David Papell is a professor of economics at the University of Houston and has published widely on monetary policy rules. David joins Macro Musings to talk about his recent paper, *Policy Rules and Forward Guidance Following the COVID-19 Recession,* as well as the origins, past uses, and current applications of monetary policy rules.
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: David, welcome to the show.
David Papell: Thank you for having me.
Beckworth: Well, it's great to have you on, and you have a great paper you have been updating starting in the COVID period. It's titled, *Policy Rules and Forward Guidance Following the COVID-19 Recession.* I love this because it has Taylor rules, several versions of them, and you update it with new data and also new SEP forecasts from the FOMC. This paper is really the motivation for our conversation today. Also, I've been following your work for some time. I think I first saw you present at a Hoover Monetary Policy Conference. I know you and John Taylor go way back. You've been friends. You've collaborated. Maybe tell us about your journey into this area of monetary policy research analysis.
Papell: I started as a PhD student at Columbia wanting to do international economics, and people who think about John Taylor and only think about Stanford forget that his first position was at Columbia and his second was at Princeton, and then, he moved to Stanford. John was hired at Columbia as an econometrician. The only course I ever took with John was Econometrics 1 in my first year and wanted to do international, really had virtually no interaction with him except saying hello in the hallway, until the end of my third year.
Papell: The summer of the third year, I did an internship at the International Finance Division of the Federal Reserve Board, and what I wanted to do is I wanted to work on estimation of models of the exchange rate and the current account with rational expectations and forward-looking expectations. I couldn't figure out how to solve it. Nobody there could figure out how to solve it, but every person there said John Taylor will know how to solve it.
Papell: At the very end of the summer, I went back and I saw him in his office, and of course, he knew immediately how to solve it, but he wasn't going to tell me immediately how to solve it. He was going to make me work through it, which was the absolute right thing to do. We progressed on that and really started with just talking with him about technique, not about the models, because he had never written an international paper at that point. He wasn't familiar with the models, and sort of evolved and ended up being my dissertation advisor on this open economy portfolio balance channel.
Papell: Then, for the next few years, I did work sort of in his spirit. I was doing maximum likelihood estimation of rational expectations models, and I actually have a couple of early papers on using monetary policy rules in the context of exchange rate overshooting, so nothing like what he was doing, no staggered contracts, not anything like that. Then, I moved, we went different directions, or I went in different directions. He stayed at the same direction, and that I got into unit roots and structural change, economic growth, purchasing power parity, and a lot of things. Then, about 25 years later, so I still had contact with him because I did the study guide for the Hall and Taylor Intermediate Macro book, I still kept in touch with him, but our research paths had diverged.
Papell: And so, about 25 years later, I was working with two PhD students at [Houston], Alex Nikolsko-Rzhevskyy, who I'm still working with on estimation of Taylor rules, and then, Tetyana Molodtsova on several exchange rate forecasting using Taylor rules. I had also worked with another PhD student of mine, Ruxandra Prodan, she's a couple of years earlier, and we've just continued the collaboration. She's the co-author of the paper that David mentioned. I basically went from doing nothing based on his work, although inspired by his techniques and a lot of other things, to doing everything based on his work, and that's pretty much where my research has gone since.
Beckworth: No, and you've been doing really interesting work. This most recent work you've been doing has been fascinating because you incorporated the new framework, FAIT framework, because as you know, David, very well, that they added this shortfall from maximum employment, whereas before it was symmetric, and so you have-- We'll come back to this, but you have this term that deals with that in the Taylor rule, or the modified Taylor rule, to kind of accommodate this new framework.
Beckworth: Also, interestingly, you mentioned how you did international work around John Taylor. I actually worked for him at the Treasury Department in International Affairs. He was the undersecretary, so he had an interest there. I recall he has done work on international monetary regimes and how-- in fact, I remember him arguing that if every country had followed a rule-based approach, you wouldn't have to worry as much about the exchange rate dynamics because they would take care of themselves. Let's start talking about monetary policy rules. We're going to end up, of course, with Taylor rules and various forms of it, but maybe walk us through the history of it, David. In my mind, I go back to the history of rules versus discretion, non-activist versus activist rules. Walk us through that.
The History of Monetary Policy Rules
Papell: I mean, when I think of that, I start with Milton Friedman. I guess everyone should start or would start when [with him] you think about monetary policy rules. Actually, when I was at Columbia, my first semester macro was Philip Kagan, which was basically a very hardcore monetarist course, but beautiful course; awesome, funny. The second semester was Ned Phelps, which was completely different, not organized at all, but to my mind, much more inspirational.
