A loan book is a tanker, not a speedboat. A EUR 6bn power-sector portfolio with an average residual life of three and a half years cannot be turned in a quarter, and every decision taken today shows up in the emissions metric two or three years later, when the old loans have run off and the new ones have been drawn. The people who plan that slow turn sit in what banks call portfolio steering: a small team, usually inside Group Strategy or the Chief Risk Officer’s area, that translates the bank’s climate commitments into sector targets, origination guidance and limits, then tracks every quarter whether the book is moving the right way.
The desk does not earn revenue. It earns its keep in three ways. First, it protects the franchise: a bank that publishes a power-sector target and misses it faces supervisory questions (the ECB treats the targets in a prudential transition plan as part of risk management), investor engagement and litigation risk. Second, it allocates scarce resources: capital, risk appetite and balance-sheet room are given to the business lines whose clients fit the trajectory. Third, it supports origination: once the trajectory is known, coverage bankers know which clients they can grow with and on what terms. The orders of magnitude are those of a strategic function: a team of five to fifteen people in a large European bank, a data and model budget of a few million euros a year, and an influence on the allocation of tens of billions of exposure.
Who is around the table.
| Function | What it brings | Typical document |
|---|---|---|
| Portfolio steering (Strategy) | Targets, pathway, gap analysis | Sector target note, quarterly dashboard |
| Credit portfolio management | Limits, concentration, hedging, sales | Sector limit framework |
| Climate risk (second line) | Scenarios, methodology, challenge | Methodology note, model validation |
| Sustainability reporting (Finance) | Year-end PCAF inventory, audited figures | Sustainability statement, Pillar 3 ESG |
| Sector coverage teams | Client transition plans, pipeline | Client climate assessment |
| ESG data office | Emissions, generation, EVIC, proxies | Data dictionary, quality scores |
A realistic week in November 2026.
WHO DOES WHAT — Portfolio steering is where the climate commitment of the annual report meets the credit committee. The analyst who can explain in one sentence why the power metric moved this quarter, and how much of the move came from clients and how much from the bank’s own reshuffling, is the person the committee listens to.
Before adding emissions, take away the climate entirely. What does it mean to steer a loan portfolio? The answer is the same for credit risk, for capital and for climate, and it explains why the sustainable layer is hard.
A bank’s balance sheet is steered at three levels. At group level, the board sets a risk appetite statement: maximum risk-weighted assets, a target CET1 ratio, a return on equity, concentration limits. At sector level, credit portfolio management turns this into sector limits (for instance, a ceiling of EUR 9bn of exposure to power and utilities) and a sector strategy that says which sub-segments to grow, hold or reduce. At client level, the credit committee approves individual limits and the coverage team allocates its balance sheet among clients according to their profitability.
The levers are always the same five:
A corporate loan book is made of facilities of different maturities. Each year a fraction matures. The bank decides whether to renew, to renew on different terms, or to let it go. The run-off profile therefore tells the steering team how much of the book it can actually influence by a given date.
IN ONE SENTENCE — The steerable share of a book at a horizon is the fraction of today’s exposure that will have matured or been repriced by then.
INTUITION — A loan book is like a car park where every car has a ticket with an exit time. If you want a car park full of electric cars in four years, you cannot tow the petrol cars away; you can only choose who gets the spaces as they free up. The longer the tickets, the slower the change.
FORMULA
is the steerable share at horizon years; the average residual maturity of the book, in years; the formula assumes maturities spread evenly between zero and ; the book size; the yearly amount that matures, and therefore the yearly origination needed to keep the book stable.
CASE — Meridian’s power-sector book is EUR 6,000m with an average residual maturity of 3.5 years. Between January 2027 and December 2030 (four years), the steerable share is 4 / 7 = 57.1%: EUR 3,429m will mature, and about EUR 857m matures every year. The other 42.9% will still be on the books in 2030 exactly as signed. Any target for 2030 has to be met by changing the 57% and by the clients themselves changing.
INTERACTIVE LAB — How much of the power book can Meridian steer by 2030?
See how the average residual maturity of a loan book limits the share that can be renewed differently before a target date.
| Parameter you can move | Range | Default |
|---|---|---|
| Book size | 1000 to 20000 EUR m | 6000 EUR m |
| Average residual maturity | 1 to 10 years | 3.5 years |
| Horizon (years to target date) | 1 to 10 years | 4 years |
| Result at default settings | Value |
|---|---|
| Steerable share at the horizon | 57.1 % |
| Exposure maturing by the horizon | 3,429 EUR m |
| Legacy exposure still on book | 2,571 EUR m |
| Exposure maturing each year | 857 EUR m |
Try it. Set the maturity to 6 years, typical of a book heavy in project finance: only one third of the book can be changed by 2030, and clients’ own decarbonisation becomes the main driver of the metric.
PITFALL — Run-off is not exit. A maturing loan to a client that remains a client is usually renewed, often at a larger size. The steerable share measures the opportunity to change terms, not a change already decided. And project finance and infrastructure loans, with tenors of fifteen years or more, barely run off at all before 2030 (see B06).
WHY IT MATTERS ON THE DESK — The first question a committee asks of any 2030 target is whether it is achievable. A steering analyst who starts with the run-off profile answers it before it is asked.
A sector book is also a credit portfolio. The power sector in Europe mixes very different risks: regulated networks rated A, integrated utilities in the BBB range, merchant independent power producers in BB, renewable project companies whose risk depends on contracts (see B06 and B07). The credit portfolio manager watches three numbers. Concentration: the largest five names in a utility book often carry half of the exposure, which means that one client decision moves the whole metric, a point that will matter for emissions too. Expected loss and RWA by grade: under F-IRB a BB exposure carries about twice the risk weight of a BBB one (82% against 39% at a 2.5-year maturity, from the CRR3 formula), so the mix by rating drives capital consumption. RAROC by client: the coverage team’s argument for keeping a client is rarely the loan margin alone; it is the hedging, payments and capital-markets business that come with it (see E03 for the full RAROC method).
Steering decisions are made in committees and recorded in documents. The risk appetite statement is approved by the board each year. The sector policy sets exclusion and eligibility criteria (for instance, no new financing of thermal coal power plants). The sector strategy note sets volume and mix objectives. The credit committee minutes record client-level decisions, including conditions linked to the client’s strategy. The ICAAP documents how all risks, including climate-related risks, are captured in internal capital. The steering metric is a management indicator, unaudited, refreshed quarterly; the reporting metric in the sustainability statement is a year-end figure subject to limited assurance. Keeping the two consistent is half the work.
The sustainable layer adds one variable to the steering problem: the emissions financed by the book. It brings a measurement standard (PCAF), target-setting methods (sector pathways, SBTi), alignment tools (PACTA, temperature scores), voluntary alliances (now largely gone) and prudential rules that turn a voluntary pledge into a supervised plan.
Module A03 teaches the PCAF attribution formula and data-quality scores in detail; this module uses them. In short, for a business loan the bank attributes to itself a share of the client’s emissions equal to its outstanding amount divided by the client’s enterprise value including cash (EVIC) for listed companies, or total equity plus debt for unlisted ones.
What was verified for this module (PCAF website, November 2026).
On the disclosure side, the ISSB issued targeted amendments to IFRS S2 on 11 December 2025, effective for periods beginning on or after 1 January 2027: an entity may limit its scope 3 category 15 disclosure to financed emissions, and may use an industry classification other than GICS to break them down. Facilitated emissions therefore remain a voluntary disclosure under IFRS S2.
A target needs a metric. Two families coexist.
Some banks add a technology-mix or energy-supply-financing ratio (low-carbon to fossil energy supply financing), which measures where the new money goes rather than the emissions of the clients.
IN ONE SENTENCE — A physical intensity target says how clean each unit of financed output must be; an absolute target says how much financed carbon is allowed in total.
