Stephen Williamson on New Monetarism and Neo-Fisherism

Stephen Williamson is a professor of economics at the University of Western Ontario and formerly served as a vice president at the Federal Reserve Bank of St. Louis. Steve joins Macro Musings to discuss his work on “New Monetarism,” a research agenda emerging out of the monetarist tradition associated with Milton Friedman. David and Steve also discuss “Neo-Fisherism,” a counterintuitive view that higher interest rates may actually lead to higher inflation.

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: Steve, welcome to the show.

Stephen Williamson: Thank you David. I'm glad to be here.

Beckworth: I'm glad to have you on. We've interacted many times before in the blogosphere and on Twitter, so it's great to talk to you in person. Let's begin as we do with all the guests on this show. How did you get into macroeconomics?

Williamson: Well, so initially I was a math student. I just liked math, but then the career possibilities didn't look so great so I was looking around for something to do and some of my friends suggested I take economics. That seemed pretty interesting and I guess the focus on macro, it kind of came later. So I went to work for the Bank of Canada. I had a master's degree. I got a master's degree in Canada, and then I went to work for the Bank of Canada. Well, what happened there is that, I kind of had a job where part of it was current analysis so we had to pay attention to data. So I got into having to worry about measurement. Then weekly the head of the department would come in and he'd give us latest update on what was going on with monetary policy. That got pretty interesting. So I got interested in the policy making and what the models had to do with the policy and etc. That was sort of really interesting.

Beckworth: It was the catalyst.

Williamson: Yeah, exactly.

Beckworth: Well, that's interesting. I didn't realize you worked at the Bank of Canada as well. So you've seen central banking on both sides of the border. You've seen Canadian central banking and US central banking.

Williamson: Oh, yeah. Sure.

Beckworth: Well, that's interesting. We'll come back to that a little bit later and we'll talk about where the Fed is taking its balance sheet because Canada has a great example, great comparison for the U.S. But before we get to that, I want to talk about some of the areas where you've made some contributions, some original work of your own, and some stuff that's controversial, interesting. I'll let you know that you've actually been requested by our listeners. So you're here by popular demand, Steve.

Beckworth: Now, I've had a number of people email me and say they want to hear you talk about Neo-Fisherism. We'll get to that in a minute. Before we do that though, I want to talk about your New Monetarism literature, I don't want to call it a school of thought, but you've pushed a New Monetarism research agenda. So tell us what it is and what is it trying to do?

The Story of New Monetarism

Williamson: The name came from ... Randy Wright was sitting next to me at dinner one night. We were in Cleveland for a conference there. Then he had this idea that we needed a name, that somehow this would help to get the research program more attention. I think it was his suggestion. We went through sort of what else was taken then we thought, well, New Monetarism that would work. Because we thought maybe we had something in common with the old quantity theorists. We were different from them in some ways, but maybe we had more of an interest in the longer run rather than the short run. We weren't entirely on board with New Keynesian economics. Not so much because of the sticky prices, but more of the lack of attention to financial detail and the explicitness say about central banks and how they work.

Williamson: So what's New Monetarism? The idea was that we thought it was important that the ideas came from back. And I guess the root of it would be kind of Minnesota monetary economics. Both of us, Randy and I and other people indirectly, say like Ricardo Lagos, [inaudible] they work in this area. Lots of other people. The idea was that, you gain a lot from being explicit about why money is useful, why other assets are useful in exchange. To try to explain those things you had to dig deeper into the fundamental frictions in the economy that create a need for asset exchange and financial intermediation and use of collateral et cetera.

Williamson: And if you are explicit about that, you've got a lot more insight into what's going on instead of ... I guess through this literature, this goes back to say the time when even like Lucas's work. So Lucas way back in 1972 wrote this paper called Expectations and the Neutrality of Money. And in that he kind of did that. Part of what he wants to do is try to explain [inaudible] correlations and the broad monetary policy in that context. What he did is, he made use of this little model. So there was this model that Samuelson wrote down in the 1950s an overlapping generations model that you could actually use as a model of money. Basic frictions that people would use this stuff and the people in the model would pick it up and use it and the model are telling them they had to use it basically.

Williamson: Then other people kind of picked up on that. Lucas kind of abandoned the thing, but then the Minnesota people, so primarily, Neil Wallace picked up on this and then a lot of his students worked on it. Then he kind of had this rationale for why you wanted to do this. Why you wanted to do monetary economics explicitly? Then there were other modeling frameworks that came along, certainly like Randy Wright's work, [inaudible] et cetera. What you think of as kind of like a search approach to monetary economics. Part of the friction was, people searching around for people to trade with and money made the trade easier. And those models evolved into what's used a lot now as Lagos Wright framework.

