Are brokers' buy and sell recommendations really worth heeding?
A few weeks ago I wrote an article on the performance of brokers' recommendations, and it seemed to generate some interest -- see Broker Buys Beat The Market.
And while it's fun (if you're that way inclined) to see what we can deduce from readily available information, it was clearly not a scientific study.
For a start, it was a snapshot in time, based on broker recommendations on a random day, and looking at the performance an arbitrary six months later. And there was no attempt to consider how new or old the recommendations were, or how they performed over other time periods.
With more resources, one could do a more complete study, using hundreds of thousands of recommendations instead of a few dozen, gathered over many years, and looking at various time periods relative to the dates when the recommendations were published.
That's the sort of work that academics have been doing for decades, as far back as 1933 in the case of brokers' recs, and that's where we should look if we really want to understand what's happening.
Are recommendations valuable?
Most of the research has focused on the US market, where a long list of studies have been conducted.
As regards the UK market, the main contributors to our knowledge have been Paul Ryan and Richard Taffler; and Elroy Dimson and Paul Marsh. The following short miscellany is compiled from more than 20 different papers on the subject, local and international.
The first thing to note is that most studies show that brokers' buy recommendations generally tend to out-perform their sell recommendations, so they do in fact have some value. There is evidence of this not just in the days following publication, but also in the following months.
Interestingly, there is also evidence of this in the days leading up to publication. Combined with increased trading volumes over these periods, it seems to imply that the information is already known privately in advance of publication.
Look for changes
Changes in brokers' opinions have the most effect, especially when they break away from the herd consensus, with new 'sell' recommendations being particularly significant. This is partly because sell recommendations are much rarer; there are seven times as many 'buys' as 'sells' in the US, although British brokers are more likely than their American colleagues to rate a share negatively.
Remember that an analyst saying 'sell' is usually sticking his neck out, and will tend to do so only if he is very sure. He also generates less business by shouting 'sell'. There is evidence that recommendations in Britain are less susceptible to conflicts of interest than those in the US and in Ireland.
It also seems to take some time for the market to assimilate new recommendations, hence the difference in performance lasting over several months. This seems to be especially true if the call is negative.
It is harder to establish the extent to which subsequent share performance is the result of brokers' recommendations; in other words, to what extent are brokers actually causing changes in share prices, rather than predicting them?
Can I make money from this?
Sadly, this does not mean that one can run a profitable trading strategy by simply acting quickly on new recommendations. The required churning of the portfolio -- approximately 400% according to one study, meaning an average holding period of three months -- would more than eliminate any profit.
It's also worth noting that in 2000 and 2001, analysts' sell recommendations significantly beat their buy recommendations, for both tech and non-tech stocks, so the general finding that analysts are more right than wrong does not always apply.
More from Padraig O'Hannelly:
