Investing Amid Low Expected Returns Ch. 3: Diversification When Everything Is Expensive

"Diversification protects you from being wrong about which asset. It does not protect you from being wrong about the price of all of them." — the central limitation of the 60/40 framework

Investment Background

The 60/40 portfolio earned its reputation honestly. For roughly forty years, a portfolio of 60% equities and 40% high-quality bonds delivered something remarkable: equity-like returns with meaningfully less than equity-like pain. In the crashes of 2000–2002 and 2007–2009, bonds rallied hard while stocks fell, and the bond sleeve did exactly the job it was hired to do.

Ilmanen's uncomfortable observation is that this record was produced under a specific and non-repeatable condition: a forty-year secular decline in interest rates. That decline was the tailwind under the bond sleeve's returns, and it was simultaneously a major contributor to the equity sleeve's multiple expansion. The same force lifted both halves.

Which produces the structural problem this chapter exists to name. When one variable — the discount rate — is driving the price of both assets in the same direction, the two assets are not independent. They are two expressions of one bet. In a rate-falling world, that shared exposure feels like magic: both sleeves win. In a rate-rising world, it is exposed for what it is: both sleeves lose, at the same time, and the diversification the retiree was counting on simply does not appear.

2022 delivered the demonstration. Global equities and long bonds fell together, producing one of the worst years in the history of the balanced portfolio. Nothing broke. The portfolio behaved exactly as its construction implied it would; investors were simply surprised because forty years of data had taught them a relationship that was conditional rather than structural.

The Wall Street Translation

Correlation Is Not a Constant

The stock-bond correlation is the single most important input to a balanced portfolio, and almost every retiree treats it as a fixed property of nature. It is not. It is a regime variable.

Regime	Dominant shock	Stock-bond correlation	60/40 behaviour
Growth-shock regime (1998–2020)	Recessions, earnings collapses	Negative — bonds rally as stocks fall	Diversification works beautifully
Inflation-shock regime (1970s, 2022)	Rising inflation and rates	Positive — both fall together	Diversification vanishes when needed

The lesson is not that 60/40 is broken. The lesson is that its protective property is conditional on which kind of shock arrives, and the retiree does not get to choose. A portfolio that only diversifies against growth shocks is undiversified against inflation shocks — and inflation shocks are the ones that damage a retiree's real spending power most directly.

The Deeper Problem: Correlated Expensiveness

Ilmanen pushes past correlation to something more fundamental. The reason returns are low across the board is not that each asset independently happens to be expensive. It is that they are all being valued off the same low discount rate.

This means the traditional response — add more asset classes — helps far less than it appears to. Real estate is priced off the same discount rate. Infrastructure is priced off the same discount rate. Private equity is priced off public equity plus a spread, and its entry multiples move with public multiples. Adding a seventh expensive asset to six expensive assets does not fix expensiveness. It diversifies the identity of what you own while leaving the underlying valuation exposure fully intact.

This is the single most misunderstood point in modern allocation. Investors respond to low expected returns by increasing complexity, and complexity does address one risk — the risk of being wrong about which specific asset. It does nothing about the risk that matters here: being right about all of them at prices that are too high.

What Actually Diversifies Correlated Expensiveness

Only a short list of things genuinely helps, and each carries a real cost that must be stated honestly.

1. Inflation-linked bonds. They pay a contractually real yield. When that real yield is positive, they are the cleanest available defence against the inflation-shock regime that breaks 60/40. Cost: when real yields are negative, they lock in a guaranteed real loss.
2. Cash and short Treasuries. Nearly uncorrelated with everything, and their expected return is low but visible. Cost: they cannot fund a 30-year retirement on their own, and long stretches of negative real cash rates are pure erosion.
3. Genuinely uncorrelated risk premia — trend-following, defensive equity tilts, carry strategies. These have return sources that are structurally unrelated to the discount rate. Cost: fees, capacity constraints, and multi-year droughts that test any investor's patience. This is Chapter 5's subject, and it is a modest hedge, never a core.
4. International and value exposure. Not because foreign or cheap stocks are guaranteed to outperform, but because the discount-rate compression was not uniform globally or across styles, so cheaper markets and cheaper styles carry a genuinely different starting yield. Cost: tracking error, and the very real possibility of a decade of underperformance.
5. Human capital and spending flexibility. These are not portfolio assets, which is exactly why they are the strongest diversifiers a retiree owns. Neither is priced off the discount rate at all.

Item 5 is the honest answer. The most powerful diversifier against correlated expensiveness in a retiree's life is not a fund. It is the ability to spend less in a bad stretch, and the willingness to earn something for a few more years.

