One of the most significant recent developments in the ongoing preparations for the cessation of LIBOR is the Bank of England's announcement that it will publish a SONIA compounded index from August 2020.

Here, we'll examine the usefulness and limitations of this index, and consider the likely effects on the wider market of such publication. The Bank of England has also published a summary of the responses from its February 2020 consultation on the index and the possibility of also publishing period averages for the index.

Calculation complexities

The drive by regulators across a number of jurisdictions to guide the development and widespread adoption of risk free rates ahead of the cession of the publication of LIBOR rates in 2021 has been hampered by the seemingly never-ending discussions as to exactly how the various risk free rates should be calculated.

With each new consultation, the extensive number of variables of methodology and complexity of calculations has become ever clearer. For example, an agent or lender seeking to calculate a backwards-looking compound SONIA rate for a three month interest period would be required to identify in the region of 65 separate SONIA rates (taking care to correctly apply the contractually agreed observation period or lag time), correctly weight weekends and bank holidays in that period, and perform a fairly complex series of compounding calculations (all the while correctly reflecting the contractually agreed business day conventions, rounding conventions and day count methodology set out in the underlying loan agreement).

The management time required to give effect to this process, administrative burden, and scope for challenge and disputes is immediately apparent.

Index publication

It was therefore a welcome development when the Bank of England announced that, from August 2020, it will publish an index of compounded overnight SONIA rates.

Starting in April 2018 with an initial principal of 1, this index takes that initial value and compounds it using each overnight SONIA rate so that on any day the index will show the effect of compounding SONIA across all previous London business days since 23 April 2018. This follows the Federal Bank of New York publishing a compounded SOFR index using the same methodology. 

The effect is that market participants will be able to use the index to calculate the compounded overnight SONIA rate between any two specific dates – significantly cutting down the laborious nature of calculating the appropriate compound SONIA rate.

The SONIA Compounded Index for any given London business day will be made available to licensees such as Bloomberg at 9 AM on that business day and will then be made available on the Bank of England's website free of charge on the next business day.

Publication of period averages

The US approach goes one step further; at the same time as publishing their compounded SOFR index, the Federal Bank of New York is also publishing period averages, being single compounded SOFR figures for 30, 60 or 90 days showing the effect of compounding a starting figure of 1 during that period. 

The Bank of England has consulted on publishing period averages, which, on any given day would show the compounded rate over the previous 90 days so that market participants would not have to calculate the compounded SONIA rate between two specific dates themselves. However, for the time being the Bank of England has decided not to publish period averages. 

A neat solution to calculation complexities?

While it is an attractive option to be able to reference a simple screen rate in order to determine the risk free rates, one issue with publishing period averages is the difficulty with aligning the periods of calculation with existing conventions for calculating interest periods. 

The 'modified following' convention that is widely used in sterling markets (including the Loan Market Association's (LMA) suite of documents) means that if an interest payment date would fall on a date that is not a business day, it will move forward to the next business day or if the interest payment date would fall in the next month as a result it will roll back to the previous business day. Thus, a notional three month compounded SONIA average figure published on any given day could have different values. 

The Federal Bank of New York's solution to this is that the compounded SOFR average figures are based on start dates that are exactly 30, 60 or 90 days prior to the publication date irrespective of whether the start date would be a business day, which was one of the solutions suggested by the Bank of England in its recent consultation.

This issue, and the lack of market consensus on the approach to take, was one of the reasons the Bank of England highlighted for not publishing period averages for compounded SONIA. Although the majority (61%) of respondents to the Bank of England's consultation were in favour of the publication of period averages, pertinently all of the trade bodies that responded were against it. A number of those respondents said that, even if published, they would not use period averages in financial contracts, with the trade bodies going to far as to state that they thought the Bank of England should not produce the period averages.

One theme that emerged from the responses was that period averages could lead to confusion and fragmentation in the sterling market, particularly as there are alternative SONIA period averages that are currently published (eg by Bloomberg) at a time when one key imperative of the transition away from LIBOR is consistency and harmonisation of the replacement rates.

The position might be further improved if moves are made to follow the provisional recommendation made by the Infrastructure and Systems Work Stream of the UK Working Group on Risk-Free Reference Rates that a calculator be developed to support the adoption of a SONIA based instrument. We will watch with interest to see whether this recommendation gathers support.