Papell: Anyway, back to Friedman. The Friedman k-percent growth rule is a monetary policy rule, but it was not an activist rule. It didn't respond to anything and that was the idea. The idea for there was that what the Fed did was to destabilize the economy by doing monetary policy. It would be better to just let the money supply growth continue on a number, and in that context, the idea of the K% was that the idea of having a number was more important than a specific number. In other contexts, he actually wrote about other specific numbers, but that's a different topic.
Papell: Then, if you think about what the implications there were for the Fed, and from my perspective, this had absolutely no impact on Fed policy whatsoever, in part because if you take this a step or two further, he was basically saying, "Let's close down the Fed, fire the 700 economists," or however many there were, then have some traders on the New York Fed desk and do open market operations and keep the money supply growth constant. You don't need all the rest of this. Surprisingly, this didn't have much impact on policy.
Papell: What I think of what happened as a transition to the Taylor rule, I think of two things. One is, I think, of Ben McCallum's work, which used the money supply as the instrument, but was activist. Then, around the same time, John Taylor's work, where he was using his staggered contract model, but with a policy rule. Again, money supply as the instrument of policy, but sort of started at the bridge, in my mind, from Friedman k-percent growth, no feedback, to McCallum and Taylor, feedback, with money supply as the instrument to the Taylor rule.
Papell: The Taylor rule, the difference is it used the interest rate, the federal funds rate as the instrument. One thing that did is it closed the gap between what the Fed was doing, which was interest rates instead of the federal funds rate, and the academic research. That's one reason I think it's been so influential, not only the Fed, but with other central banks, because it's making an argument of how the Fed should conduct policy using the instruments that they have used all along.
Beckworth: Yes. That's a great summary. Let's move on then to the motivation for rules. You gave us the history of the rules, but there's some intellectual background going on. I think probably the biggest motivation was the work on time inconsistency. What's the argument there for rules?
The Motivation for Monetary Policy Rules
Papell: Well, the argument there for rules in the context of time inconsistency, which was originally done in the late 1970s by Guillermo Calvo and Kydland and Prescott was that, if you try to run optimal policy and by optimal policy, meaning, period-by-period optimization, then what you will end up with is a suboptimal rule. Actually, there's a story, I don't think it is that well known, about the Kydland and Prescott paper and how they came up with that is they were working on this very, very complicated optimal control model. They kept on getting the wrong answer, they kept on getting the wrong answer, they kept on getting the wrong answer. At one point, they said, "Well, wait a second. Maybe it's the right answer," and they went back to square one, and that's where the paper came from.
Papell: In my mind, that richly deserves the Nobel Prize for that paper. There’s also an argument for the policy rules that you can track what it is. You have a signaling argument, you know that people know what you're doing, and then, they can evaluate it. Now, let me take one aside because I think this is something that still gets confused in peoples’ minds. Nobody has ever advocated following a policy rule exactly, except for people who are opposed to policy rules. John Taylor has never advocated doing that. The idea is you use it as a benchmark. Then, if you're going to deviate from it, then explain why you're deviating. I think a classic on this is Janet Yellen's speech in June 2012 at the Boston Economics Club, where she talks about the Taylor rule. We’ll talk about later, but it's become known as the Balanced Approach rule. She talks about optimal policy and looks at it in that context.
Beckworth: The point you just mentioned about John Taylor never saying you have to follow the rule religiously brings to mind a law a few years back before the pandemic, where they were going to require the Fed to state a monetary policy rule. Of course, the Fed and all of its supporters really got worked up about it, but if you read the bill, it said you just need to pick a rule, and then, explain why you deviated from it. You don't have to actually follow it closely. Do you remember that conversation back then?
Papell: Well, I do remember that conversation because at one of the early Hoover conferences, I presented a paper called, *Policy Rule Legislation in Practice.* So I remember that conversation really well. The idea is that-- I don't remember the exact terms they used on this, but the idea is they would have-- essentially, have a rule, his rule, which they could change, it's very explicit, and then, compare testimony to Congress. They weren't specific. I thought they should have implemented it with a monetary policy report, and they could do that and show how that could work. Then, what the idea would be, if they would deviate, is say why you're deviating. I think that there's no necessity for that rule to be legislated. The Fed could do that without legislation and, in part, they have done that. Janet Yellen did that. Rich Clarida talked about rules and not exactly hitting the rules and why when he was vice chair. That was really the sense of using it as a benchmark, not a requirement, and I think that's consistent with how policy rules have been viewed for decades.
Beckworth: Okay, so it serves as a benchmark. It should make us aware of where monetary policy has fallen short and should make the Fed accountable and do some soul-searching when things go differently than as planned. Now, one other, I think, interesting point about rules is you often hear that simple rules work well. Can you help us understand that claim?
