Jonathon Hazell on the Costs and Causes of Inflation and the Phillips Curve Debate

Jonathon Hazell is an assistant professor of economics at the London School of Economics and is a returning guest to the podcast. He rejoins David on Macro Musings to talk about the costs of inflation, the Phillips curve Debate, and the lessons learned from the post-pandemic inflation surge.

Check out our new AI chatbot: the Macro Musebot!

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: Joe, welcome back.

Joe Hazell: Thanks, David. Thank you so much for having me back. It's an honor. It's great to be back.

Beckworth: It's great to have you on, and you have been doing some traveling. In fact, you're here in the US right now as we speak. Tell us what you've been up to lately.

Hazell: It's been a great time. I've been writing quite a few papers, and it's been about inflation. So, I'm interested in two questions. I guess we'll get into it today. I think that these are the two questions of the moment. The first is the costs. So, why do people hate inflation so much? And then second, the causes. Why did the inflation happen? We all know that inflation has been very high, but why is it bad and why did it happen? I think that these are the key questions at the moment. It's a great time to be a macroeconomist right now, because all of the macro events are so exciting. So, I guess, David, you were obviously blogging a lot 10 years ago during the Great Recession, and I think that that was another time when it was all macro, all the things that were going on. There was a bit of a lull [during] maybe 2015 to 2020. Everything was too stable. Everything was fine. But now, we're back. The macroeconomy is doing terribly. Macroeconomics is back. Inflation's high. So, it’s really exciting just answering these questions. Writing these papers has been a ton of fun.

Beckworth: Well, it is really interesting to sit back and reflect. If we went back to that 2015, 2016 period, no one would have expected that the hot topic would be inflation, its causes, [and] how to manage it, but here we are. The tables have been turned, and we're back in a world where we really care about inflation. We thought we had beat that problem in the 2010s.

Hazell: Right. No, exactly. It's inverted, because every problem that macroeconomists complained about between 2015 and 2020— low inflation, falling interest rates, less than full employment— they've all been fixed, and I guess there's this real Goldilocks thing where maybe they've been fixed slightly too much. The economy is too hot instead of too cold, and these are the questions.

Beckworth: Yes, absolutely. So, you've been doing a lot of work on inflation, and thanks to people like you and others, we are better able to understand what happened during the inflation surge and how to maybe better manage it going forward. So, you're a very busy individual, but Joe, what do you do for fun? I'm just curious. What does a busy macroeconomist do with their leisure time?

Hazell: The main thing, obviously, is that I hang out with my wife and my family. Other than that, I'm a big ultra long-distance runner. So, I like to run multiple-day races, 100-mile races. So, actually, when I'm in California, I'm going to run a 100-mile race in Big Bear. It's my first 100-miler. I'm really excited. It's going to be long, it's going to be tough, but I think that you really get something out of it. So, it should be fun.

Beckworth: Joe, that is amazing. I did not know this about you.

Hazell: It's a little odd. It's a little odd of a habit.

Beckworth: No, no. You will represent us well, the econ profession, in the ultramarathon. I have followed some documentaries on the big ultramarathon race that comes out of Death Valley. It's over 100 miles. It's incredibly hot. Shoes melt on the road, I understand. I saw this one gentleman who would eat a whole cheesecake just going up the mountain to get to the other side. So, that's amazing that you're doing that and doing all of this interesting work.

Hazell: So, that one— I think it's called the Badwater Ultra. I would like to do it one day, but it's very hot. So, I don't know. Would I be able to? We'll see.

Beckworth: Well, you are competing now, and I believe that you have to do so many ultramarathons to get into that one. So, we look forward to you— when you come back in a few years, you'll tell us the story of how you survived the ultramarathon out of Death Valley.

Hazell: You're writing me checks that I can’t cash David.

Beckworth: Hey, that's what macroeconomists do. We like to get ahead of ourselves. So, as we mentioned, we are going to talk about inflation today, and one of the big puzzles is, why do people hate inflation so much? Again, this is a question that we probably weren't worried about a few years ago before the pandemic. But one thing that we really learned, or maybe we relearned, was that people dislike inflation after it reaches a certain level, and it stays, and it's persistent. It was very salient, politically. We had the elections here, midterm elections. We expected a big red wave. It never happened. But still, there was some political cost. It's still being talked about now in the election. Who's responsible for the inflation? Was it the Biden administration? Was it the Trump administration? Was it supply shocks? The pandemic? All of those things.

Beckworth: And as you know, Joe, we peaked here, in the US at least, at 9% CPI, [and] I believe [it was] around 7% PCE. Now, we're back down close to target. It looks like we're going to land on target pretty soon. In fact, the Fed, today, has cut interest rates. So, we're on a journey to normalcy in terms of inflation, but man, we learned a lot. We were reminded of a lot that inflation can take off. Inflation in terms of polls became very important. It's come back down. Even household inflation expectations became a little unanchored, at least the short term ones, which was surprising. But those are back down as well. So, it looks like we're returning to normalcy, and I'm just beginning to think, "Why do people dislike inflation?" Joe, so, what have you been working on, on that question?

Breaking Down the Costs of Inflation

Hazell: Great. I think that it's a key question and it's a puzzle, because everyone thinks that inflation is bad, just like you're saying. You look at the polls, you look at specific questions on Pew saying, "What do you think the main problems facing America are today?" You introspect, you ask your neighbors, [and] everyone agrees that inflation is really bad. But I think that one almost embarrassment for macroeconomists is that the costs that we write down and think about in our standard models suggests that inflation isn't too bad, or by introspection, the costs don't seem that big. So, what are the standard costs that macroeconomists talk about?

Hazell: We talk about things like menu costs, that firms have to change their prices more frequently, shoe-leather costs— So, you have to go to the bank more regularly because you want to optimize how much cash you hold versus how much of other assets you hold. Maybe there are greater cognitive costs, because you're trying to figure out what's going on more. But the major costs that economists talk about, to my mind, and I think to many people's minds— they don't seem that big. Rather, they seem really important during hyperinflations, but maybe not during an inflation of 10%. That's my sense of it.

