September 19, 2019 - For those that missed it – which, based on our email traffic, was no one – the published overnight SOFR rate jumped from 243 bps on Tuesday to 525 bps on Wednesday and then back to 255 bps on Thursday. So what happened to SOFR on Wednesday? We discuss below (and the ARRC explains in this helpful missive as well).

First, unlike LIBOR, SOFR is tied to the actual transactional rate on Treasury repos. As the WSJ reported, there were a number of market factors that pushed Treasury repos higher on Tuesday. These included corporates having tax payments due and therefore draining money that otherwise would be in the repo market, thus reducing demand. Another factor was Treasury issuing bonds, which increased supply. Between lower demand and higher supply, market forces pushed repo rates up. Because SOFR is a volume-weighted median of actual repo transaction data, the increase in repo rates drove Wednesday’s published SOFR to 5.25%. But the market rapidly normalized, and Thursday’s published SOFR rate was 2.55%.

So, what does this mean for cash markets – and potentially loans priced off SOFR? Not actually that much. No one is planning to price their loan off one-day’s SOFR. Instead, the proposals for SOFR in the loan market are either the forward looking term SOFR (which would be built from SOFR futures trading) or an average or compounded SOFR (which can be tracked here and will be either calculated in advance or in arrears). The term or compounded rate will smooth daily SOFR fluctuations (which, to be clear, will exist). The LSTA COW compares the daily SOFR to a one-month and three-month compounded rate from September 2018 through February 2019, including a period of substantial daily SOFR volatility around year-end. As the COW demonstrates, even though daily SOFR moved 70 bps around year-end, one-month compounded SOFR increased seven bps and three-month compounded SOFR increased just three bps. (Readers can see an updated version of this analysis in the ARRC missive.) And, in fact, our understanding is that a three-month average SOFR has increased two bps relative to rates last week whereas three-month LIBOR has increased four bps in the same time. Finally, a recent publication suggested that the SOFR spike could have a material impact on the “spread adjustment” that ISDA will publish around year-end to make LIBOR and SOFR more comparable for assets that “fall back” from LIBOR to SOFR. The article suggested that if the spread adjustment were – for example – calculated over the past 30 days, it would be 10 bps (300 bps/30 days) higher than might otherwise be. However, the spread adjustment is meant to be calculated based on the historical mean or median over years, not days. Thus, a jump like we just saw should not impact the spread adjustment materially either.