Third order efficiency implies fourth order efficiency: A resolution of the conjecture of J. K. Ghosh

Under suitable regularity conditions, it is shown that a third order asymptotically efficient estimator is fourth order asymptotically efficient in some class of estimators in the sense that the estimator has the most concentration probability in any symmetric interval around the true parameter up to the fourth order in the class. This is a resolution of the conjecture by Ghosh (1994, Higher Order Asymptotics, Institute of Mathematical Statistics, Hayward, California). It is also shown that the bias-adjusted maximum likelihood estimator is fourth order asymptotically efficient in the class.     

Loss of information of a statistic for a family of non-regular distributions, II: more general case

In the paper of Akahira (Ann Inst Statist Math 48:349–364, 1996), it was shown that the second order asymptotic loss of information in reducing to a statistic consisting of extreme values and an asymptotically ancillary statistic vanished for a family of non-regular distributions whose densities have the same values and the sum of differential coefficients at the endpoints of the bounded support is equal to zero. In this paper, the result can be shown to be extended to the case of a family of non-regular distributions without the above restriction.     

Second order asymptotic optimality of estimators for a density with finite cusps

We consider i.i.d. samples from a continuous density with finite cusps. Then we obtain the bound for the second order asymptotic distribution of all asymptotically median unbiased estimators. Further we get the second order asymptotic distribution of a bias-adjusted maximum likelihood estimator, and we see that it is not generally second order asymptotically efficient.     

Discretized likelihood methods—asymptotic properties of discretized likelihood estimators (DLE's)

Suppose thatX ₁,X ₂, ...,X _n , ... is a sequence of i.i.d. random variables with a densityf(x, θ). Letc _n be a maximum order of consistency. We consider a solution \(\hat \theta _n \) of the discretized likelihood equation $$\sum\limits_{i = 1}^n {\log f(X_i ,\hat \theta _n + rc_n^{ - 1} ) - } \sum\limits_{i = 1}^n {\log f(X_i ,\hat \theta _n ) = a_n (\hat \theta _n ,r)} $$ wherea _n (θ,r) is chosen so that \(\hat \theta _n \) is asymptotically median unbiased (AMU). Then the solution \(\hat \theta _n \) is called a discretized likelihood estimator (DLE). In this paper it is shown in comparison with DLE that a maximum likelihood estimator (MLE) is second order asymptotically efficient but not third order asymptotically efficient in the regular case. Further it is seen that the asymptotic efficiency (including higher order cases) may be systematically discussed by the discretized likelihood methods.     

Asymptotic optimality of estimators in non-regular cases

Summary The bound of the asymptotic distributions of for all asymptotically median unbiased (AMU) estimators is given in non-regular cases. It provides us with a powerful criterion for an AMU estimator to be two-sided asymptotically efficient and also useful in the cases when there may not exist a two-sided asymptotically efficient estimator since we may find an AMU estimator whose asymptotic distribution attains at least at a point, or an AMU estimator whose asymptotic distribution is uniformly “close” to it. Some examples are given. The results of this paper have been presented at the Meeting on Statistical Theory of Model Analysis at Tsukuba University in Japan, October 1979.     

Second order asymptotic comparison of estimators of a common parameter in the double exponential case

Summary The problem to estimate a common parameter for the pooled sample from the double exponential distributions is discussed in the presence of nuisance parameters. The maximum likelihood estimator, a weighted median, a weighted mean and others are asymptotically compared up to the second order, i.e. the ordern ^−1/2 with the asymptotic expansions of their distributions. University of Electro-communications     

Information Inequalities for the Bayes Risk for a Family of Non-Regular Distributions

For a family of non-regular distributions with a location parameter including the uniform and truncated distributions, the stochastic expansion of the Bayes estimator is given and the asymptotic lower bound for the Bayes risk is obtained and shown to be sharp. Some examples are also given.     

Estimation of a common parameter for pooled samples from the uniform distributions

Summary The problem to estimate a common parameter for the pooled sample from the uniform distributions is discussed in the presence of nuisance parameters. The maximum likelihood estimator (MLE) and others are compared and it is shown that the MLE based on the pooled sample is not (asymptotically) efficient.     

On a family of distributions attaining the Bhattacharyya bound

A family of distributions for which an unbiased estimator of a functiong(θ) of a real parameter θ can attain the second order Bhattacharyya lower bound is derived. Indeed, we obtain a necessary and sufficient condition for the attainment of the second order Bhattacharyya bound for a family of mixtures of distributions which belong to the exponential family. Furthermore, we give an example which does not satisfy this condition, but where the Bhattacharyya bound is attainable for a non-exponential family of distributions.