We examine the role of information from the options market in forecasting the equity premium. We provide evidence that the equity premium is predictable out-of-sample using a set of CBOE strategy benchmark indices as predictors. We use a range of econometric approaches to generate point, quantile, and density forecasts of the equity premium. We find that models based on option variables consistently outperform the historical average benchmark. In addition to statistical gains, using option predictors results in substantial economic benefits for a mean–variance investor, delivering up to a fivefold increase in certainty equivalent returns over the benchmark during the 1996–2021 sample period.

We investigate the previously unexplored herding behaviour of investors in option markets, by examining equity option contracts traded in the US between 1996 and 2012. We document strong herding effects in option trading activity that are conditional on a set of systematic factors related to periods of market stress. More specifically, we find that option investors tend to herd during periods of high market volatility risk, on dates of macroeconomic announcements, during the financial crisis of 2008, when a large number of market option positions is either opened or closed, and during periods of a large average dispersion of analysts’ forecasts.