The “animal spirits” fudge factor
Marcus Nunes directed me to an interesting quotation from Bill Gates:
Too bad economists don’t actually understand macroeconomics,” . . . It’s not like physics where you take certain inputs and you predict certain outputs. Will interest rates ever return to normal, and why aren’t they returning to normal? You won’t get a consensus between economists quite the way that if you dropped a ball out your window and called up physicists and asked, ‘What the hell happened?’ There’s so many factors including what [economist John Maynard] Keynes called ‘animal spirits’ in the economic equation that we don’t have predictability. Even today, people are still arguing about what happened in 2008. So it’s even harder to look forward. [Look at] the role of the bond rating agencies in 2008, which is completely unreformed. Why would that be? Well, there must be a lack of consensus.
I like Bill Gates, but this is an example of how people worth $100 billion get listened to on subjects outside of their area of expertise. The physics analogy is silly, and proves exactly the opposite of what Gates assumes. “Macrophysics” (predicting complex physical systems such as hurricanes and earthquakes) does just as poor a job of prediction as macroeconomics.
Nonetheless, Gates is onto something here. Macroeconomics really does have problems that macrophysics does not, and “animal spirits” lies in the center of the confusion. While we cannot predict earthquakes and hurricanes with any precision, we do understand the forces that cause them. I’d argue that most economists don’t understand what caused the 2008 recession, even a decade later.
In my view, most American business cycles are caused by demand shocks, i.e., changes in current NGDP. And those demand shocks are caused by unstable monetary policy, i.e., changes in expected future NGDP. While most economists agree with my claim that US business cycles are mostly caused by demand shocks, they disagree on the claim that the root problem is unstable monetary policy.
Because most economists don’t know how to recognize unstable monetary policy (due to their “reasoning from price changes”), they follow Keynes in putting an “animal spirits” fudge factor into their models. Their models can’t explain why aggregate spending falls, so they assume that some sort of mysterious pessimism has suddenly developed in the minds of millions of entrepreneurs.
Because of the law of large numbers, aggregate business and consumer psychology doesn’t suddenly change for no reason at all. The real problem is modeling the policymakers. We don’t know exactly how the 12 members of the FOMC will behave. Of course Trump is even more of a wild card, on the “real business cycle” side of things.
When macro policy is relatively sound (as during the Great Moderation), there is less emphasis on fudge factors such as animal spirits. When it is not sound, as in the post 2008 period, then voodoo economics takes over.
PS. Here’s another way to make my point about physics. Physicists insist that the behavior of all particles can be explained by the laws of physics (or else is random.) In that case, any failure to accurately predict any physical system that is not due to quantum uncertainty must be a failure of physics. These comparisons with physics are beyond stupid. It would be like a mathematician who studies basic arithmetic claiming his field was more successful than the study of prime numbers.
PPS. The Fed cut its fed funds target by 1/4% today. Yawn. (Should have cut by 50 basis points.)
Update: I’m no expert on reserve management, but this old George Selgin post is looking increasingly prescient.
Tags:
18. September 2019 at 13:10
I don’t see how anything you said disagrees with the Gates quote.
He says: “macroeconomics . . . It’s not like physics”, and you point out that macroeconomics is actually a lot like “Macrophysics”. But I’m sure Gates would agree with that. I don’t think he meant to include “macrophysics” in the contrast between macroeconomics and physics (especially since “macrophysics” is usually referred to by other names, such as meteorology). So yes, macroeconomics is like macrophysics, but unlike micro physics (aka physics), micro chemistry (aka chemistry), etc.
18. September 2019 at 14:22
“While we cannot predict earthquakes and hurricanes with any precision”
We certainly can predict hurricanes with a certain level of precision.
18. September 2019 at 15:10
Jeremy, Gates said:
“It’s not like physics where you take certain inputs and you predict certain outputs.”
Really? Physics can predict outcomes in highly complex systems?
LK, You said:
“We certainly can predict hurricanes with a certain level of precision.”
Once they’ve started!
