Unspanned stochastic volatility from an empirical and practical perspective

I conduct a simulation study to address concerns raised in the empirical literature on unspanned stochastic volatility (USV, i.e., interest-rate-volatility risk that cannot be hedged with bonds or swaps). Regressions have been the popular method of identifying and measuring USV, and have led to a consensus that is in favour of USV models. Despite plausible challenges to this approach, my simulations show that regressions are able to correctly identify the presence and absence of USV. This relies on a number of methodological considerations which are inconsistent in the literature. Regression results from empirical data, from several modern interest-rate markets, resemble results from data simulated from USV models. I then assess the economic significance of USV. By comparing hedged and unhedged returns of market interest-rate options, I develop quantitative guidelines around how unspanned volatility risk compares to interest-rate risk. (C) 2020 Elsevier B.V. All rights reserved.