Scott Sumner  

NGDP isn't "the economy"

PRINT
Spiderman and the trust-buster... The Auction/Bidding Metaphor: ...

Which of the following three concepts is the outlier, i.e. very different from the other two?

1. Real GDP
2. Nominal GDP
3. The $/pound exchange rate ($ price of pounds = E)

I'd guess most econ students would say "the exchange rate." If you asked most macroeconomists, however, they'd say RGDP is the outlier. Well, at least I hope they'd say that, the alternative is too horrible to contemplate.

Real GDP is (an admittedly crude estimate of) the total amount of final goods and services produced during a given period. Nominal GDP is the dollar value of that output. Both NGDP and exchange rates are nominal variables, and the difference between a real and a nominal variable is vast, much greater than the difference between two nominal variables. Indeed if I had put Keats' Ode on a Grecian Urn in the number three slot, you could still make an argument that number one is the outlier. NGDP and poems are both abstractions, RGDP is concrete.

Nominal variables (actually their inverse) can be thought of as different ways of describing the value of a dollar. Thus 1/E is the value of the dollar in term of pounds. And 1/NGDP is the share of a nation's output that can be bought with a single dollar. These are alternative definitions of the value of money. 1/CPI is the purchasing power of money in terms of a given basket of goods, and is thus a third definition. There are people who claim that the inverse of the dollar price of gold is "the" value of money. I believe that Louis Woodhill over at Forbes falls into that camp.

This distinction is easiest to see when there is hyperinflation. Back around 2008 the RGDP of Zimbabwe was barely changing, indeed it was probably falling somewhat. At the same time the Zimbabwe NGDP and the Zimbabwe$ price of US dollars was rising at trillions of percent per year. Indeed if we were able to get data for comparable periods of time, the change in the exchange rate and NGDP would have been almost identical. Ditto for the Zimbabwe price of gold and the Zimbabwe CPI.

Why does this matter? I often get commenters who accuse me of trying to control or plan "the economy." But when the news media says "the economy," they mean precisely one thing, real GDP. They do not mean nominal GDP. Then the commenters use arguments they read in Hayek to convince me that "the economy" is better off if left to market forces, free of government interference. They don't need to convince me. With a few exceptions such as pollution and other externalities, I strongly favor having the free market allocate resources. Indeed I am more pro-market than even most conservative economists. Lots of conservatives think "tests" tell us something useful about school quality. In my view if a free market voucher system leads to high schools that put 10 times more resources into Friday night football than into math and English, then that's the optimal school system, regardless of whether our students score as well as Finnish or Korean students. We are the rich country; they should be copying us.

Back to NGDP. So I'm not trying to "plan the economy," I'm a pragmatic libertarian. I'm trying to have the government plan the value of currency, a product of which it is already the monopoly supplier. More specifically I want them to minimize the harm that an unstable currency can do to the economy.

Other commenters complain about my claim that a stable path for NGDP would prevent most recessions, accusing me of engaging in a tautology. Again they are confusing RGDP with NGDP. Perhaps that is forgivable, as the two variables are highly correlated in the US. But the very same dynamic that causes that confusion (the correlation between RGDP and NGDP) is also exactly what is wrong with their complaint against NGDP targeting. The fact that NGDP and RGDP are highly correlated in the US but not at all in Zimbabwe largely reflects the fact that at low to moderate inflation rates NGDP shocks tend to cause RGDP to move in the same direction. So much so that many confuse the two concepts. But that's exactly why we need to stabilize NGDP.

BTW, the correlation at low to moderate inflation rates reflects the fact that wages and prices are sticky at those sorts of rates, and also that changes in inflation don't completely dominate changes in RGDP.

To conclude, there is no essential difference in the degree of "socialist meddling" between planning the growth path of NGDP, and planning the CPI, and planning the price of gold. I don't know how the Heritage Foundation does its monetary freedom index, but any nominal monetary target should count equally. Instead of worrying about money, libertarians need to worry about:

1. Why the US dropped out of the top ten in terms of economic freedom in the 2014 Heritage rankings.
2. Why despite the fact that we have fallen 2 places behind Denmark, the Heritage Foundation has not advocated replacing the US economic system with the Danish system. (I'm joking, given the size disparity that would be impossible.)

