LIBOR: preparing for the end

Speech by Andrew Bailey, Chief Executive of the FCA, at the Securities Industry and Financial Markets Association's (SIFMA) LIBOR Transition Briefing in New York, USA.

a bailey 340 180 migration.jpg

Speaker: Andrew Bailey, Chief Executive
Event: LIBOR Transition Briefing, SIFMA, New York, USA​​
Delivered: 15 July 2019
Note: this is the speech as drafted and may differ from the delivered version


  • Transition from LIBOR has made good progress across derivatives and securities markets, and transition in loan markets is a key next step.
  • Andrew Bailey sets out the benefits to borrowers of the move to risk-free interest rate benchmarks – UK lenders will need to begin engaging with borrowers about lending based on these rates.
  • We expect the LIBOR panels to dwindle or disappear after end-2021 – firms must be able to run their business without LIBOR from this date, and reduce the stock of ‘legacy’ LIBOR contracts.

I am grateful for this opportunity to update markets on the next important phase of transition away from London Inter-bank Offered Rate (LIBOR) to Secured Overnight Financing Rate (SOFR), Sterling Overnight Index Average (SONIA) and the other chosen risk-free rates.

1. Update on progress

The transition is happening.

Markets in the risk-free alternatives to LIBOR continue to develop.

Open interest in SOFR futures has grown to close to half a trillion dollars. In SONIA futures, open interest, which was close to zero in April 2018, had climbed steadily to £129 billion by end June.

The numbers of SOFR-referencing swap transactions are still small, but growth is clear, and continuing. Although – at the moment – LIBOR still dominates swaps markets in the United States, the growth in SONIA swaps in the UK is illustrative of how change can happen. The notional of outstanding cleared SONIA swaps now exceeds £10 trillion.

In the first 6 months of this year, SONIA accounted for a little over 45% of notional swaps trading in sterling. Importantly, the growth in SONIA volumes is particularly notable at longer maturities. Since first half (H1) 2017, there has been 380% growth in (duration-adjusted) traded notional at maturities over 2 years, compared with 150% for swaps of less than 2 years.

In cash markets, SONIA is now the market norm for new issuance of sterling floating rate notes. Since the first new SONIA-referencing issue in a decade – just over a year ago in June 2018 – we have now had £28 billion of new issuance referencing SONIA. We have not had a new unsecured listed public bond referencing sterling LIBOR and fixing past end-2021 since October last year. Issuance of US dollar bonds referencing SOFR, is even more substantial, having reached US$135 billion by end-June.

In the first half of this year, we also saw a dramatic change in securitisation markets in the UK. The first ever securitisation referencing SONIA was completed in December 2018. In quarter 2 (Q2) this year, two thirds of new public deals referenced SONIA.

Transition is not just about new business but about converting outstanding – or legacy – LIBOR contracts.

This is harder in some markets than in others.

In derivatives markets, there is a variety of ways to do so. The most forward-looking UK players have already closed out their LIBOR-referencing contracts in favour of SONIA.

The conversion challenge is perhaps hardest of all in bond markets, where consent solicitations are required. But last month one UK issuer, Associated British Ports (ABP), proved it can be done. Both issuers and investors can benefit from conversion. The successful ABP consent solicitation sets a useful precedent and model that others can follow.

Transition will need to occur not only in derivative, debt and loan market contracts, but also in systems and policies that reference LIBOR. We saw one notable example of transition in Switzerland in June. The Swiss National Bank (SNB) announced that a new SNB policy rate will replace the target range for 3-month Swiss franc LIBOR in its monetary policy strategy. The SNB will seek to keep the secured short-term Swiss franc money market rate close to the new policy rate. It has been clear that Swiss Average Rate Overnight (SARON) – the Swiss market’s chosen risk-free rate – is the most representative short-term money market rate available.

Transition is not just about new business but about converting outstanding – or legacy – LIBOR contracts.

And at the beginning of this month, Japan’s Cross-Industry Committee on Japanese Yen Interest Rate Benchmarks also released its public consultation on key transition steps for Japanese market participants needing to prepare for the discontinuation of LIBOR at end-2021.

2. Transition away from LIBOR in loan markets is a key step ahead

The market in which transition is least advanced is loans. LIBOR-referencing loans are common in lending to non-financial corporations and similar borrowers. Here there is a major transition programme to be undertaken, and on which progress is needed in the year or so ahead.

The green shoots of change are there. On 1 July, the Financial Times reported a first overnight SONIA-referencing revolving credit facility to a UK non-financial corporation, National Express. We know that other non-financial corporates have – for good reasons, which I will return to – been asking for new lending based on SONIA rather than LIBOR. Other Requests for Proposals (RFPs) have been issued. We know that a growing number of banks are now able to lend using SONIA – at least on a bilateral basis. We understand banks are now exploring putting syndicates together too.

