RAFI ETFs: A Deeper Look at the Performance


For about the past ten years, there has been an on-going debate in the world of indexing centering around whether index funds should be weighted by capitalization (the traditional method) or by fundamental factors such as dividends, sales, book value, et al. Proponents of fundamental indexing use what they call the “noisy market hypothesis” as their justification. Specifically, since the market is not populated by ultra-rational wealth-maximizing automatons but rather by human beings who are subject to behavioral pitfalls, the market will overprice some securities while underpricing others. Of course, we have no sure way of knowing which ones are which, but the proponents of fundamental indexing tell us that as long as we break the link between price (capitalization) and weighting, then we are bound to minimize the problem of being overweight in overvalued stocks while being underweight in undervalued stocks. The term “smart beta” has been used in conjunction with this approach.

The fallacy of this argument was put to bed by Andre Perold of Harvard Business School in Fundamentally Flawed Indexing. The argument can be analogized as follows. Suppose that you are shopping in a jewelry store and you see two different 1-karat diamonds. One sells for $1,000 and the other for $1,500. As far as you are concerned, they both look the same. However, the jeweler tells you that he underpriced one of them and overpriced the other. Based on your intuition, you immediately decide that the $1,500 diamond is the overpriced one while the $1,000 diamond is the underpriced one. However, a few moments of reflection would lead you to the conclusion that there is an equal probability (50%) that the $1,500 diamond is the underpriced one. Since you are not a gemologist, you have no good way of knowing. Investors (and fund managers) who believe that higher priced companies have a higher probability of being “overpriced” are making the same mistake.

While we disagree with the underlying argument of fundamental indexing, we cannot ignore the fact that it leads to portfolios that are similar to our own in that they are tilted towards small cap and value stocks. One other aspect of fundamental indexing that deserves a second look is the possible tilting towards more profitable companies, which is now a recognized dimension of expected returns.

The leading provider of fundamental index data is Rob Arnott’s Research Affiliates (RAFI), and the Powershares ETFs have been using RAFI indexes since 1/1/2006. The four factors used by RAFI for weighing the securities used in its indexes are sales, cash flow, dividends, and book value. The first two factors may result in a tilt towards more profitable companies. We believe it is important to look at live fund data because there are higher trading costs for funds that are not market cap weighted. The following charts show the difference between RAFI and similar asset classes (U.S. Large Value, U.S. Small/Mid Cap Value, International Value, International Small Value and Emerging Markets Value).


Overall, the results look favorable for the PowerShares RAFI-based ETFs, but none of them are statistically significant. Although we cannot prove it, we hypothesize that the primary explanation for the outperformance of the PowerShares ETFs is their increased exposure to high-profitability companies. Going forward, we know that the DFA funds will be fully engaged with profitability, so we do not have a compelling reason to advise the use of any of these funds at this time.

Below are the RAFI-based ETFs compared to their Morningstar assigned benchmarks. At IFA, we focus on the t-stat and look for a t-stat great than 2. We have not seen any yet.

We have taken a deeper look at the performance of several other mutual fund companies and have come to one universal conclusion: they have failed to deliver on the value proposition they profess, which is to reliably outperform a risk comparable benchmark. You can review by clicking any of the links below:

Here is a calculator to determine the t-stat. Don't trust an alpha or average return without one.
The Figure below shows the formula to calculate the number of years needed for a t-stat of 2. We first determine the excess return over a benchmark (the alpha) then determine the regularity of the excess returns by calculating the standard deviation of those returns. Based on these two numbers, we can then calculate how many years we need (sample size) to support the manager's claim of skill.