Information Aggregation Efficiency of Prediction Markets

Abstract

The increased complexity of the business environment, such as globalization of the market, faster introduction of new products, more interdependencies among firms and financial crises, has reduced the forecasting accuracy of conventional prediction methods based on historical data or experts. How can we predict the future? Where can we find information about the future?

Over the past decade, some in the business world have come to believe that the best forecasts emerge from neither past behavior patterns nor far-removed experts who analyze markets, but rather crowds; the front-line employees who are working directly with new products and services and interacting daily with buyers, sellers and customers in the field, as they have the most relevant and updated information and knowledge required for forecasting. A prediction market, an elegant and well-designed method for capturing the wisdom of crowds and predicting the outcome of a future event, has been, therefore, introduced. Its promising forecasting results have inspired much enthusiasm among both researchers and practitioners in recent years.

This dissertation adopts the information-based view to investigate the effect of information transparency on traders’ behavior and prediction market performance. The research consists of three empirical studies. The case study investigates the activity of and dynamic interactions between traders in an internal prediction market. The subsequent laboratory experiment examines the effect of price information transparency on market performance via traders’ behavior. The final field experiment further investigates different levels of price information transparency in an internal prediction market in a real business environment. The dissertation distinguishes clearly between information aggregation efficiency and market predictive accuracy for the analysis of prediction market performance by defining and developing a measurement of information aggregation efficiency. This research, as a whole, contributes to the academic literature on information transparency and prediction markets, and also demonstrates the considerable potential of prediction markets in managerial decision-making.