Papell: Well, I think that when you think about the origins of the Taylor rule, that came out of studies of macroeconomic models, a number of different models, John Taylor's models at the Fed, and looked at that… there was a book that John Taylor had edited a couple of years before on monetary policy rules, where they invited authors to have their model and then put different authors' models, different models' rules, in their own model, and see what they had. The conclusion there was that there wasn't that much robustness.
Papell: Later on, Taylor and Volker Wieland published a paper where the robustness was much better. They published that in Restat. To try to actually get to your question, I think the idea there is that simpler rules could work better than more complicated [ones], and there was some discussion there. They looked [at] putting the exchange rate in, it didn't work as well. Other things, it didn't work as well. I think there's an argument that simpler rules work better in terms of models. I think there's also an argument that simple rules are just easier to understand, and that's an advantage.
Beckworth: Yes, absolutely. Part of effective monetary policy is communication, signaling, as you noted. A simple rule is easier to communicate to the public than a really complicated rule that includes, potentially, an exchange rate. I've seen some Taylor rules that include credit terms, credit conditions, so you can add a whole laundry list of other variables to throw into the Taylor rule. Let's actually jump into the Taylor rule. We've been talking about it but let's actually spell it out. Maybe before we get into the details of it, and there's a number of terms in the Taylor rule we'll spell out, let's talk about the Taylor principle, which underlies the Taylor rule. Talk us through that.
The Basics of the Taylor Rule and the Taylor Principle
Papell: The idea of the Taylor principle is that when inflation goes up, you want to raise the federal funds rate, which is a nominal rate, more than point for point, so that the real interest rate goes up. The intuition from that comes from, I would say, the sort of late 1970s vintages of what, then, would be called new Keynesian models that John Cochrane now calls old Keynesian models, where you have an IS curve, a Taylor rule, and a Phillips curve. The idea in those models, you raise the real interest rate, the real interest rate going up depresses the GDP, it depresses primarily investment, primarily housing investment. This is not a concept that should be unfamiliar with us right now, especially in terms of housing.
Papell: That's how you slow down the economy, and that's how you stabilize. Where, for example, suppose you raised it point for point. If you raise it point for point, then you have a-- inflation goes up, nominal interest rate goes up point for point, real interest rate is unchanged, investment is unchanged, consumption is unchanged, nothing is unchanged, and nothing brings inflation down. That's the idea of the Taylor principle, is in order to bring inflation down following a shock, you have to raise the real interest rate. I think that's tremendously relevant over the last couple of years and right now.
Beckworth: Absolutely. You're invoking the Fisher identity that says the observed nominal interest rate is comprised of a real term plus expected inflation. If you want that real term to go up, you’ve got to go do more than just what inflation's doing. That's the key to this Taylor principle. Okay, we have the Taylor principle and that underlies the Taylor rule. Let's now get into the guts of the Taylor rule and spell out each of the terms in the Taylor rule. David, walk us through that. There are several terms in the Taylor rule that are important to know. What are they?
Papell: Okay, well, first of all, in the Taylor Rule formulation, you have inflation, and you have the inflation gap, which is inflation minus target inflation, and so think of it as the coefficient is on the gap. If the coefficient on the gap is bigger than zero, then the coefficient on inflation is bigger than one because you just have one plus the gap. That's one term. Second term is the output gap where you have output minus potential divided by potential, percentage deviation of output from GDP. Again, in the original Taylor Rule, it's 0.5. It just has to be positive.
Papell: Then, the other gap is the R-star, the equilibrium real interest rate. In Taylor's formulation in 1993, he set it at two based on the growth rate over the previous nine years of 2.2%. I thought that it would have been much better if he had called it 2.2 because that would have gotten the idea that this is time-varying from the very beginning rather than fixed. I think that might've helped with avoiding a lot of misunderstanding on thinking about Taylor rules when we started thinking of R-star going down.
Beckworth: The Taylor Rule tells us to adjust the interest rate based on three terms, broadly speaking, the equilibrium or R-star rate… That basically is the fundamentals of the economy determining where rates should be, all else equal. Also, the inflation gap you just described… so if inflation's above or below target, the Fed should respond to that. Then, finally, the output gap you said, where if the economy's operating below potential, above potential, rates should adjust to that. Those three things make sense. Then, maybe another way to think about it, if the Fed has inflation on target, if the economy's at full employment, then those terms drop out. The only thing we should see is the federal funds rate equal to the equilibrium target rate as well.
Papell: Actually, it's the federal funds rate, which is nominal.
Beckworth: In nominal terms, yes.
Papell: Nominal equals the equilibrium real rate plus the inflation target. It's the straight Fisher equation.