Hazell: So, if you've talked to people who have lived during hyperinflations— like I was talking to an Israeli economist who lived through the hyperinflation of the late 1980s, in Israel, who was saying, "Yes, this is a real thing.” Firms are spending all of this time thinking about how to change their costs. People are spending all of this time thinking about how often to go to the bank, how much cash to hold, and so on, because, of course, the value of cash is going to be eroded by inflation and so on. So, during inflations of like hundreds of percent a year, these menu costs, these shoe-leather costs seem really important, but maybe not during moderate inflations like 10%.

Hazell: We know that people hate inflation right now, but it's not clear that they hate them because of this menu cost, this shoe-leather cost that we write down in our conventional models that shows up in something like Mankiw's textbook. So, those are the costs in our textbooks, but people say that inflation is bad for other reasons. What people say, almost uniformly, if you go out and run surveys, is that people don't like inflation because wages don't keep up with the prices. So, there's a classic survey by Shiller making this point. He goes out [and] he asks people in the late 1990s, "Why don't you like inflation?" What people say is, "We don't like inflation because prices are going to rise. Wages aren't going to catch up. That means I'm going to be poorer."

Hazell: My experience is that if you go ask people, "Why don't you like inflation?" That's exactly what they'll say. That's not really something that economists have thought too hard about, or rather, I think that they've dismissed this as a major cause of inflation until now. And we're trying to revive this idea. We're trying to say, "Look, this is what people say is important. How do we think about that seriously?" That's my basic sense. There's a bunch of costs that we wrote down, that we thought about in macro models, but maybe these intuitively don't seem so big. Instead, people are saying that something else is going on, about wages catching up with prices, and we want to take that seriously.

Beckworth: Yes, and you could even go as far as saying that inflation shouldn't matter at all, right? If we perfectly anticipate it, if we index it, it should not be a big deal. But clearly that's not been the experience, right?

Hazell: Exactly.

Beckworth: That's not what we have seen. So, you have a really interesting paper where you begin to tackle this question. It's a co-authored paper, and the title of your paper is, *Why Do Workers Dislike Inflation? Wage Erosion and Conflict Costs.* Talk to us about that.

*Why Do Workers Dislike Inflation? Wage Erosion and Conflict Costs*

Hazell: Great. So, I'll come back to the preamble that we were discussing. So, there's this view, that Shiller found in a survey, that, subsequently, some great work by Stefanie Stantcheva and by Hassan Afrouzi recently has revitalized, which is that people don't like inflation because when inflation is high, real wages are falling. Nominal wages aren't catching up. Then, I said that traditional economists haven't taken this view too seriously. So, why not? The classic view is that maybe these costs of inflation aren't too big, despite what the people are saying. So, this is a view that, for instance, Mankiw puts forward in his textbook. His argument is that if you look at the data, nominal wages generally keep up with prices after an inflationary shock.

Hazell: So, if I go to the time series, I plot inflation, inflation goes up, and then I plot nominal wages, nominal wages tend to go up as well. So, in the data, real wages don't persistently fall. And so, people like Mankiw, in his textbook, are saying, "Well, probably, this cost of inflation doesn't matter too much,” because people say, "Look, real wages are falling. I'm getting poorer during inflation," but if you look at the data, that doesn't seem to be the case. So, that's the status quo. That's why I think maybe economists haven't taken too seriously what people are saying.

Hazell: And so, we're going to try and reconcile these two views and put a new spin on the idea. We're going to say that inflation imposes costs on workers, because during a time of high inflation, employers don't automatically give workers raises when inflation is high. Instead, workers have to fight for these raises. That places them in conflict with employers. That's the key idea of the paper. It's bringing forward this notion of conflict. It's saying that when inflation is high, for nominal wages to catch up, to catch up with prices, people need to be doing conflict. That's difficult. That's painful. So, that's the basic idea, so let me walk you through some of the details of that. But if your listeners remember one thing, that's it. If wages are going to catch up with prices after an inflation, there's going to be all of this painful process of conflict.

Beckworth: Let me just jump in here and mention a tweet that Vox's Dylan Matthews did yesterday. I think [it was] very telling, very clever, and he's talking about your paper. He goes, "Inflation is a tax on conflict-averse people who are bad at negotiating— me." I thought that was very funny and a nice summary of your paper.

Hazell: Exactly, no, that's perfect. Thank you for that. And our hope is that, this paper— it's not exactly an observation that we've seen somewhere else, but we feel that when we say it to people, it resonates. People say, "Yes, I, myself— that's why inflation is difficult."So, I'll give you a little anecdote before I get into the details. Imagine you're a young assistant professor of economics at the London School of Economics, and inflation has been pretty high [during] the last few years, so your wage is falling. How do you get a wage increase? LSE don't just gift this young assistant professor a 5%, 10% nominal wage increase. No, no, you, I won this hypothetical, professor. I'd have to go out on the job market, I'd have to search for another job, and I'd have to come back to LSE and say, "Look, I've got this high job offer. You need to give me a raise."

Hazell: That's really costly. That's painful. That's a conflict. That's the idea that we want to be thinking about. So, more systematically, we're going to start with some motivating survey evidence. We're going to survey a bunch of US workers at the start of 2024— so, just after the big inflation— and we're going to ask them about how they think about these issues, and we arrive at two findings. The first finding is that we find that this conflict is important for determining wage growth. What I mean by that is that a significant portion of workers say that they took costly actions to achieve higher wage growth than what their employer offered.

Hazell: So, what do I mean by costly actions? I mean things like having a tough conversation with employers about pay, partaking in a union activity, or soliciting job offers from other companies. So, finding a job offer at another company, having that raise to come back to your first employer and say, "Wages should be higher." The first is this fact that workers often need to take costly actions to have wages go up. Then, the second is that, well, these costly actions do lead to higher wages. So, people who say, "Yes, I took this costly action," report that their wages are much higher than they would have been without these costly actions. People who didn't take the costly action say the same thing. They say, "I think that if I had taken this costly action, this conflict, my wages would have been higher."

Hazell: But many people don't take the costly action, suggesting that there are costs, as well, to doing so that offset the benefit of a higher wage. So, that's the first survey motivating thing. Then, the second motivating survey thing is very consistent with these previous papers, like Stantcheva that I was mentioning, which is that workers are more likely to engage in conflict when inflation is high. So, people say, "Look, why did I take this costly action?— the tough conversation, and so on— "I did it because I wanted my wages to keep up with inflation." If you say to people, "Look, imagine if inflation was high. Would you be more likely to engage in conflict?" They always say, "Yes."

