Scott Sumner  

There is no such thing as "global aggregate demand"

PRINT
Unlock the School Library... Am I Out of Date? I Don't Thin...

Aggregate demand is fundamentally a monetary concept, linked to each country's monetary unit. Only in a few cases, such as the eurozone, does it make sense to talk about aggregate demand as a regional concept. Indeed in the eurozone, aggregate demand is only useful as a regional aggregate; it makes little sense to talk about "AD in Germany" or "AD in Greece."

Many economists are confused about AD, and treat it as a real concept. Thus Keynesians often point to slow real GDP growth in places like Britain as evidence that "demand" is inadequate. But that's reasoning from a quantity change. If the aggregate supply curve shifts to the left, then real GDP will fall. But obviously that's not a decline in "aggregate demand."

Here's a Reuters article discussing the views of Raghuram Rajan:

"The current non-system in international monetary policy is, in my view, a source of substantial risk, both to sustainable growth as well as to the financial sector," Rajan told an audience of economists and investors in New York.

"I fear that in a world with weak aggregate demand, we may be engaged in a risky competition for a greater share of it," he added. "We are thereby also creating financial sector risks for when unconventional policies end."

. . .

Rajan, a former IMF chief economist, said that since "the spectre of deflation haunts central bankers," it is no wonder that developed countries do not want to settle for low growth "even if that is indeed their economy's potential."

Still they should not ignore their responsibilities to developing economies, he said, adding the IMF should be the arbiter of whether accommodative policies are "in- or out-of-bounds."


Rajan seems to imply there is such a thing as global AD, that it's a pie that countries can take a larger piece of by engaging in beggar-thy-neighbor policies, such as monetary stimulus. This is almost entirely wrong. There is no such thing as global AD, and monetary stimulus either enlarges global output (it's not a zero sum game) or has no effect, and merely results in domestic inflation, with no change in the real exchange rate. Either way, monetary expansion in country X is never going to steal jobs from country Y. And when those economies are artificially depressed by a combination of low NGDP and sticky wages, then the rest of the world benefits from monetary stimulus that boosts output closer to the natural rate. It's a win-win policy, not a beggar-thy-neighbor policy.

But once again, I can't emphasize enough that there is no such thing as global AD. If you read that phrase in the linked article and thought it meant something, then you need to go back to your old college EC102 textbook. Do you see that price level variable on the vertical axis of the AS/AD diagram? There's also no global price level, or global inflation rate. There is no such thing as global aggregate demand.


Comments and Sharing


CATEGORIES: Macroeconomics




COMMENTS (19 to date)
Matt Moore writes:

Hey Scott - I have a PhD in Econ, but I avoided macro as extensively as I could throughout my studies (for some good reasons and some bad ones).

Is there a single graduate textbook you could recommend that would illuminate these monetary issues? (inflation, exchange rates, interest rates, monetary policy).

Cheers,

Nick Rowe writes:

Dunno Scott. Suppose the prices of both apples and bananas doubles, in both dollars and euros. (2 goods, 2 monies, 4 prices, 1 world). Then we could say the world price level has exactly doubled too. Sure there's an index number problem, but there always is. So we could put world price level on the vertical axis, world RGDP on the horizontal, and draw a world AD curve.

Glenn writes:

Agreed with Nick, who beat me to it. This whole post strikes me as semantic and silly. Of course there is an aggregate demand. Denominate prices in terms of apples, instead of dollars, and one would have no trouble coming to a world aggregate demand in terms of apple-equivalents.

If there is no (world) aggregate demand, there is no real GDP, world or otherwise.

E. Harding writes:

"monetary stimulus either enlarges global output (it's not a zero sum game) or has no effect"
-I'm sure someone can come up with reasons for why sufficiently strong monetary stimulus could weaken output (I'm thinking a shock to savings behavior or costly changes in inflation expectations).

E. Harding writes:

@Nick Rowe
-That's a smart and counter-intuitive (though kinda silly) way to look at things.
"If there is no (world) aggregate demand, there is no real GDP, world or otherwise."
-It does not follow.

Scott Sumner writes:

Matt, I'm afraid I can't help you there, but Barro's intro grad book is a nice intro to new classical macro.

Nick, If you compute 1/P you have the "value of money." But which money? Both the concept of the price level and the concept of AD requires a medium of account.

Suppose Zimbabwe has 1 quadrillion percent inflation and is 1/10,000th of global RGDP. So Zimbabwe goes from having 1/10,000 of global NGDP before the hyperinflation to having 99.99% of global NGDP after the hyperinflation. What does that even mean?

Also see my reply to Glenn.

Glenn, I sure hope Nick doesn't agree with you because unless you measure prices in money terms then AD makes no sense.

