Most industries have better and worse times, but when an industry matures to the point that it's mostly growing alongside GDP, and also has growth spikes and slumps that coincide with GDP, it gets tagged as a "cyclical." This can be slightly frustrating nomenclature, because some industries have a cycle that isn't related to the overall economic one: homebuilders, for example, can do well during a downturn if the policy response to the downturn is to cut real rates. Reinsurers have a cycle of their own, where high returns attract more capital that pushes down rates and lowers returns, but the return from a given year's decisions is realized at some point in the future when some new disaster hits. Energy is a tricky one, because there are also geopolitical and technological factors that affect them, both at the level of aggregate revenue and in terms of which companies face higher and lower costs.

The classic cyclical sectors are things like cars, steel, cement, home appliances, capital equipment. These tend to be either durable products or inputs into durable products, and that's the core of why they move with the economic cycle. A big driver of economic cycles is the credit cycle, but setting that aside and just thinking about the structure of the economy, and how changes in aggregate demand hit different sectors, it's straightforward to see why some companies would be more sensitive than others.

For household-facing durable goods, like cars and appliances, most of the demand will come from replacement rather than new users. That replacement can be absolutely required, as when someone's car gets totaled and they need a new one, but it can also be somewhat discretionary in the form of upgrades. During a recession, people naturally defer upgrades, and this propagates back through the supply chain: inventory piles up in stores, and then in warehouses, and even if factories cut shifts they may still have some of the components they ordered before. So there's a large swing in demand, and those manufacturers typically have a big enough base of fixed costs that they can't fully adjust, so their profits dip.

That, at least, is the historical story, and when inventory was a larger share of GDP, those inventory swings did indeed describe the contours of booms and busts: inventory would spike when demand dropped, then take a while to get worked off, and by the time it was mostly sold and demand was ticking up again, those companies were able to hire workers and get back to normal.

As inventory management has gotten better, and as manufacturing has declined as a share of GDP, these destocking/restocking cycles have gotten less important. They still matter in some sectors, and for individual companies—inventory issues are a reliable way for consumer electronics companies to run into trouble, for example, especially if their product skews heavily to holiday shopping.

A bigger driver today is that many productive assets are long-lived, and demand can change faster. There's always some game theory to expanding capacity: if every airline agrees that total available seat miles should grow a point more slowly than real GDP, they'll always have pricing power and the seats they sell will gradually get more expensive. If any one company defects from this, though, it'll gain market share in a profitable market, and amortize its fixed costs over more flights, and be able to hire more workers who start with less seniority. So it's hard for an industry to maintain this kind of discipline, and illegal for them to coordinate it.

Even when there's a secular increase in demand, the lagging factor means that there will be wobbles, which can show up in the form of industry-specific cyclical swings. In AI, for example, there basically hasn't been a case yet where a company decided to spend a lot on GPUs and that, specifically, was a bad decision. (There have been cases where they spent a lot on GPUs and used them for something that there wasn't much demand for, or just trained a model that wasn't quite good enough, but that's separate.) The only way for the GPU industry never to have a down cycle, even if the face of growth, is some combination of:

1. Every single person who can make big capital spending decisions chronically underestimates long-term AI demand. (If even one of them doesn't, that person is a larger share of the next round of capital allocation decisions.)

2. It's a straight shot from here to the Singularity.

In any other case, there will still be an air pocket or two; even if spending rises 30% annualized, there will be times when it's closer to 25% and the capex for that period supports something more like 35% growth.

And a good working explanation for why cyclicality won't go away is that Mag7 executive who had read the above description of cyclical dynamics in AI capex a few years ago, taken it to heart, and decided to spend less than planned would have ended up regretting that choice.

Many cyclical industries were born as growth industries. Airlines had great returns in the 1960s, automakers did similarly well in the 1920s (in the aggregate, with many failures made up for by a few huge success stories), and in the chip industry the slowest growth from 1950-60 was just over 40%, in 1960. (The next year, the industry shrank year-over-year for the first time, and settled into a cycle thereafter ($, Economist).) But some industries inflect in the opposite direction, when there's suddenly a new source of growth. If they can't commit to capex fast enough, and demand is rising quickly enough, they can go post-cyclical; utilities had a big drawdown in the 1930s when demand slowed and overall equity prices were low, but that was also when their bonds started yielding less than equivalently-rated railroads. It was clear that they'd been a growth industry (and would actually continue to grow until the 1960s), but it was also clear that demand was going to stabilize, and that in their case the regulatory environment would make them closer to fixed income.

The transition from growth to cyclical is more common than the other direction, but both can happen. A company that extends its revenue collection over the life of the customer relationship is making itself less cyclical in a GAAP sense, even though investors will respond to second-derivative changes and keep the price aligned with cycles. But mostly, it's the curse of any successful company to succeed, to the point that they've identified every plausible customer, priced things well, and otherwise captured as much value as they're ever going to capture. At that point, the next driver of ups and downs in the business is whatever's happening in the overall economy.

The Diff writes a lot about cycles, especially when they emerge:

• Early in the pandemic, we looked at how more of the upside from Covid-compensating policies would be captured by big companies, though the policies in general were pretty necessary.

• In early 2022, a look at how SaaS revenue is increasingly from other SaaS companies, so their performance correlates ($).

• The 2021 boom was a particularly strange cycle ($).

• And, two months after ChatGPT, thoughts on AI as a potentially cyclical business because of the high capital expenditures and the lag between starting to train a model and finally releasing it.

Join the discussion!