Customer Logins

Obtain the data you need to make the most informed decisions by accessing our extensive portfolio of information, analytics, and expertise. Sign in to the product or service center of your choice.

Customer Logins

ARR You Really Safe with the ISDA Fallbacks Protocol? Non-linear derivatives are not so straightforward

20 May 2021 Hiroshi Tanase

One major problem to tackle in the transition from LIBOR to alternative reference rates (ARRs) such as SOFR, €STR and SONIA (which are also referred to as risk-free rates, or RFRs) is how to handle legacy LIBOR positions when LIBOR is no longer published. It is fundamentally a challenging issue. While LIBOR is a credit-sensitive rate, the new risk-free ARRs such as SOFR are risk-free rates without a credit component. However, with the ISDA Fallbacks Protocol the problem is largely resolved. Or is it?

The mechanics of the ISDA protocol is now well understood for vanilla swaps. After the cessation of the relevant LIBOR fixings (e.g., GBP 6M LIBOR) referenced by legacy positions in a portfolio, each LIBOR fixing will be replaced by its replacement ARR such as SOFR, €STR and SONIA with an adjustment spread that is pre-determined for each LIBOR fixing. Since the new rates are daily rates, they will be compounded through the coupon period to form a term rate in time for payment (i.e., by the Cut-off Time such as two days prior to the payment date). If the forward LIBOR rates are traded, prior to cessation, at levels consistent with the ARR forwards plus the ISDA fallback spread, the application of the Protocol will not result in value transfer.

The situation surrounding non-linear derivatives is much more complicated. By design the ISDA Fallbacks Protocol is meant for all products. Therefore, it can handle non-linear products including LIBOR exotics, which increases the utility of the Protocol as a safety mechanism. However, the Protocol may or may not produce a desirable outcome. So, the following questions must be asked:

  • Can the protocol procedurally work for a given trade?
  • Will the 'fallen back' trade be economically equivalent to the original trade?

The answer to the first question is a reassuring yes in most cases. The protocol can transform even those products whose LIBOR-linked payoffs do not seem naturally compatible with compounded-in-arrears rates. However, the fact that the Protocol can successfully render a LIBOR trade into an 'equivalent' one that references an ARR does not mean that the resulting new ARR trade is economically similar, let alone equivalent, to the original trade. In the case of some LIBOR exotic trades, the Protocol can totally alter the economics of the trade. Let's look at a few examples.


The fallback protocol works well for swaptions with all settlement variations. For a physically-settled swaption, the existence of the Protocol means that the underlying LIBOR swap can be deemed equivalent to an ARR swap with a fallback spread (e.g., 10-year SONIA swap with a spread on the compounded SONIA rate). Therefore, a physically-settled LIBOR swaption becomes economically equivalent to a swaption to enter into an ARR swap with a spread on the floating rate. By moving the spread on the floating leg onto the fixed leg, a legacy LIBOR swaption that is physically-settled can be shown to be equivalent to an ARR swaption (e.g., SONIA swaption, SOFR swaption, etc), with a suitable adjustment on the fixed coupon rate, or the strike level.

There are some minor caveats, however. If the original term sheet specifies that the underlying swap is cleared but the legacy LIBOR swap is no longer eligible for clearing when the fallback is applied, then the swaption must be settled in cash following the Collateralized Cash Price method.

Cash-settled swaptions can be treated in a similar fashion because the transformation of the underlying swap will essentially be the same. The main issue is with the publication of the benchmark rate such as the ICE Swap Rate (formerly known as ISDAFIX) for the underlying legacy LIBOR swap. Since ICE Swap Rate, which is for LIBOR swaps, will (most likely) stop being published after LIBOR is retired, an alternative method must be established to determine the settlement amount.

To date, both ARRC and the Working Group on Sterling Risk-Free Reference Rates have proposed a formula to produce an approximate value of the discontinued ICE Swap Rate for a LIBOR swap from the corresponding RFR ICE Swap Rate. It is anticipated that the industry will embrace these formulae for the settlement of legacy LIBOR swaptions. If the market participants believe that these formulae produce results that are consistent with the present values of corresponding underlying legacy LIBOR swaps to which the ISDA Protocol is applied, the pricing of cash-settled swaptions (using Collateralized Cash Price) will not attract any additional valuation adjustments.


The situation is more involved with caps and floors. While the fallback protocol can replace the underlying LIBOR rate with a new rate such as SOFR and €STR in the same fashion as vanilla swaps, the difference in the timing of rate fixing alters the economics of the trade. Unlike the conventional LIBOR cap/floor, the payoff of a new ARR (or RFR) cap/floor is not determined until the end of (e.g., two days prior to) the coupon period. In other words, a LIBOR cap (or floor) gets transformed into a compounded-in-arrears ARR (or RFR) cap/floor. This has two implications: (1) the time-to-expiry gets extended and therefore the option value will increase; (2) the payoff will become dependent on multiple (daily) fixings, so a LIBOR cap (or floor) will become an Asian option in respect of the daily rates. RFR cap/floors must be valued considering these features that are distinct to them. In conclusion, while the Protocol will work well procedurally for caps and floors, it alters the economics of the trade and cause value transfer.

