[Week 50 of 2025] Forecasts, Funds, and Flexes

Welcome back to Price and Prejudice with a few musings from Week 50 of 2025.

Forecasting with Hindsight

Apollo put out a piece projecting near zero annual equity returns over the next decade, based on today's elevated P/E ratios. The logic rests on an accounting identity: if prices are high relative to earnings today, and earnings growth is basically unpredictable (which is empirically true), then the variation in price ratios has to come from changes in expected returns. So high valuations mean low discount rates (expected return), which means low future returns.

But the math checking out and the forecast being useful are different questions. For one, valuation ratios are highly persistent (they move slowly), which creates a statistical problem where the regression slope gets inflated in finite samples. There's also research showing that these regressions, however impressive they look over the past century, consistently fail when you try to use them in real time. The regime keeps shifting, and the forecast keeps missing.

And then there's the possibility that this time really is different—not in the bubble sense, but in the narrow sense that maybe AI or platform economics or whatever actually does justify higher valuations. If expected growth has structurally shifted upward, then today's high P/E ratios tell you nothing about future returns. The accounting identity still holds; it's just that growth, not returns, is doing the work.

None of this means you should pull your money out of equities (contrary to Twitter advice). The forecast might be right, but it also might not be, and the opportunity cost of being wrong is large. Sitting in cash because a regression line told you to is a bet that the next decade will look like the past century, which is a bigger forecasting claim than it sounds. The R² of 0.73 is impressive until you remember that R² measures how well you fit the past, not how well you'll predict the future.

Organizational Trust as Portfolio Tilt

This paper asks whether slapping a name on an index fund changes how people invest in it, even when the name tells you nothing about what's inside. The answer is yes, and the effect is large. The setup is clever: take two identical index funds—same benchmark, same fees, zero information content—and label one with a trusted organization's name and the other with a distrusted one.

This has obvious implications for retirement plan design. White label funds were supposed to strip out branding and focus investors on fundamentals, but anonymity only works if people trust the plan sponsor. Attach a distrusted name, and you've made things worse. The broader point is that organizational identity acts like a non-pecuniary asset characteristic. People aren't optimizing over abstract return distributions; they're deciding which institutions they're comfortable giving their money to. That's a reasonable thing to care about, even if it doesn't fit neatly into a mean-variance framework.

It'll be interesting to see whether AI unravels this. Trust is often a substitute for information processing—you rely on an organization's reputation because parsing fund documents is tedious and you're not sure what matters anyway. But if AI agents can actually read the prospectus, compare fee structures, and verify that two index funds are genuinely identical, maybe the trust heuristic becomes less necessary. Or maybe not: People might just prefer giving their money to institutions that feel safe, even when they know the products are the same. If that's true, AI won't fix it and instead give people better reasons to ignore what the AI tells them.

The Veblen Treadmill

This article highlights the unique problem of the ultra-rich: their status symbols keep getting democratized. Veblen goods stop working when everyone can fake them: a Birkin signals wealth only if other people can tell it's real, but most people can't. The whole point of conspicuous consumption is that it's expensive and therefore hard to imitate, but once Walmart enters the chat, the signal collapses.

The solution, from the perspective of the wealthy, is to move on to new status symbols that are harder to fake: privacy, leisure time, and self-expression. Having time for pickleball or baking fresh bread signals you're not working multiple jobs. Buying novel, expressive clothing instead of logo-heavy Prada says something about you, not just about your bank account. It's a treadmill, and luxury brands are discovering that you can't win by just raising prices. You have to stay ahead of Walmart, and Walmart is very good at catching up.

The question is whether we will eventually run out of things to flee to. Privacy and leisure time are harder to fake than a handbag, but they're not immune to imitation. At some point the treadmill might just stop, not because the wealthy give up on signaling but because there's nowhere left to run. Or maybe the endgame is that status becomes so abstract and illegible that only other rich people can decode it, which would be fitting. The whole point of wealth signaling is to be understood by the right audience, and if that audience shrinks to just the top 0.1%, then mission accomplished. Everyone else is locked out not because they can't afford the symbol, but because they can't even see it.