The search for alpha, the current grail of active investment managers, has obscured the fact that alpha can be either positive or negative. Identifying and removing negative alpha can be as profitable for investors as finding positive alpha.

In the current institutional investment climate, that is the strongest argument in favour of fundamental indexing, says Robert Arnott, chairman of Pasadena-based Research Affiliates.

Arnott, the former chairman of quant house First Quadrant, has made fundamental indexing part of the investment vocabulary. While conventional indexes weight stocks according to their market capitalisation, fundamental indexing weights them in relation to four fundamental factors: total cash dividends, free cash flow, total sales and book equity value.

Last year Research Affiliates and the FTSE Group launched the FTSE RAFI Index, a series of 24 fundamental indices covering global markets. One of the main attractions of the series, he says, is that it enables investment managers to identify negative alpha.

“Alpha is a two edged sword,” says Arnott. “It can be positive or negative. Most people think of alpha as if it presumes to be positive, and that’s not always true. In fact, it’s negative more often that it’s positive.”

This has often been overlooked by investment managers, he says. “Most people focus considerable attention on trying to find positive alpha, yet it’s much easier to find sources of negative alpha and remove them.”

An important source of negative alpha is the conventional indexing system, which weights securities in the index according to their market capitalisation. This creates what Arnott terms performance drag. “One major source of negative alpha is the performance drag attached to cap-weighting,” he says. Performance drag is the phenomenon by which the index over-weights over-valued stock and under-weights undervalued stock.

Arnott acknowledges that there have been concerns about cap-weighting, and the performance drag produced by over-reliance on high multiple companies, since the first cap-weighted index, the S&P 500, was created in 1957. The main concern has been that a company that is above its eventual true fair value is going to be over-weighted in a cap-weighted index: “If a company doubles in price, the investor suddenly owns twice as much. What’s the sense in that?” he asks.

The problem has been compounded by the fact that investment theory appears to support cap-weighted indexing. Seven years after the S&P 500 index was launched William Sharpe rolled out his Capital Asset Pricing Model (CAPM). Among other things, CAPM said that investors could not beat the cap-weighted index.

Sharpe’s endorsement of cap-weighted indices slowed thinking about alternatives, Arnott argues. “As an industry we settled on the notion that in the S&P500 we had accidentally stumbled on the right answer, seven years before the theory that proved it. We said then that now we know the right answer, we can cap-weight everything.

“The problem is that it was not the right answer, in the sense that it’s only the right answer if all the underlying assumptions are true. And we know that the underlying assumptions categorically are false, and that we are simplify assumptions that help make the mathematics work.”

What has been lacking, he says, is an intellectually satisfying alternative to cap-weighting. “The issue is a simple one, in that every company has a price which is the market’s best guess at it’s true fair value. But it’s only a guess, and the true fair value something we cannot know, cannot see. The share price may be far above or far below that eventual true fair value.”

Indexers have responded to this criticism by saying that unless the critics can identify the over-valued or under-valued stocks in an index, their criticism is worthless. Yet Arnott says this is too glib

“If you take the weight in the portfolio and tie it to price and tie it to market cap you will over -weight the overvalued and under-weight the under valued, without exception.

“But if you sever that link, if you de-couple the weight from price, market cap or valuation multiples, so that that the weight in the portfolio is valuation-indifferent, you will eliminate the tendency to over-weight the over-valued and under-weight the under-valued.” This de-coupling is worth between 2% and 3% a year.

How asset managers de-couple weight from price is immaterial, he says. They can use a dartboard to weight stocks if they wish, although most pension fund boards would blench at the idea. More seriously, they can equal weight the stocks. Here, the evidence against cap-weighting has been compelling. In the five years between 2001 and 2005 the S&P 500 returned 1%, according to figures from Research Affiliates. In the same period, the S&P 500 equal weighted returned 51%. “That simple difference was the genesis of my work in fundamental weighting,” says Arnott.

Yet there are drawbacks to equal-weighting, he warns. “With equal-weighting, you have the complicating factor of what roster of comparatives are you equal weighting. If they have been first selected based on market cap then there will be a slightly disproportionate number of over-valued companies in the list. That is why fundamental indices empirically outperform equal weighted indices where you would expect them not to.”

