In a speech and interview last Thursday, Governor Powell significantly clarified the existing guidance about the size of the normalized balance sheet and gave plausible estimates–that is, in line with ours–about its size under alternative long-run operating procedures. He said he favors the current “floor” system and discussed the interaction between balance sheet developments and funds rate decisions. While his remarks are not official guidance like that presented by Yellen or in official communications such as statements, minutes, or press releases, Powell is a centrist, generally mirrors the Chair’s views, doesn’t talk often, and measures his words very carefully when he does. We, therefore, take his remarks as a preview of what the official guidance will be.
Clarifying the Prevailing Guidance about the Size of the Normalized Balance Sheet
The prevailing guidance on the size of the balance sheet, set out in September 2014, is unclear, even misleading, and certainly stale:
The Committee intends that the Federal Reserve will, in the longer run, hold no more securities than necessary to implement monetary policy efficiently and effectively…
That had been interpreted, especially when first set out, as suggesting that the Committee would shrink the balance sheet to a level consistent with the pre-crisis regime of so-called reserve scarcity. And Board staff who have written about the normalization of the balance sheet have consistently assumed that reserves would return to their pre-crisis level, though perhaps adjusting for the trend growth in reserves over time.1
We have emphasized that the level of reserves (and hence the size of the balance sheet) would depend on the long-run operating framework the Committee decides on. There are two choices under consideration: the pre-crisis regime where reserves are “scarce” and the current regime, referred to as a floor system, where reserves abundant.
Powell’s updated guidance clarifies the September 2014 guidance (italics our emphasis):
In the long run, the balance sheet should be no larger than it needs to be to allow the Committee to conduct monetary policy under its chosen framework.
Powell’s footnote to this sentence points the reader to the FOMC’s guidance from the September 2014 Policy Normalization Principles and Plans. But the caveat, “under its chosen framework,” is not in the original September 2014 guidance. Powell’s addition clarifies the earlier guidance, and we see it as an update to that guidance, though unofficial. That removes the uncertainty about whether the previous guidance suggested
1 For example, see Engen, Eric, Laubach, Thomas, and Reifschneider, David. 2015. “The Macroeconomic Effects of the Federal Reserve’s Unconventional Monetary Policies,” FEDS 2015-005, January; Bonis, Brian, Jane Ihrig, and Min Wei (2017). “The Effect of the Federal Reserve’s Securities Holdings on Longer-term Interest Rates,” FEDS Notes. Washington: Board of Governors of the Federal Reserve System, April 20, 2017, https://doi.org/10.17016/2380-7172.1977; and Carpenter, Seth, Ihrig, Jane, Klee, Elizabeth, Boote, Alexander, and Quinn, Daniel. 2012. “The Federal Reserve’s Balance Sheet: A Primer and Projections,” Finance and Economics Discussion Series no. 2012-56, Federal Reserve Board, August.
that the Committee would return to the pre-crisis operating procedure. Now the guidance explicitly notes that the size of the balance sheet would differ under each of the two potential operating regimes.
Long-run Operating Framework
The discussion in the minutes of the November 2016 FOMC meeting, while indicating that the long-run operating framework could be either the current floor system or the pre-crisis system, clearly emphasized the advantages of the current floor system. That is why we have assumed there is no question that the eventual decision will be to maintain the current floor system. However, participants seemed to be reluctant to give their views on which regime they preferred. Powell, likewise, wouldn’t go as far as committing himself to either regime.
On the one hand, he said, “We have not made a decision about the long-run framework and we are not going to [before beginning runoff]…That’s a key driver of the size of the balance sheet — if you’re going to have a floor system, reserves have to be abundant.” But Powell also made clear he favors maintaining the current floor system: “This approach, sometimes referred to as a ‘floor system,’ is simple to operate and has provided good control over the federal funds rate…[The] ‘corridor’ framework remains a feasible option, although, in my view, it may be less robust over time than a floor system.”
This seemed more like a statement of his personal views than an update of the Committee’s official guidance, but it reinforced our strongly held view that the Committee will choose to remain in the floor system. An important reason that the current system is more “robust” is that the pre-crisis system is incompatible with carrying out asset purchases when the funds rate hits the effective lower bound.
Estimates of the Size of the Normalized Balance Sheet
Now let’s get to the numbers, the size of the normalized balance sheet in dollars.
