On the last time and the number of times an estimator is more than epsilon from its target value

Abstract

Suppose ${\hat{\theta }}_n$ is a strongly consistent estimator for ${\theta }_0$ in some i.i.d. situation. Let $N_{\epsilon }$ and $Q_{\epsilon }$ be respectively the last $n$ and the total number of $n$ for which ${\hat{\theta }}_n$ is at least $\epsilon$ away from ${\theta }_0$. The limit distributions for ${\epsilon }^2{N}_{\epsilon }$ and ${\epsilon }^2{Q}_{\epsilon }$ as $\epsilon$ goes to zero are obtained under natural and weak conditions. The theory covers both parametric and nonparametric cases, multi-dimensional parameters, and general distance functions. Our results are of probabilistic interest, and, on the statistical side, suggest ways in which competing estimators can be compared. In particular several new optimality properties for the maximum likelihood estimator sequence in parametric families are established. Another use of our results is ways of constructing sequential fixed-volume or shrinking-volume confidence sets, as well as sequential tests with power 1. The paper also includes limit distribution results for the last $n$ and the number of $n$ for which the supremum distance $\Vert F_n-F\Vert \ge \epsilon$, where $F_n$ is the empirical distribution function. Yet other results are reached for ${\epsilon }^{5/2}{N}_{\epsilon }$ and ${\epsilon }^{5/2}{Q}_{\epsilon }$ in the context of nonparametric density estimation, referring to the last time and the number of times where $|f_n(x)f(x)|\ge \epsilon$. Finally it is shown that our results extend to several non-i.i.d. situations.

18 pages, no figures; Statistical Research Report, Department of Mathematics, University of Oslo, from April 1991, now arXiv'd March 2026. The paper has appeared in Annals of Statistics, 1992, vol. 20, pages 469-489, at this url: projecteuclid.org/journals/annals-of-statistics/volume-20/issue-1/On-the-Last-Time-and-the-Number-of-Times-an/10.1214/aos/1176348533.full