A target is only science-based if it is derived from a scenario. In the power sector, banks have mostly used the IEA’s Net Zero Emissions by 2050 scenario (NZE). The power sector is the hinge of that scenario: Roncalli (2026, section 10.4) shows that a net-zero economy requires both massive electrification and decarbonisation of generation, with an illustrative fall of the average carbon intensity of electricity from about 500 gCO2/kWh in 2020 to about 20 gCO2/kWh in 2050. In the IEA’s 2023 update of the roadmap, low-emissions sources rise from 39% of global generation in 2022 to 71% in 2030.
The sectoral decarbonisation approach (SDA), developed for the SBTi, converts a sector pathway into a target for an individual company or portfolio. Its idea is convergence: by 2050 everyone reaches the sector’s 2050 intensity, and in between each actor closes its gap to that end point at the same proportional pace as the sector. Section 5.3 applies it.
NOTE — The NZE scenario is updated regularly and its detailed data are licensed. Banks rebase their pathway when a new edition is published and disclose the change. The figures used in this module’s case are a stylised pathway, not IEA data.
Three tools measure whether a book is on a pathway.
Net-Zero Banking Alliance. Launched in 2021 under the UN Environment Programme Finance Initiative, the NZBA required members to set intermediate sector targets within 18 months of joining. After a wave of departures of large US, Canadian, Japanese and British banks in late 2024 and 2025, the remaining members voted to stop operating as a member-based alliance; the decision was announced on 3 October 2025. Its target-setting guidance remains public and banks may still use it. Statements written before that date about “NZBA commitments” are therefore historical.
SBTi. The Science Based Targets initiative published its Financial Institutions Net-Zero Standard (FINZ), version 1.0, in July 2025. It sits beside the Financial Institutions Near-Term Criteria (FINT) version 2, which the SBTi confirmed in October 2026 will remain available with no end date. FINZ requires a fossil-fuel transition policy, introduces phase-out requirements for new fossil-fuel financing activities, requires coverage of defined segments of lending and investment (100% of the highest-emitting segments for near-term targets), and allows sector targets in physical intensity for emission-intensive sectors such as power, or climate-alignment targets using approved methodologies. Validation is voluntary.
Prudential rules. In the EU, the voluntary layer has been overtaken by prudential law. The Capital Requirements Directive as amended in 2024 (CRD VI, Article 76) requires institutions to have plans with quantifiable targets to address ESG risks over the short, medium and long term, including in the transition to net zero by 2050. The EBA’s Guidelines on the management of ESG risks (EBA/GL/2025/01) specify the content of these plans, including metrics and targets, and apply to large institutions from 11 January 2026. The ECB has made clear that it enforces its expectations: on 10 November 2025 it announced periodic penalty payments of EUR 187,650 on ABANCA for failing to complete an adequate climate and environmental materiality assessment within the deadline it had set.
Disclosure. The Pillar 3 ESG templates under CRR Article 449a show the bank’s exposures and financed emissions by sector. The EBA published final draft amended standards on 22 June 2026, with a first reference date of 31 December 2026 for large institutions (31 December 2027 for small and non-complex ones), subject to adoption by the Commission. The Taxonomy-related templates (GAR and BTAR) are stopped for large institutions; according to public summaries, the financed emissions template is kept and now splits scope 1, 2 and 3, and the template on exposures to the twenty most carbon-intensive companies are removed. The CSRD sustainability statement, narrowed in scope by the 2026 Omnibus, still requires a bank in scope to disclose its gross GHG emissions including scope 3 category 15 where material, and its targets.
WHY IT MATTERS ON THE DESK — Since 2025 the question “why does the bank still have a power-sector target if the NZBA is gone?” has a clear answer in Europe: because the transition plan required by CRD VI and the EBA guidelines needs metrics and targets, the supervisor reviews them, and the sustainability statement discloses them.
The “deal” of a steering desk is the annual target cycle: set or confirm a sector target, translate it into origination guidance, monitor, report. The cycle below is the one Meridian runs for its power-sector book; other sectors follow the same calendar.
INTERACTIVE LAB — The annual cycle of a power-sector target at Meridian Bank
Follow a sector target from data collection to the sustainability statement, with the role, the document and the control at each step.
| # | Who | Does what | Deliverable |
|---|---|---|---|
| 1 | ESG data office | Loads clients’ reported emissions and generation from annual reports and data vendors; assigns PCAF data-quality scores | Power-sector data file, cut-off 30 September |
| 2 | Sustainability reporting (Finance) | Computes the attributed emissions and generation of each client using year-end outstanding amounts and EVIC | Year-end financed emissions inventory |
| 3 | Portfolio steering (Strategy) | Computes the book intensity, compares it with the target pathway and decomposes the change | Variance analysis: clients versus portfolio mix |
| 4 | Climate risk (second line) | Checks the scenario edition, the methodology and any rebasing; challenges proxies | Second-line opinion |
| 5 | Energy coverage team | Collects client transition plans and capex programmes; scores credibility | Client climate assessments |
| 6 | Portfolio steering (Strategy) | Projects the book to the target year and sizes the gap; proposes levers | Steering plan with options |
| 7 | Credit portfolio management | Translates the chosen plan into sector limits, growth budgets and watch lists | Updated sector limit framework |
| 8 | Group Climate Committee | Approves the steering plan and the origination budget for low-carbon assets | Committee minutes |
| 9 | Credit committee | Applies the plan in each client decision: renewal terms, use-of-proceeds conditions, engagement milestones | Credit approvals with climate conditions |
| 10 | Board risk committee | Reviews progress against the transition plan and the risk appetite | Board risk report |
| 11 | Sustainability reporting (Finance) | Publishes the target, progress and methodology changes; the auditor performs limited assurance | Sustainability statement and Pillar 3 ESG tables |
| 12 | Joint Supervisory Team (ECB) | Reviews the transition plan and targets in the supervisory dialogue | SREP findings or follow-up letter |
Step 1: data. The power-sector metric needs, for each client, three numbers: scope 1 emissions of generation, net generation in MWh, and EVIC (or equity plus debt). Listed utilities report the first two in their sustainability statements; unlisted independent power producers are often estimated from plant-level data (capacity, load factor and fuel), which PCAF scores 3 to 4 (see A03). The data office freezes the file at a cut-off date, because a metric recomputed every time a vendor updates a figure cannot be explained to a committee. The lag is structural: in November 2026 the latest complete year is 2025.
Step 2: attribution. Finance runs the year-end inventory with the outstanding amounts on 31 December. The steering metric uses the same engine but on quarter-end drawn amounts. Reconciliation differences come from undrawn facilities, from EVIC updates and from clients whose sector code changed.
Step 3: variance analysis. This is the heart of the job. A change in book intensity between two years comes from two sources: clients becoming cleaner (or dirtier), and the bank changing its mix of clients. Steering teams decompose the change, because the first is real-economy progress and the second is a portfolio decision. Section 5.6 shows why this matters.
Step 4: second-line challenge. Climate Risk validates the method like any other model: scenario edition, interpolation between scenario points, the treatment of new clients, proxies. A rebasing (changing the baseline or the scenario) needs approval and disclosure.
Steps 5 and 6: client plans and projection. Coverage bankers collect clients’ transition plans: plant closures, renewables capex, fuel switching. The steering team turns them into a projected intensity per client for the target year, discounted for credibility. The projection is compared with the target, and the gap is what portfolio choices must close.
Steps 7 to 9: decision and execution. The plan becomes limits and budgets. A client may be placed on an engagement watch list with milestones (for example, a coal unit closure by a date), with consequences at renewal. The credit committee is where the plan bites: a renewal request is assessed against the client’s credit and against its fit with the trajectory. In practice, the strongest tool is the structure of renewals (use of proceeds, tenor, conditions), not outright refusal.
Steps 10 to 12: oversight and disclosure. The board risk committee reviews progress; the sustainability statement discloses the target, the baseline, the progress and any methodology change; the auditor gives limited assurance on the reported figures; the supervisor reads all of it in the context of the CRD VI transition plan.
PITFALL — Do not present the target as a reporting exercise. If the only owner of the power target is Finance, the target is a number that is observed, not managed. The control that matters is step 9: whether credit approvals actually reference the trajectory.
