Forecasting the Vote: A Theoretical Comparison of Election Markets and Public Opinion Polls

Michael E. Sobel421 Fayerweather Hall, Columbia University, New York, NY 10027 e-mail: mes105{at}columbia.edu The dominant methodology for short-term forecasting of electoraloutcomes uses trial-heat polls, where respondents report theircurrent electoral preferences (not their election-day predictions).Election markets, where self-selected participants trade sharesof candidates at prices predictive of election-day results,provide an alternative method that often produces more accurateforecasts. Consequently, increasing attention is being paidto this methodology. However, it is poorly understood and lackstheoretical justification. Surprisingly, the rationale for forecastingusing trial-heat polls has not been completely developed either.We develop the justification for using both election marketsand public opinion polls to forecast electoral outcomes, givingconditions under which each method performs ideally. For theideal case, we prove (under the reasonable assumption that marketparticipants are aware of the poll results) that the mean squareprediction error for the market forecast is smaller than thatof any forecast based on one or more polls. The case in whichthe assumptions supporting each method fail is also considered.It is often reasonable to expect that the best case resultshold approximately, in which case the market forecast shouldalso beat any poll-based forecast. We also compare the biasand variance of market and poll-based forecasts; our resultssuggest the utility of using the series of market prices tostudy the course of campaigns.