Was the recent excess inflation an aggregate supply or an aggregate demand shock? Tyler Cowen weighs in on the issue:
A year or so ago I recall telling Bari Weiss in a podcast that the inflation was perhaps half real shocks, half an aggregate demand problem. Don’t let the revisionists talk you into the hardcore RBC view!
I didn’t know that hardcore RBC types had a view on this question, but today I’m more interested in Tyler’s comment on aggregate demand.
Back in 2022, I also thought it was reasonable to talk about high inflation representing a roughly equal mix of supply and demand factors. As of today, supply problems have eased and almost all of the cumulative excess inflation since 2019 is demand driven—as NGDP has overshot its trend by roughly the same amount as prices have overshot their 2% inflation target path. These facts support Tyler’s claims about the excess demand. So why don’t all economists see things in this way?
Here’s where things seem to get weird. As far as I can tell, economists do not actually have any substantive debate about the existence or non-existence of a recent aggregate demand shock. There is pretty general agreement as to what has happened; the debate seems to be over the meaning of the specific term “aggregate demand”. Thus if I point to very rapid growth in nominal spending, those who disagree with me won’t deny that NGDP rose sharply, they’ll typically say that NGDP is not aggregate demand.
Others point to modest growth in real spending (which has been close to trend since 2019), as evidence that there was no big aggregate demand surge. I respond that real GDP is not aggregate demand, as a rise in aggregate supply also causes real GDP to increase.
In this imaginary debate, we aren’t disagreeing as to whether some well-defined concept like “aggregate demand” increased or did not increase, we are disagreeing over how to define AD. It’s merely a debate over semantics. But when I see these debates on the internet, I see little or no awareness that it’s merely a debate about semantics, not substance.
To be clear, I’m not saying the issue is purely subjective—“Just a matter of opinion.” We have dozens of economics textbooks with AS/AD diagrams. In that model, a sharp rise in NGDP is evidence of a rightward shift in AD. In contrast, a sharp rise in RGDP is not evidence of an increase in AD (rising AS also boosts output). So there really is reason to prefer NGDP over RGDP as a proxy for aggregate demand.
I suspect that most economists use neither NGDP nor RGDP as a proxy for AD. Instead, they have some sort of model of factors that they believe should make aggregate demand go up. If that model says AD should have gone up, then they assume that it did increase.
But that makes economics seem more akin to religion than science. Instead of just accepting as a matter of faith that certain fiscal and monetary policy stances ought to boost AD, why not look at actual empirical data and see if the model is correct? Did AD increase in response to stimulative policies? What does the data show? But this just puts us back in the original dilemma—how do we measure aggregate demand? How do we test whether fiscal and monetary policy boosted AD, if we cannot measure AD? And if we can measure it, then what’s the debate all about? Just look at the data to see if AD increased.
It is clear that if economists disagree as to whether there has been a big surge in AD, then they must, ipso facto, disagree as how best to define AD. One economist has a model showing a big rise in AD, while another has a different model showing no unusual surge in demand. In this case, each model represents a different definition of aggregate demand. Again, the debate is merely over semantics.
I believe the Bible says something to the effect:
Wherefore by their fruits ye shall know them.
Economists seem to believe we determine what happened to AD by its effects, but they cannot seem to agree as to what these effects would be. In that case, they don’t agree as to what AD actually is.
I suspect that Tyler would argue the real issue is not the role played by aggregate demand, rather the important issue is the role played by stimulative monetary and fiscal policy.
At first glance, that seems like a more tractable approach. We all accept the fact that NGDP rose sharply in 2021-23, but we don’t know if that increase was due to fiscal stimulus, monetary stimulus or a rise in animal spirits (say due to revenge spending after Covid.)
Unfortunately, that doesn’t actually help at all. I wrote an entire book explaining that economists frequently refer to the stance of monetary policy being “easy” or “tight” without having any sort of coherent definition of what they mean by “monetary policy”. All you’ve done is to replace a meaningless semantic debate over aggregate demand with a meaningless semantic debate over monetary policy.
I wish we lived in a world where all economists agreed that aggregate demand was nominal GDP, but we do not. I wish we lived in a world where all economists defined the stance of monetary policy in terms of whether the expected growth rate of NGDP was above or below the central banks implicit target, but we do not.
