Josh Hendrickson on Measuring Money in the Economy

Josh Hendrickson is an assistant professor of economics at the University of Mississippi. He specializes in monetary economics and blogs of the everyday economist. Josh's research has focused on the role money plays in monetary policy. He has also worked on the role monetary policy played during the Great Moderation and on ways to make the financial system safer. Josh joins David to discuss the trend of monetary economists pay less attention to the money supply, preferring instead to look at economic slack, inflation, and interest rates. Hendrickson pushes back against this trend by making the case that money does still matter both theoretically and empirically. Josh and David also discuss the importance of Twitter and blogging for building a professional brand.

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: Josh, welcome to the show.

Josh Hendrickson: Thank you very much.

Beckworth: All right, Josh, all my guests, I always ask them, how did they get into macroeconomics. You're an assistant professor. You're going to pretend you're soon. You're a little bit different in some of the guests but I still like to hear how you got into macro and what is the path of your journey?

Hendrickson: I think maybe the thing that got me most interested in macros, I remember as an undergrad, I started liking my econ classes. I remember one day, we talked about the Great Depression and sort of the stories that you heard in your econ classes about the Great Depression are much different in sort of like the conventional wisdom.

Elsewhere, it was always something like, "Oh, well, the stock market crashed," and then you had the depression or you had these other sorts of explanations. I don't know. All of a sudden, I got to, I think I was taking an intermediate macro and they kind of talked a little bit about Friedman and Schwartz and I was kind of really intrigued because all of a sudden, I was like, "Nobody ever talked about the money supply." They sort of got me interested.

I think from there, I always had kind of a bigger interest in macro than anything else but I sort of just really liked economics. Then, I sort of, once I was in graduate school, I kind of just gravitated towards macro.

Beckworth: Okay, did you read Friedman and Schwartz, and intermediate macro? That'd be amazing.

Hendrickson: No, no. I didn't-

Beckworth: Okay.

Hendrickson: Well, hold on. Maybe I should say yes and then so I look more impressive. No, I didn't. I did, however, pick up a copy of Friedman's the Optimum Quantity of Money and other essays. I thought, maybe this would be like a simpler introduction to some of these topics or something. I don't know if that was really simple or not, but I did get a lot out of it and it sort of piqued my interest.

Beckworth: Okay. Now, you're going to pretend you're soon. You're also someone who has been involved in blogging and on Twitter. I'm curious, what would be your recommendation to a freshly minted PhD in terms of engaging in social media? There's a lot to learn from being in the econ blogosphere and following people on Twitter, who are financial, journalists, economist and so forth. There's also a cost, you're going to publish. You got to get tenure. How do you handle that trade off and what would you recommend to a young tenure track professor?

Hendrickson: Well, I mean, I guess, for me, social media kind of served two purposes. It was a way for me to interact with other economists, but also a way to sort of build a brand. I didn't really have a name brand associated with me. I sort of needed to branch out and talk to people and discuss things and promote myself, I guess.

I guess, to me, the biggest thing about blogging is in terms of the writing, actually is, a lot of times you're thinking about something or you think something is one particular way and then you see somebody else write something different and then you think to yourself, "No, that's wrong. I need to write a blog post about it," but the thing is, is when you actually sit down to write the blog post, I think you actually learn a little bit of something about what you think and what the other person thinks because you have this instinct that you disagree, but then when you actually have to write out why you disagree, it forces you to be a little more careful and to really sort out the issues where you disagree, because typically, we're just arguing with other economists and I think most of us have all the same background.

We've read most of the same material. We have different opinions. It's not because one of us hasn't read something that the other one has read or something. It's typically there's something that one of us thinks is more important than something else. I think blogging helps you to work out some of those differences are really identify where those differences are and I think that's kind of the benefit.

In terms of publishing and getting tenure, I don't know. I haven't got it yet so I guess I shouldn't give advice but, to me, I guess what happens is I sort of just disappear for a while. If I'm working on a paper, I just don't blog or get on Twitter very often or the thing for me, too, is Twitter sometimes as a way to interact with fellow economists and kind of debate the sorts of things but also without having to put in the amount of effort that you have to write a good blog post.

Sometimes it's just a good outlet, like when you're working on a paper, and you need to take a little half hour break or something, you just get on there and see what everybody's talking about. If you're interested in it or if you have something to contribute, then you chime in. If not, you just go right back to what you were doing but...

Beckworth: Twitter is therapy, huh?

Hendrickson: I mean, I think I've met a lot of people through there. I think it's a useful tool. As much as people sort of complain about the format and the sense that you're arguing two or three sentences at a time, I actually think that's a good thing because it forces you to be concise and to get to the point and not to just filibuster your argument, I guess.

