Is Climate Really a Financial Stability Risk or Solvency Risk for Large Banks?

Introduction

Over the past several years there has been an increased focus from regulators, policymakers and academics regarding the role that climate change may play in generating financial risk within the financial system.  As part of the solution to limit global warming, banks have been called upon to build out their climate risk management capabilities and work to align greenhouse gas emissions in their lending and investment portfolios in line with the Paris Agreement.  In response to this, banks have worked diligently to meet regulators’ risk management demands and many banks and end users have made commitments to achieving “net zero” in line with the Paris Agreement by 2050. 

The risk of climate change to the planet and the changes that will ensue in the global economy either through a transition to an economy that is less reliant on fossil fuels or, if no transition takes place, the losses that will be born through increased physical climate events, may not, however, lead to financial stability risk at the macro-prudential level or safety and soundness risk at the micro-prudential level for banks.  Over the past several years, central banks and supervisors have been evaluating climate-related financial risks to banks focusing on physical and transition risks.  This work has largely been organized through the Network for Greening the Financial System (NGFS), which is made up of central banks and supervisors.  The work has been further formalized through the Financial Stability Board’s Climate Roadmap, which lays out the G20 commitments and deliverables in relation to climate-related financial disclosures, monitoring and measuring climate-related risks, the development of scenario analysis and supervisory tools and practices in relation to climate risk management for financial institutions.

These official sector efforts have resulted in a myriad of climate risk management proposals by supervisors, numerous climate scenario analysis exercises and a host of disclosure requirements across jurisdictions.  This of course would all be appropriate if climate-related financial risk is a financial stability risk or material bank safety and soundness risk; however, over the past year the results of more research and practical analysis undertaken through climate scenario analysis calls into question whether such risks are material and in fact suggests that the near-term risks are entirely manageable for large banks.[1]  This is not to say that climate events or changes in policy to mitigate climate change will not have an impact on banks’ balance sheets and will not need to be understood and managed, but rather that they may not rise to the level of systemic risks or material safety and soundness risks.  Notably, in the May 2022 Financial Stability Report published by the Federal Reserve climate fell off the list of most cited risks for the next 12-18 months, whereas climate ranked 6th in near-term risks in the 2021 Financial Stability Report.

Despite the emerging evidence that suggests banks are well-placed to manage climate-related financial risks whether they are generated by physical risk events or through transition-related events, regulators continue to push a plethora of very detailed and intensive requirements on banks—and large banks in particular.  For example, this past year we have seen two consultations from the OCC and FDIC on climate risk management; the finalization of the Basel Committee’s climate risk management principles; a consultation from the FSB on climate risk management; and European regulators continue to push for more granular assessments and integration of climate-related risks into the individual capital adequacy assessment process (ICAAP).  At the same time that risk management principles are being developed by regulators and international bodies, scenario analysis exercises are proliferating.  Currently, climate scenario analysis exercises are being run or have been completed in over 25 jurisdictions.[2]  As outlined in the recent research analysis and scenario analysis outputs, which are discussed in further detail below, this level of regulatory focus and requirements would seem to be disproportionate to the risk that climate may pose to large banks.

Recent Findings in Relation to Physical Risks

A staff study by the New York Fed released in January 2022 titled, “How Bad Are Weather Disasters for Banks?”, evaluated how banks fared in response to FEMA-level climate disasters in 1995-2018.[3]  Their answer was “not very.”  The authors found that FEMA disasters over the last century had “insignificant or small effects on bank performance and stability.”  In particular, they found that banks’ stability seemed to be endogenous rather than a mere reflection of federal aid—namely disasters increased loan demand, which then offset losses and boosted profits at larger banks.

Similarly, an FDIC staff study released in June 2022 on “Severe Weather Events and Local Economic Conditions,” available here, found no evidence of bank distress resulting from severe weather events and little evidence of a deterioration in loan performance.  In particular, the study found that Hurricane Katrina in 2005 reduced loan performance at community banks in the affected area for a time, but no banks failed.  The study found no impact on banks from Harvey (2017), Irma (2017), the Midwest drought in 2012-2013, the Camp Wildfire (2018) or the Tubbs Wildfire (2017).

