Andrew Levin on the Costs and Benefits of QE4 and the Future of the Fed’s Balance Sheet

Andrew Levin is a professor of economics at Dartmouth College and a former long-time Fed official. Andy is also a previous guest of Macro Musings and rejoins the podcast to talk about the costs and benefits of the Fed’s QE4 program. David and Andy also discuss the Fed’s recent record on inflation, QE4’s impact on market functioning, the present and future of the Fed’s balance sheet, and more.

Read the full episode transcript:

Note: While transcripts are lightly edited, they are not rigorously proofed for accuracy. If you notice an error, please reach out to [email protected]

David Beckworth: Andy, welcome back to the show.

Andy Levin: Well, thanks. It's great to be with you, David. And as I said earlier, really honored to be back with you again for a second time to talk about some perennial issues and some really significant new ones.

Beckworth: Yeah. Andy, it's been a few years since you've been on the show. Did you ever get your nominal GDP targeting mug?

Levin: I can check. I have a mug collection, so…

Beckworth: Okay. Okay. Well it's important that you have that from that earlier show and we'll provide the link to the show so listeners can go back and take a look at that. It was an interesting discussion about the governance and structure of the Fed, but today we want to look at the Fed's most recent QE program, or large scale asset purchase program. And you and your co-authors have a new paper out. You Bill Nelson, Brian Lu, and you call it QE4. So we'll call that QE4 in this program as well. I think I saw you last at the Hoover Monetary Policy Conference in May. So we were listening to people give presentations including some Fed officials. And I just wanted to get your take, because you had some views then. But what was your take on their response and what you've seen since in terms of, how did the Fed fall behind on inflation? And we'll use that as a segue into your paper later.

Andy’s Take on the Fed’s Inflation Woes

Levin: Okay. Well, I think a useful starting point, which can be another program that you host is that in principle the Federal Reserve was revisiting its monetary policy framework, which included its goals, its strategy, and its tools. And they started that in 2019. They adopted a new framework in August of 2020. And it's remarkable, really extraordinary I would say, that less than two years after that new framework is adopted, it became effectively obsolete. And I think there must be important lessons to be learned.

Levin: How could you get together a whole institution of the Board of Governors and the regional Federal Reserve banks… and it was called Fed Listens, so in principle it was gathering input from outside the Fed, and end up with a framework that could become extinct that quickly? And I think that lesson learned would probably draw out some of the kinds of issues that you and I have talked about previously, David. The Federal Reserve tends to be too insular. They tend to be too focused on a mobile outlook. They don't think hard enough about what they're missing, what could go wrong, alternative scenarios, risk management. And many of those things were really unfortunately missing from the new framework in August, 2020. And so part of what I would say, where things have been very difficult over the last couple years, really traces back to those missing elements in the foundation of the framework.

The Federal Reserve tends to be too insular. They tend to be too focused on a mobile outlook. They don't think hard enough about what they're missing, what could go wrong, alternative scenarios, risk management. And many of those things were really unfortunately missing from the new framework in August, 2020. And so part of what I would say, where things have been very difficult over the last couple years, really traces back to those missing elements in the foundation of the framework.

Beckworth: And you think that was what caused them to fall behind the curve when it comes to inflation in 2021?

Levin: So put simply, people talk about fighting the last war, and so the whole framework renewal was oriented toward the challenge of trying to bring inflation up to your target and the challenge of facing a long, sluggish recovery and the challenge of how to measure and assess maximum employment. That was the whole focus of the framework renewal. There wasn't really much focus on ... especially because by mid 2020 we were already in the pandemic. In fact, we were starting to have a rapid recovery by summer and fall of 2020, very rapid recovery in employment, to try to think about how might this be different and how could the Fed be prepared for a very different situation that might have involved very different balance sheet actions. We'll talk about that in a minute.

Levin: But in terms of risk management, the Federal Reserve put all of its chips on the outcome in which inflation was transitory. And that's completely inconsistent with the Fed's previous approach. Going back really to the days of Paul Volcker and Alan Greenspan thinking hard about risk management, if we thought that there was a 10% or 20% or 30% chance that the inflation pickup wasn't going to be transitory, the Federal Reserve should have been building that into its policy plans and its strategy. So I think it's a mistake to say, "Oh, this has all just been over the past year," or that, "Oh, this was just because the Federal Reserve was too slow in starting to taper." The truth is the risk management would've already started to happen in late to 2020 and early 2021 when those initial signs of a real surge of inflation were starting.

Beckworth: Okay. Now a big part of that outcome from that framework was the Fed's approach to QE. It kept it going for a long time. And many of us began to wonder in 2021, if it's been going on too long. And you have a paper that really delves into that question, that issue. And that's what we want to get into now. And the title of your paper is, *Quantifying the Cost and Benefits of Quantitative Easing.* And as I mentioned before, your co-authors are Bill Nelson, Brian Lu. Bill Nelson's been on the show a few times, so listeners will know him. But tell us about the motivation for the paper. Why did you guys write it?

Levin: Yeah, by the way, Bill is a great colleague. He and I worked together at the Federal Reserve in years past. Brian Lu is a recent graduate of Dartmouth College who worked on this project when he was a senior at Dartmouth, and really made a significant contribution to the project. We think it's important to do cost benefit analysis as a general matter, it's a kind of basic idea and economics, I would say. Marginal cost equals marginal benefit. And sometimes you can't differentiate, it's not always a smooth function, but the general principle of trying to think hard about cost and benefits.

Levin: And in some way, I think one of the disappointments looking through the FOMC minutes over the last couple years and other Federal Reserve communication speeches, statements on monetary policy, and reports to Congress, is there just wasn't any noticeable serious discussion about cost and benefits of the program, which might have led to kind of, well, how big should it be? How long should it go? What are the trade offs that are being faced here? And now of course in retrospect it's a glaring huge problem. We’ll talk about that in a moment, but what I'm saying is ex ante. Even before you start a major program and as you're embarking on it, to start to think very hard about what are the costs and benefits and how can we make sure that the program adapts over time to reflect that?