Hazell: And in the real world, so, in observational data, consistent with what we find in this survey of US workers, [is] when inflation is high, again, there seems to be a lot of conflict. So, specifically, what we do is, we go back to the data— this is something that we've seen before— and we look at the rate of union strikes. This is a good proxy for conflict out there in the real world. We find that when inflation is high in cross-country data, unions are much more likely to strike. So, this is sort of known, but sort of new, and it's very consistent with what we're saying. When inflation is high, there's going to be more union action. Again, this is something you see in the US and the UK. During the so-called “cost of living crisis,” there's been lots of union action, which is one proxy for conflict.

Hazell: So, that's the first part of the paper [which] is purely motivational. It's saying, in a way that hopefully resonated from the anecdotes that we were telling you, that conflict and its relation to inflation seems to be a key part of how to think about inflation and wage setting and its costs. So then, we want to sit down and formally model it, and the model that we're going to write down, what we're calling a "conflict cost model," is designed to capture these features of wage setting and then deliver some insights about the costs of inflation via this conflict channel. So, the way that we're going to think about it— it's really simple.

Hazell: You're a worker, you're going to receive what we call a default nominal wage offer from your employer. That offer probably isn't going to be fully indexed to inflation, which means that when inflation is high, the default wage that your employer is offering to you is going to be lower in real terms. In response, if you're the worker, you can either choose to accept the offer or engage in conflict. Now, if you engage in conflict, your wage is going to keep up with inflation, but you have to pay these costs. What are the costs? Exactly what we were saying before— the costly job search, the tough conversation, and so on.

Hazell: As you'd expect, when inflation rises, conflict is more likely. Intuitively, that's because when inflation is higher, the employer's wage offer isn't catching up with inflation, isn't keeping up. So, workers' real wages would fall, and then they're more likely to conflict in order to have their wages catch up with inflation. That's the background, that's the model that we set up. Then, our main result is to characterize the welfare cost of inflation in this model. What we find is that the path of real wages is no longer enough to inform worker welfare in this setting. So, we were going back to the start, we were talking a lot about real wages, and we're shifting the focus and we're that saying real wages aren't enough because of these conflict costs.

Hazell: So, in particular, what we see in our model is that wages might catch up with inflation when inflation is high, but that catch-up is achieved through higher conflict. That conflict has all of these costs. And so, the wage catch-up that happens after inflation doesn't necessarily raise welfare. Why not? Because the wage catch-up that's happening after inflation is being affected by all of these extra conflict costs that workers are paying in order to ensure the higher wages, and these extra conflict costs that workers are paying offset the benefit of higher wages. So, again, if I look at the data, I see inflation rises, I see wages catch up with inflation, so I might think, “Nothing to see here, no costs.”

Hazell: Actually, in the background, all of that wage catch-up is associated with this process of conflict. Workers are really having to fight hard in order to get this wage catch-up, and so, the benefits of the wage catch-up are limited. Instead, what we're saying is that the impact of inflation shocks on worker welfare is determined by what we call wage erosion. So, we show, we prove, formally, that wage erosion is what determines workers' welfare. What's wage erosion? It's how inflation would affect real wages if workers' conflict decisions weren't moving around with inflation. So, that's the analytical result, that the welfare cost of inflation in the labor market can be really high even if real wages don't fall, because in the background, in order for real wages not to fall, workers are having to do this process of costly conflict.

Beckworth: That actually sheds some light on the questions that we see being asked today. A lot of economists are like, "What's the big deal? Why do people keep complaining about inflation? Wages are growing. Everything should be fine." Paul Krugman loves to go off on this. But what you're saying is, "Look, on net, there's a welfare loss. People are feeling something real. It's that conflict cost." Correct?

Hazell: That's exactly right. I'm sympathetic to Paul Krugman, because before writing this paper, I had the same feeling. “Nothing to see here. Real wages are actually doing pretty well. Unemployment's low. Things look great.” Then, you talk to people, and they have this sense of anxiety of, "No, actually, it's a really tough time," and I think the reason is because, exactly, that in order to have real wages doing okay, in the background, they're having to take all of these difficult actions.

Hazell: So, we have this motivating evidence, and then we have this model with this, we think, pretty nice analytical result. Then, of course, the natural, final part of the paper is to say, "Quantitatively, how important is this likely to be?" And we're going to figure this out by calibrating the model using the survey. We're going to take some survey questions to directly inform what turns out to be the key parameters affecting conflict in the model. In particular, we're going to ask people, using our survey, "How much wage growth would you sacrifice in order to avoid taking conflict, in order to avoid these costly actions?" We find that the typical worker is willing to sacrifice 1.75% of their wages in order to avoid conflict.

Hazell: Obviously, this is going to map pretty tightly onto the costs of conflict, the cost of inflation, due to conflict in our model, because we're finding that workers are willing to sacrifice quite a bit of wages in order to avoid conflict. So, that's the first thing that we find with our survey. The second is that workers believe that, in the absence of conflict, firms wouldn't change their wage offers very much when inflation is high. So, workers believe that, when inflation is 10%, unless they take conflict, firms are going to still make the same old wage offer that they would if inflation were 2%, suggesting that workers need to take conflict in order to have the real wages rise, just as we're hypothesizing in the data.

Hazell: Then, the final part of the paper is just to say, "Okay, let's take our model, let's take these obviously quite informative statistics in the data about the cost of conflict, about the likelihood that wages rise absent conflict, and let's take these together and quantify how big these costs are." Basically, they're quite big— hopefully, as you're persuaded by this stage— And the headline way that we summarize this is that we say that the conflict doubles the cost of inflation to workers. So, in a world without conflict, workers would dislike inflation half as much as they do. And that's pretty big. We think that this is a pretty reasonable number. And so, that, hopefully, wraps together a pretty reasonable message.

Beckworth: I want to go back to your 1.75% number. So, is that saying that if inflation goes up 1.75%, people are willing to tolerate that much erosion in their wages before they actually do something?

Hazell: Exactly, and we need the full model to know if that's big or small, but just looking at that number, it's not peanuts, right? It's quite a lot.

