BONUS: Noah Smith on the State of Macroeconomics

Establishing more credible and ubiquitous microfoundations is one important way to improve macroeconomic modeling in the present and future.

Noah Smith is a former columnist for Bloomberg and is now a popular writer at his own Noahpinion Substack. In this bonus segment from the previous conversation, Noah rejoins the podcast to talk about the nuts and bolts of macroeconomic modeling. Specifically, David and Noah discuss why macroeconomics is still in its infancy, how we can improve macro modeling moving forward, how to spot “nutty” macroeconomic theories, and more.

Read the full episode transcript:

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

David Beckworth: All right, Macro Musings listeners, we're back. Noah Smith has been gracious enough to give us a bonus segment this week. Noah, thank you for sticking around for an extra segment.

Noah Smith: Hey, thanks for having me back.

Beckworth: Yes, it's great to have you on. Now, I wanted to get on macroeconomics with you in the main program, but we ran out of time. But you, again, have graciously agreed to stay on and you've written several posts recently that were very interesting and this is where I cut my teeth, is in macroeconomics. You've written a lot on this over the years. Your original blog, you wrote about macro. Of course back then in that time that was the thing, to write about macro and why we have the slow recovery. But you had an article titled, *Macroeconomics is Still in Its Infancy.* And you spend a lot of time on Ed Prescott, real business cycle models, DSGE models. What was the argument you were making in that piece?

*Macroeconomics is Still in Its Infancy*

Smith: Well, I think we had this big explosion of macro theory in the eighties. Everything changed. We went from these pretty simple models to what we now call DSGE models, which the first DSGE model was Prescott and Kydland's original classic RBC model. That was a big methodological revolution. I think it became pretty clear early on that the actual models that Prescott was making were not very good, in terms of their ability to fit whatever small, scant data we had, and in terms of just making sense. They didn't make any sense and they didn't work very well.

I think it became pretty clear early on that the actual models that Prescott was making were not very good, in terms of their ability to fit whatever small, scant data we had, and in terms of just making sense. They didn't make any sense and they didn't work very well.

Smith: And the sort of quasi empirical justifications that Prescott and company had put out for them, “oh look, we can replicate a time series that sort of has the same volatility as in real time,” ... it was not convincing. It wouldn't convince any serious econometrician. You had to hold your nose to think that this justified these models in any way. I think that Prescott got a Nobel not because he got things right about how the economy worked. Recessions aren't due to engineers forgetting how to do things. Well, I do think engineers sometimes forget how to do things, but I don't think they forget on that short a time scale.

Smith: It's not due to that. Really, that just didn't make any sense. The reason Ed Prescott won a Nobel was because that methodology, that mathematical methodology seems to satisfy people's desire to address these theoretical critiques, the Lucas critique and whatever and whatnot, in ways that made them enduringly popular. After that, people kept making DSGE models even after people decided RBC wasn't the right model for the economy. They made new-Keynesian models and other types of models. You even had some models in the RBC tradition like [inaudible] models or whatever Angeletos is working on, the multi-producer, multi-sector models, or whatever. You had lots of RBC inspired models, but mostly you had new-Keynesian models. That's what took over.

Beckworth: That's the workhorse of macro.

Smith: Right. That's like a potted history of macro in the last few decades. New-Keynesian models, of course, had to receive a major update after the great recession, financial crisis. Basically, people added the financial sector and added a zero lower bound. We know that conventional monetary policy only goes so far. There's reasons to think why unconventional monetary policies are in some sense weaker. Mike Woodford did a lot on this. We're like, "Okay, so we need to do some fiscal stimulus when we hit the zero lower bound. That became a big deal. Then I know you've written massive amounts about this, so of course, all your readers and listeners are familiar with this stuff. The other thing was obviously, finance can crash the economy. Whoa, who knew? Who could have predicted?

Smith: So based on the Great Depression or Japan or Sweden or any of these things, it was obvious finance can crash the real economy, but only recently have people started trying to shoehorn that into the new-Keynesian models that had prevailed. Those are the two big things that people updated. But still people basically used DSGE models and basically used new Keynesian models because new Keynesian models are all about the role of monetary policy. That became very influential at central banks.