PS. When presenting my ideas at the AEI, I made an offhand remark about how the Fed needed to "steer the nominal economy." There seemed to be a rumble of disquiet in the audience. When I figured out what was wrong I reassured them that I did not favor planning "the economy."


Comments and Sharing





COMMENTS (24 to date)
John Becker writes:

I understand that the dollar price of output is essentially meaningless. However, you can see how real events would affect nominal outcomes like NGDP or exchange rates over the short to medium term right?

About the Fed, institutional arrangements matter and having one person or a committee in charge of producing the most important good in the economy is a recipe for disaster. These are people with political motivations outside of doing what's right for the economy. Read Greenspan's book "The Age of Turbulence" if you don't believe me.

A non-discretionary policy rule is better than a discretionary policy but you couldn't trust them to stick to it anymore than you can trust the government to safeguard the 4th Amendment. Trusting government to stick to their limits is completely utopian.

I think it's a disgrace we aren't at the top of the Heritage Foundation list when you look at the arrangements of other countries.

Thomas Sewell writes:

Thanks, Scott. This is actually one of your more helpful posts for placing your NGDP views into context. Sometimes it's easy to get lost in the details and forget to write about the big picture, but where the details provide evidence, they don't always convince as well as seeing the big picture laid out.

Scott Sumner writes:

John, Whether real shocks affect nominal variables like NGDP and the exchange rates depends on the monetary system. Under NGDPLT they don't affect NGDP very much, and under Bretton Woods they have almost no impact on exchange rates.

Thanks Thomas.

Eli writes:

Scott,

The fact that NGDP and RGDP are highly correlated in the US but not at all in Zimbabwe largely reflects the fact that at low to moderate inflation rates NGDP shocks tend to cause RGDP to move in the same direction.

How can you be so sure that causality runs from NGDP to RGDP? Does this hold generally in your estimation (at least within the bounds of low to moderate inflation regimes)?

My concern is that if NGDP becomes the target variable, the correlation between NGDP and RGDP could weaken.

bill woolsey writes:

Eli:

If nominal GDP level targeting were perfectly successful, the correlation between real GDP and nominal GDP would disappear. Real GDP would vary from trend due to supply shocks, inflation would have -1 correlation to real GDP growth. Real GDP deviations from trend, would not be correlated with the nonexistent deviations for nominal GDP from target.

Market monetarists not only recognize this, we celebrate it. The whole point is to get rid of changes in real output caused by nominal shocks.

Scott Sumner writes:

Eli, Good points. It's not always causation running from NGDP to RGDP. For instance if the central banks successfully targets inflation it runs from RGDP to NGDP.

On the other hand when we know NGDP is being affected by a monetary shock, we still see the correlation.

And you are right that the correlation would weaken under NGDP targeting, but I also believe the business cycle would moderate.

Rajat writes:

Isn't RGDP the abstraction? You can observe NGDP by observing every transaction that occurs in a year and counting up all the values. RGDP is an abstraction derived from dividing changes in observed NGDP by some estimate of changes in prices. While NGDP is exactly $X, RGDP can be anything depending on which base year one uses. It's only good for estimating changes in the value of output.

JGRA writes:

Have you actually been to countries like Finland, Denmark, Germany, Switzerland, Austria, Japan? Most of these feel and are a lot richer than the US. Again a problem with nominal GDP. According to the media US growth has surpassed other Western economies for decades, yet most of what I see is rotten infrastructure and public and private debt. Go to developed Western economies in Scandinavia, German central European block and it feels richer AND more evenly distributed.

JGRA writes:

We are the rich country??
Have you actually been to countries like Finland, Denmark, Germany, Switzerland, Austria, Japan? Most of these feel and are a lot richer than the US. Again a problem with nominal GDP. According to the media US growth has surpassed other Western economies for decades, yet most of what I see is rotten infrastructure and public and private debt. Go to developed Western economies in Scandinavia, German central European block and it feels richer AND more evenly distributed.

Eli writes:

Bill and Scott, thanks for your replies. That does help clarify the MM perspective for me (Scott I have followed your blog for a while now, but some aspects of MM were/are not entirely clear to me).

