Scott Sumner  

Nominal GDP is not real (and is really tiny)

PRINT
Preferences in The Warriors... How Economics Helps Us Underst...

Sometimes I argue that nominal GDP is like Coke, it's the "real thing." By that I mean it's a well-defined concept, the dollar value of all output of final goods and services. Of course I exaggerate, there are some conceptual problems in how to define NGDP. But real GDP has all those problems, and much deeper ones.

Some government statistics define real GDP as "volume" of output. But that's nonsense; it would put a country in the midst of heavy industrialization (like China) far ahead of a high tech society like ours. The next defense of real GDP is the price index, but how is that to be measured? The change in the price of a given basket of goods? But which basket, and how to we address new goods, and changes in quality? There are no good answers to these questions, which is why no one agrees as to whether US living standards have stagnated since 1970, or risen rapidly (as I claim.) On the other hand, we all pretty much agree as to what has happened to NGDP since 1970.

But there is another sense where real GDP really is very much real, and NGDP isn't. RGDP is something that can be pictured in one's mind---"the economy." You can picture all those factories churning out cars, all those homes housing people, all those barbers cutting hair. But NGDP can't really be pictured. Don't believe me? Try to picture the explosive growth in Zimbabwe's NGDP during the early 2000s in your mind's eye. My claim is that either you cannot, or you end up picturing bushels of worthless currency. And that's because NGDP is essentially a monetary concept, not real.

By analogy, if Switzerland had used gold as its money, its NGDP would have plunged much lower since the early 2000s, although it would have rebounded somewhat in the last couple years. But that fact would tell us nothing about the Swiss economy we picture in our mind's eye. It tells us about the international market for gold, where prices soared in relative terms, then fell back somewhat.

In a recent comment section there was skepticism about my claim that a couple of small interest rate increases in 2011 could have plunged the eurozone into another recession, associated with a sharp slowdown in NGDP growth. There are two problems with that criticism:

1. I don't actually believe the higher interest rates were the best way of describing the problem. I prefer "tight money" and just threw out the interest rate increases as a "concrete step" for what Nick Rowe calls the "people of the concrete steppes." Interest rates are not a good indicator of the stance of monetary policy, and eurozone money was much tighter than the increase from 1% to 1.5% would have suggested.

2. The other problem with the skepticism is the implicit assumption that NGDP is some vast thing, and that a central bank would have trouble nudging it this way or that. No, real GDP is a vast thing, and as I said can be pictured in one's mind's eye. NGDP is a tiny thing, no larger than a dollar bill. Suppose the Fed depreciates the one-dollar bills that it has a monopoly on producing, from 1/17,000,000,000,000th of a year's output, to 1/18,000,000,000,000th of a year's output. Then they've just boosted NGDP from $17 trillion to $18 trillion.

I think the problem is that people think of the central bank as being like a tugboat, slightly nudging a huge ocean liner this way and that. This sort of analogy would be sort of OK (although not perfect) for central banks trying to influence real GDP. They can do so in the short run, to some extent. But when we shift over to NGDP the analogy completely breaks down. Now the central bank is no longer the tugboat trying to nudge the ocean liner, it is the ocean liner itself. It steers the nominal economy.

The path of NGDP is monetary policy. Through errors of omission or commission, the ECB has created the path of NGDP that we observe since 2007.

Do those shifts in NGDP also cause the truly vast real GDP to move around? Most conventional Keynesians and monetarists (and Austrians?) would say yes, but only in the short run. This is usually attributed to sticky wages and prices (or perhaps misperceptions.)


Comments and Sharing






COMMENTS (23 to date)
RPLong writes:

Does this suggest that the argument for NGDP level targeting is circular?

