Abstract
Macroeconomic forecasts are used extensively in industry and government even though the historical accuracy and reliability is questionable. Over the last couple of years prediction markets as a community forecasting method have gained interest in the scientific world and in industry. An arising question is how to detect valuable user input and identify experts in such online communities. Detecting such input would possibly enable us to improve the information aggregation mechanism and the forecast performance of such systems. We design a prediction market for economic derivatives that aggregates macro-economic information. Using market-based measures we find that user input can be evaluated ad-hoc. Further analysis shows that aggregated measures outperform established methods -such as reputation- in identifying forecasting experts. Moreover, using data from a two year field-experiment we find that expertise is stable for longer time horizons.
Recommended Citation
Teschner, Florian and Weinhardt, Christof, "Identifying Experts in Virtual Forecasting Communities" (2012). AMCIS 2012 Proceedings. 4.
https://aisel.aisnet.org/amcis2012/proceedings/VirtualCommunities/4
Identifying Experts in Virtual Forecasting Communities
Macroeconomic forecasts are used extensively in industry and government even though the historical accuracy and reliability is questionable. Over the last couple of years prediction markets as a community forecasting method have gained interest in the scientific world and in industry. An arising question is how to detect valuable user input and identify experts in such online communities. Detecting such input would possibly enable us to improve the information aggregation mechanism and the forecast performance of such systems. We design a prediction market for economic derivatives that aggregates macro-economic information. Using market-based measures we find that user input can be evaluated ad-hoc. Further analysis shows that aggregated measures outperform established methods -such as reputation- in identifying forecasting experts. Moreover, using data from a two year field-experiment we find that expertise is stable for longer time horizons.