Beckworth: Right, so those are the terms in the Taylor rule. That's very baseline, that's the 1993 equation, right? Or our version of it?
Beckworth: Okay, so there's been some innovation, some changes since that initial Taylor rule was put out there. Let's start with one that you've mentioned already, the balanced approach. How is the balanced approach version of the Taylor rule different than the original Taylor rule?
Innovations and Changes to the Taylor Rules
Papell: It's very simple. It doubles the coefficient on the output gap. Instead of having a coefficient of 0.5 on the output gap, you have a coefficient of 1 on the output gap. Now, Taylor wrote about this in 1999. He called it, “the rules others have used.” There was a lot of churning, including back and forth between Janet Yellen and John Taylor, on should you be calling this the Taylor 1999 rule, because Taylor did not like that terminology at all because he says, "That's not my rule. That's rules that others have used." I think in 2012, we got the Balanced Approach rule, and that terminology has stayed, which is good. That's really the first change in there.
Beckworth: Okay, so it responds more aggressively to the output gap. As you note in your paper, this balanced approach became more popular after 2008, 2009. We had the weak recovery, and so people begin to think more seriously about slack in the economy as something that should drive what the Fed is doing, not just inflation. Although, we could also point to inflation being below target, too, for a while after that great financial crisis.
Papell: Can I just add something to that? I think the biggest reason it became more popular was because the Taylor rule, at the time with an R-star of two, did not have an interest rate below zero, which means it didn't prescribe quantitative easing. The Balanced Approach rule did prescribe it, and at the time with really large unemployment [where] there could be a five-percentage point difference between sort of zero and minus five for what the target would be. I think that was the biggest reason why the Balanced Approach rule… and the Balanced Approach rule became the favorite rule from the Fed with good reason because otherwise, if you think, okay, here's the rule, and then all the Fed does was go down to zero, you're completely forgetting about anything else they do, and there's a lot of important things they do with quantitative easing with the effective lower bound.
Beckworth: Okay, that's another change then. We've noted a change on the output gap. The weight went up to two. The balanced approach means we're going to respond more aggressively to slack in the economy. But then, the view that you just outlined is that we should be mindful that R-star can change. It's time-varying. It's not fixed as in the original rule. And David, this brings us to some current debates we're having right now. It's a nice segue, I'll bring this up. But there's a lot of discussion right now of, where is R-star? If you go to a famous R-star model from the New York Fed by John Williams and Laubach, it has it relatively low. It hasn't changed, in fact, a whole lot from the pre-pandemic value of it. If you go to another estimate of R-star from, say, the Richmond Fed, it's actually gone up quite a bit. If you look at market TIPS measures of five-year, five-year forward rates, it also has gone up. There's a lot of debate now of where is that R-star? Depending on where that R-star is would definitely have a bearing on what the Taylor rule would tell you. Correct?
The Importance of Determining the Present Value of R-Star
Papell: That's correct. If you, say, follow what the Fed does in their quarterly Summary of Economic Projections then, basically, the R-star, which— you can just back out from their federal funds rate projections in the longer run and their inflation projections in the longer run, which are always two, of course— it went down to from 2 to 0.5 by certainly 2019, so before the pandemic, and it hasn't changed since then. That's actually what we use in our work. That's actually pretty-- related to the Laubach and Williams, except Laubach and Williams went down earlier. I think that made a difference, but I think now, the Laubach and Williams is around the same.
Papell: I think you could say, okay, the neutral real interest rate, people say that maybe it's going to go up in the future. You could have another argument that if you look at the CBO, they're now projecting long-run growth to go down because of climate change. Well, if long-run growth goes down [because of] climate change, every model of R-star is long-run growth and other factors. That would go in the other direction, say, it could even be lower. Jim Bullard will say maybe it should be zero, maybe it's slightly negative now that we may have to be living in that world rather than the Taylor '93, the 2 world. I guess my feeling is that since it may go up for good reasons, since it may go down for good reasons, I'm staying with 0.5 until I hear otherwise.
Beckworth: Well, I think it's also important, if you're following the Fed, to know, what are they thinking? They're thinking 0.5, still. Even if you think R-star is different, what are they doing? How are they going to react? You have got to look at their numbers and their estimates. That's an important point. What is the FOMC viewing as the 0.5? I know we had a little exchange on Twitter about this because the median is at 0.5, but there is some movement in the tails towards a higher value, but not enough people yet to really move the needle.
Papell: Yes. That's true, but I think that goes really in accord with what we just talked about, is if the median is 0.5 and the uncertainty is going up, and maybe one reason we see that uncertainty is going up is that different people in the FOMC have different views. I agree with that, and I also think that having people in the FOMC having different views is very helpful.
