Analyzing the Analysts

How useful are analysts' recommendations? The argument has been raging for years. There have been numerous academic studies and even actual matches between “random” and real-life stock pickers — remember the Wall Street Journal's dartboard contest, in which the paper would put its randomly selected stocks against the picks of professional money managers and newsletter writers? During the 12-year life of the contest, which ended in September 2002, analysts outperformed the darts by 9.6 percent to 2.9 percent. And the coup de grace? The pros also beat the Dow Jones Industrial Average, which averaged a 5.1 percent annual gain during the same period. Score one for the proponents of active money management.

In the real world, on real time, following the advice of an analyst, however, isn't as easy as it seems. Brokers must try to divine which analysts were truly helpful in understanding a company or industry and which are shameless underwriting hounds. Figuring out who adds value and who does not is of obvious significance to brokers who act as portfolio managers for their clients. One solution is to hit up independent research services such as First Call and Reuters' Multex. The trouble is that they bring so much research to the desktop — literally thousands of analysts' notes published by hundreds of research houses on scores of companies — it can be overwhelming.

A rapidly expanding upstart, StarMine, now offers filters to investors to help sort through the flood of research. The San Francisco-based company, with offices in New York and London, began tracking analyst recommendations in 1999. Today, it consolidates data from 400 brokerages worldwide and provides subscribers with notes on more than 3,900 domestic and over 6,000 foreign companies based in more than 50 countries.

StarMine's subscriber list now totals 30,000 and comprises 150 institutional investment firms and 26 of the 100 largest U.S. asset managers, including Merrill Lynch, UBS, Lehman Brothers, Deutsche Bank and US Bancorp Piper Jaffray. Nearly half of the brokerages involved in the New York State/SEC settlement have enlisted the company's services to help meet the new standards set forth for analyst compensation.

“Because these brokerages are now required to base analyst compensation on objective measures of performance, not the number of investment banking deals secured,” says StarMine's vice president of marketing, David Lichtblau, “they have adopted our rating system due to its objectivity and transparency.”

Robert Peterson, director of equity research at US Bancorp Piper Jaffrey, one of the brokerages who settled, says, “StarMine is the only independent solution available and is far more powerful and cost-effective than building a system internally.”

Buy-side firms use it differently, of course. “It's not that StarMine selects stocks for me,” explains subscriber Robert Bartels, portfolio manager at Chicago-based Northern Trust Global Investment. “But by sifting through a wide swath of the research universe, these folks can tell me which analysts have been more consistently on target for any given company. And that helps me know to which analysts I should most listen.”

Illustrating how this has helped him, Bartels points back to last summer, when he became concerned about his position in Alcoa. “There was still fairly decent analyst support for the company,” he recalls, “but there was one analyst who had been consistently dead-on about Alcoa's performance. And when he voiced material concerns about the company losing pricing power in August, I got out. Within a month, the stock had lost nearly one-third of its value as the rest of the world subsequently realized the same thing.”

StarMine relies on several simple methods for grading analyst performance. The first is an absolute measure, which is helpful in comparing analysts who are tracking a similar group of stocks. It does this by using an analyst's recommendations to generate a theoretical portfolio. A “buy” takes a $100 position in the stock. A “strong buy” doubles the bet. A “hold” sets aside $100 in cash. A “sell” would short $100 worth of the stock. And a “strong sell” would short $200. This long/short portfolio gives an indication of how well an analyst has called the direction of his stocks.

In contrast, the Wall Street Journal conducts an annual performance review of analysts across 49 industries. It is less refined, however, than StarMine's approach, relying on three tiers of rank and restricting itself to U.S.- and Canadian-based analysts. Because it's published only once a year, the Journal's presentation is static; StarMine's, on the other hand, constantly revised, interactive system.

StarMine also relies on a second, more relative measure that helps compare analysts from different industries. It first constructs a long-only portfolio that puts more weight on the stocks with the analyst's highest recommendations and under-weights the stocks with lower recommendations. Then it compares the return of this portfolio to a custom benchmark for the analyst — a portfolio that equally invests long in all of the analyst's stocks. The difference between the two gives a measure of the analyst's “excess return.” This number is then divided by the average monthly standard deviation of returns of all the analyst's stocks.

“This calculation,” explains StarMine's Lichtblau, “generates figures that permit a fair comparison between analysts covering high beta industries and an analyst tracking a less volatile group of stocks.”

StarMine also refines the Street's consensus earnings estimates to anticipate earning surprises — deviations from consensus estimates — that may move a stock. Working with data provided from Thomson Financial, StarMine develops its own so-called “SmartEstimates” by overweighting the most recent estimates and those of analysts who have strong track records for being right. The company claims that this process accurately identifies earning surprises 75 percent of the time.

Full access to the site runs $3,000 per year per user when at least five terminals are hooked up. A single user would otherwise be hit with a $15,000 annual fee. For those who think the cost is a little pricey, portfolio manager Robert Bartels would remind advisors that, with an average investment position typically representing 2.5 percent of a portfolio's assets, one good idea multiplied across dozens of clients could easily pay for the entire year — and then some.

Where They Stand

Shares of full-service brokerage houses at a glance