Beckworth: Oh, no, it's absolutely— right. I want to tie that into some other literature, because I think that that's an interesting number. So, if we view 2% as the baseline target inflation rate, and we add that 1.75 to it— let's just round up to 2— we get to 4. What's really striking is that there are these articles on inflation and attention, and they find that there's a threshold. People really start to care, at least in the US data, when you get to 4%, and it's pretty striking that your results fall right in line with that. So, you've got to start worrying about this. So, Joe, let me ask you this follow-up question. What would be a policy implication of this? If I'm at the Federal Reserve, and I'm thinking about inflation targeting, price level targeting, nominal GDP level targeting, is there an implication here for me?

Outlining the Policy Implications

Hazell: I think that there are two implications. So, let me make a distinction between steady-state and shocks. So, what I mean by that is how the Fed should think about an inflation shock and how the Fed should think about changes in steady-state inflation, changing the inflation target from 2% to 4%. Regarding the shocks, I think what this basically means is that, relative to a world where this channel doesn't exist, inflation is more costly. So, if you're a policymaker, and you're worried about something like some mix of inflation deviations and unemployment deviations, you should be a little more worried about inflation. I should caveat here that I don't want my co-authors to eat me alive. These are follow-up questions that we're working on right now, so we haven't fully figured out the subtleties, the details.

Hazell: This is my understanding of what we've done so far, but I don't want to commit to this. So, relative to the rest of the paper, which is all done and all figured out, this policy implication stuff is a little bit more off the cuff. That notwithstanding, I think the clear implication is that, if you're a policymaker, you should worry a little bit more about inflation and a bit less about unemployment. And that accords with what we were saying before, that if you talk to people out there in the real world, even though unemployment has fallen a lot, they still seem a little more worried about inflation. So that's just in a shock sense. If you're the policymaker, how should you respond to shocks?

Hazell: The second one is at the steady-state. So, previously, there was a lot of discussion about whether or not we should raise the inflation target from 2% to 4%, or something like that. Now, if you raise the inflation target to 4%, what that might mean is that there's just a lot more conflict all of the time, and that could be bad for workers. It's not totally clear, so I think that if you raised inflation to 10% or 15%, you would probably start entering a world where there's something like indexation— some institution to resolve these issues a bit more efficiently. But I think that, at the margin, if you raise inflation a bit more, it's possible that people don't find it worth their while to reorganize the whole labor market and change all of these contracts.

Hazell: Nevertheless, they're going to have to deal with the inflation somehow that could be providing more conflict. So, again, this is a little more speculative. Our paper's really about the shocks, not the steady-state, because it's a little bit harder to think about these broad questions of reorganizing the labor market and so on. But that's my intuition, that at the steady-state or the shock, inflation's a bit worse than you would think if you hadn't been thinking about this channel.

Beckworth: This is very interesting, and it reinforces, again, that inattention in the literature that I just mentioned, where they say 4%, because it says that if you raise the inflation target, there's going to be more conflict, and there's going to be a welfare loss that we don't directly observe that we need to think carefully about. So, I don't think that the Fed is even considering raising the 2% target during the framework review, but it has been brought up multiple times by prominent macroeconomists. I think it's good that you're adding or contributing to this debate. Another thing— it strikes me, at least, as an advocate of make-up policy or level targeting of some kind, whether it's a price-level target or a nominal GDP-level target— It requires to run the economy hot if you're falling below.

Beckworth: So, if you've missed your inflation target, you're below 2%, and you've got to run it above, 3%, 4%— Well, this suggests to me that maybe those policies may not be so easy to implement. There might be more conflict. There might be some welfare loss associated. So, the payoff to make-up policy and level targeting better be pretty darn good if we're going to do it, because there are these added costs to them.

Hazell: Sure. No, I think that's right. The lessons that we've learned from the financial crisis about, for instance, the benefit of level targeting, of nominal GDP [targeting], of price level [targeting]— that's still really important. And I think we just want to neither over-learn nor under-learn those lessons and acknowledge that an inflation overshoot also has pains, too. Then, the question of how to balance these things out is exactly what we need to figure out in our next paper, hopefully. We'll see.

Beckworth: Absolutely. One last implication, I think, that flows out of this, your work, and again, these threshold papers, [is] if the costs get high enough, the conflict costs get high enough, people start to negotiate, after this pain, after this welfare loss. And again, the 4% number seems to be right, pay attention. This also, to me, sheds light on why a supply shock, which we're supposed to normally look through if I'm a central banker, can actually exacerbate inflation expectations. Typically, you say, "Oh, we've got things anchored. Look through it," but if people start to really negotiate, begin to care at 4%, they can both be a cause and a determinant of higher inflation, just having to go through this process.

Hazell: Exactly. So, actually, an early version of this paper— this ended up not being in the paper, but it's an interesting observation that we might want to do something with— is that you can imagine that, as you transition to this world of higher inflation, people just start negotiating much more regularly, and that has its own momentum and continues to increase the inflation. One thing that you see in the 1970s— as far as I know, this is a new fact that we found, but we haven't made much of yet, but intuitively, or in the news, I think it's known— was that in the 1970s, when inflation was high, indexation clauses became much more likely. So, unions started to change towards saying, "Actually, what we really want is cost of living indexation clauses." And that's exactly what you're saying. Once inflation gets high, there's more negotiation, and then that extra negotiation, that extra conflict, can have its own momentum. And so, sometimes people call this a wage-price spiral.

Beckworth: Yes. Okay, well, those are very fascinating findings. Your paper is out on the— it's an NBER working paper, is that right?

Hazell: It's an NBER working paper. It's on my website. It's on everyone's website. 

Beckworth: We'll have it in the show notes as well.

Hazell: Perfect. Thanks, David.

Beckworth: So, we look forward to the contributions and ongoing discussions surrounding this paper.

Hazell: I'm excited.

Beckworth: Alright, let's segue into something else that you really love, your bread and butter, and that's Phillips curves.

Hazell: Right, the Phillips curve.

Beckworth: We talked about them before on our previous show, so go back and check [it] out. But it's a good time maybe to step back and take stock of what we've been through, what we've learned about Phillips curves. What are the lessons learned? Because we seem to have accomplished, at least in the US, a soft landing of sorts. The Fed's cutting rates, as I mentioned earlier. So, what have we learned? What are the implications for policy? I know that there's been a lot of research. There's a lot of debate. What type of Phillips curve? What tweak to a Phillips curve? So, walk us through that conversation and where you land with it.

The Recent Phillips Curve Conversation: What Have We Learned?