Williamson: So it's not like the principles that we kind of think of as important in this literature tied to this model particularly. It's not like the model itself is the thing. That's a vehicle for doing the work. And then a lot of people work in this area.

Beckworth: So the Lagos Wright model 2005 was kind of the seminal model where they took money seriously. They'd be in a kind of a search framework. Is that right?

Williamson: Yeah. Though eventually it's not specifically search. So what will you learn? I guess the most important thing we learned in kind of post, and I guess so post 1980 monetary economics was that, it's like the information frictions that are important. So there's this idea of [inaudible] kind of tack the name onto this. So the idea is that money is memory and the idea is that record keeping is important and safer. If you thought of the alternative to monetary exchange as being credit, part of what makes credit work is information credit histories. When the credit histories aren't all there, there's a role for assets in exchange and what we might call what we might call money.

Beckworth: So when I think of New Monetarism, what I hear you saying is, you and the other folks in this literature were early on taking seriously the explicit modeling of financial intermediation, banks and firms, medium of exchange assets that facilitate transaction. And I've tinkered with it with Josh Hendrickson. I find that really refreshing when you look at the alternatives. In standard DSG models, you have the money in the utility function or cash in and advanced constraint. Those don't really motivate me, they just put them in there, they just use them and they assume them. My understanding of what you guys are doing is that you actually justify a reason for money and then put it to use. Is that right?

Williamson: That's right. Yeah. There's a limit to that. For models to be useful, they have to be simple and so you're trying to be explicit but there is a limit to how much explicitness it can have. At some stage, you have to make some assumptions and then proceed from there. But the idea is that, in a lot of contexts you could go very wrong if you stuff assets into the utility function and then proceed from there.

Beckworth: Now, how are you different than Old Monetarism? What specific ways? You still buy into the quantity theory, right? Or not?

Williamson: Oh, gee. That's not clear. So here's what I think. So it's the-

Beckworth: The Steve Williamson's version.

Williamson: ... Yeah. So when we use this word New Monetarism, some people don't even like it. I was talking to Ricardo Lagos, he doesn't even like using the name. Randy Wright and I we might disagree about stuff.

Beckworth: That's okay.

Williamson: There's not like anything anybody's necessarily entirely agreed on here, but one thing we might agree on would be that, say if you looked at Milton Friedman. So Milton Freeman, he was not that interested in explicit theory. He had a pretty sharp mind. You read some of his work and the way you kind of flushes out these ideas like optimum quantity of money or something, that paper. He kind of verbally lays out a theoretical framework but doesn't write it down explicitly.

Williamson: He's pretty good at kind of deriving all these implications and things that people later fleshed out explicitly. But he wasn't a big fan of delving into explicit theoretical structures. He had an empirical mind. There's this kind of what people would've thought of is that, the monetarists black box. You could somehow measure this stuff we call money, and then that was important for everything. It's whatever it is. We aggregate a bunch of assets and we call that money, and that's driving business cycles and that's the thing that the central banks should focus on. They should target the growth on that thing and that'll do wonderful stuff, et cetera.

Williamson: The difference say what we might call New Monetarism is more of an interest in the theory and also like say the financial intermediation. It certainly wasn't Friedman's thing. If you look at how he thought about the great depression, he wasn't interested in the kind of the intermediation process and what might've gone wrong with it. That kind of turns up like some of Bernanke's work under depression or something. He starts thinking about kind of intermediary frictions and what that happened and might have to do with what was going on.

Beckworth: One way to think about this, Milton Friedman was really concerned about the medium of exchange role and what happened to that were Bernanke brings out the financial intermediation shocks. What I think, I see New Monetarism, you're at both of those. You take both of those ideas seriously. What is the transaction asset, how does it work, but also financial intermediation and how do they compliment each other, how are they important to the economy? Is that right?

The Transaction Asset and Financial Intermediation

Williamson: Yeah, exactly.

Beckworth: So you would be then also broader in your perspective than kind of a standard New Keynesian model. So in those models, monetary policy is solely a function of the gap between the expected path of the short term interest rate and the natural short term interest rate and that expected path of interest rates versus natural interest rate. That spread there is what shapes monetary policy. And I think the old monitor would say, "Well, it's much more than just that." It's a wide span of assets and relative prices. And I sense that you're sympathetic to that point, that monetary policy is much more than just short term interest rates relative to the natural rate.

Williamson: Yeah. Some of the New Keynesian economics kind of gets away from this. Like what Mark Gertler works on. He pays more attention to the intermediation, et cetera.

Beckworth: Right. So credit has become more important since the crisis, but kind of going into 2008.

