Learning from trading activity in laboratory security markets with higher-order uncertainty

Abstract

This experiment examines forecasting behavior under varying information conditions to assess the extent to which traders in security markets incorporate information in trading activity to resolve fundamental uncertainty and to resolve higher-order uncertainty. Fundamental uncertainty refers to a trader's uncertainty about liquidation value of the asset while higher-order uncertainty refers to uncertainty about the beliefs of other traders about liquidation value of the asset. I find that in an experimental security market, subjects incorporate the information contained in trading activity to the extent of about 88% to resolve both fundamental uncertainty and higher-order uncertainty. When a piece of public information (the information content of which is comparable to the information contained in trading activity) is made available to the subjects, then also the incorporation of available information remains in about the same range as reported with the first set of experiments. The inability and / or refusal to incorporate 100% of the information contained in trading activity is almost entirely attributable to an inability and / or refusal to Bayesian update. The refusal to Bayesian update is consistent with several other theories and allows post-trading forecasts to be significantly correlated with pre-trading forecasts.