Term SOFR or not Term SOFR
My last paper made the case for the wider use of Term SOFR. My argument was based on the fact that the potential users of Term SOFR would only occupy approximately 1% of USD derivative turnover and would, therefore, hardly represent a systemic risk. Wider Term SOFR use, together with robust fallbacks, could significantly ease the operational risk and pressures on the end-users, i.e., the counterparties paying spreads to dealers.
On 21 April 2023, ARRC published ‘Summary and Update of the ARRC’s Term SOFR Scope of Use Best Practice Recommendations’ which reiterated and described their approach to the use of Term SOFR. The ARRC continues to recommend the restrictions on the use of Term SOFR which are reflected in the CME license agreement (CME is the administrator of the most commonly referenced Term SOFR).
The current restrictions will most likely continue to support the basis between Term SOFR and compounded SOFR, specifically Term SOFR is around 3 basis points above compounded SOFR for term (say 5 year) derivatives. This is because the restrictions encourage a one-way derivatives market where users can only pay Term SOFR but are restricted from receiving Term SOFR.
This paper looks at some alternatives to Term SOFR which may help end-users achieve the outcomes of Term SOFR without referencing Term SOFR and potentially offending the CME license conditions.
Let’s start with an example
Clients often have had loans linked to LIBOR and fixed debt issues swapped to LIBOR. The combination of floating and fixed liabilities has been managed using derivatives to the desired mix of LIBOR and fixed rates.
As assets are added or changed, the derivative market was a convenient and efficient way to trade into the desired exposure for the asset/liability balance. But this is about to change as USD LIBOR is discontinued from 30 June 2023. From July 2023 the USD market transition will force market users to alternate reference rates, including compounded SOFR, Term SOFR among others.
In the current system, a loan referencing Term SOFR can be hedged to a fixed rate using a derivative, so in this situation there is essentially no change from the situation faced under LIBOR.
However, using a derivative to swap a fixed rate debt issue to Term SOFR is not encouraged (ARRC recommendations) or permitted (CME license).
This is very clear from Scenario 7 in the ARRC publication reproduced below.
This is a significant change from the LIBOR experience.
The end-user may have a preference for Term SOFR referencing liabilities which is allowed via a loan but cannot be achieved by swapping a fixed rate debt issue to the same Term SOFR. The only way forward is to swap the fixed rate to compounded SOFR and run the basis risk between Term SOFR and compounded SOFR.
This is hardly ideal and potentially introduces new interest rate and operational risk.
Can you manage this risk using a basis swap from SOFR to Term SOFR?
The ARRC recommendations appear to allow this activity. Specifically, Scenario 8 shows the following transaction:
The accompanying commentary seems to imply this activity is permitted but the two scenarios below would not be in the spirit of the ARRC recommendations:
swap fixed coupons to compounded SOFR; then
swap compounded SOFR to Term SOFR (Scenario 8).
This does not appear to be permitted in the CME license as the Term SOFR risk is not embedded in a cash instrument, i.e., a loan.
So, the following arrangement of an end-user issuing fixed debt, receiving fixed/pay compounded SOFR and simultaneously receiving compounded SOFR/pay Term SOFR in a basis swap would not, in all likelihood, be allowed or, at the very least not be encouraged.
So, we appear to be left in a difficult situation where the end users may be forced to manage Term SOFR and compounded SOFR where previously they were only exposed to LIBOR.
Other ways forward for end-users
Many clients do not wish to manage the Term SOFR/compounded SOFR risk which they would acquire as part of a loan and fixed rate debt liability combination.
However, all is not lost. This could be managed with an active approach to the coupon dates of the debt and the calculation dates of the derivative.
Approach 1 – manage compounded SOFR to Term SOFR for a fixed debt issue:
Trade with the dealer as receive fixed/pay compounded SOFR (permitted) with terms and dates matching the debt issue.
Each rollover (coupon date), pay the USD OIS (1, 3 or 6-month as appropriate) to swap the compounded SOFR to a fixed rate to replicate Term SOFR.
While this approach achieves the desired outcome of fixing the SOFR rate for 1, 3 or 6-months, there are some challenges:
The operational risk of diarising and executing the USD OIS trades needs to be closely managed.
Term SOFR is calculated from rates across the CME trading day (see note below) and there may be slippage between Term SOFR, and the USD OIS rate traded on the day.
It is important to carefully match the settlement dates for each leg to ensure there is minimal impact on the cash management which could attract significant costs.
This approach is actually quite effective and could be a practical alternative to Term SOFR under circumstances where Term SOFR cannot be referenced.
Approach 2 – choose a point in time for your firm to fix the rate
In this approach, the firm decides on a specific time to transact the USD OIS and accepts there will likely be a difference from Term SOFR.
This is operationally much easier but can provide tracking errors if other components of the portfolio are referencing Term SOFR.
Term SOFR calculations and timing
Term SOFR is calculated in a very different way to the current USD LIBOR.
USD LIBOR is determined at a point in time, specifically 11 am London. If a firm were trying to replicate LIBOR, then trading at 11 am London in a market closely connected to LIBOR (e.g., a single period swap) would likely have little slippage.
However, Term SOFR is not calculated at a point in time.
Term SOFR is calculated as follows (CME description):
‘A set of Volume Weighted Average Prices (VWAP) are calculated using transaction prices observed during several observation intervals throughout the trading day. These are then used in a projection model to determine CME Term SOFR Reference Rates. Full details of the calculation methodology are available on the Term SOFR webpages.’
If you wish to replicate Term SOFR, you will need to trade a proportion of the USD OIS risk at each time CME accesses the price for the VWAP. This is made even more difficult because CME uses random times in each time bucket!
In practice, CME uses fourteen , 30-minute intervals with random time sampling which is not possible to exactly replicate without knowing the timing of each sample.
In practice, this makes it difficult to manage a USD OIS process to replace Term SOFR
USD OIS trades to replicate Term SOFR
Our analysis of USD OIS and Term SOFR since May 2019 (when Term SOFR was first published) gives some comfort that a practical approach could be found.
Example 1 - Trade half the USD OIS risk CME near market open and the other half near the close.
Average slippage is 1 basis point (bp).
Standard deviation is 4 bps.
Maximum is 26 bps.
Minimum is -22 bps.
The average and standard deviation are quite acceptable to many firms, but the outliers (maximum and minimum) are less attractive outcomes as the slippage is substantial.
The average slippage is within 2 standard deviations for 99.9% of time which may provide some comfort.
Example 2 – Performance over a 3-year quarterly swap traded each day.
Over the twelve rate fixes in a 3-year quarterly swap, the statistics are:
Average slippage is 2 bps.
Standard deviation is 1 bps.
Maximum is 4 bps.
Minimum is -1 bp.
The averaging process appears to be quite effective, with the maximum and minimum much reduced.
Over a 3-year swap, there may be some significant slippage days but, on average, the total outcome may be acceptable.
Summary
The recent ARRC announcements include recommendations which may allow end-users to hedge a wider group of Term SOFR exposures. But CME licensing still appears to restrict the use of Term SOFR to cash instruments and derivatives directly linked to those cash instruments.
Should the CME position change, firms could reconsider the Term SOFR use case.
In the meantime, there are practical and effective ways to replicate Term SOFR, some of which are described above.
But there are some risks and caveats:
Term SOFR is difficult to match exactly using USD OIS trades because of the timing and random nature of the Term SOFR calculation.
There can be slippage between USD OIS and Term SOFR which may cause tracking issues if parts of the portfolio reference Term SOFR.
There are operational aspects of trading USD OIS which need to be closely managed including date and settlement timing.
On the positive side:
The slippage appears to average out over time and across the rolls for a multi-roll swap.
A process could be put in place to manage the USD OIS trading and achieve quite acceptable rate fix and operational exposure.
In summary, if you need Term SOFR but are not able to access it for licensing reasons, there are alternatives.
How can Martialis assist?
Martialis can assist in this process by:
Designing appropriate operational procedures to manage the USD OIS trading and risk management;
Providing practical advice on establishing trading relationships;
Transacting and recording the USD OIS trades to ensure date and settlement matching;
Calculating any slippage to Term SOFR; and
Monitoring risks and pricing requirements.
While replicating Term SOFR appears to be complex, our experience shows this can be managed effectively and methodically if processes are robust and established proactively.
The Case for Wider Use of Term SOFR
In my last paper I looked at the trends in USD derivative turnover from BIS Triennial Surveys 2010 – 2022. In this paper I focus on the 2022 Survey and how it supports the wider use of Term SOFR than is currently permitted under existing use limits.
Many of our clients have concerns that SOFR is difficult to use compared with LIBOR. Most of their issues are related to operational problems, forward cash management and accounting calculations. LIBOR, as a forward-looking rate allows clients the time to manage their processes before settlement, whereas SOFR is a daily update which causes additional effort and forecasting. Term SOFR appears to be an easier transition as it shares many of the more desirable attributes of LIBOR.
Simply put, Term SOFR presents an easier transition from LIBOR than is compounded SOFR: CME’s own data tends to support this.
In the past I have looked at the use case for Term SOFR. Risk.net has published on this topic a few times in 2021, 2022 and 2023 outlining the various rules which are currently restricting the use of Term SOFR.
While many people, including myself, support the ARRC and Fed view that widespread use of Term SOFR may be problematic, I also believe that some easing of the restrictions may benefit end-users and not present systemic risks. This is supported by the Survey results as I explain below.
This paper looks at the 2022 BIS Triennial Survey for a breakdown of the users of SOFR and LIBOR. We can reasonably assume the LIBOR users will become SOFR users after 30 June 2023 (i.e., LIBOR cessation) so we can look at the total as a reasonable approximation of market turnover and user mix for SOFR.
Also, we could assume SOFR turnover is at or probably greater than in 2022 based on the trend in USD turnover to higher turnover over the past 10 years.
Firstly, let’s look at the product breakdown per participant grouping.
Turnover of USD derivatives by market participant groups
The Survey does provide some breakdown of the turnover of USD per market participant type. The trades are reported by Reporting Dealers and they separate their trades with other counterparties as follows:
Reporting Dealers – other banks reporting turnover;
Other Financial Institutions – financial institutions which are not Reporting Dealers (e.g., investment funds); and
Non-Financial customers – End users not included in the 2 groups above (e.g., corporates).
There may be derivative transactions between firms who are not Reporting Dealers: these trades would not be included in the Survey. Such trades could be between Other Financial Institutions so their market share may be even larger than that in the Survey.
The following chart shows the breakdown and the market percentage of each group.
The main points I see here are:
Overnight Index Swaps (SOFR) and LIBOR derivatives have comparable turnover.
Other Financial Institutions are the dominant payers with 80% of the total (SOFR + LIBOR) turnover.
Trading between Reporting Dealers is19% of market turnover and considerably less than the trading between Reporting Dealers and Other Financial Institutions.
Trading with Non-Financial customers is a minor part of market turnover and only represents 1% of the market.
Where is the turnover located?
USD is the largest derivative currency by turnover in the Survey as shown in the following chart.
Focusing on the USD, the turnover per country is varied as shown in the next chart.
USD derivatives are predominately traded in USA and UK with some turnover in other countries as well.
Conclusions from the Survey
The 2022 Survey is very clear that:
USD derivative trading is the largest by currency (44%).
The majority of that trading is in USA (68%) followed by UK (24%).
Trading between Reporting Dealers and Other Financial Institutions dominated the trading (80%).
Trading between Reporting Dealers and Non-Financial customers is a very minor component (1%).
As the USD derivatives markets, and presumably debt markets, move from LIBOR to SOFR, I reasonably expect the four points above to continue to be relevant.
How does this impact the use of Term SOFR?
The ARRC and Fed have reiterated their view that the use of Term SOFR should be restricted to a narrow range of products as described by Risk. The CME license terms for referencing Term SOFR reflect the ARRC recommendations and effectively prohibit using derivatives except between dealers to customers to hedge Term SOFR debt to a fixed rate.
The reasons provided by ARRC include to restrict the use of Term SOFR include:
avoiding a repeat of LIBOR with Term SOFR;
inter-dealer trading of Term SOFR could grow rapidly if permitted; and
trading Term SOFR could cannibalise trading SOFR itself.
The main fear is that easing restrictions on Term SOFR to allow for more use by end-users could unleash a torrent of inter-dealer and dealer to customer trading.
The 2022 Survey results do not appear to support this if trading Term SOFR had some restrictions to support dealer-customer and limited dealer-dealer trading.
The trading with end users in LIBOR (no restrictions and the vast majority of this trading based on my last paper) and/or SOFR (minimal in 2022) is only 1% of total market turnover in USD derivatives.
If all the LIBOR trading for Non-Financial customers is replaced by Term SOFR then it still only represents 1% of the market.
If interbank trading of Term SOFR was allowed (under certain restrictions) then it may also be around 1% of the market to clear offsetting risks between dealers.
So, even with a relatively unrestricted approach to allowing Reporting Dealers to trade with Non-Financial customers, the percentage of market turnover could be expected to be in the 1% – 2% range which unlikely to create a systemic problem if Term SOFR is discontinued sometime in the future.
Summary
The BIS 2022 Triennial Survey has many interesting features.
Among these is the interesting fact that trading between Reporting Dealers and Non-Financial customers is approximately 1% of market turnover in USD derivatives.
This has important implications for the use case for Term SOFR.
If trading in USD derivatives referencing Term SOFR is restricted to Non-Financial customers, then it is likely to be similarly around 1% of USD derivative turnover if all LIBOR and SOFR trading references Term SOFR.
This is not significant and would be unlikely to present systemic issues if Term SOFR was discontinued at some time.
Of course, contracts referencing Term SOFR would have fallbacks to accommodate a permanent cessation of Term SOFR.
I believe there is a good argument to allow wider use of Term SOFR for end users. A moderate relaxation of the CME licensing rules would allow a more balanced market (i.e., Term SOFR to fixed and fixed to Term SOFR derivatives) and address the reasonable concerns of the end users in the transition from LIBOR.
Bidding credit sensitivity adieu…
Credit sensitive benchmarks have been fundamental to corporate finance for many years. The loss of this sensitivity adds a systemic burden to deposit taking institutions. Why? Because corporate loan books no longer adequately reflect and compensate for the funding risk in the manner of the LIBOR-based system.
The magnitude of this mispricing depends on the extent to which corporate assets pervade bank balance sheets and the adjustment that bankers have made to lending margins. With banking sector credit risk returning to prominence recently, it is timely to look at this annoying, obscure, yet potentially impactful problem.
Our analysis concludes that while US banking market capitalisation has notably rebounded from GFC lows and appears capable of withstanding a major credit induced event; post-Libor risk-free rates may not adequately reflect the risk undertaken by lenders. This potential interest forgone is of a material magnitude.
Hypothetical Interest Forgone under SOFA during the GFC
We described some important work done by Professor Urban Jermann in our post of August 2022.
