Integrating ESG Scores into SME Credit Risk Models
In Scenario Planning for Inflation and Demand Shifts, we examined how small and medium-sized enterprises (SMEs) can connect uncertainty to cash decisions. This first Lending edition asks the corresponding question for lenders: what evidence should change a credit decision when sustainability risks enter the picture?
Our recommendation is to keep the financial assessment as the anchor. Use environmental, social and governance (ESG) information to investigate specific risks to repayment, then validate any proposed influence on the credit model. A sustainability ranking should never be treated as a ready-made estimate of default.
An overall sustainability ranking does not explain why a borrower might struggle to repay. In a historical comparison of six ESG rating providers, Berg et al. (2022) attributed average rating divergence to differences in measurement (56%), scope (38%) and weights (6%). Those results concern the study’s common-company sample, not every provider or today’s SME market. Our recommendation is to examine the underlying indicators before giving an aggregate score credit-model influence.
The numbers describe context rather than a model coefficient
The European Central Bank’s (ECB) Survey on the Access to Finance of Enterprises for the second quarter of 2026 reported that a net 42% of euro area firms that had applied for bank loans, credit lines or bank/card overdrafts reported higher bank-loan interest rates. Separately, SMEs that had used or considered using bank loans reported a net deterioration in loan availability, with a balance of −4% (ECB, 2026, sections 2.1–2.2 and Charts 1–2). Net percentages represent the percentage of firms reporting an increase minus the percentage reporting a decrease. These figures are survey response balances, rather than percentage changes in interest rates or loan rejection rates. We use them to describe the financing environment, not to estimate an ESG effect.
EFRAG’s 2025 VSME Market Acceptance Survey received 282 responses across preparers, users and other stakeholders (EFRAG, 2025, paragraphs 15–23). We treat this online stakeholder survey as early implementation feedback, not as a representative estimate of reporting across all European SMEs. As an analytical precaution, our recommendation is to distinguish financing pressure, reporting capacity and borrower risk before attributing a result to an ESG score.
1 Start with the lending decisionStart with the lending decision before choosing an ESG score
We recommend defining the intended use before choosing indicators and applying a higher evidence standard when a component changes approval, pricing or a credit-risk parameter than when it only helps organise an analyst’s questions.
The Basel Committee’s June 2022 principles call for material climate-related financial risks to be incorporated throughout the credit lifecycle, including client due diligence at onboarding and ongoing monitoring of clients’ risk profiles (BCBS, 2022, Principle 8, para. 35, p. 6). These principles provide a common baseline for climate-related financial-risk management and supervision for internationally active banks and supervisors (Introduction, para. 5, p. 1); their stated scope does not constitute a comprehensive framework for all social and governance issues.
As an additional implementation recommendation, we propose a written use statement for each component identifying the decision, borrower population, outcome, horizon, permitted impact and circumstances requiring human review. We also recommend keeping screening, monitoring and model calibration as distinct uses until evidence supports moving between them.
For an SME finance team, the practical question is equally specific. Ask your lender which information influences the assessment and what evidence would resolve a concern. A clear answer should identify the risk being examined, rather than asking you to improve an unexplained overall score. Build the supporting evidence around that decision. A lender evaluating business interruption needs different information from one testing the financing of an energy-efficiency investment.
2 Collect proportionate data and show its confidence
EFRAG describes the Voluntary Sustainability Reporting Standard for non-listed SMEs (VSME) as a voluntary framework intended to provide a common reference for sustainability data requests (EFRAG, 2026, paragraph 1). Linklaters’ commentary explains the modular approach: micro-undertakings can use selected parts of the Basic Module, while that module is the minimum for other undertakings choosing the standard (Voskoboinikova et al., 2025).
Drawing on this proportional approach, we recommend tailoring the sustainability information requested to the borrower and the lending decision.
EFRAG’s February 2026 Explanatory Note describes its digital template and converter for producing human- and machine-readable Inline XBRL sustainability reports. It also identifies limitations of that release: the template does not include a greenhouse-gas emissions calculator, and the template and converter do not support comparative information (EFRAG, 2026, paragraphs 12, 14 and 28).
Respondents to EFRAG’s market-acceptance survey reported implementation difficulties, including reliable data collection and unclear calculation methodologies, and requested more practical guidance and examples (EFRAG, 2025, paragraphs 38–41 and 67–71). In light of these findings, our recommendation is to use the template as an intake aid, then inspect the supporting evidence for each field relevant to the assessment.
To implement this approach, we propose recording the source, reporting period, unit, verification status and responsible owner alongside each indicator relevant to the assessment. The record should distinguish information drawn from borrower documents from estimates and external proxies. Unavailable information should be explicitly marked as missing; missing data should not, by itself, be treated as evidence of poor sustainability performance.
KPMG’s May 2025 commentary identifies stronger climate-related and environmental data governance and quantitative climate-risk integration as priorities for European banks still developing these capabilities (KPMG, 2025). Building on this emphasis on data capabilities, we recommend automating collection and consistency checks first, while keeping material judgements documented and reviewable. For SMEs, we recommend maintaining a reusable evidence folder and checking its reporting period and updating changed information before reuse. The pack should also be checked against each lender’s specific request.
