Eric Sims is the chair of the economics department at the University of Notre Dame and is a research associate with the National Bureau of Economic Research and the Cleveland Federal Reserve Bank. Eric, along with his colleague, Cynthia Wu, have a number of recent papers addressing monetary policy in low interest rate environments, including a keynote paper presented this past summer at the Chicago Fed Conference that was part of the Fed's big review this year. He joins the show today to talk about this work, focusing on the latest developments in New Keynesian modelling and the current state of macroeconomic research. Specifically, David and Eric discuss the Four Equation New Keynesian Model, the Desirability of NGDP Targeting, and the welfare and cyclical implications of moderate trend
David Beckworth, Eric Sims considers the following as important:
This could be interesting, too:
Tyler Durden writes “Eat The Rich”: Davos Beefs Up Security As Protesters Gather
Tyler Durden writes US Officials Admit Covert Tech Program Is Fueling Iran Protests
Tyler Durden writes Davos Elites Warn “Climate Action Failure” Biggest Global Risk In 2020
Eric Sims is the chair of the economics department at the University of Notre Dame and is a research associate with the National Bureau of Economic Research and the Cleveland Federal Reserve Bank. Eric, along with his colleague, Cynthia Wu, have a number of recent papers addressing monetary policy in low interest rate environments, including a keynote paper presented this past summer at the Chicago Fed Conference that was part of the Fed's big review this year. He joins the show today to talk about this work, focusing on the latest developments in New Keynesian modelling and the current state of macroeconomic research. Specifically, David and Eric discuss the Four Equation New Keynesian Model, the Desirability of NGDP Targeting, and the welfare and cyclical implications of moderate trend 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: Our guest today is Eric Sims. Eric is the chair of the economics department at the University of Notre Dame and is a research associate with the National Bureau of Economic Research and the Cleveland Federal Reserve Bank. Eric, along with his colleague, Cynthia Wu, have a number of recent papers addressing monetary policy in low interest rate environments, including a keynote paper presented this past summer at the Chicago Fed Conference that was part of the Fed's big review this year. Eric joins us today to discuss this work and others.
Beckworth: Eric, welcome to the show.
Eric Sims: Thanks for having me, David.
Beckworth: Great to have you on.
Sims: I'll offer one minor correction, if I may. I am not yet the department chair. I've been named the department chair. I'm the department-chair-in-waiting, so it hasn't quite ramped up, but I'm on the way.
Beckworth: Okay. Okay, and that's why we're getting you now when you still have some time to do interviews with people like me. Fantastic. Now, you've been very active, looking at your work, and you have a lot of papers, and sadly we can't get through all of them in the time for our show today, but you've done a lot of work, a lot of modeling work, a lot of practical policy work as well to bridge the gap between those two, and, as I mentioned earlier, you're a part of a big conference this past summer in Chicago where Fed officials, people in the central banking community got together and evaluated options for the Federal Reserve as part of this big review, and it was great to see you and Cynthia present there.
Beckworth: Before we get into all that though, I want to look at your journey into economics. How did you get interested in economics originally and then, from there, how did you decide to become a macroeconomist?
Sims: That's a great question, and I don't know that I have a great story for this. My father has a Ph.D. in economics, in old-school monetary economics. My dad is Grant Sims. He got his Ph.D. at Texas A&M. His advisor was Akira Takayama, and he has a paper published in the Review of Economics and Statistics titled, “On Measuring the Nearness of Near-Monies,” which I can just see my dad coming up with that title and being pleased with himself. He did not end up being a practitioner of academic research. He ended up going to work in Washington, D.C., for Wendy Gramm. His Ph.D. is from Texas A&M. Wendy and Phil Gramm were at A&M, when Phil Gramm went to Congress, my father went there and ended up working in the oil and gas industry in Houston.
Sims: My dad has a Ph.D. in economics, and when I was applying to college, I didn't really have a particularly good thought as to what I wanted to major in and had had no exposure to economics, but, because my father had been an economics Ph.D., I thought maybe I'll major in economics. The summer before college, I went to Trinity University, which is a liberal arts school in San Antonio, Texas, that actually hosts a Nobel laureate series where Nobel Prize winners in economics come through, and so I had the good fortune to meet some of them as an undergraduate, but the summer before I went to college, I went to Barnes & Noble and picked up like a Economics for Dummies book, one of these little things.
Beckworth: Interesting. Yeah.
Sims: I really had no exposure. My dad wasn't sitting around doing Euler equations with me or what have you, and there was really no pressure there, but I just thought, because my dad had done this, maybe I'll do it, and I was pretty good at math, but I liked humanities style stuff as a student in high school, and, in fact, I thought I was going to major in something like history, and, here, I'm borrowing a quote from my late great colleague, Tim Fuerst, that said something to the effect of, "If you want to study philosophy, study philosophy. If you want to study history, study history. If you want to study mathematics, study math. If you want to do all of these things, study economics," and so I picked up this book, and I was like, "This is pretty neat," because it's... economics is drawing... It's highly relevant. It's technical, and it has a focused way of thinking about reality and forces you to hone your thoughts and be rigorous and defend your assumptions and think about how assumptions lead to results, but it can be used to address a bunch of interesting questions that go beyond what we think about or what the lay public often thinks about as economics per se.
Sims: I found this little book pretty intriguing actually, and so I decided to enroll in principles of micro and signed up as an economics major and was reasonably good at it, and it went from there, I suppose. And then in terms of how I ended up going to graduate school, I again don't have a great answer for this question. I was pretty good at being a student, and I liked being a student, and it's easy to keep being a student, I guess. At that point in my life, it's like you just applied to the next thing, and so it seemed natural to apply to graduate school.
Sims: In some sense, I'll oftentimes tell my undergraduates now, and I work with very many good, really good undergraduates, and we've had the great fortune to send a number of our undergraduates from Notre Dame to very top Ph.D. programs, I went into the process not knowing what I was getting into, and so I oftentimes say that I'm the exception that proves the rule. It worked out very well for me, but I didn't have very serious research experience as an undergraduate. I didn't really understand what that entailed, and so I went into the process somewhat blind and just got lucky that it worked out for me and that it was something that I enjoyed and something that I was good at, but if I had to go back 20 years in time to my college self and advise myself on what to do, I think I would have done things somewhat differently and maybe would have tried to get some more serious research experience before going on to graduate school, because that was really something I didn't know what I was getting into.
