By Padhraic Garvey, ING’s Global Head of Debt and Rates Strategy
ISDA has provided fixed spreads to be added to SOFR. So, use them? Probably, but alternative bank rates add a complexity, risking a perception of value transfer. Or, use lower market spreads that can be locked in now? Or use super low variable bank spreads? The latter presents potential alpha but also heightened market risk. There are choices to be made
The simplest approach to conversion is to adapt the fixed spreads of the ISDA protocol
Holders of most USD Libor-linked product technically have until June 2023 to take action on transitioning to something other than USD Libor. This can be achieved by inserting language into documentation that takes the product from referencing USD Libor to referencing something else. In the vast majority of cases, that something else is the Secured Overnight Financing Rate (SOFR). For derivatives it is typically SOFR in arrears; that’s where the liquidity is expected to be from 2022 onwards. For loans, the solution is often to transition legacy product to term SOFR.
For players that have signed up to the ISDA protocol, there is a good rationale to transition underlying loans under the same terms. As derivatives transition at end-2023 on fixed spreads that have already been determined (see below), using the same fixed spreads on the loans would maintain a perfect hedge. Such a solution can be inserted into loan documentation and is perhaps the most seamless solution for both derivatives and loans. This also applies to any derivatives written since 25 January 2021, which came under the ISDA protocol on initiation, and so the transition of loans underlying such swaps could be dealt with in the same way.
What happens if the market spreads don't converge on the fixed ISDA ones?
A key question is whether there are advantages of doing an early transition. To better grasp the answer, we start by making the assertion that the spot basis that we see today between SOFR and Libor should converge on the ISDA fixed spreads as of end-June 2023. Theoretically, this should be the case, as the ISDA spreads are a fixed point in the future that should be hit. A large rump of the market for derivatives will transition at these fixed spreads at end-June 2023, so the theory is that the wider market should also converge on the same spreads, effectively as realised market forwards. Provided that happens, there should be no value transfer between borrower and lender, at the moment of transition.
But then again there is the nagging feeling that perhaps market SOFR to Libor spreads in fact don’t converge on the fixed ISDA spreads. The deviation is quite large at the moment, where SOFR is 5bp and Libor is at 13bp, so the market spread is 8bp. This compares with 26bp on the transition from 3mth Libor to SOFR at mid-2023. Based on the current level of SOFR, this pitches implied synthetic Libor at 31.2bp (SOFR plus the spread). That is quite a difference versus today’s 3mth Libor rate at 12.5bp, almost a 20bp deviation in absolute terms. The bulk of that 20bp differential should dissipate between now and mid-2023, but this is where the existence of the likes of Bloomberg Short-Term Bank Yield Index (BSBY) or Ameribor, etc potentially muddies the waters.
Libor split between SOFR and Bank Spread (bp)
The presence of other bank rates make a perfect convergence on the ISDA spreads more difficult
Simplistically, BSBY is calculated based on where banks fund themselves. It trades a few basis points through USD Libor, and that deviation has been reasonably stable over time. With BSBY a few bps below USD Libor and Ameribor a few bps above, they will always provide a reasonable guide as to where USD Libor would be, even when USD Libor is long gone. Now roll on to mid-2023. If at that time banks are funding themselves at levels that are commensurate with a 26bp bank spread, then we are all good. But, if banks are funding themselves either above or below this spread, then there is an implied value transfer risk, at the moment of transition from Libor to SOFR.
For example, if banks found themselves funding at levels close to today, then application of the fixed ISDA fallbacks would imply that borrowers transition at a credit spread adjustment that is too high. It could also work in the opposite direction, where a spike in bank credit funding costs could mean that borrowers optically transition at a spread that is too low. Now, this does not matter much if we are talking about a loan plus derivative combination, as they offset each other, and while the spread matters, there is no value transfer then (at the moment of transition). But for a stand-alone loan or swap, the spread employed at transition matters more.
Market spreads are available as alternatives, effectively as breakevens
In the end, we may be just talking about just a few basis points and it may not matter much. Also, it is less of an issue for longer-dated product. The reason for that comes down to where the market discount pitches the spread between SOFR and Libor into the future. This varies by the maturity of the underlying (see below), with the tightest spreads for shorter maturities (bullied by tight spot spreads). Longer maturities see a convergence towards the fixed ISDA spreads, which makes a degree of sense as the fixed ISDA spreads are based off a 5-year median calculation, done this way to get at the essence of a fair long-term valuation.
