Every sustainable product in this course rests on a few numbers about the client: tonnes of greenhouse gases, an intensity, a rating, a taxonomy percentage. A sustainability-linked margin is tested on them (see B03), a green bond framework quotes them (see C01), the bank’s own Pillar 3 tables add them up (see E06). This module is about where those numbers come from, who types them into which system, and how wrong they can be.
Where it sits. At Meridian Bank the work is split between three places. In the CIB, an ESG data and analytics team of about fifteen people sits next to the sustainable finance structurers; it builds and maintains the “carbon file” of each large corporate client and answers questions from coverage, credit and DCM. In Risk, the ESG risk team owns the internal transition score and the sector policies. In Finance, sustainability reporting runs the year-end inventory of financed emissions and produces the bank’s sustainability statement and Pillar 3 ESG tables. The Chief Data Office sets the rules that apply to all three: data dictionary, lineage, controls.
Clients. The team has no external client. Its customers are internal: the coverage banker who needs a one-page carbon profile before a pitch, the credit analyst who must fill the ESG section of a credit memo, the DCM originator who needs the issuer’s intensity against peers, the structurer who needs a baseline for a KPI, and Finance, which needs one attributable emissions figure per exposure on 31 December.
How it earns money. It does not; it is a cost centre. Its budget is made of people, vendor licences (ESG ratings, emissions data, taxonomy data) and IT. The cost is recharged to the business lines, usually per client or per unit of exposure. In the case of this module we assume a recharge of EUR 15,000 a year for a large corporate carbon file; section 6 shows what that does to the RAROC of a loan. The value the team creates is indirect and of three kinds: revenue that would not exist without credible data (labelled loans and bonds), regulatory disclosures that would be qualified by the auditor or challenged by the supervisor without it, and credit decisions that price transition risk.
The team and its neighbours.
| Role | What it produces | System |
|---|---|---|
| ESG data analyst | Carbon file, data quality score, peer comparison | ESG data platform |
| Coverage banker | Client questionnaire, access to the client’s sustainability team | CRM |
| Credit analyst | Spread financials, ESG section of the credit memo | Spreading and rating tool |
| ESG risk officer | Transition score, sector policy opinion | Risk rating system |
| Finance sustainability reporting | Financed emissions inventory, Pillar 3 templates | Finance data warehouse |
| DCM business management | List of deals and bookrunner shares for facilitated emissions | Deal database |
| Data steward (CDO) | Data dictionary, lineage, control results | Data governance tool |
Model validation and internal audit review the methods; the statutory auditor gives assurance on the published figures.
A realistic week in November 2026. Monday: the team loads the half-yearly vendor feed and reviews the 140 exceptions where a vendor figure differs from the bank’s own extraction by more than 10%. Tuesday: an analyst builds the carbon file of a cement group for its annual credit review on Thursday (this is the case of section 5). Wednesday: a call with a DCM originator who wants to tell investors that an issuer is “top quartile on carbon”; the analyst explains that this is true on revenue intensity and false on physical intensity. Thursday: credit committee; the ESG risk officer presents the transition score. Friday: dry run of the year-end financed emissions inventory with Finance, six weeks before the reference date of 31 December 2026, the first expected under the revised Pillar 3 ESG templates.
WHY IT MATTERS ON THE DESK — A banker who cannot read a carbon figure cannot defend a sustainability-linked target in front of a credit committee, cannot answer an investor’s question on a roadshow, and will sign compliance certificates without understanding them. The data team produces the numbers, and the front office remains accountable for using them.
The “product” of this module is a data file. Before any sustainability content, a bank already builds a data file on every borrower; the carbon file is a second ledger kept with the same discipline. This section sets out the financial file, then the greenhouse gas inventory as an accounting object, without any reference yet to financed emissions or ratings.
When a corporate client is onboarded, the credit analyst spreads its financial statements: each line of the balance sheet, income statement and cash-flow statement is keyed or imported into a standard template, so that EBITDA or net debt mean the same thing for every client. From the spread the rating model produces a probability of default (PD); the facility data give the exposure at default (EAD) and the loss given default (LGD). Three properties of this file matter for what follows.
First, every figure has a source and a date. The spread records the document (annual report 2025, page 187), the reporting date and the auditor’s opinion. A figure without a source is not accepted by the rating tool.
Second, lineage is a regulatory requirement. Since the Basel Committee’s principles for risk data aggregation (BCBS 239, 2013), large banks must be able to trace any reported risk number back to its origin and show the controls on the way. Supervisors apply the same expectation to climate figures.
Third, the file already contains the denominator of carbon accounting. Market capitalisation, book value of debt, equity, revenue: these are the inputs of the attribution factor of section 3. A carbon file adds a numerator to data the bank already has.
INTUITION — Think of the financial spread as the client’s medical record held by the bank. The carbon file is a new page in the same record, filled in by a different specialist under the same rules: date, source, signature.
IN ONE SENTENCE — A greenhouse gas inventory is an annual account of the gases a company emits directly (scope 1), through the energy it buys (scope 2) and through the rest of its value chain (scope 3), expressed in tonnes of CO2 equivalent.
The reference is the GHG Protocol Corporate Accounting and Reporting Standard (revised edition 2004), completed by the Scope 2 Guidance (2015) and the Corporate Value Chain (Scope 3) Standard (2011). European and international disclosure standards (ESRS E1, IFRS S2) build on it.
Scope 1 covers direct emissions from sources the company owns or controls: fuel burnt in furnaces, boilers and vehicles, process emissions from chemistry (the calcination of limestone in cement, the reduction of iron ore in steel), and fugitive emissions (methane leaks, refrigerants).
Scope 2 covers the emissions of the power plants and heat networks that produce the electricity, steam, heat and cooling the company buys. They are someone else’s scope 1.
Scope 3 covers everything else in the value chain, in fifteen categories: eight upstream (purchased goods and services, capital goods, fuel- and energy-related activities, upstream transport, waste, business travel, commuting, upstream leased assets) and seven downstream (downstream transport, processing of sold products, use of sold products, end-of-life, downstream leased assets, franchises, investments). For an oil company, category 11 (use of sold products) is the bulk of the footprint; for a carmaker too; for a bank it is category 15 (investments), which is where financed emissions sit.
INTUITION — A restaurant’s scope 1 is the gas burnt in its kitchen. Its scope 2 is the power station behind its fridge. Its scope 3 is the farm that raised the beef, the lorry that delivered it and the customer’s drive home. The kitchen is easy to measure, and most of the emissions are on the farm.
PITFALL — Scopes are designed to avoid double counting within one company, and they produce double counting by design between companies: the cement producer’s scope 1 is the builder’s scope 3. Adding the scope 3 of all borrowers in a loan book counts the same tonne several times. This is why financed scope 3 is reported separately from financed scope 1 and 2 and never summed across a portfolio without a warning.
The inventory covers seven families of gases: carbon dioxide, methane, nitrous oxide, hydrofluorocarbons, perfluorocarbons, sulphur hexafluoride and nitrogen trifluoride. They are converted into one unit with the global warming potential (GWP).
FORMULA
is the emissions in tonnes of CO2 equivalent; the mass emitted of gas in tonnes; the warming caused by one tonne of gas over a chosen horizon (100 years by convention) relative to one tonne of CO2.
The GWP values come from the assessment reports of the IPCC and change with each report. Over 100 years, fossil methane is worth 28 in the Fifth Assessment Report (AR5, 2014) and 29.8 in the Sixth (AR6, 2021); nitrous oxide 265 then 273; sulphur hexafluoride 23,500 then 25,200. Over 20 years, fossil methane is worth 82.5 in AR6, because methane is a short-lived gas whose effect is concentrated in the first two decades.
INTERACTIVE LAB — From tonnes of gas to tonnes of CO2 equivalent
Convert an inventory of four gases into CO2e under the AR5 and AR6 100-year GWP values, and see what a 20-year horizon does to methane.
| Parameter you can move | Range | Default |
|---|---|---|
| Carbon dioxide emitted | 0 to 2000000 t | 500000 t |
| Fossil methane emitted | 0 to 20000 t | 4000 t |
| Nitrous oxide emitted | 0 to 1000 t | 50 t |
| Sulphur hexafluoride emitted | 0 to 2000 kg | 200 kg |
| Result at default settings | Value |
|---|---|
| Total with AR5 GWP-100 | 629,950 tCO2e |
| Total with AR6 GWP-100 | 637,890 tCO2e |
| AR6 against AR5 | 1.26 % |
| Share of methane in the AR6 total | 18.7 % |
| AR6 total with methane at its 20-year GWP | 848,690 tCO2e |
Try it. Set methane to 20,000 t, the profile of a gas network with leaks: the 20-year reading more than doubles the inventory, which is why methane-heavy sectors are sensitive to the choice of horizon.
For the default inventory (a gas network operator, illustrative figures), the total is 629,950 tCO2e with AR5 values and 637,890 tCO2e with AR6, a difference of 1.26% with no change in the physical emissions. With methane at its 20-year value the same inventory reads 848,690 tCO2e.
PITFALL — When a client’s emissions fall by 1% between two years, check the methodology note before congratulating anyone. A change of GWP set, of emission factors or of perimeter can move the figure by more than a year of real effort. Serious reporters restate the base year, and the analyst’s job is to record which one applies.
A group owns 100% of some plants, 50% of a joint venture, 20% of an associate. Which emissions are “its” emissions? The GHG Protocol offers three answers.
| Approach | Rule | Typical user |
|---|---|---|
| Equity share | Emissions in proportion to the economic interest | Oil and gas majors, alongside another approach |
| Financial control | 100% of entities whose policies the group directs for economic benefit; in practice the fully consolidated perimeter | Companies aligning with their financial statements |
| Operational control | 100% of sites where the group has authority to set operating policies | Industrial groups operating joint ventures |
A joint venture owned at 50% and operated by the group counts for 50% under equity share, 0% under financial control if it is equity-accounted, and 100% under operational control. The emissions that fall outside scope 1 and 2 under the chosen approach should reappear in scope 3 category 15. They often do not.
