In recent years, the IMF has released a growing number of reports and other documents covering economic and financial developments and trends in member countries. Each report, prepared by a staff team after discussions with government officials, is published at the option of the member country.


In recent years, the IMF has released a growing number of reports and other documents covering economic and financial developments and trends in member countries. Each report, prepared by a staff team after discussions with government officials, is published at the option of the member country.

The Operational Framework for Monetary Policy

A. Introduction

75. Unconventional policies have been used for a longer period and in greater volume than originally envisaged, with focus now moving toward the Fed’s preparedness to manage financial conditions during normalization. The Fed commenced Large Scale Asset Purchases (LSAPs) in 2009, and has increased its balance sheet from $800 billion to $4.3 trillion. Tapering commenced in early-2014 with the expectation that the LSAP program would gradually be wound down and finish towards the end of 2014, if economic conditions evolve as expected.

76. The FOMC has indicated that it wants to reduce the size of its balance sheet over time and return to targeting short-term interest rates. It first outlined its Exit Strategy Principles in June 2011, and suggested that it wanted to return the quantity of bank reserves to ‘the smallest levels that would be consistent with the efficient implementation of monetary policy.’40 It reviewed its Exit Strategy Principles and normalization plans in May 2013 and April 2014, noting that the Federal funds rate may not be the best indicator of the general level of short-term rates, and that new tools may be needed to improve control over short-term rates.41

77. This paper assesses operational issues during normalization and for the longer term. The pre-crisis framework is described, preparedness for exit is assessed, and suggestions offered on the shape of the post-normalization framework.

B. Some History

The Pre-crisis Framework

78. Pre-crisis, the Fed targeted the federal funds rate, an overnight unsecured interest rate.42 Typically policy announcements had an immediate impact on money market rates, and there was little long-term relationship between excess reserves, which were very small ($1 billion to $2 billion), and trends in interest rates (Figure 1). There was however a tight connection between the level of excess reserves and interest rates within the reserve maintenance period. Excess reserves were low, reflecting a low level of demand given a high opportunity cost (excess reserves were not remunerated) in the context of a well functioning money market and efficient payments infrastructure.

Figure 1.
Figure 1.

Excess Reserves of Depository Institutions and Selected Interest Rates

(Monthly average)

Citation: IMF Staff Country Reports 2014, 222; 10.5089/9781498331104.002.A004

Sources: Federal Reserve Bank of St. Louis and IMF calculations.

79. The Fed conducted monetary policy in the context of a small structural liquidity deficit using a standard set of instruments.43 The bulk of the deficit came through the Fed’s policy of increasing its holdings of U.S. Treasury securities as the stock of Federal Reserve notes rose over time, as well as the imposition of the reserve requirement on depository institutions. In addition, the discount window (rarely used) was available to depository institutions, and fine-tuning open market operations (OMO) to adjust the supply of reserve balances with fluctuations in demand, were conducted on a daily basis with a small set of primary dealers.44

Figure 2.
Figure 2.

Composition of Reserve Requirements

(Monthly average)

Citation: IMF Staff Country Reports 2014, 222; 10.5089/9781498331104.002.A004

Post-Lehman Balance Sheet Expansion

80. In response to the financial crisis, and in an effort to stabilize financial markets and promote economic recovery, the Fed introduced major changes to its operational framework:

  • To maintain control over short-term money market rates while adding liquidity to counter growing financial stability risks, the Fed started remunerating reserves balances in October 2008, which was earlier than an initially planned date of 2011.45

  • In December 2008 the fed funds target was changed from a point target (1.0 percent) to a range target (0-0.25 percent).

  • Initially various Fed lending programs and then ongoing LSAP programs significantly increased the volume of excess reserves (Figure 3). The interest rate paid on excess reserves (IOER) has been 0.25 percent for some time, while the expansion in excess reserves put downward pressure on both the fed funds rate and repo rates resulting in both trading below this level.46 The fed funds rate has generally remained in a range of 0.05 to 0.15 percent in recent years.

Figure 3.
Figure 3.

Reserves of Depository Institutions and Selected Interest Rates

Citation: IMF Staff Country Reports 2014, 222; 10.5089/9781498331104.002.A004

81. In parallel to those moves, fed funds transactions contracted sharply and with a change in the nature of activity. Initially, markets fragmented as uncertainty about the health of the banking system led to a flight to quality and less trading between financial institutions. Subsequently, the introduction of the IOER and the increase in excess reserves reduced institutions’ need to actively manage liquidity. As a result, fed funds daily activity fell from an estimated $250 billion in 2006 to $60 billion at the end of the 2012. The nature of the lending to the market also changed, with Federal Home Loan Banks (FHLB), that do not have access to the IOER, now estimated to provide 75 percent of fed funds lending, up from a pre-crisis estimate of 40 percent.47 The other large entities that do not have access to IOER—Fannie Mae and Freddie Mac—appear not to have been active in the fed funds market since 2011. U.S. branches of foreign banks now represent a bigger share of borrowing in the market. This is attributable in large part to differences in capital requirements and the FDIC’s expansion of its deposit insurance assessment base (2011) 48 that increased the effective cost of fed funds for domestic banks and undermined the latter’s ability to arbitrage the IOER. Deposits held by U.S. branches of foreign banks are generally not insured.49 Despite all of these changes, the Fed considers that fed funds rate is still connected to other money market rates and remains a good indicator of banks’ marginal funding costs.

C. Normalizing the Policy Stance

The Challenges Ahead

82. The Fed has made a number of changes in preparation for rate increases, the timing of which they have stressed will be data dependent—the markets are pricing in the first rate rise in mid-2015. The changes (below), including the testing of new instruments, are to ensure the Fed is operationally prepared to tighten conditions when the time comes:

  • The number of reverse repo counterparties has been increased to 139 (18 banks, 6 Government Sponsored Enterprises, 94 money funds, and the 21 primary dealers).50

  • Testing of term reserves draining instruments: Deposits51 and reverse repos.

