Wander through a few web sites or read a few books on personal investing and you are sure to come across references to the 1986 study Gary P.Brinson, L. Randolph Hood, and Gilbert L. Beebower, Determinants of Portfolio Performance, The Financial Analysts Journal, July/August 1986 (henceforth BHB) or to its later update Gary P. Brinson, Brian D. Singer, and Gilbert L. Beebower, Determinants of Portfolio Performance II: An Update, The Financial Analysts Journal, 47, 3 (1991) (which I have unfortunately been unable to find posted online ... perhaps its lack of wide availability itself contributes to the misuse?).
The studies are typically summarized something like this: "asset allocation accounts for 90% of a portfolio's return" or "asset allocation decisions are the most important factor affecting portfolio returns". This is then used to justify things like the need to develop an asset allocation plan, to invest based on passive index investing, to buy ETFs and not mutual funds, to not worry about which mutual fund to buy as long as it fits into the right asset class. Wrong!! Much as I subscribe to many of these ideas, the study provides no such support as subsequent follow-up studies have pointed out.
Let's do a myth vs reality rundown to disentangle truth from lore or illogic.
Myth: BHB says asset allocation, aka the choice amongst asset classes of stocks, bonds or cash in the portfolios they looked at, determined the level or amount of return achieved e.g. 8% vs 10% return in a year
Reality: BHB looked only at the variability over time, in crude terms the amount of up and down, of individual portfolios relative to the benchmark for the type of asset;
Researchers Ronald Surz, Dale Stevens and Mark Wimer in Investment Policy Explains All also examined investment policy or asset allocation at mutual and pension funds and came to this conclusion: "... investment policy, on average, accounted for 104% of the total return for mutual funds and 99% of the pension fund results." Then there was Roger Ibbotson and Paul Kaplan's (IK) Does Asset Allocation Policy Explain 40, 90, or 100% of Performance? in which they conclude yes to all three depending on the precise question being asked.
As Surz et all show with a brilliant simple example, if a portfolio moves in perfect lockstep with its index year after year consistently returning either 2% above or 2% below the index, it would have the exact same performance in BHB terms; BHB says nothing about the absolute level of returns; as Surz et al put it, "... their study tells us only that the average plan in the sample adhered very closely to its policy targets and used broad diversification within asset classes."
Myth: for every fund (or the bulk of, or the typical fund) asset allocation determined variability of returns
Reality: BHB took an average of many funds; to understand the logical fallacy, think of the old joke, if your feet are in the oven and your head is in the freezer, on average you will be comfortable; IK found a wide dispersion amongst balanced mutual funds - only 40% of the variability of their results arose from differences in asset allocation policy: "... the remaining 60 percent is explained by other factors, such as asset-class timing, style within asset classes, security selection, and fees.", in other words "... the relatively low R2 of 40 percent must be the result of a large degree of active management". If a fund chooses to be aggressive and deviate from the index it can easily do so and its results will vary much more. A big deviation isn't necessarily good or bad - results can be way above or below the index.
Myth: the results of these studies say that no one can beat the market and therefore index, passive investing is the only way to go and
Myth: the asset allocation decision is the only one that matters
Reality: though I subscribe heartily to passive index investing based on an asset allocation policy, such studies imply nothing of the sort; the leap from description of what actually happens to saying that is what one should do, or that it is inevitable, doesn't make sense - if most countries of the world have corrupt, despotic governments does that imply we should aim for the same or that it is not or cannot be otherwise?
The various studies conclude that as a whole (i.e. on average) mutual funds detract from the total return compared to an index - when it is said that policy accounted for 104% of total returns, it means the mutual fund managers detracted 4% due to market timing, security selection and costs. That doesn't say no one can beat the market, it says not everyone has, few have and most have not.
It is very dangerous to take a sentence such as this in Surz et al out of context: "Manager selection matters, but not to any great extent." The proper expansion of the sentence would be: "Manager selection has not mattered to any great extent over the average of the many funds we looked at, so do not be surprised if it proves not to be so for any specific fund you choose, and considering that the sum of investors as whole ARE the market, it should be no surprise either that such is the case; however, manager selection has mattered greatly in certain cases." The authors are somewhat culpable in leading us down the slippery slope from description to prescription by the use of the present tense 'matters' as if to say it cannot be otherwise and by omitting the word 'average' as if to imply that statement is true in every case.
Here is what I think the prescriptive statement should be: If you are picking amongst actively managed mutual funds, then manager selection is your critical task. You should not just pick a fund at random in an asset class since you probably will get a fund that does worse than the index of that asset class.
IK say it this way: "An investor who has the ability to select superior managers before committing funds can earn above-average returns." Going with an actively-managed fund means you are swapping the challenge of picking the best stocks or bonds with that of picking the best manager.
Myth: the 90% figure matters or suggests that an investor should pay attention to asset allocation
Reality: though an investor should pay attention and implement an asset allocation, it is for other reasons, not because of these studies; the 90% figure means nothing more than the fact that a portfolio was invested in the markets. Meir Statman's The 93.6% Question of Financial Advisors (Spring 2000, Journal of Investing) models a hypothetical US portfolio which assumes perfect foresight and uses extreme asset switching going 100% from stocks to bonds or to cash each year with perfect predictive ability of the highest yielding asset class; this portfolio still has 89.4% of its returns accounted for by a balanced portfolio with a constant asset proportion policy, i.e. asset allocation could be thought to be the determining factor but the high correlation is deceiving. IK noted the same thing and concluded: "Hence, the high R2 in the time-series regressions result primarily from the funds’ participation in the capital markets in general, not from the specific asset allocation policies of each fund". Statman's artificial portfolio did achieve hugely greater total or absolute returns - 9% per year more compounded, indicating the value of perfect foresight!