Basel’s cherry-picking toughens IRRBB shock scenarios

European banks want higher outlier thresholds to offset higher confidence level in new standard

Ever since the Basel Committee on Banking Supervision’s post-2008 masterplan was completed a decade later, international regulators have tried to limit themselves to relatively small updates. But sometimes even a small update can quickly turn into a larger remodel. Bank treasurers are worried a supposedly minor redesign to supervisory tools used for gauging the interest rate risk in deposit and loan books could turn into a major headache. 

The latest proposed modification affects a series of interest rate shock scenarios, and it involves upping a confidence level that is key for determining the severity of those shocks. Banks say the Basel Committee has essentially cherry-picked the parameters used for the calibration to ensure shock scenarios increase in size. This reverse-engineered calibration is viewed as arbitrary, rather than risk-based.

“They don’t substantiate why they would change the confidence level, they only do it because with the previous confidence interval the shocks would be too low, which is like goal-seeking,” says a head of asset and liability management (ALM) at a European lender.

On December 12, the Basel Committee released a consultation paper to adjust its outlier test for identifying banks whose deposit and loan books are most at risk from changes in interest rates. Banks must calculate the impact of six different interest rate scenarios on the economic value of equity in their banking books, as well as the impact of two of those six scenarios on their net interest income. Those found to be most at risk from interest rate changes face heightened scrutiny from their supervisors, along with the possibility of higher capital requirements. 

Following the dramatic rise in interest rates across western economies over the past two years, the Basel Committee announced in December 2022 it would review the shock scenarios in the outlier test. Most interest-rate shocks set for European currencies – including euros, pound sterling, Swedish krona and Swiss franc – increase by 50 or 100 basis points under the proposed changes.

Increasing shock sizes can make the balancing act between the two measures – economic value of equity (EVE) and net interest income (NII) – harder to maintain because they have an inverse relationship. Often, EVE is most painful when rates increase, as longer-dated fixed-rate assets lose value. When rates decrease, NII is most painful, as the possible interest banks can earn on assets falls, particularly if a bank has variable-rate or short-dated fixed-rate assets. In theory, actions to reduce the decline for one metric will lead to a larger decline for the other metric. 

The US has not yet implemented the Basel interest rate risk on the banking book (IRRBB) framework in full. European lawmakers have, and banks in Europe are now urging a rethink of the threshold for banks to be judged as outliers, in order to reflect the non-linear implications of the Basel Committee’s recalibration. 

“The supervisory outlier test thresholds would have to be recalibrated because having higher shocks will be another tightening effect and you’d produce more outliers than now,” says Christian Saß, an associate director at the Association of German Banks.

In the meantime, European banks are revisiting their approach to hedging IRRBB, in order to prepare for the possible adoption of the modified Basel standards in the European Union (see box: Finding the right hedge). 

The outlier test for EVE is embedded in level one legislation in the EU, so it can only be changed by lawmakers. By contrast, the European Banking Authority can alter the threshold for the NII test at its own discretion. A spokesperson for the EBA says the agency will review the threshold once the new Basel standard is finalised, based “as usual” on a quantitative impact assessment of the new shock scenarios.

The Basel Committee did not respond in time for the publication of this article. Its consultation on the IRRBB updates closes on March 28, 2024 and, although the committee has not specified when it intends to produce a final standard, the update is part of a two-year work programme intended to be completed by the end of 2024.

The interest derivatives market already started in the early 2000s to account for the normal distribution characteristics of interest rates by recalibrating shocks in the wake of the recent low interest rate period

Andreas Bohn, McKinsey

Computer says ‘no’

The Basel Committee’s initial intention had been only to expand the data series underpinning the calibration, to include changes to interest rate curves between 2016 (when the original standard was finalised) and 2022. However, regulators found a flaw in their methodology used to calibrate shocks in the original Basel standard, which forced them to consider a more far-reaching overhaul. 

The previous methodology had relied on relative changes in interest rates and set the shock at a 99% confidence level of those observations, which means no more than 1% of interest rate changes should prove more severe. The problem with this methodology based on relative increases is that when the starting rate is very low (as it was for most developed economies in 2016), a very small increase in interest rates quickly propels the move beyond the 99% confidence level.  To overcome this defect, the Basel Committee decided in the updated methodology to base its calculation on absolute changes in interest rate levels instead. 

“The interest derivatives market already started in the early 2000s to account for the normal distribution characteristics of interest rates by recalibrating shocks in the wake of the recent low interest rate period,” says Andreas Bohn, a Frankfurt-based partner at consultancy McKinsey and former ALM manager at Barclays and Deutsche Bank. “The normal distribution pattern has now been adapted by the Basel Committee.”

Andreas Bohn

However, this switch from relative to absolute increases to derive the confidence level had an unintended side-effect. Out of a total of 63 scenarios across all currencies covered in the revised IRRBB standard (including three new scenarios added for the Argentine peso), the Basel Committee found 15 of the shocks became smaller, while 14 became larger. Shocks applied to assets and liabilities linked to Canadian, Indian, Mexican, South African, South Korean and US rates all decreased, while European rates were either unchanged or increased by 50bp. 

“But that’s not what they wanted, so they proposed to change the confidence interval from 99% to 99.9%,” says a senior treasurer at a European bank, who describes the consultation as “cheeky”. They continue: “If you have 1,000 observations, whether you cut off at 990 or 999 makes a difference.”

The confidence level represents how much of the probability distribution curve is covered by the shock that banks are asked to assume. A higher level means the scenario will represent a more severe shock that is less likely to occur in a given period.

