The new tools of monetary policy

Chairman Ben S. Bernanke responds to a reporter the question-and-answer portion of the press conference on September 18, 2013. The event followed the September 17-18 meeting of the Federal Open Market Committee (FOMC).
Editor's note:

Ben Bernanke, a Distinguished Fellow in Economic Studies at Brookings Institution, delivered the 2020 American Economic Association (AEA) Presidential Address on the new tools of monetary policy on January 4, 2020.

Since the 1980s, interest rates around the world have trended downward, reflecting lower inflation, demographic and technological forces that have increased desired global saving relative to desired investment, and other factors. Although low inflation and interest rates have many benefits, the new environment poses challenges for central banks, who have traditionally relied on cuts to short-term interest rates to stimulate sagging economies. A generally low level of interest rates means that, in the face of an economic downturn or undesirably low inflation, the room available for conventional rate cuts is much smaller than in the past.

This constraint on policy became especially concerning during and after the global financial crisis, as the Federal Reserve and other major central banks cut short rates to zero, or nearly so. With their economies in freefall and their traditional methods exhausted, central banks turned to new and relatively untested policy tools, including quantitative easing, forward guidance, and others. The new tools of monetary policy—how they work, their strengths and limitations, and their ability to increase the amount of effective “space” available to monetary policymakers—are the subject of my American Economic Association presidential lecture, delivered January 4, 2020, at the AEA annual meetings in San Diego. As I explain below, my lecture concludes that the new policy tools are effective and that, given current estimates of the neutral rate of interest, quantitative easing and forward guidance can provide the equivalent of about 3 additional percentage points of short-term rate cuts. The paper on which my lecture is based is here. Below I summarize some of the main conclusions.

Central bank purchases of longer-term financial assets, popularly known as quantitative easing or QE, have proved an effective tool for easing financial conditions and providing economic stimulus when short rates are at their lower bound. The effectiveness of QE does not depend on its being deployed during a period of market turbulence.

Quantitative easing works through two principal channels: by reducing the net supply of longer-term assets, which increases their prices and lower their yields; and by signaling policymakers’ intention to keep short rates low for an extended period. Both channels helped ease financial conditions in the post-crisis era.

Studies that control for market expectations of the level and mix of planned asset purchases find that later rounds of QE remained powerful, with effects that did not diminish during periods of market calm or as the central bank’s balance sheet grew.

Early rounds of QE, like the Fed’s QE1 program in early 2009, had big market impacts when announced, but announcements of later rounds, like that of the QE2 program in November 2010, involved much smaller market moves. This difference led some to conjecture that QE is only effective when deployed during periods of market turbulence, as in early 2009, but not at other times—casting into doubt the usefulness of QE for normal policymaking. However, as my paper documents, most of the research literature supports an alternative explanation, that later rounds of QE were largely anticipated by market participants; as a result, the expected effects of the programs were already incorporated into market prices by the time of the formal announcements. Studies that control for market expectations of the level and mix of planned asset purchases find that later rounds of QE remained powerful, with effects that did not diminish during periods of market calm or as the central bank’s balance sheet grew.

Other research, based on models of the term structure of interest rates, finds that the effects of asset purchases on yields were long-lasting and economically significant. For example, a particularly careful study found that the cumulative effect of the Fed’s asset purchase on 10-year Treasury yields exceeded 120 basis points at the time that net purchases ended. Studies of asset purchase programs in the United Kingdom and the euro area find financial-market effects that are quantitatively similar to those in the United States. The signaling channel of QE was also important, and it remained powerful—as shown in a negative way by the taper tantrum of 2013, when hints from me, as Fed chair, that asset purchases might slow led market participants to expect much earlier increases in the short-term policy rate.

Forward guidance, though not particularly effective in the immediate post-crisis period, became increasingly powerful over time as it became more precise and aggressive. Changes in the policy framework could make forward guidance even more effective in the future.

Forward guidance is central bank communication about its economic outlook and policy plans. Forward guidance helps the public understand how policymakers will respond to changes in the economic outlook and allows policymakers to commit to “lower-for-longer” rate policies. Such policies, by convincing market participants that policymakers will delay rate increases even as the economy strengthens, can help to ease financial conditions and provide economic stimulus today.

The Fed’s forward guidance in 2009-10 was qualitative in nature and did not succeed in convincing market participants that rates would stay lower for longer.  However, beginning in 2011, more explicit guidance that tied rate policy first to specific dates, then to the behavior of unemployment and inflation, persuaded markets that rates would stay low. Guidance became more explicit, sophisticated, and aggressive in other major central banks as well. For example, the Bank of England also tied rate guidance to economic conditions, and the European Central Bank has used guidance to help the public understand the relationships among its diverse set of policy tools.

