Every insurer, pension fund and asset manager in Europe now publishes numbers about the carbon content and the climate alignment of its portfolio. Very few of them compute those numbers alone. The positions sit with a custodian, the issuer data are bought from three or four specialised vendors, the methodologies come from standard setters, and the report is checked by an auditor and read by a supervisor. Somewhere between these parties, someone must map every security to an issuer, pick the right emissions figure, apply the right formula, reconcile the totals with the accounts and produce a document that survives a supervisory review. That is the job of the ESG data, analytics and reporting desk of a securities services division.
At Meridian Bank the desk sits inside Meridian Securities Services, next to custody and settlement (see D01), fund administration (see D02) and depositary (see D03). It is a young business: most banks built it between 2020 and 2023, when the EU Sustainable Finance Disclosure Regulation (SFDR) and the French Article 29 of the Energy-Climate Law (Article 29 LEC) required investors to publish portfolio-level indicators. It reuses two assets the division already owns: the client’s positions, which the custodian holds every day, and an industrial production line for regulated reports.
Clients. Three groups. Insurers and mutuals with a general account to report under Article 29 LEC and Solvency II; asset managers who need the SFDR periodic templates and the principal adverse impact (PAI) statement for their funds and their management company; pension funds and other asset owners who need a climate report for their board, their members or a net-zero alliance. Large clients buy modules; small clients buy a full outsourced report.
How the desk earns money. Revenues are fees, in four shapes. A set-up fee for onboarding (mapping the client’s holdings, agreeing methodologies, building templates), usually EUR 50,000 to 200,000 for a complex insurer. An annual subscription per report or per module (carbon metrics, alignment, taxonomy, PAI), which for a full Article 29 report of a mid-sized insurer is in the low to mid hundreds of thousands of euros. Per-fund charges for asset managers (a few thousand euros per fund per year for an SFDR template set). And daily monitoring of ESG investment rules, priced per portfolio. These orders of magnitude are internal benchmarks of the case, not market statistics: fees are private, and you will not find a published price list. What is public is the size of the raw material: Opimas estimated that the global ESG data market would exceed USD 2bn in 2024, after very fast growth in 2021 and 2022, and expected growth to fall to single digits from 2026 as regulatory pressure eased.
The team. A desk of 30 to 60 people at a large bank is organised around the production chain.
| Team | What it does | Typical output |
|---|---|---|
| Product and methodology | Chooses metrics, writes methodology notes | Methodology document |
| Data management | Buys, loads, maps and controls vendor data | Golden source, data-quality report |
| Quantitative analytics | Builds models: alignment, temperature, scenarios | Model documentation, validation |
| Reporting production | Runs the cycle, assembles reports | Draft report, templates |
| Compliance monitoring | Codes and checks investment rules | Daily alerts, breach log |
| Client service | Onboarding, questions, change requests | Service reviews, SLA reports |
Adjacent functions matter as much as the team. Custody and fund accounting deliver positions and market values. Procurement and legal negotiate data licences, which decide what the bank may redistribute to clients. Operational risk and model risk review the processes and the models. Compliance checks that the service does not become an unauthorised ESG rating activity. IT runs the platform, often a combination of an in-house data lake and a bought analytics engine.
A week on the desk in late May. Monday: vendor A has restated scope 3 data for 400 issuers; the data team quantifies the effect on twelve client portfolios and decides whether to freeze the previous version for reports already in review. Tuesday: a client’s auditor asks for the lineage of one figure in last year’s report, from the published number back to the vendor file and the position file; the analyst rebuilds it from the audit trail in two hours. Wednesday: quantitative analytics presents to the methodology committee a change in the temperature model, which would move three client portfolios by more than 0.1°C. Thursday: the monitoring team investigates an alert on a fund whose weighted carbon intensity has crossed its internal warning level after a rebalancing. Friday: the production team delivers the third draft of an insurer’s Article 29 report, five weeks before the 30 June filing date, and books a review call with the client’s chief investment officer and chief risk officer.
WHY IT MATTERS ON THE DESK — On this desk the product is a number that someone else signs. The client’s board signs the report, the supervisor reads it, an NGO may dissect it. Your value is that the number is right, explained and reproducible; your risk is that it is wrong in a way nobody noticed before publication.
Before carbon, a securities services data business is a regulatory reporting and investment analytics business. It existed long before ESG: insurers outsource Solvency II quantitative reporting templates, asset managers outsource performance measurement, fund administrators produce prospectus and periodic reports. The ESG desk inherits the economics and the controls of that older business.
IN ONE SENTENCE — A data and reporting service turns the client’s positions and third-party reference data into controlled metrics and documents that the client is legally responsible for publishing.
The service has four layers, and clients can buy any subset.
INTUITION — Think of a hospital laboratory. The doctor (the investor) takes the decision and signs the diagnosis. The laboratory owns the machines, the reagents and the procedures, and it guarantees that the blood test is right. Nobody expects the laboratory to treat the patient, but a wrong test result is the laboratory’s problem.
All portfolio analytics, green or not, rest on three joins.
Security to issuer. Each holding is identified by an ISIN. The ISIN must be linked to the legal entity that issued it, identified by its Legal Entity Identifier (LEI). Many corporate bonds are issued by financing vehicles (a “Finance BV” in the Netherlands, a “Capital Ltd” elsewhere), guaranteed by the parent.
Issuer to parent. Data vendors usually publish emissions at the level of the consolidated group, not the financing vehicle. The service must therefore climb the corporate tree to the ultimate parent or to the entity that publishes the data. Mapping is the single largest source of error in portfolio analytics: a bond mapped to its financing vehicle has no emissions, no revenues and no target.
Issuer to data. Each vendor uses its own entity identifiers. The service keeps a cross-reference table between LEIs and vendor identifiers, and a rule for conflicts.
Once the joins are done, any metric is a weighted sum. Let be the market value of holding and the value of the perimeter. The weight is . Most portfolio metrics are of the form , where is an issuer characteristic. This is why the choice of perimeter (what is in ) matters as much as the data.
INTERACTIVE LAB — Why the security-to-issuer join matters: one bond, two answers
See how mapping a bond to a financing vehicle instead of its parent changes a weighted portfolio metric, and how the error grows with the issuer’s intensity and weight.
| Parameter you can move | Range | Default |
|---|---|---|
| Portfolio value | 1000 to 20000 EUR m | 9000 EUR m |
| Holding of bonds issued by the financing vehicle | 0 to 3000 EUR m | 900 EUR m |
| Parent’s carbon intensity | 0 to 1500 tCO2e / EUR m revenue | 300 tCO2e / EUR m revenue |
| Weighted intensity of the rest of the portfolio, per euro of total portfolio | 0 to 400 tCO2e / EUR m | 159.5 tCO2e / EUR m |
| Result at default settings | Value |
|---|---|
| Portfolio metric, holding mapped to the parent | 189.5 tCO2e / EUR m |
| Portfolio metric, holding mapped to the vehicle (no data) | 159.5 tCO2e / EUR m |
| Understatement | 15.8 % |
Try it. Raise the parent’s intensity to 1,000: one mapping error now removes more than a third of the portfolio’s carbon intensity.
The default values anticipate the case in section 5: an oil and gas bond of EUR 900m held through a financing vehicle in a EUR 9bn portfolio. The error is not exotic. It is the first thing an experienced reviewer checks.
PITFALL — “Coverage” is often reported as the share of holdings with a data point. A holding mapped to the wrong entity can have a data point (zero emissions for the vehicle, if the vendor reports one) and count as covered. Coverage must be measured after mapping to the parent, and the share of holdings that inherit data from a parent must be disclosed.
A data and reporting service is a high fixed cost, low marginal cost business. The fixed cost is the platform (data lake, calculation engine, report generator), the methodology and validation team, and the enterprise data licences. The marginal cost per client is onboarding, the analyst hours in each reporting cycle, and any per-client licence uplift the vendors charge for redistribution.
Data licences deserve attention because they shape the product. A vendor licence typically distinguishes internal use (the bank uses the data for its own purposes), derived data (the bank may give clients metrics computed from the data, such as a portfolio WACI, but not the underlying issuer values) and redistribution (the bank may pass issuer-level values to clients, usually for a fee per client or with proof that the client holds its own licence). A client that wants to see the emissions of each issuer in its report needs a redistribution right; this is why many service providers ask their clients to bring their own vendor licence.
The risks are operational, not financial: wrong positions, wrong mapping, wrong data version, wrong formula, wrong template, late delivery. They translate into client compensation (if the contract says so), regulatory exposure for the client (a misstatement in a published report), reputational damage, and, for monitoring, a breach that the service should have detected. Contracts limit the provider’s liability, usually to a multiple of annual fees, and exclude it when the client supplied the faulty input. Section 6 shows how these risks enter capital.
The documents of the business are: a master services agreement with a service description and service-level agreement (SLA: delivery dates, accuracy, availability); a methodology document for each metric, version-controlled; a data licence schedule stating which vendor data the client may see; a data-quality report delivered with each production; a sign-off pack (the client approves methodology choices and the final report); and, for monitoring, a rules book translating the client’s investment guidelines into coded rules.
