My partner, Antti Ilmanen, is launching a new series, Understanding Return Expectations. This should come as no surprise. Antti has already written two books on expected returns, and his team at AQR maintain our annual capital market assumptions (i.e., expected returns at the 5-10 year horizon).

So, what’s new this time? Prior to this, Antti focused on providing reasonable forecasts of expected future returns (never fear, he will continue to do so). Now Antti focuses on how investors actually form their own beliefs about expected returns (which may, or may not, correspond with what he finds reasonable!).

A little secret (or at least something not stressed enough) is that expected returns found in academic research (e.g., Antti’s first and second books, AQR’s Capital Market Assumptions, and some of my own writing) can differ starkly from what surveys reveal about investors’ subjective return expectations. The differences can get extreme at extreme valuations — for example in 2000 or 2021 (or even in 2024), where high valuations imply low expected returns in actuality (Antti’s usual day job are these forecasts), are often the same times surveys report high subjective expected returns (and the opposite in 2009).

One reason is academics and practitioners come at the problem from opposite directions. When academics talk about expected returns, they focus on the required returns in the denominator of the present value relationship. 1 1 Close P = E(CF)/(1+DR) – an asset’s fair price reflects its expected cash flows discounted by a risk-appropriate discount rate.  ^,  2 2 Close We at AQR (and academics in general) focus on “objectively feasible” expected returns which are often anchored to market yields or valuations because they have some objective ability to predict future returns. Not great, but some (and more at longer horizons like 5-10 years than at shorter or at super long horizons).   In contrast, many practitioners focus on the numerator, and thus associate high expected cash flows with high expected returns, regardless of price paid.

Another reason is consensus forecasts of individual investors and equity analysts extrapolate too much from recent years’ returns or growth. 3 3 Close An extreme, though common, example is forecasters who directly use trailing realized returns to forecast the future. Even very long-term trailing realized returns often lead to backwards forecasts — going up mechanically when returns have been very strong for a long enough period when future expected returns are likely going down. Good times in the rearview mirror are over-extrapolated into optimistic earnings growth or return predictions looking ahead. Growth optimism in 2000 or 2021 (even 2024? 4 4 Close Though as we write this the trade-war risks are perhaps bringing these forecasts down. ) may push equity valuations too high and lock in objectively low future returns.

Antti argues this “rearview mirror mindset” is:

• more pronounced if over-extrapolators recently got it right and became more confident (unfortunately, this effect can be cumulative when the over-extrapolators are “more right for more years” and their forecasts get even more over-confident)
• more typical for stocks than bonds
• more vulnerable when looking back 3-10 years (acting like a momentum investor at reversal horizons)
• more misleading when past outperformance mainly reflects richening valuations, now at extreme levels
• more dangerous when the macro backdrop changes

Most readers of this blog will know why Antti’s series matters today. There’s quite a bit of nuttiness in markets, whether looking at the relative pricing of stocks, or at a more macro level in US versus non-US equity markets, or (admittedly sarcastically) just about everything else that investors have flocked to over the past 5-10 years.

Given it’s Antti, this series is comprehensive. He summarizes the best research on survey expectations, the contrast between objective and subjective expected returns (split into rational and irrational camps), the evolution of expected return analysis among academics and practitioners, and provides extensive historical evidence on average returns and both objective and subjective return expectations.

He also highlights exceptions to the rearview-mirror mindset: institutional investors and bond investors 5 5 Close I have marveled for years how people seem to get that when bond yields go down (bond prices go up) future expected (certainly nominal, often real) returns go down, but fail to see that equities have a very similar yield effect, it just takes 10 more seconds to understand it.   in general appear more rational and less extrapolative than individual investors and equity investors. He shows that equity analysts are perennially overoptimistic but their overextrapolation tendency is worse as it makes analyst growth forecasts predict negatively future returns. He acknowledges that long-run growth forecasts can vary over time and across countries or single stocks, but given the poor predictive ability of those forecasts we should keep our growth expectations tightly anchored. As ever the two-handed economist, while Antti documents the usefulness of price-to-earnings ratios as objective return predictors, he also emphasizes their limits and lists several possible improvements.

Studying long histories, we find that it pays to bet against unrealistic growth optimism at the market level or with single stocks – this is why the hard contrarian strategies beat the easy multi-year extrapolation in the long run, and it’s why Antti’s new series matters today.