Here's another way of making my point. The entire AS/AD model makes no sense without some form of wage or price stickiness. We generally assume wages and prices are sticky in terms of the medium of account. If you use a different medium of account then what's the point? So we've had 50% inflation in the past year in the US measured in gasoline terms. Sure, you could say that, but why? That's not the inflation rate that matters for the AS/AD model.

Nick Rowe writes:

Scott: 1/P is the price of money in terms of goods. The question "which money?" is no different in principle from "which goods?". Both are index number problems. We take a weighted average of all the Pij , where i is a good and j is a money.

I think I could answer your Zimbabwe question, if I had an RA. I would mess up the math. It's no different in principle to "what happens to the price index if the price of apples goes up by one trillion"? You get a silly answer if you use a fixed weight index like the CPI, but a more sensible answer if you use a Fisher ideal index (or something, IIRC, which I probably don't).

I don't have any Econ degree. Math B.A., here. Seems to me the definition of "demand" changes in the course of the extended discussion (from Econ 101 to Econ 666 or whatever). People start with some idea that "demand" means "want" or "desire" and end with the idea that "demand" means a physical thing which government agents can push into any shape they want. Hutt (__The Keynesian Episode__ criticizes the concept of "aggregate demand" at length). Attempts to compare "demand" from one person to another are iffy. Does it make sense to say that you like ice cream more than I like cake? Attempts to sum individual desires (let's add your desire for rocky road ice cream to my desire for choccolate cake) rely on wild assumptions.

I suspect that defense of "aggregate demand" proceeds from the pleasure that people derive from playing with pretty models. But then I don't have an Econ degree.

Harold Cockerill writes:

Is deflation a bar to growth? It seems to me a large increase in overall productivity would lead to prices dropping but that's not a defect. Of course deflation is bad if you owe a lot of money. Gee, who owes a lot of money?

Glenn writes:

"It does not follow."

Why not?

"Glenn, I sure hope Nick doesn't agree with you because unless you measure prices in money terms then AD makes no sense."

How does this make any sense? Prices in "money" versus "goods" terms as a distinction is alien to me. Money is a unit of account. There is ONLY a price in terms of goods, denominated in whatever currency is convenient and available. I learned in my 2nd econ course that I could index prices to anything I wanted, and denominate the price of everything else in the model accordingly.

"Here's another way of making my point. The entire AS/AD model makes no sense without some form of wage or price stickiness. We generally assume wages and prices are sticky in terms of the medium of account. If you use a different medium of account then what's the point? So we've had 50% inflation in the past year in the US measured in gasoline terms. Sure, you could say that, but why? That's not the inflation rate that matters for the AS/AD model."

Again, not sure I understand this either. Of course it matters. It doesn't matter very much, because non-perishable gasoline is not the typical store of value or unit of account, has substitutes, and is, what, 2% of the typical consumers basket? What if it were different? Suppose we'd had 50% inflation measured in, say, real property terms in the past year. Would that matter?

"Attempts to sum individual desires (let's add your desire for rocky road ice cream to my desire for choccolate cake) rely on wild assumptions. "

Aggregate demand does not mean what you think it means. Fix a price level. Now how much ice cream do you want (I bet if that price is positive, it is both finite and quantifiable)? Sum that across people. This is aggregate demand. AD doesn't reference a fairyland where resources are suddenly not scarce.

Scott Sumner writes:

Nick, OK, I see your point. You could construct some sort of index that would allow the calculation of AD (or perhaps first differences) at the global level, but I'd make the following arguments:

1. Even if so, it doesn't affect any of the substantive points I made in criticism of Rajan. AD would not be a fixed pie that countries could take more of via expansionary monetary policy.

2. Global AD would not be a useful concept. As evidence I'd point to the fact that I don't recall ever in my entire life seeing any economist or journalist refer to a global inflation rate. The Fisher ideal index would make the Zimbabwe problem smaller, but unless I'm mistaken it would not go away. Or perhaps I should say that Zimbabwe would make the inflation number rather strange looking.

Alternatively, imagine a world where 50% of the world had 2% deflation, and 50% of the world had 20% inflation. You could say that global inflation was 9%, but would that be a useful statement about the global macroeconomy? What would it tell us?

Malcolm, AD can be defined in many ways, I find the AD=NGDP definition to be the most useful.

Harold, I prefer to think of deflation as an outcome, not as the cause of macro changes.

Glenn, Sorry but I just don't agree. Money matters because prices are sticky in money terms not in gasoline terms. It has nothing to do with the amount of gasoline. In 2007 the monetary base was only 6% of GDP, but money was really important because everything was priced in those terms.

You said:

"Aggregate demand does not mean what you think it means. Fix a price level. Now how much ice cream do you want (I bet if that price is positive, it is both finite and quantifiable)? Sum that across people. This is aggregate demand."

The amount of ice cream you want at a fixed price level is quantity demanded. One of the most basic concepts in economics is that one should never confuse demand and quantity demanded.

