The Fed is Stuck on the Floor: Here’s How It Can Get Up.

On Jan. 30, 2019, the Federal Open Market Committee of the Federal Reserve System announced to little fanfare a momentous change in how it conducts monetary policy.  The change was adopted without any formal notice or request for public comment, nor with any formal input from Congress or the Administration.[1]  Nevertheless, the change has had and will continue to have a profound effect on the role of the Federal Reserve in the United States financial system.  To implement policy in its new regime, the Fed not only has to be much bigger but also must continuously grow larger and expand the breadth of its counterparties.  The record indicates that the FOMC did not appreciate the consequences of its decision at the time, and the question now is whether the decision will be revisited given how manifest and serious those consequences are. 

Specifically, the Fed announced that it would conduct monetary policy by over-supplying liquidity to the financial system, driving short-term interest rates down to the rate that the Fed pays to sop the liquidity back up.  Previously, the Fed had kept reserve balances (bank deposits at the Fed) just scarce enough that the overnight interest rate was determined by transactions between financial institutions; those transactions consisted of banks with extra liquidity lending to those that needed it.  Now the rate is determined by transactions between banks and the Fed.  Moreover, the Fed has committed to providing so much extra liquidity that it would not need to adjust the quantity of reserve balances it is supplying in response to transitory shocks to liquidity supply and demand. 

Fed officials argued in 2019 that providing a buffer above the structural demand for reserve balances would make it unnecessary for the Fed to engage in fine-tuning operations; they did not anticipate that the implementation framework would be inherently unstable.  As argued below, over time, banks adjust to whatever amount of reserves the Fed supplies, and any sharp drop in that level is disruptive.  As a result, the amount that the Fed needs to supply to maintain the buffer also grows over time.  Because the supply must keep growing, the interest rate the Fed pays to get banks to hold the reserve balances must rise relative to market rates.  To avoid that premium, the Fed has been forced to expand its counterparties beyond banks.  Thus, the policy automatically results in the Fed’s continued growth and ever-widening entanglement with the financial system.

The Fed’s operating regime also increases the Treasury Department’s borrowing cost.  The Fed is, in effect, converting Treasury securities into Fed liabilities.  Because the Fed has become so huge, it must pay up to borrow – the interest rate the Fed pays on bank deposits is above the interest rate on Treasury bills.  That is, the Treasury, and thus taxpayers, are forced to pay more to borrow when the borrowing is done by the Fed than by the Treasury.

It’s time that Congress provided oversight of the Fed’s decision on its new method to conduct monetary policy and prompted a public debate about its merits.  Thus far the 2019 approach is failing in its purported benefit of allowing the Fed to take a passive approach to monetary policy implementation, is transforming the Fed’s role in society and is costing taxpayers money.  By contrast, the Fed’s previous operating system allowed it excellent control of interest rates with a light touch on the financial system. 

To return to that previous system, the Fed needs to wean the financial system off of so much Fed-supplied liquidity.  To do that, it needs to do four things:  1) make reserve balances relatively expensive by gradually but steadily reducing the amount it is supplying so that market rates rise above the interest rate the Fed pays on deposits; 2) adjust the supply of reserves in response to shocks to supply and demand as the supply trends down; 3) eliminate regulatory and examination biases toward reserve balances and away from Treasury securities as a source of liquidity; 4) keep the new standing repo facility stigma-free by allowing banks to consider rather than ignore its existence for liquidity regulation purposes. 

Policy Implementation

The Fed, like all central banks, implements monetary policy by adjusting its policy rates and the size and composition of its balance sheet.  For the purposes of understanding monetary policy, it is sufficient to think of the Fed’s assets as Treasury securities and its liabilities as currency and reserve balances – the deposits of depository institutions (henceforth “banks”) at a Federal Reserve Bank, as depicted in Figure 1. Because the Fed’s balance sheet must balance, and currency is essentially fixed, when the Fed increases or decreases its holdings of Treasuries, it also increases and decreases the amount of reserve balances.  Those reserve balances can move around the banking system, but the total amount isn’t increased or decreased by those interbank transactions.  At the same time, each individual bank can pick whatever level of reserve balances it wants.  As a result, interest rates on bank liabilities and assets must adjust to leave all those voluntary choices by banks adding up to the total supply of reserves created by the Fed.  That’s how monetary policy works.

