Pillar politics: which part of the Basel framework is best for tackling climate risks?

February 25, 2022

Two recent papers offer contrasting views on Pillar 1 and Pillar 2 as tools for addressing banks’ exposures to climate risk

Three ‘Pillars’ undergird the global banking rules known as the ‘Basel framework’: Pillar 1 covers rules on minimum loss-absorbing capital requirements for all lenders; Pillar 2, supervisory review measures, which include firm-specific capital add-ons above and beyond those set under Pillar 1; and Pillar 3, rules on public risk disclosures. Together, these three pillars support a safe and sound banking system — at least, that’s the theory.

Right now, banking experts and climate activists are debating how these three pillars should be enhanced to properly guard the banking system against climate-related financial risks. Two recent papers — from two very different sources — highlight a particular divide: over whether Pillar 1 or Pillar 2 should be used to ensure banks are adequately capitalized against these risks.

The first, released last Thursday, is the product of experts at the Financial Stability Institute (FSI), a think tank created by the Bank for International Settlements and Basel Committee on Banking Supervision (BCBS) — the group behind the Basel framework. This argues that Pillar 2 is “the natural candidate” for making sure banks handle their climate risks and maintain enough capital to defend against them. 

The second is a response to the BCBS’ own proposed principles on climate risk management put together by the Climate Safe Lending Network (CSLN), an interest group made up of financial institutions, NGOs, and policy experts. This says it is too much to assume Pillar 2 approaches “will be sufficient on their own” and that authorities “cannot afford to overlook” Pillar 1 when it comes to climate risk.

What to make of these contrasting perspectives? It helps to understand their different interpretation of the purpose and capacity of each Pillar, and of capital requirements generally. 

The FSI paper expresses what could be termed the ‘orthodox view’ — that Pillar 1 requirements should be calibrated to reflect each bank’s actual risk of incurring losses over a set time horizon, specifically a one-year time horizon. This microprudential view of capital requirements also holds that risk-based calibration should be rooted in historical loss experience, rather than forecasts.

This interpretation has plenty of holes in it. First, it ties together the purpose of Pillar 1 (capital requirements based on risk assessments) with the current practice of Pillar 1 (setting requirements based on set time horizons, historical loss experience). When it comes to climate risk, it’s true that its impacts can’t be extrapolated using historical loss experience — because most of the financial effects haven’t materialized yet, and frankly can’t be modeled precisely, either. But if the purpose of Pillar 1 is to capitalize banks against risks — and climate change is recognized as a risk — then the practice of setting the requirements has to evolve to accommodate this. 

Second, the history of the implementation of Pillar 1 suggests that what makes a requirement ‘risk-based’ is not some objective formula, but the subjective interpretations of regulators and firms. Here, it’s worth breaking down how risk-based Pillar 1 requirements are computed today.

The Basel framework allows banks to calculate the Pillar 1 requirements for their credit portfolios (where the bulk of their risks lie) using a standardized approach (SA) and/or an internal ratings-based approach (IRB). Banks using the former multiply their exposure amounts by BCBS-set risk weights to generate their risk-weighted assets (RWAs). Capital requirements are then set as a fraction of these RWAs. 

These risk-weights are defined by the BCBS — not a complex mathematical formula tailored to each bank or exposure. Yes, the risk-weights are informed by external credit ratings, but these are the inventions of fallible entities, not hard science, and there are plenty who view credit rating agencies as having failed to factor climate risk into their evaluations of companies to date.

Furthermore, at the jurisdictional level, these standardized risk-weights have been altered in the past to suit political ends. Most significantly, the European Union lowered the risk-weights for bank exposures to small and medium-sized enterprises in an effort to encourage lenders to extend more credit to them.

The IRB, on the other hand, makes use of banks’ own credit rating models to determine appropriate risk-weights. The resulting figures are supposed to reflect a borrower’s actual probability of default and a bank’s loss given default. However, as these values are determined using banks’ own data and models, there’s room for discretion — a little too much, in the eyes of some regulators. The European Central Bank, for instance, conducted a Targeted Review of Internal Models (TRIM) between 2016 and 2021 which found over 5,800 weaknesses in banks’ models. This in turn led to “unwarranted” variability in RWAs across lenders. Put simply, banks were lowballing their actual credit risks.

This is all to say that while Pillar 1 requirements may be risk-based, what counts as risk-based is not set in stone, and banks and regulators have adjusted the risk-weights of exposures in the past to account for new data or in the pursuit of political objectives.

Furthermore, to claim as the FSI paper does that using Pillar 1 to captialize banks against climate risks would be “suboptimal” because of their “longer time horizons and the high degree of uncertainty as to how and when … [these] will materialise” is a technocratic argument that sidesteps the overriding aim of Pillar 1 in the first place. 

