This chapter discusses the practical challenges supervisors face when assessing climate‑related risks associated with net-zero commitments, and how internal data, available to supervisors, may help address them. It examines the main constraints on publicly disclosed information, which include: for banks, differences in the metrics institutions use for the same sector exposures; for pension funds and insurers, the absence of comparable sector-level reporting; and, across all sectors, limited data quality and reliance on estimated emissions.
Net‑Zero Commitments and Prudential Risks in the Dutch Financial Sector
Abstract
This chapter discusses the practical challenges encountered when assessing the climate‑related data that are relevant in the context of net-zero commitments for supervisory purposes, building on the analysis presented in the preceding chapters. While this analysis is largely based on public disclosures, the report considers the supervisory and internal data that prudential supervisors may have access to and use.
First, this chapter introduces how publicly available data – which are a good proxy for which information is currently available – can inform prudential risk assessment. It then highlights limitations in the comparability and suitability of publicly disclosed information, as well as methodological gaps and potential data blind spots that may affect the monitoring of financial institutions’ net-zero commitments and associated transition risks.
The chapter then presents how the DNB, and by extension other supervisors, can address these challenges using supervisory and internal data, often collected directly from financial institutions and complemented with data from commercial data providers. Further, it highlights challenges that internal supervisory data may not be able to address. It then develops considerations for supervisors as well as policymakers. While recommendations for policymakers are not directly addressed to supervisors, they can inform supervisory decision making and engagement with other public authorities, policymakers and the private sector.
The insights derived from this analysis and considerations inform the supervisory framework for assessing risks proposed in Chapter 5.
4.2.1. Risk categories and publicly available information
Financial institutions’ net-zero commitments are relevant from a prudential perspective insofar as they affect relevant risks. This section assesses the suitability of publicly disclosed information for evaluating legal and reputational risks, as well as other prudential risks, including credit, market, liquidity and other operational risks. It focusses on established and widely reported information primarily based on the regulations and reporting practices explained in Chapters 2 and 3.
The usefulness of publicly disclosed information varies across financial institutions. Disclosure frameworks for banks, such as those developed by the EBA, request relatively granular data on exposures, emissions, and credit quality. By contrast, public disclosures by pension funds and insurers typically provide less insight into underlying portfolio risks. This limits their direct use for detailed risk assessment.
Internal data and analyses may address this challenge. DNB, for example, collects asset-level exposure data from financial institutions and combines these with emission estimates, enabling assessment of exposures to emission-intensive activities by economic sector, as further explained in section 4.4. Such analyses may draw on supervisory reporting collected under frameworks such as the CRD VI, as well as on data integrated from other supervisory authorities – such as the ECB – or commercial data providers.
Noting that reputational risk can act as a cross-cutting risk driver across traditional prudential risk categories, the analysis below first considers credit, market and liquidity risk. It then turns to operational risk, where legal risks related to net-zero commitments are most directly situated, before focussing in more detail on legal and reputational risks related to net-zero commitments.
4.2.2. Credit, market and liquidity risk
Prudential risks driven by climate‑related risk drivers may arise where misalignment with net-zero pathways exposes financial institutions to transition pressures, meaning policy, market or technological changes that affect the value, profitability or viability of carbon-intensive activities. These pressures may affect borrowers, issuers, collateral values or financial assets through higher carbon costs, declining demand, technological substitution, stranded assets, market repricing or lower asset liquidity. Supervisors, therefore, need information that combines transition risk indicators with financial exposures and other risk indicators.
Credit risk may arise where transition pressures affect borrowers’ or debt issuers’ ability to meet financial obligations. Market risk may arise where investors reprice equities, corporate bonds or real estate assets before actual credit losses occur. Liquidity risk may arise where climate‑related repricing reduces the marketability or collateral value of assets, increases margin or collateral needs, or affects funding costs and outflows.
Two types of information are particularly relevant. First, sector-level alignment with physical emissions-intensity pathways helps supervisors assess whether exposures are likely to become more vulnerable to transition pressures over time. Second, information on the size and characteristics of sectoral exposures helps assess the potential materiality of the risk. For banks’ loan books, EBA ESG Pillar 3 Templates 1 to 4 are therefore the most useful public information source. They combine sector exposures, financed emissions, credit-quality indicators, maturity, collateral information, counterparty concentrations and sector-level alignment metrics. For pension funds and insurance portfolios, comparable public information is not available. Section 4.4. explains how supervisory data may fill this gap, and what further actions supervisors may consider.
Sector-level analysis is important because climate‑related risks differ across economic sectors. Where financial institutions’ exposures consist of activities that do not transition towards available low-emission alternatives, this may affect borrowers’ profitability, asset values and debt-servicing capacity. Over time, these effects may be reflected in sector-level credit-quality indicators, such as Stage 2 exposures or non-performing loans.