Andrew_FL writes:

Scott- I can't presume to speak for all Austrians, and especially not the Rothbardians, but the issue I have with the concept of "real gdp"-and thus whether it is driven by changes in NGDP in the short or long run, is that the concept is ill grounded from the view of subjective value theory. Or rather, the concept of the price level is. There are ways to reformulate these concepts in subjective value terms, i think, but they must involve removing pretensions to measuring "real value." It meaningful to talk about, for example, the price a set of goods was tending to exchange for at some past time and thus speak of the size of total income in "equivalent number of market baskets." But it would be wrong to say that people's income in total objectively had greater or lesser "value" only that they made more or less things or the equivalent of said things. I happen to think abundance is good and i think people generally agree, so in that sense long run growth of "real gdp" meaningfully reflects a good thing, one is tempted to say objectively good: increased abundance.

With regard to short run influence on "real gdp" the answer is yes, as you would understand it. An expansion of M not in response to increased money holdings (decreased V) in the short run of the Austrian story, involves a diversion of resources to early stages of production away from later stages, which involves usually an increased volume of investment-at the same time as making actually deferring consumption less attractive than it otherwise would be, meaning consumption increases to. If P is in consumer prices, and consumer prices rise some time after M increases, it must be the case that Q increases in the interim. That's standard Quanity theory, of course. The logic of the Austrian exposition is predicated on the increase in M being through loan markets. Through transfer payments or a helicopter it would look more like Friedman's short run Phillips Curve story.

I actually have pretty extensive thoughts on synthesizing Austrian and Monetarist ideas in this respect, including a rambling essay about the equation of exchange that I mentally drew on heavily for this comment. But I don't presently have a good place to put it. Oh well.

Scott Sumner writes:

RPLong, No. It's based on the theory that a more stable path of NGDP leads to a more stable path of RGDP.

Andrew, Yes, I assumed that Austrians believed in short run non-neutrality, but I hedged as I didn't know if some denied that claim.

I agree that RGDP is not well grounded in subjective value theory.


Rob Dawg writes:
[N]o one agrees as to whether US living standards have stagnated since 1970, or risen rapidly (as I claim.)

Why not both? Since 1970 I doubt you could find anyone claiming stagnation. That was 44 years ago. Two generations. But what about since 1992, one generation previous? I'm sure there are quite a few would would opt for an unquestioned slowing. What about more recently? What about since 2003? Half again. Here I doubt you'd find many asserting "rising rapidly" nearly all choosing stagnation with a sizable majority opting for decline. If the argument were of medians and not averages then the field tilts even more strongly towards decline.

The problem is one of choosing the appropriate time frame with which to measure improvement.

Michael Byrnes writes:

Andrew,

To me, concerns about the validity of RGDP or the price level are all the more reason to favor NGDP targeting. At least relative to other regimes such as inflation targeting.

It is certainly feasible that life is more complicated than "RGDP growth = good!". It is harder to imagine, though, that driving RGDP down via monetary policy is ever a good idea.

If you target NGDP (value the dollar as a fraction of output) rather than targeting inflation (value the dollar as a fraction of the CPI or PCE basket), you don't need to concern yourself with what estimates of inflation and RGDP really signify.

Andrew_FL writes:

@Michael Byrnes-I don't disagree that the concerns about RGDP make it make more sense to focus on NGDP. In fact I focus more on it myself lately. I'm not in favor of targeting anything per se, if targeting means that the central bank actively decide about how to attempt to manage the money supply with some behavior of NGDP in mind. I am, I think, in favor of implicit targeting of NGDP as would occur if banks were not regulated and allowed to issue their own notes. And I would consider the idea of setting up a bitcoin style computer program to manage the monetary base-instead of a Central Bank-with particular behavior of NGDP in mind-say, Selgin's "productivity norm"-if it proved to be an improvement on the stability that would be achieved with free banking with a fixed monetary base.

And if driving down RGDP via monetary policy means allowing NGDP to contract, I agree that it is not a good idea. I think it's probably not a good idea to drive it up either. But that's what happens with Central Banks.

