Heterogeneous forecasters and nonlinear expectation formation in the US stock market

We use a Panel Smooth Transition Regression (STR) model to study nonlinearities in the expectation-formation process in the U.S. 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 market conditions summarized by stock-market misalignments and recent returns change. We find that survey participants form stabilizing expectations in the long run. Short-run expectations, in contrast, are consistent with weak mean reversion of stock prices.