Compound decision in the presence of proxies with an application to spatio-temporal data

TL;DR

This paper introduces a non-parametric empirical Bayes method for estimating means in a compound decision problem with covariates, extending Fay-Herriot, and demonstrates its effectiveness on spatio-temporal data.

Contribution

It proposes a novel non-parametric empirical Bayes approach that generalizes Fay-Herriot for covariate incorporation in small area estimation.

Findings

The method has proven optimality properties.

It outperforms Fay-Herriot in semi-real data experiments.

Effective in estimating proportions in small areas with spatio-temporal covariates.

Abstract

We study the problem of incorporating covariates in a compound decision setup. It is desired to estimate the means of $n$ response variables, which are independent and normally distributed, and each is accompanied by a vector of covariates. We suggest a method that involves non-parametric empirical Bayes techniques and may be viewed as a generalization of the celebrated Fay-Herriot (1979) method. Some optimality properties of our method are proved. We also compare it numerically with Fay-Herriot and other methods, using a `semi-real' data set that involves spatio-temporal covariates, where the goal is to estimate certain proportions in many small areas (Statistical-Areas)