Hazell: Totally. Excellent. I'd love to do that. So, I'm going to give you 30 seconds of background, which people who have listened to our previous episode will know, but at least it's a good refresher. So, what's the Phillips curve? It's a relationship between inflation, inflation expectations, slack, and supply shocks. So, three things can cause inflation— supply shocks, slack, and inflation expectations. The background is that, before 2020, after 1978 or so, the Phillips curve seemed to be flat in the sense that the slack term from unemployment, from labor markets, and so on, didn't seem to affect inflation very much.

Hazell: So, we liked this because this Phillips curve— it could explain the missing disinflation during the Great Recession. Inflation didn't fall very much. [During] the missing reinflation [of] the late '90s and the late 2010s, inflation didn't rise very much, and also, the fall in inflation during periods like the 1980s. So, there, the idea is that inflation expectations fell a lot, and that's why inflation fell a lot. Now then, the question is, this flat Phillips curve, does it work after the pandemic? And I should say, full disclosure, I was one of the people who was estimating flat Phillips curves. So, me, I'm haunted by this classic quote by Stock and Watson.

Hazell: So, these famous econometricians and macroeconomists, Stock and Watson— they observed that the history of the Phillips curve was one of apparently stable relationships falling apart upon publication. And so, the question is, was the flat Phillips curve one of these stable relationships falling apart, or was it something that still worked after the pandemic? And to me, this is really important, because if we have this Phillips curve that we really believe and it just falls apart on the next data point, [then] this is a real problem for macroeconomics writ large. Of course, every time a new data point comes in, we can always tweak the theory to fit it.

Hazell: But if we can't, ex ante, come up with models that, ex post, work well, then what's the value of models at all? The Phillips curve and maybe other such relationships would be of questionable value. So, that's the background. Then, if the Phillips curve doesn't work well, if the flat Phillips curve doesn't fit, perhaps we would need to come up with different shocks or explanations. So, there are many people saying, perhaps prematurely— I think prematurely, as I'll explain— there are many people saying, "Look, the flat Phillips curve doesn't work because inflation was very high in the 2020s, so we need to can it and come up with different models."

Hazell: So, one would be like a nonlinear Phillips curve, the Phillips curve becoming much steeper in the 2020s. One would be like some new measure of slack, for instance, maybe something involving vacancies. Another would be introducing new shocks, for instance, bottlenecks, or maybe new theories altogether. So, for instance, Leeper, Cochrane, and others— they've been advancing the fiscal theory of the price level. So, many people are saying, "Look, the flat Phillips curve doesn't work because inflation has been very high, so we need these new theories." Now, it turns out, I think— and I'll try and explain this a bit more— It turns out that the flat Phillips curve of pre-2020 actually fits the post-2020 period pretty well and out of sample.

Hazell: So, here, I'm drawing on this great paper by Beaudry, Hou, and Portier— by Sev Hou, Franck Poitier, and Paul Beaudry that was presented at the NBER [Macroeconomics Annual]. They show, basically, that the flat Phillips curve from before 2020 does a pretty good job of fitting the inflation data after 2020. Now, I want to stress that this is quite surprising. You estimate this pretty simple relationship. You estimate it before 2020. You have this giant move in all sorts of variables. And what they find is that that Phillips curve estimated before does a pretty good job after. Why is that the case? Basically, what happens just at the end of 2020 [and] at the start of 2021— Inflation expectations go up by a lot, and through this Phillips curve mechanism, actual inflation is going to go up by quite a lot, too. So, that's what they find.

Hazell: And that means that, potentially, the flat Phillips curve is not such a crazy thing to think about, because what's going on here is that inflation expectations are rising a lot and that's increasing inflation, too. So, for now, that's my tentative view, is that, actually, the flat Phillips curve does a fairly good job over this recent period. But what's really interesting is that inflation expectations are moving around a lot over this recent period, and that motivates, I think, just understanding why inflation expectations move around so much. So, you mentioned this really interesting work on inflation attention and thresholds and so on, and maybe that's part of the story. Maybe it's the case that once inflation just inches over 4%, suddenly it's going to be really volatile, and that could be one explanation for what's going on.

Beckworth: Yes, that's very interesting, and this is very consistent with your earlier work, your well-known co-authored work, where you argue that Paul Volcker's disinflation, really, was largely tied to the Fed and Volcker re-anchoring inflation expectations.

Hazell: That's right. There's one caveat. I think that this is why our work doesn't age as badly as I might have worried, but it doesn't age perfectly either. And so, the caveat is the following. What matters in terms of the story that I just told you, in terms of the flat Phillips curve doing well, is that one period ahead, inflation expectations rose a lot. Over the next year, two years, three years, household inflation expectations were rising a lot. That can explain the rise in actual inflation. What we previously emphasized in our earlier work was long-term inflation expectations, like 5 or 10-year inflation expectations. Those have actually been quite stable in the recent period. So, long-run inflation expectations have been flat.

Hazell: And so, I think, because we'd emphasized those long-term inflation expectations, I, for one— it took me a little while to realize exactly how to think about this. But what you had over the recent period is that long-term inflation expectations were flat, shorter-term inflation expectations rose a lot, and then still, via this Phillips curve mechanism, because inflation expectations rose, current inflation rose, too. So, my takeaway, which is a slight caveat to our earlier paper, is that when one is thinking about the Phillips curve and the flat Phillips curve, you do need to be looking at a range of inflation expectations measures— the long-run, the short-run, [and] the medium-run. And that's something that I sort of appreciated more after the current episode than beforehand from my earlier work.

Beckworth: Okay, so it is useful to look at a broad spectrum of measures of inflation expectations.

Hazell: Absolutely.

Beckworth: So, I guess, going forward then, the Fed should be mindful of household expectations as well as the consensus, the bond markets, and other things.

Hazell: Exactly. So, I think that if you're the Fed, one lesson from this is that you want to be monitoring expectations a lot, as they do, and a whole range of expectations— So, long-run, short-run, household, firm, the whole gamut.

Beckworth: And I think that you would say that even if the Fed was late to the game in terms of tightening, the fact that it was so aggressive, [that] it front-loaded these big interest rate hikes, was just reconfirming or maybe re-anchoring— to the extent that the short-run expectations were going up, it did what it had to do to say, "Hey, we're still in this game. We're still serious about inflation expectations." And so, they did the right thing.

Hazell: I think so. I think that's right.

