How fast will the Fed drain the punchbowl?

Deutsche with the note. To my mind, if get anywhere near this outcome then stocks are going to crash. Put another way, stocks are going to crash to prevent it:


Last week we adjusted our Fed view to bring forward liftoff to March and accelerate quantitative tightening (QT) to begin during the summer (see Leaving behind “behind the curve”). This hawkish adjustment reflected the signals from the December FOMC meeting, the minutes to that meeting released last week, as well as recent labor market data that showed the unemployment rate cut through the Fed’s median estimate of NAIRU. Related to the balance sheet in particular, the minutes confirmed that, while drawdown will come after liftoff, it would be “closer to that [timing] of policy rate liftoff than in the Committee’s previous experience”.

In addition, “many participants judged that the appropriate pace of balance sheet runoff would likely be faster than it was during the previous normalization episode.”

Importantly, the minutes did not offer a dovish counterweight to these hawkish indications.

Having provided the broad contours of our updated view on balance sheet unwind, in this note we discuss the details of how quantitative tightening could proceed over the next several years. In particular, we start by providing our assumptions for the parameters of shrinking the balance sheet, which admittedly are uncertain at this point given limited guidance from the Fed. Based on these parameters we then present our baseline projections for the size and composition of the balance sheet through 2025. Finally, we conclude by discussing the implications of this drawdown for rate increases by providing a set of estimates for the equivalence between rate increases and balance sheet reductions.

Cap design and implementation

The Fed will likely lean on the principles and plans it developed in 2017 as a blueprint for quantitative tightening. In particular, several key elements from the previous episode could be retained, such as caps for controlling the pace of runoff, a focus on the level of reserves as the main determinant for the long-run size of the balance sheet, and potential adjustments at a later point to achieve optimal composition of asset holdings in the longer run.

With regards to caps, the December minutes noted officials’ support for this approach by stating “Many participants also judged that monthly caps on the runoff of securities could help ensure that the pace of runoff would be measured and predictable, particularly given the shorter weighted average maturity of the Federal Reserve’s Treasury security holdings.” Recall that during the last QT episode the Fed began with relatively low caps of $6bn and $4bn per month for Treasuries and MBS, respectively, and gradually increased these every three months over a span of twelve months to a maximum of $30bn and $20bn.

While the last episode of QT was intentionally gradual to avoid market disruptions, Fed officials, including Chair Powell this week, have indicated that the unwind will be faster this time. A faster unwind is motivated by a need to tighten financial conditions given the stronger macroeconomic backdrop, as well as a desire to more quickly bring the Fed’s elevated balance sheet back closer to historical levels. As such, we expect the maximum caps to be considerably higher this time, around $60bn and $45bn per month for Treasuries and MBS, respectively. These levels would be roughly consistent with scaling the previous experience to the growth in the Fed’s portfolio holdings and projected runoff during this episode. A shorter portfolio duration and higher caps would be consistent with the assertion from the minutes that “the balance sheet could potentially shrink faster than last time if the Committee followed its previous approach in phasing out the reinvestment of maturing Treasury securities and principal payments on agency MBS.”

We anticipate that the Fed will once again phase in these caps, but do so over a shorter time horizon. Specifically, our baseline sees initial caps of $20bn for Treasuries and $15bn for MBS for securities maturing from its portfolio each month.

Principal payments received up to the caps would be redeemed, and balances in excess of the caps would be reinvested. The caps would increase in steps of $20bn for Treasuries and $15bn for MBS every two months until they reach their maximum sizes after four months. They would then remain in place until the balance sheet has shrunk sufficiently to approach its longer-run size. We also expect the full set of cap sizes will be announced ahead of their implementation, likely at the May FOMC meeting. We currently assume the first caps will be established in August, the transition caps in October, and the maximum caps in December.

There is a risk that the Fed could tilt in a more hawkish direction by either shortening the caps’ transition period to every month or bringing forward the start of QT from August to June. The latter scenario would likely be perceived by the market as more hawkish and hence could lead to a stronger market reaction.

