Investing Amid Low Expected Returns Ch. 1: Why High Past Returns Predict Low Future Ones

"The most reliable way to raise realized returns is to lower expected returns — and the bill for that arrives later." — Antti Ilmanen

Investment Background

Antti Ilmanen spent a career at Salomon Brothers, Brevan Howard, and AQR Capital building the cross-asset return-premia framework that this library already covers in Expected Returns. That earlier book asked a horizontal question: for each asset I hold, what risk am I being paid to bear? This 2022 sequel asks a much harder and much more personal one: what do you do when the honest answer, for every asset simultaneously, is "not very much"?

The setup for that question was the 2009–2021 era. Central banks pinned policy rates near zero, then bought bonds outright to push long yields down as well. That did exactly what it was designed to do: it raised the price of every asset that could be valued by discounting future cash flows. Equity multiples expanded, credit spreads compressed, property capitalization rates fell, and private-market valuations followed. Investors experienced this as a golden decade — US equities compounding in the mid-teens, bonds delivering positive real returns while yielding almost nothing.

Ilmanen's central and deeply unwelcome point is that most of that performance was not earned; it was borrowed. A return that comes from cash flows growing is income. A return that comes from the market agreeing to pay a higher multiple for the same cash flows is a repricing — and a repricing is a one-time transfer from the future owner to the present owner. When you buy a bond at a 1.5% yield, you can compute your future return precisely: roughly 1.5% nominal, forever, no matter how well bonds performed last decade. The good decade did not create future return. It consumed it.

This is the intuition every retiree needs and almost no retiree has. High past returns and low future returns are not two independent facts that happen to sit next to each other. They are the same fact, described from two ends.

The Wall Street Translation

The Decomposition That Changes the Conclusion

Any realized return over a period can be split into three components, and the third one is the trap.

Component	What it is	Repeatable?
Income	Dividends, coupons, rents actually received	Yes — but only at the current yield, not the historical one
Growth	Real growth in earnings, dividends, or rents	Yes, roughly, at long-run economic growth rates
Repricing	The market paying a higher multiple for the same cash flow	No — and it reverses the sign of future returns

Run US equities through this for 2009–2021 and the answer is uncomfortable. A meaningful slice of that period's return came from the multiple expanding rather than from earnings compounding. Run a long Treasury through it and the picture is starker still: essentially all of the excellent 40-year bond bull market from 1981 to 2021 was the repricing of a yield falling from double digits toward zero.

The practical consequence: an investor who extrapolates the trailing decade is double-counting. They are treating the repricing as if it were income — as if the market will keep paying an ever-higher multiple indefinitely. It cannot. A multiple that rises forever implies a forward return that falls to zero and then goes negative.

The Yield Is the Message

Ilmanen's estimation discipline is deliberately unglamorous: the best available forecast of a long-horizon return is a currently observable yield, not a historical average.

• For a bond held to maturity, this is nearly exact. The yield to maturity is the return, absent default.
• For equities, the analogue is the earnings yield or the dividend yield plus a conservative real growth assumption. It is noisy over any five-year window and informative over a fifteen-year one.
• For credit, it is the spread minus expected losses — not the headline yield, which flatters by ignoring defaults.

None of these are forecasts in the pundit sense. They are readings. The reason people resist them is not that they are inaccurate; it is that they are disappointing. A trailing-return spreadsheet tells a 62-year-old that 7% real is normal. A yield-based reading tells the same person that a global 60/40 at today's starting point plausibly delivers something closer to 2–4% real. That difference is not academic. Over a 30-year retirement it is the difference between a plan that works and a plan that quietly runs out at 84.

Why This Feels Like Pessimism and Is Not

The emotional structure of this chapter is the hardest part. Investors hear "low expected returns" as a market call — a bearish prediction that stocks are about to fall. It is not a prediction of a crash. It is a statement about the price you are paying for a stream of cash flows.

A high-priced market can go on being high-priced for a very long time. What it cannot do is deliver the same return from a high price as it delivered from a low one. The low-return environment does not announce itself with a crash; it announces itself with a decade of adequate-but-thin performance that quietly fails to fund the plan. That is the far more common and far more dangerous outcome, because nothing ever happens that is dramatic enough to make the retiree reassess.

Execution Rules

1. Replace the trailing-return assumption in every plan you own. Open whatever calculator or spreadsheet drives your retirement plan and find the expected-return input. If it was set from historical averages — 7% real equities, 3% real bonds — rebuild it from current yields instead: current bond yield to maturity for the bond sleeve, current dividend yield plus one to two points of real growth for the equity sleeve, weighted by your actual allocation. Write down the resulting number and use it everywhere.