  • Testing of Overnight Fixed Rate Reverse Repos (ONRRP). This instrument, first introduced in September 2013, could be considered either as similar to a standing facility—if accessible without restriction, or an OMO—if offered with an allotment cap. The instrument has been tested at fixed rates from 1 to 5 basis points and up to a recently increased cap of $10 billion per counterparty. Testing is partly aimed at identifying how the instrument impacts money market rates and intermediation flows, and its effectiveness in establishing a floor for overnight market rates.52 The evidence to date suggests that the instrument has been effective in setting a floor under rates (Figure 4).

Figure 4.
Figure 4.

Selected Interest Rates and the Fed’s ONRRP

Citation: IMF Staff Country Reports 2014, 222; 10.5089/9781498331104.002.A004

Source: Bloomberg and IMF estimates.

83. In preparation for normalization, decisions and clear communication regarding two key parameters are vital: the operating target and its positioning.

  • The fed funds rate should remain as the operational target during most (if not all) of the normalization period. The prior is to change the operating parameters only where there is a clear benefit in doing so. At this point, there seems no clear benefit in moving away from the fed funds target, while there could be risks in doing so, given the uncertainties about the behavior of interest rates in different market segments going forward. The use of the fed funds rate in financial contracts would also suggest that any change should be communicated well in advance, to allow market participants sufficient time to adjust their contractual arrangements.

  • The Fed should target a single rate in the context of a floor system; with the target rate set equal to the IOER. While a fed funds target range (0–0.25 percent) makes sense at interest rates close to zero, continuing with a policy range once the tightening phase has begun could undermine the clarity of the policy signal. If a policy range were to be retained, there is a question of whether the Fed would be indifferent to the fed funds rate trading anywhere within the announced range, or whether the mid-point of the range was in fact the implicit target. And if it were the latter, then why not return to a single-point target; and in the context of a floor system, the fed funds target would equate with the rate set on the IOER.53

  • The ONRRP, with an allotment cap to mitigate financial stability risks, would be the primary monetary policy tool to engineer an increase in the fed funds rate to, or slightly above, the IOER (Box 1). With uncertainty about the ONRRP rate that is consistent with a fed funds rate trading at the IOER, the Fed could gradually increase the ONRRP rate until it achieved the fed funds objective. Periodically, given seasonal influences in the net-issuance of securities, the ONRRP rate may need to be changed to control the fed funds rate as desired, but any such change in the rate would not signify a change in policy – to be clear, the ONRRP is a monetary policy tool and not a target.

  • Using the new tools in an environment of abundant liquidity could result in more variation in the fed funds rate from the target than in the past, which should not be of concern given clear communication. Moving back to a single rate policy target could be challenging but would not need to be achieved immediately upon announcement. A modest first step in the tightening cycle would be from the current target of 0–0.25 percent to 0.25 percent. That the fed funds rate does not immediately adjust to the target should not be a concern given an initial small change in the policy target, when combined with a credible commitment to achieve the target over the short-term (maybe one to three weeks). The Fed should also communicate clearly about the likely challenges of controlling the fed funds rate, suggesting that somewhat more volatility is possible, especially early on in the tightening cycle given the uncertainty about the demand for its new tools.

84. Preparations for the effective control of money market conditions during normalization appear to be on track, but there are uncertainties ahead:

  • The ONRRP may not always be sufficiently effective to move the fed funds rate to the IOER. It is not certain that the counterparty list has been sufficiently expanded to control rates in a rising rate environment. Testing so far suggests that ONRRP rate has placed a floor under repo rates but volumes have been low with rates still compressed close to zero. It is not clear whether significantly larger volumes would be required once tightening begins, in order to have the desired impact on the fed funds rate; and if larger volumes were required, potentially more counterparties may also be needed. Further, limited access (12.45pm–1.15pm) may undermine the efficacy of the ONRRP once policy rates are increased, as up to half of tri-party repo (the bulk of the repo market) activity is reportedly conducted after 3pm (although as noted earlier the ONRRP has thus far set an effective floor).

The Mechanics of Draining Reserves: the ONRRP and the Fed Funds Rate

The fed funds rate is determined by the interaction between the supply and demand for fed funds (Fed Funds Figure below) and the Fed’s policy interest rates.1 For any given demand curve, Fed operations that drain liquidity—whether overnight or for a term—move the supply curve left, resulting in an increase in the fed funds rate.2 Currently the fed funds rate trades below the IOER because of: 1) abundant liquidity (S1), and 2) constraints on some banks’ ability to arbitrage between the providers of fed funds that are not able to receive the IOER and the IOER. These constraints are largely regulatory in nature—capital, leverage, liquidity, and the FDIC levy—but may also reflect an unwillingness of the GSEs to make placements with certain banks.

If the Fed wanted to increase the fed funds rate to a level at or slightly above the IOER, then it would have to reduce the supply of fed funds by magnitude “A.” This objective could be achieved in principle by offering the fixed volume ‘A’ of term sterilization instruments, but in doing so the Fed would have to accept the market clearing rate on the instruments. This approach is problematic from two perspectives: 1) significant uncertainty about the demand curve for fed funds makes it difficult to assess “A,” and 2) term instruments would incur a term premium.

The ONRRP is a fixed rate instrument offered by the Fed. The Fed’s demand for funds (i.e., the scale of ONRRP operations) is perfectly inelastic while the market’s supply curve of funds is upward sloping (ONRRP Figure below). There are two impacts here: 1) The ONRRP rate sets a floor under repo rates; participants would not deal at a rate lower than what the Fed was offering. The solidity of this floor is however contingent on the breadth of the counterpart list – which has been addressed through the increase in the number and type of counterparties. 2) Funds placed in the ONRRP are funds that are withdrawn from banks, therefore the supply of fed funds falls (the fed funds supply curve moves left), putting upward pressure on the fed funds rate.

