Separate borrower records from loan facilities before requesting financed-emissions data
Three loan facilities can belong to two borrowers. Keep the three financial records, but do not create three independent borrower emissions records simply because the loan extract has three rows.
This guide helps lending and sustainability teams prepare a financed-emissions data request. Its worked example tests record matching and a document-coverage statistic. It does not calculate financed emissions, assess credit risk or recommend a lending decision.
Agree the method before building the request
PCAF’s December 2025 Part A distinguishes general-purpose business lending from financing with a specified use of proceeds. Section 5.2 attributes part of a borrower’s annual emissions using outstanding exposure and a company-value denominator; the denominator differs for listed and private companies. That makes a borrower’s total emissions a different quantity from the lender’s attributed result.
The same section covers borrower Scope 1, Scope 2 and Scope 3 emissions, with Scope 3 disclosed separately. A record containing only Scope 1 and Scope 2 cannot establish complete coverage.
Ask the method owner to record the chosen edition, asset-class treatment, portfolio cut-off and financial inputs needed before contacting borrowers. Put unresolved classification questions in a decision list. Do not let the layout of an export choose the accounting method.
Give each object its own identifier
The following record design is an original working aid, not a prescribed PCAF template. It separates the financial extract, the borrower’s response and the link between them:
Facility record: facility ID, borrower legal-entity ID, source-system reference, currency, balance date, amount and its stated basis. Retain the original row even where several facilities share a borrower.
Borrower record: legal-entity ID, legal name and the contact responsible for clarifying its data. Keep a group or trading name as an additional label, not a replacement for the identifier.
Emissions document: document ID, supplied version, reporting entity, period, units, covered scopes and the location of the underlying calculation. Record unknown fields as unknown.
Matching record: facility ID, borrower ID, document ID if available, matching decision, reviewer and unresolved question. A link says where to look; it does not approve a calculation.
When the document names a parent group but the loan names a subsidiary, flag the mismatch for the method owner. Do not silently relabel the group document as a subsidiary inventory. Keep both names visible while the boundary question is resolved.
Test a three-facility extract
All entities, IDs, amounts and documents below are invented. The sample is a deliberately selected three-facility extract, not a complete institution portfolio. Assume one currency, a common 31 March 2026 balance date, unique facility IDs and amounts already reconciled by the fictional finance team. Amounts are outstanding drawn balances, in crore rupees.
F01 links to borrower A and contains ₹8 crore.
F02 links to borrower A and contains ₹4 crore.
F03 links to borrower B and contains ₹6 crore.
Borrower A supplies document E1, labelled as A’s entity-level Scope 1 and Scope 2 inventory for the year ending 31 March 2026. The example stipulates that its name, entity identifier and period match. E1 states 1,000 tonnes CO2e. No Scope 3 record or attribution denominator has been supplied.
Borrower B supplies no emissions document at the review cut-off. Its emissions remain unknown. Neither a zero nor A’s result belongs in B’s record.
The matching record has three lines: F01 → A → E1; F02 → A → E1; F03 → B → missing. The document store contains one E1. The two references do not create two independent measurements or two separate 1,000-tonne inventories.
Retain both A facilities and the link to the same E1. This document-store check does not decide whether a calculation engine aggregates exposures or calculates separately by facility. That decision belongs to the approved method and its treatment of the portfolio. No financed-emissions result has been established here.
Name the denominator before reporting progress
The selected drawn balances total 8 + 4 + 6 = ₹18 crore. Rows linked to an emissions document total 8 + 4 = ₹12 crore. Thus 12 ÷ 18 × 100 = 66.7%, rounded to one decimal place.
Label the statistic precisely: “66.7% of the selected extract’s drawn balance is linked to a received emissions document at the review cut-off.” This only measures that document-link test. E1 has not been approved as a complete input.
Two of three facility rows have a document link, while one of two borrowers has supplied a document. Those counts happen to produce 66.7% and 50%. They answer different questions. A percentage without its population and test is not useful progress information.
Do not describe any of these percentages as the share of emissions measured, a data-quality score, an eligible portfolio share or completion of a financed-emissions inventory. The missing attribution inputs and Scope 3 record remain visible. No tonnes are calculated by multiplying 1,000 by 66.7%.
Send one coordinated request for each unresolved borrower record
For A, attach the F01 and F02 references together so the recipient can see why one response is being requested. Ask the method owner to identify the remaining calculation inputs and confirm the request before it is sent. Avoid two teams independently asking A to recreate E1.
For B, ask whether an entity-level inventory exists and who can explain its coverage. If it does not exist, record that answer and refer the estimation route to the method owner. Keep B’s ₹6 crore in the selected population while the question is open.
Use the organisation’s approved secure exchange route and request only the necessary records. Do not place borrower financial statements or confidential loan details in a public tracker. A reference to an access-controlled source is often enough for the working index.
Hand over the join and its exceptions
Deliver the three-row facility extract, two borrower records, E1, the three matching lines and the two coordinated request records. Include the exact definition of the 66.7% statistic. List missing data and unresolved method decisions separately so a received document cannot silently become a calculation approval.
When E1 is replaced, keep its earlier version and update the references for both F01 and F02. When a facility changes borrower, investigate the identity change before moving the link. A quick check for duplicate facility IDs, orphan borrower IDs and missing document versions helps catch broken joins.
Primary source checked 14 October 2026. The record design, extract and arithmetic are original fictional examples. They do not establish PCAF conformity, regulatory compliance, independent assurance or a real borrower’s environmental performance.
Source
PCAF Part A: Financed Emissions, Third Edition. December 2025; section 5.2, printed pages 55–57 (PDF pages 55–57 one-based).
Comments