3 Translate the indicator into cash flow or collateral
For each proposed indicator, explain how it could affect the borrower’s cash flow or collateral value. What could change in revenue, costs, investment, working capital or recoverable collateral value? When could the effect arise, and is any part of it already reflected in existing financial variables?
Frank Elderson’s May 2026 ECB supervisory blog describes some banks translating scenario assumptions, such as carbon prices and green investment needs, into company financial ratios and feeding those ratios into internal ratings to derive stressed probabilities of default (Elderson, 2026, “Transition risk modelling at the counterparty level”). It also describes physical-risk analysis connecting asset location and hazards to business interruption or collateral effects. These are illustrative, non-binding good practices that establish no new legal or regulatory requirements. The blog describes modelling approaches; it does not provide ready-made coefficients for an SME portfolio.
Consider a hypothetical manufacturer seeking finance to replace equipment with a more energy-efficient alternative. We would examine the investment cost, financing schedule, expected savings, installation disruption and available cash headroom separately. The transition plan might reduce a future vulnerability while creating a near-term funding need.
An aggregate score should not determine the lending assessment on its own. Our recommendation is to retain the underlying risk and cash-flow evidence, including uncertainty around projected savings. Identify where each effect belongs—in the financial forecast, collateral assessment, monitoring plan or a separately validated risk adjustment—and avoid counting effects already reflected elsewhere. For the SME, useful preparation means an implementation plan showing funding sources and an adverse case, without assuming that greener operations will secure cheaper credit.
4 Separate sustainability progress from repayment capacity
The supplied OECD July 2024 draft policy note argues for combining finance with tools, advice and capacity building to support SME sustainability reporting and investment (OECD, 2024, pages 11–12). It is a draft policy discussion document, not a binding legal instrument. Drawing on this rationale, we recommend assessing transition capacity alongside sustainability performance: identify committed funding, an accountable owner and milestones, and distinguish an ambition from expenditure already undertaken.
Assess three dimensions separately: sustainability risk, quality of evidence and repayment capacity. A borrower may have an ambitious plan, incomplete reporting and adequate financing. Another may have extensive disclosures but an investment schedule its cash flow cannot support. These examples illustrate questions to investigate; they do not establish a relationship between disclosure quality and default risk.
In the EIB Investment Survey 2025, 39% of EU SMEs reported investment in energy-efficiency measures in the previous financial year, compared with 62% of large firms (EIB, 2025, page 20). Here SMEs means firms with five to 249 employees; interviews ran from April to July 2025 and the report uses value-added weighting. These findings motivate our recommendation to examine actual expenditure and funding; these figures do not assess disclosure scores or default risk.
For SMEs, our advice is to present both the sustainability objective and the financing mechanics. Explain what is committed, what remains dependent on funding and which milestone would trigger a revision to the plan. For lenders, retain existing management-quality and financial assessments, and check whether sustainability score components duplicate factors already captured in those assessments, to avoid rewarding or penalising the same underlying factor twice.
5 Validate the additional information before giving it influence
Bonacorsi et al. (2024) found that selected ESG factors added explanatory information beyond accounting variables for an adjusted Altman Z-score credit-risk proxy in a 2019 cross-section of European listed companies; they discuss potential endogeneity and disclosure-related selection bias.
Ferriani and Pericoli (2024) found that higher ESG scores were associated with lower Moody’s EDF estimates of default probability among European listed non-financial firms in the Eurostoxx 600 over 2014–2022, with larger estimated associations at longer horizons.
Kažytė et al. (2026) analysed 548 firm-year observations from Euro Stoxx 50 firms over 2005–2024 and found a positive association between ESG disclosure scores lagged by two years and Altman Z-scores. ESG disclosure, financial-distress proxies and model-estimated default probabilities are distinct constructs and measures. The observational results do not establish causality or provide a calibrated loan-level default model for non-listed SMEs.
Our recommendation is to start with a comparison against the existing financial assessment. Test whether the proposed component improves credit-risk assessment beyond that baseline across relevant borrower groups and periods, using data separate from those used to develop or tune it. Review predictive performance, stability, missing-data handling, analyst overrides and the reasons for changed decisions. An improvement on development data alone is insufficient evidence for deployment.
Ragulskis et al. (2026) propose sector-specific SME ESG scoring using questionnaire data drawn from public disclosures and weights optimised against expert sustainability rankings. Their applications cover meat/dairy and agricultural SMEs, not observed loan defaults. The authors acknowledge voluntary-disclosure and sector limitations. Our preferred initial use is structured analyst challenge or a monitored trial. Greater influence on approval or risk-based pricing would require independently validated, repeatable improvements in credit-outcome prediction beyond existing underwriting, reliable scoring across relevant portfolio segments, and evidence of improved decision outcomes. Trial success criteria and acceptable uncertainty should be set before outcomes are observed.