I must confess I have always been sympathetic to the Austrian Business Cycle story -- not that it is the only one to tell, but that it ought to at least be considered when evaluating monetary policy proposals -- that monetary expansion has the potential to create relative price distortions across sectors and across time which can result ultimately in R and N GDP collapses. And this in part motivated my previous comment.

Given NGDP=P*RGDP, a policy of NGDPLT would call for declines in RGDP to be met with monetary expansion, and I fear that this expansion could exacerbate the problem of resource misallocation and thereby further inhibit RGDP, or at least its growth path (I get that for MMers expectations is the name of the game, but this inverse correlation still typically holds under NGDPLT I would think).

Or put differently: it seems to me that MMers take business cycles to be a function of both real (supply) shocks and nominal shocks which can be largely treated as independent from one another (at least in the evaluation of NGDPLT). If we can eliminate nominal shocks with NGDPLT, my interpretation of MM thinking goes, then that is one less source of fluctuation and therefore it will lead to less severe business cycles. My concern is that the effort to reduce monetary shocks may end up amplifying (what will be viewed as) supply shocks if ABCT has any merit -- that real and nominal shocks cannot be evaluated independently under any monetary policy.

That said, please don't mistake my probing of NGDPLT for an absolute confidence in ABCT. In general I find that the more I read about macroeconomics, the more confused I become about everything.


Don Geddis writes:

Scott, typos: Para 1 "...following THREE concepts...". Para 7 "...often GET commenters..."

Excellent post, BTW. Both clear and non-obvious.

Eli: "...if ABCT has any merit...". Fortunately, it doesn't, so: problem solved! :-)

Michael Byrnes writes:

Eli wrote:

"I must confess I have always been sympathetic to the Austrian Business Cycle story -- not that it is the only one to tell, but that it ought to at least be considered when evaluating monetary policy proposals -- that monetary expansion has the potential to create relative price distortions across sectors and across time which can result ultimately in R and N GDP collapses.

Given NGDP=P*RGDP, a policy of NGDPLT would call for declines in RGDP to be met with monetary expansion, and I fear that this expansion could exacerbate the problem of resource misallocation"

I probably shouldn't comment on ABCT, about which I am far from an expert, but it seems to me that Austrians should prefer NGDPLT to flexible inflation targeting. For a couple of reasons:

1. A lot of people (not just Austrians) oppose NGDPLT on the grounds that it would call for looser monetary policy. But this is not always the case! Scott Sumner advocates a 5% NGDP growth level target... for example, NGDP growth was consistently higher than 5% throughout the 1970s (often it was greater than 10%). A policy of 5% NGDPLT would have meant tighter money in the 1970s. Other market monetarists prefer a level target of less than 5%.

2. Doesn't ABCT really argue that business cycles are driven by loose money during the boom? I thought the Austrian narrative about our current crisis was that the Fed inflated the housing bubble with easy money - its inevitable collapse is what drove us into recession in 2008. (I'm skeptical of this, in part because we had a worldwide recession, not one limited to the US).

But if the Austrian explanation were true, that would have to be thought of as a failure of flexible inflation targeting (since that is the policy that the Fed and many other central banks followed during the housing bubble.) In this view, flexible inflation targeting was a procyclical. (i.e. under this policy regime, the Fed provided too much liquidity during the boom).

NGDPLT would arguably not have this flaw. During a boom, we would see faster RGDP growth. But under NGDPLT, faster RGDP growth would mean tighter money and lower inflation. In contrast, under flexible inflation targeting, faster RDGP growth would mean more NGDP growth (beyond what would be allowed under a NGDP level target.


James writes:

[Comment removed pending confirmation of email address and for rudeness. Email the webmaster@econlib.org to request restoring your comment privileges. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Matt D. writes:

Just curious:

1. How often should NGNP be measured – and then policy adjusted? Can NGNP really be accurately measured – and in a manner timely enough to be useful? The free market is constantly in flux moving prices toward the right direction, so how can some centralized targeting measurement system ever hope to keep up with this?

2. How are injections of cash and withdrawals of cash supposed to take place? Say real GNP is sinking, then isn't the response to inject a lot of cash into the economy? Who gets this money first? Doesn't this potentially encourage government profligacy? Moreover, what if there was an NGNP that was too high? How easy will it be to withdraw cash from the system? Will the government just cut spending at the drop of a hat?