Beckworth: For sure. We need more of that. The interesting question then is, where will interest rates go in the long run? Right now, they're up and it could be because of uncertainty. It could be because markets are pricing in the Fed keeping rates higher for longer in terms of actual market rates. There's an argument to be made that the same structural forces in the global economy that kept rates low before the pandemic are going to be around after we're finally out of the woods, all of the imbalances are taken out of the system, and we'll be back to a low-rate world. Now, David, the one pushback I think someone could argue, and I'm thinking of John Cochrane who you invoked a few minutes ago, would be that we seem to have larger structural primary deficits baked in. It seems like they're getting larger and they're not going to go down and that could crowd out, and maybe that could push up R-star. What do you think about that argument?
Papell: I totally agree. We're having larger primary deficits and they sure don't look like they're going down. I totally agree with that. I could see reasons for R-star going up. I said that I'm not convinced that it's going to be 0.5 forever. In fact, I think one thing I'm completely convinced of is it's not going to be 0.5 forever. I just don't know where it's going to be. It's like a unit route where your best prediction is that you'll be at your current level, but you never get there. R-star is not a stationary process.
Beckworth: Okay, okay. You're being very conservative and cautious in using this. Then, again, the other point I think is important, again, is you're just doing what the Fed is doing, too. You're looking at, what is the Fed thinking as well as yourself being cautious? Alright, one other question about this, the Taylor rule, before we get to where we are today, and that is, why is it so popular, so widely used? You mentioned Bennett McCallum. As you know, we recently had a conference for him because he passed away late last year. He had this McCallum rule, as you alluded to. It had the monetary base as an instrument, and it responded to, not only a certain growth rate, but to velocity. It was activist. It responds systematically to changes in the economy.
Beckworth: It was popular for a while, but then it really took a beating when the Taylor rule emerges on the scene in 1993 and loses its luster. Of course, in 2008, it becomes irrelevant because we now have interest on reserves, the monetary base isn't a great instrument for thinking about monetary policy. But why do you think the Taylor rule became so popular? And the name Taylor, John Taylor, right? The Taylor rule, why is it so widely known? You ask any macroeconomist who's gone through a standard macroeconomics program, and they're going to know the Taylor rule. Why is that?
The Popularity of the Taylor Rule and the Significance of the 1970s Period
Papell: Well, I think one initial reason is from his '93 paper. Now, his '93 paper has a lot of normative aspects to it, but the thing that got the most attention in the '93 paper was the graph of the prescribed Taylor Rule and the federal funds rate from '87 to '92, which was spot on, almost indistinguishable. That was viewed as a good period. I think that helped. I think another thing, that we talked about earlier, was that it accords more with how the Fed operates with interest rates than with money supply. But, I think a bigger thing which could come out of John Taylor's '90 paper, work by Rich Clarida, Jordi Gali, Mark Gertler, came from looking at historical periods where we had big deviations from the Taylor rule and bad economic performance.
Papell: In particular, if you look at the 1970s where you have the biggest deviations— the federal funds rate was below the prescribed Taylor rule in any form… Taylor rule, balanced approach rule, whatever— for almost the entire decade during the Great Moderation, it looked better. Then, we have the lower for longer period preceding the Great Recession. I think that had a lot of impact on it, is the association there, and as you alluded to, this paper, *Policy Rules and Economic Performance,* which does a more statistical analysis of those types of things. So I think it's the pictures that have more of the impact than the statistics.
Beckworth: So as they say, a picture is worth a thousand words, and that '93 article [from] John Taylor had a wonderful picture in it that was very stunning and convincing, I guess, and created some momentum, a the life of its own. You mentioned the Clarida, Gali, and Gertler paper, 2000, which is also a big paper. Now, I want to jump to someone else who wrote around that time. I know you've engaged with this person, and that's Athanasios Orphanides. He responds to this literature as growing on John Taylor. He goes back and says, "Look, I don't know about the Clarida, Gali, and Gertler claim that during the '70s, the Fed was not following a Taylor Rule," because if you go back and plug in real-time estimates, not exposed, cleaned-up data, you see that they actually were doing the best they could do. But you, I believe, have gone back and responded to Orphanides, in turn. Is that right? What did you find about that period?
Papell: That's great. First, let me say, I think Orphanides deserves a huge amount of credit for bringing in real-time data into Taylor rules or into policy rules. There was work on real-time data before, Simon Van Norden, other people did on this, but just the idea that if you're going to evaluate policy, you have to evaluate it based on what policymakers knew, not what we know 20 years in the future, is, in one sense, obvious, but in another sense, wasn't done. He completely changed that.