“There are a lot of myths about fundamental indexing,” he says. “One of them is that the turnover is high in the index. Cap-weighting has 6% turnover against 10% for fundamental indexing. But relative to pretty much any strategy out there, whether it’s enhanced index or active management , it’s extremely low.”

Critics of fundamental indexing have argued that it is simply closet value investing, with a tilt towards value rather than growth stocks. Arnott agrees that the critics have at least half a point. “The argument is valid and not valid at the same time. Cap-weighting puts twice as much money in a company if it’s at twice the market multiple, which means that cap-weighting, structurally and inevitably, will put most of your money in growth stocks.

“There’s nothing inherently wrong with that. Growth won’t beat value, so it won’t help you either. But it does mean that, relative to a cap-weighted index, the fundamental index will look like a value-tilted index.”

Critics also argue that there is a small cap tilt which could account for the outperformance of fundamental indexes. Again Arnott agrees that the tilt is there. “You are going to have a bias towards the segments of the market that are trading at below market multiples. Often that is small cap, because small cap historically tends to trade below the market multiple. But that turns out to be a very mild bias and one that is episodic.”

One of the main attractions of fundamental indexing for institutional investors is that it is a mainstream rather than alternative strategy, says Arnott. “Fundamental indexes put most of your money into large companies which are inherently mostly large cap, mostly highly liquid and mostly highly tradable.”

This resolves the problem of capacity, he says. “A trillion dollar fundamental index in the US would have capacity issues on only about 20 to 30 of the largest names. So the idea is scalable on a vast scale.”

How important are market conditions for fundamental indexing? Are there some conditions when it works better than others? In particular, does fundamental indexing provide higher returns than cap-weighted indexing when there is more ‘noise’ than information in the market?

Keith Ambachtsheer, the international pensions consultant and head of KPA Advisory Services in Canada, has suggested that the respective merits of cap-weighted versus cap indifferent indexing strategies depend greatly on the noise/information ratio of stock price changes.

Ambachtsheer argues that “if the ratio is significantly biased towards noise, logic tells us that additional returns can be generated by re-balancing protocols that are cap-weight indifferent (such as fundamental indexes).Conversely if the ratio is significantly biased towards information returns will be degraded by unproductive rebalancing protocols that are cap weight indifferent.”

He points out that the evidence of Arnott and his colleagues suggests that for the past 43 years in the US, the bias has favoured noise over information in stock market price changes. Similar results have been found in other markets.

Arnott agrees that Ambachtsheer has put his finger on a significant factor in the cap-weighted/fundamental index debate. “On one level, it’s a very important insight if price under-reflects the available information. That is to say, if good news is assigned too small a decrement, then the cap-weighted indexes will derive just as much incremental benefit from growth stocks outperforming and value stocks underperforming as it will lose from the structural pattern of overvalued stocks underperforming.

“It’s a very subtle point, but it’s an important one if price differs from true fair value in a fashion that’s different from company size. Then you will have a size effect because the companies with positive error are more likely to percolate into the largest market cap spectrum. So his observation is a good one.

“I think that the reality of the market doesn’t do that. The market, if anything over-reacts to information. So I think the market tendency to over-react to information augments the alpha of fundamental indexing.”

Fundamental indexing is a concept that has empirical but not theoretical backing. It works, but nobody know mathematically why it works. Arnott is now working on the maths to back the empirical evidence that fundamental indexing performs better than cap-weighted indexing.

“We’ve had finance theory for 50 year predicated on the notion that price, structurally and inevitably, identically equals true fair value, that if the price suddenly jumps 10% the present value of future cash flows has jumped 10%,” he points out. “Intuitively we know that’s not true. Intuitively we know that the price is an estimate and probably a pretty bad estimate of the true fair value of the company.

“And if the weight that one allocates top portfolio is directly linked to market cap price-to-valuation multiples, it means that the error in that price is going to be directly correlated with, and a key driver in, the weight.”

“The work that we’re doing at the moment recognises that price and true fair value may be different - something that anyone in the investment management community knows their heart to be true.”

A mathematical theory of why fundamental indexing works may not be far off, Arnott suggests. “What we are finding is that if price and fair value are different. the price constantly seeks true fair value. We find that the noise creates the size effect, the noise creates value effect. The mean reversion creates the momentum effect and the noise creates the value-added associated with value indifferent indices such as fundamental indexing .

“So they are all tied to one factor. That, I think, is very interesting - that each of these factors turns out to be an artefact of one factor - noise.”