Description: Job Market Practice Talk Mahendra Singh (Iowa State University)

Location: 368A Heady Hall

Title: Market Stress in Agricultural Markets: Can Alternative Implied Volatility Measures Predict It?

Abstract: In this paper, we provide a comprehensive assessment of return and volatility predictive ability of option-implied information for corn, crude oil, soybean, and wheat futures options using daily settlement option premiums. In particular, we investigate the information content of four option-implied volatility measures namely: VIX (Demeterfi et al. 1999), SVIX (Martin, 2017), Black’s (1976) implied volatility, and risk-neutral variance measure (Bakshi et al. 2003). Furthermore, we examine the incremental information embedded in the gap of VIX and SVIX which could arise due to the violation of a standard asset pricing theory assumption–that the stochastic discount factor and returns are conditionally jointly log-normal–during the market stress time, i.e., a period of high volatility (Martin, 2017). In addition, we also augment the forecasting exercises with volatility smirk (Xing et al., 2010), risk-neutral skewness and risk-neutral kurtosis (Bakshi et al., 2003). We find several interesting results: (i) option-implied measures have very limited predictive ability to forecast daily returns and could explain up to 2% of the total variation; (ii) volatility smirk picks up consistent and strong statistical significance to forecast daily returns (impacts negatively) for all the four commodities considered in the sample; (iii) option-implied variables have strong predictive ability to forecast realized volatility (up to 60% of the total variation); (iv) all of the implied volatility measures mentioned above impacts realized volatility positively for all of the commodities; (v) the gaps between VIX and SVIX show limited predictive ability for the realized volatility although those gaps do widen during the period of high market volatility; (vi) volatility smirk, risk-neutral skewness, and risk-neutral are statistically significant predictors for corn and soybean futures. Overall, our findings seem to have some relevance for the real-world applications, i.e., for the risk management agencies.