If they are right, it's bad news for stockpickers, because it means it is far harder to beat the market without taking on extra risk than you might think. But are they right?
In one sense, they have a point. If you back enough horses you will have some winners including some at long odds. If you shout about your winners and keep quiet about your losers you might therefore earn a reputation as a tipster, even if you actually know nothing. A similar thing might be happening in financial economics. There are hundreds of things that might predict share prices: indicators of risk, accounting and balance sheet measures, investors' errors of judgment or even the companies' names themselves. Simple probabilities tell us that if you look for enough patterns, you will find some, and some will even appear to be statistically significant. But it doesn't follow that investors should bet that these patterns will persist, any more than that they should follow our tipster's recommendations.
Academic economics is not unusual here. In fact, Professor Harvey was consciously echoing a claim made about medicine by John Ioannidis at Stanford School Medicine: "most published research findings are false".
All this poses a question for investors: are there any systematic mispricings of equities at all? I suspect there are. The evidence for them, though, must be stronger than merely one piece of research. To paraphrase Thomas Aquinas, beware the man of one study.
To continue the horsey metaphor, any alleged mispricing must jump over three hurdles.
First, it must exist in several data sets. Momentum has cleared this hurdle. It exists not just in UK and US equities but in stock market indices, currencies and commodities. And it also seems to have existed as far back as the 19th century.
Defensive investing also clears the hurdle. Low-beta shares outperformed from the 1930s to 1960s in the US, and around the world since the 1990s. And again, this isn't confined to equities: low-beta assets do well in bond and commodity markets, too.
Some alleged anomalies, however, fall at this hurdle. Perhaps the most significant is the so-called size effect - the tendency for small stocks to outperform big ones. This appeared to be well-attested in the 1980s. But small stocks then underperformed horribly for the next 10 years and have done about as well as the FTSE 100 since then.
The second hurdle is: is there a strong theoretical reason why the anomaly should exist? Why do people leave money on the table?
Here, I'm sceptical of accounting-based anomalies simply because there are hundreds of smart fund managers and analysts who are looking for them. They should therefore have spotted them and so traded them away.
What I find more plausible are anomalies based on cognitive biases. For example, momentum assets can do well because investors underreact to news - they cleave too much to their prior beliefs - with the result that prices don't respond fully and immediately to such news but instead drift up or down in the following weeks. Similarly, I find it plausible that investors are overconfident about their ability to spot future growth, with the result that they pay too much for 'growth' stocks, leaving dull defensives underpriced.
My preference here, though, isn't fixed. If most investors were to pay more attention to cognitive biases and less to company accounts, then anomalies based on the latter would become more plentiful while those based on the former might disappear.
The third hurdle is: why don't well-informed investors exploit the anomaly and in doing so bid it away?
One good answer is that they are prevented from doing so by limits on their ability to short-sell shares. These limits might be institutional or merely practical: it's difficult and dangerous to short-sell high volatility stocks, for example. However, while this explains why some shares stay overpriced, it doesn't explain why others stay underpriced.
But something else does - benchmark risk, the danger that a fund will underperform its peers costing the manager his job. Defensive stocks carry this risk. They might well underperform if the market rises strongly, which makes them unattractive to fund managers. Victoria Dobrynskaya at the LSE has shown that the same might be true for defensive stocks.
What we have here are three tough hurdles. I suspect that although momentum and defensives clear them, very few other alleged anomalies do. For example, value stocks' good performance might be merely a reward for taking on the danger that many such stocks would do badly in a recession rather than a genuine mispricing.
This is not to rule out other possibilities, but merely to say that many of these horses haven't yet completed the course.
For now, though, this means that there are very few well-established deviations from the efficient market hypothesis.