Beckworth: Well, let's build upon this discussion about the Phillips curve and the explanations for inflation and go to a paper that you've co-authored with your grad student, and the title of the paper is *Do Deficits Cause Inflation? A High Frequency Narrative Approach.* Tell us about that paper.

*Do Deficits Cause Inflation? A High Frequency Narrative Approach*

Hazell: Great, thanks, David. So, this is with Stephan Hobler, who's just a great PhD student at LSE. He'll be going on the academic job market, not this year, but next year, and I expect great things for him. He's been a fantastic co-author. Okay, so we've been speaking about the Phillips curve before, and that's like an intermediate relationship between slack and inflation. But we can also ask the question in maybe a more primitive way, which is to say, there were a bunch of shocks and then inflation was the outcome. So, which shocks really seem to matter? Then, maybe it's going to go via the Phillips curve, but we can also just ask the more fundamental question.

Hazell: And one version of this question, one classic version of this question, is, did deficits cause inflation? This is a classic question. Macroeconomists have been asking this for decades or centuries, but of course, it looms large because recently, inflation has been very high, like we've been talking about, but, also, deficits have been high. That's true around the world. It's particularly true in the US, where in December 2020 and March 2021, deficits were very large, like 13% of GDP. This is associated with the start of the Biden Administration. Then, inflation rose a lot afterwards. So, we want to ask, did deficits cause inflation? And there's a big empirical debate about this question.

Hazell: Some people think that deficits were the primary cause, one of the most important causes. Other people are suggesting that other factors matter, too— supply constraints, bottlenecks, commodity prices, and so on. Fundamentally, it's going to be hard to tell these stories apart. Why? Because the other factors— they're acting as omitted variables. They're making it hard to single out the effect of only deficits on inflation, because how do you single out the effect of deficits from the other shocks, from the bottlenecks, and so on? And of course, it's difficult, because it's just one episode. It's one episode. Many shocks are hitting at the same time. How can we disentangle these different shocks?

Hazell: But our view is that it's essential to understand the cause of an influential episode like the post-pandemic inflation, because, ultimately, it's these key episodes that really forge how we think about macro models. To give you an analogy, the Great Depression and the 1960s to '80s inflations— these were pivotal events that really changed how we thought about the prevailing model. And perhaps the post-pandemic inflation might be just as influential. It's this big inflation, it's associated with big deficits, and so on. And so, we really need to figure out what's going on to understand if we're on the right track with the current way that we think about macro or not.

Hazell: So, what we're going to try and do is estimate the causal effect of deficits on inflation using what we're calling the high-frequency narrative approach. And this narrative approach— it basically has two parts. I'll just say what these two parts are very quickly and then fill out the details. The first part is what we call the narrative bit. This is following the method of Friedman and Schwartz as the classic reference, which is identifying a key event which released lots of news about deficits. Then, the second is a slightly more modern high frequency approach, which is to look at the response of inflation forecasts from asset prices. And as I'll say more, the hope is that by looking at the response of high frequency forecasts from asset prices, one can separate out the effect of the deficits from the other shocks. So, that's the high level. If it's okay with you, let me get into some of the details.

Beckworth: Please do, yes.

Hazell: Excellent. Okay. So, the narrative approach— The narrative idea is to identify an event that released news about deficits, and then calculate the size of the associated shock. We're going to focus on a specific event. That event was the Georgia Senate election runoffs of early 2021. So, I'll remind you, probably, you remember it somewhat, but maybe not fully. It took me a while to get all of the details of this together. So, let me give you the potted history of what happened. In November 2020, Biden won the presidency, the Democrats won the presidency, they held 48 Senate seats. Now, in early January, both Georgia Senate seats were to be decided by runoff elections. So, the implication for these elections was fiscal policy.

Hazell: So, if Democrats won both seats, they'd have a majority of the Senate for fiscal stimulus. However, if they didn't win both seats, if they lost at one seat, then Republicans would be able to block Democrat legislation. On the Senate procedure, if Democrats are the majority, they'd be able to pass fiscal legislation, but they wouldn't be able to pass non-fiscal legislation. What you need for non-fiscal legislation is 60 seats. That's out of reach either way. In the event, Democrats won by January 7th. That was known. Afterwards, in March, they passed this very big stimulus, the American Rescue Plan. That's about $1.9 trillion worth of deficits or 8.8% of GDP that added to another very large bill that was passed in December 2020, so just before Democrats won in Georgia. That's the key event.

Hazell: We're then going to measure the size of news about deficits due to the Democrat victory, due to this event. The key challenge here is figuring out how much deficit spending was expected on the eve of the election. Because what you need to know— you need to know the amount of news that's released to markets from the deficit spending. And for that, you need to know not only how much markets thought Democrats would spend, but also how much markets thought Republicans would spend, and what the chances are of Democrats versus Republicans winning. It turns out that investment banks write lots of very detailed reports discussing exactly these outcomes.

Hazell: So, what we do is we collect, from about 20 investment banks, a bunch of detailed narrative information that can basically tell us how big was the shock to deficits after Georgia, how much did markets update towards big deficits from just before versus just after Democrats winning in Georgia. And I should say that this is a pretty painful process, because you got to ring up or email 20 investment bank chief economists and get access to all of their portals. You've got to read 100 reports for each portal. But when you're done with that, you have this amazing resource.

Beckworth: But Joe, you're good at this. You did this with that famous paper of yours about the Volcker inflation. So, this is well-traveled area for you.

Hazell: Well-traveled. So, at least there's elbow grease, there's lots of reading. At least these are things that I can do. If nothing else, your audience can think of me as someone who can put in the hours. But this is actually with the help of some amazing predoctoral researchers as well. I'm really grateful to them. So, we've got the reports. Then, with that, we can size that news from the Georgia runoff. So, we find that the typical investment bank— they thought that Democrats were going to win with 50% probability. They would spend about $900 billion worth of stimulus if they won. That means that there's a shock to news about deficits of about $450 billion. So, it's a 50% chance of winning going to 100% chance of winning, multiplied by, if Democrats win, how much they're going to spend, $900 billion. So, that's the main shock.

Hazell: And I should say, the other really useful thing that we get from the reports is that we read, "What is the main consequence of Democrats' victory?" It might be the stimulus, it might be something else. You might think, "Well, Democrats are going to do all manner of things," but what markets were really interested in, what they really expected was going to happen after a Democrat victory was the stimulus, which is why it's sort of the main event that we're studying. This is the first part, the narrative part. So, we have a measure of the shock from the narrative. Then, the second step is a high frequency step.

