It is a pleasure to be able to comment on this year’s U.S. Monetary Policy Forum paper. As always, what I have to say here today reflects my own views and not necessarily those of the Federal Open Market Committee (FOMC) or the Federal Reserve System.1
To say that I was interested in this year’s topic is an understatement. The use of the Federal Reserve’s balance sheet has been an important element of U.S. monetary policy over the past decade. The start of the reversal of that expansion—the so-called process of balance sheet normalization—has been a key focus of the FOMC’s monetary policy deliberations over the past year. I am pleased that the start of this process—in contrast to the bond market taper tantrum of 2013—has been non-disruptive. The normalization process is now running in the background, essentially on autopilot, with short-term interest rates serving as the primary tool for adjusting the stance of monetary policy.
Let me start with my main point: Although I agree with many of the authors’ findings in this year’s paper, I would take a much less skeptical view of the Federal Reserve’s balance sheet as a tool of monetary policy. While additional study of the effects of large-scale asset purchase (LSAP) programs should be encouraged—as it furthers our understanding of the use of these unconventional monetary policy tools—the paper’s findings do not, in my mind, invalidate the use of LSAPs when the Federal Reserve is operating at or close to the zero lower bound for short-term interest rates. That is the key issue—not the magnitude of the effects of LSAPs or whether short-term interest rates should be the primary tool of monetary policy. On the latter point, which is consistent with the FOMC’s statements about policy normalization, I see broad agreement.
Concluding that LSAPs are less powerful than suggested by some of the estimates from the event study literature does not imply that there is no role for LSAPs at the zero lower bound. In such circumstances, LSAPs can be used to provide additional monetary accommodation by depressing bond term premia and the spread between agency mortgage-backed securities (MBS) and Treasury securities, as well as by strengthening the credibility of forward guidance on the path of short-term interest rates. This can provide support to asset values more generally and make financial conditions more accommodative. Thus, discarding such a tool or ruling out its use seems counterproductive for two reasons. First, to the extent that such a tool can provide accommodation, ruling out its use would raise the risks of insufficient monetary policy accommodation when interest rates are pinned at the zero lower bound. Second, expectations matter. If households and businesses believe that the Fed has run out of tools to provide monetary policy accommodation or is unwilling to use certain tools for that purpose, the risk of inflation expectations becoming unanchored to the downside would increase, which would make it even more difficult for the central bank to achieve its goals. For these reasons, LSAPs should be viewed as a viable tool in our arsenal to be used when the zero lower bound is a relevant policy concern.
In terms of calibration, if LSAPs are not as powerful as some of the event studies imply, the answer is not to simply discard the tool, but instead to look for ways to enhance its efficacy and use it more aggressively. In this respect, I believe that LSAPs work best when they are open-ended. In such circumstances, the use of the tool is likely to be helpful in reducing tail risks and making the bad states of the world less likely rather than signaling to market participants a significant deterioration in the Federal Reserve’s outlook. LSAPs may also be a useful means of making forward guidance more credible. If the monetary authority were to commit to a program of asset purchases until some economic objective is reached and, at the same time, were to make it clear that the policy rate would not begin to rise until the asset purchase program were completed, these steps might help anchor expectations about the likely path of short-term rates more effectively. The Federal Reserve’s third round of LSAPs, which was open-ended, was an important innovation of policy that the paper might have discussed in greater detail.
I also don’t agree with the paper’s conclusions about agency MBS purchases. Such purchases do not represent “credit easing” as argued in the paper. The government’s interventions to support Fannie Mae and Freddie Mac have made it clear that there is no credit risk in the Federal Reserve’s purchase of such obligations. Also, the authors don’t acknowledge that the ability to purchase agency MBS has two benefits. First, it enables the Federal Reserve to purchase longer-duration assets at a faster pace without disrupting market functioning. This facilitates larger LSAP programs, when a larger program is viewed as necessary. Second, it opens up another channel by which asset purchases can provide stimulus—by narrowing the spread between agency MBS and Treasuries, the Federal Reserve can drive down long-term mortgage rates and provide support to the housing sector and to home prices. When monetary policy is constrained at the zero lower bound for interest rates, being able to purchase agency MBS seems like a good tool to have available.
I am also a bit baffled by the conclusion that the Fed should “consider larger and looser caps, possibly specified in terms of quarterly rather than monthly changes, with caps perhaps removed completely by 2019.” It seems to me that the current process of balance sheet normalization is proceeding very smoothly. We placed caps on redemptions because we wanted to limit the potential for market disruption that might occur if the amount of securities that had to be absorbed by private markets fluctuated unduly. The cost of the caps in terms of delaying the shrinkage of the balance sheet is very low. Even if the caps were to be removed completely at the end of 2018, it would most likely pull forward the timing of normalizing the balance sheet size by only two quarters. The delay due to the caps is small because, when they are fully phased in, they are likely to be binding only for Treasuries and then only in the middle month of each quarter.
While I don’t agree with some of the broader conclusions of the paper, I do agree that a healthy skepticism about event studies is warranted. The major theme of the paper is that event studies that have sought to measure the impact of the Federal Reserve’s three major LSAP programs are flawed for a number of reasons. These include potential selection bias of the events chosen and the fact that event studies may not capture all the factors that influence how market prices respond to LSAP announcements.
The first problem with event studies, of course, is the issue of how much new information is revealed by the actual policy announcement. I would argue that LSAPs were always a possibility, so the probability of their use never went from 0% to 100%, even for the first LSAP program. This suggests that event studies could lead to downwardly biased estimates of the impact of LSAPs. In fact, if the announcement indicated that the program was smaller than expected, then bond yields could rise on the program’s announcement. This scenario could occur even though the overall program—including its anticipation—was effective in reducing yields. This makes interpreting the direction of changes in bond yields on the day of an announcement difficult.
The second problem with event studies is that there are always other things going on. When the Fed announces an LSAP program, long-term rates might drop because of the announcement, or because market participants were revising downward their views about the state of the economy and the economic outlook in light of the Fed’s actions. Or, conversely, long-term yields could conceivably even rise as market participants became more confident that the policy actions would be sufficient to push the economy toward the central bank’s goals of maximum sustainable employment and price stability.
The third problem with event studies is the difficulty in disentangling the various factors behind the moves in long-term yields. If an open-ended LSAP program pushes back the perceived timing of short-term rate hikes, it will be difficult to apportion the impact on long-term yields between LSAPs and an expectation of a flatter path for short-term rates.
These conclusions argue for putting less reliance on event studies to measure the impact of LSAPs, rather than more. In this regard, I would advocate taking a different tack: rely more on an evaluation of bond term premia to assess the potential impact of LSAPs. Bond term premia currently are unusually low, and I suspect that LSAP programs here and abroad have been an important factor. Evaluating the factors that have been affecting bond term premia over time might be a good alternative way of estimating the impact of LSAPs.
To sum up, while the impact of LSAPs on long-term bond yields is likely lower than what is implied by some of the event study literature, this does not mean that LSAPs aren’t a useful tool when the FOMC is constrained by the zero lower bound. LSAPs can not only narrow bond term premia, but also make forward guidance more credible. Thus, LSAPs remain useful to have in the toolkit for those times when the short-term interest rate tool may not be available.