Information Processing in Prediction Markets: An Empirical Investigation

This paper analyzes the direction and timing of information flow be- tween prices, polls, and media coverage of events traded on prediction markets. We examine the race between Barack Obama and Hillary Clin- ton in the 2008 Democratic primaries for presidential nomination and ask the following questions: (i) Do prediction market prices have infor- mation that is not reflected in contemporaneous polls and media stories? (ii) Conversely, do prices react to information that appears to be news for pollsters or is prominently featured by the media? We construct time se- ries variables that reflect the ‘pollster’s surprise’ in each primary election as well as indices that describe the extent of media coverage received by the candidates. We carry out Granger causality tests between the day-to-day percentage change in the price of the ‘Obama wins nomination’ secu- rity and these information variables. There seems to be two-way Granger causality between price changes and the surprise element in the primary results. There is also evidence of one-way Granger causality in the short run from price changes towards media indices. These results suggest that prediction market prices anticipate at least some of the discrepancy be- tween the actual outcome and the latest round of polls before the elec- tion. Nevertheless, prices also seem to be driven partly by election results, suggesting that there is an element of the pollster’s surprise that is gen- uine news for the market. Furthermore, it seems prices capture informa- tion earlier than its revelation in the news media, although the apparent strength of this effect depends on modeling decisions such as the inclu- sion autoregressive conditional heteroskedasticity terms in the price pro- cess.