One potential advantage of several different markets in several different jurisdictions working on transition in parallel has been the opportunity to increase alignment of conventions.

The Loan Market Association in Europe, and its equivalent in the US (the Loan Syndications and Trading Association (LSTA)) are making progress on essential work to develop new standardised documentation for syndicated loans referencing overnight RFRs rather than the forward-looking LIBOR term rates that have been used to date.

Delivery of that documentation – and its use – will be key next steps in the transition.

One potential advantage of several different markets in several different jurisdictions working on transition in parallel has been the opportunity to increase alignment of conventions. We have seen, for example, use of the daily compounding already standard in derivative markets now used also in bonds on both sides of the Atlantic. We have also seen compounding being built into new SONIA-linked loan products. Seizing these opportunities to align helps to remove basis risk between different contractual obligations, and increase the efficiency of hedges.

3. The transition case for borrowers

I am not at all surprised to see the first corporates actively seek loans referencing overnight SONIA, rather than term LIBOR.

The level of LIBOR is a combination of several different factors, including expectations of central bank policy rates, but also including a credit premium seeking to reflect the cost of unsecured wholesale funding to banks.

Borrowers paying LIBOR are therefore taking on an exposure to credit premiums that are determined by reference to a very thin market, which accounts for a very small portion of overall bank funding, and bank funding costs. One may wonder why many borrowers should be asked to take on this kind of risk, and how many of them really wish to do so.

In previous episodes of financial stress in which lending based on LIBOR has been widespread, the interest paid by those with LIBOR debts has been subject to downward adjustment because of the decrease in central bank policy rates, but upward adjustment because the price of bank credit risk has risen. To take an example, the 6-month sterling LIBOR rate fell by around 3 percentage points over the course of 2008. That was driven by a 3.5 percentage point fall in the Bank of England’s policy rate. The credit premium, however, moved in the opposite direction. The spread of 6-month LIBOR over 6-month overnight indexed swap (OIS), one measure of that spread, hit 3.25 percentage points at one point. That compares with an average of just 7bp between 2000 and end-2006, and 15bp at end June this year.

While much has been done to ensure that confidence in banks’ credit worthiness is not shaken in the way that happened in 2008, it would be brave to predict how much the term unsecured credit spread derived from LIBOR submissions will increase given the already low levels of borrowing in this market. If there are borrowers still on LIBOR at a future cyclical low, might they face the possibility of their interest burden increasing rather than decreasing as economic conditions deteriorate? More generally, as I have said before in the context of LIBOR reform, the market for term unsecured wholesale borrowing in amounts over £10 million (or equivalent in other currencies) – the market which LIBOR seeks to measure – is very thin. Even in the relatively benign conditions of recent years it has been thin. In times of economic stress, we can expect it to be thinner still. Is it really wise for borrowers to tie their interest payments to the cost of such lending at such times?

This is particularly true for consumers. In the United States, most floating-rate mortgages are tied to LIBOR. In UK home-lending markets there are relatively small numbers of individual consumers borrowing at LIBOR rates. Mortgages referencing LIBOR make up only a small proportion of UK mortgage lending. It is more commonly fixed rate. Where it is floating, it is more usual for mortgages to reference variable rates determined by the individual lender, arguably akin to the Prime rate in the United States, or directly to reference Bank Rate – the Bank of England’s overnight policy rate. Bank Rate has the advantage of being easily understood by a wide range of borrowers. It also has the advantage of being tracked very closely by SONIA, meaning there are readily available hedges for those lending on this basis. Importantly, it means consumers do not assume the risks inherent in LIBOR.

I also note the advantages that borrowers may consider from paying interest on a compounded basis. This tends to smooth out any spikes in interest rates that may occur as a result of unusual supply and demand factors affecting a benchmark rate on a particular day.

So LIBOR’s future disappearance is not, in my view, the only reason why we have seen bond issuers move quite quickly to use compounded SONIA, or SOFR. There are a number of other advantages for borrowers to consider. These advantages are not confined only to bond issuers, but also to other borrowers, including small and medium sized enterprises, and retail consumers.

4. Expect to hear from your bank about LIBOR transition

So for participants in UK markets I have 2 messages as the next phase of LIBOR transition approaches.

If you are a LIBOR borrower, you should be expecting your bank to contact you about the change ahead. Put simply, banks should be discussing with you loan products that do not rely on LIBOR. If you are a bank lending LIBOR to corporate and retail customers, you should have plans to explain the forthcoming change to your customers, and be explaining the potential risks of continuing with LIBOR.

