GreenCalculusGet a free key
  1. Home
  2. Methodology
  3. Product Carbon Footprints (LCA)
  4. Spend-Based Emission Factors (MRIO/EEIO)
v1.2Last reviewed August 2026

Spend-Based Emission Factors (MRIO/EEIO)

Methodology hero for spend-based MRIO factor landscape: spend base (multi-currency, mixed-region portfolio) × MRIO database (5 major sources plus 3 in-house) = method-dependent CO₂e with provenance disclosure required. Worked example: $4.2M steel spend returns 1730 tCO₂e via USEEIO, 2010 via EXIOBASE, 2356 via EORA — a ±18% provenance delta. Provenance lineage: EPA Supply Chain, EXIOBASE 3.8.2, EORA and WIOD via the GreenCalculus MasterBrain factor library.
MB v2026.264 · updated 18 Oct 2026

The complete method for converting procurement spend into a defensible Scope 3 Category 1 inventory using multi-regional and environmentally-extended input-output (MRIO/EEIO) emission factors: where spend-based sits in the GHG Protocol data-quality hierarchy (Score 4/5 secondary data — a fallback, not a default), a full comparison of the major MRIO databases (GLORIA, EXIOBASE, EORA, USEEIO, and the EPA / UK ONS / Eurostat factor sets behind the GreenCalculus engine), the monetary-flow calculation chain with currency-deflator and price-vector handling, the GWP-basis rule that must never be silently normalised across sources, and two worked examples showing exactly when to abandon a spend-based estimate for supplier-specific data. Aligned to the GHG Protocol Scope 3 Standard, the GHG Protocol Technical Guidance for Calculating Scope 3 Emissions (2013), PCAF data-quality scoring, and CSRD ESRS E1.

This page is the factor-selection layer beneath the Scope 3 Category 1 spend-based methodology. The parent page tells you how the spend-based method works in principle; this page tells you which MRIO/EEIO database to read a factor from, why the same purchase produces materially different numbers across databases, and when a spend-based estimate is no longer good enough. The defining editorial challenge of spend-based accounting is not the arithmetic — it is one multiplication — but the factor-provenance discipline: a single sourcing decision (which database, which reference year, which currency base) can move a reported total by more than the operational reductions a company spent a year achieving. Jump to the decision tree, the database comparison, or open the Scope 3 Category 1 Spend-Based Calculator.

⚠ The silent restatement — switching MRIO database mid-baseline

The same procurement spend, run through different MRIO databases, produces emission totals that diverge by tens of percent — and the divergence has nothing to do with anything the company did. An organisation that built its Scope 3 baseline on one database and then quietly moved to another in a later year can manufacture a phantom “reduction” (or a phantom increase) of a size that crosses materiality at almost any disclosure scale. Published comparisons of import emission factors across databases integrated into the same modelling framework found the spread between databases reaching double-digit percentages for the same sectors. A worked steel-procurement example below shows a 1.48× spread across four databases on an identical $4.2M spend — purely from database choice. The discipline this page enforces: fix the database, the reference year, and the currency base at baseline, document all three, and treat any change to them as a Major recalculation under GHG Protocol base-year recalculation rules — never as a year-on-year improvement. The GreenCalculus engine pins a single coherent factor set (EPA Supply Chain v1.3.0 for US spend, UK ONS for GBP spend, Eurostat for EUR spend) precisely so this drift cannot hide in plain sight.

Why Spend-Based Is Structurally Different from Activity-Based Accounting

An activity-based emission calculation is a closed physical statement. You burned 10,000 litres of diesel; diesel has a published combustion factor; the product is your emission. The activity datum (litres) and the factor (kg CO₂e per litre) are both physical quantities, and the result is reproducible to the decimal. Scope 1 stationary combustion and Scope 2 electricity both work this way — the accounting question is “did you pick the right factor for the right physical flow.”

Spend-based accounting inverts the foundation. The activity datum is not a physical quantity at all — it is money. You spent $4.2M on steel; an MRIO database publishes an average emission intensity per dollar of output for the steel-producing sector of an economy; the product is an estimate of the emissions embodied in everything that sector did to earn your dollar. The chain is monetary in, physical out, mediated by an economic model of the entire economy. Three structural consequences follow, and every one of them is a place where two competent analysts produce different numbers from the same purchase order:

  • The factor is a sector average, not your supplier’s reality. An MRIO factor for “primary metal manufacturing” blends every producer in that sector — coal-route blast furnaces and scrap-fed electric-arc furnaces, efficient plants and inefficient ones — into one intensity per dollar. Your actual supplier may be far cleaner or far dirtier than that blend. The factor cannot see your supplier; it can only see the sector.
  • The result moves with price, not just with physical throughput. Because the input is dollars, the same physical purchase generates a different emission estimate depending on what you paid. Buy the identical tonne of steel at a premium and your spend-based emissions rise, even though nothing physical changed. This is the price-vector problem, and it is the single most counter-intuitive property of the method for practitioners arriving from activity-based accounting.
  • The model itself is a choice with consequences. The number of sectors, the number of countries, the reference year, the currency base, and the construction assumptions differ across MRIO databases — and those differences propagate into the factor. There is no single “correct” spend-based factor for steel; there is GLORIA’s, EXIOBASE’s, EORA’s, the EPA’s, each defensible, each different.

This is why spend-based factors are classified as secondary data in every serious data-quality framework, and why this methodology spends as much time on when not to use them as on how to apply them.