Three things change. The data are less standardised (two vendors give two emissions figures for the same company, see A03). They are often estimated rather than reported, especially for scope 3 and smaller issuers. And they feed legal documents that did not exist ten years ago and that name the investor’s targets, so an error is not only a wrong number but potentially a misleading statement.
The core metrics of a portfolio climate report come in three families.
Absolute and attributed emissions. Financed emissions apportion each issuer’s emissions to the investor in proportion to its share of the issuer’s financing. For listed equity and corporate bonds, the PCAF standard uses the enterprise value including cash (EVIC) as the denominator.
FORMULA
is financed emissions in tCO2e; the market value of the holding in issuer ; the issuer’s enterprise value including cash; the issuer’s emissions (scope 1 and 2, and scope 3 when reported separately); the carbon footprint in tCO2e per EUR m invested; the value of the perimeter in EUR m.
Intensity. The weighted average carbon intensity (WACI) is the portfolio-weighted average of issuers’ emissions per unit of revenue. It is the metric of the lecturer’s Helios case, step 7, and of TCFD-style reporting.
FORMULA
is the weight of issuer in the portfolio; its carbon intensity in tCO2e per EUR m of revenue; its emissions; its revenue in EUR m. In the Helios benchmark of five issuers, tCO2e per EUR m.
Roncalli shows why practice settled on the WACI: the revenue-based intensity of a portfolio computed as total financed emissions over total financed revenue is not additive across holdings, whereas the weighted average is. The price of additivity is that the WACI is not the carbon intensity of anything real; it is an index of exposure to carbon-intensive business models.
Forward-looking alignment. Footprints describe the past. Alignment metrics ask whether issuers’ emission paths are compatible with a temperature goal. They include the share of the portfolio with science-based targets, sector pathway comparisons for high-emitting sectors (tonnes per tonne of cement, grams per kWh), and the implied temperature rise (ITR), which converts an issuer’s projected emissions or targets into a warming figure. Roncalli’s participation, ambition and credibility (PAC) framework is a useful lens: participation is the past trend, ambition compares targets with a sector scenario, and credibility compares targets with the trend. Most ITR products are built on those three pillars, with different weights.
| Report | Who must publish | Desk deliverable |
|---|---|---|
| Article 29 LEC report | French investors, incl. insurers | Full report and annexes |
| SFDR entity PAI statement | EU managers and insurers (over 500 staff, or by choice) | PAI table, 30 June |
| SFDR product annexes | Article 8 and 9 products | Pre-contractual and periodic templates (see D02) |
| IFRS S2 / TCFD-style | Listed investors, voluntary users | Metrics and targets section |
| Solvency II inputs | EU insurers | Data for ORSA, SFCR, sustainability risk plan |
Article 29 LEC. Article 29 of the French Energy-Climate Law of 8 November 2019, codified in Article L.533-22-1 of the Monetary and Financial Code, and its implementing Decree 2021-663 of 27 May 2021 replaced the earlier Article 173-VI regime. Investors (asset managers, insurers, mutuals, pension institutions) publish annually an “ESG and climate” report on how they take environmental, social and governance criteria and climate and biodiversity risks into account. The decree requires a strategy of alignment with the temperature goals of the Paris Agreement, with quantitative greenhouse-gas reduction targets set every five years up to 2050 and a target for 2030; a strategy of alignment with long-term biodiversity goals; information on the integration of ESG risks, including physical and transition risks, into risk management; and the share of assets in EU-taxonomy-aligned activities and in fossil-fuel activities. The most demanding sections apply above EUR 500m of balance sheet or assets under management. The report is due within six months of the financial year end, so by 30 June for calendar-year entities, published on the website and sent to the supervisor (ACPR for insurers, AMF for asset managers), and centralised by ADEME on the Climate Transparency Hub.
SFDR. The entity-level PAI statement covers mandatory indicators including GHG emissions (PAI 1), carbon footprint (PAI 2), GHG intensity of investee companies (PAI 3) and exposure to fossil-fuel companies (PAI 4), under Delegated Regulation (EU) 2022/1288. The reform known as “SFDR 2.0” was proposed by the Commission on 20 November 2025; the Council agreed its negotiating mandate on 24 June 2026, the Parliament’s ECON committee adopted its report on 10 September 2026 and the plenary its mandate on 5 October 2026, and the first trilogue was scheduled for 22 October 2026; no agreement had been reached by mid-November 2026. The proposal replaces the de facto Article 8 and 9 labels with product categories. Application is expected around 2029, so the 2027 reporting cycle runs on the current templates.
IFRS S2 and TCFD. The TCFD disbanded in October 2023 after the ISSB published IFRS S1 and S2; the IFRS Foundation took over the monitoring of climate disclosures. IFRS S2 requires financed emissions for asset managers and insurers with investment activities. In November 2026 the IFRS Foundation reported that companies in 12 jurisdictions report on the basis of ISSB Standards for FY2025 and in 6 more for FY2026, with nearly 50 jurisdictions on the path to adoption. The EU uses its own standards (ESRS) instead.
Solvency II. Since 2 August 2022, Delegated Regulation (EU) 2021/1256 requires insurers to integrate sustainability risks into their risk management and investment policy (prudent person principle). Directive (EU) 2025/2, the Solvency II review, adds sustainability risk plans, with effect from 30 January 2027; EIOPA consulted on technical standards in December 2024, but the Commission deprioritised those Level 2 measures in October 2025, so firms apply the Level 1 requirements without final technical standards. On capital, EIOPA recommended on 7 November 2024 an additional charge for fossil-fuel assets, up to 17% on equities on top of the existing charge and up to 40% (multiplicative) on spread risk for bonds; the Commission decides whether to implement it. For the desk, Solvency II is a client of data: the insurer’s ORSA and its sustainability risk plan need the same issuer data as the Article 29 report.
CSRD after Omnibus I. Directive (EU) 2026/470, published on 26 February 2026, limits the CSRD to undertakings with more than 1,000 employees and EUR 450m of net turnover, with transposition by 19 March 2027. Fewer investee companies will publish audited emissions under ESRS, so investors will rely longer on vendor estimates. That is good business for data vendors and bad news for data quality.
Regulation (EU) 2024/3005 on ESG rating activities applies from 2 July 2026. Providers of ESG ratings in the EU must be authorised and supervised by ESMA; providers active on 2 January 2025 could continue if they notified ESMA by 2 August 2026 and applied for authorisation or recognition by 2 November 2026. The regulation covers ratings (opinions or scores on an entity’s ESG profile or risk built on a defined methodology and ranking system), not raw ESG data or emissions figures. For a bank’s securities services desk the question is precise: delivering portfolio metrics computed from vendor data to a client is a reporting service; selling the bank’s own issuer-level ESG scores to clients may be a rating activity. Meridian’s compliance function has drawn the line: the desk distributes metrics and vendor data under licence, and does not publish issuer scores of its own.
The PCAF Global GHG Accounting and Reporting Standard, Part A (financed emissions), was updated on 2 December 2025 with four new methods (use-of-proceeds structures, securitisations, sub-sovereign debt, optional undrawn commitments); the existing listed-equity and bond method based on EVIC was not changed. The PCAF data-quality score (1 = audited reported emissions, 5 = estimate from revenue and sector averages) is used by most Article 29 and PAI reports. The EU Climate Benchmarks (Delegated Regulation (EU) 2020/1818) give the reference trajectories used by many investors: a 7% average annual reduction of intensity, from a starting point 30% (Climate Transition Benchmark) or 50% (Paris-Aligned Benchmark) below the investable universe. The Net-Zero Asset Owner Alliance published the fifth edition of its target-setting protocol in March 2026, adding “transition targets” (share of financed emissions from companies with credible transition plans) alongside emission-reduction, sector, engagement and financing targets; the alliance then had 87 signatories. The Net Zero Asset Managers initiative relaunched on 25 February 2026 with more than 250 signatories and a revised, less prescriptive commitment.
A data and reporting mandate has two lifecycles: the sale and onboarding, which happens once, and the annual production cycle, which repeats. The flow below follows an insurer from the request for proposal to its first filed Article 29 report.