Forward guidance could be made even more effective if it were incorporated into the central bank’s formal policy framework.

Forward guidance could be made even more effective if it were incorporated into the central bank’s formal policy framework. For example, the Fed’s current consideration of “makeup policies,” which promise to compensate for periods in which inflation is below target with periods of overshoot, amounts to putting forward guidance in place in advance of the next encounter with zero rates. This advance preparation would make the guidance clearer, more predictable, and more credible when it is needed.

Some major foreign central banks have made effective use of other new monetary policy tools, such as purchases of private securities, negative interest rates, funding for lending programs, and yield curve control.

Each of these tools has costs and benefits but has proved useful in some circumstances. The Fed has not used new tools other than QE and forward guidance, but, within the bounds of its legal authorities, it should not rule out other options. For example, yield curve control—at a shorter horizon than used by the Bank of Japan, say two years—could be used to augment the Fed’s forward guidance, as recently suggested by Fed Governor Lael Brainard. Funding for lending programs might be of value in situations in which constraints on bank lending and credit availability are impeding monetary policy transmission. The Fed should also consider maintaining constructive ambiguity about the future use of negative short-term rates, both because situations could arise in which negative short-term rates would provide useful policy space; and because entirely ruling out negative short rates, by creating an effective floor for long-term rates as well, could limit the Fed’s future ability to reduce longer-term rates by QE or other means.

For the most part, the costs and risks of the new policy tools have proved modest. The possible exception is risks to financial stability, which require vigilance.

Uncertainty about the costs and risks of the new policy tools made policymakers cautious about their use, at least initially. For the most part, these costs and risks—including the possibilities of impaired market functioning, high inflation, difficulty of exit, increased income inequality, and capital losses on the central bank’s portfolio—have proved modest. For example, worries about high inflation were based on a crude monetarism, which did not adequately appreciate that the velocity of base money would fall in the face of low interest rates. If anything, of course, inflation has recently been too low rather than too high. The international literature on the distributional effects of monetary policy finds that, when all channels of policy influence are included, monetary easing has small and possibly even progressive distributional effects. The risk of capital losses on the Fed’s portfolio was never high, but in the event, over the past decade the Fed has remitted more than $800 billion in profits to the Treasury, triple the pre-crisis rate.

There is more uncertainty about the linkages between easy money and low rates, on one hand, and risks to financial stability, on the other. Monetary easing does work in part by increasing the propensity of investors and lenders to take risks—the so-called risk-taking channel. In periods of recession or financial stress, encouraging investors and lenders to take reasonable risks is an appropriate goal of policy. Problems arise when, because of less-than-perfectly rational behavior or distorted institutional incentives, risk-taking goes too far. Vigilance and appropriate policies, including macroprudential and regulatory policies, are essential.

A related but still important question is whether the new monetary tools pose greater stability risks than traditional policies or, for that matter, than the generally low rate environment expected to persist even when monetary policy is at a neutral setting.  There is not much evidence that they do. For example, QE flattens the yield curve, which reduces the incentive for risky maturity transformation; removes duration risk, which increases the private sector’s net risk-bearing capacity; and increases the supply of safe, liquid assets.

The amount of policy space the new monetary tools can provide depends importantly on the level of the nominal neutral interest rate.  If the nominal neutral rate is in the range of 2-3 percent, consistent with most estimates for the United States, then model simulations suggest that QE and forward guidance together can add about 3 percentage points of policy space, largely compensating for the effects of the lower bound on rates. For this range of the neutral rate, using the new policy tools is preferable to raising the inflation target as a means of increasing policy space.

The neutral interest rate is the interest rate consistent with full employment and inflation at target in the long run.  On average, at the neutral interest rate monetary policy is neither expansionary nor contractionary. Most current estimates of the nominal neutral rate for the United States are in the range of 2-3 percent.  For example, the median projection of FOMC participants for the long-run federal funds rate is 2.5 percent. Models based on macroeconomic and financial data currently give estimates of the U.S. neutral rate between 2.5 and 3.0.

My paper reports results of simulations of FRB/US, the Federal Reserve Board’s principal macro-econometric model, which aim to compare the long-run performances of alternative monetary policies. When the nominal neutral rate is low, traditional policies (which rely on management of the short-term interest rate and don’t use the new tools) perform poorly in the simulations, consistent with previous studies. The problem with traditional policies is that they run out of room when the short-term rate hits zero.