Williamson: But the baseline New Keynesian models don't have any assets in them. They take it all out. So not only does it not have any medium of exchange in there, there's no money. Also the markets are all complete. The only friction in there is the sticky prices. So it's purely about these sticky prices and sticky wages and the wedges they create. If you've got these distortions that the monetary policy, so-called monetary policy can mitigate somehow but monetary policy in those models is kind of a funny thing. It's like just the central bank could go out and dictate a short term nominal interest rate.

Williamson: They just say what it is and then there's nothing in there about how they accomplish that, which to me is the interesting part. Looking at like say taking the central bank and looking at it for what it is, it's a financial intermediary. So it's doing some kind of asset transformation. It's got a portfolio of assets and it's got from liabilities. The question is, what's that asset transformation about? And what exactly should this institution be doing? How should you design it and then what's monetary policy all about? How does it work? How it works is going to have to do with that asset transformation that it does.

Beckworth: Right. So you and your fellow New Monetarists are very careful, very explicit about the theory behind central banking monetary policy. You want good deep theory. Now, we recently interviewed Bill Barnett. He's the father of the division of monitoring measures and he goes on the other side of this question and says we need theory to have good measurement and that the way money's being measured is not very precise simple sum, it's nonsense. If you're treating, for example, currency and time deposits equivalence in terms of purchasing power and liquidity. So he's gotten really rigorous with the measurements side. And I'm just curious, what are your thoughts on using Divisia monetary aggregates?

Williamson: The extent I'd been exposed to it, for me it's not too helpful than the whole approach I guess. I can see a motivation. So it's like more macro economists and we have to ... Somewhere we have to aggregate. This is still going back to this monetarists idea that somehow we can separate what's money from what's not. The truth is, there's a whole spectrum of assets in terms of liquidity and other characteristics and these assets play very different roles sometimes, sometimes they're roles that are related. I don't know how you can maybe think seriously about trying to aggregate a bunch of stuff, call it money and think that you've got something that's useful for anything.

Beckworth: I think he would reply that he acknowledges your point. Not everything is equally liquid. They're very different assets and his attempt, his work has been to try to somehow use index theory, aggregation theory in a scientific way to kind of combine them in a meaningful measure. Kind of like CPI does for the price of goods and services. But anyway let's move on. Let's move on to Neo-Fisherism where you've kind of made a big splash past few years. And again, as I mentioned earlier-

Williamson: … splash, I don't know.

Beckworth: ... Well, you've definitely left an impression with many people. Again, I've had people email me and say, "When are you getting Stephen Williamson on the show to talk about Neo-Fisherism?" So to all those listeners you've emailed, here we are. It's going to begin right now. So tell us, Steve, what is Neo-Fisherism?

The Basics of Neo-Fisherism

Williamson: Well, what is it? The name, this is like Noah Smith came up with the-

Beckworth: Oh, he did? Okay, interesting. I didn't know that.

Williamson: ... He came up with a name. Where did it come from? Exactly. John Cochrane was talking about some of these things, but for me it was like just trying to understand this recent experience with inflation from the time of the financial crisis. So we had this period where the Fed started to expand the size of the balance sheet. And I started looking at this, I thought, "Well, surely this has to cause inflation." And then I started to think up a story for why that was. Like I'm doing this in blog posts, I'm kind of not writing it down. And then I kind of think I got something, but then what was helpful about that is, people keep asking me questions. So then I keep trying to explain it and then actually, it's kind of not making sense.

Williamson: And they started exploring this in various models and then roughly looking at the data. So it's like the basic idea is that we've always thought ... I think it's been considered kind of just conventional wisdom that central bank increases phenomenal interest rate and that's going to make inflation go down in the short run. But anyway, we kind of always knew about this Fisher Effect that sets in and sets in in the long run. And usually everybody is used to the causality running from inflation to nominal interest rates. So inflation, high inflation causes nominal interest rates to be high. Now, I was also familiar with this. There's literature on liquidity effects.

Williamson: So people started doing this stuff in, I guess, well, or some kind of early models in the 1980s but it's kind of around 1990. Lucas promoted the ideas and Tim Fuerst was one of his students worked on this. [Inaudible] did a bunch of empirical work on it. So there were kind of whole class models that went with this. So people thinking about what quiddity effects of monetary policy. Now, the way they thought of this was in terms of like central banks controlling monetary aggregates. Suppose central bank controls money growth, now what we knew was that in conventional models that a lot of people use run of the mill kind of cash in advanced models. It's kind of an anticipated inflation matters in these models. So money growth goes up, inflation goes up and people are anticipating higher inflation.

