Alex Tabarrok and Tyler Cowen are doing a series of podcasts on the economy of the 1970s. A few weeks back, I commented on one of their previous podcasts, which discussed the tricky problem of establishing causality for changes in inflation. Their most recent podcast discusses oil shocks and the business cycle, an area where causality is even harder to establish:
TABARROK: Now, let’s talk about some puzzling economics because the price of oil goes up. War starts in October of 1973. The US goes into a recession in November of 1973. Unemployment doubles from 4.5 percent to 9 percent. Now, I think most of our listeners will say, “Well, what’s puzzling about that? Price of oil goes up and you go into a recession. That seems entirely normal.”
Yet for economists, this is still quite puzzling because even though oil is obviously of relative importance, it’s not that big a feature of the economy, and there in fact are pretty sophisticated theorems, which say that if you have—this is Hulten’s theorem—if you have a shock to a sector of say 10 percent, something goes up, productivity goes down 10 percent, price goes up 10 percent or something like that, and that sector is a relatively large share of the economy, say 5 percent, then the effect on GDP should just be those two things multiplied together. Ten percent times 5 percent, which is just 0.5 percent on GDP.
COWEN: Those theorems are wrong, right?
TABARROK: Yes.
The link between oil shocks and recessions seems pretty strong. And yet, I’m not entirely sure that those theorems are wrong. So how can we explain why recessions often follow oil shocks? Here are two possibilities:
1. Induced monetary tightening (a nominal shock.)
2. Reallocation of resources (a real shock.)
Oil shocks often occur at a time when the global economy is booming. In many cases, this is preceded by excessively expansionary monetary policy. In the short run, the oil shock makes the pre-existing inflation problem even worse. Monetary policymakers respond vigorously with tight money, slowing NGDP growth. With less nominal GDP and sticky nominal wages, unemployment rises sharply. I call this the musical chairs model of recessions.
In this scenario, the actual cause of the recession is tight money, but the oil shock partly explains why policymakers make this mistake. In a counterfactual scenario where NGDP keeps growing on trend, there is no significant recession after an oil shock.
In reality, oil price shocks can have an impact beyond their indirect effect on monetary policy and NGDP growth. As Arnold Kling has emphasized, the public will respond to sharply rising oil prices by re-allocating consumption and production toward less energy intensive parts of the economy. During the transition period, the unemployment rate may rise. This is a real shock to the economy, which can affect employment even if monetary policy maintains steady growth in NGDP.
How important is the real channel for oil price shocks? Later in the podcast, Alex and Tyler provide some suggestive evidence provided by the Ukraine War:
TABARROK: Yes. A lot of people, including German politicians, predicted that Germany would have to ration gas, that people would freeze to death, that the economy would go into a deep recession. In the end, the German economy adapted to a much lower supply of natural gas by using less and finding substitutes. The spot price of gas rose by a factor of more than eight at peak, but instead of price controls and rationing, the German government let the price rise, but they did protect German consumers with a lump sum transfer based upon the past use of natural gas.
That meant everybody had an incentive to listen to the signal of the higher price of natural gas. In the end, the German economy rode out this massive decline in the quantity of natural gas. To me, this is a sign that maybe economists at least have learned some lessons.
COWEN: I was shocked that went as well as it did. You may recall, I think it was Deutsche Bank forecast a major recession for Germany. I’m not sure they had a recession at all, but if they did, it was just a marginal recession, and they nailed it.
Tyler’s memory is correct; Germany had only a very modest rise in unemployment, from 5% to 6%:
Why were the pessimistic forecasts wrong? Why did Germany experience such a small rise in unemployment? Monetary policy in the Eurozone remained expansionary, allowing for a strong rise in NGDP:
In contrast, large increases in unemployment such as 1980-82 are associated with tight money policies that sharply contract the rate of growth in NGDP.
Non-economists tend to underestimate the extent to which free markets can find substitutes when one commodity becomes more scarce. (Even economists may briefly forget the importance of substitutes, before coming to their senses later in a podcast.)
One final point. In previous posts, I’ve argued that the number of people with the talent to become a great artist or scientist far exceeds the number that actually achieve greatness, mostly because you must also be in the right place at the right time. This conversation caught my eye:
TABARROK: Many of these lessons, which we’ve been talking about in the 1970s, you can say the 1970s led to Milton Friedman. Milton Friedman became a much more important spokesperson, representative of Free to Choose, and so forth, but Milton Friedman’s been dead for some time. People forget. People forget Milton Friedman, and they forget what caused Milton Friedman to come into being, which is all of the mistakes which we made in the 1970s.
COWEN: One of my takeaways is simply the 1970s was a great time to learn economics. The lessons were very visible.