Question. In November 2026 Meridian’s Group Climate Committee asks the steering team to propose a 2030 intensity target for the power-generation book, consistent with the transition plan the bank must hold under CRD VI, and a steering plan to reach it. The committee wants to know three things: where the book stands, what the clients’ own plans deliver, and what the bank must change, at what cost.
Data. The book is concentrated in six groups (case assumptions; all figures are for financial year 2025, the latest complete year, and all six report emissions and generation, so data quality is good).
| Client | Profile | Rating | Outstanding (EUR m) |
|---|---|---|---|
| Helios Renewables | Wind and solar producer | BBB | 900 |
| Castilla Energía | Integrated utility, gas and renewables | BBB | 2,200 |
| Baltica Power | Coal-heavy generator | BB | 700 |
| Rheinland Gaskraft | Gas-fired independent producer | BBB | 1,000 |
| Nordfjord Hydro | Hydro producer | A | 600 |
| Aegean Power | Gas and oil island systems | BB | 600 |
| Client | EVIC or E+D (EUR m) | Generation (TWh) | Intensity 2025 (kg/MWh) | Plan 2030 (kg/MWh) |
|---|---|---|---|---|
| Helios Renewables | 6,000 | 8 | 5 | 5 |
| Castilla Energía | 40,000 | 90 | 250 | 120 |
| Baltica Power | 7,000 | 30 | 820 | 600 |
| Rheinland Gaskraft | 5,000 | 18 | 360 | 330 |
| Nordfjord Hydro | 9,000 | 20 | 8 | 8 |
| Aegean Power | 3,000 | 10 | 520 | 300 |
The “plan 2030” column is the steering team’s credibility-adjusted reading of each client’s published transition plan: Castilla’s renewables build-out, Baltica’s closure of two of its oldest coal units, Rheinland’s limited efficiency gains, Aegean’s island interconnections. The book totals EUR 6,000m. Helios is the issuer of the green bond used as the course’s reference case (see C01); here it appears as a borrower.
Reading. Before computing anything, look at the structure. Baltica is 11.7% of the exposure but runs the dirtiest plants of the book; Castilla is the largest exposure but a small share of a very large company. Both facts will dominate the results.
Question. What is the physical emission intensity of the book in 2025?
Method. For each client, the attribution factor is the outstanding amount divided by EVIC (A03). The same factor is applied to emissions and to generation, so that the bank attributes to itself a slice of the client’s output and the emissions that came with it.
FORMULA
is the book intensity in kgCO2e per MWh; the attribution factor of client ; the outstanding amount; its EVIC or equity plus debt; its scope 1 emissions of generation; its net generation in MWh; its own intensity. One TWh at one kg per MWh is one thousand tonnes, so attributed emissions come out in kt when generation is in TWh.
INTUITION — The book intensity is the carbon content of “Meridian’s share” of the electricity its clients produce: imagine every client handing the bank a slice of its output proportional to the bank’s share of its financing, and weighing the carbon in the pile.
Calculation. Baltica: AF = 700 / 7,000 = 10%; attributed generation 3.0 TWh; attributed emissions 3.0 × 820 = 2,460 kt. Castilla: AF = 2,200 / 40,000 = 5.5%; 4.95 TWh; 1,237.5 kt. Rheinland: 20%; 3.6 TWh; 1,296 kt. Aegean: 20%; 2.0 TWh; 1,040 kt. Helios: 15%; 1.2 TWh; 6 kt. Nordfjord: 6.67%; 1.33 TWh; 10.7 kt. Total: 6,050 kt over 16.08 TWh.
INTERACTIVE LAB — Meridian’s power book in 2025: attributed emissions and intensity
Compute the PCAF-attributed emissions and generation of the six clients, the book intensity, and compare with an exposure-weighted average.
| Parameter you can move | Range | Default |
|---|---|---|
| Outstanding to Baltica Power | 0 to 2000 EUR m | 700 EUR m |
| Baltica EVIC | 3000 to 14000 EUR m | 7000 EUR m |
| Outstanding to Castilla Energía | 0 to 4000 EUR m | 2200 EUR m |
| Result at default settings | Value |
|---|---|
| Attributed emissions | 6,050 ktCO2e |
| Attributed generation | 16.08 TWh |
| Book intensity (attributed) | 376.2 kg/MWh |
| Exposure-weighted average of client intensities | 300.9 kg/MWh |
| Baltica share of attributed emissions | 40.7 % |
| Baltica share of exposure | 11.7 % |
| Attributed emissions: Baltica | 2,460.0 ktCO2e |
| Attributed emissions: Castilla | 1,237.5 ktCO2e |
| Attributed emissions: Rheinland | 1,296.0 ktCO2e |
| Attributed emissions: Aegean | 1,040.0 ktCO2e |
| Attributed emissions: Helios | 6.0 ktCO2e |
| Attributed emissions: Nordfjord | 10.7 ktCO2e |
| Castilla weight in exposure-weighted average | 36.7 % |
Try it. Double Baltica’s EVIC to EUR 14,000m, as a rally in its share price would: attributed emissions fall by 1,230 kt, yet the book intensity only falls from 376.2 to 330.5, because Baltica’s generation falls in the same proportion. Then set Baltica’s outstanding to zero.
Result. The book intensity is 376.2 kg/MWh on 6,050 kt of attributed emissions and 16.08 TWh of attributed generation. Baltica alone carries 40.7% of attributed emissions for 11.7% of the exposure.
Reading. Three lessons for the desk.
First, the weighting choice matters. A simple exposure-weighted average of the six clients’ intensities gives 300.9 kg/MWh, markedly lower. It gives Castilla, a large exposure to a large company, a weight of 36.7% when Meridian finances only 5.5% of it, and it underweights Rheinland and Aegean, where Meridian finances a fifth of the company. Both conventions exist in published bank targets; the attributed one is consistent with the PCAF inventory and with the absolute emissions disclosed, which is why Meridian uses it. A bank must state which one it uses and never switch silently.
Second, intensity is more robust to market noise than absolute emissions. If Baltica’s share price doubles its EVIC, attributed emissions fall by 1,230 kt (the EVIC effect discussed in A03), but attributed generation falls in the same proportion, and the book intensity moves (from 376.2 to 330.5 kg/MWh) only because Baltica’s weight in the book shrinks. The metric is not immune, but it is less exposed.
Third, a single client can dominate. With Baltica out of the book, intensity would be 274.4 kg/MWh, a fall of about a hundred kg/MWh at a stroke. Remember this number: it is the temptation section 5.6 is about.
PITFALL — Do not compute a book intensity by dividing attributed emissions by the euros lent (tCO2e per EUR m). That is an economic intensity: it falls when power prices and valuations rise, and it cannot be compared with any physical scenario. For power, the target metric is kg per MWh.
Question. Which 2030 intensity is consistent with a 1.5°C pathway for this book?
Data. The steering team uses a stylised net-zero power pathway (case assumption, shaped like published net-zero scenarios but not taken from IEA data): 440 kg/MWh in 2025, 190 in 2030, 70 in 2035, 15 in 2040, 0 in 2050.
FORMULA
is the target intensity of the book in year ; the baseline intensity (376.2 in 2025); the sector pathway intensity in year ; the base year. The fraction is the share of the sector’s remaining distance to its 2050 end point that is still left in year .
INTUITION — Picture runners on a track who must all reach the same finish line in 2050. The SDA asks each runner to cover, each year, the same proportion of its own remaining distance as the pack. A runner who starts closer to the line still has to move at the pack’s pace; it is not allowed to wait for the others.
Calculation. With , the fraction is . The target is kg/MWh, a reduction of 56.8% in five years, or 15.5% a year compounded.