Most of all, I wish we lived in a world where economists showed some awareness that their debates over things like “aggregate demand” and “monetary policy” are actually largely a debate over semantics, but we do not.
Here are some debates that would actually be useful:
1. Was the rapid 2019-2023 growth in NGDP appropriate, or undesirably fast?
2. If NGDP growth was excessive, was there some alternative fiscal and monetary policy path that would have delivered appropriate NGDP growth?
3. If Congress was determined to do an excessively large fiscal stimulus, was there an alternative monetary policy path that would have offset this stimulus, delivering appropriate NGDP growth?
4. If there was a monetary policy path that would have prevented excess NGDP growth, should the Fed have adopted that policy in 2021?
Those are relatively clear and interesting questions with meaningful policy implications. In contrast, the entire “Was is supply or demand?” debate is so poorly defined as to be almost meaningless.
In my new book, I basically argued that, “The emperor has no clothes”. I called out the economics profession for making “monetary policy” a key part of macroeconomics, without ever offering a coherent definition of monetary policy. I could write another book offering a similar critique of “aggregate demand”.
I suspect that most economists would regard my new book as boringly pedantic. But the current very depressing debate over the role of aggregate demand in inflation shows that there’s never been a greater need to clearly define our core macroeconomic concepts. Until we do so, we’ll continue talking past each other. And non-economists will be justifiably scornful of a profession that cannot even answer a question as basic as “Was the recent inflation demand or supply-side?”
If the supply and demand model cannot answer a question that basic, then of what use is the supply and demand model? What is its purpose?
READER COMMENTS
JP
Jan 23 2024 at 10:17am
I know your point is broader but Cowen and Tabarrok define aggregate demand in their textbook as “all the combinations of inflation and real growth that are consistent with a specified rate of spending growth,” which seems to be nominal GDP given that they use a dynamic Exchange Equation to arrive at this.
John Hall
Jan 23 2024 at 12:15pm
I’m sympathetic to this point. In the AS/AD framework, NGDP is determined through the interaction of the Aggregate Demand and Aggregate Supply curves. Aggregate demand isn’t literally NGDP in this framework. I just assume that whenever Scott says aggregate demand, he usually just means NGDP.
Scott Sumner
Jan 23 2024 at 11:35pm
According to JP, Cowen and Tabarrok use the monetarist definition of AD, in which case NGDP is determined solely by AD, and AS has no impact on NGDP. Think of it as a rectangular hyperbola in the P–Y space.
John Hall
Jan 24 2024 at 11:44am
After thinking on it for a while, you’re basically right that “Again, the debate is merely over semantics” so I have nothing much more to say…
Lizard Man
Jan 23 2024 at 1:15pm
Global respiratory pandemics with mortality rates similar to COVID seem like fairly rare events. The US Fed decided to err on the side of NGDP growth being too high instead of too low. That seems like a good choice to me, given that there should be some level of uncertainty of the impact of a central banks actions during a once per century event, and because there is a ton of public pressure to bring down inflation if it is too high, but very little public pressure to correct for inflation that is too low. If the Fed had a NGDP level targeting regime in place before the pandemic, I think that would have reduced uncertainty about the Fed’s ability to maintain steady NGDP growth, but not entirely eliminated it. I think that the Fed performed pretty well.
Scott Sumner
Jan 23 2024 at 11:41pm
Even if you are right about the Fed doing the right thing (and I don’t agree), it has no bearing on this post. Right or wrong, the Fed caused the high inflation by boosting AD.
Lizard Man
Jan 24 2024 at 11:16pm
I was responding to questions 1 through 4. It does seem strange to me that economists would have different definitions of aggregate demand. My assumption is that professionals use jargon in order to communicate more precisely and efficiently to other professionals. So it seems weird to me that economists would keep on using the term “aggregate demand” if they disagree on the definition of the term, instead of coming up with new jargon to precisely delineate what it is that they are talking about. Do you have a sense as to why economists aren’t using more precise language if the term AD has a contested definition? Do they just not realize that different economists mean different things when using the term?
Scott Sumner
Jan 24 2024 at 11:31pm
I find this puzzling, as the same ambiguity applies to “monetary policy”. The profession doesn’t seem to be aware of the problem, as far as I can tell.
vince
Jan 23 2024 at 5:17pm
What’s wrong with the Wikipedia definition? … the amount of goods and services that will be purchased at all possible price levels. … plotted with real output on the horizontal axis and the price level on the vertical axis.