The result of the higher confidence level is 26 scenarios increasing in size compared with the previous shocks set under the old methodology, and only eight shocks shrinking. When compared with the shocks generated by the new dataset and methodology at the lower confidence level, 36 of the scenarios are larger under the higher level. 

“They wanted to ensure a sufficient level of conservatism, but I didn’t read any actual arguments based on empirical data whatsoever,” says Saß at the Association of German Banks. “It doesn’t seem justified to change that percentile in the calibration.” 

 

Long debates

Other sources are more sympathetic about part of the Basel Committee’s justification for altering the confidence level – namely, the recent volatility in interest rates. A head of ALM at a third European bank says it is possible to make the case for the alteration, because the shocks will be “more comprehensive” in terms of how many risk scenarios they cover.

A senior market risk manager at a fourth European bank points out that the change in 10-year Euribor swap rates over six months from mid-December 2021 to mid-June 2022 was almost the same size as the Basel Committee’s revised parallel shift scenario – whereby the whole curve moves by the same amount – for euro rates. The actual size of the yield curve change during that period was slightly smaller than the IRRBB requirement of 250bp, but the Basel Committee rounds shocks to the nearest 50bp in any case.

A theoretical discussion about defining the right confidence level could go on for a long time, says the market risk manager but, given the actual rise in rates during 2022, in practice “it doesn’t seem outrageous to prescribe 250bp” as the shock scenario.

 

The Basel consultation paper states the confidence level should increase “to maintain sufficient conservatism” and also cites the “accommodative methodological change”. The paper doesn’t provide a direct numerical comparison between the old methodology and the new methodology at the same confidence level and on the same dataset, so it is impossible for bankers to know how far the methodological change would have eased the shock sizes without the alteration to the confidence level.

The paper does provide analysis showing the effect of a higher confidence level and the change in the dataset. Expanding the dataset under the new methodology at the 99.9 percentile leads to 10 shocks becoming larger than under the old dataset, while two shrink. 

Moving thresholds

Upping the size of shocks used in the outlier test will push banks closer, or even above, thresholds set by regulators for spotting outliers. The head of ALM at the first European lender says supervisors should take this into account when interpreting the results of a revamped outlier test with the higher confidence level.

“The correct conclusion would be: ‘We changed the confidence interval, we changed the dataset, so we are using more severe shocks. And in a severe shock, this bank would be more sensitive as a result’,” says the head of ALM. “It’s not that the balance sheet changes overnight. If you make [the test] stricter, obviously the outcome will be higher.”

The senior treasurer at the second European bank says the thresholds should be altered to ensure the test produces the same number of outliers as before. The Basel Committee leaves the EVE threshold unchanged from the original standard. 

The Basel Committee only set the threshold for the scenarios gauging EVE risk. Any bank facing a decline in their EVE of more than 15% of their total Tier 1 capital in any of the six scenarios is judged to be an outlier. For the test gauging NII risk, the Basel Committee left individual jurisdictions free to determine their own thresholds, so long as they are as severe as the one for the EVE scenarios. 

In the EU, the outlier threshold for the EVE scenarios decreased from a transitional 20% to 15% of Tier 1 capital in June 2021 under the capital requirements directive. For the NII scenarios, the threshold for catching outliers is set at a decline equivalent to 5% of Tier 1 capital, or higher. 

Altering the thresholds doesn’t make the whole exercise of updating shock sizes pointless, because the impact of shocks on a banking book’s value is not exactly linear. A larger shock should lead to a larger decline, but the exercise of optionalities built into products changes the more interest rates move.

“The impact of a 100bp shock on EVE is not necessarily half of the impact of a 200bp shock,” says Berend Ritzema, an Amsterdam-based manager in Deloitte’s risk advisory practice. “There could be nonlinear effects such as convexity, that are more pronounced when the shock that is applied is higher.” 

He gives the example of mortgages with an early repayment option. The level of prepayments depends on the level and movement of interest rates. When a bank uses linear hedging of its mortgage portfolio, such convexity effects remain unhedged, and more extreme interest rate shocks can lead to relatively bigger EVE losses. 

“[It is] an exponential effect so to say, where losses can become exponentially bigger if the shock that is applied is larger,” says Ritzema.

Finding the right hedge

Faced with uncertainty about if and when the EU will adopt the new, tougher IRRBB outlier tests devised by the Basel Committee in December 2023, European banks are examining how they should change their hedging strategy. The aim would be to reduce the decline in their NII metric from a fall in rates, while having as little impact as possible on their EVE sensitivity. 

To that end, banks have been eyeing interest rate derivatives with embedded optionality or floors, or fixed-receiver swaps that last for two years. The former would only activate in the circumstance of rates falling. The latter is cheaper to enter than an options trade, and while two years is long enough to protect NII in the one-year downward shock scenario, it is also short enough not to cause too much damage to EVE in the one-year upward shock scenario. 

“The discrepancy between [NII and EVE test results] will go up by 25% to 50% depending on what currency you are holding,” says a senior treasurer at a second bank. “It will simply mean whatever you thought you need to do for NII, you might need to do a bit more.”

The result would be higher hedging costs for banks. That’s why, if the EU implements the new Basel shock scenarios, there are calls for the outlier thresholds to be increased as well, to ensure the same number of banks are found to be outliers. This would ease the pressure by reducing the level of hedging banks would need to stay within the regulatory threshold.

Editing by Philip Alexander