INTERACTIVE LAB — From a sector pathway to Meridian’s 2030 target (SDA)
Apply the sectoral decarbonisation approach to Meridian’s baseline and compare the target path with the sector pathway and with a simple point target.
| Parameter you can move | Range | Default |
|---|---|---|
| Sector intensity 2025 | 300 to 600 kg/MWh | 440 kg/MWh |
| Sector intensity 2030 | 50 to 400 kg/MWh | 190 kg/MWh |
| Sector intensity 2050 | 0 to 50 kg/MWh | 0 kg/MWh |
| Baseline convention (stage 1) | attributed / exposure-weighted | attributed |
| Result at default settings | Value |
|---|---|
| SDA target 2030 | 162.4 kg/MWh |
| Required reduction 2025 to 2030 | 56.8 % |
| Compound annual change | -15.5 % a year |
| Reduction if the target were the sector’s 2030 point | 49.5 % |
| Share of the sector’s distance to 2050 left in 2030 | 43.2 % |
Try it. Switch the baseline to the exposure-weighted convention: the target falls to 129.9 kg/MWh, while a point target of 190 would then ask for only a 36.9% cut. Then set the 2050 intensity to 20: the target shifts slightly, because everyone now converges to a non-zero end point.
Result. Meridian’s 2030 target is 162.4 kg/MWh (−56.8%). Had the committee simply adopted the sector’s 2030 point (190 kg/MWh) as its target, the required cut would have been only 49.5%.
Reading. The SDA preserves Meridian’s head start. A book cleaner than the sector cannot use its starting advantage to slow down. This is why a “point target” taken from the scenario is a red flag when a bank starts below the sector average: it looks ambitious in level and is lenient in effort. Conversely, a bank above the sector would find the SDA target lower in effort than a point target would demand. Note that when the sector end point is zero, the SDA reduces to applying the sector’s percentage reduction to the bank’s baseline.
The committee also asks how the stylised pathway compares with what peers used. When HSBC set its power and utilities target in February 2022, on IEA’s net-zero pathway, it moved from 0.55 MtCO2e per TWh in 2019 to 0.14 in 2030, a 75% cut over eleven years, on-balance-sheet lending, scopes 1 and 2. Meridian’s −56.8% over five years is in the same family; the methodology note must cite the exact scenario edition before publication.
PITFALL — A target has a baseline year, a scenario edition and a perimeter. Change any of them and the “progress” changes. Rebasing to a newer scenario that is less steep, or redefining the perimeter to exclude a client, are the two classic ways in which a target is met on paper.
Question. If Meridian keeps every client at today’s exposure (renewing at maturity) and each client executes its credible plan, where does the book land in 2030?
Method. Same formula as stage 1, with 2030 intensities. Generation and EVIC are held constant: the purpose is to isolate the effect of clients becoming cleaner, what Roncalli (2026, section 11.3) calls self-decarbonisation of the portfolio, as opposed to decarbonisation obtained by rebalancing.
INTERACTIVE LAB — Business as usual: the book in 2030 if only clients change
Project the book intensity from the clients’ credible transition plans, keeping exposures constant, and measure the gap to the SDA target.
| Parameter you can move | Range | Default |
|---|---|---|
| Castilla 2030 intensity | 50 to 250 kg/MWh | 120 kg/MWh |
| Baltica 2030 intensity | 200 to 820 kg/MWh | 600 kg/MWh |
| Rheinland 2030 intensity | 100 to 360 kg/MWh | 330 kg/MWh |
| Aegean 2030 intensity | 100 to 520 kg/MWh | 300 kg/MWh |
| Sector pathway 2030 (2025 = 440) | 100 to 300 kg/MWh | 190 kg/MWh |
| Result at default settings | Value |
|---|---|
| Projected book intensity 2030 | 261.1 kg/MWh |
| Attributed emissions 2030 | 4,199 ktCO2e |
| Self-decarbonisation 2025 to 2030 | 30.6 % |
| Gap to target | 98.6 kg/MWh |
| Share of required reduction delivered by clients | 53.9 % |
Try it. Move Baltica from 600 to 400 kg/MWh (closing two more units): the book gains 37 kg/MWh. Moving Castilla from 120 to 60 gains less than you would expect, because Meridian finances only 5.5% of it.
Result. With clients’ plans alone the book reaches 261.1 kg/MWh in 2030, a self-decarbonisation of 30.6%. The gap to the target is 98.6 kg/MWh: clients deliver 53.9% of the required reduction.
Reading. This is the typical finding in practice: credible client plans take a power book a little over half way. The remaining gap is what the steering plan must close, and it is large. Roncalli’s distinction is exactly the one the committee must keep in mind: a portfolio can always be pushed onto a pathway by rebalancing if its assets are liquid, but that does not make it net zero; what counts is how much of the path is self-decarbonisation. Loans are not liquid, which makes the point sharper: the bank’s rebalancing happens at renewals, slowly, and each choice is visible.
The projection is also a credibility exercise. Baltica’s 600 kg/MWh assumes two coal units close by 2028; Aegean’s 300 assumes an interconnector commissioned in 2029. The PAC lens (Roncalli, section 9.3) applies client by client: past participation (has the client actually cut emissions over the past five years?), ambition (is its target below the sector path?), credibility (is the target consistent with its trend and capex?). The coverage team’s file on each client should answer these three questions explicitly.
WHO DOES WHAT — When you present a projection, show it client by client. A committee will accept a gap; it will not accept a number it cannot trace back to named clients and named plants.
Question. Which combination of levers closes the gap of 98.6 kg/MWh, and what does each lever do?
Data (case assumptions).
FORMULA
is the attributed generation of renewable project loans in MWh a year; the outstanding amount in EUR m; the load factor; the capex in EUR m per MW; 8,760 the hours in a year. The attribution factor (loan over project equity plus debt) cancels against the project value, so only the loan amount matters.
The use-of-proceeds point. Under the third edition of PCAF Part A, a use-of-proceeds structure is attributed the emissions and production of the assets it funds, not those of the borrower. For the EUR 350m of the Baltica facility documented as funding wind farms, Meridian attributes to itself the wind farms’ generation (EUR 350m × 1,752 MWh = 0.61 TWh at near zero), and Baltica’s corporate intensity applies only to the EUR 350m left as general corporate purpose. Whether a given loan qualifies depends on its documentation and allocation reporting (see B02); this is an assumption of the case.
INTERACTIVE LAB — Closing the gap: engagement, use of proceeds, renewables and exit
Combine the four steering levers and see their effect on the 2030 book intensity, attributed emissions and attributed generation, against the SDA target.
| Parameter you can move | Range | Default |
|---|---|---|
| Baltica 2030 intensity (600 plan, 400 if engaged) | 200 to 820 kg/MWh | 400 kg/MWh |
| Share of Baltica facility refinanced as use of proceeds (wind) | 0 to 100 % | 50 % |
| New renewable project finance | 0 to 4000 EUR m | 2000 EUR m |
| Exit Baltica at 2028 maturity | no / yes | no |
| Wind load factor | 20 to 45 % | 30 % |
| Wind capex | 1 to 2.5 EUR m per MW | 1.5 EUR m per MW |
| Sector pathway 2030 (2025 = 440) | 100 to 300 kg/MWh | 190 kg/MWh |
| Result at default settings | Value |
|---|---|
| Book intensity 2030 | 160.4 kg/MWh |
| Attributed emissions 2030 | 2,999 ktCO2e |
| Attributed generation 2030 | 18.70 TWh |
| Distance to target (negative = met) | -2.1 kg/MWh |
| Generation per EUR m of wind project | 1,752 MWh a year |
| Attributed generation of the use-of-proceeds tranche | 0.61 TWh |
| Attributed generation of new project finance | 3.50 TWh |
Try it. Set the use-of-proceeds share to 100% with EUR 1,500m of renewables: the book reaches 141.6 kg/MWh while Baltica still burns coal, because none of its corporate emissions is attributed any more. Compare with exit.