Professor Urban is Safra Professor of International Finance and Capital Markets at the Wharton School of the University of Pennsylvania, and we showcased his major work from 2021:
Interest Received by Banks during the Financial Crisis: LIBOR vs Hypothetical SOFR Loans.
This estimated the amount of interest that would likely have been forgone by US banks had US business loans been indexed to SOFR during the GFC:
“The cumulative additional interest from LIBOR during the crisis is estimated to be between 1% to 2% of the notional amount of outstanding loans, depending on the tenor and type of SOFR rate used”.
With cumulative interest forgone estimated at:
U$32.1 billion if loans had instead followed compounded SOFR; or
U$25.0 billion had they followed Term SOFR
To recap, these numbers represent the hypothetical interest income foregone if corporate facilities had been indexed against the major SOFR variants over the period 1st July 2007 and 30th June 2009.
To contextualise the GFC moves, 3m LIBOR versus 3m SOFR Overnight Index Swap (OIS) rates are plotted on the following graph. This credit sensitivity proxy is generally known as the LOIS spread.
The LOIS spread averaged 10.7 basis points for the five years prior to July 2007. This average LOIS spread rose to 89.1 basis points through the GFC, as defined by Professor Jermann.
A non-trivial amount
In a follow-up article in Knowledge at Wharton, editor-writer Shankar Parameshwaran highlights what Professor Jermann was driving at:
“The $30 billion in interest income due to the credit sensitivity of LIBOR is not a trivial amount”.
Inviting hurried back-of-the-envelope calculations, Parameshwaran calculates that:
“On March 6th, 2009, when bank share prices tanked, the top 20 commercial banks from 2007 had a combined market capitalization of $204 billion.”
According to NYU-Stern School of Business, forward price-earnings ratios for money center banks today sit at around 9x earnings.
Those wanting to interpolate what might have been should take care to note that Professor Jermann’s calculations were for hypothetical interest foregone over two years, whereas P/E ratios are based on annualised earnings measures.
Nonetheless, sector annualised interest forgone of between $12 and $17 billion (i.e., halving Professor Urban’s numbers) in an environment where stocks are trading on 9x multiples is very worrying, especially considering the sector market cap of only $204 billion.
2023 – What of the situation today?
Taking Professor Urban’s raw calculations and assumptions and applying sector asset growth, US banks across 2023/24 would likely forgo:
U$59.4 billion if loans follow compounded SOFR; or
U$46.3 billion if loans follow Term SOFR
This is a simple estimate if the credit moves of the GFC period are replicated in the coming two-year period; but system risk appears much lower given the expansive sector market caps.
For the sake of comparison, the market capitalisation of the 20 largest financials in the US sits today at around U$1.54 trillion at mid-March 2023 by Martialis calculations (U$1,541.29 billion to be precise). This represents an impressive rebound and growth of 755% since the height of the GFC.
Corporate lending has grown far less quickly.
According to the St Louis FED, total financial assets of the domestic US financial sectors grew only 85% over the corresponding period (a surprisingly large lag). For the sake of framing, it’s worth noting that the US economy is only a third larger than March 2009 in terms of annualised real GDP, so we’re within reasonable ballpark.
This suggests that the system could handle GFC like conditions, but that the interest likely forgone if GFC-like conditions re-emerge, is still quite material.
What of the latest credit stress event?
To simplify how we look at credit stress we’re starting to favour Invesco/SOFR Academy USD Across-the-Curve Credit Spread Indexes, known more generally by their acronym ‘AXI.’
AXI is starting to gain interest given its constituent make-up and methodology, and we like the clean picture of credit sensitivity it provides in USD:
AXI is a weighted average of the credit spreads of unsecured US bank funding transactions with maturities ranging from overnight to five years, with weights that reflect both transaction volumes and issuances.
AXI can be added to Term SOFR (or other SOFR variants) to form a credit-sensitive interest rate benchmark for loans, derivatives, or other products.
The historic picture of AXI across 1m, 3m, and 6-month tenors from 2018 is as follows:
Credit sensitivity is highlighted over the period of COVID market stress and more recently as bank funding costs have risen with the Silicon Valley Bank and Credit Suisse events of early 2023.
For those interested in further AXI resources, please refer to:
How is AXI tracking the current stress?
Here I compare 3-month AXI with 3-month LOIS (LIBOR-OIS) over an analogue period, starting 90-days prior to the largest jump above one standard deviation in LOIS from 2007, which occurred on 9th August 2007.
While the base of AXI commences somewhat higher than LOIS, at +19.3 versus +8.8 basis points in the ninety days to Day-0 (identified by the dotted red line), the credit sensitivity of the subsequent period shows remarkable similarities at the start of both periods of credit deterioration.
Conclusions
What is clear is that despite the move to risk-free rates (RFRs), rational investors remain rational; demanding higher risk premiums to compensate for the risk of funding banks.
What’s less clear is the extent to which banks are able to pass-on a higher cost of funds to cover their various assets; and this is a systemic problem.
While higher rates are almost uniformly beneficial across mortgage and smaller variable finance segments, it’s not clear how banks compensate for the absence of LIBOR-like credit sensitivity in their corporate assets.
With these points in mind, we encourage bankers to ask three questions:
How long will banking sector credit remain elevated?
Are we receiving sufficient compensation for corporate lending in a risk-free-rate world?
Should corporate loan margins be recalibrated accordingly?
Widespread failure to address these seems to us to have rather obvious consequences.
Term rates are crucial market infrastructure
Late in 2022 the Canadian Alternative Reference Rate working group (CARR) announced it would accede to market pressure and develop a Term-CORRA interest rate benchmark.
Somewhat unusually, CARR’s announcement was made prior to release of summary responses to their consultation on a potential new term rate. This was released on January 23rd.
Leaving aside the unusual delay, it is instructive to review the summary of the consultation’s responses. As I hope to explain, survey responses to regulatory surveys such as these give us important insights into the financial infrastructure requirements of a modern economy.
The market demand for Term-CORRA is telling. It tells us that Term Rates are crucial market infrastructure, not just a addition.
Recapping CDOR’s demise
We took a look at Canadian benchmark reform last year, see here, but to recap, the principal elements within Canadian benchmark reform are:
The mid-2024 cessation of Canadian Dollar Offered Rate, CDOR, the primary interest rate benchmark in Canada since market inception, with the proposed replacement rate being:
The Canadian Overnight Repo Rate Average, CORRA.
CORRA is Canada’s official risk-free rate based on daily transaction-level data on repo trades. These measure “the cost of overnight general collateral funding in Canadian dollars using Government of Canada treasury bills and bonds as collateral for repurchase transactions,” and is thus quite similar to USD-SOFR.
Similar to the approach taken in both the UK and US, and mirrored elsewhere, CARR has published its own well-laid transition roadmap to guide participants:
Source: https://www.bankofcanada.ca/wp-content/uploads/2022/05/transition-roadmap.pdf
What had been lacking from the Canadian plan was the development of a term rate similar to Term-SONIA and Term-SOFR.
Respondents emphatic call for Term-CORRA
To address the absence of a term rate solution, on May 16th CARR surveyed Canadian market participants to “seek feedback on the need for a potential forward-looking term rate (i.e. Term CORRA) to replace CDOR in certain loan and hedging agreements.?”
The survey responses were emphatic and instructive, for example:
Question 1) Does your institution need a Term CORRA rate?
All non-financial firms “wanted a Term CORRA benchmark”.
As did a majority of financial firms.
Overall, a clear majority (37 of 42 respondents) supported its creation.
Which to our mind is as emphatic a statement of end-user demand as any consult on this topic we have seen, clearly prompting CARR’s response as announced in October:
‘In response to the overwhelming support for a Term CORRA, CARR members agree to try to develop a Term CORRA, so long as a robust and IOSCO compliant rate can be created’, my emphasis added.
However, it was the feedback received on why firms felt they needed Term-CORRA that was most instructive.
To help summarise the feedback, I have bracketed the stated respondent reasoning into two areas:
1. Functional/Operational Reasoning:
cash flow predictability, including the need for accurate cash flow forecasting given hedge accounting considerations;
operational simplicity, thus obviating the need for treasury system overhauls, and reducing operational burdens on staff, particularly among smaller firms;
reducing the liquidity risk where a significant change in interest rates were to occur towards the end of an interest-period, firms may have difficulty acquiring adequate cash to meet interest by the due date; and
as a market basis for discount calculations where a present value is needed (e.g., financial reporting).
2. Commercial/Product Reasoning:
Non-financial companies noted:
hedging loan facilities (e.g., for derivatives products to hedge term SOFR-denominated USD borrowings to a CAD term equivalent) and other term rate exposures;
transfer pricing for inter-company loans;
as a reference rate when negotiating contracts with third parties (e.g., vendors, JV partners, customers) or related entities (e.g., partnership loans);
as a rate for inventory/receivables financing;
securitization products with floating rate tranches (asset-matching); and
financial leases.
Financial companies noted:
clients’ hedging activity (i.e., where firms require hedges for Term CORRA interest rate swaps, caps and floors based on the same benchmark);
derivative products, including the hedging of Term CORRA derivatives in the inter-dealer market (as US banks offering Term SOFR products are facing substantial issues managing the associated risks);
securitisations of Term CORRA-based lending; and
where operational constraints on some of the parties to the contract limit their ability to use overnight rates.
To which we are inclined to add:
Sharia-compliant products fundamentally require forward-looking rates (i.e., to establish rates of profit in-advance);
import/export financings of capital projects, in order to forecast cash flows or arrange outgoing foreign currency denominated payments; and
trade and commodity prepayments, for a range of calculations that by definition require forward-looking rates.
Which are all reasons why Term-RFR benchmarks can’t not exist…
We have delved into the topic of Term-RFR rates from several dimensions over the past year:
All of which have served to buttress our view that term rates simply can’t not exist in post-LIBOR finance; they are fundamental to the proper working of whole industry segments.
Taking this further, we see the proper evolution of term rates (bother in RFR and credit sensitive formats) as a desirable goal for the global regulatory family, particularly given trade and capital linkages between jurisdictions.
Then there’s the question of risk dispersion
As the CARR consult respondents noted, Canadian financial respondents wanted the ability to access an inter-dealer market:
Derivative products, including the hedging of Term CORRA derivatives in the inter-dealer market (as US banks offering Term SOFR products are facing substantial issues managing the associated risks) – my emphasis again.
Our evolving view is that while use-case limits are a not unreasonable consideration, they are a) likely unnecessary, and b) possibly responsible for a build-up of essentially un-hedgeable basis risk in dealer swap books that cannot be a positive for financial stability.
And understanding the Use-Case Limits
As I explored last year in Term RFR use limits; what use are they?, if use-case limits are here to stay, firms should start considering the use-case environment and their response quite carefully:
Ensure that use-case controls prevent dealing activity that contravenes regulator or industry preferences, and/or licensing requirements.
Establish a rules-based exceptions mechanism for allowance of Term-RFR use where use is warranted, and processes that document and track artefacts where use-case exemptions are approved.
Understand the peculiar reset risk that exists where Term-RFR’s are hedged via traditional OIS products.
Define and tag Term-RFR products as such, including within risk systems where unexpected basis and reset risk can emerge (between Term-RFR and traditional OIS hedges).
Ensure key staff understand term rates, their role in finance, their use-case limits, particularly where such staff are customer facing.
For Canadian-dollar risk, we would add that it will be important to understand the subtle variation that CARR has flagged already:
This appears to open up the possibility that a Canadian term rate could trade in interdealer markets, though how that might be ‘policed’ is another matter.
To our mind it would be better if the wish list of CARR’s Canadian respondents were met fulsomely, and preferably with a generous relaxation of use-case limits – a topic which we intend to explore in greater detail through 2023.
Summing up
Canada continues to surprise markets with the evolution of CORRA, and now a Term-CORRA benchmark solution. This is in a helpful alignment to developments in the US, where CME Term-SOFR activity continues to develop apace.
Other jurisdictions should be thinking about the market infrastructure developments that the Canadians see as crucial building blocks for finance and financial markets in an IBOR world.
From where we sit it really is hats off to the Bank of Canada and their Canadian Alternative Reference Rate Working Group.
Challenges in using SOFR for all loans
Bank funding risks
Many of our clients are asking about the impact of using risk free rates such as SOFR and SONIA in loan products which allow for discretionary drawdowns. This is especially challenging in the case of revolving credit facilities and particularly those with multi-currency options.
The choices of when, how much and which currency a bank can allow a borrower client to draw down in certain products has always been a challenge to manage. For example, in times of significant stress (e.g., GFC and COVID) the clients can suddenly draw cash from revolving facilities which can create liquidity and pricing problems for the banks.
In the past, the pricing of these facilities was somewhat easier because the reference rates were intrinsically linked to credit-sensitive benchmarks such as LIBOR. When liquidity and credit was stressed, LIBOR typically rose faster than Fed Funds (i.e., the TED spread increased) and the pricing of the loan facilities reflected the increased cost for the banks. This was automatically transferred (in the most part) to the borrower as the reference rate (LIBOR) closely tracked the bank’s borrowing rate.
This affected the bank profitability, the cost of the loans based on that expected cost/return and the behaviour of the borrowers. Banks could reasonably accurately price the facility based on expected costs (additional spread) and borrower drawdowns (expected liquidity requirements) and so were generally prepared to offer competitively priced products to clients.
The recent paper published on 22 December 2022 on the Federal Reserve of New York site and authored by Harry Cooperman, Darrell Duffie, Stephan Luck, Zachry Wang, and Yilin (David) Yang looks at the possible costs and therefore pricing challenges for banks when using SOFR rather than LIBOR for certain loans.
For those whose holiday calendar resulted in them missing this important paper, I really urge you to read the full transcript. In the meantime, I will outline some of the main points and my view of how these may impact banks and their clients.
One word of caution: the authors do note that the paper does not necessarily represent the views of the NY Fed or the Federal Reserve.
Bank Funding Risk, Reference Rates, and Credit Supply – Federal Reserve of New York – 22 December 2022
The paper references some very interesting data and does some really great analysis. The work focusses on the revolving credit loans that give borrowers the option to draw funds up to the credit limit at any time under agreed terms and pricing.
These credit facilities are widely offered by banks and are often used by their clients to guarantee funding for short periods of time even when liquidity conditions in the broader market may be challenging. It is this optionality and the very real likelihood that clients will require the funds at short notice in difficult markets that dictates the terms and pricing of the facilities.
The facilities tended to use credit-sensitive reference rates (e.g., LIBOR) which reflect the actual borrowing rates for the banks at a point in time. In this case, the underlying market conditions are automatically included in the LIBOR pricing for the clients and the banks can offer these products at competitive prices and terms.
The authors look at the impact of replacing LIBOR with risk free rates such as SOFR. They run some simulations for GFC (2008) and COVID (2020) and compare the performance of the products (LIBOR plus a fixed credit spread versus SOFR plus a fixed credit spread).