Beckworth: Yeah, one of the impressions I got from your paper is that the previous QE rounds, FOMC officials had bigger sense of humility or they acknowledged their uncertainty, that they were going forward with it, but they weren't as clear as to what it would achieve. Whereas the most recent round, what you sensed was that they were completely focused, it was going to work and just pedal to the metal type attitude. Is that a fair reading?

Levin: You could ask Federal Reserve officials, But I think, again, my reading is simply from what's publicly available, which is FOMC statements and minutes and speeches and testimony to Congress and reports to Congress and so forth. I think it was intentional, probably, that Federal Reserve officials wanted to avoid talking about costs this time around because they were worried about the politics of it and they prefer not to have much active oversight from Congress. I think that's a pretty strong preference of many Federal Reserve officials going back quite a ways. And so if the communications emphasized the potential benefits and downplayed the costs, that's a way to make sure that your boss doesn't ask too many questions or start looking into it.

Beckworth: And to be fair to the Fed, this was a sudden shock that hit that quickly threw the economy into the zero lower bound. So they quickly resorted to this tried and true tool, QE. But your point is there should have been more thought ex ante about the limits of it, what it can accomplish, and maybe even a roadmap as to where they hope to go with that. So let's look at this. In your paper, it's interesting because you provide a very detailed look at this type of securities. You also do a 10-year projection, you also do a counterfactual analysis. So you check off all the important, I think, points you need to cover in a paper like this. Before we get into all that though, let's maybe step back and, Andy, walk us through the overview of the program. Just for those who don't know, I'm sure most listeners do know, but for those who don't, just remind us what happened under QE4, how big did the balance sheet get? What happened to the components of it, and things like that?

An Overview of QE4 and the Fed’s Balance Sheet

Levin: Okay, so one important thing to say right up front is that when we talk about quantitative easing, QE, we're talking about a monetary stimulus that's intended to promote economic recovery and help push inflation back upward toward the target. That's what we mean by QE. That's distinct from emergency actions that are taken in the midst of a financial crisis or to address severe financial strains. Now, what happened in March of 2020 is that when it became clear that the US and the global economy was going to be hit very hard by the pandemic and that there were lockdowns, shutdowns, people staying home from work, firms having to put workers on temporary layoff or even just closing their doors, created very significant strains in key financial markets, including the Treasury market and the mortgage markets. And the Federal Reserve took very strong actions during those few weeks in March and early April, 2020.

One important thing to say right up front is that when we talk about quantitative easing, QE, we're talking about a monetary stimulus that's intended to promote economic recovery and help push inflation back upward toward the target. That's what we mean by QE. That's distinct from emergency actions that are taken in the midst of a financial crisis or to address severe financial strains.

Levin: Darrell Duffie has written about this, and others, that those actions were important in stabilizing markets. And there may be lessons learned about market reforms that may be needed so that the markets can function without necessarily having such strong intervention by the central bank. I just want to say those are distinct issues because our paper is focused on quantitative easing as a monetary stimulus tool. So in the latest version of the paper, we actually make this more clear than it might have been in the draft that you were looking at. All the tabulations essentially start as of mid-April, and the reason we picked that date was because during those first four weeks, Darrell Duffie and others have concluded that there was a large amount of purchases, and more than a $1 trillion in Federal Reserve purchases of US Treasury securities during those four weeks that was judged to be necessary and probably quite effective in stabilizing markets.

Levin: But those strains really subsided in late March and early April. So I think the real question that Bill and Brian and I are asking in this paper is okay, after that, after you get past those first four weeks of emergency intervention, you kind of imagined a counterfactual where the Fed kind of wound that down. And as the market frictions were subsiding, the scale of the Fed's purchases shrank close to zero and the Fed started unwinding those actions maybe over a period of months so that by the end of 2020 its balance sheet looked pretty much the way it had in 2019. That would be a perfectly reasonable counterfactual to consider.

Levin: What the Fed did instead was they basically continued the same pace of purchases of Treasury securities and mortgage backed securities all the way through the end of the year in 2020. And then essentially all the way through 2021, they finally started tapering at the end of 2021. Then of course, quickly wrapped up the program in early 2022, but for essentially for eight more months in 2020 and close to 12 months in 2021, purchasing at almost the same pace as they had during the emergency phase of those first four weeks. And that's the part where I think our paper's really raising questions here about cost benefit analysis.

Beckworth: If you had to pick a time period where they should have stopped, so in real time, it’s 2020, you're beyond the emergency period. So maybe you're in April, you're going into summer, into fall and maybe there's a lot of uncertainties still. I mean, at what point if you were in charge, would you have said, "Okay, enough is enough. Time to wind down QE." Would it have been late 2020, early 2021?

Levin: Okay. I think that's where we come back to this point about risk management. So I was privileged to coauthor a paper with Michael Bordo and Mickey Levy, we presented it in May and June of 2020, that had three scenarios. One was, we'll call it a severe adverse scenario where this really did turn out to be something like the Great Recession a decade ago. And there was a benchmark scenario that was more hopeful. And then there was a benign scenario. The benign scenario was one where the development of vaccines happens very quickly. Vaccines start to be distributed by the end of 2020, the lockdowns are over, the economy's recovering quite briskly. And that was a scenario where inflation might actually pick up above the Fed's target due to the huge fiscal stimulus that, in effect, aggregate demand might exceed aggregate supply. And that was essentially built into that benign scenario.

Levin: So the reason I emphasize this is because sometimes officials want to say, "Oh, well it's all 20/20 hindsight." And I would say no, that actually what's critical for a public agency like the Federal Reserve is ex ante try to identify the risks and identify material risks and do scenario analysis and contingency planning. Okay, So then to answer your question, by mid 2020, and certainly if I remember correctly, by around September of 2020, it was becoming clear that these vaccines were in fact going to be ready by the end of the year. And of course, plus or minus a month or two, okay, but there was already very strong evidence that these vaccines were going to be effective and that they would facilitate a rapid recovery. And by fall of 2020, we were starting to get very rapid, just extraordinary, unprecedented improvements in the employment situation, the labor market.

