Using Markets To Make Decisions

Over the last decade considerable attention has been given to the possibilities of predicting political and other events by observing the outcome of appropriately designed experimental markets. In the Iowa Presidential Stock Market (cf. Forsythe et al. [1992]) the outcome of the U.S. presidential election was predicted successfully through observations of a stock market, where individual experimental subjects traded shares in each of the presidential candidates. Since then similar experimental markets have been used to predict different electoral outcomes in the U.S. as well as a number of other countries (Berg, Nelson, Rietz [2001]). A remarkable feature of these experimental asset markets is that they often yield forecasts more reliable than conventional techniques, including opinion polls, expert opinions and crude statistical analysis. A number of reasons have been advanced why opinion polls and expert opinions may yield unreliable, even biased, predictions of electoral outcomes. Among these are judgement bias, inappropriate sampling of the population, and the problem of low voter turnout (cf. Forsythe et al. [1994] for a discussion of these and other possibilities). However, very little formal analysis has been devoted to the reasons why markets appropriately designed may yield reliable predictions. In an experimental asset market each trader is subjected to a variety of information regarding the outcome of the event being predicted. In the case of an electoral outcome opinion polls are arguably the most prominent