Provider directory

Now it has to hold up all year.

See who does this work. Every listing names the standards it works to, and paid placements are labelled. Including The Carbon Trust and Anthesis Group.

Browse 6 carbon accounting & inventory providers →

Do this work? A listing is US$390 a year. Get listed →

Where Spend-Based Sits in the GHG Protocol Data Hierarchy

The GHG Protocol Technical Guidance for Calculating Scope 3 Emissions establishes a clear preference order for Category 1 (purchased goods and services). Spend-based sits near the bottom — it is the method of last resort that guarantees completeness when nothing better is available, not the method you reach for first. The PCAF Standard formalises the same ordering as a numeric data-quality score (1 = best, 5 = worst), and spend-based MRIO factors land at Score 4 or 5.

Scope 3 Cat 1 data-quality preference order (GHG Protocol Technical Guidance · PCAF score in brackets)
Best · PCAF 1
Supplier-specific (primary) data. A product carbon footprint (PCF) for the exact good purchased, measured by the supplier to a recognised standard (ISO 14067, the GHG Protocol Product Standard, or a WBCSD Pathfinder-conformant exchange). Reflects the supplier’s actual decarbonisation. Always preferred where available.
Better · PCAF 2–3
Average-data (activity-based) method. Physical quantities of purchased goods (mass, volume, units) multiplied by secondary process-based emission factors per physical unit (for instance an embodied-carbon factor per kilogram of a material). More physically grounded than spend, but requires quantity data the procurement system often does not hold.
Fallback · PCAF 4–5
Spend-based (MRIO/EEIO) method. Procurement spend by category multiplied by an emission factor per unit of currency. The subject of this page. Guarantees full Category 1 coverage from data the finance system already holds, at the cost of supplier-level and physical resolution. Use it to map the whole spend base, then progressively replace the largest and most material lines with supplier-specific data.

The practical posture this hierarchy demands: spend-based is a screening tool and a completeness backstop, not a destination. A mature inventory uses spend-based factors to size the entire Category 1 footprint quickly, identifies the hotspot categories that dominate the total, and then retires those hotspots to supplier-specific PCFs over successive reporting cycles — leaving spend-based factors only on the long tail of small, immaterial lines. The GreenCalculus engine treats every spend-based output as PCAF Score 4/5 and flags it as a supplier-specific-data candidate accordingly.

Regulatory & Framework Matrix — What Each Regime Says About Spend-Based

Before reading a factor, confirm what the framework you report to actually permits. No major framework prohibits spend-based — its completeness value is universally recognised — but they diverge sharply on whether it is acceptable for material categories and on how aggressively they expect migration to primary data.

Framework Spend-based permitted Stance on material categories Migration pressure to primary data
GHG Protocol Scope 3 Yes — explicitly Permitted for all; recommends prioritising data quality for the most significant categories Implicit — via the data-quality hierarchy and base-year recalculation rules
EU CSRD / ESRS E1 Yes Permitted, but requires disclosure of the share of Scope 3 from primary vs secondary data, and the methodology and assumptions High — the primary/secondary split is a mandatory datapoint; reviewers track year-on-year improvement
CDP Climate Change Yes Permitted; the data-source and methodology fields capture spend-based vs supplier-specific per category Medium — supplier-engagement and data-quality questions reward primary-data migration in scoring
SBTi Corporate Net-Zero Yes (baseline) Acceptable for target baseline and screening; supplier-engagement targets push the largest categories toward primary data High — Scope 3 reductions tracked against a spend-based baseline are scrutinised for genuine vs methodological change
PCAF Yes (Score 4–5) Permitted as the lowest-quality tier; explicitly a fallback when supplier or physical-activity data is unavailable High — the data-quality score is itself a disclosed metric, so spend-based use is visible by construction

The pattern: every framework permits spend-based and every framework wants you to move off it where the category is material. The disclosure regimes (CSRD, CDP, PCAF) increasingly make the primary-vs-secondary split a visible, scored datapoint — so a company reporting 95% of Scope 3 from spend-based factors is not non-compliant, but is transparently signalling an immature inventory. The methodology consequence is to deploy spend-based deliberately: comprehensive at first pass, then surgically replaced where it matters.

Decision tree — spend-based vs supplier-specific vs hybrid

Five questions, in order. Run them per procurement category, not once for the whole inventory — the right answer for steel may differ from the right answer for cloud software.

1 · Is a supplier-specific PCF available for this category?
Yes — a PCF measured to ISO 14067 / GHG Protocol Product Standard, or a Pathfinder-conformant exchange. Don’t use MRIO Use the primary data — it is PCAF Score 1 and reflects the supplier’s actual footprint. Spend-based would discard that signal.
No — continue to question 2.
2 · Do you hold physical quantity data (mass, volume, units)?
Yes, with a process factor available — prefer the average-data (activity-based) method with a per-physical-unit factor (e.g. embodied-carbon per kg). Hybrid More physically grounded than spend; use spend-based only for the residual lines where quantities are missing.
No — only spend — continue to question 3.
3 · Is this category material to your total Scope 3?
Yes — a hotspot — spend-based is acceptable for this reporting cycle, but flag it for supplier-engagement and primary-data migration next cycle. Spend-based (interim) The larger the category, the larger the error a sector-average factor can carry.
No — long tail — spend-based is the right permanent home. Spend-based Migrating an immaterial line to primary data is effort better spent on hotspots.
4 · Which factor set matches your spend geography & currency?
US-dollar spend — EPA Supply Chain factors (AR5, per 2022 USD). GBP spend — UK ONS GHG intensities (AR5, per £ GVA). EUR spend in the EU — Eurostat-derived NACE intensities (AR5, per € GVA). Spend outside these — a global MRIO (GLORIA, EXIOBASE, EORA) is the fallback; pick one and pin it. See the database comparison.
5 · Have you fixed the reference year and currency base?
Yes — proceed; record the database, version, factor reference year, and currency-deflation base in the audit trail. No — stop. An unfixed reference year or currency base is the mechanism behind the silent-restatement risk in the editorial hook above. Pin all three before computing.

