Title: Market maker inventory, bid/ask spreads, and the computation of option implied risk measures
Authors: Daniela Osterrieder - Rutgers Business School (United States) [presenting]
Bjorn Eraker - Wisconsin School of Business (United States)
Ivan Shaliastovich - Wisconsin School of Business (United States)
Abstract: Option implied risk measures (OIRMs) such as the VIX index are computed from the midpoint of best bids and asks. For these measurements to be unbiased, the midpoints are implicitly assumed to be unbiased estimates of a ``true'' underlying option price. We derive a model where market makers adjust their spreads according to the size of their inventory. The model implies that negative inventory leads to upwardly biased midpoints, highly volatile asking prices, and right skewed spreads. We rely on a generalized method of moments (GMM) estimation procedure to show that these properties are true in the data set, which is comprised of detailed option-market observations and high-frequency measures of the underlying stock market. We further show that the OIRM's are very sensitive to the spread. For example, the ability of the variance risk premium (VRP) and higher order option implied tail measures in predicting stock returns is very sensitive to whether the they are computed from the bid, ask or midpoint.