Why does the mix explain most of what a portfolio does?

There is a famous claim in this field, and it is quoted wrongly more often than it is quoted correctly. Getting the precise version right is worth a lesson, because the sloppy version leads people to conclusions the research never supported.

What the research actually found

In 1986 Brinson, Hood and Beebower studied large US pension funds and reported that policy asset allocation explained about 93.6% of the variation in returns over time within a given fund. Ibbotson and Kaplan revisited the question in 2000 and separated three questions that had been getting mashed together:

• How much of a single fund's variability over time is explained by its allocation? Roughly 90%. This is the number everyone quotes.
• How much of the difference between one fund and another is explained by allocation? Roughly 40%. Much lower — because funds with similar allocations still differ through security selection, timing and fees.
• How much of the level of return is explained by allocation? Roughly 100% on average, by construction: across all investors, active decisions are a zero-sum game before costs.

The sloppy version — "asset allocation determines 90% of your returns" — merges all three and is simply not what the studies say. The defensible version is: the mix is the dominant driver of how a portfolio moves through time.

Why this has to be true

You already computed the mechanism in Unit 1. Take a mix of 60% equity (15% volatility) and 40% bonds (5% volatility). Equity supplies about 95% of the portfolio's variance. Now imagine a heroic stock-picking year in which the equity sleeve beats its index by 3 percentage points. The portfolio gains 0.6 × 3 = 1.8 percentage points from that skill.

In the same year, the equity market itself might move −18% or +26%. The selection decision moved the needle by under 2 points; the allocation decision determined exposure to a swing of 40 points or more. That ratio is the whole finding.

A worked comparison

Three illustrative portfolios in a year when equities return +20% and bonds return +2%:

• 80/20: (0.8 × 20) + (0.2 × 2) = 16.0 + 0.4 = +16.4%
• 60/40: (0.6 × 20) + (0.4 × 2) = 12.0 + 0.8 = +12.8%
• 30/70: (0.3 × 20) + (0.7 × 2) = 6.0 + 1.4 = +7.4%

A 9-point spread in outcome, produced by nothing but the mix. Now run the same three through a year of equities −18% and bonds −13%: −17.0%, −16.0%, −14.5%. Note what happened — in that year the mix barely helped at all, because both classes fell together. The mix is the dominant lever, and it is not an omnipotent one.

What this means in practice

The professional consequence is one of sequencing, not of superiority. The allocation decision gets made first and reviewed on a slow cadence, because it is the decision with the largest effect. Security selection sits underneath it, because it is the decision with the smaller effect. That is a statement about where the variance lives — not a claim that any particular mix is correct, and not a claim that selection is worthless.

Try it now

1. The two sleeves are below, drawn on adjusted closes so what you read is total return rather than price. Switch both to Monthly and Measure each calendar year of the last ten on both charts. You now have ten pairs of annual returns.

2. For each year, compute the return of an illustrative 60/40 and an illustrative 30/70 mix using the weighted-average arithmetic above. Which years show the widest gap between the two, and which show almost none?
3. Identify the single year in your table where the mix mattered least. What was happening to both classes at once? One neutral sentence.