Sentiment, Convergence of Opinion, and Market Crash

I introduce a novel proxy of investor sentiment and differences of opinion among trendchasing investors to forecast skewness in daily aggregate stock market returns. The new proxy is an easy-to-construct, real time measure available at different frequencies for more than a century. Empirically I find that negative skewness is most pronounced when investors have experienced high sentiment. The role of differences of opinion depends on the states of average investor sentiment: it positively forecasts market skewness in an optimistic state, but negatively forecasts it in a pessimistic state. Conceptually, I provide an explanation for the role of differences of opinion based on the theory of Abreu and Brunnermeier (2003). I argue that convergence of opinion in an optimistic state indicates that the price run-up is unlikely to be sustained since fewer investors can remain net buyers in the future. Therefore rational arbitrageurs coordinate their attack on the bubble, leading to a market crash. Vice versa, the convergence of opinion in a pessimistic state promotes coordinated purchases among rational arbitrageurs, leading to a strong recovery.