How does informational heterogeneity a ect the quality of forecasts ? ?

We investigate a toy model of inductive interacting agents aiming to forecast a continuous, exogenous random variable E. Private information on E is spread heterogeneously across agents. Herding turns out to be the preferred forecasting mechanism when heterogeneity is maximal. However in such conditions aggregating information e ciently is hard even in the presence of learning, as the herding ratio rises signi cantly above the e cient-market expectation of 1 and remarkably close to the empirically observed values. We also study how di erent parameters (interaction range, learning rate, cost of information and score memory) may a ect this scenario and improve e ciency in the hard phase.