Estimation for Discretely Observed Small Diffusions Based on Approximate Martingale Estimating Functions (Q4677112)

This paper deals with a family of \(d\)-dimensional diffusion processes defined by the following stochastic differential equations \[ dX_{t}=b(X_{t},\theta)+\varepsilon \sigma(X_{t})dw_{t},\;t\in[0,1],\;\varepsilon\in(0,1],\;X_0=x_0, \] where \(\theta\in\bar\Theta\) the closure of an open-bounded convex subset \(\Theta\) of \(R^{p}\). It is assumed that the drift \(b\) is known apart from the parameter \(\theta\). The author considers discrete observations of \(X\) at \(n\) regularly spaced time points \(t_{k}=k/n\) on the interval \([0,1]\). It is shown that an \(M\)-estimator obtained from the approximate martingale estimating function is asymptotically efficient under the general assumption that \((\varepsilon n^{l})^{-1}\to0\) for a positive integer \(l\), as \(\varepsilon\to0, n\to\infty\).