Presidents Harker and Williams have already given estimates, but Powell’s estimates are more consistent with what we have said, and he provided estimates under the two different operating regimes. First, we note that any figure for the eventual size of the balance sheet must be associated with a date because the normalized balance sheet will grow over time. Harker has said “in the twos sounds about right to me” and Williams has mentioned a range of $2 trillion to $2½ trillion: “It’s not a question of are we going to reduce the balance sheet significantly, it’s really like at the end, is that balance sheet going to have something like two trillion or two and a half trillion, or something like that.” On the other hand, Powell said it is hard to see a balance sheet below $2½ trillion to $3 trillion, setting that as perhaps the level that he sees as being consistent with returning to the pre-crisis framework. Recently, estimates from surveys of primary dealers put the normalized balance sheet at $3.1 trillion at the end of 2025.2
Powell noted that the pre-crisis level of reserves was only about $15 billion. Using that as the starting point, and then assuming growth at the rate of growth of nominal income (as we did in our analysis [link]), reserves would be roughly $30 billion in 2022. Powell, however, used $100 billion in his low reserves scenario, which we see as a return to the pre-crisis regime, and that seems to be the number that the staff and Committee associate with such a regime as well. That is also the level that the New York Fed staff assumed in its projections of the balance sheet under its “lower reserves” scenario for balance sheet normalization.3
The $500-billion figure used by the New York Fed staff was chosen because, while there was uncertainty about which regime the FOMC would choose for its long-run operating framework, this was “a level that falls within a range of estimates amid uncertainty about the future operating framework and demand for reserve balances.” Powell used a similar figure, $600 billion, as an intermediate case, citing responses to the New York Fed’s May 2017 primary dealer survey. This roughly splits the difference between the $100 billion in his low reserves case, which we interpret as a return to the pre-crisis regime, and the $1 trillion in his high
2 “Responses to Survey of Primary Dealers,” Markets Group, Federal Reserve Bank of New York, April 24, 2017. [link] 3 “Domestic Open Market Operations During 2016,” prepared by the Markets Group of the Federal Reserve Bank of New York. April 2017. [link]
reserves case, which we interpret as keeping the floor system. In Powell’s scenario using $600 billion, the balance sheet would be at $2.9 trillion in 2022. We estimated that the level of reserves would be $800 billion in 2023, which led to our estimate of a $3.4-trillion balance sheet.4 The high reserves assumption by NY Fed staff and Powell of $1 trillion would put the balance sheet in 2023 at $3.6 trillion.
So, we conclude that if the FOMC chooses to remain in the current regime, the normalized balance sheet at the end of 2023 will be near $3.5 trillion.
Interaction of Balance Sheet and Rate Normalization
We have emphasized that balance sheet normalization and rate normalization are not merely separate and independent developments proceeding at the same time. Rather, the FOMC is in the process of normalizing monetary policy via two interrelated instruments, the balance sheet, and short-term rates. Therefore, there are many combinations of eventual levels, as well as paths towards those endpoints, for both the balance sheet (and reserves) and the fund’s rate.
Powell’s remarks reflected this theme:
Although the process of normalizing the size of the balance sheet will be in the background, that process will interact with the Committee’s decisions regarding the federal funds rate. As the Fed’s balance sheet shrinks, so debt held by the public will grow, which in theory should tighten financial conditions by putting upward pressure on long-term rates. Any such tightening could affect the Committee’s decisions on the federal funds rate.
He provided the most detailed and clearest discussion to date of the interaction of balance sheet normalization and rate normalization by giving estimates, based on Fed Board staff work, that can be used to infer what we refer to as the substitution effect: what balance sheet developments, including but not limited to runoff, are equivalent to in terms of 25-basis-point funds rate hikes.
He noted, “Model-based approaches to that question estimate changes to financial conditions through increases in the term premium as the balance sheet shrinks. These estimates vary but are generally modest.” But that seems to contradict his remarks about the size of this effect. He cited one Fed staff estimate that the effect of the earlier balance sheet programs was to lower the term premium by 100 basis points as of 2016:Q4. That effect was expected to diminish by 15 basis points by 2017 year-end (to about -85 basis points), even in the absence of the beginning of phasing out of reinvestment.5 This reflects the anticipation effect we have talked about, as well as the ongoing decline in the weighted average maturity of the balance sheet after outright asset purchases were completed. That 15-basis-point rise in the term premium in 2017 might not be much, but the remaining 85 basis points to come are much more substantial! Actually, though Powell didn’t venture into this subject, it would be less than that if the FOMC were to stay with a floor regime and the associated higher level of reserves and a much larger balance sheet. In our view, this would leave a permanent downward effect on the term premium, which we estimate as about 20 basis points. So while that would imply less of a rise in the term premium to come, about 65 basis points, that’s still an awful lot to offset with a lower path of the fund’s rate!
In an earlier commentary, we used the estimates from an earlier Fed staff paper (Engen et al), which also estimated that balance sheet developments would raise the term premium by 15 basis points this year, but starting from -57 basis points as of 2016:Q4 [link]. Once we take into account the permanently higher level of reserves in the floor system, that would leave only about -37 basis points of the effect to go once runoff
4 Both Powell and the New York Fed reported levels of SOMA holdings. There are other assets outside of SOMA securities holdings on the Fed’s balance sheet, which we assumed would total $88 billion in the normalized balance sheet, as they did in 2007. 5 Bonis, Brian, Jane Ihrig, and Min Wei (2017). “The Effect of the Federal Reserve’s Securities Holdings on Longer-term Interest Rates,” FEDS Notes. Washington: Board of Governors of the Federal Reserve System, April 20, 2017, https://doi.org/10.17016/2380- 7172.1977.
begins.6 While participants seem likely to use the more recent estimates in the Bonis et al paper rather than the earlier estimates in the Engen et al paper, we frankly feel more comfortable with the Engen et al estimates.