TABARROK: Yes. I would put it the following way. I think Milton Friedman was not the smartest economist ever. Maybe that’s Ken Arrow, but Milton Friedman was right on the greatest number of things. The reason he was right on the greatest number of things was that he was lucky enough to come to fruition at a time where we were doing everything wrong.
COWEN: That’s right.
A very astute observation. Overall, a very insightful podcast.
READER COMMENTS
John R. Samborski
Oct 24 2024 at 8:39am
Mr. Sumner,
Can you give links to the posts where you argued that the number of people with the talent to become great artists or scientists far exceeds the number who actually do? Thanks.
Scott Sumner
Oct 24 2024 at 7:24pm
https://scottsumner.substack.com/p/italian-mannerism-david-lynch-and
David Henderson
Oct 24 2024 at 9:13am
Alex says that Hulten’s Theorem is wrong. I actually hadn’t known that literature. To say a theorem is wrong, you need to say what the wrong step in the proof is. Alex doesn’t say. Do you know why he thinks it’s wrong?
Craig
Oct 24 2024 at 9:51am
https://marginalrevolution.com/marginalrevolution/2017/12/not-gdp-equal.html
Tabarrok’s thoughts on Hulten’s Theorem
David Henderson
Oct 24 2024 at 9:58am
Thanks, Craig. So Alex is challenging its assumptions, which makes sense. I’m just not used to people saying that a theorem is false if it follows from the assumptions. But thanks for providing that clarification.
Student
Oct 24 2024 at 12:55pm
I think they make this way more complicated than it has to be, even though the issue remains the assumption that there are no spillovers. In other words, the change in productivity in one sector has no impact on the productivity of other sectors. That is a very problematic assumption (and in fact, is very reminiscent of a problem I often see in spatial econometrics where people estimate a spatial model and then use the betas as if it’s OLS. This mistake is essentially the same issue we see with Hulten’s theorem).
Let’s start with Hulten’s theorem (sorry the math looks ugly but formulas are difficult to do in this comment box. It looks a lot better in a properly formatted version):
delta ln TFP_agg = sum(i) [ s_i * delta ln TFP_i ]
Now, if a change in TFP in one sector (or firm) has an indirect effect (a spillover effect) on the productivity of other sectors (or firms), the simple formula becomes insufficient because it ignores the secondary effects of productivity changes spreading across sectors.
If we introduce spillovers, the aggregate TFP growth might look something like this:
delta ln TFP_agg = sum(i) [ s_i * delta ln TFP_i ] + sum(i) sum(j≠i) [ s_i * delta ln TFP_j * beta_ij ]
Where:
beta_ij represents the spillover coefficient from sector j to sector i.
The term sum(i) sum(j≠i) [ s_i * delta ln TFP_j * beta_ij ] captures the spillover effects across the economy.
In this case, the effect of a change in TFP in sector j is not just weighted by s_j, but also has a spillover effect weighted by beta_ij, the influence it has on other sectors i.
Examples of Spillovers in Practice
Tech Industry: If a tech company introduces a revolutionary algorithm, other firms in the same sector or even across sectors can benefit from the knowledge spillover, enhancing their productivity. We fully expect this to be true.
Automobile Supply Chain: A car manufacturer’s improvement in assembly line technology can increase the demand for higher-quality inputs from suppliers, leading them to improve their own productivity. We expect this is also true.
Regional Spillovers: Productivity improvements in one geographic region can affect neighboring regions, especially in areas with high labor mobility or trade connections. We know this to be true empirically.
A Two-Sector Counterexample to Hulten’s Theorem
To see this more concretely, consider a simple counterexample with two sectors (A and B):
Assume no spillovers: beta_ij = 0. Then Hulten’s Theorem holds, and:
delta ln TFP_agg = s_A * delta ln TFP_A + s_B * delta ln TFP_B
Now introduce a positive spillover from A to B with beta_ij > 0:
The observed TFP change in sector B is now: delta ln TFP_B + beta_BA * delta ln TFP_A.
The aggregate TFP change is now:
s_A * delta ln TFP_A + s_B * (delta ln TFP_B + beta_BA * delta ln TFP_A)
We can simplify this to:
(s_A + s_B * beta_BA) * delta ln TFP_A + s_B * delta ln TFP_B
The presence of the spillover term s_B * beta_BA modifies the contribution of delta ln TFP_A beyond its revenue share s_A, contradicting Hulten’s original formula.
Spencer Hall
Oct 25 2024 at 10:40am
The real world is distanced from the theoretical world by actual numbers.
Student
Oct 25 2024 at 7:14pm
They used numbers in the underlying paper. I am just saying their motivation for it was overly complicated. A spillover model is all you really need to show it.
Craig
Oct 24 2024 at 9:37am
Directly as energy goes up business income statements begin to experience some form of margin compression depending on how energy intensive they are. Gas at the pump is also a special price inasmuch as many buy it, its highly visible, ie when gas spiked in FL to $4.86 that might’ve cost me what? $5-$10 additional that week? Not thay big of a deal, but I remember seeing the price and being a bit rattled by it. As such I’d suggest it might have an outsized psychological impact that is difficult to quantify.