Without the credit sensitivity (i.e., using SOFR), the banks would be very likely to have a lower margin (NII – Net Interest Income). This was particularly evident in GFC (-6.48 billion) while the COVID experience was lower (-1.59 billion).
This is demonstrated in the following chart used by Ross Beaney previously.
These results can be found in Section D of the paper – Accounting Counterfactual Assumptions & Additional Results, Appendix D.
Why does this matter?
The authors conclude that the choice of reference rate (LIBOR or SOFR) affects the supply of revolving credit lines.
Under LIBOR referencing facilities, the bank and the client share the embedded liquidity risk. As the market supply of funds decreases, the cost increases (i.e., LIBOR rises relative to Fed Funds) are largely passed on to the borrower. The lender (the bank) still retains some risk as their individual ability to borrow funds from the market may be impacted by their own credit compared with other banks and this may diverge from LIBOR.
Point 1
This could result in the banks increasing the pricing for revolving credit facilities linked to SOFR or restricting the supply of these products. Because the risk is now firmly with the bank, the pricing and supply must reflect a ‘worst case’ scenario where the bank has limited ability to borrow from markets and/or an increased cost.
Note that this is not such a major issue for standard, term loans. These products can be term funded at a known margin and do not have the optionality for drawdown. In this case, there is little or no uncertainty about the principal and timing of the drawdown and repayment of the loan.
The authors do not consider multi-currency revolving credit facilities in their work. In my experience, the multi-currency option creates even more issues for banks and their clients. If you use a credit sensitive reference rates for the optionality within one currency (e.g., AUD BBSW) and a risk-free reference rate for another currency (e.g., USD SOFR), then under certain market conditions, one alternative could be far better than another for the borrower. For example, with a fixed spread to SOFR, if the actual market rate is above SOFR plus the fixed spread, then the rational borrower would logically opt for the USD.
Point 2
Multi-currency revolving credit facilities present additional problems for pricing and supply of the product. A bank must reasonably use the worst-case spread for a risk-free rate or risk being drawn on the facility when liquidity spreads exceed the fixed spread in one or more currencies referenced in the facility.
Is there another way to resolve this?
Fortunately, several credit-sensitive alternatives exist which could be used to replace LIBOR and be used with or instead of SOFR. We have previously written on this subject here, here, and here.
In the USD market these include:
Across the curve Spread Indices – SOFR Academy and Invesco
Bloomberg Short-Term Bank Yield
Credit Inclusive Term Spread/Credit Inclusive Term Rate – S&P Global – not yet available for licensing.
Published by AFX
I will not describe each of these alternatives, but the links are provided for further information if you are not already familiar with the reference rates. In all cases, they have reasonably tracked LIBOR in the past and add back a level of credit sensitivity.
If Messrs Cooperman, Duffie, Luck, Wang, and Yang are correct in their conclusion that SOFR-linked revolving credit facilities may be affected (I.e., negatively compared with the outgoing LIBOR equivalents), using the newer credit-sensitive rates above may help restore the risk balance. This may allow the banks to continue to offer these products as they do now.
Summary
The 22/12/2022 paper (and I appreciate the alliteration!) highlighted some particularly important challenges in the transition from LIBOR to SOFR for revolving credit facilities. The authors rightly point out that pricing and supply of these facilities may be negatively impacted.
While they did not look at multi-currency facilities, these present even more problems for banks who now provide these to clients.
Many end-user borrowers have such facilities to ensure they can have liquidity available under all circumstances. They may be expensive but are essential components of good risk management for many corporate and investor firms.
If the supply and/or price of these products is negatively impacted, there is a real risk that the end-users may face increased risk to liquidity crises in the future.
Why debate over credit sensitive rates won’t go away…
It’s not a debate being played out around weekend barbeques, and it’s not likely to gain regular billing on 6 o’clock bulletins, but the question of whether finance needs credit sensitive benchmarks is one we’ve looked at many times – and yes, debated.
In my latest blog post I look at a rather obvious pitfall that seems likely to arise if corporate finance doesn’t settle on a credit sensitive lending solution.
Urban Jermann
Urban Jermann is Safra Professor of International Finance and Capital Markets at the Wharton School of the University of Pennsylvania. In December 2021, Professor Jermann released a topical research paper titled: Interest Received by Banks during the Financial Crisis: LIBOR vs Hypothetical SOFR Loans, which was a formal study into the cost to the US banking sector that might have been incurred (interest earnings foregone) had US business loans been indexed to SOFR reference rates (instead of LIBOR) during the two years of the GFC.
Jermann’s work is interesting from several standpoints:
He uses the expression “insurance payout” to describe the credit sensitive component of the lift in LIBOR rates through the GFC, which is both interesting and, in our view, a healthy development since banks may like to consider actuarial approaches in assessing their long run risk to credit sensitivity, and,
He calculates estimates of the “payout” LIBOR lenders received over what they would have had their loan books referenced SOFR in both compound and term formats.
He notes that:
The cumulative additional interest from LIBOR during the crisis is estimated to be between 1% to 2% of the notional amount of outstanding loans, depending on the tenor and type of SOFR rate used.
And the amounts estimated that would have been forgone by banks on loans whose economy-wide balance “may have been as high as $2trln” are meaningful:
U$32.1 billion if loans had instead followed compounded SOFR; and
U$25.0 billion had they instead followed Term SOFR.
It’s important to note, these numbers are the hypothetical reduction in interest income US banks would have borne if loans had been struck against compound or term SOFR, instead of LIBOR between June 2007 and end June 2009, and incorporate the quantum of the following US loan-types:
Syndicated loans;
Corporate business loans (bilateral),Noncorporate business loans;
Corporate Real Estate loans
Professor Jermann obtained his data from the Shared National Credit (SNC) Program for syndicated loans, and the Financial Accounts of the Federal reserve for the other categories.
All of which unearths an important question: does the loss of LIBOR’s insurance-like credit sensitivity reduce banking sector returns if not compensated?
Our answer is probably, but not if average bank funding costs settle below their long run spread to risk-free rates and stay there indefinitely.
Which makes us wonder what kind of numbers Professor Jermann might have found had he chosen to analyse the post-GFC environment since major bank funding mixes been altered so dramatically in the aftermath of the GFC?
We think it’s also interesting to note that the interest foregone by banks would have been worsened under compounded SOFR compared to Term SOFR.
Credit sensitivity since the GFC
The post-GFC numbers are relatively easy to calculate, and while we’re not going to attempt to research the quantum of loans financed through that period, we will use post-GFC USD LIBOR/SOFR OIS spreads to estimate the potential loss in basis points per annum.
We find that mean and median loss of credit sensitivity as follows:
In other words, the average annual loss of credit sensitivity yield since the end of the GFC (June 30th, 2009), was 21.4 basis points (BP) for USD facilities.
Market anecdotes
Our most recent market soundings indicate that in loan markets in the US and elsewhere lenders are:
increasingly moving towards CME (Chicago Mercantile Exchange) Term-SOFR as the basis for new USD deals, in a clear move away from compounding;
that there is no clear consensus on credit adjustment spreads (CAS); and
that lenders appear to be simply adjusting credit margins where they can do so.
By the way, this latter practice results in the same outcome as lenders negotiating a fixed CAS and simply adding this to compound or term SOFR, i.e., analogous with the basic mechanic used in the ISDA (International Swaps and Derivatives Association) Protocol for derivative transitions.
The key point is that lenders who adopt this approach are fixing the credit sensitive component of the facility, and therefore foregoing the potential “insurance payout” of a credit sensitive alternative.
Our point is (and has been) that this practice may prove problematic depending on the extent to which recent credit spreads are reflective of future.
A long-term 3-month LOIS (LIBOR/OIS spread) proxy
To prove the possible problem, we have constructed an indicative 3-month USD LOIS proxy based on an adjusted TED-Spread (3-month USD LIBOR less Treasury Bill yields) going back to 1986.
It supplies an interesting picture of the peaks and valleys of a proxy credit-sensitive rate going back well into the distant years of modern finance.
The long run statistics on the proxy are as follows:
But we think it’s important to consider the historic levels of the proxy across differing periods and differing credit environments to prove how times can change; hence we have broken the data into 5-year periods to give a perspective based on different historic periods:
This approach shows that based on 5-year bucket analysis, banking sector credit was more expensive than the ISDA 3-month spread (26.15 basis points) in all prior periods and has been relatively rarely below the mid-20s.
It also appears somewhat correlated to the general level of interest rates, which makes intuitive sense. This should give those in favour of static CAS pause for thought (a possible topic for a future blog).
AXI has gone live
Which is why we can’t see the credit sensitivity debate disappearing anytime soon.
It’s also why we took a keen interest in the recent official release of Invesco Indexing and SOFR Academy’s Invesco USD Across-the-Curve Credit Spread Indices (AXI):
The AXI and FXI indices are forward-looking credit spread indices designed to work in conjunction with the Secured Overnight Financing Rate (SOFR).
AXI and FXI work to form a credit-sensitive interest rate when used in combination with Term SOFR, Simple Daily SOFR, SOFR compounded in arrears, or SOFR Averages.
AXI is a weighted average of the credit spreads of unsecured bank funding transactions with maturities out to multiple years.
The relevant methodology can be found here.
AXI’s release joins BSBY, and IHS-Markit’s USD Credit Inclusive Term Rate (CRITR) & Spread (CRITS), in the credit sensitive reference-rate stakes.
In conclusion
It’s impossible to accurately predict which mode markets will eventually settle on, fixed estimates, or dynamic credit. Our hunch is that credit sensitive rates will become an economic necessity once the unprecedented recent compression of financial market spreads and interest rates abates.
The use of fixed spreads based on recent LOIS history could be thought to weaken the overall financial system, since it’s hard to see how the lost revenue of Professor Jermann’s “insurance payout” gets made up for in any repeat of GFC-like credit conditions, but we are not forecasters.
There is, though, the possibility that like El-Nino and La-Nina we have recently traversed a period of an unusual credit risk drought in what is a cyclical system. In that case credit sensitivity should be considered of long-run benefit.
If banking serially misprices its own costs, there can be few actual winners.
Term SOFR – Moving forward because the demand is real
We have posted quite a few blogs on this subject, notably in May 2022 and April 2022. When these blogs were published, the use cases for Term SOFR very restrictive and are defined by the Term SOFR administrator (CME).
More recently, the ARRC has reconvened the Term Rate Taskforce which was reported in the July 13 Readout. The important aspect of this group’s work is to review certain aspects of the Term SOFR uses as in the Readout:
‘The Term Rate Task Force provided an update on its discussions around term SOFR derivatives. In particular, the Task Force has been discussing participants’ views on issues/questions around term SOFR derivatives and the overnight SOFR/term SOFR basis, including clearing, capital, and accounting considerations. It was noted, however, that the Task Force is distinctly not being reconvened to materially relax the substance of the ARRC’s best practice recommendations regarding scope of use for the term SOFR rate.’
The important word is ‘materially’. This implies some changes to the use cases could be supported and therefore able to be reflected in the CME licensing rules for Term SOFR. Risk magazine has covered this topic only recently (7 July 2022) and has provided insightful perspectives on what additional use cases may address participant requirements while still adhering to the basic ARRC principle of proportionality (i.e., not allowing exponential expansion of the use of Term SOFR to keep the volumes referencing the benchmark in proportion to the underlying transactions used to calculate it).
Current CME licensing
The CME website has the following data for Term SOFR use:
Licenses from the CME are divided into 3 categories:
Category One applies to cash instruments like loans, mortgages, bonds, notes, and money market instruments.
Category Two is necessary for applying Term SOFR as a reference in any derivative product. However, the catch is that the license only covers derivatives that are directly linked to cash instruments which reference Term SOFR.
Category Three is used by specialist providers in products or services they develop and license to external clients. We do not cover this category in the blog as it only applies to service providers.
The licensing for Category Two is quite restrictive, appears to create a one-sided market and excludes potentially compliant uses for Term SOFR. In practice, Category Two requires the derivative to only use reference Term SOFR if it applies to a cash debt exposure.
Why is Category Two restrictive?
Category Two, in practice, only allows for derivatives to be used to swap floating rates (Term SOFR) associated with cash products to fixed rates. In other words, a end user can only pay fixed and receive Term SOFR in a standard interest rate swap because the Term SOFR must be associated with the cash instrument to allow a derivative to reference Term SOFR.
One example where there may be an end-user use for Term SOFR is in a derivative swapping fixed debt coupons for floating USD. SIFMA does publish the corporate bond issuance statistics for USD as summarised in the following table:
The corporate bond market is large and a significant percentage (78 – 91%) issue fixed rate. A proportion of the fixed rate issuers may decide to swap the fixed coupons to SOFR and many of these issuers are possibly in a similar operational position to the issuers of floating bonds who are permitted to use Term SOFR.
However, the fixed rate issuers cannot access Term SOFR in the swap to achieve a floating rate liability as it is not allowed in CME Category Two licenses.
Is there any real difference between the floating rate and fixed rate issuers? Arguably there is no difference except the preference of investors at a specific time.
Perhaps there is a legitimate use case for corporate fixed rate issuers to use derivatives to swap to Term SOFR and therefore an amendment to the CME Category 2 licence may be appreciated to allow a fixed to term SOFR swap.
The challenge for the ARRC and the Term Rate Taskforce
The Term Rate Taskforce has been reconvened to consider changes to the recommended use of Term SOFR given how the market has evolved since 2021. (I very much respect their challenges in finding a way to balance the proportionality with clear end-user demand!)
One advantage of allowing for derivatives which swap fixed to Term SOFR is that it helps balance the current market where only Term SOFR to fixed rate is, in practice, permitted. This could remove or reduce the current additional costs of around 1.5 to 3.5 basis points for end user borrowers in referencing Term SOFR (see the Risk article above). I do note that investors could provide the balancing side of the swap, but they could just as easily buy floating rate products (as they seem to be doing in 2022!)
Should the use cases be expanded to allow for derivatives such as the example above?
Should there be inter-dealer trading (perhaps with some restrictions) to allow banks to clear risk and provide more competitive pricing for their customers?
We will have to await the outcomes from the ARRC, but the market is showing there is demand for a less restrictive approach to Term SOFR which would need to be reflected in the CME licensing arrangements.
Summary
While we have written blogs previously on this topic, markets have changed considerably, and the tensions are very clear. The Term SOFR derivatives market is currently one-way as reflected in the basis cost of swapping floating for fixed rate.
Allowing a more balanced market where end-users can have equivalent access to fixed to floating as well as floating to fixed swaps may address some of the pricing imbalances and also provide bank customers a more complete service.
If there is a (limited) inter-dealer market then risk can be cleared more readily and banks can move the Term SOFR exposures to their trading books, thereby reducing operational and capital costs.
There is a compelling argument that end users of Term SOFR derivatives would benefit from some adjustments to the ARRC-recommended use of Term SOFR.