Hazell: So, we're looking at inflation forecasts from financial markets to figure out what the market thought would happen to inflation. So, the second step is looking at inflation forecasts. What we estimate is that after Democrats won, markets expected that prices, over two years— the two-year inflation rate— would increase by 38 basis points. I don't yet know, I'll tell you, but so far, you don't know if that's big or if it's small. So, markets today— they have the shock from deficits, and then they think that inflation is going to rise after the deficits. Then, the question is, the mix of the shock and the response— are these big numbers?

Hazell: I should say, at the same time, markets thought that there would be a big increase in real GDP growth. Investment banks also thought that there'd be this big increase in real GDP growth. So, it looks very much like this demand shock that raises inflation and raises real output. I'll say one thing— at least some of your listeners probably thought, "Hang on a second. January, early January— wasn't there this other major political event that took place, which was the January 6th Capitol Hill riots?" So, indeed, that was a real problem for our strategy, but we do a bunch in the paper, and we're very confident that the January 6th riots isn't confounding things.

Hazell: So, that's what we've got. We've got two things. We've got the narrative shock. We've got the high frequency response. We're going to combine these to come up with a back-of-the-envelope measure of how much the overall fiscal deficits— not just from this narrow window, but everything, the whole 13% of GDP— how much inflation that caused. And so, there are two steps to take to go from these numbers that I gave you, the response and the shock, to the overall effect of deficits. The first one is the premise of the paper, which is that changes in inflation forecasts from swaps are an unbiased measure of what actually happens to inflation. This is the premise of the paper, and in some ways, it's the— I don't want to say the weakest part, but it's like the leap of faith.

Hazell: We have this effect on inflation expectations from the market, and we want to say, "Look, let's extrapolate that towards actual inflation." And there's a there's an advantage and a disadvantage to doing so. The disadvantage, of course, is that you don't fully know if inflation forecasts actually accurately measure the world. But there's a real advantage, which is that the response of inflation forecasts in this narrow window— they're plausibly not contaminated by other shocks affecting the economy. So, overall, there's lots of other shocks affecting the economy— oil prices, bottlenecks, and so on— these might be affecting inflation. But the change in inflation forecasts in this narrow window around the Georgia election— they're really only being affected by deficits. So, that's why we think that it's a worthwhile exercise.

Hazell: So, that's the first bit. We're going from inflation forecasts to actual inflation, with caveats abounding. Then, the second thing that we're going to do is we're going to combine the numbers I just told you into what we call an inflation multiplier. So, the idea here is that with a shock worth about 2% of GDP, we've got the response of inflation from asset prices, and we're going to apply the implied effect of deficits on inflation from this nice eventsstudy. We're going to apply that marginal effect— what we call an inflation multiplier— we're going to apply that inflation multiplier to the whole 13% of deficits, because this 13% of deficit shock, if it has the same effect as it did around the Georgia election, [then] we can figure out the whole effect of deficits.

Hazell: So, this is the final exercise, the final quantitative number, which is, if we take this deficit shock of 13% of GDP, we multiply it by, we think, [is a] cleverly identified inflation multiplier, [then] what do you get? You get about 30% of the post-pandemic inflation as explained by these deficits of December 2020 and March 2021. So, that's a big number. It's also not all of it. It suggests that deficits were an important cause of inflation, but they weren't the only cause. I'll say one other caveat. So, here, I'm talking about the deficits of late 2020 and early 2021, the Biden deficits. There was another very big deficit finance stimulus, in March 2020, that was called the CARES Act. If you add the CARES Act to these other stimuluses, the total effect on inflation is more like 50% of the total between 2021 and 2022. So, you're getting numbers, again, that are suggesting that some large share, but definitely not all of the inflation in the early 2020s, was caused by deficits.

Beckworth: That is so fascinating and very interesting, the magnitude. I want to come back to that in a minute when we wrap this up in terms of the big takeaways. But I think what's really neat about your paper is that it deals with this eternal question that haunts macroeconomists, and that's identification. How do we get this exogenous event? How do we identify— and you've worked really hard to do that. And the Georgia Senate is a natural experiment for macroeconomists, and you jumped on that, and you utilized it. So, kudos to you and your grad student. The other observation I have though, Joe, is that you are mapping the deficits to inflation, which makes complete sense to me, but someone like John Cochrane or Eric Leeper might say, "Aha, you've just proven the fiscal theory of the price level," or maybe you want to tell the story through a Phillips curve. So, how would you navigate those two perspectives?

Navigating the Alternative Macroeconomic Perspectives

Hazell: So, actually, there's one more thing in the paper. Thank you. Thank you for that. So, the question is, what's the right model to think about this? I think that there are two models that, to my mind, are in play. One is the fiscal theory of the price level. A second is what I'll call the standard HANK model. So, what's the standard HANK model? It's the standard heterogeneous agent New Keynesian model. This is a model that has a Phillips curve. Then, it has a consumer who, when you give the consumer a bunch of transfers, is going to spend a bunch. So, the jargon— a consumer who has a departure from Ricardian equivalence.

Hazell: So, the second model, the standard HANK model— that's the model that, I think, is probably the consensus of the field. But the first model, the fiscal theory of the price level, is also a very interesting model, [and is] also a model in which deficits lead to inflation. Now, my top-line takeaway is that I think that both models are still in play after what we find. So, it turns out that both models, the FTPL and the standard HANK model, can fit the data pretty well, can fit the response of deficits to inflation pretty well. And so, we're not going to resolve this question of, “Was Cochrane right versus modern HANK?” The paper isn't resolving that question, but it's still making progress, because you would want to know if this thing that we'd estimated had wiped out one of the models and favored one of the others— you'd want to know that too. And so, that's the final part of the paper.

Hazell: We're going to set up and calibrate two models. The main one is the standard HANK model. The other model that we have in the appendix of the paper is the FTPL. And we do the best to match the salient details of this Georgia shock. When we do so, both models could basically fit the data. So, unfortunately, at some point, maybe we wanted some splashy takeaway of, “FTPL is right. The mainstream is wrong,” or vice versa. That's not what we find. But nevertheless, it suggests that, to some extent, both models could be along the right track, and that's still a pretty valuable takeaway.