After all, it’s clear that climate change will have unprecedented implications for banks and the broader economy. How devastating these will be may be unknowable in the here and now. But faced with this uncertainty, the prudent course of action is to take what certain climate activists — and groups like CSLN — call a ‘precautionary approach.’ This means setting capital requirements with the aim of preventing activities that would push the world to exceed the crucial 1.5°C warming threshold, and not getting caught up with overly complex forecasting and nuanced debates on just how different climate impacts will hit specific exposures.

Unsurprisingly, the CSLN view of Pillar 1 is more expansive than that held by the FSI authors. The group’s letter says that Pillar 1 measures “based upon financial stability risk” should be the cornerstone of regulators’ responses to banks’ climate risk exposures. This reference to financial stability fits with a view that Pillar 1 measures serve a macroprudential function in protecting the banking system as a whole from financial risks. This view is supported by the fact that Pillar 1 requirements are made up of more than the exposure-specific calculations described above. They also include broad add-ons — namely, the capital conservation buffer and countercyclical buffer. These are fixed as specific ratios of banks’ RWAs set at the BCBS- and individual regulator-level to serve as additional loss-absorbing cushions.

But the letter also argues that Pillar 1 is the proper place for “asset-specific micro-prudential capital for fossil fuel assets” to be applied, too. Indeed, the CSLN proposes what it calls a ‘one-for-one’ rule — one popularized by the European NGO Finance Watch — under which any bank exposure that represents an expansion or exploration of oil, gas, and/or coal is tagged with a risk-weight of 1,250%. This effectively means that a bank would have to hold one dollar in equity capital for each dollar of exposure to these fossil fuel expanders.

The idea is that this would make financing fossil fuel expansion so capital intensive that it stops being a profitable enterprise, while keeping the risk-weighting of all other lending activities the same. It’s a blunt approach, but one not completely at odds with how Pillar 1 has been wielded in the past. Certain securitization exposures are subject to a 1,250% risk-weight in order to deter banks from investing in structures they do not fully understand. In addition, as CSLN itself argues, the BCBS recently proposed slapping cryptocurrency exposures with the 1,250% risk-weight because of their “unique risks.”

Still, the one-for-one rule has its limits — not least because it would only apply to a small slice of carbon-intensive assets. This seems out-of-sync with climate activists’ broader goal of having climate risks adequately capitalized across banks’ portfolios. After all, borrowers in emissions-intensive manufacturing sectors may also be misaligned with climate goals, and subject to transition risks. What would a one-for-one rule do to shield banks from these kinds of exposures? 

In addition, the rule would only impact banks using the SA, as standardized risk-weights don’t hold sway under the IRB approach (except through an aspect of the Basel framework known as the ‘output floor’, which places a limit on how much a bank’s modeled capital requirements can fall below those that would be produced under the SA).

A wider-ranging climate reform of Pillar 1 would adjust the risk-weights of all exposures by the degree to which they contribute to climate change. This could be calibrated using a borrower’s carbon footprint. The higher that borrower’s emissions, the higher the multiplication factor applied to their baseline risk-weighting. 

Of course, this approach would require banks to have sufficient actual or proxy data on each and every client’s carbon profile — a tall order, for now at least. However, it would perhaps be more in keeping with the view that Pillar 1 requirements should capture risks across portfolios, rather than incorporating them for some exposures and not for others.

The FSI and CSLN see Pillar 1 requirements very differently. What about Pillar 2? The FSI authors claim the “flexibility” of Pillar 2 “offers more scope” for tackling climate risk. This is because Pillar 2 gives supervisors an array of tools to address risks missed under Pillar 1 — including capital add-ons — that can dialed up and down based on a bank’s behavior. The CSLN, meanwhile, argues that while these tools exist, they are not being used. One reason may be that regulators lack the confidence and/or competence to deploy them in response to climate risk. This, in CSLN’s view, makes the case for the more “formulaic” approach offered through Pillar 1.

What’s somewhat overlooked in both papers is how Pillar 2 measures are generally used to address banks’ risk management deficiencies, rather than unpriced risks that lurk in banks portfolios — like climate risks. Put another way, supervisors don’t see Pillar 2’s purpose being to address risks not covered in Pillar 1, but as remedying bank-specific issues managing those risks already identified under Pillar 1. Pillar 2 capital add-ons, therefore, are unlikely to be applied at a size and scale necessary to capture the true extent of the physical and transition risks posed by banks’ clients, or to fulfill the ‘precautionary approach’ envisioned by CSLN. 

This makes Pillar 2 of limited use when it comes to the big picture of facing down climate risks to the banking system. One reason why some may see it differently, though, is because of the discretion it offers individual supervisors. Applying Pillar 2 measures in an aggressive manner to tackle climate risk management deficiencies is in the gift of each jurisdictional authority. It does not require a contentious rewriting of the Basel framework, an exercise that typically takes years.

This convenience, though, cannot contend with the much broader scope Pillar 1 affords supervisors to combat climate risks. Nor can bank-by-bank Pillar 2 add-ons, tailored to each firm’s risk management profile, do the job of ensuring the banking sector is protected against systemic climate risk.

Only Pillar 1 requirements, applied universally, have a chance to achieve this.