EIB SONIA-linked bond

The move to the use of overnight benchmarks as the main interest rate benchmarks is progressing. After the wave of new overnight-linked futures in the last months by CME, ICE and CurveGlobal, we now have a large overnight-linked bond issuance. The EIB has successfully issued a one billion 5-year SONIA-linked bond.

The coupon payment is based on the backward-looking overnight SONIA composition plus a spread of 35 basis points (Maturity 2023-06-29, quarterly payments, ISIN XS1848770407). Note that a small 2 millions bond had been issued in January (quarterly SONIA + 25 bps, maturity 2023-03-20, ISIN XS1889459713). Thanks to an acute observer of the market for pointing that to me.

Backward-looking overnight is viable. Certainly for some issuers and investors. There was never a doubt about it for some market participants. This does not prove it is viable for all market participants, and this is where the question lies. Note that the coupon will be paid with a five-day lag after the last overnight benchmark fixing. This is a serious discrepancy with the current overnight-linked derivative market which pays with a zero day lag in Sterling (and 2 in EUR and USD). Some elements of the issuance are discussed in a Risk article 'EIB shrugs off term RFR worries with Sonia bond plan' (subscription required) and a Financial Times article 'EIB to offer Libor alternative with Sonia-based bond' (subscription required).

The question of backward-looking or forward-looking — i.e. OIS benchmark — overnight-linked coupons is heavily discussed in the potential LIBOR discontinuation and fallback provisions. The EIB issuance, even if linked to backward-looking overnight benchmark, does not bring any substantial positive information regarding the fallback discussion. If anything, the five-day lag expresses the difficulty of that approach. The backward looking payments have been used for many years in the derivatives market and there is not doubt that they are working for products designed with its requirements in mind.

Regarding the LIBOR fallback, the products linked to LIBOR have been designed with the term rate at their core. Can we replace a term rate by a backward-looking rate? That is the question. The answer is: in some case yes and in some cases no. The obvious 'no' are FRA with FRA discounting settlement and LIBOR in-arrear that require a payment at the start of the term period. As the fallback procedure should be the same for all LIBOR usages in derivatives, the global answer to the fallback procedure question has to be 'no, a backward-looking fallback is not possible'. The above answer is not only a personal opinion, it is a physical impossibility, except if one can realize backward time travel.

I will provide more technical details and opinions about the LIBOR fallback in a forthcoming notes. Hopefully I will have time to finish it by mid-July.


Multi-curve framework: double or quits?

This blog entry is a reaction to the column by Bakkar and Brigo in Risk titled Model risk in the transition to risk-free rates. While I agree with most of the column, there is a secondary issue, which is close to my heart, I would like to react to. And I certainly concur with the last sentence of the column which starts with "This should involve the quant community in an active role" as this has also been my opinion for a long time.

I would nevertheless like to present a different perspective on some side elements of their argument. To my opinion, the expanded use of overnight benchmarks will (should) not reduce the multi-curve complexity.

The pre-2007 single curve "simplicity" was coming from a combination of error and domination. Error in confusing two effects, the (pseudo-)discounting and the reference benchmark. This error was helped by the domination of a rate family: the IBOR benchmarks — the LIBOR king. The crisis emphasised the difference between the overnight benchmark used for collateral remuneration and the IBOR benchmarks and lead to the overnight discounting. A second crisis impact is that the king is dying.

Moving to overnight benchmark as the dominant benchmark may simplify the valuation of vanilla instruments by simplifying the vanilla contracts(1); both the collateral/discounting and the reference benchmark will be the same. But it will only simplify the bank balance sheets on the vanilla derivative side, there is no guarantee that it will simplify the balance sheet globally. The funding term structure of a bank will become the LIBOR of tomorrow but without its liquidity. The generalised VM/IM and its associated costs has pushed the funding issue into the valuation of derivatives (FVA, MVA). There is no need to see the LIBOR or the funding cost to appear directly in the instrument term sheet to need it in derivatives valuation.

I certainly concur with the authors that transition to new benchmark could be an opportunity to review models related to funding but I don't think it will reduce multi-curve complexities. It may lead to a name different from Multi-curve Framework, but the issue of different curves to be used for different purposes will continue to be relevant, maybe even more so with the introduction of new benchmarks. Maybe it will be called the Funding Framework(2).

Even if I concur with the authors that the transition could be an opportunity, I'm afraid that the market will have its back against the wall due to time and regulatory constraints — like in the margin changes case —, that only the minimal to survive will be realistically achieved and that the bank managements, market associations and regulators will not let the quant community play an active role.

Like for other similar issues, I sincerely hope that I will be proved wrong on the quants involvment. Don't hesitate to let me know that I'm wrong, I will enjoy it.

Don't hesitate to contact me for projects on the subject of benchmarks transition.

(1) Contract here as to be understood as a combination of the specific instruments term sheet and the generic master agreement, which govern the now mandatory associated daily variation margin.

(2) By the way, this was the name I used in my March 2007 paper on Discounting Irony, the first paper on what is today called multi-curve framework.