Dave writes:

I'm probably missing something obvious to everyone else. Or maybe this is Nick's point in other terminology.

But, couldn't you use the purchasing power parity exchange rate to determine a measure of aggregate demand, in a standardised PPP 'currency'?

And world inflation could just follow the same PPP standardisation?

Prakash writes:

This sounds a bit like the moneyness problem at a global level. My instinct would be for granting harder currencies a little more weight and other currencies a lesser weight, but would have no way to quantify it.

My tendency is to agree with Prof. Sumner on the fact that there is no global fixed pie, but in the short term, the number of people with disposable income who can afford to buy the output of the world is not very variable. So, I don't think that Rajan's point is totally invalid.

The hard currency money CB's printing more money brings higher stock prices in the developing world, thus it could be said that the developed world's CB's effectively control the "global AD" or size of the pie.

Toby writes:

Scott,

Quick Question: Are you saying that AD would have no meaning in the US if instead of the dollar the US would adopt a regime with competing currencies?

Scott Sumner writes:

Dave, I think the problem here is one of utility. One can define AD any way one likes. You could define it as a real variable. The question is whether the definition is useful in terms of the AS/AD model. The only definitions that are useful in that model are monetary definitions.

Prices are not sticky in standardized PPP currencies, they are sticky in actual real world currencies. My point is that the AD used in textbooks refers to national AD, not global AD, and for good reason. Global AD would not be a very useful concept (except perhaps under an international gold standard, or something similar.)

Prakash, You said:

"My tendency is to agree with Prof. Sumner on the fact that there is no global fixed pie, but in the short term, the number of people with disposable income who can afford to buy the output of the world is not very variable. So, I don't think that Rajan's point is totally invalid."

You are missing the key point. Yes, AD may not be very variable in the short run, but to the extent it does vary it varies with monetary policy. But Rajan is imaging a changing monetary policy affecting the distribution of the pie, without changing the size of the pie.

Toby, Then you'd want to focus on the AD provided by each currency. Just to be clear, it would be OK to say "On average, individual central banks adopted policies that lowered AD during 2008-09, in many countries around the world. Thus global real output fell."

J.V. Dubois writes:

"Alternatively, imagine a world where 50% of the world had 2% deflation, and 50% of the world had 20% inflation. You could say that global inflation was 9%, but would that be a useful statement about the global macroeconomy?"

You can say the same about regions in US and/or about different goods purchased and sold.

But how can it be useful? Imagine that all countries experienced monetary contraction. Is it OK to say that AD in every country fell? Then we may say that Global AD fell.

Now imagine the world in 1930ies. Most countries had their local currencies tied to gold. Do you think there was no AD just because countries were using different currencies?

I can name a lot of different examples but I am more with Nick here. The shape of AD curve depends a lot on monetary policy. Based on this shape there may be some very interesting things happening. I think it is quite extreme to say that just because something you are interested in does not show as strongly (e.g. price sitckiness) then the whole concept does not exist or does not make sense.

Yes, the World AD now is very strange. It would look different if every country was on gold standard. Or if most of the relevant countries had their currency tied to US dollar.

But no, I do not think that global GDP is a fixed pie. But maybe this question alone should be argument enough to actually start thinking about it.

Brian Donohue writes:

Scott, you keep using extreme examples. What about a world where all countries generated 4% inflation versus one where all countries generated 2% inflation?

I'm inclined to say global AD/NGDP was 2% higher in the first case.

That seems much nearer to the reality among major economies.

Joe Eagar writes:

Aren't we really talking about global consumption demand here? I do think there is merit to discussing that, as nations seem to have a lot more trouble trading investments opportunities between them than real goods, so the distribution of consumption demand does seem to matter.

That said, it mystifies me when accomplished central bankers talk as if monetary policy has anything to do with that. The American CA deficit exploded in the 2000s not because the Fed raised interest rates, but because Asia was running a large savings surplus, arguably the result of deliberate government policies in East Asian countries.

I suppose it just isn't very dramatic when one says things like "it isn't fair to American workers when the Chinese government discourages Chinese workers from increasing consumption/getting pay raises [that they could use to buy our products]", so people talk about money printing instead.

Joe Eagar writes:

"as nations seem to have a lot more trouble trading investments opportunities between them than real goods, so the distribution of consumption demand does seem to matter."

Eek, I didn't quite condense this part right (I didn't want to post a long comment so I tried to condense everything). That's not supposed to be a declaration of fact. I do think that generally speaking, this tends to be true, but obviously you could navigate an oil tanker (or a certain East Asian economy populated by one billion people) though the holes opened by "generally" and "tends to be." I was trying to give a potential justification for why the distribution of global consumption might matter, not argue that cross-border capital flows are everywhere and always a bad thing.

Comments for this entry have been closed
Return to top