These market spreads assume that current tight SOFR to Libor spreads converge on the ISDA fixed spreads by mid-2023. The market spreads are effectively breakevens, so the SOFR to Libor spread for, say, a 5yr product is set at a spread that breaks even against an alternative route where the borrower continues to pay Libor at market levels and then pays SOFR plus the fixed ISDA spread from end-June 2023 onwards. We’ve done the calculations (see below), and can confirm that SOFR to Libor spreads by tenor indeed act as break-evens (or within a half of a basis point of it). If that is the case, there should be broad indifference between transitioning early at the lower market spread versus transitioning at end-June 2023 at the fixed ISDA spread.
The big question is whether there is any arbitrage to go for here? Well, it depends.
So is there an arbitrage at all? An arbitrage can be found to the extent that the forwards are deviated from. It could be argued there has been a structural change where bank funding costs remain structurally tight versus the risk-free rate, helped by central clearing that reduces system risk, banks that are better capitalized than ever, and e.g. a persistent excess of demand over supply for bank commercial paper. If so, locking in at the fixed bank adjustment spreads available today results in higher borrowing costs, relative to latching on today’s lower bank spreads. And note here that there is no distinction here between the ISDA lock-in versus the fixed market spread lock-in, as they break even against one another.
So, there is a potential arbitrage for players that are willing to choose a variable bank spread e.g. as provided by the likes of AXI, etc, and using that. The switch would be from Libor to SOFR plus a variable bank credit spread adjustment. And here, there could be value for borrowers to choose to transition early. In this way, the product is safely transitioned to SOFR plus this low but variable spread. And by the way, if it were term SOFR it would feel just like the Libor rate being paid today. One important caveat is that term SOFR and AXI or IHS spread product are in their infancy, and the latter, in particular, is not in a fully tradeable state. In consequence, this is something to be considered in the coming few months, rather than here and now.
So, choose from these three:
The baseline stance is to enjoy the low Libor rates being quoted now, and transition at the fixed ISDA spread to SOFR by end-June 2023. That is a defendable strategy, provided the caveats noted above are taken on board. It is certainly the more uncontroversial approach, and one that links nicely with the ISDA protocol, and is in any case the approach to be taken for derivatives written since 25 January. It is also a nice fit for portfolios of loans overlayed by derivatives, where derivatives can sign up to the ISDA protocol and loans can employ the same treatment (which is typical where loan language has been hardwired).
2. Market spreads
Alternatively, do the switch to SOFR sooner at fixed spreads in the market. The advantage of doing an earlier switch to a lower fixed spread (e.g. 20bp for a 5yr product) is not so much for full tenor savings in funding costs, as these spreads breakeven against a convergence on 26bp by end-June 2023. Advantage comes more from cleaning up the issue early, say at the same time as transition in other product, a lot of which gets done by end-December 2021. An additional argument is early transition avoids Libor spike risk between now and mid-2023; no reason to expect a Libor spike per se, but the early fixed transition avoids the risk completely.
3. Variable spreads (coming soon)
Or, latch on to low bank spread that will become increasingly available. This approach takes on more risk, but at the same time can result in structurally lower funding costs should banks continue to print at tight spreads over the risk free rate. In a similar vein, a link to BSBY or ICE, etc (rather than SOFR) could be considered as an all-in solution. Should banks be in a position to fund themselves at or around current funding levels, then by definition the likes of BSBY would trade well through a synthetic Libor constructed from term SOFR plus the fixed ISDA spread.
The counter-argument is that bank spreads could overshoot, and the borrower ends up with more expensive funding, and subject to funding spikes during times of stress. This is where the official sector has issues, and may well be where borrowers do too; choices to be made. We like latching on to super-low variable bank spreads, but fully understand the simplicity of hitching to the ISDA fixed spreads.
Finally, other relevant pieces on choices between SOFR and alternative bank rates can be found here. And in addition, a piece on the choice between SOFR in arrears versus term SOFR can be found here.
Source: ING