Climate capital in the balance as EBA rejects green risk weights

European regulator suggests climate change must be factored into existing risk categories

  • The European Banking Authority has recommended avoiding both green-supporting and penalising factors in Pillar 1 capital requirements.
  • Instead, the EBA proposes incorporating climate into existing risk categories.
  • But sources say climate risk-weighting may yet have a second future in the macroprudential framework within the systemic risk buffer.
  • Concerns remain, however, over the suitability of existing capital models to capture climate risks – in particular, probability of default and loss given default – given the mismatch in time horizons and lack of past data.
  • The EBA’s protracted timeline also divides opinion; some sources warn that delayed action will cause worse physical risks to be baked in.

You might say bankers’ concerns about climate risk-weighting are evergreen. Banks have long been sceptical about bespoke risk weights that benefit green-friendly deals and penalise those that aren’t. And now, the European Banking Authority has arrived at the same conclusion.

But that doesn’t mean banks are happy with the regulator’s proposed alternatives. Far from it.

“It’s all pushed back to the banks,” says Michel van den Berg, a sustainability adviser who has worked with the major Dutch banks. “With the EBA saying: ‘We want you to start building up historical data, but we’re not really giving you the toolbox to apply climate risk in a conservative fashion’.”

The EBA originally consulted in May 2022 on how to factor climate risk into the prudential framework, as mandated under the capital requirements regulation (CRR). In October 2023, it published a final report incorporating responses to its consultation. In it, the EBA rejected the idea of green risk weights in the Pillar 1 capital regime, arguing it would cause “design and calibration challenges” inconsistent with the existing framework.

The EBA will say: ‘We want you to start building up historical data, but we’re not really giving you the toolbox to apply climate risk in a conservative fashion’
Michel van den Berg, sustainability adviser

Six sources who spoke to for this article agree with the EBA’s decision to reject the green-supporting factor. They argue the Pillar 1 capital framework is designed to absorb shocks rather than to direct capital allocation. They also note that a green asset is no less susceptible to physical harm than a so-called brown asset. 

It is the location of the asset that determines whether that hazard is going to be subject to either acute or chronic physical risks,” says van den Berg.

Instead, the EBA has recommended targeted enhancements to the Pillar 1 framework that are intended to capture environmental and social risks while preserving the risk-based approach. 

This recommendation requires banks to begin including environmental risks as part of the internal ratings-based (IRB) approach to credit risk, over the next three years. There are also proposals to reflect climate change in the standardised approach to credit risk (see box: Relying on ratings). In the “medium-to-longer term”, banks may be expected to develop forward-looking prudential tools such as scenario-analysis and transition plans, as well as incorporating concentration risk metrics introduced under Pillar 1.

Most large banks have not yet started this process, according to the head of climate risk at one global bank – given the more immediate challenges around sourcing data, advancing methodologies, developing processes and allocating budget. Most “have not even begun to set up the infrastructure”, says the climate risk head. 

Meanwhile, those who want quicker action on climate change are concerned that the EBA’s approach borders on perfectionism, potentially delaying the whole process. The data collection needed to fulfil regulatory expectations will take time, and the EBA has earmarked 60% of the proposed changes for the medium-to-long term, which sources anticipate could be as many as five or 10 years away.

“We ought to calibrate the probability of default before climate risk happens,” says Pierre Monnin, a former financial stability expert at the Swiss National Bank. “Because otherwise, we’ll be 10 years too late.”

Apocalypse ... when? 

One of the overarching challenges for banks and regulators is that climate risk is of a different nature to typical financial risks. It is nonlinear, idiosyncratic and has unknown second order effects – features that are not easily captured in existing approaches for capturing credit risk.

The EBA’s move indicates that it wants credit assessments to remain fundamentally unchanged, but that firms should consider additional risk transmission mechanisms resulting from physical and transition risks. “They are clear in saying it would not be a sound credit analysis if these were not factored in,” says Judson Berkey, advocacy lead for the chief sustainability office at UBS. 

However, bankers have long warned that the timeframe for climate-related risks to materialise – particularly chronic physical risks – does not match the one-year horizon of a probability of default (PD) estimate. The climate scenarios banks are running are also longer-dated. 

We ought to calibrate the probability of default before climate risk happens. Because otherwise, we’ll be 10 years too late
Pierre Monnin, former financial stability expert at Swiss National Bank

The head of climate risk says climate scenario analysis, despite continuing efforts of the Network for Greening the Financial System to refine its models, is fraught with difficulties, particularly due to the time horizon mismatch between PD calculations and the scenarios themselves.

