Robust Application of Information from Option Data

• Applicants: Professorin Dr. Antje Mahayni, since 8/2017; Professorin Dr. Judith Christiane Schneider
• Subject Area: Accounting and Finance
• Term: from 2016 to 2019
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 316058991

Practitioners as well as academics show considerable interest in extracting information on future stock price developments out of market option prices. When pricing complex financial products, practitioners calibrate different theoretical models to observed option prices. This standard pricing approach is thus prone to model risk. In practice, this risk is mostly ignored or reduced to parameter uncertainty. The first goal of this proposal is to link a non-parametric worst case approach inspired by the methods used in robust control with market option prices. This robustifies the pricing of exotic and illiquid options and allows to calculate a direct risk margin for model risk - in line with recent regulatory proposals. The second area of this research proposal will add to the literature which analyzes option-implied information to gain more insights into characteristics of the risk-neutral distribution, thus extracting qualitative features of beliefs held in the market. Typical measures in the literature are the first four moments or linear correlations, both cross-sectionally and over time. The methods for estimating these measures rely, however, on structural assumptions like extrapolation of the market option prices. The following research proposal addresses the question whether the existing methods can be robustified. Concerning the characteristics of the risk-neutral distribution, we suggest to apply measures which capture the shape of the risk-neutral distribution but do not heavily rely on tail behavior which needs to be extrapolated. The performance of these measures like quantile-based skewness will be tested theoretically and within an economic applications. We also intend to study alternative dependence measures beyond linear correlation such as mutual information. Again, we are interested both in the economic content of these measures, and their robustness properties regarding estimation from option price data.

• DFG Programme: Research Grants
• Ehemaliger Antragsteller: Dr. Nikolaus Schweizer, until 8/2017