On efficient stopping times (Q1060518)

The author considers the problem of sequential estimation for a stochastic process where the sufficient statistics form a process of the exponential class. A sequential plan (\(\tau\),f,h) consists of a stopping time \(\tau\), a real function h of the unknown k-dimensional parameter and an unbiased estimator f of h. A sequential plan is called efficient if it yields equality in the Cramér-Rao-inequality for all parameters in some k-dimensional interval. In this paper, the author characterizes all stopping times for which there exist f and h as above such that (\(\tau\),f,h) is efficient. Some examples are given which illustrate this characterization.