Figure 1: Simplified Federal Reserve Balance Sheet

Figure 2 provides the workhorse model used by economists for thinking about how the process works.  The purple line is banks’ demand for reserve balances.  The vertical orange line is the supply of reserve balances (the difference between Fed assets and currency).  The y-axis is the federal funds rate (the overnight, unsecured interest rate at which banks lend each other) that results from interaction of supply and demand.  On any given day, banks will demand a certain quantity of reserve balances for payment needs and precautionary demand.  If the Fed supplies an insufficient quantity of reserves, banks bid aggressively for funding in the fed funds market, driving the rate up.  In theory, the funds rate should not go much above the discount rate, because banks can borrow from the Fed at that rate.[2]  If the Fed supplies more reserve balances than banks want, the fed funds rate will fall to the Interest on Reserve Balances (IORB) rate and no further because no bank would lend money in the funds market for less than it can earn from simply leaving the money on deposit at the Fed.

Figure 2: Reserve Balances and the Federal Funds Rate Corridor System

In a nutshell, the decision the Fed made in January 2019 was to conduct monetary policy by setting reserve supply on the flat part of the curve to the right.  The new implementation method is called a “floor system” because the federal funds rate ends up at the floor of the policy corridor defined by the discount rate and the IORB rate.  In the two decades prior to the Global Financial Crisis, the Fed conducted policy by setting reserve supply in the middle of the steep part of the curve (a “corridor system”).  The primary purported advantage of the shift in how the Fed implemented policy was that it would supposedly make monetary policy implementation much easier.  Shocks that moved reserve demand left or right or moved reserve supply left or right wouldn’t change the fed funds rate where the two lines intersect – the IORB rate.

Figure 3: Reserve Balances and the Federal Funds Rate Flor System

Reality proved quite different, however.  First, it should be noted that the Fed had not found it difficult to conduct monetary policy before the GFC.  Each morning a handful of staff at the New York Fed and the Board would estimate what quantity of reserves would meet banks’ demand that day given other known developments.  Errors in that determination were smoothed out because banks were only required to meet reserve requirements on average over a two-week period.  The staffs in the two Fed groups were actually much smaller under the old regime than they are today.  Policy implementation was so successful that the Fed was criticized by some for not allowing enough volatility in the fed funds rate.

Second, as this note discusses, in practice, setting reserve supply on the flat part of the reserve demand curve has proven far from simple.  The model that guides how the Fed thinks reserve supply and demand interact to determine interest rates is based on intraday market behavior during a period when the Fed did not pay interest on reserves and reserve balances were in the tens of billions of dollars.  That model has not been a helpful guide for how the process works over longer time periods, with the interest rate on reserves at or above market rates, and the quantity of reserves in the multiple trillions of dollars.

Growth

In April 2008, when Federal Reserve staff first considered the possibility of operating policy with an abundant-reserve framework, staff estimated that the level of reserves that would be needed “…might be on the order of $35 billion but could be larger on some days.”[3]  Staff judged that $35 billion (with a “b”) would be sufficient because it observed that on any given day a relatively small overestimate of reserve demand could drive the fed funds rate down to zero.  As a result, staff estimated that a level of reserve supply only slightly above normal would put reserves on the flat part of the demand curve.

The Fed’s estimate of the quantity of reserve balances required to implement a floor system has increased dramatically over time.  In March 2016, the New York Fed raised its assumption about the level of reserve balances needed in equilibrium to $100 billion.[4]  In 2017, it raised it to $500 billion.[5]  The FOMC did not specify the quantity of reserves it thought would be necessary when it made its decision to use a floor system in January 2019, but in March 2018, the New York Fed put its estimate of that level at $600 billion; by September 2019, it had more than doubled its estimate to $1.3 trillion.[6]  It is reasonable, therefore, to guess that the level the FOMC judged to be necessary when it adopted a floor system was somewhere in between those two estimates, perhaps about $1 trillion.

