Review of 2023: A hard road to a soft landing

Central banks were trying to engineer something of an economic miracle in 2023, by taming inflation without causing a recession. The aim was a soft landing, but the glide path was bumpy. 

The historic series of US interest rate hikes contributed to a string of bank failures. The collapse of Silicon Valley Bank on March 10 sparked a crisis of confidence that spread as far as Credit Suisse, which was forcibly sold to UBS five working days later. 

The collapse of Silicon Valley Bank on March 10 sparked a crisis of confidence that spread as far as Credit Suisse, which was forcibly sold to UBS five working days later

Regulators faced tough questions about their handling of the crisis. The US had not fully implemented global standards for measuring interest rate risk in the banking book, which could have alerted supervisors to problems at SVB and other regional lenders. The Swiss authorities did not use the internationally agreed playbook for the wind down of ‘too-big-to-fail’ banks, and spooked the young market for additional Tier 1 (AT1) capital, after wiping out bondholders while billions of dollars were paid to shareholders. One silver lining was that derivatives markets remained stable – thanks in part to central clearing – meaning a 2008-style meltdown was never on the cards. 

Confidence was ultimately restored and the market’s focus swung elsewhere. The phaseout of US dollar Libor was completed in June, and prudential regulators unveiled their long-delayed Basel III ‘endgame’ proposal in July. 

Rising rates and regulatory changes were not the only concerns for financial firms in 2023. Earlier in the year, trading in cleared derivatives was disrupted for several days following a ransomware attack on financial software firm Ion Group, while compliance concerns stopped banks from jumping on the generative AI bandwagon. 

Here, we look back at some of the year’s biggest Risk.net stories, and those with the greatest potential to shape the coming 12 months. 

PRE-FLIGHT CHECKS

While SVB’s failure was largely due to gross mismanagement, regulation also played a role. 

US regulators did not fully implement a Basel standard – known as the economic value of equity, or EVE – for measuring interest rate risk in the banking book (IRRBB), which could have alerted them to potential losses at SVB more than a year before it collapsed.

As Risk.net reported in March, SVB’s EVE at the end of 2021 – which the bank voluntarily included in annual reports until this point – showed its Tier 1 equity capital would drop by more than 35% if its fixed rate assets had to be sold following a 2% rise in interest rates. An attentive regulator armed with this information might have encouraged SVB to either shift its investments to floating rate assets or hedge with derivatives paying a variable rate. Instead, SVB was allowed to unwind nearly $15 billion of swap hedges in its AFS portfolio over the course of 2022. 

From March 2022 to February 2023, the Federal Reserve hiked interest rates by 4.5%. The resulting losses effectively wiped out SVB’s equity base, leaving it especially vulnerable to a run on uninsured deposits. The EVE measure was missing from its final annual report.

Have US regulators learned their lesson? An internal review conducted by the Fed blamed gaps in supervision rather than regulation. The report suggested that “supervisors could reconsider how to best reflect interest rate risk in regulatory capital” but stopped short of directing them to fully implement the Basel standards, which include a mechanism for applying capital add-ons for banks running excessive interest rate risks.         

“There are no formal regulations in the US concerning the management of interest rate risk” – Scott Cammarn, Pierce Atwood (Missing Basel metric could have revealed SVB risks, March 15)

“It’s frowned upon in the industry to have volatility in P&L that comes with undesignated interest rate hedges, so generally, it’s not a common risk management practice” – Leanne Fredericks, Riverside Risk Advisors (Did US hedge accounting rules contribute to SVB’s recklessness?, April 4)

“Every time there’s been a rising interest rate environment in the United States, there’s always breakage in the system” – Keith Noreika, Patomak Global Partners (SVB failure signals Fed’s need for a speed rethink, May 2)

BAILING OUT 

The collapse of SVB – seemingly out of nowhere – triggered depositor runs at other banks that suddenly looked suspect. Most were US regional lenders, but the crisis leapt the Atlantic after the largest shareholder in Credit Suisse ruled out a further capital injection for the troubled bank, which had already endured months of bleeding deposits. When its stock price tanked anew, local authorities decided enough was enough. But the terms of its shotgun wedding with UBS caught the market by surprise and further undermined confidence in banking regulators.