MRIO/EEIO Database Comparison — GLORIA, EXIOBASE, EORA, USEEIO, and the In-House Factor Sets

This is the section that determines your number before any arithmetic happens. The major environmentally-extended MRIO databases differ on three axes that all propagate into the factor: how many countries they resolve, how many sectors they distinguish, and what reference years and currency bases they carry. Coverage figures below are hardcoded point-in-time citations from each database’s documentation (verified May 2026) and are flagged for refresh on each database’s next release.

Database Countries / regions Sectors Time series Developer Strength Weakness for Scope 3
GLORIA 164 (160 + 4 RoW) 120 1990–2021 Univ. of Sydney / IELab (for UN IRP) Broadest country resolution with harmonised sectors; recent years; UN-backed Built for material/resource footprints, not procurement Scope 3; commercial licensing via FootprintLab
EXIOBASE 3 49 (44 + 5 RoW) 163 ind / 200 prod 1995–2022 Consortium (NTNU, TNO, others) Highest harmonised sector detail; SEEA-compatible; long time series; strong satellite accounts EU-centric — most of the world is aggregated into 5 RoW regions, weak for emerging-market supply chains
EORA ~190 26–500 (variable) 1990–2015 Univ. of Sydney / IELab Most complete country coverage; full historical series; reliability estimates included Sector detail varies by country (as few as 26), impeding cross-country comparison; less current
USEEIO (EPA) US only (single region) ~411 commodities (v2.0) 2010–2016 base US EPA ORD High US commodity resolution; BEA-anchored; free; regulatory familiarity in US reporting US domestic economy only; imports modelled, not natively resolved; not for non-US spend
WIOD 43–44 ~56 (35 in earlier) 2000–2014 Univ. of Groningen (EU-funded) Clean OECD-economy coverage; well-documented; widely cited in trade studies Low sector detail; OECD-centric; the public series is not actively extended beyond 2014
EPA Supply Chain v1.3.0 (in-house, US spend) US only NAICS-3 (90 rows) 2022 USD base US EPA (in GreenCalculus MB) AR5 basis; kg CO₂e per 2022 USD purchaser price; primary-source EPA factors, kept current US-dollar spend only; sector-average like all MRIO
UK ONS 2025 (in-house, GBP spend) UK only SIC section (18 rows) Per £ GVA UK ONS (in GreenCalculus MB) Official UK GHG intensities; AR5 basis matching DEFRA convention Coarse SIC-section grain; UK spend only
Eurostat NACE A*64 (in-house, EUR spend) 27 EU + EU27 aggregate NACE A*64 (2,039 rows) Per € GVA Eurostat AEA + GVA (in GreenCalculus MB) Per-member-state EU resolution; AR5 basis; official European statistics EU spend only; GVA-denominated, not purchaser-price

The GreenCalculus engine does not ship GLORIA, EXIOBASE, EORA, or WIOD as live factor sets — those are documented here as the external landscape so you can reconcile a third-party inventory built on them. What the engine does carry is a coherent trio matched to spend geography: EPA Supply Chain factors for US-dollar spend, UK ONS intensities for GBP spend, and Eurostat-derived NACE intensities for EUR spend. For example, the food-manufacturing intensity used for US food procurement renders live as 0.467 kg CO₂e/$ (AR5) and the UK manufacturing-section intensity as 0.280000 kg CO₂e/£ (AR5) — the (AR5) suffix is generated automatically from each row — the EPA, ONS and Eurostat spend-based sets are all AR5 GWP-100, so the in-house family is internally consistent, and the label is what tells you when a set you add later is not (see §7).

When GLORIA wins

GLORIA is the strongest choice when your spend reaches deep into emerging-market supply chains that EXIOBASE collapses into “rest of world” and that USEEIO does not cover at all. With 160 individual countries at a harmonised 120-sector resolution and a series running to 2021, it gives genuine per-country factors where the EU-centric and US-only databases give aggregates or nothing. Its origin as a UN International Resource Panel material-footprint model also makes it the natural choice when the same study spans carbon and raw-material or land footprints. The trade-off is that GLORIA was engineered for resource accounting rather than procurement Scope 3, and commercial use runs through a licensing route.

When EXIOBASE wins

EXIOBASE is the strongest choice for European-weighted supply chains that need fine sector resolution. Its 163-industry / 200-product classification is the most detailed harmonised scheme among the global databases, and its long, SEEA-compatible time series with rich satellite accounts makes it the academic and consultancy default for EU footprinting. The weakness is symmetrical to GLORIA’s strength: outside its 44 explicitly modelled economies, the world is five rest-of-world regions, so an EXIOBASE factor for a deeply emerging-market input is an aggregate, not a country-specific figure.

