Abstract

This article considers elastic net and fused Laplacian priors for random effects in the Fay-Herriot model. Our model captures spatial patterns and accounts for heterscedasticity of random effects. We demonstrate that the proposed model outperforms existing methods via simulations and a case study.