The longer the current U.S. expansion drags on, and the more unapologetic Jerome Powell’s Fed comes across, the more obsessed the market becomes with curve inversion.
And when I say “curve inversion”, I mean “pick a curve, any curve.” Swaps, corporates or the old standbys in the Treasury complex. It all works if you’re running down stories. Anything to put the word “inverted” next to the word “curve” in a headline. “It’s provocative. It gets the people goin’!”
On Thursday, Morgan Stanley is out with an expansive new note that flags mid-2019 for inversion. To wit:
We lower our net Treasury supply forecast by $690 billion through 2020 and we lower our Treasury yield forecasts: 10y yield at 2.75% by year-end 2018 and at 2.50% with an inverted yield curve by mid-2019.
While that’s the headline, the rationale is what’s interesting.
In the note – which spans a truly “impressive” 77 pages – Morgan takes the view that the Fed will ultimately call an end to balance sheet rundown by September of next year with SOMA sitting at ~$3.7 trillion. They also say the balance sheet will then start to grow again, hitting $3.8 trillion by the end of 2020. Here’s Morgan:
Our view contrasts with the average view of primary dealers and market participants. Exhibit 2 shows the percent chance of possible outcomes for the par value of the SOMA portfolio at the end of 2020, conditional on the Fed not moving to the zero lower bound at any point between now and the end of 2020, according to the New York Fed’s primary dealer and market participant surveys conducted ahead of the June 2018 FOMC meeting.
Primary dealers and market participants placed a 68% and 60% probability on average, respectively, that the SOMA would be smaller than $3.5 trillion at the end of 2020. Primary dealers and market participants placed only an 8% and 11% probability on average, respectively, that the SOMA portfolio would be larger than our base case at the end of 2020. At the heart of our view is the imbalance we project between the demand and supply of the liabilities that underly the assets in the SOMA portfolio: reserves.
Note that bit at the end there about what informs the bank’s assessment. Basically, they argue that “too much” (and those are my scare quotes, not meant to denote an actual verbatim quote from the bank’s note) normalization risks driving the effective federal funds rate above IOER, an outcome which increases the chances that EFFR rises above the target range. In other words, this is an extension of the discussion about the tweak to the interest the Fed pays on reserves. Consider this:
Exhibit 11 shows a history of the level of excess reserves and the distance between the upper bound of the Fed’s target rate range and the EFFR. As excess reserves have decreased since 2014, the EFFR has moved higher in the target range as shown in Exhibit 12 . The EFFR moved as close as 5bp to the upper bound in May 2018. In an attempt to move the EFFR back towards the middle of the range, the Fed lowered IOER by 5bp within the target range at the June 2018 FOMC meeting, and this pulled EFFR lower as well. As of now, the EFFR is only 4bp below IOER and is 9bp below the upper bound. We expect that the EFFR would continue ticking higher towards the upper bound as balance sheet normalization continues and that the relationship between the EFFR and the level of reserves might prove non-linear.
That non-linear relationship could occur since, as the level of reserves in the system shrinks, the likelihood that a bank would have to resort to borrowing from the Fed at the discount window—in order to meet the 10% reserve requirement—increases. If borrowing at the discount window is notably more expensive than the prevailing fed funds rate, the bank would bid reserves ever more aggressively as the likelihood of having to resort to the discount window increases. This would lead to a non-linear relationship whereby changes in the level of reserves in the system would have an increasingly larger impact on the EFFR.
Obviously, that scenario would be exacerbated in a situation where excess reserves are concentrated in the hands of too few banks, who may or may not meet demand with sufficient supply. Thus, “the concentration of reserves is a key determinant of when balance sheet normalization will end”, Morgan goes on to say, before adding these important clarifying remarks:
By bidding up reserves in the fed funds market, smaller banks that start to run short on required reserves would be able to obtain the reserves from the large holders such as JP Morgan. However, such borrowing would likely have to occur at rates noticeably higher than IOER since GSIBs are unlikely to become active lenders in the fed funds market simply because EFFR is higher than IOER.
There’s a lengthy discussion of that bolded point, but if what you’re looking for is evidence, Morgan simply says that “if GCF repo, a secured lending rate, can accelerate beyond IOER, then it is reasonable to expect EFFR, an unsecured lending rate, to also accelerate past IOER rather than plateau once it crosses IOER.”
So what to do? Well, that’s where the balance sheet normalization argument comes in. Morgan says the Fed will likely try to lower IOR again by another 5bp, and if that doesn’t work, well then they’ll “commence the end of balance sheet normalization”.
They float the idea of the Fed lowering IOER versus IORR in order to incentivize banks to lend out their excess reserves without punishing them them for meeting their requirements, but Morgan notes that because “the majority of reserves in the system are excess reserves”, the distinction probably wouldn’t make any difference.
Morgan then lays out a number of possible strategies, with the “conservative” approach seen as the most likely. Here’s what that approach entails (and here they’re using “IOR” to mean both IOER and IORR):
The FOMC maintains IOR 5bp below the upper bound as long as the EFFR remains about 5bp below IOR. At some point, the EFFR rises to within a couple basis points of IOR. The FOMC lowers IOR to 10bp below the upper bound and, at the same meeting, announces a gradual end to balance sheet normalization.
Past that point, the discussion gets even further into the weeds and while there are undoubtedly some folks out there who would be interested in the mechanics, in the interest of brevity, we’ll just excerpt the following passage which explains how a decision to halt normalization early ends up flattening the curve. Here’s Morgan:
The early end to balance sheet normalization means the Fed will begin rolling over more of its Treasury holdings sooner than we thought previously. We also think it means that the Fed will remove Treasury supply from the secondary market when it reinvests agency debt and MBS principal into the Treasury market instead of the agency MBS market. In turn, the US Treasury will have to issue fewer Treasuries to the public than what we thought before. As a result, we lower our forecast for Treasury supply by $226 billion in 2019 and $464 billion in 2020. Less supply should push Treasury yields lower this year and next, and we now forecast a 2.75% 10y yield by year-end, and a 2.50% 10y yield with an inverted yield curve by mid-2019.
The only lingering question here is when the Fed will start to tip its hand on this and for their part, Morgan Stanley says that’s likely to come in the minutes from the December 2018 meeting.
Trade accordingly and do note (again) that this entails the Fed’s balance sheet will start to grow again in 2020.