INTERACTIVE LAB — From request for proposal to the filed Article 29 report
Follow the actors, documents and controls of an ESG reporting mandate from tender to filing.
| # | Who | Does what | Deliverable |
|---|---|---|---|
| 1 | Insurer’s CIO and CRO | Issue a request for proposal for an outsourced Article 29, PAI and climate-risk data service | RFP with scope, perimeter and timetable |
| 2 | Securities services sales and product manager | Scope modules, check data licences, estimate effort and price | Proposal, fee grid, draft service description |
| 3 | Legal, procurement and outsourcing risk | Negotiate the master services agreement, liability cap and data licence schedule; client runs its outsourcing assessment | Signed MSA, SLA and licence schedule |
| 4 | Onboarding analyst and data management | Load positions from custody and client books, map ISINs to LEIs and ultimate parents | Mapping file, exception list |
| 5 | Methodology lead and client’s sustainability team | Agree perimeter, metrics, data waterfall, base year and target definitions | Signed methodology note |
| 6 | Quantitative analytics | Run the dry run on the 30 September positions, including alignment and temperature models | Dry-run results and data-quality report |
| 7 | Compliance monitoring team | Code the client’s ESG investment rules and start daily checks | Rules book, first alert log |
| 8 | Reporting production | Freeze year-end positions and the data vintage; compute all metrics | Frozen dataset, metric pack |
| 9 | Four-eyes reviewer and model risk | Reconcile totals with the accounts, check movers, validate model changes | Review checklist, sign-off |
| 10 | Client’s sustainability and finance teams | Draft the narrative, challenge figures, approve methodology choices | Approved draft report |
| 11 | Client’s board or risk committee | Approve targets and the report | Board minutes |
| 12 | Client’s investor relations and regulatory reporting | Publish on the website and transmit to the ACPR before 30 June | Published report, filing receipt |
| 13 | Client service and product manager | Run the post-mortem and the annual service review; agree changes for next cycle | Service review minutes, change requests |
Request for proposal. The insurer usually tenders the service when it outgrows spreadsheets or when its supervisor criticises its report. The ACPR’s reviews have been direct: in its analysis of the 2023 reports it found that none fully met the requirements for completeness, accuracy or precision, and that the standard reporting plan was not always followed. A client that has received such feedback buys reliability and audit trail more than metrics.
Scoping and licences. Before quoting, the product manager checks which vendor data the bank may redistribute to the client and whether the client holds its own licences. A proposal that promises issuer-level emissions in the annex without redistribution rights is a contractual problem waiting to happen.
Contract. For an insurer the service is an outsourcing in the sense of Solvency II and EIOPA’s guidelines on outsourcing; if the client considers it important, it must notify the ACPR and keep audit and access rights. The contract fixes the liability cap, the client’s responsibility for its own published statements, and the procedure for data restatements.
Onboarding and mapping. This is where most of the effort lies. The analyst maps every line to an issuer and a parent, flags financing vehicles, funds (which need look-through), sovereigns and sub-sovereigns, and real assets. The exception list is reviewed with the client.
Methodology note. The client decides; the bank proposes. The note fixes the perimeter (which assets are in the denominator), the data waterfall (which vendor first, which estimate next), the treatment of missing data, the base year and the target metrics. It is signed, because every later dispute will be settled by it.
Dry run. A first run on the 30 September positions, six to nine months before filing, shows the client its numbers early, tests the mapping and reveals the movers. Clients who skip the dry run discover in June that their footprint jumped because a vendor changed its scope 3 model.
Monitoring. In parallel, the ESG rules of the client’s investment guidelines are coded and checked daily (see D03 for the depositary’s version of this control for funds).
Production, review and approval. After year end, positions are frozen, the data vintage is frozen (the vendor files of a stated date), all metrics are computed and reviewed by a second analyst who reconciles the market values with the accounts and explains every large movement. The client’s teams write the narrative; its board approves the targets. Filing is the client’s act.
WHO DOES WHAT — The bank produces, the client publishes. Every page of the report belongs to the client, and the methodology choices are the client’s, documented in a signed note. The bank owns accuracy against that note, timeliness and the audit trail.
Question. What exactly does Garonne buy, and on which assets will the numbers be computed?
Data (case assumptions). Garonne Assurances is a fictional French mixed insurer based in Bordeaux: life savings, property and casualty, and health. Its balance sheet is EUR 21bn and its investments EUR 19.5bn: EUR 6.3bn of sovereign and supranational bonds, EUR 7.4bn of corporate bonds, EUR 1.6bn of listed equities held directly, EUR 1.9bn of real estate, EUR 1.5bn of investment funds and EUR 0.8bn of cash. Meridian has been its custodian for years. In September 2026 Garonne awards Meridian a mandate for four modules: the full Article 29 LEC report, the SFDR entity PAI statement, a Solvency II climate data pack for the ORSA and the sustainability risk plan, and daily monitoring of its ESG investment rules. The first filing is due on 30 June 2027, on the positions of 31 December 2026.
Garonne is far above the EUR 500m threshold, so every section of the decree applies: Paris alignment strategy with a 2030 target, biodiversity strategy, integration of ESG risks, taxonomy and fossil-fuel shares. Its previous reports, produced in-house, set a target to reduce the carbon footprint of its directly held corporate portfolio by 50% between 2019 and 2030. The base-year footprint, recomputed on the current methodology, is 210 tCO2e per EUR m invested.
The perimeter. The carbon metrics of the case are computed on the directly held corporate perimeter: corporate bonds and listed equities, EUR 9.0bn. Sovereign bonds are reported separately with a country intensity (PAI 15 in SFDR); funds need look-through to their holdings; real estate is measured in kgCO2e per square metre. Mixing these in one number is a classic error, because the denominators are not comparable.
For teaching purposes the corporate perimeter is represented by seven issuer lines. Each line stands for Garonne’s aggregated holdings in one fictional issuer, bonds and shares together.
| Issuer | Sector | Holding (EUR m) | EVIC (EUR bn) | Revenue (EUR bn) |
|---|---|---|---|---|
| Rhône Énergie | Integrated utility | 1,500 | 60 | 30 |
| Nordwind Energy | Oil and gas | 900 | 120 | 150 |
| Bastide Matériaux | Cement | 600 | 24 | 15 |
| Auriga Automobiles | Autos | 1,500 | 75 | 120 |
| Lyra Télécom | Telecoms | 2,000 | 80 | 40 |
| Maison Vesta | Consumer goods | 1,300 | 52 | 26 |
| Solenne Renouvelables | Renewables | 1,200 | 24 | 5 |
| Issuer | Scope 1+2 (ktCO2e) | Source | PCAF score |
|---|---|---|---|
| Rhône Énergie | 12,000 | Reported, verified | 1 |
| Nordwind Energy | 45,000 | Reported, unverified | 2 |
| Bastide Matériaux | 15,000 | Reported, verified | 1 |
| Auriga Automobiles | 3,000 | Reported, unverified | 2 |
| Lyra Télécom | 1,200 | Reported, verified | 1 |
| Maison Vesta | 1,560 | Vendor B estimate | 4 |
| Solenne Renouvelables | 250 | Reported, unverified | 2 |
Reading. Three of the seven lines (utility, oil and gas, cement) are a third of the money and, as we will see, most of the carbon. Two are held through structures that need care: Nordwind’s bonds are issued by Nordwind Finance BV, and Maison Vesta does not publish emissions.
Question. How sensitive are the reported figures to the two data decisions the onboarding analyst must take: the mapping of Nordwind’s financing vehicle and the source for Maison Vesta?
The data waterfall. Meridian’s standard waterfall, written into the methodology note, ranks sources: (1) emissions reported by the issuer and verified, taken from vendor A; (2) reported but unverified; (3) vendor B’s company-specific estimate, built from the issuer’s physical activity; (4) a sector proxy, revenue multiplied by an average sector emission factor. The PCAF scores follow the same logic, from 1 to 5. The waterfall is applied line by line, and each line keeps the score of the source actually used.
FORMULA
is the weighted PCAF data-quality score of the perimeter; the score of line (1 best, 5 worst); its weight; the coverage, the weight of the lines in the set of lines with an emissions figure.
For Maison Vesta, three choices exist: use vendor B’s estimate (1,560 kt, score 4), apply the sector proxy (revenue of EUR 26bn times a consumer-goods factor of 110 tCO2e per EUR m, so 2,860 kt, score 5), or exclude the line and report reduced coverage.
INTERACTIVE LAB — Garonne: mapping and data-waterfall choices and what they do to the report
Replay the two data decisions of the onboarding: map Nordwind’s bonds to the parent or to the financing vehicle, and choose the source for Maison Vesta.
| Parameter you can move | Range | Default |
|---|---|---|
| Nordwind Finance BV bonds mapped to | parent / vehicle | parent |
| Source for Maison Vesta | vendor B estimate / sector proxy / exclude | vendor B estimate |
| Result at default settings | Value |
|---|---|
| Financed emissions, scope 1+2 | 1,154.0 ktCO2e |
| Carbon footprint | 128.22 tCO2e / EUR m invested |
| WACI (re-weighted on covered lines) | 189.50 tCO2e / EUR m revenue |
| Weighted PCAF data-quality score (covered lines) | 1.833 1 to 5 |
| Coverage by value | 100.00 % |
Try it. Map Nordwind to the vehicle: coverage stays at 100% while financed emissions fall by 29%. Then exclude Maison Vesta: the data-quality score improves and the WACI rises.
Results. With the correct parent mapping and vendor B’s estimate, financed emissions are 1,154 ktCO2e, the footprint is 128.22 tCO2e per EUR m and the WACI 189.5 tCO2e per EUR m of revenue, with a weighted data-quality score of 1.833 and full coverage.
