Performing Climate Risks to Financial Stability

Are climate change models the source of our next financial crisis?

As the dust of the global financial crisis of 2007–08 settled, many of those looking for someone to blame for the economic disaster zeroed in on something called the Gaussian copula model, a family of mathematical functions used to estimate the probability distribution of investment losses. In a lament on Wired, for example, the journalist Felix Salmon wrote that the model had become “so deeply entrenched—and was making people so much money—that warnings about its limitations were largely ignored.”

Other observers, like the financial sociologists Donald MacKenzie and Taylor Spears disagreed, arguing in an article for Social Studies of Science that the popular focus on Gaussian copula was superficial. The problem was not the models per se. Rather, the issue was the ways in which disparate social and organizational communities in the derivatives market used them to game the regulatory system.

MacKenzie and Spears showed that financial analysts knew the models were inaccurate but felt compelled to use them because common ways of compensating for its inaccuracies actively “performed” the market; the model created, shaped, and organized the phenomena it intended to describe. The Gaussian copula model created consensus. As one of the traders interviewed by the two researchers put it, “once everyone was using it, you have to use it as well, because it then becomes a good guide to prices.”

But so, too, did it become a shared guide for risk managers, auditors, and accountants in ensuring compliance to regulatory and financial objectives. In turn, traders’ reports of their activities and investments were structured in ways that produced favorable model results while masking the full nature of the underlying risk.

In their article, MacKenzie and Spears concluded that the financial crisis “was caused not by ‘model dopes’, but by creative, resourceful, informed and reflexive actors quite consciously exploiting the role of models in governance.” Their ability to do so underscores the extent to which financial markets are actively designed. Reflecting on the earlier 1987 stock market crash, in which models were also implicated, MacKenzie mused that “the design of financial markets is always implicitly political: it influences who will perform which transactions with whom and with what effects.” In turn, he explained, “it prompts a question: what sort of a world do we want to see performed?”

This is a question Mark Carney, who was governor of the Bank of England and chair of the Financial Stability Board, could well have asked himself in 2015, when he charged former New York Mayor Michael Bloomberg with leading the Task Force for Climate-related Financial Disclosures (TCFD) to develop regulatory guidelines for treating climate change as a risk to financial stability such that they must be disclosed. At the time, a major politically oriented risk modeling project funded by Bloomberg and colleagues had just wrapped up. The project was reportedly intended to “mak[e] the climate threat feel real, immediate and potentially devastating to the business world.”

Risky ideas about climate

Falling in line with Carney and Bloomberg’s herding of financial entities towards the management of climate-related financial risk, last week the U.S. Federal Reserve released results of a pilot banking stress test exercise.

The test was designed to align with modeling methods established by the Network for Greening the Financial System, a coalition of central banks co-founded by the Bank of England with Carney as its lead. NGFS members “pledge their support” for Bloomberg’s TCFD recommendations. Meanwhile, Bloomberg’s philanthropy funds the development of NGFS model scenarios.

The Fed was tasked with conducting the test after U.S. President Joe Biden’s 2021 Executive Order on Climate-Related Financial Risk. Climate-related financial risk, according to the order, threatens “US companies and markets, the life savings and pensions of US workers and families, and the ability of US financial institutions to serve communities.” And by managing climate-related financial risk through disclosure and emissions reporting, the nation will “achieve our target of a net-zero emission economy.”

No actual energy policy needed, then. Just some financial reporting, perhaps, and money would rain down on the good and just, bringing forth spontaneous decarbonization of the world’s largest economy.

This multi-tiered conflation is the Carney-Bloomberg vision. Carney said as much in 2015 when he pitched the idea of TCFD to an audience at Lloyd’s of London, the historic insurance market. Carney wanted to make “climate policy a bit more like monetary policy” and, in so doing played to insurers’ interest in risk to foster urgency.

And so it is, that deep within the technical details of the Bloomberg funded NGFS scenarios are outdated and extreme assumptions about emissions driving inflated views of climate risk.

Conceptual sins

Climate change is real and important. Even so, making connections between climate change and weather extremes is deeply nuanced, and effects marginal.

Warming is certainly associated with extreme heat getting hotter and extreme cold getting milder, but according to the Intergovernmental Panel on Climate Change, many supposed changes in extremes, such as river floods, precipitation droughts, and tropical cyclones, are not yet even detectable, which certainly undermines the idea that changes in these weather phenomena are having a large impact on society.

The science of climate change and weather extremes does not support the common conception that upward trends in weather-related economic losses are a result of climate change. Rather, increasing economic losses are a product of social change, including changes in population, wealth, and inflation. Trends in insured losses reflect this basic fact. When disaster losses are normalized to account for socioeconomic changes, there is no upward trend in loss. Hands down, the most meaningful factor is growing exposure: more people and their stuff in harm’s way.