Beckworth: Yeah. No, I found Twitter useful too for the same reasons in terms of connecting with people that you never would have otherwise and to see kind of what is the current conversation and a chance to float some of your own ideas.

Let's move to some of your research and the key area of your research is monetary economics. You have taken a novel approach, when I say novel, a different approach than most people do in the field of monetary economics because you still take money seriously in your research, both theoretically and in your empirical work. Most monetary economists don't think very long or hard about money. I want to talk about that.

First, I want to focus on the empirical side. I think, at least for me, it's the easiest ones to wrap my mind around because we get into theoretical one, there's a little bit more involved but on the empirical side, most monetary economist, as I said, including central bankers and prominent academics, don't look at monetary aggregates anymore. They do monetary policy without money. There is no money in monetary policy, ask someone on the street, they might find that strange if you said the Federal Reserve doesn't care about money.

Money is more of a byproduct or we would say in economics, it's endogenous to the system. The Fed worries about output gaps, inflation, and so forth but you've taken a different tack. Before we get into what you've done exactly, why do you think the profession has gone down this path where money isn't taken seriously anymore and money isn't really looked at closely. What's the evolution of thought that has brought us this far?

Hendrickson: Well, I don't think it's ideological. I think it's grounded in what people think is solid evidence, I guess. I mean, like I said, my interest in macro to start with was based on these arguments that were described to me from Friedman and Schwartz. To me, being interested in macro, I sort of thought, "Oh, money is really important." Then, you get to graduate school and you start learning monetary models and you're reading Woodford and there is no money. Woodford also cites these papers to justify this. Woodford cites several papers where he says, "Well, look-"

Beckworth: Just to be clear, this is Michael Woodford...

Hendrickson: Yeah, Michael Woodford, yeah, sorry.

Beckworth: ... and he's the kind of the godfather of graduate macro textbooks, but go ahead.

Hendrickson: Yeah. He cites a couple of papers where basically people have taken monetary aggregates and put them in certain empirical models and tested to see whether or not money plays any important role above and beyond to say the short-term nominal interest rate or something like that. There's quite a few of these papers in the '90s that couldn't find any evidence that money mattered at all. They estimate these empirical models and the empirical model would basically say something like, "Look, money growth isn't helping us to forecast inflation and money's not helping us to predict fluctuations in output," and things like that.

I think sort of like motivated by those, I don't know, empirical results, people like Michael Woodford and others working in the similar line of work kind of developed these models without money because they sort of said, "Well, look, if the quantity of money doesn't seem to matter and we look at central banks and central bank's look like they're just using interest rates as their sort of instrument of policy, then maybe we don't actually need to think about money at all. Maybe we should just think about it entirely in terms of interest rates."

I mean, and in fact, Woodford's original model, if you go back and you look at his first model where he doesn't have money, he sort of motivates it by saying, "Well, look, one of the reasons why money it probably isn't as important as it used to be is that we're just getting much more technologically sophisticated. Eventually, what's going to happen is that the monetary base is going to become arbitrarily small." Eventually, you get to a point where the central bank isn't even going to be able to use the monetary base to conduct monetary policy. What do they do at that point?

He kind of viewed his model as sort of like, "Hey, let's take the limit as the monetary base goes to zero and then see what monetary policy looks like and then when it was sort of, when you combine this with the empirical evidence that suggested that money wasn't important, money just sort of gradually disappeared from the models.

Beckworth: Right. Now, you've pushed back against that view, a few others have as well but you've done some work. You had an article in the Journal of Macroeconomic Dynamics and it was titled Redundancy or Mismeasurement? A Reappraisal of Money. Tell us what you uncovered in that study and what it suggests about the consensus sphere in the profession now?

Divisia Measures of Money

Hendrickson: Well, one of the things that I was always kind of interested in is why these empirical results existed. Why is it that something changed and there were all these stories about technological change and how just the structure of banking change and money have changed and there's all these assets that didn't exist before. That's sort of what people will use to explain it but to me, it wasn't clear why this would immediately mean that this stuff would break down, especially if these were identifiable changes.

I sort of got attracted to this literature that was started by William Barnett and Barnett in the '80s basically derived these Divisia monetary aggregates. Basically, what a Divisia monetary aggregate is, is it takes all the components of the monetary aggregate and then it weights them by something called an expenditure share. That expenditure share is itself a function of like the user cost of money.

The basic idea here is, is that if you think about different types of money, they don't all provide the same flow of services. I can give currency to virtually anybody. Not everybody will let me write a check. I can't necessarily spend out of my savings account immediately or something like that.