I hope the ARRC can adapt the recommended uses for Term SOFR and that this will be quickly incorporated into the CME Category Two licensing.
CDOR, CARR, CORRA, CAG? What’s really happening in Canada?
The Canadian Alternative Reference Rate working group (CARR) was first established in March 2018 with a remit not dissimilar to the ARRC of the US Fed, to “guide benchmark reform efforts in Canada.”
In this blog I attempt to cut an explanatory path through the maze of acronyms that loom in Canadian finance, and distil what’s really going on with Canadian benchmarks, and try to gain a better understanding of the fundamental, underlying problem with CDOR.
Are there important lessons from the Canadian experience?
Why so many acronyms?
In every jurisdiction where benchmark reform has been attempted, new market acronyms abound.
This has been without exception, though we’d concede that the Bank of England Working Group on Sterling Risk-Free Reference Rates has seemed to defy market ‘acronymisation’!
Canada has been no Robinson Crusoe in this evolution, though the list of reform acronyms could be said to have been taken to new lengths in terms of sheer number. We’re sure this was not deliberate, but with so many new terms mixing with old terms it’s somewhat instructive to recap what they all stand for, and what role they play:
Canadian Dollar Offered Rate, CDOR, is administered and published daily by Refinitiv Benchmark Services. The rate was developed in the 1980s as the basis for pricing Bankers’ Acceptance supported credit facilities (short-term corporate promissory notes backed by an unconditional payment obligation of a bank). It has remained the primary interest rate benchmark in Canada since inception and is referenced in over $20 trillion of gross notional exposure across the Canadian wholesale financial system.
Canadian Alternative Reference Rate Working Group, CARR, was created through the sponsorship of the Canadian Fixed-Income Forum, to coordinate Canadian interest rate reform and “ensure Canada’s interest rate benchmark regime is robust, relevant and effective in the years ahead,” with objectives to:
o support and encourage the adoption of, and transition to, the Canadian Overnight Repo Rate Average (CORRA) as a key financial benchmark for Canadian derivatives and securities; and
o analyse the current status of the Canadian Dollar Offered Rate (CDOR) and its efficacy as a benchmark, as well as make recommendations on the basis of that analysis.
Canadian Overnight Repo Rate Average, CORRA, Canada’s official risk-free rate based on daily transaction-level data on repo trades that measure “the cost of overnight general collateral funding in Canadian dollars using Government of Canada treasury bills and bonds as collateral for repurchase transactions”. The Bank of Canada considers CORRA a public good and publishes rates at no cost to users and data distributors each business day at 11:30 am Canada Eastern Time.
CORRA Advisory Group, CAG, was initially established to advise the Bank of Canada’s CORRA Oversight Committee on potential adjustments to the CORRA methodology, “stemming from changes in repo market functioning and from any emerging methodology issues, as well as on any changes undertaken as part of regular methodology reviews.” A key role of CAG is to assess if CORRA continues to represent the overnight general collateral funding rate where Government of Canada securities are posted as collateral.
While to-date no formal Term CORRA rate has emerged, I note that Canadian industry demand for a CAD term rate is like that found in other jurisdictions (high, reasonable, and in our view very rational), and that CARR has a Term CORRA subgroup reviewing the need for “a complementary term rate to overnight CORRA for loan and related hedging products.” This welcome group is also expected to develop an appropriate methodology.
So, what’s the plan?
Similar to the approach taken in both the UK and US, and mirrored elsewhere, CARR has published its own well-laid transition roadmap.
Source: https://www.bankofcanada.ca/wp-content/uploads/2022/05/transition-roadmap.pdf
Readers with experience of the Bank of England roadmap to GBP LIBOR’s cessation will note the similarities:
announce a formal cessation trigger;
establish a ‘no new LIBOR (CDOR)’ exposure cut-off date; and
set a formal publication cessation date in stone.
We see no reason for Canadian markets to struggle with any of this, and market disruption should be minimal; the global template for cessation has essentially been set.
What’s the problem with CDOR?
Helpfully, CARR’s formal review of CDOR is available on-line, and I am leaning on it heavily through this section
After consulting with industry through 2021, and as outlined in the transition roadmap, CARR formed the view that Refinitiv Benchmark Services (RBSL) should cease the calculation and publication of CDOR after June 30, 2024.
My key question is, why was this?
CARR is explicit here, saying that “there are certain aspects of CDOR’s architecture that pose risks to its future robustness.”
What risks?
Here I paraphrase the two key risks CARR found, and which we believe are fundamental, to make summary easy:
Input rates crucial to CDOR’s publication cannot be directly tied to observable transactions, and are hence “based predominantly on expert judgement”. CARR decided that this was “not consistent with evolving global best-practices” nor, we would add, was this consistent with IOSCO Principles for Financial Benchmarks.
CARR noted that Bank funding has evolved to “better match the term of their funding to the term of their loans, and this practice is now codified in Basel III regulation”. Further, CARR noted that “BA loans are “term” or “committed” facilities, bank treasuries no longer fund them through the issuance of BA securities that are generated through the loan drawdown”. The reduction of bank acceptance issuance tolls a rather obvious bell.
CARR also noted that the move away from CDOR “aligns Canada with the heightened standards other jurisdictions began adopting in 2018,” and that the rate’s “contributing member banks may decide they no longer wish to continue submitting rates voluntarily.”
These are interesting other riders; hinting at a desire on behalf of the working group, and presumably others across Canadian finance, to follow the benchmark modernisation path seen across the US, UK, Japan, and Switzerland.
Boiling it all down?
When CARR tell us that “there are certain aspects of CDOR’s architecture that pose risks to its future robustness,” I am at once reminded of the fall in interbank (LIBOR) lending that has occurred since the mid-90s, and prominently since the GFC, as displayed in the classic St Louis Fed graphic on the topic:
Source: https://fred.stlouisfed.org
In Canada, the pattern has been familiar; bankers’ acceptances are simply no longer playing the significant role they once played in Canadian finance.
Short-term paper held on financial balance sheets has fallen to around 1% of the national balance sheet since Y2k:
Source: Statistics Canada. Table 36-10-0580-01 National Balance Sheet Accounts
Perhaps more interestingly, since the GFC short-term Canadian paper has hardly registered as meaningful liabilities on the non-financial sectors cumulative balance sheet either:
Whichever way CARR may like to couch it, dire levels of actual BA activity really has spelled the end of CDOR.
Important Lessons?
Our interest in Canada as a benchmark test-case stems from the similarities between Australian and Canadian finance, where bank accepted securities play and have played a key role in benchmark formation in both countries.
We will look more closely at the Australian benchmarks in the next blog.
Perhaps the most important lesson from Canada is that times change, and that rational actors will act rationally. Market evolution, some pushed along by natural market evolution, and some pushed along by regulators, needs to be carefully watched, and shouldn’t be ignored.
We continue to watch this space with deep interest.
Term RFR use limits; what use are they?
It’s a long time since anyone disputed trade liberalisation as a force for economic good.
Likewise, it’s a long time since anyone questioned the value of liberalised financial markets in promoting the efficient movement of global capital.
So why are global regulators defining use-case preferences or limits on the use of Term RFRs? Why are lay out hurdles that restrict us?
In this blog I explore this topic and present some suggestions on how to negotiate the use-case minefield that has emerged.
Liberalised markets
It’s a long time since anyone seriously disputed trade liberalisation as a key driver of economic growth. Likewise, it’s a long time since liberalised financial markets were claimed to detract from the efficient allocation of capital and resources across economies.
Financial markets work best when they’re liberalised. For one thing, risk transfer and risk dispersion take place more efficiently when there are deep pools of market activity. For another, pricing transparency is advanced in markets where participation is not discouraged; and these are both aspects of what the Big Bang was ultimately all about.
Step forward to the 2020’s and the freedom of a truly ubiquitous financial benchmark is making way for a collection of replacement rates, including a serious suite of new term rates that we’ve written about before, (see here, here, and here). What seems somewhat at odds with liberalisation is that these rates appear to be emerging somewhat less free depending on the jurisdiction they are based.
We notice that all of the Term-RFR variants outside Japan are accompanied by regulator preferences and licencing restrictions that are likely to govern how they’re used. We’re not convinced these are helpful, but we can’t deny their existence.
Regulatory Pressures
Paraphrasing key quotes and/or summarising measures from across the regulatory landscape, the use-freedom of term rates appears to have been deliberately crimped:
They (term rates) are by their nature a derivative of RFR markets. Because these RFR-derived term rates would be based on derivatives markets, their robustness will depend on derivatives market liquidity. Activity in these derivative markets may, however, be relatively thin and vary significantly with market conditions, including on expectations about central bank policy changes.
Moving the bulk of current exposures referencing term IBOR benchmarks that are not sufficiently anchored in transactions to alternative term rates that also suffer from less liquid underlying markets would not reduce risks and vulnerabilities in the financial system. Therefore, because the FSB does not expect such RFR-derived term rates to be as robust as the overnight RFRs themselves, they should be used only where necessary.
Bank of England Term-SONIA
In January 2020, the Sterling Working Group recommended a limited use of Term SONIA.
If use of a TSRR became widespread, there is a risk of reintroducing structural vulnerabilities similar to those associated with LIBOR. While several hundred $trillion worth of financial contracts reference LIBOR, the underlying market determining the rate was comparatively smaller.
These risks can be avoided by limiting the use of TSRRs.
The areas identified as being potentially appropriate for Term SONIA uses were:
smaller corporate, wealth and retail clients. However, the Task Force noted that other rates such as fixed rates or the overnight Bank Rate (see further below) should be considered as well;
trade and working capital financing, which use a term rate or equivalent to calculate forward discounted cash flows to price the value of assets in the future;
export finance and emerging markets, where the customer typically requires more time to arrange and make payments; and
Islamic financing which can pay variable rates of return, so long as the variable element is predetermined.
FMSB Standard on use of Term SONIA reference rates
Noting that FMSB members agree to abide by FMSB Standards in their business practices, we note the wide reach of this standard in UK term rate use.
Term SONIA is derived from executable quotes for SONIA-based interest rate swaps. Its robustness therefore depends on the liquidity in such swap markets, so it is in the interest of all potential users of Term SONIA for those markets to remain primarily based on overnight SONIA. If the volume of swaps data available is not consistently as large as in overnight funding markets, then Term SONIA cannot be as robust as overnight SONIA.
The standard then loosely defines possible use cases across lending, bonds, and derivative markets, repeatedly imploring market participants to assess use case limits “in a manner consistent with this Standard.”
It makes little attempt to be prescriptive.
FED ARRC Term-SOFR
Use of the SOFR Term Rate should be in proportion to the depth of transactions in the underlying derivatives market and should not materially detract from volumes in the underlying SOFR-linked derivatives transactions that are relied upon to construct the SOFR Term Rate itself over time and as the market evolves. Like all the ARRC best practices, the extent to which any market participant decides to implement or adopt any benchmark rate is voluntary.
On 29 July 2021, ARRC formally recommended CME Group’s forward-looking SOFR term rates.
ARRC did not impose similar restrictions on the use of Term SOFR as seen with Term SONIA, instead leaving CME Licencing rules to govern use limits (see below).
CME Licencing CME Term-SOFR
The CME’s licencing regime maintains three categories of use (described within Use Licences) which combined, broadly conforms to the ARRC’s Best Practices guide:
Category 1 – Use in Cash Market Financial Products;
Category 2 – Use in OTC Derivative Products; and
Category 3 – Use in Treasury, Risk & Transaction Admin Services.
However, Term-SOPFR’s free use in a traditional liberal OTC setting is not assured, and this is crucial:
Use of CME Term SOFR Reference Rates only as a reference in an OTC Derivative Product that is tied or linked to a licensee and End User hedging against exposure from one or more Cash Market Financial Products that references the same CME Term SOFR Reference Rate.
We read this as a serious control, likely to crimp activity.
SNB National Working Group on Swiss Franc Reference Rates
Imposing a complete term-rate limit, the SNB considered it unlikely that a robust SARON term rate could ever be feasible and recommended that market participants use compounded SARON wherever possible.
Working Group on Euro Risk Free Rates (ECB / ESMA)
While the joint working group recommended a forward-looking term rate for €STR, derivative markets based on €STR are not yet sufficiently liquid to permit the endorsement of a forward-looking term rate, despite having announced an RFP process in July 2019. No target date for the publishing of such a term rate exists at this stage.
To an extent this is less problematic in EUR, since Euribor continues to be published.
TORF becomes the outlier
Somewhat surprisingly, Japan has taken the liberal path to term-rates.
In Tokyo, the Tokyo Term Risk Free Rate (TORF) is published by Quick Corp. Since May 2020, this has been based on uncollateralised overnight call rate which calculates the interest rate from JPY Overnight Index Swap (OIS) transaction data.
Unlike other major jurisdictions, neither Quick nor the Bank of Japan has (so far) placed any limitation on TORF’s use, which would appear to make Japanese financial markets the most liberalised on the planet for Term-RFR rates.
So, why limits?
Our reading as to why a collection of use-limits has unfolded is quite simple; the robustness of new benchmarks is of paramount importance to a regulatory family bruised by the fact that LIBOR robustness fell so far. They don’t wish to see a repeat.
What’s also clear is that the robustness of Term-RFR is always and everywhere dependent on the derivatives market liquidity of related OIS and RFR futures markets, which are implied-forward RFR markets. The proper functioning of the term rate system therefore depends on the viability of these implied forwards, which themselves must be kept robust.
It is not hard to imagine what could happen if a particular Term-RFR market started to cannibalise the liquidity of its own underlying rate.
We view this as a particularly unlikely scenario since OIS and RFR-Futures markets are developing and have almost limitless liquidity potential (they are implied, not physically supplied). We also see Term-RFR fallbacks as an important risk-mitigant. Nonetheless, we have to acknowledge that the regulators have a serious point to make, and they appear to have made it via their use-case preferences.
Practical Implications
While expounding its use-case preferences for Term-SOFR, the FED’s ARRC noted that: “each market participant should make its own independent evaluation and decision about whether or to what extent any recommendation is adopted.” Which seemed to throw the use-case ball into the welcoming arms of individual firms, but with the CME licencing regime taking the shape that it has the ARRC’s use-case wish-list has emerged more practically within the licence.
This effectively swamps all sense of “independent evaluation and decision” (making) on the USD term RFR scene.
Conceding that participants are generally still grappling with the wider ramifications of LIBOR cessation, and the uncertain new world of compound or simple RFR’s, the use of term rates is patchy. Nonetheless, firms should start considering the use-case environment quite carefully to:
Ensure that use-case controls prevent dealing activity that contravenes regulator or industry preferences, and/or licencing requirements.
Establish a rules-based exceptions mechanism for allowance of Term-RFR use where use is warranted, and processes that document and track artefacts where use-case exemptions are approved.
Understand the peculiar reset risk that exists where Term-RFR’s are hedged via traditional OIS products.