Hazell: One thing I will add, and this maybe goes back to your roots, David, is we find that monetary policy is pretty important for the effect of deficits on inflation. In particular, what we find is that around this deficit shock, short-term interest rates don't respond. So, markets believe that the Fed is not going to offset this fiscal multiplier. Why do we care about that? We know, from your blogs back in the day and other work besides, that when we think about fiscal stimulus, we're also thinking about the monetary response. We know that if the Fed hikes enough, they can eliminate fiscal stimulus. If they accommodate, then fiscal stimulus is going to have a big effect. In the data, and also interpreted through our model, the Fed was fairly accommodative, and that's one of the reasons why inflation was relatively high.

Beckworth: Yes, for sure. Both hands of macro policy were involved in the inflation surge to the extent that macro aggregate demand policies did contribute. So, I imagine that each camp will claim victory here, the HANK modelers. Ben Moll, your colleague, will say, "Good job, Joe." Then, John Cochrane will say, "Good job, Joe." So, I will say this, though. One of the challenges of the fiscal theory of the price level is being able to empirically show or to quantify, because there's this really tough thing to measure called the net present value of primary surpluses. We don't know that, so they might really like this. So, they’re like, "Man, Joe, you finally found a way to get a hold of a quantitative measure."

Hazell: No, exactly. So, David, I'm a data-first person. I'm not ideological. So, I think that there's plenty of great stuff to behold in both theories. And I think, precisely, one of the challenges with the fiscal theory of the price level, [is that it is] very hard to know what people think is going to happen to deficits in the future. What we have here is that we have a really sharp measure of what people think is going to happen to deficits. The fact that the FTPL does a pretty good job is, I think, nice, and it means that there's some coherence between the FTPL and the data.

Beckworth: Okay, well, Joe, in the time we have left, I want to wrap this up by painting a big picture here. What have we learned in terms of macroeconomic policy? So, today, the Federal Reserve cut interest rates. Just a month ago, Jay Powell gave a talk at Jackson Hole. And in this speech, it was almost a mission-accomplished tone to it, maybe a more modest version of that. But he acknowledged that there were both supply side factors, disturbances caused by the pandemic, as well as demand side. And he said, "Our tightening played a role." And depending on who you are, you read your story into his speech.

Beckworth: So, I want to invoke Paul Krugman again. He said, "Ah, Jay Powell's told us that it was mostly supply side. Transitory was right. We win. Team transitory wins after all," but then if you go and look [at] some of the authors he cites for the demand side, they argue for a very strong demand story, and you just mentioned that you could arguably attribute at least half of this inflation surge to excess aggregate demand pressures. So, how can we land this plane? Let's get a soft landing in terms of the debate. Where should we go in terms of the role caused by supply disturbances versus demand shocks?

Evaluating the Supply vs Demand Story for the Post-Pandemic Inflation

Hazell: So, David, I think you're setting me up for failure, which is that, in the closing minutes, I'm just going to solve the debate between Krugman, Summers, Blanchard, Powell, Waller, and everyone else, but I'll do my best. So, let me tell you what I learned from our paper about this episode. And again, the headline seems to be that deficits were important, but not the sole cause— maybe 30%, maybe 50%, something like that. Also, to be clear, that's because of deficits and also the contribution of monetary policy allowing these deficits. So, what I learned from this is that policymakers, I think, probably stimulated slightly too much, but were very unlucky. What do I mean by that?

Hazell: Imagine that there had been no other shocks except this deficit shock and its monetary accommodation. Inflation probably would have maybe topped out at 4% or 5% in 2022. That's what the numbers from our paper say. That's slightly high, but it's not that high. At the same time, and this is also something that we find in this paper, there's a tremendous amount of real GDP growth. Real GDP growth is great. The US had a roaring recovery, much better than the countries that didn't do fiscal stimulus. And we find, in our paper, that you can causally attribute a lot of that to the big deficits.

Hazell: So, absent these other shocks, if there's only the deficit shock, [then] you have maybe a slight over-stimulus, [and] inflation is a bit too high, but you get excellent real GDP growth, you get a great recovery, and that's great. Alternatively, if you hadn't done this big stimulus— maybe you're something like the UK, maybe you're Europe— [then] you're in the doldrums, and so on. On the other hand, it turns out, I think, that the Fed, the Biden administration, were slightly unlucky because, at the same time as they did the stimulus, which was maybe a little too large, but not grossly too large, there were other very inflationary shocks at the same time. So, the Russia-Ukraine war at the start of 2022, bottlenecks in early 2021, and so on— these happened at the same time.

Hazell: And so, I think, in the end, the Fed and government policymakers did a pretty good job— lots of GDP growth, maybe slightly too much inflation, combined with bad shocks, which is to say, at the same time, that there's this other big inflationary shock. So, I guess my closing thought is a quote from Napoleon. So, I think, Napoleon, he says something like— and I'm paraphrasing, "The best thing for a general to be is lucky." Obviously, it's a joke, because how could generals be lucky? And I think that the best thing for a policymaker to be is lucky. And here, the policymakers weren't that lucky, because at the same time as doing what arguably seemed like a pretty sensible policy, they were very unlucky, because these other big inflationary shocks happened at the same time. Then, the Fed had to figure out how to deal with those as well. A little difficult to know exactly what to do, but the fact that inflation came down pretty quickly afterwards suggests that the Fed managed to navigate that pretty well. So, that's my sense. It's pretty even-handed, pretty milquetoast, pretty moderate, but hopefully you agree.

Beckworth: No, I like that, and I appreciate your humility here, Joe. And I think it's something that we should be cognizant of and be grateful for. We should take the win, right?

Hazell: Right, right.

Beckworth: And when you look at the US economy relative to other economies, we are doing relatively well. So, we need to take the win, because what is the true counterfactual? We don't know, but it could be a whole lot worse. We could have had a financial crisis. Imagine no CARES Act, for example. It could have been a catastrophe. So, let's take the win, but let's also learn the lessons, and that's what you have showed us today. How can we thread the needle so that next time we have a more precise support without overheating? And I think that's what you guys are doing, so, great work, and I appreciate you coming on the show again today. Our guest has been Joe Hazell. Joe, I appreciate your time.

Hazell: Thank you so much, David. It was a ton of fun.