We see that it is still uncertain how environmental risk will impact financial risk over time. Among reasons, there is the non-linear and forward-looking nature of environmental risks. Environmental risks are expected to increase gradually over time, and this could coincide with environmental shocks that cannot be predicted.
However, we see that it is too early to establish whether the risk within the industry overall will grow, leading to different assumptions on optimal capital levels, or whether environmental risks would prompt a reassessment of risk profiles and a consequent reallocation of capital requirements with neutral effects on the overall capital needs.
One of the key challenges for assessing the double materiality of climate risks from a prudential perspective is the fact that climate risks show an endogenous component, i.e. the decisions that investors, firms and policy makers may take today about climate change affect the realisation of risk across climate scenarios.
This aspect has been analysed in a recent working paper of the ECB (2665/ May 2022) which illustrated how if financial actors have climate sentiments, i.e. they form expectations about the impacts of climate scenarios on their investments, they could adjust their investment decisions, thus influencing the outcome of the transition. In contrast, if financial actors do not trust the introduction of climate policies, and do not adjust their risk assessment, they would maintain their exposures to high-carbon firms, or may even increase them. Since capital allocation depends on financial risk assessment, a delay in capital reallocation from high-carbon to low-carbon activities would occur, increasing the costs of the transition, and decreasing its feasibility.
The ECB paper suggests that this should lead to embed endogenous macro-financial feedback loops, and firms’ expectations, in climate stress testing exercises and that double materiality principle could not only enhance disclosures but also support the calibration of prudential instruments. While we could see the rationale for reflecting double materiality in disclosures and for embedding feedback loops in stress testing, we do not believe that double materiality should be used for Pillar 1 purposes.
Pillar 1 requirements are notably set to ensure that banks have sufficient amounts of regulatory capital to cover unexpected financial losses caused by the financial risk categories of credit, market, operational and counterparty credit risk. Climate risks are drivers of existing risk categories and could hardly be disentangled in a manner that could produce reasonable benefits.
Those risk are also by nature forward looking and institution specific, therefore most apt to be addressed in the supervisory dialogue that takes place as part of the SREP (which is a holistic, forward looking and bank specific exercise as well).
Also, it should not be underestimated that this would constitute a fundamental conceptual intervention on the mechanics of the Pillar 1 framework, which could place EU banks at a disadvantage compared to other lenders.
We appreciate the risk-based approach being followed by the EBA, and we would like to emphasize the following aspects:
▪ Green activities (i.e. those which are aligned with the EU Taxonomy) do not automatically translate into less risk for credit institutions. It is stated in the DP that prudential requirements should reflect the risk profiles of exposures and should not be used for other policy purposes. We fully support this view. The same is true for activities that are not green in terms of Taxonomy alignment, these cannot be automatically assumed to bear more risk.
▪ Currently available data make it rather complex to reflect ESG risks e.g. into internal models. The uncertain and forward-looking nature of ESG risks over longer than usual time horizons add to this problem. Therefore, the requirement to integrate ESG risk into Pillar 1 prudential requirements and internal models would have to be done very carefully in order not to distort the picture of risks a credit institution is exposed to. The current risk categories are still the best suited for risk capture.
The recently concluded ECB climate stress test1 has already highlighted areas where banks need to improve, much before even considering any action on the Pillar 1 framework.
In fact, the situation is still heterogeneous, and banks still have to enhance their climate risk stress-testing frameworks to account for various transmission channels and asset classes, covering both physical and transition risks. This would also mean incorporating climate risk scenarios into stress-testing models, reflecting both physical and transition risks, long- and short-term horizons – and integrating climate risk stress tests into their ICAAPs where that has not yet been done.
Certainly, a lot of efforts will also go into improving climate-relevant data collection by engaging with customers and improving proxy assumptions.
Often the necessary environmental data may not be available. This would result in portfolio-level estimation based on national or regional averages. In addition, this could lead to unjustified reputational risk for the bank if its assessments confer a lower level of risk and are considered/perceived as greenwashing.
