Categorical Learning and Investor Attention

Investment based on asset categories (aka style investing) is considered a cause of major financial anomalies, such as the abnormal stock price gains of companies that changed to dotcom names during the Internet bubble without other changes in strategy. However, the exact reasons for this categorical behavior are unclear. In this project, we will conduct the first experimental test of the hypothesis that attention constraints are responsible for style investing. The efficient allocation of limited attention can lead to learning in categories, because group-level data provides information about all category members, whereas asset-specific data provides information only about the individual asset. This theory predicts that attention to categories is driven by the confluence of psychological elements, like information processing ability, and economic elements, like the number of assets within a category. We plan to directly investigate these predictions by measuring the amount of time participants spend looking at category-level vs asset-specific information in simulated financial markets, varying these psychological and economic predictors. Moreover, we can measure individual psychological traits in other tasks and see if they relate to individual differences in financial behavior. Thus, we hope to shed light on how people attempt to solve such computationally challenging problems in practice, and how limits on our ability to do so may result in economic anomalies that persist and even grow at large scales.