As discussed above, physical emissions-intensity metrics, such as those reported under EBA ESG Pillar 3 Template 3, allow supervisors to assess whether sectoral exposures are aligned with net-zero pathways, implying financing of available low-emission technologies. Deviation from such pathways indicates potential misalignment and therefore provides a clear basis for further supervisory assessment. Where deviation is material and persists over time, supervisors may assess, for instance, whether this is accompanied by deteriorating sector-level credit-quality indicators, including Stage 2 exposures and non-performing loans reported under EBA Template 1. Maturity of exposures and sectoral targets may provide supervisors with further insight into the forward-looking risks and mitigating measures of financial institutions.
Different sectors face distinct economic circumstances. Real estate exposures may be affected by energy efficiency standards and renovation costs. Transport, power, steel and oil and gas exposures may be affected by carbon prices, technology shifts, demand changes and phase‑out policies. Carbon-intensive sectors with long-dated exposures, deteriorating credit quality and weak targeted alignment with net-zero pathways may therefore warrant closer supervisory scrutiny. Further, sector-level assessment is important as aggregate information may hide risk. Exposure to carbon intensive sectors may be obscured at the aggregate level, especially in large portfolios.
Liquidity risk assessment should also consider collateral channels. Template 2 is relevant where energy-inefficient real estate collateral may face valuation pressure, reducing its usefulness as security. This can matter where mortgages are securitised or included in covered-bond pools. If the underlying properties are seen as more exposed to transition risk, the related securities may become less attractive to investors, harder to sell, or subject to higher haircuts when used as collateral. In simple terms, the same mortgage portfolio may raise less funding if markets or central banks assign a lower value to the collateral.
This channel may become more relevant as central banks factor climate risk into collateral valuation. When banks borrow from the central bank, they must pledge assets as collateral. The central bank then assigns a value to these assets and applies a discount, or “haircut”, to protect itself against losses if the asset falls in value. In July 2025, the ECB announced that it would introduce a climate factor in the Eurosystem collateral framework (ECB, 2025[1]). This factor is intended to reflect the risk that certain assets may lose value under adverse climate transition shocks. Where the climate factor applies, the ECB may assign a lower collateral value to the asset, meaning that banks can borrow less against it. While the ECB measure initially applies to marketable assets issued by non-financial corporations, it illustrates the broader liquidity channel: climate‑related valuation risks can reduce the funding capacity of assets used as collateral.
4.2.3. Legal and reputational risk
Legal risk constitutes a form of operational risk, as it may arise from failures in internal processes, controls, disclosures or governance that expose institutions to claims, litigation or regulatory action. Reputational risk is not a standalone prudential risk category in the same way, but can act as a cross-cutting risk driver across prudential risk categories.
Legal and reputational risks may arise where financial institutions’ net-zero commitments are public and their activities are perceived to be inconsistent with their net-zero commitments. While this risk may ultimately depend on how courts and the public assess progress towards net-zero commitments, financial institutions can contribute to shaping this assessment. By providing transparent and decision-useful information, they can support a more consistent and informed understanding of transition progress, reduce the risk of misinterpretation and strengthen the credibility of disclosed net-zero strategies.
As explained in Section 3.6, physical emissions-intensity alignment information (EBA ESG Pillar 3 template 3 for banks; no similar information for pension funds and insurers) allows for the clearest assessment against net-zero pathways and thus for assessing legal and reputational risks tied to net-zero commitments. This is because alignment assessment allows supervisors to understand (past and) forward-looking sectoral exposure trajectories in relation to the net-zero pathways that financial institutions target in the medium term. Based on this information, supervisors can assess whether material deviations from these pathways arise. Where substantial deviations occur in carbon-intensive sectors, supervisors may further quantify the potential risk using publicly available information, such as exposure size (in financial and emissions terms), credit quality indicators (EBA ESG Pillar 3 template 1 for banks; no comparable publicly disclosed information for pension funds and insurers) and other relevant financial data.
In the absence of such information, supervisors and financial institutions may use WACI, including decompositions (see Section 4.4), as a second-best alternative. Additionally, absolute financed emissions and their relative reductions should also be considered, as they are visible to the public and may shape perceptions of progress in the short and medium term. Finally, exposure to high-emitting sectors may affect how the reputation of financial institutions is perceived but is secondary to net-zero pathway alignment.
4.2.4. Other operational risk
Operational risk (other than legal risk) may arise where financial institutions lack adequate data, systems, processes or governance to identify, measure and manage risks related to net-zero commitments. The main risk is not the emissions exposure itself, but that weak data or controls lead to inaccurate disclosures, poor risk assessment, ineffective transition planning or greenwashing-related failures. This risk is relevant because prudential risk assessment depends on data quality, as well as the availability of adequate sectoral data.
Such data, including sector-level alignment metrics, require reliable emissions data, sector classification, activity data and pathway comparisons. Exposure‑based risk assessment requires accurate mapping of financial exposures, counterparties, collateral, maturities and credit-quality indicators. Weaknesses in either part can affect supervisors’ ability to assess both whether transition pressures may become material and how large the exposure could be.