Swiss regulators avoided classifying the deal as a resolution, which would have required shareholders to be wiped out, and instead came up with a rescue plan that involved writing down $17 billion of Credit Suisse’s AT1 bonds. 

Bondholders were outraged. Spreads on AT1s blew out – temporarily closing the new issue market – and regulators in the EU and UK rushed out statements that shareholders would be wiped out first in their jurisdictions.

The handling of Credit Suisse’s failure cast new doubt on the global resolution regime for large banks. The Swiss finance minister declared after the fact that it wasn’t possible to wind down a bank the size of Credit Suisse under those rules. Regulators elsewhere dispute this, but how they will manage the next major bank failure is anyone’s guess. 

“Burning the AT1 but not the shareholders is absolutely against the rules of the Financial Stability Board that everybody has adopted, including the Swiss” – Ignazio Angeloni, former member of the ECB’s supervisory board (How Finma milked Credit Suisse’s CoCos to close UBS deal, March 21)

“There’s going to be scar tissue around this event, and investors are going to be much more demanding of what’s put into, or what’s left out of, [AT1] documents” – Hunter Hayes, Intrepid Capital Management (Investor wish-list offers no quick fix for Swiss CoCos, May 4)

“Maybe what the crisis has taught us is that while it is good to have a resolution strategy, that is not enough. At the end of the day, the crisis may force you away from this strategy and you may need a more articulated tool-kit” – Francesco Mauro, European Banking Authority (Lessons on bank resolution, from Silicon Valley to Zurich, June 20)

“What the Swiss finance minister said about resolution as not fit for purpose for a G-Sib like Credit Suisse – we don’t think it’s a valid statement from our point of view” – Dominique Laboureix, European Union’s Single Resolution Board (SRB head asks for extra tools to restore faith in resolution, July 20)

DEPOSIT FLIGHT

The speed of the digitally-driven run at SVB – and the prolonged pain at Credit Suisse – raised serious questions about the assumptions used for deposit modelling.     

The Basel liquidity coverage ratio (LCR) uses runoff factors to calculate the amount of high-quality liquid assets (HQLAs) banks must hold to meet outflows. Banks are expected to lose 5% of their insured deposits in stress periods, while uninsured deposits have a 10% run-off rate. 

The runs in March shattered those assumptions. SVB lost 30% of its deposits on the day prior to its collapse, while two-thirds of Credit Suisse’s deposits were withdrawn over two quarters. 

Regulators briefly debated imposing more conservative outflow assumptions for the digital age, but ultimately decided tougher supervisory stress tests should be enough to catch outliers. 

The year’s bank failures also highlighted a conflict between prudential rules on liquidity risk and the accounting treatment of bonds on bank balance sheets. The HQLAs banks hold to comply with the LCR are typically recorded as held-to-maturity for accounting purposes and do not need to be marked at fair value. That can lead to problems if those assets are sold to meet outflows, as the bank would then need to recognise a loss that depletes its capital and potentially triggers more outflows. Reconciling the liquidity rules with global accounting standards should be high on regulators’ to-do lists for 2024. 