The minutes of the FOMC meetings in late 2018 indicate that the Committee viewed the main disadvantage of implementing monetary policy with a floor system was that it required the Fed to operate with a larger balance sheet. Indeed, at least some members of the Committee indicated that if the needed balance became too large, they would reconsider the policy:

Participants judged that if the level of reserves needed for a regime with abundant excess reserves turned out to be considerably higher than anticipated [likely about $1 trillion], the possibility of returning to a regime in which excess reserves were limited and adjustments in reserve supply were used to influence money market rates would warrant further consideration.[7]

In September 2019, a mismatch in the supply of and demand for repo financing led to a sharp spike in repo rates and unsettled repo markets for a week.  In “Reserves were not so Ample After All,” Copeland, Duffie, and Yang (2021) attribute the dislocation in part to an insufficient supply of reserve balances, then $1.45 trillion.[8]  The New York Fed has estimated that a buffer of at least $350 billion above the structural demand for reserves is needed to account for variation in reserve balances, which would put the amount needed to conduct a floor system at $1.8 trillion.[9]

That estimate has not been tested.  In response to the September 2019 turmoil the Federal Reserve expanded the supply of reserve balances, and the Fed’s massive purchases of government securities in response to the COVID-19 pandemic increased the supply further.  Reserve balances are currently $4 trillion, and reverse repurchase agreements, discussed further below but essentially deposits of money funds and GSEs, are $1.5 trillion.

Figure 4: Postulated Shape of the Reserve Demand Curve in 2008 and 2019

Why Did Demand for Reserve Balances Grow?

Why did the Fed’s estimate of the quantity of reserves necessary to conduct a floor system rise from $35 billion to $1.8 trillion, more than 50 times higher?  The model of reserve supply and demand underpinning the Fed’s intuition was built on observations about bank behavior within a trading day at a time when the interest rate on reserve balances was zero.[10]  Banks developed procedures to minimize their holdings of reserves. If aggregate supply ended up below the minimum level banks needed on that day, interest rates rose sharply in the afternoon; if supply exceeded that level, interest rates fell as every bank sought to shed the expensive-to-hold reserve balances.

Perhaps not surprisingly, that inflexibility within a trading day does not accurately describe banks’ demand for reserves when they have time to adjust their procedures.  It never has.  For example, in the early 1990s, when high interest rates made reserve balances expensive to maintain, banks developed procedures that swept funds out of demand deposits, which had a 10 percent reserve requirement, into MMDAs, which had no requirement, at the end of each day, reducing banks’ need to maintain reserve balances to satisfy reserve requirements.  

Conversely, during and shortly after the Global Financial Crisis, when the Fed had driven the supply of reserve balances up to about $2.75 trillion and paid banks 25 basis points on those balances when repo rates were 10 basis points, banks adjusted their procedures to make use of the cheap and abundant reservoir of liquidity.  In particular, banks used reserve balances rather than available alternatives to satisfy regulatory and supervisory liquidity requirements, and banks held an extra supply of reserve balances to reduce to near zero the possibility that they might need to borrow from the discount window. 

In general, banks switched from making use of the interbank market for liquidity to maintaining large stockpiles of reserve balances.  At least one other central bank understood this phenomenon.  The Norges Bank (the central bank of Norway) judged such a dynamic to be undesirable, and in 2010 switched from a system with abundant reserves to a system with scarce reserves (the reverse of the FOMC’s decision in January). When seeking comment on their decision, they indicated:

When Norges Bank keeps reserves relatively high for a period, it appears that banks gradually adjust to this level…With ever increasing reserves in the banking system, there is a risk that Norges Bank assumes functions that should be left to the market. It is not Norges Bank’s role to provide funding for banks…If a bank has a deficit of reserves towards the end of the day, banks must be able to deal with this by trading in the interbank market.[11]

Moreover, bank regulators and bank examiners also grew used to banks meeting their liquidity needs by maintain large stockpiles of reserve balances.  This was a major change from past practice.  My notes from examiner school in 1994 include no mention of reserve balances as something to consider when evaluating bank liquidity.  Back then, no bank would hold zero-interest-bearing reserve balances as a source of liquidity.  The number one characteristic we were instructed to consider when assessing a bank’s liquidity was the bank’s access to the interbank market.  Indeed, examiners would criticize a bank for holding excess reserve balances that detracted from profitability.  (See A Case Study in Holding Excess Liquidity – Bank Policy Institute (bpi.com)).  However, by 2009, when Basel III was designed, reserve balances were designated as the preeminent liquid asset. 

But at super-abundant levels just as at low levels, banks develop procedures that are costly to adjust quickly that make use of reserves.  One CIO of a large bank described how demand at high levels can get locked in.  When the IORB rate was above the repo rate, the bank decided to hold five days’ worth of liquidity in the form of reserve balances.  In 2017 when the situation reversed – the repo rate rose above the IORB rate – the bank considered reducing its reserve balances to a three-day supply, holding Treasury securities for the remainder.  Although that plan is compliant with regulations, the bank’s treasury staff did not want to have to explain the change to their examiner. 