Even the authors of the Fifth National Climate Assessment, some of whom were engaged in Bloomberg’s climate modeling project, could not articulate a clear impact of climate change on losses. “Although climate change has played a significant role in [loss] trends,” the writers assert, “consensus is lacking on the extent to which increased losses are attributable to climate change versus other factors.”

This type of misdirection—we are sure there is a climate change signal here, but we just can’t find it—is remarkably common. For example, a recent press release from reinsurer Swiss Re quoted an executive in saying that “As weather hazards are intensifying due to climate change, risk assessment and insurance premiums need to keep up with the fast-evolving risk landscape.”

The advertised report gives special attention to increased losses from severe convective storms attributing the increase to macroeconomic factors. But of the 8% increase over time, a “high level” estimate of 1% was labeled as climate change. And a dig through the technical details suggests this figure was more of a guesstimate than an estimate.

Dollar dollar bill, y’all

If the connection between climate change, extreme weather, and losses is so indiscernible as to be reportable only in riddles, then why the sudden agreement that climate change threatens the stability of the financial system?

One reason may be the value of the data and modeling intensive climate change sciences to the makers of financial markets. Erica Thompson, a data scientist at the London School of Economics, reflects on the merging of climate science with financial business practices:

"future oriented climate modelling results are largely of use to the most quantitatively sophisticated stakeholder: insurance and finance companies, engineering consultants, multinational companies and so on…quantification of current and future climate has translated into the financialization of current and future climate."

No model dopes here.

In 2023 testimony during a Senate Budget Committee hearing on climate change and insurance, climate scientist Judith Curry, who works closely with the insurance industry, explained:

"Privately, insurance industry leaders state that climate change isn’t taken into account since 90% of property and casualty policies last 12 months. Any change in climate related impacts can’t be effectively measured on the time scale of a year. However, the industry is under pressure from investors that are reacting to an exaggerated public narrative surrounding climate change, as well as confusing weather with climate."

That investors play an outsized role in how the world understands climate change risk is nicely underscored in University of Illinois Professor Morton Lane’s scholarly history of the catastrophe bond market.

He recounts an episode in 2006 when a financial risk model vendor developed a new view of future hurricane activity to take into account climate change. Because science could be pieced together to support both the new view and the old, the company allowed users to choose. “The choice of two measures confronts the investor (and the issuer) with a valuation dilemma. Which estimate is most consistent with their belief?” In practice, he continued, “the marginal buyer determines price, and it is the cautious one—the one who expects the worst of two [loss] estimates—who will determine price.”

When this new climate changed hurricane risk became standard in certain sectors of the insurance industry it created havoc in the coastal property insurance market. Overnight, the industry needed to raise an additional $86 billion to remain solvent in the eyes of the rating agencies.

And so, a public narrative harping on risks to financial stability from climate change can stir demand for climate change risk estimates—a handy thing if you happen to be selling climate analytics or investments tied to weather risk or perhaps purchasing real estate recently devalued by risk adjustments. Or if you’re looking for news lines of business; the banks engaging in the Fed’s climate stress test hired modeling consultants for the exercise.

A market in his image

Last fall, Senators Sheldon Whitehouse (D-RI), who is chairman of the Senate Budget Committee, and Ron Wyden (D-OR), who is chairman of the Senate Finance Committee, announced an investigation into the ways in which insurance companies are managing “mounting risks from climate change.”

The senators foresaw climate change leading to a systemic financial crisis as rising insurance costs put heavy pressure on the mortgage market. “A widescale decline in coastal and wildland-urban interface (WUI) community property values would present a systemic risk to the U.S. economy,” they noted, “similar to what occurred in the 2007-2008 mortgage meltdown.”

The senators’ evidence for this looming catastrophe?

Insurers’ own climate change risk models.

Whitehouse and Wyden’s insurance investigations come after a series of hearings last summer that kicked off with a familiar character: Carney, along with Robert Litterman, a former asset manager and member of government advisory groups on climate related financial risk. Both argued that climate change is causing increasing frequency and intensity of weather extremes and losses creating risks to financial stability. Both argued for managing emission to control losses. For his part, Litterman, referenced his work as chairman for the development of a report of the Commodity Futures Trading Commission on managing climate risk. The report, of course, made ample use of Bloomberg funded modeling projects.

In a later interview about the investigation, Whitehouse zeroed in on climate risk and financial risk. “There’s a core underlying reason for the insurance problems that Florida’s experiencing right now and for the risks it faces,” Whitehouse said. “And that is the persistent failure to deal with the problem of climate change.”

This is wrong. The underlying problem is the failure of policymakers to inspire a critical debate about urban development and risk mitigation—and about the misguided investor risk perceptions that may be inflating insurance costs.

And there you have it: the performance of climate change risk to financial stability.