Papell: I think that's great, and some of the other work on real-time data, basically showing that the revisions are not forecastable, also helps with it. That's basically the best that you could do. Now, the argument is not on using the real-time data, but it's on the measure of the output gap. The measure of the output gap that he used was done by the Council of Economic Advisors from '62 on, and this measure of the output gap in the 1970s was extremely large. In 1975, it was a -16.2%, where the revised number was -5% or something. It was much, much smaller. And actually, in January of 2000, I believe, John Taylor discussed one of Orphanides' papers at the AEA meetings and made the argument that this measure became politicized in the 1970s. Serious economists like Arthur Burns and Alan Greenspan paid no attention to it.
Papell: It was like, “assume a can opener.” It was a very colorful discussion. What we did to say is, well, let's ask the question of what would people have thought about output gaps? How would we do output gaps that would do detrending? You do linear detrending or quadratic detrending, and actually, before then, John Taylor and Andy Levin had a paper where they did Hodrick-Prescott detrending, which, of course, the technology didn't exist, but you could do approximately the same thing, it would just have some curvature there.
Papell: Their numbers for that were much closer to the revised. What we did is, we went back to Brookings Papers and Brookings Papers started publishing in the 1970s, and almost each paper had an article or something on the actual rate of unemployment or the full employment rate of unemployment. The idea is you take Okun's law, which at that point was about a coefficient of -3 to go from the unemployment gap to the output gap and say, given these perceptions, what would a perception of an output gap be?
Papell: Basically, these Okun's law perceptions would be, say, let's take 1975 as the biggest example. We also did '72, but middle of '75 is the big one. Orphanides was on -16.2. The Okun's law one, and again, this is based on papers by Bob Hall, George Perry, Michael Wachter, a whole bunch of papers published, which of course were… Brookings Papers was designed to be read by academic and non-academic audiences, so people at the Fed were well aware of this work.
Papell: And again, [they were] well aware of the idea of what… clearly, the view of the natural rate of unemployment. So, by that point, it was at about 5.5. And so, if you did that, you get an output gap of about -9.9, linear and quadratic, the trends were -10.4, 10.8, something like that. Those constructs were much better than either the -16 or the -5. And so, what you get is, if you do the -5, you just go back to basically-- the -5 with the HP detrending is very similar to revised data, but it also gives you things like an output gap of zero in the peak of the 1971 recession, which just makes no sense whatsoever.
Papell: So, the other thing about the CEA numbers is that the official unemployment rate from the CEA started in '62 at 4%. The output gap numbers didn't change from that 4% until ‘77 or '76. So, basically, our result there is you've got something in between. The reason I think we got something in between is that there were two factors. One, is we had a productivity growth slowdown we didn't really understand by then. It started in '73, '74, we didn't understand that in '75, '76, but we had a rise in the natural rate of unemployment that was well understood. You still get the deviations, just not as big as with the revised data.
Beckworth: So, measurement matters. We've got to get our data right if we're going to really get the proper policy implications and lessons from past behavior. David, this has been a fun conversation so far. One more detail about the Taylor Rule that I failed to ask earlier, but we need to settle before we get into your paper on the current state of monetary policy, and that is, what is the inertial Taylor rule versus the non-inertial Taylor rule, and what does it mean for the Fed policy?
The Inertial vs. Non-Inertial Taylor Rule and Implications for Fed Policy
Papell: A non-inertial or just the Taylor rule would be that you have a prescribed federal funds rate and the perception is you just move there immediately. An inertial rule says that you would move over time to get there. Clarida, Gali, and Gertler were the first people to do the inertial rule, that in terms of estimation of Taylor rules, inertial rules are just-- you have to do that. Everybody does that. In terms of normative things, if you look at what the inertial rule… and, basically, the standard inertial rule would put a coefficient of 0.85 on the one-quarter lagged federal funds rate, and then 0.15 on the target level, what you get from the non-inertial rule.
Papell: Then of course, over time, it adjusts to the non-inertial rule. Now, inertial rules use-- certainly when Greenspan raised the federal funds rate by this whole series of 25 basis point increments over several years, could think of that as clearly inertial. On the other hand, we don't want to think of policy rules always being inertial. Policy in the fall of 2008 was clearly non-inertial. We went down to zero very quickly. Policy in March, 2020 was clearly non-inertial, we went down. The idea is that we want to think about inertial rules as inflation is going up and how you respond to that.
Beckworth: So, in normal times, it's better to think of an inertial rule as gradual changes being made, like back when they did 25 basis points, it was always gradual. But in unusual times, like you mentioned, 2008, 2020, would you also include the recent rate hikes that happened relatively quickly? And they went up quite a bit. Would that also be a non-inertial period?