Levin: So we could quibble about whether the QE could have continued in May, June, July, but probably by some time in the summer of 2020 and certainly by fall 2020, risk management would've led the Federal Reserve to start deliberating about tapering the QE. And that taper probably would've been appropriate to happen in fall of 2020. Winter, we can quibble again about the precise timing here, but I think the risk management would've probably led to tapering a whole year earlier and potentially to lift off from the zero bound by winter or spring of 2021 because that was the point at which the Fed was being dismissive of the pickup in inflation as transitory. Whereas a risk management approach, I think it would've been at that point to say, okay, it might be transitory, but it might not. And so we probably shouldn't be having the pedal at the metal in a situation where there's a possibility that we could be facing a persistent pickup in inflation.

We could quibble about whether the QE could have continued in May, June, July, but probably by some time in the summer of 2020 and certainly by fall 2020, risk management would've led the Federal Reserve to start deliberating about tapering the QE. And that taper probably would've been appropriate to happen in fall of 2020.

Beckworth: So we've been speaking on really the first key finding in your paper, and that's program design where the thought and the processes that went into it weren't as clearly laid out, ex ante risk analysis as you just mentioned. And there weren't any, as you said, substantive discussions of cost benefit analysis at any stage of the program.

Levin: By summer and fall of 2020 we were in a housing boom, because mortgage rates were extraordinarily low and frankly there was a lot of people with extra money who were thinking about moving to outside of cities and lots of other things going on that contributed to a housing boom. I think that should have led the Federal Reserve to seriously reconsider whether it was appropriate to be continuing to buy a large amount of agency mortgage backed securities. That part of the QE1 was done in 2008 and in the early 2009 at a point when the housing market was imploding. Okay, and maybe there was some risk of that in spring of 2020, but by summer 2020 and fall 2020, even if the Federal Reserve had a different judgment about the broader macro economy and the risks, the idea that they would be adding fuel to the fire in the housing market is really hard to understand. And again, it wasn't addressed clearly at the time. It hasn't really ever been addressed since to explain why did they continue to buy mortgage backed securities in the middle of a housing boom.

Beckworth: So that's program design. And before we move on to the next key point you make, let me just summarize some of the numbers here on the Fed’s balance sheet. So the assets grew from about $4.1 trillion before the pandemic to about $8.9 trillion, almost $9 trillion in size. Treasuries grew from about $2.3 trillion to $5.7 trillion. So we added a little over $3 trillion. Mortgage backed securities, $1.3 trillion to $2.7 trillion. That's on the asset side. But you also note there's some big changes in the liabilities side, that the Fed's liabilities grew to about two thirds of the liabilities being interest bearing liabilities. So reserves and overnight reverse repo, which is kind of a big deal, we'll get to later because it has the Fed doing interest rate management, public debt management that really belongs in the Department of Treasury. But we'll come back to this a bit later.

Beckworth: Let's move on to another point you make, your second one. That is, what are the consequences for market functioning? And let me read a brief excerpt here from this point. You and your co-authors note, “QE4 markedly increased the Federal Reserve's footprint in the markets for treasuries and agency MBS with potentially adverse consequences for the functioning of these markets over time.” Explain.

QE4’s Impact on Market Functioning

Levin: Okay. Well, so a useful starting point here is that actually in the late '90s and early 2000s, there were very careful consultations between the Federal Reserve and the US Treasury Department. The Treasury Department has a very strong motivation to try to make sure that Treasury markets are liquid. And the reason for that is because if they can reduce the liquidity premium, then they can reduce the financing cost to taxpayers. So that led to consultations between the Federal Reserve and the Treasury about how could the Federal Reserve contribute to facilitating the liquidity of the Treasury markets. And the answer was, the Federal Reserve should limit its holdings of individual Treasury securities so that it effectively had a small footprint in the market for any given security. If you think about the analogy to the stock market, someone who holds more than 10% or 12% of a given company, okay, it's called a controlling interest and they're subject to special regulations by the SEC because it's understood that once you hold more than about 10% or 15% of a company, that you have a big footprint, you have an influence on the board and so forth. And the same kind of rationale, I think, was true in the thinking about the Treasury market. Okay, so that was the judgment that was made 20 years ago.

Levin: During QE3, they had to go beyond those thresholds in order to expand the Federal Reserve's balance sheet in a context where inflation was lower than the target and the economy was recovering in a sluggish way and policy makers recognized that there were some downsides to this, but their overall judgment was that QE was appropriate. Now we fast forward to QE4. What's happened, because the Federal Reserve made a huge amount of additional purchases here in a relatively short period of time, is the Fed now has a very large footprint in both the Treasury market and the agency mortgage backed security market. And just to be concrete, over the period of QE4, on net, the Federal Reserve effectively bought all of the issuance of mortgage backed securities during those two years.

Levin: So the Fed is no longer a price taker in that market. It's no longer a kind of marginal player that's occasionally coming in from the sidelines. The Federal Reserve now owns 40% of all the outstanding agency backed mortgage backed securities. Okay? It's got a huge footprint in that market. And of course then the questions going forward [are], well, what kind of precedent does that set? What kind of moral hazard does it create? How does that affect the market functioning when the Fed pulls back? And a bunch of questions that ideally would've been thought about in the spring and summer of 2020 that now those things are out there.

The Federal Reserve now owns 40% of all the outstanding agency backed mortgage backed securities...It's got a huge footprint in that market. And of course then the questions going forward [are], well, what kind of precedent does that set? What kind of moral hazard does it create? How does that affect the market functioning when the Fed pulls back? And a bunch of questions that ideally would've been thought about in the spring and summer of 2020 that now those things are out there.