In other words, the idea that banks will just park excess reserves above some level at the Fed and be indifferent about whether reserve balances decline from that level is incorrect.  When the Fed began to gradually shrink its balance sheet in 2018 from its post QE-3 peak, as noted above, staff expected reserve balances to begin to become scarce at some level below $1 trillion.  Instead, the repo rate moved below the IORB rate – an indication of scarcity – at $1.9 trillion in June 2018.  At that point, the Fed needed to control downward swings in reserve balances caused by factors such as rapid increases in the Treasury General Account.  When it didn’t, in September 2019, money market rates rose sharply, a spike that is easy to understand if the “steep part of the reserve demand curve” is understood to reflect adjustment costs from the normal, prevailing level rather than a time-invariant measure of structural demand.

Figure 5: Spread Between the Fed Funds Rate and the IORB Rate

The Ratchet

If that were all there was to it, the Fed would still be able to shrink gradually, but it would meet resistance sooner than expected and would then have to control volatility in reserve supply to shrink further.  But the situation is worse than that.  As described above, the Fed’s plan is to maintain a buffer of about $350 billion above the level where reserve scarcity begins.  Because the short-run structural demand will rise up toward the prevailing level, to maintain the buffer the Fed will have to steadily increase reserve balances over time by always expanding its balance sheet by more than the growth in currency.

Of course, reality is a bit more complicated.  When the Fed expanded reserve balances by $2 trillion in six weeks in spring 2020, no doubt even short-run structural demand was unable to keep pace.  In fact, the experience illustrated another important element of the demand-supply situation.  Banks’ demand for reserve balances is not perfectly elastic as the quantity of reserves increases.  Instead, market rates end up below the IORB rate by an amount that widens as reserve balances grow.

Banks’ limited ability to absorb reserve balances became clearly evident at the end of March 2021.  On April 1, 2020, the Fed temporarily excluded reserve balances and Treasury securities from the denominator of banks’ leverage ratios – a regulatory measure calculated by dividing equity by assets.  On March 31, 2021, the exclusion ended.  Repo and fed funds rates fell, and money market mutual funds began to invest more in the Fed’s Overnight Reverse Repurchase Agreement facility.  Put another way, the Fed was unable to fund its balance sheet from bank lending (reserves) and turned instead to other lenders.[12]

The ON RRP facility was created by the Fed in 2014 to strengthen the ability of the Fed to raise rates when it decided to do so despite high levels of reserve balances.  The facility effectively extends the Fed’s ability to pay interest on deposits beyond banks to money market mutual funds and GSEs.  Those institutions provide the Federal Reserve funds at the end of the day and receive the funds back with a fixed and known rate of interest (the ON RRP rate) the next day.  The Fed has kept the ON RRP rate below the IORB rate, so the facility has acted as a backstop.

In the spring of 2021, the IORB rate was 10 basis points, and the ON RRP rate was zero.  With the balance sheet continuing to grow, and the Fed now requiring banks to fund their loans to the Fed (a.k.a. reserve balances) in part with capital, repo rates dropped to zero or even below at times.  Money funds began to invest in the Fed at the ON RRP facility, and the facility grew.

Figure 6: Federal Reserve Borrowing: Reserve Balances and Repo

Eventually, money funds reached their capacity to hold zero-interest-bearing investments and the Fed was faced with the prospect of interest rates declining further.  To avoid that outcome, the Fed raised the IORB rate to 15 bp and the ON RRP rate to 5 bp on July 29, 2021. 

The episode illustrates the second horn of the Fed’s dilemma when conducting policy with a floor system.  Once the monkey is on the Fed’s back, to keep a buffer above the steep part of the short-run demand curve, the Fed must keep increasing the supply of reserve balances.  But as the supply moves to the right, the Fed must pay an ever-higher IORB rate to hit any given interest rate target.  With the IORB rate above the interest rate on Treasury securities, the Fed loses money on each additional transaction.  (Correctly) viewing the Fed as part of the government, the Fed is converting Treasury securities, which can be held by anyone, into reserve balances, which can only be held by banks.  At a certain level, reserve balances become a more expensive source of funding for the government than Treasury securities, so taxpayers lose money.