Scott Sumner
Oct 24 2024 at 7:26pm
But the business cycle is not driven by psychology, it’s driven by NGDP growth (i.e. monetary policy.)
Thomas L Hutcheson
Oct 25 2024 at 10:52pm
By the settings of the instruments aiming at (targeting) at inflation, i.e. by monetary policy.
Jim Glass
Oct 25 2024 at 2:46am
I have no opinion on who was “smarter”, but as Milton was born in 1912 and Ken in 1921, it doesn’t seem like their times were so different to me. Anyhow, there were a whole lot of other economists working at the same “lucky” time as Milton who didn’t produce anywhere near the same results.
One could make the same argument that Einstein was lucky to be working just when physics was arriving at the concepts of relativity and quantum mechanics, and Shakespeare was lucky to find himself located at the brief peak of the Elizabethan theater. But that doesn’t explain why they produced such a “greater number of things” than any of their equally lucky contemporaries.
Of course, to be “world best” at anything requires both a whole lot of both ability *and* luck. If Albert and Will had been born in farm country somewhere, we’d never have heard of them. Tom Brady easily could have been picked in the 6th round from Michigan by the Detroit Lions. If so, then after sitting on the bench behind the next year’s 1st-rounder Joey Harrington as the team lost 72 games in 6 seasons, today instead of being the GOAT part-owner of the Raiders with a $375 million Fox contract, he’d likely be the answer to a Lions fans’ trivia question.
TGGP
Oct 25 2024 at 1:13pm
Arrow is best known for his impossibility theorem on voting, a rather timeless thing. Friedman is best known (among economists) for monetarism, which was relevant when Keynesian demand-management broke down. So perhaps it’s not just about the time they were in but the subjects within economics they focused on.
Jim Glass
Oct 26 2024 at 5:54pm
Exactly. The point of the Brady example was that in addition to being at the right time and place there are always a host of choices, contingencies and random chance events that must break one’s way to become the “best ever”.
TGGP
Oct 25 2024 at 1:21pm
Also, William Shakespeare was not born in the big city of London, where he became famous. His father was a glover, but had started out as a farmer. So Shakespeare wasn’t from farm country per se, but just one generation removed from it. Among econ Nobelists, James Buchanan actually was born on a farm, hence the title “Better Than Plowing” for his collection of personal essays.
Scott Sumner
Oct 25 2024 at 7:08pm
“Of course, to be “world best” at anything requires both a whole lot of both ability *and* luck.”
I am pretty sure that was Alex’s point. Einstein and Shakespeare were obviously great minds, but also benefited from being at the right time.
Craig
Oct 26 2024 at 12:36pm
“Of course, to be “world best” at anything requires both a whole lot of both ability *and* luck. If Albert and Will had been born in farm country somewhere, we’d never have heard of them. ”
Indeed, Jim, and honestly just to expound on your point here a bit when you discuss luck here with respect to famous minds, I often wonder about that which is unseen, how many unlucky minds were ground to dust in wars. Right place, right time, right natural abilities and I’m not sure how to calculate the odds of being in the right place at the right time, but IQ is a distribution and we can say that out of 1 million people X% will have an IQ of 100, 110, 120, 130, 140, 150, etc. Quick google on Einstein IQ and I’m seeing 160-180. 160 is 3 in 100k, 180 is 3 in ten million. So when a war chews up millions of people we know some % have to be very talented. Could they have found themselves in the right place at the right time as well? We’ll never know…..
Spencer Hall
Oct 25 2024 at 10:45am
Isn’t the substantial substitutability of labor and products differentiated by the elasticity of demand and supply? Isn’t that how to measure shocks?
Thomas L Hutcheson
Oct 25 2024 at 10:47pm
A real shock calls for above target inflation to maximally facilitate the re-allocation of resources, to facilitate the adjustment of relative prices in the face of some prices being sticky downward. That is why the Fed has a Flexible Average Inflation Target, to allow for temporary busts of inflation above the targeted, forward looking average, as it did in 2021-2023 (but failed to do 2008-2020). Of course there is no guarantee that the shock may not me so large as to produce a fall in real income/rise in unemployment. The Fed can only minimize this.
Thomas L Hutcheson
Oct 26 2024 at 8:28am
The rest of the world should take a lesson from the German adjustment to Putin’s cut off of natural gas imports.
Allow prices to adjust to the tax on net CO2 emissions needed to achieve net zero and compensate consumers from the revenues of the tax.
Thomas L Hutcheson
Oct 26 2024 at 2:57pm
I expanded on these comments in my Substack post
https://thomaslhutcheson.substack.com/p/recession-and-real-shocks
Comments are closed.