Sims: I knew that I liked economics. I knew that I thought it provided a useful means by which to analyze social and policy problems, and I was attracted to that and I had some vague sense that there were employment opportunities for economists, but I didn't go to graduate school with a particular topic that I really wanted to work on, because I hadn't really worked on any topics. I had taken classes, had done a very little bit of what I would call rudimentary research, but this is just research for writing a term paper in a class, not real academic research, so, in a sense, I went to graduate school with an open mind, which is good and bad in the sense that you're going to expose yourself to new ideas and see where that stuff takes you, but I didn't go with a very concrete idea of what I wanted to do or really even why I wanted to do it. I just liked what I was doing and I thought being a college professor would be fun, but I can't say that I had a really concrete idea as to what I wanted to do research in or what I wanted to use that research for other than I thought being a college professor sounded like a cool idea.
Beckworth: Yeah, and now you're the soon-to-be-chair of the economics department at Notre Dame. We've had one of your colleagues on the show, Rudi Bachmann, another-
Sims: I can't get away from Rudi.
Beckworth: Oh, has he followed you along in your career path?
Sims: Yeah, so there's an interesting story on that. I was Rudi's TA when he first got a job as an assistant professor.
Sims: Yeah. Rudi was hired from Yale while I was still a graduate student at Michigan, and I was assigned as the head TA for his intermediate macro class, and so that's where I met Rudi, and so we did a couple of years together while I was in graduate school and then, when I finished, we started working on papers and then we ended up hiring him here some years after that once I had been tenured, and, now, I'm going to be Rudi's boss, so I'm excited about that. He started as my boss, and now I'm going to be his boss, but... so there's a close connection with Rudi, and we go back, gosh, now like 12 years.
Beckworth: Yeah, you guys got a pretty strong macroeconomics program there. You've got Cynthia Wu, another colleague. I mentioned earlier she's done work with you as well, so a lot of fun things are happening there, and I imagine your dad, after you took this career path and showed him you were successful, was happy you chose it.
Sims: Yes, I think very much so, and, they, my parents had told me, "This is why you listen to your kids. Your kids know what they want in life better than you do," so that's all worked out, and I've been very blessed to be at an institution you mentioned. Notre Dame is... We could do an entire podcast on the history of economics at Notre Dame. I realized that's not the point of this, but there's an interesting and long history of economics at Notre Dame that really, when I came, I'm now in my 11th year here, the department was new. We started in 2006. I want to say there were 14 people on the faculty. We're now up to, believe it or not, 39 tenured faculty in the economics department. It's unclear where the cap, end, is. We seem to be expanding faster than the universe without end, but, at some point, that expansion is going to cease and we'll be more of a mature department in the sense that you're always hiring, but you're not necessarily growing in net terms, but it's been a lot of fun to be at a place where a majority of my colleagues are younger.
Sims: A vast majority of our faculty have PhDs within the last 20 years, and we've got a group of people that have been committed to trying to build up something from a department that did not exist in the mid-2000s, a Ph.D. program that did not exist in the mid-2000s, and so it's very interesting and humbling to have been able to work here, and it's humbling going forward to have some leadership role and how evolve going forward into the future.
Beckworth: Let's move into your work, and I want to begin more broadly about macro. We're going to get to some of your papers, and you have a lot of them, but we'll get to some of them, they're very interesting, but I want to maybe step back and talk more broadly about macroeconomics, the state of macro, since you're in the trenches in academia, you're doing scholarly work, you see what's going on, and my question to you is what do you consider to be cutting-edge macro?
Beckworth: I know macro is a huge, huge field, but this is David Beckworth, right? I'm immersed in the policy world of central banking and stuff, so, when I ask this question, I have a narrower view probably than what macro encompasses entirely, but what do you see as where macro is today?
Current State of Macroeconomics
Sims: That's a great question, and I think it's not an easy question to answer because what people would identify or define as macro is going to be very broad, and there's going to be a lot of folks that have different taste mechanisms going forward. I think that, actually, I've come up in the profession in an interesting time. I got my Ph.D. in 2009 and, right at that time, the AEJ macro journal came out, and Olivier Blanchard said, "The state of macro is good," and then the financial crisis happened, so... And then we've been thinking about that for the last 10 years and tried to think about how to be policy relevant and the questions about, “Did macro fail?” Et cetera, et cetera.
Sims: I think that there's, prior to the crisis, and I think there's been some of this after the fact, and you mentioned... We can talk about the HANK models a little bit, which I have not worked with directly, but I have some interest in. I think that the real key and what the best macro research going forward is going to focus on, relatively small scale models that you can tease out intuition and cleanly communicate policy advice to folks. And I think that in some sense we had gone in the opposite direction.
I think that the real key and what the best macro research going forward is going to focus on, relatively small scale models that you can tease out intuition and cleanly communicate policy advice to folks.
Sims: If you think about the intellectual development of macro over the last several decades, there was this outcry against medium scale or large scale macro-econometric models, and then there was this focus on developing microeconomic underpinnings and emphasizing general equilibrium and market clearing coming into the real business cycle literature, and then there were New Keynesian models that incorporated different frictions into that, and these things developed where you get these medium scale, the DSGE models, which I've worked with, which gets lots of bells and whistles, and at some level, it's not that obvious how far we've gotten from the immediate... from the large scale macro-econometric models. We've dressed these up with some micro underpinnings, but it's not obvious that's sort of, say, a Smets-Wouters type of model, what we've gained relative to a large scale macro-econometric model, and I think that there's been a movement, and it's something that I would like to do more of in my own work, to distill things down to things that really highlights the intuition and highlight the channels and highlight the frameworks that you're thinking about for policy advice as opposed to this mangled mess of 35 equations, et cetera.
Sims: I can't say that I've always been great at doing that because, like I said, you mentioned the paper with Cynthia. The paper we presented in Chicago was one of these mangled messes. There's a lot of stuff going on, but we've since written a follow-up paper that I think I'm more excited about in a sense that it's trying to distill things down to the core and trying to emphasize mechanisms and develop some intuition perhaps at the expense of some realism, but trying to get at that, and so that's where I think we need to go, and I think we run the risk that, if we don't distill things down to a level that you can communicate to folks, that you run the risk of being policy irrelevant.