Smith: I still think it's tempting to see this as a great march of progress, but I think that I'm much more of a pessimist about this. First of all, because the whole idea of using DSGE models in the first place was that these models were supposed to provide credible micro foundations for macro behavior. And the micro foundations of DSGE, of new Keynesian and DSGE models, and the micro foundations of the original RBC models often don't generally make a lot of sense. When we start looking at these micro foundations in the micro data it doesn't seem to fit. So for example, Euler equations, the consumption Euler equation. Basically the idea that if interest rates are high you'll, I don't know, save more money, whatever. The idea that basically consumption decisions are made-

Beckworth: Based intertemporally.

Smith: Based on interest rates intertemporally.

Beckworth: Yeah.

Smith: When you look at the actual data on this, it looks like it's the exact opposite of what Euler equations would predict. It's not just that you can't find a signal, it's not just a weak signal, it's the exact opposite. It flips the sign. And Eichenbaum, who's one of the main new Keynesian DSGE modelers was the guy who told me about this. He said it using profanity that I will not repeat on this podcast. That is how he described the Euler equation that is at the core of his own models that have made his career. That is scientific honesty, I think. And it is a scientific honesty that many sort of shallower boosters of the research literature lack. So, my idea comes from Eichenbaum, that macro is in its infancy.

Beckworth: Okay. And he’s a big name.

Smith: He's a big name.

Beckworth: Yeah, he's a huge name. But let me ask this question then. You're talking about the Euler equation and the workhorse standard new-Keynesian model.

Smith: Right. Calvo pricing.

Beckworth: Yeah, sticky prices and all that-

Smith: Do we really think that macroeconomic fluctuations are determined by a fairy that tells you when you're allowed to change your price? No.

Beckworth: No. It's an approximation. It's a shortcut. But let me ask this question about this.

Smith: But not a structural one?

Beckworth: Right. That's fair.

Smith: A structural shortcut would describe how people change their prices in response to actual events. And people have tried to make those models.

Beckworth: Fair. There has been endogenous price setting behavior.

Smith: Oh, tons. It's just really hard.

Beckworth: It's very hard. I guess that's two comments I want to make. Number one, I think one contribution that Prescott did make is just really emphasizing general equilibrium. You note this in your article. That's an important consideration when you're thinking, what's the general equilibrium effect? You got to consider all the pieces moving together, not this sort of ceteris paribus isolated argument.

Beckworth: The other thing though is, one could argue there has been some progress. The models you mentioned, the representative agent new-Keynesian model, you have that one Euler equation, but you know the work on the new heterogeneous agent new-Keynesian, so you have different types of households. You have poor hand of mouth, even rich, someone who gets a lot of money, but they just live paycheck to paycheck. When you get these different types of households, you might find Euler equations that make more sense.

Beckworth: I don't know. Do you have any thoughts on the HANK models, the Heterogeneous Agent New-Keynesian models? And for those who don't know, I'm sure most of you do, but DSGE is dynamic stochastic general equilibrium models. But do you find those a productive endeavor? Are we making progress using HANK models?

Are We Making Progress Using HANK Models?

Smith: Well, I mean, it's a good thing to try out. I had to do heterogeneous agent models for my class. I know that there's a lot of mystery meat in the way you get those things to converge. There's no real proof of existence of solutions for a Krusell-Smith kind of model and that's the most simple possible heterogeneous agent RBC model that you can make. Now you bring in imperfect competition and blah, blah, blah. It just increases the parameter space. The answer is that these models maybe are a good exercise, but you can't validate these things with data.

Smith: And to validate these things with data, you've got to really push the boundaries of what's credible mathematically. A lot of times you have mystery meat code that ends up converging to something and saying, "Look at this result we get." And I don't have confidence in that result because I know how you change one little assumption about how approximate aggregation works in this value function iteration kind of thing. You can also solve them with finite element method. I'm sure people have actually improved the solution methods a lot in the last decade. I'm sure the solution methods have improved, although I don't think they've gotten an actual existence proof for these things. But these technical details all come down to the fact that we don't have a lot of macroeconomic data.

The answer is that these models maybe are a good exercise, but you can't validate these things with data. And to validate these things with data, you've got to really push the boundaries of what's credible mathematically.