Beckworth: So this has a bearing on QT, as it unwinds, and how does it do it without creating ripples and minimizing volatility, but that's a big number, 40% of outstanding agency MBS held by the Fed. So maybe the Fed’s holding some small piece of my mortgage here at home.

Levin: The good thing is that the New York Fed is very transparent about almost every aspect of its balance sheet. And I'll say what the exception is, but almost every aspect of it. So we know exactly which CUSIP securities are held by the Federal Reserve's open market account. And in principle, you could look and see which mortgage backed security your mortgage is in, and then we can find out whether that's held by the Federal Reserve or not.

Beckworth: That's an interesting point. I'll have to check that out sometime. So this affects market functioning. And another side of this would be simply the Fed's footprint is so big in the repo market. The Fed is a big, big player in the repo market. I mean, the other side of purchasing all these assets is the liability side. And we see how big of a player it is in the repo market via the reverse repo facility. So yeah, that's something that some people really wrestle with is, how big is too big? Let me go back a bit on this idea of the Fed being neutral in its asset purchases. And we've had guests on the show before. We had a gentleman from the ECB, Ulrich Bindseil, and he had a role similar to the New York Fed's role that actually interacts with the marketplace for the ECB. He brought this point out that there's different views of neutrality.

Beckworth: So like the US view, the UK view is you only buy government securities because from a consolidated government balance sheet perspective, they net out. So anything the Fed purchases, it's a wash when you combine the Fed's balance sheet and the Treasury balance sheet. That's the US, UK view. And he'd say, "Well, some people would disagree that you're still a big player in those markets, even if from a consolidated perspective it's a wash." The other view was the German view where you want to take a view that's the average portfolio, if there's such a thing. You want to buy assets that mimic the average portfolio on the market. So you actually buy some corporate bonds, some treasuries, maybe even some stocks. And that way you're minimizing your footprint because you're just like the typical, if there's such a thing, or average investor out there. What do you think about those two views?

Competing Views of Central Bank Balance Sheet Composition

Levin: I'm an economist here so you would have to start with, what's the goal? And then, what are the constraints? But what I would say is that if you are talking about say, a sovereign wealth fund, then the logic would be to try to have a diverse a portfolio as possible and try to minimize your footprint in any single market, any narrow security or any single company. Again, because if you have a bigger footprint in that market, then your purchasers or sales are going to be affecting the price and that's probably going to be detrimental over time. So for a sovereign wealth fund, absolutely, I think what you described as the German view. I'm not sure all Germans would agree with it, but at least that particular guest would make very good sense for a sovereign wealth fund. Another situation is where the central bank is trying to help facilitate credit, and in the case of the European Central Bank, they took some pretty direct actions to try to facilitate credit markets in Europe. And I think it makes sense to leave that for some other program for multiple reasons.

Beckworth: Fair enough.

Levin: But the stated goal of QE for the Federal Reserve, going back certainly to QE2 and to QE3, and it came up again in QE4, was by buying longer term securities, which means longer term Treasury notes and bonds and these agency mortgage backed securities… the Federal Reserve could put downward pressure on longer term yields and thereby reduce borrowing costs. And that would facilitate a faster economic recovery and help put upward pressure on inflation to bring it back up to the target, if that was the goal. So given that goal, it wouldn't make sense necessary for the Federal Reserve to be buying a little bit of everything because in a sovereign wealth fund, you want some short term stuff and some medium term stuff and some long term stuff. And you want private sector stuff and public sector. In this case, the clear stated goal of QE is to buy the long end to put downward pressure on term premia and reduce borrowing costs.

Levin: Now, a question we can ask, and we do look at this in the paper, is in terms of benefits of QE4, did it work? Okay, did it actually put significant downward pressure on the term premium? Because they bought a lot of medium and longer term Treasury notes and bonds and they bought a lot of agency mortgage backed securities. Did it work in terms of reducing the term premium? And the simple answer is no, it didn't really work.

In terms of benefits of QE4, did it work? ...Did it actually put significant downward pressure on the term premium? Because they bought a lot of medium and longer term Treasury notes and bonds and they bought a lot of agency mortgage backed securities. Did it work in terms of reducing the term premium? And the simple answer is no, it didn't really work.

Levin: And part of the reason it didn't work is because Treasury yields were already very low in 2020 and early 2021, generally below 1%. And so you're trying to compress the Treasury yields even closer to zero. And so even if QE works as desired in a situation like in 2012 where the Treasury yields were up at two and a half and they were putting downward pressure from two and a half down to two… in 2020, the Federal Reserve might say, "Well gee, we're not sure it's going to work the way it's supposed to in this situation because the Treasury yield's already quite close to zero. We may not be able to push it down much more." And so our conclusion of that part of our paper is [that] the benefits are minuscule. They're very hard to see. And we've used two different measures of the term premium, by the way. One from the Federal Reserve Board, the other one published by the Federal Reserve Bank of New York. And both of those measures, the term premium don't show any significant decline after the launch of QE4.

Beckworth: This raises so many interesting angles and directions I could go, Andy. Such as, what is the theory of QE? Which version do you buy into? But I'll refrain from that because we've had a number of guests on this show in the past to debate and discuss that. But I do want to ask another question related to this, and that is when the Fed did begin to talk up rate hikes and tightening policy, we did see a significant increase in long term Treasury yields. Now, was that more from the forward guidance or rate hike talk, less so from QT then? Is that your sense?

Levin: Yes, I think that's actually completely fair. And there's deeper things here that we should talk about at some point, and I know we're going to run out of time today. But as of a year ago, and I'm going to say a year ago is September, October of last year, 2021, Chicago Mercantile Exchange, which has trading of fed funds futures and options contracts… They have a really nice website where they show the implied probabilities. Markets expected that the Federal Reserve was only going to hike once or twice this year, one or two notches.