Williamson: The increase in money growth is permanent and that just gives you a Fisher Effect, then nominal interest rate goes up. So everybody thought, "Well, gee. That's not what happens." What happens is that central banks increase money growth and nominal interest rate goes down in the short round. That's a liquidity effect. So then people wrote down some models that would give you that. They were kind of like models with distribution effects in them. It's like a model where the first round-

Beckworth: The limited participation models.

Williamson: ... Limited participation, segmented markets type models. Fernando Alvarez, he worked on these things a lot, for instance. So what's true in that model? What happened was, I kind of explored what's actually going on in some of these models. So the way that they'd tell the story was in terms of money growth. Again, suppose money growth increases permanently, well, these models will have the property that in the short run, nominal interest rate goes down. But in the long run, the Fisher Effect kicks in and then nominal interest rate will go up. So some models real interest rate isn't affected in the long run. And then what'll happen is, you kind of get all the long run one for one increase in the nominal interest rate from an increase in money growth.

Williamson: Well, so the question is what happens if you've got a central bank which is pegging nominal interest rates in the very short run? Which is how they operate, right? Since this kind of quantity, what central banks did in the seventies and eighties was to do a quantity theory experiment where they were targeting money growth rates and then eventually that didn't work so well to do what they were trying to accomplish. They start targeting overnight nominal interest rates and that's the way they formulate policy. And then Taylor comes along with this rule and gives people a way to think explicitly about nominal interest rate targeting. And this is what central banks do.

Williamson: So the question is, how is monetary policy going to work in the context of nominal interest rate targeting? So here's an experiment. So experiment might be a nominal interest rate goes up 1% permanently. What happens? Well, it turns out in one of these liquidity effect models [inaudible] things got the darn liquidity effect in there, the response to higher money growth is in the short run, the nominal interest rate goes down. But if you increased the nominal interest rate permanently and in the background is that central bank has got to be manipulating quantities in order to accomplish that.

Williamson: They got to be doing the open market operations in the background that supports the higher nominal interest rate permanently. What happens in those models in response to that is that, inflation doesn't go down, it just goes up. Not one for one initially because you got this liquidity effect in there. But got to have a really big liquidity effect in order for the inflation rate to go down in response to an increase in the nominal interest rate. And eventually Fisher Effect takes over in the long run and you get the one for one increase. And then in this case, it's like, this is the way we kind of have to think about it. When we think more generally, not just in terms of that experiment, but in terms of policy rules and how they work. That if you're in this world where you've got central banks following policy rules, where the instrument they're working on is a short term nominal interest rate, indeed it is.

Williamson: It's the nominal interest rate that's causing inflation. So what do we know about this? It's like there's some other results in the work to do with Taylor rules and how they work. So that's kind of the crux of it. The question is, what's the right policy rule for the central bank? Is it kind of conventional Taylor rule or what is it?

Beckworth: Before we get too far into the policy implications, let's just summarize again the basic idea behind Neo-Fisherism. And that is contrary to maybe conventional wisdom and what you argue Neo-Fisherism is, if you permanently raise the short term rate, eventually inflation will also permanently go up maintaining that Fisher relationship. Is that right?

Williamson: Yeah. And even further than that, it's like the conventional Taylor rule tells you that what the central banks should do is that, if inflation goes up 1%, the nominal interest rate target should go up more than 1%.

Beckworth: Right. The Taylor principle.

Williamson: Taylor principle. Now this Neo-Fisherian idea would be that, not only is the magnitude of that response wrong but the signs wrong too. In fact, the right rule would be one where in response to an increase in inflation, you go the other way. You actually reduce the nominal interest rate.

Beckworth: Now, obviously and you've gotten this feedback, you've gotten it from me and many other people though, that that just seems so counterintuitive to everything we've been taught. That's just so against the grain. And that's why it's managed a big splash. I know you've written about this formerly and John Cochran's written. I know John Cochrane is taking a New Keynesian model to show this actually happens in the model too.

Williamson: It happens in a New Keynesian model with a Phillips curve in it.

Beckworth: Yeah. Which is surprising.

Williamson: So the Phillips curve is there and this is what happens. This is what happens in conventional rational expectations models.

Beckworth: Let me flush this out a little bit. I was talking to David Andolfatto about this and he's trying to wrap his mind around it as well. And you guys used to be colleagues at St Louis Fed. And he came to this conclusion and this made it a little easier for me to understand. I want to hear your take on his perspective on how this can be true. He argued that the story you described can happen, you can permanently raise short term rates and inflation eventually goes up, but only if you've got fiscal policy in the background supporting that action.

Beckworth: So in other words, the knee jerk conventional view would be, "Man, if you raise rates, you're tightening policy." But what David said was, "Well, if you got fiscal policy providing stimulus, whatever is needed to get that inflation up, you can do this." Is that kind of a missing piece of the story? Do you buy that or?