Supervisors may use public disclosures of financial institutions, mostly available for banks, that provide insight into operational risk by explaining the scope, methodology and data sources used for climate‑related metrics. In EBA Pillar 3 reporting, this information is mainly found in the narrative accompanying quantitative templates, especially financed emission disclosures, and in the qualitative information on governance, business strategy and risk management.
Comparable public information is less consistently available for pension funds and insurers. Where disclosed, PCAF data-quality scores can provide additional insight into the reliability of financed emissions’ estimates, which serve as the basis for both WACI and physical emissions-intensity metrics. These scores are not disclosed only by banks: pension funds, insurers and other financial institutions may also publish them when applying the PCAF standard. However, disclosure remains voluntary and uneven. Supervisors may therefore need to assess data collection, validation, governance and board oversight in more detail, as discussed further in Chapter 5.
4.3.1. Diverging benchmark pathways
Physical emissions-intensity data and targets are only decision-useful for supervisory risk assessment when compared against adequate net-zero pathways.
Net-zero pathways vary across countries and asset types within economic sectors. Since the portfolios of banks and non-bank financial institutions differ in their country and asset type exposures, their benchmark pathways may, and likely should, also differ accordingly. However, financial institutions usually do not report the subsector composition of their sectoral exposures. Section 5.2. analyses practical challenges related to the assessment of financial institutions’ sectoral pathways in detail.
Without adequate information on financial institutions’ subsector exposures and the net-zero pathways used to assess them, supervisory assessment becomes less accurate.
4.3.2. Diverging data across financial sectors
Differences in reporting requirements and business models make aggregate comparisons across financial subsectors difficult and potentially misleading. Supervisors may consider assessing which parts of the financial sector harbour the largest climate‑related risks, both in aggregate form and by economic sector. This may include comparing risks in pension fund asset portfolios with those in bank loan books or assessing the aggregate financing of carbon-intensive sectors. However, differences in reporting requirements create uneven data across financial sectors, as discussed in Section 2.5. Banks are subject to more granular disclosure requirements than pension funds or insurers: while pension funds often disclose high-level indicators such as WACI, banks provide more detailed sector-level information for their loan portfolios.
Aggregate indicators can support high-level companions, but they do not provide a sufficiently precise view of prudential risk. Aggregate indicators such as absolute financed emissions or WACI may help compare broad carbon exposure across financial subsectors. However, they do not capture economic sector-level dynamics or the specific context of each financial sector.
Because financial subsectors differ in business models, instruments, maturities and economic sector exposures, comparing climate‑related risk at aggregate financial sector level may be of limited use. Pension funds and insurers may invest in liquid equities and corporate bonds issued by transitioning carbon-intensive companies, while banks may provide long-term, less liquid loans and face stricter supervisory expectations for credit exposures. Insurers may also underwrite carbon-intensive but essential economic activities through short-term policies, which entail relatively lower risk dynamics.
Supervisors should prioritise economic sector-level assessment, as climate‑related prudential risks are economic-sector specific. For example, transition risk in real estate may arise through tightening energy efficiency standards and renovation costs, while transition risk in aviation or shipping may arise through fuel costs, technology shifts or demand changes. Supervisors therefore need information on financial institutions’ exposures by economic sector, rather than only aggregate indicators at financial-institution or financial-sector level.
More uniform economic sector-level reporting would allow supervisors to compare climate‑related risks across financial subsectors more effectively. If banks, pension funds and insurers reported comparable information for their major sector exposures – such as physical emissions-intensity information, or sector-level WACI – supervisors could compare climate‑related risks across loan, equity and corporate bond portfolios. For example, if all institutions reported financed emissions per tonne‑kilometre for transport exposures, supervisors could identify whether one financial sector is more exposed to net-zero misaligned transport assets than another. If the financial sector supports a transitioning economy across economic sectors, an individual financial subsector should not deviate materially from net-zero pathways compared to other financial subsectors. Therefore, observing such economic sector-level differences across financial subsectors may be insightful to supervisors.
At present, publicly available information does not support such cross-sectoral assessment sufficiently (see sections 2.5 and 3). Internal data available to supervisors could however fill part of this gap, enabling more consistent sector-level analysis (see Section 4.4).
4.3.3. Diverging data between institutions in the same financial sector
Business models, objectives and mandates may vary even between financial institutions within the same subsector, for example between promotional banks and traditional commercial banks. As a result, institutions may differ in their sectoral focus, risk appetite and exposure profiles. For example, promotional banks may focus on economic sectors based on public-sector-led activity and investment such as healthcare, education and social housing, and they may have little or no exposure to carbon-intensive sectors such as manufacturing or non-renewable energy generation. Commercial banks may also show varying involvement in specific economic sectors. While some banks may have medium-term exposures to fossil fuel sectors, based on their engagement strategy, others may choose to divest. Based on strategic decisions and built expertise, financial institutions in the same financial sector may also have exposure to different carbon-intensive economic sectors such as agriculture and manufacturing. Therefore, as argued in Section 4.3.2, aggregate comparisons may obscure economic-sector level risks.