Beckworth: Yeah. I'm guessing you're referring to your four equations model.
Beckworth: We'll come to that in a minute. That is I think a pretty cool innovation, and we'll talk about that in a minute, but I just want to for a second stop and look at these Heterogeneous Agent New Keynesian models. They call them HANKs, and you made an observation that I actually was thinking or wondering about, and that is how practical is a HANK model for someone like Jay Powell, and I think these HANK models are very interesting. I listened to an interview with Greg Kaplan from University of Chicago where he got into great detail talking about the richness, the heterogeneity, the different types of households, there's the poor hand to mouth, there's wealthy hand to mouth, and you end up getting results where monetary policy doesn't always work through intertemporal substitution, and maybe it works indirectly through labor markets and labor income. It's a much richer story and, in some ways... and it was interesting he mentioned some of the stories that are coming out of these HANK models are similar to some of the simple macro stories you tell in Macro 101.
The HANK, TANK, and RANK Models
Sims: There's a sense in which, and this is something that I would... Provided I have more time, I really would like to enmesh myself in this literature a little bit, but there's a sense in which HANK has come full circle back to old school Keynesian models emphasized in MPC channels and things like that, which is probably not a bad first approximation to how someone like Jay Powell thinks about how the world works. I think that the criticism, and I levied this criticism in different contexts as the modern canonical New Keynesian model and is extremely forward-looking in intertemporal substitution is at the heart of the mechanism, and if you take a step back, it's just not obvious to me that that's a good description of reality, so I'm very sympathetic to the work that the HANK modelers are doing, trying to think about marketing completeness and heterogeneity, which is obviously going to be important for a number of different questions that you want to address.
Sims: I think one downside of this literature is that these models are a bit of a black box in the sense that there's a lot going on in the background and there's a lot of details that get swept under the rug that might be really important for thinking about transmission, so, for example, you have these fiscal transfer rules that emerge in a lot of these models, and the one I'm most familiar with is the McKay, Nakamura, Steinsson paper in the AER about forward guidance and HANK, a potential HANK resolution to that, and you have these high productivity folks and these low productivity folks, and you have the government issuing transfers back to these folks, and how exactly those transfers work is really important quantitatively for how successful these models are at resolving different paradoxes, and so I think there's a lot of details there that can come across as a black box.
Sims: I like the TANK approach, the Two Agents New Keynesian model which is in some sense a convex combination, if you will, of a full-on HANK model where you really model heterogeneity in the distribution of assets and et cetera, et cetera, and the RANK model where everybody is a forward-looking, full information, rational expectations, no liquidity constraints, intertemporal optimizing robots in some sense. The TANK models, and I think this is an opinion, for example, someone like Jordi Gali has really tried to push. I think they're a nice compromise in the sense that they're fairly transparent. They're fairly easy to work with. You can reduce these things down to relatively small scale, and you can capture both the intuition of the intertemporal substitution channels that the RANK models emphasize, as well as some of these MPC mechanisms that the older school models emphasized without having these enormous computational details and who gets transfers and et cetera, et cetera, that you run into with some of the HANK literature, so I think it's going to be interesting to see where the literature coalesces.
Beckworth: Just to be clear for our listeners, the HANK model is Heterogeneous Agent New Keynesian models. The RANK model which, Eric, you were talking about, is Representative Agent New Keynesian model, the one household, and then the TANK model, is Two-Agent New Keynesian.
Sims: The TANK models have two types of agent. One type is a hand-to-mouth consumer where MPC channels are really important, intertemporal subscription is shut down, and the other one is the full-information, rational expectations, representative agent consumer, and, that way, you can get some of both worlds in a way that is relatively easy to communicate, and so I think there's some useful insights there.
Beckworth: Okay. This is fascinating. Because one of the implications actually that comes out of the HANK models is that monetary policy may be very limited. Fiscal policy is much more important, or at least more coordination between monetary policy and fiscal policy.
One of the implications actually that comes out of the HANK models is that monetary policy may be very limited.
Beckworth: The other thing though that comes out, you've highlighted is that there's a lot of devil in the details in those HANK models, and so that could be an issue for some, but I like this TANK approach you mentioned, and it reminds me of some other work that's being done by Jim Bullard and people like him who've worked on simple heterogeneous agent models. I don't think they're New Keynesian necessarily, but we have a saver or someone endowed with a lot of wealth, someone who isn't, and at least some interesting stories about risk sharing in the economy.
Sims: I think that that's exactly right, and I think that the future is going to be incorporating some degree of heterogeneity in a way that we can think about in a policy relevant way in an aggregate framework, but in models that are relatively simple and transparent and easy to communicate to folks like Jay Powell, and so I think some of this stuff that Jim... that Bullard has done on his regime switching stuff and trying to think about why we're stuck, seem to be stuck in a low inflation regime, et cetera, is really interesting.
I think that the future is going to be incorporating some degree of heterogeneity in a way that we can think about in a policy relevant way in an aggregate framework, but in models that are relatively simple and transparent and easy to communicate to folks
Beckworth: Okay. Let's move on from models then to the perception of macro, and, recently, your colleague, Rudi Bachmann, got into an exchange on Twitter with a number of people, and, again, I watched from the sidelines, but I was very sympathetic to Rudi's position. There were a number of folks on Twitter, Ph.D. economists, who were questioning whether macro should be in the first year of your graduate work in the PhD program. I was a little taken back. I think Rudy was, too, and a number of individuals. But, I guess the reason they brought this up is they felt like we're in such different worlds. Macro is so far removed from what applied micro folks are doing. They weren't sure what this usefulness was. But, it struck me as very feeble ground to stand on. If you want to understand general equilibrium effects, understand some of the big contributions we make to policy, you'd want to have at least some understanding of macro.
The Place for Macro in the Discipline
Sims: I think that's absolutely right. I mean, I think that watching this somewhat from the outside, applied micro broadly defined ... If we can call that a field. I think of applied micro more as a tool than a field, but let's think about it as a field for a second. I mean, this big identification, credibility revolution, or the identification revolution in applied micro, where folks are really hyper focused on trying to isolate exogenous variations in some treatment and seeing what happens to various different outcome variables is really scientifically appealing. I think the big limitation of that is it's hard to think about how to scale that stuff up, and to scale that stuff up to address questions that really matter in the aggregate. I think you simply have to think about general equilibrium effects, which is where macro comes in and where macro's really important.