The Fed is acquiring information about the market’s supply curve for the ONRRP by varying the rate but the information to date is limited given the highest rate offered is 5 basis points, and the volumes have been capped. Consequently, there is considerable uncertainty about market behavior when ONRRP rates approach the IOER. However, when the time comes for an interest rate rise, the Fed can move slowly, increasing the ONRRP rate in small steps until the desired fed funds rate is achieved. This process would further reveal the supply curve for ONRRP and the demand curve for Fed funds, recognizing that these curves move over time. Market commentators have variously suggested that the ONRRP rate could be set at 0 to 15 basis points below the IOER. A driver of this spread is the FDIC levy which is around 12 basis points; a spread lower than this level provides the money market mutual funds with an advantage (since they don’t pay the levy) allowing them attract more funds from banks thereby putting upward pressure on the Fed funds rate. And a higher spread could well result in the fed funds rate remaining below the IOER. But only when the time comes, will it be clear where ONRRP rate will need to be, to exert the desired pressure on the fed funds rate.

The ONRRP offers Fed counterparties with a safe asset given the use of Treasuries and Government guaranteed mortgage backed securities, and with the Fed as counterparty. The supply of this safe asset is limited only if the Fed limits access to the ONRRP, for example with the allotment cap (currently set at $10 billion per counterparty). Financial stress would increase the demand for safe assets—moving the ONRRP supply curve to the right (S2), and the fed funds supply curve to the left. The more acute the financial stress the further the curves move, with potential to push banks to the Fed’s Primary Credit Facility (S3)—the demand curve for fed funds would also move up exacerbating the impact on the fed funds rate. Counterparty caps limit the extent to which the ONRRP supply curve could move to the right, and as such are useful to contain disintermediation pressures in the banking sector. And lowering the rate on the ONRRP also could reduce flows into the instrument. However, it is unclear that fine-tuning the counterparty cap and/or the ONRRP rate would provide a sufficiently effective response in a severe stress scenario; a more targeted approach may be required depending on the circumstance.

1 The demand curve is a stylized illustration of what is may look like.2 To the extent the demand curve is sloping; at current levels of oversupply it is virtually flat.
  • Financial stability risks arising from disintermediation of the banking sector.54 These risks could arise if deposits are attracted out of banks and into money market mutual funds—increasing the size of the less-tightly regulated shadow banking system.55 Further, in times of financial stress there could be runs into money funds—exacerbated by some funds having publicized their access to an unlimited volume of safe assets (i.e. the ONRRP). Risks of disintermediation increase if the instrument is offered without an allotment cap.

  • Operational challenges of conducting large volumes of overnight transactions. Each day, transfers of securities and cash will be required. While only small volumes of transactions are currently conducted, much larger volumes may be required to move the fed funds rate closer to the IOER (maybe in excess of $1 trillion), which could potentially double the current tri-party repo volumes. Such a surge in volumes would increase the operational and financial risks related to settlement.56

Steps to Improve Implementation and to Mitigate Risks

85. Measures could be considered to increase the solidity of the interest rate floor:

  • Further expanding the number of counterparties, if rates traded below the ONRRP rate. There could be important segments of liquidity unable to access the instrument (e.g., a large number of investment intermediaries which fall below the threshold to become counterparties); this could be addressed by further expanding the counterparty list. The resulting increased risks of disintermediation (i.e., funds flowing out of banks into money market funds) would need to be mitigated by:

    • Maintaining and managing the allotment cap on the ONRRP. The allotment cap impacts only on the non-banks, as banks will not use the ONRRP because of access to IOER (which pays a higher interest rate). The caps should be maintained to mitigate financial stability risks (see Box 1) and be set at a level that allows the Fed to meet its operating objectives. A uniform cap across counterparties is the easiest to administer, while recognizing proportionally larger counterparties may be disadvantaged with this approach.

    • Managing the ONRRP rate, relative to the IOER. A significant move of funds out of banks would push the fed funds rate up which could be countered by a lowering of the ONRRP rate. The ONRRP will probably be set below the IOER to compensate for the FDIC levy and other regulatory costs that fall on banks but not on money market funds.

  • Extending the timing of the operation. When the ONRRP is not available there would be no effective lower bound on rates. Therefore, it may be necessary to extend the ONRRP operation to much later in the day when a significant volume of overnight deals are done.

Table 1.

Summary of Instruments’ Costs and Impacts

article image

This assumes the current policy of remunerating required reserves at the same rate as excess reserves, although the Fed could choose to apply different rates to each category of reserves.

A standing facility is an arrangement offered by the central bank that can be utilized at the discretion of depository institutions and generally without restriction. Payment of interest on excess reserves (IOER) has these characteristics.

The FOMC has suggested there could be residual MBS sales after normalization.

86. Term sterilization instruments would reduce the heavy reliance on overnight operations but would carry additional cost.57, 58 A mix of term deposits and term reverse repos could be used to tighten liquidity conditions, thereby reducing segmentation and operational risks. Term deposits, would incur a term premium above the IOER, while term reverse repos would incur a term premium above the ONRRP (but could still be below the IOER). The liquidity premium on term reverse repos is to some extent, a function of the liquidity constraints on money market mutual funds, something that could be mitigated by attaching a 7-day put option to the instrument; something the Fed has also tested.

D. Considerations for the Future Shape of the Operating Framework

87. The current juncture provides an opportunity for a broader review of the operating framework. The post-exit framework will be different to that of today, and probably that of the pre-crisis period. Abundant liquidity currently limits operational choices, yet it should still be possible to implement changes along the path to normalization; i.e. that being the point where the Fed’s balance sheet reverts to a steady state.

