Next steps in transition from LIBOR

Speech by Edwin Schooling Latter, Director of Markets and Wholesale Policy at the FCA, delivered at the Risk.net LIBOR Summit, 2019.

edwin schooling latter speech.jpg

Speaker: Edwin Schooling Latter, Director of Markets and Wholesale Policy
Event: Risk.net LIBOR Summit, London​​
Delivered: 21 November 2019
Note: this is the speech as drafted and may differ from the delivered version

Highlights

  • Key next steps in reducing the risks from continued use of the LIBOR benchmark include ending use of LIBOR in new sterling loans from Q3 2020, and making it standard to quote based on SONIA in sterling swap markets.
  • The FCA’s Director of Markets and Wholesale Policy describes how LIBOR could cease or fail the Benchmarks Regulation ‘representativeness’ test at end-2021, and how robust contractual fall back triggers can protect market participants from risks in both scenarios.

The need for transition away from LIBOR has been well rehearsed. The end-2021 date by which market participants need to be ready for life without LIBOR will be familiar to most, if not all in the room. Today, therefore, I would like to focus on some key next steps at this point in the transition from LIBOR to near Risk Free Rates (RFRs).

Progress made

I will pause only briefly to note what has already been achieved. We now have liquid markets for swaps and futures based on SONIA – the market’s chosen sterling Risk Free Rate. SONIA is now the norm in new issuance of floating rate sterling bonds and securitisations. There have been a number of other important new landmark achievements in the past weeks. These include the successful conversion to SONIA of £4.2 billion in previously LIBOR-referencing securities – with some of these consent solicitations achieving 100% investor agreement – as well as new firsts for the conversion to SONIA of a LIBOR loan, and a first SONIA swaption. The sterling RFR Working Group is working to publish a document that will share more widely some of the lessons learned in these conversions of ‘legacy’ LIBOR instruments.

There is no room whatsoever for complacency. There is much more to do. This week ISDA recorded the 1000th SOFR-swap transaction in the year to date, but SOFR swaps are still a very small part of the US dollar swaps market. And while we have seen some landmark firsts, beyond a handful of pioneering examples, we are yet to see loan markets move from LIBOR to the RFRs.

Key next steps in sterling swaps and loan markets

New sterling LIBOR public bond issuance appears to have ceased. But in sterling, as well as in US dollar, Japanese yen and Swiss francs, significant volumes of new LIBOR swaps maturing after end-2021 are still being struck. In sterling IRS (interest rate swap) markets, we will be encouraging market makers to make SONIA the market convention from Q1 2020. That does not, at this stage, mean no more sterling LIBOR swap transactions for those who have a particular reason to prefer LIBOR, but it does mean making it standard to quote and offer swaps based on SONIA rather than LIBOR. As infrastructure and liquidity to support SONIA swaps are already in place, this should be achievable with relatively little cost.

In sterling IRS (interest rate swap) markets, we will be encouraging market makers to make SONIA the market convention from Q1 2020.

In the loan market, the task ahead is a bigger one. Looking at UK markets, we can draw some comfort that use of LIBOR is rare in new mortgages. It is getting even rarer as the handful of smaller lenders still using LIBOR have completed, or near completion of their transition programmes. Alternatives, including both fixed rates and alternative variable rates are already widely used. One of these is Bank rate (which past data show SONIA to have tracked closely).

Bank rate is also used for some variable-rate SME lending. But LIBOR continues to be common in corporate lending, including in syndicated loans. The sterling RFR Working Group has set a target of Q3 2020 to stop new lending using LIBOR. This will involve significant infrastructure and documentation preparation, customer communication and staff training exercises for some banks. The main IT system providers in this area have described to the Working Group their planned new product releases. One of the largest providers plans to make available its SONIA loans product on 29 November.

The sterling RFR Working Group has set a target of Q3 2020 to stop new lending using LIBOR.​​​​​

One message we have been keen to communicate is that delaying transition of new business away from LIBOR until the production of forward-looking term rates based on SONIA may not be the best approach. The loan market, like some others, has become accustomed to the forward-looking LIBOR benchmark. In those other markets, notably bonds and securitisations, we have also in the past heard arguments that forward-looking term rates are needed. But those arguments have largely evaporated in those markets, in the face of decisive shifts to overnight rates compounded in arrears. For similar reasons, including hedging costs and effectiveness, many expect to see this happen in large parts of lending markets too, particularly the wholesale end of those markets.

