Academic researchers armed with ever more complete data sets, computing capability and more sophisticated statistical and mathematical techniques have been pushing forward the boundaries of what is the best. The original US equity index, the Dow Jones Industrial Average, has long been discarded as inadequate. Now the S&P 500's limitations as an index are too apparent. It simply doesn't do a good job compared to alternatives.
A couple of papers from the EDHEC - Risk Institute (an arm of the largest French business school by the same name) test the cap-weighted S&P 500 against a number of alternatives and find large, statistically significant differences in favour of the alternatives. The papers:
- A Comparison of Fundamentally Weighted Indices: Overview and Performance Analysis by Noël Amenc, Felix Goltz and Véronique Le Sourd (March 2008)
- Efficient Indexation: An Alternative to Cap-Weighted Indices by Noël Amenc, Felix Goltz, Lionel Martellini and Patrice Retkowsky (January 2010)
- Equal Weight S&P 500
- Various Indices using Fundamental Financial data to select and weight stocks for each index -
- FTSE GWA US index
- Dow Jones US Select Dividend
- AMEX Dynamic Market Intellidex
- WisdomTree Domestic Earnings- or Dividend-Weighted Indices - 4 permutations of each kind
- FTSE / Research Affiliates RAFI 1000 Index
- Mergent/AMEX Broad Dividend Achievers Index
- VTL Associates / Standard & Poors Revenue-Weighted Large Cap Index
- Every Fundamental index had higher returns than the standard S&P 500 and the Total Market index; the best outperformed by a statistically significant margin: the RAFI, the Intellidex and the WisdomTree HighYield Equity Index (now called the Equity Income Index)
- None of the Fundamental indices does better than the Equal Weight index (most do worse), in fact a couple of WisdomTree indices underperform the S&P 500 by a statistically significant margin.
- Fundamental indices had long periods - almost 16 years for the worst index - when their performance lagged the S&P 500, and the lag was quite large - over 50% in the worst case. Among the indices with a long history, the RAFI was the best with only 88 months underwater and 22% maximum performance lag. The killer period for relative performance of all the fundamental indices was the 1990s, the time of the tech bubble!
- Some of the indices provide true alpha (excess return) compared to the S&P 500 - the RAFI, Intellidex, WisdomTree Dividend and HighYield Equity.
- When taking account of exposure to small cap size, value (based on stock book value) and price momentum, it turns out that most of the out-performance is due to a value tilt by all these indices. There is still a bit of statistically significant alpha in the RAFI, the FTSE GWA and the Intellidex. The WisdomTree HighYield comes close but doesn't quite make it.
- Analysis of sector weightings show that almost all of the variability of fundamental indices is due to sharp divergences in weightings with what the S&P 500 contains - much less Tech, Telecom and Health Care and much more Utilities, Consumer Discretionary and Finance. As the authors say, the fundamental indices underweight typical growth sectors and overweight value sectors. Despite taking this into account, a few indices still show positive alpha - RAFI, Intellidex, WisdomTree HighYield and Dividend 100.
- Net selectivity measures the success in adding value after adjusting for total risk in addition to the alpha, which took account of systematic risk. Almost all of the indices had positive net selectivity over the S&P 500 even after after deducting the alpha.
- By constructing a theoretical optimal portfolio (in terms of the highest Sharpe ratio aka greatest bang for the buck, or return divided by standard deviation) , they conclude that none of the tested indices is optimal.
They call the indices "nothing but value-tilted active indices". That is not a condemnation or a dismissal of their value. Indeed, they say, "... the main value added of these indices may be to provide investors with a liquid, systematic, and relatively cheap alternative to other value-tilted strategies."
The opportunity for an investor to incorporate in a portfolio the proven value of a value tilt with one fund instead of buying another to provide it is intriguing. In fact, one index, the Intellidex also includes a small cap tilt and a momentum tilt and both these tilts have been found by researchers to be beneficial.
The indices are tangible and based on real data but that is not the end of the process to decide to jump into the ETFs or mutual funds that implement the indices as their guide for investment.
"In practice, of course, management fees, transaction costs and out-of-sample performance may reduce the value of characteristics-based indices." The alpha must be at least greater than the expenses difference to compensate. Other features of funds that affect real investor returns like index tracking error, transaction costs from rebalancing and reconstitution, bid-ask spreads and divergences between NAV and market price of a fund must also be taken into account to figure out if there is a net advantage for an investor at the moment.
'Out-of-sample' means the future may not be like the past. The data used in the study went up to December 2006 so we are now in the test period that will answer whether the past out-performance results will continue. Are we in period when the fundamental approach will lag, another time of irrational exuberance and a bubble? It is impossible to know for sure.
Bottom Line Takeaway
I'm impressed and convinced that some of the fundamental indices offer a better solution than the S&P 500. The indices that seem to be worth pursuing, based on the paper's results and comparisons, from most to least promising, are:
- RAFI - long back data, probably more reliable results, consistently shows positive statistically significant alpha across many measures, though its alpha at 1.9% is less than the others; fell in the low end below the S&P 500 during its period of under-performance from 1993 to 2001; very highly correlated to the S&P 500, which means it will play the same asset allocation role in a portfolio
- WisdomTree HighYield Equity (now called Equity Income) - another with data going back to the 1960s, seems to have lowest downside risk though a longer period of under-performance; achieved higher alpha than RAFI
- Intellidex - amazingly higher return difference over the S&P 500 with positive statistically significant alpha of 0.48% per month (c.5.8% per year) but has much shorter history with data only going back to 1992 (will it continue to work in other market environments?); lowest lag and shortest period of under-performance relative to S&P 500; has exposure to all excess return risk factors - momentum, value and small cap, unlike the other fundamental indices, which only include the value factor
- Equal Weight - since none of the fundamental indices performed any better
Paper 2 describes how to construct the optimal efficient index. The Fundamental indices assessed in Paper 1 certainly fall short of that goal but until some fund company puts one together and offers it on the market, they are appreciably better than the venerable cap-weighted indices like the S&P 500.
One question of importance to international investors and not addressed in these papers, is whether Fundamental indices would work equally well for markets outside the USA. The answer is almost surely yes, since their inherent nature is building in exposure to value stocks and the value premium has been found across other markets.
In the next post, I'll compare the funds, mostly ETFs, that implement the above best indices.