“You’ve got short-dated scenarios over the zero-to-three-year time horizon, 10-year for mid-term, and a 30-year time horizon for the longer-dated. Somehow you need to translate that into what's not an overly conservative perspective of probability of default and loss given default over the next year,” says the climate risk head. “So that is quite challenging.”

In effect, the EBA is expecting banks to adjust their portfolios ahead of time before those risks start materialising on their balance sheets. 

“They expect the banks to be prudent about which assets to finance or not to finance, including based on their assumption of how material climate risks are going to be in the next investment cycle,” says van den Berg.

Typically, in PD, banks look at several financial factors – such as a company’s net earnings over the last three years – to determine the probability of default. Now, banks will need to decide whether they should change their models to specifically incorporate physical risk.

Before those risks begin to materialise and affect the financials of companies, it is very difficult to pinpoint causality,” says van den Berg. “In academic research there is a lot of suggestion of correlation, but correlation is not equivalent to causality.”

Judson Berkey
Judson Berkey, UBS

The EBA appears to be signposting that, in the medium term, there may be sufficient data collected to enable backtesting that would ensure the impact on capital through changes to PD and loss given default (LGD) is appropriate, says the climate risk head.

Until then, however, banks that use IRB have been told to consign the consideration of environmental risks, such as climate risks, on defaults and loss rates to a medium-to-long-term action. Only when these risks become evident should banks reflect them in PD and LGD estimates “via a redevelopment or recalibration of the rating system”. 

 “You can’t just sort of start playing around with your algorithms to change the way your PD is calculated,” says van den Berg.

Back(test) to the future

In the meantime, IRB model validation for climate risk is difficult, as banks need to identify the specific causality behind a given loss.

“There’s a whole challenge in identifying the specific losses due to a climate event and decoupling that from broader credit losses,” says the global bank climate risk head. “There will be a lot of challenges in having the data and infrastructure to be able to disaggregate climate impacts and to be able to separate them out so that you can do backtesting specifically for climate impact.” 

And yet, the EBA is discouraging IRB banks from liberally using expert judgement that deviates from the historical data, warning that “a large number of overrides of the results of the model might be an indication of a model weakness”. Instead, it says that the adjustment of estimates during the risk quantification stage should be “based on a sufficient number of observed and reliable data”.

Consequently, the global bank's model validation team has shifted focus from judgement overlays to a statistical approach

“Because we can’t use backtesting, the technique we used involved our analysts doing a calibration exercise on about 100 counterparties, so it’s a huge time and effort commitment,” says the climate risk head at the global bank. “But that’s the only way we could get model validation teams comfortable with it – we’ve been running that on a quarterly basis.”

Some evidence-based research is beginning to emerge. UBS’s Berkey points to recent research from the UK into a price premium for energy-efficient properties in the country that can be used to differentiate credit. Banks would be able to determine whether a new mortgage is exposed to a higher loss rate in the future and could be worth less; and whether it faces a higher chance of default because the property can’t be resold as easily.

Alexandre Petrov, executive adviser for risk models at Nordea Bank, says many IRB banks in Europe are members of the bank-led Global Credit Data consortium, which pools data for PD models and is currently developing a common data warehouse of environmental information.

Petrov suggests some banks might want to develop, alongside the official capital modelling, a function similar to a climate-adjusted PD or LGD, which they can compare with their official capital modelling. Banks could run their own scenario analysis to determine the impact of climate risk drivers on parts of their portfolios, and as that gap starts to become material, he points out, “it should give you a sense where you should prioritise your management actions beyond the average duration of your portfolio”.

Thierry Philipponnat
Thierry Philipponnat, Finance Watch

Given the absence of adequate historical evidence to do accurate modelling and calibration of climate-related risks for future PD, there is a risk that climate-related risks are currently being under-estimated. Petrov and Monnin agree that firms ought to apply a margin of conservatism. Indeed, the Basel Committee on Banking Supervision recommends this.

Monnin notes, however, that the Basel Committee is not clear on how this should be done. In a November 2023 newsletter on climate-related financial risks, the BCBS indicated banks could use proxies and assumptions where hard data is not available, but they would need to establish “necessary guardrails” to account for the resulting model risk.

Petrov says firms could deploy simulation-based models using various scenarios. He expects banks to start using stochastic methods, such as a Monte Carlo approach, within the next year or two. This would involve modelling a spread of values representing the bank’s confidence in certain losses – for example, a most probable outcome and a conservative one. He suggests the initial data for the simulations might come from counterparties’ transition plans. This could then be used to increase capital requirements for a specific segment.