Basically, these different assets have different liquidity properties as it relates to spending. As a result, they also have different rates of return. One of the things that's in the weighting system with these Divisia aggregates is this user cost. With the user cost, the less liquid the asset is, the higher the rate of return that it's going to have and the lower the weight that it's going to get in the aggregate.

Basically, in a way, you're sort of reweighting these assets in terms of the degree of liquidity of the asset so that the sort of flow of services that you're getting is being properly aggregated across these different assets because they're... Yup, go ahead.

Beckworth: No, I'm sorry, contrast that with what we typically look at, when I go to FRED and I download the M2 Money Supply, you've called that a simple-sum. Why is that called simple-sum and how is that different than the Divisia as you just described?

Hendrickson: The simple-sum aggregates are called simple-sum because they are just the summation of all the monetary values of these assets. Then, the Divisia aggregates are our weighting each component of the aggregate based on these expenditure shares.

The basic thing is that Barnett's derivation of these Divisia aggregates just follows from standard sort of demand theory. What he basically showed is that the only way that a simple-sum aggregate is the appropriate way to aggregate is if all of these assets were perfect substitutes and we know that they're not perfect substitutes. I mean, we know both intuitively and if you go to the data, if you estimate elasticity and substitution and things like that, we know that they're not perfect substitutes.

I'm interested in this, if we're really concerned with things like technology and we're really concerned with the new products that are coming about because of financial innovation, these aggregates would seem to be able to incorporate those substitution effects.

In other words, if I take, say M2U or something like that and I switch from holding a checking account to a savings account, if people are doing this on average in the economy, then you're going to get more people with savings accounts and less people checking accounts, but also, that's going to change the rates of return on these assets. When it changes the relative rates of return, it's also going to change the weights. The numbers are going to change and the weights... The quantities are going to change. The weights are going to change. That's going to be incorporated in the asset.

Maybe that's not particularly important if you think about checking accounts versus savings account but it is important when you start thinking about things that have been added to monetary aggregates. For example, in the early '80s, things like Money Market Mutual Funds and things like that were add to the aggregates.

Well, if you think of a simple-sum aggregate, if you expand the simple-sum aggregate to include Money Market Mutual Funds, then in the month that you insert Money Market Mutual Funds into that aggregate, that aggregate is going to increase at least by the amount of Money Market Mutual Funds that exist, like the dollar value of those funds.

In other words, what you'll see is like this giant spike in the monetary aggregate because you've added these assets, whereas if you think about the Divisia aggregate, if you put Money Market Mutual Funds into the Divisia aggregate, all of the other weights on all of the other assets have to change. Then, the Money Market Mutual Funds are also included, but they're given a weight as well. Those weights adjust.

In fact, Barnett actually has a paper about this. Actually, I think it was a speech that he gave that was published, I think, maybe an economic journal or something but in the speech, he described how in the early '80s, Milton Friedman had a column in Newsweek and Milton Friedman's column in Newsweek talked about how the Fed wasn't really committed to stopping inflation and this was in the early '80s. He said, "They're not really committed to stopping inflation because if you look at money growth, money growth spiked in the most recent data release. Inflation is going to be back."

Barnett talks about how on the same day, he had a op ed in Forbes where he basically said, "A lot of people are making a big deal about the spike in money growth and claiming it's going to lead to inflation but if you look at these Divisia aggregates that I came up with, that spike in money growth just isn't there." This had a lot to do with that financial innovation that was going on at the time and these revisions to these aggregates.

If you look at the data, Barnett turned out to be right and if you look at the simple-sum measures, there are these spikes that exist in those simple-sum measures in the early '80s, that just don't exist in the Divisia aggregates.

This comes back to what you asked me about my paper. In that macroeconomic dynamics paper, the idea that I got is that these aggregates are theoretically superior. Barnett showed, using standard sort of microeconomics that these Divisia aggregates were preferable theoretically and one of my questions was, "Well, how much do they matter empirically?" Maybe these studies that have showed that money is not important, they all use simple-sum aggregates. Maybe the reason they're getting these results is due to the fact that they're using these simple-sum aggregates and they're not using these Divisia aggregates.

Essentially, what I did in that paper is I replicate the results on these previous papers on the simple-sum aggregates and by and large, my conclusions agreed with their conclusions. Even though I had more data than these other papers, their conclusion still sort of held up with these simple-sum aggregates for the most part. Then, but then when I looked at the Divisia aggregates, basically, all these puzzling results just sort of were overturned.

