Risk and Returns, Before and After the Fact

All Projections Are Extrapolations, But Be Careful About What You Extrapolate

This week's edition of Capital Gains is brought to you by our sponsor, Shortform!

Know someone who might like Capital Gains? Use the referral program to gain access to my database of book reviews (1), an invite to the Capital Gains Discord (2), stickers (10), and a mug (25). Scroll to the bottom of the email version of this edition or subscribe to get your referral link!

There are two ways to talk about long-term returns from investing in a given asset class. One is easy because it has limited utility, and the other is a useful intellectual exercise, but impossible to get right.

This might shock some of my younger readers: when I first started learning about personal finance in the 90s, the statistic I heard was that the stock market, on average, returns about 11% a year. This was very much a 90s thing to believe; in the mid-2000s, the stylized estimate was closer to 9%, and for the last decade that figure has been closer to 6%.

Why do these generally accepted truths change? For one thing, the longer the time series, the more prone it is to survivorship bias. We have very long-term data for countries, like the US and UK, which haven't been successfully invaded and have mostly come out ahead in military conflicts1 , but the time series for Russian equity returns has a big gap from 1917 through the early 1990s, and Chinese stock market data suffers from a similar deficiency.

You might treat countries like the US and the UK as the upper bound—but you'd also want to note that valuations reflect different levels of the risk of catastrophe. Fortunately there’s another option:

Long-term returns are, in a simplistic model, the current dividend yield plus long-term growth in dividends per share—where “long-term” is very long-term, like a lifetime or longer. And growth in dividends per share is, also in the very long term, a function of growth in revenue per share: margins tend to be surprisingly stable over long periods, and multiples can only go so far in one direction or another. The per share criterion is important, because some companies return capital through buybacks as well as dividends, and others are net issuers of stock.2 (Also, for what it’s worth, while this model looks like it's ignoring acquisitions and IPOs, those transactions are actually incorporated into the "per share" model: if you treat the entire market as a single company, an IPO is just another share issuance, and an acquisition is a buyback.)

This model looks too easy, and in some ways it is. For example, it's correct over a timescale long enough to be irrelevant to individuals; there can be long swings towards lower margins, like the one that took place in the US from the 60s through the 90s. And there can be swings in the opposite direction, like the rise in corporate margins that's taken place since then. But these trends can't go on forever—each one of them could easily be the majority of the lifetime of a specific individual investor. Swings in valuation can happen, too; P/E ratios went from ~20x in the late 1920s (on low-quality earnings that were increasingly from financial engineering) to the mid-single digits in the late 1940s, back to 20+ in the 60s, back to single digits by the late 70s, and so on.3

This note points to another issue with using historical data to predict future returns: starting periods matter! From 1980 to 2020, long-term bond yields went from 16% to 0.6%. Since then, they've risen to 4.6%. Government bonds have the wonderful trait that they tell you exactly what the expected return is: the current yield on a ten-year treasury is also the market's live estimate of what an investor will earn over the next ten years by owning 10-year treasuries. The long bull market meant that people were buying a decent expected return and realizing a better one. Every repetition of that made the historical rate of returns from holding long-term bonds better, but also made the expected return from continuing to hold them worse.

Anyone allocating money to an asset class needs to construct some view on what that asset's returns are, and why. Historical data supplies the "what," but it's market participants' behavior and underlying economics that create the "why." The only way you can get a return from buying something is by finding a seller who is willing to forgo that future return. Occasionally, this will be an obvious and clear-cut opportunity, where the market is pricing in a scenario so apocalyptic that capital losses will be the least of your concerns, or, in the other direction, when the market is pricing in so much growth that you'll be well-off even if you forgo some equity exposure and move money into bonds and commodities instead.

But it's usually a messy process. The discussion of fundamental drivers of equity returns hammered home the idea that valuation is not a driver of returns, because it moves so slowly. Even if the market switched from trading at ~10x earnings in the early 20th century to ~25x today, that's 70 basis points per year of annual return. But the flip side of this is that over shorter periods, valuation's role as a return driver goes way up. Year to year, the market's moves are typically the result of multiple expansion rather than earnings growth or dividends (there are occasions when that's untrue because earnings are distorted by big writedowns; technically, the S&P's trailing twelve month earnings rose 9x from Q1 2009 to Q1 2010, but most of that was from huge writedowns in Q4 2008; usually investors thinking about the market multiple are using some kind of smoothed, cyclically-adjusted PE). Over even shorter periods, valuation becomes even more important; an intraday swing or the change in price from one trade to the next are, almost by definition, overwhelmingly driven by valuation multiples rather than some tiny incremental update to a company's fundamentals.