Further, institution-level aggregation or comparison of total exposures across similar financial institutions can be challenging and not necessarily insightful. Financial institutions’ different sectoral exposures, as well as their targets, are measured using different metrics. For example, power generation, when reported, is typically measured in gCO₂/kWh, while real estate is measured in kgCO₂/m². Aggregating these metrics to a single institution-level figure would obscure the distinct transition dynamics and risks within each sector. It would also make it difficult to assess whether one financial institution’s portfolio is more or less aligned than another’s.
For the banking sector, current reporting practices already allow for meaningful sector-level comparison. Concerning real estate exposures, all major Dutch banks have significant residential mortgage exposures and commonly report emissions per square metre. These data can be aggregated and compared against an appropriate benchmark, such as CRREM. However, even residential real estate pathways differ across countries because buildings’ emissions depend on national energy mixes, floor area assumptions and property types. Accurate supervisory assessment therefore requires information on the geographical composition of sector portfolios where banks hold material cross-country exposures.
The latest EBA proposal partly addresses this issue for some exposure templates. For Template D 01.00 (as per the EBA’s April 2026 consultation package), for instance, the EBA proposes a country breakdown for loans and debt securities measured at amortised cost. This breakdown is subject to materiality thresholds. Institutions report country-level information where non-domestic exposures exceed 10% of total exposures and separately identify countries representing more than 1% of total exposures, while smaller countries’ exposures are aggregated. However, Template D 03.00 on sector-level emissions intensity and targets is not subject to the same country-breakdown requirement. As a result, supervisors may know the country composition of certain sector exposures, but not necessarily the country composition of the emissions-intensity metrics, targets and the relevant pathway for assessment of alignment with net-zero goals.
Reporting-related challenges also remain pronounced in sectors with heterogeneous subsector composition. Commercial real estate is one example. One bank may be more exposed to warehouses, while another may be more exposed to hotels or retail property. These property types can require different emission reduction pathways because their energy use, operating hours, heating and cooling needs, and current emissions per square metre differ. Differences in subsector exposures may also be the result of portfolio allocation choices. However, without subsector-level breakdowns (over time) it is not possible to assess to what extent banks’ exposures reflect deliberate strategic positioning versus underlying asset characteristics. In April 2026, the EBA’s proposal of new reporting requirements do not foresee mandatory subsector breakdowns based on materiality thresholds. Supervisors therefore may not expect such information to be publicly available in the near term.
Even for similar sectors, banks may use different reporting metrics. For agriculture, Bank A uses a financial intensity metric, reported as mtCO₂e per EUR million financed, and its benchmark is based on the Netherlands’ 2022 Climate and Energy Outlook, rather than a sector-specific 1.5°C physical emissions-intensity pathway. However, Bank B reports selected agricultural exposures, such as soy and dairy, separately using physical intensity metrics, such as tCO₂e per tonne of product. Similarly, for transport-related exposures using mtCO₂/tkm, whereas Bank A provides more granular reporting by separating car and van transport in mtCO₂/vkm and truck transport in mtCO₂/tkm. Comparison of economic sector-level exposures between institutions can therefore be challenging where reporting practices vary.
Varying portfolio coverage across similar financial institutions may also hinder accurate assessment. Financial institutions are still expanding the scope of their reporting as data availability improves. Although coverage varies, large banks already report on much of their carbon-intensive exposure.
Smaller banks, for which similar reporting is not yet mandatory, may need additional time to provide comprehensive and high-quality data. If disclosures omit parts of carbon-intensive assets, comparisons across institutions may become misleading or inconsistent. The same challenge persists for insurance companies and pension funds.
4.3.4. Limited data quality and availability
Data quality
Limited and weak climate reporting data quality puts overall data quality (including supervisory data) into question. Financial institutions rely on emissions data from borrowers, investees and insured counterparties, which is often estimated or unverified.
Data quality is generally poorer for exposures to SMEs, private companies and activities outside public capital markets, such as bank loans and insurance underwriting, where counterparties are less likely to report under the CSRD. By contrast, listed equities and corporate bonds often benefit from broader company-level disclosure and data provider coverage.
Reported data quality varies materially by institution, asset class and emission scope. For large Dutch and other European banks, financed emissions data quality for most loan portfolios is typically around 3.0‑3.5 on the PCAF scale, indicating reliance mainly on counterparty-specific or sector-level estimates rather than verified reported data. For large Dutch pension funds’ investment portfolios, Scope 1 and 2 data for listed equities and corporate bonds can be stronger, with funds often reporting weighted PCAF scores below 2.0 for listed equities and corporate bonds, indicating a relevant share of verified data. However, data quality weakens for Scope 3 emissions, private equity, mortgages and other exposures, as institutions rely more on proxies, sector averages or external estimates for these segments.
Further, the Dutch Authority for the Financial Markets also notes that PCAF data-quality scores should be interpreted with care. Emissions data reported by an intermediary, such as an external asset manager reporting fund-level emissions to an insurer or pension fund, may receive a relatively high score even where the underlying investee company data rely on estimates or sector averages, merely because the data is considered to have been “reported” by the asset manager rather than “estimated” by the pension fund in the example (AFM, 2025[2]).