If you are a LIBOR borrower, you should be expecting your bank to contact you about the change ahead.

One question close to front of minds is of course how to ensure that those already on a LIBOR-related loan, but maturing after end-2021, are treated fairly. I think the Alternative Reference Rates Committee’s (ARRC’s) recommended fallback language for syndicated and bilateral loans, and the wide consensus that has emerged across buy-side and sell-side in response to International Swaps and Derivatives Association’s (ISDA’s) consultation on how to calculate a fair replacement value for LIBOR helps to provide a path forward on this. But the best way to avoid the complications of calculating, and explaining, replacement rates is to avoid new LIBOR contracts.

5. RFR term rates

One reason that loan markets have been relatively slow to move may be the way that forward-looking term rates are embedded in practice and systems. Some may have persuaded themselves to wait until forward-looking term versions of SONIA or SOFR can be produced.

We used to hear the same from sterling bond markets. Yet, these have now moved to overnight SONIA, compounded in arrears.

And we used to hear the same from our securitisation market. Yet this too has now moved to overnight SONIA, compounded in arrears.

Indeed I think the prevailing view on our Risk Free Rate Working Group – the UK equivalent of the ARRC in the United States – is that overnight SONIA, compounded in arrears will and should become the norm in bilateral and syndicated loan markets too.  On 4 June, the Financial Stability Board (FSB) published details of how overnight rates can be used across a range of different markets. And the desire to use a common reference rate across linked markets, so that risk management is easier, hedges are less costly and basis risk is avoided, can be expected to exert strong gravitational pull towards the overnight rates in all markets. This context is one reason why we think that any firms still delaying transition until term rates arrive are making a mistake.

Notwithstanding that, we think there may be parts of some markets that do want a forward-looking term reference rate. Some borrowers, for example, may prefer precise cash flow certainty months in advance even if it would be less costly to use the overnight rate in arrears.

In the United Kingdom, we are therefore determined to push forward the production of a term rate based on SONIA. The ARRC is seeking the production of a SOFR-based term rate for similar reasons. This forward-looking term version of SONIA should be useful to some niche users in cash markets. It can also provide a key building block for market participants to address some of the issues presented by legacy LIBOR contracts in cash markets. But the use of these forward-looking term rates is meant to be limited. These term rates cannot and will not be the primary avenue to transition. The risk-free rates themselves, SONIA and SOFR, should serve that purpose.

we think that any firms still delaying transition until term rates arrive are making a mistake.

The support that has been offered from banks in discussions to date in relation to term SONIA, makes us optimistic we will make good progress on its production. But I cannot guarantee that these efforts to produce term rates will be successful, or the precise timing of their arrival. That depends in part on ample supporting liquidity in SONIA and SOFR derivatives markets. That is a second reason we think that delaying transition until term rates arrive is mistaken.

6. The end-game for LIBOR

This brings me to the difficult issues of legacy LIBOR contracts.

I can offer no certainty to those who have not taken steps to move off LIBOR by end-2021. Many market participants strive for certainty in their contractual arrangements. In order to achieve it, you do need to transition. But let me say a little more about what might happen after end-2021, and how that might affect remaining LIBOR users.

We do expect panel bank departures from the LIBOR panels at end-2021.

That is why we keep stressing that the base case assumption for firms’ planning should be no LIBOR publication after end-2021. This is not a low-probability tail event.

It may well be that so many banks leave the LIBOR panels at end-2021 that it is simply not feasible to produce a rate based on panel submissions any longer.

But what if some banks are prepared to carry on for a further time-limited period, even while others leave?

The first obvious observation is that this will not be continuity. It could change the properties of the rate. It could affect the level of the rate to a degree. It should also be expected to make the published rate more volatile, with fewer observations and even fewer transactions across which to average.

We do expect panel bank departures from the LIBOR panels at end-2021.

But such a panel contraction will also pose a more fundamental question. Is the rate still sufficiently representative of an underlying market or economic reality?

The European Benchmark Regulation gives us at the FCA a duty, as supervisor of the LIBOR administrator, to reassess whether this ‘representativeness test’ is met every time a panel bank announces that it intends to leave.

One follow-up question is how we would judge this. It is not as simple as setting a precise minimum number of panel banks in advance. We would also assess how active panel banks are in the underlying market and whether their activity is representative of that market. That also raises the question of the state of the underlying market itself. There needs to be an underlying market or economic reality to be measured. It is quite plausible that for any one or more of three reasons – a reduced number of panel banks, the remaining panel banks no longer being a sufficient or representative share of such market as does exist, or the market itself not being substantial enough to measure – that the representativeness test will not be satisfied after end-2021. I urge you not to have misplaced confidence that LIBOR as it exists today will survive. The FCA will not hesitate to make the representativeness judgments that it is required, under law, to make.