I compare the traditional policies to policies supplemented by a combination of QE and forward guidance. (In my simulations, the forward guidance consists of an inflation threshold policy, under which the Fed promises not to raise the short-term rate from zero until inflation reaches 2 percent.) When the nominal neutral rate is in the range of 2-3 percent, then the simulations suggest that this combination of new policy tools can provide the equivalent of 3 percentage points of additional policy space; that is, with the help of QE and forward guidance, policy performs about as well as traditional policies would when the nominal neutral rate is 5-6 percent. In the simulations, the 3 percentage point increase in policy space largely offsets the effects of the zero lower bound on short-term rates.

Another way to gain policy space is to increase the Fed’s inflation target, which would eventually raise the nominal neutral interest rate as well. However, to match the 3 percentage points of policy space achievable with QE and forward guidance would require an increase in the inflation target of at least 3 percentage points, from 2 percent to 5 percent. That approach would involve both important transition costs (including the uncertainty and volatility associated with un-anchoring inflation expectations and re-anchoring them at the higher level) as well as the costs of a permanently higher level of inflation. At least when the neutral interest rate is in the 2-3 percent range or higher, active use of the new monetary tools seems preferable to raising the inflation target.

There is, however, an important caveat: If the nominal neutral interest rate is much less than 2 percent, then the new tools don’t add enough policy space to compensate for the effects of the lower bound. In that case, other measures to increase policy space, including raising the inflation target, might be necessary.

My relatively upbeat conclusions about the new monetary tools depend importantly on the neutral interest rate being in the range of 2-3 percent or above. In the simulations, when the nominal neutral rate is much below 2 percent, all monetary strategies become significantly less effective. In this case, although QE and forward guidance still provide valuable policy space, the new tools can no longer compensate fully for the effects of the lower bound. Moreover, in that case, any monetary policy approach, with or without the new tools, is likely to involve extended periods of short-term rates at the lower bound, as well as longer-term yields that are often zero or negative—a situation that may pose risks to financial stability or other costs.

However, as I’ve noted, estimates of the neutral interest rate are in the 2-3 percent range, implying that policy today is mildly accommodative.

Currently, actual short-term and long-term rates in the United States are below 2 percent. However, as I’ve noted, estimates of the neutral interest rate are in the 2-3 percent range, implying that policy today is mildly accommodative.  My simulation results depend on the neutral rate, not the current level of rates. There is, nevertheless, considerable uncertainty about the current and future levels of the nominal neutral rate. If it does ultimately prove to be lower than 2 percent or so, then there would be a case for a modest increase in the inflation target and perhaps a more central role for fiscal policy in responding to economic slowdowns as well. For now, a cautious approach could include making plans to increase the countercyclicality of fiscal policy, for example, by increasing the use of automatic stabilizers.

A bottom-line lesson for all central banks:  Keeping inflation and inflation expectations close to target is critically important.

My simulations apply only to the United States, and the quantitative conclusions can’t be directly extended to other countries. Two conclusions do apply elsewhere, though: that (1) the new monetary tools, including QE and forward guidance, should become permanent parts of the monetary policy toolbox; and (2) monetary policy in general is less effective, the lower the neutral interest rate. In Europe and Japan, where monetary policy is straining to achieve its objectives, much of the problem arises from inflation expectations that have fallen too low, which in turn has depressed nominal neutral interest rates and limited the available space for monetary policy. In those jurisdictions, fiscal as well as monetary policy may be needed to get inflation expectations up. If that can be done, then monetary policy, augmented by the new policy tools, should regain much of its potency.

In the last decades of the twentieth century, the principal challenges for monetary policymakers were high inflation and unstable inflation expectations. Fed chairs Paul Volcker and Alan Greenspan won that war, bringing inflation down to low levels and anchoring inflation expectations. Benign inflation in turn promoted economic growth and stability, in part by giving policymakers more scope to respond to fluctuations in employment and output without worrying about stoking high inflation. We have come almost full circle: In a world in which low nominal neutral rates threaten the capacity of central banks to respond to recessions, low inflation can be dangerous. Consistent with their declared “symmetric” inflation targets, the Federal Reserve and other central banks should defend against inflation that is too low as least as vigorously as they resist inflation that is modestly too high.  Although the new monetary tools have proved their worth and can be made more effective in the future, keeping inflation and inflation expectations close to target is critically important for preserving or increasing available policy space.