Results. Lever by lever, starting from 261.1 kg/MWh:
| Strategy | Book intensity 2030 | Attributed emissions | Target met? |
|---|---|---|---|
| Clients’ plans only | 261.1 | 4,199 kt | No |
| + Baltica engaged (400) | 223.8 | 3,599 kt | No |
| + EUR 1.5bn wind PF | 192.3 | 3,599 kt | No |
| Plan M: engage, 50% use of proceeds, EUR 2bn PF | 160.4 | 2,999 kt | Yes |
| Exit Baltica + EUR 1.5bn PF | 152.7 | 2,399 kt | Yes |
Reading. Each lever works through a different part of the ratio. Engagement lowers the numerator through a real change at a client. Renewable project finance raises the denominator: it adds clean megawatt-hours, so it dilutes the intensity without reducing any emission; the EUR 1.5bn of wind loans adds 2.63 TWh at near zero and takes 31 kg/MWh off the book. Use of proceeds moves part of the Baltica exposure from the numerator-heavy corporate bucket to a clean asset bucket. Exit removes Baltica’s numerator and denominator altogether.
Plan M reaches 160.4 kg/MWh, 2.1 kg/MWh below the target, with EUR 2bn of new renewables origination, a renegotiated Baltica facility and a client that closes more coal than it planned. Exit plus EUR 1.5bn of renewables reaches 152.7 kg/MWh with less origination and no negotiation. On the target metric alone, exit is the better plan. The next stage explains why the committee should not choose on that metric alone.
PITFALL — Origination of renewables is necessary and welcome; it is also the easiest way to “dilute” an intensity metric. A bank that grows renewables while keeping every coal client unchanged can meet an intensity target with its absolute financed emissions flat. That is why sector intensity targets are usually paired with an absolute or technology-mix indicator, and why an analyst always reports the numerator and the denominator separately.
Question. What do Plan M and the exit strategy do to emissions in the real economy, and how much of each plan’s book improvement is real?
Method. Two measures. First, Baltica’s own emissions in 2030, which is what the atmosphere sees. Second, a variance decomposition of the book’s change between 2025 and 2030 into a client effect (the change that would occur with 2025 exposures and 2030 client intensities) and a portfolio effect (the rest: mix changes, use of proceeds, new origination, exit). The client effect is the measurable part attributable to clients’ decarbonisation.
FORMULA
is the book intensity computed with attribution weights (outstanding amounts, use-of-proceeds split, new loans) and client intensities ; subscripts are years. The client effect holds the book fixed; the portfolio effect holds clients fixed at their 2030 intensity.
INTERACTIVE LAB — Exit or engage: the book versus the atmosphere
Compare what exit and engagement do to Meridian’s metric and to Baltica’s real emissions, decompose the book change into client and portfolio effects, and see where facilitated emissions sit.
| Parameter you can move | Range | Default |
|---|---|---|
| Baltica intensity 2030 under its own plan | 400 to 820 kg/MWh | 600 kg/MWh |
| Baltica intensity 2030 if engaged and refinanced | 200 to 820 kg/MWh | 400 kg/MWh |
| Baltica bond issue (2027) | 0 to 1500 EUR m | 500 EUR m |
| Number of equal bookrunners | 1 to 8 | 4 |
| PCAF facilitation weighting | 0 to 100 % | 33 % |
| Result at default settings | Value |
|---|---|
| Baltica real emissions 2030 if Meridian exits | 18.0 MtCO2e |
| Baltica real emissions 2030 if engaged | 12.0 MtCO2e |
| Extra real reduction from engagement | 6.0 MtCO2e |
| Exit plan: client effect on book intensity | -115.1 kg/MWh |
| Exit plan: portfolio effect | -108.4 kg/MWh |
| Exit plan: share of book change that is client effect | 51.5 % |
| Plan M: client effect | -152.4 kg/MWh |
| Plan M: portfolio effect | -63.4 kg/MWh |
| Plan M: share of book change that is client effect | 70.6 % |
| Facilitated emissions of the bond mandate (2025 data) | 145.0 ktCO2e |
Try it. Set the engaged intensity equal to the plan (600): engagement brings nothing real, and Plan M’s advantage over exit disappears. The case for engagement rests entirely on the credibility of that extra closure.
Results.
Reading. This is the attribution trap in numbers. The target metric ranks exit first (152.7 against 160.4). A real-economy view ranks it last: it buys a better-looking book with no emission reduction, and possibly worse outcomes if Baltica’s new lenders impose no conditions. The decomposition gives the committee a way to see it on one line, and it is what a supervisor or an informed investor will ask for.
Two caveats keep the analysis honest. First, the “6.0 Mt” is a claim about causality: engagement only matters if the lender’s conditions actually change the client’s decisions. A bank that is one of fifteen lenders to Baltica has little leverage; a house bank refinancing a large share of its debt has a lot. The steering team should write down why it believes the engagement is causal (milestones, covenants, use-of-proceeds reporting) and test it at the next renewal. Second, use of proceeds is itself a perimeter effect: in Plan M, half of the Baltica exposure is attributed wind-farm emissions while the money is fungible inside Baltica. That is what the PCAF method prescribes for use-of-proceeds structures, and it is why the third edition requires such loans to be disclosed with their method; a steering team that pushed the use-of-proceeds share to 100% would obtain 141.6 kg/MWh with Baltica still burning coal, which is exit by another name.
Facilitated emissions. If Meridian exits the loan but remains one of four bookrunners on a EUR 500m Baltica bond in 2027, PCAF Part B attributes to it 500 / 4 / 7,000 × 33% of Baltica’s emissions, 145.0 kt on 2025 data (439.3 kt unweighted). That figure is reported separately, once, in the year of issuance, and it sits outside the lending target. A bank can therefore move a client from its balance sheet to its capital-markets franchise and see its target metric improve while its relationship, its fees and its role in financing the client continue. The red-flag list in section 7 includes this.
WHY IT MATTERS ON THE DESK — The trap is not a theoretical worry. Since 2021 NGOs and academics have pointed out that financed-emissions targets can be met by divestment, and the SBTi’s FINZ and the EBA guidelines both put weight on transition planning with clients rather than on portfolio exclusion alone. A steering analyst who reports the client effect next to the headline number protects the bank from that criticism.
Question. How would an alignment tool score the book under each strategy, and what does the score add?
Method. Roncalli (2026, section 9.3) defines the carbon budget of a trajectory as the area under it, approximated by a sum of yearly values. A transparent temperature mapping interpolates between two reference trajectories: a 1.5°C-consistent path (Meridian’s SDA path from 376.2 to 162.4) and a current-policies path (case assumption: the book falls only 25% by 2030, to 282.1 kg/MWh, a path assumed to be consistent with 2.7°C). Each path is linear between 2025 and 2030.
FORMULA
is the intensity budget of path (kg/MWh-years, left Riemann sum over the five years); the book’s projected path; the SDA path; the current-policies path; the temperature assumed for the current-policies path (2.7°C); the implied temperature of the book.
INTERACTIVE LAB — Implied temperature of the book by strategy
Compute intensity budgets 2025 to 2030 for each strategy and map them to an implied temperature by interpolating between a 1.5°C path and a current-policies path.
| Parameter you can move | Range | Default |
|---|---|---|
| Strategy | Clients only / Engage / Plan M / Exit + PF | Clients only |
| Reduction 2025 to 2030 on the current-policies path | 5 to 45 % | 25 % |
| Temperature of the current-policies path | 2 to 3.5 °C | 2.7 °C |
| Result at default settings | Value |
|---|---|
| Book intensity 2030 for the strategy | 261.1 kg/MWh |
| Intensity budget of the book 2025-2029 | 1,651 kg/MWh-years |
| Budget of the 1.5°C (SDA) path | 1,453 kg/MWh-years |
| Budget of the current-policies path | 1,693 kg/MWh-years |
| Implied temperature rise | 2.49 °C |
| Current-policies path, 2030 end point | 282.1 kg/MWh |
Try it. Change the current-policies path from a 25% to a 15% reduction: every score moves towards 1.5°C, although nothing in the book has changed. Temperature scores are as much a statement about the reference paths as about the portfolio.
Results. Implied temperatures: clients’ plans only 2.49°C; engagement alone 2.12°C; Plan M 1.48°C; exit plus renewables 1.40°C.