The volume of RFR cap/floors is still limited. But markets are evolving as methodologies for these new cap/floors have been proposed, and major broker-dealers gradually increase their trading of this new instrument type.

LIBOR-in-arrears (LIA)

LIBOR-in-arrears swap is one of the classic variations of the vanilla interest rate swap. Unlike the vanilla interest rate swap, the floating rate is fixed at the end of the reset period. As the swap references LIBOR at a later date (e.g., 6 months later on a semi-annual LIA), it allows the investor/hedger to take a view on the future changes in the slope of the yield curve (together with a consideration for convexity).

Since there is no gap between rate fixing and payment, the structure of the swap is not suitable for the use of an arrears-fixing rate. However, Section 6(d)(i) of the ISDA Fallbacks Protocol provides language that allows the determination of a fallback rate nevertheless. The following text from the accompanying FAQ document describes the provision in a slightly more accessible fashion:

"However, if a Calculation Period is shorter than the length of the Designated Maturity (for example, if 6-month sterling LIBOR was originally to be used for a four month Calculation Period), then the fallback rate for the 'Original IBOR Rate Record Day' that corresponds to the Relevant Original Fixing Date would not be published/provided by the Applicable Cut-off Time such that the most recently published fallback rate for the most recent Original IBOR Rate Record Day will be used instead."

In short, if a compounded-in-arrears fallback rate cannot be formed because there is not enough time between the LIBOR fixing date and the cut-off date (e.g., two business days prior to the payment date), the Protocol will use the most recent rate that is known by the Cut-off Time. But the brute force application of this rule will have the effect of degenerating a LIA into a vanilla interest rate swap. That is ironical.

In other words, although procedurally the ISDA fallback works with LIA, the economics of the trade will be entirely altered. As a result, the fallback protocol will cause value transfer for legacy LIA positions. More fundamentally, LIA's product design is not sustainable post LIBOR retirement. It is likely to stop trading in the future in the absence of LIBOR.

Range accrual

This is yet another popular LIBOR exotic product, which has been in existence for multiple decades. In a range accrual trade, the coupon accrues only on days when the relevant LIBOR is within a pre-determined range (e.g., 0-3%). Since the daily observations of the relevant LIBOR are used for the calculation of the coupon amount, the fallback protocol must prescribe how each of the daily LIBOR fixings should be replaced with the ARR (with a suitable spread).

However, except for the very first LIBOR fixing, no replacement compounded-in-arrears rates can be used because they are not known in time by the Cut-off Time (e.g., two days prior to the coupon payment date). So, the provision we cited for LIA is invoked again to resolve the issue. As a result, from the second fixing onwards the "last known fallback rate" is used for the determination of the range accrual coupon amount. In other words, the same fallback rate, which essentially corresponds to the LIBOR fixing set at the beginning of the reset period, is used as the replacement rate for all the daily fixings.

As a consequence, a range accrual will fall back to become economically equivalent to a single-look LIBOR range bet (or a combination of a LIBOR cap and a floor from a risk perspective). Due to the change in the economic properties, the fallback will trigger value transfer. The averaging effect of daily fixing will be lost, and the timing of fixing will be brought forward, which will bring the effective forward fixing levels lower in an upward sloping curve environment.

Issues with range accruals do not end there. Since there is no strict standardization of language used to describe range accrual trades, some term sheets that are not documented under the ISDA range accrual template might not be able to apply the fallback.

The only way to remove these uncertainties is to bilaterally negotiate a mutually agreeable conversion of the existing positions into alternative trades. While there is no consensus on the future format of range accruals, possibilities include (1) using the daily RFR rates for accrual calculation (without any compounding), (2) using short swap rates such as CMS 1Y instead of LIBOR as a reference index. However, since all these alternative payoffs will have a risk and valuation impact, ample time should be allocated to review the details thoroughly and to agree appropriate new terms to avoid inadvertent value transfer.


Having reviewed the potential consequences, it is evident that simply letting the ISDA Fallbacks Protocol handle legacy non-linear LIBOR positions is not going to guarantee a desirable outcome. Even for vanilla swaps, the industry best practice is to proactively convert legacy IBOR swaps ahead of LIBOR cessation so that the use of ISDA Protocol is minimised. Clearinghouses such as LCH and CME are converting existing LIBOR swaps en masse into RFR equivalents ahead of 31 December 2021. If vanilla trades require this much attention to avoid surprises, why shouldn't non-linear trades that are fundamentally more complex deserve more attention?

Undoubtedly, the value of the ISDA Fallbacks Protocol for the industry is immeasurable. But it is only a seatbelt. Although a seatbelt increases the chances of survival in the event of a car crash, it does not help avoid a crash. Why should a firm leave things to chance instead of proactively converting legacy LIBOR positions, especially non-linear positions, into ARR equivalents on terms that make sense economically?


Posted 20 May 2021 by Hiroshi Tanase, Executive Director

IHS Markit provides industry-leading data, software and technology platforms and managed services to tackle some of the most difficult challenges in financial markets. We help our customers better understand complicated markets, reduce risk, operate more efficiently and comply with financial regulation.

Follow Financial Services


Follow Us

Filter Sort