March 31, 2023

Bank runs and tipping points: Parallels between finance and climate

**UNLISTED**

P#0 •  ePUB    •    PDF    • 12 min  •   comments 

The previous post on this blog on March 1st analyzed the climate stress for US banks proposed by the Federal Reserve. Just over a week later, Silicon Valley Bank (SVB), a medium-sized bank in the US, experienced a financial stress test — a bank run fueled by social media.¹ Responding to rumors that the bank was in trouble, customers rushed to withdraw their money. The bank ran out of cash and was shut down. Other banks have also since failed since then.

The failure of SVB was driven, in part, by rising interest rates. Regulators failed to assess the relatively simple financial risk of rising interest rates. Can we trust them to assess the more complex climate risk of rising temperatures? Perhaps not. As argued in the previous blog post, we should keep climate risk assessment as simple as possible for the sake of transparency.

A bank run is an example of a social tipping point. There’s been a lot of talk about climate tipping points in recent years, with media headlines saying that crossing 1.5 degrees of warming could trigger global tipping points.² The recently released IPCC Synthesis Report was heralded with headlines such as: World is on brink of catastrophic warming. ³

The best time to discuss tipping points is perhaps when they are actually happening! We can use the SVB collapse as an excuse to analyze tipping points in both finance and climate. This post argues that global average thinking is a bad way to describe either financial or climate tipping points. There is no precise global threshold that is guaranteed to trigger tipping behavior. The IPCC models do not predict climate tipping points occurring anytime soon. Nevertheless, we cannot rule out them out completely.

Feedback vs. feedback

The tipping point is a powerful visual meme. Consider a coffee cup on a table. If you tilt a coffee cup slightly, it will return to the normal position. If you tilt it a bit too much, it will topple over. A tipping point is a phenomenon where a small push beyond a threshold gets rapidly amplified and leads to a big response. (Two previous posts in this blog also discuss tipping points and thresholds.⁴)

Tipping points are features of a nonlinear system. The balance between stabilizing (or negative) feedbacks and destabilizing (or “positive”) feedbacks maintains the system in equilibrium.⁵ These feedbacks can act on different timescales. In the case of bank transactions, customer withdrawals can destabilize the system on short-time scales but availability of liquid cash acts as a fast stabilizing feedback. Long-term assets held to maturity, such as 10-year treasuries, can act as a stabilizing feedback but only on 10-year timescales.

In the case of global warming, there are several destabilizing feedbacks operating at multiple timescales (Figure 1). These include the greenhouse effect of increasing water vapor in a warmer climate on short timescales or the greenhouse effect of methane released from thawing permafrost on longer timescales. But there are also many stabilizing feedbacks that counter them. Perhaps the most important stabilizing feedback, is the fast feedback known as the Planck feedback: A warmer atmosphere radiates more heat to space, which leads to cooling in a matter of months.

Figure 1 Global impacts of destabilizing climate processes plotted against their approximate timescales, with confidence estimates [Figure 16 of Wang et al., 2023]⁶

Let us return to SVB, which had been operating as a bank since 1983. This means that stabilizing feedbacks dominated over destabilizing feedbacks for near 40 years to maintain the equilibrium. On March 9th, the fast stabilizing available-cash feedback failed to counter the fast destabilizing effect of customer withdrawals amplified by social media pressures. The stabilizing feedback of generating cash through a gradual sale of long-term assets was too slow to counter the destabilization. In fact, the rushed sale of some long-term assets at a loss acted as a fast destabilizing feedback when the word of the sale got out.

To understand the bank run tipping point, we have to carefully consider all feedbacks, both stabilizing and destabilizing, and the associated timescales. To maintain stability, we need fast stabilizing feedbacks to counter fast destabilizing feedbacks and perturbations. Either fast or slow stabilizing feedbacks can counter the slow destabilizing feedbacks.