88. Differences in interest rate targeting frameworks come down to a few key areas: (1) Specification of the operating target; (2) positioning of the operating target; (3) instrument design; (4) counterparty selection; and (5) collateral policy. Each of these issues except collateral policy is discussed in subsequent sections and Table 2 provides a country comparison.59

Table 2.

Operational Features of Interest Rate Targeting Regimes

article image

More recently liquidity has been withdrawn at the weekly operation owing to a structural surplus.

Open Market Operations (OMOs) comprise a range of transactions conducted by the central bank including repos, reverse repos and outright purchases and sales of securities.

Specification of the Operating Target

An implicit or explicit operating target?

89. Central banks can influence financial conditions by targeting interest rates, either implicitly or explicitly.60

  • With an implicit target there is no market rate announced as a target. The policy target is tied to a central bank instrument in the expectation that short-term money market rates will remain close to the level at which there is commitment to add or withdraw liquidity.61 Idiosyncratic movements in individual market rates therefore matter less. However, market rates may move away from the policy rate when the interbank market is fragmented and some banks have a higher demand for liquidity than others (e.g., Euro area).

  • An explicit target involves a commitment to guide an identified market rate (or rates) consistent with an announced policy target. This option may require a more active approach, as operational credibility relies on central bank actions to contain deviations of the targeted rate from the announced target.

90. The Fed should continue to announce an explicit operating targeting. Central banks have successfully employed both implicit and explicit approaches to the operating target (Table 2). As neither approach has been proven demonstrably superior, the Fed should retain an explicit target recognizing there is no clear advantage in changing.

Options for an explicit operating target

91. Post-normalization, the choice of operating target will be guided by an assessment of the relevance of interest rates in different segments of the money market. And, with more regulatory changes to be phased in over the years ahead, there is uncertainty about how the different segments will function. Three main options for targeting short-term interest rates are available—the fed funds rate (an unsecured overnight rate), a repo rate (a secured overnight rate) and a composite (reflecting money market conditions with indicators of which rates are most important).

Maintaining the fed funds rate

92. The fed funds market has shrunk and it is uncertain by how much it will recover. Risk aversion during the crisis reduced activity in the unsecured markets globally, in absolute terms and relative to activity in secured markets. In the U.S. a number of regulatory changes—capital, leverage, liquidity, and the FDIC levy—require banks to have more stable funding, relative to short-term financing. And there is evidence of diminished transmission of the fed funds rate to longer rates, but this should be treated with caution given the prolonged period of compressed rates.62 Undermining the case for retaining the fed funds target is the likelihood that activity in the unsecured market is likely to remain subdued. It could be retained if activity did recover, and providing there was a sufficiently close relationship with movements in other money market rates.

The General Collateral Repo Rate (GCRR)

93. The treasury GCRR is an alternative to the fed funds rate.63 The value of tri-party repo transactions is around $1.7 trillion with about one third relating to the treasury general collateral category, although the breakdown between overnight and term transactions is not available.64 However, using the FICC-GCF repo data65 as a proxy, around one third of the transactions may be overnight—equating to $190 billion (compared to fed funds estimated at $60 billion). With money market mutual funds active in tri-party repos and other segments—short-term US treasury and corporate securities and the bi-lateral repo market—arbitrage ensures that changes in the GCRR are widely transmitted. Also relevant is the transmission from the GCRR to the repo yield curve, which appeared to remain more stable when compared to the unsecured lending curve.66

94. The Fed could exert control over the GCRR—albeit with added complexity given the role of collateral. Pre-crisis, the GCRR responded predictably to movements in the fed funds rate, but this relationship weakened at the onset of the crisis as risk premiums and liquidity increased.67 Repo rates are affected not just by changes in liquidity, but also by seasonal influences affecting the net-issuance of treasury securities.68 While abundant liquidity has weakened the short-term liquidity effects, when excess reserves run down, the liquidity effects should strengthen, with arbitrage activity recovering. With a balance sheet back at steady state, the Fed should then be able to manage the GCRR in a similar way to the fed funds rate pre-crisis, albeit with added complexity given the role of collateral and the uncertainty about the extent to which arbitrage activity recovers.

A composite approach could be considered

95. Individual rates—including the GCRR—have been hit by idiosyncratic shocks complicating policy implementation.69 Going forward, the GCRR is certainly a viable option for a single rate operating target, yet it was impacted during the financial crisis (2008) when risk aversion resulted in a sharp fall in the rate, and again during the debt ceiling negotiations (2013) when, as the default risk on short-term treasuries spiked, so too did the rate. Rigid implementation responses in these circumstances, when a single rate is targeted, could exacerbate volatility in other markets undermining policy signals and effectiveness.

96. To get around this problem the Fed could target the general level of short-term rates, while providing clear guidance on which rate was the most important indicator. When a single-targeted rate moves but other short-term rates do not (or move by much less), from a policy perspective, there is no need for an immediate response to bring the rate back to the target; yet failure to act could pose communication challenges and risk credibility. To get around this problem, a combination of money market rates could be considered. The GCRR could be of primary importance—and communicated as such—but the fed funds rate and EURO dollar rates could also be used with lesser emphasis, as well as term rates (repo and bank certificate of deposits).

97. A less rigidly specified target rate would need to be clearly communicated. In choosing between a single rate target and one more loosely defined, the trade-off is clear: signaling clarity vs flexibility. The signaling challenges can be met by communicating which rates are important, and why. And it should be noted that a moderate amount of volatility in all rates is expected and would not undermine the operational objective.