“If you are contemplating using the securities for liquidity – if and when you need liquidity – then it’s just completely inconsistent with the spirit of classifying things as HTM” – Stephen Ryan, New York University Stern School of Business (SVB opens floodgates on liquidity buffers debate, March 30)

“There is also a question to be discussed on the international stage on the calibration of the liquidity coverage ratio” – Carolyn Wilkins, external member of the Bank of England’s Financial Policy Committee (Why tougher liquidity rules may not reduce the risk of bank runs, April 25)

“You could have a bank which is objectively not exposed to a large risk. But if enough depositors think it is [exposed], the bank will default. Once the deposits disappear, the liquidity isn’t there” – Rama Cont, Oxford University (The Tweet and the trust collapse: how banks can fall on a dime, April 28)

“If there are some clients reacting in one or two days, would the same speed go on for 30 days in a row? It is not plausible – maybe for the first five or 10 days, but then definitely slowing down” – Gábor Koncz, Raiffeisen Bank International (Risk modellers navigate fearful new world of depositor behaviour, May 25)

HIGH PRESSURE

While the US has mostly ignored Basel’s IRRBB standards, regulators in Europe spent much of 2022 working on new tests to identify growing risks in banks’ loan and deposit books. 

The standard Basel test looks at whether a hypothetical rates shock would cause the value of assets and liabilities to decline by more than 15% of a bank’s total Tier 1 capital. In October 2022, the European Banking Authority proposed an additional test that hinged on whether a rates shock would cause net interest income to decline more than 2.5%.

Then came the backlash. Analysis shared with Risk.net revealed around half of European banks would fail the test. The EBA backtracked in April this year by doubling the threshold for outliers to 5%. While some wanted the bar raised further to 7.5%, others grudgingly accepted the EBA’s compromise. The European Commission adopted the revised proposal on December 1. The first public disclosures under the rule are expected in June 2024. 

The Basel Committee also faced calls to update its IRRBB standards. The biggest potential change involves applying a Pillar 1 capital charge for IRRBB. This idea was part of the original rules drafted in 2016, but was dropped after pushback from banks and US regulators. Other elements of the current rules may also need an update. Research conducted by a group of bank risk managers and shared with Risk.net in November showed how banks could take excessive interest rate risks and still comfortably pass the existing outlier tests. 

“If you actually apply the 2.5% threshold, it means that about half of the banks in Europe are outliers” – Adrian Docherty, BNP Paribas (We’re all outliers now: Europe’s unflattering IRRBB test, Feb 2)

“The test, as currently calibrated, will most probably identify as outliers many banks that are actually perfectly OK from a regulatory perspective” – Paul Newson, IRRBB consultant (Why risk managers don’t trust the EU’s new IRRBB test, February 26)

“Based on the current draft of the test, it could be billions in notional, depending on the size of the balance sheet and business model of the bank” – Andreas Bohn, McKinsey (EU banks need ‘billions’ in hedges to pass new NII test, March 17)

“Even if it’s in a technical standard it shall not mean that the threshold should be fixed forever. We are able to amend it on a regular basis if need be” – Delphine Reymondon, EBA (Europe’s new IRRBB test: the riddle with no answer, May 17)

“It is possible that the idea would now have better chances in view of recent events” – Mark Branson, BaFin (Regulators’ remorse: SVB and the case for IRRBB capital charges, June 6)

“Systems in banks for calculating EVE and NII can be disjointed, and you have different teams working on them. Some banks do and some don’t push through the same scenarios” – Eric Schaanning, UBS (Filling the gaps in Basel’s interest rate risk measures, November 1)

HARD LANDING FOR SOFR

As if regional banks didn’t have enough on their plates, the long-planned cessation of US dollar Libor on June 30 created another headache. Loan portfolios had to be transitioned to the secured overnight financing rate, or SOFR, before the deadline. Hedging this new exposure proved costly, however. 

Most borrowers had opted to link their loans to a term version of SOFR. But a ban against interdealer trading of term SOFR swaps meant that while dealers could provide hedges to regional banks, they were prevented from offsetting this risk with other dealers. This created a one-sided market, with term SOFR swaps trading at a 2-3bp premium compared with overnight versions. The regional banking crisis made matters worse, with bid-offer spreads widening as much as 10bp due to concerns about counterparty risk.

The demise of Libor also marked the end of forward rate agreements, which could not function with the new benchmark, leaving regional banks without an obvious way to hedge against rising funding costs.   

