"The problem with quantitative easing is that it works in practice, but not in theory."
Ben S. Bernanke
On December 16, the Federal Open Market Committee is poised to hike interest rates, putting an end to the near-zero interest rate policy that began in December 2008. So, it’s natural to step back and ask what this episode has taught us about monetary policy at the near-zero lower bound for nominal interest rates. This is not merely some academic exercise. The euro area and Japan are still constrained by the zero bound. And, in this era of low inflation and low potential growth, policy rates in advanced economies are likely to hit that lower bound again (see, for example, here). How the Fed and other central banks respond when that happens will depend on the lessons drawn from recent experience.
While policies that utilize the central bank’s balance sheet to manage the quantity of money and credit are as old as central banks themselves, there is something different about what is now commonly known as “unconventional monetary policy.” First, a number of the tools on which leading central banks have come to rely since the Great Financial Crisis (and earlier for Japan) were introduced precisely because policymakers could no longer reduce short-term nominal interest rates further. Second, despite the experience gained in recent years, uncertainty about how these policy actions affect financial conditions and the economy remains much greater than it is for short-term interest rate policy.
To understand unconventional policy, we find it useful to divide the tools into three categories: quantitative easing (QE), targeted asset purchases (TAP), and forward guidance. QE involves an expansion of the central bank’s balance sheet that supplies reserves (which are just the deposits of commercial banks at the central bank) beyond the level needed to hit the target policy rate. TAP involves a change in the mix of central bank assets—keeping the balance sheet scale and supply of reserves unchanged—in order to alter the relative prices of different assets. If, as in the United States, the traditional central bank balance sheet is composed primarily of short-term government debt, any shift toward other assets is an example of TAP.
Forward guidance is a form of communication by the central bank about its future policy rate path; the message can range from a simple forecast about the economy to a strong statement (“whatever it takes”) to an announced trajectory. Critically, it alters neither the mix nor scale of the balance sheet when the announcement is made. And, because of its focus on the policy rate, forward guidance is the most conventional of the unconventional trio. It has been a weapon in the central bank arsenal for many years, and its use is independent of the zero bound.
The history of the Fed’s asset holdings over the past decade reveals its outsized balance sheet interventions (see chart). Following the run on Bear Stearns in March 2008, the Fed increased its lending to investment banks and other stressed intermediaries at the same time that it lowered its holdings of short-term Treasuries. That is, they changed the mix of assets, but not the size of their asset holdings—an example of TAP without QE. Another instance of TAP without QE—not shown in the chart—was the Fed’s 2011-2012 maturity extension program in which they sold short- and medium-term Treasuries and used the proceeds to purchase long-term Treasuries in an effort to flatten the yield curve.
Federal Reserve Assets by Type (Billions of U.S. Dollars), 2007-15
The evolution of the Fed’s balance sheet depicts several episodes involving a mix of QE and TAP. After Lehman failed in September 2008, the Fed’s immediate provision of credit to intermediaries and to key markets (TAP) was accompanied by a rapid expansion of its balance sheet (QE). Similarly, QE1 (announced in November 2008 and March 2009), QE2 (November 2010) and QE3 (September 2012) all presaged episodes of an expanding balance sheet and a changing composition. The QE1 and QE3 periods included increased purchases of mortgage-backed securities (MBS), while QE2 focused on buying long-term Treasuries.
It is important that we understand each of these tools. Let’s start with TAP. Aside from maturity extension programs, central banks have come to use TAP primarily to restore the function of impaired markets. And it can be a powerful tool (see Gertler for how central bank intermediation can substitute for private intermediation in a crisis). In the United States, the Fed’s late-2008 and early-2009 purchases of MBS provided liquidity in a market that had seized. The impact was to narrow MBS yield spreads over Treasuries and support the housing market. In the euro area, ECB purchases of (and promises to purchase) the debt of peripheral sovereigns under stress sharply narrowed the yield spreads over German debt from their peak in 2012, and helped prevent renewed contagion this year despite widespread expectations of a Greek exit from the euro.
TAP poses two classic problems. First, by design it favors assets with one set of liquidity, maturity and default risk characteristics over others. The Fed’s purchase of mortgage-backed securities, the purchase of corporate liabilities by the Bank of England and the Bank of Japan, and the ECB’s securities market program that favored one country’s sovereign debt over another are all examples. Because every one of these has important distributional effects, they involve aspects of fiscal policy. As such, they pose a risk to central bank independence and—as a result—to the long-run effectiveness of policymakers in achieving their economic stability objectives (see our recent primer on central bank independence here). Second, TAP involves the purchase of assets that are often less liquid than short-term government debt; consequently, these assets can be difficult to sell when the central bank wishes to tighten monetary policy.