Williamson: No. I don't buy it. So we would argue endlessly-

Beckworth: You and David would ... okay.

Williamson: ... about that. So I think it's not right.

Beckworth: So you don't think it depends on supportive fiscal policy? It's just-

Williamson: Mm-mm (negative). No, no, no. I was thinking about some of these things recently to do. It's kind of, if you're familiar with fiscal theory at the price level, it's kind of that idea. What David is talking about is a bit of like the fiscal theories. I think there are various shades of the fiscal theory. It's a little hard to understand sometimes, but it seems to be. The extreme view is that, it's like fiscal policy is driving everything.

Beckworth: ... So you're saying David is invoking the fiscal theory for himself to make [inaudible].

Williamson: I think that's what he's saying.

Beckworth: But you don't buy that. You're saying no that's…

Williamson: The argument might be that, if you're in a world with central bank conducting open market operations. In the background, like I said, it's like somehow the nominal quantities have to be increasing. You want sustained higher inflation. It's true, in fact the nominal assets have to be growing over time. So I guess David's thinking, that also means that nominal government debt has to be growing over time if you've got a sustained inflation.

Beckworth: That's probably the hardest disconnect. For most people it would be, on one hand you're raising rates, on the other hand you're going to get this permanent increase in inflation. Typically we think of a permanent increase in inflation because government liabilities are growing faster. But you have found a way to reconcile those two things. Is that what I hear you saying?

Williamson: Yeah. One way to look at it might be, so the question you might ask is, people used to say that hyperinflations came from the central bank basically monetizing the government debt. Well, a lot of these QE operations. But that's being monetized. Japan, they monetize the huge portion of the government debt. The Fed monetized a bunch of debt, but it didn't create inflation. How come? If you read some of these accounts of hyperinflations, then people are describing what's going on.

Williamson: It appears that central bank is monetizing the government debt. So sometimes it appears to create inflation, sometimes not. How come? So the answer might be that the instances where that, you get high inflation like in Argentina or hyperinflations like we've seen historically is cases where it's like the government is actually forcing the central bank to make a lot of transfers to it. It's the transfers that kind of matter. And what creates transfers from the central bank to the government is a big interest rate spread. Historically that's been the spread between the nominal yield on currency, which is zero and the nominal yield on the central banks, the assets. What's going to create these large transfers is a high nominal interest rate.

Williamson: It's another way to think of it in terms of just the asset pricing relationships that, if a nominal interest rate goes up and then something has to adjust, what is it? So it's like, ultimately the asset markets need to collaborate. So ultimately part of what has to adjust is the inflation rate. You can think of some of that happening through the interface between asset markets and the markets for goods and services. But ultimately you have to have this adjustment in inflation in response to the response to nominal interest rates which is being paid by the central bank.

Williamson: So it's like a central bank can create ... I'm convinced a central bank can create as much inflation as it wants unless the fiscal arm of the government or the government itself can ultimately control things if it wants to dictate what the central bank does. But the independent central bank, even if it doesn't have a ... Suppose it's like doing open market operations. So it's like creating the money through open market operations requires having a ready supply of nominal government debt to buy to generate the inflation. But it can do it other ways. It can do it through central bank lending.

Beckworth: Let me ask this question. Before I do that, you asked earlier, sometimes it appears several banks can create inflation, hyperinflation, monetizing debt. Other times like QE in Japan hasn't. My answer, I know you know this, I'll just throw this out there and move on though. Is it depends on whether that monitoring injection is expected to be permanent or not if it's temporary. Of course that depends on what fiscal policy is doing as well. But I think that's one plausible explanation. But let me come back to your story, the Neo-Fisherism story. Do you have some good case studies or examples you can point to that support this take?

Neo-Fisherian Case Studies

Williamson: That's an example of Japan. So you want to interpret that in terms of temporary and permanent. Today it looks to me like these, we'll, say the QE experiments, why aren't those permanent? We understand it's kind of well-known that you can have ... If you think in terms of monetary injections, the effects depend how the money Is injected and then temporary and permanent can matter. If you make the promise, you make the commitment that you're going to withdraw the cash later on. Sometimes if done in a particular way that can have no effect at all, but I don't think that's what's going on here.

Williamson: We've got this line history in Japan. If it were true that low nominal interest rates eventually makes inflation go up, surely after 22 years of low nominal interest rates in Japan inflation should have gone up and it never did.

Beckworth: Well, I'll play devil's advocate here. The New Keynesian would say, "Well yeah, but they didn't lower rates low enough. The natural rate have fallen." But look I hear your point. So what you point to as empirical evidence is not necessarily a case where central bank raised rates and you had higher inflation, but kind of the opposite of that. Where they kept it low and you didn't get the inflation that was expected.

