Information Processing in a Transparent Market: Evidence from a DeFi Protocol

We take advantage of a rare setting in a decentralized finance market (the MakerDao Lending Protocol) in which each trader’s activities are made publicly available in real time and recorded permanently on the blockchain to examine traders’ learning behavior in this extremely transparent market. We find robust evidence of cross-sectional persistence in traders’ performance. Despite the public availability of all traders’ past and real-time activities, investors’ learning and following decisions incorporate easier-to-process indicators (such as past liquidations) but fail to leverage indicators that are more difficult to keep track of (such as past loan returns).