Estimation methodologies may also differ materially. PCAF notes that limited data is often the main challenge in calculating insurance‑associated emissions and provides data-quality scoring to improve transparency (PCAF, 2022[3]). However, PCAF guidance for insurance‑associated emissions is still evolving and does not yet cover all lines of business. The first PCAF insurance standard covered commercial lines and personal motor insurance, while the 2025 update added project insurance and treaty reinsurance. By contrast, guidance for financed emissions in loans and investments is better established.
Therefore, the Dutch Association of Insurers has set up sector teams to review participating insurers’ CSRD reports, test them against PCAF guidance, discuss differences in assumptions and working methods, and agree on calculation steps for ESG sector standards. These methodological differences reduce comparability, even where institutions rely on the same overall standard. The sector is therefore developing voluntary standards for insurance‑related emissions to improve consistency in reporting (The Dutch Association of Insurers, 2025[4]). Therefore, the DNB may expect more consistent climate‑related reporting of Dutch insurers.
Data availability
Even when banks provide extensive climate reporting on their loan books, disclosure on non-loan balance‑sheet assets often remains limited. These portfolios are often dominated by government bonds but may also include corporate bonds and equities. For example, Bank A discloses EUR 47 billion in “financial investments”, comprising government and corporate debt securities. The same bank reports the WACI of these holdings and benchmarks it against an index but does not provide specific emissions-reduction targets for this portfolio.
For real estate exposures across financial sectors, data-quality issues may persist where energy-efficiency labels are missing, and institutions rely on estimated EPCs or energy-consumption data.
Separately, insurance underwriting and pension fund investments are currently not subject to standardised quantitative climate reporting comparable to EBA ESG templates for bank loan portfolios. While select insurers provide detailed disclosures, this limits public information available for supervisory risk assessment across the sectors.
The AFM’s 2025 review of four major Dutch banks and four major Dutch insurers reached a similar conclusion: differences in scope, definitions, metrics and methodologies continue to limit comparability of financed emissions reporting and transition plans (AFM, 2025[2]).
4.4.1. Internal data
Central banks and supervisors tend to have more data available than what is publicly reported. This internal data helps address some of the comparability and data quality challenges identified above. Unlike public disclosures, supervisory data can provide more granular information on financial institutions’ holdings and positions, including asset-level exposures. This allows supervisors to assess exposures by economic sector, instrument type and, where available, counterparty characteristics.
Internal data refers to both the reported data received by DNB’s supervisory division from institutions under supervision and to data received by DNB’s statistics division from financial institutions with reporting obligations. DNB receives asset-level exposure data from financial institutions, including pension funds and insurers. These data are matched with data on emissions and financials of counterparties, allowing the calculation of consistent institution- and sector-level carbon footprint indicators, such as financed emissions and WACI.
DNB’s internal data are used to calculate carbon footprint indicators, which are also published as part of the ECB climate change indicators (ECB, 2025[5]). This allows for more detailed and economic sector-level analysis of financial exposures in carbon-intensive sectors. Such data also permits DNB to overcome the limitation of WACI usually being disclosed only on the portfolio level (see Section 3.6) and it allows for the calculation of harmonised indicators, meaning institutions can be compared to each other and, by construction, to a financial sectoral benchmark. This data is illustrated in Figure 4.1.
Further, DNB and the ECB use decompositions to better interpret changes in WACI over time (DNB, 2025[6]; ECB, 2025[7]; DNB, 2023[8]). The decompositions differentiate portfolio-weight, emissions, revenues, inflation and exchange rates, helping supervisors better understand changes in WACI. In practice, these decompositions partly overcome the third and fourth limitations (asset and revenue‑driven fluctuations) of WACI, as explained in Section 3.6. Decomposing the portfolio-weight effect allows supervisors to better understand the contribution of emissions reductions and the revenue efficiency. Additionally, adjusting revenues for inflation and exchange rates allows for a more accurate assessment of the real revenue contribution to WACI.
The relevance of such adjustments is illustrated by DNB’s analysis of Dutch pension funds and insurers. Between 2012 and 2019, WACI of pension funds and insurers declined by 34.5% and 31.0%, respectively. After adjusting for inflation and exchange‑rate effects, the decline was smaller, at 24.1% and 23.7% (DNB, 2021[9]).
Internal analyses and data allow supervisors to decompose WACI, substantially improving potential interpretation of emissions trajectories at economic-sector and financial-sector level
Note: Financial sector-level figures are produced as part of the ESCB climate change‑related indicators, while institution-level figures are internally available to DNB users. Emissions data for 2024 are based on nowcasting techniques.
Source: ECB (2025[5]), Climate change‑related indicators https://www.ecb.europa.eu/stats/all-key-statistics/horizontal-indicators/sustainability-indicators/html/index.en.html; DNB (2025[6]), Climate risks for the financial sector https://www.dnb.nl/en/statistics/dashboards/sustainability-in-the-dutch-financial-sector/climate-risks-for-the-financial-sector/, and internal DNB data.