Positioning of the Operating Target

98. Interest rates are targeted either at the floor or the mid-point of the corridor. Prior to the financial crisis many central banks targeted the mid-point of the corridor, but liquidity injections aimed at addressing financial stability concerns subsequently pushed rates to the bottom of the corridor (a floor system) in a number of cases (e.g., BOE, ECB, and U.S.). In the U.S., bank reserves are currently remunerated at the top of the announced policy range (0.25 percent) with discount window borrowings available at a 50 basis points higher rate (0.75 percent). During normalization there is little choice for the Fed other than to adopt a type of floor system—determined by the interest rates applied to IOER and the ONRRP. As structural liquidity declines however, there will be the option to revert to a mid-corridor system.

99. There is a good case to retain the floor system beyond normalization because it is operationally less complicated and more robust through the risk cycle (Appendix I). The potential loss of bank information from subdued trading—an often-cited benefit of retaining a mid-corridor system—is countered through the stringent reporting requirements of the new liquidity regulation. And retaining the floor system post-normalization reduces the need for change once the Fed’s balance sheet has reverted to a steady state.

Instrument Design

100. The supply of reserves should be just sufficient to keep rates at or slightly above the floor, to ensure institutions have some liquidity risk to manage. To illustrate (Figure 5), with a mid-corridor system and an assumed policy rate of 50 basis points and amount of reserves at S2 would be required to meet the policy target. In the case of a floor system and with a policy rate of 25 basis points, then S1 volume of reserves is required. Strictly defined, implementation of a floor system would require the level of reserves to be at some point to the right of S1; thereby always pushing rates always to the floor. However, more liquidity means less liquidity risk, and less market activity. Therefore, a less rigidly applied approach to the floor system with liquidity provided at S1, would retain some liquidity tension, increasing trading, but with a consequence that the rate may on occasion trade above the floor.

Figure 5.
Figure 5.

Floor versus Mid-Corridor Systems

Citation: IMF Staff Country Reports 2014, 222; 10.5089/9781498331104.002.A004

101. Standing facilities at the floor and the ceiling of the corridor are needed:

  • The IOER will again become effective as the large structural surplus shrinks, allowing for a withdrawal of the ONRRP in its current form. Active arbitrage will reduce market segmentation, thereby establishing the IOER as an effective floor under rates. While high liquidity and segmentation necessitated the introduction of ONRRP—to fix the floor—a return to tighter conditions means this instrument is no longer needed, and given the disintermediation risks it should be removed as soon as possible.

  • The Primary Credit Facility (PCF) establishes a ceiling rate, but this may involve stigma. The PCF is for highly rated banks and intended to provide a safety valve for liquidity pressures. 70 It still may not be fully effective because of stigma as the Fed must, with a two year lag, disclose its lending activity.

102. Instruments to manage short term liquidity fluctuations are still needed. Fine-tuning operations will need to offset seasonal influences given the GCRR may respond to changes in both reserves, and the net-issuance of collateral. The Fed would need to use its portfolio of treasury securities through repo operations best offered as a variable rate/fixed volume format. This approach differs to its operations pre-crisis in that while the operations were variable rate, it did not announce an amount, thereby providing greater flexibility to determine an outcome consistent with its interest rate objective. If the Fed wishes to continue with this approach, information should—as before—be provided post-auction on the results of operation, to help market participants better assess liquidity conditions in support of more stable market conditions. Noted is that the Fed’s auction facilities during the crisis involved detailed pre- and post-auction information on volumes and rates.71

103. The reserve requirement no longer meets any monetary policy or liquidity management objective and can be withdrawn (Appendix II). Eliminating the reserve requirement would free up reserves, thereby increasing excess reserves and reducing the need to inject additional liquidity.72 There would also be administrative savings.

Counterparty selection

104. Counterparty selection can be differentiated on the basis of instrument type:

  • Standing facilities: The IOER and the discount window facilities are limited to depository institutions, and this should not change.

  • Open market operations: Regulatory developments (LCR and the supplementary leverage ratio) are likely to have undermined the Fed’s ability to implement policy through the traditionally narrow set of primary dealers. However, beyond normalization, and in the context of an increased focus on the GCRR, the Fed may need to transact in larger volumes. With these considerations the expanded counterpart list is appropriate and should be retained, with an additional benefit of enhanced competition in the Fed’s OMOs. A caveat being, only banks should have access to standing facility-type instruments (i.e. fixed rate instruments with or without limits) to mitigate disintermediation and subsequent financial stability risks.

E. Summary

105. The recommendations for the normalization period are summarized:

  • Continue with the fed funds rate as the operating target of monetary policy.

  • Announce that a floor system will be used during most, if not all, of the normalization period, and therefore the rate set on the IOER will equate with the fed funds target.

  • Use the ONRRP, with counterparty allotment caps, as the primary monetary policy tool to manage the fed funds rate at or slightly above the IOER.

  • Assess the need for further expanding the counterpart list if the ONRRP is not sufficiently effective.

  • Manage the dis-intermediation and shadow banking risks of the ONRRP by announcing that it is unlikely that the instrument will be used post-normalization and that allotment caps may be changed to contain flows.

  • Re-instate the single rate policy target; with the first modest move from the current 0-0.25 percent to 0.25 percent. Communicate the prospect that given abundant liquidity, there could be somewhat higher variation around the policy target than in the past.

  • Explore the use of term sterilization instruments—deposits and reverse repos—to lessen operational risks. Small term premiums could be justified to reduce the burden of a daily rollover of large transaction volumes.

106. Post-normalization considerations:

  • Consider alternate operating targets after it is clear how markets have adapted to the regulatory changes. Consider de-emphasizing the importance of a single rate in favor of focus on the general level of money market rates, while highlighting what the Fed considers to be the most important indicators (e.g., GCRR, Fed funds rate).

  • Continue with the floor system.

  • Withdraw the ONRRP: because the IOER will provide an effective floor once liquidity conditions tighten.

  • Use short-term repos and reverse repos to manage the operating target close to floor.

  • Abolish the reserve requirement, as it provides no benefit and is administratively cumbersome.

