LIBOR transition and contractual fallbacks

Speaker: Edwin Schooling Latter, Director of Markets and Wholesale Policy
Location: International Swaps and Derivatives Association (ISDA) Annual Legal Forum, London
Delivered on: 28 January 2019

Highlights:

• On a monthly basis, cleared notional in Sterling Overnight Index Average (SONIA) swaps is now higher than that for sterling London Inter-bank Offered Rate (LIBOR).
• The best and smoothest transition from LIBOR will be one in which contracts that reference LIBOR are replaced or amended before fallback provisions are triggered.
• Market participants should not rely on the availability of an option to use Libor for legacy contracts.

Note: this is the speech as drafted and may differ from delivered version.

The subject I will address today is what the final steps in transition away from the London Inter-bank Offered Rate (LIBOR) might look like. This is, of course, directly relevant to many of the key legal documents that underpin the global derivatives market.

There is now wide recognition that LIBOR will come to an end. Thanks to the agreement reached with 20 panel banks to continue submitting until end 2021, LIBOR is not expected to cease before that point. But, when I spoke at the International Swaps and Derivatives Association’s (ISDA’s) annual Europe conference in September last year, on this same stage, an audience poll found just over 50% thought LIBOR would stop before end-2022. Today, I would like to explore in a bit more detail not whether LIBOR will end, but how it will end.

This has important implications for contractual design. It is relevant in particular to how ’fallback‘ language in outstanding contracts that continue to reference LIBOR will and should work. Understanding the way in which the end of LIBOR will play out is key to choosing the right trigger point for moving to a new or replacement ‘fallback‘ rate.  

But before I get to that detail, I would like to give an update on the progress made on transition.

The share of cleared sterling swaps referencing the Sterling Overnight Index Average (SONIA) grew to 19% in the second half of 2018, from 11% in the first half of 2016. That’s on a duration-adjusted basis. In notional terms it is a far higher share. In fact, on a monthly basis, cleared notional in SONIA swaps is now higher than that for sterling LIBOR. London Clearing House (LCH) data also show growth has begun in use of swaps referencing the near risk-free-rates (RFRs) identified to replace yen, Swiss franc and US dollar LIBOR rates. On the yen side, LCH has cleared US$ 78 billion in TONA swaps in 2019 to date, with US$ 657 billion outstanding. In Swiss franc, it has cleared US$ 21.7 billion in Swiss Average Rate Overnight (SARON) swaps so far in January, with US$ 79 billion outstanding. Of the relatively small US$ 17 billion Secured Overnight Financing Rate (SOFR) notional outstanding, US$ 7 billion was cleared in the past 2 weeks. The smaller numbers in SARON and SOFR reflect that there was not the same advantage of a previously existing market in these swaps. But even in SOFR, LCH has registered the highest levels of activity to date on a number of measures since the start of 2019.  

Turning to futures, daily trading volume in SONIA futures, and SOFR futures, both averaged 15,000 contracts per day in December.

This progress is not just in derivatives markets. In bond markets, where in sterling too we had, in effect, to start from a zero base, we saw a total of £6.9 billion in SONIA referencing sterling floating rate notes (FRNs) issued between June and December 2018. Already in 2019, by Wednesday last week, we had seen a further £7.2 billion. If past weeks are a guide, new sterling FRN issuance is only SONIA-based now. We have also seen US$ 46.3 billion in SOFR-referencing US dollar FRNs since mid-2018.

Also last year the market made a very significant step forward on the fundamentally important task of calculating a fallback rate that could replace LIBOR in outstanding contracts. It is ISDA that took the lead on this, at the request of the Financial Stability Board’s Official Sector Steering Group (OSSG), which is the international forum where authorities from across the globe have co-ordinated their work on interest rate reform. Through its consultation, ISDA has successfully established a wide consensus across different types of market participant, and across jurisdictions, on an appropriate, and appropriately fair, way of calculating a fallback rate that could substitute for sterling, yen and Swiss franc LIBOR in contracts.

This fallback rate relies, sensibly in our view, on the RFRs identified for these 3 currencies (ie SONIA, TONA and SARON). ISDA’s consultation proposals[1] attracted wide consensus that the term element of LIBOR should be replicated by compounding of the observed overnight rates at the end of the relevant term.

