Counting what we cover: a pragmatic path to climate accounting

Summary

Export credit agencies (ECAs) face growing demands to account for the climate footprint of the trade and investment they enable. Yet dominant carbon accounting standards for financial institutions were developed primarily for lenders, while most of the exposure of the ECAs in the leading climate alliance (NZECA) takes the form of guarantees and credit insurance. As a result, different accounting approaches can lead to vastly different carbon footprints for the same transaction, ranging from zero to tens of thousands of tonnes of carbon dioxide. This Policy Brief draws on recent research, public disclosures of leading ECAs and emerging NZECA guidance to set out a lean, credible sequence for institutions that are building a climate framework from a standing start: commit, measure, disclose, set targets, review. It argues that transparency about method, coverage and data quality matters more in the early years than precision, and it sets out priorities for ECA boards, their governments and standard setters.

The accounting gap at the heart of export finance

Export credit agencies and export-import banks support large infrastructure, energy and manufacturing projects that shape emissions for decades. In the energy sector alone, ECA commitments ranged between about USD 22 and 43 billion per year (in constant 2020 USD) over 2013–2023. The proportion of renewables increased to around 40 per cent by 2022–2023, yet fossil fuel support remains substantial (Censkowsky et al., 2025). ECAs provide far more public finance for fossil fuels than multilateral lenders, although they have historically received little scholarly attention (Peterson & Downie, 2024). As with other financial institutions, their climate impact is mostly indirect and extends beyond own operations to the portfolio emissions of the projects and companies they support (NZECA, 2024). Members of the UN-convened Net-Zero Export Credit Agencies Alliance (NZECA) have committed to align these attributable greenhouse gas (GHG) emissions with net zero pathways by 2050 and to disclose them annually (NZECA, 2025).

The main challenge is methodological. Around 74 per cent of members' exposure is in insurance and guarantees, 24 per cent in direct lending and 2 per cent in other products (UNEP FI, 2025; NZECA, 2025). The PCAF Standard for financed emissions covers loans and project finance, but its earlier editions attributed no emissions to a guarantee until it was called and paid; its third edition, published in December 2025, adds new methods, but none of them addresses export credit guarantees or insurance (PCAF, 2025a). The PCAF insurance standard, now in its second edition, covers several types of insurance, including primary trade credit and political risk insurance, but keeps structured trade credit out of scope and relies on premium-based attribution, which can produce relatively low estimates and understate materiality (PCAF, 2025b; UNEP FI, 2025). Thus, there is no single standard tailored to the core products of ECAs, and the alliance itself notes that methodological guidance for these instruments remains fragmented (NZECA, 2025).

This gap is not a technicality. Figure 1 illustrates how alternative attribution approaches change the result for one hypothetical guaranteed project, from zero to 50,000 tonnes of CO2e per year. Two of the four approaches, amount at risk and expected loss, are applied in published ECA methodologies; the first reflects the PCAF Part A treatment of uncalled guarantees, while maximum liability is shown as a sensitivity case only. Published figures confirm the scale of the effect: one agency's 2025 Scope 1–2 portfolio emissions amount to 4.8 million tCO2e on an amount-at-risk basis but 0.9 million tCO2e on an expected-loss basis (UKEF, 2026). An institution that does not state its choice, and the reasons for it, makes comparisons with peers and assessment of its own progress more difficult, and risks undermining the credibility of its reporting.

Figure 1: Illustrative impact of alternative attribution approaches on one transaction (hypothetical project: 100,000 tCO2e/yr, project value USD 500m; ECA cover outstanding USD 200m, commitment USD 250m; PD 5%, LGD 45%)

• PCAF Part A rule (uncalled guarantee, no attribution): 0 tCO2e per year
• Expected loss (EAD × PD × LGD ÷ project value): 900 tCO2e per year
• Amount at risk (outstanding cover ÷ project value): 40,000 tCO2e per year
• Maximum liability (full commitment ÷ project value; sensitivity case): 50,000 tCO2e per year

*Source: authors' illustration with hypothetical values, based on PCAF (2025a), UNEP FI (2025) and UKEF (2026).*

A conceptual distinction applies throughout this Brief: attributed portfolio emissions are an accounting measure of an ECA's emissions associated with its financial exposure, not a direct measure of emissions caused or avoided by its intervention. Year-on-year changes can reflect portfolio growth, amortisation, shifts in credit risk, better data or methodological restatements as much as real-world decarbonisation (UNEP FI, 2025; UKEF, 2026).

What leading ECAs have learned

A small group of agencies has moved ahead of the standards and published their own approaches. Their choices differ in terms of attribution basis, product scope and comparability, but the four approaches reviewed here share three features.