Papell: No, I think, currently, you can't understand Fed policy in terms of non-inertial rules. I think what I was trying to say is that when you're raising rates, you're raising them slowly. At times, in two particular times, we cut rates in 2008 and 2020. You don't want to do that slowly. In normal times, non-inertial is fine, because there's not much difference between non-inertial and inertial rules, so do the simple one.
Beckworth: So, David, how do you interpret the most recent rate hikes over 2022, 2023: inertial or non-inertial?
Papell: Definitely inertial, but still behind the curve.
Beckworth: So, David, the paper that's really fascinating and really motivated this conversation is titled, *Policy Rules and Forward Guidance Following the COVID-19 Recession,* and as I mentioned, you've updated it through this period as new SEP forecasts come out and new economic data becomes available. Maybe talk us through that paper and also what FAIT, the new framework, meant for thinking about Taylor rules in that paper.
*Policy Rules and Forward Guidance Following the COVID-19 Recession*
Papell: The idea of the paper is to evaluate policy rules and a range of policy rules, traditional ones and rules inspired by the new framework, inertial and non-inertial, using real-time data up to the point of the FOMC meeting starting in September of 2020, and now going up through September of 2023. And so, two things, we use the data available and use the data that the Fed uses, core PCE inflation, how they define unemployment, all of those things. Then, we use the Summary of Economic Projection’s forecasts for the next few years, which are now through December of 2026. So, to go back to August of 2020, this FAIT or Flexible Average Inflation Targeting, is the idea is that when inflation is below target for a long period of time, you want to raise it above target temporarily, because otherwise you get stuck the way we got stuck the decade before, in having the 2% as the upper bound of inflation, not an average of inflation.
Papell: The second part was the idea that we should be responding to shortfalls of unemployment from U-star, not deviations, and that came from the idea of the flat Phillips curve, that you can run the economy hot, we can lower unemployment below 4%, and still not have a big effect on inflation. Of course, what the Fed does, of course, is they don't define these terms. Exactly, that's part of what we do is to define terms. So, the first aspect of this is FAIT, and I think that was very important early on. Let's look at, what are the rules we use? We look at the Taylor rule, we look at balanced approach rules, we look at Taylor shortfall rules, which just has the unemployment part. We look at balanced approach shortfall rules. We look at Taylor and balanced approach, what we call consistent rules to be consistent with the revised statement, that also has the FAIT part plus the unemployment part, and we look at non-inertial versions and inertial versions.
Papell: So, we look at 12 different rules for each of these time periods. Everything is just graphs. We're not estimating anything. We're just looking at that and we're comparing it with these prescriptions of these 12 rules with the federal funds rate. Initially, that inertial, non-inertial was irrelevant. Initially, what mattered most was really the consistent rule, was the FAIT plus the shortfall. If you go back to September of 2020, these rules were predicting liftoff from the effective lower bound as early as March of 2021, which was the original Taylor rule, and as late as June of 2023, and at that point in time, the FOMC projections were to be at the effective lower bound through the end of 2023. And so, the consistent rule, I think, was important there, because it was the closest to the Fed predictions, which is also because it was the closest to what the Fed was doing there.
Papell: Now let's move forward to early 2021. In March of 2021, inflation was about 1.5% for annual core PCE, and by June of 2021, it was about 3.5%. Now, a big part of that is you took out the March 2020 number for that, but there's this big jump in the annual core PCE. Maybe more importantly, the Fed prediction, the median prediction for core PCE inflation for 2021, went from 2% to 3%. We basically went from being in the FAIT world, where we want to go above the inflation, to the world where that's just not optimal. It doesn't matter anymore because inflation has gone too high. So, basically, you can always say things like, “the FAIT has met its fate,” or my favorite was Ben Bernanke's, “a simple twist of FAIT,” to quote the Bob Dylan song. But then, so what about shortfalls? To have a difference between traditional and shortfall rules, you have to have unemployment being under 4%.
Papell: That started in March, 2022. So, before that, it didn't matter. Even now, it makes some difference, but not a huge difference there, because if you look at unemployment— now, we talked about the coefficients on the Taylor rule using Okun's law equal to two. The coefficient for the Taylor rule is one on the unemployment gap and two for the balanced approach rule on the unemployment gap, so it makes some difference, especially for the balanced approach, but not that big. What makes the big difference is inertial versus non-inertial. And so, simply starting in September of 2021, all of the rules prescribe an increase above the effective lower bound; inertial, non-inertial, all 12 of them prescribe an increase above the effective lower bound. And again, this is using the information available at the time and the measures that the Fed use, so when Fed officials say, "Well, if we had known this at the time, we would've acted differently," my response was, “You did know this at the time and you didn't act.”
