Information Aggregation Mechanisms and Environmental Complexity

We experimentally test four different information aggregation mechanisms. Each mechanism is tested in a simple ‘two-state’ environment and a complex ‘eight-state’ environment where the number of securities is large (relative to the number of traders) and the Bayesian inference problem is significantly more difficult. We find that the Double Auction market mechanism performs well in the simple environment, but performs poorly in the complex environment. An Iterative Polling mechanism performs the best in the complex environment but exhibits failures in the simple environment. Thus, the optimal mechanism choice depends crucially on the complexity of the environment. JEL: C72, C91, D82