[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries

Stockmarket efficiency revisited (again)

Throwing bricks at the efficient market hypothesis is an activity that has been popular for almost 30 years. Robert Clarkson’s article (The Actuary, August 1999) is merely the latest and no doubt not the last in this continuing saga. Despite these attacks, the efficient market hypothesis continues to survive as a benchmark model and it is useful to ask why.

It is important at the outset to stress that there is no generally agreed definition of market efficiency. A number of definitions have been suggested. The most enduring is that it is not possible to generate excess returns systematically by investing in securities unless you possess information that is unknown to other market participants. A common-sense notion is enshrined in this doctrine: if others know about profitable opportunities then they will act to exploit them. The outcome is that prices reflect what is known and change when new information becomes available. There is no necessity for investors to be correct in their judgements. On some occasions investors will overvalue new information, on other occasions they will undervalue new information, but on average there will be no systematic error.
Two aspects of this discussion are important. The caveat, on average, and the vagueness of the concept of information. The efficient market hypothesis is about what happens on average. It does not claim that prices reflect information accurately. It simply asserts that, if there are profit opportunities, investors will exploit them, and if there are systematic pricing errors, investors will learn and adjust to them. It seems difficult to argue with these simple propositions. The generality of the concept leads to difficulties in testing. The normal procedure is to look for consistency of profits over time.
The nature of information is not specified by the hypothesis and most tests of market efficiency suggest that the market is efficient with respect to a set of information. Information is not constrained to be a collection of facts. A superior method of processing information can also be information. Superior investors may exist who can process information more effectively and profitably than others. The problem for the investigator is to distinguish between the successful investment survivors who by chance have consistently turned in superior performance, and the skilled that have superior abilities. Given several million investors, there are bound to be some lucky survivors.
A typical test of market efficiency examines share returns, given a particular set of information. To illustrate, consider the announcement of a takeover. Can profits be made by buying shares in the target company immediately an announcement is made? The efficient market hypothesis suggests that it will not be possible. As soon as the announcement is made market forces will raise the company’s share price. Many thousands of academic studies have investigated different sets of information. On the whole these studies suggest that it is difficult to make excess returns.

A few anomalies have been found. Various seasonal effects have been discovered (returns on particular days and in particular months are higher/lower than on other days or months) but, despite the widespread nature of these effects, there is little evidence of them offering profitable opportunities for investors. A few mechanical decision rules have been found (for example relating to closed-end funds) that appear to offer excess returns but there are always problems in replicating actual trading conditions spreads, marketability, etc. One of the most famous anomalies relates to over-reaction by investors. This is particularly useful in displaying the scientific process at work.
De Bondt and Thaler revealed that by investing in stocks that had fallen dramatically and selling stocks that had increased dramatically, large excess returns were possible. Subsequent papers have systematically reduced the profitability of the strategy by introducing increased realism into the assumed investment process, although complete realism remains impossible. One of the most intractable problems in studies of this kind relates to the assessment of risk. Modern finance asserts that return is related to risk and generally uses some sort of simple linear model to correct for changes in risk. These models are almost certainly incorrect and the risk adjustment inadequate. A particular problem with over-reaction studies is that companies that fall dramatically in value probably experience large increases in risk. Modelling this increase is difficult and much of the profitability from such decision rules may be largely illusory.

A healthy dose of scepticism
So where does all this leave us? It suggests that challenges to the efficient market hypothesis should be viewed with caution. This is not to assert that the market is always right (the efficient market hypothesis does not in any event claim this) but that profitable anomalies are few and far between. The efficient market hypothesis is a convenient working assumption. Challenges to it should be viewed with a healthy dose of scepticism. Casual empiricism is no substitute for careful empirical work, replicated, corrected, and pored over.