Money growth did predict inflation. Real money balances do have predictive power about the future output gap. The demand for money is stable over long periods of time. A lot of these results sort of were overturned just by using Divisia aggregates. My argument in that was that these things are not only the superior, but empirically, they also seem to be very important and provide a better indicator than the simple-sum aggregates.

Beckworth: Had the Fed been looking at these Divisia measures in the '70s and '80s, maybe things would have turned out differently?

Hendrickson: Yeah, I mean, I think so. William Barnett actually has a book called Getting It Wrong, where he actually documents... It was at the time when he actually came up with these Divisia aggregates. He was at the Board of Governors. He was in their research division. He wrote this book Getting It Wrong.

I think it came out just after the recession, where he basically talked about how the major failures of monetary policy have been because not only they ignore money, but even if they were to look at money, they wouldn't be looking at the proper measurement. Yeah, he makes that case in that book. I largely agree.

Beckworth: One of the issues in doing monetary policies is having the right amount of knowledge. In fact, I call this the knowledge problem. It's hard for people who are running the central bank to know exactly what's the best thing to do in real time. We see that today with inflation targeting and should a central bank respond to a sudden change in inflation. Is it due to commodity prices? Is it due to demand pressures? It's hard to know in real time.

Does this knowledge that you get, this added knowledge, this added insight, I guess, from using the Divisia measure of money, is it something that you could completely rely on? I mean, how would you use it? Would you use it in conjunction with some other indicators or would it be your sole indicator? How would you operationalize the knowledge you get out of these Divisia measures of money?

Operationalizing Knowledge from Divisia measures

Hendrickson: Well, I certainly think it should be used as an indicator at the very least. I think, in general, as a central bank, what you want to do is you want to have some target and you want to have some target that you can actually control. You want to pick a nominal variable that you can control. You don't want to pick real variables but maybe we want to push that conversation aside for a minute, but you want to pick a nominal variable target that's actually what you want to achieve.

For me, I tend to go towards nominal GDP but in a way, I think Bennett McCallum said this once or maybe it was Tobin, I don't know, I should be better prepared, but they basically said if you think about nominal GDP, nominal GDP is basically a velocity corrected measure of the money supply. I think that's probably right but if you set the, I mean, I think that's the right way to think about it but I think if you set your goals for policy, you want to set it as a goal that actually is consistent with some broad objective.

In other words, you target nominal GDP because you think that it's going to provide the right sort of macroeconomic stability but the thing with nominal GDP is that we only get nominal GDP numbers every quarter. When you're looking at this every quarter, policymakers need to have some idea about whether or not they're going to hit that target.

I think if you think of nominal GDP as this velocity adjusted monetary aggregate, then, the monthly data that you get on these Divisia monetary aggregates would help you to kind of achieve that goal. If you're using these Divisia monetary aggregates as intermediate targets, for example, to accomplish your ultimate goal, which is stable nominal GDP, then I think there's a useful role there.

Beckworth: No, that because they... You mentioned how they're constructed and by the nature of their construction, do they better reflect velocity changes then? Is that what you're suggesting and therefore, they can better get at nominal GDP?

Hendrickson: Well, one of the things I think that's important is if you think about... One of the reasons why people thought that money wasn't important is that velocity appeared to be doing weird things in the '80s. If you look at the data from the '80s, there's like these sort of jumps in velocity and things like that but the thing about this is, is that those changes in velocity just come from the way that velocity is being calculated.

If you're using simple-sum monetary aggregates and that simple-sum monetary aggregate jumps up because you now include Money Market Mutual Funds, for example and nominal GDP, that change in money is not going to affect nominal GDP. Nominal GDP stays the same. The money supply jumps up. Well, velocity has to jump down so that...

Beckworth: I see.

Hendrickson: ... equality holds. You get this weird behavior in velocity that can have nothing to do with changes in people's actual demand for money. It can actually just be due to changes in how the monetary aggregate is measured or substitution effects that were there's some assets that are included in one aggregate and not the other. If people are switching from holding one asset to another, then it's going to make one monetary aggregate look like it's going down and the other one look like it's going up and things like that.

Beckworth: Okay. You get a more stable, as you mentioned, more stable, long-run money demand relationship. You don't have these crazy experiences. For example, in the '70s, there's the case of the missing money, which was really what you would explain as this poor measurement issues.

All right, we have talked about Divisia as one way we can think of taking money more seriously. One reason to take money more seriously, at least empirically, better measurement of money and how it is actually used. I want to look at a different way that I think has maybe caused some people to reconsider money being an important part of macroeconomic analysis and that is this recent crisis we went through.