In the case of the climate system, we know that the stabilizing feedbacks must have dominated over the destabilizing feedbacks for many thousands of years. Otherwise, we would not have had the relatively stable climate for over that period. Global warming is disturbing this stable equilibrium, just as customer withdrawals disturbed the stable equilibrium of SVB. Destabilizing feedbacks such as the decreased surface reflectivity due to melting of continental ice sheets can amplify the disturbances. But these are slow processes and any changes will happen over long timescales. The fast Planck feedback, and slow feedbacks such as absorption of carbon by land and ocean sinks, can act to counter the destabilizing feedbacks.

Will the stabilizing feedbacks fail to counter the destabilizing feedback at some point, as it happened with SVB? To answer this question, we use complex models of the climate system. Our most comprehensive models, the global climate models used for IPCC assessments, predict steady warming for plausible emission scenarios (Figure 2). They do not predict abrupt changes in global surface temperatures. This means that stabilizing feedbacks are likely to hold the destabilizing feedbacks at bay for the rest this century. Figure 2 Global surface temperature change projected by IPCC models for different emission scenarios. (“Very likely” ranges are shown for scenarios SSP1-2.6 and SSP3-7.0) [Figure SPM8a of IPCC AR6 WG1 report]⁷

The simplistic allure of global-average thinking

The tipping point meme strongly influences how many people think about climate change. This includes both ordinary people and influential people. In April 19, 2022, US Treasury Secretary Janet Yellen made the following statement when addressing the Coalition of Finance Ministers for Climate Action:⁸

Existing fossil fuel infrastructure is working against us, driving us instead towards reaching climate tipping points this decade.

Presumably, Yellen was motivated by headlines saying that global average temperature will likely cross the 1.5°C threshold for warming this decade, along with other headlines saying that global average temperature crossing this threshold could trigger climate tipping points. Is this a valid interpretation of such headlines?

We cannot completely rule out the possibility that 1.5°C warming could trigger global climate tipping points, but as we noted earlier, it is unlikely. IPCC global climate models do not predict any abrupt global climate change occurring through the rest of this century.

Focusing on global average temperature exceeding a precise threshold is a poor way to frame climate risk. Global average temperature is a useful diagnostic measure of warming, but it is not a physically relevant quantity that directly determines climate impacts (except perhaps for sea level rise). This may help explain some of the confusion regarding climate tipping points, as illustrated by Yellen’s precise-sounding statement.

Climate impacts are determined by local and regional temperatures; not global average temperature.⁹ A banking analogy can help explain this better. Although the spate of customer withdrawals was the immediate cause of the SVB bank run, the ultimate cause was higher interest rates. Higher rates decreased the value of long-term assets held by SVB. The bank had to sell these assets at a loss to generate cash, thus compounding its financial woes. With this in mind, consider the following fictitious headline in, say, the Financial Times:

Global average central bank interest rate crossing 4.5% could trigger global financial collapse

Of course, one cannot rule out the possibility that the “global interest rate” exceeding this threshold could trigger a global financial collapse. However, there are two problems with how the headline frames the issue. The first problem is that the average of interest rates of all central banks may be a useful diagnostic metric, but it does not directly affect individual national economies. Some nations are used to much higher sustained interest rates than others. The second problem with the headline is that there is no way we can define a precise “global interest rate” threshold beyond which the financial system will collapse. We can consider the following two fictitious headlines to be just as informative as the previous headline:

Global average central bank interest rate crossing 5.15% could trigger global financial collapse

Global average central bank interest rate crossing 4.11% could trigger global financial collapse

The point being that the “global interest rate” threshold is not a very precise metric with which to predict whether banks will collapse. What determines whether a bank collapses is the details of its balance sheet and the national interest rate. The collapse could happen for many different values of the “global interest rate”. Similarly, regional temperature exceeding a threshold could trigger a regional destabilizing feedback. However, the same regional warming could happen for a range of values of the predicted “global average warming”.

