The once hugely popular LIBOR, underpinning approximately US$350 trillion in financial contracts, came under the spotlight in 2012 after evidence emerged that some of the largest global banks were falsely inflating or deflating their rates to profit from trades, or to give a healthier picture of their credit quality.
These manipulations led to financial regulators around the world imposing fines of over US$ 9 billion and revising the basis of calculation of LIBOR, designed to make the process more objective and to minimise the risk of interference.
Regulators and industry groups across jurisdictions have since recommended the adoption of overnight RFRs, offering a robust alternative rate anchored in a sufficiently liquid traded market. RFRs are backward-looking overnight rates, based on actual transaction data.
The transition to daily-compounded RFRs is a paradigm shift, where collating and compounding a multiplicity of rates taken on all days within the relevant period on which such a rate is published will replace a single observation of an interest rate on a screen at the start of an interest period.
As the daily RFR is typically published at the end of the day or on the following day to which it refers, this necessitates a “lookback” or a payment delay at the end of the interest period to ensure that the interest due can be calculated and paid on time.
Up to this point, the lookback has been utilised as a business day lookback, in order to ensure that the first and last day of the observation period fall on business days.
Market participants have so far used several conventions in conjunction with the business day lookback approach, and especially to determine how to weight the rates associated with any business day that precedes a non-business day (whether a weekend or a bank holiday) and for which a published RFR is unavailable.
One such convention is the so-called “Observation Period Shift”, where the interest period is indeed shifted into the past by a specified number of business days in order to derive the relevant observation period, according to which the daily rates are grabbed, weighted and compounded.
The use of a published daily compounded RFR index considerably simplifies the process of documenting, calculating and reconciling payments by using only two published reference points in the calculation of the applicable compounded rate for any interest period: one pertaining to the start of the observation period and the other to the end of the observation period.
These benefits of the use of an index for RFR cash products have been recognised by both the Alternative Reference Rate Committee convened by the Federal Reserve Board (New York Fed) in the United States and the Bank of England to facilitate the use of the secured overnight financing rate (SOFR) and the sterling overnight index average (SONIA), respectively.
The New York Fed started publishing a daily compounded index for SOFR in March, and the Bank of England started publishing a daily compounded SONIA index in August this year.
It is essential to note that the aforementioned Observation Period Shift is the only methodology compatible with the use of an RFR index, but it has one flaw when paired with the business day lookback approach. The number of days in the shifted observation period can differ from the number of days in the interest period.
This can cause some anomalies, and in some rare instances, it may result in negative daily accruals even in a positive rate environment. This is a relatively minor concern when using a 5 business day lookback, as the problem can only occur around bank holidays if rates are very close to zero, and they will cure with the passage of time, so that it will not be visible over a whole interest period of one-month or longer.
However, it is important to keep in mind that, when using a lookback number of business days that is not a multiple of five the mismatch in days is more frequent, and therefore negative daily accrual calculations in a positive rate environment are no longer linked solely to Bank Holidays.
A solution to the mismatch of the number of days in the interest and observation periods would be to switch from a business-day lookback to a calendar-day lookback structure.
This approach would obviously ensure that the number of days in the Interest and observation periods always match, which is desirable in and of itself, while assuring that daily interest cannot be negative when RFRs are non-negative.