The conclusions of the FDIC and New York Fed reports are consistent with those found in many previous studies.  Noth and Schuewer (2018) find that the probability of default increases at U.S. banks following weather disasters, but the effects are short-lived and relatively small. Looking across countries, Klomp (2014) finds that only geological disasters, and not weather-related disasters, have a negative effect on bank stability in developed countries. Brie and colleagues (2019) find that following hurricanes banks face deposit withdrawals and experience a negative funding shock, but they find no signs of deterioration in loan performance or bank capital. Barth et al. (2019) find lending and profitability go up after natural disasters and that “…banks located in the disaster-prone areas contribute to helping communities recover from natural disasters.” (p. 1)   Similarly, Koetter et al. (2019) find German banks lent more in reaction to flooding along the Elbe River, but that the expansion in lending is not associated with higher insolvency risk or higher loan impairment.

These studies evaluate observable, past events and their impacts on banks.  Regulators are of course trying to get a better sense of the risk that future climate events might have on bank balance sheets, which is where scenario analysis has come in.  Climate scenario analysis typically looks at a variety of physical and transition risk scenarios over many years.  While climate scenario analysis is an evolving tool and the results are far from precise, it is somewhat reassuring that recent exercises have not resulted in huge near-term losses for banks.  For example, in the Bank of England’s recent Climate Biennial Exploratory Scenario (CBES) the No Additional Action Scenario (NAA or physical risk scenario), in which global temperatures are 3.3C higher relative to pre-industrial levels at the end of the 30-year period resulting in increased severe weather events, sea level rises, and drought; bank loan impairment rates increased only by half.  Furthermore, the NAA scenario results in “substantially lower losses” for banks over the scenario period as compared to the transition scenarios where the price of carbon rises significantly.

Similarly, the ECB also looked at physical risk through its 2022 Climate Risk Stress Test.  Under the “hot house world” scenario, the ECB frontloaded physical risks that were expected beyond the 2050 timeframe.  Given this frontloading, one could say this is the most severe physical risk scenario that has been run to date.  However, projected losses were not much more than the CBES NAA scenario.  While these losses are not insignificant, they must be taken in context—that is, they result from future, very severe physical risk climate scenarios that would occur over time.  Modeling them is informative, but not reflective of short-term risk or resilience of banks, particularly in relation to capital adequacy.

Recent Findings in Relation to Transition Risk

When it comes to assessing the impact of transition risk on banks, there is limited empirical analysis; however, there are practical case studies one can look to as well as the results of recent scenario analysis exercises, which again have shown relatively muted impacts on large banks.

In terms of empirical analysis evaluating the consequences of carbon taxes—a type of transition risk—Metcalf and Stock (2022) examine the economic impact of carbon taxes on the 31 European countries that are part of the EU emissions trading scheme, and the 15 of those countries that have implemented a carbon tax.  They found “…no evidence of adverse effects on GDP growth or total employment.” (p. 2).  Although they do not explicitly examine the consequences for banks and the level of carbon taxes imposed was limited (e.g. approximately $30/ton CO2), the lack of any measurable effect on the economy suggests banks’ conditions must have been essentially unaffected.

Given the limited empirical analysis, it is useful to examine case studies and practical exercises in the context of transition risk.

Case Studies

Mid 2010’s Drop in Oil Prices

Transition risks are losses resulting from changes in public policy relating to climate change, especially a carbon tax.  Because a substantial carbon tax could, at least in principle, be enacted quickly and surprisingly, it could generate substantial losses on the liabilities of carbon-intensive firms.  While there is a dearth of historical examples of the government implementing policies rapidly and surprisingly that lead to significant loan losses, we can look to some past examples as indicators of what could transpire.  For example, the rapid drop in oil prices over the second half of 2014 from over $100/barrel to less than $50/barrel could be informative.  It was only the fourth time in 30 years that the price of oil fell 50 percent in six months.  A consumed barrel of oil produces 0.43 metric tons of C02 (see EPA website).  A carbon tax of $116 would therefore be equivalent to $50 drop in oil prices (over six months).  The peak drop of $80 would be equivalent to a $186 carbon tax (over 18 months but not sustained).

