However, I'm not sure he's entirely correct – though I would say that, wouldn't I? This is because, in its response to news, the market is sometimes nowhere close to being efficient.
Evidence for this lies in the phenomenon of post-earnings announcement drift. Shares that enjoy good news about earnings tend on average to drift upwards in the days after the news, whilst shares suffering bad news continue to fall. This is one reason why momentum investing has been so successful. Norman Strong, a professor at Manchester Business School, has studied the performance of UK stocks and concluded: "Returns continue to drift upwards for stocks with extreme positive earnings surprises and they continue to drift downwards for stocks with extreme negative earnings surprises."
One reason why this happens is that, contrary to the more simple-minded versions of efficient market theory, investors are not wholly rational. Instead, they under-react. If they see what they believe to be a bad company posting a good result, they cling too much to their Bayesian prior and attach too little weight to the news. This under-reaction holds the share price down immediately after the earnings announcement, causing it to drift upwards later as those bearish priors get revised away.
There's another way in which news can lead to shares rising days later – through attention bubbles. Zhi Da and Pengjie Gao of the University of Notre Dame and Joseph Engelberg at the University of North Carolina have found that, in the US, stocks which are heavily searched for in Google tend to rise in the subsequent two weeks. To the extent that a high search volume is correlated with news about a stock, news thus predicts returns.
There is, however, a big difference between these two mechanisms. With post-earnings announcement drift, the rise in share prices after news is long-lasting. But stocks that rise because of an attention bubble tend to fall back after their initial rise.
So, what determines which of the two mechanisms operates?
One thing is herding. If a stock is being widely discussed and bought, other investors might feel emboldened to buy it themselves. Such a bandwagon effect can cause a share to rise too fast and so subsequently fall.
The other factor is what some economists at the University of Georgia call "moderated confidence." The idea here is that investors underreact to reliable signals but overreact to less reliable ones. Imagine a small or volatile stock gets some genuinely good news - a reliable signal - which is not widely reported or commented upon. Investors might then underreact, believing that the signal is weak relative to the normal noise surrounding the share. This will cause the share to rise too little in response to the good news, with the result that it drifts up later as investors cotton onto the share's better prospects. However, if a big-name stock gets good news of similar magnitude investors might put lots of faith into this signal, believing it to be strong relative to the lower amount of noise accompanying the share. This can lead them to overreact, pushing the share price up to a level from which it subsequently falls. This is especially likely to happen if the news is accompanied by lots of press comment and attention from other investors.
These two factors have the same implication. We are more likely to see post-earnings announcement drift than attention bubbles in circumstances where earnings news is not much noticed – such as for smaller stocks which are out of favour with investors.
For these reasons, news can be useful for investors because it is not immediately embedded into prices as efficient market theory predicts.
But there's a problem here. We know, thanks to the work of Brad Barber and Terrance Odean, that very many retail investors under-perform the market, as indeed, do the professionals; most unit trusts in Trustnet's database of UK all companies' funds have under-performed Legal & General's UK index fund in the last five years. This suggests that, if active investors do trade on news, most of them do so badly – or at least not well enough to offset other sources of under-performance.
This might not be entirely due to personal shortcomings. It could be that it is much harder to distinguish between attention bubbles and post-earnings announcement drift in practice than it is in theory. And the failure to do so might be expensive. If you bet on post-earnings announcement drift when in fact there’s an attention bubble, you’ll lost money as the bubble deflates. And if you bet on an attention bubble when in fact there’s drift, you’ll sell too soon and miss out on profits.
The message here is one that perhaps generalizes. Although researchers have uncovered many deviations from the efficient market hypothesis which suggest that it might be possible to trade successfully on apparently old news, doing so in practice is fiendishly hard – and beyond many investors.
In this sense, whilst Mr Smith is wrong, he isn't grievously so.
