Before the financial crisis, tightening monetary policy was straightforward. The Federal Open Market Committee (FOMC) would announce a rise in the target for the federal funds rate in the overnight interbank lending market, and the open market desk would implement it with a small reduction in the quantity of reserves in the banking system.
Matters are no longer so simple. The unconventional policies designed first to avert a financial and economic collapse, and then to spur growth and employment, have left the banking system with reserves that are so abundant that it would be impossible to tighten policy in the conventional manner.
So, as the FOMC moves to “normalize” monetary policy after years of extraordinary accommodation – eventually raising the federal funds rate to their projected long-run norm of nearly 4% – how, precisely, will the Fed tighten monetary policy?
The answer is that the mechanics will be fundamentally different from previous Fed tightening cycles. While the nature of the prospective policy tools will be familiar to long-time specialists, their use will be radically different. As a result, the chapter on Fed operations in money and banking textbooks (including ours) will once again be substantially amended to explain this new framework to the next generation of students aiming to understand the U.S. central bank.
This post summarizes why the Fed’s policy mechanics must change and the basics of how the operating framework will function going forward. For those interested in a more detailed version of this discussion, Fed researchers Ihrig, Meade, and Weinbach recently published an excellent primer that is likely to be a reference for years to come.
To understand what is happening, we need to analyze the changing market for bank reserves, where the federal funds rate is determined by the supply and demand of overnight funds. Reserves are the deposits that banks hold at the Federal Reserve. Starting with the situation before the Lehman bankruptcy, at the end of August 2008 reserves in the U.S. banking system totaled only $10.5 billion. Of this, a miniscule $1.9 billion were the excess reserves that banks held as a buffer against unforeseen fluctuations in the liquidity needs of their clients. As one would expect, the demand for these reserves was negatively related to the opportunity cost of holding them. And, since one of the closest alternatives to holding reserves is to make an overnight loan to another bank at the federal funds rate, this interest rate is a good proxy for the opportunity cost. Consistent with the red line in Figure 1, banks demand fewer reserves as the market federal funds rate rises.
The Fed is the monopoly supplier of bank reserves. On any particular day, the Fed chooses the amount of reserves to supply, so that the supply curve is vertical at that level (like the blue line in Figure 1). In August 2008, the Fed faced the kind of downward-sloping demand for its product that typically confronts a monopolist. By buying or selling securities in the market through an open-market operation (OMO), the Fed could increase or decrease the supply of reserves (shifting the supply curve to the right or the left) in order to lower or raise the market federal funds rate. The use of OMOs in small volume – no more than a few billion dollars on any given day – allowed the central bank to hit its federal funds rate target with reasonable accuracy.
Figure 1. The Market for Bank Reserves
Source: Authors.
That was the state of affairs before September 2008. In the aftermath of Lehman’s failure, the Fed lowered its policy target close to zero – near the lower bound for nominal interest rates – and then engaged in several rounds of large-scale asset purchases (LSAPs) that boosted the supply of reserves well beyond the level needed to keep the federal funds rate near zero. Figure 2 depicts the scale of this QE as the gap between points A (where reserves would be sufficient to keep the funds rate close to zero) and B on the horizontal axis. As of July 2015, point B reflects total reserves of $2,591 billion, of which $2,499 billion are excess reserves. That is, total reserves are nearly 250 times what they were seven years ago, and excess reserves are 1,300 times larger!
Awash in excess reserves, banks are now largely indifferent between holding a few hundred billion dollars more or less. Graphically, this indifference is reflected in the flat portion of the banks’ reserve demand curve. Note that in Figure 2 the Fed’s supply of reserves now intersects banks’ demand for reserves far to the right of the level at which reserve demand flattens – so moving the supply curve a bit right or left has no impact on the rate.
Figure 2. The Market for Reserves with Quantitative Easing (QE)
Source: Authors.
Two additional changes occurred at the height of the financial crisis. First, in contrast with the pre-crisis FOMC, which targeted a specific level of the federal funds rate, since December 2008 policymakers have targeted a range (0.00% to 0.25%). And, second, in October 2008 the Fed began paying banks interest on excess reserves (IOER). Since December 2008, this IOER rate has been set at 0.25%.
That’s the backdrop for today: reserves are so abundant that, unless the Fed is planning to sell more than $2 trillion in securities, it cannot get to the point where small OMOs would have any impact on the federal funds rate. Attempting to go back to the pre-crisis framework faces two insurmountable obstacles. First, given the introduction of the payment of interest on reserves, combined with a broad array of other regulatory changes, no one knows much about the demand for reserves, either where the kink is that is shown in Figure 3 or the steepness of the line to the left of the kink. And second, sales of securities on such a massive scale would almost surely lead to unpredictable and potentially serious disruptions in financial markets.
So instead, the FOMC has said that it will employ an alternative set of tools when it chooses to tighten policy and drive up interest rates in financial markets. It also has tested these tools to see how well they work. As a result, the outline of the new policy framework is clear. Nevertheless, there remain notable uncertainties about how well the system will function in practice, so market practitioners (and textbook authors!) will be paying close attention as it is rolled out.
According to the Fed’s plans, the principal device for raising the federal funds rate will be the IOER rate. By raising this rate, the Fed will raise the floor (or “reservation”) rate at which banks are willing to lend. No less important, a higher IOER rate will encourage banks to bid aggressively for funds from other money market participants in order to re-deposit them at the Fed’s riskless IOER rate. So, when the Fed hikes the IOER, other short-term rates should rise as well. Graphically, raising the IOER raises the flat portion of the reserve demand curve shown in Figure 3 (see the dotted red line labeled “New Demand Curve”). This moves the equilibrium between supply and demand in the market for bank reserves from the old target range near point B to the new target range near point C. (And, at least for now, we expect the target to remain a range.)