The Fed avoided, or at least diffused, that outcome by borrowing from a wider set of counterparties than merely banks.  By borrowing $1.5 trillion from money funds, the Fed avoids having to raise the IORB rate even further to maintain the fed funds rate in the mid-part of the FOMC’s 0-25 basis point target.  About one-third of money fund assets is now lending to the Fed.  Eventually, even money funds would reach capacity, however.  When money fund capacity to finance the Fed is reached, the Fed could consider opening the ON RRP facility up to corporate treasurers as a place where they can invest their cash.

That step may not be necessary, however, at least for a while, because the FOMC appears likely to begin reducing the size of the Fed’s balance sheet relatively soon.  When it does so, the interpretation of events presented in this note will be put to the test.  If the Fed is able to reduce reserve balances to $1.8 trillion and experience occasional swings down to $1.4 trillion without unwanted increases in money market rates, $1.4 trillion may indeed be an immutable level of structural demand for reserves. 

Reducing the Size and Entanglement of the Fed

If, however, signs of reserve scarcity – market rates edging up above the IORB rate – appear at a higher level of reserve balances, then the Fed will have to make a choice.  On the one hand, the Fed could decide to implement its floor system by continuously expanding its balance sheet so that reserve balances stay ahead of banks’ increasing demand, meanwhile borrowing from an ever-expanding set of counterparties so that the premium over market rates it has to pay to borrow remains manageable.  On the other hand, it could regularly offset shocks to reserve supply to avoid interest rate volatility at whatever new higher level of reserve demand asserts itself.  In that case, though, there would not be any particular reason to stick with a floor system, at least at such a high level of reserve balances, so the Fed may decide it wants to shrink.[13]  To do so successfully, without a replay of September 2019, the Fed would need to do four things.

First, the Fed would need to let reserve balances shrink gradually as its securities mature (or as it also slowly sells assets).  Once scarcity begins to reappear, money market interest rates will move up above the IORB rate, as they had in 2018.  With IORB rates lower than market rates, banks will begin to find ways to economize on their holdings, and bank examiners will also become accustomed to lower levels.

Second, as noted, the Fed would need to engage in more frequent fine-tuning operations.  Sept. 16, 2019, for example, was corporate tax day, a day when reserve balances shrink because money flows out of banks and into the Treasury’s account at the Fed.  While there was nothing new about such flows, since the fall of 2008, the Fed hadn’t needed to add reserves to offset them.  But once reserve scarcity reappears, such sharp drops will have to be prevented, as they used to be.

Figure 7: Intraday Demand for Reserve Balances

Third, to facilitate a decline, the Fed would need to remove any bias toward holding reserve balances as opposed to other sources of liquidity that have become built into examination practices and regulations.  For example, even though banks maintain collateral to cover daylight overdrafts in their accounts at the Fed, and the Fed does not charge for such overdrafts, they have been instructed by their examiners that they cannot expect such an overdraft during resolution or in their internal stress tests.  If they can’t expect an overdraft in resolution, then they had better not run one in normal times, so they have to hold higher levels of reserves to ensure overdrafts won’t happen.  As another example, the primary liquidity requirement included in the post-crisis set of bank regulations, the liquidity coverage ratio (LCR), treats reserve balances and Treasury reverse repos the same, but examiners have made it clear to banks that reserve balances are preferred.

Fourth, the Fed needs to reduce the stigma associated with borrowing from it. Banks maintain trillions of dollars of collateral at the Fed to support borrowing, and the Fed has encouraged banks to include borrowing in their contingency plans.  Nevertheless, banks are not allowed to anticipate borrowing in their stress tests and report that examiners would view borrowing negatively (Bank Treasurers’ Views on Liquidity Requirements and the Discount Window – Bank Policy Institute (bpi.com)).  If banks are more comfortable borrowing when they need to, they will be willing to hold lower levels of reserve balances to ensure they never have to.

The value of making banks comfortable borrowing, thereby reducing their demand for reserve balances, is especially apparent with the Fed’s new Standing Repo Facility.  At the facility, primary dealers and large commercial banks are able to convert Treasury securities, Agency securities, or Agency MBS for reserve balances by executing a repo transaction with the Fed.  One of the reasons the Fed implemented the facility is to attempt to create a stigma-free borrowing option to make it easier for the Fed to shrink its balance sheet – if banks can hold Treasuries or Agency MBS and be confident that they can convert them to reserve balances, banks do not have to hold reserve balances.[14]  Despite this objective, the Fed has not indicated that banks can incorporate use of the SRF into their liquidity stress test projections and resolution liquidity plans.

