Information Aggregation Efficiency in Virtual Prediction Markets

Prediction markets have shown excellent information aggregation and forecasting abilities. Markets can efficiently consolidate information that is widely dispersed among a number of individuals. In such markets, securities of future outcomes are traded on virtual trading platforms. Resulting market prices reflect a market consensus about the likelihood of future outcomes. The development of electronic markets has enabled a broader application of markets and improved market design options for the purpose of business forecasting and decision support. Prediction markets are expected to become a central information management instrument of organizations in the future. However, little is known about the mechanism of information aggregation in such markets. A predominant theory of information aggregation based on differences in trader types has been proposed in literature. It states that a certain trader type shows superior ability to identify relevant information and show less cognitive biases. It states further that this trader type is responsible for efficient information aggregation. Empirical observations made in several experimental markets militate against such a theory. This thesis seeks to challenge the trader-based theory by testing its assumptions with regard to prediction accuracy and by proposing and testing an alternative theory based on allocative efficiency.