I think Gary Gorton has been at the front of this, but he has stressed that the money used in the shadow banking system collapsed as he has documented and many people have noted, there was a run on the shadow banking system, just like there was a run on the banking system during the Great Depression.

Like the Great Depression, the run on the banking system caused the collapse in the money supply. If you look at the money assets used by these institutional investors in the shadow banking system, they also disappeared, at least the privately provided ones disappeared. Public ones, treasuries, offset some of them but what you see, if you account for both retail money assets, which are things that are included in M2 as well as these institutional money assets, the things that are used in the shadow banking system, we see this collapse in the broad money supply that never has fully recovered.

Do you see that also as a promising, I don't know, future or something that's caused people to reconsider their lack of focus on money?

Hendrickson: I don't know how many people have reconsidered it but I think that maybe they should. Not surprisingly. The data that you can get on the Divisia aggregates is actually much broader than the data that you can get on simple-sum aggregates. I think that that's actually another one of the benefits of this because one of the trends that we've seen over the, say the last 20 years, is big changes in sort of institutional demand for safe assets.

If you have this big change in institutional demand for safe assets, typically like large institutions they're not going to want to hold savings deposits or something like that because of things like deposit insurance. You have deposit insurance, but their deposit is going to be much larger than what's covered under deposit insurance, and so they're basically subsidizing all the other depositors in the bank.

They want to find something that's going to give them a rate of return, something that they think is safe. What we've seen over the last 20 years is we've seen all of these alternative assets that have that have sort of sprung up, that people hold on to and they're able to use, for transaction purposes, and things like repurchase agreements and things like that. I think this is kind of a big story that maybe is unappreciated and it's not going to show up in the standard statistics because these repurchase agreements and things like that are like money in a way, because you're using the asset like money because you're able to purchase by engaging in these repurchase agreements.

These institutions that are holding these things, we've seen the demand increase substantially over the past 20 years. If you look at the Divisia aggregates that are calculated by the Center for Financial Stability, they actually have broad measures of the money supply so they actually go beyond things like M2U and they have M3 and they have M4. M4 includes all of these different types of assets that are in standard monetary aggregates, but it also includes things like repurchase agreements and T-Bills and all these other sorts of things that can be used by institutions like money. I think that these are important because empirically, they also look important.

In theory, you think if there's this change in trend and this trend is important, then maybe we should focus on it and then if you go to the data, the data actually seems to suggest that it's important because when you look at the data, there were much sharper declines in the growth rate of these broader measures of the money supply than there were in the narrower measures because a lot of what happened was, I mean, Gorton sort of draws the correlation between depression and what happened in recession, where instead of a run on traditional banks, it was a run on the shadow banks.

If you buy that story, then you have to be interested in these broader measures of Divisia aggregates because they're the aggregates and they're the variables that are going to show you actually what's going on in that market.

Beckworth: Is it frustrating for you, as a teacher, when you open up your principles of macroeconomics textbook and they're talking about M2 still, I mean, maybe there's some books that now talk about these broader measures of money, the shadow banking system, but it's been my experience that most textbooks stop at M2.

Hendrickson: It bothers me less that textbooks do this. It bothers me substantially more that central banks ignore them.

Beckworth: Good point. All right. Well, let's move from the empirical side to the theoretical side. We've talked about how better measurement basically does show an important role for money. Let's move now to the modeling side. There's been a huge literature on how to model monetary economics but the approach that has been taken has been unsatisfactory. You've written about this. Can you explain the standard approach and then tell us what is wrong with it?

Critiquing the Standard Approach to Monetary Theory

Hendrickson: To me, there's a long history of economists who have written about the importance of money and sort of why we use money, why we should care about money, why money is important. Then, a lot of what you see in the literature where people are looking at money for macroeconomic analysis, sort of ignores all these things that we think are important.

Essentially, they say, "Well, we know that people hold money. They hold money for some purpose. You're out of way to get it into the model so that we can figure out how important it is.

Typically, what people do is they put money in the utility function, for example. Real money balances make people happy, apparently but if you read the monetary literature, this is never the argument. Nobody ever argues that, well, people hold money because money makes them happy. People are holding money because they want to make transactions. I mean, if we could live in a pure credit economy where nobody ever had to hold money, I don't think people would be sad. I don't think that's going to reduce our on welfare. I just approach to sort of ignore the important money.

Also, if you think about cash in advance constraints, if you have a cash advance constraint, it's basically this idea that we just append another constraint to the model. The consumer has a budget constraint, but then they also have this additional constraint that says that any consumption that they do has to be done with money. They have to use money too.