The identification revolution in applied micro is really scientifically appealing. I think the big limitation of that is it's hard to think about how to scale that stuff up, and to scale that stuff up to address questions that really matter in the aggregate.
Sims: So, I mean, I could list a laundry list of applied micro papers where there's some quasi random variation isolated in some treatment, and they look at the effects on some outcome variable, and you say, "Well, this is the effect of X on Y," but it's a very, very partial equilibrium construct. The external validity is easy to question. If you want to think about the implementation of aggregate policies, I think you have to think about general equilibrium stuff. I think this is what macro, ultimately, teaches you to think about, right?
Beckworth: Absolutely. Yep.
Sims: ... is trade offs and what's happening there. So, as an example ... This is not to put it down one way or another. One of my colleagues, his job market paper was about measuring the effects of evictions ... measuring the effects of homelessness on economic outcomes, basically. So, people get evicted from their apartments. What happens? How do you isolate exogenous variation in evictions? It was a treatment that there was random assignment to judges in an evictions kind of court. And, some judges ... There's a fixed effect, where some judges are more lenient than others, and that's the quasi random variation in what happened, with your propensity for being evicted if you're late on a rental payment.
Sims: You can look at what happens, and not surprisingly, being evicted from your home is bad. It has a number of bad outcomes. But, scaling that up to a national level, this is ultimately a macro question, right? If we enacted policies that made it more difficult for landlords to evict tenants that were late on their payments, this would make it more difficult for potential renters to get access to housing in the first place. There would be more stringent credit screenings, et cetera, et cetera, before landlords would enter into agreements. At the end of the day, that's the general equilibrium macro thing that we need to think about, right?
Sims: I'm obviously pro having macro being in the first year core. More generally, I think there's this push to remove theory, broadly defined from economics, and just think about isolating these quasi natural experiments and looking at treatment effects, et cetera, et cetera. I think that's great, and it has a place, and we need to use the results from those kind of works as inputs into our modeling, but ultimately, economics is thinking about incentives and thinking about how price changes affect behavior, and macro is aggregating that up, right?
Sims: And, thinking about spillover effects between different markets, et cetera, et cetera. I think that, at the very least, people need to have a training where they think about theory and incentive effects, et cetera, et cetera. And so, I think it's a misunderstanding of what macro is or what macro can promise. I think a lot of folks, on the outside looking in, think that macro ought to be able to eliminate business cycles, or macro ought to be able to forecast crises, et cetera, et cetera. The reality is I don't think that either of those are realistic goals.
Sims: Nevertheless, I think that good macro research and good macro policy advice is incredibly important and incredibly valuable. At the end of the day, macro's hard. It's not hard in the sense that it requires lots of mathematics. Certain kinds of macro do. It's not hard in the sense that it requires years and years to study. It's hard in the sense that it's hard to get really credible answers to really important questions.
Sims: I would love it very much if I could get Jay Powell to randomly play with monetary policy. That would make it really easy to write really good papers. And, we would get really definitive answers as to what happens when the Fed buys more bonds, or when the Fed cuts policy rates, or when we switch from a corridor to a floor system, or what have you. But, we're not going to have that, right? And so, we have to work with these toy models where we can run the experiments ourselves. And, we build up, and we're going to have debates, and not everybody's going to agree. But, these are really important questions when you think about aggregate wealth there and the future of our country's economy. These are really important questions to answer, and ones that rank and file people really care about.
Sims: And so, the fact that it's hard and the fact that the data are messy and the fact that it's difficult to get very clean, precise answers in macro, I don't think is reason to not teach it. It's all the more reason to teach it so that we can push the frontier out further and come up with better and better answers to really important questions.
Beckworth: Well, I completely agree with all that. And, I would add to that, people who say, "Let the data speak for itself," they are implicitly still using theory, right? I mean, how they set up their experiment, how they collect data ... You can't escape theory. Theory guides your research agenda, how you implement your projects. Theory is always there. You just need to be explicit about it and wrestle with these general equilibrium questions that you outlined.
Sims: Yeah, that's exactly right. I mean, I'm not a theorist. Let's be frank. But, I mean, empirics without theory is problematic. As social scientists, I think we have to be testing some kind of framework in the back of our minds, and that framework has to be guiding the empirics that we're doing. There's a constant give and take, right?
But, I mean, empirics without theory is problematic. As social scientists, I think we have to be testing some kind of framework in the back of our minds, and that framework has to be guiding the empirics that we're doing.
Sims: We develop theories. We test them. In the data, we find where they're inadequate. We revise the theories. We find empirical regularities. This causes us to try to come up with different ways to model different phenomena. And so, it's a healthy give and take, but I think it'd be very dangerous to not teach people macro. Particularly when you think about, if you ask the rank and file person on the street, "What's economics about?" They're probably going to tell you macro issues.
Beckworth: That's right.
Sims: Fiscal policy and exchange rates, trade ... These are very aggregate things. And so, just I think it'd be very dangerous to train a generation of economists that don't have any grounding at all in that.
Beckworth: Absolutely. Okay. Let's move on to your research that you've dealt with. We've touched on some of it. I want to begin with your four equation model. So, this is back to modeling for a little bit, but it's an important paper, I think. It gets at what you were talking about earlier, something that's trackable but practical for policymakers. So, tell us why it's called a four equation model as opposed to the three equation model.
The Four Equation New Keynesian Model
Sims: Yeah. To give a little bit of intellectual history on this, maybe I can talk about two papers.
Sims: You mentioned the paper that I wrote with Cynthia for the Chicago Fed Conference. One of the things that I'm interested in, and I think one of the real challenges for us as macro economists, or monetary economists more generally, is we live in a world where short-term interest rates are really low. They've been really low for 10 years. They're probably going to be really low for some time.
Sims: We live in a world where at different points in the last decade, we tried to implement different kinds of policies to circumvent the issues of the zero lower bound or the effect of lower bound, if you prefer. I think it's reasonable to think we're going to have to do stuff like this going forward into the future. And so, what Cynthia and I were trying to think about was to incorporate unconventional monetary policy tools into relatively conventional macro DSGE models that folks in central banks use for forecasting and policy advice. The first paper does so in one of these medium-scale models, where there's a lot of moving parts. There's a lot of frictions here, frictions there, some of which one may consider to be ad hoc, et cetera.
