So why does old, stale data work better? The reason is relatively straightforward. The Fundamental Index advantage comes from breaking the link between price and portfolio weight, rather than from having the “right” weight. What drives price movements? Recent changes in financial results, which alter our expectations for the future! So, by using shorter term results to construct fundamental weights, we tacitly re-introduce price back into the equation, through the “back door.” This results in demonstrably worse returns.
Multiple Measures—Avoiding the Biases of Single Factor Portfolios
Which is the largest company in the United States—Exxon, Walmart, General Electric, or Bank of America? The answer is… all of the above. Walmart ranks first on five-year average sales. Exxon ranks first on five-year average cash flow or earnings or market cap. General Electric is at the top of dividend payer list for the past five years. And, even after all of its write-offs, Bank of America remains the largest company by book value.
There is no right answer. If the goal is to build an alternative equity beta that can serve as a core portfolio position, shouldn’t the resulting portfolio be representative of the economic opportunity set? When we rely on only one metric, we begin to lose the broad representation that is a key part of index investing. Structural biases pop up in a variety of unintended ways. For example, using only sales leaves the portfolio overexposed to large companies with thin margins. Using only cash flow or profits may lead to a bias toward cyclical stocks at profit peaks. An emphasis on dividends will naturally favor mature, high-yield companies and will largely exclude growth companies. Lastly, book value may favor firms with aggressive accounting.
Unquestionably, using a single measure of firm size can expose investors to a skewed sample of companies that fails to adequately reflect the economy. The introduction of multiple measures of firm size mitigates such biases, leading to a more efficient performance profile and lower portfolio turnover. The average information ratio of the four single metric Fundamental Index portfolios from 1962–2010 was 0.43 relative to a portfolio of the top 1,000 companies measured by capitalization. The RAFI composite, which equally blends the four metrics, was 0.47, roughly a 10% increase in efficiency. The methodology also results in more stable weights over time and, thus, lower turnover. In our original research, the annualized turnover of single metric portfolios was 12.5% from 1962–2004, which fell to 10.6% when we utilized a composite approach.5
Selection Bias Strategy—Why Start with a Capitalization Roster When Building a Fundamental Index?
One of the oft-overlooked index design details is the importance of selection criteria—which securities are included within an index. Stocks with rising market capitalizations make their way into large-cap indexes, such as the S&P 500 Index, while ones with shrinking market caps drop out.
The tech bubble of the late 1990s illustrates what happens when inclusion in an index is based on relative capitalization. As Figure 1 shows, a simulated cap-weighted index of the top 1,000 U.S. stocks had 123 technology companies at the beginning of 1998. With the S&P 500 Technology Sector rising 218% over the next two years, the number of technology firms making the top 1,000 by market cap nearly doubled to 223. The cap-weighted index added these stocks because they had high prices. Even the most casual market historian knows this dynamic would have led to significant erosion in returns due to the tech-led bear market of 2000–2002.
Who gets shoved off of the list by these high flyers? Big companies that have recently depressed prices. Why? Because they have low prices. Is there a performance differential between these high-priced small companies and low-priced big companies? Yes. Our work indicates that these small-cap large companies outperformed their large-cap small companies by nearly 1,000 basis points per annum in the United States since 1962. By fundamentally re-weighting a cap-weighted starting universe—but failing to adjust the selection criteria—we miss these wonderful opportunities.
How much are those opportunities worth? A lot. As Table 2 shows the impact of selection bias ranges from 59 basis points per year to 376 basis points per year; the more inefficient the market, the greater the outperformance. Interestingly, the ratio of the value-add between selection and weighting remain relatively constant at 23% to 34% across markets. In other words, re-weighted indexes eliminate the RAFI selection effect and re-introduce a link with price which the Fundamental Index methodology seeks to avoid. This costs us one-fourth to one-third of the potential benefits of a true Fundamental Index strategy.