The current Credit Risk Mitigation (CRM) framework does not need at this stage adjustments in order to further account for environmental factors, whether they have positive or negative impact.
The framework for exposures secured by immovable property is already moving towards more granularity (and complexity) in the CRR3 proposal. The proposed clarification in Art. 208 CRR that energy efficiency improvements unequivocally increase the property value is a welcome step.
The (re)evaluation of collateral is of particular relevance for immovable properties serving as collateral for residential and commercial mortgage loans, particularly since under the new Standardised Approach, the risk weights would be dependent on the loan-to-value (LTV) ratios.
The positive relationship between the value of the property and its energy efficiency needs to be reflected in the prudential framework through the lower LTV levels and related lower risk weights, also since there may be a positive relationship between the energy efficiency of a loan and its credit performance.
We would like to stress that the framework of the Standardised Approach should be kept simple and understandable. There is an evident need for consistency and alignment in the wider framework: several different Regulations, Directives and Guidelines deal with the topic of energy efficiency and overall progress towards the goals of the Paris Agreement. For example, the Directive on the energy performance of buildings is under review at the moment, and according to the EBA Guidelines on Loan Origination credit institutions have to take energy efficiency into account when having exposures secured by immovable property.
It should be noted that high-quality Object Finance and high-quality Project Finance exposures benefit from an enhanced due diligence to ensure that financial risks are adequately covered. This includes the study of environmental factors which are carefully considered. Therefore, a risk-based approach should be followed and double-counting should be avoided. Environmental considerations, with the transition and physical risks, would be best taken into account within the cash flows projections of the borrower when possible, and within the risk assessment of the transaction. There should not be an adjustment factor on top of other factors of selection of high-quality transactions.
Therefore, we think that it would be unnecessary to add environmental criteria for the eligibility to the sub-asset classes HQ Object and HQ Project finance since environmental factors are sufficiently taken into account due to the very nature of these activities.
As stated in the paper, simplicity and risk sensitivity need to be balanced. However, if the objective is to integrate environmental risks, risk differentiation in the corporate exposure class could be justified. Exposure to environmental risks in this class can differ immensely. Differentiation in the retail class would be more challenging and perhaps too complex.
The SA framework partially relies on external credit ratings. In case the European co-legislators decide to further strengthen the role of ESG-factors in the CRAR (Regulation (EC) 1060/2009) or the ESMA decides so in a review of its 2019 Guidelines on Disclosure Requirements Applicable to Credit Ratings, these standards should not be applied for internal models.
In our view, there is no need to revise the level 1 IRB framework at this stage. Until the common data foundation has significantly matured, there should be ample flexibility for banks to integrate environmental risks into IRB models where relevant and how seen appropriate.
Banks have the best understanding of their customers’ risks and regulators should allow banks to adequately integrate these risks in their models. For example, the introduction of an override on rating classes could be a solution to adopt. The cases in which this risk is significant should be reflected in rating and funding decision. In addition, banks could integrate these factors in the definition of risk appetite with targets for progressive reduction of high-risk exposure. This would also avoid the modification of the formula (e.g. the correlation coefficients), which would be very difficult to implement and which could imply double counting of risks, since the best way to integrate these risks is within ratings.
Supervisory expectations on banks’ stepwise implementation of the C&E factors in their IRB approach should be further elaborated.
Current EBA Guidelines on PD and LGD estimations (EBA/GL/2017/16) were drafted more suitably for High Default Portfolios so that many requirements are designed for statistical models.
In particular, the lack of observed data ends in substantial margin of conservatism. This makes it difficult to capture the risk associated with environmental criteria that are not substantiated by a sufficient number of observations.
Therefore, EBA Guidelines would deserve an update to recognize expert judgement to a greater extent.
The application of the infrastructure supporting factor is already subject to numerous and strict criteria. We do not see any need to strengthen the environmental criterion for the infrastructure-supporting factor, in fact this is duly considered.
The criterion already in its current form requires the obligor to have assessed whether the project in question contributes to environmental objectives. If the obligor is subject to NFRD/CSRD requirements, information should be available.
