Some analysts are better than others. That much we know.
But how to distinguish between — and capitalise on — those who ‘get lucky’ with their last calls and those who have a genuine tendency to make ‘accurate’ recommendations?
A new academic paper, by Dieter Hess, Daniel Kreutzmann and Oliver Pucker, attempts to do just that.
The abstract:
We document that investors can actually profit from the contemporaneous link between earnings accuracy and recommendation profitability (Loh and Mian (2006)). Differentiating between “able” and “lucky” analysts we suggest an implementable, i.e. look-ahead bias free, trading strategy that yields annual excess returns of 11.5% before transactions costs during our 1994 – 2007 sample period. Rather than past track records analysts’ characteristics indicate their ability to identify undervalued stocks. We find that a reputation effect, i.e. higher recommendation announcement returns, is insignificant. This indicates that the ability is real. Able analysts can distinguish between firms that will over- or underperform.
The authors use IBES and CRSP data, plus eight analyst ‘characteristics,’ to build a model to judge and predict their earnings forecast accuracy. The characteristics being things like past earnings accuracy, the number of firms and industries the analyst covers, forecast frequency, and so on.
And (surprise, surprise) following the recommendations of analysts deemed more accurate, seems to be a more profitable strategy than following analysts judged not so competent.
By 3.89 per cent, to be exact.
For instance, investing in a ‘strong buy’ call of an ‘able’ analyst would earn you an average return of 11.44 per cent in the period, net of transaction fees. But investing in the ‘strong buy’ of a not-so-able analyst would generate 7.55 per cent.
The conclusion:
We show that investors can hence improve their trading strategy based on stock recommendations by focusing on analyst characteristics that can be used to predict earnings accuracy. The stock picking ability of the analysts that we identify as able is real, since the higher return of the recommendations is not due to a reputation effect. Our results imply that at least some analysts have the ability to discriminate between over- and undervalued stocks due to their superior earnings forecasts.
Quick, to the models — to test your analyst’s accuracy.
Related links:
Analysts know how (if not when) to hold’em – FT Alphaville
UBS analysts are mostly green, visualised - FT Alphaville
[Outlook 2010] How will analysts fare? – FT Alphaville