Levin: And the surveys of professional forecasters were consistent with that. And the Fed's own Dot Plot, which remember the Dot Plot is only the Fed's assessment of the most likely, okay, it's not a risk assessment. But the professional survey… the surveys of professional forecasters do have some risk assessments. And the market price, of course, these are options contracts, all made it very hardwired that the Federal Reserve was going to be staying essentially at the zero bound or close to the zero bound through this year and maybe slowly starting to lift off next year and the year after.

Levin: And what happened late last year was the Federal Reserve started saying, "You know what? Inflation's not transitory. We're going to have to move a lot faster." And boom, those options contracts, not just fed funds futures, but the treasuries and all the rest started moving upward very quickly. So I don't think that had much at all to do with the tapering or the QT. As you said, it really had to do with ... and we can see it in the forward contracts, the forward contracts ... of course, now here we are. We're only, what, 11 months past that, 9 months into the year, and at its next meeting, the Federal Reserve will probably raise the federal funds rate target to around 3% plus or minus a quarter point. That’s going to be close to 3%. That was simply not anticipated a year ago.

Beckworth: Yeah, I agree with that completely. And just quickly, I'll disclose, I'm a fan of the idea that the Fed's large scale asset purchases aren't that meaningful outside of crisis moments. So a kind of Wallace neutrality. Although I also appreciate a rebuttal, which I think has some merit and that is the plumbing of the system does make it have some consequence, but we can resolve that another time. Let's move on to another key finding from your paper, and that is balance sheet normalization. We've kind of touched on this already, but what do you guys see happening with the balance sheet over time? Is it going to get smaller? How small? Do we want it to get smaller?

The Evolution of the Fed’s Balance Sheet Moving Forward

Levin: Okay. And actually an interesting question here, I'll just tie it back to what you mentioned about reverse repos, is that during QE3, the Federal Reserve indicated that it might use repo transactions temporarily to try to help keep the federal funds rate within the target range. But that was intended to be temporary. And sure enough, the Fed got out of the repo market pretty much once that was over.  This time around the Federal Reserve, I just checked, as of last Thursday, the Federal Reserve releases a weekly report. It's called the H.4.1, Factors Affecting Reserve Balances. So this is as of August 31^st, currency in circulation was $2.3 trillion. The reserve balances was $3.1 trillion, and the reverse repo agreements was $2.5 trillion. So in round numbers here, what we're saying is the reverse repos now are essentially the same order of magnitude as the reserve balances held at the Fed, and larger than the total amount of currency in circulation.

Levin: Okay, so now a question for the Federal Reserve is, well, would that be unwound? Would they create a situation in which the reverse repos shrink back towards zero once the Federal Reserve lifted off, once it was successful in keeping the federal funds rate within the target range? And the answer is no, that in fact the Federal Reserve has judged that it needs to continue to be involved with the repo market on both sides. So it's created standing facilities, one on the low end and one on the high end to try to help create some protections here, a floor and a ceiling so that if there are strains in Treasury markets, the repo market, which is in fact a critical market, doesn't go haywire. So our baseline assumption in our paper is that the reverse repos are going to continue to be large. They may shrink roughly proportional to the shrinking of the bank reserve balances held at the Fed, but they're not going to go away.

Levin: Now under that view, one of our conclusions is completely consistent with what Federal Reserve officials have said, which is that the Fed's balance sheet would shrink over the next couple years, but by late 2024, it would probably need to start growing again because at that point, the reserve balances will be close to where they were at the end of 2019 and presumably start having to grow from that point onward in line with the growth of the overall economy. And likewise, the reverse repos, which would stay proportional to the bank reserves, would also start growing again, so that means the normalization process is going to probably be two to three years long.

Levin: What we emphasize in the paper is that the size of the Federal Reserve's balance sheet will be a new normal in some sense because that'll include all these new liabilities of reverse repos. But the composition of its assets is going to be very far, very, very far from where it was in 2007 and even pretty far from where it was three years ago. And the reason for that is because with much higher interest rates, there's much lower prepayments on mortgages, and with lower prepayments on mortgages, the principal payments on the mortgage backed securities the Fed holds are going to be that much, much smaller.

Reverse repos are going to continue to be large. They may shrink roughly proportional to the shrinking of the bank reserve balances held at the Fed, but they're not going to go away. Now under that view...the Fed's balance sheet would shrink over the next couple years, but by late 2024, it would probably need to start growing again because at that point, the reserve balances will be close to where they were at the end of 2019 and presumably start having to grow from that point onward in line with the growth of the overall economy.

Levin: That means those mortgage backed securities are going to be on the Fed's books for 5, 10, 15, 20 years. They're going to be paid off much, much more gradually than maybe the Fed was anticipating two years ago. And of course the Federal Reserve holds a lot of longer term Treasury notes and bonds, maturities of 5, 10, 20 years or more. And those are also going to be on the Fed's books for a long time. So even in 2025 when the Federal Reserve says, "Okay, we're done with QT. We're going to be allowing our balance sheet to start growing again," the balance sheet's going to still be very heavily weighted toward the long end.

Beckworth: And that leads us to the next finding. And that is interest rate risk. If you're heavily weighted to the long end, duration is an issue. So talk us through that. What should we be looking for when it comes to interest rate risk?

The Potential for Interest Rate Risk

Levin: Since you did it, I'm going to do it myself, which is to quote a sentence from the introduction of our paper. In effect, the Federal Reserve’s balance sheet now looks similar to that of a hedge fund whose long term assets are financed by short term liabilities, except that such funds routinely hedge their interest rate risk. That's why they're called hedge funds. Whereas the Federal Reserves portfolio is effectively naked. The Federal Reserve has not bought any options over derivatives to hedge its risk. So it's naked risk. And ex ante in 2020, the Federal Reserve may have thought, oh, those risks are really minor. But what we're seeing now, and we're not done with the story yet, is as of June, it looks like the mark-to-market losses on the Fed's balance sheet we're already $800 billion. And we can go through those numbers in a moment here, but the possibility is that number will increase, Federal Reserve officials are now talking about the federal funds rate may need to go above 4% and maybe higher.