Define and tag Term-RFR products as such, including within risk systems where unexpected basis and reset risk can emerge (between Term-RFR and traditional OIS hedges).
Ensure key staff understand term rates, their role in finance, their use-case limits, particularly where such staff are customer facing.
At Martialis we have immersed ourselves in the complexities of RFR’s and the conundrums associated with Term-RFR’s. We are well placed to guide firms as they negotiate the path through the new rates, associated products, and the various use-case minefields.
Credit Sensitivity Perspectives
The transition to multi-rate benchmarks was always going to raise some fundamental questions regarding post-LIBOR deal pricing.
In this blog I take a closer look at simple proxies for USD credit sensitivity using historic data available at the St Louis FED. The findings are quite consistent with what we should expect intuitively, but in contracting for new deals it’s worth considering the historic path of credit sensitivity and whether exposure to it is advantageous.
We believe participants need to understand the new offerings. The availability of both credit sensitive and credit insensitive offerings makes this somewhat more important, since those found to have accepted an unnecessary path run the risk of being challenged to explain their reasoning.
Historic credit sensitivity
There are many different approaches to identifying periods of financial-sector credit stress. In this piece I have chosen to rely on a very simple proxy stress measure; the historic difference between USD AA-rated commercial paper (CP) of financials and the corresponding CP of AA-rated non-financials at the 3-month tenor.
(FinCP – NonFinCP) = FCP Indicative Risk Premium
The data I’ve used comes courtesy of the repository kept by the St Louis FED, which is an excellent source for those interested in this kind of analysis.
Intuitively, the interest rate spread between AA-rated financials and non-financials should track quietly between extended periods of stability and periods of intense fluctuation in which the yield of financials jumps relative to that of non-financials.
And this is the historic pattern over the past 25 years of the data, which I’ve displayed below. I have labelled some of the key events of that time.
Which is consistent with the picture of the typical market stress measure used by dealers:
(LIBOR – SOFR OIS) = Indicative Risk Premium
Using the slightly shorter dataset available for 3-month USD LIBOR versus SOFR OIS (as implied by the FED’s Term-SOFR proxy).
I then compare the various standard statistics of the indicative risk premia, the results of which are displayed in the following table:
We find that 3m LIBOR exhibits a consistent margin above AA financial paper, at an average +21.3 BP, which is +13.7 BP higher than AA bank funding spread (as defined by 3-month financial CP) through time.
In periods of market stress LIBOR’s excess premium jumped as high as +89.5 BP in 2008, coincident with the excessive market stress event of the Lehman Brothers collapse of October 2008 (the GFC).
Replicating this with the LIBOR – SOFR OIS spread we find the excess premium somewhat more pronounced and more volatile.
Incidentally, in looking at the 3-month LIBOR – SOFR OIS data (term-versus-term rates) the five-year mean of rates to March 5th, 2021, is slightly different to the ISDA Credit Spread Adjustment announced that day; 28.2 BP versus 26.1614 BP for the ISDA. This is due to the difference between the SOFR compounded in-arrears calculations of the ISDA (backward-looking) versus the implied Term-SOFR of the SOFR-OIS rates used here (forward-looking).
Major stress events mapped
To identify historic market stress events consistently, I search for periods of more than single days where 3m financial rates exceeded non-financials by one standard deviation or more (i.e., greater than +23.6 BP (7.6 + 16.0 BP)).
The following bar graph looks more like a commercial bar-code, but each blue period is an individual stress period identified using this approach.
The dominant stress events of the GFC bouts (both 2007, and 08) are clearly shown.
Stress since 2006
Across the 25 years of St Louis FED data there have been 11 credit stress events using this simple measure, however only 10 of these seem viable.
Viable?
The stress event of the March-May 2020 COVID shock froze US commercial paper markets at a time when the FED was easing rates very aggressively, resulting in a very brief inversion of the normal stress indicator (because CP rates lagged the FED move). I have therefore chosen to ignore the 2020 COVID ‘event’ as an obvious false flag.
Statistics from the ten viable/real stress events for our financials versus are presented in the following table:
With the resulting simple averages:
When we compare the excess premium during periods of market stress to the simple averages of excess for the full data set, it suggests 3-month USD LIBOR carries between +8.7 and +13.8 BP of additional premium compared to basic CP funding.
Interestingly, for the early period of the data (the credit stress free 1997-2006 period) the average financial less non-financial spread was +3.2 BP (10 BP lower than the average of the entire 25 years). From 2006, in a foretaste of the GFC stress, a sector-wide repricing of bank risk commenced.
Credit sensitivity acceptance
What does all this tell us?
For those seeking or willingly accepting rate exposure to a LIBOR-like replacement for USD LIBOR the data analysed here is worth more than passing comment:
· LIBOR credit sensitivity has consistently exceeded that of identified financial minus non-financial credit spreads, at an average premium of +13.7 BP to CP since 1997, and +17.7 BP to risk-free (SOFR OIS) since 2001.
· During periods of market stress, historically of 47 business-days duration, the LIBOR minus CP premium has averaged +22.4 BP,
· During those same periods the 3-month LIBOR minus SOFR OIS premium averaged +31.5 BP.
· As past FED studies have indicated, it’s not clear that LIBOR has reliably tracked actual bank funding.
· New reference rates such as Ameribor, AXI, BSBY and CRITR/CRITS may behave differently compared with each other and LIBOR because they are based on different inputs.
While we remain benchmark agnostic with good reason, it’s worth the parties to new deals understanding these facets of credit-sensitivity, and at least asking whether exposure to it is advantageous.
Summary – what does this mean for me?
Averages are often quite misleading and ignore the extreme events and the impacts on us and our actual outcomes. Stress events in markets, as described above, can be defining moments and have significant implications.
We firmly believe that firms should consider:
· the impact of extreme market events and the implications for funding and/or returns;
· the choice of reference rate and whether this is appropriate for their needs; and
· planning for extreme market events and how their choice of reference rate will be likely to perform.
With more choices for reference rates and the high likelihood of markets diverging from averages periodically, our work with clients has focussed on the impacts and strategies to manage the financial implications.
Derivatives and USD Term Rates
This article continues the series on the practical uses of term risk free rates (Term RFRs) in USD. Previously, Ross and myself have both looked at the USD term rates where we are seeing significant interest in Ameribor, AXI, BSBY, Term SOFR, CRITS and CRITR. They all have uses and many market participants are actively looking at the way these new reference rates can complement[RB1] compounded SOFR and provide a more appropriate solution to their requirements.
In this article I look at derivatives and how these important hedge products are being introduced into markets referencing the term rates.
Derivatives referencing compounded SOFR
The market for interest rate swaps and cross-currency swaps referencing compounding SOFR is well established now and supported by price-makers and some end users.
However, the basis swap market is still developing with some swap pairs liquid while others are less so. Examples with significant liquidity include:
USD LIBOR/SOFR; and
Fed Funds/SOFR.
Some examples of developing markets with identifiable pricing include:
BSBY/SOFR; and
Ameribor/SOFR.
Other reference rates such as AXI, CRITS and CRITR have some pricing transparency, but it has been somewhat more difficult to quantify at this early stage.
Basis swaps are not static and can vary considerably. The chart below shows the last 11 months (since BSBY has been available) of the USD SOFR versus other reference rates. Since there is some volatility (noting Fed Funds is on the secondary axis) many users may wish to hedge the exposure!
Basis swap markets are essential and provide different market participants the ability to transform their current risks to something more appropriate. For example, the combination of a fixed-rate debt issue by a corporate could be converted to and Ameribor reference by a combination of a Fixed rate to SOFR swap and a SOFR to Ameribor basis swap.
Of course, this can be delivered by a bank in a single swap, but the pricing would typically use the combination if a fixed swap and a basis swap to achieve the outcome for the corporate and a hedge in the inter-dealer market.
Derivatives referencing Term SOFR
Derivatives referencing Term SOFR are available for end-users. An active and available derivatives markets is essential to provide effective hedges for loans, debt issues and investments.
But there is a catch: the inter-dealer market for Term SOFR derivatives does not yet exist. Risk.net has a great article on this problem titled ‘Term SOFR restrictions spark valuation debate’. The Fed and ARRC guidelines restrict the use of Term SOFR derivatives and effectively do not allow inter-dealer trades.
This means dealer books can be one-way with the end-holders and there is no way to clear this risk in a similar way to other derivatives: i.e., hedging with other dealers with opposite positions or views.
There are no on-screen prices, so dealers are obliged to use internal models to replicate the valuation curves. While this is somewhat trivial (with the Term SOFR and compounded SOFR forward curves theoretically identical) there can be small differences due to hedging, collateral and one-way adjustments which need to be considered when building the curves.
In addition, if Term SOFR derivatives cannot be identically hedged then they may be consigned to the ‘structured’ book within the dealer firm and attract additional reserves and capital.
Either way, the absence of the interdealer market is very real and has consequences for all market participants who may use these derivatives.
Why does the absence of an inter-dealer Term SOFR derivatives market matter to end-users?
It matters for at least 2 reasons: cost and availability.
Firstly, dealer costs would very likely be higher for Term SOFR than for compounded SOFR. This is based on the provisions in the structured book and the higher capital applied to the trades. These costs will be passed on to the end-users.
Secondly, as dealers develop larger one-way books then internal risk limits may start to restrict the number and size of additional trades that can be added to the book.
The inter-dealer market could address both issues by allowing dealers to offset the risks with each other. If this cannot happen, then end-users must expect greater cost and potentially product restrictions and shortages which will likely lead to price inflation (which is quite topical in the economic sense).
What are we observing and hearing?
There is very clear demand for term rates from many buy and sell side consumers of reference rates. The reasons for the preferences of term rates over compounded SOFR are varied and include:
Current processes for operational management are difficult when you only know the refence rate at the end of the period.
Cash management is done well in advance for many firms and arranging for settlements and cash needs more than a few days to organise.
Accounting and planning processes are normally conducted when the rates and cashflows are known in advance of settlement.
Changes to interest rates (e.g., the recent Fed tightening) during a period can impact the costs and make planning for offsetting income more difficult to anticipate.
We often hear the expression ‘need or want’ applied to term rates. While many system issues can be resolved and remove part of the ‘need’, many process and planning issues remain and still represent a considerable ‘need’ as well as ‘want’.
Summary
The Term rates are gaining in popularity and use as more market participants appear to find them appropriate for their businesses. The loan and debt markets are expanding their use of term rates, and this shows no signs of slowing.
But there could be additional costs and restrictions. End-users will have to weigh the costs against the benefits for term rates in many processes and situations where they are appropriate.
Sell-side firms will likewise have to carefully consider how they price and manage the term rate risks while still offering a competitive product to their customers.
Assessing the Term Rate Menu
John recently updated the case for alternative US term rates, focusing on the emerging alternatives to compounded SOFR. Market evolution has seen at least six rather new alternative offerings emerge, all of which provide welcome, forward-looking competitors to compound or Term SOFR.
In this blog I add to John’s work by sharing some of our questions and listing some of the distinguishing features that make these evolving rates all subtly different. I also look at some of the recent pricing relativities, using the period of market stress generated by the Ukraine-Russia conflict.
We believe market participants need to understand the new offerings, particularly since, as John noted, turning a whole market from LIBOR to compounded SOFR is not without challenges, and for some a transition to backwards looking rates may not even be possible.
Questioning the Menu
Use of the Swedish ‘smorgasbord’ as a metaphor for the emerging alternatives to compound or simple risk-free-rates was very deliberate in my blog of October 2021.
The point?
Financial markets are proving different post-LIBOR; moving from the familiar, ubiquitous, and operationally simple system that LIBOR delivered, to a multi-rate environment, with a menu that includes the opportunity to avoid heightened credit sensitivity if one wishes to do so.
So, what is on the term rate menu?
At Martialis, we’ve followed the evolution of the menu with great interest from the ring-side seats of the consultant. Our part is not played as end-users, and we’re not financial product providers. Properly, we are benchmark agnostic, but we have been watching this space closely and from our vantage-point the need to amplify our call for end-users to better understand the new array of benchmarks continues to be real.
Questions
Distilling our view is best achieved by recounting some of the various questions we have raised ourselves, or that clients have raised and to which we’ve responded.
What follows is not an exhaustive list but are those we feel are of salient importance.
QUESTION: Under LIBOR the credit margin for any given credit was easily distinguished, simple to convey, and readily comparable among adjacent credits. Borrowers were easily able to compare quotes in any RFQ. In a multi-rate system, where each benchmark has distinctive characteristics within a mix of credit sensitive and insensitive alternatives, how do firms properly assess the price of a distinctive credit?
This appears almost impossible to answer at present. Markets will likely take time to adapt and settle within the multi-rate environment, and we can’t discount the possibility that a particular benchmark may eventually become dominant.
What we know is that some providers continue to lean on the ISDA Credit Adjustment Spread methodology to engineer a LIBOR-like base, essentially placing a transition-tool fixed spread based on a quite aged fixed five-year lookback of LIBOR-OIS basis.
It seems to us unlikely for this to continue indefinitely. Bank funding costs are a moveable feast, as the COVID and Ukraine war credit-stress events so ably reminded us (see graph below).
What we fear is a dilution of pure credit risk may take hold. Any blurring of the lines between sector credit risk and individual credit price is a bad outcome. Efficient capital allocation is not advanced in a system in which credit risk is blurred.
QUESTION: We are seeing some use of Term-RFR plus the fixed BISL spread across corporate finance; does that mean borrowers are realising some inherent advantage exists?
It may be that a perceived advantage exists, and the attraction of Term-RFRs has both an operational dimension (they’re relatively easy to implement), as well as a ‘credit insensitivity’ dimension.
This may prove illusory in the longer run, since banks themselves face variable funding; a reality that can’t easily be ignored ad-infinitum. If an extended period of more expensive bank funding were to emerge it is hard to see how a RFR or Term-RFR plus historic fixed spread could be sustained.
Consequently, we believe some interesting ‘credit’ pricing outcomes could emerge, for example it is possible that for realised yield outcomes the following may hold in some cases, and possibly persist in some markets:
(Term-RFR + Margin X) = (Term-CSRx* + Margin Y),
where Margin X > Margin Y
Note: CSRx = some Term Credit Sensitive Rate.
If this condition were to become common, it would imply that investors/lenders and/or the sell-side were willing to discount credit margins where customers elected to accept a credit-sensitive benchmark. This is a topic we intend to explore in a coming blog.
QUESTION: SOFR Academy (who partnered with Invesco Indexing as administrator) recently published prototype rates of their Across the Curve Credit Spread Index, or AXI, a dynamic credit-spread ‘add-on’ to SOFR. Could the use of such add-ons become common-place?
Note: Any prospective user of AXI that would intend to also use CME Term SOFR in developing an interest rate for Cash Market Financial Products or OTC Derivative Products would require a license with CME Group for use of CME Term SOFR.