Conclusion

The floor system is not working well.  It has not made monetary policy implementation easier nor interest rate control better.  Instead, it has required the Fed to keep growing and expanding its set of counterparties.

The Fed can and should get much smaller.  In 2018, banks were beginning to adjust to the fact that market rates had moved above the IORB rate by reducing their holdings of reserve balances.  The money market volatility in September 2019 did not demonstrate that the Fed had encountered its minimum feasible size, just that the Fed had forgotten that it had to control huge swings in reserve balances once those balances became modestly scarce.  To get smaller the Fed needs to wean banks off of reserve balances by shifting the IORB and ON RRP rates below market rates and wean bank examiners off of reserve balances by rooting out and eliminating biases against alternative sources of liquidity.  As part of that effort, the Fed needs to announce that banks can plan on using the new standing repo facility when under stress.

The Fed is on track to stop expanding its holdings of securities by March 2022, and it is currently making plans for whether and how to reinvest payments of principal.  If the Fed allows principal to be repaid without reinvestment, its securities portfolio will decline, reducing reserve balances and the ON RRP.  Most likely, the Fed will reinvest payments of principal for a while before beginning to shrink.  If it were to reinvest in Treasury bills rather than longer-term securities, it would be able to shrink more quickly once it decides to do so.  Consequently, now is a critical time for the Fed to receive input from policymakers outside the Fed and from the public about how it should conduct monetary policy and its role in society. 


[1] For additional background on the Fed’s decision see Selgin (2018) “Floored!: How a Misguided Fed Experiment Deepened and Prolonged the Great Recession,” and Nelson (2018) “Understanding the Fed’s Implementation Framework Debate.”

[2] In reality, banks would pay very high rates to avoid borrowing from the discount window; see the BPI blog post “Bank Treasurers’ Views on Liquidity Requirements and the Discount Window,” Nelson and Waxman (2021) https://bpi.com/bank-treasurers-views-on-liquidity-requirements-and-the-discount-window/.

[3] “Interest on Reserves: A preliminary analysis of basic options,” memo to the Federal Open Market Committee, Interest on Reserves workgroup, April 2008.

[4] https://www.newyorkfed.org/medialibrary/media/markets/omo/omo2014-pdf.pdf

[5] https://www.newyorkfed.org/medialibrary/media/markets/omo/omo2016-pdf.pdf

[6] For further discussion of the Fed’s unfortunate decision to hold back its forecast of its balance sheet and income in March 2019 see “This Is Not What Transparency Looks Like,” Bill Nelson, July 30, 2019. https://bpi.com/this-is-not-what-transparency-looks-like/

[7] FOMC minutes, November 7-8, 2018.

[8] “Reserves Were Not So Ample After All,” Copeland, Duffie and Yang, Federal Reserve Bank of New York staff report, July 2021: https://www.newyorkfed.org/research/staff_reports/sr974.html

[9] “Implementing Monetary Policy: Perspective from the Open Market Trading Desk,” Lorie Logan, May 18, 2017. https://www.newyorkfed.org/newsevents/speeches/2017/log170518

[10] The model of reserve demand and supply is based on “Commercial Bank Reserve Management in a Stochastic Model:  Implications for Monetary Policy” by Bill Poole, which was published in 1968.  The model describes the decision of a bank reserve manager to borrow or lend in the federal funds market at noon on the last day of the two-week reserve maintenance period.

[11] Consultative document, Changes in “Regulation on the Access of Banks to Borrowing and Deposit Facilities in Norges Bank etc.” 1 October 2010, pp 5-6. https://static.norges-bank.no/contentassets/cfc83348f4574a719dd5a4ce70a48840/consultative_document_06102010.pdf?v=03/09/2017123145&ft=.pdf

[12] See the BPI blog “A Very Different Federal Reserve Funding Model,” https://bpi.com/a-very-different-federal-reserve-funding-model/

[13] The Fed could also be encouraged to shrink because the large interest payments it would be making to Wall Street and foreign financial and official institutions could be politically unpopular.  For example, if overnight rates rise just to 4 percent, the Fed will be making annual interest payments of about $250 billion.  The Fed could also be encouraged to shrink because it made substantial losses (easily many hundreds of billions of dollars) on its highly leveraged portfolio of longer-term securities as interest rates rise.  See “Helicopter Money, Fiscal QE, The Magic Asset, and Collateralizing the Currency,” Bill Nelson, CATO Journal, forthcoming.

[14] https://bpi.com/design-challenges-for-a-standing-repo-facility/