For the last 15 years, economist Kay-Yut Chen has been quietly toiling away in Hewlett-Packard's labs, running experiments mostly on nearby Stanford students to help determine such things as the best incentive schemes to use with the hardware maker's suppliers and distributors, and how to set the right pricing and policies for H.P.'s online store.

Though his work has generated millions of dollars for H.P., Chen's experimental-economics lab remained the only one of its kind within a major corporation for many years. But recently that's started to change, as high-tech firms in particular have come to realize the value of testing their programs internally in labs, and experimental economics as a field has gained increasing respect.

"The picture has completely changed," Chen says, pointing to the 2002 Nobel Prize given to pioneering experimental economist Vernon Smith as an event that helped legitimize the field. Before, the field had been viewed as something of a crutch for economic theories that lacked good formal models.

Within the last three years, both Google and Yahoo have built in-house economics facilities of their own to work on such tasks as optimizing their keyword auctions, and dozens of other companies have turned to outside consultants for help on specific projects. eBay used experimental economists to develop a new seller-feedback system that wound up boosting the total value of goods sold on the site by 25 percent, according to the researcher who worked on the project.

"As things get more complex, it gets more difficult to manage your business using only buzzwords," says Özalp Özer, a Stanford professor of management science and engineering who recently ran experiments for Hitachi Global Storage Technologies to determine the optimal time to introduce a new disk drive, given the trade-offs among engineering, marketing, and manufacturing. Instead, says Özer, corporations are looking for hard data to support their business moves.

Experimental economists like Özer and Chen begin with a number of theories about which programs will maximize corporate profits or other goals and then test them—typically on paid subjects.