Gaming theory
Last week, the Bank of England sketched out the details of its inaugural climate stress tests. The watchdog laid out the exercise's objectives and outlined the climate scenarios that participating banks and insurers would be subject to.
On first reading, the parameters appear strict, rigid – leaving firms little room to input their own assumptions and projections.
This may be by design to prevent the stress tests from being gamed.
If financial institutions were allowed to set their own climate change scenarios, they could select those that pose less of a threat to their business models, making the seismic climate-related changes expected over the coming decades appear of small consequence to their ongoing stability.
This is likely one reason why the BoE is imposing its own “climate pathways” for temperature, emissions, and climate policies in the forthcoming tests. In doing so, the BoE, rather than banks and insurers, will define the physical and transition risks that will stress their balance sheets. Not only should this ensure uniformity, it should also prevent firms from cherry-picking scenarios to their own tastes.
Source: Bank of England
This is important, as the first objective of the tests are to size the financial exposures of firms to climate risk. Of course, these are tough to estimate. Physical risks, such as flooding, are tricky to model and difficult to judge at a property-specific level. Transition risks, meanwhile, are subject to a volatile, politically-driven policy process and uncertainties around the pace and effectiveness of technological change.
Therefore it’s perfectly possible that similar climate risk scenarios using slightly different assumptions could produce wildly different outcomes. Given this, it’s understandable why the BoE wants to set the rules itself instead of delegating scenario development to firms themselves.
By putting forward three scenarios of its own – an “early action” scenario assuming the increase in global temperature is kept below 2 degrees Celsius, a “late action” scenario where transition is delayed, and a “no action” scenario where temperatures spiral higher – the regulator should prevent firms cooking up overly ambitions, or downright unrealistic, scenarios of their own that flatter their business models.
Source: Bank of England
But the BoE is not shutting out industry input entirely. The regulator will not provide every variable that participants would need to estimate the impacts on individual counterparties and assets. Banks and insurers will therefore have to extrapolate additional variables on their own.
As James Belmont at Baringa Partners, a consultancy, writes: “This expansion will require highly configurable modelling capability that enables firms to fill in these details in an internally consistent fashion while adhering to those parameters that are specified in the Bank’s scenarios.”
This means that firms will have a say in the tests, though they will be subject to the constraints of the BoE's own scenarios. Because of this, it’s hard to see how extrapolation can be done in such a way as to artificially lower the impact of the overall scenarios.
Freeze out
Institutions will have to assume their balance sheets are frozen as of June 30, 2020 under the tests. This bars firms from incorporating the anticipated effects of risk-reducing activities and management actions they could theoretically take to limit the financial pain projected by the stress scenarios.
These can be significant. In the BoE’s recent bank stress tests, the effects of strategic management actions – like cutting dividend payments, additional Tier 1 (AT1) bond coupons, employee bonuses, and converting bail-in debt to equity – lifted the overall projected minimum stressed capital ratio of the seven participants by 220 basis points.
Perhaps the anticipated management actions were reasonable and supportable, but in a real crisis the best-intentioned plans can go out the window. Assuming a static balance sheet is therefore a blunt, but effective, way of stopping firms from muddying the test results.
However, in another way using a static balance sheet could allow gaming. Because they know the snapshot of their balance sheets would be taken on June 30, 2020, firms could conceivably adjust their assets and liabilities ahead of that date to lower their climate risk profiles going into the tests.
For example, it’s possible a bank holding a bunch of fossil-fuel company equities could exchange these with a willing counterparty for a portfolio of climate change-resilient assets, like government bonds, for a short period covering the June 30 cut-off date. These equities, highly sensitive to transition risks, would then not be included in the assessment, despite the fact they'd only been temporarily shifted off-balance sheet.
How incentivised participants would be to engage in such skulduggery is unclear. The test results will be published by the BoE in 2021 – but to what level of detail is yet to be determined. Additionally, the BoE will not be testing banks against minimum capital requirements, as in the annual prudential stress tests. This somewhat takes the pressure off.
However, banks have gamed stress tests in the past, and no firm wants to come out of this looking bad. The reputational risk alone could provide a strong incentive to come out of the climate scenarios looking hale and hearty.
The lengths firms will go to ensure such outcomes remains to be seen.
Read the previous newsletter on the CFTC's climate subcommittee, and its first presentation to the market risk advisory committee.
If you enjoyed this post, please click the thumbs up icon above
