How Journal Found Options Pattern [WSJ]
By CHARLES FORELLE
May 22, 2006; Page A11
In preparing the latest article on options granting practices, The Wall Street Journal examined the grant patterns for five companies whose option-grant dates were regularly followed by large gains in share price. The Journal looked at all option grants to each company's top executive between roughly 1995 and August 2002, when securities-law changes curtailed the potential for backdating.
Experts say that a regular pattern of sharp stock gains after the stated grant dates could be an indication of backdating. That's because nobody authorizing grants on a specific day could know how the stock would perform in the future. By chance alone, grants ought to be followed by a mixed bag of stock performance -- some rises, some declines.
Using an identical method to that followed for an article in March, the Journal calculated how much the shares rose, in percentage terms, in the 20 trading days following each grant date. The analysis then ranked each grant date against all other days in the year, using the percentages. There are generally 252 trading days in a year. The analysis excluded any repricings of earlier-granted options.
For instance, a grant priced on April 17, 2000, to executives at Boston Communications Group Inc. was followed by a 68% leap in share price over the next 20 trading days. That was the largest 20-day jump of the year, so April 17 was ranked No. 1 for that year.
Having the grant fall by chance on such a high-ranking day is unusual, but it's extremely unlikely to see several grants over a number of years each bearing high ranks.
John Emerson, an assistant professor of statistics at Yale University, developed a computer program to quantify the probability of the grants falling as they did had the dates been selected at random. The program figured the probabilities for each executive's rank-pattern except Mr. Levy's at KLA-Tencor. Because the complexity and number of grants in Mr. Levy's pattern made a direct calculation infeasible, the Journal used an alternative method that simulates the random selection of patterns. Repeating the simulation billions of times should yield a close estimate of the probability. Mr. Emerson says the simulation method should give results similar to his program's. Several tests confirmed that it indeed does.
Write to Charles Forelle at charles.forelle@wsj.com |
