Asymptotic analysis for stochastic volatility: Edgeworth expansion (Q638406)

This paper establishes the validity of an approximation formula for European option prices under a general stochastic volatility model. Here \((X_t)\) is a diffusion driven by a two dimensional Brownian motion, specifically the volatility, i.e. the diffusion coefficient in the SDE for \(X\), is itself a solution to an SDE driven by a correlated Brownian motion. Parameter \(\eta\) controls the \textit{time scale of the volatility process}. The case \(\eta=0\) corresponds to \(X\) being a geometric Brownian motion (Black-Scholes model). The paper studies approximations of \(\mathbb{E}[f(X^\eta_T)]\) by expanding it around \(\mathbb{E}[f(X^0_T)]\) (the Black-Scholes price) in the parameter \(\eta\). Such expansions have been studied in many papers in past, in particular using singular perturbation methods, see e.g. [\textit{J.-P. Fouque}, Multiscale stochastic volatility for equity, interest rate, and credit derivatives. Cambridge: Cambridge University Press (2011; Zbl 1248.91003)] and the references therein. The contribution of this work is to provide a proof of the expansion formula and provide a precise uniform error bound, which does not depend on regularity of \(f\) but holds for all bounded \(f\). This is done using Edgeworth expansion for ergodic diffusions developed in the previous work by the same author [Probab. Theory Relat. Fields 142, No. 1--2, 1--20 (2008; Zbl 1154.60018)].