• First, all four build on the PCAF logic of attributing an entity's emissions in proportion to exposure relative to the value of the entity or project.
• Second, many treat guarantees and insurance in the same way as lending, following the argument that the ECA's cover enables the transaction in the first place.
• Third, all disclose the limitations of their data openly; where weighted data-quality scores are published, they range between about 3.6 and 4.8 on the five-point PCAF scale, on which 1 is the best and 5 the weakest. As these scores refer to different years and methods, they are indicative rather than strictly comparable.

The differences are equally instructive. The amount-at-risk approach ("follow the money") is transparent, stable and easy to explain, and for guarantees it typically yields higher attributed emissions than expected loss. A higher figure is not, however, necessarily a more accurate one, and the approach can result in double-counting emissions between ECAs and the banks whose loans are covered. The expected-loss approach ("follow the risk") better reflects the economic role of a guarantee and reduces double counting, yet its results vary with default probabilities and recovery assumptions rather than with emissions alone. Reporting both, as one agency does, makes the uncertainty visible rather than hiding it.

Research on corporate emissions data explains why this transparency matters. Emissions data from different providers diverge, particularly for Scope 3, and third-party estimates are less consistent than company-reported figures (Busch et al., 2022); estimated emissions also can behave quite differently from disclosed ones in empirical analysis (Kalesnik et al., 2022; Aswani et al., 2024). An FTSE study cited by Tang et al. (2023) found that more than a quarter of estimated emissions values deviated from reported figures by at least 200 per cent, particularly among lower-quality estimation methods. For ECAs, whose exposures are concentrated in markets where company reporting is sparse, heavy reliance on modelled data may be unavoidable particularly in the early years of reporting, but it must be visible to the reader.

Table 1: Selected public ECA approaches to portfolio emissions

• UKEF (UK): dual reporting, amount at risk and expected loss for cover, amount at risk for direct lending; Scopes 1–2 plus modelled upstream Scope 3 for all sectors
• EIFO (Denmark): outstanding amount ÷ total equity and debt or project cost, also for guarantees; disclosure split by asset class, scope and sector
• Euler Hermes (Germany): share of current outstanding risk in total project financing value; Scopes 1–2, upstream Scope 3 where data allow, downstream for fossil fuels
• Cesce (Spain): PCAF Part A with three data-dependent tiers, incl. sector EEIO factors; Scope 3, Category 15

*Sources: UKEF (2026), UNEP FI (2025), NZECA (2025), Euler Hermes (2026). Authors' compilation.*

Building a framework from a standing start

For agencies that have only recently joined a climate alliance, or are preparing to do so, the leading examples can seem out of reach. They need not be. In the authors' assessment, a credible first framework can typically be developed within one to two annual reporting cycles if it is deliberately lean, concentrates on the fields alliance members are actually asked to report, and treats data quality as something that improves over time rather than as a precondition for disclosure.

Figure 2: A phased pathway from commitment to targets

1. Commit: board-approved climate policy; base year fixed
2. Measure: own operations and portfolio baseline
3. Disclose: policy, emissions, method and data quality
4. Target: 2030 targets for the most-emitting sectors
5. Review: annual progress, restatements, review at least every five years

*Data quality improves with each renewal cycle: the PCAF score falls from 4–5 (modelled proxies) towards 1–2 (reported, verified data). Source: authors' illustration based on NZECA (2023, 2024).*

Governance first

A board-approved climate policy anchors everything that follows. It should state the commitment, the scope of products and emissions covered, the attribution basis, and how exceptions to any sector policy are decided and disclosed. A small steering group chaired by the chief executive, with finance validating data and a strategy or risk function compiling the methods log, keeps accountability clear without creating unnecessary bureaucracy.

Own operations: proportionate, but not optional

An ECA's own footprint is small compared with its portfolio, but it is the part the institution most directly controls, and peers expect it to be counted in accordance with a recognised standard. Using the operational control approach under the GHG Protocol, Scope 1 can be compiled from fuel and refrigerant records and Scope 2 from energy invoices, reported on both a location- and a market-based basis. Business travel (Scope 3, Category 6) is often among the most material categories of an institution's operational Scope 3 emissions. These must be kept separate from portfolio emissions, which an ECA reports under Scope 3, Category 15. Business travel emissions can initially be estimated from expenditure data and subsequently refined using travel-agency records. Two simple intensity metrics, emissions per employee and per unit of revenue, support year-on-year comparison.