If my oblique reference to being unable to speak for the Rothbardians didn't make it clear, I'm not a 100% reservist or a "fractional reserves are fraud" sort of person. I favor free banking. I think the Rothbardians misunderstand the theory they purport to represent in this regard. In this respect it's important to note that in the face of declining V, I favor what a free banking system would do: increasing M enough to prevent a decrease in MV, no more, no less.

Rae writes:

You assert:

...NGDP is a tiny thing, no larger than a dollar bill. Suppose the Fed depreciates the one-dollar bills that it has a monopoly on producing, from 1/17,000,000,000,000th of a year's output, to 1/18,000,000,000,000th of a year's output. Then they've just boosted NGDP from $17 trillion to $18 trillion.

This is a glib and highly questionable statement as it implied money supply growth = NGDP growth.

A realty check:

https://research.stlouisfed.org/fred2/graph/?graph_id=202137

RPLong writes:
It's based on the theory that a more stable path of NGDP [by which you mean "monetary policy" - RPL] leads to a more stable path of RGDP.
No one that I know of disagrees with the claim that stable monetary policy leads to stable RGDP, ceteris paribus. Surely we can produce examples of people who favor very volatile movements in monetary policy, but AFAIK, it is the minority of economists who favor wild swings in monetary policy, ceteris paribus.

But then a big economic shock comes along, people stop spending money, and the NGDP level-targeters suggest large increases in the money supply to ensure that, whatever might be happening to real output, NGDP follows a stable path.

So the only way this looks like "stable monetary policy" is if we look only at NGDP growth, not at anything else (like, say, the money supply).

Hence, it starts to sound like a circular argument: "NGDPLT is stable monetary policy because it produces stable NGDP levels." So it wouldn't be surprising that you would argue that NGDP is "the real thing." We have to look at it that way, otherwise "monetary policy" (NGDPLT) doesn't look as stable as you're suggesting it is.

Maybe it's just me?

Kevin Erdmann writes:

RPLong,

Think of it this way. Do CEO's go on conference calls and say that they will be able to increase compensation and pay off creditors because the money supply is way up? Or do they say they will be able to increase compensation and pay off creditors because revenues are up? At best, they might say that since the Fed is increasing the money supply, they hope that leads to higher revenues.

So, why are you concerned with money supply? Corporate revenues are basically NGDP, aren't they? Why not get rid of nominal macro fluctuations and leave firms to compete in a stable macro revenue environment?

Stable monetary policy, when stability is defined as an inflation target, most certainly causes unstable RGDP, doesn't it?

Kevin Erdmann writes:

Keep in mind that the day after the Lehman Brothers failure, when the Fed's head trader told the FOMC that a large portion of the market was signaling that "the financial system is going to implode in a major way" the Fed held rates at 2% because of a concern for inflation.

http://idiosyncraticwhisk.blogspot.com/2014/03/the-fed-in-2008.html

RPLong writes:

Kevin,

I agree that the inflation hawks were wrong. I disagree that NGDP is the only macroeconomic data point that matters.

I do believe the money supply matters. I agree that CEOs don't explain their operating plans based on the money supply, but they also don't explain their position based on NGDP. I agree that corporate profits are part of NGDP, but I disagree that corporate profits are "basically" NGDP.

This is why the argument seems circular to me. It seemingly only makes sense when you start to see everything as "basically" NGDP. But if we do that, then we aren't talking about macroeconomic stabilization theory anymore. Instead, we're just articulating a tautology: Stable NGDP --> stable NGDP.

In a sense, stable NGDP --> stable NGDP is very true. In a certain other sense, stable NGDP --> stable NGDP is not as explanatory as it needs to be, in my opinion.

Michael Byrnes writes:

Andrew wrote:

"I'm not in favor of targeting anything per se, if targeting means that the central bank actively decide about how to attempt to manage the money supply with some behavior of NGDP in mind. I am, I think, in favor of implicit targeting of NGDP as would occur if banks were not regulated and allowed to issue their own notes."

