Power of Ideas
Predicting Hit Products
Prof. Teck-Hua Ho champions prediction markets
Anyone who tried to buy the Wii, Nintendo's latest video game console, during the last holiday season knows about the inability of firms to forecast market demand.
Wii shortages abounded because Nintendo failed to accurately predict how successful it would be.
One novel way to improve such forecasts is a prediction market, says Teck-Hua Ho, the Haas School's William Halford Jr. Family Professor of Marketing. Ho recently outlined how prediction markets work in an article for the 50th anniversary issue of the Haas School business journal, California Management Review.
A prediction market is an exchange in which participants vote on a possible outcome by buying and selling shares that correspond to that outcome, similar to trading in the stock market. The most likely forecasts trade for a higher price than shares in a less likely scenario.
"The key idea behind a prediction market is pooling the knowledge of many people within a company," Ho says. "It's a very powerful tool for firms with many different pockets of expertise or a widely dispersed or isolated workforce."
Such markets have been created to predict the next president, Hollywood blockbusters, and flu outbreaks. But Ho and his co-author, Kay-Yut Chen, a principal scientist at Hewlett-Packard Laboratories, believe that prediction markets also work well for forecasting demand for new product innovations, particularly in the high-tech arena.
H-P tested prediction markets to forecast sales of several existing and new products and found that six of eight prediction markets were more accurate than official forecasts. "Prediction markets work because you get a lot of people and ask them to put their money where their mouth is," Chen says.
In a step-by-step guide on how to create a prediction market, Ho and Chen advise recruiting at least 50 participants and providing a strong monetary incentive to promote active trading - at least $500 per participant.
A firm then creates ten different forecasts -- either according to sales or units sold -- and gives each participant a set number of shares and cash to trade, buy, and sell, according to their beliefs about which forecast is most accurate.
After a product is launched and sales are observed, participants who own shares in the prediction that matches actual sales receive $1 per share.
Ho and Chen describe five principles, which they refer to as I4C (pronounced "I foresee"), that form the foundation of a successful prediction market. The principles are incentive, indicator, improvement, independence, and crowd.
Under those principles, a prediction market works in part because the price of shares for different forecasts serves as a concise "indicator" of information for all participants. The pricing process conveys information to uninformed participations and therefore helps them "improve" their knowledge about demand. Prediction markets work best when there is a large, "independent" "crowd" of participants who receive strong monetary "incentives," based on the theory that a crowd is more intelligent than a small group of experts, Ho and Chen argue.
However, Ho and Chen note a few common pitfalls of prediction markets, including too few participants and too little trading.
"The biggest challenge is getting people in the company to be active," Ho says. "Contributing knowledge to this market in the form of trading has to be part of the job."