4.4.2. Remaining comparability and data challenges
Internal data and the decomposition of WACI improve the interpretation and suitability of WACI for risk assessment, especially in light of limited public data availability, such as for pension funds and insurers. However, limitations remain.
First, decomposition methods of WACI would benefit from a clearer distinction between changes in portfolio weights caused by market-value movements and changes driven by active portfolio positioning. A financial institution’s WACI may decline because the share price of a carbon-intensive company falls, reducing its portfolio weight, even if the institution has not sold any shares. This reflects passive greening and may indicate that transition risk has already materialised. By contrast, a decline caused by reducing the number of shares or bonds held reflects an active portfolio decision that reduces the exposure of the institution’s exposure to the transition risk. Without this distinction, WACI changes attributed to portfolio weights can remain ambiguous.
Second, revenue effects remain difficult to interpret. A decline in WACI caused by higher revenues may reflect genuine carbon-efficiency gains if a company produces more or higher-value output with the same or lower emissions. However, revenues may also increase due to commodity price shocks, scarcity, or pricing power. For example, oil, gas, electricity, air travel or digital services may become more expensive without any improvement in emissions per unit of physical output. In such cases, WACI may decline even though the underlying activity has not become more carbon efficient.
Decomposed WACI also remains difficult to compare against sectoral net-zero pathways. Net-zero pathways are typically defined using physical emissions-intensity metrics that reflect sector activity, such as emissions per square metre or per kilowatt-hour. WACI is instead based on emissions per unit of revenue and weighted by portfolio value. Even when decomposed, it therefore cannot directly show whether exposures in a specific economic sector are aligned with physical net-zero pathways. This limitation is particularly relevant for legal and reputational risks, where the credibility of net-zero commitments is central, and for market risk, where misalignment may eventually be priced into asset values. Physical emissions-intensity metrics avoid several of these limitations because they relate emissions to physical activity rather than financial revenues or market values. They therefore do not require decomposition to distinguish price‑driven revenue growth from improvements in operational emissions efficiency, or market-value changes from changes in underlying activity. However, they still depend on reliable emissions data – just like the WACI – and sector-specific activity data – which is not always reported.
Therefore, data quality remains a constraint. Emissions estimates may rely on external data providers, proxies or sector-level assumptions, especially where counterparties are SMEs, private companies or otherwise outside public financial markets. Physical activity data may also be difficult to obtain for pension funds and insurers, particularly where exposures are held indirectly through funds or where counterparties do not report the relevant activity indicators. This challenge may persist if the scope of CSRD reporting is reduced, as fewer companies would be required to publish sustainability data.
Indirect investments through funds add a further data challenge. Financial institutions often invest through funds rather than holding all securities directly. To assess the underlying climate exposure, supervisors need to identify the securities or assets held by those funds (DNB, 2024[10]). If a pension fund invests in an investment fund, supervisors need to know what that investment fund owns. Look-through data can reveal indirect exposures where underlying fund-holding data are available. However, it may still leave gaps where funds are foreign, holdings cannot be attributed, or fund-of-fund structures require further look-through steps.
4.5.1. Considerations for supervisors
Legal and reputational risks related to net-zero commitments, as well as other climate‑related prudential risks, may need to be assessed on an economic sector basis. This is because net-zero pathways are generally defined for specific sectors and activities, using metrics that reflect how those sectors decarbonise, such as kgCO₂/m² for real estate or gCO₂/tkm for transport. Climate‑related risk drivers also affect sectors differently and may translate into different prudential risks. In real estate, renovation requirements, energy-efficiency standards or higher energy costs may weaken borrowers’ repayment capacity, increasing credit risk. Where mortgage exposures are securitised, included in covered-bond pools or used as collateral, lower values or weaker marketability may also create market or liquidity risk. This becomes more relevant as central banks, including the ECB, start incorporating climate transition risks into collateral valuation frameworks (ECB, 2025[1]). In power, transport, steel, oil and gas, carbon prices, technology shifts, demand changes or phase‑out policies may affect borrowers’ earnings, asset values and refinancing conditions. Additionally, supervisors may consider differentiating between business lines in insurance underwriting, as the risk profile depends on the insured activity, policy duration, claims exposure and the insurer’s ability to reprice or adjust coverage at renewal. For example, short-term motor or property policies may allow faster repricing than long-term liability or life products.
Assessing progress against net-zero commitments for prudential purposes benefits from the complementary use of climate‑related metrics. Indicators such as WACI and sector-level physical alignment metrics may be used complementarily and depending on data availability and quality.
WACI allows supervisors to compare the emission intensity of financial institutions in monetary terms and to identify institutions with relatively carbon-intensive portfolios. Where decomposed, WACI can also help supervisors distinguish whether changes in portfolio carbon intensity are driven by changes in emissions, revenues, portfolio weights or macroeconomic effects such as inflation and exchange‑rate movements. However, even decomposed WACI remains subject to the limitations discussed above. In particular, it cannot be assessed directly against sectoral net-zero pathways, which are generally defined using physical activity metrics.