Smith: We have only a few decades of macroeconomic data. The series are all correlated with each other, and the international series are correlated too, and everything's serially correlated. So you've gotten massive, massive, massive amounts of correlation in a sparse data set. How are you going to validate these models? I'm pessimistic about the top-down DSGE project. And I think that people need to actually think back to the Lucas critique and think, if you have the wrong microfoundations, do you have any confidence that putting these things together in a big tinker toy model, will give you the right answers when you're starting with the small, wrong answers? Will small wrong answers add up to a big right answer? And Lucas critique went to just, no.

Smith: If you don't know how microagents work, if you don't know how companies decide when to invest, if you don't know how and when consumers decide to consume, and all these things like this. If you don't know these things, if you don't have a very good idea of how those microprocesses work, can you trust a macro model that seems to fit aggregate data, or at least the small amount of aggregate data we have, should you trust that the results of the macro model will be structural, will be robust to policy changes? And the Lucas critique answer is no. And I feel like we've moved away from the Lucas critique answer of, well no, then you shouldn't trust it, to the Milton Friedman pool player analogy of who cares what goes into the salad as long as it tastes like salad. The mystery meat is, okay, kind of thing.

Smith: Or like, yeah, who cares what goes into the Long Island iced tea, as long as it tastes like an iced tea. Maybe you should care. But then, so that pool player analogy, which for the listeners, the analogy is a pool player doesn't know the physics of how they hit the pool ball, they just do it, and it comes out right. And that's sort of an expression for as long as we fit the macro data, we don't need to fit the micro data. But the Lucas critique says the exact opposite. It says that we need credible microfoundations or else our macro results may be good at curve fitting on past data series, but they won't be structural, they won't be predictive. And so I think that macro economists need to revisit that idea, to revisit Lucas, and to aim lower, and many are doing this, I think young macro economists are mostly doing this.

Beckworth: So what would be an example of that? What do you want them to do, like aim lower in what way?

Smith: Try to figure out how the pieces of the economy work. So there's been a lot of work on consumption in response to plausibly exogenous policy shifts. So for example, you have unemployment insurance gets extended, and then extended UE gets cut off. How do people respond to that, and how do heterogeneous classes of people respond to that? That was some work during the great recession that I thought, this is great, I'm really glad that people are doing this. Or you have, Klenow’s paper on how different classes of companies responded to the Great Recession, how they changed unemployment differently, depending on things. And so this can help you build up search-theoretic models of the labor market. And so it's just these small bore things that don't satisfy the Ed Prescott-ian desire for grand hand-waving solutions, theories of everything.

Smith: But they're more of what I call normal science, they're building up small pieces of knowledge that someday we will aggregate. And this is what I say when I mean macroeconomics is in its infancy. And I think this is a thing that most, not all, but most young macro economists would agree with me these days about, especially because I... Before COVID at least, I would talk to them a lot at conferences and whatever, and people would tell me they agreed with this. And when they showed me what they're working on, it's very humble stuff, just designed to glean facts about consumption and unemployment, and investment and all these things. I think investment's the most understudied category by the way. Knowing what makes companies invest is insanely understudied, because we have such good data on corporations, and so many possibly exogenous government policy shifts, and so little understanding of what drives investment.

Smith: Anyway, that's a tangent. But I think using the techniques of empirical economics to understand the micro-behaviors that add up, including finance, including bubbles, including bank collapses, including debt. Are there really balance sheet recessions? That question depends on whether or not people have these sort of mental modes of deleveraging that they shift into, that's Richard Koo's hypothesis. And so, you can test that, you don't have to just throw that assumption into a big meat grinder, and see if you match the macro data. You can test the micro assumptions, you can go look at how policy changes that have an exogenous cut-off, where they cut-off the debt… they forgive this person's debt, they don't forgive this other similar person's debt. You see how they change things, you can see debt forgiveness or... You can see these things, you can observe these things.

The Lucas critique says...that we need credible microfoundations or else our macro results may be good at curve fitting on past data series, but they won't be structural, they won't be predictive. And so I think that macro economists need to revisit that idea, to revisit Lucas, and to aim lower.

