The Road to Normal: New Directions in Monetary Policy

Annual Report 2015 | Federal Reserve Bank of St. Louis

The Road to Normal

What Is Monetary Policy Normalization?

Click on the terms in bold to see their definitions in the Glossary below.

In the previous section, we detailed what has been unusual about the state of monetary policy in the United States—an abnormally long period of ZIRP, a very large Fed balance sheet, a Fed asset portfolio that is unusually long in maturity, and large holdings of MBS, which are essentially private assets. So, what will normalization entail?8 A good outline of the Fed’s planned normalization approach was in its “Policy Normalization Principles and Plans,” presented in September 2014.9 Three key elements were in the normalization plan:

  1. Begin increases in short-term market interest rates—trigger liftoff, that is, an end to ZIRP. (The FOMC took this step in December 2015.)
  2. Reduce the size of the balance sheet so that monetary policy works as it did before the Great Recession.
  3. Transform the Fed’s asset holdings to a composition similar to those of pre-Great Recession times. This transformation will involve a reduction in the average maturity of assets and a transition to a portfolio consisting primarily of Treasury securities.

Interest Rate Increases

In normal times, prior to the Great Recession, the New York Fed would control the fed funds rate, according to the directive from the FOMC, through daily open market operations. Monetary economists would describe this as a variant of a channel system for monetary policy implementation. Under a channel system, a central bank sets an interest rate at which it lends to financial institutions (the discount rate in the U.S.) and an interest rate on reserve balances in excess of reserve requirements (IOER—interest on excess reserves), which is then the interest rate at which financial institutions lend to the central bank. Those two interest rates constitute, respectively, an upper bound on the overnight interest rate and a lower bound. In the U.S., no financial institution should want to borrow from another financial institution at an interest rate greater than the discount rate, nor would a financial institution lend at an interest rate less than the interest rate on reserves. Under a channel system, the central bank targets an overnight interest rate to fall between the upper and lower bounds. Prior to the Great Recession, the Fed operated under a channel system with IOER=0.

When a central bank has a large quantity of excess interest-bearing reserves outstanding, monetary policy implementation works differently, as a floor system. In a smoothly functioning overnight credit market, with excess reserves outstanding, IOER should peg the overnight rate because market participants must be indifferent between lending to the central bank and lending to another financial institution overnight. If the U.S. overnight market worked this way, then liftoff would be an easy thing for the Fed to implement. An increase in the IOER would simply increase the fed funds rate one-for-one.

But the U.S. overnight credit market is not a smoothly functioning market. Figure 6 shows the fed funds rate and the IOER since the beginning of 2009. As is evident from the figure, there is a significant gap between the IOER and the fed funds rate. There are several factors that, researchers have argued, explain this gap, including regulatory costs associated with holding reserves for commercial banks, imperfect competition and the fact that government-sponsored enterprises (GSEs) do not receive interest on their reserve balances with the Fed.10 Because the gap is not well-understood, it is difficult to predict what will happen to this gap as market interest rates increase over time. As IOER increases, will the fed funds rate increase more or less in tandem? Will the interest rate margin between the IOER and the fed funds rate increase, or will it decrease?

Figure 6

To deal with this problem, the New York Fed experimented with an ON-RRP (overnight reverse repurchase agreement) facility. Since liftoff, the facility has served as a way to restrict the rate gap. The ON-RRP facility has an expanded set of counterparties, including money market mutual funds and GSEs. In a reverse repurchase agreement, one of these counterparties lends to the Fed, usually overnight, with the Fed posting some securities in its portfolio as collateral. The goal of the ON-RRP facility is to expand the set of financial institutions that can hold interest-bearing Fed liabilities.

Under the FOMC’s normalization plans, the FOMC will continue to set a 25-basis-point range for the fed funds rate, but with the IOER set at the top of the range and the ON-RRP rate at the bottom of the range. On a typical day, the New York Fed currently conducts a fixed-rate full-allotment auction of ON-RRP borrowing, that is, the New York Fed fixes the ON-RRP rate and lets the market determine the quantity of lending to the Fed at that rate. In principle, the ON-RRP interest rate should put a floor under the fed funds rate, with the IOER determining the ceiling on the fed funds rate. The system should then be a modified floor system—a floor with a subfloor—which will allow the Fed to tightly control the fed funds rate.

As shown in Figure 7, the Fed has been successful at targeting the fed funds rate between the ON-RRP rate, currently at 0.25 percent, and IOER, currently at 0.5 percent. (The departure of the fed funds rate from the target range on Dec. 31 occurred because of temporary technical reasons that may recur at the end of each quarter and which are not important to monetary policy.)