Appendix I: Floor versus Mid-Corridor Systems

This appendix outlines the considerations when deciding between implementing a floor system, where the policy target rate is set at the floor of the corridor, and a mid-corridor system where the policy target rate is set away from the floor (but not necessarily at the mid-point).

Floor System


There may be less variability between market and target rates. The combination of abundant liquidity and an effective floor anchors rates at the floor, thereby aligning market rates more precisely with the targeted rate.1

Higher reserve balances (than required in a mid-corridor system) may facilitate an increased supply of high quality liquid assets (HQLA).2 Higher reserves levels could help alleviate a shortage of HQLA—although not a current concern in the U.S.—and support the function of the payments system.

The price/quantity nexus is removed, thereby providing more flexibility in times of stress. The IOER was introduced in 2008 at time when the Fed sought to put a floor under rates, while simultaneously increasing liquidity to counter mounting financial stresses. In these situations rates trade at the floor, facilitating a break in the price/quantity nexus—quantity can be increased—to a point—without impact on price (subject to the solidity of the interest rate floor and market segmentation issues which will impact at some point). A floor arrangement is therefore likely to be more robust across different phases of the risk cycle.

Less operational resource and fewer monetary instruments are required. Larger reserve balances would mean less need for accurate forecasting of the influences on reserves, and most likely, less fine-tuning operations to offset those influences.

There would be a positive impact on central bank profitability given the somewhat larger balance sheet. The higher level of reserves, incurring interest costs at the policy rate, would be matched against term assets earning the term premium.


Abundant liquidity reduces both liquidity risks and the incentive to trade in interbank markets, with a potential loss of information. The question arises whether less activity reduces the incentives and scope for peer-monitoring, leading to a loss of information about individual banks solvency, with negative financial stability consequences. While some literature supports this view,3 others highlight that because interbank trading is very short-term, there is little focus on the long-term solvency of the borrowing bank.4 Active interbank markets did not prevent the recent financial crisis (much of which was bank-sourced), so the benefit of active interbank markets should not be overstated.

Mid-corridor System

Two-way liquidity risk encourages trading and perhaps better transmission along the yield curve. With a greater incentive to manage liquidity, transmission along the short part of the yield curve would be stronger, given that rates are not forced to the floor.

Liquidity management is more challenging but could be mitigated by the use of a reserve requirement (or some form of contractual reserves) with averaging. With no liquidity buffer, frequent fine-tuning operations would be required to manage supply against forecasted demand in order to meet the targeted rate. Reserve requirements with full averaging would reduce interest rate volatility that would arise when supply and demand were not aligned on a particular day. A contractual reserves approach requires banks to reveal their demand for precautionary reserves, and combines incentives for them to manage their position in a manner consistent with their stated reserves position.5

Appendix 2: A Short History of and the Case Against Reserve Requirements


The use of reserve requirements in the USA can be traced back to voluntary redemption arrangements in the 1820s. Bank notes were used as a medium of exchange, but limited information on the solvency of the issuer meant the geographical coverage for an individual bank’s notes was narrow. Banks in New York and New England agreed to redeem each others’ notes at par, providing the issuing bank maintained sufficient specie (gold or its equivalent) at the redeeming bank. This first use of required reserves was, in essence, for prudential reasons.

On a nationwide basis reserve requirements were first established under the National Bank Act (1863). The charter established a network that banks could join, requiring them to hold a 25 percent reserve against bank notes and deposits. This network facilitated greater countrywide acceptance of notes of the participating banks. From that point until the establishment of the Fed, reserve requirements continued to be viewed as a prudential measure to support the liquidity of bank notes and deposits (under the prevailing gold standard, by linking them to physical gold reserves), yet financial panics still occurred.1 By 1931, after the establishment of the Fed, and with it a lender of last resort function, reserve requirements were used to influence the expansion of bank credit and were no longer viewed as a prudential tool. Membership of the Fed was optional for state-chartered banks and some began leaving the system to take advantage of lower reserve requirements imposed by state authorities. By the late 1970s, less than 65 percent of total transaction deposits were held at Fed member banks. Congress introduced the Monetary Control Act (1980) mandating the Fed to set reserve requirements universally across all depository institutions.

From the early 1980s, the Fed aimed to influence monetary and credit conditions by adjusting the cost of reserves to depository institutions. Actual reserve balances were stabilized around the minimum level needed to meet requirements and clearing purposes (as reserves were not remunerated, banks kept balances as low as possible). Individual banks could however, contract to hold more reserves if they deemed their clearing needs were higher than the requirement, and this excess earned credits that could be offset against Fed priced services. The reserve requirement was averaged over the two-week maintenance period for larger banks and one week for smaller banks.

The Fed was first able to remunerate reserve balances in October 2008.2 The rate paid on required reserves was originally set at the average of the Fed funds target over the maintenance period less 10 basis points, while excess reserves were remunerated at the lowest Fed Funds target rate during the period, less 75 basis points. These margins were removed from early 2009: both categories of reserves are now remunerated at 25bp, the top end of the fed funds target rate.

The Case for Abolishing Reserve Requirements in the U.S.A.

The current arrangements are complex and administratively burdensome.3 Depository institutions report either on a weekly or quarterly basis, depending on the size of their net transactions accounts and other accounts.4 Reserves are maintained over a two-week period, which in the case of institutions reporting weekly, is based on two reporting periods and lagged to the computational period by 17 days. The computation period for institutions that report quarterly is the 7-day period beginning on the third Tuesday of the report month, with compliance lagged by four, or in some cases five weeks. The reservable base covers net transactions accounts split in three tranches—up to $13.3 million is reservable at 0 percent (referred to as the low reserve tranche), $13.3.million to $89 million is reservable at 3 percent, and more than $89 million is reservable at 10 percent.5 The low reserve tranche is adjusted each year by 80 percent of the annual increase or decrease in net transaction accounts at all depository institutions. The impact of reserve requirements was reduced in 1990 when the ratio on non-personal time deposits and Eurocurrency liabilities was set to zero.