WHY IT MATTERS ON THE DESK — The loan documentation defines the borrower group with the financial consolidation perimeter. If the carbon figure uses another perimeter, a KPI can improve simply because an emitting plant has been moved into a joint venture. Structurers therefore write the perimeter into the KPI definition and add a recalculation clause for acquisitions and disposals.
Absolute emissions say how large the problem is. Intensities allow comparison between companies of different size.
FORMULA
is the emissions in tCO2e; the physical output (tonnes of steel, MWh, passenger-kilometres); the revenue in EUR m. is the physical intensity; the monetary (or economic, or revenue) intensity in tCO2e per EUR m.
Physical intensity measures the efficiency of the production process and compares directly with sector decarbonisation pathways (kilograms of CO2 per tonne of cement, grams per kWh). It exists only for homogeneous sectors. Monetary intensity exists for every company, which is why portfolio metrics and data vendors use it; but its denominator moves with prices, exchange rates and business mix.
INTERACTIVE LAB — Physical against monetary intensity when prices move
See how a change in selling price moves revenue intensity while the plant emits exactly the same per tonne produced.
| Parameter you can move | Range | Default |
|---|---|---|
| Emissions, year 0 | 0.1 to 10 MtCO2e | 1 MtCO2e |
| Output, year 0 | 0.1 to 20 Mt | 2 Mt |
| Selling price, year 0 | 50 to 2000 EUR/t | 500 EUR/t |
| Change in selling price in year 1 | -30 to 50 % | 20 % |
| Real change in emissions per tonne in year 1 | -20 to 10 % | 0 % |
| Result at default settings | Value |
|---|---|
| Physical intensity, year 0 | 0.500 tCO2e/t |
| Physical intensity, year 1 | 0.500 tCO2e/t |
| Monetary intensity, year 0 | 1,000.0 tCO2e/EUR m |
| Monetary intensity, year 1 | 833.3 tCO2e/EUR m |
| Change in monetary intensity | -16.7 % |
Try it. Set the price change to -20% and the real efficiency gain to -10%: the plant is cleaner, yet its revenue intensity rises by 12.5%.
With the default values, a producer emitting 0.5 tCO2e per tonne sees its monetary intensity fall from 1,000 to 833.3 tCO2e per EUR m (a fall of 16.7%) when prices rise by 20%, without any change at the plant.
WHO DOES WHAT — Rule of thumb on the desk: use physical intensity to judge a client against its sector pathway and to design a KPI; use monetary intensity to compare across sectors in a portfolio and accept that it is noisy; use absolute emissions when the question is the bank’s own footprint. State which one you are quoting, with scopes and year, every time.
A data file has its own risks: transcription error, stale data, unit error (thousand tonnes read as tonnes is the classic), perimeter mismatch, and unsupported estimates. The documentation that controls them is internal: the data dictionary (definition, unit, owner of each field), the extraction sheet with page references, the methodology note for estimates, and the evidence of a four-eyes check. Section 7 returns to these.
Section 2 described the client’s inventory. This section adds what is specific to finance: how a share of that inventory is assigned to a financial exposure (PCAF), how third parties summarise ESG performance into ratings, and which rules now govern both.
IN ONE SENTENCE — Financed emissions are the share of a client’s emissions that a lender or investor reports as its own scope 3, in proportion to the share of the client’s financing that it provides.
The Partnership for Carbon Accounting Financials (PCAF) publishes the Global GHG Accounting and Reporting Standard for the Financial Industry in three parts: Part A on financed emissions, Part B on facilitated emissions from capital markets, Part C on insurance-associated emissions. Part A has been recognised by the GHG Protocol as conforming to its scope 3 standard for category 15.
FORMULA
is the financed emissions of the portfolio; the outstanding amount of the bank’s loan to, or investment in, company ; the value of company ; the emissions of company . The ratio is the attribution factor.
The denominator. For listed companies, is the enterprise value including cash (EVIC): market capitalisation of ordinary and preferred shares at fiscal year-end, plus the book values of total debt and of minority interests, with no deduction of cash. For private companies, is total equity plus total debt at book value. For project finance it is the project’s total equity and debt; for mortgages and commercial real estate, the property value at origination; for sovereign debt, GDP adjusted for purchasing power.
INTUITION — If ten friends finance a car in equal parts, each owns one tenth of its exhaust. PCAF applies the same reasoning to a company: all providers of equity and debt share 100% of the emissions, each in proportion to its ticket. EVIC is the price of the whole car, and cash is not deducted so that the shares never add up to more than 100%.
Asset classes. The 2022 edition of Part A covers seven: listed equity and corporate bonds, business loans and unlisted equity, project finance, commercial real estate, mortgages, motor vehicle loans and sovereign debt. The edition launched on 2 December 2025 leaves these methods unchanged and adds four: use-of-proceeds structures, securitisations and structured products, sub-sovereign debt, and an optional disclosure for undrawn loan commitments. It was accompanied by a supplemental guidance on financed avoided emissions and forward-looking metrics, to be reported separately.
Data quality score. Each figure carries a score from 1 (best) to 5. For business loans and listed securities:
| Score | Emissions data | Typical source |
|---|---|---|
| 1 | Reported by the company and verified | Assured sustainability statement |
| 2 | Reported, not verified; or computed from the company’s energy use | CDP response, questionnaire |
| 3 | Computed from the company’s production data | Tonnes produced times an emission factor |
| 4 | Estimated from the company’s revenue | Revenue times a sector factor |
| 5 | Estimated from the outstanding amount alone | Loan amount times a sector factor |
The bank publishes the exposure-weighted average score for each asset class. A score of 1 means that the client’s own figure has been checked by someone; it does not mean that the bank’s attributed figure is exact.
What is not counted. Undrawn commitments are outside the financed emissions figure (the 2025 edition allows a separate disclosure). Derivatives are outside. Scope 3 of clients is reported, and kept separate from scope 1 and 2.
PITFALL — Because EVIC contains the market capitalisation, the attribution factor rises when the share price falls. A bear market increases the bank’s financed emissions with no change in any chimney. PCAF’s 2025 edition includes recommendations on this “inventory fluctuation”; section 5.6 measures it.
When a bank underwrites a bond or an equity issue, the money comes from investors and the paper leaves the bank within days. PCAF Part B (December 2023) nevertheless attributes a share to the arranger.
FORMULA
is the amount facilitated by the bank (its share of the issue according to the league-table credit); the issuer’s EVIC or total equity plus debt; the weighting factor, set at 33%; the issuer’s emissions over the year.
Two features distinguish facilitated from financed emissions. The 33% weighting recognises that the arranger is neither the owner of the exposure nor responsible for holding it. The figure is a flow: it is reported once, for the year of issuance, whereas financed emissions are a stock measured at each year-end. The two are disclosed separately and not added. The amendments to IFRS S2 issued by the ISSB in December 2025, effective for periods beginning on or after 1 January 2027, allow reporters to limit category 15 to financed emissions; facilitated emissions therefore remain a voluntary disclosure under that standard.
IN ONE SENTENCE — An ESG rating is a third party’s opinion on a company’s exposure to, management of, or impact on environmental, social and governance matters, expressed on a scale.
The main commercial providers a banker meets are MSCI, Morningstar Sustainalytics, ISS (part of ISS STOXX), S&P Global, LSEG and Moody’s, the last having handed its corporate ESG scores to MSCI under a partnership in 2024. EcoVadis rates suppliers and is frequent in mid-cap loan documentation. CDP is a disclosure platform that also scores responses from A to D. These products do not measure the same thing: some assess financial risk to the company from ESG issues, others the company’s impact on the world, some both.
Berg, Kölbel and Rigobon (2022) compared six providers and found an average pairwise correlation of 0.54, with a range from 0.38 to 0.71; credit ratings from the main agencies correlate above 0.9. They split the divergence into three sources: measurement (the same attribute measured with different indicators) explains 56%, scope (different attributes included) 38%, and weights 6%.
INTERACTIVE LAB — Two providers, one company, two ESG scores
Show how different pillar weights and a different measurement of the same pillar produce different scores for an identical company.
| Parameter you can move | Range | Default |
|---|---|---|
| Environmental pillar score | 0 to 100 0-100 | 40 0-100 |
| Social pillar score | 0 to 100 0-100 | 60 0-100 |
| Governance pillar score | 0 to 100 0-100 | 80 0-100 |
| Provider X weight on E (rest split 60/40 between S and G) | 0 to 100 % | 50 % |
| Provider Y weight on E (rest split 40/60 between S and G) | 0 to 100 % | 20 % |
| Provider Y measures E differently by | -30 to 30 points | 0 points |
| Result at default settings | Value |
|---|---|
| Score from provider X | 54.0 0-100 |
| Score from provider Y | 65.6 0-100 |
| Gap, Y minus X | 11.6 points |
Try it. Give the company the same score on all three pillars: weights stop mattering and only the measurement input can create a gap. Real divergence comes mostly from measurement.
With the default values the same company scores 54 with provider X and 65.6 with provider Y, a gap of 11.6 points, through weights alone.
How a credit committee uses them. At Meridian, a vendor ESG rating is never an input of the PD model. It appears in the ESG section of the credit memo as external information, next to the bank’s own transition score, which is built from the carbon file (see 5.10 and E01). The EBA Guidelines on loan origination and monitoring (2020) ask banks to consider ESG factors in credit decisions; the EBA Guidelines on the management of ESG risks (EBA/GL/2025/01), applicable to large institutions since 11 January 2026, ask for an internal methodology and are explicit that institutions should not rely on external ratings alone.
How DCM uses them. The originator checks the issuer’s ratings before a pitch because investors do: some funds have minimum rating thresholds, and index inclusion depends on them. The rating appears in the investor presentation. A second-party opinion on a bond framework is a different product from an ESG rating of the issuer (see A04). Using an ESG rating as the KPI of a sustainability-linked instrument is possible but criticised, since the issuer does not control the methodology.
