Expansions are getting longer, booms are not (plus Iran deal bonus)
By Scott Sumner
As the current US expansion approaches the previous record (10 years during 1991-2001), there is increased interest in the ascertaining the date of the next recession. In previous posts, I’ve cautioned readers to not assume a recession is inevitable in the next couple of years, despite the previous history of US business cycles. Two facts seem noteworthy:
1. Our expansions have recently been getting longer.
2. Other similar economies such as Australia and the UK have had far longer expansions in recent decades.
I’ve recently become interested in the issue of “soft landings”, the idea of reaching a low unemployment rate and staying there, without triggering high inflation. Can the Fed do it this time, as the UK and Australia recently did?
Of course “soft landing” is a vague concept, so let’s call a soft landing an extended period where unemployment is near the cyclical lows. I’ll define “near” as within one percent of the cyclical low point.
We’ll call this period a “boom”, which is not to be confused with an expansion. For instance, 2010-15 were years of expansion, but not boom. So how long have previous American booms lasted (using my definition)? It turns out that the answer is, on average, 2 years and 4.7 months. Or 2.4 years, if you prefer.
Here’s a graph of the US unemployment rate, with recession periods shown with grey bars:
Starting with the Korean War boom of the early 1950s, the booms lasted for 2.5 years (meaning 2 years 5 months), 2.6 years, 1.1 years, 4.4 years (the 60s and the longest), 1.4 years, 2.1 years, 1.0 years (not really a boom, just an interlude between the 1980 and 1982 recessions), 2.11 years, 3.5 years (the 90s), and 2.10 years (again, 2 years and 10 months).
Unlike with expansions, there’s really not much evidence that booms are getting significantly longer. If we are at the current cyclical low, then the current boom started in November 2016. More likely, the cyclical low for unemployment will end up being closer to the 3.4% of 1969, in which case the boom started in 2017. Thus it’s likely that we are at least 1.5 years into the boom. And booms last about 2.4 years on average. The challenge for the Fed is to keep the boom going, without triggering high inflation.
The current economy probably seems better to the public than to economists, because the public thinks in terms of levels (booms and bad times) whereas economists think in terms of growth rates (expansions and recessions). The job market is not improving more quickly than under Obama, but it’s certainly better—and that’s what people notice (not unreasonably.) Real GDP growth is somewhat quicker.
Speaking of politics, I have a quirky theory that you might be interested in. It’s possible that President Obama’s Iran deal cost Clinton the election, and it’s possible that Trump’s new sanctions on Iran are boosting GDP growth. The Iran deal contributed to sharply higher Iranian oil output, which helped depress global oil prices in 2015-16 (from roughly $100 to roughly $50/barrel.). My quirky hypothesis is that this helped the US economy, but hurt the blue-collar economy of Wisconsin, Michigan and Pennsylvania.
It’s pretty easy to argue that it helped the US economy overall, as real median family income in 2015 and 2016 saw some of the biggest increases in decades, partly due to low inflation caused by falling oil prices. No surprise there. What is a surprise is that these good times were accompanied by a sharp slowdown in RGDP growth, which had been at robust “Trumpian” levels around 2014-15. People forget that. (12-month growth peaked at 3.8% in 2015:Q1, and then fell sharply.) It’s really odd that Americans would do so well during a period like 2016, when RGDP growth was slowing sharply. Why?
Think about the Bastiat “broken window fallacy”. Residents of a country devastated by a hurricane might feel bad, while glass replacement companies in that country might feel pretty good. One problem with my theory is that the low oil prices of 2015-16 would seem to hurt Texas and North Dakota, not the electorally important Midwest. So I decided to look at data on manufacturing employment growth:
Manufacturing employment growth peaked at 1.77% in January 2015, and then plunged into negative territory in the fall of 2016. Hmmm, what else happened in the fall of 2016?
But what does this have to do with the Iran sanctions removal? Recall the subsequent fall in oil prices during 2015-16. With a slight delay, this devastated the US energy industry, just as manufacturing employment was also slowing:
This procyclical pattern is actually pretty weird. In the 1970s, high oil price would have helped Texas and hurt the manufacturing sector, as automakers got crushed. Today, the correlation is positive. Why?
My theory is that America’s manufacturing base has increasingly shifted away from consumer goods, toward bigger capital goods. In addition, we now have a vast fracking sector with a pretty elastic supply curve. In the 1970s, high oil prices just meant more rents for Exxon from its big offshore oil platforms. The rest of the country suffered. Today it means a frenzy of activity in the Permian basin and the Bakkan. As this occurs, manufacturers of capital goods in Wisconsin, Michigan and Pennsylvania ramp up production of steel pipes, pumps, portable generators, and all the other machinery used in the fracking industry. The link between manufacturing and energy has gone from negative to positive.
Just one more of the many lucky breaks that Trump received.
In America’s Electoral College system, the GOP can actually gain from losing a million votes in California and gaining 100,000 votes in the Rust Belt. Trump’s win was partly due to his getting votes in exactly the right places. So you can no longer look at the entire economy, you need to think about the energy/manufacturing nexus. With oil prices once again soaring, this factor again may help the GOP in the midterms, at least in the Rust Belt manufacturing sector. Employment in manufacturing is now growing at more than 2%/year, even higher than the 2015 peak. (So I have more than one data point!) I’m guessing that many of these workers are among the voters who swung toward Trump in 2016, even as he lost votes among well-educated suburban Republicans.
PS. As I said, this post is highly speculative. I’m actually not certain there’s enough construction of energy-related capital goods in the Rust Belt to make this argument work.
PPS. Although I favored Hillary over Trump, I still think the Iran deal was the correct move by Obama. I wish it were still in effect. Furthermore, I have no idea how Trump can justifying cozying up to Kim, even as the North Korean leader insists he won’t give up his nukes, while walking away from an Iran deal that seems to have at least slowed the Iranian development of nukes. If there’s a logical argument there, it’s way over my head.
Update: I probably should have noted that the question of the Iran deal’s role in the 2015 oil price drop is actually much less interesting that the macro implications for the US economy. The US fracking boom also contributed to the price drop. But the new Iran sanctions do seem to have pushed prices back up.