Reading the alternatives.
PITFALL — Exclusion is not neutral. Removing a line from the denominator implicitly assumes it has the average intensity of the rest. When missing data are concentrated in a low-carbon segment (consumer goods, services), exclusion overstates intensity; when they are concentrated in emerging-market industrials, it understates. Meridian’s note forbids exclusion above 5% of the perimeter and requires the proxy instead.
WHO DOES WHAT — The data manager owns the waterfall; the onboarding analyst owns the mapping file; the four-eyes reviewer runs the “large holding, zero emissions” control. The client signs the methodology note that contains the waterfall, so a later change of source is a change request, not a correction.
Question. What are Garonne’s footprint metrics, which holdings drive them, and what changes when scope 3 is added?
Calculation. Financed emissions apply the attribution factor to each issuer. For Rhône Énergie, of the company’s financing is Garonne’s, so Garonne carries kt. Doing the same for each line gives the table below. The WACI uses revenue intensities: Rhône Énergie emits tCO2e per EUR m of revenue (12,000 kt is 12,000,000 t, 30,000 is the revenue in EUR m), and its weight is .
| Issuer | Attribution | Financed (kt) | CI (t/EUR m rev.) | WACI points |
|---|---|---|---|---|
| Rhône Énergie | 2.50% | 300.0 | 400 | 66.67 |
| Nordwind Energy | 0.75% | 337.5 | 300 | 30.00 |
| Bastide Matériaux | 2.50% | 375.0 | 1,000 | 66.67 |
| Auriga Automobiles | 2.00% | 60.0 | 25 | 4.17 |
| Lyra Télécom | 2.50% | 30.0 | 30 | 6.67 |
| Maison Vesta | 2.50% | 39.0 | 60 | 8.67 |
| Solenne Renouvelables | 5.00% | 12.5 | 50 | 6.67 |
| Total | 1,154.0 | 189.50 |
INTERACTIVE LAB — Garonne: financed emissions, footprint and WACI by issuer, with and without scope 3
Compute the three footprint metrics of the case, see the contribution of each issuer, and measure what scope 3 does to the picture.
| Parameter you can move | Range | Default |
|---|---|---|
| Add scope 3 of Nordwind (360,000 kt) and Auriga (60,000 kt) | no / yes | no |
| Maison Vesta intensity used | 20 to 200 tCO2e / EUR m revenue | 60 tCO2e / EUR m revenue |
| Result at default settings | Value |
|---|---|
| Financed emissions | 1,154.0 ktCO2e |
| Carbon footprint | 128.22 tCO2e / EUR m invested |
| WACI | 189.50 tCO2e / EUR m revenue |
| Share of financed emissions from utility, oil and gas, cement | 87.7 % |
| Weight of those three lines | 33.3 % |
| Share of financed emissions from Nordwind | 29.2 % |
| WACI relative to the Helios 5-issuer benchmark (336.5) | -43.7 % |
Try it. Switch scope 3 on: Nordwind’s share of financed emissions doubles to about 60% and the WACI almost triples. Then try a Maison Vesta intensity of 110, the sector proxy.
Results. Financed emissions 1,154 ktCO2e; carbon footprint 128.22 tCO2e per EUR m invested; WACI 189.5 tCO2e per EUR m of revenue. The utility, oil and gas and cement lines are 33.3% of the value and 87.7% of financed emissions. Garonne’s WACI is 43.7% below the 336.5 of the Helios five-issuer benchmark, mainly because Garonne holds no issuer as intensive as Oil and Gas C (850) in a large weight.
Reading: three metrics, three rankings. Nordwind is the second contributor to financed emissions (337.5 kt, close behind Bastide’s 375 kt) but only the third contributor to the WACI (30 points), because its revenue is enormous relative to its emissions. Bastide and Rhône dominate the WACI. Solenne has the highest attribution factor (5%) yet almost no emissions. A portfolio manager steering the WACI will sell cement and utilities; one steering financed emissions will look at the oil major as much as at cement. Article 29 and the PAI statement ask for both, and the report must explain why they move differently.
Scope 3. With the scope 3 of Nordwind (mostly use of sold products) and Auriga (use of vehicles), financed emissions rise to 5,054 kt, more than four times the scope 1+2 figure, and Nordwind alone becomes 60.1% of the total. The WACI rises to 512.83. PAI 1 requires scope 3 to be reported, but scope 3 is double-counted across the value chain (Nordwind’s downstream emissions are Auriga’s customers’ emissions) and largely estimated. Meridian reports it separately, never added to the scope 1+2 target metric, and states its data-quality score.
WHY IT MATTERS ON THE DESK — The CIO will ask one question in the review: “why did our footprint change?” Your answer must split the movement into portfolio changes (trades), issuer changes (real emissions), data changes (new vintage, new estimates) and denominator changes (EVIC). The next stage shows why the last one matters most.
Question. Between the dry run and year end, equity markets rallied. How much of the change in Garonne’s financed emissions is real?
Mechanics. The attribution factor uses market values. If an issuer’s EVIC rises 20% and Garonne’s holding is unchanged in value (bonds do not follow equity), the attribution factor falls by , and so do financed emissions, with no tonne of CO2 avoided. Roncalli calls the related problem of financed emissions shrinking on paper an attribution effect; PCAF’s 2025 update responded to the industry’s discussion of “inventory fluctuation” with recommendations rather than a change of method.
FORMULA
and are financed emissions before and after; the change in issuers’ real emissions; the change in EVIC, holdings constant. When and the reported change is .
INTERACTIVE LAB — Garonne: separating real decarbonisation from the EVIC effect
Decompose a change in financed emissions into the real emissions change of issuers and the effect of a change in their enterprise value, holdings constant.
| Parameter you can move | Range | Default |
|---|---|---|
| Change in issuers’ EVIC | -30 to 40 % | 20 % |
| Change in issuers’ real emissions | -20 to 10 % | 0 % |
| Financed emissions at the start | 100 to 5000 ktCO2e | 1154 ktCO2e |
| Result at default settings | Value |
|---|---|
| Reported financed emissions | 961.67 ktCO2e |
| Reported change | -16.67 % |
| Real-economy change | 0.00 % |
| EVIC effect (reported minus real) | -16.67 % points |
Try it. Set the EVIC change to -25%, a market crash: financed emissions jump by a third while every issuer kept its emissions flat.
Result. A 20% rise in EVIC with flat real emissions cuts reported financed emissions from 1,154 to 961.67 kt, a 16.67% “reduction” that is entirely a valuation effect. If issuers also cut real emissions by 5%, the reported fall is 20.83%, of which 15.83 points are EVIC.
Reading. The footprint is a ratio of a physical quantity to a market price. Desks that report it without a decomposition let their clients claim decarbonisation in bull markets and suffer “recarbonisation” in crashes. Meridian’s report shows the four-way bridge of section 5.3 and, for targets, an EVIC-adjusted metric.
Question. Is Garonne on track for its 2030 target of −50% versus 2019, and does the answer survive the EVIC adjustment?
Data (case assumptions). Base-year footprint 210 tCO2e per EUR m (2019); target −50% in 2030, so 105. Average EVIC of the issuers in the perimeter is 25% higher than in 2019, holdings and real emissions held at their current levels.
Calculation. The linear target path loses tCO2e per EUR m each year from 2019, so in 2026 (seven years in) the path is at . The reported 2026 footprint is 128.22: 14.96 below the path. Now restate the footprint at 2019 EVIC: , which is 17.10 above the path. On a reported basis Garonne is ahead; on a constant-valuation basis it is behind.
FORMULA
is the footprint restated at base-year enterprise values; the average growth of EVIC since the base year (1.25 here); the constant annual rate of change needed to reach the target in year from year .
INTERACTIVE LAB — Garonne: reported and EVIC-adjusted footprint against the 2030 target path
Test the portfolio against a linear 2019-2030 target path on the reported and on the EVIC-adjusted footprint, and compute the annual reduction still needed.
| Parameter you can move | Range | Default |
|---|---|---|
| Base-year footprint (2019) | 100 to 400 tCO2e / EUR m | 210 tCO2e / EUR m |
| Target reduction by 2030 | 20 to 70 % | 50 % |
| Reported footprint in 2026 | 50 to 300 tCO2e / EUR m | 128.22 tCO2e / EUR m |
| EVIC growth since 2019 | -20 to 60 % | 25 % |
| Result at default settings | Value |
|---|---|
| 2030 target | 105.00 tCO2e / EUR m |
| Linear path in 2026 | 143.18 tCO2e / EUR m |
| Reported footprint minus path (negative = ahead) | -14.96 tCO2e / EUR m |
| EVIC-adjusted footprint | 160.28 tCO2e / EUR m |
| Adjusted footprint minus path | 17.09 tCO2e / EUR m |
| Annual change needed 2026-2030, reported basis | -4.87 % |
| Annual change needed 2026-2030, adjusted basis | -10.03 % |
Try it. Set EVIC growth to 12%: the adjusted footprint lands almost exactly on the path. That is the market rally Garonne’s progress silently relies on.