Reading. The temperature score is a communication device built on two scenario choices and a budget horizon. It ranks strategies in exactly the same order as the 2030 intensity, because with linear paths the budget is a fixed combination of the start and end points; it adds no information about the real economy, and it inherits the attribution trap: exit scores best. Changing the reference current-policies path from −25% to −15% moves every score towards 1.5°C without any change in the book. Roncalli notes that temperature ratings are often perceived as black boxes; in practice, scores published by different providers for the same portfolio differ widely, which is why supervisors and the SBTi treat them as one tool among several. A PACTA-style technology-mix view would add something the intensity does not: how much coal and gas capacity Meridian finances and whether new financing goes to low-carbon capacity at the pace of the scenario.
PITFALL — Never present an implied temperature without its reference paths, horizon and scope. A “1.48°C book” is meaningless on its own; “1.48°C on a 2025-2030 intensity budget, scope 1, power generation only, against a 2.7°C current-policies reference” is a statement a committee can challenge.
The Group Climate Committee adopts Plan M with three conditions. First, the Baltica refinancing in 2028 is conditional on the board-approved closure of the two additional coal units, with a milestone in the facility agreement and a use-of-proceeds allocation report each year. Second, the project finance team receives an origination budget of EUR 2bn of renewables over four years, subject to the usual credit and RAROC criteria. Third, the quarterly dashboard reports, next to the headline intensity, the absolute attributed emissions, the attributed generation and the client-effect share of the change. Exit remains the fallback: if Baltica misses its closure milestone in 2028, Meridian will not renew.
What the quarterly dashboard shows. The committee asked for one page per sector. For power it has five lines, each with the 2025 baseline, the latest value and the trajectory value for the date: book intensity (376.2 at baseline against a pathway that reaches 162.4 in 2030); attributed emissions (6,050 kt at baseline); attributed generation (16.08 TWh); the client-effect share of the cumulative change; and the use-of-proceeds and project finance exposures with their attributed generation. Below the lines, a client table lists the six groups with their latest intensity, their plan for 2030 and a credibility flag from the coverage team. The page is deliberately boring: the committee should be able to see in thirty seconds whether the book is moving because clients move or because Meridian moved.
The 2028 checkpoint. The plan has a single decisive date: the refinancing of Baltica’s EUR 700m facility. Three situations are prepared in advance. If Baltica has closed the units or is visibly on schedule, Meridian refinances half general purpose and half use of proceeds, with the milestone carried into the new documentation. If Baltica is late but has a credible revised schedule and the units are under a legally binding closure decision, the committee may grant a one-year extension with a smaller commitment. If Baltica has abandoned the plan, Meridian does not renew, discloses the exit as a consequence of failed engagement, and reviews whether to keep its bookrunner role, whose facilitated emissions would otherwise replace the financed ones in the bank’s overall picture. Writing these three branches before the event is what turns engagement from an open-ended conversation into a policy that a supervisor can test.
The technology-mix cross-check. Climate Risk also runs a PACTA-style view. Instead of kg per MWh, it compares the share of coal, gas and renewable capacity in the capacity Meridian finances with the capacity mix of the scenario, plant by plant from asset-level data. The two views usually agree on the direction; when they diverge, the divergence is informative. A book can improve its intensity through gas plants running more hours while coal plants stand idle in a windy year, without any change in installed capacity; the technology mix would show no progress. Conversely, a client that builds wind farms whose output is curtailed shows capacity progress before intensity progress. The committee reads both, and the methodology note says which one is the target and which one is the control.
| Figure | Value | Stage |
|---|---|---|
| Book exposure | EUR 6,000m | 5.1 |
| Steerable share by end-2030 | 57.1% | 2.2 |
| Attributed emissions 2025 | 6,050 kt | 5.2 |
| Attributed generation 2025 | 16.08 TWh | 5.2 |
| Book intensity 2025 (attributed) | 376.2 kg/MWh | 5.2 |
| Exposure-weighted intensity | 300.9 kg/MWh | 5.2 |
| Baltica share of emissions / exposure | 40.7% / 11.7% | 5.2 |
| SDA target 2030 | 162.4 kg/MWh (−56.8%) | 5.3 |
| Compound annual reduction | −15.5% a year | 5.3 |
| Clients’ plans only, 2030 | 261.1 kg/MWh (self-decarbonisation 30.6%) | 5.4 |
| Share of required reduction from clients | 53.9% | 5.4 |
| Plan M, 2030 | 160.4 kg/MWh | 5.5 |
| Exit plus EUR 1.5bn PF, 2030 | 152.7 kg/MWh | 5.5 |
| Extra real reduction from engagement | 6.0 MtCO2e a year | 5.6 |
| Client-effect share, exit / Plan M | 51.5% / 70.6% | 5.6 |
| Facilitated emissions, Baltica bond | 145.0 kt | 5.6 |
| ITR: clients only / Plan M / exit | 2.49 / 1.48 / 1.40°C | 5.7 |
The steering decision changes the composition of the power book, and therefore its capital. Under F-IRB, the CRR3 corporate formula with a 40% LGD and a 2.5-year maturity gives a risk weight of 82.1% for a BB borrower (PD 1.00%) such as Baltica and 39.0% for a BBB borrower (PD 0.20%). Replacing EUR 700m of Baltica with EUR 700m of BBB lending roughly halves the RWA of that slice. Renewable project finance follows a different route: banks without approved PD models for specialised lending use the supervisory slotting approach (see B06), and CRR3 keeps a preferential treatment for high-quality project finance in the operational phase. The capital effect of Plan M is therefore not a simple swap; it depends on how the new renewables loans are rated or slotted.
Climate does not enter the Pillar 1 formula directly. It enters through the PD (a coal generator’s rating already reflects carbon costs under the EU ETS, see E01), through Pillar 2 (the ICAAP must capture material climate risks, and the supervisor can add a Pillar 2 requirement), and through concentration: a book where one coal client carries 40% of attributed emissions is also a concentrated transition-risk exposure.
Question. What does Meridian give up if it exits Baltica and redeploys the EUR 700m into BBB renewables lending, compared with Plan M’s restructuring?
Assumptions (case). Baltica: margin over FTP 300 bp, fees 15 bp a year, ancillary revenue EUR 6.0m a year (EU ETS allowance hedging, power and FX hedging, cash management), BB, PD 1.00%. Replacement: Helios-like BBB renewables lending, margin over FTP 120 bp, fees 10 bp, ancillary EUR 1.5m, PD 0.20%. Under Plan M, the use-of-proceeds half of the Baltica facility is priced 10 bp below the general-purpose half. Cost-income 40% on lending revenue and 60% on ancillary revenue; ancillary revenue carries EUR 2.5 of RWA per euro; capital 13% of RWA; tax 25%.
INTERACTIVE LAB — RAROC of exit, replacement and Plan M for the Baltica slice
Compute the RAROC of the EUR 700m Baltica exposure, of a BBB renewables replacement and of the Plan M restructuring, and the cost per tonne of attributed emissions removed by exit.
| Parameter you can move | Range | Default |
|---|---|---|
| Exposure | 100 to 1500 EUR m | 700 EUR m |
| Baltica margin over FTP | 100 to 500 bp | 300 bp |
| Baltica ancillary revenue | 0 to 15 EUR m | 6 EUR m |
| Replacement margin over FTP | 50 to 300 bp | 120 bp |
| Replacement ancillary revenue | 0 to 10 EUR m | 1.5 EUR m |
| Margin discount on the use-of-proceeds half (Plan M) | 0 to 50 bp | 10 bp |
| Attributed emissions of Baltica in 2030 | 500 to 3000 ktCO2e | 1800 ktCO2e |
| Result at default settings | Value |
|---|---|
| Baltica risk weight (F-IRB) | 82.1 % |
| Baltica RWA | 589.4 EUR m |
| Baltica revenues | 28.05 EUR m |
| Baltica operating costs | 12.42 EUR m |
| Baltica lending revenue (margin and fees) | 22.05 EUR m |
| Baltica expected loss | 2.80 EUR m |
| Baltica allocated capital | 76.6 EUR m |
| Baltica net profit after tax | 9.623 EUR m |
| Baltica RAROC | 12.56 % |
| Replacement RWA | 276.9 EUR m |
| Replacement net profit after tax | 4.125 EUR m |
| Replacement RAROC | 11.46 % |
| Plan M restructured Baltica RAROC | 12.35 % |
| Profit given up by exit (before redeploying capital) | 5.498 EUR m |
| Capital freed by exit | 40.63 EUR m |
| Profit given up after redeploying freed capital at 12% | 0.622 EUR m |
| Profit given up per tonne removed from the book | 3.05 EUR per t |
Try it. Set Baltica’s ancillary revenue to zero: its RAROC falls below the 12% hurdle and exit looks free. Coverage bankers defend difficult clients with the ancillary line; the steering team should ask for it in writing.