Rather than pass comment on an individual alternative, I decided to ask the CEO of New York based SOFR.org, Marcus Burnett, for his thoughts on this question.
CEO Burnett: I think the vast majority of institutional liquidity will go to SOFR, whether compound, simple, or Term-SOFR, and recent data tends to support this.
It’s important to note that AXI can be applied to any variation of SOFR, so end-users have quite wide degrees of freedom as to the underlying convention that best suits them.
In launching AXI we will not create a path for banks to skip over SOFR, which is a key distinction between our offering and other credit-sensitive alternates. A loan that references AXI will also reference SOFR, so AXI is supportive of the SOFR-First initiative.
We believe SOFR plus a higher overall credit margin has the potential to be higher than SOFR + AXI + Credit Margin because banks have to include an insurance premium for their funding costs when the credit spread is embedded within an overall margin. Borrower margins traditionally only increase due to credit events such as a ratings downgrade.
We note that such an add-on affords users to retain their link to what we might regard as the official sector’s preferred rate (SOFR).
Which encourages us to think there is a long way to go before markets settle on a particular favourite, but credit add-ons present a particularly interesting dimension, and we cannot exclude major banks favouring their use. If syndicates were to follow-suit, a period of heightened market credit stress might forge their lasting use.
We’ll continue to monitor this topic with interest.
For those interested in the movement of various USD term rates through the Ukraine war credit stress event, I have taken simple daily benchmark rates and plotted them for Q1, 2022.
It’s interesting to note the relative stress differences, and to highlight these I simply use max-min to give an indicator of comparative credit stress.
We intend to look at credit stress events in greater detail in a coming blog but leave readers to draw their own conclusions on the interesting relative credit sensitivity evidenced here.
For those who’d like to receive a copy of our up-to-date Term-Rate summary paper, please e-mail us at info@martialis.com.au to request a copy.
Developments in USD Reference Rates
I have commented in the past on the emergence of alternatives for compounded SOFR in the USD market. Just as a reminder, compounded SOFR is the recommended replacement for USD LIBOR where the daily SOFR rate compounds this over a specific period and the rate is calculated a few days (typically 2 or 5) prior to the end of the relevant period.
As James Lovely pointed out to me in my previous blog, I neglected to mention Ameribor in my list of alternatives which will be corrected here! My reasoning at the time was that the target user base for Ameribor was quite different to the Martialis blog reader base but it should be included for completeness.
Current alternatives to compounded SOFR
Certain users of reference rates who have traditionally referenced the soon-to-be-discontinued USD LIBOR, can have requirements which do not naturally lend themselves to compounded SOFR. Turning a whole market from LIBOR to compounded SOFR is not without challenges and this transition may not be possible for some users.
The case for alternatives is very real for many users.
Here are the current alternatives (in alphabetical order):
Ameribor (AFX);
AXI (SOFR Academy);
BSBY (Bloomberg);
CRITS/CRITR (IHS Markit);
ICE Bank Yield Index (ICE BYI);
Term SOFR (CME); and
Term SOFR (ICE).
Right now, Ameribor and Term SOFR (CME) appear to be well supported with the other options still developing.
Firstly, let’s have a quick review of the features of the alternatives.
In general, there are 2 categories of alternative: risk-free and credit sensitive. Of the 6 alternatives above, only the Term SOFR is risk-free.
Risk-Free, Term SOFR
A risk-free rate does not include appreciable credit risk. SOFR, and therefore Term SOFR, is a secured lending rate with minimal credit risk.
In many cases, this is a particularly appropriate rate for users. For example, a corporate borrower can set the risk-free reference rate against an observable and defined rate (I.e., one which is close to the Fed target rate) and effectively pass on the liquidity and credit risk to the lender.
Any variations in the actual funding rate will be absorbed by the lender if it differs from the risk-free rate. Depending on the maturity and type of lending product, the cost or benefit will be priced into the lending rate and could be expected to reflect the risk to the lender.
For example, a 3-year bullet loan would simply include the known cost of funds that the lender could access for that maturity and principal. However, a revolving credit facility where the principal and term of the funding are not typically defined when the contract is negotiated, would have to price the uncertainty of the timing and amount of the loan as well as the potential market liquidity at the time.
While the risk-free rate may provide more certainty for the borrower, it may come at a price which reflects the risk to the lender in the product.
Credit Sensitive Rates
LIBOR has a credit and liquidity component embedded in the rate. Look at the USD LIBOR – SOFR spread since early March! It has moved up 30+ basis points. That is credit and liquidity in action.
The other non-SOFR rates have moved up as well and partly replicate LIBOR in performance. While they still replicate LIBOR much more closely than Term SOFR or compounded SOFR there are some differences. We will come back to this in later blogs.
The dynamic nature of the credit and liquidity spread within the credit-sensitive rates may make them attractive to some users. After all, LIBOR did have a use for over 30 years and did not attract any realistic competitors.
Ameribor, AXI, BSBY, CRITS/CRITR, ICE BYI and AXI are all variants of credit-sensitive reference rates. They all manage to include the same aspects of LIBOR which tracked actual market rates but ground the rates in actual transactions (unlike LIBOR).
It is this connection to the ‘real’ world which makes credit-sensitive rates attractive for some users. It is relevant to borrowers and lenders as the credit and liquidity pendulum swings both ways. Sometimes credit-sensitive rates benefit borrowers (e.g., the past 2 years where the LIBOR – SOFR spread was at a relatively low level) and lenders (e.g., now where LIBOR—SOFR spread is above average level).
Benchmark longevity and fallbacks
History suggests that benchmarks may come and go: LIBOR is a great example. Will all or any of the alternatives survive until the end of a contract?
Although this may be a risk, the management of that risk is something which should be factored into the use of any benchmark. The simplest way to manage this is to ensure each contract has an effective fallback to be used in case of a benchmark failure.
Summary
There is no right answer for which reference rate to use: it all depends on the person and the use case.
It is clear that there are alternatives to compounded SOFR and these are being used in USD transactions. The extent of their use will continue to develop as users are made aware of them and adopt the most appropriate reference rate for their needs.
Four down, more to go…
December 31st could be described as a watershed moment in finance, though it could also be described as an effortless finale. While most of the planet partied, LIBOR benchmark rates that had served markets for over thirty years were consigned to the dustbin, at least almost. While GBP and JPY LIBOR rates continue to be published in a synthetic form, their daily movement is now a function of movement in underlying risk-free rates (RFR’s). That is: they no longer move in line with term interbank lending rates set in the City of London.
That delta is different.
In this blog we look at the cessation event for GBP, JPY, CHF and EUR LIBOR, and demonstrate that while a great deal happened, with around U$180 trillion worth of financial contracts impacted, the transition-event was perfectly kept out of the limelight. Financial markets continued to price, trade and distribute risk, functioning almost seamlessly throughout. In our view global regulators got the cessation right and should feel a sense of achievement at the overall result for such a complex endeavour.
The local post-cessation scene
Over the past three years we’ve had a lot of engagement with firms across Asia-Pac in the LIBOR-cessation field, in several cases been engaged consulting to major transition-projects. With the GBP, JPY, CHF and EUR-LIBOR cessation moment having passed we asked these firms how the transition up to and through cessation finally went.
The following Q&A has delivered some anecdotes worth sharing, and not simply for the sake of promoting FCA bragging-rights, but worth it given the world’s biggest LIBOR cessation moment is still to come: USD-LIBOR, on 30th June, 2023. It’s in this context December 31st could be considered a well-planned and executed dress-rehearsal.
Q1. Was there any disruption experienced in your markets and/or corporate-finance business over this first LIBOR-cessation event?
Answers were strongly uniform, with all respondents noting that the transition was a resounding success, and relatively minor disruptions were worth mentioning:
· Very few firms reported issues with clients, though some client positions in corporate-finance (loan) deals remain outstanding, pending a transition prior to their next reset date.
· The somewhat limited corporate/buy-side adherence to the ISDA Protocol proved no barrier to transition, which we take to mean that most firms managed to strike bilateral agreements in sufficient time.
· The large stock of cleared derivative transitioned through CCP’s in early-December proved of significant benefit given this reduced risk at cessation, and gave important markers to risks at the actual cessation-date.
· A majority of firms transitioned their internal trades, a surprisingly large stock in some cases, well prior to Dec 31st.
· Interbank counterparties overwhelmingly elected to rely on CCP rules, or ISDA fallbacks to transition deals struck with other banks, a point we had predicted.
· Some firms reported that a small sub-set of all-trades had to be manually handled to ensure accurate booking, reconciliation, and representation of risk.
· Most firms noted that the need for pre/post-cessation trade reconciliations placed significant pressure on their operations teams at a difficult time of year (an issue we’d noted, but one that will not be so relevant in 2023).
To cap this off, from what we make of market moves across the year-end there wasn’t the slightest hint of market disruption.
Well laid plans came to fruition impressively.
Q2. Where your firm holds positions mark-to-market, did position-deltas, market risk-factors, and regulatory risk-gauges make sense in the immediate cessation aftermath?
Again, respondents reported no significant issues in their trading books, either in front office book management or risk management oversight:
· Several firms noted minor booking adjustments were required, but were immaterial.
· Revaluation processes and marks were not impacted.
· Money-market brokerages continue to quote USD-LIBOR curves, and revaluation issues have not arisen.
· Major risk systems providers were noted as having worked tirelessly to ensure IR trading systems could handle auto-transitions, and resulting risk-factors were well tested.
We credit this last point as being a helpful institutional feature of modern finance where standards tend to develop with a healthy degree of cross-pollination among major firms.
Q3. Was there a material reliance on Synthetic LIBOR?
Back in October I blogged about the FCA Consultation CP21/29, and how it had the potential (but not the intent) of opening floodgates on the permitted use of Synthetic LIBOR. I inferred that the FCA appeared to be going well beyond the market expectations of the time. I noted that the “proposed decision” (of CP21/29), as subsequently confirmed, “should not be read in isolation,” and that “significant practical constraints to reliance on Synthetic LIBOR remain.”
We still believe this is the case, so it was interesting to note firms had sporadic use-case examples of reliance on Synthetic-LIBOR.
· Several firms mentioned customer requests for use synthetics in loan contracts.
· Others noted modest requests for use in interest rate Cap & Floor trades, which in our opinion is quite understandable given the unique complexities of these products (versus swaptions).
· Firms appeared to have spent considerable time ensuring that customers who retained an economic link to Synthetic LIBOR did so willingly and after developing a full appreciation of the risk reward trade-off of doing so.
Q4. Other anecdotes?
· The late-year customer crush was definitely evident, but manageable for all.
· All firms noted the extent to which use of fallbacks dominated the path for customers and interbank counterparts, most agreeing that this was larger than regulators and planning estimates had suggested.
· Several firms noted the use of facility side-letters where customers elected not to do full-blown deal transitions for multicurrency facilities, a very reasonable approach where drawings in ceasing currencies were unused, and likely to remain so.
Q5. What happens next?
· Most major firms continue to monitor the LIBOR-reform space closely and have ongoing project-team involvement, indicating this is likely to remain the case to around mid-year.
· The end of free-use of USD-LIBOR for regulated firms is a point of high focus, with clear regulatory pressure for firms to cease all bar necessary customer hedging activity, and we note that since forward deltas for USD past June 2023 are an increasing function of forward SOFR, firms may start to more willingly accept basis-risk in hedges where that does not break hedge accounting relationships.
· Firms expect to conduct project post-mortems, and a couple expect regulator engagement on this topic in the first half.
· Local firms are closely watching developments with respect Canadian Dollar benchmark reform and the cessation horizon for CDOR, a market based on bankers’ acceptances.
And global markets didn’t blink
Since the first regulatory-hints that LIBOR reform may be necessary, regulators sought to avoid demanding anything that might result in market disruption. At every stage reform needed to keep markets calm. Typically, regulator fears were couched in terms related to the extent to which LIBOR is or was ubiquitous in finance: with over U$400 trillion worth of contracts likely impacted.
In our view this first major cessation event went particularly well; a credit to the no-disruption mantra, and we find no evidence of market disturbance across the classic FICC suite of markets, nor of course in global equities. All was as it should have been.
And the path of LIBOR-to-Synthetic rates through transition was smooth, though many will question why GBP LIBOR rates moved so much as they finally became synthetic (a function of Term-SONIA plus ISDA Credit Spread Adjustment):
3-Month LIBOR rates in basis-points, 31st December to 4th January
John Feeney touched on the possibility of a step-function over the new year in his blog of July 20th The Potential Wall at Cessation — Martialis Consulting.
The reality is that while there was a one-day cessation step, and visually a seemingly high one given the incredible liquidity sitting in short-term GBP instruments, those market players whose risk-systems modelled the step accurately would have had no X-factor revaluation gain or loss across the new year. Our pre-Christmas checks showed the market got this right. In very simple terms, the step was priced (or mispriced if your system wasn’t up to handling it) well in advance.
What now?
So, what can we expect as we move further into 2022?
Our role is to consult, not play at forecasting. But having been so close to markets in LIBOR reform it’s hard not to ponder and posit some thoughts around trends that seem likely to prevail in the year ahead.
1. USD LIBOR usage will not simply go away, and we see evidence of this with the ongoing ability for non-regulated participants to continue to trade in a range of venues. Discussing this with clients, it seems many continue to grapple with how to curtail all-bar essential customer-driven turnover. This will be an important consideration as regulated firms work at meeting the well-flagged expectations of the regulators.
2. Though we note ISDA reports 2 GBP LIBOR trades actually dealt in the first week of January, GBP SONIA and its term variants are settling down very well as market standards for UK finance. We expect this to solidify through 2022 with SONIA cementing its place as the dominant benchmark in UK finance.
3. By comparison, we see significant evidence of a dispersion of interest and viewpoints in competing benchmarks across USD-finance. We expect an ongoing debate as to the desirability of credit-sensitivity, see SOFR Academy, FactsetFactset, and RiskRisk.net. With the availability of BSBY, compounding with an eighteen month lead time to full cessation, there is a lot of debate still to be had in US markets, which we expect will take more than 2022 to resolve (if ever). There is plenty of scope for more than one benchmark.
4. Domestic markets are likely to start considering the benchmark reforms now underway in Canada, and the implication of post USD-LIBOR transitions in a market that is not unlike that for BBSW. This can be expected to be most pronounced with respect the 1-Month BBSW rate, where the lack of deep activity is already quite well-flagged.
Which brings us to the question of outcomes, and economics, and thus pricing.
Finally, the benchmark smorgasbord needs to be well understood
I’ve written previously about the range of benchmark choices available to market participants as we move past cessation.