When USEEIO (and the EPA Supply Chain factors) win

For wholly US-dollar spend, the EPA’s USEEIO-derived Supply Chain factors are the most defensible choice — they are anchored to US Bureau of Economic Analysis input-output tables, published free by a government agency, expressed natively in kg CO₂e per US dollar at a stated price year, and widely recognised by US reporters and assurance providers. This is exactly the set the GreenCalculus engine carries for US spend (EPA Supply Chain v1.3.0, NAICS-3, 2022 USD, AR5). The limitation is geographic: the model is a single US region, so non-US spend should not be forced through it.

Why mid-baseline switching is a recalculation event, not an improvement

Because each database produces a different factor for the same sector, changing database between a baseline year and a later reporting year changes the reported total even if the underlying business did not change at all. Under the GHG Protocol Scope 3 Standard, a change of calculation methodology of this kind is a base-year recalculation trigger: the baseline must be restated on the new basis so that the time series remains like-for-like, and the change must be disclosed. Presenting a database switch as a year-on-year reduction is the single most common — and most consequential — error in spend-based reporting. Pin the database at baseline; if you must move, recalculate the baseline and disclose.

The Calculation — Step by Step

The spend-based calculation is one multiplication per line, summed across the spend base — but two adjustments around that multiplication carry most of the methodological risk: currency deflation to the factor’s price base, and consistent currency conversion.

E = Σi ( Spendi × Deflatori × EFi )
E Total Category 1 emissions for the spend base (kg CO₂e, summed across categories i)
Spendi Procurement spend in category i, in the factor’s currency (purchaser price for EPA; GVA basis for ONS/Eurostat)
Deflatori Price index adjustment converting nominal spend in the reporting year to the factor’s reference-year currency (e.g. 2026 USD → 2022 USD). Equals 1 when spend is already in the factor’s base year.
EFi Emission factor for category i (kg CO₂e per unit currency), read from the matched factor set at its fixed reference year and currency base

Two subtleties dominate at audit. First, the spend must be deflated to the factor’s price year before multiplication. An EPA Supply Chain factor is expressed per 2022 US dollar; multiplying it by 2026 nominal dollars without deflation overstates emissions by whatever cumulative inflation occurred — a real effect with no physical basis, shown in Worked Example B. Second, currency conversion must be consistent and documented. Spend in one currency converted to the factor’s currency at a reporting-period exchange rate introduces an exchange-rate dependency into the emission estimate; the convention (period-average vs spot, which rate source) must be fixed and recorded, because changing it later silently moves the total.

GWP Basis — the Split You Must Not Normalise

Spend-based factors are pre-aggregated CO₂e at the publisher’s GWP basis. This page’s in-house factor sets now share one basis: the EPA Supply Chain, UK ONS and Eurostat sets are all computed on AR5 GWP-100. (The EPA set was labelled AR6 here until August 2026; that was a labelling error on our side, corrected in MasterBrain v2026.110 — no factor value changed.) The split has not gone away, it has moved: EXIOBASE is AR6, and so is a typical corporate inventory built on the current GHG Protocol default. The MasterBrain carries the basis on every row, and the live [gc_factor] renders it automatically via basis_label="auto" — which is why the comparison section above shows (AR5) on both the US food-manufacturing factor and the UK manufacturing factor.

The rule that follows is the most easily-missed in spend-based accounting: do not normalise across sources, and do not mix bases within a single total. A portfolio that sums an AR6-basis EXIOBASE factor and an AR5-basis EPA or Eurostat factor into one number is internally inconsistent — the two were computed against different GWP tables. The same applies to dropping any of these AR5 spend lines into an AR6 inventory total. The defensible treatments are either to report US and EU spend as separately-disclosed sub-totals on their native bases, or to convert one set to the other’s basis explicitly and document the conversion as a methodological choice. The asymmetry is by design and mirrors the platform-wide convention: corporate reporting defaults to AR6 (for methane, 29.8), while DEFRA-flavoured and European-statistics factors carry AR5 internally and are not flagged as inconsistent for doing so. See the AR6 GWP value reference for the underlying numbers.

Worked Examples — When Spend-Based Is Enough, and When It Is Not

Two audit-record snapshots at this page’s review date (2026-05-22). All numerics are hardcoded — worked examples are audit records that must reconcile to their stated inputs regardless of future MasterBrain changes. The factor values shown are illustrative bands consistent with the cited source families; they are flagged for migration to live [gc_factor] keys once the full spend_based key catalogue is published.