During this period of oil price volatility, the official sector monitored the effect of the decline in oil prices on financial institutions and financial assets closely.  According to the FSOC 2015 Financial Stability report, the high-yield bond market came under significant pressure starting in June 2014, especially the energy sector.  In the first quarter of 2015, the market stabilized.  The report also noted that

“…regional banks with larger oil and gas portfolio concentrations, smaller banks in areas whose local economies are dependent upon the energy sector, and banks exposed to countries whose sovereign debt is supported by oil production may be at increased risk.” (p. 61)

The 2016 report stated that investors shifted away from riskier corporate debt in the second half of 2015, forcing banks to hold on their balance sheet or sell at a discount some leveraged loans that they had planned to sell, especially to the oil and gas sector.  In response, banks tightened underwriting standards in the fourth quarter of 2015.  The report stated that energy exposure of the largest firms appeared manageable but cautioned that some regional banks as well as credit unions in parts of the country reliant on the oil and gas sector could come under pressure.  By the 2017 report, stresses on banks appeared to have ebbed.

The FDIC’s Quarterly Banking Profile did not report much stress on banks.  The QBP in 2015 Q4 noted that “…lower oil prices adversely affected some energy sector borrowers.”  The 2016 Q1 profile attributed an increase in the noncurrent rate on C&I loans from 1.56 percent to 1.58 percent to troubles in the oil and gas industry but observed that it was still the second-lowest rate since 2007.  Subsequent QBPs did not mention the oil and gas industry.

The banking agencies’ annual Shared National Credit reports provided a more granular look at the effect of the oil price drop on loan quality.  The 2015 SNC review included an examination of 141 oil and gas obligors representing a third of SNC oil and gas borrowers.  The report observed that the drop in oil and gas prices had impaired many companies’ ability to pay interest and principal and had led to some defaults.  The report indicated that

Banks are showing flexibility in working with borrowers by relaxing leverage covenants and allowing customer’s time to curtail borrowing base over-advances. The banks and customers are taking reasonable actions during this stressed environment. (p.10)

The 2016 report again noted that banks were working with customers, but that the reduction in liquidity and high debt burdens of oil and gas firms had “resulted in a dramatic increase in borrower defaults and bankruptcy filings…”  The report noted, however, that bank commitments to the troubled borrowers were “…primarily in a senior secured position with the lowest risk of loss.” 

The 2017 report added that the risk was concentrated in loans to the non-investment-grade and unrated energy companies, and that the loans were primarily held by regulated entities, which is the converse of the usual situation where nonbanks usually held riskier credits.  The report further observed that the energy sector stabilizing and by the 2018 report, the level of syndicated loans with adverse supervisory ratings had declined because of improvements in the sector.

There were only 25 failed banks between 2015 and 2019.  Only two of the reports of the FDIC and Federal Reserve on the bank failures mentioned oil and gas lending – Allied Bank in Arkansas, September 2016, and First NBC Bank, Louisiana, in April 2017.  In both cases, broader risk management issues rather than oil and gas lending specifically caused the failure.

Impact of Technological Advancements

One of the ways that climate change and climate policies could put pressure on banks is through Schumpeterian creative destruction – as conditions change, some old industries fade while new industries rise.  In the climate case, rising carbon taxes could cause the fortunes of climate-intensive firms to decline.

Although climate change is an unprecedented risk, changing fortunes across industries is nothing new.  To draw lessons from such episodes, Nelson (2021) determined the disposition of U.S. firms that fell out of the top 50 by revenue between 1991 and 2021.  Many of the firms that exited were victims of technological change.  Kodak, Kmart, JC Penney, and Sears failed, but most firms simply got smaller or were acquired. As the firms that ultimately failed conditions worsened, banks reduced their exposures and took collateral; ultimately credit was provided by specialized lenders rather than banks.  Nelson could find no record of a bank making a loss.