2. Decompose any track record before you extrapolate it. Whenever a fund, strategy, or asset class is sold to you on the strength of a five- or ten-year record, ask what fraction of that record came from repricing. If the strategy's yield today is far below its yield at the start of the track record, the record is a description of a repricing you already missed, not a forecast you can buy into.

3. Plan with a range, not a point. Use your yield-based estimate as the central case, then run the plan again with that estimate reduced by 1.5 percentage points. If the plan only survives the central case, it is not a plan; it is a hope. The decisions that survive both cases — a higher savings rate, a more flexible spending rule, lower fees — are the robust ones.

4. Do not convert this reading into a market-timing trade. Using current yields to set a planning assumption and using them to decide when to be in or out of the market are entirely different acts with entirely different evidence behind them. The first is well supported. The second is not, and the investor who exits a richly priced market typically spends years watching it get richer while earning cash returns.

5. Keep the low-cost index core exactly where it is. Nothing in this chapter is an argument to replace broad, cheap, globally diversified index exposure with something cleverer. A low expected return on the index is still, in almost every case, a better deal than a high advertised return on a complicated product with layered fees. This chapter changes the number in your spreadsheet, not the holdings in your account.

Retirement Application

For a retiree, the low-expected-return problem has a specific and cruel shape: it attacks the plan at the exact moment the plan has the least capacity to respond.

An accumulator facing a low-return decade has powerful compensating levers. They can save more, work longer, and — crucially — they are buying into low prices the whole way through, which raises their own future expected return. A low-return decade is genuinely good for a 35-year-old net buyer of assets.

A retiree has none of that. They are a net seller. Every year of thin returns is a year in which withdrawals come out of principal rather than out of gains, and the portfolio's capacity to recover shrinks permanently. This is the single most important asymmetry in retirement finance, and it is why the honest expected-return number matters more to a 65-year-old than to anyone else.

The correct response is unromantic. When the expected return on capital falls, the plan must be rebalanced away from things you do not control and toward things you do:

• Savings rate (for those still accumulating) — fully controllable, immediately effective.
• Spending flexibility — a retiree who can cut discretionary spending 10% in a bad three-year stretch has bought more safety than most portfolio changes can provide.
• Time horizon — working two extra years shortens the drawdown period and lengthens the accumulation period simultaneously.
• Fees — a 60-basis-point fee reduction is a guaranteed 60 basis points of return in an environment where the total expected return might be 300.
• Taxes — asset location and withdrawal sequencing are worth real basis points and require no market view at all.

What is not on that list is "reach for a higher return." That is the instinct this book spends six chapters dismantling, and Chapter 2 takes it head-on.

Risk Management

• Extrapolation risk: The single largest planning error is assuming the last decade repeats. Guard against it by rebuilding assumptions from yields annually rather than inheriting last year's number.
• False precision risk: A yield-based estimate is a central tendency with wide error bars, especially for equities over horizons under ten years. Treat it as a planning input with a range, never as a forecast to trade against.
• Despair risk: The genuine danger of internalizing this chapter is paralysis or capitulation — abandoning equities entirely because the expected return looks thin. A thin equity premium is still a premium, and cash at a negative real yield is a guaranteed loss. Low expected returns argue for more patience and more saving, not for exiting risk assets.
• Reaching risk: The most damaging response is the one that feels most active — replacing the index core with high-yield products that appear to solve the problem. Chapter 2 is entirely about that failure mode.

What This Chapter Cannot Do

This chapter cannot tell you what returns will actually be. Yield-based estimates have been wrong in both directions for a decade at a stretch, and anyone who exited equities on valuation grounds in 2014 spent seven years being punished for it. Nor can it tell you when the low-return environment resolves — starting yields have risen materially since the 2021 trough, which improves forward returns and complicates any story told purely about the zero-rate era.

What it can do is remove one specific error: planning a 30-year retirement with a number borrowed from a decade whose returns were, in substantial part, a one-time repricing. The chapter's contribution is negative and it is valuable: stop extrapolating.

---

Key Takeaway: High past returns and low future returns are the same fact seen from two ends. A decade of multiple expansion is a transfer from tomorrow's owner to today's, and the retiree planning off that trailing number is budgeting money that has already been spent. Rebuild the assumption from current yields, then respond by pulling the levers you actually control — savings, flexibility, horizon, fees, taxes — not by reaching for a return the market is not offering.