Williamson: And we kind of know. This is like kind of what we know from theory. This is a pretty robust kind of result. It's the kind of Taylor rule result. You've got the central bank that follows a Taylor principle. In the model what happens is that, eventually they end up at zero lower bound and they get stuck there.

Beckworth: They get stuck there.

Williamson: They get stuck there and that seems to be a nice description of what central banks have gotten themselves into.

Beckworth: So let me just ask. What is your sense of the acceptance of Neo-Fisherism? I know you and John Cochrane are probably the most vocal, well-known advocates of it. Are other people kind of joining the chorus...

Williamson: Well, yeah. You kind of get two reactions. One reaction is, "Oh, that's obvious." The other reaction is like disbelief. I think a lot of what explains that is just, there's this idea of ... I think there are central bankers, I think who the model they have in the back of their heads is just basic IS-LM Phillips curve with fixed expectations. What happens in that model is, nominal interest rate goes up, real rate goes up, that makes output go down and you move down the Phillips curve, you got less inflation.

Williamson: That's what they're accustomed to. So that's so well ingrained as an idea that's become everybody's intuition about how the world actually works. I don't know why they do it, but people will tend to ignore what they see in the data. They have data that's inconsistent with a Phillips curve idea and that doesn't seem to bother anybody or bothers them a little bit, but then they seem to find a way to convince themselves that everything is okay.

Beckworth: Well, let's move on to another topic in time we have left. I want to move onto your critique of QE or quantitative easing or the large scale asset purchases. The main theory that was used to justify them, there's two, but the main one was the portfolio balance channel. The signaling channel was also invoked as well. But the portfolio balance channel is kind of the main argument for why these purchases actually mattered. But you argued, for the most part, that they weren't that important. Maybe some effect on the margin, but there really weren't that consequential.

Beckworth: So tell us why you're critical of the portfolio channel theory. Maybe let's talk through the assumptions of the portfolio channel and then tell us why they're wrong and why it doesn't work.

Critiquing the Portfolio Channel Theory

Williamson: The portfolio balance, when somebody says that, and if you read Bernanke speeches, you'd have citations in there, you'd have references and he'd talk about like Tobin. So an example would be like Tobin 1969 paper. I think there's like inaugural issue of the TMCB or something. He's got a little like three asset model. It's like a static model or asset demands and asset supplies. You move quantities around in there so you have central banks do open market operations and it matters. Well, you could take one of those models. You could add something you call a long-term debt, the long-term government debt. So you could have demands and supplies for a short government debt, long government debt. So if we think of like QE, there are different kinds of QE, but one kind of QE would just be a central bank swaps reserves for long maturity, government debt.

Williamson: You right down one of these Tobin models, little static model, that's going to matter. You do swaps a like the operation twist. That'll matter. Because you're changing relative asset supplies and by assumption the markets are all segmented, we don't have financial intermediaries that intermediate across maturities in there and we don't have any dynamics. Lots of things we don't have in that kind of world. But the basic assumption is that, there's some kind of market segmentation that you can play off. That imperfect arbitrage and the central bank can somehow exploit that and very relative asset supplies and affect prices.

Beckworth: So the central bank is a special financial intermediary that can bridge the gap. They can do things that other banks can't do. Is that another assumption of this?

Williamson: Yeah. That would, so that's important. So let's start to think about that a bit. Like what's special about the central bank? So typically I think we think that conventional policy that works in part because of the monopoly the central bank has on currency. Kind of old fashioned medium of exchange we have that only the central bank can issue typically in most countries. There's some subtleties in the U.S about this. But basically the Fed has got a monopoly on the supply of currency. So the intermediation there, it's like basically the Fed's purchasing government debt which can't be used in transactions and converting that into some stuff that can be used in transactions.

Williamson: Well, it seems like that should matter. And that's the basis I guess most people think for why a central bank matters. Now, there have been all kinds of debates about what's the basis for granting the central bank a monopoly on the supply of currency. There are people around who think that that was a bad idea, that there should be private currency supply. We've had cases in the past of private banks issuing currency. Sometimes that seems to have worked, so there was some debate about that, but supposedly you just take it for granted. So now and while we're thinking about unconventional policy, well as you know, if that's going to work and work to make us better off, it better be true that the central bank somehow has an advantage in whatever it is that it's doing when it does QE. So what would that be?