The hits kept coming. On July 3, the International Organization of Securities Commissions ruled that credit-sensitive rates, which some regional banks embraced as an alternative to SOFR, did not meet its principles for financial benchmarks.

There was a silver lining in the SOFR clouds. As interest rate expectations eased in the second half of the year, regional banks were able to hedge against future rate cuts by entering receive-fixed swaps linked to term SOFR at attractive prices. Dealers also found other ways to offload their SOFR basis risks, including in structured notes and by convincing their internal lending and asset-liability management arms to buy receive-fix swaps. With the SOFR market finding its feet, attention shifted to Europe, where the legacy Euribor benchmark appears to be on borrowed time.

“It’s definitely ‘when’ not ‘if’. Maybe with the dollar path becoming clearer and Libor’s June deadline moving closer, all eyes will start to turn to Europe” – Phil Lloyd, NatWest (Is Euribor on borrowed time, or here for the duration?, March 13)

“If the market does not have an ability to adequately hedge credit risk, there could be potential gaps and exposures. And this is when, perhaps, risk-free can be risky” – Mark Cabana, Bank of America (FRA-OIS demise leaves hole in bank treasury risk management, May 8)

“Spreads have dramatically widened out. It’s very dealer dependent and we find a great deal of disparity between dealers in terms of how they’re pricing on a given day” – Chris Slusher, Derivative Path (Regional banks face soaring term SOFR spreads, May 24)

“A lack of credit-sensitive rates in financial markets will leave risks in places that can’t handle it” – a rates trader at a US bank (As Libor ends, credit-sensitive rates face day of reckoning, June 29)

FINAL APPROACH 

US regulators released their long-awaited Basel endgame proposal on July 27, implementing the final version of the global bank capital rules agreed in 2017.

While those standards called for the elimination of internal models for calculating operational risk, the US also unilaterally decided to end the use of internal models for credit risk – leaving market risk as the only area where US banks are allowed to use their own models for capital calculations.

The irony is that many banks – both in the US and elsewhere – are already choosing to ditch the use of internal models for market risk capital. The Fundamental Review of the Trading Book introduces a more stringent approval process for internal models, while the implementation of output floor caps under the final Basel rules limits the amount of capital savings they can produce. 

The double whammy is leading more and more banks to conclude that internal models for market risk are simply not worth the hassle. In November, Risk.net reported that at least three large European banks are planning to stop using them entirely, with one citing the output floor as the main reason for doing so. If US banks follow suit, they would become almost entirely dependent on regulator-set standardised capital approaches.

“The transition to standardised will definitely eliminate a substantial volume of modelling and analytical work, and may even reduce effort in the data management segment” – Brett Ludden, Sterling Point (US credit risk modellers prepare for life after IRB, February 15)

“Credit and operational risk are going to be standardised, and a lot of banks may not even apply for model approval for market risk, because it’s so much work and the benefits are not there, so I’m not sure how much the capital floor is going to matter” – a US regulatory expert at a large bank (Banks slam zombie floors in Basel endgame proposal, August 3) 

“The benefit of internal models is becoming smaller and smaller because our internal models are now being calibrated to more and more stress periods” – a senior market risk modeller at a European bank (As banks limit FRTB model use, outputs get more volatile, October 26)

“For us, it was almost an easy choice [to use the standardised approach]” – a senior risk modeller at a European bank (Party’s over as more banks drop internal models for market risk, November 13)

“It is painful to build a model that actually works and satisfies the statistical requirements” – a senior bank risk modeller (FRTB managers face hard facts about risk factors, November 20)

UNDER THE RADAR 

While the changes to the bank capital rules were well-telegraphed, a parallel effort to shore up standards for US insurers largely flew under the radar. In February, the National Association of Insurance Commissioners (NAIC) voted to press ahead with a plan to raise capital charges on riskier tranches of collateralised loan obligations (CLOs). The move threatened to derail an investment strategy popular among private equity-backed insurers, which tend to have higher than average allocations to CLOs. 

