CME One-Month SOFR futures can be used to evaluate and manage day-to-day volatility in Treasury general collateral repurchase agreement (GC repo) rates, particularly around month-end dates. In what follows, we examine a couple of recent episodes to show how.
In mid-2017, the Alternative Reference Rates Committee (ARRC) endorsed SOFR (the Secured Overnight Financing Rate, published daily by the Federal Reserve Bank of New York (FRBNY) as “the rate that represents best practice for use in certain new USD derivatives and other financial contracts, representing the ARRC's preferred alternative to USD ICE LIBOR. SOFR is a broad measure of the cost of borrowing cash overnight collateralized by U.S. Treasury securities in the repurchase agreement market.”1 “The repo market represents a liquid, efficient, tested and safe way for firms to participate in a short-term financing arrangement, providing funding for their day-to-day business operations.”2 In short, the GC repo market is the grease that keeps the wheels of the US government securities market turning.
SOFR began regular daily production in the first week of April 2018.
A month later, CME Group launched two new futures products based on it. Each Three-Month SOFR (SR3) futures contract is based on daily compounded SOFR interest during the contract’s three-month reference period. Each One-Month SOFR (SR1) futures contract references the arithmetic average of daily SOFR values over all days in the contract month. Both futures products have multiple delivery-month listings.3
For SR1 futures, daily settlement prices, trading volumes, and open interest levels are exemplified by data for Friday, 10 May 2019, shown in Figure #1.
Futures Delivery Month | Daily Settlement Price | Trading Volume (Contracts) | Open Interest (Contracts) |
---|---|---|---|
MAY 2019 | 97.5850 | 364 | 22,871 |
JUN | 97.5750 | 514 | 10,775 |
JLY | 97.6200 | 2,014 | 7,445 |
AUG | 97.6450 | 3,207 | 14,888 |
SEP | 97.6400 | 1,946 | 2,781 |
OCT | 97.6800 | 54 | 4,811 |
NOV | 97.7100 | 83 | 4,130 |
Total | 8,182 | 67,701 |
Source: CME Group
In Figure 2, which depicts recent history of the daily effective federal funds rate (EFFR) and SOFR, two features are notable. One is the generally strong correlative relationship between the two benchmarks. The EFFR-minus-SOFR spread is centered on an average of -10 basis points (bps) per year.
The other is the occurrence of transitory departures from the prevailing average, noticeably around month-end dates. Such events at year’s end, commonly known as “the turn”, are long-familiar to money market practitioners. At the close of 2018, for example, increased demand for collateralized cash balances drove SOFR to 76 basis points (bps) over EFFR.
Both of the CME SOFR futures contract mechanisms absorb and reflect such events without difficulty. Their respective designs incorporate the principle (highlighted in the ARRC’s recently published “User’s Guide to SOFR”) that financial products based on overnight interest rate benchmarks do well to reference period-averages instead of single-day values:
“There are two essential reasons why financial products use an average of the overnight rate:
Given this, how might SR1 futures be used either to manage calendar-related volatility in SOFR or to take a view on it?
To answer by example, let’s start with November 2018, for which Figure #3 presents various data for each day:
The futures contract final settlement price, shown in the next to last row, is 100 price points minus the arithmetic average of daily SOFR values during the month: 100 – 2.222 = 97.778. (In accord with the product rules, the SOFR value assigned to any weekend day or holiday is the value corresponding to the first preceding US government securities market business day, and the month-average value is rounded to three decimal places – the nearest 1/10th of one basis point (bp) per annum of SOFR interest – before being subtracted from 100.)
Figure 3 illustrates how the averaging process that informs the final settlement price smooths out daily SOFR values (eg, the intramonth high of 2.28 pct on 14 November, or the intramonth low of 2.18 pct on 19 November).
Figure 3 also suggests how SR1 futures users might use contract price information to gauge market expectations as to the magnitude of month-end SOFR volatility. To see this, suppose it is Thursday, 29 November, the next-to-last day of trading in the contract.
What might this imply for market expectations of SOFR’s behavior at month’s end? One of potentially many ways we could answer is to impose two side assumptions: First, in the absence of any calendar-related divergence, SOFR will hold steady at its latest published value of 2.19 pct/yr (highlighted in red) on each of the remaining two business days of the month; second, any calendar-related divergence will occur entirely on the last business day of the month.
With these two side assumptions, the result could be interpreted to indicate that market expectations are centered on SOFR values of:
With the data available to us and with a couple plausible assumptions, we can venture a reasonable guess that market participants collectively foresee month-end calendar effects pushing SOFR 1.5 bps higher than it would be otherwise.
In the event, this prediction would have proven tolerably good. The published value of SOFR for Thursday, 29 November, was 2.19 pct/yr, as assumed. For Friday, 30 November, it was 2.20 pct/yr instead of 2.21 pct/yr.
Two features of the outcome merit remark –
For contrast, consider an alternative scenario:
The March 2019 delivery month (SR1H9) makes an ideal specimen. As Figure 4 evidences, its last day is a Sunday, which means that the SOFR value for the last business day of the month, Friday, 29 March, exerts impact upon nearly 1/10th of the month-average value (three days out of 31) that determines the final settlement price.
To appreciate the implications, assume as before that it’s the contract’s next-to-last trading day, Thursday, 28 March.
Likewise, suppose we apply simplifying assumptions similar to the two we used earlier: SOFR holds steady at its last published value, 2.40 pct/yr, on each of the remaining two business days of the month in the absence of any month-end calendar effects; and any calendar-related divergence occurs entirely on the last business day of the month, Friday, 29 March.
The forecast path that emerges places SOFR at:
In fact, the month-end rise in SOFR turned out to be around 25 bps (equal to actual SOFR of 2.65 pct/yr minus the baseline prediction of 2.40 pct/yr). Although this is significant, it is modest relative to the 64.7 bps predicted according to the scheme described here.
The next day -- contract’s last trading session, Friday, 29 March -- is notable for at least two reasons:
As Figure 5 illustrates, a hypothetical trader who acquired a long (short) position of 500 SR1H9 contracts at 97.5325 on 28 March, and who offset this position by holding it to final settlement, at the final settlement price of 97.570, would have realized a profit (loss) of 0.0375 price points, or $78,131.25.
Given that the initial margin requirement for SR1 futures at the time was $203.50 per contract, initial margin for this position would have been $101,750.
The examples here underscore the usefulness of the CME SOFR futures mechanism. By averaging SOFR across an entire contract month, it smooths day-to-day fluctuations in the underlying SOFR benchmark. Yet it preserves enough flexibility for hedgers and speculators to effectively manage risk arising from SOFR volatility. This is especially useful around month’s end, when overnight Treasury GC repo rates often become more challenging to predict.
David Gibbs
Director of Education
Fredrick Sturm
Executive Director of Research and Product Development
Beau Parker
Rotational Analyst