Example A Steel procurement — spend-based vs supplier-specific, and cross-database spread
Your data source
US manufacturer — fictional
Reporting year: 2026  |  Framework: GHG Protocol Scope 3 + CDP
Category: Purchased specialty steel (Cat 1)
Annual spend: $4,200,000 (2022 USD purchaser price; already at factor base year)
Physical volume known: 3,200 t (implied $1,312/t — specialty grade)
Supplier PCF available: Yes — 0.62 tCO₂e/t, EAF/scrap route, EPD-backed The deliberate lesson: a supplier PCF is available, so the decision tree (Q1) says stop and use it. The spend-based figure is shown only to demonstrate the gap — and that the gap runs in the non-obvious direction here.
Pre-step — method selection: Decision tree Q1: a supplier-specific PCF exists → use primary data, not MRIO. We compute the spend-based number anyway, purely to show how far a sector-average factor sits from this supplier’s reality — and to demonstrate the cross-database spread on the same spend.
Spend-based (MRIO)
Spend (2022 USD)$4,200,000
Deflator (already 2022)1.000
EF (EPA NAICS-3 331 band)0.412 kg CO₂e/$
E = Spend × EF4,200,000 × 0.412
= kg CO₂e= 1,730,400 kg
1,730.40 tCO₂e
Spend-based · PCAF Score 4 · AR5
Supplier-specific (PCF)
Physical volume3,200 t
Supplier PCF0.62 tCO₂e/t
E = volume × PCF3,200 × 0.62
= tCO₂e= 1,984.00 t
PCAF data qualityScore 1 (primary)
1,984.00 tCO₂e
Supplier-specific · PCAF Score 1
Δ supplier vs spend-based: +253.60 tCO₂e (+14.7%). The reportable figure is 1,984.00 t (PCAF Score 1) — the spend-based 1,730.40 t is discarded once the PCF exists.
Cross-database spread — same $4,200,000 spend, different MRIO source (illustrative bands)
GLORIA (0.379 kg/$)4,200,000 × 0.379 = 1,591.80 t
EPA Supply Chain (0.412 kg/$)4,200,000 × 0.412 = 1,730.40 t
EXIOBASE 3 (0.478 kg/$)4,200,000 × 0.478 = 2,007.60 t
EORA (0.561 kg/$)4,200,000 × 0.561 = 2,356.20 t
Spread (low → high)1,591.80 → 2,356.20 t = 1.48×
Two lessons in one example. First, the supplier-specific figure (1,984 t) is higher than the spend-based estimate (1,730 t) — the gap does not always favour the supplier. A small volume of high-value specialty steel has a true intensity per dollar that the broad NAICS-3 average understates here; only the PCF resolves it, and it must replace the spend figure regardless of direction. Second, the four-database spread of 1.48× on an identical purchase is the editorial hook made concrete: had this company built its baseline on EORA and switched to GLORIA, it would have booked a 32% “reduction” having changed nothing. Open the Scope 3 Cat 1 Spend-Based Calculator → to reproduce the spend-based figure with category = primary metal manufacturing, spend = $4,200,000.
Example B Cloud / SaaS spend — when MRIO is the only defensible path, with currency deflation
Your data source
US services company — fictional
Reporting year: 2026  |  Framework: CSRD ESRS E1 + CDP
Category: Cloud infrastructure & SaaS subscriptions (Cat 1)
Annual spend: $950,000 (2026 nominal USD)
Physical volume known: No — no kWh, no compute-hours disclosed by vendors
Supplier PCF available: No standardised PCF for the bundled subscriptions The opposite case to A. No PCF, no physical quantities — spend-based is the only path that produces a defensible, complete number. Demonstrates the currency-deflation step.
Pre-step — method selection: Decision tree Q1: no supplier PCF → continue. Q2: no physical quantity data → continue. Q3: material enough to report, no alternative → spend-based is the answer. Q5: spend is in 2026 nominal USD but the EPA factor is per 2022 USD, so deflation is required before multiplication.
Step 1 — Deflate nominal spend to the factor’s price base
Nominal spend (2026 USD)$950,000
Cumulative deflator 2026→2022÷ 1.16 (illustrative CPI)
Spend in 2022 USD base= $818,966
Step 2 — Apply the matched factor
EF (EPA data-processing band)0.151 kg CO₂e/$ (AR5)
E = deflated spend × EF818,966 × 0.151
Category 1 (cloud/SaaS)= 123.66 tCO₂e
Step 3 — What skipping deflation would have reported
Nominal spend × EF (no deflation)950,000 × 0.151 = 143.45 t
Correct (deflated)123.66 t
Overstatement from skipping deflation+19.79 t (+16.0%)
The reportable figure is 123.66 tCO₂e, disclosed as PCAF Score 4/5 secondary data with the methodology, database, factor reference year, and the deflation step documented. There is no higher-quality alternative here: cloud and SaaS vendors rarely publish a product-level PCF for bundled subscriptions, and no physical quantity is available. This is exactly the long-tail case spend-based exists to cover. Note the deflation step alone moved the result 16% — the same magnitude as the cross-database spread in Example A, and a reminder that the adjustments around the multiplication carry as much risk as the factor choice itself. Open the calculator → with category = data processing / hosting, spend = $950,000, price-year deflation enabled.

Limitations and Uncertainty

Spend-based factors carry structural uncertainty that no amount of careful arithmetic removes. Naming these limitations explicitly is part of an audit-grade disclosure — assurance providers expect them acknowledged, not hidden.

Price-vector sensitivity

Because the input is money, anything that changes price without changing physical throughput changes the emission estimate. Paying a premium for the same good raises reported emissions; negotiating a discount lowers them. Inflation between the factor’s base year and the spend year does the same unless corrected by deflation (Worked Example B). The estimate is therefore a function of commercial terms, not just physical reality — the opposite of the activity-based ideal.

Sector aggregation error

An MRIO sector blends every producer it contains into one intensity. The more concentrated and the more atypical your supplier, the larger the error: a single dominant supplier whose process differs sharply from the sector mean (a clean EAF steelmaker, or a coal-route one) is misrepresented by the sector average in either direction, as Worked Example A shows. Finer-grained databases (EXIOBASE’s 200 products, USEEIO’s 411 commodities) reduce but never eliminate this — there is always a producer the average does not describe.