Results. Target 105; path value in 2026 143.18. Reported basis: 14.96 below the path, and a further 4.87% a year is enough to reach the target by 2030. Adjusted basis: 160.28, 17.10 above the path, and the portfolio would need to cut its footprint by 10.04% a year for four years, faster than the 7% a year of the EU climate benchmarks.
Reading. Article 29 asks for the target and its monitoring; it does not prescribe an EVIC adjustment. Meridian shows both figures and recommends that Garonne’s narrative say which share of its progress comes from valuation. This is not pedantry: the 2025 analysis of Article 29 reports by the Climate Transparency Hub found that 37% of the actors it reviewed showed no explicit monitoring of their targets, and that only 15% explained the causes of changes in emissions. A report that explains its bridge stands out.
PITFALL — Do not “fix” the EVIC effect by switching the target metric to the WACI mid-way. Revenue intensities have their own denominator problem (commodity prices move revenues, see A03), and a change of metric after the base year breaks the comparability that the five-year target cycle of the decree is meant to ensure.
Question. What temperature does Garonne’s portfolio imply, and how much does the answer depend on the aggregation method and on trusting issuers’ targets?
Method (case convention). Real ITR products are proprietary and differ widely. The case uses a transparent stylised version of the regression approach behind the CDP-WWF temperature rating: an issuer’s ambition is summarised by its linear annual reduction rate (LAR) of scope 1+2 emissions, and mapped to a temperature by linear interpolation between two anchors. A LAR of 4.2% a year, the minimum absolute-contraction pace that SBTi associates with 1.5°C, maps to 1.5°C; a LAR of zero maps to 3.2°C, the case’s convention for a no-reduction path. Faster reductions are floored at 1.5°C.
FORMULA
is the implied temperature of issuer in °C; the temperature of a no-reduction path (3.2°C); the issuer’s linear annual reduction rate in % per year; 4.2 the rate associated with 1.5°C.
Two LARs exist for each issuer, which is Roncalli’s PAC framework in miniature. The target-based LAR measures ambition (what the issuer promises). The trend-based LAR measures participation (what it has done over the last five years). When the two diverge, credibility is in question.
| Issuer | LAR targets (%/yr) | LAR trend (%/yr) |
|---|---|---|
| Rhône Énergie | 2.6 | 1.8 |
| Nordwind Energy | 0.8 | 0.2 |
| Bastide Matériaux | 2.0 | 1.2 |
| Auriga Automobiles | 2.4 | 2.0 |
| Lyra Télécom | 3.5 | 3.0 |
| Maison Vesta | 1.5 | 0.5 |
| Solenne Renouvelables | 4.2 | 4.5 |
Aggregation. The temperature-scoring method published by CDP and WWF and used by SBTi offers several portfolio aggregations. Two are enough to see the issue. WATS (weighted average temperature score) weights issuers by portfolio weight; ECOTS (enterprise-value-owned emissions temperature score, in the spirit of PCAF) weights them by financed emissions.
FORMULA
is the portfolio weight; the financed emissions of line and their total.
INTERACTIVE LAB — Garonne: issuer and portfolio implied temperature, targets against trend
Map issuers’ reduction rates to implied temperatures and aggregate them by portfolio weight and by financed emissions.
| Parameter you can move | Range | Default |
|---|---|---|
| Reduction rate used | targets / trend | targets |
| Temperature of a no-reduction path | 2.5 to 4 °C | 3.2 °C |
| Result at default settings | Value |
|---|---|
| Portfolio ITR, weighted by value (WATS) | 2.147 °C |
| Portfolio ITR, weighted by financed emissions (ECOTS) | 2.442 °C |
| Nordwind ITR | 2.876 °C |
| Share of value at 1.5°C | 13.3 % |
Try it. Switch to the trend basis: every issuer except Solenne warms up, and the portfolio moves by about a quarter of a degree under both aggregations.
Results. On targets, the portfolio ITR is 2.147°C weighted by value and 2.442°C weighted by financed emissions. On trends it is 2.377°C and 2.730°C. Nordwind maps to 2.876°C on its targets and 3.119°C on its trend. Only Solenne (13.3% of value) is at 1.5°C.
Reading.
NOTE — Article 29 asks for an alignment strategy and quantitative targets, not for an ITR. Many French reports nevertheless show one. The ACPR’s reviews of Article 29 reports have criticised their lack of precision, a criticism that applies with special force to model outputs such as temperatures. The 2025 Climate Transparency Hub analysis also found that at least 20% of actors used a fossil-fuel share definition different from the regulatory PAI 4 method: methodological diversity is the norm, which makes documentation the product.
Question. Garonne’s investment committee wants the WACI 20% lower within a year, without new exclusions. What does the cheapest tilt look like, and what does it cost?
Method. This is exactly step 7 of the Helios case. Moving weight from a high-intensity issuer to a low-intensity issuer changes the WACI linearly.
FORMULA
is the weight moved from to ; and their intensities; the target reduction; the starting WACI. In the Helios benchmark, moving weight from Oil and Gas C (850) to Renewable E (40) cut the WACI by 8.1 points per point of weight, so 12.5 points of weight removed 101 points of WACI.
Calculation. Here is Bastide (1,000) and is Solenne (50): each point of weight moved removes 9.5 points of WACI. The target is , a reduction of 37.9 points, so points, or EUR 359.05m sold of Bastide and bought of Solenne. Bastide’s weight falls from 6.67% to 2.68%. With Bastide bonds yielding 4.10% and Solenne bonds 3.55% (case assumptions), the yield give-up is 55 bp on EUR 359.05m, EUR 1.975m a year, or 2.19 bp on the whole perimeter. Because Bastide is also the most EVIC-intensive line (625 tCO2e per EUR m of EVIC against 10.42 for Solenne), financed emissions fall by 220.66 kt to 933.34 kt, a footprint of 103.70.
INTERACTIVE LAB — Garonne: the WACI tilt from cement to renewables and its yield cost
Reproduce the Helios step-7 logic on Garonne’s portfolio: size the weight to move from a high- to a low-intensity issuer to reach a WACI target, and measure the yield given up and the effect on financed emissions.
| Parameter you can move | Range | Default |
|---|---|---|
| Target WACI reduction | 0 to 33 % | 20 % |
| Yield of the bonds sold (Bastide) | 2 to 7 % | 4.1 % |
| Yield of the bonds bought (Solenne) | 2 to 7 % | 3.55 % |
| Result at default settings | Value |
|---|---|
| Target WACI | 151.60 tCO2e / EUR m |
| Weight moved | 3.99 % points |
| Amount moved | 359.05 EUR m |
| Bastide weight after the tilt | 2.68 % |
| Income given up per year | 1.975 EUR m |
| Yield give-up on the perimeter | 2.19 bp |
| Financed emissions after the tilt | 933.33 ktCO2e |
| Footprint after the tilt | 103.70 tCO2e / EUR m |
Try it. Push the target to 33%: the whole Bastide line is almost sold. Beyond 33.4% this single pair cannot deliver the target and the tilt must involve Rhône Énergie.
Reading. The tilt is cheap (2.19 bp) and spectacular: −20% WACI and a footprint of 103.70, below the 2030 target of 105 on the reported basis. It is also, in Roncalli’s words, a disguised form of divestment: Bastide’s emissions do not change; someone else holds the bonds. Roncalli’s chapter 11 shows the same mechanics in an optimiser: a minimum-tracking-error portfolio under a carbon constraint drops or underweights the worst-in-class issuers first. Three observations complete the picture. First, the tilt concentrates credit risk on fewer issuers and lengthens duration if renewables issue longer (in Helios the duration drifted from 6.42 to 6.58). Second, Garonne’s Article 29 report must describe the levers used; a footprint reduced by reallocation reads differently from one reduced by issuers’ decarbonisation. Third, the new NZAOA protocol’s transition targets were designed precisely to reward financing companies that transition rather than selling them; on that metric, selling Bastide (a cement company with a verified decarbonisation plan, case assumption) would be a step backwards.
Question A. Garonne’s investment guidelines, coded by the monitoring team, contain a hard limit of 200 on the WACI of the corporate perimeter and a warning level at 190. A portfolio manager wants to invest EUR 150m of new cash in Nordwind bonds. What does the pre-trade check return?
Calculation. The WACI after a cash-funded purchase of in issuer is . For Nordwind, : above the warning level, below the limit. The trade is allowed, the alert is logged, and the portfolio manager must justify it to the risk function. The same EUR 150m in Bastide would give 202.79 and be blocked.