Levin: I think that it's very plausible now that the total cost of QE4 is going to approach $1 trillion. And when I talk about cost here, we're not talking about kind of paper losses. We're saying that the remittances that the Federal Reserve pays to the US Treasury, so this is real money that really matters for the debt and for tax obligations and government spending, okay, real money. Those remittances over the next 10 years are going to be about $1 trillion lower than they would be if the Fed hadn't done QE4. I should say, there's a benchmark scenario where those costs are more like $500 billion dollars, but the $1 trillion as I'm saying it is now becoming quite conceivable because interest rates are still heading upward.

Beckworth: Some people would say that's just a loss on the books. But Bill Nelson did a good job in some of his little pieces that he writes explaining why that's not the case. And then you just mentioned as well, because those book losses are mark-to-market losses actually are being driven by interest rate changes. And so the way I think about it, it's easy for me, is that the Fed, if we go to 4%, let's say, the Fed has to pay, on the liabilities side, 4% to those overnight liabilities. It has reserves, overnight reverse repo facilities. And so it's paying more on the liability side, literally paying more, but it's holding these securities with fixed interest rates and those aren't going up. So there is a growing space between what they're earning and what they're paying out. And that's being reflected on the asset values as well.

I think that it's very plausible now that the total cost of QE4 is going to approach $1 trillion.

Levin: In fact, I generally would avoid talking about the mark-to-market losses because that would only be relevant in a case where the Federal Reserve actually sold some of its securities, but those losses are never going to be realized. What our paper emphasizes is the net interest income of the Federal Reserve over the coming decade. And so, exactly as you just said, okay, that net interest income is reasonably predictable because we know what these securities are, we know what the coupons they pay… the Treasury, the coupons are paid semi annually and it's completely fixed. But the mortgage backed securities, there's principal and interest payments every month. So it's a little bit more complex, but you can still make reasonable projections.

Levin: And as you said, the securities the Federal Reserve bought in 2020 and 2021, were at a time when interest rates were at extraordinary historic lows, Treasury yields below 1%. That means that many of the securities the Federal Reserve holds now are paying an interest income of 1% or less. And on the counterpart of the liability side of the Federal Reserve balance sheet, those purchases were financed by reverse repos and bank reserves that are, as of next two weeks from now, are going to be ... the Federal Reserve's going to be paying 3% on those liabilities.

Levin: And if Loretta Mester and others are right, within another few months, the Federal Reserve's going to be paying 4% on those liabilities. And that differential, which is roughly three percentage points, you can multiply by $4 trillion in purchases. And you start to see, wow, this was a very expensive program, not just kind of paper losses. This is real money. In principle, if Congress had approved it and said ... now we're getting into the policy implications, by the way. If Congress had said, "Okay, we understand what you're doing, We understand the risks and it's okay with us, just go ahead with this program," then the Federal Reserve comes to after Congress later and says, "Well, we warned you about these risks. The risks are materializing." Congress says, "Okay, well that's too bad, but you know, you win some, you lose some."

Levin: By the way, New Zealand, that's essentially what's happened. The Parliament had agreed the Treasury has underwritten the QE program that was done in New Zealand, and the Treasury is paying for the realized losses in costs. When the Federal Reserve goes to Congress and says, "This is going to cost the US taxpayers $500 billion or $1 trillion," some members of Congress may be quite unhappy that they were not warned about this ahead of time, that they didn't have an opportunity to weigh in, even though the Congress is supposed to be the Federal Reserve's boss under the Constitution. This was action taken by the Federal Reserve without alerting the boss.

Beckworth: Yeah, and you can see some people getting really worked up about this because the optics look bad. You're paying this high interest payment to the banks we bailed out in 2008? So the political optics could get really interesting here, really quickly. Let me step back though and ask this question. Is this more just an issue of Congress getting upset or are you also concerned that it could affect and impair the ability of the Fed to do monetary policy?

When the Federal Reserve goes to Congress and says, 'This is going to cost the US taxpayers $500 billion or $1 trillion,' some members of Congress may be quite unhappy that they were not warned about this ahead of time, that they didn't have an opportunity to weigh in, even though the Congress is supposed to be the Federal Reserve's boss under the Constitution. This was action taken by the Federal Reserve without alerting the boss.

Impacts on the Fed’s Ability to Conduct Policy

Levin: Well, the Federal Reserve should be able to set short term interest rates to foster price stability. When the academic literature on central bank independence started, that's what it was all about, was making sure that political pressures don't hold the central bank back from taking actions needed to preserve price stability. I think a policy implication of our analysis here is that it's appropriate for the public ... and Congress works for the public, for the Congress ... which under the Constitution is responsible for regulating the money supply, the Congress should understand and be informed about what the Federal Reserve was doing. The appropriate arm of the Congress that was created in the 1970s is the Government Accountability Office, the GAO. And in the 1970s, when the GAO was created, the Federal Reserve was effectively exempted from GAO oversight. Now you have to ask, why? And the simple answer at that time was not anything to do with central bank independence. It was because the Federal Reserve had such a simple balance sheet, it had paper cash [inaudible]-

Beckworth: Oh, that's interesting-

Levin: And those paid no interest.

Beckworth: It was a boring place to go look, huh?

Levin: And it's assets were just Treasury securities. And so in effect, the Federal Reserve officials said to members of Congress and members of Congress agreed, that there's nothing really for the GAO to do here. There's no reason for them to be monitoring us. And so they were exempted. But now we're in this situation where two thirds of the Federal Reserves liabilities are interest bearing, where its assets are quite complex, maturity composition and the mortgage backed securities and where the Federal Reserve is doing emergency actions every few years. I think there's a very strong case now for Congress to say [that] now it's appropriate for the GAO to have oversight to be able to monitor what the Federal Reserve is doing and to write reports and to keep Congress informed about that.

