Institutional Forecasting: The Performance of Thin Virtual Stock Markets

We study the performance of Virtual Stock Markets (VSMs) in an institutional forecasting environment. We compare VSMs to the Combined Judgmental Forecast (CJF) and the Key Informant (KI) approach. We find that VSMs can be effectively applied in an environment with a small number of knowledgeable informants, i.e., in thin markets. Our results show that none of the three approaches differ in forecasting accuracy in a low knowledge-heterogeneity environment. However, where there is high knowledge-heterogeneity, the VSM approach outperforms the CJF approach, which in turn outperforms the KI approach. Hence, our results provide useful insight into when each of the three approaches might be most effectively applied.