This only makes sense to me if you are opposed to having a central bank at all, which seems to be your position.

It makes less sense if you are arguing that, given the existance of a central bank with a monopoly on issuance of base money, it should not target NGDP.

To me, those are two different arguments. (I think you are making the first one, which may be a good idea, but doesn't really say anything about the second.)

Kevin Erdmann writes:

Not corporate profit - corporate revenue. Corporate revenue is a much closer proxy to NGDP than corporate profits. Shocks to nominal corporate revenue are more important to labor and credit markets than either inflation or real production.

Andrew_FL writes:

@Michael Byrnes-I am indeed trying to argue for the "first best option," that is, free banking with no Central Bank at all.

As for the question, given the existence of a Central Bank, what the Central Bank should try to do: I do indeed think it should target NGDP in some way, according to some rule. I also don't think you can get Central Banks to do that. That is, as long as policy is decided by actual human beings, they will exercise discretion, rather than follow a hard and fast rule.

Michael Byrnes writes:

RPLong wrote:

"I do believe the money supply matters. I agree that CEOs don't explain their operating plans based on the money supply, but they also don't explain their position based on NGDP. I agree that corporate profits are part of NGDP, but I disagree that corporate profits are "basically" NGDP."

Are there reasons to be concerned about growth of the money supply... beyond the concern that it might destabilize NGDP? After all, it is excess spending (NGDP), not excess money, that causes excess inflation.

As for businesses... I don't think they pay close attention to NGDP. But they don't have to - they pay very close attention to spending (on their product).

Andrew_FL writes:

@Michael Byrnes-

Are there reasons to be concerned about growth of the money supply... beyond the concern that it might destabilize NGDP? After all, it is excess spending (NGDP), not excess money, that causes excess inflation.

I would say, yes and no. An increase in M which increases funds loaned out for investment is a concern if the change is not justified by an increase in voluntary deferrence of consumption (increased "money holdings"). Of course, if MV is stable in the sense of being constant, there's no cause for concern-all income that is not spent on consumption must be invested and vice versa. Changing NGDP could (must?) imply otherwise. This a problem even if P is unchanged.

RPLong writes:

@ Michael Byrnes - My concerns about growth in the money supply are: (1) that it is an avoidable market distortion, and (2) that it increases the potential for systemic corruption.

@ Kevin - I apologize for misspeaking, but that revision doesn't appreciably alter my point. Let me try to make it one more time, the long-winded way:

There are people out there who attempt to "manage" their Body Mass Index, but these people are fools. Why? Because BMI is calculated from height and weight, and because height is a scalar for most adults. It's a monotonic transformation invented to get around the colinearity between height and weight. So talking about BMI is mostly just a roundabout way of talking about weight.

Despite what Prof. Sumner has said above, RGDP is not tangible. It is even less tangible than NGDP. NGDP is at least a number we can add up. RGDP is nothing more than that same number adjusted for inflation. It is an econometric invention, even moreso than NGDP.

So when Prof. Sumner starts talking about "stable NGDP growth" leading to "stable RGDP growth," he is technically putting NGDP on both sides of the equation.

What happens when you put the same variable on both sides of an equation? One of two things:

  • The variable cancels out, or
  • The variable is reduced, but we're left with something else to explain
In either case, we haven't told the whole story. I feel as though we haven't told much of a story at all. It's too circular to be descriptive.

Scott Sumner writes:

Rob, I frequently see people claiming living standards have not risen since the 1970s.

Rae, I certainly didn't mean to imply money growth equals NGDP growth. That would be absurd, as you say.

RPLong, You said:

"But then a big economic shock comes along, people stop spending money, and the NGDP level-targeters suggest large increases in the money supply to ensure that, whatever might be happening to real output, NGDP follows a stable path."

That is certainly not what I am advocating. I do not favor any sort of money supply targeting, and I doubt a big increase in the money supply would be needed when a "shock" came along.