INTERACTIVE LAB — Garonne: pre-trade check of the WACI limit
Compute the post-trade WACI of a cash-funded purchase and the status returned by the compliance engine (0 pass, 1 warning, 2 blocked).
| Parameter you can move | Range | Default |
|---|---|---|
| Issuer bought | Nordwind / Bastide / Rhone / Lyra / Solenne | Nordwind |
| Amount bought | 0 to 1000 EUR m | 150 EUR m |
| Warning level | 150 to 250 tCO2e / EUR m | 190 tCO2e / EUR m |
| Hard limit | 150 to 260 tCO2e / EUR m | 200 tCO2e / EUR m |
| Upward restatement of Bastide’s emissions by the vendor (no trade) | 0 to 50 % | 30 % |
| Result at default settings | Value |
|---|---|
| Post-trade WACI | 191.31 tCO2e / EUR m |
| Status (0 pass, 1 warning, 2 blocked) | 1 |
| Maximum amount before the hard limit | 945.0 EUR m |
| WACI after the restatement alone (passive breach test) | 209.50 tCO2e / EUR m |
Try it. Buy Bastide: the headroom is only EUR 118m. A limit on an average is consumed fastest by the issuers furthest above it.
The monitoring engine needs the same data as the report, every day, with the same mapping. A data vintage change can create a breach without any trade: if vendor A restates Bastide’s emissions upwards by 30%, the WACI rises by points to 209.5 and crosses the limit overnight. The rules book must therefore say whether a “passive” breach caused by data must be cured, and in what time (see D03 for the equivalent in fund depositary oversight).
Question B. Over the first nine months of 2026, Garonne’s corporate portfolio returned 4.27% against 4.32% for its benchmark. How much of the 5 bp under-performance comes from the carbon tilt?
Method. A Brinson-Fachler attribution with two buckets, high-carbon (utilities, oil and gas, materials) and low-carbon (the rest), splits the active return into allocation (being under- or overweight a bucket), selection (choosing better or worse bonds inside it) and interaction.
FORMULA
and are the weights of bucket in the portfolio and the benchmark; and their returns; the benchmark’s total return; , , the allocation, selection and interaction effects, which add up to the active return.
Data (case assumptions). High-carbon bucket: portfolio weight 33.33%, benchmark 45%; benchmark return 5.2%, portfolio 5.0%. Low-carbon bucket: benchmark return 3.6%, portfolio 3.9%.
INTERACTIVE LAB — Garonne: attribution of active return with a carbon tilt
Split the active return of a carbon-tilted portfolio into allocation, selection and interaction effects across a high-carbon and a low-carbon bucket.
| Parameter you can move | Range | Default |
|---|---|---|
| Portfolio weight, high-carbon bucket | 0 to 100 % | 33.33 % |
| Benchmark weight, high-carbon bucket | 0 to 100 % | 45 % |
| Benchmark return, high-carbon | -10 to 15 % | 5.2 % |
| Benchmark return, low-carbon | -10 to 15 % | 3.6 % |
| Portfolio return, high-carbon | -10 to 15 % | 5.0 % |
| Portfolio return, low-carbon | -10 to 15 % | 3.9 % |
| Result at default settings | Value |
|---|---|
| Benchmark return | 4.320 % |
| Portfolio return | 4.267 % |
| Allocation effect (carbon tilt) | -18.67 bp |
| Selection effect | 7.50 bp |
| Interaction effect | 5.84 bp |
| Active return | -5.34 bp |
Try it. Set the high-carbon benchmark return equal to the low-carbon one: the allocation effect disappears. The tilt only costs or earns when the two buckets diverge.
Results. Benchmark 4.32%, portfolio 4.267%, active return −5.34 bp: allocation −18.67 bp, selection +7.5 bp, interaction +5.84 bp.
Reading. In a period when high-carbon credit outperformed, underweighting it cost Garonne about 19 bp, partly recovered by good bond selection. The attribution lets the investment committee see the price of its climate policy separately from the skill of its managers. Two cautions: an attribution on two buckets is crude, and a factor model (carbon intensity as a factor alongside duration, spread and sector) gives a cleaner reading; and one year says nothing about the long-run cost, which academic work finds small for moderate decarbonisation but which depends on the period.
Question. Why does Garonne outsource, and at what price does the mandate pay for Meridian?
Garonne’s side (case assumptions). Producing the report in-house would require 3 FTE at EUR 110,000 fully loaded (EUR 330,000), vendor data licences of EUR 280,000 a year and a calculation and reporting tool at EUR 120,000 a year: EUR 730,000. Meridian’s fee for the four modules is EUR 480,000 a year: EUR 300,000 for the Article 29 report, EUR 60,000 for the PAI statement, EUR 60,000 for the Solvency II data pack and EUR 60,000 for daily monitoring, plus a one-off set-up fee of EUR 150,000. The annual saving is EUR 250,000, 34% of the in-house cost. Garonne also buys an audit trail, a methodology reviewed across many clients and a team that will not resign in May. It keeps the responsibility for the report and must retain enough expertise to challenge it, which Solvency II outsourcing rules require.
Meridian’s side (case assumptions). The platform costs EUR 9m a year (methodology, quantitative analytics, data management, enterprise licences, IT run and amortisation). Each additional client costs EUR 150,000 a year (analyst time in the cycle, per-client licence uplifts, client service). The business plan counts on 45 clients of Garonne’s size.
FORMULA
is the average annual cost per client with clients; the fixed platform cost (EUR 9m); the marginal cost per client (EUR 150,000); the annual fee per client (EUR 480,000); the break-even number of clients.
INTERACTIVE LAB — Build against buy: the client’s saving and the provider’s scale
Compare the client’s in-house cost with the fee, and show how the provider’s average cost per client falls with the number of clients.
| Parameter you can move | Range | Default |
|---|---|---|
| Number of clients on the platform | 10 to 100 | 45 |
| Fixed platform cost | 3000 to 20000 EUR k | 9000 EUR k |
| Marginal cost per client | 50 to 400 EUR k | 150 EUR k |
| Annual fee per client | 100 to 1000 EUR k | 480 EUR k |
| Client’s in-house cost | 200 to 2000 EUR k | 730 EUR k |
| Result at default settings | Value |
|---|---|
| Client’s annual saving | 250 EUR k |
| Saving as share of in-house cost | 34.2 % |
| Provider’s average cost per client | 350.0 EUR k |
| Provider’s cost-income ratio | 72.92 % |
| Break-even number of clients | 27.27 clients |
Try it. Lower the client count to 25: the average cost exceeds the fee. Then raise the fee to 730, the client’s own cost: the business works with 16 clients, but no client would sign.
Results. Garonne saves EUR 250,000 a year (34.2%). At 45 clients Meridian’s average cost per client is EUR 350,000, a cost-income ratio of 72.92%; break-even is 27.27 clients. At 30 clients the cost-income ratio is 93.75%.
Reading. The fee is set between two walls: the client’s in-house cost above, the provider’s average cost below. Scale decides everything. This is why banks without a large custody base often partner with specialist vendors instead of building, and why the fee is usually bundled with custody: the custodian already holds the positions, which removes the costliest step of onboarding. It is also why vendors’ licence policies matter strategically: a vendor that raises redistribution fees moves the marginal cost and therefore the break-even.
| Item | Value | Source |
|---|---|---|
| Corporate perimeter | EUR 9,000m | 5.1 |
| Financed emissions, scope 1+2 | 1,154 ktCO2e | 5.3 |
| Carbon footprint | 128.22 tCO2e / EUR m | 5.3 |
| WACI | 189.5 tCO2e / EUR m rev. | 5.3 |
| WACI vs Helios benchmark (336.5) | −43.7% | 5.3 |
| Weighted PCAF score | 1.833 | 5.2 |
| Financed emissions, Nordwind mapped to vehicle | 816.5 ktCO2e (−29%) | 5.2 |
| WACI if Maison Vesta excluded | 211.36 | 5.2 |
| Financed emissions with scope 3 (two issuers) | 5,054 ktCO2e | 5.3 |
| EVIC +20%, flat emissions | 961.67 ktCO2e (−16.67%) | 5.4 |
| 2030 target / 2026 linear path | 105 / 143.18 | 5.5 |
| EVIC-adjusted footprint | 160.28 | 5.5 |
| Annual reduction needed, reported / adjusted | −4.87% / −10.04% | 5.5 |
| ITR on targets, WATS / ECOTS | 2.147°C / 2.442°C | 5.6 |
| ITR on trends, WATS / ECOTS | 2.377°C / 2.730°C | 5.6 |
| Tilt for −20% WACI | EUR 359.05m, 3.99 points | 5.7 |
| Yield give-up | 2.19 bp, EUR 1.975m a year | 5.7 |
| Post-trade WACI, EUR 150m Nordwind | 191.31 (warning) | 5.8 |
| Attribution: allocation / active | −18.67 bp / −5.34 bp | 5.8 |
| Client saving | EUR 250k (34.2%) | 5.9 |
| Provider cost-income at 45 clients | 72.92% | 5.9 |
| Break-even clients | 27.27 | 5.9 |
A data and reporting service lends nothing, holds no client asset on its balance sheet and takes no market position. Its economics are fee revenue, operating cost, operational-risk capital and contractual liability. This section applies Meridian’s framework to the Garonne mandate.