Portfolio: choose, state and stick to a method

For a pure-cover ECA, an amount-at-risk basis is a transparent, exposure-based starting point. The choice should, however, reflect the purpose of reporting: portfolio coverage, risk perspective or comparability with peers. Whether exposure is measured gross or net of reinsurance should be decided explicitly, because risk transfer is an established feature of ECA business models and affects attribution (Krummaker & Klasen, 2025; NZECA, 2025). Where an entity's emissions and value are unknown, sector emission factors can be applied to exposure, provided price-year and currency adjustments are documented and the resulting PCAF score of 4–5 is disclosed. Some product lines, such as short-term whole-turnover policies, may be excluded initially because of their transaction volume, short duration and limited leverage over emissions (NZECA, 2024), but such exclusions should be named, justified and given a timeline for inclusion. Avoided emissions may be shown separately; they should never be netted against portfolio totals or used, together with offsets, to meet targets (NZECA, 2024; Fankhauser et al., 2022).

Disclosure that builds trust

The first report is a baseline, not a performance claim. What makes it credible is transparency about its own limitations. Table 2 sets out a minimum disclosure set that can be produced with modest resources and distinguishes alliance requirements from recommended good practice.

Table 2: Minimum disclosure set for an entry-stage ECA

• Policy and governance: board-approved climate policy; oversight and management responsibilities. *Required, all members (Protocol 8.1).* Anchors accountability.
• Portfolio emissions: Scope 3, Category 15 absolute tCO2e (Scopes 1–2 of supported activity), base year, snapshot date. *Required, all members (8.1).* Core reporting field and basis for targets.
• Method: attribution factors, data used and emission scopes; treatment of reinsurance and commitments. *Required for full members (6.1, 8.1); reinsurance and commitments recommended.* Enables comparison and avoids misreading of trends.
• Coverage and carve-outs: share of exposure covered; excluded products and timeline for inclusion. *Required for full members where targets cover part of the portfolio (6.1).* Prevents selective reporting.
• Sector hotspots: most-emitting sectors, sector definitions and taxonomy used. *Required for full members: approach to sector selection and boundaries (6.2).* Directs targets to where leverage is greatest.
• Intensity: tCO2e per million of exposure; physical intensities for key sectors where possible. *Required where intensity targets are set (7.1, 8.1); otherwise recommended.* Separates decarbonisation from portfolio growth.
• Own operations: Scope 1, Scope 2 (location- and market-based), material operational Scope 3; tCO2e per employee and per revenue. *Encouraged; outside Protocol scope (6.3).* Shows internal stewardship.
• Data quality: weighted PCAF score by scope. *Recommended (PCAF; peer practice).* Signals reliability and improvement path.
• Restatement policy: triggers and significance threshold for recalculating the baseline. *Recommended (PCAF; peer practice).* Protects comparability over time.

*Source: authors' compilation. Requirements based on NZECA (2024), with section numbers of the Target-Setting Protocol; recommendations based on PCAF (2025a), GHG Protocol (2004, 2015) and peer disclosures.*

Data: improve through the business cycle

The most immediate risk is not the absence of data but the absence of a plan to obtain it. Simple climate-data requests can be embedded in onboarding and renewal processes, contractual clauses can require operational-phase emissions for large new transactions, and the largest emitters in the portfolio can be approached first for verified figures. Each renewal cycle then replaces a proxy with reported data and progressively lowers the weighted PCAF score towards 1, the highest quality.

Targets without haste

Credible targets follow the baseline; they do not precede it. Under the NZECA Target-Setting Protocol, members commit to set 2030 targets within twelve months of joining as full members (18 months for founding members), prioritising the most emission-intensive sectors and using no- or low-overshoot 1.5°C scenarios from the IPCC, IEA or NGFS (NZECA, 2023, 2024). The alliance asks for absolute or intensity targets, preferably both. Intensity metrics keep progress visible when portfolios fluctuate with demand, whereas absolute targets can be met, or missed, through the amortisation of legacy transactions alone. Recent disclosures illustrate how differently the two can move: one agency's oil and gas emissions stood 49 per cent above its 2021 baseline in 2025, while its power-sector intensity was 73 per cent below baseline (UKEF, 2026). Supplementary engagement and transition-finance targets can show how an agency is helping exporters and buyers to decarbonise, which is ultimately where its influence lies.

Recommendations

For ECA boards and management

• Decide the attribution basis early and publish the rationale. Use amount at risk as a transparent, exposure-based starting point; add expected loss as a second, risk-based view where risk models allow, and explain the purpose of each approach.
• Start lean. Concentrate on the information alliance members must report, keep templates short and limit method changes.
• Disclose coverage and data quality alongside reported emissions figures. Name exclusions and a timeline for addressing them.
• Build data collection into the business cycle. Use onboarding, renewals and contract clauses rather than one-off surveys.
• Keep offsets and avoided emissions out of emissions accounting for targets. Report them separately, if at all.