Beckworth: So let me try some pushback, I think this is what a macro economist would say. They would agree, yeah, we do need to look at the building pieces, better understand those parts much more thoroughly, investment, consumption, things like that. But, the danger is if you look, and you're too narrow in your focus, you may overlook the fact that there's still general equilibrium effects going on. And you might find a really robust finding in certain types of businesses. But how do they respond in the midst of a recession, a global slowdown? The empirical work is really subject to this general equilibrium critique. You can't just assume up that everything's going to approximate and be the same.

How Important Are Micro Considerations in Macro Models?

Smith: Well, no. So there's a couple of responses to that. So first of all, when you do the micro studies showing how people respond to say, the end of extended unemployment insurance, the consumption response to that, right? You do a micro study about that, you know that everybody's in the same macroeconomic situation of the Great Recession, right? So, you haven't gotten a final answer to the question of how people behave in response to this kind of thing. You've only got a conditional answer to how people respond individually, in the context of a larger general equilibrium environment. Right?

Smith: And so you're not done, it's just a start. So then you have to look at other macroeconomic environments, you have to look at the Great Moderation era, and other policies that did things. So basically, you have to build up an understanding of consumption behavior by looking at various sort of regimes of recessions and booms and stability and inflation and deflation and all these things, to get a more plausible picture of the microeconomics of consumption. And then, when you take those pieces and once you're really confident of those pieces, you can start adding them in to DSGE type models. And, they may look ugly, a consumption Euler equation looks really nice.

Beckworth: Elegant. Yeah.

Smith: Yes, it's super elegant. But, I think we've passed well beyond the point where we expect elegance from DSGE models at all. Like, HANK models are just giant tinker toy nightmares. And we're okay with that, we accept that. And so the point is that, yes, it's going to look nightmarish when we get all these realistic models of consumption and investment and all these things. Once we get those realistic micro foundations, not Prescott sprinkles fairy dust on his favorite fable micro foundations, but actual real micro foundations that are probably going to have behavioral effects. They're going to be complex, they're going to be situational, they're going to have a lot of hard to estimate things. And when we get those things and we start building back up again, it's going to look ugly as heck, right? It's going to look really ugly. But that's just the way it is, because that's the complexity of the system we're dealing with.

Smith: And, if we're going to pretend that that super complex system is actually a super simple system, you might as well do it with just a couple lines on a graph, rather than a Prescott model. As long as we're pretending things are simple, we might as well use lines on a graph. And you do this all the time, right? You prefer lines on a graph to whipping out the old DSGE model and calibrating it, for a very good reason, which is that as long as we're in the realm of simplistic, heuristic, directionally correct kind of stuff, then we might as well use models where the inputs and outputs are very, very simple, rather than hard stuff.

Smith: And if we're going to make a model that we have more confidence in… a structural model, a real microfounded structural model with good microfoundations, that we have more confidence in than the couple of lines on a graph, it's going to be ugly as heck, and we shouldn't worry about that. And we should just try to get it right, and we might not be able to get it right for 50 or 80 years. But you know what? That's how science works. And there's still a lot of natural science we don't know. And it's hubris to assume that, because I can write down some equations for something, that I understand it.

Beckworth: Okay, so let's take Noah Smith and put him in charge of the Fed. Alright, you're the chair.

Smith: I'm the chair?

If we're going to pretend that that super complex system is actually a super simple system, you might as well do it with just a couple lines on a graph...You prefer lines on a graph to whipping out the old DSGE model and calibrating it, for a very good reason, which is that as long as we're in the realm of simplistic, heuristic, directionally correct kind of stuff, then we might as well use models where the inputs and outputs are very, very simple, rather than hard stuff.

Beckworth: What type of model would you use on a practical level? So, you don't have much confidence in a HANK model, some of the other DSGE models. So, what do you use? You're trying to feel your way through the economy. Let's say you're chair right now, trying to figure out how much of the inflation you should respond to. Some of it's supply driven, some of it's demand driven. What do you do? What framework can you rely upon to give you some meaningful advice?