Beckworth: Is that the date then, when you would say they fell behind the curve?
Beckworth: Okay, continue.
Papell: Yes, they definitely were behind the curve by September, and it's pretty similar with inertial and non-inertial. What happens by March of 2022, when they first raised the federal funds rate… by March of 2022, if you used non-inertial rules, you were about 800 basis points too low. That's why I look at non-inertial rules and go, non-inertial rules just became irrelevant. The Fed's not going to raise the federal funds rate by 800 basis points between September and March. That's just nonsensical. If you look at the inertial rules, it was about 200 basis points. So, to get between September '21, to 200 basis points by March of 2022, you could do it by, I think it was a 150 basis point increase and everything else could be 25 basis points; so certainly doable in terms of what the Fed does.
Papell: Now what happens starting in March of 2022? First of all, the Fed started very aggressively raising the interest rates. They started raising interest rates and if you look right now, inertial and non-inertial prescriptions and the federal funds rate are really close to each other. If you look, say, for prescriptions for December of 2023, the Fed is actually 25 basis points, would be 25 basis points above the prescribed rule with the balanced approach and 50 basis points above with the Taylor rule. One thing that's interesting now— and this is using SEP projections from September, which would prescribe one more 25 basis point increase. Now, a lot of this between September and now, and certainly with the November meeting, is to say that there's certainly an equal chance, maybe a more than an equal chance, that they're going to pause, that they're not going to raise the interest rate in December.
Papell: Jay Powell at the press conference left it open whether they might raise it, they could still raise it later. Well, if they don't do that, they're even closer. They're basically exactly on the target with the balanced approach of 25 with the Taylor rule, which makes no difference. So, I think on one of the Econbrowser posts, we wrote that the Fed is following a Taylor rule. They're really right, and that's pretty much with the inertial or the non-inertial. Now you project going forward, and there's still the federal funds rate forecasts, and the inertial Taylor rule prescriptions using the FOMC projected forecasted data are just spot on. They're never more than 25 basis points. Interestingly, the non-inertial rules say that they should start cutting earlier, which is, of course, what markets are saying.
Papell: So, in some sense, as an approximation, markets are saying that the Fed is going to not use these inertial rules, which would keep it higher for longer, but use the non-inertial rules— well, not exactly, which will put it lower faster, which I think is in accord with how the Fed has cut rates before. They haven't been constrained to have slow rate decreases the way they've been constrained to have slow rate increases. The non-inertial rules, again, whether you believe the September SEP or the markets, the non-inertial rules may become a better description of Fed policy for the next couple of years, and again, using Fed forecasts, these median forecasts of inflation and unemployment.
Beckworth: That is really interesting, and I'll encourage listeners to go check out figure 14 in David's paper that we're talking about, because you provide a nice chart of the non-inertial rule and then the inertial rule. You can see exactly, in a graph form, what you just described, that they are right now roughly where they should be per these rules. But going forward, the non-inertial rule shows steeper cuts than the inertial rule, and markets seem to seem to think along the lines of the non-inertial rule. So, that is really fascinating, and again, the chart, figure 14, really summarizes this nicely. It's a powerful picture.
Papell: Can I just add one thing, which is that much easier than reading the paper, by the time this is on, there should be an Econbrowser post with it so you could get the two-page version instead of the 40-page version, which I would highly recommend, with the same graph that David talked about.
Beckworth: We will provide a link both to the paper and to the Econbrowser post, but yes, a picture is worth a thousand words, and this is a really powerful picture. So David, in the time we have left, what would you tell the FOMC? What advice, suggestions, because we know some of them listen to the podcast. What would you leave with them as we close the show?
Papell: I guess my advice would be that, almost everyone in the FOMC believes now that they fell behind the curve, but their falling behind the curve wasn't because they-- the sort of what we call the “we didn't know” hypothesis, or what we call the Larry Summers “they should have known” hypothesis, that if they had taken policy rules using their own data, their own rules, their own benchmarks, looked at the non-inertial path, they would've known earlier that they should've raised, and we would have gotten to where we are now with a much smoother path, mostly 25 basis point increases, a couple of 50s, no 75s, none of the churning, none of all the craziness that we've had as the rates have gone up. So, I guess my response on that is, yes, you should look at the policy rules, think of it as a benchmark, and then think, if you don't want to follow it, think about why. Then, if you think about why and you don't want to follow it, don't follow it. That's why we don't have policy being determined by a rule, but think about it in terms of the rule.
Beckworth: Okay, well with that, our time is up. Our guest today has been David Papell. David, thank you so much for coming on the podcast.
Papell: Thank you.