The often-heard argument that borrowers who value certainty of payments need forward-looking term rates often doesn’t hold much water. For those whose priority is payment certainty, fixed rates may be best. Other variable rate options, for example SONIA observed over an earlier period, can also provide payment certainty before the end of the interest period. Moreover, the benefit of compounding may mean such rates are less volatile than a forward-looking term rate polled from market transactions on a single day may prove to be.

The place for forward-looking SONIA term rates is the central issue in a forthcoming sterling RFR Working Group publication on term rate use cases. The Group expects overnight compounded rates to be the norm in derivatives, securities and wholesale loans. In some more retail markets, where simplicity or payment certainty is the key factor, fixed rates, or Bank rate may be preferred. There may, however, be some places where a forward-looking term rate is a good fit with business needs. Moreover, term rates may be particularly useful in calculating fair replacement rates for legacy LIBOR contracts that cannot be amended to work on overnight rates compounded in arrears.

The sterling RFR Working Group and its members are also working closely with counterparts in the United States and elsewhere to try and agree the final elements of common conventions that can best support an international syndicated loans market.

The conduct risks of striking new LIBOR-referencing transactions that endure beyond end-2021 are rising.

Further progress on transition in loans and swaps markets is the key task in the year ahead. The risks of continuing to rely on contracts that reference LIBOR beyond end-2021 are rising. The prudential risks are rising given uncertainty about the existence of the rate and how the properties of the rate (eg its volatility) or the depth of liquidity in LIBOR instruments will change as the end of LIBOR approaches. The conduct risks of striking new LIBOR-referencing transactions that endure beyond end-2021 are also rising. It is no longer credible for any regulated firm to claim it did not know LIBOR might not survive this date. This week we published some initial answers to some of the most commonly raised conduct questions. We will continue to engage with market participants on key questions as transition progresses.

Contractual fall backs in global derivatives markets

Let me now turn to more global markets, and issues relevant to all the LIBOR currencies. Moving away from LIBOR altogether is likely to be the best way of avoiding LIBOR risks.  But another important safety belt is ensuring that contracts that do rely on LIBOR have clear fall backs that deal effectively with the prospective end of LIBOR.

This is true across all markets, but is perhaps of particular systemic importance in the swaps market, given the sheer size of exposures. The International Swaps and Derivatives Association (ISDA) last week made a further announcement on how best to calculate a fair replacement value for LIBOR in contracts agreed using ISDA documentation. The market has coalesced around a methodology using the median of the LIBOR-RFR spread over a 5-year look back period. That is an important step forward.

These fallback provisions would need, of course, to be triggered when publication of LIBOR ceases.

But the FCA and other authorities, working through the Financial Stability’s Board Official Sector Steering Group (OSSG), have also been keen to ensure market participants are aware of the need to be prepared for other possible variations on how LIBOR will end. The global derivatives industry currently faces an important decision point in this regard.

Loss of representativeness

At this stage we do not know precisely how the LIBOR ‘end-game’ will play out. It may be that for all 35 LIBOR currency-tenor pairs a final cessation date can be announced comfortably in advance, and transition away from each of these rates can be substantially completed in an orderly manner before then. This has been done in the past for other LIBOR currency-tenor pairs.

Many will argue this is the best possible way for LIBOR to end. Our own efforts to encourage transition before end-2021 are intended to make this possible. We have also warned that firms must not assume LIBOR will continue beyond end-2021 even if transition is not substantially complete. I repeat that warning today. But another possible variation of the end-game is that LIBOR’s final cessation is preceded by a period in which it is still published but no longer passes the key regulatory test of being capable of measuring the underlying market or economic reality. In other words the benchmark is no longer ‘representative’. The most obvious reason for failing this test would be the departure of panel banks.

This representativeness requirement is set out in the European Benchmarks Regulation, which sets the legal framework for critical benchmarks such as LIBOR and EURIBOR. The Regulation also sets out substantial powers for competent authorities such as the FCA to take action if representativeness is at risk. These include compelling continued panel bank contributions for a period. Currently the maximum length of that period is 24 months.

Back in 2017 we faced the uncomfortable reality that despite concerns of panel bank departures, it did not appear feasible to complete a transition away from LIBOR in just 2 years. Hence, the FCA agreed with all 20 LIBOR panel banks that they would continue voluntarily to contribute for a further 4 and a half years. Our aim was to push back the risk of cessation or loss of representativeness until end-2021, and secure the window necessary for an orderly transition. I am pleased to say that none of those panel banks has withdrawn that agreement. This should provide us continued stability for the next 2 years.