In a sense, this seems like a reasonable way to incorporate money because you're requiring money to be used in transactions, which we observe in reality so that seems like a good idea. The problem is, is that you're adding another constraint to the consumer's problem. If you're adding a constraint to the consumer's problem, you're basically implying that the consumer would be better off if there was no money because then they wouldn't have this additional constraint.

Beckworth: Money enters as a friction almost, right? It's almost like a nuisance.

Hendrickson: Yeah. My big problem is that money basically seems to be this thing where it's like, we don't really know how to put it in the model but we think maybe it's important or maybe we want to test whether or not it's important. We need to figure out a way so that we can act appropriately with how important it is.

If you look at the standard, like New Keynesian literature, there's this sort of circular logic that goes on. They take the standard, like New Keynesian model that doesn't have any money. They have this New Keynesian model and they can close the model with an interest rate rule for policy and so they don't have to include the monetary aggregate. Then, initially, this approach was critiqued because they were leaving out money and people thought, "Well, money is important. Maybe you shouldn't leave it out of the model even if it's less important than it used to be. Maybe it still needs to be in the model."

The response to this was basically to say, "Well, hey, look, we'll put money in the model and then we'll see whether it's important," but the problem is, is if you start from a model where you already have a closed form solution and you just add an equation, a money demand equation, by definition, that's not going to affect the dynamics of the model because money is not playing an important role. You're just adding. You have three equations, now you have four equations.

The three equations have three variables in them. Those three equations with three variables, you can use system reduction and you can solve for those three variables all by themselves, and then when you solve for the for money, you just take those equilibrium values of the other three variables, plug them into the money demand equation, and there you go. Now, you have money too.

It's completely endogenous they haven't taken seriously money. It's a circular logic because it's like, "Well, we don't think money is important. Money's not in the model." To show you that money's not important now, given that we already have this model, we'll add another equation to the model and show you that it's not important and see it's not important. Well, yeah, it's not important because you add it to a model where you assumed that it wasn't important.

Beckworth: They've kind of set money up for a fall from the way the model is designed. This is the traditional approaches. You mentioned the money in the utility function, the cash and advance constraint. What are you doing that's trying to remedy this deficiency?

Hendrickson: I've kind of switched to thinking about things in terms of these search models of money. I'm not going to say that the search models have, that all the characteristics of these search models are desirable or preferable, but the one thing that these monetary search models do is they force you to motivate why particular assets exist and also to explain why different assets coexist.

In other words, if you want to have currency in your model, you have to have some reason why currency exists. The search models motivated these things because you assume that people sort of meet at random and they trade, and these trades are anonymous. You don't know who you're trading with. Credit's not feasible and you have to use currency.

But then if you want to add more assets to the model, so maybe you want to have banks. If you want to have banks, then you have to explain why the bank exists and you have to explain why anybody would ever hold currency when a deposit pays interest. This really forces you to think about the microeconomic aspects of money and all the different types of money that exist and why assets coexist.

To me, these monetary search models are really the first type of model to really try to integrate the microeconomic literature that's been going on for a very long time with macroeconomic models that are capable of being used for policy analysis.

Beckworth: How widespread are these monetary search models?

Hendrickson: I don't know how widespread they are. I mean, there's a conference every year at the Chicago Fed that's fairly large in size but I think, in general, it's still small compared to sort of like the New Keynesian consensus, I think, that you see. Typically, people are going to do monetary policy analysis.

I think it's still much more common for people to work within a sort of New Keynesian-type model with interest rate rules and things like that than it is to work with these models but I think it's growing. I think that these models provide... They provide a reason to be optimistic that people will start taking money more seriously and that people will think a little bit more deeply about some of these issues.

Beckworth: Okay, well, let's take this understanding that you have that money does still matter. We need to be careful when we build our models, when we measure money. Let's apply it to some monetary history.

You've written on the Great Moderation, that period between the early '80s and 2007 when macroeconomic stability was very relatively high, recessions were mild, and a lot of people have attributed to monetary policies, some have attributed to other factors, good luck, supply shocks, globalization, but you've written on the Great Moderation and the role monetary policy may have played in creating that. Can you speak to that?

Explaining the Great Moderation Period

Hendrickson: One of the things that I noticed when I was reading this literature on the Great Moderation is that there was a lot of emphasis on monetary policy but the literature sort of seemed to come to this kind of odd conclusion.

In other words, early on, John Taylor had done work that showed that in the '70s, if you estimated a Taylor rule, for example, the central bank wasn't responding sufficiently to inflation. The coefficient on inflation and their Taylor rule was too low. In fact, if you estimated these Taylor rules and then you plug in those values to a standard, like New Keynesian model, you would get indeterminacy. There was multiple equilibria.