The proposal split the industry. Private equity-backed firms criticised the regulator’s plan as flawed, while others backed the effort to rein in what they viewed as a growing risk to the industry.

But the regulator managed to unite the industry with a follow-up proposal that would also give it the power to overrule credit ratings set by external agencies. Industry sources universally blasted that idea as “nonsensical” and an overreach. 

The NAIC is pressing ahead with its plan to build an internal model to set capital requirements for CLOs. The final calibration of that model could well determine whether the strategies of private equity-owned insurers remain viable.   

“Nobody knows how the industry is going to be affected” – Dmytro Mukhin, AllianceBernstein (US insurance regulators move to kill CLO arbitrage, February 27)

“This is effectively anti-competitive behaviour. They’re acting as judge, jury and executioner” – a CLO manager (An NAIC plan to second-guess bond ratings is ‘nonsensical’, insurers say, July 31)

“Once you establish that an asset class has both a very high correlation of loss and potential for near total loss, there’s no good capital charge to ensure the resilience of an insurer with a large exposure” – Aaron Sarfatti, Equitable (Equitable backs concentration charge on riskier ABS, October 3)

“The mutuals went and lobbied their regulatory compadres to round up pitchforks and torches to go on this mission against CLOs” – a senior executive at a private equity-backed insurer (How US insurers went to war over CLOs, November 15)

CRACKS IN THE BODYWORK

The resurgence of the US Treasury basis trade in the second half of the year raised concerns that a rapid unwinding of leveraged arbitrage positions could once again rattle the wider markets. 

While some dismissed the concerns as overblown, Risk.net reporting in October highlighted a potential weakness in the most recent iteration of the trade. 

Rising term repo rates meant hedge funds were almost exclusively reliant on overnight repo to fund the cash leg of the basis trade, raising the chances of another disorderly unwind if they are unable to roll the funding during a stress event – perhaps due to a lack of dealer capacity, or because a jump in the repo rate makes it uneconomic to do so.

Hedge funds defended the trade on the basis that it provides crucial liquidity to the market at a time of rising rates, volatility and issuance. Some pointed fingers at asset managers for creating the arbitrage by piling into Treasury futures, seemingly in an effort to profit from an inverted yield curve without lowering the effective duration of their funds. US mutual bond funds held more Treasury futures at the end of September – over $800 billion – than at any time in history, though a lack of transparency into funds’ risk exposures makes it difficult to say conclusively whether they were taking excessive risks.

In December, the SEC adopted its long-awaited clearing mandate for US Treasury cash and repo transactions. Once the mandate takes effect at the end of 2025, repo transactions will be subject to standard haircuts set by clearing houses. This is likely to reduce the amount of leverage available for basis trades. If the hedge funds are right, US Treasury market liquidity could also take a hit.     

“If you get some sort of hiccup that makes it very expensive for the funding on these trades to continue, there could start to be unwinds – and that could be quite ugly” – Gennadiy Goldberg, TD Securities (How repo roll risk could spoil US Treasury basis trade, October 3)

“The trade has very low profit margins to begin with, so increasing the price after a certain number of trades is likely to make further trades uneconomic” – Jennifer Han, Managed Funds Association (Margin failings raise concern over Treasury basis trade, November 27)

“We have been extending durations in the portfolios where we have discretion. We are not using leverage in any of our funds” – Matt Nest, State Street Global Advisors (The unknown risk on the flip side of the basis trade, December 19)

SYSTEM FAILURE

Cleared derivatives markets were disrupted for several days in February after a ransomware attack on Ion Group blocked access to its widely used trade processing and matching services. 

Clients were furious with the way the incident was handled. Ion shared limited information about the cyber attack – refusing to say how the hackers infiltrated its systems or what vulnerability they exploited – and missed its initial targets for restoring services. 

