Market Sentiment and Paradigm Shifts in Equity Premium Forecasting

Abstract

There is a recent debate and even a doubt about whether fundamental economic variables can predict equity premium or not. Some remedies seem working well and help in restoring the confidence on predictability. However, we show that those remedies are fragile and irrelevant in some sense. The predictability is gone again, even with those remedies utilized, once market sentiment kicks in to distort the fundamental link between economic variables and equity premium. In contrast, without using any remedies, economic variables still show predicting power as long as sentiment stays low to not distort the link. In addition, we show that many non-fundamental predictors, such as time-series momentum and 52-week high, lose their power when sentiment is low since their power depends on behavioral activities significant only in high sentiment periods. As about 80% (20%) times can be classified as low (high) sentiment periods in our framework, fundamental predictors seem a more prevalent force than non-fundamental predictors in terms of forecasting equity premium. Nevertheless, investors can be better-off by utilizing both type of predictors though need to conduct a paradigm shift between fundamental predictors in low sentiment periods and non-fundamental predictors in high sentiment periods.