We thought it was dead, but the controversy over a widely used database of Wall Street research has popped up again. The debate began last fall when a group of researchers questioned the integrity of Thomson Financial’s I/B/E/S service, a clearinghouse of analysts’ stock picks. The researchers appeared to backtrack after Thomson disputed their methodology, but last month they published the paper and are standing by its conclusions.
“Is it sleazy fraud or inadvertent error?” a headline on Slate asks. “You be the judge.”
In their paper titled “Rewriting History,” professors Alexander Ljungqvist of New York University, Christopher J. Malloy of the London Business School and Felicia C. Marston of the University of Virginia say they found 55,000 changes to the database from 1993 to 2002 that tend to make certain stock analysts look good.
The database is widely used by fund managers, academic researchers, by regulators to track questionable activity on Wall Street — and by The Wall Street Journal to assemble its annual list of the best analysts.
Specifically, the professors say, the analysts’ recommendations were altered to make them appear more conservative toward the end of boom cycles, causing returns on theoretical portfolios based on those picks to be 15 percent to 42 percent better. The alterations, according to the professors, strongly correlate to the best performers on the Wall Street Journal’s “Best on the Street” ranking.
The professors are careful to say they aren’t alleging malfeasance, only that the alterations are not random. It’s hard to see, though, why else the data would be changed this way.
Thomson disputes the findings. Mike Thompson, the head of research for Thomson Financial, told Barron’s that the professors “really did a hatchet job,” adding, “We’re a little stunned.”
Some outside observers, though, think the paper strongly suggests that there is something funny going on. The professors “make a good case in their paper that these are changes that do not seem to be random,” Brad Barber, a finance professor at the University of California, Davis, told Barron’s.
The Wall Street Journal ranking is meant to be an objective, data-driven ranking, unlike other rankings, like the one produced by Institutional Investor magazine, that are essentially popularity contests. But how and why could the I/B/E/S data be compromised?
According to the Wall Street Journal and Thomson, analysts are allowed to view the data through a special Thomson Web site and request changes to correct inaccuracies.
Thomson said that all requested changes are carefully reviewed by in-house analysts. Nevertheless, Mr. Barber said, allowing stock analysts to have any say in how the data are presented can’t help but skew results. It “can only lead to bias,” he told Barron’s. And a former Thomson executive challenged the company’s characterization of how carefully the analysts’ requests are screened.
Last fall, in an earlier version of the paper, the professors asserted that some analysts’ names had been stripped from the database, often in cases where their stock picks proved to be way off. At the time, Thomson attributed the deletions to database maintenance. But the professors now say that about a quarter of those deletions still are not accounted for.
Thomson insists that the correct data are all there, but are contained in different data sets, which would have to be reconciled for the results to be fully understood. A spokesperson told Slate’s Daniel Gross that Thomson has offered to walk the professors through the complex process.
“It’s hard to know what to conclude,” Mr. Gross writes. But “Wall Street executives—stock analysts among them — have shown that there’s virtually nothing they won’t do, and nobody they won’t corrupt, to advance their own careers and portfolios.”
Go to Article from Slate »
Go to Article from Barron’s (Subscription Required) »
Go to Earlier Item on DealBook »