In the very long run, stocks tend to go up. But that's only sustainable if that long-term gain is compensation for some shorter-term variability. You can know roughly what you're getting into by looking at broad valuation metrics, which tell you that stocks are fairly expensive right now, especially in the context of higher rates. But you can't get much more information than that—the difficulty of timing the market is a function of how all the easy ways have been arbitraged away, and what's left is a bedrock of uncertainty that, over short timescales, dominates fundamental drivers and long-term trends.

A Word From Our Sponsors

Read a book a day with Shortform’s POWERFUL nonfiction book guides.

Have a shelf full of unread books? Shortform can help you learn (and retain) the world's best ideas in a fraction of the time with their super-detailed book guides.

Take your understanding of important subjects to the next level with Shortform’s comprehensive coverage of must-reads in every nonfiction genre - from finance to self-improvement.

But don’t mistake Shortform for the other summary services out there that just give a brief overview. Their book guides include chapter breakdowns, insights, and even exercises.

Start your FREE Trial and get 20% off at shortform.com/capital.

Read More in The Diff

The Diff has addressed the question of where equity returns come from, and what drives them, in a few different places:

  • "Should We Expect Valuations to Mean-Revert Over Time?" focuses on how changes in the composition of the market make long-term valuation metrics suspect.

  • In “Passive” as a Factor, we look at whether passive investing itself drives changes in market prices.

  • Investing in Times of Uncertain Expected Returns ($) is a review of the book by the same name, an update to Expected Returns—if you read this post and wished it were 500 pages longer with hundreds of academic references for every claim, pick up a copy.

  • In Why Do Equities Build So Much Wealth? we look at a surprising paradox: the risk-adjusted returns of stocks, bonds, and many simple systematic investment strategies are similar—a portfolio of bonds can be levered to the point that its risk and return are similar to what you'd get from the Nasdaq—but far more people get rich either picking stocks or owning shares of a company they started than trading or buying-and-holding bonds.

1. You can quibble with both: the US obviously didn't achieve its goals in Iraq or Vietnam. On the other hand, from an investor's perspective, there's a major qualitative difference between going halfway around the world to lose a war and losing one that went halfway around the world to you. Meanwhile, the UK won the Second World War in political terms, but absolutely ended up losing economically afterwards, squeezed by lower-cost manufacturing from recovering continental economies, US dominance in higher value-added manufacturing, and an overextended financial system whose gradual decline was continuously deflationary. Losing an empire slowly is an expensive proposition.

2. That can be because they do repeated follow-on offerings, but there's bad signaling there, and many companies try to avoid it. There is very little signaling cost in skewing compensation to equity rather than cash, and in some industries it's standard. In the very long run, spending 1% of market cap on equity compensation in order to drive a 1% annual increase in sales should, assuming stable margins, lead to the same returns as spending less and growing less. That won't be true in practice, because the industries with high equity comp are also industries with high fixed costs and high incremental margins. It also isn't true because, if a company is clearly going to win its segment, reducing dilution gets them a few more years of above-average growth before they win. If the market they're in is competitive and has returns to scale, though, faster growth can be mandatory. So the dilution question is a strategic one at the level of individual companies, but matters mostly because it affects the gap between revenue growth and revenue per share growth in the aggregate.

3. There is an argument to be made that there can be secular changes in P/E ratios based on the composition of the market and the broader economy. Specifically, there are more long-term growth stocks in the US market today than decades ago because so many tech companies can perfect a model in the US market and then deploy it at low marginal cost globally, while cyclical industrial and materials stocks are a smaller share of the market because they're a smaller share of GDP. And it may be that the optimal way to finance some low-multiple companies is not through taking them public, but by owning a portfolio of them with a private equity firm. In that case, though, the economic exposure can still end up in the public markets, just refracted through PE firms' fees.

Share Capital Gains

Subscribed readers can participate in our referral program! If you're not already subscribed, click the button below and we'll email you your link; if you are already subscribed, you can find your referral link in the email version of this edition.

Help us learn more about you!

We're building a media kit and need some anonymous aggregated subscriber data. If you work in finance and have time for a 3-question survey, that would be very helpful:

Join the discussion!

Join the conversation

or to participate.