Authorities discussed this at the end-November meeting of the OSSG. The OSSG welcomed this choice.

There was also wide market consensus that the best way of calculating a spread to reflect term credit premia is to converge to an average of previously observed credit premia. This will result, in the end, in a fixed credit spread. That’s understandable given the difficulty of measuring these bank borrowing costs in a dynamic, forward looking way. Again, the OSSG welcomed the choice that market participants have made.

There are consultations yet to come on US dollar LIBOR, and with regards to euro LIBOR and Euribor. We shall see what those consultations reveal, but the same logic may well lead to similar results.

Transition from LIBOR

Let me be clear, however. We think that the best and smoothest transition from LIBOR will be one in which contracts that reference LIBOR are replaced or amended before fallback provisions are triggered.

It will likely be in market participants’ best interests to have moved away from LIBOR contracts to ones based on the RFRs before liquidity in LIBOR-referencing contracts has significantly declined. In the early days of transition there was the question of who would have the boldness and agility to be among the first to move away from LIBOR to initially less liquid alternatives. When we approach the end of that transition, we may instead find reluctance to be the last to remain on LIBOR. Many, I hope, are therefore likely to close out or convert their LIBOR contracts before fallbacks are needed. 

But ISDA’s work on these fallbacks significantly reduces the risks of widespread disorder in derivatives markets when LIBOR does end. We also welcome the confirmation from LCH and Chicago Mercantile Exchange (CME)[2] that they intend to adopt the same fallback rates. There is clearly merit in a consistent approach across contracts between which there are hedging relationships.

I would like to thank ISDA for their continuing work in this regard. Wide adoption of the new fallbacks will help reduce the risk of market participants finding themselves in disagreement or costly dispute on their rights and obligations attached to LIBOR-referencing contracts, or with positions split across multiple different fallback arrangements. For new transactions, adoption will be by way of the ISDA definitions for interest rate derivatives. For legacy contracts it will be achieved by signing the protocol implementing this fallback arrangement.

But the phrase I used a few moments ago – ’when LIBOR ends‘ – raises a number of questions about precisely how LIBOR will end.

One scenario is that Intercontinental Exchange (ICE) Benchmark Administration, or IBA, ’the administrator‘ of LIBOR, announces in advance that it will cease the publication of particular currency-tenor combinations. Such a process has been followed before. For example, for the now discontinued LIBOR reference rates once published for various other currencies. By way of another example, publication of the 2-month Yen Tokyo Interbank Offered Rate (TIBOR) rate will cease on 1 April this year, having been announced in February 2017. But cessation is easier for currency-tenor pairs that are little used. For the most commonly-used LIBOR rates, there may be loud demands to continue publication from those who were slow, and in some cases, unable, to transition away from LIBOR in their outstanding contracts. That demand for continued publication may be increasingly difficult to satisfy after 2021 as the number of panel banks willing to contribute to determining the rate declines. With every bank that leaves the LIBOR panels, the problem of capturing enough transactions to underpin robust calculation of the rate becomes more severe.

At this point, the requirements of the European Benchmark Regulation become relevant. Amongst these requirements, is a clear and unambiguous requirement for not only the administrator, but also for the supervisor of the benchmark administrator to assess the capability of a critical benchmark to be representative of an underlying market and economic reality. In the case of LIBOR, the supervisor is of course the FCA. The FCA is required to make this assessment of representativeness each time a supervised contributor – ie a panel bank – announces that it intends to stop submitting data.

So, it is entirely plausible that the end-game for LIBOR will include an assessment by the FCA that one or more panels have shrunk so significantly in terms of number of banks or the market share of the banks remaining, that it no longer considers the relevant rate capable of being representative. The FCA would announce such a view if and when it is reached. Market participants would then need to consider the many potential negative ramifications of using a rate when its regulator had found it not to be reliably representative of the underlying market.

It seems sensible, then, to consider this scenario when choosing the design of fallback triggers.