What Cynthia and I were trying to think about was to incorporate unconventional monetary policy tools into relatively conventional macro DSGE models that folks in central banks use for forecasting and policy advice
Sims: The so-called three equation New Keynesian Model is the textbook model that a lot of folks use within the central banking sphere to think about policy. And, here, I'm thinking about Woodford's 2003 very long book, or Gali's textbook that spells this out. And, the three equation rank model ... There's households that are forward-looking, subject to no frictions, supply labor, and buy goods. There's firms that are monopolistically competitive, that are subject to some kind of pricing friction. There's the central bank. And, this model has been really useful for thinking about the desirability of things like inflation-targeting, and potentially thinking about gains from commitment, as opposed to discretion, when thinking about monetary policy.
Sims: But, it's very ill-suited to thinking about stuff that central banks around the world have done over the last 10 or so years, which are quantitative easing or large-scale asset purchases, forward guidance, implementation of negative interest rates in different parts of the world. More generally, when I'm thinking about the innards of the banking system, issues about liquidity and how we target short-term funds rates ... It sweeps all that stuff under the rug.
Sims: And so, our objective function in the four equation model, in a sense, was to build a model that featured non-trivial financial intermediation, non-trivial credit intermediation, and a potential scope for central bank credit policies. And, I think credit policies is actually a term that somebody like Bernanke would prefer to use, as opposed to quantitative easing, although the press has run with QE. And, try to think about those policies in a framework that's as close as possible to this textbook model, and in a framework that, in fact, under a certain parameter restriction, reduces back to it.
Sims: And so, it is a RANK/TANK model. It's not a HANK model, going back to our earlier discussions, in the sense that there's non-trivial credit intermediation. In the standard rank model, everybody's the same, and there's credit markets, but there's no credit flowing from one party to another in some sense. Households are just consuming ... In the standard framework, at least, they're just eating all of their endowment. Y is equal to C in aggregate, right?
Sims: And so, that's problematic, right? I think that's not what's happening in the real world. In the real world, there's credit funneling from savers on to investors, and we think the popular narrative of what happened in the crisis is that there was a breakdown in the intermediation process, and that a lot of the unconventional policy tools that were tried, including the lender of last resort, opening the liquidity facilities in the immediate wake of the crisis, were ways to try to deal with that breakdown.
Sims: We wanted a model where there was actually credit flowing. And so, it's a two-agent model in the sense that you've got one household that wants to borrow, one household that wants to save, and there's a rudimentary financial intermediary or bank that stands in between them. And, that bank is subject to frictions, which gives rise to credit spreads. Those frictions can be time varying. And, you can think about a credit crunch or a credit crisis as an exacerbation of those frictions, and, maybe, potentially, things like quantitative easing or bond purchases or ways in which central banks can deal with the exacerbation of those frictions in a crisis, and think about how substitutable that is with conventional short-term interest rate adjustment policy, which is the framework that the Fed and other central banks have moved to. And so, that's the background there.
Beckworth: Okay. So, you have this model, your four equation model, and this is different than the one that you presented this past summer. But, you mentioned it came out of that. It was related to your thinking on that paper. So, why don't you tell us about your paper you presented this summer to the Fed?
Sims: Yeah. So, this paper was a benchmark, medium-scale, New Keynesian DSGE model, where we modeled financial intermediaries. And, we modeled friction. So, markets ... We have a rudimentary term structure in the model. Firms are required to finance some fraction of their investment by issuing long-term debt. Long-term debt has to be bought by banks who fund themselves with short-term deposits from households.
Sims: This intermediation process can get screwed up. And, there was a bunch of other bells and whistles in there. And, the model was a quantitative evaluation of three different kinds of unconventional tools. And so, we thought about forward guidance, quantitative easing, and negative interest rate policy, and trying to think how substitutable those things are with conducting monetary policy by adjusting short-term interest rates, according to something like a Taylor rule, for example.
Sims: So, it was really ... That paper was in a quantitative framework, trying to think about how substitutable are unconventional policies for conventional policies, and trying to say something quantitatively about how effective do we think some of the Feds unconventional policy actions were in the wake of the crisis, when short-term policy rates went down to zero, and take a conclusion there.
Sims: My read of the literature, which, again, is my read, not necessarily everybody's read, is that the zero lower bound was not the constraint on policy that it might've otherwise been, or that one might think on the basis of a simple three equation model, or even one of these more souped up models that has lots of bells and whistles, and different frictions here and there, where the zero lower bound is really very costly, where when you get to zero, central bank can't do anything. Weird things can happen. Output caps can be really big. They cannot close. You can get stuck in deflationary traps. You can have contractionary beneficial supply shocks, et cetera.
Sims: I think we quibble with the details of what central banks did in the wake of the crisis, but I think that, by and large, none of those ominous forecasts came to pass. And so, I think that the missing ingredient from those models is modeling intermediation more generally and modeling unconventional policy. And so, that's what we were trying to get at. We come away with the takeaway there that we think that quantitative easing policies, in particular, were a reasonably effective substitute for interest rate policy. And, we tied this into some of Cynthia's earlier work on the so-called shadow rate, which is based on stuff from the behavior of the term structure of interest rates.
Sims: We tried to provide some quantitative background for that. And so, in some sense, we didn't set out to write this, but in some sense, I think it's a fairly pro-Fed paper, in the sense that I think they don't want to move away from a framework in which they adjust short-term policy rates as their normal operating procedure. But, they're going to have to reckon with the fact that they're going to run into constraints on that. I think that those constraints are not as tight as one might think, from the basis of a standard textbook model. And so, that paper's trying to quantify that.
Sims: But, it's got the bells and whistles, and as I was saying earlier in our discussion today, I think the way forward is to have transparent, smaller-scale stuff that might rely on ... crazy, if you will, assumptions, but nevertheless, which facilitates clean aggregation and clean results. And so, that's what we set out to do in the second paper, to build in the core frictions, market segmentation, and a rudimentary term structure, and timed varying credit frictions from our larger-scale model, into something that's smaller scale, where you can do some more transparent policy analysis. And so, that's what we were after. That was the want operator, if you will.