Then, what they showed is that if you looked at the post like Greenspan era, that the estimated parameters of the Taylor rule actually looked a lot like Taylor's theoretical rule. Not only that, but that the parameter attached to inflation and the Taylor rule was now sufficient to get a unique equilibrium.

The basic argument here then was, is that, "Well, what changed in monetary policy is that policymakers actually started taking inflation more seriously. They started responding more towards inflation when setting policy." Because they were more responsive to inflation and setting policy, that can explain why we got better policy with Greenspan than we did with, like, Arthur Burns.

That was interesting, but there were two kinds of lines of literature that cast a little bit of doubt on this story. The first one was the work that was done by Robert Hetzel. Hetzel, essentially was arguing that this view of monetary policy was just odd. On the one hand, what people are arguing is that the central bank just wasn't responsive enough to inflation.

What that implies is that the central bank sort of has to respond to inflation. What Hetzel said is that this is kind of like a weird argument and argument because central banks have to respond to inflation. Why are they responding to something that they control that's a weird way of thinking about policy. That got me thinking about this evidence and how to interpret this evidence differently.

Then, on the other hand, you had this empirical literature where Orphanides had basically said that, "Well, if you look at the data, you're estimating Taylor rules with the data that we have now, but policymakers didn't have that data at the time." What we really need to do is we really need to look at real time data. If we're going to estimate how central banks responded, then when you estimate how central banks responded to inflation, for example, you have to estimate that equation with the data that the central bank had at the time. Otherwise, you might have sort of these biased conclusions because data gets revised.

What Orphanides did is in a series of papers, he showed that when you use real time data, this significant difference between how central banks responded to inflation in the '70s and then after the '70s, wasn't very different at all, actually. In fact, the way that the central bank responded to inflation in real time in the '70s, was sufficiently high that if you plug it into these New Keynesian models, you would have got that unique equilibrium solution that other authors had claimed you couldn't get.

You have these two strands of literature. It basically started making me think because I had this idea that monetary policy seemed like it least played a role in the Great Moderation but at the same time, I didn't really think that this way of looking at monetary policy from the Taylor rule what was really the right way to do it.

I started looking at nominal GDP. I noticed that if you look at nominal GDP growth, nominal GDP growth was just very, very stable during the Great Moderation and that it was not stable, prior to the Great Moderation, in the '70s, for example. My basic idea is that if you want to the simplest possible explanation, if you think about how you teach principal students about fluctuations in the economy, you teach them in terms of either supply shifts or demand shifts, right?

When aggregate demand changes, price and quantity are moving in the same direction. The price level is going up, GDP is going up, but when you have a decrease in aggregate supply, the price level is going up, but real GDP is going down. What I started thinking about is if you think about the quantity theory and you think about monetary policy, then if the central bank is just being too expansionary, then you should see high nominal GDP growth. If the central bank is trying to actually maintain low stable rates of inflation, you should see stable nominal GDP growth. When you look at the data, that's what you saw.

What I did is I went to went to the data and I basically estimated some monetary policy rules where instead of responding to inflation in the output gap, I had the central bank responding to fluctuations in nominal GDP growth.

What I found is that even when you use real time data, using the Federal Reserve's forecast of nominal GDP growth, you could show that there was this very significant change from the '60s and '70s into the Great Moderation where they became much more responsive to these changes in nominal GDP growth. It essentially looks like policy. The argument that I made is that it looked like policymakers realize that they're the ones who create high inflation, right?

Beckworth: Mm-hmm (affirmative).

Hendrickson: By accepting responsibility for it, when they saw indications that inflation was rising and when they saw indications that nominal GDP growth was rising, they knew that that was driven by factors that they control. Even if you're only concerned with inflation, if nominal GDP is rising, you know that it's because of demand side factors, which as a central bank, you can you can have some control over whereas if you see inflation but nominal GDP isn't really changing, then you know that that's supply side factors and you can sort of let that pass through sort of like one time, the price level, and then sort of ignore it going forward.

If that's what central banks were really concerned with and really concerned with maintaining low stable rates of inflation, then they should be much more responsive to things like nominal GDP growth because it provides them with evidence about whether or not the inflation is their fault or whether it's being driven by these temporary supply shocks.

Beckworth: What you're saying then is your results confirm the point Taylor is making but in a different way where he was arguing central banks became more responsible, more responsive to macroeconomic changes, but you argue they did it through nominal GDP versus looking at inflation itself directly. This, I guess, then maybe undermines Orphanides' critique because I think I heard you say that this works even with real time data, is that correct?