ESG Ratings Regulation. Regulation (EU) 2024/3005 has applied since 2 July 2026. Providers operating in the Union must be authorised by ESMA (EU providers) or come in through equivalence, endorsement or recognition (third-country providers). Providers already active had to notify ESMA by 2 August 2026 and file their application by 2 November 2026; they remain on a temporary register until ESMA decides. Small providers benefit from a lighter regime for three years. The regulation requires separation from conflicting activities such as consulting to rated entities, credit ratings and benchmark administration, transparency on methodology, including whether the rating addresses financial risk, impact or both, and it gives the rated company a short notice period before publication to flag factual errors. It encourages separate E, S and G ratings; a provider that issues one aggregate rating must disclose the weights. Credit ratings, second-party opinions, external reviews of European green bonds and scores produced by a bank for its own use are outside its scope. ESMA stated on 1 July 2026 that third parties may go on publishing and distributing the ratings of existing providers during the transition and should then consult its register.
Corporate disclosure. The supply of client data depends on the CSRD. The Omnibus I Directive (EU) 2026/470, in force since 18 March 2026, restricts mandatory sustainability reporting to undertakings with more than 1,000 employees and net turnover above EUR 450m, from financial year 2027. The simplified ESRS were adopted by the Commission as a delegated act on 3 July 2026, with about 60% fewer mandatory datapoints; gross scope 1, 2 and 3 emissions remain. A “value-chain cap” protects companies with up to 1,000 employees: a reporting undertaking, including a bank, cannot require from them more than the voluntary standard provides. For the data team this means that reported and assured data will exist for large caps and that proxies will remain the rule for the mid-cap book (see A02).
Bank disclosure. Article 449a of the CRR, as amended by CRR3, extends Pillar 3 ESG disclosures to all institutions. The EBA published its final draft implementing standards on 22 June 2026 (EBA/ITS/2026/02), submitted to the Commission and not yet adopted in November 2026: first reference date 31 December 2026 for large and other institutions, 31 December 2027 for small and non-complex ones. Template 1 keeps the financed emissions of counterparties by sector; the taxonomy templates (Green Asset Ratio and BTAR) are stopped for large institutions; the GAR remains in the Taxonomy Article 8 disclosures, themselves simplified by Delegated Regulation (EU) 2026/73, published on 8 January 2026 (see E06).
Standards in motion. The GHG Protocol is revising its corporate suite. The public consultation on scope 2, which proposed hourly matching and a same-grid condition for market-based claims, ran from 20 October 2025 to 31 January 2026. The development plan of 29 July 2026 merges the workstreams into one Corporate Standard version 3.0, with a consultation draft expected in the second quarter of 2027 and publication in the fourth quarter of 2028. Inventories for 2026 and 2027 are therefore prepared under the current texts.
Companies increasingly publish avoided emissions: the difference between the emissions of their product and those of a reference solution it replaces (an insulation panel against no insulation, a low-clinker cement against an ordinary one). The figure depends on a counterfactual chosen by the company. Under the GHG Protocol, ESRS E1 and PCAF, avoided emissions are reported separately and are never deducted from scope 1, 2 or 3. The same holds for carbon credits bought and retired: they are disclosed next to the gross inventory. A bank’s financed emissions are computed on the client’s gross figures.
WHY IT MATTERS ON THE DESK — A client presentation that shows “net emissions” after avoided emissions and credits is the most common way a wrong baseline enters a term sheet. The analyst’s first reflex is to find the gross scope 1 and 2 in the audited statement and work from there.
The “deal” here is a data point. The flow follows one figure, the scope 1 and 2 emissions of a corporate client, from the client’s annual report to the bank’s published tables.
INTERACTIVE LAB — Life of a carbon data point, from the client’s report to the bank’s Pillar 3
Follow one emissions figure through the bank, with the role that touches it and the document or record produced at each step.
| # | Who | Does what | Deliverable |
|---|---|---|---|
| 1 | Client: sustainability department and statutory auditor | Publish the annual report with the sustainability statement and the assurance report | Annual report, assurance conclusion |
| 2 | Coverage banker | Sends the bank’s ESG questionnaire and obtains a contact in the client’s sustainability team | Completed questionnaire, CRM entry |
| 3 | ESG data analyst | Extracts emissions, activity data, perimeter and methodology notes into the carbon file with page references | Carbon file, draft |
| 4 | Data vendor | Delivers its own reported or estimated figures through the periodic feed | Vendor feed, exception report |
| 5 | Second ESG data analyst | Performs the four-eyes check, reconciles with the vendor and rebuilds scope 1 from activity data where possible | Validated carbon file, data quality score |
| 6 | Credit analyst | Spreads the financial statements, computes EVIC components and writes the ESG section of the credit memo | Credit memo |
| 7 | ESG risk officer | Assigns the internal transition score and checks the sector policy | ESG risk opinion |
| 8 | Credit committee | Decides on the limit and sets information undertakings on sustainability data | Credit decision, minutes |
| 9 | Loan middle office | Books the facility with LEI, NACE code, purpose and label flags | Facility record in the loan system |
| 10 | DCM business management | Provides the list of the year’s issues with bookrunner shares | Deal list for facilitated emissions |
| 11 | Finance sustainability reporting | Runs the year-end engine: outstanding amounts times attribution factors times emissions | Financed and facilitated emissions inventory |
| 12 | Model validation and data steward | Review methodology changes, proxies and control results; data quality committee signs off | Validation report, committee minutes |
| 13 | Statutory auditor of the bank | Tests samples back to source documents and gives limited assurance | Assurance report on the sustainability statement |
| 14 | Finance and Investor relations | Publish the sustainability statement and the Pillar 3 ESG templates | Published disclosures |
Collection (steps 1 to 4). The primary source is the client’s own report. For a company under the CSRD the sustainability statement is part of the management report and carries a limited assurance conclusion from the statutory auditor or another accredited provider. The analyst reads four things before any number: the basis of preparation (consolidation approach, GWP set, emission factor sources), the list of scope 3 categories included and excluded, the restatements of prior years, and the scope of the assurance. The coverage banker’s questionnaire fills the gaps: activity data, the split of electricity contracts, the treatment of joint ventures. The vendor feed arrives in parallel. Vendors are faster and cover thousands of names, but they mix reported and modelled figures, and the flag that distinguishes them must be loaded with the number.
Validation (step 5). A second analyst checks every keyed figure against the page reference and reconciles with the vendor. Where the client publishes activity data, scope 1 is rebuilt from fuel and production volumes as a plausibility test, exactly as a credit analyst reconciles the cash-flow statement with the balance sheet. The data quality score is assigned at this step and justified in one line.
Use in credit (steps 6 to 9). The credit analyst owns the financial side, including market capitalisation and debt at the client’s year-end. The ESG risk officer converts the carbon file into a transition score. The credit committee can attach conditions: at Meridian, large corporate facilities carry an information undertaking to deliver the sustainability statement within 150 days of year-end. The middle office’s booking matters more than it seems: a wrong NACE code sends the exposure to the wrong sector line of Pillar 3 Template 1, and a missing LEI prevents the automatic match with the carbon file.
Year-end run (steps 10 to 14). Finance takes outstanding amounts from the accounting ledger at 31 December and joins them with the latest validated carbon files. For clients without a file, the engine applies proxies in a fixed order of preference. DCM business management supplies the bookrunner shares; they must agree with the league-table credit, which is the allocation PCAF prescribes. Changes of method go through model validation. The auditor selects a sample of exposures and walks each one back to the client’s report, which is only possible if step 3 recorded the page.
WHO DOES WHAT — Timeline for the 2026 inventory at Meridian: client reports for financial year 2025 arrive between February and May 2026; carbon files are validated by end of September; the dry run takes place in November; the final run in the third week of January 2027; the auditor’s fieldwork in February; publication with the annual report in March 2027. The emissions in the bank’s 2026 statement are therefore the clients’ 2025 emissions. The one-year lag is standard practice and is disclosed.
Question. What does Meridian need to know about Arvernia, and from which document?
Arvernia Matériaux SA is a fictional French cement and aggregates group listed in Paris, rated BBB by Meridian. On Thursday 12 November 2026 its limits come up for annual review. Meridian has three relationships with it: a EUR 135m participation, fully drawn, in a 5-year syndicated term loan; a EUR 22.5m holding of Arvernia’s outstanding bond in the treasury’s credit portfolio, bought in the secondary market; and a joint bookrunner role, with three other banks, on a new EUR 500m 7-year bond issued in June 2026. Meridian also has a EUR 100m undrawn commitment in the group’s revolving credit facility.
The ESG data analyst has the 2025 annual report, published in March 2026. The data below are case assumptions.
| Item, financial year 2025 | Value | Source in the report |
|---|---|---|
| Cementitious product sold | 12.0 Mt | Operating review |
| Clinker produced | 8.4 Mt | Sustainability statement |
| Revenue | EUR 2,400m | Income statement |
| Market capitalisation at 31 December 2025 | EUR 2,700m | 90m shares at EUR 30 |
| Total debt, book value, including leases | EUR 1,750m | Balance sheet and notes |
| Minority interests | EUR 50m | Balance sheet |
| Total equity, book value | EUR 2,250m | Balance sheet |
| Cash and equivalents | EUR 400m | Balance sheet |
The sustainability statement reports scope 1 of 6.60 MtCO2e, scope 2 of 0.396 Mt (location-based) and 0.317 Mt (market-based), and scope 3 of 2.00 Mt. The statutory auditor gives limited assurance on the statement. The analyst’s task is to verify that these figures hold together, to record them in the carbon file, and to hand Finance and DCM the attributable tonnes.
Question. Is the reported scope 1 of 6.60 Mt consistent with what the plants physically do?
Cement is a convenient sector for this test because its emissions follow from two physical facts. Clinker, the active ingredient, is made by heating limestone to about 1,450°C: the chemical reaction itself releases CO2 (process emissions), and the kiln burns fuel to reach the temperature (combustion emissions).
