Abstract
This thesis studies the information content of the implied volatility surface in options markets.
In Chapter 2, we study the implied volatility curves in four commodity markets by adopting a quadratic function on a standardized moneyness measure as in Zhang and Xiang (2008). First, we document the term structure and dynamics of implied volatility curves. Overall, the commodity implied volatility curves are negatively skewed with a positive curvature. Then we analyze the commodity and S&P 500 returns’ predictability based on in-sample and out-of-sample tests and find that the information embedded in implied volatility curves can significantly predict monthly commodity and S&P500 returns. For example, the risk-neutral fourth cumulant (FC) from the crude oil market outperforms all of the standard predictors in predicting the S&P 500 returns.
In Chapter 3, we examine the information inferred from the Carr and Wu (2020) formula based on a new option pricing framework in the United States Oil Fund (USO) options. We first document the term structure and dynamics of the risk-neutral variance and covariance rates which lead to a negatively skewed shape of the USO implied volatility smile with a positive curvature. We then investigate the return predictability of the innovations in the risk-neutral variance and covariance rates (DRNV and DRNC) and their term structures (TRNV and TRNC) and find that DRNC is a significant and robust predictor to forecast daily, weekly and monthly USO excess returns in both statistical and economic terms based on in-sample and out-of-sample tests.
Finally, in Chapter 4, we introduce a new stock market return predictor, covariance risk premium (CRP), which is defined as the difference between the historical and the risk-neutral covariance rates (HC and RC) of the implied volatility changes and the market index returns. CRP is positively and significantly related to future stock market returns at horizons from 1 month to 24 months. This chapter empirically documents that CRP has significant in-sample and out-of-sample predictive ability, generates sizable economic value for a mean-variance investor and outperforms many well-known predictors. In addition, CRP can predict cross-sectional stock returns at the portfolio level.