The article includes this criticism: Austrian theorist Henry Hazlitt argued that aggregate demand is “a meaningless concept” in economic analysis.
Scott Sumner
Jan 24 2024 at 11:34pm
That’s a verbal description of the AD curve. But it’s so generic as to be almost useless.
vince
Jan 25 2024 at 12:39am
Say a country produces 100 widgets priced at $10 each. Based on production, income, or expenditure, NGDP is $1000. If no one wants to buy the widgets at any price, is NGDP still $1000? Is AD zero?
Jerry Melsky
Jan 23 2024 at 5:38pm
What if, in a candid moment among friends, Powell were to see your questions 1-4 and say:
Look, in an ideal world we wanted to get NGDP back on the pre covid trend quickly with minimal overshoot. But lacking a crystal ball on just when the Fed should begin a monetary policy that started working to override fiscal policy, we decided that a little more inflation (OK, quite a bit more) than we would ordinarily tolerate isn’t such a bad thing, especially given the huge debt burden on the Federal government. But, don’t ever expect us to admit that in public. That would be political suicide. We knew as we were watching NGDP nose above trend in 2021 that we could always find cover by saying that the amount of the aggregate supply shortfall and the speed of it’s recovery was impossible to forecast given the unprecedented nature of the global pandemic.
Scott Sumner
Jan 24 2024 at 1:11pm
Powell has admitted that they screwed up.
Kenneth Duda
Jan 23 2024 at 7:32pm
Scott,
At the risk of lecturing someone who knows 100 times more economics than I…
If I’m understanding, then I think(?) most economists would agree that:
the aggregate demand curve is a function AD(P) expressing quantity demanded (total quantity of real goods/services purchased) as a function of the price level
the aggregate supply curve is a function AS(P) expressing quantity supplied as a function of the price level
Y (real output) and P are determined by the intersection of the AS and AD curves
Y*P = M*V = NGDP
So the question is, what does “aggregate demand” mean in a claim like “excessive aggregate demand causes excessive inflation”? I would assume that it means that given the shape of the AD and AS curves, P is going to increase more than desired (we’re going to exceed our 2% inflation target), so someone better do something (like tighten the money supply) to shift the AD curve to the left. Here, “aggregate demand” is shorthand for “the shape of the AD curve given everyone’s expectations of everything”. Then, people will blame inflation on “demand” or “supply” depending on whether the source of blame is the kind of thing that shifts AD right or shifts AS left. Like, an effective oil embargo shifts AS left, and the Fed credibly announcing a permanent 4% inflation target shifts AD right.
Since the AD curve is not observable, claims like “there’s been a big surge in aggregate demand” can be supported by noting a big/sudden increase in the price level without much change in real output (say, a modest increase along trend, matching population growth). You can infer there was no significant AS curve movement from the lack of change in real output (if we can assume we’re near the vertical portion of the AS curve). We reason that if the inflation had been due to the AS curve shifting left, then real output would have dropped from trend. So the increase in P must have been from the AD curve shifting right, i.e., “excessive aggregate demand”. I think this framing is coherent.
In other words, I completely agree when you write “I suspect that most economists use neither NGDP nor RGDP as a proxy for AD.” Indeed. I think by “AD” they mean the full aggregate demand curve, not a single scalar quantity like NGDP or RGDP, and the phrase “increasing AD” means an AD curve shifting right.
In this framing, one could view the standard model as something like, “the Federal Reserve should run monetary policy so that the AD curve is continually getting a little closer to the AS curve,” meaning if AS shifts a little to the right (population/productivity growth), the Fed should move the AD curve even more to the right, so that P is always increasing steadily. If the AS curve shifts left, then the Fed should yank the AD curve left to preserve P — who cares about Y. This is not my policy preference — I agree with you that we are all better off if the Fed stabilizes P*Y (total nominal income) instead of stabilizing P alone. Of course, we can only have that debate coherently if we can agree on what things like “aggregate supply” mean.
Thanks,
-Ken
Scott Sumner
Jan 23 2024 at 11:59pm
Ken, Good comment. Two points:
If AD is viewed as a rectangular hyperbola, then AD is simply NGDP. That’s my preferred approach, and also the approach used in some textbooks (like Cowen and Tabarrok.)
