Asymptotic Inference and Testing in Two-Dimensional Random Coefficient Autoregressive Models

Read full text Summarise with VIRAT
Authors: Ouerdia Arezki
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

We develop hypothesis tests for two-dimensional random coefficient autoregressive models 2D-RCAR(p 1 , p 2 ). The main goal is to assess whether the autoregressive coefficients exhibit random spatial variability around fixed means, i.e., to test H 0 : Λ = 0, where Λ is the covariance matrix of the coefficient perturbations. Because H 0 lies on the boundary of the parameter space defined by Λ ⪰ 0, standard likelihood ratio and Wald procedures may fail. We propose (i) a score-type test evaluated under H 0 with a standard χ 2 limit, and (ii) a boundary-aware statistic built from residual-based estimation of Λ, leading to a chi-square mixture limit. Under local alternatives with Λ positive definite, the second procedure yields higher asymptotic power. Monte Carlo experiments illustrate finite-sample size and power.

Fields of Study & Subjects

Spatial and Panel Data Analysis Statistical Methods and Inference Advanced Statistical Methods and Models

Publication details

Journal / source
HAL
Publication date
27 March 2026
Language
EN
Document type
Preprint