Sims: Within the context of that, we say things about what is optimal monetary policy look like in a world like that? How much QE do you need to do if you can't move short-term policy rates to achieve different objectives, et cetera, et cetera. I think, nevertheless ... Maybe you wanted to go here. There's some limitations to that. We presuppose in that work that there's a zero lower bound on short-term interest rates, but that longer term interest rates never get close to that. And so, you've always got the ability to manipulate those.
We presuppose in that work that there's a zero lower bound on short-term interest rates, but that longer term interest rates never get close to that. And so, you've always got the ability to manipulate those.
Sims: I think that's a concern going forward, and I think that's a potential limitation. What happens if we end up in a world where long-term rates go to zero, and these unconventional policies, which work on the long end of the yield curve, or the risk-adjusted end of a risk structure of interest rates ... We're running out of ammunition there, and I think that's something that particular model doesn't really speak to. But, that's something to worry about going forward, I think.
Beckworth: Yeah, that's my concern, too, on a practical level, is that long-term yields across advanced economies are marching down, and because of global capital markets and arbitrage, we're all marching down together. And, I'm not sure that it's over. I mean, there's ebbs and flows, and Trump's trade wars may ease one day, and that may take off some of the immediate pressure. But, I think the long-term structural forces driving the downward push of interest rates, they're still with us ... demographics, globalization…
Beckworth: All those things, I think, are still with us, so I do worry if we have a recession, and I hope we don't, but if we have a recession, say, in two years ... In two years, where will a 10-year treasury yield be?
Sims: Yep. No, that's right. Yeah. So, we obviously don't have much room to maneuver the Fed funds rate down 500 basis points, or whatever we typically do in recessions. But, we also don't have 500 basis points for the 10-year anymore.
Sims: So, I think as long as this research program ... I think that manipulation of the long end of the yield curve can, in principle, work ... can, in principle, work in doses. I think that there's a broader question about whether ... I don't think we would want to do that is the main means by which we implement policy, but I think is a substitute, or is an antidote to an inability to move short-term rates. It's reasonably effective, but I think that the broader question there is what if we run out of room there? And, that's what you're getting at.
Sims: And, at the end of the day ... At some level, that becomes a fiscal issue, but it becomes a fiscal issue that's relevant for monetary policymakers, right?
Sims: It's a fiscal issue in the sense of I don't think monetary policy is behind these lowering trends in interest rates, both at the short and long end of the yield curve, but we have to reckon with that. And, that opens up a whole new set of questions, and it's exciting to think about them. I can't claim to have got great answers for that, but it's exciting to think about.
Beckworth: I'm glad you brought that up because so many people do blame central banks, the low interest rates, and I always try to tell them, "Central banks are following equilibrium rates down, not the other way around."
Sims: I completely agree. I think a worrying concern is that there's increasingly broad ... what I would call mission creep, when you think about central banks. In central banks, thinking about inequality or the environment or things like that, I don't think central banks are very well equipped to deal with that.
Sims: So, in spite of the fact that I've advocated in some of my work for creep, in terms of the instruments that central banks use, I think be very focused on targets, about thinking about price stability and full employment, et cetera, and focusing on just what I think central banks can credibly control, I think is important. And, I think when you start thinking about why are neutral rates falling so much, this leads you into this mission creep thing, where central banks can't address the reasons why neutral rates are falling, but they've got to deal with it. They have to reckon with it.
Beckworth: Right. And then, they get blamed for it, too, even though it's not their fault. Let me ask a related question. So, the Bank of Japan is doing yield curve control, where it targets its 10-year treasury yield. I'm wondering, do you see any usefulness for something like that in the United States?
Sims: I think it's certainly a possibility. And, I mean, I think that some of ... We never called it yield curve control, and we never targeted long-term yields, but at the end of the day, it's the flip side of the same coin, as to what we were doing with large-scale asset purchases, which was buying up quantities of stuff, trying to push the price up and the yield down.
Sims: So, again, this goes back to the question of how much can we move long-term yields down? But, I think that there's certainly some scope for thinking about yield curve control, or targeting longer term interest rates, as opposed to short-term interest rates, particularly in a low short-term rate environment. I think there's certainly some scope for that.
Beckworth: It's interesting. In the case of Japan, the Bank of Japan actually used yield curve control to keep the 10-year yield up. It was falling too low, causing financial intermediation problems, but in practice, you could apply it by pushing it down.
Beckworth: All right. One last question about QE, and this is where I'm going to put on my QE skeptic hat. I asked this question to you at the conference. We were both at a conference in November, at the Cato monetary policy conference. You presented a paper. It's a derivative of what we've been talking about. And, what I want to get at is Michael Woodford and Gauti Eggertsson's irrelevance results.
Beckworth: I know it really goes back to Neil Wallace, but they've made the case in the past, and I know Woodford is modified some of this in subsequent work, but I think the critique still applies more broadly. And, that is just irrelevance result that QE is not going to have a big impact unless you've got really strong forward guidance tied to it. And, I look back at the past decade. I look at the recovery, coming out of the Great Recession. No, we did not experience another Great Depression. It could've been far worse. I completely acknowledge that, but we didn't also have a great, healthy, V-shaped recovery. And, I wonder to what extents the irrelevant result applies to the use of QE. Any thoughts on that?
Benefits and Limits of Quantitative Easing
Sims: Yeah. So, I mean, the extreme Wallace neutrality result, going back further in time to the early 1980s ... I don't think that holds. I think it's like a Ricardian equivalence result, where it's a useful benchmark, and you want to think about deviations from that. The mechanisms through which we emphasize QE, potentially working ... It's, in essence, in the models that I've worked with Cynthia ... It's an anecdote to credit market frictions.
Sims: So I guess, my perspective on this would be that I think the QE as, and I think as I kind of phrased this earlier, I think is a short-term, temporary antidote to monetary policy impotence engineered by the zero lower bound, I think makes sense, and I think can work in small doses. I think that in enormously large doses, I'm skeptical as well. I think as you are.
Sims: So I think something that our papers don't really address is sort of like what's the optimal size of a central bank's balance sheet going forward? We're thinking about deviations of that, and I kind of think that in the situation where private credit markets are impaired, basically the central bank by doing QE can step in and do some of the intermediation that intermediaries are unable to do.
