The Modern Glidepath: De-Risking Within Equities

The consensus, stated properly

The classic glidepath rests on solid foundations, and we start by making its case rather than caricaturing it.

The premise is right. As the horizon shortens, the arithmetic of Chapter 26 becomes binding: equities need time to convert volatility into return, and money that will be spent within a few years cannot ride a −40% drawdown. Reducing risk as spending approaches is not a preference, it is a constraint.

Bonds have historically been the right stabilizer. Over most of modern history, high-grade government bonds delivered positive real returns, low volatility relative to equities, and, critically, a negative correlation with equities during crises. In 2008, long Treasuries rallied hard while equities fell by half. That crisis-rally property is the single strongest argument for long duration in a de-risking portfolio, and it is not a small one.

The implementation is well-engineered. Target-date funds automate the entire path, rebalance internally, and eliminate the decision entirely (Chapter 27's "discipline as a product"). For most people, a target-date fund is a genuinely excellent default, and this chapter's disagreement is with one component of it, not with the concept.

Our view, and its reasoning

We think the calm side of a de-risking portfolio is better held in short-duration instruments than in long bonds, and that part of the de-risking should happen inside the equity sleeve.

Three arguments.

1. Duration is a large risk, and it is not the one you were de-risking for. A 20-year government bond fund has a duration near 16: a one-point rise in yields costs roughly 16% (Chapter 22's see-saw). An investor five years from retirement who shifts from equities into long bonds has not reduced risk so much as swapped equity risk for interest-rate risk, in similar magnitude. 2022 made this concrete: long government bond funds fell 25–40%, an equity-sized drawdown, in the sleeve that existed to prevent equity-sized drawdowns.

2. The correlation that justifies long bonds is conditional, not structural. Equity-bond correlation has flipped sign across history: reliably negative in the low-inflation decades from the late 1990s, positive through much of the 1970s and again in 2022. The mechanism is intuitive: bonds hedge growth shocks (recession, deflation) and fail against inflation shocks, because inflation hurts both assets simultaneously. Building a retirement plan on an assumption that only holds in one inflation regime is a concentration of a subtler kind.

3. Short-duration assets now do the job with far less machinery. Money-market and 0–1 year instruments (Chapter 44) deliver the policy rate with duration near zero. When short rates are meaningfully positive, the investor gets most of the "safe asset" return with essentially none of the drawdown risk. The classic argument against cash, that it earns nothing, was an artifact of the zero-rate decade rather than a permanent truth.

The within-equities part. Rather than moving the whole reduction into fixed income, some of it can happen inside the growth sleeve: from concentrated to broad (a single-country or mega-cap-heavy holding to an all-world fund), and optionally toward lower-beta exposure. This keeps more of the portfolio's long-run return engine intact while genuinely reducing the depth of drawdowns, and it costs nothing in fees or complexity.

What this looks like in practice

A worked structure for an investor approaching a spending phase (adapt the ages, the shape is the point):

• Growth sleeve: stays in broad global equity throughout. What changes as the horizon shortens is its share, not its quality. Any concentrated positions (single-country, sector satellites from Chapter 61) are unwound first: the cheapest risk reduction available, because it costs no expected return you were reliably being paid for.
• Calm sleeve: money-market or 0–1 year government funds for anything needed within about three years; short-to-intermediate high-grade bonds for the three-to-seven-year band; duration never materially exceeding the horizon of the money it backs (Chapter 43's matching rule, the operating principle here).
• Dated liabilities: target-maturity bond ETFs for known future spending (Chapter 43), which convert the entire duration question into a date.
• Long bonds: not banned. Held deliberately by investors who specifically want the crisis-rally property and can tolerate the 2022 scenario, and sized as a decision rather than a default.

Notice that this structure never requires a rate forecast. It matches duration to horizon and lets the yield curve do whatever it does.

The case against our view

Printed at length, because a house view without its counterarguments is marketing.

1. It leans on one episode. 2022 was a single, unusually violent inflation shock. A view constructed largely from the most recent painful year is the definition of recency bias, and the classic glidepath survived many decades before that year. This is the strongest objection, and we accept it as unresolved rather than defeated.

2. It gives up real crisis protection. In a deflationary crash such as 2008 or March 2020, long bonds rallied hard and cushioned portfolios substantially. Short-duration assets do not rally; they simply fail to fall. An investor who de-risks into cash-like instruments has traded a potential large gain in a crisis for the certainty of not losing. Over a full cycle, that trade may cost real money.

3. Yields change the argument. Much of our case is stronger when short rates are positive and long yields are low or rising. Should long yields sit at high real levels, locking them in for a decade becomes considerably more attractive, and the classic approach improves accordingly. Our view is regime-sensitive and we say so: it is a judgement about instruments under conditions, not a law.

4. Reinvestment risk is real. Short-duration holdings must be rolled continuously at whatever rate exists then. A retiree in cash-like assets through a sharp rate-cutting cycle watches income fall, while the long-bond holder locked theirs in. Duration matching cuts both ways.

5. The consensus is implemented better. Target-date funds do the classic path automatically, cheaply, and without you. Our view requires slightly more attention, and an approach that demands attention has a failure mode the automated one does not.

How to decide for yourself

The question is not which view is correct in general, but which failure you are less able to absorb:

• If a −30% fall in your "safe" assets five years before retirement would derail your plan, the modern approach removes that scenario.
• If missing a large bond rally during an equity crash would leave you materially short, the classic approach retains it.
• If you will not maintain the structure, take the target-date fund and stop reading this chapter; an automated good plan beats a manual better one.

Whichever you choose, three points are not in dispute and matter more than the debate: risk must fall as the horizon shortens, duration must be matched to the horizon of the money it backs, and the decision belongs in the written policy from Chapter 19 rather than in your reaction to whatever happened last year.