiries.
A desk holds EUR 5m of EU allowances (daily volatility 3%) and bonds with a CS01 of EUR 20,000 per bp (daily spread volatility 5 bp). The correlation between the EUA price and the spread is −0.3. Compute the ten-day 99% VaR, the ten-day 97.5% ES, and the ES with a 20-day horizon for carbon and 40 days for the spread.
Solution. One-sigma P&L: carbon EUR 150,000; bonds EUR 100,000. Because a wider spread loses money on the bonds, the P&L correlation is +0.3. Standard deviation: thousand. VaR: EUR 1.499m. ES: EUR 1.506m. Credit-only ten-day ES: EUR 0.739m. With liquidity horizons, EUR 2.373m. The longer horizon on the spread adds 58% to the ten-day ES.
A one-year client swap has expected exposures at the end of each quarter of EUR 4m, 5m, 5m and 3m. The client’s spread is 300 bp, market LGD 60%, discount rate 2.50%. In a two-state model, the stress state has a 25% probability, triples the default intensity and gives exposures of EUR 7m, 8m, 8m and 6m; the normal state gives EUR 3m, 4m, 4m and 2m (the averages equal the profile above). Compute the CVA under independence and with WWR.
Solution. Market intensity . Independence: summing gives EUR 122,700. States: , . CVA in the normal state EUR 63,000, in the stress state EUR 408,300. Weighted: EUR 149,400, a ratio of 1.22. The averages of exposure and of default probability are unchanged; only their co-movement adds 22%.
Meridian sells a cement producer 1 Mt of allowances forward for delivery in one year at EUR 80 (notional EUR 80m), with a mid-market value to Meridian of EUR 0.5m. No CSA. The client is BBB (PD 0.20%, LGD 40%), M = 1 year. Annual net revenue EUR 400,000, costs 60%, operational RWA 2.5 per euro of revenue. The desk hedges with a future of the same expiry but keeps 30,000 tonnes unhedged. Compute the SA-CCR EAD, the RAROC of the client trade, and the FRTB standardised own funds on the open position.
Solution. Add-on ; EAD EUR 20.86m. Risk weight 26.69%, credit RWA EUR 5.567m, plus EUR 1.0m operational, total EUR 6.567m; capital at 13% EUR 853,800. Expected loss EUR 16,700. RAROC , just above the hurdle. The open 30,000 tonnes are worth EUR 2.4m; FRTB standardised own funds EUR 1.44m, or EUR 1.296m after the 0.9 multiplier, that is EUR 2.106m of allocated capital: an unhedged sliver of inventory consumes more than twice the capital of the client trade.
1. What is the difference between VaR and expected shortfall, and why did the FRTB switch? VaR is the loss exceeded with a given probability; ES is the average loss beyond that point. VaR is blind to the shape of the tail: two books with the same VaR can have very different losses on bad days. ES also adds up better across sub-portfolios, which matters for capital allocation. The FRTB chose 97.5% ES because, for normal distributions, it is close to the 99% VaR, so the level of capital did not jump for that reason alone.
2. How would you integrate climate risk into a trading book’s risk framework? Not through VaR, which is calibrated on the past and cannot contain a policy shock that has not happened. I would define a small set of climate scenarios, run in both directions, translated into instantaneous shocks plus stressed volatilities and correlations, with documented calibration. I would set stress-loss limits on them, add sensitivity limits on the main climate drivers (carbon delta, sector CS01), and report stressed counterparty exposures for carbon-intensive clients. Finally, I would feed the stress losses into the ICAAP.
3. Is a carbon-intensive client always wrong-way risk? No. Wrong-way risk depends on the direction of the trade relative to the driver that hurts the client. A cement company buying allowances forward is right-way in a carbon spike, because the bank owes it money when it suffers. A chemicals company that fixed its gas price is wrong-way if the transition scenario lowers gas prices while hurting its business. The flag must be set trade by trade, not by sector.
4. Why is the FRTB standardised charge on EU allowances controversial? The 60% risk weight is among the highest of the commodity buckets and implies a stressed volatility above what ISDA measured for EUAs. The 99% tenor correlation also penalises hedges between different December contracts, which are almost perfectly correlated in practice. Supporters reply that carbon is a policy-driven market prone to jumps. The EU’s 2026 Delegated Act adjusted the tenor correlation but not the risk weight.
5. What does collateral do against a climate shock? Variation margin removes the gradual build-up of exposure, and initial margin covers a move of a given size over the margin period of risk. Neither covers a jump larger than the margin model’s calibration, and a policy jump is exactly that. Collateral can also lose value in the same scenario if it is issued by companies in exposed sectors. So collateral reduces the problem but must be combined with stressed-exposure limits and careful eligibility.
6. Walk me through the CVA formula. CVA is the sum, over future dates, of the expected exposure times the probability that the counterparty defaults in that period times the loss given default, discounted. The exposure comes from simulating the trade’s value; the default probabilities come from the counterparty’s credit spread through the credit triangle. The formula assumes independence between exposure and default, which wrong-way risk breaks. On the desk, the XVA trader charges it in the price and hedges the spread sensitivity with CDS where possible.
7. Why might a bank accept a client hedge below its RAROC hurdle? Because the hedge is part of a relationship that includes lending, bond issuance and cash management, and the client chooses its banks for the package. The test is the RAROC of the relationship, with the hedge’s capital included. But the bank should also look at the stressed RAROC: a hedge that is marginal in normal times and deeply negative in a climate scenario is a concentration the relationship must justify explicitly.
8. What changes for market risk on 1 January 2027 in the EU? The FRTB replaces the old rules: ES with liquidity horizons, desk-level model approval with backtesting and PLA, a new standardised approach, and a default risk charge. The Commission’s Delegated Act of June 2026 softens the impact for three years, with an optional multiplier for banks with internal models, a 0.9 multiplier on standardised requirements and relief on the PLA test. For carbon desks the standardised charge roughly triples compared with the current simplified approach, which pushes towards matched hedging and small inventories.
Basel standards and principles
EU law and regulatory texts
Supervisors and central banks
Industry
Books and academic papers