But what about end-2021? We have already said publicly that we do not expect all the panel banks to be willing to continue to contribute after that date. We do expect departures from the panels. For each panel bank that leaves, the discomfort for the remaining panel banks increases. We therefore have to plan on the basis that continued production of panel bank LIBOR will not be possible. But what if, despite other panel bank departures, some are willing to remain even though they are too few in number, or too inactive in the underlying market, for a benchmark produced from their contributions to be capable of being representative?

This might be in the context of loud calls for more time to complete transition from those market participants who failed to heed our warnings about being prepared for LIBOR’s cessation at end-2021. To these stakeholders, a period of continued publication of even an unrepresentative LIBOR, prior to final cessation, might appear the least bad option.

What are the consequences of LIBOR becoming unrepresentative?

There are at least two things I can say with confidence.

First, reaching the point of unrepresentativeness in this scenario would be an irreversible step towards the end of panel bank LIBOR. The FCA does not plan to compel banks to join or rejoin LIBOR panels after end-2021. We have already made our agreement with the panel banks. In any case, the more obvious time to use a panel bank compulsion tool would be before the rate became unrepresentative rather than afterwards.

Reaching the point of unrepresentativeness [...] would be an irreversible step towards the end of panel bank LIBOR.

While the Benchmarks Regulation also gives us powers to require the administrator of LIBOR to change the methodology or rules of the benchmark, we see no way of doing so which would restore LIBOR’s representativeness. As in ISDA’s careful work on calculating a fair replacement rate for LIBOR, we too cannot see a robust way of continuing to calculate a dynamic credit spread for every business day.

Second, while a short period of continued publication of an unrepresentative panel bank LIBOR might pre-empt or delay greater disruption for some, it would not be a comfortable place for anyone.

It would not be comfortable for those still using LIBOR – the behaviour of the rate is difficult to predict. Moreover, with the forthcoming demise of LIBOR clear from this point even to those who had previously resisted this message, and with the forbidding ‘unrepresentative’ label attached to it, appetite to buy, or even to continue to hold, let alone to issue new LIBOR-referencing instruments is likely to be very significantly diminished.

Even if some panel banks were willing to continue, it would not be comfortable for them – I imagine they would want to keep the period in which they continued submitting as short as they could.

It will not be comfortable for the administrator of the rate.

Those are all reasons why it would be unwise to assume a period in which an unrepresentative rate is published. Certainly relying on this scenario is not a safe alternative to transition. It just means that transition would have to be completed at greater haste and in probably less favourable conditions than now.

With liquidity in LIBOR-linked instruments severely diminished, with firms reluctant to take on new LIBOR obligations or assets, and, indeed with EU-supervised firms potentially prohibited from entering into at least new LIBOR transactions, it would not be viable for central counterparties to continue to clear LIBOR swaps and futures. Risk management of LIBOR derivatives portfolios would be impaired. Sourcing hedges for and managing an auction of a defaulter’s LIBOR position might be all but impossible.

Thus, for good reason, the major CCPs clearing LIBOR swaps have been clear that they would view loss of representativeness as a trigger to using the discretion in their own rules to shift all of their cleared LIBOR portfolios to the new RFRs adjusted by the spreads identified by ISDA.

How should market participants prepare for this?

Let me repeat that the best way to avoid LIBOR-related risks is to move off LIBOR altogether. Lenders and borrowers using SONIA or other variable rates, or fixed rates, will not face LIBOR risks. Banks and asset managers with SONIA or SOFR swaps and futures will not face these risks.

But for those not quite ready to make this move to SONIA, SOFR or equivalents yet, contractual fall backs offer a safety net.

The Alternative Reference Rates Committee (ARRC), which is leading transition from LIBOR to SOFR in the United States, consulted on and has already subsequently published its recommendation on including in cash market contracts fall backs with a so-called ‘pre-cessation trigger’. This is activated if and when the FCA finds LIBOR unrepresentative.

But given the scale of exposures, at least in notional terms, and given the need to be able to enter new transactions to risk manage outstanding books, the importance of considering pre-cessation triggers is perhaps even greater in swaps markets.

ISDA published an important consultation on pre-cessation triggers in May this year. A substantial majority of respondents to that consultation said they would not want to continue to reference LIBOR in new or existing derivatives contracts once its supervisor – the FCA – has said it is no longer representative of the underlying market. We think they are right. 

