Option-based assessments of expected price distributions.

摘要

Option premia are used to solve for implied expected future price distribution parameters via a no-arbitrage option pricing model. The parameters of the implied distributions are then modeled as functions of possibly stochastic structural variables to detect non-stationarities in the underlying price processes. Transactions data are used consisting of daily matched futures and options prices on the S&P 500 and Soybeans futures contracts. The use of synchronous data avoids the problem of non-simultaneity of quoted prices from different markets. Calibration tests indicate that the markets are well calibrated over intermediate time periods and are not well calibrated immediately prior to expiration and at distant times to expiration. Switching regression results indicate an acceleration of the decrease in variance as time to maturity decreases. Also detected are the presence of distinct regimes of influence along a time index that govern parameters of the switching regressions. Event-study benchmarks are generated as a consequence of techniques used, and monthly intervals appear to be the most prominent events in these markets. The ability of options markets to provide complete probabilistic descriptions of uncertainty is judged to be highly useful and fairly reliable over intermediate length time periods.