Using 719,830 analyst recommendations from 1994 to 2017, we construct various portfolios based on levels and changes in analyst recommendations and examine how the value of those recommendations in predicting the abnormal stock returns has changed over time. We find that the predictive value of...

We use a Panel Smooth Transition Regression (STR) model to study nonlinearities in the expectationformation process in the US stock market. To this end, we use data from the Livingston survey to investigate how the importance of regressive and extrapolative expectations fluctuates over time as...