Response to discussion on the simplification and assessment of the credit risk framework
Q1. For the purpose of reporting under CRR Article 430a, which definition of loss should be used?
Referring to Box 1 there are two suggestions regarding the numerator and the denominator.
Numerator: For the calculation of the losses we use specific provisions and write offs as loss estimate. These figures are easily obtainable and easily compatible with other data used for the RWA calculation. A new definition would probably make the calculation more difficult. We respectfully ask the EBA to consider the proportionality between the benefits and costs of a new calculation
Denominator: The Standardised and IRB approaches define the exposure value differently. A new definition, which would prefer one of the approaches would be to the disadvantage of the other approach. The exposure values are the same for RWA calculation and the IP Losses which is easily comparable for the banks. In the interest of standardization and simplification, the current calculation method of the respective approach should not be changed.
Q2. Should the loss data (CRR Article 430a) be used for the assessment of RWs of real estate exposures under CRR Article 126(4) and CRR Article 465(11)?
The weaknesses of the calculation under article 430a CRR mentioned by the EBA would therefore not only be the basis for the hardtest but also be used for the risk weights of commercial immovable property under 126 (4) CRR. The IP Losses data is only collected once a year. We kindly ask the EBA to consider more regular sources like COREP C07 or more granular sources like AnaCredit Data in the first analysis of this topic.
Q3. Which elements of the real estate framework should be further simplified?
It’s highly complex to differentiate between an IPRE and a Non-IPRE Exposure and the scope of ADC exposures is too extensive broad.
In order to accelerate the building of public houses, exposure to public housing companies or not-for-profit associations that are regulated by law and exist to serve social purposes and to offer tenants long-term housing should be excluded from the scope of ADC exposures. Non-profit building associations do not build speculative projects. Instead, they focus on urgently needed affordable housing which corresponds to the European Affordable Housing Plan published in December 2025.
Simplification always includes a clear description of the regulations. The regulator has given banks leeway in interpretation, particularly in the area of real estate. The discussions between risk software providers and banks are time consuming. Furthermore, differing interpretations of the regulations lead to distortions of competition between the banks with different software providers. For example, how should the ETV, article 125 (2) or 126 (2) CRR be calculated if there are mixed securities (one RRE and one CRE)? One software provider uses only the security with the better risk weight, another software provider split the exposure and uses both securities.
We respectfully request the regulator to answer the questions in the rejected EBA Q&A 2025_7422 (Calculation of the Exposure-to-Value (ETV) ratio and risk weighting when dealing with complex collateral structures, https://www.eba.europa.eu/single-rule-book-qa/qna/view/publicId/2025_7422)
Transparency ‘Hard Test’
Currently, it’s very complex to verify for each real estate market in every Member State, whether the so-called “hard test” set out in Article 125(2), third subparagraph, and Article 126(2), third subparagraph, CRR has been met. To enhance transparency in the application of the so called ‘hard test’, we recommend introducing (i) a reporting obligation requiring the competent authority of the relevant Member State to report its assessment to the EBA or the ESRB, and (ii) disclosure obligations requiring the EBA or the ESRB to publish the outcome of that national assessment on its website.
[1] 2025_7422 Calculation of the Exposure-to-Value (ETV) ratio and risk weighting when dealing with complex collateral structures | European Banking Authority
Q4. Which other clarifications do you consider necessary to apply the new ECAI framework?
In our opinion, the use of ECAI Ratings is clearly described in the CRR.
General Comments - Standardised approach: In the KSA there’s high potential to simplify the framework for credit risk. This is especially true for the calculation of real estate exposures and the so called hard test.
Q5. Should the consolidation of regulatory products for credit risk be a priority or should the regulatory stability be preferable instead? Have you identified any redundancies in IRB products?
Yes. From our perspective, a consolidation of the various regulatory products related to credit risk would be highly beneficial and is preferred over stability. Streamlining these products can enhance clarity, reduce complexity, and improve efficiency in risk management. The current landscape consists of multiple documents issued by different authorities—including the ECB’s Guide to Internal Models (2024), the EBA’s Guidelines on PD and LGD Estimation (2017), the EBA’s Report on the Role of Environmental and Social Risks in the Prudential Framework (2023), the EBA’s Guidelines on ESG Risk Management (2025), and the ECB’s Guide on Climate‑Related and Environmental Risks (2020).
These documents partly overlap in scope yet differ in emphasis, depth, and timing of expected implementation. This can result in inconsistencies, interpretational ambiguities, and supervisory expectations that are not fully aligned—particularly with respect to the treatment of ESG risk drivers in internal models.
A more streamlined and integrated set of regulatory products would therefore support greater coherence, reduce redundant or conflicting requirements, and enhance regulatory certainty for institutions. While regulatory stability remains important, we see clear value in prioritizing consolidation where it improves consistency and reduces interpretational complexity, especially for IRB‑related frameworks.
Q6. Do you consider that the integration of environmental and social risks into the credit risk framework could be further enhanced without undermining its simplicity? Which areas, if any, would you prioritise for further work or clarification?