18. September 2019 at 15:15
[…] or nominal GDP growth,” Sumner believes. Nonetheless, for his own reasons, Sumner declared on his blog today both that he considers Fed actions like this a feckless “yawn” (he thinks Fed moves […]
18. September 2019 at 15:19
[…] or nominal GDP growth,” Sumner believes. Nonetheless, for his own reasons, Sumner declared on his blog today both that he considers Fed actions like this a feckless “yawn” (he thinks Fed moves […]
18. September 2019 at 15:29
[…] or nominal GDP growth,” Sumner believes. Nonetheless, for his own reasons, Sumner declared on his blog today both that he considers Fed actions like this a feckless “yawn” (he thinks Fed moves […]
18. September 2019 at 16:03
[…] or nominal GDP growth,” Sumner believes. Nonetheless, for his own reasons, Sumner declared on his blog today both that he considers Fed actions like this a feckless “yawn” (he thinks Fed moves have less […]
18. September 2019 at 16:06
Let’s take the physics analogy even further and ask why their models can’t predict the behavior of a complex physical systems like the U.S. economy?
18. September 2019 at 16:52
I do not disagree with anything in this post, but I would like to see some commentary on the fact that the Federal Reserve just digitized $125 billion to buy securities, in a two-day period.
In the old days, the Federal Reserve would engage in a QE program of about $50 billion a month.
I understand that the purpose of the present QE program is to provide liquidity to a complicated (and evidently fragile) interbank lending system. Yet is not QE QE?
If a $50 billion a month QE program is stimulative, what are we to make out of $125 billion in QE in two days?
18. September 2019 at 18:56
Scott,
” “Macrophysics” (predicting complex physical systems such as hurricanes and earthquakes) does just as poor a job of prediction as macroeconomics.”
Bingo. There is a whole book along that kind of reasoning, called “How the laws of physics lie” by a physics professor, Nancy Cartwright. She even starts with a perfect reflection of Bill Gates’ example, except that it shows that physics has the same problem: “Imagine you go to a tall tower and drop a dollar bill. Can physics predict where it will land with any more precision than the social sciences can predict human beings’ behavior?” (quoting from memory – this is not the exact phrase she’s using). Point being, the laws of physics face the same problem as economics, i.e. fundamental laws are well understood when studied in isolation, but complex behavior in complex real life situations are just as hard to model in physics, as they are in economics. Let’s not even go to the three-body-problem.
This nonwithstanding your point on misunderstanding aggregate demand, where progress may be made with simple means after all.
https://www.amazon.com/How-Laws-Physics-Nancy-Cartwright/dp/0198247044
18. September 2019 at 19:51
mbka: I like your example, on the other hand one could probably develop a bell curve of where a dollar bill would land after leaving a 100-foot-high tower.
What looks like chaos in physics can often be reduced to probabilities, with enough observations.
I hate to say it, but in macroeconomics we seem to be moving towards less confidence in our probable estimations, rather than more. Also, in physics, there is not often a political component coloring observations.
18. September 2019 at 20:03
[…] or nominal GDP growth,” Sumner believes. Nonetheless, for his own reasons, Sumner declared on his blog today both that he considers Fed actions like this a feckless “yawn” (he thinks Fed moves […]
18. September 2019 at 20:33
Benjamin Cole,
point taken on probabilities, but this does not always work. The dollar bill example was well chosen because air turbulence is chaotic, and likely the folding behavior of the bill too, so you may not end up with a normal distribution, but with flat one… In addition, we’re interested in prediction of this one special bill – so the main takeaway here is that even in macro every day physics, we can’t predict simple behavior of simple objects unless we abstract away the real world.
I like the example for another reason: while physicists may struggle to determine where the dollar bill will land, sociologists will have a near 100% accuracy in predicting what will happen to the bill once it lands on a busy sidewalk [joke].
18. September 2019 at 22:27
mbka: where will the dollar bill land? I feel 100% confident in my prediction that the dollar bill will land on the ground (barring nearby bodies of water).
19. September 2019 at 02:37
Now the Fed plans another $75 billion in Treasury bond-buying, bringing a three-day total to $200 billion.
So…what is the Fed’s monetary policy again? Is QR stimulative? Or does $200 billion in 72 hours count for nothing?
If it counts for nothing…is this not a great way to reduce the national debt?
Or…is this MMT in drag?
19. September 2019 at 05:45
What, no one is going to make a joke about NOAA saying that it was reasonable to expect Hurricane Dorian to potentially hit Alabama? I mean, this is a leading national meteorological agency being laughably wrong and contradicting the National Weather Service! And this is not even to mention the disagreement about the state/cause of global warming among leading policy-makers who are in a position and have a responsibility to be highly informed.
