Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Quirks in interest-rate benchmarks always make for a fun time here at Alphaville.
So we were intrigued by Barclays’ recent note highlighting a corner of money markets that may be unnecessarily excluded from calculations of the Secured Overnight Financing Rate, or SOFR, the US’s regulator-selected replacement for Libor.
Long story short: SOFR’s underlying transactions exclude a type of trade with volume that totals 10 to 15 per cent of what’s currently included. These types of trades are also expected to get more popular in coming years, mostly because they help reduce a type of capital constraint imposed by the expansion of clearing requirements.
OK now let’s take a step back. SOFR measures the cost of repo agreements that basically involve pledging Treasuries as collateral to borrow cash short-term. (Nb: It’s really a sale with a promise to repurchase. That does matter for various reasons, but isn’t super relevant here.) Libor, in contrast, is the cost of unsecured lending between banks, but after a massive scandal and an equally massive change in the way regulators view bank capital, there hasn’t been much action in that market.
Unlike unsecured interbank loans, there is a whole ton of volume in the market for Treasury repo.
And yet there are still some transactions that might be missing from the data underlying SOFR, says Barclays. SOFR doesn’t include forward-starting or “reg” repo transactions, which occur on a different timeline than the daily morning rush of financial institutions’ most urgent financing needs.
Barclays describes the schedule quite nicely:
Cash [repo] trading is an early morning market. On December 11, $15bn worth of cash settled repo across 193 transactions occurred in the first second of trading. By the end of the first minute, 20% of the day’s cash trades had already occurred (Figure 5). While activity tapers off somewhat after the first 10m of trading, about half of the roughly $500bn in daily trades was already completed by 7:15-7:20am on both days (Figure 6). By 10:30, cash trading is mostly done for the day and just 6% ($30bn) traded after the 1:15pm RRP close.
Reg activity occurs simultaneously but at a slower pace. Reg volumes did not breach their 50% threshold until after 9am.
This likely reflects traders’ priorities. Traders are more occupied with their principal task early in the morning (that is, getting funded that day). Once this is finished, they can turn to their “homework” — reg trading. By trading reg, they also reduce the volume of securities that need to be funded when they come back to the office the next morning.
Reg trades are basically repo traders working ahead for the coming days. Maybe they’re inspired by what Barclays calls the “beach effect”, or trying to get out of the office in time to reach the Hamptons each Friday.
That could make reg repo pricing a little more arbitrary, we guess? And the trades probably include some type of premium to account for any uncertainty about interest rates one day in the future. On the other hand, Barclays argues that reg repo trades are helpful because of those timing differences. They help provide a starting point for repo traders in the busy morning hours.
So should they be excluded from the US’s interest-rate benchmark?
For context, the other group of trades are “specials”, which are repo trades done for a particular security, usually to cover a short. Those trades often deviate pretty far from the going rate for repo financing, making them unhelpful to include with the benchmark.
In contrast, the economic argument for excluding reg repo trades is tougher to make, even if they’re just traders trying to extend their weekends. After all, they still reflect the market information available at the time of the trade. And there are plenty of other reasons to trade reg repo:
First, reg trades are often done to reduce the pile of general collateral securities the dealer has to fund the next morning. This frees up time to focus on more profitable activity such as specials trading. Likewise traders may want to lock in funding a day in advance if they expect rates to climb throughout the next day.
What’s more, reg repo may capture financing disruptions that don’t hit during the busiest repo-trading hours. For example, the night of Dec. 28 saw unusually high reg rates — a full 13 basis points above the next day’s rate — which prompted talk among Barclays clients that SOFR was too low on Dec. 29.
But the analysts concluded that reg rates’ inclusion in SOFR wouldn’t materially or consistently affect its daily levels the way that including repo “specials” would.
Even on Dec. 29, the SOFR rate with reg volumes included was just around 1bp above the reported rate, says Barclays. And on the average day (Barclays studied Dec. 11), and presumably most days, including reg repo trades hardly affected SOFR at all.
And most important, Barclays says repo traders might start shifting more of their activity into the reg repo market as central-clearing requirements expand.
This is because reg trades (along with longer-term repo trades) reduce the burden of firms’ required commitments to the clearinghouse’s emergency liquidity facility, says Barclays. So even though the material effects of reg repo are small today, the trades could become more important to money markets, says Barclays:
At a minimum, we think these results suggest that it might be time to re-evaluate the exclusion of overnight reg trades from the SOFR calculation — especially as reg trading is expected to increase ahead of the July 2026 repo clearing mandate as direct members move to minimize their share of the FICC’s CCLF . . .
Without more data on cleared sponsored bilateral activity, we cannot be certain whether overnight reg trades should be included in the benchmark rate. To solve this we would need sponsored repo volumes broken out by their maturity, settlement type (cash or reg) as well as their collateral type. Calculating volume-based medians would also require (anonymized) transaction level data.
In other words, US regulators might want to keep an eye on reg repo and see what data they can find.