Sims: But I think as a longer run proposition that's not going to work. So I think then we're getting closer into the irrelevance results. So I want to be clear, I don't think that QE is a be all end all, I don't think it should be only line of defense and I don't think it's sort of a good way to think about conducting monetary policy as a baseline. I think that it is an emergency tool, in environments in which credit markets are impaired and frictions are high. I think it can be a relatively effective substitute.
I don't think that QE is a be all end all, I don't think it should be only line of defense and I don't think it's sort of a good way to think about conducting monetary policy as a baseline. I think that it is an emergency tool, in environments in which credit markets are impaired and frictions are high.
Beckworth: No, I think we agree with that. I mean, and when I look back at the empirical record, at least my take on it, like QE1 I think fits that description very well. Markets were impaired and the Fed stepped in and did kind of act as a backstop to the markets. It was the financial intermediary because markets weren't. But QE2, QE3, I think there's more room for debate as to what it accomplished. I mean, I definitely lowered the yields. I don't dispute that, but how much oomph did it give the economy and that's where I begin to wonder.
Sims: No. I completely agree with that. That's a good empirical question and yeah.
Beckworth: Let me phrase my point this way. I think one takeaway from this discussion, Michael Woodford and Oregon's discussion is, if you had a level target which did provide stronger forward cut ins, QE would be much more effective. Is that a fair interpretation?
Sims: I think that is a fair interpretation. By the way, I'm sympathetic and I know you are too to the notion of some kind of level target which builds in some element of forward guidance. So I know you're interested in nominal GDP targeting. That's something else that I've worked on. I think there was a bunch of desirable features of that precisely because it builds in this sort of automatic forward looking anchoring that goes on that helps stabilize people's expectations.
Beckworth: We'll speak on that. Let's move to that paper. So you have a paper with several coauthors is called “On the Desirability of Nominal GDP Targeting,” and walk us through your findings. What is desirable about nominal GDP targeting?
Desirability of NGDP Targeting
Sims: So it turns, so this was a paper and we didn't do this in any kind of strange framework. We basically just did a horse race. We took kind of a standard new Keynesian model and we did a horse race between different kind of policy rules and to our surprise, nobody, the paper in the context of that kind of model saying, "Well what does nominal GDP targeting look like in a quantitative sense?"
Sims: So we just posed nominal GDP targeting against things like inflation targeting, output gap targeting, a Taylor rule, et cetera, et cetera. It turns out nominal GDP targeting works quite well. Now we kind of knew and there's been some of the analytical results in the literature about the desirability of having forward-looking anchors in there, that sort of builds in these, you were describing some of the Woodford results, that builds in some forward guidance.
Sims: So we found, I guess somewhat to my surprise, that nominal GDP targeting does really well. For example, in the baseline model that we work with, is a whole heck of a lot better than a straight inflation target, which is if you will, a straight inflation target is sort of more backward looking, less forward-looking. Bygones will be bygones, has less of the stabilizing features that nominal GDP targeting does.
Beckworth: Yeah, I was thrilled to see this paper when it came out a few years ago. I do want to highlight one interesting part of your findings. You do mention that output gap targeting sometimes can do better than nominal GDP targeting. However, when there is any kind of uncertainty about the output gap, nominal GDP targeting does better. This reminded me of Michael Woodford's kind of pragmatic reasons for why he picked nominal GDP targeting in his Jackson Hole paper. He said, "Look, what I really want is an output gap adjusted price level target." That's a mouthful. But he said, "But in practice, nominal GDP level targeting does as good of job as one could do given our uncertainty of the outfit gap." I thought that kind of echoed here with what you guys have-
Sims: That is exactly right. So if we assume that the central bank can perfectly observe what the efficient level of output is, in principle, it could try to implement that and that would look something different than a nominal GDP target. But in practice, I don't know what the efficient level of output is in the US right now. I don't think you do either. We might have an idea, but it's not observed, and it's difficult to communicate to people. Whereas something like, I think the big advantage of nominal GDP targeting, and it's somewhat difficult to build this into a model, and to be frank, it's not in the model in the JDC paper that you were talking about, it's just an issue of communication, right?
Sims: We're used to nominal quantities all the time in our daily lives. So telling people we're going to target a growth rate of nominal GDP of 4% per year, I think is transparent. I think it's easily understood by folks. I think it's something that you can easily communicate and because of that and you can achieve better outcomes. It kind of gets at some of the ideas or some of the, we would like to do this is Woodford's terminology, "This efficient output level adjusted price level target," or whatever the mouthful was. In practice, it's getting at that. So I think has a lot of desirable features because of that.
Beckworth: Yeah, and I think on a practical level, and listeners will know, I've talked about this extensively on the shows. I won't spend too much time on nominal GDP level targeting, but on a practical level for the Federal Reserve, for the European central bank. Well particularly for the Federal Reserve, given the dual mandate, it really takes the burden of trying to divine what the Phillips curve is doing. If it's broken, if it's not linear.
Beckworth: I mean you can kind of just put that to the side and you just focus on nominal GDP and if it starts to grow above target and you know, "Okay we're getting to a point where we need to hit the brakes." Whereas right now there is so much obsessing over what is the Phillips curve really tell us. Is there a deeper structural parameter we're missing? Is it non linear? And I think life would be a whole lot easier for them if they did that approach.
Sims: I completely agree. It much more transparent.
Beckworth: Okay. Well let's move on to some other work that you've done and I want to look at one that kind of was a flash in the pan idea, and maybe it has more lasting value. Maybe I'm being a little harsh here, but Neo-Fisherism. It was talked about a few years back quite a bit. I've had Steven Williamson on the show. He's really big proponent of it. John Cochran's done some work on it. Interestingly you have a paper that shows this and they show it as well, and it kind of naturally falls out of the standard new Keynesian model. But you show, you provide some fixes that really undermines the case for Neo-Fisherian wisdom. So walk us through that.
Sims: Yeah, so just to reiterate, the idea of Neo-Fisherism is, what happens to inflation when you move interest rates? The conventional wisdom is, "Oh we cut interest rates, we stimulate demand. This pushes inflation up." But of course if you think about the Fisher relationship, if you think about the real interest rate is being fixed, we know there's a one to one positive relationship between nominal rates and inflation, at least over the long run. That actually holds exceptionally well in the data.
