The U.S. economy continues to heal even though the COVID-19 pandemic persists. GDP growth was 6.5% in the second quarter of the year, and the economy has regained its end of 2019 level of GDP. The labor market continues to rebound, with over 900,000 jobs created in July, which drove the unemployment rate down to 5.4%. The speed of recovery has led to an unexpected large spike in inflation, although we believe this will be transitory. While both the level of activity and the number of jobs are still below the pre-COVID-19 trend, the economy is on a solid path to recovery (see "Credit Conditions North America Q3 2021: Looking Ahead, It's Looking Up," June 29, 2021).

With these developments in growth, inflation, and the labor market, the policy narrative has shifted--from aggressively fighting the economic effects of the pandemic, to exit strategies. Key is formulating an end game and a sequence of steps to get there in a way that minimizes potential market disruptions. In the case of monetary policy, this means a shift from extraordinary measures--policy rates at the effective lower bound (ELB), ongoing asset purchases, and an abnormally large balance sheet--to a post-COVID-19 normal. What seems clear at this point is that in the post-COVID-19 steady state, monetary policy will not be a familiar sight. In particular, discussions of the terminal balance sheet have been missing, and this article seeks to fill that gap.

Not Your Father's Central Bank Balance Sheet

To employ an overused phrase, post-COVID-19 monetary policy will feature a new normal, particularly regarding the balance sheet. To see this, consider as a reference point the pre-global financial crisis period. Then, the central bank policy rate was always above the ELB and the balance sheet was relatively small, at 6.0%-6.5% of GDP. This balance sheet comprised Treasury bills on the asset side and mostly currency in circulation on the liability side. Bank reserves, government deposits, and other liabilities were fractional (see tables 1 and 2).

Importantly, at that time, reserves held by banks were small (0.1% of GDP), and the Fed intervened in the interbank market to ensure its target fed funds rate prevailed. That meant selling short-term Treasury bills to retire/lower reserve balances and drive interbank rates higher or buying Treasury bills to create/raise reserve balances and drive rates lower. Quantitative easing (QE)--buying longer-term Treasuries and creating "excess" bank reserves--was not needed since the policy rate had never reached the ELB. At that time, QE was used only in Japan and was seen as an anomaly.

Table 1

The global financial crisis changed the old normal, and the COVID-19 shock further enlarged the Fed's balance sheet. Beginning in late 2008, the policy rate has spent a cumulative period of over eight years at the ELB. Since more monetary stimulus was required by the Fed to achieve its policy objectives during this period, it employed four rounds of QE, taking the size of the balance sheet to $8.2 trillion at the end of July 2021 (36% of GDP). Owing to QE, the size of excess reserves has ballooned. And to make things more complicated, the "desired" holdings of bank reserves by banks has increased as well owing to regulatory requirements and a reassessment of their own liquidity needs in a post-global financial crisis world.

These developments have changed the end game for the post-COVID-19 balance sheet. In a 2019 speech, Fed Chair Jerome Powell identified the forecasted year-end 2019 balance sheet as a potential "new normal" (the smallest level consistent with conducting monetary policy efficiently and effectively), comparing it with the pre-COVID-19 balance sheet at year-end 2006 and the then-maximum sized balance sheet of year-end 2014. In addition to this data, we have inserted data for mid-2021 to get a sense of the sharp, further increase in the balance sheet in the COVID-19 era (see table 2).

Table 2

Before we present our assumptions and simulations, a few observations on Chair Powell's end game balance sheet are in order. Overall, in terms of GDP, the balance sheet will be at least 10 percentage points higher post-COVID-19 than pre-global financial crisis. While some of this corresponds to currency demand, which typically grows a bit faster than GDP, most of it does not. In particular, reserve balances will not revert to the fractional pre-global financial crisis levels for reasons noted above. The Fed will also carry higher Treasury deposits and other liabilities, including reserve repurchase obligations.