Line by line for Baltica. Lending revenue: EUR 700m × (300 + 15) bp = EUR 22.05m; ancillary EUR 6.0m; total revenues EUR 28.05m. Operating costs: 40% × 22.05 + 60% × 6.0 = EUR 12.42m. Expected loss: 1.00% × 40% × 700 = EUR 2.80m. RWA: 82.06% × 700 + 2.5 × 6.0 = EUR 589.4m; capital 13% = EUR 76.6m. Net profit after tax: (28.05 − 12.42 − 2.80) × 75% = EUR 9.62m. RAROC 12.56%, just above the 12% hurdle.
Replacement. RWA EUR 276.9m, net profit EUR 4.13m, RAROC 11.46%: thinner margins on investment-grade renewables lending make it slightly below hurdle on a stand-alone basis, a familiar finding (see B06 and E03).
Comparison. Exit gives up EUR 5.50m of after-tax profit a year and frees EUR 40.6m of capital. If the freed capital is redeployed at the 12% hurdle, the net give-up falls to EUR 0.62m. Per tonne of attributed emissions removed from the book (1,800 kt in 2030), the gross cost is EUR 3.05 per tonne: far below any carbon price. Exit is cheap for the bank, which is exactly why the metric’s bias towards exit is dangerous. Plan M’s 10 bp concession on the use-of-proceeds half lowers Baltica’s RAROC only from 12.56% to 12.35%: the sustainable feature costs little; what costs effort is the negotiation and the monitoring.
WHO DOES WHAT — When a committee member says “let’s just exit”, the steering analyst should answer with two numbers: what it costs the bank (here EUR 0.6m to 5.5m a year) and what it does to real emissions (here nothing). The first number is small; the second is the reason not to do it lightly.
All margins above are quoted over funds transfer pricing. For a five-year corporate facility Treasury charges the swap rate plus a liquidity premium of about 30 bp (see C15). The renewables loans of Plan M are longer: a fifteen-year project finance loan, even with mini-perm structures, carries a much higher term liquidity premium (of the order of 90 bp on Meridian’s 6 bp per year scale) and consumes stable funding under the NSFR. Steering the book towards renewables therefore lengthens the bank’s assets and raises its funding cost, which must be priced into the origination budget. The partial offset is that renewables loans are eligible collateral for the bank’s own green bonds (see C15): if Meridian can fund them with green issuance, the small greenium on its own bonds reduces the cost, though not by enough to change the decision (the Helios case in C01 puts a greenium of the order of a few basis points in proportion).
Under IFRS 9 the loans are measured at amortised cost if they pass the solely-payments-of-principal-and-interest (SPPI) test. The IASB’s May 2024 amendments, effective for annual periods beginning on or after 1 January 2026, clarify how contractual features linked to ESG targets are assessed under that test; a use-of-proceeds loan with a fixed margin raises no SPPI question, while a margin linked to Baltica’s closure milestone would need the clarified analysis (see B03). Expected credit losses for Baltica should reflect forward-looking transition risk: a coal generator facing rising carbon costs may need a stage 2 classification or a management overlay, with the climate risk team providing the scenario input (see E01). Nothing in the steering decision changes the accounting of the loans already on the books; an exit at maturity is a simple repayment, and a sale before maturity would create a gain or loss on derecognition.
A steering framework fails in three ways: the number is wrong, the number is right but means something else than the bank says, or the number is right and meaningful but the bank does not act on it. The controls below address each.
| Risk | Control | Owner |
|---|---|---|
| Wrong or stale client data | Data cut-off, quality scores, four-eyes on top 20 clients | ESG data office |
| Inconsistent weighting or perimeter | Methodology note approved; changes disclosed | Climate risk (second line) |
| Silent rebasing to an easier scenario | Rebasing policy; committee approval; restatement | Portfolio steering |
| Target met by exit, not by clients | Client-effect share reported next to the headline | Portfolio steering |
| Dilution by growth in renewables | Absolute emissions and fossil capacity reported too | Sustainability reporting |
| Use-of-proceeds misclassification | Allocation reports; documentation check | Loan agency, credit |
| Shift from lending to facilitation | Facilitated emissions reported by client | DCM business management |
| Targets not used in credit decisions | Trajectory section in credit memos; sample review | Credit risk, internal audit |
| Over-claiming in public statements | Legal and compliance review of disclosures | Compliance (see E10) |
| Supervisory finding on the transition plan | Gap analysis against EBA guidelines | Group risk |
Red-flag checklist for a bank’s power-sector target.
PITFALL — A target that is on track is not proof of a credible plan. A supervisor looking at the transition plan will ask how the bank got there, which clients moved, and what happens at the next renewal of the largest emitters.
Dated facts, verified in November 2026.
Orders of magnitude.
What changed recently.
Open debates.
INTERACTIVE LAB — Financed emissions and net-zero steering: self-assessment
Twelve questions on book intensity, SDA targets, steering levers, the attribution trap and the current status of standards and alliances.
Q1. Meridian’s power book intensity is computed as:
Answer: B. The physical intensity applies the same attribution factor to emissions and generation; an economic intensity per euro cannot be compared with a physical scenario.
Q2. Why is the exposure-weighted average (300.9) lower than the attributed intensity (376.2)?
Answer: C. Exposure weighting ignores how much of each client Meridian finances; attribution weights by the share of each client’s output.
Q3. Under the sectoral decarbonisation approach with a sector end point of zero, the bank’s target equals:
Answer: B. With S(2050) = 0 the SDA fraction is S(2030)/S(2025), applied to the bank’s own baseline.
Q4. A bank starting below the sector average adopts the sector’s 2030 point as target. This is:
Answer: C. Meridian would need a 49.5% cut instead of 56.8%: the point target lets a cleaner-than-average book slow down.
Q5. Adding EUR 1.5bn of wind project loans lowers the book intensity mainly by:
Answer: B. Renewables origination dilutes the ratio through the denominator; it removes no emissions from the atmosphere by itself.
Q6. If Meridian exits Baltica, what happens to Baltica’s real emissions in the case?
Answer: C. Exit changes who finances Baltica, not what Baltica emits; that is the attribution trap.
Q7. In the variance decomposition, the client effect is:
Answer: A. Holding the book fixed isolates clients’ own decarbonisation; the rest is the portfolio effect.
Q8. Under PCAF Part B, the facilitated emissions of a bookrunner are weighted at:
Answer: C. Part B (December 2023) sets a 33% weighting, with optional separate disclosure of unweighted figures.
Q9. The third edition of PCAF Part A (December 2025) added methods for:
Answer: B. Existing methods were unchanged; four additions were made, with supplemental guidance on avoided emissions and forward-looking metrics.
Q10. What is the status of the Net-Zero Banking Alliance in November 2026?
Answer: B. Members voted to stop operating as an alliance after a wave of departures; the announcement was made on 3 October 2025.
Q11. In the EU, which instrument now requires banks to have transition plans with quantifiable targets?
Answer: B. Prudential transition plans are required by CRD VI and specified by EBA/GL/2025/01, applicable to large institutions from 11 January 2026.
Q12. In the case, the gross profit given up per tonne removed from the book by exiting Baltica is about:
Answer: A. EUR 5.50m of after-tax profit for 1,800 kt of attributed emissions: about EUR 3.05 per tonne, far below any carbon price, which is why the metric’s bias towards exit is tempting.
Q13. An implied temperature score computed by interpolating budgets between two reference paths:
Answer: C. The score is a transformation of the budget gap; it inherits the attribution trap and depends on the chosen references.