3. What if the level of cash were kept constant, but regulation were decreased on banks, such as reserve requirements, capital requirements, and so on. We could also eliminate deposit insurance and let banks really fail. Then banks could expand or decrease the money supply in this manner – would that not be preferable to the government attempting via brute force to adjust the NGNP? Why can't the monetary base just be kept constant?

4. Isn't the real problem that we don't have enough of a free market in banking anyway? We have central planning in monetary policy – so the suggestion here is that we should at least exercise this central planning in as predictable a manner as possible – we should avoid any discretionary measures – but act according to a simple predictable rule – that at least in theory makes some sense. Yet, how can this compare with a genuinely liberated system – where there is no centralized authority at all? Can we not see the problem? It's as if we want to say, well, yes, I know the system sucks, but at least for what it is, this would be the best policy. But maybe that's not enough – maybe the criticism needs to be directed at the deeper underlying problems. Focusing on monetary policy might unintendedly serve the end of those who *want* a monetary policy – by suggesting if only the right policy were in place, we can get it to work. We're talking about brute force adjustments anyway you cut it, splice it, or dice it. I don't see how this view is compatible with libertarianism. It's *not* a market solution.

5. Basically, politicians will always be happy to see money coming into the economy – and they'll cite whatever theory is convenient to justify this activity. But then when that same theory calls for the opposite action, they'll just shift to another theory ... just again, realistically, can NGNP targeting even become a political reality? How?

mbka writes:

Scott,

you should direct everyone towards this kind if post before they read further on your take on monetary economics.

I read your blog for years and thought I understood at least some of it... But only when in some offhand remark you defined 1/NGDP as the value of output that a single dollar can buy, only then did the whole concept of NGDP targeting suddenly make sense to me. Once I understood this, stabilizing NGDP crystallized effortlessly as the simplest and most logical measure to anchor the whole monetary system, similar to inflation targeting but immune to supply shocks, and independent from a lot of assumptions.

"1/NGDP = the 'real' value of money" in terms of the real economy, whatever it currently does, this is the starting point for the entire discussion. You should proselytize more on this. Say, both friends of the gold standard and inflation "hawks" intend nothing more insidious than a stable, predictable value for "money". It shouldn't be hard to convince either one of the MM concept, IF and only if you get the idea across that the most natural anchor for money is not gold or some artifically computed value, but the actual economy, 1/NGDP.

mbka writes:

Hmm, just realize that my 2nd paragraph doesn't make complete sense because at the end I talk about NGDP and the real economy in the same breath once more instead of using your term "the share of real output". It's a dollar's share of real output of course that would become the monetary anchor, not the actual real output.

Evelyn Zhang writes:

[Comment removed for supplying false email address. Email the webmaster@econlib.org to request restoring your comment privileges A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Eli writes:

Michael Byrnes,

I agree generally with the points you make and prefer NGDPLT to inflation-targeting and discretionary monetary policy regimes. What I am not sure about is if I prefer NGDPLT to growing the monetary base by X%/yr, with X being either a definite number between (and including) 0 and 5, or tracking population growth, or some measure of credit availability (which could effectively end up as growing M2 by some fixed %/yr), or something else.

John Becker writes:

Scott,

You said, "Whether real shocks affect nominal variables like NGDP and the exchange rates depends on the monetary system. Under NGDPLT they don't affect NGDP very much"

If there were a serious problem like bank failures such as we had in the early 1930s or 2008, I very much doubt that the Fed could maintain NGDPLT even over a period of say 5-10 years. Do you really think people would be willing to tolerate say 7% inflation, -2% real GDP growth, and an 8% unemployment rate? I think that inflation getting that high while people were suffering would become a serious political issue.

In my view, stabilizing Nominal GDP is still a form of central planning that has a deep impact on real/microeconomic issues. A 5% NGDP target builds in a fair amount of inflation each year. This will affect how people save money, how businesses plan for the future, borrowing costs, how people have to do accounting, the distribution of income, etc. If money is really neutral in the long run, why not 0% NGDP growth or 3%,2%,4%,1%, etc,

Scott Sumner writes:

Rajat, RGDP is a very crude attempt to measure something concrete. NGDP is a much more precise attempt to measure something abstract.

JGRA, I've been to Switzerland and Germany, and I agree they have better infrastructure. I don't think Germany is as rich as the US, but of course it's a judgment call.

