For many investors, index investing is where they want to be. It is inexpensive, allows broad participation in the market, and can be finely diced into sector and other sub-index pieces. Index investing has traditionally been linked to capital-weighted indexes, whereby a company's market cap determines how much of a company actually got into an index. And that's the fatal flaw in conventional index-linked investing, according to Robert Arnott, chairman of Pasadena-based Research Affiliates and editor of the Financial Analysts Journal.
Arnott, the former chairman of First Quadrant, and global equity strategist at Salomon Brothers, has developed the radically different concept of fundamental indexing, which uses a company's fundamentals: sales, profits, book value, and dividends to determine its weighting in an index. Arnott says the problem is that cap-weighted index investing "overweights a portfolio with overvalued stocks and underweights it with undervalued stocks."
Things are moving fast for Research Affiliates: last November, FTSE started publishing 37 Research Affiliates Fundamental Indexes (FTSE/RAFI Indexes). In the U.S., PIMCO has developed the first fundamental index-based mutual funds--Fundamental IndexPLUS and Fundamental IndexPLUS TR. In addition, PowerShares launched broad market FTSE/RAFI ETFs last fall, and now advisors can choose from nine new FTSE/RAFI sector ETFs. Other countries are adding foreign RAFI-linked ETFs as well. Arnott estimates that there is about $4 billion in assets invested in RAFI strategies worldwide, up from $100 million in December 2004. While that may be a drop in the bucket compared to the estimated $2 trillion worldwide linked to cap-weighted indexes, fundamental indexing has clearly been noticed.
How did fundamental indexing start?
S&P introduced the first cap-weighted index in 1957. It is astonishing that in 50 years no one had tested the structuring and weighting of an index based on other measures of company size.
People have looked at equal-weighting and they have puzzled over why it's better and they've come up with answers about a small-cap tilt, a value tilt, and so forth but the real reason equal-weighting wins is because it's not drawn into a weighting structure that guarantees that you're overweight the overvalued and underweight the undervalued; that directly links the weight in the portfolio to the error in the price. If your weight is proportional to price and that price itself ever departs from the true fair value--and of course we know it does--then price will be over or under true fair value, and the weight of every company overvalued is going to be too high, so you wind up chasing every bubble that comes along, you wind up doubling your weight in an investment just because it's doubled in price--well that makes no sense. We decided to test other ways of indexing and as we did those tests, [weighting the fundamentals] the relative performance was shocking: 2% to 3% value added, per year, compounded over long periods of time.
This stimulated some research into the theoretical backdrop--why does it work so well? It works so well because the link between portfolio over- and undervaluation and the weight in the portfolio has been severed; you're not overweighting the overvalued and underweighting the undervalued. You don't know which companies are over- and undervalued but you do know weighting by market cap assuredly overweights the overvalued and that weighting by simple measures of a company's size isn't going to expose you to that, because price, P/E ratios, valuation multiples, market cap are not part of the decision as to what weight to put into a stock.
Explain equal-weighting versus fundamental weighting.
Equal weighting is you take the S&P 500 and put 1/5th of a percentage in each stock. That adds a lot of value. It also is relatively higher volatility, higher turnover and it's expensive to run, but it does work. Fundamental indexing works better because with equal-weighting the S&P you're still left with a universe that is biased--you're still left with companies that are popular, trendy, comfortable, and therefore probably aren't priced attractively.
With fundamental weighting what are you looking for?
For the U.S. FTSE/RAFI 1000 we choose the 1,000 largest companies--not by market cap--but by measures of the financial size of a company. What measures would people on Main Street think of as sensible measures of a company? Not market cap. People on Main Street would think: What are the company's sales? What are its profits? What's its book value? How much could it afford to pay out in dividends every year?
We look at a company's sales, profits, book value, and dividends. If a company is 4% of the economy by sales, 3% by book value and profits, and 2% by the dividends that it pays, then, we could argue is it 2% or 3% or 4% of the economy or we could just take the average, and say "It's about 3% of the economy and therefore it will be 3% of our index." We simply equal-weight the four measures; if a company has never paid a dividend we just weight the other three--we don't want to saddle a company with zero weight for dividends. It's a simple process and that's what makes it doubly amazing that it hadn't been explored decades ago.
The genesis for this was in early '02; the market was down 25% on its way to being down 47%, the average stock was up 20%! Most companies went up in 2000, 2001, most companies went up in the first quarter of 2002, so for two years after the bubble burst, for most companies it was a bull market. The bull market of the '90s didn't end until April of 2002--for most companies. How many people out there think of 2000 and 2001 as bull market years? They don't--because the indexes were all down, and for far too many people their portfolios were all down. Why? Because they were loaded up in these ultra-high-flying, ultra-high-multiple growth stocks that, in retrospect, were revalued.
So, what you're doing is kind of what a value manager would look at for an actively managed portfolio, but you're including a much larger universe in the portfolio?
We're including a larger universe, and staying away from subjective judgments about what a company is worth. We're also staying away from the value manager's temptation to say, "Google is too expensive and I won't own it." We look at it and we say, "Perfectly good company, here are its sales, its profits, here's its book value; doesn't pay dividends so we'll ignore that. Its footprint in the economy is, I don't know, .03%, so let's weight it .03%." We don't make the secondary judgment of, it's going to be worth .5% of the economy in five or 10 years and let's go ahead and pay for that now as if that's already happened--that's what cap-weighting does.
Then what happened?
We tested in all 10 of the S&P sectors going back 17 years, found it adds value in 10 out of 10, no exceptions. We've tested in some niche markets--small companies--our [RAFI] 2000 against the Russell 2000; [the RAFI] adds 3.5% per year, going back 27 years. In small company space it's not because of a small company tilt or value tilt. Why?
Think about the companies in the Russell 1000 and not the RAFI 1000: they'll be companies that are large cap, but small companies. Small companies that are large cap are trading at lofty enough multiples to take them into large-cap space. Those drop into our [RAFI] 2000 because they are small companies, so we have some ultra-high-flyer, ultra-high-multiple, growth stocks that are big cap companies that drop into our small company universe. That means we've got some big-cap names and some very-high-multiple names.
These stocks neutralize what would normally be a value tilt in the index. And they actually flip the small-cap bias to large-cap, because there are enough large-cap names that drop into this list. What it means is our small company list is larger cap and is not value-tilted, therefore according to standard metrics, which say small beats large and value beats growth, we should slightly underperform the Russell Index. We don't, we outperform by 3.5% a year, because the [Russell] Small-Cap Index is cap weighted; it overweights the overvalued and underweights the undervalued, and we don't. So that's the best evidence that I've got that it's not just a small-cap and value tilt that's creating the alpha.