PCAF Data Quality Score — Definition and GHG Accounting Context
Two banks can report identical financed-emissions totals and mean entirely different things by them. One built its number from audited, counterparty-reported emissions; the other estimated everything from sector averages and regional economic data. The PCAF data quality score exists to make that difference visible — and to stop a precise-looking number from concealing how much of it is estimate.
The score grades the evidence behind a financed-emissions figure, not the figure itself — and on the PCAF scale, 1 is best and 5 is worst.
The PCAF data quality score is a 1-to-5 grade on every financed-emissions figure under the PCAF Standard. Score 1 is verified reported emissions; score 5 is estimated from economic proxies. Lower is better.
Definition and the 1–5 Scale
The data quality score is a mandatory companion to every financed-emissions figure under PCAF. It does not change the emissions number — it grades the evidence behind it, so users of the disclosure know how much rests on primary data versus proxy estimation. The scale inverts the intuition many bring to it: 1 is best, 5 is worst.
The ordering follows a single principle: the closer the figure is to the counterparty’s own verified emissions, the lower (better) the score; the more it relies on generic estimation, the higher (worse). PCAF arranges this as a descending hierarchy from reported-and-verified data, through physical-activity estimation, to economic-activity proxies.
The score grades evidence, not emissions. Two portfolios can carry the same financed-emissions total with very different score distributions — the score tells a reader which total to trust and where the estimation risk sits. A figure without a score is not a complete PCAF disclosure.
The Score at a Glance
The defining characteristics of the PCAF data quality score, as set out in the PCAF Global GHG Accounting and Reporting Standard for the Financial Industry.
| Property | Value | Notes |
|---|---|---|
| Scale | 1–5 | 1 = highest quality (verified reported); 5 = lowest (generic estimate) |
| What it grades | Evidence | The data behind the figure, not the tonnage itself |
| Assigned at | Per exposure | Aggregated to a financed-emissions-weighted portfolio average |
| Direction (all classes) | reported > physical > economic | Constant across asset classes; qualifying inputs differ |
| Mandatory | Yes | A financed-emissions figure without a score is incomplete |
| Asset classes scored | 9 | Each with its own scorecard; most run 1–5 |
| Source standard | PCAF (2024) | Global GHG Accounting Standard for the Financial Industry, v2 |
What Each Score Means
The hierarchy below is the PCAF data quality direction common to every asset class. The score descends from counterparty-reported, verified emissions to estimates built from economic averages.
| Score | Basis of the emissions figure | Evidence type |
|---|---|---|
| 1 (best) | Reported emissions, verified / audited | Primary, assured |
| 2 | Reported emissions, unverified | Primary, unassured |
| 3 | Physical-activity-based estimate (e.g. energy use × emission factor) | Estimated, physical proxy |
| 4 | Economic-activity-based estimate using counterparty-specific data (e.g. revenue) | Estimated, entity-specific economic proxy |
| 5 (worst) | Estimate from sector / regional averages, no counterparty-specific data | Estimated, generic proxy |
The exact wording and the inputs that qualify for each score are set per asset class in the PCAF Standard. For listed equity and corporate bonds, for example, Score 1 additionally requires that the issuer’s EVIC is known alongside verified reported emissions. The direction — reported > physical > economic — is constant across classes.
One Asset, Three Scores — a Worked Example
The score is not a property of the asset — it is a property of the data you have about that asset. The same property, financed identically, lands on a different score depending on how its emissions were established. The figures below are for one UK residential mortgage: £180,000 outstanding on a £300,000 origination value, giving an attribution factor of 0.600. Only the emissions evidence changes.
| Evidence available | Score | Building emissions (tCO₂e) | Attribution | Financed emissions (tCO₂e) |
|---|---|---|---|---|
| Actual metered energy (12,000 kWh gas + 3,000 kWh electricity) | 2 | 2.72652 | 0.600 | 1.63591 |
| EPC band + floor area (EPC D, 90 m²) | 3 | 2.49875 | 0.600 | 1.49925 |
| Property type only (semi-detached, no EPC) | 5 | 2.73420 | 0.600 | 1.64052 |
Building emissions = (heating kWh × gas factor) + (electricity kWh × grid factor); financed emissions = building emissions × attribution. The same £180k-on-£300k mortgage produces three different figures and three different scores purely from the evidence behind the energy estimate — metered consumption (Score 2), an EPC-and-floor-area estimate (Score 3), or a property-type average with no building-specific data (Score 5). Engine-confirmed PCAF Mortgages worked examples on DESNZ NEED 2025 benchmarks with DEFRA 2025 natural-gas (0.18296 kg CO₂e/kWh) and UK location-based grid (0.177 kg CO₂e/kWh) factors, hardcoded here as a worked-example snapshot.
Notice the financed-emissions figures barely differ (1.499 to 1.641 tCO₂e) while the scores span 2 to 5. The score is not telling you the emissions are wrong — it is telling you how much confidence the number deserves. A Score 5 estimate can be close to the truth; you simply have no building-specific evidence that it is.
Why the Scorecard Differs by Asset Class
The evidence available is not the same across asset classes — a listed company files audited accounts; a mortgaged home does not. PCAF therefore publishes a tailored scorecard for each of its asset classes (listed equity and corporate bonds, business loans and unlisted equity, project finance, commercial real estate, mortgages, motor vehicle loans, sovereign debt, and the associated- and facilitated-emissions classes), while holding the 1–5 direction constant. Not every scorecard has the same number of rungs — some lines, such as commercial-lines insurance, define fewer than five.