Cash market participants have already noted the benefits of a fallback trigger that would convert contracts at the point LIBOR rates no longer satisfied the representativeness test. In response to the consultation launched by the Alternative Reference Rates Committee in the United States in September 2018, there was strong support for such a representativeness trigger. I would envisage that major central counterparties (CCPs) would have serious concerns about having large books of contracts resting on an unrepresentative rate. LCH, for example, sensibly has provisions to change the reference rate in contracts it clears if the existing rate is no longer sufficiently robust and no longer fit for purpose. A regulatory finding and announcement that LIBOR was no longer representative would surely be relevant to those provisions. This sensibly helps CCPs to avoid a situation in which they are trying to manage a legacy portfolio based on a rate in which there is no liquid market for new contracts (and possibly even a prohibition, at least for EU regulated firms, in entering such contracts).

Market participants have said repeatedly that to avoid basis risks we should aim for triggers in non-cleared derivatives that align with triggers in cleared derivatives, and with triggers in cash markets too. This is because, of course, of the hedging relationships between these contracts. The OSSG at the end of last year heard very strong calls from chairs of some of the national working groups leading work on transition from LIBOR to new RFRs that we should aim for a common representativeness trigger across all these contract types. The OSSG agreed that this had merits, where relevant regulations provide a foundation for such a trigger.

The FCA would not come to a conclusion of unrepresentativeness prematurely. We authorised IBA in April 2018. Market participants may assume from that decision, that, notwithstanding issues around the small number of transactions underpinning LIBOR – issues that do have a bearing on representativeness – we did not consider these so severe that the benchmark did not satisfy the requirements of the Regulation. The FCA would be very clear in any future announcement about the precise date at which a particular LIBOR rate would no longer be capable of being representative – for example upon a combination of panel bank departures taking effect, or following the last in a sequence of such departures. The actual date would be fact, event and analysis dependent – so not necessarily knowable far in advance – but also unlikely to be a complete surprise.

The departure of panel banks may be relatively gradual. This might allow us to communicate that while we did not yet consider the representativeness test to be failed, the prospects of this, for example in the event of further panel bank departures, were increasing. I do not think this would go as far as setting a precise minimum number of panel banks. The size of panels is undoubtedly relevant, but so too are other factors such as which banks remain, and, in particular, how active they are in the market. But, we would expect to get some notice before a panel bank departed – and, if absolutely necessary, the Benchmark Regulation means they can be required to continue to submit for up to 4 weeks before they depart. That may well mean that, the point at which non-representativeness is reached could be announced at the least some weeks in advance, rather than materialise overnight or intraday.

Contractual reliance on LIBOR

There is a powerful logic to avoiding contractual reliance on a benchmark that is no longer representative of an underlying market, at least for those market participants that can avoid that reliance. That’s one clear reason to consider including a representativeness trigger in contractual fallbacks. With very small numbers of panel banks, and the disproportionate impact of individual transactions on the published rate that this would lead to, the properties of the rate – its level and its volatility – could also deviate from previous expectations.

In some scenarios, the finding that the rate is or will no longer be representative upon departure of panel banks may lead rapidly to cessation of publication. In this case, the timing difference between a trigger based on a regulatory announcement of loss of representativeness and a trigger based on cessation would be small.  

But that short period could be important. IBA’s LIBOR Reduced Submissions Policy[3] lays out the possibility that the last available value of LIBOR might be published for some days before publication of LIBOR ceases. If this were the case, reliance only on cessation triggers would subject derivatives seeking to hedge exposure to floating rates to an uncertain period of fixed-rate payments. A representativeness trigger would avoid this because contracts would change reference rate before any period of LIBOR publication based on a fixed rate.

Where a non-representative finding leads quickly to cessation, there should be little difference in spreads determined in line with ISDA’s proposed methodology in one circumstance versus the other. But under a cessation trigger, if the cessation is not announced before any fixed-rate period under the Reduced Submissions Policy takes effect, the historical average spread would include what would essentially be a non-market fixed rate. Under the representativeness trigger, the spreads to be applied would more clearly be market-based throughout the look-back period. That seems a second reason why market participants may prefer to include a trigger for fallback based on an announcement of non-representativeness rather than triggers based on cessation alone.  

