Option-based Equity Risk Premiums

Abstract

We construct the term structure of the (forward-looking, US market) equity risk premium from SPX option chains. The method is "model-light". Risk-neutral probability densities are estimated by fitting $N$-component Gaussian mixture models to option quotes, where $N$ is a small integer (here 4 or 5). These densities are transformed to their real-world equivalents by exponential tilting with a single parameter: the Coefficient of Relative Risk Aversion $\kappa$. From history, I estimate $\kappa =3\pm 0.5$. From the inferred real-world densities, the equity risk premium is readily calculated. Three term structures serve as examples.

44 pages, 12 figures, revised Appendix A, added reference to an application to COVID-19 pandemic, supplemented conclusions with a remark about Ross Recovery