Local asymptotic mixed normality property for nonsynchronously observed diffusion processes (Q888473)

The problem of estimating a parameter affecting the drift and diffusion coefficient of a planar diffusion process with non-synchronously generated observations is studied in the framework of local asymptotic mixed normality, a notion introduced by \textit{P. Jeganathan} [Sankhyā, Ser. A 45, 66--87 (1983; Zbl 0574.62035)] who also proved that it implies an asymptotic min-max property that characterizes asymptotically efficient estimators. Under suitable regularity conditions on the coefficients of the diffusion and appropriate asymptotic behavior of the sampling instants (assumed to be independent of the diffusion), the aforementioned property is established and, as a consequence, asymptotic efficiency of a quasi-maximum likelihood estimator and a Bayes type estimator are recovered. Regularity results for transition densities based on Malliavin calculus play a major role in this development.