Currency-conversion compounding

Multi-currency spend bases compound the price-vector problem with exchange-rate dependency. Converting GBP or EUR spend to a USD factor’s currency at a reporting-period rate embeds that rate into the emission estimate; a different rate convention next year moves the total with no underlying change. Where possible, match spend to a factor set in its own currency — the in-house trio (EPA for USD, ONS for GBP, Eurostat for EUR) exists to avoid forcing cross-currency conversion in the first place.

The hybrid downgrade path

The GHG Protocol permits hybrid approaches that combine process-based and input-output data — using primary or physical data for the parts of a supply chain that are known, and spend-based factors only for the residual upstream that is not. A mature inventory drifts down this path over time: as supplier PCFs and physical quantities are obtained for the largest categories, spend-based factors are progressively confined to the immaterial tail. Treat the share of Scope 3 still on spend-based factors as a quality metric to drive down each cycle — it is exactly the datapoint CSRD ESRS E1 and PCAF make visible.

Standards Alignment

Six standards govern spend-based Scope 3 Category 1 accounting. Each supplies a different input to the same audit-grade output.

Standard Role for spend-based reporting
GHG Protocol Scope 3 Standard + Technical Guidance (2013) Defines Category 1, the spend-based and average-data methods, the data-quality preference order, and the base-year recalculation rules that make a database switch a restatement event. The methodological foundation.
PCAF Financed Emissions Standard Formalises the data-quality hierarchy as a 1–5 score; spend-based MRIO factors are Score 4–5. The score is itself a disclosed metric, making spend-based reliance visible.
IPCC AR6 Supplies the GWP-100 values behind the CO₂e aggregation. The in-house EPA, ONS and Eurostat sets are all AR5; EXIOBASE is AR6 — the split that must not be normalised (§7).
CSRD / ESRS E1 EU mandatory disclosure requiring the primary-vs-secondary data split, the methodology and assumptions, and value-chain coverage — making spend-based share a tracked datapoint.
WBCSD Pathfinder Framework Defines the primary-data (PCF) exchange that sits above spend-based in the hierarchy — the target state when migrating hotspot categories off MRIO factors.
ISO 14064-1 Sets the inventory-quality and documentation requirements that the audit checklist below operationalises for reasonable-assurance review.

What the Calculator Handles vs What You Decide

The Scope 3 Category 1 Spend-Based Calculator automates the mechanical steps. This methodology page covers the upstream decisions the calculator cannot make for you.

⚙ Calculator handles automatically
  • Factor lookup across the in-house spend-based sets (EPA Supply Chain for USD, UK ONS for GBP, Eurostat NACE for EUR)
  • Per-category spend × factor multiplication, summed to a Category 1 total
  • Currency-deflation to the factor’s price base year where enabled
  • Per-row GWP-basis labelling so mixed-basis sub-totals are surfaced, not silently merged
  • PCAF data-quality scoring (Score 4/5) tagged on every spend-based line
  • Hotspot ranking — sorting categories by contribution to flag primary-data migration candidates
  • Audit-trail export with factor source, version, reference year, and currency base per line
✎ You must decide before using the calculator
  • Method selection per category — spend-based vs supplier-specific vs hybrid, per the decision tree
  • Database choice — which factor set matches your spend geography and currency; pin it at baseline
  • Reference year and currency base — fix and document; treat any change as a recalculation event
  • Currency-conversion convention — which exchange-rate source and basis for multi-currency spend
  • Materiality judgement — which categories are hotspots that warrant primary-data migration
  • GWP-basis treatment — whether to report AR5 and AR6 sub-totals separately or convert explicitly
  • Spend-data hygiene — mapping the chart of accounts to factor categories, excluding non-emitting spend (e.g. intra-group transfers, taxes)

Ready to compute? Open the Scope 3 Category 1 Spend-Based Calculator →

Audit Checklist — Spend-Based Scope 3 Cat 1 Pre-Flight

Eight items, in order. Run this list before submitting any spend-based Category 1 disclosure for reasonable-assurance review under ISO 14064-1 or equivalent. Each maps to a specific failure mode auditors look for first.

  1. 01
    Database and version pinned. Confirm a single database and version is named for the baseline and carried consistently into the reporting year. A mid-baseline switch is a recalculation event, not a reduction — verify the baseline was restated and disclosed if any switch occurred.
  2. 02
    Reference year and currency base documented. Every factor carries a price year (e.g. 2022 USD). Confirm spend was deflated to that base before multiplication, and that the deflation index and source are recorded. Skipping deflation overstates by cumulative inflation (Example B: +16%).
  3. 03
    Currency-conversion convention fixed. For multi-currency spend, confirm one exchange-rate basis (period-average vs spot) and source is used and documented. Match spend to a same-currency factor set where possible to avoid the dependency entirely.
  4. 04
    GWP basis not mixed. Confirm AR5-basis spend factors (EPA, ONS, Eurostat) and any AR6-basis factor (EXIOBASE, or your own inventory default) are not summed into a single total without explicit conversion. Report native-basis sub-totals or document the conversion (§7).
  5. 05
    Data-quality score disclosed. Every spend-based line is PCAF Score 4/5. Confirm the score is recorded per line and the primary-vs-secondary split is reported — mandatory under CSRD ESRS E1 and scored by CDP and PCAF.
  6. 06
    Hotspot migration plan present. For material categories still on spend-based factors, confirm a supplier-engagement / primary-data migration plan is documented. Reviewers expect the largest lines to have a path off MRIO (the decision tree Q3 outcome).
  7. 07
    Spend-base hygiene. Confirm the chart of accounts was mapped to factor categories deliberately, and that non-emitting spend (taxes, intra-group transfers, financial transactions, depreciation) was excluded. Mis-mapped or double-counted spend is the most common quantitative error.
  8. 08
    Primary data takes precedence where held. Confirm that any category with a supplier-specific PCF uses it rather than the spend-based factor (decision tree Q1). A spend-based number standing in for available primary data is a quality regression, even when the spend figure is conservative.

