Allocation and the Rebalancing Habit

The mix is the decision that does most of the work

How a portfolio is split between asset classes explains most of its long-run variance, which is why it deserves more attention than any single holding. The famous attribution study by Brinson, Hood and Beebower put the figure above 90%, and it has been argued about ever since — the number depends on what you are attributing, and later work has been more careful. The part that survives the argument is the ordering: the mix explains far more of the ride than manager selection does, and it is the only one of the two that a household fully controls. The rule that sets it is the calendar. Money needed within a few years belongs in cash and short bonds whatever the market looks like, because a drawdown in a balance you need is a crisis rather than a paper loss. Money with a decade or more belongs mostly in equities, because it needs growth and it has time to recover. General rules of thumb (“100 minus your age”) are shortcuts for people who have not thought about the horizon, and they can be badly wrong in both directions — a 30-year-old saving for a house deposit in three years is not a high-equity investor, and a 70-year-old with a pension covering their spending can rationally hold more equity than the rule suggests. “Risk tolerance” is worth splitting into two questions, because households routinely confuse them. **Ability** to take risk is about the horizon and the other resources available — it is a fact about the situation. **Willingness** is about what you will actually do when the portfolio is down 30%: hold, or sell. A plan has to satisfy the smaller of the two, because a portfolio that is right on paper and abandoned in a drawdown delivers the return of the abandonment, not of the paper. The mix, by horizon — Needed in under 3 years: cash, money market, short T-bills · Needed in 3–10 years: bonds and cash, weighted toward the date · Needed in 10+ years: broad global equities, with bonds for ballast ← · What the rule of thumb misses: the horizon is the input; age is a proxy for it The “100 minus age” shortcut assumes a retirement horizon. Applied to a house deposit it produces the wrong answer, and applied to a household with a pension covering its spending it produces a needlessly timid one.

Bonds are ballast, not returns

In a portfolio, bonds do three jobs and none of them is “win”. First, they dampen the ride: high-quality bonds usually fall less than equities in an equity crash, which keeps the total drawdown survivable and the household invested. Second, they fund near-term spending without forcing a share sale at the wrong moment — the same logic as the emergency buffer (PF3), extended across a long horizon. Third, they are rebalancing ammunition: when equities fall, the band rule sells bonds to buy equities, which is a mechanical version of buying low. The mistake is expecting bonds to deliver equity returns, particularly after a long period of low rates. Duration cuts both ways: a long-dated government bond fund lost roughly 30% in 2022, which was a reminder that “safe” describes the credit quality, not the price over a short window. Bonds with a horizon matched to the money they support behave; long bonds bought for yield behave like a volatile asset. A practical structure follows from this. Cash equivalents are the emergency layer and the next-two-years layer; short and intermediate high-quality bonds are the spending-in-retirement layer; equities are the long-horizon layer. Name the job of each sleeve and the allocation designs itself, which is far more useful than a single stock/bond percentage chosen by feel. The jobs of the sleeves — Cash and T-bills: spendable now — the buffer and the next two years · Short / intermediate bonds: dampen the ride; fund retirement spending and rebalancing · Broad global equities: long-horizon growth · Long-dated bonds bought for yield: a volatile asset with a credit-quality halo ←