Hendrickson: I wouldn't necessarily say undermines his critique. I mean, I guess, in the sense, these are two different ways of looking at policy. I kind of view the way that I'm looking at policy more like the way that Hetzel was describing policy. If you actually read the minutes of the FOMC meetings and things like that, you actually get this impression that that's what was going on because you see the Federal Reserve really take responsibility for creating inflation. It's not just a matter of being vigilant enough to fight high inflation. You're the one that's creating it. You don't have to be vigilant enough to fight it. You're the one who created in the first place.

I'm not saying that the Federal Reserve necessarily consciously targeted nominal GDP growth but what I'm saying is that responding to nominal GDP growth, thinking about monetary policy is responding to nominal GDP growth is much more consistent with this Hetzel line of thinking where you're taking responsibility for causing inflation and so you're adjusting monetary policy when you have some indication that that inflation is rising because of your actions.

Beckworth: They took ownership of the fact that they're the ones who create these inflation changes coming from demand shocks.

Hendrickson: Right.

Beckworth: All right, well, let's move to the Great Recession, what are your views on what caused the Great Recession?

Hendrickson: My view is this mix of Gorton and then the views that you and Scott Sellner have articulated, I guess. To me, I see the real the real problem as happening with these assets that people thought we're safe. Gorton and David Andolfatto like to talk about whether assets are information sensitive and I think maybe that's a good way to think about it.

A lot of these assets that were considered safe assets, in good times, they were not inflammation-sensitive. If something was changing in sort of like the underlying assets, in the short run, you didn't really worry about it if it wasn't going to affect anything in the long run. Everybody sort of operates as though nothing is changing in the short run. Everybody just sort of operates in accordance with the long run.

The problem is, is that when you had the collapse in the housing market, some of these safe assets were things like mortgage-backed securities and things like that. Now, there was something sort of similar to like a lemons problem. People know that there's this problem in housing. They know that there's going to be an increase in these defaults. Some of these mortgage-backed securities aren't going to be worth as much as they were before.

Suddenly, these assets become information sensitive. Now, they're constantly being priced and the way that they're being priced is if people want to use these things in transactions, they had to take very substantial haircuts.

Well, if you think about what happens when they have to take haircuts on these assets is if I'm going to engage in a repurchase agreement and let's say I have an asset that everybody thinks is worth $100 and I want to engage in the repurchase agreement, then, I sell that asset to you at par and then I buy it back from you later on for a little bit more, but I'm selling it to you at par. Holding that asset that's worth $100 is giving you $100 worth of purchasing power.

Whereas once these assets become information-sensitive, now all of a sudden, I go to you and I want to engage in repurchase agreement, you might be afraid that the reason I'm engaging in this repurchase agreement is that I have this bad asset and I'm just trying to get rid of it and I'm trying to stick you with a bad asset. I'm going to sell the asset to you, promise to repay you and then go, "Oh, yeah, you know what, I can't repay you. You can just keep the asset but the asset is not worth as much as you paid for it. I'm better off, you're worse off.

If you think about what happens is that in that scenario, you're going to want to make me take a haircut, right? You're going to say, "Look, I know that this asset is worth $100 but I'm going to treat it as though it was worth $75 or $50 or something like that. If you want to engage in the repurchase agreement, I'm not going to give you as much for that asset as I would in normal times."

Well, the problem is, is that basically what this does is that, in effect, this reduces the amount of transaction assets that I'm holding. Even though I'm holding the exact same assets that I was holding before, the actual value of those assets is much lower. I can't make the same volume of transactions that I had before.

I think, that's broadly consistent with sort of Gorton's story but if you look at the data, we sort of see this, when you look at the very broad measures of the Divisia aggregate, so if you look at Divisia M4, for example, Divisia M4 growth collapses, it's actually negative for a short time. If standard monetary theory suggests that if you have this sudden collapse in money growth, you should have a collapse in nominal spending and real GDP in the short run, at least.

I think that when people talk about the housing market, the housing market sort of indirectly caused the recession but only because of all these sort of assets. Of course, all of this is endogenous, so whether or not you can have the housing bubble is maybe conditional on whether or not these assets exists, et cetera but in general, my thinking is that this sort of collapse in the value of these sorts of things was like an exogenous decline in the money supply, basically. Without a corresponding increase, then you're going to get the adverse consequences that go with it.

Beckworth: Well, we're out of time. Our guest today has been Josh Hendrickson. Josh, thank you for being on the show.

Hendrickson: No problem. I appreciate it.

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.