Keep scenario analysis distinct from evidence of empirical predictive validity. Elderson (2026) describes stressed default probabilities derived by translating scenario-specific developments into financial impacts and applying rating models. Our recommendation is to report the scenario or pathway, the PD horizon and key assumptions and uncertainties, and to assess empirical calibration and predictive performance separately. A stressed scenario output should not be presented as an empirically validated one-year default estimate without evidence supporting that interpretation.
6 Govern providers and decisions throughout use
An ESMA Q&A containing the European Commission’s answer dated 18 March 2026 explains the Article 2(2)(b) exemption for ESG ratings issued by regulated financial undertakings and used exclusively for in-house or intragroup financial services or products, provided the ratings are not published or distributed to third parties. Examples include risk management, internal lending processes, and using an undisclosed rating to inform whether to offer a client a financial service or product (ESMA, 2026). Application depends on the actual use and disclosure arrangements; an internal ESG rating does not automatically require ESG rating provider authorisation. The Q&A addresses regulatory scope, rather than suitability for a particular lending decision.
To govern the use of ESG ratings and assessment components in lending, we propose maintaining a record for each external provider and internally developed assessment component.
The record should include the methodology version, covered entities, assessment horizon, missing-data treatment and permitted uses. Document how changes are communicated and whether underlying fields remain accessible. Require a documented impact review before adopting a methodology change that could materially alter borrower assessments.
Basel’s 2022 climate-risk principles call for clearly assigned responsibilities across the three lines of defence and explicit communication of relevant data limitations (BCBS, 2022). Consistent with this guidance, we recommend assigning application to the credit team, independent challenge and monitoring to risk management, and independent assurance to internal audit. As an implementation of the responsibility-assignment guidance in Principles 2 and 4, document who may approve changes to the component or reduce or suspend its influence.
We recommend monitoring use through a short dashboard covering evidence coverage, proxy use, stale fields, overrides, decision changes attributable to the component and subsequent borrower outcomes. Set review triggers locally against the portfolio and approved use. If a component cannot be explained or its performance deteriorates, route affected cases to review and reconsider its influence. Alongside this monitoring, provide SMEs with a practical route to correct inaccurate information, supply missing evidence and understand the reason for a material assessment change.
Conclusion
Our position is that integrating ESG scores into SME credit risk models is a design and evidence task. Start with a defined lending decision, proportionate information and an explanation of how each material indicator could affect repayment or collateral. Keep the financial assessment as the anchor, and assess sustainability risk, evidence quality and repayment capacity separately.
We recommend beginning with structured analyst challenge or a monitored trial in a defined borrower segment. Agree the intended use and success criteria before observing outcomes, document the data and compare the proposed component with the existing financial assessment. Greater influence on approval or risk-based pricing should depend on independently validated, repeatable improvements beyond existing underwriting. Record changed decisions, their supporting evidence and subsequent borrower outcomes.
For SME finance teams, prepare a reusable evidence pack and a transition plan showing committed funding, financing dependencies, milestones and explicit assumptions. This preparation should make the credit discussion more concrete without implying approval or preferential pricing. The outcome to seek is a decision whose reasons both lender and borrower can understand and challenge.
Key takeaways
Define the lending decision and permitted use before selecting an ESG score.
Use proportionate VSME-aligned intake and retain supporting evidence for information material to the assessment.
Assess sustainability risk, evidence quality and repayment capacity separately.
Connect each material indicator to cash flow or collateral, and avoid counting effects already captured in the financial assessment.
Test additional value against existing underwriting using data separate from development, before increasing influence on approval or pricing.
Assign owners, review methodology changes, monitor decision outcomes and provide a practical route for borrower corrections.
Strategic implication
For boards and credit committees, our proposed approval question is: can management demonstrate why this component belongs in this lending decision and what evidence supports its permitted influence? Require a documented use, traceable data, validation proportionate to the intended impact and ongoing monitoring. Assign authority to reduce or suspend influence when the component cannot be explained or its performance deteriorates. These are our proposed controls for keeping ESG integration accountable to its purpose and evidence.
Ready to pressure-test your risk framework?
How would your organisation explain the role of ESG information in a lending decision? Whether you assess SME borrowers or prepare evidence for your lender, consider where uncertainty remains—in the data, the financial impact or the decision itself. Explore those questions through Amaranth Brose’s Free Risk Discovery, using this article’s framework as a starting point for discussing your own approach.
Discuss the data, validation and governance of your SME lending approach with Amaranth Brose. Book a risk advisory consultation or explore our advisory work.
What is next
The next Brave Horizons Lending edition, on 27 October 2026, is Managing Geopolitical Sanctions Risk in Trade Finance. We will explore how sanctions exposure can affect a transaction’s counterparties, payment routes and financing arrangements—and what lenders and SME finance teams should examine before committing.
The central question: when a transaction appears commercially sound, what geopolitical risks could still interrupt payment or make the financing unworkable? Expect a practical discussion of due diligence, escalation and the evidence needed to support a defensible decision.
Comments