An entropy approach to the Stein and Stein model with correlation (Q2488487)

The authors study optimal investment strategies and pricing of derivatives in the context of the \textit{E. M. Stein} and \textit{J. C. Stein} [Rev. Financial Stud. 4, 727--752 (1991)] stochastic volatility model with correlation, i.e., when the Brownian motions in the price and volatility processes are correlated. First, some properties of the minimal entropy martingale measure are reviewed. This measure plays an important role in the utility indifference approach to valuation of derivatives. The problem of existence and integrability properties of martingale measures are studied. A necessary and sufficient condition under which the stochastic exponentials are actually martingales and the resulting equivalent probability measure has finite entropy, are proposed. This condition is sufficient to guarantee existence and uniqueness of the minimal entropy measure. Then, a martingale duality approach for the calculation of the minimal entropy martingale measure in a continuous semimartingale framework is outlined. This approach leads to a certain nontrivial martingale representation equation which is studied in the specific case of the Stein and Stein model. As a consequence, in the case where the mean reversion level and the correlation coefficients are nonzero an investor who can use trading strategies adapted to Brownian filtration may achieve a higher expected exponential utility from terminal wealth than an investor who can only observe the price process.