Eli, As for ABCT, recall that Hayek also favors NGDP targeting, so it's not clear to me that the two theories have different policy implications. NGDP targeting would not eliminate distortions, but it would greatly reduce them, compared to inflation targeting.

Don, Thanks, I'll fix it.

Matt, We should target future expected NGDP. I'd prefer the government measure it monthly, but quarterly will still work.

The government should inject money by buying T-bonds. But it makes no difference if they use another technique, like paying government salaries in cash. It doesn't encourage big government because no more money is injected with NGDPLT than inflation targeting.

A constant base would be a complete disaster. During recession when interest rates fall the demand for base money would rise, causing depressions.

Thanks mbka.

John, There are two problems with your argument:

1. The fall in RGDP and the bank failures of 1930-33 were mostly caused by the fall in NGDP. With a sound monetary policy those problems don't happen.

2. Even if they do 2008 proves you wrong. Inflation rose and RGDP growth fell in 2008, and yet the Fed cut rates in 2008 and it was not controversial. People often make the exact opposite argument, asking me if I think the Fed could get away with letting RGDP fall without doing more monetary stimulus.

Many NGDP proponents do favor lower rates. I'd be fine with 4%, if done with level targeting. Even 3% might work.

Michael Byrnes writes:

John Becker wrote:

"Do you really think people would be willing to tolerate say 7% inflation, -2% real GDP growth, and an 8% unemployment rate? I think that inflation getting that high while people were suffering would become a serious political issue."

It's hard to imagine this happening for any sustained length of time. Prices can only rise if people are buying goods at those prices. (If you run a business, you wouldn't increase prices to a point where no one bought anything from you and them keep them there until you go under.) You could see prices rise in certain sectors (see oil, 2008), but if oil prices stayed that high for any length of time people, to the extent they needed oil, would cut back on purchases of other items in the CPI basket in order to buy it, which would attenuate the rise in the CPI.

Sustained inflation cannot happen without sustained inflation of wages - wages are the means with which people are able to pay higher prices. Periods of high inflation such as the 70s were characterized by high inflation of wages.

"In my view, stabilizing Nominal GDP is still a form of central planning that has a deep impact on real/microeconomic issues. A 5% NGDP target builds in a fair amount of inflation each year. "

If you take this view, what you are really criticizing is having a central bank with a monopoly on issuance of base money. But you criticism would apply to any and every monetary regime run by a monopoly central bank.

It's not much of a basis for arguing that one central bank monetary regime is superior to another.

In this case, there's a strong argument to be made that NGDP targeting would have less impact on the real economy than any other central bank monetary regime that has been tried or seriously contemplated.


Farid Elwailly writes:

Nice post. Thanks.

You should give a name to the quantity 1/NGDP, "Savings Share" for example to imply the share of the economy a saved lump sum of dollars is worth.

Then you could advocate level targeting the Savings Share and it would mean something concrete to someone with a dollar savings account balance.

TallDave writes:

Comparisons of 2014 RGDP with 1954 RGDP are extremely problematic due to the cumulative hedonic adjustments, to say nothing of times earlier and places more distant.

John Becker writes:

Michael Byrnes,

"If you take this view, what you are really criticizing is having a central bank with a monopoly on issuance of base money."

Exactly right. I don't get why otherwise sane economists keep supporting this insane system. Also, I don't think that one fiat money government monopoly plan such as inflation targeting is better than another like NGDP targeting.

Scott,

I'll address point two first because there seems to be something weird about it. You said, "People often make the exact opposite argument, asking me if I think the Fed could get away with letting RGDP fall without doing more monetary stimulus."

If you're targeting NGDP and RGDP falls, don't you have to create more inflation by definition? Don't you have to do monetary stimulus to get some type of inflation? I get what you're saying that people would want the Fed to "do something" if RGDP declined but they would have to "do something" to keep NGDP stable.

My view on the great depression is that the Smoot-Hawley tariff-important dates in the passage of this bill coincide with stock market declines as predicted by the EMH-particularly hurt farmers who were tied to small, undiversified midwestern banks. This set in motion bank failures that the Fed could do little to stop at the time. I see the correlation going from bad supply side policy to Nominal results rather than the other way around.

Comments for this entry have been closed
Return to top