The list of criteria for the application of the infrastructure-supporting factor is already extensive and it rather represents an operational challenge for the banks to fulfil it.
The infrastructure supporting factor is an important feature of EU regulation and should not be tightened further if it has to perform its role of promoting infrastructure projects to the maximum extent possible. In addition, the EBA approach for risk-based treatment is already adequate. We would also recommend that the infrastructure supporting factor is not linked to the taxonomy (as currently under discussion in the EU banking package).
As noted by the EBA in the DP, there are numerous “methodological” challenges to address in this area: such as the low availability of data, albeit this is gradually improving (among other things due to the disclosure requirements), the nature of ESG risks (including requirements to take a long-term, forward-looking perspective) and the uncertainty regarding the behaviour of companies and other market participants. Therefore, it is difficult to assess whether environmental risks could be better integrated into Pillar 1 through an adjustment factor than via already existing Pillar 1 instruments. This would strongly depend on the design and applicability of the solution envisaged.
Any change or adjustment should be risk based and should be basically guided by an openness to methods, it is important that banks, regulators and supervisors work together to develop the methodologies that can include the forward-looking aspect of climate risk and long-term horizons in the risk management framework.
Non-risk-based Pillar 1 add-ons to increase capital requirements, such as a Brown Penalizing Factor would be purely political measures, whose goals should rather be pursued via other public policies or taxation and not through banking regulation. In addition, a non-risk-based approach would be counter-productive in terms of risk management and transition financing.
If the Pillar 2 approach is to be complemented also with a view to foster the transition in a risk-based manner, consideration could be given in the first place to assess whether a Sustainability adjustment factor (SAF) could be envisaged for 1) energy efficient/green mortgages and 2) other suitable exposures to be identified, on the basis of clusters of taxonomy aligned exposures via sample based forward-looking methodologies and that are characterized by a reduce prospective financial risk. In this respect, a more precise mandate would be needed for EBA to study other asset classes in view of an extension of the adjustment, based on evidence of reduced riskiness of exposures by virtue of their environmental sustainability.
Such a factor clearly would not affect risk management activities: its application would not exempt banks from a prior creditworthiness analysis, applying only after having calculated their capital requirements as usual (similarly to the SME support factor).
We reject the idea to impose surcharges to risk weights based on future-oriented scenarios for various reasons. Such an approach breaks with the Pillar 1 approach and would be based on very uncertain and subjective assessments of scenarios, which can have very different effects on individual institutions depending on the market risk positions concerned. An addition of scenario considerations seems to us to be a suitable instrument for risk management, but this should be purely subject to Pillar II. The more detailed derivation of the risk weight allocation based on the addition of certain relevant factors for the environmental risk (e.g. dependence of part of the risk factor on e.g. economic activity and sector) would have to be clearly derivable empirically, but would then also be a strong generalisation due to the lack of reflection of the concrete market price risk positions of an institution.
The introduction of a new risk factor specifically for environmental risks cannot be decoupled from empirical proof of the effects of environmental risks on relevant instruments of the trading or banking book. In general, intensive impact studies would be indispensable to verify the effects on capital adequacy and the possible adequate calibration of the capital approaches.
Given the methodological freedom in the application of internal models, the changes under the "Fundamental Review of the Trading Book" were introduced to have stronger limitation of discretionary scope.
We are therefore critical of the inclusion of environmental risks within the models by adjusting the observed historical data for possible future dynamics due to the increasing subjectivity and the possible double counting in light of the difficulty of extracting particular effects from environmental risks from the volatilities observed anyway. This also applies to a possible mapping outside the models through adjustments in the consideration of event risks. Due to the early stage of empirical observations and model developments, it would be advisable to first gain more experience with mapping in Pillar 2 before adjustments are made to the Pillar 1 methodology of capital approaches. On this basis, a "range of practices" could be created to enrich the discussions before moving any further in adjusting the Pillar 1 framework.
Credit institutions with significant market risk positions should be given the opportunity under Pillar 2 to gradually develop their models further when a larger empirical database is available, before a decision is made on an adjustment of Pillar 1.