Methodology Metadata — for GHG Inventory Documentation

Copy verbatim into your GHG inventory methodology statement for ISO 14064-1 transparency compliance. Adjust the database, reference year, and category lines to match your inventory choices.

Methodology GreenCalculus Spend-Based (MRIO/EEIO) Emission Factors Methodology v1.0 (May 2026). Parent: Scope 3 Category 1 spend-based methodology.
Method Spend-based method per the GHG Protocol Scope 3 Standard and Technical Guidance for Calculating Scope 3 Emissions (2013). Applied per category after the data-quality hierarchy is exhausted (primary → average-data → spend-based).
Factor source [EPA Supply Chain GHG Emission Factors v1.3.0, NAICS-3, kg CO₂e per 2022 USD purchaser price (AR5) — US spend | UK ONS Atmospheric Emissions GHG Intensity 2025, SIC section, per £ GVA (AR5) — GBP spend | Eurostat AEA + GVA NACE Rev. 2 A*64, per € GVA (AR5) — EUR spend | external MRIO (GLORIA / EXIOBASE / EORA) where spend falls outside the in-house sets — name database + version + reference year]. Pinned at baseline; any change is a base-year recalculation.
Currency handling Nominal spend deflated to the factor’s price base year using [stated price index]; multi-currency spend converted at [period-average | spot] rates from [source]. Conventions fixed for the time series.
GWP basis Factors carry the publisher’s GWP-100 basis: EPA, UK ONS and Eurostat all on IPCC AR5. For comparison the AR6 per-gas value for CH₄ fossil is 29.8. Bases not normalised across sources; mixed-basis totals reported as separate sub-totals or converted explicitly. See AR6 GWP values.
Data quality All spend-based lines classified PCAF Score 4–5 (secondary data). Primary-vs-secondary split disclosed per CSRD ESRS E1. Material categories carry a documented supplier-specific (Pathfinder / ISO 14067) migration plan.
Dark green Pinterest pin titled METHODOLOGY · SCOPE 3 · CAT 1. Serif headline: “Four global models, one method — choose on coverage and currency.”, a GreenCalculus summary. A light card compares four spend-based MRIO models: USEEIO (US, 411 industries, $), Exiobase (49 regions, 200 sectors, €), Eora (189 countries, widest reach), WIOD (40 countries, 1995–2011, dated). Source bar: USEEIO · Exiobase · Eora · WIOD.
Save to Pinterest Download · 1000×1500 JPG

Frequently Asked Questions

Supplier-specific is always better where it exists. The GHG Protocol data-quality hierarchy, and the PCAF 1–5 score that formalises it, both rank a supplier’s measured product carbon footprint (PCAF Score 1) above any spend-based MRIO factor (Score 4–5). A PCF reflects the supplier’s actual emissions and decarbonisation; a spend-based factor is a sector average that cannot see your supplier. The role of spend-based is completeness and screening: use it to size the entire Category 1 footprint quickly from finance data you already hold, identify the hotspot categories, then progressively replace those hotspots with supplier-specific data over successive cycles. A mature inventory keeps spend-based factors only on the immaterial long tail. The one firm rule: where a supplier PCF exists, use it — a spend-based number standing in for available primary data is a quality regression even if it happens to be conservative.

Because each is a different model of the world economy. They resolve different numbers of countries (GLORIA 164, EXIOBASE 49, EORA around 190), different numbers of sectors (EXIOBASE 163 industries / 200 products, GLORIA 120, EORA a variable 26 to 500), different reference years, and different construction assumptions — and all of those propagate into the emission factor for any given sector. Published comparisons that ran identical spend through different databases within the same modelling framework found double-digit-percentage differences in the resulting factors. The worked steel example on this page shows a 1.48× spread across four databases on one $4.2M purchase. None of them is “wrong” — they are different defensible estimates of an inherently uncertain quantity. The practical consequence is that you must pin one database at baseline and treat any switch as a base-year recalculation, not a reduction.

Match the factor set to your spend’s geography and currency first. For US-dollar spend, the EPA Supply Chain factors (USEEIO-derived, per 2022 USD, AR5) are the most defensible — government-published, BEA-anchored, and the set the GreenCalculus engine carries. For GBP spend, UK ONS GHG intensities; for EUR spend in the EU, Eurostat-derived NACE intensities. Only when spend falls outside those — deep emerging-market supply chains in particular — does a global MRIO become the answer, and then GLORIA wins on country resolution (160 countries, harmonised 120 sectors, recent to 2021) while EXIOBASE wins on sector detail for European-weighted chains. Whatever you choose, pin it: the database, version, reference year, and currency base must be fixed at baseline and carried consistently, because changing any of them changes your number with no underlying business change.

