Strong herd instinct among analysts

By WONG WEI KONG


WITH the reporting season now underway in earnest, consensus estimates will increasingly come into focus. Companies will be measured against the consensus forecasts made by analysts, and their shares will be bought or sold based on the extent to which they have exceeded or fallen short of those expectations.

Such market behaviour, while firmly established, hinges on a big assumption. It presumes that each analyst independently arrives at a forecast based on individual conviction, and considers the average of these forecasts as the collective representation of a group of independent opinions.

That's the theory; in practice, analysts are often accused of 'herding', or going with what seems to be the majority view. If so, the value of the consensus estimate will be diluted, because this will not then be a representation of independent opinions but rather an outcome shaped by other factors, such as analysts who issue more frequent reports, or reports issued by the bigger and more influential investment banks or broking houses.

Corroborating studies

But, first, is it fair to say that analysts tend to herd? There is, in fact, an established body of work on this subject.

One such recent study is Analysts' Herding Propensity: Theory and Evidence from Earnings Forecasts, prepared by Murugappa Krishnan (Yeshiva University), Steve C Lim (Texas Christian University) and Ping Zhou (Lehman Brothers).

While prior studies focus on herding at the forecast level, the trio also focused on herding at the analyst level (that is, if an analyst is likely to herd). The study used 14 years of IBES annual earnings forecast data and put it through an empirical model which looked at actual earnings announcements, analysts' earnings forecasts, analysts' true posterior beliefs (based on all available private and public information available), analysts' forecast bias (choice variable of the analysts), and prior consensus (function of preceding analysts' forecasts).

'We find that herd behaviour is pervasive at both the aggregate and the analyst levels. At the aggregate level, yearly estimates of herding propensity suggest that the result is not due to any particular year . . . At the analyst level, we find that roughly 75 per cent of the analysts in our sample tend to herd,' the study concluded.

It went on to say: 'Our results suggest that investors and researchers should exercise caution when using analysts' earnings forecasts as proxies for the market's expectations. This observed stability in the degree of herding propensity means that investors and market participants could also benefit potentially by adjusting for herding.'

Other studies also arrived at similar conclusions. Herding Among Security Analysts, by Ivo Welch of the School of Management at Yale University, found that 'the buy or sell recommendations of security analysts have a significant positive influence on the recommendations of the next two analysts'.

In Do Analysts Herd? An Analysis of Recommendations and Market Reactions, by Narasimhan Jegadeesh (Goizueta Business School, Emory University, Atlanta) and Woojin Kim (School of Public Policy and Management, Seoul), the authors concluded: 'Our empirical results support the herding hypothesis . . . We find that analysts from larger brokerages and analysts following stocks with smaller dispersion across recommendations are more likely to herd.'

Over-optimistic

In Experts' Earning Forecasts: Bias, Herding and Gossamer Information, Olivier Guedj and Jean-Philippe Bouchaud from Science & Finance, Capital Fund Management, Paris, studied the statistics of earning forecasts of US, EU, UK and Japanese stocks during the period 1987-2004. 'We confirm, on this large data set, that financial analysts are on average over-optimistic and show a pronounced herding behaviour,' they said.

One caveat for Singapore investors is that these studies are based mostly on overseas market data. However, the herding instinct, like investor and market behaviour in general, is likely to cut across cultures. It may even be possible to argue that in markets with a relatively small analyst community, the propensity to herd is even stronger.

So it comes down to this: the consensus estimate, while still useful as a general indicator of the forecasts analysts are publishing, should not be viewed as the definitive benchmark of earnings performance. And it may be worthwhile for investors to pay more attention to the outliers, and the track record of individual analysts.

Let us check every quarter whether is herd correct or not? Collective wisdom of analysts