Williamson: So QE is typically taking ... just take the simplest one, which is purchasing long treasury. Say you're the Fed, you purchase long maturity treasury securities, and you turn those into reserve. So the Fed is taking long maturity treasuries, converting that stuff into overnight assets. Now, what advantage does it have and doing that? Well, first thing you might want to ask us is, did it actually do anything useful there? Are the reserves actually more useful in some sense than the long maturity treasuries are? So there's some question about that. People will talk about financial plumbing, worry about that. And they worry about the fact that the reserves are only held by a fraction of financial institutions.

Williamson: There's the things that people have talked about to do with what they call balance sheet costs that are associated with reserve holding. You're going to write down models of this and those are actually real resource costs in some sense. So the Fed in doing a QE operation like that is taking useful treasuries out of the financial market. Treasuries are supporting the repo market for instance and even long maturity treasuries can be highly liquid. 10 year treasuries. That's a very useful asset. You're taking those out of the market and giving the market reserves, which maybe isn't that so great. But then you could think of, what's the private sector doing? So there are things I call shadow banks and other financial institutions that could do the same thing.

Williamson: They can hold long maturity treasuries and convert those into overnight repos. So they can do that. They're taking the long treasuries, converting them into overnight assets. Why is it that they can't do such a great job of that and the Fed is so good at it? So it's not clear if you think about it that way that this would actually accomplish anything. In fact it might go the other way. You might actually be doing harm. Not that things like this couldn't work. It could be that if you think about what a QE might accomplish, some of what I talked about was just that maybe reserves are kind of a crappy asset.

Williamson: Well, maybe if the Fed wants to get into the business of debt management, which is kind of what this is. What they seem to think they could do was to shorten the maturity of the outstanding liabilities of the treasury plus the Fed rate and that would do some good. Maybe they could do that. If they wanted to do debt management, then maybe if they really wanted to get into that business, which maybe there should be a public discussion about that because debt management is something that was assigned to the treasury. If the Fed really wanted to get into that business, maybe it needs some other liabilities. Maybe it should have the ability to issue things that look like T-Bills.

Williamson: Some other central banks can do that. Not that they do a lot of it necessarily, but Swiss National Bank can issue central bank bills. There's an argument, this is like a Jeremy Steiner argument that the Fed's reverse repos are actually pretty good assets. There would be some advantage to it. Like expanding the reverse repo program.

Beckworth: He wants to keep a large balance sheet because the reverse repo opens up access to the Fed's balance sheet for these non-bank national firms. But if we're going to do that, I think that's something Congress should probably approve. As you said earlier, legally, fiscal policy has been delegated to the Treasury. I've mentioned just the whole debt management angle, the Fed is in a losing battle against the treasury. The Fed holds about 19% to 20% of outstanding marketable treasury securities. This kind of goes back to your earlier point. It's doing a lot of what other financial intermediaries are doing.

Beckworth: Maybe a little bit bigger than some, but it's still a bit player compared to 80% of the other holders of treasuries. Another point I want to raise with you, and this is the point Michael Woodford raised in his 2012 Jackson Hole paper and I think it's in line with your point that the Fed isn't really that unique. At least in normal circumstances, it's not that unique. Its only unique role is swapping treasuries for currency. When it's swapping reserves for treasury it really isn't that unique. And he goes on and talks about ... he kind of tells a story. I'm kind of summarizing his words here, but he goes, the argument and another way of thinking of the argument for QE is, that the Fed was going to pull duration risk or interest rate risk off of the private sectors balance sheet and it was going to take it. It was going to free up the private sector, they could then go do fun things, balance portfolios to have riskier assets and be a wonderful world.

Beckworth: But he notes that duration risk hasn't been eliminated, it's simply been transferred from the private sector to the Fed's balance sheet. And should the Fed have problems because of that, who's going to [inaudible] out the tax payers? So the taxpayers really haven't gotten rid of the duration risks…

Williamson: No, no, no. It's not like he can get rid of it.

Beckworth: Yeah. They just moved it around. They've masked it, they've put it in somewhere. The aggregate balance sheet that the U.S economy still holds the same amount of risk is what his point is. And I think that speaks to what you're saying too. That the Fed really isn't that unique in normal times. Now, I do think during QE1 you could make that argument that maybe the Fed was doing something that the rest of the financial system wasn't doing because markets were literally freezing up.

Williamson: Yeah, yeah. So in the content, it's like once you're in a crisis, all bets are off. Inflation control goes out the window, that's a very important role for the central bank and then you may do unusual things.

Beckworth: But your critique and Woodford's critique I think applies to QE2, QE3 very acutely in the sense that, by that point we hadn't gotten back to more normal conditions. Financial markets were working, so at that point, these critiques do come in to play. Let me move on real quick because we're running out of time here. The Fed has been on a path now to reduce its balance sheet. It is now shrinking it by letting these securities runoff. When they mature, they don't reinvest the funds. So we know it wants to shrink. It's outlined a specific plan to shrink how much it's going to do.

