Bank of England System-wide Exploratory Scenario Stress Test
This article reviews the recently specified 2023 Bank of England System-wide Exploratory Scenario (SWES) stress test exercise. See https://www.bankofengland.co.uk/financial-stability/boe-system-wide-exploratory-scenario-exercise.
When is it happening, whom is taking part, how does it compare to prior stress test exercises and what features of this new exercise stand out?
When?
The existence of SWES was announced in June 2023. PRA has been information gathering and building out the participant list since then.
The scenario was released last Friday (10 Nov). This starts Round 1 of the stress scenario phase. I describe and assess this scenario later in this post. Participants should submit their round 1 content by January 2024.
Round 2 of the scenario phase is currently intended to be launched in Q2 2024, with the exercise to conclude in Q4 2024, with a BoE report on results to follow before end 2024.
Whom?
A list of 54 participants has been published by BoE. See Table A in The Bank of England’s system-wide exploratory scenario exercise | Bank of England.
It includes major UK banks, UK life insurers, asset managers, hedge funds, pension funds, and market-infrastructure entities (eg clearing houses).
The list of participants may now be fixed. Full details of the SWES including response templates are only available to participants.
Context: BoE Exploratory scenarios
This is the 2023 version of BoEs biennial exploratory scenario exercise.
This biennial process was put in place in 2015 in https://www.bankofengland.co.uk/-/media/boe/files/stress-testing/2015/the-boes-approach-to-stress-testing-the-uk-banking-system.pdf.
The first BoE exploratory scenario exercise was in 2017. It covered only six major UK banks & one building society. It focused on stagnating global trade, prolonged subdued GDP growth, prolonged very low interest rates and increased competition in UK banking. Results were published in Nov 2017 with a headline that “the UK banking system is resilient” to a scenario involving losses that “would have wiped out the common equity base of the UK banking system 10 years ago”. See https://www.bankofengland.co.uk/-/media/boe/files/stress-testing/2017/stress-testing-the-uk-banking-system-2017-results.pdf .
The second BoE exploratory scenario exercise was in 2019. It covered the same seven entities as the 2017 exercise. It incorporated an instant decrease in gilt prices of 5%, one month of stress-then-closure of foreign-exchange swaps markets and 3 months of accelerated liquidity outflows from banks/building societies. It was a two-phase exercise with the second phase to include feedback elements based on first-round submissions. The first phase was completed, but the second phase was cancelled due to the Spring 2020 “dash-for-cash” as the first global wave of Covid19 lockdowns led to massive real-world liquidity stress. The original scope can be found from p53 of https://www.bankofengland.co.uk/-/media/boe/files/financial-stability-report/2019/july-2019.pdf, while limited conclusions from March 2021 can be found in paras 89-99 of https://www.bankofengland.co.uk/-/media/boe/files/financial-policy-summary-and-record/2021/march-2021.pdf.
The third BoE exploratory scenario exercise was in 2021. It focused on physical and transition climate risks. It covered the same seven bank/buildings societies as previously, five main UK life insurers, the UK parts of six main general insurers and ten Lloyds or London syndicates. It comprised three scenarios: early policy action, late policy action, and no additional policy action. The primary elements of this exercise had been scoped out in a December 2019 Discussion Paper. This exercise used a 31/12/2020 t0 valuation date with a 3 month “round 1” period from June-Sept 2021. A second round launched in February 2022 included a doubling of climate-related losses versus round one. A final report on this exercise was published in May 2022.
The themes for both the 2019 and 2021 exploratory exercises were both published in July 2019 (see link above). Given the form of the 2007-9 global financial crisis, a severe liquidity stress was a clear element that hadn’t been covered in the 2017 exercise. That the second phase of the 2019 exercise was cancelled due to an event of its type being in progress demonstrates the relevance of the 2019 theme. Focusing the 2021 exercise on climate risk was fully consistent with Mark Carney’s focus on this risk (within BoE and globally) during his period as BoE governor, and with the UKs role as chair of COP26 in Glasgow, held in H2 2021.
Round 1 of the BoE 2023 SWES exercise
See Launch of the scenario phase of the system-wide exploratory scenario | Bank of England.
The 2023 exploratory scenario exercise has many more participants, spanning a much broader range of market participants and providers than any other.
The scenario has a clear top-down narrative with a day 1 “significant geopolitical shock”, large-scale sovereign bond sales by sovereign wealth funds, default of a mid-sized hedge fund and downgrades of some financial institutions, with no improvement in economic outlook by day 10. This is a fuller narrative than previous BoE exploratory scenarios. The specific geopolitical shock is not specified, but further Russian military aggression, further expansion of the Israel/Gaza conflict, potential China/Taiwan conflict and a Trump re-election are all candidate trigger events.
The SWES references the spring 2020 first-wave-of-Covid “dash for cash” stress and the H2 2022 LDI episode, ie it acknowledges and builds upon two recent system-wide events to which the SWES could be seen as a big-sibling. The three US mid-sized bank failures in Spring 2023 may also have informed the SWES scenario design.
Unlike any comparable prior BoE exercise it uses a base valuation date, 31/10/2023, which is not a standard quarter-end, which may pose practical challenges for any firms that carry out only an approximate valuation process at non-quarter-end month ends. It applies a scenario that is defined over 10 days, with 5 separately reportable stages. No get-out of applying the stresses instantaneously, as was embedded in the IST2022 exercise (see below), appears to have been publicly offered, but use of such an approximation by life insurers seems likely.