Noah’s Ideal Policy Model for Current Macro Conditions

Smith: I have a very good answer to this. It comes from Tetlock, Tetlock's examination of forecasters and who wins at forecasting. Now, I know the Fed does a lot more than just forecast. They have to look at conditional policy responses and other kinds of things. They do much more than forecast. But when Tetlock looks at who succeeds and who fails, the people with the one big theory fail and the people who glean a lot of little bits of information from a lot of different paradigms, whom he calls foxes, though he didn't invent the term, he calls the single theory people hedgehogs. And I forget who actually invented this term, it's some older writer. But then he calls the eclectic people foxes, and the foxes always win the forecasting competition. And so I think that although Fed policy making isn't just forecasting, there's an element of forecasting, but there's lots of other elements too, I would say that they should be foxes. And in fact, this is exactly what they do. When you talk to Fed people, you quickly understand this is exactly what they're doing.

Smith: So, you should make the models that are a couple of lines on paper. You should have some nerds who know how to run some big DSGE models, including Smets-Wouters model, augmented with a financial sector and a zero lower bound. You should have those people who know how to run that model somewhere in your bureau. I'm not saying that's the greatest model in the world, but I'm saying that you should have that somewhere. You should have your little FRB/US, basically giant spreadsheet with unrealistic assumptions of correlations between things. And you should have your VARs, you should have your SVARs. Somewhere, you should have some people running some SVARs, and run several different SVARs. And then you should have your green book sort of business conditions, tell me this, reading the tea leaves sort of people. And you have all these different sources of thoughts and modeling information, and then you debate it out amongst yourselves and see whether the sources agree or whether they disagree, and try to understand why you might trust one over the other in a certain situation. And then think about it a lot and then use discretion.

Beckworth: Okay.

Smith: So, I'm sorry, rules versus discretion people. Again, Prescott, John Taylor, I'm sorry, you guys. Discretion versus rules is not up to our discretion. The rule is that you must always use discretion because... There's papers showing this, by the way, showing that because of lack of ability to signal credibility, you'll always be using discretion because people will just assume you are. And so there's papers about this. And then if all else fails, just ask Emi Nakamura what to do.

Beckworth: Okay, we will do that. Alright, Noah, we've gone through mainstream macro. We see there's plenty of opportunity for growth and improvement. So, we will take the challenge, Noah, and plot ahead and do our part to make this world a better place, advancing science one economist at a time. All right, well, let's turn to something that you've also written on, beyond mainstream macro. And this is an article titled, *Nutty Macroeconomic Theories Will Ruin Your Economy.* So, talk us through these other nutty macro theories.

Breaking Down “Nutty” Macroeconomic Theories

Smith: Right. So, the most famous nutty macroeconomic theory that everyone loves to pick on, justifiably, is called MMT, and it’s this sort of made-up thing that's not really a theory because it's a gang of people who are extremely aggressive online and extremely good at finding new journalists and politicians to listen to them for a little while before the new journalists and politicians realize, "Wait a second, I'm talking to crazy people." But often then it's too late, and you've written your glowing New York Times profile of the MMT people. And they're just absolutely horrible people if you meet them. They're just real pieces of work. But then their ideas are not just nutty but vague. So, they'll never write down their theory. There's no theory. And they'll never admit their theory ever having made a single wrong prediction, which is how there's no actual theory. Because if there's an actual theory, no theory ever gets everything 100% right. Sorry, standard model of physics, maybe. But no. No theory, we know, ever gets everything right. And so if you never get anything wrong, you don't actually have a theory.

Beckworth: So, it's an unfalsifiable theory.

They'll never admit their theory ever having made a single wrong prediction, which is how there's no actual theory. Because if there's an actual theory, no theory ever gets everything 100% right...And so if you never get anything wrong, you don't actually have a theory.

Smith: Yes, it's unfalsifiable because you can't write it down. It's like my theory is that [inaudible]. Now, take my recommendation of always borrowing more money. And that is the best summary of MMT I can give. And so MMT people will just tie themselves in knots. It's like a non-economist's imagination of what economists do, but with no actual relation to what actual economists do. Once in a while, they slip up and they say an actual substantive thing. And this is when you catch them. And so if you don't know how to recognize this BS, then you catch the BS when they sometimes slip up and say a very specific thing. So, there's an exchange between Stephanie Kelton, one of the main MMT proponents, and Jason Furman, former White House economic advisor under Obama. When Stephanie Kelton says, "Have you considered the possibility that raising interest rates might actually make inflation worse?" Or something like that, and Furman just replies, "No."