IN ONE SENTENCE — The only regulatory capital the mandate consumes is operational-risk capital, and under CRR3 that capital is proportional to the revenue the mandate brings, not to its risk.
Since 1 January 2025, CRR3 computes operational-risk capital with a single standardised approach: the business indicator component (BIC) applies marginal coefficients of 12%, 15% and 18% to slices of the bank’s business indicator, in which fee and commission income enters the services component. The EU set the internal loss multiplier to 1, so the bank’s loss history does not change the requirement. Module D02 derives Meridian’s figures: with a business indicator of about EUR 20bn, the BIC is EUR 2.97bn and the marginal coefficient 15%; allocating at that marginal coefficient (the course convention, see E03) and 13% of RWA gives EUR 0.24375 of capital per euro of fee revenue. We use the same convention here.
FORMULA
is the operational-risk capital allocated to the mandate; Meridian’s allocated BIC coefficient (its marginal coefficient, 15%); the mandate’s annual fee revenue; 12.5 converts own funds into RWA; 13% is Meridian’s capital allocation on RWA.
There is no credit exposure in this mandate: fees are invoiced quarterly in advance, and Garonne’s custody overdraft is allocated to the custody relationship (see D01). The platform itself raises a separate question. Software developed in-house is an intangible asset, which the CRR deducts from CET1; Delegated Regulation (EU) 2020/2176, in force since December 2020, exempts from deduction software that is prudently amortised over at most three years, with the exempted part risk-weighted. A EUR 12m capitalised platform therefore costs capital either through deduction or through RWA, and that cost belongs to the platform’s fixed cost , not to any one client.
Assumptions. Revenue: the EUR 480,000 annual fee of section 5.9 (the set-up fee is treated in 6.4). Operating costs: the average cost per client at 45 clients, EUR 350,000, a cost-income ratio of 72.92%, above Meridian’s standard 60% for fee businesses. The difference is real and explained: a data business pays vendors for licences that a custody business does not, and its production is labour-intensive in May and June. Expected operational losses: 2% of revenue, the convention of D02 (compensation, rework, write-offs). Tax 25%.
INTERACTIVE LAB — RAROC of the Garonne mandate: regulatory view, liability view and scale
Compute the RAROC of a data and reporting mandate with operational-risk capital, compare it with a capital equal to the contractual liability cap, and show how it depends on the number of clients sharing the platform.
| Parameter you can move | Range | Default |
|---|---|---|
| Annual fee | 100 to 1000 EUR k | 480 EUR k |
| Number of clients on the platform | 20 to 100 | 45 |
| Fixed platform cost | 3000 to 20000 EUR k | 9000 EUR k |
| Marginal cost per client | 50 to 400 EUR k | 150 EUR k |
| Expected operational losses | 0 to 10 % of revenue | 2 % of revenue |
| Allocated BIC coefficient | 12 to 18 % | 15 % |
| Liability cap, multiple of annual fees | 0.5 to 5 x | 1 x |
| Result at default settings | Value |
|---|---|
| Operating costs allocated | 350.00 EUR k |
| Expected operational losses | 9.60 EUR k |
| Risk-adjusted profit before tax | 120.40 EUR k |
| Risk-adjusted profit after tax | 90.30 EUR k |
| Operational-risk RWA | 900.00 EUR k |
| Allocated capital (13% of RWA) | 117.00 EUR k |
| RAROC, regulatory capital | 77.18 % |
| RAROC, capital equal to the liability cap | 18.81 % |
Try it. Lower the number of clients from 45 to 30: the regulatory RAROC falls from 77% to 13%, close to the 12% hurdle, without any change in risk. At 28 clients the mandate loses money.
Line by line. Revenue EUR 480,000. Operating costs EUR 350,000. Expected operational losses EUR 9,600. Risk-adjusted profit before tax EUR 120,400; after tax EUR 90,300. Operational-risk RWA EUR 900,000; allocated capital EUR 117,000. RAROC 77.18%, more than six times the hurdle. With a capital equal to the liability cap of one year of fees (EUR 480,000), the RAROC is 18.81%.
Why this RAROC differs in nature from a lending RAROC.
What the ESG content changes. Against a plain regulatory reporting service (Solvency II templates from the same positions), the ESG layer adds three things. It adds cost: vendor licences and analyst time on methodological judgement. It adds liability of a new kind: a misstated footprint or temperature is not measured by a tolerance threshold like a NAV error, and it can fuel a greenwashing complaint against the client, who may then turn to the provider. And it adds model risk: ITR and alignment models are models in the sense of the bank’s model risk policy and need validation, which is part of the fixed cost .
PITFALL — A 77% RAROC does not mean the mandate is well priced. It means the capital model sees almost no risk. The real tests are the cost-income ratio at a realistic client count and the RAROC on a liability-based capital; a desk that prices to the regulatory RAROC will under-price its liability.
There is no funding to transfer-price: the service has no assets, and fees are paid in advance, which creates a small contract liability (deferred income) of one quarter of fees, about EUR 120,000, worth a negligible FTP credit. The liquidity contribution of the relationship comes from Garonne’s custody cash balances, which Treasury values as operational deposits and which D01 attributes to custody. The data mandate’s relationship value is indirect but important: it makes the custody mandate stickier, because switching custodian would now also mean rebuilding the climate report.
Fees are revenue from contracts with customers under IFRS 15. The annual subscription is a series of services transferred over time and is recognised as the service is rendered, typically on a straight-line basis over the reporting cycle. The set-up fee of EUR 150,000 is the classic IFRS 15 question: onboarding (mapping, templates) does not transfer a distinct service to Garonne, which cannot benefit from the mapping file on its own, so the fee is deferred and recognised over the expected period of service, including likely renewals. Onboarding costs that generate resources used in future performance can be capitalised as costs to fulfil a contract. The platform is an intangible asset under IAS 38 when development costs meet the capitalisation criteria (technical feasibility, intention and ability to use, probable future benefits, reliable measurement), amortised over its useful life; research and methodology work are expensed. Vendor data licences are expensed over the licence period, unless they meet the definition of an intangible asset. A probable compensation payment after a reporting error is provisioned under IAS 37.
The service has almost no footprint in Meridian’s own sustainability indicators, and that is worth understanding precisely.
The real link between this business and Meridian’s risk profile is liability and reputation. A securities services division that computes the climate reports of 45 insurers holds, in effect, a concentration of methodological risk: a systematic error repeats across all clients at once.
| Risk | Example | Control | Owner |
|---|---|---|---|
| Mapping error | Bond mapped to financing vehicle | Large holding with zero emissions reviewed | Data management |
| Data vintage | Vendor restatement mid-cycle | Frozen vintage, restatement log | Reporting production |
| Perimeter error | Funds or sovereigns mixed in | Perimeter reconciled to accounts | Four-eyes reviewer |
| Formula error | WACI re-weighted wrongly | Independent recalculation | Quantitative analytics |
| Model risk | ITR anchors changed silently | Model inventory, validation | Model risk |
| Licence breach | Issuer data sent without rights | Licence schedule in report templates | Product, legal |
| Unauthorised rating | Own issuer scores distributed | Compliance review of outputs | Compliance |
| Passive breach | Data change crosses a limit | Rules book cure period | Monitoring team |
| Late delivery | Report after 30 June | Production calendar, escalation | Client service |
| Client misstatement | Narrative contradicts figures | Sign-off pack, consistency check | Client, reviewer |
Two of these deserve a comment. The perimeter error is the most frequent finding of supervisors: a footprint per euro invested whose denominator includes sovereign bonds or cash looks better than it is, because those assets carry no corporate emissions. The model risk of temperature metrics is underestimated: they are presented as data, while they are model outputs that depend on scenario, anchor and aggregation choices, as section 5.6 showed.
Red-flag checklist for a reviewer of a portfolio climate report.
| Data point | Value | Date and source |
|---|---|---|
| Global ESG data market | Over USD 2bn | 2024 estimate, Opimas |
| ESG data spending growth | Single digits from 2026 | Opimas, October 2025 |
| Insurer Article 29 submissions collected by ACPR | 228 | 2025 reports, Climate Transparency Hub |
| Average taxonomy alignment in Article 29 reports | About 5% | 2025 reports, Climate Transparency Hub |
| Actors without explicit target monitoring | 37% | 2025 reports, Climate Transparency Hub |
| NZAOA signatories | 87, over USD 9trn | March 2026 |
| NZAM signatories at relaunch | Over 250 | 25 February 2026 |
| ESG Ratings Regulation application | 2 July 2026 | Regulation (EU) 2024/3005 |
| Solvency II sustainability risk plans | From 30 January 2027 | Directive (EU) 2025/2 |
What changed recently. Four movements in 2025 and 2026 reshape the desk’s work. First, the simplification wave: Omnibus I narrowed the CSRD to the largest companies, so fewer investees will publish audited ESRS data and estimates will stay central for longer. Second, SFDR 2.0 is in negotiation; the desk must maintain the current templates through at least the 2028 cycle while preparing for product categories. Third, ESG ratings became a regulated activity on 2 July 2026, which forces providers and banks to draw a clear line between data, analytics and ratings. Fourth, alliances changed: the Net Zero Banking Alliance ceased operations in 2025, NZAM relaunched with a looser commitment and NZAOA moved towards transition and engagement targets. Clients’ reports increasingly describe engagement and transition financing, not only portfolio footprints.
