Nonlinear Expectation Formation in the U.S. Stock Market

This research applies data from the Livingston survey to study the time variation in the sentiment of U.S. stock-market forecasters. A Panel Smooth Transition Regression (STR) model is estimated to identify the importance of market conditions summarized by stock-market misalignments and recent returns for the formation of regressive and extrapolative expectations. We find that survey participants expect little mean reversion in times of large misalignments reflecting the observed substantial and persistent swings in stock prices. Recent returns are negatively extrapolated depending on the sign and the size of the return revealing a contrarian behavior of forecasters in the presence of market exuberance.