Turning to QE, trial and error has taught us a great deal. For example, some critics viewed the Fed’s second balance sheet expansion (QE2) as risking “currency debasement and inflation.” Yet, even after QE3, with the Fed’s balance sheet at nearly 5 times its mid-2007 size, the U.S. dollar’s value today is at its highest point in a decade and inflation remains below the Fed’s 2% objective. More generally, in a world where the central bank pays interest on reserves—as the Fed has since October 2008—the quantity of central bank money (the sum of reserves and currency in circulation) is a poor indicator of broad money growth and inflation, let alone the value of the currency.
So, we can conclude that policymakers took the lessons of the Great Depression to heart: when the financial system is under stress, don’t worry about inflation. Instead, saturate the banking system with reserves to prevent a collapse of credit to the real economy (see our earlier post). Unfortunately, this may be all that we currently know with any degree of confidence.
Recent studies of central bank balance sheet policies typically focus on policy actions that combine QE and TAP that the Fed has labeled large-scale asset purchases (LSAPs). Researchers examine the impact of LSAPs by measuring the change of government bond yields over a narrow time window around the announcement of the program. These “event studies” typically detect a significant decline of yields. For example, in his summary of 15 LSAP studies (see the table on page 10 here), Federal Reserve Bank of San Francisco President John Williams concludes that the announcement of a $600-billion asset purchase (equivalent to QE1 in scale and asset mix) can be expected to lower the 10-year U.S. Treasury yield by 15 to 25 basis points, roughly equal to the impact of lowering the federal funds rate by 75 to 100 basis points. However, the estimates of the impact on the long-term rate are imprecise, and their range extends from zero all the way to 100 basis points. Moreover, since the event studies focus on a very narrow slice of time—usually several days—they say little about the ultimate impact of the policy on inflation and growth.
Perhaps most important is that—as the initial quote from Ben Bernanke highlights—event studies of LSAPs do not reveal how QE operates. Is it the expansion of the Fed’s balance sheet that somehow affects financial conditions and the real economy? Is it the targeted purchases of long-term Treasuries that reduce their relative supply? Was the primary channel of transmission through the 10-year Treasury risk premium, which turned negative for some time during the Fed’s repeated LSAPs? Or, are QE and TAP simply complements to accommodative forward guidance, strengthening the signal that central bankers intend to keep policy rates low in the future? And what about the fact that, as Greenwood et al document, as the Fed was buying long-term debt, the U.S. Treasury was busy issuing so much that the net effect was an enormous increase in the supply to the private sector?
The upshot is not encouraging. We don’t know to what extent the three unconventional tools, QE, TAP and forward guidance, are complements or substitutes in promoting monetary accommodation at the zero lower bound. The best assessment to date of the impact on the U.S. economy since 2008 looks at their combined impact, rather than distinguishing their effects. Engen, Laubach, and Reifschneider estimate that the maximum effect of all post-2008 Fed unconventional policies combined—including balance sheet actions and forward guidance—was to lower the unemployment rate by 1.2 percentage points and raise the inflation rate by 0.5 percentage point from the baseline path.
In addition to this quantitative assessment, one key takeaway from the Engen et al analysis is that the impact of unconventional policies depends greatly on how quickly people and policymakers recognize the depth of the downturn and the changed policy response. Clear, early and compelling policy actions matter because they lower term premia and the path of expected future policy rates when the potential impact is greatest.
Our view is that, in addition to being the closest thing we have to an official Federal Reserve statement on the subject, the Engen et al paper is very clearly at the frontier of what current economic modeling can produce. And, as is often the case, the state of the art means complexity. Here, the implication is that the estimates of the impact of unconventional policy researchers produce—even the very best research—come with a high degree of uncertainty.
Will future studies increase our confidence in both the manner by which unconventional policy influences inflation and growth and the size of the effect? After all, with the benefit of hindsight we now see that policies influencing money and credit can be very important even at the zero bound—a view that unfortunately was not held by the Federal Reserve in the Great Depression. Even 50 years ago, our knowledge of both the transmission mechanism for monetary policy and its quantitative impact on the real economy remained woefully inadequate. And, as recently as 1980, monetary economists and central bankers had not yet agreed that the interest rate was the primary policy tool.
We hope that our understanding of the tools—both their quantitative impact and the mechanism through which they work—will progress more quickly this time around. For now, though, what we can say is that, while we do know more about unconventional policies than we did before the financial crisis, we still don’t know very much. As a result, central banks will remain reluctant to use these tools unless they lack alternatives. Unfortunately, that is the situation still facing the Bank of Japan and the European Central Bank. And, while U.S. prospects have improved, the odds are that the Fed will return to the zero bound in the future—hopefully not soon.