A bank holds three power clients. Client X: outstanding EUR 400m, EVIC EUR 4,000m, generation 20 TWh at 700 kg/MWh. Client Y: EUR 1,000m, EVIC EUR 20,000m, 60 TWh at 200 kg/MWh. Client Z: EUR 600m, EVIC EUR 3,000m, 6 TWh at 10 kg/MWh. (a) Compute the attributed intensity and the exposure-weighted intensity. (b) With a sector pathway of 450 kg/MWh in 2025 and 200 in 2030 and an end point of zero, compute the SDA target and the required reduction.
Solution. (a) Attribution factors: X 10%, Y 5%, Z 20%. Attributed generation: 2.0 + 3.0 + 1.2 = 6.2 TWh. Attributed emissions: 2.0 × 700 = 1,400 kt; 3.0 × 200 = 600 kt; 1.2 × 10 = 12 kt; total 2,012 kt. Attributed intensity: 2,012 / 6.2 = 324.5 kg/MWh. Exposure-weighted: (400 × 700 + 1,000 × 200 + 600 × 10) / 2,000 = 486,000 / 2,000 = 243.0 kg/MWh. (b) SDA fraction 200 / 450 = 44.4%; target 324.5 × 0.444 = 144.2 kg/MWh; required reduction 55.6%. Client X carries 69.6% of attributed emissions for 20% of the exposure; the target cannot be met without a change at X.
Continue with the book of exercise 1, keeping 2025 intensities. (a) How much solar project finance (capex EUR 0.8m per MW, load factor 15%) must the bank add to reach the target by dilution only? (b) What happens to absolute attributed emissions? (c) Comment.
Solution. (a) Generation per EUR 1m of project: 8,760 × 15% / 0.8 = 1,642.5 MWh. To reach 144.2 kg/MWh with 2,012 kt, attributed generation must be 2,012 / 144.2 = 13.95 TWh, i.e. 7.75 TWh more. Required lending: 7.75 TWh / 1,642.5 MWh per EUR m = EUR 4,720m, more than twice the existing book. (b) Unchanged at 2,012 kt. (c) Dilution alone is both implausible in scale (a bank cannot more than triple its sector book in five years within its limits) and empty in substance: no emission falls. The exercise shows why the SDA intensity target must be read with absolute emissions and with the client effect.
A bank holds a EUR 500m loan to a BB coal generator (PD 1.00%, LGD 40%, M 2.5, risk weight 82.06%), margin over FTP 260 bp, fees 10 bp, ancillary revenue EUR 3.0m. The replacement is BBB (PD 0.20%, risk weight 39.02%), margin 110 bp, fees 10 bp, ancillary EUR 1.0m. Cost-income 40% on lending, 60% on ancillary; ancillary RWA EUR 2.5 per euro; capital 13%; tax 25%. Attributed emissions of the coal loan: 1,200 kt. (a) Compute both RAROCs. (b) Compute the profit given up per tonne.
Solution. (a) Coal loan: lending revenue 500 × 2.70% = 13.50; total revenue 16.50; costs 0.4 × 13.50 + 0.6 × 3.0 = 7.20; EL = 1% × 40% × 500 = 2.00; net profit (16.50 − 7.20 − 2.00) × 0.75 = 5.475; RWA 0.8206 × 500 + 7.5 = 417.8; capital 54.3; RAROC 10.08%. Replacement: lending revenue 500 × 1.20% = 6.00; total 7.00; costs 2.40 + 0.60 = 3.00; EL 0.40; net profit 2.70; RWA 0.3902 × 500 + 2.5 = 197.6; capital 25.7; RAROC 10.51%. (b) Profit given up: 5.475 − 2.70 = EUR 2.775m; per tonne: 2,775,000 / 1,200,000 = EUR 2.31. Here the coal loan is below hurdle and less profitable on capital than the replacement: exit improves both the metric and the RAROC, so the business case for engagement must rest on the client relationship and on real-economy impact, not on returns.
1. What is the difference between financed and facilitated emissions, and why does it matter for a bank’s targets? Financed emissions are the share of a client’s emissions attributed to the bank’s own on-balance-sheet lending and investments, measured as a stock at year-end under PCAF Part A. Facilitated emissions are attributed to the bank’s role in arranging bond and equity issues and syndications, measured once in the year of the deal with a 33% weighting under Part B. Most published lending targets cover only financed emissions. A bank can therefore exit a loan and remain a bookrunner for the same client, improving its target metric while its role in financing the client continues. A serious disclosure shows both, by client and by sector.
2. Walk me through how you would set a 2030 power-sector target. I would start with the perimeter (power generation, on-balance-sheet lending), the metric (kgCO2e per MWh on attributed emissions and generation), and a baseline from the latest audited inventory. I would choose a reference scenario, in practice the latest IEA net-zero pathway, and apply the sectoral decarbonisation approach so that the book closes its gap to the sector’s 2050 end point at the sector’s pace. Then I would check feasibility: run-off of the book, clients’ credible plans, and the levers left. Finally I would document the methodology, the rebasing policy and the governance so that the second line can validate it.
3. Your book intensity fell by 30% in a year. What questions do you ask? First, how much of the fall is client effect and how much portfolio effect: did clients become cleaner, or did the book change? Second, did any large emitter leave the book, and if so, was it engaged before? Third, did the perimeter, the scenario or the data change, for instance a client’s EVIC or reporting boundary? Fourth, what happened to absolute emissions and to facilitated emissions for the same clients? Only after these questions would I present the fall as progress.
4. Is exiting a coal client good climate policy for a bank? It depends on what follows. If the client refinances easily elsewhere, exit changes the lender, not the emissions, and the bank loses the leverage of renewal conditions. If the bank is a significant lender and uses that position to obtain time-bound commitments, such as plant closures backed by milestones, engagement can change real emissions. Exit is the right answer when engagement has failed or the client has no credible plan, and it should then be disclosed as such. A good framework makes exit the consequence of failed engagement, not the default.
5. Why might a bank prefer an intensity target to an absolute target for power? Electricity demand must grow with electrification, so the sector needs more clean generation, not less generation. An intensity target lets the bank finance that growth while asking each financed MWh to be cleaner. An absolute target would penalise a bank for financing new renewables attached to utilities that still own fossil plants. The weakness is dilution: intensity can fall while absolute emissions stay flat, so the bank should report both.
6. What does the end of the Net-Zero Banking Alliance change for a European bank? It removes a voluntary commitment framework and its reporting deadlines, but not the substance. In the EU, CRD VI requires plans with quantifiable targets to address ESG risks including the transition to 2050, and the EBA guidelines, applicable since January 2026 for large institutions, specify metrics, targets and governance. The ECB reviews these plans and has shown it will use penalties. The sustainability statement still discloses targets. What changes is the label and the peer pressure; the supervisory and disclosure pressure remains.
7. How does the PCAF third edition affect transition finance to high emitters? It introduces a method for use-of-proceeds structures that attributes to the bank the emissions of the funded assets rather than the borrower’s whole footprint. A bank can therefore finance a coal generator’s wind farms without importing the generator’s intensity into its book for that tranche. That supports transition finance, but it can also be used to make a book look cleaner while the money is fungible. Good practice is to report use-of-proceeds exposures separately, check allocation reports, and keep the borrower’s corporate emissions visible in the client analysis.
8. How do you price the cost of a steering decision? I compute the RAROC of the exposure the bank would give up and of its replacement, with all revenues including ancillary business, the expected loss, operating costs and allocated capital. The difference in after-tax profit, adjusted for the capital freed and redeployed at the hurdle, is the cost. I then express it per tonne of attributed emissions removed, which is usually a few euros. The low figure is a warning, not a comfort: it means the metric can be improved cheaply without real effect, so the decision must also be judged on its real-economy consequence.
Standards and guidance
pacta.loanbook documentation: https://rmi-pacta.github.io/pacta.loanbook/articles/cookbook_advanced_use_cases.htmlRegulation and supervision
Scenarios and market data
Academic and textbook references