Normalization: The Sequence

Now that we have a sense of the end game, what does monetary policy normalization entail and how will it be carried out? There are three components: i) tapering, which brings new bond purchases down to zero; ii) policy rate liftoff, which raises the federal funds rate above its ELB; and iii) rightsizing the balance sheet, which means unwinding QE and presumably bringing bank reserves down to a steady state that Chair Powell (in 2019) identified as 5.7% of GDP.

A strict ordering of these three components is not necessary, although they are often presented as such. As argued separately by both us and economist Benjamin Friedman, money policy is now two dimensional. That is, the central bank has two (active) instruments--the policy rate and asset purchases--at its disposal at all times to obtain its policy objectives. These two instruments can be used independently of each other, although for many strategic objectives, they should be directionally aligned.

Indeed, clear communication from central banks on the joint use of these two instruments is important. And the Fed has, in broad terms, laid out the sequence for normalization noted above--tapering, liftoff, and rightsizing. The first two of these have been discussed actively, while the third has not. Factors determining when the balance sheet normalization is achieved are based on:

• Level: Target reserve balances held at the Fed (deemed operationally "efficient and effective"), and
• Speed: Whether the Fed elects to taper reinvestment (passive quantitative tightening [QT]), fully stop reinvestment, or allows for outright sales (active QT).

The Fed has yet to communicate fully on both fronts. Does Chair Powell's 2019 identification of reserves level at steady state (5%-6% of GDP) still hold? Active QT was not part of the balance sheet normalization announced in the previous tightening cycle. Will time-contingent or state-contingent caps be employed on the reinvestment amount? What will the maturity composition of asset holdings look like once normalized?

In the next section we look at how the rightsizing of the balance sheet might play out. We will not consider the path of the fed funds rate. Since the policy rate and asset purchases are independent instruments, by not addressing the former we implicitly assume that its path toward a neutral rate in the post-COVID-19 steady state is not affected by the paths toward the balance sheet end game.

Simulating The Balance Sheet End Game

Rightsizing the balance sheet is the final step of monetary policy normalization post-COVID-19. We define this step as ending when QE is unwound to the point bank reserves are at their assumed steady state of 5%-6% of GDP. Regarding the steady state, Fed officials learned in the 2018-2019 QT period that the short-term rate could rise by more than they intend as they continue to reduce the size of the balance sheet, since the amount of steady state reserves is unobservable. This may make them likely to err on the side of caution regarding the speed of reducing the balance sheet. (See more in "The Federal Reserve's Balancing Act," published Nov. 19, 2019.)

How do we get there? The dynamics of the bank reserves can be explained by two variables--currency in circulation and assets. Other balance sheet items, such as changes in government deposits and other minor liabilities, are more or less steady and can be ignored, along with the central bank's capital and other minor items. The central bank balance sheet identity is:

Starting with the (largely) autonomous element, cash in circulation has typically risen slightly faster than nominal GDP. If the central bank's assets are fixed, then as currency rises, excess reserves will decline pari passu, all else constant. Moving to the second element, the number of excess reserves in the system is the counterpart to QE, so the central bank's policy on asset unwind is a key variable in the simulations. As assets roll off, bank reserves are extinguished. Asset rolloff can be achieved through tapering the reinvestment of maturing assets, fully stopping reinvestment, or outright sales of bonds. The first two options are passive run-off (the Fed's preference from the last QE unwind), and the last option is active shrinkage. The faster the rolloff of assets, the sooner QE ends.

How long does the normalization process take? The transition period lasts as long as there are reserves in excess of the steady state level in the system. The elimination of such excess reserves is the end state, but that state will likely leave a reserve balance at 5%-6% of GDP, in tandem with short-term rates managed using the current system of floor rates that includes the reverse repo program and the interest on excess reserves. A complete set of assumptions appears in the Appendix.