We are also mindful that, in the case of LIBOR, there are some markets in which changing contractual references to LIBOR is not practicable. While contracts of this kind should be a diminishing share of the total over time, there may, because of historical issuance, still be a material volume of such contracts in cash markets into the 2030s.

Here I note that the Benchmark Regulation allows the administrator to continue to publish even an unrepresentative rate for a ’reasonable period of time‘. The length of this period is not defined in the Regulation. Moreover, some provisions of the Regulation explicitly foresee the situation where cessation of a benchmark would result in a force majeure event and frustrate financial contracts that reference that benchmark. Cessation of LIBOR would likely have those consequences. The Regulation therefore allows continued publication and use of the benchmark to allow for orderly cessation and avoid frustration of financial contracts, but restricts the use of the benchmark by supervised entities only to contracts that already reference the benchmark in certain circumstances.

Whether or not the relevant provisions will apply in the case of LIBOR’s cessation, the problem these provisions seek to address – that continued use in new contracts is not sensible, but legacy contracts remain dependent on the rate – will very likely exist for LIBOR. The potential solution of allowing continued publication of LIBOR for use in legacy instruments that do not have mechanisms to remove their dependence on LIBOR could help to prevent otherwise unavoidable disruption in cash markets. Possibly, even some derivative contracts that are designed and entered to hedge such cash market contracts would also find this useful. It is important that the way critical benchmarks come to an end is not unnecessarily disruptive to markets. But for the major derivative markets, not being able to enter, or clear, new contracts that reference LIBOR would make managing and hedging a back book of legacy LIBOR contracts risky and complex. It would, we think, be extremely problematic for large back books of derivatives to remain on a non-representative LIBOR in this circumstance.

We cannot be certain that this route to use LIBOR only for legacy contracts will be available. Certainly, no one should rely on this option being available. But the possibility should be considered in design of contractual fallbacks. The problems of retiring a benchmark that is no longer able to meet the requirements of the Benchmark Regulation (for example the Euro OverNight Index Average (EONIA)), without causing systemic disruption have been a major theme in the early years of that Regulation. The provisions of the Regulation which touch on this scenario may well be ones that attract close attention in the review of the Regulation that is due by 1 January 2020. For markets such as those which ISDA serves, however, there is no need to risk remaining on a non-representative rate. Contracts can be amended to avoid this. This is true for any new LIBOR-referencing contract being struck in any market. This is a third set of reasons why market participants may prefer to choose fallback triggers based on a regulatory announcement of failing to meet the representativeness test rather than one based on cessation alone.

Conclusion

It is important that trigger events are clear and unambiguous. Cessation of publication has intuitive appeal in that regard. Clearly there should be a cessation trigger. I have set out above, however, that the process of cessation may not be straightforward – it could involve an undetermined period in which the rate is published, but not representative, or even a period when the published LIBOR rate is a fixed one. A regulator’s announcement that the representativeness test will no longer be satisfied upon the date of a forthcoming departure of a particular bank or set of banks could offer a suitably clear and precise additional and alternative event on which to activate a trigger. The FCA is mindful that contracts could therefore choose to include a trigger based on such an announcement. We are also clear on the consequent responsibilities we would have to ensure that any such announcement is communicated to the whole market in an appropriate manner with appropriate clarity.  

I hope these remarks have identified issues to think about in the important next stages in the design of contractual fallbacks, and how they are triggered.

As I hope is also clear from my remarks, however, one thing I cannot provide to you today is certainty about what the end-game for LIBOR will look like. There is uncertainty. That means uncertainty for those who continue to hold or write contracts that reference LIBOR. This continued uncertainty is one reason why we continue to urge those still creating new contracts that reference LIBOR, and have a contract life beyond end-2021, to move rapidly instead to the new RFRs whose continued publication beyond that date can be relied upon.

Having the right fallback triggers means having the right safety belt. But, as Andrew Bailey said last July[4], the best transition will be one in which that safety belt is never needed. That’s why the progress I reported at the beginning of my remarks today on the development of new markets based on the RFRs is so welcome. It is why our efforts are so focused on securing transition before fallbacks are needed. And it is why market participants must implement their transition plans too.