Data from the sustainability statement: 8.4 Mt of clinker; a process emission factor of 0.525 tCO2 per tonne of clinker; a thermal energy consumption of 3,500 MJ per tonne of clinker; an average fuel emission factor of 70 kgCO2 per GJ, net of the biomass share of the fuel mix; and 0.132 MtCO2e from quarry vehicles, on-site transport and other sources.
Calculation.
is the clinker production; the process factor; the heat consumption per tonne of clinker; the emission factor of the fuel mix; the remaining sources.
Process: 8.4 Mt × 0.525 = 4.41 Mt. Thermal energy: 8.4 Mt × 3,500 MJ = 29.4 million GJ; at 70 kg per GJ this gives 2.058 Mt. Adding 0.132 Mt, scope 1 is 6.60 Mt.
INTERACTIVE LAB — Arvernia: scope 1 rebuilt from clinker and fuel
Recompute scope 1 from production and energy data, as a plausibility test of the reported figure, and derive the gross physical intensity.
| Parameter you can move | Range | Default |
|---|---|---|
| Cementitious product | 5 to 20 Mt | 12 Mt |
| Clinker ratio | 50 to 95 % | 70 % |
| Process emission factor | 0.5 to 0.55 tCO2/t clinker | 0.525 tCO2/t clinker |
| Thermal energy | 3000 to 4500 MJ/t clinker | 3500 MJ/t clinker |
| Fuel emission factor, net of biomass | 30 to 100 kgCO2/GJ | 70 kgCO2/GJ |
| Other scope 1 sources | 0 to 0.5 MtCO2e | 0.132 MtCO2e |
| Result at default settings | Value |
|---|---|
| Clinker produced | 8.40 Mt |
| Process emissions | 4.410 MtCO2e |
| Combustion emissions | 2.058 MtCO2e |
| Scope 1 | 6.600 MtCO2e |
| Share of process emissions | 66.8 % |
| Gross scope 1 intensity | 550 kgCO2e/t cementitious |
Try it. Lower the clinker ratio to 60%: scope 1 falls by 14% without touching the kilns. Substituting clinker is the main lever of the sector, which is why a clinker-ratio KPI is a credible one.
Result. The rebuilt figure is 6.60 Mt, equal to the reported one. Process emissions are 66.8% of the total. The gross scope 1 intensity is 550 kgCO2e per tonne of cementitious product.
Reading. The reconciliation tells the analyst three things. The reported figure is plausible. Two thirds of it comes from chemistry, which no change of fuel can remove; a decarbonisation target that assumes a fast fall in scope 1 without clinker substitution or carbon capture is not credible. And the phrase “net of biomass” hides a convention: CO2 from burning biomass is reported outside the scopes, so a kiln that switches to waste wood shows lower scope 1 while the chimney emits as much. The analyst records the biogenic figure in a memo field.
Question. Which of the two scope 2 figures should enter the file?
Data. Arvernia buys 1,320 GWh of electricity (110 kWh per tonne of cementitious product). The average grid emission factor of its countries of operation, weighted by consumption, is 0.300 tCO2e per MWh. 40% of the volume is covered by power purchase agreements and guarantees of origin with a contractual factor of zero; the remaining 60% is accounted at a residual-mix factor of 0.400 tCO2e per MWh.
Calculation. The location-based method applies the grid average to all consumption: 1,320 GWh × 0.300 = 0.396 Mt. The market-based method applies the factor of each contract and, for uncovered volumes, the residual mix, which is the grid average after removing the green electricity already claimed by others: 1,320 × 60% × 0.400 = 0.3168 Mt.
INTERACTIVE LAB — Arvernia: scope 2 under the two methods
Compute location-based and market-based scope 2 from electricity volume, grid factor, contractual coverage and residual mix.
| Parameter you can move | Range | Default |
|---|---|---|
| Electricity purchased | 500 to 3000 GWh | 1320 GWh |
| Average grid factor | 0.02 to 0.8 tCO2e/MWh | 0.3 tCO2e/MWh |
| Share covered by zero-carbon contracts | 0 to 100 % | 40 % |
| Residual-mix factor | 0.02 to 0.9 tCO2e/MWh | 0.4 tCO2e/MWh |
| Result at default settings | Value |
|---|---|
| Scope 2, location-based | 0.3960 MtCO2e |
| Scope 2, market-based | 0.3168 MtCO2e |
| Market-based against location-based | -20.0 % |
Try it. Set coverage to 100%: market-based scope 2 is zero although the plants draw the same electrons from the same grid. Set it to 0%: market-based is higher than location-based, because the residual mix is dirtier than the average.
Result. 0.396 Mt location-based, 0.3168 Mt market-based, 20% lower.
Reading. Both figures are legitimate and both are required by the GHG Protocol’s dual reporting. They answer different questions: the location-based figure describes the physical grid; the market-based figure describes the company’s purchasing choices. Meridian’s data policy stores both and uses the market-based figure for attribution when the contractual instruments are documented in the assured statement, which is the case here. The analyst notes that 40% coverage relies partly on unbundled guarantees of origin, the instrument the GHG Protocol’s scope 2 consultation of 2025-2026 proposed to restrict through hourly and same-grid matching. If that proposal enters the future standard, Arvernia’s market-based figure will move towards the location-based one. For a cement group scope 2 is 4.6% of scope 1 and 2; for an aluminium smelter or a data-centre operator the same choice would decide the whole footprint.
Question. What does the scope 3 figure contain, and what is missing from the inventory altogether?
Data. Arvernia reports scope 3 by category.
| Scope 3 category | MtCO2e |
|---|---|
| 1. Purchased goods and services (clinker, additives, explosives) | 0.62 |
| 3. Fuel- and energy-related activities (extraction and transport of fuels) | 0.48 |
| 4. Upstream transport | 0.36 |
| 9. Downstream transport | 0.41 |
| Other reported categories (2, 5, 6, 7) | 0.13 |
| Total reported | 2.00 |
The basis of preparation states that the inventory follows the financial control approach. Note 14 of the financial statements shows a 50% joint venture, Tangier Cement, accounted for by the equity method and operated by Arvernia under a management contract. The joint venture’s own report gives a scope 1 of 1.00 Mt. It appears nowhere in Arvernia’s inventory.
Calculation. Under financial control the joint venture is outside scope 1, which stays at 6.60 Mt. Under equity share Arvernia would add 50% of 1.00 Mt, giving 7.10 Mt. Under operational control it would add 100%, giving 7.60 Mt. Under the approach Arvernia has chosen, its share of the joint venture belongs in scope 3 category 15: 0.50 Mt, which would raise scope 3 from 2.00 to 2.50 Mt.
INTERACTIVE LAB — Arvernia: what the joint venture does to the inventory
Compare scope 1 under the three consolidation approaches and compute the scope 3 category 15 amount missing under financial control.
| Parameter you can move | Range | Default |
|---|---|---|
| Scope 1 of consolidated entities | 1 to 20 MtCO2e | 6.6 MtCO2e |
| Scope 1 of the joint venture | 0 to 5 MtCO2e | 1 MtCO2e |
| Economic interest in the joint venture | 0 to 100 % | 50 % |
| Scope 3 as reported | 0 to 20 MtCO2e | 2 MtCO2e |
| Result at default settings | Value |
|---|---|
| Scope 1, financial control | 6.60 MtCO2e |
| Scope 1, equity share | 7.10 MtCO2e |
| Scope 1, operational control | 7.60 MtCO2e |
| Operational against financial control | 15.2 % |
| Missing scope 3 category 15 | 0.50 MtCO2e |
| Scope 3 including category 15 | 2.50 MtCO2e |
Try it. Raise the joint venture’s emissions to 3 Mt, as if Arvernia transferred two more plants into it: reported scope 1 does not move under financial control while the plants it operates emit 45% more.
Result. Scope 1 is 6.60, 7.10 or 7.60 Mt depending on the approach, a spread of 15.2%. Category 15 should show 0.50 Mt and shows nothing.
Reading. The assured figure is correct under its stated convention, and the convention leaves 1.00 Mt of emissions from a plant the group runs outside every line of the inventory. The analyst does two things. For attribution, the file keeps the reported figures: PCAF works from the company’s inventory, and a bank that adjusts client data name by name loses comparability and auditability. For credit, a flag is raised: “joint venture of 1.00 Mt operated by the group outside the reporting perimeter; category 15 not reported”, and the coverage banker is asked to raise it with the client. If Arvernia later proposes a sustainability-linked facility, the KPI definition will have to include the joint venture or freeze the perimeter (see B03).
On the rest of scope 3, the categories are those expected for the sector; there is no category 11 because cement does not burn fuel in use. The figure is 22.4% of the total footprint, and the statement says that categories 1 and 4 are partly estimated from spend.
Question. Which headline figures go on the first page of the file?
Calculation. Scope 1 and 2 (market-based) is 6.60 + 0.3168 = 6.9168 Mt. With scope 3 the total is 8.9168 Mt. The physical intensity on scope 1 and 2 is 6.9168 Mt over 12.0 Mt of product, or 576.4 kg per tonne. The monetary intensity is 6,916,800 t over EUR 2,400m, or 2,882 tCO2e per EUR m of revenue.
Arvernia’s investor presentation shows a different headline: “net emissions of 5.87 Mt”. It is obtained by deducting 0.90 Mt of avoided emissions, claimed for low-carbon cements against a standard cement, and 0.15 Mt of carbon credits retired during the year.
