Traders' heterogeneity and bubble-crash patterns in experimental asset markets

We propose a heterogeneous agent model for experimental closed-book call markets with speculators, fundamental and noise traders. We provide structural estimates of the parameters of the model using new experimental data, which allow us to track individual behavior, cognitive reflection abilities, and accuracy of price forecasts. Based on the model's predictions for individual behavior we identify different types of traders in the data. We find that fundamental traders and speculators have higher terminal wealth and perform better on a cognitive reflection test and price forecasting than noise traders. More importantly, we find that all three types of traders are important to understand the mechanics of bubbles and crashes. In the initial period, fundamental traders buy from noise traders. Next, speculators buy from fundamental traders during the boom. Finally, speculators generate the crash by selling to noise traders. Our model predicts smaller bubbles if the cash and asset endowments are higher, keeping the cash-to-asset ratio constant. Our theory has predictive power as we confirm this prediction with additional out-of-sample data. (C) 2015 Elsevier B.V. All rights reserved.