An economist, like Yellen, would likely refrain from making precise-sounding pronouncements like the fictitious global interest rate headlines above. Yet, Yellen seems comfortable using a precise time period—a decade—in her statement about the imminence of climate tipping points. People outside a discipline (be it climate science or economics) often attribute more precision to predictions within that discipline than do people within the discipline. Outsiders are typically unaware of the unquantified uncertainties that are implicit in predictions within a discipline. They may also not understand the many terminological subtleties intrinsic to a discipline, such as the distinction between predictions and projections, or between full-complexity and intermediate-complexity models. This prevalence of superficial certainty among outsiders can be characterized as the uncertainty trough (Figure 3). It can help explain why aspirational warming thresholds that should be taken seriously end up being taken literally as precise physical thresholds. Figure 3 The Uncertainty Trough: As distance from knowledge production increases for a complex technology, the uncertainty first decreases, but then increases again among those hostile to the technology [Figure 15.1 from D. MacKenzie, 1998]¹⁰

Just as there is no single global bank that will collapse at a specific interest rate, there is not a single global climate process that will collapse at a precise warming threshold. There’s a range of possible thresholds across different regions and nations. Given our imperfect knowledge of the complex system (banking or climate), we cannot predict precisely when the harmful impacts will occur. But we can say that a higher interest rate will increase the risk of banks failing and greater warming (due to continued carbon emissions) will increase the risk of harmful regional climate impacts.

The downsides of tipping point headlines

Although the IPCC climate models do not exhibit global tipping point behavior in their predictions, articles with headlines about the planet being on the “brink” of 1.5/2°C warming appear regularly in the media. The articles often suggest that we may have already crossed, or that we might soon cross, global tipping point thresholds for irreversible change. There are two reasons tipping points continue to generate interest, one scientific and one cultural:

1. Even our state-of-the-art global climate models are scientifically incomplete. Our most complex global climate models could be overestimating the strength of the stabilizing feedbacks and/or underestimating the strength of the destabilizing feedbacks. Some longer-term feedbacks, such as ice sheet melting, are often omitted from these models. Simpler climate models which address these longer-term feedbacks (but in a cruder fashion) are more prone to exhibiting tipping point behavior than the global climate models. But the simpler models may not represent all feedbacks correctly.

2. Readers love a gripping story. Dramatic narratives describing tipping points are more engaging than pedestrian pieces about steady warming or moderately interesting articles about extreme heat waves, even if the latter are more firmly grounded in scientific certainty.

The risk of global climate tipping points is real.¹¹ However, that risk lies more in the realm of unknown unknowns, rather than known unknowns. Additional research may not yield any more precise risk estimates due to “irreducible imprecision” in the highly complex and nonlinear climate system.¹² Our imperfect knowledge of the risk adds to the urgency of curbing carbon emissions within the next few decades. We also need to monitor the climate system carefully for incipient tipping point behavior.

At this point a climate activist reader may wonder: What’s wrong with overplaying the risk of tipping points? Wouldn’t it just motivate us to take stronger action to mitigate climate change? Since our current mitigation efforts are far from adequate, that would be an upside. But there are also downsides. Loose talk of inevitable tipping points can trigger despondency or doomerism. Fear of imminent tipping points provides a justification for billionaires to support geoengineering.¹³ There is also the potential reputational damage if tipping points touted as being imminent/inevitable turn out to be neither when aspirational warming limits like 1.5/2.0°C are breached.

Articles about tipping points typically focus on a single destabilizing feedback as the protagonist driving the narrative. Remember that there are also stabilizing feedbacks that work behind the scenes to counter the protagonist. Pay careful attention to caveats, conditionalities, and timescales buried in the article discussion. For example, when a geologist talks about “abrupt climate change”, they may be referring to something that happens over many centuries or even millennia.

If you want to stay informed about new research on climate tipping points without becoming despondent, there’s a simple — if somewhat impractical — solution: Read articles about tipping points all the way to the end, while completely ignoring the headline! If you just skim the headline and the lede paragraphs, you’ll miss the nuanced discussion that is usually present in the rest of the article.