Rebalancing in bands

A target is a range, not a number. Rebalancing on every drift would mean trading constantly, paying spreads and — in a taxable account — creating realised gains, which costs more than the small risk it corrects. The convention that has survived is the **5/25 rule**: rebalance a sleeve when it drifts five percentage points from its target or 25% of it, whichever comes first. On a 70% target, five points is 65–75%; on a 20% target, 25% of it is five points too. Households can also simply look annually, which is enough for most portfolios. The direction is the part that requires discipline rather than arithmetic. Rebalancing sells the asset that has just done well and buys the one that has lagged, so it feels like a mistake at the moment it is executed and it is the only mechanism in the plan that enforces buy-low-sell-high without anybody knowing anything about the future. It is not free money — in a mean-reverting market it adds a little, in a trending one it costs a little, and its real job is risk control: keeping the portfolio at the risk level the household chose. Three implementation rules keep the cost near zero. Rebalance with **new contributions** first, since redirecting money costs nothing. Do it **inside the tax-advantaged accounts**, where a trade has no tax consequence. And use **wide bands** when the account is taxable, because a realised gain is a permanent cost while the drift is usually temporary. None of that requires a forecast, and none of it requires watching the market daily — the opposite, in fact. One year of drift, three points wide — Before: $175,000 equity · $75,000 bonds · $250,000 · After +20% / +2%: $210,000 equity · $76,500 bonds · $286,500 · Equity weight: 73.3% — inside a five-point band · Action: none required · new contributions can go to bonds instead ← The trade is small and the drift is three points, which is why the habit is a glance and a band rather than a monthly event. Doing nothing is a legitimate rebalancing decision.

The currency exposure hiding in the allocation

The stock-and-bond split gets the attention, and there is a second decision inside it that most households never make deliberately: how much of the equity sleeve is foreign, and whether the currency exposure that comes with it is a feature or a bug. The mechanics first. An unhedged investment in another country’s shares carries two return streams: the local price move and the currency move. A fund holding European equities and quoted in dollars will rise when European shares rise, fall when the euro weakens, and produce a total return that is the combination. For a household whose spending is in dollars, an unhedged foreign holding is therefore not a pure equity exposure — it is an equity exposure plus a long position in another currency, and that second position has its own volatility, which is roughly comparable in magnitude to equities over a year. What that implies depends on the horizon, and the evidence is reasonably consistent. Over a single year, currency moves can dominate: a foreign equity fund’s return can be substantially different from the underlying market’s simply because of the exchange rate. Over long periods, the currency effect tends to average toward zero for major pairs — a currency cannot depreciate forever without rebalancing the trade balance that drives it — so the long-horizon investor is mostly holding equity exposure with a lot of short-run noise attached. The noise is real and it is not a reason to avoid the exposure; it is a reason to know where it sits. Then the two decisions that are usually conflated. **How much foreign equity** to hold is a diversification question, and the case for a substantial allocation abroad rests on the fact that markets are not perfectly correlated and no single country’s share of world earnings is a natural neutral point. **Whether to hedge the currency** is a separate question: currency-hedged share classes exist for most developed markets, they remove the short-run noise, and they cost something — the hedging has a price that is roughly the interest-rate differential, which can be several percent a year for currencies whose rates differ sharply from the dollar. So a hedged position gives up the interest carry to remove the currency volatility, which is a fair trade for a bond portfolio and a much less clear one for an equity portfolio held for decades. The practical shape for a household is unglamorous. Decide the foreign share as part of the mix and write it down, so it does not drift with whichever market has just run. Leave developed-market equity unhedged if the horizon is long and the spending is in one currency at the end, because the hedging cost is a certainty against a noise that averages out. And hedge anything with a short horizon and a fixed liability, because there the currency move is not noise, it is the difference between having the money and not. • Unhedged foreign equity is equity plus a long position in another currency. • Currency moves can dominate a one-year return and average toward zero over decades. • The foreign share and the hedging decision are separate questions. • Hedging costs roughly the interest differential — worth it for a fixed liability, less so for a long horizon. A simple diagnostic for a portfolio that holds foreign funds: compare the fund’s return with the underlying index’s local-currency return over a year. The gap is the currency and the hedging cost, and seeing its size once usually settles whether it belongs in the plan.

Sources

• Asset allocation and the majority of return varianceBrinson, Hood & Beebower attribution research (with the well-known caveats on its interpretation)
• Glide paths and target-date allocationVanguard — target-date methodology; SEC investor bulletin
• Rebalancing bands, thresholds and the 5/25 conventionBogleheads wiki — rebalancing; practitioner guidance
• What long bonds did in 2022Bloomberg U.S. Aggregate / long Treasury index returns, 2022