INTERACTIVE LAB — Arvernia: first page of the carbon file
Assemble the headline figures of the file from the scopes, production and revenue, and show the ‘net’ figure of the investor presentation next to the gross one.
| Parameter you can move | Range | Default |
|---|---|---|
| Scope 1 | 0 to 20 MtCO2e | 6.6 MtCO2e |
| Scope 2, market-based | 0 to 5 MtCO2e | 0.3168 MtCO2e |
| Scope 3, reported | 0 to 20 MtCO2e | 2 MtCO2e |
| Cementitious product | 1 to 30 Mt | 12 Mt |
| Revenue | 500 to 10000 EUR m | 2400 EUR m |
| Avoided emissions claimed | 0 to 3 MtCO2e | 0.9 MtCO2e |
| Carbon credits retired | 0 to 2 MtCO2e | 0.15 MtCO2e |
| Result at default settings | Value |
|---|---|
| Scope 1 and 2 | 6.9168 MtCO2e |
| Scope 1, 2 and 3 | 8.9168 MtCO2e |
| Scope 2 in scope 1 and 2 | 4.6 % |
| Scope 3 in the total | 22.4 % |
| Physical intensity, scope 1 and 2 | 576.4 kgCO2e/t |
| Monetary intensity, scope 1 and 2 | 2,882 tCO2e/EUR m |
| ‘Net’ figure of the investor presentation (not used) | 5.8668 MtCO2e |
| Understatement if the net figure were used | 15.2 % |
Try it. Raise avoided emissions to 3 Mt: the ‘net’ figure falls below 4 Mt. The counterfactual is chosen by the company, so the deduction has no natural limit, which is the reason standards forbid netting.
Result. The file records 6.9168 Mt for scope 1 and 2, 8.9168 Mt with scope 3, 576.4 kg per tonne and 2,882 tCO2e per EUR m. The net figure of 5.8668 Mt is recorded in a memo field; using it would understate scope 1 and 2 by 15.2%.
Reading. Avoided emissions may be real. A cement with less clinker does emit less than the one it replaces. They are kept outside for three reasons the analyst should be able to give: the reference scenario is hypothetical and chosen by the reporter; the same avoidance is claimed by the cement producer, the builder and the building’s owner; and the atmosphere responds to gross emissions. They remain useful as a commercial argument and as an indicator of where revenue is going, and the 2025 PCAF supplemental guidance lets a bank report “financed avoided emissions” in a separate line. Carbon credits are a different object: a purchase from a third party, disclosed as such. EU ETS allowances surrendered by Arvernia are a cost of compliance and do not reduce the inventory either (see C10).
Question. How many of Arvernia’s tonnes belong to Meridian’s balance sheet?
Data. Outstanding at Meridian’s year-end: EUR 135m of term loan and EUR 22.5m of bond at amortised cost. The EUR 100m undrawn commitment is excluded. Meridian’s convention pairs the client’s emissions and the client’s EVIC at the same date, 31 December 2025.
Calculation. EVIC is market capitalisation plus total debt plus minority interests: 2,700 + 1,750 + 50 = EUR 4,500m. Cash is not deducted; the usual enterprise value would be EUR 4,100m.
The attribution factor of the loan is 135 / 4,500 = 3.00%. Financed scope 1 and 2: 3.00% × 6,916,800 = 207,504 tCO2e. Financed scope 3: 3.00% × 2,000,000 = 60,000 tCO2e, reported separately.
The bond holding is attributed in the same way: 22.5 / 4,500 = 0.50%, or 34,584 tCO2e. PCAF treats a corporate bond and a corporate loan to the same listed company identically, since both are claims on the same enterprise value.
In total Meridian carries 3.50% of Arvernia, that is 242,088 tCO2e of scope 1 and 2 and 70,000 tCO2e of scope 3, for EUR 157.5m of exposure: 1,537 tCO2e per EUR m lent or invested.
INTERACTIVE LAB — Arvernia: EVIC, attribution factor and financed emissions
Compute EVIC, attribute scope 1 and 2 emissions to the loan and the bond holding, and measure how a move in the share price changes the result.
| Parameter you can move | Range | Default |
|---|---|---|
| Market capitalisation | 500 to 10000 EUR m | 2700 EUR m |
| Total debt, book value | 0 to 5000 EUR m | 1750 EUR m |
| Minority interests | 0 to 500 EUR m | 50 EUR m |
| Cash (for comparison only) | 0 to 2000 EUR m | 400 EUR m |
| Loan outstanding | 0 to 500 EUR m | 135 EUR m |
| Bond holding | 0 to 200 EUR m | 22.5 EUR m |
| Client scope 1 and 2 | 0 to 20000000 tCO2e | 6916800 tCO2e |
| Client scope 3 | 0 to 20000000 tCO2e | 2000000 tCO2e |
| Change in share price next year | -50 to 50 % | -20 % |
| Result at default settings | Value |
|---|---|
| EVIC | 4,500 EUR m |
| Enterprise value net of cash (not used) | 4,100 EUR m |
| Attribution factor, loan | 3.00 % |
| Financed scope 1 and 2, loan | 207,504 tCO2e |
| Financed scope 3, loan | 60,000 tCO2e |
| Attribution factor, bond | 0.50 % |
| Financed scope 1 and 2, bond | 34,584 tCO2e |
| Financed scope 1 and 2, total | 242,088 tCO2e |
| Financed scope 3, total | 70,000 tCO2e |
| Financed scope 1 and 2 per EUR m of exposure | 1,537 tCO2e/EUR m |
| Loan, after the share price move | 235,800 tCO2e |
| Change caused by the share price alone | 13.6 % |
| Loan, with book equity plus debt as denominator | 233,442 tCO2e |
Try it. Move the share price by +40%: the bank’s financed emissions fall by 19% while Arvernia emits the same. Then ask yourself what a decarbonisation target on absolute financed emissions really measures in a bull market.
Result. EVIC is EUR 4,500m. The loan carries 207,504 tCO2e and the bond 34,584 tCO2e, 242,088 tCO2e in total, or 1,537 tCO2e per EUR m. If Arvernia’s share price fell by 20% with unchanged debt and emissions, EVIC would be EUR 3,960m and the loan would carry 235,800 tCO2e, an increase of 13.6% caused by the stock market alone. With book equity plus debt (EUR 4,000m) as the denominator, the loan would carry 233,442 tCO2e.
Reading. Three remarks for the banker.
The order of magnitude first. One loan of EUR 135m to a cement producer carries about 207,500 tonnes of scope 1 and 2. A loan of the same size to a software company with a monetary intensity of 10 tCO2e per EUR m and a similar attribution factor would carry a few hundred. A handful of names in cement, steel, power and oil make up most of a bank’s financed emissions, which is why sector targets concentrate on them (see E07).
Then the nature of the figure. Meridian neither emits nor controls these tonnes. Attribution is a convention for sharing responsibility among capital providers, useful for steering a portfolio and for comparing banks, and unsuitable as a measure of what would happen if Meridian withdrew: another lender would take the ticket and the kilns would keep running.
Finally the volatility. An absolute financed emissions figure moves with the share price through EVIC, with the client’s leverage, with exchange rates and with the bank’s own share in the syndicate. Practitioners therefore explain the annual change in a waterfall: change in client emissions, change in attribution, change in exposure, change in method. Only the first describes the real economy. PCAF’s work on inventory fluctuation exists because the other three often dominate.
PITFALL — Mixing dates produces nonsense quietly. Emissions of 2025 divided by an EVIC computed with the market capitalisation of November 2026 attribute the wrong share. Keep numerator and denominator of the client at the same year-end, and disclose the lag with the bank’s own reporting date.
Question. What does Meridian report for its role on Arvernia’s EUR 500m bond of June 2026?
Data. Four joint bookrunners with equal league-table credit. Weighting factor 33%. The emissions and EVIC are those of the file.
Calculation. The facilitated amount is 500 / 4 = EUR 125m. The attribution factor before weighting is 125 / 4,500 = 2.78%. After the 33% weighting, 0.917%. Facilitated scope 1 and 2: 0.917% × 6,916,800 = 63,404 tCO2e. Facilitated scope 3: 18,333 tCO2e.
INTERACTIVE LAB — Arvernia: facilitated emissions of a bookrunner
Compute the facilitated emissions of a bond mandate under PCAF Part B and compare them with the financed emissions of the loan.
| Parameter you can move | Range | Default |
|---|---|---|
| Issue size | 100 to 2000 EUR m | 500 EUR m |
| Bookrunners with equal credit | 1 to 10 banks | 4 banks |
| Weighting factor | 0 to 100 % | 33 % |
| EVIC of the issuer | 1000 to 20000 EUR m | 4500 EUR m |
| Issuer scope 1 and 2 | 0 to 20000000 tCO2e | 6916800 tCO2e |
| Issuer scope 3 | 0 to 20000000 tCO2e | 2000000 tCO2e |
| Financed emissions of the loan, for comparison | 0 to 1000000 tCO2e | 207504 tCO2e |
| Result at default settings | Value |
|---|---|
| Facilitated amount | 125.0 EUR m |
| Attribution before weighting | 2.78 % |
| Attribution after weighting | 0.917 % |
| Facilitated scope 1 and 2 | 63,404 tCO2e |
| Facilitated scope 3 | 18,333 tCO2e |
| Scope 1 and 2 without the weighting | 192,133 tCO2e |
| Facilitated as share of the loan’s financed emissions | 30.6 % |
Try it. Set the weighting to 100%: the mandate would carry 192,133 t, almost as much as a loan held for five years, for a week of underwriting. The 33% is a negotiated convention, and it is the most debated number in Part B.
Result. Meridian reports 63,404 tCO2e of facilitated scope 1 and 2 for 2026, equal to 30.6% of the annual financed emissions of its loan, and 18,333 tCO2e of scope 3.
Reading. The figure is reported once, in the 2026 inventory. In 2027 the bond is still outstanding and Arvernia still emits, but Meridian reports nothing for it unless it holds some of the paper, in which case the holding is financed emissions. This asymmetry is deliberate, since a flow business is measured as a flow, and it makes facilitated emissions volatile: a strong year in high-yield energy issuance raises the figure, a quiet year lowers it, without any change of policy. It also means that league-table credit, a marketing statistic, becomes an input to a sustainability disclosure; DCM business management must keep the allocation evidence (see C01 for the fee side of the same mandate).
Question. How far from the truth would Meridian be if Arvernia published nothing?
This is the situation for most mid-caps, and after the Omnibus it will remain so. PCAF sets an order of preference among estimation options. The analyst tests them on Arvernia, where the true answer is known. Sector factors below are case assumptions, of the kind found in the PCAF emission factor database or in environmentally extended input-output tables.
