Investors talk about market regimes, like periods when growth stocks gracefully trend upward or times when the main thing traders fixate on is a company's ability to withstand inflation. And more narrowly, they talk about themes, where there's one big bet that can be expressed across sectors and categories—you could play the rise of GLP-1s by shorting candy and alcohol stocks, but also by buying clothing retailers (everyone has to replace their wardrobe) or, in the era of peak GLP-1 extrapolation, arguing that it would save some airlines tens of millions of dollars in fuel costs.¹ And there's something in between these two: loose, mostly qualitative heuristics about the way things work, that feel permanent but aren't really.

The best way to illustrate these is with an example: when I started working on the buyside, I came across a reference to how treasury bonds move inversely with equities, thought about it for a second, and realized how obviously true it was: stocks go up when the economy grows faster, but that leads to higher rates, so long-duration bonds suffer. And when growth expectations come down, the government cuts rates (or just buys lots of bonds directly if rates are already at zero). So, stocks plus bonds plus leverage can deliver a better-than-stocks return with equivalent risk. But, as it turns out, this only started to be true around 1998, and earlier generations of investors didn't think that way. Someone active in finance in the 70s and early 80s would have been more likely to think of the big two categories of inversely correlated assets as financial claims like most stocks and bonds, or real assets like real estate and commodities (equity in commodity producers did work as an inflation hedge). When inflation is high and uncertain, central banks don't mechanically respond to slower growth by cutting rates, because they have a more complicated balancing act. But gold does fine whether the bank's response to stagflation is to cut rates and let inflation rip or to raise rates and deal with a brutal recession.

This kind of narrative isn't really one you can make a single directional bet on, but it's a correlation that persists for long enough that people can forget what drives it. Until very recently, the big question about China's growth in any given year was whether it would be around the typical 10% or a worrisome number like 6% or 7%. Every year that China had above-average growth was a year where a) China's growth got more important to the global economy, and b) questioning how long it could last got less and less credible. And then there was a reset, partly driven by the Trump/Biden/Trump administrations (which were pretty consistent on decoupling), and probably driven by Covid creating an explainable break in the growth time series that made it less embarrassing to reset to a lower and more volatile number.² An investor with the perspective of ten years ago, who looked at China's GDP growth today, would be astounded that copper prices had tripled over the intervening decade.

There are other periods that have other almost-unstated assumptions. The 90s were the era of equity maximalism; the market was regularly hitting all-time highs, and celebrity fund managers were back after a generation-long hiatus. Nobody in 1980 was all that interested in reading about the astounding investing talents who'd managed to lose only 10% of their investors' money in real terms over the intervening decade, though readers of trade magazines like Institutional Investor heard about Soros in the early 80s, back when he was a relative neophyte in macro and had spent most of his career in equities.

It's hard to remember now, but there was a widespread view in the 90s that people were underweight equities and, similarly, that it was hard for stocks to go down. (I remember asking my dad in the 90s if a crash like 1929 could ever happen again. His answer was that every two weeks, millions of people automatically put a set fraction of their paycheck into retirement plans, so there's always a buyer. While it was true that we haven't had anything like 1929 again, flows like that get priced in after a while and we can still have drawdowns.)

Sometimes, shifts in these narratives catch people by surprise. The default way of talking about software companies from the earliest period that they started going public through just the last few years was that they were a high-quality business because they needed so little capital. All you needed to start one was a working computer, an Internet connection, and maybe an O'Reilly book. But today, they're the biggest capital spenders around—once again, someone transported from the mid-2010s to the present who looked at big tech's capex as a percentage of sales would have to assume that something terrible had happened, and that all of these companies would be trading at a discount to the market. They would also be very surprised to see that despite record high capex, many of these companies have simultaneously printed record ROICs (which also happen to be leagues above broader market ROICs).

What often happens with these shifts is that there's a zombie correlation that sticks around in stock prices for a while even when fundamentals have changed, but then reverses violently when there's a big drawdown and people start pricing out the downside scenario. Railroad economics were permanently impaired by the rise of the car, which permanently impaired their passenger business. But the rise of the car, and of some other new manufactured goods, created more freight demand, which made railroads look fine, albeit a bit dowdy, in the 1920s. And then early in the Depression, investors reassessed them and decided that they weren't the safest businesses any more, and that there wouldn't be mean-reversion back to a market that was mostly railroads ever again. The current generation of investors won't think of treasury bonds as a universal equity hedge, because they remember 2022. There are probably some investors out there who think of consumer staples as good companies to own during a flight to safety, since they've historically outperformed in tough markets; you might cancel a vacation because you got laid off, but you're probably not going to stop buying laundry and toilet paper. But you might switch brands, and if stores see that consumers are more price-sensitive, one way to preserve their spending power in high-margin areas is to get them to buy the cheap store brand for these kinds of repeat purchases. Tech is the odd exception here; nobody is calling Microsoft a capital-light business these days. But even that illustrates just how extreme the change in fundamentals has to be before it changes investor perceptions: yes, investors can change their mind about these kinds of unconsidered assumptions before it costs them money, but usually when the deviation from that older model is greater than a percentage point of GDP.

In The Diff, we often look at cases where these narratives work, and where they shift.

• Even if there's a good theoretical argument for a given approach, you need some kind of heuristic you can actually apply ($).

• One way to get around this problem is to maximize your idea velocity.

• A good example of betting against the easy secular model is the long winning stream of Nexstar, which happens to be well-positioned in a dying industry ($).

• And betting on one: Google’s growth rate has been surprisingly durable over long periods ($)

Join the discussion!

1  The problem with that one is that airlines tend to add to their fleets, and to price-discriminate more to fill the last economy-class seat, if their margins go up. So, if it's true, it's true over a short period. On the other hand, airlines trade at low enough multiples that a small one-time gain does move the valuation. Just not by very much.

2  This happens in public-facing careers, where sometimes an exogenous change causes someone to reset their public persona. It was going to be hard for Martha Stewart to keep presenting as an unironic idealized upper-class suburban wife after she'd done time in prison. So she reinvented herself as a relaxed, self-aware parody of the earlier Martha. Which had to be a relief.