As you correctly say, most economists assume a different AD curve, where AS changes will impact NGDP. But even in that case, economists need to find real world data points to ascertain what must have happened to AD. As you note, they might look at changes in P and changes in Y and make inferences about what happened to AD. I agree with all of that. My point is that when mainstream economists do exactly that, they reach radically different conclusions as to what happened to AD in recent years, even though they are all looking at exactly the same government macro data. That can only mean one thing—these economists each use different formulas when inferring what happened to AD, different weights on P, Y and other variables. But their formulas are essentially definitions—i.e. they are defining AD in radically different ways, and thus are talking past each other. That’s my criticism.
marcus nunes
Jan 24 2024 at 3:07am
It´s not an either or (supply or demand) question.
(44) Figuring out why 2023 was so surprising – by Marcus Nunes (substack.com)
Rajat
Jan 24 2024 at 7:55am
I can’t help thinking that apart from practising academic macroeconomists (like Reis), most economists still think according to a Keynesian framework, either old or new. In old Keynesianism, AD is a real variable, as Ken says. The AD curve slopes down against price, because in an old Keynesian model, higher prices mean a lower real money stock, which means higher real interest rates and lower investment. Inflation happens when real AD exceeds productive capacity (AS). Almost all economic journalists and finance- or lawyer-type generalists (eg Christine Lagarde?) think like this. In New Keynesianism, similarly inflation follows low unemployment. As far as I can tell, many models have terms for real output, interest rates etc but don’t have a variable representing nominal demand. It’s depressing, but also empowering if one feels like one understands the way things really work.
Scott Sumner
Jan 24 2024 at 1:10pm
“Inflation happens when real AD exceeds productive capacity (AS). Almost all economic journalists and finance- or lawyer-type generalists (eg Christine Lagarde?) think like this.”
OK, but lots of Keynesians are denying that the recent inflation is caused by AD. That means they are clearly not “thinking like this”.
Again, I’m not complaining about the shape of the Keynesian AD curve, I’m complaining that mainstream economists who are debating whether inflation was caused by AS or AD are engaged in a meaningless exercise, as there is no generally accepted definition of AD. If there were, there would be no debate. Right?
spencer
Jan 24 2024 at 11:30am
Aggregate supply is a component of aggregate demand. M*Vt = AD.
spencer
Jan 24 2024 at 1:48pm
N-gDp’s momentum during C-19 was fueled by a change in the composition of the money stock, more transaction accounts relative to gated deposits, increasing AD.
vince
Jan 24 2024 at 2:04pm
What’s wrong with the Wikipedia definition? … the amount of goods and services that will be purchased at all possible price levels. … plotted with real output on the horizontal axis and the price level on the vertical axis.
The article includes this criticism: Austrian theorist Henry Hazlitt argued that aggregate demand is “a meaningless concept” in economic analysis.
Philippe Bélanger
Jan 24 2024 at 5:11pm
Like Kenneth Duda above, it seems to me that the most straightforward way to think about aggregate demand is to define it analogously to how we define demand in microeconomics, which is essentially a counterfactual notion: how much would the population spend if the price level had a certain value. I can’t see what is gained by defining AD as NGDP. Under that second definition, a claim like “NGDP rose because AD increased” becomes a tautology. Whereas under the first definition it becomes an empirical claim that offers an explanation as to why NGDP rose.
I am also inclined to say that the interesting question is the role played by fiscal and monetary policy. But I disagree with the idea that this replacing “a meaningless semantic debate over aggregate demand with a meaningless semantic debate over monetary policy.” That is only true if you carry out this debate in terms of easy or tight monetary policy. But the fact that these terms have no clear definition is a good reason to avoid using them. Instead, we should talk about monetary policy in a clear way by describing it in terms of the interest rate of reserves or the balance sheet of the Fed, in such a way that there is no semantic ambiguity over what we are talking about. “Would inflation have risen as much as it did if the Fed had increased its policy rate earlier?” is a perfectly sensible question about which we can have a substantive debate. And of course the same applies to fiscal policy.
Philippe Bélanger
Jan 24 2024 at 9:33pm
I typed this a little too quickly. Obviously, aggregate demand should tell us how much real output would be bought (not how much would be spent) if the price level had a certain value.