Figure 7

Balance Sheet Reduction and Transformation

In the FOMC’s normalization plans, balance sheet reduction was projected to start taking place sometime after liftoff. Further, reduction will occur through the end of the reinvestment program; when reinvestment stops, the assets on the Fed’s balance sheet will mature over time, and the balance sheet will gradually shrink in size. In the process, reserves will fall; the target balance sheet size will have been achieved when reserves fall to a small amount, on the order of what was outstanding before the Great Recession. Balance sheet reduction could occur through outright sales of the Fed’s assets, but there are no plans for this.

How do reserves fall as the Fed’s assets mature? Consider two possible cases. First, suppose that an MBS held by the Fed matures (either because the underlying mortgages mature, a mortgage holder refinances or a mortgage defaults), then the issuer of the MBS makes a payment to the Fed. Supposing that issuer is a GSE (Fannie Mae, for example), the Fed would then debit the reserve account of the GSE at the Fed by the amount of the payment. Effectively, the Fed tears up the IOU of the GSE to the Fed (the MBS), and the Fed tears up the IOU of the Fed to the GSE (reserves); so, there are equal reductions in the Fed’s assets and in its liabilities.

If the asset that matures is a Treasury security, then what happens is a little less obvious. When the Treasury security matures, there is a payment from the Treasury to the Fed, which occurs through a debiting of the Treasury’s reserve account with the Fed. Again, there are equal reductions in the Fed’s assets and liabilities, but in this case there is no reduction in reserve balances held in the private sector, which is what we care about. Such a reduction would occur, for example, if the Treasury wanted to replenish its reserve balances after paying down its debt with the Fed. The Treasury could do this, for example, by issuing new Treasury securities to a private-sector financial institution, which pays for those securities with reserve balances. This would then increase the reserve balances of the Treasury but reduce reserve balances held in the private sector. The ultimate effects would then be the same as for the mortgage-backed security example.

How long will the balance sheet take to normalize once reinvestment stops? Studies by economists within the Federal Reserve System suggest that this process could take seven years or more.11 It will take even longer to reduce the average maturity of the Fed’s assets to what it was before the Great Recession. And this is under the assumption that the Fed will not engage in more QE programs during the normalization process; so, potentially, normalization of the balance sheet could take a very long time.

BACK: The Origins of Unconventional Monetary Policy in the U.S.

NEXT: Why Normalize?


Channel system: a monetary policy system that involves controlling a key market interest rate by bounding it between two central bank-controlled rates; the U.S. version of the channel system involves bounding the federal funds rate between the discount rate and the interest rate on excess reserves. [ back to text ]

Discount rate: the interest rate at which the Federal Reserve lends to financial institutions in its role as lender of last resort; this rate serves as the ceiling for a channel system. [ back to text ]

Floor system: a monetary policy system that exists when there is a large quantity of excess interest-bearing reserves outstanding, implying that the interest rate on excess reserves will theoretically serve as a floor that pegs the federal funds rate. [ back to text ]

Interest rate on excess reserves (IOER): the interest rate received on reserves held at the Federal Reserve in excess of an institution’s reserve requirement; this rate serves as the floor in both a channel system and a floor system. [ back to text ]

Liftoff: the date at which the Fed raised the fed funds rate after seven years at practically zero. The FOMC took this action—the start of so-called normalization of monetary policy—on Dec. 16, 2015. [ back to text ]

Reverse repurchase agreement (RRP): the borrowing of funds by a central bank from a financial institution, generally overnight (ON-RRP), with central bank-owned securities held by the financial institution as collateral until the funds are returned; this monetary policy tool serves as a way to expand the set of institutions that can hold interest-bearing Federal Reserve liabilities. The ON-RRP rate serves as a subfloor under the IOER in the U.S. floor system. Often referred to as reverse repos. [ back to text ]

Reserve requirement: the share of certain types of deposits that must be held in the form of reserves with the Fed by a depository institution. [ back to text ]


  1. See Williamson’s 2015 work “Monetary Policy Normalization in the United States” for further discussion of the normalization process. [ back to text ]
  2. See the FOMC’s 2014 “Policy Normalization Principles and Plans.” [ back to text ]
  3. See Martin, McAndrews, Palida and Skeie, as well as Williamson’s 2015 work titled “Interest on Reserves, Interbank Lending, and Monetary Policy,” for examinations of imperfect competition as an explanation for the interest rate gap. Also, an example of a GSE is the Federal National Mortgage Association, generally referred to as Fannie Mae. [ back to text ]
  4. See Williamson’s 2015 work titled “Monetary Policy Normalization in the United States.” [ back to text ]