It seems to me like an important difference between macroeconomics and macrophysics, at least until the past few years, was that macrophysical characteristics were relatively unpoliticized, while macroeconomics has not really ever been free of ideological taint.
19. September 2019 at 06:36
LK Beland: “We certainly can predict hurricanes with a certain level of precision.”
S Sumner: “Once they’ve started!”
I agree. 2 points:
1- Economics cannot forecast the length/depth of recessions–once they’ve started–to a degree comparable to physics-based forecast of hurricanes.
2- Physics-based weather forecast continuously improve, thanks mostly to improvements in computational power–smaller mesh-sizes and whatnot. To a large extent, we know how to solve the macro-physics problem in terms of theory. We simply lack the computational resources. The same cannot be said of macro-economics. More computing resources would not increase the accuracy of economic predictions.
19. September 2019 at 08:15
Of course, the macro-physics and macro-economics problem have a fundamental difference. NOAA-based and market-based weather predictions do not influence the trajectory of hurricanes. On the other hand, Fed-based and market-based predictions have an effect on the trajectory of recessions.
19. September 2019 at 08:46
Dr. Sumner,
Is not a reduction in expected NGDP essentially the same as the development of mysterious pessimism? “The market” develops expectations based on monetary policy trajectories (maybe your quibble is with the term ‘mysterious’), but once that trajectory doesn’t look as sufficient as before, expectations adjust…pessimistically.
And in a world of primarily demand shocks, what is the economic benefit of NGDPLT over inflation targeting- just “objectivity” of the measure? (I understand you’ve also raised non-economic benefits in terms of public understanding and so on)
I feel like your mental model of the macroeconomy is rather in line with the orthodox view, just with a different preferred target for the Fed.
19. September 2019 at 10:35
Dollar bill drop: I bet physicists could do a decent job of predicting the shape of the statistical distribution parameterized by some variables: wind speed and direction, shape and height of the tower, lay of the surrounding land, type of foliage, position and orientation of other buildings, fences etc… and all without ever dropping an actual dollar bill.
Earthquakes: we can predict where they will be (ring of fire, etc) and probably the long term statistics of their magnitudes, etc, parameterized by location. Similar statements can probably be made regarding Hurricanes. And in addition, once a Hurricane starts we can make some OK predictions about where it will and won’t go, how long it will last, what will happen to it’s power once it makes landfall, etc.
Can we do anything comparable regarding macro? I don’t know.
If we were to select many small teams (say of 3) of physicists, meteorologists or geophysicists at random with the requisite expertise and ask them to make these predictions, would they come up with similar results? Or would we find many of these teams dominated by “schools of thought” which cause the teams predictions to vary wildly from each other?
Now repeat for macro economists.
Any differences?
19. September 2019 at 14:24
If you hugely stretch analogizing our economic system to our climate system, you could say that EPA is to our climate system what the Fed is to the economic system. To make the analogy better you’d need to give the Fed the power to pump sulfur dioxide into the atmosphere, seed clouds, sequester carbon dioxide, (insert favorite pet AGW theory here) to make the EPA’s powers more analogous to the Fed’s power to conduct Open Market operations. After you gave the EPA those powers, you’d start to hear physicists bickering like economists about how much sulfur dioxide to pump etc. to hit temperature targets because they would forever be missing their targets.
(For my part, I don’t think we need to do any of these things about AGW, which I guess makes me the equivalent of a Free Banker.)
19. September 2019 at 14:48
Bill’s specific statement is wrong, but his general hunch is right. He should have said: Economists often don’t even agree on fundamentals. So it’s even worse than Bill suspected.
19. September 2019 at 16:08
The financial media reports that the US Federal Reserve will buy another $75 billion in Treasuries Friday in so-called repo. This is about $275 billion in 4 days. That equals about six months of QE in the old days.
Another way to look at it, we have just transferred one quarter of this year’s deficit to our central bank and interest payments from that debt will flow back to taxpayers. That happened in just four days.
Chairman Powell has more or less admitted the Fed will not ever sell these bonds back, in that he has said the central bank balance sheet will resume growth.
Who knew that the MMT crowd would rule America?
19. September 2019 at 19:44
Jeff, Yes, that’s what I was trying to get at in my PS.
mbka, I like that example.
