Tipsters have been behind the curve since the credit crunch began in earnest last summer, with earnings estimates from 1,800 Wall Street analysts over 33 per cent too high for the fourth quarter 2007, writes Proinsias O'Mahony.

'IN BULL markets, you don't need 'em; in bear markets, you don't want 'em." That was legendary investor Gerald Loeb's caustic assessment of stock analysts way back in the 1930s. New data from Bloomberg suggests little has changed in the meantime, with investors who followed analyst recommendations in the financial arena losing out badly over the last year.

The research shows that investors who bought financial firms on "buy" recommendations, sold when the analysts downgraded to "hold" and bet on a down move when instructed to "sell" suffered an average decline of 17 per cent over the last 12 months. This was twice the decline seen by the SP 500 during that time. Thirty-two of the 39 analysts surveyed by Bloomberg produced investor losses.

Despite a doubling in the number of "sell" ratings of financial stocks, analysts' reputation as eternal bulls remains intact, with just 10 per cent of ratings recommending offloading beleaguered banks. As for the market as a whole, over 94 per cent of ratings are "buy" or "hold", slightly above the 92 per cent reading recorded a year ago.

Analysts have been consistently behind the curve since the credit crunch began in earnest last summer. Earnings estimates from almost 1,800 equity analysts on Wall Street were more than 8 per cent too high for the third quarter of 2007. That ballooned to 33 per cent for the fourth quarter - analysts had predicted growth of 11 per cent but companies reported a profit decline of 22 per cent.

Analyst over-optimism is not a new phenomenon, however. One study that looked at 12-month price targets slapped on stocks between 1997 and 2002 found that the forecasters were wildly optimistic, with actual returns 35 per cent below estimates and only 24 per cent of stocks hitting their target by the end of the 12-month period.

Older research by renowned fund manager David Dreman confirms that large earnings surprises are the norm, not the exception - Dreman calculated that there's a one-in-170 chance that consensus forecasts will be within 5 per cent of actual results for four consecutive quarters.

Unsurprisingly, there is a large copycat tendency in the analyst community, with a recent academic paper finding a "leader-follower relation in analyst recommendations". It noted analyst "herding" behaviour, with recommendations moving "towards both the consensus and the most recent recommendation".

Investigating whether this was a case of mimicry or "independent reactions to common news", the evidence led the authors to conclude that "analysts do issue similar recommendations because they want to follow leaders".

Ironically, the few stocks that are not plastered with analyst buy ratings may be the ones worth buying. For example, Merrill Lynch's Richard Bernstein found that, out of 40 quantitative trading strategies tracked by the firm in 2006, the best was to buy the SP 500 stocks that were least covered by the analyst community!

More recently, the same firm announced that it was looking to depart from the usual sunny consensus by ensuring that at least 20 per cent of the companies they cover attract a "sell" rating - that's four times the average on Wall Street.

Merrill's move has attracted titters from those who point out the irony in the financial community having to set limits on the number of "buy" recommendations being issued in the midst of a bear market and a possible US recession. Nevertheless, commentators have generally welcomed the move, belated as it was. Merrill's own research shows that about 37 per cent of stocks across the globe decline each year.

The bank is likely positioning itself to secure business from hedge funds who seek ideas on the short as well as the long side. It may be too late, however. Hedge funds increasingly use in-house research or independent research firms to generate trading ideas.

Analysts' reputations took a battering in the wake of the dotcom aftermath, with 10 Wall Street firms shelling out $1.4 billion (€901 million) in compensation in 2003 on the grounds that analysts were using research to tout investment banking clients. Despite the negative publicity, research quality has, if anything, dropped in the intervening period.

In 2006, for example, a report found that two-thirds of the research received by Britain's top 50 fund managers was deemed irrelevant. Four years earlier, more than half of the information received by institutional investors was regarded as valuable.

Furthermore, the bullish bias among analysts is even more evident today, with just half as many stocks receiving "sell" ratings as they did in 2003.

Irish analysts tend to be particularly culpable in this regard, with a recent newspaper survey finding that 92 per cent of recommendations over the last 12 months were either "buy" or "hold" ratings. During that period, the Iseq fell by almost 40 per cent.

Of course, the meaning of the word "hold" has always been slightly ambiguous. A "hold" or "neutral" rating is "to subtly indicate a negative view, avoiding the word 'sell' in order to preclude adverse reaction from company management and major institutional owners of the stock". That's according to Scott McClellan, former Wall Street analyst and author of Full of Bull, a new book that takes a no-holds barred look at the industry he worked in for more than 30 years.

McClellan writes that most research is aimed at professional investors who "understand the nuanced manner in which the game is played", but there is a "massive disconnect" with retail investors, most of whom "take Wall Street literally". As for price targets, they're "terrific if you like fiction", he scoffs.

McClellan adds that analysts used to do "decent, unbiased research" before they became "diluted and distracted". Nowadays, there is an "upside-down emphasis of marketing over research analysis". He estimates that he spent "about 20 per cent" of his time "actually conducting research activities".

Despite this, McClellan says that research can still be useful in that it provides investors with detailed information on companies. He advises investors to take note of the information but ignore the actual conclusions and recommendations.

So is it ever worth looking at the actual buy and sell recommendations? Maybe. A new study has found that one group of analysts can help investors outperform. The paper, entitled Sell Side School Ties, found that analyst stock recommendations outperform when the author went to the same university as a senior officer of the company being followed.

"School ties facilitate information sharing," the authors say, adding that "a simple portfolio strategy of going long the buy recommendations of analysts with school ties and going short the buy recommendations of analysts without ties" earns abnormal returns. The authors also looked at UK data and found the phenomenon to be even greater than in the US.

In other words, the "best" analysts may not be the brightest; as the old saying says, it's who you know, not what you know.