I did a presentation on this to the guys at my work.
Here is a quick cut and paste job from it.
_____________________________________________________________________
Modern portfolio theory often quotes that investment returns are largely derived as a result of the asset allocation decision rather then the stock selection or market timing decision. This leads us to ask the following two questions:
1. How much of the variation in returns (across time) is explained by asset allocation?
Studies (by Brinson, Hood and Beebower) have examined the effect of asset weights on investment performance, using data from a variety of funds from the early 70’s to the late 90’s. The study found 93.6% of a funds return over time can be explained by its target asset allocation.
Unfortunately, this study is often misrepresented. The 93.6% result has also been commonly used to explain how much impact asset allocation has on the different returns between funds (i.e. across funds).
Also, some investors quote that the over 90% result means that asset allocation will explain over 90% of any portfolio’s return – regardless of the make up of the portfolio. Individual investors portfolio’s typically aren’t as well diversified as big pension funds. The 93.6% result was based on a study that only sampled highly diversified pension funds.
2.How much of the variation of returns (across funds) is explained by differences in asset allocation?
Studies (by Ibbotson & Kaplan) show about 40% of the variation of returns across funds is explained by asset allocation policy. For example, if one fund returns 5% more than another, then on average about 2% of the difference (i.e. 40% of 5%) is explained by a different asset allocation, while the remaining 3% difference (i.e. 60% of 5%) is explained by stock selection, timing and fee differences between the funds.
Opponents to the Brinson, Hood and Beebower (BHB) study
The BHB study demonstrated that the entire group of pension funds failed to add value through either market timing or stock selection. Therefore if you had purchased the entire group, you would have received (minus costs) roughly the returns of their asset allocations.
Opponents to the study and the notion that “93.6% of returns over time can be explained by asset allocation” argue that the study only examined the long term total-return averages for a group of pension funds.
The opponents claim that the BHB study only looked at large pension funds, most of which hired multiple sub-advisors and thus were conglomerates of money managers. They claim “the result shows more about the nature of money management in pension funds, rather than what asset allocation means to a single investor in practice”.
BHP arrived at their result by computing the quarterly performance of the pension funds to an indexed version of that asset allocation, and then plotted that against the fund’s actual quarterly returns. The average R-squared statistic for the funds was 93.6. The opponents claim it is very difficult to draw firm conclusions from an R-squared statistic.