Stresses are defined for most “obvious” liquid tradeable markets: government bonds, interest rate swaps, inflation swaps, govt bond futures, repo rates, bond credit spreads, swaption vols, equities, VIX, gold, spot FX, FX cross-currency swaps.
No stresses are defined for real estate, nor of how non-tradeable assets should move compared to their tradeable-market nearest equivalents. Exclusion of real estate shocks may leave insurer exposure via unit-linked real estate funds, commercial real estate loans, equity release mortgages etc under-captured. Exclusion of shocks to off-market assets is more understandable, but it is to be hoped that BoE will provide sufficient additional guidance (eg re participant-driven FAQs) on this aspect to ensure coherence between exposures within a firm and consistency between firms.
Zero stresses are defined for energy, emissions, metals, agricultural commodities, freight. While BoE may be aware that this exercise is already complex, some may consider there to be a coherence gap between the language of its scenario description (eg a day 1 “significant geopolitical shock”) and no upward shock to food and energy pricing.
Zero stresses are defined for non-US equity volatility. It is curious that the same equity value shock applies for all equities but that option-implied equity volatilities in UK/Europe/Japan remain flat while the US equity volatility indicator, VIX, increases sharply.
Interest rate swap shocks are defined as parallel shifts. This may be an understandable simplification to reduce the quantum of assumptions that BoE has to provide, but a) large shock to interest rate yield curves tend not to be parallel and b) the shift can’t be parallel in both forward space and spot space.
Only market shocks are specified. It is left to firms to assess the impacts of such a scenario on their wider experience (eg withdrawals) and pricing (eg what to assume about whether bulk annuity NB quotes they had “out” at 31 October 2023 were taken up or not, differing from what happened to these quotes over these 10 days in reality).
The shock is assumed to occur within the 31 October 2023 to 10 November 2023 window. So within SWES the current well-developed HMT/PRA agenda of Solvency II reform has not occurred. There is a risk that results life insurers, particularly those with large matching adjustment businesses, may see this exercise as being of reduced relevance because of this. On the flip side, such life insurers may see this exercise as a further means of influencing those parts of Solvency II reform that are still not finalised. Managing what-if analysis for wide-ranging Solvency II reform (eg working out what MA attestation means in reality) at the same time as carrying out such a detailed statutory stress test exercise (in which the question of how a firm might attest its MA in severely stressed conditions would have been highly relevant if SII/MA reform were reflected in SWES) could be particularly testing.
Round 1 of the exercise is about more than just a set of output numbers. Firm’s choices of management actions, assumptions about internal and external constraints on deliverability of those actions, and coherence between those assumed actions and their firm’s Internal Model, ORSA, Recovery Plan and any emergent Resolution Plan are likely to be material areas of focus by BoE.
BoE has indicated that Round 2 may involve updated price paths in the Round 2 shock scenario, and amendments and additional detail in the Round 2 narrative. Given that Round 1 spans so many different participants in the UK financial markets the potential for Round 2 to involve significant alterations to scenarios, reflecting BoEs learnings from aggregated Round 1 submissions, appears to be substantial.
Wider context
This section provides a round-up of links re various recent stress test exercises. But they tend to focus on either banks or on insurance. SWESs breadth of scope is a key potential strength, but will likely make the exercise harder for BoE to manage.
EIOPA ran an EU-wide insurance stress test in 2021, using a “lower for longer” scenario. See https://www.eiopa.europa.eu/browse/financial-stability/insurance-stress-test/insurance-stress-test-2021_en.
ECB ran an EU-wide banking stress test in 2023 using a scenario with high interest rates, high inflation and low growth, covering 57 banks. See https://www.bankingsupervision.europa.eu/press/pr/date/2023/html/ssm.pr230728~a10851714c.en.html.
Bafin/Deutsche Bundesbank ran a 2022 stress test exercise for smaller German banks. See Results of the 2022 LSI stress test | Deutsche Bundesbank.
Deutsche Bundesbank DP 19/2021 is a good review of the state of play of system-wide stress testing in different jurisdictions globally at that time https://www.bundesbank.de/resource/blob/867670/aa6f1d612da1a8a1aab8814555e99b6a/mL/2021-06-11-dkp-19-data.pdf but this is restricted to the banking system.
The Federal Reserve ran a 2023 stress test exercise on major US banks. See https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2023.htm.
The BoE runs regular insurer stress testing exercises, the most recent of which was IST2022. It ran from May to September 2022. The life insurance stress scenario for that exercise is summarized in figure 2 of https://www.bankofengland.co.uk/-/media/boe/files/prudential-regulation/letter/2022/may/ist-2022-life-insurance-scenario-specification-guidelines-instructions.pdf. The IST2022 life insurance scenario was one that notionally occurred over 1 year from 31/12/2021, but which was applied as a sequence of four instantaneous shocks to the 31/12/2021 position. Over the four stages it featured widespread market shocks (interest rates down, equities and real estate down, credit spreads up, interest rate/inflation/equity vols up) plus a simple credit downgrade shock and a longevity increase shock, and phased assumptions as to which types of trading were possible at each stage. By the time of the IST2022 submission deadline, 10yr gilt yields were >260bps higher than at 31/12/2021, making the IST2022 interest rates down from 31/12/2021 scenario appear to be of pretty low ongoing relevance from the perspective of 2023 conditions.
What next?
If your firm is working out how to resource its SWES now the spec is out, please get in touch with your usual APR contact, or with me, alan.reed@aprllp.com.
Alan Reed
November 2023