This is the price-vector problem, and it is the most counter-intuitive property of spend-based accounting. The method multiplies money by an emission factor expressed per unit of currency, so the input is what you paid, not what you physically received. Buy the identical tonne of steel at a premium and your spend-based emissions rise; negotiate a discount and they fall — even though nothing physical changed. Inflation does the same thing over time unless you deflate spend to the factor’s price base year before multiplying (a step that moved the cloud example on this page by 16%). It is a genuine limitation, not a bug: it is the price of estimating physical emissions from monetary data. The two mitigations are to deflate consistently to the factor’s base year, and to migrate price-sensitive material categories to physical-quantity or supplier-specific data where the input is mass or a measured footprint rather than money.

Yes, whenever your spend year differs from the factor’s price base year. An EPA Supply Chain factor is expressed per 2022 US dollar. If you multiply it by 2026 nominal dollars, you inflate the result by whatever cumulative price change occurred between 2022 and 2026 — an increase with no physical basis. The fix is to deflate the nominal spend to the factor’s base year using a price index before multiplying. On the cloud-spend example on this page, deflating $950,000 of 2026 spend to its 2022 base reduced the result from 143 to 124 tCO₂e — a 16% difference from the deflation step alone. Record the index and its source in the audit trail; the deflation convention, like the currency-conversion and database choices, must be fixed for the time series so it cannot silently move the total year to year.

On GWP grounds, the three in-house sets are now safe to sum: EPA, UK ONS and Eurostat are all AR5 GWP-100. (EPA was labelled AR6 here until August 2026 — our labelling error, corrected in MasterBrain v2026.110, with no change to any factor value.) What is still not safe is mixing any of them with an AR6-basis set such as EXIOBASE, or folding them into an AR6 inventory total. The currency and denominator bases remain a separate and larger trap: these are per-dollar of purchaser price, per-pound of gross value added and per-euro of gross value added respectively, which is why the defensible treatments are to report US, UK, and EU spend as separately-disclosed sub-totals on their native bases, or to convert one set to the other’s basis explicitly and document the conversion as a methodological choice. The GreenCalculus engine carries the basis on every factor row and labels it automatically, precisely so a mixed-basis sum is surfaced rather than silently created. This mirrors the platform-wide convention: corporate reporting defaults to AR6, while DEFRA-flavoured and European-statistics factors carry AR5 internally by design.

Score 4 or 5 — the lowest-quality tiers in the PCAF 1–5 scale. PCAF reserves Score 1 for verified primary (supplier-reported) data, the middle scores for physical-activity data multiplied by process factors, and Scores 4–5 for economic-activity (spend) data multiplied by sector-average emission factors, which is exactly what a spend-based MRIO calculation is. The score is not a penalty — it is an honest signal of the estimate’s resolution, and it is itself a disclosed metric under PCAF, CDP, and CSRD ESRS E1. A company reporting most of its Scope 3 at Score 4/5 is not non-compliant, but is transparently flagging an early-stage inventory. The improvement path is to move material categories up the scale by obtaining physical quantities (Score 2–3) or supplier PCFs (Score 1), leaving spend-based factors only where nothing better is obtainable.

No — and treating it as one is the single most consequential error in spend-based reporting. Because each database produces a different factor for the same sector, moving from one database to another between your baseline year and a later reporting year changes the reported total even if nothing about the business changed. Under the GHG Protocol Scope 3 Standard, a change of calculation methodology of this kind triggers a base-year recalculation: you must restate the baseline on the new basis so the time series stays like-for-like, and disclose the change. Booking the difference as a year-on-year reduction would manufacture a phantom improvement — the steel example on this page shows a 32% swing between two databases on an unchanged purchase. Pin the database at baseline; if a genuine reason to switch arises (a database is discontinued, or a better-matched set becomes available), recalculate the baseline and disclose the methodology change rather than letting it flow through as performance.

Calculator

Apply this methodology with audit-grade output: per-category spend × factor with automatic currency deflation, the in-house EPA / ONS / Eurostat factor sets matched to spend currency, per-row GWP-basis labelling, PCAF Score 4/5 tagging, hotspot ranking for primary-data migration, and full provenance export: Open the Scope 3 Category 1 Spend-Based Calculator →

Spend-based factor sets (MasterBrain v2025.52, verified 2026-05-22): US-dollar spend uses EPA Supply Chain GHG Emission Factors v1.3.0 at NAICS-3 grain, kg CO₂e per 2022 USD purchaser price, AR5 basis — e.g. food manufacturing renders 0.467 kg CO₂e/$ (AR5). GBP spend uses UK ONS Atmospheric Emissions GHG Intensity 2025 at SIC-section grain, per £ GVA, AR5 — e.g. 0.280000 kg CO₂e/£ (AR5). EUR spend uses Eurostat AEA + GVA derived NACE Rev. 2 A*64 intensities, per € GVA, AR5. GLORIA, EXIOBASE, EORA, and WIOD are documented as the external database landscape and are not carried as live factor sets.

Method: GHG Protocol Scope 3 Standard + Technical Guidance for Calculating Scope 3 Emissions (2013), spend-based method. Data-quality scoring per PCAF (Score 4–5, secondary data). All spend-based outputs are a fallback to be migrated toward supplier-specific (Pathfinder / ISO 14067) data for material categories.

Reviewed: May 2026 by Jeremiah Say. Refreshed within 30 days of each EPA Supply Chain / UK ONS / Eurostat factor release.

Corrections: Spotted an error? Email jeremiah@greencalculus.com — corrections welcomed and contributors credited in the public changelog.

Scroll to Top