Beyond Pillar I

Firms could undertake such simulations as part of their own economic capital calculations, where banks have more freedom to apply their own models, not predefined by regulators, as under Pillar 1.

Equally, supervisors could apply Pillar 2 capital add-ons to offset any climate risk they believe is being undercapitalised. The European Central Bank has already indicated it is heading down that path. But Thierry Philipponnat, chief economist at the public interest group Finance Watch, says Pillar 2 is limited in what it can achieve “in the absence of ambitious Pillar 1”. 

If a risk-weighting for a certain oil company is 40%, for example, then a supervisor who thinks the risk is underestimated may propose increasing it 50%. But that would still be inconsistent with the risk weighting of 150% that Basel recommends for assets such as real estate and private equity.

Further, since Pillar 2 is reliant on the judgement of the bank and then its supervisor, it is likely to lead to inconsistent outcomes, particularly when there is so much uncertainty about the underlying risk.

Adding buffers

Beyond the microprudential framework, three sources see potential in the EBA’s suggestion to mitigate the build-up of climate-related risk through the systemic risk buffers, which unlike Pillar 1 can be more forward-looking.

The EBA suggests three options: adaptations to the general systemic risk buffer; a sectoral systemic risk buffer; and a concentration measure.

Monnin favours a general approach, warning that a sectoral systemic risk buffer could disadvantage firms within a sector that are more advanced in their transition efforts than their peers. 

It should be at the firm level, and you should have an indicator that summarises how well a firm is engaged in the transition or not,” says Monnin. “That's where transition plans play an important role, and we might think of defining macroprudential capital buffers as linked to this transition plan rather than linked to sectors.”

The non-profit public interest group Finance Watch meanwhile suggests a version of a concentration measure that applies the loan-to-value concept from risk management in real estate to the financing of fossil fuel assets. While the typical mortgage loan-to-value is 80%, Finance Watch research suggests that the figure for financing fossil fuel assets, considering the stranded asset logic, is above 400%. 

“The banks are lending to fossil fuel companies more than four times what will be exploitable,” says its chief economist Thierry Philipponnat. “Because we know that the majority of the remaining proven reserves of oil and gas – approximately 75% – will have to be left in the ground.”

UBS’s Berkey agrees that macroprudential instruments provide jurisdictions with more flexibility, and that they can be used in a more ad hoc or targeted way, but he urges that such policies be “based on data, facts and scientific findings and that these tools are used in a proportionate way.”

Relying on ratings

The majority of EU banks do not use an internal ratings-based approach (IRB) but are instead on the standardised approach for credit risk. This can be calculated in one of two ways: using a table of regulator-set risk weights for different types of exposures; or, using external credit ratings.

Sustainability adviser Michel van den Berg is concerned by the apparent lack of responsibility that characterised feedback on environmental risks from banks using a standardised approach, which were published in the EBA report. For him, it creates an impression that they don’t see the point of understanding climate risk themselves, and instead rely on “the outside world” to make that determination for them. 

“They are not as risk-aware, I think, as the IRB banks, so they run the risk of falling behind in terms of developing their understanding and knowledge of how climate risk will affect their portfolios; there are a lot of standardised banks,” says van den Berg.

Some public interest groups still take issue with the regulator-set standardised risk weights allocated to fossil fuels, given their risk of being stranded. Echoing calls from the European Parliament to increase fossil fuel risk weighting to 150%, public interest group Finance Watch says fossil fuel exposures currently receive something between 20% and 50% capital requirements.

“More to the point, prudential regulation should be in coherence with itself and if assets with similar levels of risk do not receive the same risk weights, then we have a distortion in the system’s ability to allocate capital where it needs to be allocated,” says its chief economist Thierry Philipponnat, comparing the fossil fuel risk weights with the 150% applied to other assets such as real estate.

With regard to external credit ratings, the EBA report recommends that rating agencies should be obliged to consider environmental and social risks as part of their methodology within the next three years. Some sources believe this could present an even greater challenge, given the difference in rating agencies progress so far.

Former Swiss National Bank official Pierre Monnin says, “If the credit agency doesn't do their work properly then their credit ratings are also biased, and [they] are at very different stage in terms of integrating climate risk. There is no consensus on how to do it, though some have more clarity about how they do it than others.”


Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact or view our subscription options here:

You are currently unable to copy this content. Please contact to find out more.

You need to sign in to use this feature. If you don’t have a account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here