Identifying Analysts' Echo Chambers: Evidence from Multilayer Information Networks
Abstract
This study investigates the formation and economic consequences of "echo chambers" within sell-side analysts' information networks. Departing from the view of analysts as independent agents, we construct a novel multilayer network that integrates three distinct channels of professional interaction: team cooperation, intra-firm affiliation, and stock co-coverage. We aggregate these layers using an optimization procedure that recovers their relative importance in shaping belief similarity, and then apply community detection to identify tightly-knit analyst cliques. We document that these cliques are structurally persistent and function as potent echo chambers: analysts within the same clique exhibit significant convergence in their forecast biases and optimism, an effect distinct from strategic herding. We also find that firms covered by a greater number of distinct cliques experience a deteriorated information environment, manifested as higher forecast dispersion and increased stock price crash risk. Our findings highlight that the social structure of information intermediaries can foster insular belief systems, with significant negative externalities for market efficiency.