GRSAC: from evidence to reviewable disclosure

Short answer and objective

Reviewable disclosure lets another person reconstruct where information came from, how it was transformed, who assessed its limitations and who authorized publication. This guide proposes an original evidence-control workflow for institutions and counterparties. It is neither an official layout nor a compliance promise.

Context reference: the BCB-hosted GRSAC file includes governance narratives and quantitative exposure and emissions information. BCB-LAYOUT:governance BCB-LAYOUT:climate Its adopted version and the new regime’s operative provisions remain subject to confirmation. BCB-772:layout BCB-586:transition

Four states that must not become the same number

Source contents, with version caveat: CRFR1 distinguishes quality levels for independently verified primary data, unverified primary data and proxy estimates. BCB-LAYOUT:quality Do not automatically map those three levels to PCAF scales. Our table describes information states; it does not replace the official classification.

Original worksheet: record through approval

Enverium practice: a row should not become “approved” merely by passing format validation. The reviewer must assess definition, coverage, calculation and interpretation; the approver must have disclosure authority.

Institution workflow: prepare, review, approve

First confirm applicability. Then freeze a working population with a cutoff date. Reconcile identifiers and units before aggregation. A duplicated customer, kilograms mixed with tonnes or a head-office address used as an asset location can change a result without any real economic change.

During review, compare numbers and narrative. With partial coverage, “the entire portfolio is measured” cannot survive merely because a total reconciles. Preserve a discrepancy record with owner and resolution, without erasing previous values.

At approval, present material limitations and distinguish public data from restricted evidence. Do not publish customer attachments for convenience. A later change requires assessing correction, versioning and communication under the applicable regime, not silently overwriting the file.

Company workflow: supply reusable data

Maintain a base evidence package with entities, facilities, periods, units and method. For each bank, record the requested slice and transformation. This enables evidence reuse without pretending every request has the same definition.

For an estimate, supply the assumption alongside the result. If external verification occurred, describe exactly the object and period covered; do not extend the conclusion to new facilities. Name someone able to answer technical questions and provide a secure channel for restricted records.

Fictional case: coverage is not zero emissions

Fictional example and teaching arithmetic, not a regulatory calculation: a test portfolio has three counterparties with exposures of R$40 million, R$35 million and R$25 million. The first two supply emissions data; the third does not.

Exposure-weighted coverage is (40 + 35) / (40 + 35 + 25) = 75%. This describes coverage of this fictional population, not an officially mandated field or financed emissions. Counting customers gives 2/3, approximately 66.7%; the denominators answer different questions.

If only the first counterparty has independent verification, the exposure share supported by that evidence is 40%, not 75%. The third counterparty’s total emissions remain unknown. Any estimation method must be documented, not concealed as zero.

The company supplies the slice and limitations; the bank records the denominator, quality and appropriate use. Review prevents a narrative from turning “we have data for 75% of exposure” into “75% of the portfolio is sustainable.”

Review checklist and simple tests

• Does each number have a unit, period and perimeter?
• Does the population reconcile to an identified control total?
• Are duplicate and exclusion treatments documented?
• Is missing data separated from zero and not applicable?
• Do estimates preserve factors, versions and assumptions?
• Does the narrative respect coverage and limitations?
• Are preparer, reviewer and approver real people with defined roles?
• Can method changes be distinguished from performance changes?

A useful test is asking another person to reconstruct a result using only the evidence index. If they must guess a unit or informally hunt for the author, the package is not sufficiently documented yet.

Limitations and monthly updates

This workflow does not automatically assign official quality, materiality, assurance obligations or disclosure deadlines. Those depend on the rule and context. The consulted PDF has editorial caveats; confirmation of its adoption remains outstanding and is necessary before operational use of its instructions.

In monthly review, check sources and methods, not only links. A page still returning HTTP 200 may have changed version or contain only an empty application. Update the check date after reading; change the substantive date only when content changes.

Further reading

See the GRSAC overview, applicability and PCAF. The IFC worksheet addresses evidential sufficiency for a diligence question, not the same disclosure regime.