RGDP and NGDP are completely different things, as different as poetry and elephants. Check out the data for Zimbabwe.

RPLong writes:

Prof. Sumner, you lost me. When did I mention money supply targeting?

You don't mean that RGDP and NGDP are different things. Rather, you mean that real economic output, for which we use RGDP as a proxy, is different than nominal spending, for which we use NGDP as a proxy. This is important because it highlights my point: You can't get to an RGDP number without first computing an NGDP number and then combining it with an inflation estimator.

Thus, using an NGDP level target to influence RGDP means (by arithmetic fact) that central banks must either (1) do something to inflation or (2) express RGDP in terms of something other than NGDP, and then do something to one of those other things.

I think we agree on this much, right?

Don Geddis writes:

@RPLong: "by which you mean "monetary policy"" LOL. No, Sumner meant what he wrote originally, "stable NGDP". How silly that you change the words in the quote, and then complain about the new phrasing which happened only in your own head.

"When did I mention money supply targeting?" You said: "big economic shock ... the NGDP level-targeters suggest large increases in the money supply." You're the one who mentioned money supply increases. You've missed the importance of forward guidance. NGDPLT would not necessarily require more M. (The commitment to NGDPLT can change -- and/or stabilize -- V, instead of changing M, in order to stabilize NGDP = MV.)

"You can't get to an RGDP number without first computing an NGDP number and then combining it with an inflation estimator." False. You can look at prices of various goods during a previous year, and then multiply by quantities produced during the current year, to get an estimate of RGDP during the current year. It is not necessary (although it is often convenient) to compute the current NGDP first. In fact, my alternative computation completely ignores whatever prices happen to be today (and completely ignores whatever inflation might have been between the two periods).

RPLong writes:

Hi Don,

"Stable NGDP" is a monetary policy, according to Prof. Sumner. I think we're all on the same page here, so please try to argue in good faith.

Yes, I did mention changes to the money supply. What I did not mention was "money supply targeting." Again, we all know what we are talking about here, so let's try to discuss this in that spirit.

Your RGDP estimation is nearly equivalent to the arithmetic definition I gave above. Engaging in silly sleight-of-hand tricks does not nullify my point.

Don Geddis writes:

@RPLong: You think I was being unfair to you? Just nitpicking, and not responding to the "spirit" or "good faith" of your comments? I was trying to show you the specific point where your reasoning went off the rails.

But ok. If we ignore what you actually wrote as reasoning, and just look at some of your conclusions, we see things like this: "No one ... disagrees with the claim that stable monetary policy leads to stable RGDP ... it is the minority of economists who favor wild swings in monetary policy" But Sumner was arguing specifically for stable NGDP, not for "stable monetary policy" in general. So your change of phrase is hugely important. Because you're unlikely to find economists who favor "volatile monetary policy". But it's easy to find economists who favor "a laser-like focus on low inflation", even if that happens to lead to volatile NGDP. So you're not actually responding to Sumner's point.

And then you say: "Sumner starts talking about "stable NGDP growth" leading to "stable RGDP growth," he is technically putting NGDP on both sides of the equation." Which is totally false. But you've just confused yourself, because you've been sloppy about what you see as an arithmetic manipulation, that RGDP is typically computed by first adding up NGDP, and then doing an adjustment.

But the idea of RGDP as a concept, is not something that is a mere restatement of NGDP. Sumner has talked about Zimbabwe, where they differ greatly. And even the general observation that they're typically correlated, is something that we cannot take for granted but must instead explain.

You're completely on the wrong track, thinking that NGDPLT is essentially a tautology, that all it means is "stable NGDP leads to stable NGDP". RGDP is something completely separate from NGDP, and the goal of stable NGDP is to achieve stable RGDP. You glossing over that difference is completely confused.

Commander writes:

It's quite revealing to compare RPLong's measured comments with Don Geddis's foul-tempered responses.

Comments for this entry have been closed
Return to top