Election Stock Market - Reliability of Election Stock Markets

Reliability of Election Stock Markets

Election stock markets are prediction markets for a particular purpose: elections. Even though election stock markets have been conducted for almost twenty years, the accuracy of these markets is nearly always judged by comparing the election stock market prediction (closing prices) on election eve with final pre-election polls and actual outcomes. Evidence that election stock markets perform remarkably well predicting election outcomes is found in a string of academic papers, mostly based on data from the Iowa Electronic Markets and the UBC Election Stock Market. Accuracy is typically measured as the average absolute forecast error for vote shares and seat shares. A more rigorous attempt to assess the performance of election stock markets is found in Berg et al. (2008); they report that for five recent elections covered by the Iowa Electronic Markets, the average absolute error in the market's prediction of the major-party presidential vote share across the 5 days prior to the election was 1.20 percentage points, while opinion polls conducted during that same time had an average error of 1.62 percentage points. Berg et al. (2008) also report evidence that election stock markets outperform polls for longer time periods before the election date.

Erikson and Wlezien (2008) challenge the view that election stock markets outperform polls. They argue that polls only measure preferences on the polling day, whereas election stock markets forecast the outcome on election day. When poll leads are discounted using statistical techniques, they find that poll-based forecasts outperform vote-share market prices.

A critical feature for the proper functioning of election stock markets is market liquidity. As prediction markets function through aggregation of beliefs and opinions into market prices, high trading volume and/or a continuous stream of new investments are essential for prices to provide an accurate forecast of the election outcome. Signs that liquidity is lacking in an election stock market include wide spreads (large differences between bid and ask prices) and arbitrage opportunities (where the sum of bid prices exceeds the value of a unit portfolio, or where the sum of ask prices is lower than the value of a unit portfolio). As election stock markets are opinion aggregators, the accuracy of such markets can be expected to increase with the number of market participants. Investment caps (as maintained by the Iowa Electronic Markets and the UBC Election Stock Market) level the trading opportunities among traders. Whether investment caps help with prediction accuracy has not yet been determined conclusively. However, without an investment cap, commercial election stock markets may be dominated by a small number of traders. The existence of transaction costs for investing and trading in commercial election stock markets may also reduce their efficiency.

Sunstein (2006) argues that prediction markets are often more accurate than deliberating groups because prediction markets create strong incentives for revelation of privately held knowledge and succeed in aggregating widely dispersed information. Contrastingly, deliberating groups often amplify individual errors, and group members may fall victim to a bad cascade, either informational or reputational. Deliberators may emphasize shared information at the expense of uniquely held information.