A mortgage scores on whether actual energy-performance data exists for the property versus a floor-area or building-type proxy. A listed-equity holding scores on whether the issuer’s emissions are reported and verified and whether its EVIC is known. The inputs differ; the meaning of “1 is best” does not. The PCAF Data Quality Score Calculator applies the correct scorecard for each class and derives the score from the inputs, rather than leaving it to manual interpretation.
The estimation rungs map to specific data sources. A Score 5 listed-equity estimate typically uses a spend-based intensity factor applied to revenue; moving to Score 3–4 means sourcing the counterparty’s physical activity or entity-specific economic data; reaching Score 1–2 means obtaining the counterparty’s own reported emissions. Knowing which rung an exposure sits on tells you exactly what data to go and get.
Two Axes — Data Quality vs Attribution Quality
A financed-emissions figure has two distinct uncertainty axes, and the data quality score grades only one of them. The first axis is the quality of the emissions data — the score’s subject. The second is the quality of the attribution factor: how reliably the institution’s share of the counterparty’s emissions is established. For listed equity that share rests on EVIC, which may itself be known precisely or estimated.
The PCAF scorecards bundle both axes into the single 1–5 score for each asset class — which is why, for listed equity, Score 1 requires both verified reported emissions and a known EVIC. But it helps to hold the two apart conceptually: a figure can have excellent emissions data sitting on a shaky attribution denominator, or vice versa. When a score moves, knowing which axis drove it points to the right remedy — chase the counterparty’s emissions report, or pin down the valuation denominator.
A Score 5 is not a defect to conceal. PCAF’s design assumption is that institutions begin with broad estimates and improve over time — disclosing a high score transparently is correct practice, not a weakness. The reportable position is “here is our number and here is how much of it is estimate”; suppressing or inflating the score to look better defeats the purpose of having one.
The Portfolio Weighted-Average Score
The score is assigned exposure by exposure. A portfolio-level figure is then reported as the weighted-average data quality score, weighting each exposure’s score by its share of total financed emissions. This is why a single large, poorly-evidenced exposure can drag a portfolio’s average score upward even when most positions are well-evidenced — the weighting is by emissions, not by count.
Assigning one score to an entire portfolio is a category error. The score is computed per exposure and aggregated as a financed-emissions-weighted average. Reporting a single blanket score hides exactly the distribution — which exposures are well-evidenced and which are proxies — that the metric exists to reveal.
The Score as an Improvement Roadmap
Because the score is exposure-by-exposure, a portfolio’s score distribution doubles as a prioritised to-do list. The largest exposures sitting at score 4–5 are where action moves the needle: engaging counterparties for reported emissions, or upgrading estimation from an economic-activity proxy to physical-activity data. Improving a small, already well-scored exposure changes the weighted average very little; improving a large 5 changes it materially.
This makes the score distribution a year-on-year progress measure in its own right. A falling weighted-average score signals improving evidence quality independent of whether the absolute emissions number moved — the two tell different stories and are reported together.
Regulatory Context
The data quality score travels with the financed-emissions figure into the frameworks that consume PCAF output.
| Framework | Role of the score | Reference |
|---|---|---|
| PCAF Standard | Defines the 1–5 scale and the per-asset-class scorecards; requires every financed-emissions figure to carry a score. | View standard → |
| IFRS S2 / ISSB | Financed-emissions disclosure for financial institutions is expected to be accompanied by a data-quality indication. | View standard → |
| CDP | The financial-services questionnaire collects the weighted-average score alongside the absolute financed-emissions number. | View standard → |
Five Common Mistakes
Frequently Asked Questions
It is a 1-to-5 grade attached to every financed-emissions figure, indicating how much of the figure rests on reported, verified data (score 1) versus estimation from economic averages (score 5). It is required by the PCAF Standard and grades the evidence behind the number rather than the number itself.
1 is the best score — reported, verified emissions. 5 is the worst — estimated from sector or regional averages with no counterparty-specific data. The scale runs the opposite way to school grading, which is a frequent source of confusion when reading or comparing disclosures.
No. The score grades the evidence behind the figure, not the figure itself. Two portfolios can have the same financed-emissions total with very different score distributions — the score tells a reader which total to trust and where the estimation risk concentrates.
Each exposure is scored individually, then the portfolio figure is reported as the weighted-average data quality score — each exposure’s score weighted by its share of total financed emissions. Because the weighting is by emissions rather than by number of positions, one large, poorly-evidenced exposure can raise the portfolio average even when most holdings score well.
Because the evidence available differs. A listed company files audited accounts; a mortgaged home does not. PCAF publishes a tailored scorecard for each asset class — a mortgage scores on energy-performance data versus a floor-area proxy, a listed-equity holding on whether issuer emissions are reported and verified and EVIC is known. The 1–5 direction is constant; the qualifying inputs are not.
No. PCAF assumes institutions start with broad estimates and improve over time, so disclosing a Score 5 transparently is correct practice. A high score signals where building-specific or counterparty-reported data is missing — not that the emissions number is necessarily wrong. Suppressing or inflating the score defeats its purpose.