This leads to a second follow-on question. What are the consequences of LIBOR no longer being considered representative by its regulator?

I think such a judgment must be the point at which use of the rate in new contracts stops. The Benchmark Regulation itself reaches for this conclusion. Some key provisions already set out certain circumstances where benchmarks do not satisfy the Regulation, or where a benchmark administrator’s authorisation is suspended. The Regulation accepts that the disruption caused by ceasing publication immediately, or preventing use of the benchmark in particular legacy contracts, would be undesirable. But use in new contracts by EU-regulated firms is nevertheless prohibited in these circumstances.

I think the EU’s co-legislators were wise to anticipate this situation. As the ways in which LIBOR and other critical benchmarks may end become clearer, I think it will be sensible to ensure these provisions work effectively across the full range of possible ‘endgames’ for these benchmarks. There is an opportunity to do this in the forthcoming review of the Benchmark Regulation.

In the meantime, we urge market participants to be prepared for the scenario in which loss of representativeness will lead to a legal or regulatory restriction on use of LIBOR for new contracts. To the extent the Benchmark Regulation needs to be fine-tuned to achieve this, we will be supporting such clarifications and/or modifications either or both at EU and UK level.

I am often asked about how to manage the conduct risk associated with the coming end of LIBOR. I think the risk firms have in mind here is whether they will be seen to have done right by their customers. One part of my answer to that is that once FCA has determined LIBOR not to be representative, writing a new contract still referring to the rate involves more conduct risk than I would countenance.

writing a new contract still referring to the rate involves more conduct risk than I would countenance.

This is also why we have suggested market participants very carefully consider their fallback provisions to the extent that LIBOR continues to be used in contracts written now, or can practicably be added into outstanding contracts. In particular, we have encouraged market participants to look at provisions that would work not only when LIBOR ceases, but also in the event that FCA announces the rate will no longer be representative. I am pleased to see the US ARRC has recommended such triggers for use in contracts here in the US. I am also pleased to see the major central counterparty clearing houses (CCPs) ensuring they are ready to switch from LIBOR at this point. Of course, in many cases the even better option will be to avoid use of LIBOR altogether, or to convert those contracts before fallbacks are triggered.

7. The ‘tough legacy’

Even if LIBOR does continue on this unrepresentative basis for a period, the issue about how long the now depleted panel bank rate survives will not have gone away. The period could be quite short. Quite likely it will be more uncertain than ever. We will still face the question of what happens at the end of that period, and in particular what happens to legacy contracts that could not convert, and could not add fallbacks. I have previously called this the ‘tough legacy’ question.

There is a range of options here. None is easy.

One end of that range is that parties to such contracts are left to whatever arrangements their contracts include. That may be reverting to fixed rate. Some contracts may be silent on what happens when LIBOR ceases – leaving only negotiation between counterparties where feasible, or the acting party making their decision on how best to perform on their obligation, but facing the risk of legal dispute.

Also in that range of options is what the benchmark administrator considers viable as part of its own contingency plans, and whether the administrator and its customers see scope to adapt these. We as supervisor of LIBOR’s administrator will also have a view. But today we see no prospect of the administrator being able to continue with a dynamic credit spread – the likely choice would be between a risk-free rate plus fixed spread, or nothing. In other words this does not provide a route to making LIBOR representative again.

Market participants will also ask whether legislation could help. For example, could legislators redefine LIBOR as RFRs plus fixed spreads for those tough legacy contracts? Or could they create safe harbours for those adopting consensus industry solutions which enjoy authorities’ support such as compounded RFRs and fixed spreads? These measures are not in the gift of regulators, but it is sensible to consider their pros and cons.

One task for the second half of this year will be to see if and where consensus exists, so relevant authorities can share and consider the feasibility and consequences of each path. But I want to be very clear – none of the options except that of cessation can be relied upon to be deliverable. Those who can transition should do so.

8. Conclusion

Let me emphasise some points again. The base case assumption should be that there will be no LIBOR publication after end-2021.

Even if LIBOR does continue for a further period after end-2021, it would have changed. There is a high probability it will no longer pass regulatory tests of representativeness. Markets for LIBOR-related contracts are likely to have become highly illiquid. It may not be usable in new contracts. The ability to hedge outstanding LIBOR obligations and claims is likely to have been impaired. The future for those still on LIBOR will be more uncertain than ever. Transition – and transition comfortably before end-2021 – is a better choice.