An investigation by Risk.net concluded the hackers likely gained access to Ion’s systems through a phishing attack and then exploited an unpatched vulnerability in its virtualisation servers. The article raised questions about Ion’s cyber security practices – including the adequacy of its user authentication and endpoint protection systems – as well as its business strategy of rapidly acquiring rivals, stitching their systems together and cutting costs to drive profits. 

Clients of Ion were not without fault, either. Many of them failed to classify it as a critical third-party vendor that needed close monitoring. Some argued that current regulations that put the onus on financial firms to monitor critical third parties were outdated. There were calls for more direct supervision of technology vendors, including cloud service providers. This may be one outcome of new EU rules on digital operational resilience, but regulators in the US seem unmoved. In December, the US Commodity Futures Trading Commission reiterated that it was the responsibility of regulated financial firms to identify, monitor and manage their critical third-party vendors for potential risks.

“Some banks have moved to manual clearing, whereas before they were reasonably automated because of Ion’s software” – a clearing executive at a European bank (Ion cyber outage continues as banks rely on workarounds, Feb 3)

“There are a lot of really angry people out there, up to and including the regulators. There was just a gaping black hole of information that left everyone absolutely gobsmacked” – a derivatives lawyer (Hacked off: banks demand answers after Ion cyber attack, March 23)

“Many firms that were onboarded [by] Ion didn’t use the highest level of scrutiny that they use for their most critical third-party vendors” – Todd Conklin, US Treasury Department (Ion wasn’t deemed a ‘critical’ vendor by most clients, March 24)

AI TEST FLIGHTS

Artificial intelligence went mainstream in 2023 with the public release of ChatGPT, a chatbot with an unnerving ability to complete all manner of menial tasks in a human-like fashion. Within two months, ChatGPT had more than 200 million users. But few among them were bankers. 

Several large banks – including Bank of America, Citi and JP Morgan – banned employees from using ChatGPT until they could address risk and compliance concerns. One of the main hurdles to adoption is explainability. Regulators expect banks to be able to explain the inner workings of their models – a near impossibility when dealing with generative AI models that have complex architectures and huge numbers of inputs. 

Deutsche Bank’s chief innovation officer told Risk.net in November that regulatory expectations may need to shift to observability – where banks must understand the inputs and outputs of models, but not necessarily their whole inner working – before generative AI can be widely used in financial services.

Meanwhile, adoption of more traditional forms of AI that follow set rules and well-understood computational processes continued apace, with use-cases ranging from anti-money laundering to derivatives hedging. Researchers at JP Morgan even began testing the use of quantum computers to train their machine learning-based hedging strategies, finding that this nearly doubled their effectiveness. Risk.net also reported on an AI model built by two hedge fund quants that has the capacity for abstraction and analogy, allowing it to learn more like a human. 

“By stressing the importance of using AI in fighting financial crime, the Dutch court paved the way for long-term positive change in the industry” – spokesperson for bunq (OK regulator? How AI became respectable for AML controls, March 7)

“Now we actually have a solution that gives us confidence that in the future this work will be usable in production” – Marco Pistoia, JP Morgan (JP Morgan is testing quantum deep hedging, April 3)

“We are still at an exploratory stage to ensure the technology aligns with our risk appetite and regulatory requirements. We want to use it in a way that we can explain the technology internally to ourselves and create real economic values for our clients” – head of AI at a global bank (Banks’ internal watchdogs bark back at ChatGPT, July 13)

“We need to transition to observability of models and work with regulators to address their concerns. Otherwise, generative AI and LLM will have limited adoption within the financial industry” – Gil Perez, Deutsche Bank (Generative AI is changing debate on explainability, says Deutsche, November 7)

“We’ve evolved to be able to process gigantic amounts of inputs without facing the curse of dimensionality. But the price we pay is cognitive fallacies, as has been demonstrated in the nascent field of quantum cognition” – Dario Villani, Qognitive (AI model uses quantum maths to learn like a human, December 4)