Considering the situation in the US, the demise of USD-LIBOR has brought forth a real range of alternates beyond the regulatory favourite (SOFR, compound or simple interest in-arrears):
Term-SOFR
With/without a specific credit add-on – e.g., AXI
SOFR-in advance
BSBY
Ameribor
What’s clear is that five or more replacements are vying for attention in US markets, and markets are unsure about all of them. As these evolve it’s hard not to see market auctioning processes leading to real changes in the spread-to benchmark for each variant. If it were possible to imply an absence of further banking system stress (it isn’t) credit sensitive benchmarks should closely resemble SOFR/Term-SOFR at point of price.
Thus, as we’ve made a point of saying: end-users really need to understand the nature of the benchmark they are contracting to. How it performs over the long-haul matters.
Using Compounded Rates – A comparison of different methodologies in high volatility
My previous blog looked at the ways to use SOFR to calculate a term rate. Because SOFR is an overnight rate, users are obliged to transform this into a term rate such as 1-month to align with the settlement frequencies on financial products. This is typically done by averaging or compounding SOFR over the relevant period.
This blog looks at the impact of different methodologies during the periods of high SOFR volatility in 2008, 2019 and 2020. These dates were selected because they have large moves in SOFR over a short time frame resulting in significantly different outcomes for the compounded term rate. While this may be a relatively rare event, should we enter a period of increasing US overnight rates (Fed Funds) then this will be reflected in SOFR. Even minor differences in the methodology can lead to significant differences in outcomes.
Of course, you can set the SOFR rate upfront in the same way as LIBOR. This is Term SOFR and is available for licensing from CME. I will look at developments in Term SOFR in following blogs.
The term SOFR (small t) here is set in arrears once the final SOFR input is known.
A recap on the methodologies
I used this table in the previous blog to describe the basic methodologies used in creating a term SOFR.
In the previous blog I showed Lookback and Observation Period shift are quite similar in outcome for each hi.
In this blog I look at the Payment Delay and Lookback options only to show their relative performance in different conditions and for different lookback days.
2008 – the GFC effect
SOFR was not published until 2018 but the NY Fed has kindly provided a set of proxy SOFR rates derived from market inputs from much earlier dates including 2008. While this is not actually SOFR, it appears to be a good representation of the repo rates (SOFR proxy) from 1 August 2008 to 31 December 2008.
The following chart shows the rates over that period followed by 2 tables with the results for 17 September and 19 September 2008. I selected these 2 dates to demonstrate how a few days difference in the number of days for the lookback can give very different outcomes for the 1 and 3-month term rates.
SOFR was quite flat around 2.00% until 11 September 2008 (LHS red circle) when it spiked on 12 September and fell to 0.25% on 17 September. It then bounced to 1.82% on 19 September (RHS red circle) before continuing a volatile fall to near zero by 31 December 2008.
The grey shading for the 5-day lookback demonstrates how volatility can impact the outcomes for the methodologies.
Both the 1 and 3-month term SOFR rates differ because the input rates to the calculation are very volatile, and the difference of a few days can include or exclude outliers in the data set.
2019 – the SOFR spike in September
SOFR was setting at around 2% for most of September 2019 but spiked to 5.25% on 17 September on technical liquidity issues. While this was unusual it still had an impact on the outcomes for term SOFR rates which did or did not include the spike. The chart and table are below.
The payment delay does not include the spike but the lookbacks both have this feature in the data sets. This results in a 17-19 basis point increases in the 1-month term SOFR and 8-9 basis point increases in the 3 -month term SOFR.
Again, we see the impact of the choice of methodology on the outcomes for the term SOFR.
2020 – the COVID-19 impact in March
This period was quite volatile and presented 3 rapid falls in SOFR from 1.6% before 2 March 2020 to 0.00% by 18 March 2020. The red circle outlines the largest of these falls in SOFR between 13 and 16 March where SOFR moved from 1.20% to 0.26% over the weekend. The impacts are captured in the chart and table below.
Yet again, the methodology does matter. The 1-month difference between the payment lag (o days) and the 5-day lookback is almost 25 basis points! The 2 and 5-day lookbacks both include SOFR rates at 1.6% while the payment delay does not use that data. The outcomes are as expected with the lookbacks showing higher term SOFR rates than the payment delay.
Does this matter?
Last time we saw that the choice of methodology was not important when SOFR is not volatile: this has been the case in the recent past.
But the methodology does matter in periods when SOFR is volatile (2008), subject to a liquidity event (2019) or has a sudden change based on Fed activity (2020).
In these cases, we can see that the use of a 0, 2 or 5-day lookback can have very significant impacts on the term SOFR rate applied to a contract or trade.
Read the fine print carefully and align your hedges
Although the actual methodology does not significantly impact the term SOFR rate when markets are calm and the Fed is not moving the target band, this is not always the case. We have seen in this blog that a simple change of the lookback days can lead to important and meaningful differences in the final term SOFR.
So, as we stated in the previous blog, read the contractual terms very carefully, it does matter! And make sure you align the hedges as well.
Do you really want the risk of a mismatch in the timing of the SOFR data used for the calculation of the term SOFR? As we have seen, this can matter a great deal and can be eliminated by correct alignment of the hedge to the risk.
The next blog will look more at CME Term SOFR and how it performs relative to the compounded term SOFR.
Using Compounded Overnight Rates – a basic guide
As markets move towards overnight rates and away from LIBOR (and other IBORs) there are a variety of ways to transform theses rates into a simple interest rate that can be used for the relevant period, say 1-month. My previous blog on the alternatives for term rates in USD looked at how these could be used to replace LIBOR.
This blog addresses the ways in which the overnight rates can be turned into a rate which is applied for a period such as 1-month (or 3, 6 etc. months). and how these are accessed in the USD SOFR market example.
As many have noted previously, when you use overnight rates to calculate a ‘term’ rate (i.e. one for that period) the actual result is not known until the end of each relevant period. The final rate is not published until the day following the final rate fix and many firms need a number of days to arrange for calculation and payment resulting in a variety of methodologies to accommodate this ‘inconvenience’.
Note that the term rate is not a Term rate such as Term SOFR. A Term rate is set at the start of the period whereas the term rate is not known until the end of the period. The terminology is a little confusing but it is commonly used.
Some basic concepts
Using SOFR as an example, the basic compounding and averaging calculations for each period are:
Some common methods for allowing for payments to be arranged
As mentioned above, the final SOFR rate is not known until the day after it is set and many firms need a few days to arrange for payments. The more common variants are payment delays, lookbacks (observation shifts or ‘lag’) and observation period shifts (‘shift’).
Each variant is in current use and often one is favoured for a particular product.
The difference between the lookback and the observation period shift is important.
A lookback preserves the relevant period and simply uses the ri from a certain number of days prior. This can have the disadvantage that the ri intended for a particular number of days may not be applied to that number in the calculation. This is the main reason a 5-day lookback is often favoured because it has a better chance of alignment than a 2-day lookback.
The observation period shift avoids this problem by moving the whole calculation back a number of days thereby ensuring the ri continue to align with the correct days.
NY Fed SOFR, SOFR Averages and Index publication
The NY Fed publish the Daily SOFR and a number of other calculations each day at 0800 ET. These are excerpts of the screens.
The SOFR rate (0.05%) is per day in the first screen and the 30, 90 and 180-day averages plus the index is also provided.
The ‘Average’ is actually the simple interest rate calculated using the compounding methodology but is referred to as the average. This can be confusing but it is accepted by markets.
The index is the discount factor calculated each day using the SOFR rates and measures the cumulative value since 2 April 2018. The index can be used for the observation period shift methodology by finding the dates and using the formula below. Note the index cannot be used for a lookback because the dates are no longer aligned to the appropriate SOFR rates.
The following tables are snapshots of the NY Fed pages for SOFR and then the Averages and Index.
Does this matter?
The simple answer is ‘not at this time and with these rates’.
In this table I have used 8 decimal places to show there is little difference in the rates. Most product and contracts round to 5 decimal places so there would be no difference. But be aware that the NY Fed Averages are actually the compounded 90-day and do differ from the more commonly used 3-month period.
It is also worth noting that the compounded and averaged calculations of the rate are different.
Read the fine print
It is clearly important to know if the methodology is compounded or averaged because this will almost always lead to different rates. Both calculations are used in USD products.
Likewise, it is important to use the correct dates: 90-day or 3-month as this also matters to the final rate.
So, read the contractual terms very carefully, it does matter!
The next blog will look more into how the methodologies can impact the rates when rates are SOFR higher and/or have greater volatility. In these cases, the choice of lookback or observation period shift can matter.
Lessons on the road to cessation – the JPY capital markets
In our last post we resorted to a smorgasbord metaphor to help carry the point that in the post-LIBOR world market participants have so many alternate benchmarks they should carefully evaluate for choice. Given some of the feedback, we thought it worth following up with a close look at progress in Japan. It’s a jurisdiction with a massive savings pool where benchmark choices appear somewhat different to USD and GBP.
What we found was more than interesting.
The USD market – real progress since 2019
On July 31st 2019, the world’s first SOFR deal settled; JP Morgan was issuing another piece of its regular preferred stock offerings. As Risk.Net sagely noted at the time, the issue had a “first-of-its-kind provision buried in the small print.” What they were pointing to was that the floating leg would “pay a forward-looking term rate based on SOFR, the secured overnight financing rate.” The $2.25 billion deal was set to pay a fixed-coupon of 5% through to August 1, 2024 thereafter switching to “a floating rate of three-month term SOFR, plus a spread of 3.38%.”
Which was quite brave of JP Morgan; term-SOFR had not yet hurdled regulatory approval let alone seen the light of day, and there could have been no certainty regarding the future viability of Term-SOFR.
Today, USD issuance in a SOFR format is more common than LIBOR, so much so that the CME run a SOFR Issuance Tracker, offer a wide suite of futures, and OTC market-makers have followed the obvious trend. Market conventions appear to have largely settled.
Looking into issuance data, what we note is that the hallmarks of a successful USD product transition are falling into place. Likewise, UK markets have carried-off a similar transition to their SONIA-based future, with markets now almost entirely absent GBP-LIBOR dealings. Thus, the crucial USD and GBP-based capital and financial markets appear to have the necessary features that promote their proper functioning:
For the sake of building up a point of comparison from which to judge the ARR scene in Japan, we review USD issuance since mid-2019.
Which to us is an impressive backdrop.
The Yen market – limited progress
In our comparative look at Japanese markets, we asked a simple question; what’s been happening in the market for JPY alternate reference rates (ARRs)?
Here is what we found for the same period (Mid-2019 onwards):
The Bank of Japan’s Cross-Industry Committee on Japanese Yen Interest Rate Benchmarks was established in August, 2018, and in November 2019, Quick Corporation’s forward-looking TORF benchmark was announced as the rate “most supported by public consultations.” The availability of an OIS-based TORF (in production since April 26th 2021) offers market participants a Japanese Term-SONIA lookalike in most respects, and it should be operationally easier than compounded alternates.
And JPY issuers have a range of other choices they can tap:
TIBOR, the Japanese Bankers Association administered rate, has been calculated and published as "Japanese Yen TIBOR" since November 1995 and is NOT scheduled to cease when JPY-LIBOR does.
Compound or Simple Interest TONA, could be thought of as Japan’s version of SONIA (unsecured, unlike SOFR), the modern form of TONA was introduced in 2016 and is administered and published by the Bank of Japan.
Digging a little further, we found that the single instance of TONA use in issuance was actually deferred until September 2026, coming as it does in the form of a convertible security issued by Mitsubishi Corporation:
What this points to is that notwithstanding the Mitsubishi callable (which we note could be called before it ever converts), there is currently not a single floating issue referencing compound or simple TONA, and none referencing TORF whatsoever (further queries on this with major data providers confirmed that there have been no TORF security issues).
We will follow-up this analysis with a look at JPY-based corporate and syndicated facilities in coming posts.
Digging still further we took a deeper look at historic behaviour of the various choices, and noted that the annual volatility (a favourite measure, expressed in basis-points of underlying yield) of different ARR’s, including those where a particular convention might be used can be vastly different.
Consider the following distinctive range of volatilities over the past 10 years data:
JPY-LIBOR is clearly the most volatile, which is not unreasonable given LIBORs credit sensitivity, whereas TONA variants and TORF do not. But this opens the question of: why does TIBOR not travel at a similar volatility to JPY-LIBOR? And looking across time, why has TIBOR remained where it is despite the collapse in JPY-LIBOR and TONA to negative rates?
This suggests that those looking to Japan’s capital markets need to tread carefully around benchmark liquidity as well as expected benchmark performance, and those offering JPY drawings have to be able to find appropriate points of funding.
Are Japanese institutional features at play here?
To help us answer the question of why JPY markets have largely ignored the development of ARR’s we turned to macro-economic texts and some research on the features of Japanese finance.
These indicate that there are a range of features at play:
Existing Bank of Japan interest rate settings continue to compress short-dated JPY yields, e.g., the JPY OIS Curve is negative to 4-Years, 9-Months
Forward yield projections are partially implied by the extent to which the Bank of Japan is expected to hit its “Price Stability Target” of 2% at a time when Japanese Core Consumer Prices have been negative for all but one of the past twelve months
The size of the household saving ratio in Japan (estimated at 11.4% percent for 2020) is the highest since 1994, adding to the existing Japanese savings pool
To the extent that Japanese Corporations require funds there is a long-run preponderance of bank funding over bond funding
To gain a stronger feel for whether these features are right we turned to a Japanese based interest-rate dealer, Stuart Giles a Japan-based interest rate dealer
RB – Stuart, you’ve traded Japanese markets and rates for an extended period, can you give us a feel for why Japanese floating-rate issuance is exceptionally low?
SG – Traditionally there’s been an over reliance on bank funding vs domestic market issuance, but this has changed marginally in the Covid-19 era. Demand for floating-rate paper remains low as it provides yield starved local investors little in terms of carry or potential for capital gain. With essentially no prospect of higher floating rates here I don’t expect this to change. Floating-rate issuance also serves no practical purpose for hedging ALM mismatches
RB – Can you see the TORF benchmark developing momentum and liquidity in the manner of say Term-SONIA or SOFR?
SG – With the recent pickup in liquidity in TONA, this increases the robustness of the TORF benchmark given that TORF is based on the uncollateralized overnight call rate. There are still some important issues with the use of TORF to be addressed such as the establishment of the governance structure and an improvement of transparency in the calculation process however.
RB – If you had to guess, what would be the prominent Japanese benchmark in 10-20 years’ time from now?
SG – A further shift away from TIBOR may require additional pressure from regulators, so near term I’d expect it remains the dominant reference rate. The momentum is clear however that risk-free reference rates are the global norm, and Japan will likely keep moving in this direction also. TORF is well designed with advance fixing and thus predictability of cash flow like Tibor/Libor fixings. Once we have the availability of cleared TORF products I think this benchmark can become the dominant reference point for markets.
Lessons
Our main take-away from these surprising JPY findings is that big assumptions can be quite dangerous. The Japanese economy and it’s institutional features and pressures have all ‘conspired’ to ensure that Japan’s capital markets are different to GBP and USD, and other currencies, and thus the path to JPY-LIBOR’s cessation is quite distinctive.