Figure 3. The Market for Reserves with Interest on Excess Reserves (IOER) and Overnight Reverse Repo (ON RRP)
Source: Authors.
This approach allows the FOMC to raise the interest rate at which banks borrow and lend overnight without altering the supply of reserves. Indeed, the Fed has found a clever way to set independently both the price (the interest rate) and the quantity of the product (aggregate reserves) that it supplies. Most monopolists have to choose one or the other.
The Fed does not plan to shrink its asset holdings and reduce the supply of reserves until well after it has begun raising the target range for the federal funds rate. And, once it does, the plan is to do so by halting the reinvestment of proceeds from its holdings, including those securities that mature. Normalizing its balance sheet and reducing the supply of reserves in this fashion is likely to take quite a few years (a year ago, the average duration of the Fed’s securities was nearly seven years). While the Fed hasn’t ruled out asset sales to speed up this process, it has presented no plans for them either.
The other policy tools that the Fed has said it will utilize are designed to keep the market-determined funds rate in a range close to the IOER rate, or at least not far below it. We know of plans to do two things: engage in overnight (ON) and term reverse repurchase agreements (RRPs), and auction term deposits. (See our earlier post for a description of the workings of RRPs.) Both are free of default risk, since the Fed is the counterparty.
In addition to 22 primary dealers who transact regularly with the Fed, more than 140 intermediaries – including the government-sponsored enterprises (GSEs), money market mutual funds (MMMFs) and a variety of other institutions – have become eligible to participate in RRPs. Since RRPs are essentially a risk-free loan to the Fed and will be offered at an interest rate that is about 25 basis points below the IOER, offering them in sufficient volume will set a floor on the rate at which this broad array of money market participants is willing to lend. In addition, by auctioning term deposits (say, with a maturity of one week and a rate slightly higher than the IOER rate) to banks, the Fed can absorb some overnight bank reserves. Aside from a possible early-withdrawal penalty, these term deposits will have virtually the same characteristics as an equivalent-maturity Treasury bill.
Notice that in Figure 3 we have drawn the flat portion of the reserve demand curve slightly below the IOER rate. This may seem odd, because no bank will lend to a risky counterparty below the riskless IOER rate. Yet, the market federal funds rate has fluctuated below 25 basis points ever since the IOER was set at that level in December 2008. The reason is that some lenders in the money markets are not banks, so they are not entitled to receive interest on deposits at the Fed. (While GSEs can make deposits at the Fed, they cannot receive interest.) So, these nonbank lenders are willing to lend to banks at a rate below the IOER, and it is this lending – from nonbanks to banks – that determines the market federal funds rate when the system is awash in excess reserves. Furthermore, because the FDIC insurance premium that banks pay rises with their reserve holdings, they are unwilling to bid up the rate paid to these nonbank lenders all the way to the IOER rate. Put differently, the borrow-and-deposit-at-IOER process is not costless for the banks, and the resulting market federal funds rate remains lower in equilibrium than the interest paid on reserves. (The paper by Fed economists Ihrig, Meade, and Weinbach that we referenced earlier describes these complex details.)
We think of this operational framework as a new channel or corridor system. In the system used by many central banks prior to the crisis, the overnight interbank lending rate would fluctuate in a channel between the central bank deposit rate (a floor much like the IOER rate) and the central bank lending rate (a ceiling equivalent to the Fed’s discount rate; for a U.S. version of this system, see here). In the case of the Fed going forward, the channel for the market federal funds rate will be between the ON RRP rate (the floor below which no intermediary will lend to a risky counterparty) and the IOER rate (the ceiling).
At first glance, it may be confusing that the IOER rate is both the floor rate for loans made by banks and the ceiling for the market-determined federal funds rate in the new channel system. How are these two facts compatible? The resolution of this apparent conflict is that the funds rate is determined in the market by banks’ willingness and ability to borrow from intermediaries that cannot earn interest on deposits at the Federal Reserve and then re-deposit the borrowed funds at the Fed. Banks’ willingness to undertake this apparent arbitrage is capped by the rate that they receive on excess reserves less the deposit insurance premium that they have to pay on the funds they borrow. As the market federal funds rate rises toward the IOER rate, the profit opportunity shrinks.
As an aside, the discount (or primary credit) rate has little policy relevance in this new channel system. It is still the rate at which banks borrow from the Fed. But since it is 0.5% above the IOER, and the market federal funds rate is below the IOER, few banks avail themselves of that opportunity or are likely to do so in normal times. A quick look at borrowings over recent years (see here) suggests that only small banks borrow, and when they do, the amounts are tiny.
This new channel system sounds quite neat. It allows the Fed to tighten policy when excess reserves are abundant without selling assets. Only time will tell if there are notable flaws in the new design. For example, the Fed is appropriately wary about making ON RRPs available without limit to nonbanks for an extended period. As we noted in an earlier post, doing so could lead to a destabilizing flight from bank deposits in an episode of stress on bank assets. As a result, the Fed hopes to reduce the use of ON RRPs as soon as that is feasible after policy tightening has begun. However, a limited supply of ON and term RRPs may prove insufficient to put a floor under the market federal funds rate.
The bottom line: the old mechanism for policy tightening is no longer functional, and will not be for some years to come, if ever. The principal new tool for policy tightening will be the IOER rate, supplemented by additional instruments designed to absorb funds from banks (term deposits) and nonbanks (RRPs). How well this new mechanism works will only become clear when the Fed actually tightens, so fasten your seatbelts and get ready for the ride.