Informational linkages and stock market comovement

Abstract

We empirically study how excess comovement in stock returns can be explained by stockto-stock interdependence through informational links. Informational linkages emerge from anomalous interdependence in agents’ beliefs about stocks’ economic performance. We propose a novel measure for informational linkages based on agents’ learning process, which is biased towards common information due to learning frictions, generating correlated beliefs across stocks. We empirically measure these correlated beliefs based on analysts’ forecast errors. Our results show that informational connections explain stock returns after cleaning for fundamental connections and controlling for various explanations already studied in the literature. We use our estimated informational linkages to study the propagation of a climate event and simulated shocks through the stock market, finding quantitatively important indirect effects.