We love prediction markets. And we love Google. So we were pretty excited to stumble on a little article on internal prediction markets at Google in the New York Times. The article is a great summary of a recently published report by Eric Zitzewitz of Dartmouth College, Justin Wolfers of Wharton, and Bo Cowgill of Google. The report (which you can download HERE) is the result of a two-year study of Google’s prediction markets and how they relate to internal information transfer. While you won’t find out how effective Google’s prediction markets are in the report’s 30-something pages, you will learn some fascinating things about prediction markets and about Google. Among the findings:
- Employees’ micro-geography was a significant factor in how employees feel about/view certain events. The strongest correlation in betting was found among people who sat in close proximity to one another, trumping friendship and other close social ties. The correlation weakens as employees get farther away from one another on the same floor, and almost disappears when employees are on different floors from one another. In fact, employees on different floors of the same building were no more correlated than employees in different cities. The lesson? Open, high-density office environments foster greater information transfer. Yes, this finding is a little obvious, but it’s great to have the data to back it up.
- Google employees typically overpriced securities tied to optimistic outcomes (e.g. Google will have a 99% market share by the end of the quarter). This bias was also heightened during the days following substantial growth in the price of Google’s real-world stock. What’s even more interesting is that employees that had a longer tenure with Google (and more experience trading) tended to exhibit less of this optimism bias. This indicates that prediction markets may work better as collective experience increases.
- Google employees change offices very frequently (on average, once every 90 days during the course of the study). It’s almost as if Google is constantly watching information flow and readjusting employee office assignments accordingly.
- Google had a mix of “serious” securities and “fun” securities in its markets. Serious securities directly related to business at Google (things like market share or product release dates); fun securities had to do with completely unrelated events (the quality of Star Wars Episode III or the federal funds rate). The study found that volume in fun and serious markets were highly correlated, suggesting that fun markets help create liquidity and activity in serious markets.