Institutions can satisfy their reserve requirements through holdings of vault cash and reserve balances.6 Vault cash covers in excess of 40 percent of the requirement for the system as a whole, and more than 100 percent of the requirement for some banks. A penalty-free band is calculated as the greater of $50,000 or 10 percent of the reserve requirement. When an institution has more than the sum of its reserve requirement plus the penalty free band, it is deemed to be in excess.7 If it has less than the reserve requirement minus the penalty-free band, then it is in deficit and will be subject to penalty set at the rate on the Primary Credit Facility plus one percentage point.

Reserve requirements in the U.S.A. no longer play a prudential or monetary role and may not be needed for liquidity management. Many central banks now operate an effective operational framework without reserve requirements.8 Other central banks that use them aim to stabilize the demand for reserves, thereby improving liquidity management outcomes. The key features of these arrangements are requirements that result in reserve levels close to that which is voluntarily demanded for a given remuneration rate (typically at the policy rate or up to 25bp below it), a relatively long maintenance period (around a month), and full averaging.

Monetary policy could be implemented effectively in the USA using a floor system and without reserve requirements.9 Liquidity management in a floor system involves keeping interest rates at, or close to the floor. Consequently there is a one-way risk of not meeting the operational target—assuming that there is an effective interest rate floor, the market rate can only be too high. With rates pushed to floor of the corridor, there is less of a requirement to fine-tune operations, and as a result, less need to accurately forecast and stabilize the demand for reserves. However, there is still a need for periodic operations, perhaps in both directions (injecting and withdrawing liquidity) as the objective should be to over-supply liquidity only by a small margin, to ensure that money market activity is not undermined by a high volume of excess liquidity (as it is currently). A flexible form of required reserves (contractual reserves) where banks reveal their demand for reserves could however be useful in a mid-corridor system to reduce uncertainty.10


  • Armantier and Sporn, Federal Reserve Bank of New York Staff Report 635 September 2013: Auctions implemented by the Federal Reserve Bank of New York during the Great Recession.

    • Search Google Scholar
    • Export Citation
  • Basle Committee on Banking Supervision, Basle III: The Liquidity Coverage Ratio and liquidity monitoring toolsJanuary 2013.

  • Bech and Keister: BIS Working papers No 432. October 2013: Liquidity regulation and the implementation of monetary policy.

  • Bech, Klee and Stebunovs 2011: Arbitrage, liquidity and exit: The repo and federal funds markets before, during and emerging from the crisis.

    • Search Google Scholar
    • Export Citation
  • Bernhardsen and Kloster, Norges Bank, 2010: Liquidity Management System: Floor or Corridor.

  • Board of Governors of the Federal Reserve System: Reserve Maintenance Manual November 2013.

  • Dudley, William, President FRBNY Speech May 20, 2014: The Economic Outlook and Implications for Monetary Policy.

  • ECB Monthly Bulletin April 2013: Liquidity regulation and monetary policy implementation

  • Feinman Joshua: Reserve Requirements: History, Current Practice, and Potential Reform: Federal Reserve Bulletin June 1993.

  • Gagnon, Joseph E. and Sack B, 2014, “Monetary Policy with Abundant Liquidity: A New Operating Framework for the Federal Reserve,” (January).

    • Search Google Scholar
    • Export Citation
  • Gray, Simon: IMF Working Paper WP/11/36: Central Bank Balance and Reserve Requirements.

  • Klee and Stebunovs December 2012: Target Practice: Monetary policy implementation in a postcrisis environment.

  • Liberty Street Economics December 2, 2013: Who’s Lending in the Fed Funds Market?

  • Liberty Street Economics December 9, 2013: Who’s Borrowing in the Fed Funds Market?

  • Potter, Simon, Executive Vice President Federal Reserve Bank of New York, December 2, 2013. Recent Developments in Monetary Policy Implementation.

    • Search Google Scholar
    • Export Citation
  • Rochet and Tirole 1996: Interbank Lending and Systemic Risk.

  • The main features of the monetary policy frameworks of the Bank of Japan, the Federal Reserve and the Eurosystem: Conference paper May 2000.

    • Search Google Scholar
    • Export Citation

Minutes of the FOMC meeting held June 2011.


Minutes of the FOMC meetings held April 30–May 1, 2013 and April 29–30, 2014.


Simon Potter: Executive Vice President Federal Reserve Bank of New York, December 2, 2013. Recent Developments in Monetary Policy Implementation: (Potter speech).


The main features of the monetary policy frameworks of the Bank of Japan, the Federal Reserve and the Eurosystem: Conference paper May 2000.


Primary dealers, usually numbered between 20 and 25, and were required to be a banking organization or a registered securities dealer in good standing with the regulator. Their duties included making markets to the Fed Trading desk, supporting the U.S. Treasuries market, and providing market intelligence.


The Financial Services Regulatory Relief Act 2006 authorized the Fed to pay interest on reserves beginning October 2011. This authority was superseded by the Emergency Economic Stabilization Act 2008 bringing forward the authority to October 2008.


The IOER does not act as a firm floor to the Fed Funds rate because only depository institutions have access to the IOER. Government Sponsored Enterprises do not have access to IOER and are large sellers of Fed funds.


Liberty Street Economics December 2, 2013: Who’s Lending in the Fed Funds Market?


In December 2013 U.S. branches of foreign banks held $1 trillion of the reserves at the Fed representing 43 percent of total reserves but accounted for only 13 percent of banking assets in the U.S.A.


Liberty Street Economics December 9, 2013: Who’s Borrowing in the Fed Funds Market?


These counterparties account for 25 percent of all overnight Treasury tri-party repo volume—(Potter speech).


All banks are able to participate in term deposit auctions and not just those in the reverse repo counterparty list.