Sector-level physical emissions-intensity metrics assessed against relevant reference pathways remain the clearer tool for assessing alignment with net-zero trajectories, and thereby legal and reputational risks that financial institutions may face. They can also support the assessment of other prudential risks driven by climate‑related risk drivers, especially where misalignment points to future transition pressure in material sector exposures. Where reported in a comparable manner, they also enable sector-level comparisons across institutions.
The analysis in this report shows that existing disclosure frameworks support such assessments unevenly. Large banks subject to the EBA’s ESG Pillar 3 disclosure requirements currently provide the most detailed and decision-useful information, including sector-level physical emissions-intensity metrics and targets, as well as information on sectoral financial exposures and financed emissions. Public disclosures by pension funds and insurers are less granular, reflecting differences in regulatory scope and European supervisory expectations.
Where sector-level emissions information is not available and physical emissions intensity metrics are not reported, supervisors may consider assessing sector-level emissions and trajectories based on internal data such as through decomposition of indicators like WACI. However, supervisors need to be aware of the limitations of such analysis as discussed in Section 4.4.2. Supervisors may build such indicators through a combination of supervisory data and emission estimates from commercial data providers. The DNB, for instance, collects information on positions and holdings from pension funds and insurers (supervisory data) and matches this information with emissions estimates and publicly available financial data such as market values, sourced from commercial data providers.
Supervisors may partly bridge these differences with supervisory data and internal data analysis, while recognising the limitations discussed above. Where climate‑related risks in large pension funds or insurers appear material and insufficiently managed, supervisors may consider encouraging these institutions to report sectoral physical emissions-intensity information to DNB, where such information can be obtained at reasonable cost. This may be more feasible in sectors where the underlying physical activity data are already used by companies for business management and reporting. In real estate, for example, floor area is commonly used to report portfolio size or rental performance, making emissions per square metre a relatively accessible metric. Similarly, in steel or cement, tonnes produced are core production indicators, making emissions per tonne of output relatively well available. Where such data are already used in ordinary business steering, supervisors may expect better availability and potentially higher data quality. Where they are not, physical emissions-intensity metrics may be more costly to collect, of lower data quality, and less comparable across institutions.
Data quality, reporting coverage and reliance on estimated emissions affect the extent to which disclosed information can be compared across institutions and sectors. These challenges are likely to remain relevant. Smaller institutions face lighter disclosure requirements, and recent revisions to the CSRD may reduce the expected scope of corporate sustainability reporting. This may limit anticipated improvements in data availability and quality.
Supervisors may therefore need to consider how these developments affect their monitoring strategies and engagement with policymakers. If data quality and reporting scope do not improve sufficiently, supervisors may place greater weight on transition plans, governance arrangements, internal controls and risk management practices.
Supervisors may also consider engaging with regulators and policymakers where the wider economy is observed to deviate substantially from net-zero pathways, as this may affect supervisory expectations towards financial institutions. For example, if economies rely on higher-emitting energy sources for longer than net-zero pathways allow, related activities may become difficult to finance or insure where they are no longer aligned with supervisory expectations. This would require supervisors to consider whether risks stem mainly from weak risk management by individual institutions, or from a broader misalignment between economic activity, policy pathways and net-zero objectives.
In this context supervisors may treat climate‑driven risks arising from insurance underwriting separately, given their distinct business nature. In insurance underwriting, climate‑related risks arise through the liabilities insurers assume when covering policyholders’ risks. Higher underwriting-related emissions, or slower reductions in such emissions, do not necessarily mean that the insurer faces higher transition risk. They may reflect slower transition in the insured real economy. However, prudential concerns may arise where climate‑related risks are not adequately identified, priced or managed in underwriting practices. By contrast, investments held by insurance companies may create climate‑related risks that are directly relevant for prudential supervision.
Where decarbonisation levers depend mainly on public policy or real-economy infrastructure, supervisors should not be expected to address these alone. However, supervisors may communicate relevant risk insights to relevant public authorities, especially where delays in sectoral transition pathways affect the credibility of financial institutions’ transition plans or the prudential risks arising from their exposures. By contrast, if an individual financial institution’s misalignment is institution-specific and substantial, it may warrant stronger supervisory engagement.
4.5.2. Considerations for policymakers
Some challenges extend beyond the direct supervisory mandate. For example, diversified companies may face difficulties determining the appropriate physical normalisation metrics for emissions reporting across multiple sectors. Inconsistent approaches may limit comparability across financial institutions’ portfolios. Clearer guidance on subsector reporting could therefore improve the usability of climate‑related disclosures for financial risk assessment.
Where reasonable and proportionate, regulators could prioritise economic-sector-level information and physical emissions-intensity metrics. Transition risks are sector-specific: real estate, power, transport, steel or oil and gas face different policy, technology and demand pressures. Sector-level disclosures, including physical emissions-intensity alignment metrics, are therefore more informative for assessing progress against net-zero pathways and identifying exposures that may become vulnerable to transition pressures.
