At any given time, some things will look historically great, and some pretty bad, judged mostly by where you chose to start your time series.

You see this in financial data all the time. In the late 90s, when people talked about the stock market, they'd say its average return was 11% per year. A few years later, the number people threw out was usually more like 9%. Choose some point in the 20th century when the time series of S&P 500 performance and dividends is within your budget, and roll things forward to a few different dates, and you'll see returns bob around a bit. They'll still fluctuate around some central average, but it's hard to know exactly what that mean is. For a market like Japan, returns looked structurally high and strategists drove themselves crazy trying to rationalize it—maybe cross-shareholdings meant that if you were buying equity in any one Japanese company, you were getting a piece of the rest of the market, and maybe there was a lower risk premium for a country that had mastered sustainable growth. It turned out that all of these stories were basically wrong, and the stocks were just overpriced.

That same start-date sensitivity also shows up in politics and social commentary. Compared to the 1950s, manufacturing jobs pay less, it's harder to have a single-earner household, the media are more fragmented and less responsible, and politics has sharp partisan divides. But the 1950s were a high-water mark for many of these phenomena! Postwar, the US had a surplus of manufacturing capacity and a shortage of workers, so it made sense to offer generous wages, particularly in scale- and utilization-driven industries like auto manufacturing. The media environment was less confusing when there were fewer media outlets, partly because of another kind of scale economy—it's easier to maintain one newspaper distribution network rather than two, and because there were only three TV networks. As for politicians, they had an easier time getting along for a few reasons. Most of them had recently served in the same war, but they were also united in their fear of either losing the Cold War to, or being mistaken for, communists. And some other issues, like segregation, cut across party lines—the divides were between Northern and Southern Democrats, and between Western/Midwestern and Eastern Republicans. That leads to what looks a lot like bipartisan cooperation, but which was actually a different flavor of coalitions.

The easiest way to get this wrong, as alluded to above, is to be born at the right time. People who started following news in the 90s will never shake the sense that a mostly peaceful world benignly ruled by the United States was the default expectation, or at least the rough direction things were headed. People who started following news in the 70s, or the 2000s, tend to have a darker view by default.

You'll sometimes see this evolution in other dimensions, like the size of organizations. Over my lifetime, the expected lifespan of big companies has dropped a lot—GM was a big company when I was born, and a big company a generation earlier. It was a pretty good-sized growth stock a generation before that! But now, while GM is an objectively big company that still produces lots of cars and pays lots of workers, it isn't the kind of unbeatable business it used to be. You could look at this as an example of the economy getting more dynamic, and companies going through faster rises and falls, but that, too, is a cyclical phenomenon. There are periods where small companies can scale quickly, like the new industrial giants of the late 19th century, or the railroads before that. (Railroads had a strange kind of scaling: the minimum capital investment required to enter the industry meant that some of them were born big, and didn't grow especially fast after.) And there are times when the list of big companies is static—for a long time, the dominant company in the computer industry was IBM, which had been founded as a roll-up of scale, calculator, and time clock companies in 1911. There were new entrants, but IBM had accidentally backed into a model where they installed computers upfront and earned recurring revenue, which meant that competing in that business was capital-intensive, and the financial sector wasn't equipped to fund it appropriately. Still , IBM went from being the largest company in the S&P by a large margin in 1985 to out of the top 10 within 30 years.¹ An environment where the biggest companies were mostly founded within the lifetimes of the people analyzing them is historically unusual—it's a function of a venture capital ecosystem that can fund capital-intensive new businesses, if they're promising enough, and an equity market that lets those companies attract talent with equity compensation.

If you look at a general trend—more globalization, more bandwidth, cheaper electronics, more paywalls, etc.--it's actually pretty reasonable to extrapolate. They tend to go on for a long time! But if you want to be early to spotting changes, you have to treat the trend as a measure of some underlying reality that might not endlessly trend in the same direction.

These timing-related questions are an important part of framing pieces on long-term trends. A few examples:

• In The Tyranny of the Long Generation, we look at the pattern where an industry stops growing and its workforce starts aging, leading to a feedback loop.

• There are some time periods long enough that these effects wash out. Like 10,000 years.

• What happens when you buy at an all-time high?

• One of the long-running cycles to pay attention to is demographic ($).

• The consumer credit cycle matters ($), but isn’t the only driver of cycles.

• Sometimes, what looks like an intrinsic feature of an industry’s economics turns out to be temporary, as with capital-intensity ($).

Join the discussion!

1  The reason IBM was obsessed with ARR so early compared to everyone else is that their older business model was to sell cheap computers and then make money marking up punchcards, so they effectively had usage-based pricing. As the punched-card system gave way to digital displays and printers, they realized this threatened their highest-margin business, so they just rejiggered the computer business to be the economic equivalent to selling cards, by way of leases.