Yes. We are convinced that the integration of environmental and social risks into the credit risk framework can be further enhanced. From our perspective, the areas requiring additional clarification relate primarily to the treatment of ESG risk drivers in internal models, where current regulatory expectations across different guidance documents are not fully aligned.
For example, the ECB’s Guide to Internal Models (2024) explicitly expects institutions to incorporate all relevant and material ESG risk drivers into PD and LGD models, supported by modelling data that sufficiently captures these risks. At the same time, the guide requires the application of margins of conservatism (MoCs) where information is missing or uncertain, and the use of (conservative) overrides if ESG drivers cannot yet be directly included in the models.
Conversely, the EBA’s Guidelines on PD and LGD estimation (2017) and its Report on the role of environmental and social risks in the prudential framework (2023) take a more cautious approach. They emphasise that ESG risk drivers should only be integrated into the risk differentiation model when sufficient and reliable information is available and model performance is not adversely affected. The EBA further stresses that model redevelopment or recalibration should occur only in the medium to long term, once the actual impact of ESG risks on defaults and losses becomes empirically observable. Until then, MoCs and selective, temporary overrides may be applied, but only within the boundaries of existing guidelines and only for a limited number of cases.
Additional regulatory documents highlight further operational challenges for the industry, such as the limited availability and representativeness of historical ESG‑related data, the uncertainty surrounding transmission channels, and the need to consider both micro‑ and macroeconomic effects of environmental risks. They also underline that ESG risks may affect both default likelihood and loss severity, and that institutions must ensure robust data processes, scenario-based assessments, monitoring of ESG-related losses, and sustainable collateral valuations over the life of the exposure.
As highlighted in paragraph 26 of the discussion paper, the primary challenge in effectively integrating environmental and social risks into the credit risk management framework is the insufficient availability of adequate data, particularly historical data. Clearer guidance on transitional requirements—such as the extent to which human judgment may be applied until sufficient data history is gathered, and the expected threshold for including or excluding factors analytically—would be beneficial.
Given these partly diverging expectations, it would be highly beneficial for the sector if regulators provided a more harmonised and pragmatic framework for integrating ESG risks into internal credit risk models. In our view, priority should be given to:
- clarifying when ESG risk drivers are expected to be directly integrated into PD/LGD models versus when MoCs or overrides are more appropriate;
- specifying minimum data quality and availability standards that would justify model redevelopment;
- providing guidance on how to balance model simplicity with the increasing complexity of ESG risk considerations; and
- ensuring consistency between short‑term supervisory expectations and the longer‑term empirical evidence required for model stability and performance.
Such harmonisation would reduce interpretational differences, avoid unnecessary supervisory discussions during IMIs and OSIs, and support a more consistent, efficient, and proportionate integration of ESG risks into the credit risk framework.
Additionally, creating a shared data source for physical risk accessible to all institutions could help harmonize assessments and reduce inconsistencies.
Q7. Which requirements should apply in relation to the measurement of the performance of continuous models (e.g. Back-testing)? How could testing requirements be facilitated and enhanced for continuous models that are compliant with CRR, Part three, Title II, Chapter 3, Section 6 (Requirements for the IRB approach)?
To ensure comparability with existing IRB requirements, the discretization applied to continuous models should mirror the minimum standards currently applicable to rating grades. In particular, the discretization framework should support an equivalent level of homogeneity and heterogeneity across segments and enable back‑testing at a grade‑like level of granularity. This would allow continuous models to be assessed using performance measurement techniques that are consistent with established IRB validation practices.
Q8. Which requirements should apply in the application phase of continuous models (e.g. overrides)?
Overrides in the application phase of continuous models should be implemented as log‑odds shifts, applied continuously across the risk spectrum. This approach ensures that overrides remain consistent with the underlying functional form of the model and preserves the inherent rank‑ordering of obligors. By expressing adjustments as shifts in log‑odds rather than as discrete grade movements, institutions can avoid distortions in risk differentiation while maintaining full transparency and traceability of the override decision.
Q10. Should a consistent and single facility definition be applied across all risk parameters?
The consistency discussion should be expanded to cover the definition of the client level. This might help institutions to avoid inconsistencies when interpreting legal requirements relating to the joint credit obligations (e.g. in risk discrimination and quantification).
Q11. Are adjustments proposed in the representativeness requirement for the CCF parameter also suited for PD and LGD risk parameters? Which amendments would be needed to accommodate PD and LGD specificities?
Yes. We would strongly welcome further harmonisation, simplification, and clearer specification of the representativeness requirements, according to the CCF parameter. At the same time, certain amendments are necessary to adequately reflect the specificities of PD and LGD models. In particular, the formulation in paragraph 52 of EBA/CP/2025/10 regarding key risk characteristics would require adjustment for PD models.
Generally, emphasizing model performance over strict sample representativeness should also be applied to the PD and LGD guidelines, particularly when evaluating the representativeness of data used in model development.