Derek, I wish I had thought of that.
LK, If we could predict recessions then we could prevent them. And in fact recessions are becoming less frequent.
Daniel, Whether it is “essentially the same” depends on the policy implications of your view. If the policy implication is “use monetary policy to target NGDP and don’t use fiscal policy” then I’d say they are the same. But I don’t see many economists saying that.
20. September 2019 at 09:07
I see a few problems with the field of economics.
Economics is highly mathematized which creates an illusion of rigor and suggests a precision that does not actually exist.
Models are based on parameters that cannot actually be measured. What is the marginal propensity to consume? But, it drives the Keynesian multiplier (which also cannot be measured). The Keynesian multiplier is estimated to be somewhere between slightly less than 0 to significantly greater than 1.
Old theories don’t die, e.g. the Phillips Curve. In part, they cannot be definitively falsified. But, if a proposition cannot be falsified, it is not a valid hypothesis.
For the physicists, they have more humility. They are quick to acknowledge the limits of their models. Physicists have a better appreciation for chaos. They admit that they cannot model the 3 body problem.
20. September 2019 at 09:18
@Benjamin Cole
Well he should say that. But he’s not credible, hence no inflation.
20. September 2019 at 10:24
Scott — Looking at the FRED NGDP numbers, the Fed expectations vs inflation graph, and the Fed inflation expectations graph, am truck by the fact while NGDP has been on a 4-5% trend Fed has consistently expected higher inflation/UE… small sample but one that seems to vindicate NGDP targeting as a tool, and perhaps even suggest they might be taking it more seriously.
20. September 2019 at 10:25
Sorry about that!
Scott — Looking at the FRED NGDP numbers, the Fed UE vs expectations graph, and the Fed inflation vs expectations graph, am struck by the fact while NGDP has been on a 4-5% trend Fed has consistently expected higher inflation/UE… small sample but one that seems to vindicate NGDP targeting as a tool, and perhaps even suggest they might be taking it more seriously.
20. September 2019 at 10:31
Also wanted to mention b/c it was surprising — since 2008 QE has already saved taxpayers around a trillion dollars in debt from revenues returned to Uncle Sam by Fed. That’s real money!
Also, I declare a tax-free government funded by QE to be the new libertopian ideal.
20. September 2019 at 12:08
Bill gates sees the economy in real time, practically, and the obstacles that this country faces. And he see’s a theory that is driving this country into bankruptcy. Look, I am not trying to troll you, but this country is over 20 trillion in debt. Debt to GDP is one of the highest in the world. You and your cronies have shipped jobs and wealth abroad for over 25 years and have significantly weakened this country. The discipline of economics is fairly new. And your generation is simply wrong. Forget about bill gates. Even a layman can come to the conclusion that your wrong. It is staring us right in the face. Your pride is simply blinding you. Any businessman is more of an expert than you. We don’t need economists. We were just fine with political theory, which covers a wide range of issues. We don’t need you guys to tell businessmen how the economy functions. That is like some taking the subway to work telling the boss how to run his business. You have no clue how to innovate and how to implement that innovation. You have zero practical experience in the real world where variables like “corruption” are accounted for in real-time. You just don’t get it, and the country would be in much better shape if people like you listened to people like bill gates.
21. September 2019 at 11:58
Xu, Your comment is so full of silliness that I hardly know where to begin. Our new “businessman” president is running up debt far more recklessly than any non-business president we’ve ever had, just to take one example.
As far as international trade shipping out jobs, that the typical comment you get from a business person who knows nothing about the subject.
You said:
“Any businessman is more of an expert than you. We don’t need economists. We were just fine with political theory”
Oh really? Like Mao’s political theory? Hitler’s? Stalin’s? Trump’s? Which brilliant political theory are you referring to?
As far as your defense of Bill Gates, do you even understand the difference between economics and business? Gates relies on Keynes’s theories–is Keynes and economist?
21. September 2019 at 20:16
Scott, as I tried to make clear in my original comment, the common use of “physics” does include weather forecasting. Physics generally studies relatively simpler, but easier to understand/model systems than other fields of science.
But even despite this, the answer to your rhetorical question “Physics can predict outcomes in highly complex systems?” is still yes. Physics can predict the outcomes of *specific* extremely complex systems, which are in practice extremely useful. For example, the behavior of rocket engines, or integrated circuits.