There was, however, some variation of views on how best to implement that trigger.

Some consultation respondents preferred an optional approach. But a larger number of respondents saw risk management and data challenges, as well as off-putting complexity in the optionality approach. When we have discussed these issues among authorities, we have found those reservations about the optionality approach compelling.

Although there were also some respondents who expressed reservations about including both triggers as standard, some of the main reservations are relatively easily addressed. At least one suggested that further clarification from the authorities would be helpful. I will try and respond to that now.

One reservation was that calculations of spread adjustments made upon a cessation event would differ from those on a pre-cessation event. If many contracts had included the pre-cessation triggers while a significant volume of others had not, and two different spread calculations applied, that could complicate pricing and risk management. That pitfall can, however, be avoided by uniform inclusion of the pre-cessation trigger. As with the ARRC’s language, there would then be only one spread-adjustment, based on the time of whichever trigger event (pre-cessation or cessation) occurred earliest. In other words, there would be the same spread adjustment for all covered ISDA contracts.

A second reservation was whether the loss of representativeness that activated a pre-cessation trigger would be a clear and unambiguous event. Respondents rightly attach importance to avoiding ambiguity. I hope they can draw comfort that the requirements on when authorities must make the representativeness test for a critical benchmark such as LIBOR are set out clearly in law – ie in the Benchmarks Regulation. Similarly, I hope they draw some comfort that the authority making and declaring the results of such a test – a securities regulator, and for LIBOR, the FCA – is well placed to make market sensitive announcements in an appropriate and unambiguous way to the whole market at the same time.

A third reservation related to the prospect, at least for a short period, of having two rates side by side – ie the fallback rate, based on RFRs plus fixed spread, could be compared with the unrepresentative LIBOR rate while the latter is still being published. Counterparties could calculate whether they would have been better off under one than the other. That comparison could indeed be made. But as others point out, firms would be choosing to use the fall back. Moreover, one of the two rates – LIBOR, could not be said to be an accurate ongoing representation of the market. It would have been officially found unrepresentative.  

The fourth reservation related to hedging relationships. Some derivatives hedge other exposures in cash markets, say on bonds or loans. Many of those cash market instruments, especially older contracts, will not have pre-cessation triggers. If the derivative moves to fall back rates on loss of representativeness, but the bond does not, basis risk will reduce the effectiveness of the hedge.

Of course these older cash market contracts which do not have appropriate triggers anyway pose a hedging problem. In many cases they will revert to fixed rates when LIBOR disappears. Derivatives covered by new LIBOR cessation triggers, or even existing ISDA language, will not. Deciding not to adopt a pre-cessation trigger in derivatives will not solve this hedging problem. It will just delay it, and potentially not for long. Addressing the fundamental flaw in the cash product is likely to be a more effective solution.

Indeed, the volume and proportion of cash market instruments relying on LIBOR and without pre-cessation triggers may change significantly between now and end-2021 when these triggers would be needed. In the UK, real inroads into the stock of LIBOR bonds and securitisations maturing post end-2021 are now being made – by converting them to SONIA. We have seen the first LIBOR to SONIA loan conversion too. 

In the United States, where bond conversions are harder because they often require unanimous consent, the ARRC has been exploring whether there is a legislative option to build pre-cessation triggers into these contracts.

The largest subset […] thought the pre-cessation trigger should be an integral, not an optional, part of standard language: their arguments seem persuasive.

Moreover, when it comes to minimising basis risk in hedges, it is also important to look at the extensive hedges between uncleared and cleared derivatives. Given the prospective CCP actions when LIBOR is found unrepresentative, these hedges are likely to be best preserved if pre-cessation triggers are included in uncleared contracts too.

Perhaps for these reasons, the largest subset of those who wanted a pre-cessation trigger thought it should be an integral, ie not an optional, part of the standard language. The arguments made by this group seem persuasive. Hence, last week, the OSSG co-chairs wrote to ISDA to encourage them to consult on taking forward this approach.

Conclusion

Progress on transition is continuing. We welcome the new landmarks being achieved. These include firms pioneering new transactions, and system providers rolling out upgrades that work on the new RFRs. They include ISDA’s progress on robust fallbacks for derivative contracts. But the months ahead will be critical ones in taking the next steps to a successful and smooth transition from LIBOR by end-2021.