Q12: Do you consider further simplification of the representativeness requirement, as proposed for the CCF parameter, as necessary for PD and LGD and if so, what kind of simplification?
Yes. We would strongly welcome further harmonization, simplification, and clearer specification of the representativeness requirements, according to the CCF parameter. At the same time, certain amendments are necessary to adequately reflect the specificities of PD and LGD models. In particular, the formulation in paragraph 52 of EBA/CP/2025/10 regarding key risk characteristics would require adjustment for PD models.
Q13. Should these simplifications be pursued? Do you have any preferred approaches with respect to these simplifications?
Simplifying the assessment dimensions—such as the scope of application and the distribution of key risk characteristics—and adopting the streamlined analysis approach similar to that in the CCF guidelines would be beneficial.
Overall, we strongly welcome simplification of the regulatory framework and believe that it should be pursued. However, the situations for which the envisaged simplified approaches can be used shall be clearly defined to avoid additional complex discussions on whether their application is allowed. In case they require strong justification, or the back-testing requirements are strict, such methods do not constitute simplifications but only trigger additional discussions that increase the complexity of the regulatory framework.
We are fine with the specific points in the simplification apart from the following two:
- C10. For MoC C, we do not believe that a standardisation is possible, given the heterogeneous portfolios that it is applied to.
- C15. The introduction of yet another approach for calibrating LRACCF is only helpful if that is the only approach. If differences between the approaches must be explained, this triggers additional analyses and creates additional modelling burden on banks; this proposal comes at an unfortunate point when it is clear to most banks that the subsample of observations 12-months prior to default is not a random subset of all facilities defaulting within the next 12 months. At the same time, the CRR clearly prescribes the fixed-horizon approach.
Q14. Do you have any comments and suggestions with reference to the calibration of the fall back approaches?
Please explicitly state that the simplified approaches are conservative by nature. In other cases, they risk being treated as benchmarks rather than fallback values and might only trigger additional discussions.
Defining fallback values for PD and LGD presents a greater challenge compared to the CCF parameter. Simple methods, such as applying a relative increase by a fixed percentage, could offer an easy-to-implement solution and be calibrated according to the current MoC level of supervised entities. Additionally, quantile-based approaches may also be worth considering.
Q15. Do you see other potential simplification areas where the modelling burden is not commensurate to the gain in risk sensitivity?
If the goal of simplification is to effectively reduce the bank’s modelling burden without materially impacting their risk sensitivity, the concept of non-compliance with negligible effect shall be introduced.
According to par. 50 of the EBA GL on PD/LGD estimation, institutions should “correct the models to ensure their full compliance with the requirements of Regulation (EU) No 575/2013” and that “all data and methodological deficiencies as well as any other potential sources of additional uncertainty shall be rectified within a reasonable timeframe”. However, in certain situations, rectifying a deficiency or eliminating non-compliance with regulatory standards causes extra work that can double the workload for the bank, even where the impact on model outcomes is negligible.
For example, it is a standard practice in many European banks to use the outstanding amount at default to isolate outliers in the LGD model. However, this stands in direct contradiction with par. 122 and 162 of the EBA GL on PD/LGD estimation, which mandate the use of appropriate risk driver values within the year before default to isolate outlier values. At the same time, even if such risk drivers were identified for a certain sample, the stability of such relationships over time would be questionable.
Further on, we observe a common practice among supervisors to challenge discriminatory power or homogeneity in any sub-segment of the application portfolio by arguing non-compliance, even if the relevant sub-range is considered immaterial and any effect on the estimates is negligible. In this context, the criteria for homogeneity and heterogeneity of rating grades could be further refined, potentially through standardized methodologies, to avoid disproportionate modeling effort.
Several aspects of modeling low default portfolios—such as backtesting, using alternative targets instead of observed defaults, and demonstrating risk differentiation—significantly increase modeling complexity. However, current guidelines generally lack specific guidance for these special cases. Providing tailored guidance for low default portfolios would therefore reduce unnecessary complexity while maintaining risk sensitivity.
Some institutions employ master scales to ensure comparability across different rating models, but meeting regulatory requirements (e.g. backtesting, homogeneity, heterogeneity, and adequate population of rating grades) often conflicts with this approach, resulting in modeling challenges and supervisory issues. Providing guidance on the use of master scale models or introducing tailored requirements would be helpful.
Additional simplification is also necessary in the rating process area for exposures with partial information. Currently, overly conservative ratings are mandated when certain information is not available, even though the rating is still conducted based on other available information (e.g. for newly founded companies—start-ups—financial information might not be available for a limited period of time during which the rating is derived from behavioural information).
Q16. What do you perceive as challenges in your capacity to collect appropriate data, in particular in relation to indirect costs?
Collecting appropriate data for indirect costs is not considered an unattainable challenge for some institutions.
On the other side, enforcing the full replication (back-simulation) of the current Definition of Default to cover the entire data history (spanning more than 20 years) creates additional complications, especially since those data are in most of the cases not available or only available to the limited extent.