Levin: I think that would be a very significant positive step. I don't see how that would undermine or interfere with the Federal Reserve's independence to be able to set interest rates. And if needed, that legislation could specifically say that GAO shall not comment on the Federal Reserve's setting of the federal funds rate target, and just exempt that. But everything else would be within the realm of what the GAO could be looking at.

Beckworth: Yeah, and that makes sense to me. I mean, the Fed is effectively becoming another party that manages US public debt and they need to be accountable for how they do that and what's being done on their balance sheet. My question was more, is that $1 trillion, that upper bound number, is that large enough to make it tough for the Fed to control interest rates? And you're saying no, it's not that big, it's just more of the democratic legitimacy question versus do they have the ability to control interest rates? In other words, are they getting near fiscal dominance if this number gets too big?

Levin: Here's a way to reframe the question a little bit that I think helps clarify answering your question. The Federal Reserve when it was created was given monopoly to be able to create paper currency, and that's a valuable monopoly. And the reason is because the paper currency doesn't pay any interest. The Federal Reserve can use the proceeds from issuing paper currency to hold Treasury securities that do pay interest. I call it the golden goose that lays the golden eggs because the Federal Reserve over decades would use those golden eggs and a little slice of it would go to pay for the operating costs of the Federal Reserve itself. And the remainder of the golden eggs was sent over to the Treasury Department to contribute to reducing the government debt. That was the deal. And in the 1970s, it was still the deal. The Federal Reserve lays the golden eggs, slice it off a little bit, and gives the rest to the Treasury Department.

I think there's a very strong case now for Congress to say [that] now it's appropriate for the GAO to have oversight to be able to monitor what the Federal Reserve is doing and to write reports and to keep Congress informed about that. I think that would be a very significant positive step. I don't see how that would undermine or interfere with the Federal Reserve's independence to be able to set interest rates.

Levin: That's why the Federal Reserve itself was never on the federal budget. That Congress doesn't appropriate funds was because the Federal Reserve makes money through the issuance of paper currency and just uses a small slice of it for its expenses and everything else goes back to the Treasury. That was the whole premise of the whole thing. Now the Federal Reserve is still laying those golden eggs. We still have more than $2 trillion in paid currency outstanding. The effective interest the Federal Reserve earns against that, call it now 3% multiplied by $2 trillion is about $60 billion a year if you use that calculation. What's happening now is the losses or the net interest expense of QE4 is reducing. So instead of the Federal Reserve paying $60 billion a year in remittances to Treasury, the Federal Reserve is going to pay zero remittances to Treasury and it's going to use that $60 billion of golden egg to pay the net interest expense. And that will happen for probably 5 or 10 years. It could happen for longer than that because in some cases the net interest expense is more than $60 billion.

Levin: Well then the Federal Reserve can effectively borrow against future golden eggs. Bill Nelson calls this the magic asset. Okay, because you can just create it on the balance sheet and say, "We're going to count this one against the golden eggs that are coming in 2032, and we're coming this one against 2034." Okay, so to answer your question, the only way that the Federal Reserve would literally have to get a bail out from Congress is either if these golden eggs started to shrink, if paper currency started to diminish, okay, which could eventually happen.

Levin: Bill seems confident, more confident than I am about when that might happen. But the other question, kind of digital payment sort of stuff, but the other possibility ... you said $1 trillion, I would just emphasize $1 trillion is not the upper limit to the cost of QE4. According to the Taylor rule, if inflation is running at around 5%, which is more or less the Michigan survey expectation of inflation over the coming year, the New York Fed is sort of similar, a little bit higher over the coming year, then under the Taylor rule, the federal funds rate should probably be 6% or 7%. And we'll see over the next three or four months, if in fact inflation starts to come down a lot, which some people think, or if it stays high, or even goes up higher, which is what I'm worried about. Again, risk management.

Levin: It could well turn out to be the case that the Federal Reserve has to raise rates to five or six or seven percent, and not just temporarily, but that it might have to maintain them over a period of years. That's what happened with ... people forget about this, the Volcker Fed, it wasn't just the recession of 1981, '82, real interest rates were high for years after that in order to contain the aggregate demand to make sure that the inflation pressures didn't reemerge, real interest rates stayed very high. So my point is that in a situation where interest rates have to go up to 6%, that's going to be another $500 billion, maybe $800 billion depending on what happens to the prepayments on mortgage backed securities.

Levin: Fortunately, this is still contained. Even if it turned out to be $2 trillion, that's 10% of the US GDP. So no one should worry… This is not going to bring down the US economy or the US financial system, but it's a real cost to a program where there wasn't cost benefit analysis done when the program was launched or during its first year of the program, as far as we know, and it cost the taxpayers $1 trillion or $2 trillion. That should have consequences in terms of the oversight of the Federal Reserve and the accountability of the Federal Reserve.

Beckworth: Okay, fair enough. So we can rest assured that the net present value of future seigniorage flows will be more than enough in most likely scenarios to keep the Fed going. Let me do a little pushback, Andy, on these points you've been making about these losses, and I'm going to refer to a paper by Don Kohn and Bill English, and the argument they make is if you look at QE4 in its totality, look at the benefits. Look at what we avoided. We could have had lower nominal GDP, so depth of GDP would've been higher, maybe even run more deficits, things could have been a lot worse, and therefore the consolidated balance sheet of the US government would've looked a lot worse had we not done QE4. And even if there are these losses, in the grand scheme of things, were still better off. How would you respond to that?

Levin: I have the highest respect for Don Kohn. He hired me into the Monetary Affairs Department at the Federal Reserve. So I'm treading lightly here to say that anything that he writes could possibly be anything less than perfect. But take that as a given. First of all, there was a huge fiscal stimulus this time around. Much, much bigger than the fiscal stimulus during the Great Recession. And it's not clear still how the Federal Reserve adjusted and adapted its monetary stimulus to reflect the fiscal stimulus.

