Noncausality and asset pricing

Misspecification of agents' information sets or expectation formation mechanisms may lead to noncausal autoregressive representations of asset prices. Within the class of linear (vector) autoregressions, annual US stock prices are found to be best described by noncausal models, implying that agents' expectations are not revealed to an outside observer such as an econometrician observing only realized market data. A simulation study shows that noncausal asset prices are observed when the data are generated by asset-pricing models featuring heterogeneous expectations.