The Fed’s Balance Sheet: Floor or Corridor?

Beckworth: What it hasn't given us clearly is this destination. Where are we going to take the bow? How small will it get? We all know that no matter how small it gets, it'll be bigger than it was in 2008, because currency demand growth has occurred. But the question is, is it going to stay large enough so that it keeps a floor system or small enough that it could become a true corridor system? And speak to those differences, what has Canada done? Because it briefly had a large balance sheet too, and it's come back down.

Williamson: Yeah. Just for a year. And that was kind of interesting. The institutional setups very different. So it's like the overnight markets are different here to what's called the large value transfer system in Canada just has if I got my numbers right. There's [inaudible] like 15 or 16 participants. Those are the institutions that are actually potentially holding a reserve balances. Of course in the U.S that's entirely different. There are thousands of institutions holding reserve balances, there are a lot of traders on the Fed funds market. That's a lot of traders in the repo market, et cetera. And so it's kind of a different setup.

Williamson: But here's what happened. So it's like over, what is it? 2009 to 2010, I think spring 2009 to spring 2010 they operated a floor system here, but they did it with not much reserves in the system. They targeted, I'm thinking $3 billion overnight in reserve balances. So you take Canada, it'd be about 10th of the U.S. It's Canadian dollars but roughly that's like 30 equivalent or like $30 billion U.S. Canada has no reserve requirements, so they could actually operate a floor system with not much reserves. I think that's true in the U.S too. My guess is, if you've got the balance, you've got it down to like $100 billion. Now, the question is, will you account the reverse repos or not? I don't know.

Williamson: I guess you account the total I guess. But total of reverse repos, flash reserves, I think if you got that down to $100 billion you could still operate this floor system, which is a funny floor system too.

Beckworth: What do you think is better? A floor system or a corridor system?

Williamson: It depends. Some people at the New York Fed made the argument. So they thought that the floor system would be good partly because of what happens in the daytime. This is a bit in the weeds maybe, well, maybe people need to think more about this sort of thing, but it has to do with daylight payments. So what people noticed was that, when reserves were scarce, that's the usual terminology. I don't know if that's the greatest word for what's actually going on when you're operating a corridor. Effectively though, before the financial crisis, the U.S was a corridor system. It's just they're paying zero on reserves overnight and the discount rate was above the Fed funds rate and they're targeting the Fed funds rate in between.

Williamson: So there's no argument like pre-financial crisis. You get these kind of congestion effects in the overnight market, that financial institutions would delay payments till the end of the day for efficiency reasons, from their point of view. For the whole market, the argument was that it wasn't efficient because there'd be congestion in payments toward the end of the day. There's an argument that if you had the large balance sheet, the system flush with reserves that there's not the same concern so that you wouldn't get this kind of congestion. So you'd say something on that margin.

Williamson: I think there may be other ways to correct that problem even if you had the kind of small balance sheet. Operationally, it's not like it makes so much difference. So one argument people made before, way back, so this is like something you could read and say some of the stuff that Marvin Goodfriend wrote way back, he would have said, "The floor system would be great in the U.S, because all you have to do is announce an interest rate on reserves on that thesis that'll take the overnight rate and you're done." So the procedure before the financial crisis was this kind of funny thing. Where the Fed would target the Fed funds rate, but they weren't actually intervening in the Fed funds market, they were actually intervening in the repo market.

Williamson: The question is, why the Fed funds rate and then why wouldn't they target a repo rate if they're intervening in the repo market? So some of these things don't necessarily make sense. And it's sort of historical accident that created all these ways of this particular method of implementing monetary policy in the U.S. So one argument was that, it would just be easy, the implementation would be easy in a floor system. The implementation might be easy in the floor system, maybe you'd have some improvement in these daylight payments among financial institutions with a lot of reserves in the system, et cetera.

Williamson: Now, I don't know, you might be able to shoot down all those arguments. It's kind of a matter of concern of what the outstanding liabilities of the central bank are that are going to determine what we call a floor system and what's not a floor system for instance. It could be that maybe you do want to have these reverse repos outstanding. Maybe Jeremy Stein's argument is correct. That those are useful assets. His argument was that, treasury is not issuing enough T-Bills basically and the Fed can correct that problem by having a big balance sheet and having a lot of reverse repos outstanding.

Beckworth: Well, of course there's a cost to that as well and that's a bigger footprint for the Fed on the financial system. But on that note, we've run out of time and we have to end unfortunately. Our guest today has been Stephen Williamson. Steve, thanks for being on the show.

Williamson: Yeah, thanks a lot David. Glad to have done it.