Levin: For example, by December, 2020, it was clear that Biden administration was coming in. By early January, it was clear that Democrats were going to control the Senate. It was clear that there was going to be a large additional fiscal stimulus. That was the point at which it might have been, again, in conjunction with the risk management we talked about before regarding inflation, might have been quite natural for the Federal Reserve to say, "Well, we need to balance this by adjusting the stance of monetary policy," but they didn't do that.  So the point is that we saw a very fast recovery in 2020 and 2021. Most analysts think that that recovery was driven, number one, by the rapid dissemination of vaccines and the extent to which the economy was able to reopen and people were able to get back to work. That was probably by far the single most important factor. But number two, this enormous amount of fiscal stimulus with a multiplier somewhere between one and two probably, contributed to an extraordinary unprecedented speed of recovery.

It could well turn out to be the case that the Federal Reserve has to raise rates to five or six or seven percent, and not just temporarily, but that it might have to maintain them over a period of years...This is not going to bring down the US economy or the US financial system, but it's a real cost to a program where there wasn't cost benefit analysis done when the program was launched...and it cost the taxpayers $1 trillion or $2 trillion. That should have consequences in terms of the oversight of the Federal Reserve and the accountability of the Federal Reserve.

Levin: Is that a situation where you say, "Oh, well, it's really critical that we need to add a QE to get an additional..." imagine it's a car going down a hill and you say, "Well, we're going to press on the gas to get the car going even faster down the hill." No, I think maybe this might be a good time to just let the car go on its own momentum, maybe even start to tap on the brakes a little bit. So I can see how the argument that Don Kohn's made in his Brookings piece is very relevant for thinking back to QE2 and QE3, which is of the time that Don was at the Federal Reserve and shortly thereafter. But I just don't see that case for QE4.

Beckworth: Okay. Well, let's move on to the policy implications of your article. And you've already touched on one of them, and that is the GAO should look and do reports on the Fed's balance sheet. What else would you recommend going forward?

Policy Recommendations for the Future

Levin: Well, I think a sensible approach, which is actually,, essentially the approach that the Bank of England and the Bank of New Zealand and the Bank of Canada took is going forward, if the Federal Reserve's going to do major operations that affect the maturity composition of the consolidated government debt. And that means kind of net of between the Treasury and the federal, the consolidated federal debt held by the public, the Secretary of the Treasury should be involved in that process.  And it could be quite parallel to what Congress initiated under Dodd-Frank with emergency credit facilities. Under Dodd-Frank, if the Federal Reserve is going to launch emergency credit facilities, the Secretary of the Treasury has to sign off on it and notification has to be sent to Congress within a relatively short period. I think it's 30 days or something along those lines. Okay. You could imagine something like that essentially applying to QE types of operations where large amounts of security purchases substantially affect the maturity composition of the consolidated federal debt.

Levin: And so in that sort of circumstance, I think the Secretary of the Treasury would've almost certainly signed off on it during March and April of 2020 because those purchases were an emergency action that was intended to stabilize financial markets. And again, by the way, part of what Don Kohn and Bill English were talking about here was that initial intervention, it was more than $1 trillion in purchases. It wasn't free. The Federal Reserve still holds those securities and those are also low yields, but that part almost certainly was worth ... it was unacceptable to allow the US Treasury market to implode in March and April of 2020. But then if you fast forward to May, June, July, August, September, at some point the Secretary of the Treasury might be hesitating and the reports are going to Congress and the GAO is looking at this and assessing risks and risk management, that at some point there's a general agreement that it's time for the Fed to wind this down as opposed to just continuing sort of inertial continuance of a program for a whole other year after that.

I'm hopeful that the Congress could adopt rules like what they did with Dodd-Frank. The next time around when we get to QE5, the program would have to be approved by the Secretary of the Treasury, and it would have to be alerted to Congress quickly, and GAO could weigh in. I'm hopeful about those things.

Beckworth: Well, in the time we have left, I want to ask you one final question about the Fed's balance sheet, and that is, should it be in the business of managing US public finance? I've had a previous guest on, Peter Stella, and he worked at the IMF for many years, and he went to countries that had a very similar situation to what we have now in the US and one of the things he would encourage him to do, and he is encouraged the Fed to do, is to let the Treasury take on that debt. So move that debt off the Fed's balance sheet back to Treasury. So he has some clever ways of doing this, but shrink the Fed's balance sheet, put the liabilities back to the Treasury where he thinks they should be managed, more public accountability, better democratic legitimacy for the Fed. What do you think about that?

Levin: Well, I guess I try to be a political realist. It's not always easy. I think to me it seems plausible and I'm hopeful that Congress would ask the GAO to start overseeing all aspects of the Federal Reserve's operations, including its securities purchases. I'm hopeful that Congress would actually strengthen the independence of the Inspector General of the Fed. That's a whole other question we can come to another day. I'm hopeful that the Congress could adopt rules like what they did with Dodd-Frank. The next time around when we get to QE5, the program would have to be approved by the Secretary of the Treasury, and it would have to be alerted to Congress quickly, and GAO could weigh in. I'm hopeful about those things.

Levin: The likelihood that there's trillions of dollars of transfer of mortgage backed securities from the Federal Reserve to the Treasury, or even that the Federal Reserve gives a lot of longer term Treasury notes and bonds back to Treasury in exchange for short term bills… That part, it's really hard for me to see that happening in the next few years. So this is, again, where in a small country where a few people understand things, they can make quick decisions and do it. In a big country like the United States with a complex financial system, and where the Treasury markets are really important. In a situation where our debt is a 100% of GDP, these kind of things would be very complex to work out, and in the end it might not make much difference. I think it's much better to just strengthen the oversight of the Federal Reserve to make sure this never happens again.

Beckworth: Well, Andy, it's been a real treat to have you back on the show. And for listeners, the title of his paper is *Quantifying the Costs and Benefits of Quantitative Easing.* It's both a Hoover working paper and NBER working paper. We'll provide a link in the show notes. Andy, thank you again for being a guest.

Levin: Thanks a lot.

Photo by Drew Angerer via Getty Images