The hypothesis stated

In financial economics, the efficient-market hypothesis states that asset prices fully reflect all available information. A direct implication is that it is impossible to "beat the market" consistently on a risk-adjusted basis, since market prices should only react to new information or changes in discount rates.

The hypothesis was developed by Professor Eugene Fama, one of the true “greats” of financial markets analysis. Fama argues for the purest form of the hypothesis: namely, that shares trade at their fair value, making it impossible for investors to either purchase undervalued shares or sell shares for inflated prices.

There is some variation of the hypothesis, however. There are, for example, some who argue that the theory works on a macro level and also that there is no discernible bias in the prices of shares. A share trading at a high price earnings ratio, for example, is as likely to be under-priced as one trading at a low PE ratio. This thought means that shares can be priced at more or less than their fundamental values, but that such variations are random and cannot be exploited by any trading styles.

Empirical evidence

There has been an abundance of academic articles testing the hypothesis. Some of these have focussed on some of the oddities in the markets, while others have looked at different types of transaction.

Cooper, Dimitrov and Rau of Purdue University published a paper in 2000 entitled, rather elegantly, and with apologies to the Bard, “A Rose.com by any other name”. (The preceding phrase in Romeo and Juliet is, “What’s in a name?” and this is the theme of the paper). They looked at 95 different companies with internet connections which changed their name in the frothy period from June 1998 to July 1999. The name changes in each case were to include the phrases “dot.com”, “dot.net” or “internet”. The researchers found that there were very significant and sustained increases in share prices following such name changes.

However, the authors were concerned to point out that only one of these companies was on Nasdaq, with the other 94 being on the Nasdaq Bulletin Board. These were therefore very modestly-sized entities with low market capitalisations, thin trading in their shares and relatively little research coverage.

In addition to this, some of the name changes at least had some commercial benefits. For example, the decision by Computer Literacy Inc. to change its name to Fatbrain.com meant that there were fewer mis-spellings in the searches for the company’s website.

The Chipotle paradox

A rather more telling example is known as the "Chipotle paradox": Chipotle Mexican Grills Inc was a spin off from McDonald's in September 2006. The company operated a chain of Mexican-style restaurants. There were two classes of shares in issue:

A shares which had one vote per share;

• B shares which had 10 votes per share.

The normal expectation would be that the B shares would trade at a modest premium above the A shares. However, in the period from October 2006 to February 2008, the A shares actually traded at a premium over the B shares. The premium was initially 9% and this then gradually widened to 21%. The two share classes then converged in price.

This is an example of a sizeable, well researched company, with institutional investors. It appears that the market was beguiled for a period of 16 months into the belief that A shares were to be preferred to B shares.

The authors hypothesised as to why this market oddity was not removed by those who would see the opportunity to buy the B shares and short the A shares. It appears that the main reason was friction in the market in respect of the cost and risk associated with shorting the stocks.

Other examples of unusual market pricing include the closed end fund puzzle: why do closed end funds generally trade at a discount from underlying net asset values? A unit trust with identical investments would realise net asset value.

Active or passive investment management?

However, the greatest, all-pervasive empirical test of the theory comes in the performance of active investment funds. If the theory holds, in its strong or weak form, it should not be possible for active investment managers consistently to beat the market. Fama and French did a study in 2012 and found that the distribution of abnormal returns of US funds was very similar to the distribution which would be expected if no fund managers had any skill. They found that from 1984 to 2006 the average active fund underperformed the market to the extent of 0.8% a year.

This study recognised that there are outliers in any normal distribution. As an example, if 1,000 people each toss an unbiased coin ten times there is a 97.7% probability that one person will produce ten heads.

There is a documented case of one fund which had a very long record of success, the Legg Mason Value Trust. This fund outperformed the market every year from 1991 to 2005. Its investment style was to invest in value stocks, namely those companies with strong asset backing to support the market capitalisation, relatively low pricing multiples and relatively high rates of dividend. However, virtually all of that relative over-performance was then lost in the meltdown of 2008.

It does appear from the literature that there can be trading styles which can beat the market in the relatively short term. However, as soon as this is identified it ceases to provide an opportunity. This is what might be expected from reasonable efficient markets.

There have been many studies undertaken of the relative performance of active investment funds with the market. The studies generally make grim reading: a minority of funds may overperform the market in one year, but this minority shrinks to very small percentages once they are measured over five years.

Rusty pin?

There is a surprising consistency in the findings of the various studies: only some 20% of the actively managed funds beat the market in any year. On a random basis some 50% might be expected to outperform the market for one year. According to the research, it is the management fees charged by such funds which means that, on average, 80% underperform the market.

Andrew Strickland, Consultant, Scrutton Bland
Valuation Group, December 2016