Beckworth: Yes, famous tweet.

Smith: Right. And so he just says, "No." And that's good because some macroeconomists have considered this and they're called the neo-Fisherians. So, back in the days of quantitative easing, people who didn't like quantitative easing were trying to say, "Oh, actually these low interest rates are going to bring us into deflation, and that's why we shouldn't like them." I don't think they ever really believed it. They were just toying around at the idea, possibly for political purposes, but also just for fun, because that's what macroeconomists get paid to do.

Beckworth: That's what the model says.

Smith: That's what the model says. And so you had these people, I think that [inaudible], they sort of did entertain the real concept, but then John Cochrane was probably just trolling.

Beckworth: Stephen Williamson's another one.

Smith: Stephen Williamson doesn't know whether he is being serious or trolling because he is Canadian. And being Canadian means you exist in a quantum superposition of seriousness and sarcasm. And that's Steve Williamson. But anyway, here's the point. They tried it, Erdogan tried it in Turkey, and Erdogan really, really got this very, very strong idea that lower interest rates will make inflation lower, and higher interest rates will make inflation higher. So, in response to spiraling inflation, he continues to keep interest rates low based on… he didn't read Neo-Fisherian theories or papers. And he didn't listen to the MMT people either. But he just came up with the same silly idea on his own. And currently, Turkey is just spiraling into hyperinflation because, guess what? That's not how things work. And so these nutty ideas, and you have other nutty ideas.

Smith: I consider Austrianism to be another one of these nutty ideas. I know lots of people like it, but certainly the version of Austrianism that relies on praxeology, which it's that meme where they say, "Source? I made it up." That's basically praxeology. And so there's these other nutty ideas floating around, and when you see them actually tried, they blow up economies really badly. Yes, macroeconomists, they haven't nailed down the facts about how things work quite yet. But the heuristics, they've developed. These little couple lines on a graph will easily beat whatever Stephanie Kelton is serving up. A couple lines on a graph, IS-LM will outperform Stephanie Kelton every time. You just follow Stephanie Kelton's advice, you'll just blow up your economy. And then Stephanie Kelton will say, "Well, of course MMT got this right too." Because remember, they always take a victory lap no matter what happens. And they'll wave away and qualify their tweets when they slipped up and said something specific because that's how charlatans work. And so anyway, these nutty ideas are consistently worse than mainstream macro, despite all the weaknesses of mainstream macro. So, stick with mainstream macro, just don't think it's anything like real science yet, settled science. Macro is in its infancy. But it's better to be in your infancy than to be dead.

Beckworth: Okay, well let me make a charitable case for some of these groups here. I agree that they have not been borne out and practiced, nor do they always have good theory, although I would say neo-Fisherians, they do rely on a model. They do have some reason, and I think neo-Fisherian-

Smith: There's a model. We know why it's wrong.

Stick with mainstream macro, just don't think it's anything like real science yet, settled science. Macro is in its infancy. But it's better to be in your infancy than to be dead.

Beckworth: Well, I would say neo-Fisherians confuse the long run for the short run, because it's a Fisher relationship, as the name implies. But the MMTs, I mean, they are really a subset of Post-Keynesian economics.

Smith: They're a cartoon version of post-Keynesian economics.

Beckworth: Okay. Well I'm just saying post-Keynesian economics, you can find some reasonable people who make arguments. And I want to mention one, I think it's Marc Lavoie, he came on the podcast and we talked about, for example, Paul Volcker's disinflation and the recession he created to bring inflation down. And he said, "Look, I would've preferred price controls over jacking up interest rates to double digits like Paul Volcker did." And I said, "Well, okay, but don't we have evidence that price controls don't always work out so well?" And he was honest. He goes, "Yes, you're right. They can cause distortions and can cause problems, but I'm going to err on the side of price controls," because he believed they were less harmful. But at least he was open, transparent about it. And they come from a very different place. Post-Keynesians view inflation as a power struggle between capital and labor. Whereas mainstream would view inflation, demand and supply kind of driven by spending, whatever it may be.

