ETFguide.com - August 29, 2007
The idea that market indexes can be fundamentally weighted by important financial metrics hasn't just shaken the ETF industry - it's shaken the entire world of modern finance.
It's sparked a heated debate between supporters of traditional market cap weighted indexes and a newer generation of indexers that think they have a better solution.
At the epicenter of this controversy is Robert Arnott, Chairman of Research Affiliates. His firm has been pioneering new ways of constructing market indexes and he takes time to visit with us.
Q: Are fundamentally weighted portfolios really "indexes?"
RA: It depends entirely on how one defines “indexes.” From a CAPM perspective, anything that’s not cap-weighted is neither passive, nor is it an index (although the indexing community leans towards the decidedly non-CAPM-compliant perspective that float-weighting is somehow preferable). By this definition, "Fundamental Index" is decidedly neither passive, nor an index. If, alternatively, we define an index as something which is formulaic, objective, transparent, historically replicable and low-turnover, Fundamental Index qualifies on all counts. Ironically, many of the cap-weighted don’t qualify for this simple pragmatic definition of an “index.”
I don’t really care whether people think that Fundamental Index is an index or an active portfolio. Either way, it’s an utterly simple concept with powerful practical and theoretical implications. Indeed, the empirical evidence surrounding Fundamental Index calls some well-respected theories like efficient markets and CAPM into some question. Rather than arguing about whether these are indexes or not, shouldn’t we (1) enjoy the risk-adjusted alphas that come from the concept, and (2) recognize that the elegant theories of modern finance may need some tweaking to better fit the real world?
Q: Regarding the turnover of holdings and rebalancing, how do RAFI's fundamentally weighted indexes compare to market cap-weighted ones?
RA: The Fundamental Index concept can be anything from extremely low turnover to extremely high turnover. If one wanted to rebalance back to the fundamental weights for individual company’s every single day, the turnover could easily exceed 100 percent per annum. If one does it quarterly, the turnover tends to be 30-50 percent. If one does this annually, the turnover tends to 15-25 percent. And if one does it annually based on smoothed long-term measures of the fundamental scale of the company, the turnover drops to the 10-15 percent range. This compares quite respectively with the average turnover of the capitalization weighted indexes, which range from roughly 6 percent (the 45-year avg. turnover for the S&P 500) to 30-40 percent for niche indexes likes Russell 2000 value.
Q: Fundamental Indexes, can use many financial measures (book value, earnings, dividends, etc.) - is there any one single metric that's more important than the next?
RA: Empirically, some fundamental measures are more powerful than others, but the range is surprisingly narrow. The gap between the best and worst single-metric approach, a sales-weighted Fundamental Index and a dividend-weighted Fundamental Index, is barely 90 basis points averaged annually over the past 45 years. Sales is the wellspring from which profits, book value and dividends must flow; so, it makes sense that sales works best and dividends works worst. But there is also a link with risk. A sales-weighted Fundamental Index is the most volatile and works best in bull markets, while a dividend-weighted Fundamental Index is the lowest-volatility and works best in bear markets (while failing miserably in most bull markets).
This is why we favor using a blend of measures. It results in a lower tracking error than any single-metric Fundamental Index, lower volatility than the average Fundamental Index, higher returns than the average Fundamental Index and the highest information ratio of any Fundamental Index. It also has, by good margin, the lowest turnover.
The most important single metric is not even part of the Fundamental Index: it’s the cap-weighted index. While no Fundamental Index is even 0.5 percent away from the average return of RAFI, cap-weighting is 220 basis points off the pace. Accordingly, avoiding cap-weighting (or indeed any price-sensitive weighting scheme) is vastly more important than the selection of which Fundamental Index to rely upon.
Q: Many of RAFI's stock indexes have a bias toward value and smaller cap stocks. What happens when these areas are out of favor?
RA: RAFI weights companies by their economic footprint. Accordingly, it is utterly neutral relative to the composition and weightings of business enterprises in the economy. Cap-weighting, in contrast, have a stark growth tilt. Companies at twice the market multiple get double their economic weight in the cap-weighted portfolio, while companies at half the market multiple get half their economic weight.
In 1997, Cisco was 0.5 percent of the market, at thirty times earnings. By 2000, it was 4% of the market, at one hundred thirty times earnings. Did it comprise 8 times as much of the market as the peak of the bubble because it was 8 times as attractive at a PE of 160 than it was at 30 times earnings? Of course not. The weighting went up 8 fold because the stock went up 8 fold relative to the average stock, and because it was now at 160 times earnings.
Our Fundamental Index doesn’t get drawn into this sort of bubble, though it also doesn’t weight growth companies more heavily than value companies even when they deserve the higher multiples.
Q: What about other securities besides stocks? Can fundamental weightings be applied?
RA: The short answer is “yes.” I can’t go into detail on this question, because it’s an area of active research at Research Affiliates. That work is hopefully protected by the same pending patents as RAFI. We have a draft paper with Harry Markowitz and Jun Liu, which explores the mathematics behind Fundamental Index. So long as price is randomly distributed around value, not the other way around (which Efficient Markets Hypothesis assumes), the value that is forfeited by cap-weighting is proportional to the square of the pricing error.
Accordingly, if you double the pricing error, you quadrupled the value-added in a valuation-indifferent index. This has implications for the bond world, where pricing errors are small, so a bond-based Fundamental Index should add a trivial increment to returns; reciprocally, in emerging markets, the pricing errors are presumably huge, and the impact of RAFI on performance is quite remarkable.
Q: Thanks Rob.