The Climate Transparency Hub’s analysis of the 2025 Article 29 reports gives a picture of French practice: 2019 is the most common base year and 2030 the most common target year; average ambition is about 46% reduction, with many actors at 50%; average taxonomy eligibility is about 20% and alignment about 5%; nearly 40% of actors refer to the Kunming-Montréal biodiversity targets. These are the numbers Garonne’s report will be compared with.
Open debates.
INTERACTIVE LAB — Self-assessment: ESG data, analytics and reporting services
Check the key mechanics and regulatory points of the module.
Q1. A bond of EUR 500m is mapped to its financing vehicle, for which the vendor reports zero emissions. What happens to reported coverage?
Answer: B. The vehicle has a data point (zero), so the line counts as covered. Only a line-level control catches the error.
Q2. Why is the WACI preferred to total financed emissions divided by total financed revenue?
Answer: C. Roncalli shows that the revenue-based portfolio intensity is not additive, whereas the weighted average of issuer intensities is.
Q3. EVIC of all issuers rises by 25% and real emissions are flat. Reported financed emissions change by:
Answer: A. 1/1.25 - 1 = -20%. The fall is a valuation effect only.
Q4. Which threshold triggers the most demanding sections of the Article 29 LEC report?
Answer: B. Decree 2021-663 applies the alignment, biodiversity and risk sections above EUR 500m.
Q5. Excluding a low-intensity line with missing data from a WACI calculation will usually:
Answer: C. The WACI is re-weighted on remaining lines, which are on average more intensive.
Q6. From what date does the EU ESG Ratings Regulation apply?
Answer: B. Regulation (EU) 2024/3005 applies from 2 July 2026, with a transitional regime for existing providers.
Q7. In a portfolio where heavy emitters have weak targets, which ITR aggregation gives the warmer result?
Answer: D. ECOTS gives most weight to the heavy emitters, which have the highest temperatures.
Q8. Moving weight from an issuer with CI 1,000 to one with CI 50 changes the WACI per point of weight by:
Answer: A. (50 - 1,000) / 100 = -9.5 points per percentage point of weight.
Q9. Under CRR3, what drives the operational-risk capital allocated to a fee mandate at Meridian?
Answer: B. The BIC is a function of the business indicator, in which fees enter the services component; the EU set the loss multiplier to 1.
Q10. Do Garonne’s holdings in custody at Meridian enter Meridian’s financed emissions?
Answer: D. Assets under custody are not the bank’s investments; including them would double-count.
Q11. A set-up fee for onboarding that transfers no distinct service is recognised under IFRS 15:
Answer: A. The onboarding is not a distinct performance obligation, so the fee is deferred over the period of service.
Q12. When do Solvency II sustainability risk plans take effect?
Answer: C. Directive (EU) 2025/2 applies from 30 January 2027; the related Level 2 standards were deprioritised in October 2025.
Q13. In a Brinson attribution, the effect of underweighting high-carbon issuers is the:
Answer: C. Allocation measures the effect of bucket weights different from the benchmark.
Q14. Why does the RAROC of a data mandate depend so much on the number of clients on the platform?
Answer: B. Capital per euro of fee is constant; average cost per client falls with scale, so profit and RAROC rise.
A pension fund holds EUR 400m in issuer X (EVIC EUR 20bn, revenue EUR 8bn, scope 1+2 emissions 4,000 kt), EUR 300m in issuer Y (EVIC EUR 30bn, revenue EUR 15bn, 600 kt) and EUR 300m in issuer Z (EVIC EUR 10bn, revenue EUR 2bn, 100 kt). Compute financed emissions, the footprint and the WACI. Which issuer dominates each metric?
Solution. Portfolio value EUR 1,000m. Attribution factors: X , Y , Z . Financed emissions: X kt; Y kt; Z kt; total 89 kt. Footprint: tCO2e per EUR m invested. Intensities: X ; Y ; Z tCO2e per EUR m of revenue. WACI . X dominates both: 90% of financed emissions and 88% of the WACI. Z has the highest attribution factor but contributes almost nothing.
A fund’s footprint was 180 tCO2e per EUR m in 2020 and the target is −40% by 2030. In 2026 the reported footprint is 120, and its issuers’ average EVIC is 30% higher than in 2020. Is the fund on its linear path, on a reported and on an adjusted basis? What annual reduction is needed from 2026 on each basis?
Solution. Target . Linear path: in 2026. Reported: , ahead. Adjusted: , which is 19.2 above the path, behind. Annual change needed over four years: reported ; adjusted . The conclusion flips with the adjustment, and the adjusted pace is faster than the 7% of the EU climate benchmarks.
A provider’s platform costs EUR 6m a year; the marginal cost per client is EUR 100,000; the fee is EUR 250,000. Expected operational losses are 2% of revenue, tax 25%, capital EUR 0.24375 per euro of fee. Compute the break-even number of clients, and the RAROC with 60 clients and with 45 clients.
Solution. Break-even clients. With 60 clients the average cost is k; expected loss 5k; pre-tax profit k; after tax 33.75k; capital k; RAROC . With 45 clients the average cost is k; pre-tax profit 11.67k; after tax 8.75k; RAROC 14.4%, just above the 12% hurdle. Same capital, same risk, a 41-point difference from scale alone.
1. What is the difference between financed emissions, carbon footprint and WACI? Financed emissions attribute each issuer’s emissions to the investor in proportion to its share of the issuer’s EVIC, and are expressed in tonnes. The carbon footprint divides them by the value invested, giving tonnes per million invested. The WACI is the portfolio-weighted average of issuers’ emissions per unit of revenue. The first two depend on market values through EVIC, the third on revenues. They rank holdings differently, so a report should show and explain all three.
2. A client’s footprint fell 18% this year. How do you explain it? I would build a bridge with four components: trades (the client changed its holdings), real emissions (issuers reported lower emissions), data (new vintage, new estimates, restatements) and valuation (EVIC changes). In a rising market the valuation effect alone can explain most of the fall. I would quantify each piece and show the EVIC-adjusted figure so that the client does not claim a decarbonisation it did not achieve.
3. How do you handle missing emissions data? With a documented waterfall: reported and verified first, then reported, then company-specific estimates, then sector proxies, each tagged with its PCAF score. I avoid excluding lines because exclusion implicitly assigns them the average intensity of the rest and can bias the result either way. I disclose coverage and the weighted data-quality score together, since one without the other can mislead.
4. What is an implied temperature rise, and would you put it in a regulatory report? It translates an issuer’s projected emissions or targets into a warming level, often through carbon budgets or a regression on scenarios, and aggregates it at portfolio level. It is a model output that depends on scenario, anchors, the choice between targets and trends, and aggregation. I would include it only with its method, coverage and both aggregation views, and present it as one alignment indicator among others, not as the answer.
5. Why is the RAROC of a reporting service not comparable with that of a loan? In a loan, capital follows exposure and default risk, and pricing improves the RAROC directly. In a fee service, operational-risk capital under CRR3 is proportional to revenue, so the RAROC is driven by the cost-income ratio and platform scale. Regulatory capital also ignores the liability cap and the quality of controls. I would look at the cost-income ratio, the break-even client count and a RAROC on liability-based capital.
6. Where is the line between a data service and an ESG rating under the new EU regulation? The regulation covers opinions or scores on an entity’s ESG profile, risks or impact, produced with a defined methodology and ranking system. Raw data and emissions figures are not ratings, and portfolio metrics computed for a client’s own reporting are a service. If the bank distributed its own issuer scores to clients, it could fall within the scope and would need authorisation by ESMA. That is a compliance decision to take before launching a product, not after.
7. A client asks you to cut its WACI by 30% for next year’s report. What do you advise? I would show the cheapest mechanical route, a tilt from the most to the least intensive issuers, with its yield and risk cost. I would also say clearly that it changes nobody’s emissions and that the report must describe the lever used. Depending on the client’s commitments, transition or engagement targets may serve its objectives better, and the new NZAOA protocol points in that direction.
8. What makes an Article 29 report credible to the ACPR? Following the standard plan, complete sections, quantitative targets with a base year and monitoring, and consistent definitions such as the PAI 4 fossil-fuel method. Figures that reconcile with the accounts and are explained from one year to the next. A clear statement of methods, data sources and limits. The supervisor’s past reviews of insurers’ reports criticised above all incompleteness, imprecision and departures from the standard plan.
Standards and methodologies
EU and French law
Supervisory papers and market analyses
Academic and textbook