Data. Production-based factor: 0.62 tCO2e of scope 1 and 2 per tonne of cement, sector average. Revenue-based factor: 2,300 tCO2e per EUR m of revenue for cement manufacturing. Asset-based factor: 1,100 tCO2e per EUR m of loan outstanding to the sector.
Calculation. With production known (score 3): 12.0 Mt × 0.62 = 7.44 Mt; attributed at 3.00%, 223,200 tCO2e. With revenue known (score 4): 2,400 × 2,300 = 5.52 Mt; attributed at 3.00%, 165,600 tCO2e. With nothing but the loan amount (score 5): 135 × 1,100 = 148,500 tCO2e.
INTERACTIVE LAB — Arvernia: financed emissions of the loan under each PCAF data option
Compare the financed emissions obtained from reported data with those from production-based, revenue-based and asset-based proxies, and measure the error of each.
| Parameter you can move | Range | Default |
|---|---|---|
| Reported scope 1 and 2 | 1000000 to 15000000 tCO2e | 6916800 tCO2e |
| Production | 1 to 30 Mt | 12 Mt |
| Production-based factor | 0.3 to 1 tCO2e/t | 0.62 tCO2e/t |
| Revenue | 500 to 10000 EUR m | 2400 EUR m |
| Revenue-based factor | 500 to 5000 tCO2e/EUR m | 2300 tCO2e/EUR m |
| Loan outstanding | 10 to 500 EUR m | 135 EUR m |
| EVIC | 1000 to 20000 EUR m | 4500 EUR m |
| Asset-based factor | 100 to 3000 tCO2e/EUR m lent | 1100 tCO2e/EUR m lent |
| Result at default settings | Value |
|---|---|
| Score 1: reported and verified | 207,504 tCO2e |
| Score 3: production-based | 223,200 tCO2e |
| Error of the production-based estimate | 7.6 % |
| Score 4: revenue-based | 165,600 tCO2e |
| Error of the revenue-based estimate | -20.2 % |
| Score 5: asset-based | 148,500 tCO2e |
| Error of the asset-based estimate | -28.4 % |
Try it. Raise revenue to EUR 2,900m, as if cement prices had risen by a fifth: the revenue-based estimate becomes almost exact by accident. A proxy can be right for the wrong reason.
Result. Against 207,504 tCO2e from reported data, the production proxy gives 223,200 (7.6% too high), the revenue proxy 165,600 (20.2% too low) and the asset proxy 148,500 (28.4% too low).
Reading. The ranking of errors follows the ranking of scores, which is the logic of the scale: the closer the proxy is to physical activity, the better. The production proxy errs on the high side because Arvernia’s clinker ratio of 70% is better than the factor’s sector average; the revenue proxy errs on the low side because Arvernia sells cheaper products than the sector average embedded in the factor. Neither proxy can see a company’s own progress: a client that halves its emissions keeps the same estimate until it reports. A bank whose book is mostly at scores 4 and 5 is therefore publishing a sector statistic weighted by its exposures. The practical conclusion for coverage is that obtaining two numbers from a non-reporting client, production volume and energy consumption, moves the score from 5 to 2 or 3 and brings the estimate much closer to the truth. Under the value-chain cap the bank may ask and the client may decline.
Question. How does Arvernia sit in Meridian’s building-materials book, and how good are the book’s data?
Data. The book has four names (the three others are fictional case assumptions).
| Borrower | Exposure (EUR m) | Intensity (tCO2e/EUR m revenue) | Data quality score |
|---|---|---|---|
| Arvernia Matériaux (cement) | 157.5 | 2,882 | 1 |
| Granulats du Rhône (aggregates) | 200.0 | 400 | 2 |
| Béton Préfa Ouest (precast concrete) | 150.0 | 250 | 4 |
| Plâtres Occitans (plaster) | 92.5 | 600 | 5 |
Calculation. The weighted data quality score is the exposure-weighted average of the scores: (157.5 × 1 + 200 × 2 + 150 × 4 + 92.5 × 5) / 600 = 2.70. The weighted average carbon intensity (WACI) is the exposure-weighted average of the monetary intensities: 1,045 tCO2e per EUR m.
INTERACTIVE LAB — Building-materials book: weighted data quality score and WACI
Compute the exposure-weighted data quality score and the weighted average carbon intensity of a four-name book, and Arvernia’s share in each.
| Parameter you can move | Range | Default |
|---|---|---|
| Exposure to Arvernia | 0 to 500 EUR m | 157.5 EUR m |
| Exposure to Granulats du Rhône | 0 to 500 EUR m | 200 EUR m |
| Exposure to Béton Préfa Ouest | 0 to 500 EUR m | 150 EUR m |
| Exposure to Plâtres Occitans | 0 to 500 EUR m | 92.5 EUR m |
| Data quality score of Plâtres Occitans | 1 / 2 / 3 / 4 / 5 | 5 |
| Data quality score of Béton Préfa Ouest | 1 / 2 / 3 / 4 / 5 | 4 |
| Result at default settings | Value |
|---|---|
| Total exposure | 600.0 EUR m |
| Weighted data quality score | 2.70 1-5 |
| WACI | 1,045 tCO2e/EUR m revenue |
| Arvernia’s share of exposure | 26.2 % |
| Arvernia’s share of the WACI | 72.4 % |
Try it. Set the exposure to Arvernia to zero: the WACI falls by 63% and the data quality score worsens to 3.31. The book looks greener and is known less well.
Result. The book’s weighted score is 2.70 and its WACI 1,045 tCO2e per EUR m. Arvernia is 26.25% of the exposure and 72.4% of the WACI.
Reading. The best-documented name is also the one that weighs most in the carbon metric. This is typical: large emitters are large companies that report, and the weak data sit on small names with small footprints. The portfolio manager’s temptation is visible in the simulator: selling down Arvernia improves the intensity metric immediately and worsens the average data quality. The data quality score is also a work plan: moving the two weakest names to score 2 by collecting energy data would bring the book to 1.74.
WACI is the metric investors know from the TCFD recommendations. It needs no EVIC and is therefore stable against share prices, but it inherits every weakness of revenue intensity seen in 2.5.
Question. What goes in the ESG section of the credit memo?
Data. Three ESG rating providers cover Arvernia. To compare them the analyst converts each rating into a percentile rank within the provider’s own construction-materials peer group (100 is best): 35 for provider A, 55 for provider B, 70 for provider C. Meridian’s ESG risk team scores transition risk on four criteria from the carbon file, each on 100: physical intensity against the sector (45), three-year trajectory (60), credibility of targets and investment plan (70), data quality (90), with weights of 40%, 30%, 20% and 10%.
Calculation. The mean external percentile is 53.3 with a range of 35 points. The internal score is 0.40 × 45 + 0.30 × 60 + 0.20 × 70 + 0.10 × 90 = 59.
INTERACTIVE LAB — Arvernia: three external ratings and Meridian’s transition score
Summarise the dispersion of external ESG ratings expressed as peer percentiles and compute the internal transition score from four criteria.
| Parameter you can move | Range | Default |
|---|---|---|
| Provider A, peer percentile | 0 to 100 0-100 | 35 0-100 |
| Provider B, peer percentile | 0 to 100 0-100 | 55 0-100 |
| Provider C, peer percentile | 0 to 100 0-100 | 70 0-100 |
| Intensity against sector (weight 40%) | 0 to 100 0-100 | 45 0-100 |
| Three-year trajectory (weight 30%) | 0 to 100 0-100 | 60 0-100 |
| Credibility of targets and capex (weight 20%) | 0 to 100 0-100 | 70 0-100 |
| Data quality (weight 10%) | 0 to 100 0-100 | 90 0-100 |
| Result at default settings | Value |
|---|---|
| Mean external percentile | 53.3 0-100 |
| Range between providers | 35 points |
| Internal transition score | 59.0 0-100 |
Try it. Raise the intensity criterion from 45 to 70, the score Arvernia would get with a clinker ratio of 60%: the internal score reaches 69, close to the threshold of 70 that Meridian uses for ‘aligned’.
Result. External percentiles of 35, 55 and 70: mean 53.3, range 35 points. Internal transition score: 59, in Meridian’s “transitioning” band (50 to 70).
Reading. A cement producer is, for one provider, a third-quartile company and, for another, a top-third one. Reading the methodologies explains it: provider A scores absolute exposure to carbon regulation, where any cement group does badly; provider C scores management quality relative to peers, where Arvernia’s targets and disclosure do well. Neither is wrong. The memo therefore presents the three ratings with one line on what each measures, and does not average them. The internal score is the one the committee discusses, because every component can be traced to a figure in the carbon file and challenged. The DCM originator, for the June bond, used the same file in a different way: the investor presentation quoted the physical intensity of 550 kg per tonne against a sector pathway, which fixed-income investors in the sector follow more than any rating.
| Figure | Value | Stage |
|---|---|---|
| Scope 1, rebuilt and reported | 6.60 MtCO2e, of which process 66.8% | 5.2 |
| Gross scope 1 intensity | 550 kgCO2e/t cementitious | 5.2 |
| Scope 2, location-based / market-based | 0.396 / 0.3168 MtCO2e | 5.3 |
| Scope 1 under equity share / operational control | 7.10 / 7.60 MtCO2e | 5.4 |
| Scope 3 reported / with category 15 | 2.00 / 2.50 MtCO2e | 5.4 |
| Scope 1 and 2 retained in the file | 6.9168 MtCO2e | 5.5 |
| Physical / monetary intensity, scope 1 and 2 | 576.4 kg/t / 2,882 tCO2e per EUR m | 5.5 |
| “Net” figure of the investor presentation, not used | 5.8668 MtCO2e | 5.5 |
| EVIC | EUR 4,500m | 5.6 |
| Attribution factor: loan / bond holding | 3.00% / 0.50% | 5.6 |
| Financed scope 1 and 2: loan / bond / total | 207,504 / 34,584 / 242,088 tCO2e | 5.6 |
| Financed scope 3, total | 70,000 tCO2e | 5.6 |
| Loan after a 20% fall in the share price | 235,800 tCO2e (+13.6%) | 5.6 |
| Facilitated scope 1 and 2, bond issue | 63,404 tCO2e | 5.7 |
| Proxy errors: production / revenue / asset | +7.6% / -20.2% / -28.4% | 5.8 |
| Book: weighted data quality score / WACI | 2.70 / 1,045 tCO2e per EUR m | 5.9 |
| External percentiles, range / internal score | 35 points / 59 | 5.10 |
A data module has no deal of its own. The economics below are those of the exposure the data describe, Meridian’s EUR 135m term loan to Arvernia, and the question is what the carbon file changes in them.