Bill portfolio = spigot for accelerated QT

The Fed currently owns a $326bn T-bill portfolio alongside its Treasury notes and bonds and MBS holdings. These bills were acquired predominantly through reserve management purchases in 2019-2020 during a time when the Fed sought to raise its reserve balances to support the effective implementation of monetary policy. Since March 2020, the Fed has kept its bill portfolio steady by rolling over maturities at bill auctions. The reinvestment of bills and coupon securities are separate processes from each other, and thus we don’t think bills will be subjugated to the Treasury cap during the runoff.

We anticipate that the Fed will unwind its entire bill portfolio as part of the normalization strategy. By running off its bills, the Fed could shrink its balance sheet by a further $326bn than relying on caps alone. The bulk of runoff could occur within the first few months of QT: about 30% of the Fed’s bill holdings mature within one month, 50% mature within two months, and 90% mature within six months.

A more hawkish scenario, though not our baseline expectation, is the Fed will begin running down its bills before implementing the caps. The Fed could frame such a move as a technical adjustment to reduce the amount of excess liquidity for supporting effective policy implementation during the rate liftoff. At the time of these bills’ purchases, the Fed had asserted that its operations were strictly technical and had no material implications for the stance of monetary policy the same argument should apply for their unwind.

A benefit to running off bills before the other securities is that it would allow for a smoother reduction in liquidity, especially if the bulk of the bill portfolio is unwound before the maximum Treasury and MBS caps are reached. Using bills to jump start QT could also give the Fed an early look into money markets’ reaction to declining excess liquidity, which could in turn help the Fed design more effective caps. The risk is that this could trigger a negative market reaction and cause financial conditions to tighten more than desired. This could particularly be the case if the Fed’s assertion that the size of its bill portfolio is uncorrelated to the stance of monetary policy is not understood or rejected by market participants.

Balance sheet details

Based on the caps presented and the bill runoff assumption, our baseline projection is the Fed balance sheet could shrink by $560bn in 2022 and another $1tn in 2023.

Together, these amounts match up to the $1.5tn that Atlanta Fed President Bostic indicated he would like to reduce the balance sheet by relatively quickly. A key variable is MBS redemptions, which we forecast will run at an average monthly
pace of $30bn after QT begins. Slower MBS redemptions could mean less balance sheet runoff, and vice versa. By the end of 2023, we project the Fed balance sheet will fall to around $7.3tn.

For the hawkish scenarios we outlined earlier, shortening the caps’ transition period could result in $620bn of runoff this year and a Fed balance sheet size of $7.2tn by end-2023. Bringing forward the start of QT to June could result in $750bn of runoff this year and a Fed balance sheet of $7.1tn by end-2023.

The terminal size of the Fed’s balance sheet, according to the most recent Balance Sheet Normalization Principles and Plans, will be determined by the amount of reserves “necessary to efficiently and effectively implement monetary policy”. The December minutes suggest that this principle has not changed, although what that optimal level should be is uncertain because of difficulty in estimating the underlying demand for reserves by banks.

The September 2019 repo market ruction revealed that number to be around $1.5 trillion reserves. That number may have increased since then, as payments and hence intraday liquidity needs for banks have risen. On the other hand, that number may have declined as the standing repo facility, for which depository institutions can apply for access, can increase the intraday monetization assumptions banks place on Treasury securities, thus reducing their need for holding reserves.

Complicating matters is the large fluctuation in non-reserve liabilities, which behave autonomously and have direct impact on the level of reserves. Ultimately, the level of reserves for effective policy implementation may be only discovered through persistent usage in the standing repo facility. We will have more to say on this topic in future notes.

In our projections through the end of 2025, the Fed balance sheet would peak just a shade under $9tn after QE completes in March. At the end of the projection period, the Fed balance sheet would be about one-third smaller and approach 20% of GDP, a level more consistent with the pre-covid experience and a potential target highlighted by Governor Waller. Reserve balances would fall to $2.2tn, a decline of $1.8tn from the current level, while the combined liabilities of currency in circulation, repo balances and TGA balances would be $0.9tn smaller.