We are of the view that climate-related risks are not a new category of risk but drivers of existing risks. It would be very helpful, if ESG flagging of internal and external loss data would become a standard procedure, so Consortium and other databases would be comparable.
Likewise, EBA should provide data for distribution and expectation values of environmental risk categories as triggers for losses per Basel event type based on EBA available data.
As to the practical challenges, one of the main issues is the lack of an official categorisation of environmental risk drivers (as well for social and governance risks) predefined at supervisory level similar to the Basel event type categorisation (including as well an example level 3). We would also note still a deficit of regulatory incentive for good ESG risk management and this particularly under the standardized approach.
As OpRisk management has included forward looking components for a long time (e.g., risk self-assessment scenarios) the question would rather be to define a more precise period for the forward looking perspective (3 years, 5 years, 10 years, even longer?), with the questions in terms of reliability that go with it.
Depending on this requirement, regulatory recommended data sources for respective ESG “forecast” data (expected loss development from ESG risk) would be appreciated.
When it comes to capital requirements, it is worth noting that a forward-looking perspective has not been integrated into the general operational risk framework. Therefore, it should not be done with environmental risk, that is very new and for which banks do not currently dispose of sufficient information.
We fully agree, the impact of environmental risk factors on strategic and reputational risk should remain under the scope of the Pillar 2 framework. We support the argumentation given under para. 194.
From our perspective, from the point of view of risk and ESG management it could be limiting in this regard to set ILM to 1.
The question of the ILM, along with other elements of the operational risk framework, is currently being addressed by the co-legislators in the review of the CRR that will implement the Basel reforms in the EU.
In this regard, we believe that it should be explicitly indicated that, within the same jurisdiction, it should be possible for banks that are able and willing to calculate the historical losses to do so. For banks that do not choose to calculate the ILM a presumption of ILM=1 should apply.
In order to avoid any cherry picking, institutions should have to stick to the chosen approach for at least three years before they might switch to the other method. Institutions would have to report this decision to the competent authority.
Additionally, it would be very important to highlight the necessity of an overarching approach for ESG risk management taking into account all relevant second lines (compliance, risk, legal, business continuity and so on).
The EACB welcomes the opportunity to comment on the EBA Discussion Paper on the role of environmental risks in the prudential framework.
We appreciate the stance of the EBA in clarifying that the approach followed will be first and foremost risk based. Indeed, we believe that it is key to clarify that the aim is not to embed in the framework prudential incentives/disincentives to redirect capital as this might in an early phase lead to the buildup of risk in counterparties that are still economically uncompetitive and lack credible long-term strategies and/or hamper the financing of transition activities that would help other sectors to become more sustainable.
We would also highlight that climate and environmental risks are, as of today, beginning to be covered by Pillar 2 (SREP) and Pillar 3 (disclosure) requirements. Supervisory expectations and guidance documents were published giving institutions direction on how to better arrange processes and strategies in terms of climate risks (e.g. ECB guide on climate-related and environmental risks for banks, EBA Report on management and supervision of ESG risks etc). The exposure towards environmental and climate risk is sufficiently addressed at institution specific level, and as it is mentioned in the discussion paper, the Pillar 1 framework already includes mechanisms that allow the inclusion of new types of risk drivers such as environmental risks (e.g. via credit ratings or the valuation of collateral).
As the discussion paper focuses on the Pillar 1 framework, it must be born in mind that environmental risks could also be tackled via the buffer framework. While a review of the macroprudential framework is in the pipeline, and some elements are already being addressed as part of the CRD 6/CRR 3 legislative process, it is worth noting that macroprudential measures such as the sectoral systemic risk buffer must not lead to any double counting of risks, especially when they are already covered by P2R/P2G (as stated also in the EBA Call for Advice for the review of the EU macroprudential framework).
Finally, and from a general standpoint, it is important to ensure that only new exposures originating after a measure has come into effect – or a change of legal basis – are targeted by certain measures. This would prevent unwarranted increases in capital demands.