For those considering tapping one of the world’s largest savings pools the lesson here is that it will pay to fully understand the range of factors that drive your outcome; which is likely to rest in the fundamental nature of the underlying benchmark.
It really is a Smor Gås Bord…The decision on which post-LIBOR benchmark to use
In this blog we explore the increasing degrees of product-freedom available across post-LIBOR financial markets and the corporate finance suite. With the dizzying array of options seemingly available, we pose important questions for market participants to consider.
It’s not quite A Table Alphabeticall, and, somewhat amazingly, the Cambridge dictionary has only been available on-line since 1998, but they helpfully define the original Swedish noun ‘Smorgasbord’ as: a mixture of many different hot and cold dishes that are arranged so that you can serve yourself.
Why on earth is this relevant in 2021 finance? Well, we’ve come to the conclusion that the Swede’s mix of smor (butter), gås (goose), and bord (table) is the near-perfect metaphor for the world of post-LIBOR finance.
We field a lot of finance questions from many different clients, and increasingly note that there is a wide and increasing array of ways to contract new deals. All of these have quite different ‘gives versus gets,’ and component pieces. Some are refined, some are quite raw and participants appear to have to figure out how to serve themselves.
And it may be difficult for bankers to accurately describe what’s on all those platters.
It’s now over four years since Andrew Bailey gave his momentous ‘The future of LIBOR’ speech. Despite the opportunity that many of us expected that LIBOR reform would bend market conventions toward conformity, the lived reform experience has been of the market response leading us to a world of far greater complexity.
Corporate finance and financial markets products are not going to click together with the simple LEGO-like efficiency of LIBOR products: new benchmarks are really quite different. Not just a little different; a lot different (think volatility).
So, what kind of basic questions should one consider when approaching the post-LIBOR deal ‘bord’?
Do I really understand the underlying nature of the Alternative Reference Rate (ARR) I’m about to reference and use in contracts?
How have ARR’s, or proxies, performed in terms of realised outcomes, annual ranges and historical volatility?
What do ARR’s actually do at times of market disruption and heightened market volatility?
Am I willing to accept the credit sensitivity of the financial sector remaining at play in my all-in cost outcome? The credit sensitivity is a feature of LIBOR and ARRs such as BSBY, Ameribor, CRITR and AXI.
Could there be an all-in-rate advantage to accepting credit sensitivity?
s the choice of ARR different for a borrower or investors?
Then, excuse fingers, there are real questions about what type of cutlery is available to help consume all these trimmings?
Term RFR’s seem like a viable solution, but can I access them if regulators impose use-case limits?
If I go term rate (credit sensitive or RFR) and need a derivative hedge, am I sure I can get one?
If I get a hedge, will it work like traditional LIBOR hedges did in all cases?
Speaking of that hedge, what about in three- or five-years’ time if use-case limits change and liquidity dries up?
Cross currency swaps are two-day lagged, my cash instrument is five-day lookback, how am I (and my auditor) going to handle the exposure to basis risk and short-term liquidity?
Can I get a convention match?
We’re not sure what part of the smorgasbord metaphor we can use for the other important considerations, but perhaps these are best described as condiments?
My bankers insist on imposing an interest rate floor, isn’t that now at-the-money?
Shouldn’t that floor also be embedded in the hedge?
What’s this difference between period-floors and daily floors in the RFR facility, and why does it matter?
Bloomberg Index Services Limited rates show some large differences in credit-adjustment-spreads between currencies and tenors in the ISDA fallback conventions, but my bankers want a one-size fits all margin in my multi-currency facility. What the gives here?
To gain a better sense for the practical realities we spoke on this topic with Managing Associate, Yu Zhang, a specialist in the banking at corporate finance practice at leading legal firm, Allens Linklaters in Sydney.
“I think the smorgasbord analogy is quite apt.
The range of possible ARR/RFR conventions is somewhat daunting, then there’s the choice between credit-sensitive and essentially credit-risk free rates, and we think clients need to evaluate each of those areas very carefully. Then, for corporate finance and securities issuance, there are a range of risk-free-rate contractual facets that need to be considered, also carefully given the way these can influence economic outcomes.
Whereas a standard LIBOR corporate facility of old might contain three or four contractually meaningful elements (i.e., that can impact economic outcomes), when we deal with RFR-based deals, there are a lot more.
People should pick their meal from this smorgasbord carefully, and I think that view, that caution, is generally better understood than it was only a few months ago.”
Yu Zhang, Allens Linklaters
What’s dawning on market participants is this: if you attempt major global cross-jurisdictional reforms and invite market regulators, the sell-side and the buy-side, and the loan-markets and derivative markets industry bodies to the same dinner party, not everyone wants a Big Mac.
Which is a recipe for confusion.
For those charged with determining efficient capital costs or maximising returns we think it will be important to be able to cut through the confusion, and soon.
And lastly, try to avoid being the Gås!
Synthetic LIBOR - More devils in the detail?
Counterintuitively, the suite of LIBOR regulatory reform measures has often served to actually stoke market confusion. Judging from client reactions, the FCA’s latest consultation adds to the list of measures in the ‘confusing’ category. In this blog we explore the CP21/29 consultation, place it in its proper context, explain what may be causing the confusing, and work on demystifying it. All is not as it seems!
FCA CP21/29
On 29th September the UK’s FCA issued a number of important market notices, which included a seemingly innocuous consultation paper CP21/29, Proposed decisions on the use of LIBOR (Articles 23C and 21A BMR).
CP21/29 follows on from previous, related consultations:
§ CP 21/15 of May 2021, on the FCA use of critical benchmark powers, and
§ CP 21/19 of June 2021, on GBP and YEN LIBOR settings.
The key to understanding CP21/29 is that it pertains to two quite specific areas of benchmark reform, to use the FCA’s own language:
1. The decision on whether and how the FCA might permit legacy use of GBP and YEN Synthetic LIBOR settings after January 1st, 2022.
2. The question of whether and how the FCA might prohibit new use of overnight, 1m, 3m, 6m and 12m US dollar LIBOR settings after January 1st, 2022.
While we will explore the prohibition of use of US dollar LIBOR settings in a later post, we focus here on the linked questions around whether and how the FCA might permit legacy use of GBP and YEN Synthetic LIBOR settings?
Grounds for confusion?
Since the sweep of LIBOR reforms commenced major global regulators have uniformly urged market participants to deal with legacy LIBOR contracts in two broad ways:
1. amending contractual fallbacks; or
2. conducting active LIBOR-ARR transitions.
For those contracts that might “prove unable to convert or be amended” regulators in major jurisdictions have focussed extensive past-year efforts trying to determine how best to a) define, and b) accommodate post-cessation treatment of “Tough Legacy” contracts.
Exemplifying this, the Bank of England established a Tough Legacy Taskforce within the Bank’s Working Group on Sterling Risk-Free Reference Rates while the UK Parliament passed laws in early 2021 giving the FCA wide powers to deal with this issue. Similar US legislation passed the New York State Assembly in April 2021.
The combination of these various efforts and widespread common use of the descriptor “tough legacy” has not unreasonably led market participants to develop LIBOR-cessation plans on the assumption that the sub-set of products deemed “tough-legacy” would likely be narrowly defined in any jurisdiction. Further, that reliance on a regulatory response for contracts other than those deemed “tough” would be unwise.
Consider the FCA’s language under section 3, The Article 23C legacy use power:
Our proposed decision
3.1 We propose to permit legacy use of these 6 synthetic LIBOR settings in all contracts except cleared derivatives (whether directly or indirectly cleared).
3.2 We do not propose to apply any limitations or conditionality to the above permissions, at least before the end of 2022.
While the CP21/29 is not a foregone conclusion (consult responses are requested by 20th October), LIBOR exposed parties could infer that it hints at a 180-degree turn on the part of the regulator, and we wouldn’t blame them:
§ Far from being narrowly defined, the availability of a Synthetic LIBOR for legacy GBP and YEN contracts will be exceptionally broad.
§ Parties to non-ISDA LIBOR-referencing contracts that have not yet transitioned (in any way) can now consider a third option: doing nothing.
§ Parties that elected to repaper contracts, particularly via common switch provisions, or who have actively transitioned deals already, may feel they were somewhat misled in the lead-in to their decision-making.
These reactions are not unreasonable, but that is not to concede that they’re warranted.
Martialis View
The “proposed decision” outlined in Point 3. CP21/29, should not be read in isolation.
Significant practical constraints to reliance on Synthetic LIBOR remain and market participants should carefully evaluate whether to utilise it, particularly where more robust alternative paths are clearly available.
Synthetic LIBOR could be thought of as a makeshift substitute to robust fallback or active transition, but important considerations, not limited to the following, should be carefully taken into account:
1. The FCA’s preparedness for an all-contract use case is not open ended
While they have the power to compel IBA publication of a Synthetic LIBOR for up to 10 years, the extent of a commitment outlined in CP21/29 is to end-2022, to wit: they do not propose to apply any limitations or conditionality to the above permissions, at least before the end of 2022.
To be specific, under ARTICLE 21(3) BENCHMARKS REGULATION – NOTICE OF FIRST DECISION, of September 29th, the FCA state that: the compulsion period shall be for 12 months starting immediately after the final publication of the 6 LIBOR Versions (GBP and YEN) on 31 December 2021 and before the 6 LIBOR Versions would otherwise cease.
The FCA will regularly assess Synthetic LIBOR (annually is likely) and may discontinue its publication prior to the 10-year maximum allowed.
This should give significant pause for consideration to those who might otherwise accept what presents as an easier path past the end-2021 cessation event.
1. Term-RFR use case limits across derivative markets remain unsettled
While a jurisdictional divide on this question appears to have opened up between the UK and US (and Japan), it’s clear that UK regulators have misgivings over unlimited Term-SONIA use.
For those who may subsequently seek term-rate hedges to replace Synthetic LIBOR in legacy deals, a downstream lack of use-case availability or liquidity should be a consideration.
However, we do note the use of Term SOFR in large syndicated deals in the US is becoming more common.
2. Parties with fixed-rate products
In examples such as the hedged-facilities common across asset and infrastructure-finance, active reliance on Synthetic LIBOR for both facility and hedge (provided it is uncleared) should be carefully considered.
Subject to points 1 and 2 (above) this approach may hold advantages where parties attempt to preserve existing fixed rates, a topic we are particularly familiar with.
3. New products and reference rates
With what appears like a ‘stay on execution’ of GBP and YEN LIBOR the markets will have additional time to develop new products and alternatives which may include term rates. These may be more suitable for some participants.
Firms will need to remain alert to these developments as many end-users have shown a preference for term rates such as Term SOFR and BSBY.
Conclusion
For those who feel they could have avoided transition or repapering costs/disruption, a degree of frustration at the FCA’s poorly telegraphed announcement could be expected. We believe it will be important for participants to better understand this topic, and how it actually adds to choices that need to, like so much in the post-LIBOR finance, be carefully evaluated.
The Potential Wall at Cessation
Is this the way LIBOR ENDS?
As we approach the date when pre-cessation triggers become effective for GBP, JPY and CHF the transition for derivatives subject to the ISDA Fallback Protocol or Supplement may face the impact of hitting a wall where the reference rate changes suddenly.
This has been written about previously at Clarus and by the Murex team and in Risk.net. Then it was a theoretical problem about revaluations and curve construction; now it looks very real! And rather ironically, when this potential issue was first raised, the wall looked like a cliff where there may be a sudden fall in reference rates over the cessation date whereas now it looks more likely there will be a sudden increase in rates.
In this blog I look at the GBP and USD markets and how they are predicting the reference rate changes from LIBOR to the ISDA Fallback Rate on 31st December 2021 and 30th June 2023 respectively.
LIBOR is not necessarily Risk Free Rate plus Spread Adjustment (Fallback Rate)
The ISDA Fallback Rate is the Adjusted Reference Rate plus the Spread Adjustment for the relevant tenor of that period of the trade. Although this is commonly applied in derivatives there is some use in other products such as loans and securities.
The Fallback Rate replaces LIBOR after the cessation date or the effective date of the pre-cessation announcement, in this case after 31 December 2021 for GBP (and other non-USD LIBORs) and 30 June 2023 for remaining USD LIBORs.
Note that the Fallback Rate is a replacement and is not required to equal LIBOR. The LIBOR on the last day will behave as LIBOR always has and be subject to issues of liquidity and credit which may not equal the 5-year median in the ISDA Spread Adjustment.
GBP on 31 December 2021
The table on the right is where the markets are at this time and where they are predicting LIBOR and the Fallback Rate may be on 31 December 2021.
The yellow highlights show the potential difference in the reference rate applied to contracts. The current differences are around 3 bps (1-month), 9 bps (3-month) and 23 bps (6-month).
If we look forward to the market prediction on 31 December 2021, the the differences are smaller but still there, respectively approximately -1 bps (1-month), 6 bps (3-month) and 19 bps (6-month). This indicates contracts with a cessation date for GBP LIBOR on 31 December 2021 could see a jump in rate on 3 January 2022.
But a word of caution, the Fallback Rate is only known at the end of the relevant period and could differ substantially from the market predicted rate.
Note: the forward OIS and LIBOR rates for GBP and USD are derived from futures prices. In particular, the USD 1 and 3-month LIBOR rates may be too low dues to ‘turn of year’ effects.
USD on 30 June 2023
A similar outcome emerges in the USD LIBOR which is destined for effective cessation on 30 June 2023. Again the yellow highlights show the differences now on those predicted on 30 June 2023.
The 1-month difference is relatively minor however there are more substantial predicted differences in the 3 and 6-month tenors.
Potential impacts
Costs for borrowers and returns for investors
the sudden change in rate will give a potentially large difference between the reference rate on 31 December 2021 and 3 January 2022
small mismatches in reset dates between exposure and hedge may create large accrual differences
Modelling the gap
such a discontinuous change in rates is notoriously difficult to model
many systems will attempt to create a smooth curve around the dates which can impact revaluations on many days either side of the change (see the Murex article)
Differences in outcomes
LIBOR fixes at the start of the period while Fallback Rate has daily fixes during the period
the actual outcome may differ from the projected outcome at the start of the period leading to accrual differences
Summary
I end this blog very simply: LIBOR is not the ISDA Fallback Rate and it may evolve differently to the market expectations.
Changes to liquidity and credit spreads will impact LIBOR and the final outcome may not be consistent with the 5-year median approach in the ISDA Spread Adjustment and therefore the Fallback Rate.
So monitor the markets and be prepared for a potential surprise on cessation dates!
And finally a reminder of how much the Spread Adjustment can vary under extreme conditions. This is the GBP 3-month for the year up to June 2020 through the COVID-19 time.
The 5-year median is 0.1193% bit the spread went to approximately 0.80% before falling to its present level.
Source: Bloomberg