Sims: So it becomes sort of a topic of conversation from folks like Williamson or Cochrane in the last several years because the puzzle of our time in some sense is why is inflation so low? Right? We have a coexistence of very low interest rates and low inflation. According to this conventional wisdom, low interest rates are stimulative. So we shouldn't be seeing this low inflation, and going back to something we were talking about earlier, we would like to give ourselves more space to cut interest rates the next time we have a crisis. So what can we do to get that? So there's this extreme, Neo-Fisherian idea that we should actually raise interest rates and that this would result in higher inflation, move us further away from these problems.
Sims: So the idea in that particular paper is to just say, does this come out of these standard textbook models that we work with? (A.) And (B), to the extent to which it does what's driving that? The answer to the first part of the question part A, is yes, a standard new Keynesian model does exhibit Neo-Fisherian behavior.
Sims: And B, the reason why is that that model is very, very forward looking and it relies on intertemporal substitution. So this goes back to something we were talking about much earlier with regards to Hank models and kind of the idea there is pretty simple. If you're going to persistently raise inflation targets, this is going to move inflation expectations in these models, which is going to result in not much movement in the nominal interest rate and hence not much movement in output. So we work through a couple of different fixes in that model where if you just mechanically make it somewhat less forward looking, some of these Neo-Fisherian results go away.
Sims: So this is something that I struggle with in my own work, and I think about is, that I'm attracted to, if you will, the elegance of fully micro founded rational models, et cetera. But I don't think that people in the real world are all that forward looking. So do I think, would I call for Neo-Fisherian policy changes now? No, I would not think that raising interest rates would actually raise inflation. Why is that? I think that expectations are much stickier and so that's kind of what we play with in that particular paper.
So do I think, would I call for Neo-Fisherian policy changes now? No, I would not think that raising interest rates would actually raise inflation.
Beckworth: Well thank you for that work. I think it's important that folks like you are showing when you get things like bounded rationality, and people not being so forward-looking, that the standard intuition does hold.
Beckworth: We have a little bit of time left. We have one more article I'm going to look at before we have to end the show. This, I think, is a very topical paper and it's titled “On the Welfare and Cyclical Implications of Moderate Trend Inflation.” This paper I think is important because there's been a lot of suggestions that one way to deal with the zero lower bound is to have a higher inflation target, raise it to three or 4%, so you would ultimately change the trend inflation rate. Then you know via the Fisher equation you get higher nominal rates and more interest rate room to the cut in times of recession. What did you find in your paper?
Implications of Moderate Trend Inflation
Sims: So the argument there is that there's lots of folks that have said, and we kind of touched on this earlier, is let's be simple. Let's just raise the inflation target and this would raise the nominal interest rate by the same amount in the long run and give us more space to cut interest rates. I think that the cost of moving from, say two to 4% inflation, are much higher than folks that are just wrapped into the black box of a new Keynesian model might otherwise think.
Sims: So that's one of the things that we do in that paper. So we kind of take a new Keynesian model. We add in what we call their roundabout production or firms networking, which in essence generates a source of strategic complementarity, which for a given degree of price rigidity, actually exacerbates the cost of trend inflation and kind of make the argument there that moving from two to 4% might actually be pretty costly. So may not be such a good idea to try to do that, to move away from the zero lower bound and to give us more space.
Sims: This touches on with something I was talking about earlier, just that, thinking about reality and not so much the model, we can write down a model where the Fed announces a 4% target instead of 2%, everybody understands that, everybody adjust to it, and voila, no costs. I think that misses something in reality. I think there's transition dynamics. I do think that 4% would entail costs that we currently don't think about with 2% and I mean after all, one of the big benefits of the conquering of inflation over the last three decades or so is that we typically don't think about it much when we're making decisions. Then all of a sudden when inflation gets even just a little bit higher, all of a sudden you got to think about it more. So there's much higher processing costs.
Sims: So this kind of ties into some of my work on QE, et cetera, is that, I don't think the zero lower bounds, given the availability of alternative policy tools is as costly as one might otherwise think. So this sort of necessarily means that we need to reevaluate proposals to try to avoid the zero lower bound, and raising inflation targets is one such proposal. So that's the proposal that I view with some skepticism, and thankfully I think the Fed views with skepticism, and this review that they've been doing the last year, they never once put on the table moving away from the 2% inflation target as something that they were considering.
Sims: You know, should it be 2%? Should it be 3%? Should it be 1%? We could quibble about this, but I think low and stable is a good thing. I think that that's been a major success of central banking, and I think that if we started to play with that, as a way to try to avoid the problem of the zero lower bound, I think opens a Pandora's box of problems that we don't really want to deal with.
Beckworth: Yep, and as we mentioned earlier, nominal GDP level targeting would solve many of those problems. If you needed some inflation in a recession, nominal GDP targeting would provide it at exactly the right time, as opposed to a permanent trend increase necessarily.
Sims: That's exactly right. I mean it shares some of the features with like, you know Bernanke has called for sort of average inflation targeting. Nominal GDP targeting kind of gets some of the weight of that in a way that I think is actually easier to communicate than something like average inflation targeting.
Beckworth: Yeah. I completely agree, as you know. I've recently heard Chair Jay Powell talk about this very thing, this communication challenge of inflation targeting. It was a listening tour, it was at the Board of Governors and I happened to attend and there was a trade group there, a labor group there. He brought up this question, he goes, "Do you want to hear me say we want to raise inflation?" And both sides said, "No." Right?
Beckworth: So I shot my hand up in the audience, "Please, please let me, let me respond to this." I said, "What you want to say is you want to raise income, you want to raise sales." Or stabilize and better yet. So communication is a big part of monetary policy being effective, in communicating to the average person on the street. So I do think there's good reasons to move towards nominal GDP level targeting, although I have no illusions that we're going to do that anytime soon. It does look like average inflation targeting is the direction that we're going as you mentioned.
Sims: And that's going part of the way, like some insights from nominal GDP targeting are picked up there. So I think that's something of a success, and for proponents of nominal GDP targeting, moving in that direction towards average inflation targeting I think is something of a success.
Beckworth: Okay. Well with that, our time is up. Our guest today has been Eric Sims. Eric, thank you so much for coming on the show.
Sims: Thanks David. I really enjoyed it.
Photo by Win McNamee/Getty Images