Essay by Eric Worrall
Professor Andy Pitman wants central banks to hire climate scientists to help them build high resolution risk models, to properly evaluate regional climate risk. But the Aussie RBA disagrees their role is to evaluate climate risk.
Trillions of dollars at risk because central banks’ climate models not up to scratch
Climate research finds modelling used cannot predict localised extreme weather, leading to poor estimations of risk
Wed 10 Aug 2022 03.30 AEST
Prof Andy Pitman, director of the Australian Research Council’s Centre of Excellence for Climate Extremes, said regulators are relying on models that are good at forecasting how average climates will change as the planet warms, but are less likely to be of use for predicting how extreme weather will imperil individual localities such as cities, however.
Pitman said. “We need to take this issue seriously – not just access information flying around and think we can package it to do proper economic assessments.”
“We do not evaluate risks on behalf of the entities that we regulate,” the spokesperson said.
…Read more: https://www.theguardian.com/environment/2022/aug/10/trillions-of-dollars-at-risk-because-central-banks-climate-models-not-up-to-scratch
Professor Andy Pitman’s study is available here.
Of course, to meaningfully model regional climate change, national reserve banks and central banks would need climate models which deliver useful results.
Professor Andy Pitman himself let the cat out of the bag in 2019, about climate model utility, when he revealed climate scientists can’t even tell us whether global warming will increase or decrease rainfall – a pretty important metric for determining climate impacts, especially in a dry country like Australia.
There is also a moral hazard associated with central banks being too prescriptive about risk assessments. If central banks demand regulated entities use specific models, those same central banks will carry moral liability if banks suffer financial losses, because of defects in the central bank risk models.
Betting any imposed model would be defective is a very safe assumption, because all non trivial financial models contain defects and limits.
For example, many financial models in the leadup to the 2007-10 global financial crisis ignored liquidity risk, by assuming that there would always be a buyer for financial instruments. The assumption of infinite market liquidity spectacularly broke down during the 2007-10 subprime crash, when bankers who played it safe, who stuck with the recommendations of their financial models, got stuck with catastrophic loss making subprime mortgage investments they couldn’t offload in time to save their own skins.
One of the great survivors of the GFC, Greg Lippmann of Deutsche Bank, famously made a vast profit during the subprime crash by overruling his computer model recommendations, and betting everything on an imminent total collapse of the market.
What about climate models? I suspect bankers would love to find a way to create better risk models, and would be happy to further integrate climate risk models into their financial risk projections, if such integration yielded value. But are climate risk models ready for incorporation?
Former HSBC responsible banking head Stuart Kirk recently complained he was being pressured to add unrealistic assumptions to climate financial models to make them interesting. “Even with a carbon tax, even with growth, they couldn’t make climate risk move the needle, so they had to get their clever little wonks in the back room to put a gigantic interest rate shock in their models to make headlines”.
Of course some scientists make far more interesting predictions than predictions of mild financial losses, like predictions of imminent catastrophic tipping points or even imminent extinction. But how do you incorporate a wild range of scenarios, from near business as usual to an imminent extinction event, into a financial risk model, and still produce meaningful output? Especially since even climate scientists producing more pedestrian projections sometimes admit their models are running “implausibly hot”?
How do Aussie banks for example model future changes in rainfall into their regional farm mortgage credit score models, when Professor Andy Pitman himself recently admitted that nobody can tell whether global warming will cause continental scale rainfall to increase or decrease?
At least bankers experimenting with market liquidity models have something tangible to work with – they can hindcast their models against high quality historical price and liquidity data, to see if their models generate the right buy / sell signals. But as Anthony Watts has demonstrated, historical global climate data has serious quality issues.
Until climate scientists come up with climate models which produce a narrow range of useful predictions, models which add quantifiable value to risk calculations, they can continue to expect a less than enthusiastic reception outside their immediate fan base, when they try to push bankers to make more use of their product.
I have no problem with Professor Andy Pitman approaching bankers and asking for funding. I was impressed Professor Pitman’s candour, at his admission of model uncertainty in 2019, even if he later appeared to back off and qualify his original statement. I think Professor Pitman would find bankers receptive to the idea of helping to fund the development of better risk models. But it seems a big stretch for Professor Pitman to insist that climate models in their current form are bank risk ready, that models are mature enough for bankers to start hiring climate scientists to help integrate climate models into their risk projections.
via Watts Up With That?
August 10, 2022