FOMC Minutes: September 17–18, 2013.


The Fed should move away from the current approach of paying interest at the top of announced interest rate range. In all other cases, central banks pay interest on reserves to signal and set a floor under market rates.


William Dudley, President FRBNY Speech May 20, 2014: The Economic Outlook and Implications for Monetary Policy.


Regulatory arbitrage is likely given that banks are subject to capital, liquidity and FDIC costs and other counterparties are not.


Because haircuts are not applied to reverse repo transactions, the Fed is exposed to market risk if a counterpart defaults and if the value of the security has increased – albeit a small risk in the context of overnight transactions. Further large volumes of overnight deals increase the intra-day exposures in the tri-party system, something that has been the focus of the Tri-party Repo Infrastructure Reform Task Force to reduce.


Some central banks use foreign exchange swaps as a liquidity management instrument but this is not feasible for the Fed as it has few foreign reserves to swap into dollars and borrowing the reserves could be costly.


Higher reserve requirements could complement the toolkit during normalization. However statutory limitations on the types of reservable accounts and the maximum level restrict the potential impact.


With Fed’s balance sheet returning a small deficit and with abundant supply of US government securities, there no requirement to expand the collateral pool used for normal monetary operations. Further, the collateral policy for the Discount Window (LOLR) is well defined and sufficiently broad to meet current needs.


Conventional monetary policy works through actual and expected real short-term risk free rates while unconventional policy aimed to impact expectations of short-term rates (through forward guidance) and the term premium through the asset rebalancing channel (LSAPs).


This excludes transactions at central bank standing facilities which are priced at penalty rates.


Klee and Stebunovs December 2012: Target Practice: Monetary policy implementation in a post-crisis environment—page 22.


The GCRR is based on a daily survey conducted by the FRBNY which is not publicly available. The DTCC General Collateral Finance (GCF) repo rate for treasury securities is suggested as a good proxy for the GCRR: Klee and Stebunovs 2012.


FRBNY Tri-party repo data April 9, 2014. Data is obtained on the 7th business day each month, selected as being typical business day.


Fixed Income Clearing Corporation (FICC)—General Collateral Finance (GCF) repos are transactions between dealers.


Klee and Stebunovs December 2012: Target Practice: Monetary policy implementation in a post-crisis environment—page 22.


Bech, Klee and Stebunovs 2011: Arbitrage, liquidity and exit: The repo and federal funds markets before, during and emerging from the crisis—Tables 4, 5 and 6.


Klee and Stebunovs 2012: Target Practice: Monetary policy implementation in a post-crisis environment – page 20.


Incorporating Financial Stability Considerations into a Monetary Policy Framework Jeremy Stein, March 21, 2014: Highlights the frequency of events that impact term and credit risk premiums.


A Secondary Credit Facility is available to lesser rated institutions with restrictions on the use of fund—it is therefore more akin to a lender of last resort arrangement. It is currently priced 50 basis points above the PCF.


Armantier and Sporn: Federal Reserve Bank of New York Staff Report 635 September 2013: Auctions implemented by the Federal Reserve Bank of New York during the Great Recession


The amount of reserve balances used to meet reserve requirements in March 2014 was $78 billion.


An important operational decision is the extent to which liquidity is oversupplied, as greater oversupply reduces the prospect of rates moving away from the floor.


The impact on HQLA will depend upon how the reserves are supplied to the system. If they are supplied through the purchases of HQLA, then there will be no net change in the volume of HQLA. However, HQLA will increase when non-HQLA transactions are undertaken (e.g., repos with non-HQLA or foreign exchange swaps).


Rochet and Tirole 1996: Interbank Lending and Systemic Risk.


Bernhardsen and Kloster, Norges Bank, 2010: Liquidity Management System: Floor or Corridor?


The BOE pre-crisis, allowed banks to nominate their reserves levels at the start of each maintenance period.


The U.S.A. experienced a number of financial panics during 1869, 1873, 1893, and 1901.


The Financial Services Regulatory Relief Act (2006) originally authorized the Fed to pay interest on reserves beginning October 2011. This authority was superseded, as a result of the financial crisis, by the Emergency Economic Stabilization Act (2008), bringing forward the authority to October 2008.


Board of Governors of the Federal Reserve System: Reserve Maintenance Manual—November 2013.


There are two further categories of reporting—annual and non-reporters. Institutions in these categories have net transactions accounts of less than the amount specified as the low reserve tranche.


Net transactions accounts are total transaction accounts, less amounts due from other depository institutions, and less cash items in the process of collection. Total transactions accounts include demand deposits automatic transfer service accounts, NOW accounts, telephone and pre-authorized transfer accounts, and others.


There are also pass-through arrangements that allow institutions to meet their requirements through a correspondent bank.


The distinction between required reserves and excess reserves is less relevant now that both categories of reserves are remunerated at the same rate.


From the IMF – ISIMP 2013 survey, 10 central banks implement monetary policy without using reserve requirements, these include Australia, Canada, Denmark, Mexico, New Zealand, Norway and Sweden.


Current examples of counties using the floor system are Norway, and New Zealand.


In 2006 the Bank of England (BOE) operated a mid-corridor system and introduced a contractual reserves, system which forced banks to reveal their demand by nominating the amount of reserves they intended to hold during a given maintenance period. The maximum amount a bank could nominate was limited to the lower of; two percent of their sterling liabilities, or £3 billion. Banks were remunerated at the policy rate when they held average balances through the maintenance periods of within +/- 1 percent of their nominated amount. No remuneration was paid if the balances were above this range, and a penalty was charged for balances below this range. This system was suspended during the financial crisis, as the BOE embarked on a policy of quantitative easing which pushed rates to the floor of the corridor.

United States: Selected Issues
Author: International Monetary Fund. Western Hemisphere Dept.