Essays on Option Valuation and Empirical Asset Pricing

Date

2019-05

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

This dissertation consists of three essays on option valuation and empirical asset pricing. In the first essay, coauthored with Kris Jacobs, we propose an new approach based on Adaptive Metropolis-Hasting MCMC and the Implied Spot Variance Particle Filter to estimate dynamic models using large option panels. The estimation of option valuation models is challenging due to the complexity of the models and the richness of the option data. Many existing studies limit the time-series and especially the cross-sectional dimension of the option data, which may complicate the identification of model parameters. We address these computational constraints by filtering the state variables using particle weights based on model-implied spot volatilities rather than model prices. Some of our estimation results differ substantially from the existing literature. We show that samples restricted to at-the-money and especially short-maturity options may result in serious identification problems. The composition of the option sample also critically affects the relative importance of returns and options for parameter estimates when both are used in estimation.

In the second essay, coauthored with Guanglian Hu, we investigate the relation between variance risk premiums and option returns. Empirically, the out-of-the-money (OTM) S&P 500 call and put options have large negative average returns, and the literature interprets these results as inconsistent with asset pricing theory. We show that these negative OTM option returns are primarily due to the pricing of market volatility risk. With a negative volatility risk premium, expected option returns in a stochastic volatility model are consistent with average call and put option returns across all strikes. The volatility risk premium also predicts future option returns. The predictability of index option returns is both statistically and economically significant. Lastly, we find that some portion of OTM put option returns is attributable to the jump risk premium. Overall our results suggest that the pricing of volatility risk has a first-order effect on the cross-section of index option returns.

In the third essay, I propose a novel method to identify the physical measure parameters and the latent spot variances for dynamic asset pricing models from option data. It is well known that we can estimate physical parameters from underlying returns and risk-neutral parameters from options, but the existing literature mainly focuses on the option price levels and therefore ignores that option price changes contain information about both the physical and risk-neutral distributions. I find that most physical parameters are better identified using options than using returns. The parameters estimated exclusively from options provide the best fit for the monthly option returns. Moreover, By re-examining the index option return puzzle based on the parameters estimated from options, I find that the option return puzzle disappears when jumps are included in the model.

Description

Keywords

Particle Filter, Bayesian MCMC, Option Price, Option Return

Citation