Scott Sumner
Jan 24 2024 at 11:41pm
I disagree on both points. The micro analogy doesn’t work because demand curves for individual goods slope downward due to the substitution effect. But there is no substitution effect for all goods. In that case you are left with only the income effect, and an AD curve that is a rectangular hyperbola—the AD=NGDP “tautology”.
On monetary policy, the current setting of the fed funds target or the current setting of bank reserves plays at best a trivial role in the overall setting of monetary policy. Monetary policy is 99% forward guidance.
Philippe Bélanger
Jan 25 2024 at 6:52pm
I don’t think we can infer that the AD curve is a hyperbola from the fact that there is no substitution effect. That’s only true if we assume that people’s nominal spending doesn’t change if the price level changes, but I don’t see why we should assume this. And I don’t think the absence of a substitution effect means that the curve isn’t downward sloping. Suppose that expectations about the long-run price level are anchored at the current level. Then it’s plausible that people would wisely reduce their spending for a while if the price level increased temporarily. (Which is really a form of inter-temporal substitution.)
I agree that monetary policy is mainly about expectations (although I think the current setting of monetary policy has an important impact on expectations about future monetary policy). But I don’t see why this means that a debate over monetary policy must be a meaningless semantic debate. We can characterize monetary policy in terms of expected NGDP or the expected price level, which are well-defined concepts.
Thomas L Hutcheson
Jan 24 2024 at 8:50pm
The whole question of how much of 2019-2023 inflation was due to excess aggregate demand and how much to negative supply shocks (inadequate aggregates supply) is, and so Tyler Cowen’s 50/50 answer is, nonsense. Since it is the Fed’s job to regulate aggregate demand, given any positive or negative demand or supply shocks, any amount of excess – if there was “excess” – was 100 % the Fed’s doing.
Summer tries to suggest some non-nonsense question in place of the nonsensical one that I like I’ll take a stab at answering them. [“I think” is implied throughout without my saying so again and again]
Was the rapid 2019-2023 growth in NGDP appropriate, or undesirably fast?
I’d like to rephrase that as was 2019-2023 growth in inflation undesirably fast? I fear that doing the analysis in NGDP terms (or “aggregate demand”) implies a one-good, one labor input, one good/wage relative price when that is necessary is disaggregation to the level that disparate sectoral shocks occur
The answer depends on what was desirable. The labor response to COVID both voluntary and involuntary was certainly a huge negative shock with differentiated impacts on different sectors and different wage groups. The “supply chain disruptions” were smaller negative shocks. The shift from demanding services to demanding goods was a positive demand shock for goods and as negative demand shock for services. Similarly supply change disruptions was felt more for some goods than others.
These shocks change the market clearing relative prices of virtually all prices and wages in the economy. Given that the prices of some good and some wages cannot adjust downward, how much increase in the prices of things that can adjust is necessary to achieve market clearance? It was “desirable,” the Fed’s job to engineer enough inflation to achieve market clearance; i.e. avoid recession. There is a NGDP growth that is the sum of the product of the changing market clearing quantities of each good and its changing price, but it’s just arithmetic, not a variable in the multi-good model.
If NGDP growth was excessive, was there some alternative fiscal and monetary policy path that would have delivered appropriate NGDP growth?
Again, I’d re-direct this to inflation. Everybody understands that inflation is an average of the increases in multitudes of prices. [Let’s leave fiscal policy out of this; inflation is what ever Fed policy makes it regardless of fiscal policy. The failure to finance expenditures with taxes instead of borrowing reduces private investment ad future real gdp, but has not effect on inflation.] The alternative monetary policy path was the Fed NOT delaying from September 2021 (when TIPS expectations first rose above target) until March 2022. In that scenario inflation would have peaked at a lower (but still over-target) rate, the Fed would not have had to raise the EFFR as high as it did and inflation would have returned to target sooner.
If Congress was determined to do an excessively large fiscal stimulus, was there an alternative monetary policy path that would have offset this stimulus, delivering appropriate NGDP growth inflation?
Yes. To have raised the EFFR sooner
If there was a monetary policy path that would have prevented excess NGDP growth inflation, should the Fed have adopted that policy in 2021?
Yes, that is its Congressional mandate
Comments are closed.