Chapter 2 showed that the regulatory framework and supervisory expectations remain fragmented across financial subsectors. This partly reflects differences in business models. However, policymakers may still seek to make disclosure requirements more consistent. This could include reducing reliance on indicators with limited prudential relevance and focussing instead on physical emissions-intensity information and sectoral views where these are proportionate to the size and capacity of financial institutions. Supervisors would then need to bridge fewer differences and remaining data gaps through supervisory data. While supervisory data can be insightful and suitable for risk assessment, it also faces limitations, as explained in Section 4.4.2.
High regulatory asymmetries may also create incentives for risk migration and regulatory arbitrage within the financial system. Financing of carbon-intensive activities may shift from more strictly supervised institutions, such as large banks, to less regulated parts of the financial sector, including pension funds, asset managers or private credit companies. In such cases, transition risks may spread within the financial system while declining more slowly than required for an orderly transition.
For the reasons discussed in Section 3.6, policymakers may also consider deprioritising taxonomy-alignment metrics if the insight they provide is limited for several stakeholders, such as supervisors (for risk assessment) or investors (for capital allocation). This is consistent with the EBA’s April 2026 consultation on ESG supervisory reporting, which proposes removing taxonomy-related reporting templates previously included in the ad hoc ESG reporting exercise. The analysis in this report supports this direction, as taxonomy-alignment metrics are not designed to assess whether financial exposures are aligned with net-zero pathways. Furthermore, financial institutions may incur disproportionate costs to gather this data.
The EBA also proposes replacing the existing template on exposures to the top 20 carbon-intensive firms with a new debtholder-level template on environmental corporate exposures. This change can provide insights beyond sector-level exposures, especially where risks are concentrated in a small number of large counterparties.
The EBA proposal also uses materiality thresholds to limit the reporting burden while preserving relevant country-level information. For Template 1, which includes a detailed sector-level view of financial and emissions exposures, combined with credit quality indicators, institutions would report country-level information where non-domestic exposures exceed 10% of total exposures. Within that breakdown, only countries representing more than 1% of total exposures would be reported separately. Smaller country exposures would be aggregated into an “other countries” category.
Policymakers may consider encouraging a similar approach for physical emissions-intensity metrics where non-domestic exposures are material. For example, if a financial institution has material real estate exposures outside its domestic market, country-level emission intensity information would help supervisors assess whether the reported pathway reflects the underlying portfolio. Similarly, where a subsector accounts for more than 10% of a sector exposure and more than 1% of the total portfolio exposure, institutions could be encouraged to provide a more detailed breakdown. In commercial real estate, this may be relevant where a large share of exposure is concentrated in higher-emitting subsectors, such as certain logistics or warehouse assets, which may follow different decarbonisation pathways than offices or residential property.
References
[2] AFM (2025), Towards transparent reporting on climate transition plans, https://www.afm.nl/en/sector/actueel/2025/dec/sb-klimaattransitiesplannen-emissies.
[6] DNB (2025), Climate risks for the financial sector, https://www.dnb.nl/en/statistics/dashboards/sustainability-in-the-dutch-financial-sector/climate-risks-for-the-financial-sector.
[10] DNB (2024), Adding toes to the carbon footprint, https://www.dnb.nl/media/lc5e2uwt/dnb-analysis-adding-toes-to-the-carbon-footprint.pdf.
[8] DNB (2023), Dynamics of the carbon footprint of financial institutions: a decomposition approach, https://www.dnb.nl/media/oasdsjjg/dnb-analyse-decompositie-duurzamheidsindicatoren-def2.pdf.
[9] DNB (2021), Misleading Footprints. Inflation and exchange rate effects in relative carbon disclosure metrics, https://www.dnb.nl/en/publications/research-publications/occasional-study/nr-1-2021-misleading-footprints-inflation-and-exchange-rate-effects-in-relative-carbon-disclosure-metrics/.
[5] ECB (2025), Climate change indicators: November 2025, https://www.ecb.europa.eu/press/stats/cci/html/ecb.cci251127.en.html.
[1] ECB (2025), ECB to adapt collateral framework to address climate-related transition risks, https://www.ecb.europa.eu/press/pr/date/2025/html/ecb.pr250729_1~02d753a029.en.html?.
[7] ECB (2025), Technical annex: Climate change-related statistical, https://www.ecb.europa.eu/stats/all-key-statistics/horizontal-indicators/sustainability-indicators/data/shared/files/Technical_annex.en.pdf.
[3] PCAF (2022), Insurance-Associated Emissions, https://carbonaccountingfinancials.com/files/downloads/pcaf-standard-part-c-insurance-associated-emissions-nov-2022.pdf.
[4] The Dutch Association of Insurers (2025), Association completes 1st ESG standards and takes next step towards other insurances, https://www.verzekeraars.nl/en/publications/news/association-completes-1st-esg-standards-and-takes-next-step-towards-other-insurances.