For governments and shareholders

• Align ECA mandates with national transition strategies, including clear and published rules for any sector exceptions.
• Treat first-year figures as a baseline, not a verdict, recognise that restatements may reflect improving data quality, and do not interpret changes in attributed emissions as a direct measure of climate impact.
• Resource the work. A small, dedicated team and access to sector emission data can help to avoid the reputational cost of unreliable numbers.

For alliances and standard setters

• Develop product-specific guidance for guarantees, medium- and long-term credit insurance and short-term business, building on existing member practice.
• Address double counting between ECAs and the banks whose loans they cover, and clarify how overlapping emissions attribution should be treated in aggregate reporting.
• Create shared tools: a common library of sector and country emission factors for emerging markets, and informal peer review of first-time disclosures.

Conclusion

Export credit agencies sit at the intersection of trade, industrial and climate policy. Their guarantees and insurance shape which technologies are built and which assets are locked in for decades. Counting the emissions associated with the transactions they support is the precondition for steering portfolios towards climate goals. The leading agencies show that this can be done before the standards are complete. For those starting now, the lesson is clear: choose a sound method, say what it can and cannot show, and let data quality improve with every reporting cycle. Robust accounting also provides the measurement backbone for the broader agenda of scaling climate finance through export credit (Hale et al., 2021; Klasen et al., 2022) and of aligning international rules with climate goals (Jansen, 2022). Credibility comes from transparency, not from precision on day one.

Bibliography

• Aswani, J., Raghunandan, A. and Rajgopal, S. (2024) Are carbon emissions associated with stock returns? Review of Finance, 28(1), 75–106.
• Busch, T., Johnson, M. and Pioch, T. (2022) Corporate carbon performance data: Quo vadis? Journal of Industrial Ecology, 26(1), 350–363.
• Censkowsky, P., Waidelich, P., Shishlov, I. and Steffen, B. (2025) Quantifying the shift of public export finance from fossil fuels to renewable energy. Nature Communications, 16, 900.
• Euler Hermes (2026) Greenhouse Gas Accounts of the Export Credit Guarantee Portfolio. Hamburg: Euler Hermes Aktiengesellschaft on behalf of the Federal Government.
• Fankhauser, S. et al. (2022) The meaning of net zero and how to get it right. Nature Climate Change, 12, 15–21.
• GHG Protocol (2004) A Corporate Accounting and Reporting Standard, Revised Edition. Washington, DC/Geneva: WRI/WBCSD.
• GHG Protocol (2015) GHG Protocol Scope 2 Guidance. Washington, DC/Geneva: WRI/WBCSD.
• Hale, T., Klasen, A., Ebner, N., Krämer, B. and Kantzelis, A. (2021) Towards Net Zero Export Credit: Current Approaches and Next Steps. Working Paper BSG-WP-2021/042, Blavatnik School of Government, University of Oxford.
• Jansen, M. (2022) Managing the green transition: The role of the OECD export credit arrangement. Global Policy, 13(4), 554–556.
• Kalesnik, V., Wilkens, M. and Zink, J. (2022) Green data or greenwashing? Do corporate carbon emissions data enable investors to mitigate climate change? Journal of Portfolio Management, 48(10), 119–147.
• Klasen, A., Wanjiru, R., Henderson, J. and Phillips, J. (2022) Export finance and the green transition. Global Policy, 13(5), 710–720.
• Krummaker, S. and Klasen, A. (2025) Reinsuring export credit agencies: What drives demand? Journal of World Trade, 59(3), 343–372.
• NZECA (2023) The Net Zero Export Credit Agencies Alliance Commitment Document. Geneva: UNEP FI.
• NZECA (2024) Target-Setting Protocol. Geneva: UNEP FI.
• NZECA (2025) First Progress Report: Charting the Path to Net Zero. Geneva: UNEP FI.
• PCAF (2025a) The Global GHG Accounting and Reporting Standard, Part A: Financed Emissions, Third Edition. Partnership for Carbon Accounting Financials.
• PCAF (2025b) The Global GHG Accounting and Reporting Standard, Part C: Insurance-Associated Emissions, Second Edition. Partnership for Carbon Accounting Financials.
• Peterson, M. and Downie, C. (2024) The international political economy of export credit agencies and the energy transition. Review of International Political Economy, 31(3), 978–994.
• Tang, K., Shrimali, G. and Christiaen, C. (2023) Data Quality Considerations for Estimating Financed Emissions. Discussion Paper, Oxford Sustainable Finance Group, University of Oxford.
• UKEF (2026) Financed Emissions Report 2025 and Basis of Reporting; Annual Report and Accounts 2025 to 2026. London: UK Export Finance.
• UNEP FI (2025) GHG Accounting at Export Credit Agencies and Export-Import Banks: State of Play and Case Studies from NZECA Members. Geneva.