"Marketwide Memory"

Abstract: We propose a novel measure that allows us to study memory associations in financial markets over the course of several decades. Using our measure, we construct memory-based beliefs and show that they can explain return expectations from surveys as well as higher-moment beliefs implied by the VIX. We also show that memory associations drive trading decisions of individual investors: when investors are more likely to recall a past positive (negative) trading experience with a stock, they are more (less) likely to repurchase that stock. Our measure builds on two well-established regularities of associative recall: similarity and interference. For each point in time, it captures the probability that a representative investor recalls past episodes of a stock. Without any further assumptions, our measure generates signature patterns of cued recall, such as the recency effect and the recall of past crises during extreme episodes. We validate our measure using actual recall patterns extracted from transcripts of corporate events, like earnings calls, and show that our measure predicts which historical periods are mentioned during these events. Overall, our results show that theories of human memory can be broadly applied in financial markets.

Read the paper