The drive for consistency

A topic that has been extensively debated is the method by which a compound SONIA rate can be calculated prior to an interest payment date, given SONIA is a backwards looking overnight rate varying each day. Therefore, unlike LIBOR, the final rate is only ascertainable at the end of the interest period if the compounded SONIA rate is calculated based on the actual overnight rates for each day within the relevant interest period. 

In order to publish a compounded SONIA index rate, a consensus has to be reached on this debate and one approach adopted over the other possible options. 

To give the payor of the interest sufficient notice of the quantum of the interest payment, the calculation mechanism has to allow for the final rate to be settled in advance of the interest payment date. A period of five business days has been used in most of the SONIA transactions to date (in alignment with the approach taken in the LMA's 'exposure draft', published in September 2019).

There are two commonly-used mechanisms for calculating a compounded SONIA rate in advance of the interest payment date: 

• The 'observational shift' mechanism, which calculates the compounded rate by shifting back the period over which the rate is calculated by five business days, the 'observation period'. This mechanism calculates the compounded SONIA rate by compounding on each business day in the observation period using the overnight rate for that business day.
• The 'observational lag' or 'reset days prior' mechanism, which calculates the compounded rate using the actual number of days in the interest period. This mechanism calculates the compounded SONIA rate by compounding on each business day in the interest period, but uses the overnight rate five business days prior to each business day in the interest period.

Each methodology gives the payor five business days' notice of the final compounded rate (although, in practice, because the SONIA rate for any one day is published on the next business day at 9 AM, the payor has four and a half days' notice).

However, in both methods, the assumed SONIA rate for any non-banking days (ie weekends and public holidays) is set by the rate on the preceding banking date. The observational shift method weights non-banking days by reference to the observation period, whereas the reset days prior method weights non-banking days by reference to the actual interest period.

Although this is unlikely to yield significantly different compound rates, it will produce different results where bank holidays fall within the calculation periods (as different overnight rates will be applied to the non-banking days over the bank holiday), and therefore certainty of approach is key to avoiding disputes and creating liquidity.

The Bank of England's SONIA Compounded Index will adopt the observational shift method, however, to date the floating rate note market (where there have been a considerable number of SONIA-linked issuances) has largely used the reset days prior or 'lag' method. It will be interesting to see whether the publication of a compounded SONIA index drives more consistency of approach across debt products.

In fact, some respondents to the Bank's consultation felt that the SONIA Compounded Index would act as a catalyst for a switch to the observational shift convention for products that currently use the lag methodology. 

The adoption of the observational shift methodology for the SONIA Compounded Index is consistent with the approach taken by ARRC for SOFR and should therefore assist market participants in developing products and internal processes that align across multiple currencies. 

From a borrower's viewpoint, the SONIA Compounded Index should give borrowers and issuers the ability to independently verify compound SONIA interest calculations, thereby avoiding disputes and generally helping accounting and reporting processes. 

Final thoughts

While it may not be the panacea to the LIBOR transition challenges, the Bank of England's publication of a compounded SONIA index is a very welcome development, and should help push the market towards adopting a universal methodology. This will in turn increase the marketability of SONIA products, as market participants can engage in financial products that are underpinned by consistent compound SONIA rates. 

Indeed, HSBC – one of the arrangers on the first syndicated loan linked to both SONIA and SOFR that was made to British American Tobacco in March 2020 – admitted that the decision to use the observational shift mechanism in that facility was motivated in part by the Bank of England's decision to produce a compound SONIA index. Hopefully, this will give lenders greater confidence to write new SONIA-based products in the expectation that the market will adopt the Bank of England's SONIA Compounded Index compounding methodology. 

The Bank of England did not completely shut the door on publishing period averages in the future, particularly if the market becomes more unified on how to define the periods over which the averages are published. Given the majority of respondents were in favour of the Bank publishing period averages, it will be interesting to see if this garners greater support following the Tough Legacy Taskforce's paper calling for consideration of a legislative fix.

Get in touch

Please reach out to a member of our Banking and Finance team if you would like to discuss any of the information covered in this article in greater detail. Taylor Wessing acted for Deutsche Bank on one of the first adoptions of a loan referencing a compounded SONIA rate.