Two examples (one moderate and another aggressive unwind) illustrate the interaction of changes in currency and assets in determining the post-COVID-19 normalized balance sheet and reserves:

A more plausible--and, in our view, effective--scenario is one where the central bank unwinds gradually by tapering reinvestments (a playbook the Fed developed during the last monetary cycle).  After the end of tapering in 2022, we assume that the central bank keeps its balance sheet roughly constant (by reinvesting maturing assets) during 2023 to monitor policy rate liftoff from the ELB. Following this period, the Fed starts balance sheet normalization in a gradual manner, keeping in view the maturity schedule of its asset holdings. Assets begin to roll off in 2024, beginning with $25 billion in the first quarter and rising by $25 billion in subsequent quarters to $75 billion per quarter by third-quarter 2024. This $75 billion per quarter pace, or $300 billion per year, of rolloff is assumed through 2030, after which policymakers would likely decide to reduce the pace of rolloff the next several quarters as they approach the steady state, at which time the balance sheet will gradually expand again (organically with currency liabilities). In this scenario, the balance sheet reaches its new normal level by the end of 2032, when reserves reach 5.7% of GDP, far above the 0.1% of GDP before the global financial crisis but close to the year-end 2019 level of 7% (see chart 1).

Chart 1

Chart 2

A more aggressive scenario is one where the central bank actively unwinds its assets.  As in the first scenario, after the end of tapering in 2022, we assume that the central bank keeps its balance sheet roughly constant (by reinvesting maturing assets) during 2023 to monitor policy rate liftoff from the ELB. Following this period would be a faster escalation of asset rolloff--rising by $30 billion per quarter beginning in 2024, to $360 per quarter at the end of 2026. (This equals $120 billion per month, the current buying rate of assets, as tapering commences.) Policymakers would likely decide to ease the pace of rolloff the next several quarters as they approach the steady state, at which time the balance sheet will gradually expand again. We assume assets decline gradually in 2027 and bank reserve balances will be close to $1.7 trillion (5.7% of GDP) by the end of the year (see chart 3).

Chart 3

Chart 4

In both scenarios above, with reserves at a steady state level, central assets will be managed to ensure the target fed funds rate prevails in the interbank market and will rise in line with currency demand. Variations in Treasury deposits and other liabilities will have an impact as well, but those are assumed to be basically constant in our scenario.

An aside: In an implausible, illustrative scenario where the central bank adopts a completely passive strategy of no changes on the asset side, rolling over and reinvesting assets as they mature, and reverting back to a "zero" excess reserves operational framework, excess reserves would eventually be eliminated from the system but could take decades. For example, if the currency-to-GDP ratio rises by 0.2 percentage point per year, and excess reserves amount to 19% of GDP at the start of the simulation period in 2023, it will take 95 years of this passive strategy to bring excess reserves down to zero.

It's Time To Consider The Path Toward Policy Normalization

The macro picture has improved markedly in 2021, albeit with some setbacks on the pandemic front. As such, it is time to commence laying the groundwork for policy normalization. Monetary policy must now aim to ensure a credible recovery from COVID-19 while achieving monetary objectives. This means shifting from damage control mode to normalization mode.

The bottom line is that balance sheet normalization--essentially unwinding QE--will take time. In our main scenario, excess reserves (above the operationally efficient level, assumed at 5.7% of GDP) are eliminated in 2027, five years after tapering ends. More passive scenarios could take a decade or more. Even with aggressive asset reduction and a much larger balance sheet target than before the financial crisis, attaining this goal will take years.

The length of the transition is also important from a macro-credit perspective. QE was a necessary distortion designed to raise asset prices and lower the term structure of interest rates once the effective lower bound was reached. On balance, it helped to cushion the blows of both the global financial crisis and COVID-19. But continued QE beyond the crisis weakens credit discipline by potentially prolonging the life of marginally viable firms, thereby dampening dynamism, holding down innovation, productivity, and potential growth.

Related Research

S&P Global Ratings' research

• Credit Conditions North America Q3 2021: Looking Ahead, It's Looking Up, June 29, 2021
• The Federal Reserve's Balancing Act, Nov. 19, 2019
• A Future For QE: Monetary Policy In Two Dimensions, June 12, 2019

Other research

• Has the Financial Crisis Permanently Changed the Practice of Monetary Policy? Has It Changed the Theory of Monetary Policy?, Benjamin M. Friedman, National Bureau of Economic Research