Finally, an important consideration is the phasing out of MBS and returning the Fed’s portfolio to hold primarily Treasuries. The minutes showed that some participants expressed such a preference and favored making the switch “relatively soon”. As the last QT program wound down, in August 2019 the Fed adjusted its reinvestment policy to use MBS redemptions under the cap to buy Treasuries instead of letting those balances mature uninvested. This continued until March 2020. The same approach remains available, but given the much larger overall Fed holdings today and a longer anticipated time to unwind assets, the Fed may also opt to increase the pace MBS are redeemed relative to Treasury securities by raising MBS caps, or decide to reinvest a portion of MBS redemptions above the cap into Treasuries. If MBS redemptions slow down drastically later on, outright sales of MBS is also a possibility. These remain open questions, and the Fed has provided little guidance on this front so far.

More QT = fewer rate hikes, but how much?

A natural question given the magnitude of the balance sheet runoff we anticipate, is how might this QT-driven tightening of financial conditions substitute for rate increases? In a post-mortem of the December meeting we postulated that one reason the Fed’s SEP was dovish over the longer-term by having the fed funds rate still below neutral in 2024 was that they built in additional tightening via the balance sheet (see A dovish perspective of the Fed’s hawkish pivot). The minutes seemed to confirm that this was indeed one factor behind the SEP projections for some officials, noting that “Some participants commented that removing policy accommodation by relying more on balance sheet reduction and less on increases in the policy rate could help limit yield curve flattening during policy normalization.”

Thus, to get a better sense of the substitutability of QT for rate increases and thus how the Fed’s tightening strategy could impact the yield curve shape, we review some earlier analysis on the topic.

In a 2019 report, we used a variety of approaches to equate Fed balance sheet runoff to rate increases (see Fed balance sheet: Lurking in the shadow (rate)?). Among these approaches were: through our DB shadow rate, through the historical comovements between the balance sheet and the Wu-Xia shadow rate, through the lens of Treasury issuance and the term premium, and finally by equating the two through simulations in the Fed staff’s model of the US economy (FRB/US). These approaches provided a variety of estimates for the equivalence between QT and rate increases. Fortunately, however, these estimates clustered around a finding that an approximately $650-700bn drawdown in the balance sheet was equal to roughly one 25bps rate increase.

Fed staff work focused on this topic yields remarkably similar estimates. For example, analysis from the Fed Board staff which augmented FRBUS to include a role for SOMA found that a 2% of GDP reduction in the Fed’s balance sheet was equivalent to about 20bps in rate increases. 2

Two percent of nominal GDP in 2023 is equal to roughly $525bn on our forecast. Scaling that up to achieve a 25bps rate hike, we would find approximately $650bn in QT is needed to be equal to one increase. Similarly, analysis from the Kansas City Fed staff, which built the balance sheet’s term premium effects into an empirical model of r-star, found that a $675bn reduction in the balance sheet is equal to a 25bps rate hike. 3

Taken together, from a dollar amount perspective, the roughly $1.5tn reduction in the SOMA portfolio we anticipate through the end of 2023 would be consistent with nearly two and a half 25bp rate increases. Alternatively, the roughly 9 percentage point reduction in Fed assets / GDP would amount to about three and a half 25bp rate increases. Despite the relative consistency of these estimates, it is critical to emphasize the uncertainty about the quantitative trade-off between QT and rate hikes. Indeed, the December minutes indicated that officials continued to view the fed funds rate as the Committee’s primary policy tool, in part because “there is less uncertainty about the effects of changes in the federal funds rate on the economy than about the effects of changes in the Federal Reserve’s balance sheet.” Beyond the higher degree of uncertainty around balance sheet policy in general, differences in the Fed’s balance sheet and in the Treasury market could also suggest that this prior experience is not directly applicable to the current episode. For example, the duration of the Fed’s portfolio is lower, consistent with less term premium compression per dollar of QE. In addition, the tightening of financial conditions that will come from a higher implied term premium will depend on how Treasury adjusts its issuance to fund securities maturing from SOMA. As such, we would treat our above calculations as rules of thumb rather than inalterable rules.

Houses and Holes

Comments are hidden for Membership Subscribers only.