Carbon data do not enter the capital formula. Under the F-IRB approach the risk weight of a senior unsecured corporate loan depends on PD, LGD (40%) and maturity (2.5 years); under the standardised approach it depends on the external credit rating, 75% for the BBB bucket under CRR3. There is no green supporting factor and no brown penalising factor in Pillar 1 (see E02 for the debate).
INTERACTIVE LAB — Arvernia term loan: risk weight and capital, F-IRB against standardised
Compute the CRR3 risk weight, RWA and allocated capital of the loan, and see how a climate overlay on the PD would move them.
| Parameter you can move | Range | Default |
|---|---|---|
| Exposure at default | 10 to 500 EUR m | 135 EUR m |
| One-year PD | 0.05 to 4 % | 0.2 % |
| LGD | 10 to 75 % | 40 % |
| Standardised risk weight | 20 to 150 % | 75 % |
| Result at default settings | Value |
|---|---|
| F-IRB risk weight | 39.02 % |
| RWA, F-IRB | 52.67 EUR m |
| Allocated capital at 13%, F-IRB | 6.85 EUR m |
| RWA, standardised | 101.25 EUR m |
| Allocated capital at 13%, standardised | 13.16 EUR m |
Try it. Move the PD from 0.20% to 0.30%, a one-notch overlay for transition risk: capital rises by 24%. This is the only channel through which carbon data reach Pillar 1, and it passes through the rating.
At a PD of 0.20% the risk weight is 39.02%, RWA EUR 52.67m and allocated capital EUR 6.85m. The standardised approach would give EUR 101.25m of RWA and EUR 13.16m of capital; this matters at bank level through the output floor, which rises in steps to 72.5% of standardised RWA.
Where carbon data do bite is Pillar 2. The ECB’s supervisory expectations and the EBA Guidelines on ESG risk management ask banks to identify concentrations in high-emitting sectors and to reflect transition risk in ratings and in the ICAAP. An analyst override of one notch for a cement producer with no credible plan is the usual mechanism (see E01).
Assumptions. Margin 150 bp over EURIBOR; FTP liquidity premium 30 bp for 5-year funding; upfront fee 50 bp spread over five years; cost-income 40% on lending; PD 0.20%, LGD 40%. Ancillary revenue from the relationship: EUR 0.4375m for Meridian’s quarter of a 35 bp fee on the EUR 500m bond, plus EUR 0.60m of cash management and FX, total EUR 1.0375m at a cost-income of 60% and EUR 2.5 of RWA per euro. Carbon file recharge: EUR 0.015m a year.
| Line | Loan alone (EUR m) | Relationship (EUR m) |
|---|---|---|
| Net margin over FTP, 120 bp on 135 | 1.620 | 1.620 |
| Upfront fee, one fifth of 0.675 | 0.135 | 0.135 |
| Ancillary revenue | n/a | 1.0375 |
| Operating costs | 0.702 | 1.3245 |
| Carbon file recharge | n/a | 0.015 |
| Expected loss, 0.20% × 40% × 135 | 0.108 | 0.108 |
| RWA | 52.67 | 55.27 |
| RAROC after tax | 10.35% | 14.04% |
INTERACTIVE LAB — Arvernia: RAROC of the loan and of the relationship, with data cost, PD overlay and a shadow carbon price
Work out RAROC line by line and test three ways carbon data could enter it: the cost of the file, a PD overlay, and an internal price on financed emissions.
| Parameter you can move | Range | Default |
|---|---|---|
| Margin over EURIBOR | 50 to 300 bp | 150 bp |
| FTP liquidity premium | 0 to 80 bp | 30 bp |
| Upfront fee, flat | 0 to 100 bp | 50 bp |
| One-year PD | 0.05 to 2 % | 0.2 % |
| Ancillary revenue | 0 to 3 EUR m | 1.0375 EUR m |
| Carbon file recharge | 0 to 0.1 EUR m | 0.015 EUR m |
| Shadow price on financed scope 1 and 2 | 0 to 5 EUR/tCO2e | 0 EUR/tCO2e |
| Result at default settings | Value |
|---|---|
| FTP liquidity cost, already deducted from the margin | 0.405 EUR m |
| Loan revenues | 1.755 EUR m |
| Loan operating costs | 0.702 EUR m |
| Expected loss | 0.108 EUR m |
| Loan RWA | 52.67 EUR m |
| RAROC, loan alone | 10.35 % |
| Relationship operating costs, before data and carbon | 1.3245 EUR m |
| Relationship RWA | 55.27 EUR m |
| Shadow carbon charge | 0.000 EUR m |
| RAROC, relationship | 14.04 % |
| Shadow price that brings the relationship to the 12% hurdle | 0.94 EUR/tCO2e |
Try it. Set the shadow price to EUR 1 per tonne: the relationship drops below the hurdle. Then compare this with the price of an EU allowance, which is tens of euros, and ask who already pays it.
Reading. The loan alone returns 10.35%, below the 12% hurdle; the relationship returns 14.04%, which is why the loan is made. The carbon file enters in three ways.
The cost of the file. Without the EUR 15,000 recharge the relationship would return 14.20%. Data cost 0.16 point of RAROC on a large corporate; the same recharge on a EUR 10m mid-cap loan would be prohibitive, which is the economic reason proxies survive.
The PD overlay. If the transition score led the analyst to notch the PD to 0.30%, the loan would return 7.88% and the relationship 10.98%, below the hurdle. This is the legitimate channel: the data change the bank’s view of credit risk, and price and capital follow.
A shadow carbon price. Some banks have tested an internal charge per tonne of financed emissions. At EUR 1 per tonne the charge is EUR 0.208m and the relationship falls to 11.87%; the break-even with the hurdle is EUR 0.94 per tonne. Any price resembling a market carbon price would end the relationship. The result shows why this instrument is little used for heavy industry: the client already pays for its emissions through the EU ETS, attribution is a sharing convention and not a cost borne by the bank, and the charge would push the bank out of exactly the sectors where transition finance is needed. Banks steer instead with sector intensity targets and client-level transition assessments (see E03 and E07).
The EUR 135m term loan is funded for five years; treasury charges 30 bp, EUR 0.405m a year, already deducted from the margin above. The undrawn EUR 100m commitment in the revolving facility consumes liquidity buffer and capital without appearing in financed emissions (see B03 and E04). Carbon data do not affect funding cost today. They affect eligibility: loans to Arvernia cannot enter the asset pool of Meridian’s own green bonds, which requires taxonomy or framework data on the use of proceeds, while a client’s inventory is not sufficient for that purpose (see C15).
Under IFRS 9 the loan is held at amortised cost; the bond holding is at amortised cost or fair value through other comprehensive income depending on the treasury’s business model. Carbon data matter for expected credit losses: forward-looking information must include climate-related factors where material, and a transition-driven notch can move an exposure towards stage 2. Financed emissions themselves are not an accounting item. They are a sustainability disclosure under ESRS E1 (as scope 3 category 15) and, for groups applying ISSB standards, under IFRS S2, with the industry breakdown that the December 2025 amendments made more flexible. The audit trail expected for these figures is converging with that of financial statements, and the statutory auditor tests it (see E05).
Financed emissions. Arvernia adds 242,088 tCO2e of scope 1 and 2 and 70,000 tCO2e of scope 3 to Meridian’s 2026 inventory, at data quality score 1 for scope 1 and 2. Meridian’s policy assigns score 2 to this client’s scope 3 because of the estimation the client itself declares, a conservative choice that goes beyond the PCAF table.
Facilitated emissions. 63,404 tCO2e for the 2026 bond, disclosed separately in the sustainability statement.
Pillar 3 ESG. In Template 1 the gross carrying amount of EUR 157.5m is reported on the line of Arvernia’s NACE sector (manufacture of cement, within section C), with the financed emissions and the share derived from company-specific reporting. The first reference date under the revised templates is 31 December 2026.
Green Asset Ratio. A general-purpose loan to a cement group counts in the numerator only for the share of Arvernia’s turnover or capex that is taxonomy-aligned, taken from its own Article 8 disclosure. The carbon file does not feed the ratio; the taxonomy file does (see A02 and E06).
Sector target. Meridian steers its cement book on physical intensity, in kg of CO2 per tonne, weighted by exposure. Arvernia enters at 550 kg per tonne gross scope 1. This metric is immune to EVIC and to prices, which is why banks chose physical intensities for the sectors where they exist.
| Risk | Control | Owner |
|---|---|---|
| Transcription or unit error (kt read as t) | Four-eyes check against page reference; plausibility test on intensity against sector range | ESG data team |
| Perimeter mismatch between carbon data and borrower group | Consolidation approach recorded; joint-venture flag; reconciliation with notes to the accounts | ESG data analyst, credit analyst |
| Stale data | Reporting year stored with each figure; ageing report; maximum lag in policy | Data steward |
| Vendor estimate taken for reported data | Reported or modelled flag loaded with every vendor figure; exception report | ESG data team |
| Inconsistent dates between emissions and EVIC | Engine rule pairing both at the client’s year-end | Finance sustainability reporting |
| Proxy misuse or unapproved factor | Hierarchy of options in policy; factor library under change control | Model validation |
| Wrong sector code or missing LEI | Booking controls; monthly unmatched-exposure report | Loan middle office |
| Double counting between financed and facilitated, or loan and bond | Separate ledgers; instrument-level identifiers | Finance sustainability reporting |
| Misleading use of figures in marketing | Review of client-facing ESG claims; mandatory statement of scope, year and metric | Compliance, DCM syndicate |
| Reliance on an unregistered ESG rating provider | Check of ESMA register at onboarding of a vendor | Procurement, Compliance |
| Loss of audit trail | Immutable log of changes with user, date and reason; archived source documents | Chief Data Office |
Red flags when reading a client’s carbon data.
