User talk:Cumulant

LIBOR-OIS
Hi, I definitely agree with the SOFR edit. However, the argument you discarded is supported by the reference. You presented a counter-argument using the same reference. Can you find a source that supports your claims?

I disagree (also in the original text) with the tenor discrepancy because we're not talking about OIS but about the OIS-3m swap. Are you thinking of LIBOR-3m receiver + OIS-3m-payer? — Preceding unsigned comment added by Hroptatyr (talk • contribs) 08:50, 27 March 2020 (UTC)

I think the reversion is problematic. First, SOFR is secured whereas LIBOR is unsecured; that's a major difference that I know multiple counterparties are not willing to accept. It's literally the difference between collateralized and uncollateralized rates: they are not the same and never will be. As for the tenor mismatch, OIS is an overnight rate whereas LIBOR is typically quoted at 1M, 3M, and 6M tenors. If I enter into an OIS-3M swap, I still have a tenor mismatch in that swap: one leg is exposed to overnight rates while the other leg is exposed to 3M rates. These are at different points on the yield curve. Making one leg the OIS-3M spread does not resolve this issue either. Making the other leg LIBOR-3M does not resolve this because then I'm assuming the 3M is a UST (risk-free) rate -- in which case these are just bets on a credit plus yield curve slope (OIS-3M) versus a pure credit spread (3M LIBOR - 3M USTs).

Both of these issues, secured vs unsecured and overnight vs 3-month, are issues counterparties are unhappy about. I've discussed these at length with other people in the market; and, none of them sees any resolution apart from forcing people to change (which they admit will lead to legal fights). Even people at the CME admit that their SOFR futures are suffering because averaging an overnight rate still does not get you to a 3-month rate. Having been through swap compression and renegotiations, these are not trivial issues.

You can read plenty about this in Morgan Stanley's page on the "transition" (which they admit is problematic). An excerpt is telling: First of all, SOFR relies entirely on transaction data, whereas LIBOR is based partially on market-data “expert judgment." Secondly, SOFR is purely a daily rate—what’s called an overnight rate—vs. LIBOR's seven varying rates on terms of one day to one year. Finally, LIBOR incorporates a built-in credit-risk component because it represents the average cost of borrowing by a bank. In contrast, SOFR represents a “risk free” rate because it is based on Treasurys.

Given these differences, USD LIBOR can’t simply be “swapped out” with SOFR in existing contracts that reference LIBOR—at least not without appropriate adjustments. Can LIBOR be based on transaction data? Sure; just use trades in the hyper-liquid Eurodollar futures market. So issue #1 is surmountable. However, those other two issues? Not fixable. Cumulant (talk) 23:56, 1 July 2020 (UTC)