Execution Rules

1. Stop scoring your portfolio on number of holdings and start scoring it on number of distinct risks. List your holdings, then write beside each one what would have to happen in the world for it to lose money. If most of the answers collapse to "rates rise" or "growth disappoints," you own two risks in twelve wrappers, not twelve risks.

2. Own the inflation defence explicitly rather than assuming nominal bonds provide it. Nominal Treasuries diversify growth shocks; they amplify inflation shocks. A retiree with a 25-year horizon should hold a deliberate allocation to inflation-linked bonds — sized to cover the non-discretionary portion of spending — rather than assuming the nominal bond sleeve covers both jobs. It does not.

3. Hold two to three years of spending in cash and short Treasuries and treat it as a strategic asset, not a drag. In an environment where every long-duration asset is priced off the same discount rate, the cash sleeve is the one holding that is not. Its job is not return; its job is to make sure no forced sale ever occurs during a correlated drawdown. That job is worth its low yield.

4. Do not add complexity to solve expensiveness. Before adding any asset class, ask: is this priced off the same discount rate as what I already own? If yes, it changes the label on your risk, not the amount. Most alternatives, real assets, and private-market products fail this test.

5. Keep the low-cost index core as the centre of gravity and let every diversifier orbit it. Nothing in this chapter argues for replacing broad global index exposure. Inflation-linked bonds, cash, and any uncorrelated sleeve are additions of a defined, modest size — hedges alongside the core, never substitutes for it. A retiree who dismantles the index core in pursuit of better diversification has traded a known, cheap, transparent risk for an unknown, expensive, opaque one.

Retirement Application

For a retiree, the practical translation is a change in what the bond sleeve is for.

Under the old regime, the bond sleeve did two jobs simultaneously: it produced meaningful income, and it rallied when equities fell. When starting yields were near zero, it could do neither well — the income was gone, and the capacity to rally was mathematically exhausted because rates could not fall much further. A 40% allocation to an asset that provides neither income nor crash protection is not a defensive sleeve; it is a large position held out of habit.

The honest reconstruction splits the bond sleeve's jobs across the instruments that can actually perform them:

• Crash protection comes from intermediate high-quality nominal government bonds, sized for the growth-shock scenario, and from the cash buffer.
• Inflation protection comes from inflation-linked bonds held to match liabilities, not traded.
• Income comes from accepting that in a low-yield world, spending is funded by total return, not by natural yield — the Chapter 2 discipline.
• Sequence protection comes from the two-to-three-year cash sleeve, which is the only thing that reliably prevents selling into a correlated drawdown.

Note that starting yields have since risen materially, which restores much of the nominal bond sleeve's capacity to do both jobs. That does not invalidate the analysis; it demonstrates it. The bond sleeve's usefulness was always a function of its starting yield, which is precisely the chapter's point. The retiree's discipline is to re-examine that starting yield annually rather than to hold 40% bonds because a rule of thumb said so.

Risk Management

• Regime risk: A portfolio built for the negative-correlation regime is silently short the inflation-shock regime. Assume both will occur within a 30-year retirement, because historically both have.
• Diversification illusion: Holding twelve funds that all lose money when discount rates rise is concentration wearing a diversification label. Audit by shared risk driver, not by fund count.
• Complexity cost: Every asset class added brings fees, tax complexity, rebalancing burden, and behavioural difficulty. If a diversifier does not address a risk driver you actually hold, its costs are certain and its benefit is not.
• Cash drag misjudgment: The most common error is dropping the cash buffer because it "yields nothing." Its return is not measured in yield; it is measured in the equity shares you did not have to sell at the bottom.
• Home bias: Concentrating in the most expensive market in the world because it has performed best is the specific version of this chapter's error most retirees are actually making.

What This Chapter Cannot Do

This chapter cannot deliver an allocation that escapes low expected returns. There is no combination of assets that produces 1990s returns from 2020s prices — that is arithmetic, not pessimism, and any product claiming otherwise is selling leverage, illiquidity, or a fee-shrouded version of the same beta.

Nor can it tell you when the inflation-shock regime will arrive or how long it will last. Regime identification is reliable only in hindsight, and a retiree who repositions on a regime forecast is market timing with extra steps.

What it can do is stop the specific error of responding to low returns with more expensive complexity, and redirect that energy toward the two diversifiers that genuinely work and cost nothing: a cash buffer sized to survive a correlated drawdown, and spending flexibility.

---

Key Takeaway: 60/40 was never diversified against the risk that mattered most — it was two bets on one falling discount rate, which felt like magic for forty years and stopped feeling like magic in 2022. Adding a seventh expensive asset does not fix expensiveness. The diversifiers that genuinely work against correlated expensiveness are inflation-linked bonds, a real cash buffer, and spending flexibility — all held around a low-cost index core, never in place of it.