Dated facts, as of November 2026.
What changed recently. Three shifts shape the desk’s work. The supply of mandatory corporate data has narrowed with the Omnibus, so banks rely more on bilateral collection and proxies for mid-caps. ESG rating providers have moved under supervision, which adds transparency on methods but does not make ratings converge. And bank disclosure has been rationalised: fewer taxonomy tables in Pillar 3, while financed emissions stay.
Open debates.
INTERACTIVE LAB — Check your understanding of carbon and ESG data
Twelve questions on scopes, intensities, PCAF attribution, facilitated emissions, proxies and ratings.
Q1. A cement producer buys electricity from the grid. The emissions of the power plants are, for the cement producer:
Answer: B. Purchased electricity is scope 2. Upstream extraction and transport of the fuels behind it go to scope 3 category 3.
Q2. Arvernia’s process emissions are 66.8% of scope 1. What follows?
Answer: C. Process CO2 comes from limestone chemistry, not from fuel; only less clinker or capture addresses it.
Q3. A company covers 100% of its electricity with guarantees of origin. Its location-based scope 2 is:
Answer: B. The location-based method applies the grid average whatever the contracts; only the market-based figure reflects them.
Q4. A 50% joint venture operated by the group and equity-accounted counts in the group’s scope 1, under operational control, for:
Answer: C. Operational control consolidates 100% of the emissions of sites the group operates.
Q5. Selling prices rise by 20% with unchanged production and emissions. Monetary intensity:
Answer: D. Intensity is divided by 1.2, a fall of 1 - 1/1.2 = 16.7%.
Q6. EVIC is:
Answer: A. Cash is not deducted, so that the attribution factors of all capital providers sum to 100% at most.
Q7. The share price of a borrower falls by 20%; its emissions, debt and the bank’s loan are unchanged. The bank’s financed emissions:
Answer: C. EVIC falls, the attribution factor rises. In the case the loan’s figure goes up by 13.6%.
Q8. Under PCAF Part B, the facilitated emissions of a bond are reported by a bookrunner:
Answer: B. Facilitated emissions are a flow measure, weighted at 33% and attributed by league-table credit.
Q9. A PCAF data quality score of 4 on a business loan means that emissions were:
Answer: C. Score 4 is the revenue-based economic estimate; production-based is score 3, verified reporting is score 1.
Q10. A client claims 0.9 Mt of avoided emissions. In the bank’s carbon file they are:
Answer: C. Standards require gross reporting; avoided emissions rest on a counterfactual and are disclosed apart.
Q11. According to Berg, Kölbel and Rigobon (2022), the largest source of ESG rating divergence is:
Answer: A. Measurement explains 56%, scope 38% and weights 6%.
Q12. Since 2 July 2026, an ESG rating provider operating in the EU must be:
Answer: B. Regulation (EU) 2024/3005 gives supervision to ESMA; existing providers had to apply by 2 November 2026.
Q13. Which metric would you use to compare a cement client with its sector decarbonisation pathway?
Answer: D. Sector pathways are expressed per unit of output; revenue intensity is polluted by prices.
Ferralia, a fictional steelmaker, produces 4.0 Mt of crude steel through the blast-furnace route with a scope 1 factor of 1.9 tCO2e per tonne. It buys 2,000 GWh of electricity; the grid factor is 0.25 tCO2e per MWh; half the volume is covered by zero-carbon contracts and the residual-mix factor is 0.35. Revenue is EUR 3,200m. Compute scope 1, both scope 2 figures, and the physical and monetary intensities on scope 1 and market-based scope 2.
Solution. Scope 1 = 4.0 × 1.9 = 7.60 Mt. Location-based scope 2 = 2,000 × 0.25 / 1,000 = 0.50 Mt. Market-based scope 2 = 2,000 × 50% × 0.35 / 1,000 = 0.35 Mt. Scope 1 and 2 = 7.95 Mt. Physical intensity = 7.95 / 4.0 = 1.99 tCO2e per tonne of steel. Monetary intensity = 7,950,000 / 3,200 = 2,484 tCO2e per EUR m. Scope 2 is 4.4% of the total, so the choice of method matters little here; it would matter greatly for an electric-arc producer.
Ferralia’s market capitalisation is EUR 1,800m, its total debt EUR 1,100m and its minority interests EUR 100m. Meridian holds a EUR 60m loan and is one of five equal bookrunners on a EUR 1,000m bond. (a) Compute the financed and facilitated scope 1 and 2. (b) The share price rises by 25%; recompute the financed emissions.
Solution. (a) EVIC = 1,800 + 1,100 + 100 = EUR 3,000m. Loan attribution = 60 / 3,000 = 2.00%; financed emissions = 2.00% × 7,950,000 = 159,000 tCO2e. Facilitated amount = 1,000 / 5 = EUR 200m; attribution = 200 / 3,000 = 6.67%; after the 33% weighting 2.20%; facilitated emissions = 2.20% × 7,950,000 = 174,900 tCO2e. The mandate carries more than the loan because the ticket is more than three times larger. (b) Market capitalisation becomes EUR 2,250m and EVIC EUR 3,450m; attribution = 60 / 3,450 = 1.739%; financed emissions = 138,261 tCO2e, a fall of 13.0% without any change at the blast furnaces.
Meridian lends EUR 12m to a private food processor with revenue of EUR 80m and total equity plus debt of EUR 48m. (a) Estimate the financed emissions with a revenue factor of 500 tCO2e per EUR m. (b) The client then provides its energy consumption: 60 GWh of natural gas (0.202 tCO2e per MWh) and 30 GWh of electricity (0.25 tCO2e per MWh). Recompute and give the proxy’s error. (c) The rest of the book is EUR 28m at score 3. Compute the weighted data quality score before and after.
Solution. (a) Attribution = 12 / 48 = 25% (book values, since the company is private). Estimated emissions = 80 × 500 = 40,000 tCO2e; financed = 10,000 tCO2e, score 4. (b) Gas: 60,000 MWh × 0.202 = 12,120 t. Electricity: 30,000 MWh × 0.25 = 7,500 t. Total 19,620 tCO2e; financed = 25% × 19,620 = 4,905 tCO2e, score 2. The proxy overstated by 10,000 / 4,905 - 1 = 104%. (c) Before: (12 × 4 + 28 × 3) / 40 = 3.30. After: (12 × 2 + 28 × 3) / 40 = 2.70. Two figures obtained from the client halve the reported tonnes and improve the score by 0.6.
1. Explain scopes 1, 2 and 3 to a client’s CFO in one minute. Scope 1 is what comes out of your own chimneys and exhausts. Scope 2 is what the power plants emit to produce the electricity and heat you buy. Scope 3 is everything else in your value chain, from your suppliers to the use of your products, in fifteen categories. The first two are under your control and well measured; the third is usually the largest and the least precise. Lenders look at scope 1 and 2 for targets and at the dominant scope 3 category of your sector for credibility.
2. Why does PCAF use EVIC and not enterprise value? The attribution factor must share 100% of a company’s emissions among all its providers of capital. Enterprise value deducts cash, so the sum of equity and debt holders’ shares could exceed 100%, and a cash-rich company could even have a tiny or negative denominator. EVIC keeps cash in. The price paid is volatility, since market capitalisation moves the factor every year. For private companies, book equity plus debt plays the same role.
3. Your financed emissions rose by 8% this year. How do you explain it to the board? I would decompose the change into four effects: clients’ real emissions, attribution (EVIC and leverage), exposure (new lending and repayments) and methodology or data quality changes. Only the first reflects the real economy, and the second often dominates in a year of falling equity markets. I would show the same analysis on physical intensity for the sectors with targets, where EVIC plays no role. The board should take decisions on the basis of the first and third effects.
4. Should avoided emissions be deducted from a client’s footprint? No. They depend on a reference scenario chosen by the company, they are claimed by several actors along the chain for the same product, and climate depends on gross emissions. Standards require separate reporting. I would still look at them, because they show whether revenue is shifting towards low-carbon products, which matters for the credit story. In a KPI I would prefer a direct measure such as the share of low-carbon products in sales.
5. Three ESG ratings disagree on your client. What do you put in the credit memo? I report the three, with provider, date and scale, and one line on what each measures: risk exposure, management quality, impact. I do not average them, because they are not estimates of the same quantity. I then give the bank’s internal transition score with its components, each traceable to the carbon file. If the disagreement points to a real issue, for instance a controversy one provider captured, I investigate that issue itself.
6. What does the EU ESG Ratings Regulation change for a bank? Providers must be authorised or recognised by ESMA and be transparent on methodology, including whether they rate risk, impact or both. A bank that uses ratings in client-facing material should check that the provider is on ESMA’s register after the transition. The bank’s internal scores used for its own credit decisions are outside the scope. Divergence between providers will remain, but it becomes easier to explain.
7. Why is a facilitated emissions figure weighted at 33% and reported only once? An arranger does not hold the exposure; it brings issuer and investors together and exits within days, so full attribution would equate a week of underwriting with years of lending. The 33% is a convention agreed after consultation, and remains contested. Reporting once reflects the flow nature of the business: capital markets revenue is also booked in the year of the deal. The consequence is a volatile figure that follows issuance volumes.
8. A mid-cap client refuses to fill in your ESG questionnaire. What do you do? Since the Omnibus, a company with up to 1,000 employees is protected by the value-chain cap, so I cannot require more than the voluntary standard contains. I would ask for the two or three figures that improve the estimate most: energy consumption by type and production volume. I would explain what is in it for the client: a better data score, a more accurate footprint than the sector proxy, and access to labelled products. If nothing comes, the engine applies the revenue proxy at score 4 and the file records the refusal.
Standards
Regulation
Academic and reference works