Rediscovering price discovery

We propose a comprehensive structural analysis that generalizes and categorizes numerous existing models. Our approach is characterized by minimal assumptions, offering a flexible and fully identified model that enables economic interpretations and empirical applications. Building upon this theoretical foundation, we introduce a novel measure of price discovery: the Covariance Information Share (CovIS). Quite intuitively, the CovIS is the covariance between the shocks on observed market prices and the permanent shocks that impact the common efficient price. Our measure accommodates the investigation of low frequency data and correlated residuals. It also proves to be well-suited for both static and dynamic analyses of the price discovery process. We compare the CovIS measure with well-known information shares and connect dots between works that have hardly been related to each others up to now. This enables us to resolve some of the paradoxes raised by the conflicting findings sometimes encountered in empirical studies. Moreover, failures with known methods are reanalyzed as informative features. Finally, we apply our measure to both simulated and real data and revisit previous empirical studies.