The experiences of banks in reaction to industrial change are supportive of the conclusion of former Fed Vice Chair for Supervision Randal Quarles when he was asked what risks climate change and climate change policies present to the financial system:

Now, I think that a dispassionate analysis of that question is that climate change is not much of a risk to the financial system. And it’s hard to see how it can be… [as the risks] evolve over such a long period of time, and exposures in the financial system turn over so rapidly that the financial system will adjust.

So, I live in Utah. Skiing’s big industry here. It’s big recreation. I’m a big skier. Banks here that have a loan out to a ski resort, well, in 30 years, a ski resort may be much less profitable. One can take that into account, but nobody makes a loan to a ski resort for 30 years. You make a loan to a ski resort for three years or a year. And in that time, there’s going to be little difference between the profitability of the ski resort at the time you made the loan versus the time the loan is rolling over. And as it rolls over, you evaluate then. “Well, what’s the situation?” That’s a homely example, but multiply that by thousands through the financial system.[4]

Transition Scenario Analysis

Former Vice Chair Quarles’ statement above notes one of the challenges that regulators face when trying to evaluate transition risk—the short-dated nature of banks’ balance sheets.  The ECB and the Bank of England aimed to deal with this in two different ways, with the Bank of England holding the balance sheet static over the period of the scenario, while the ECB frontloaded the risks into a three-year traditional stress testing horizon.  Interestingly, neither of these approaches resulted in catastrophic results for large banks.

For example, in the Bank of England’s climate stress test under the most severe scenario (“Late Action” also called “Disorderly Transition”), annual loan losses doubles from a baseline rate of £3.7 billion per year to £7.5 billion per year.  By comparison, in the 2021 stress tests, banks reported a loan loss level of £18.3 billion per year, five times higher than the baseline in the climate projections.[5]

While the Late Action Scenario losses are significant, there is still plenty of headroom between the projected losses and the losses projected in the regular stress test, which would suggest large banks have financial resources more than sufficient to remain solvent through the contemplated transition scenario.

In the ECB climate stress test, cumulative losses across the three-year scenario including a disorderly transition combined with the two physical risk scenarios (a severe drought and flooding), credit and market losses totaled about €70 billion for the 41 banks undertaking the exercise.  In the EBA’s 2021 stress test, the three-year credit and market risk losses were €382 billion for 50 banks. 

Comparing the outputs of climate scenario analysis with traditional stress testing results is informative, because despite the fact that much is unknown in climate scenario analysis in terms of second-round effects, non-linearities and data gaps, the magnitude of losses is still substantially less than the outputs of normal stress testing against which banks are capitalized for resiliency.  These scenario exercises suggest that climate-related risks may not present safety and soundness risks for large banks and certainly are not indicative of a need for more capital to buffer against climate losses under the normal capital planning time horizon. 

Conclusion

Based on recent studies, practical examples and the results that are starting to come from climate scenario analysis, it is reasonable to conclude that climate-related financial risks do not present a near-term risk to financial stability or material safety and soundness risk for large banks.  This is not to say that banks should not build climate-related risk management capabilities into their risk frameworks or that further work should not be undertaken to better understand the impact of climate-related risks on particular sectors or geographies, but recent studies and regulatory tests suggest this is not a material risk for large financial institutions and certainly should not have direct implications for the bank capital framework.


[1] The majority of scenario analysis exercises and regulatory requirements to date have been focused on large banks and not smaller institutions.  Therefore, the conclusions drawn in this blog do not necessarily apply to smaller banks, where exposures may be more concentrated and portfolios may be less geographically diverse.

[2] NGFS 2021, “Scenarios in Action: A progress report on global supervisory and central bank climate scenario exercises” (October 2021).

[3] sr990.pdf (newyorkfed.org)

[4] https://www.mercatus.org/bridge/podcasts/07182022/randal-quarles-inflation-balance-sheet-reduction-financial-stability-and

[5] In the climate stress test projection, the losses from 2030 to 2040 are about £12.5 billion per year, come closer to the losses reported in the macro stress test.