Beckworth: So there are people now, I guess with MMT, I believe one thing that maybe really undermined their credibility is that when we had this high inflation, no one acknowledged this is what they had been talking about, that inflation was the real constraint. If you recall, that's one of the claims they made when we had this inflation. Then they weren't ready to say, "Okay, now it's time to start tightening policy." And some of the MMTers also have, I think, a good understanding of how the plumbing of the monetary policy system works. But your point is these ideas have been tested and they have not worked out in practice.

Smith: Well, right. So I think with ideas like price controls, that's not obviously nutty because you see some cases where we've done price controls and it didn't blow up the economy.

Beckworth: Right, World War II.

Smith: World War II. And so it's not nutty on its face either. So MMT, it's nutty because when you ask them to tell you the model, they never will, you can't find the model. Price controls, the rationale for price controls is very simple, it doesn't depend on the model and it's pretty straightforward. Stop raising prices because I tell you to. And you can point to some examples where this was done. And so it's not obviously as nutty. I think that in general it's a pretty bad way to control inflation and that it has extreme dangerous downsides and relatively little upside if you look at how well it's actually done. So I think it's a bad idea, but I don't think it's a nutty idea.

Smith: So I think there's a difference between bad and nutty. And any physicist is able to tell the idea between a physics theory that's wrong and a physics theory that's gibberish. And so people in the economic sphere need to be able to tell the difference between a theory that's wrong and a theory that's gibberish. And I think that MMT and praxeology are gibberish, whereas the idea of price controls and a lot of these things are not gibberish.

Beckworth: Right. So they need to come up with a formal model in order for it to be falsifiable, to be checked.

People in the economic sphere need to be able to tell the difference between a theory that's wrong and a theory that's gibberish.

Smith: Yeah. And it doesn't have to be like a math model. You don't have to write equations. But you have to say, "Okay. My theory of how the economy works is if this happens, then that happens. If you do this, then that happens." Some way for you to evaluate and test even in a heuristic, even in a non-quantitative way, but just a directional way of, "I did this in an inflation went down." You don't have to predict that it went down by 3.3752 points or whatever.

Smith: But you'd have to say, "Yes, when we do this, inflation goes down.” And to a certain extent we're working off a similar alchemy with fiscal stimulus. We don't exactly know how fiscal stimulus works. We just observe that there appears to be a Keynesian multiplier, especially in deep recessions. We don't know how it works yet. It is alchemy. We have models of the zero lower bound. Those models are almost certainly mis-specified at the micro level. It's alchemy. And so price controls to the extent that they work or don't work or are another form of alchemy. And macroeconomics isn't a real science yet, it's still alchemy. And that's okay because you can't pretend, right? You shouldn't pretend to be more scientific than you are. It's okay to not know things yet. Right?

Smith: And there's many things we just don't know yet about the world and how it works. But when you see people like MMT or the praxeology people, when you see these people, they claim to already know everything. They have an immaculate theory that has solved everything, an immaculate theory, which we cannot relay the particulars of at this exact time. But trust us, it always works every time exactly, because based on accounting identities or undeniable axioms about human action, it's based on absolute truth, never fails. That's gibberish, that's nonsense. And that's very different than saying price controls will reduce inflation. That's something you can test by doing some price controls.

Beckworth: It's a testable theory. Okay.

Smith: Testable. Let us remain anchored to reality instead of pretending that we know everything.

Beckworth: Right. Right. This has been great, Noah. We're going to have to wrap it up. But our friend from the days of blogging, Nick Rowe, he attempted to re-engineer the IS-LM to make an MMT model out of it. So for those who are interested in an attempt by a non-MMTer to formalize the MMT view, you can go check his work out. In fact, we'll provide a link to that. He basically redid the IS-LM and MMT view. So at least someone's trying to give them a model. Well, Noah, thank you again for coming on the show, spending the extra time with us and we appreciate it.

Smith: Alright. Thank you very much, really appreciate having you on.

Check out our new Macro Musings merch here!

About Macro Musings

Hosted by Senior Research Fellow David Beckworth, the Macro Musings podcast pulls back the curtain on the important macroeconomic issues of the past, present, and future.