COMPSTAT 2022: Start Registration
View Submission - COMPSTAT2022
Title: Model structure identification in spatial econometrics Authors:  Maria Lucia Parrella - University of Salerno (Italy) [presenting]
Francesco Giordano - University of Salerno (Italy)
Marcella Niglio - University of Salerno (Italy)
Abstract: Spatio-temporal data is often analyzed by means of \emph{spatial econometric models}. In the last decade, several versions of these models have been proposed, each one based on specific assumptions and different properties of the estimators. We propose a strategy to identify the structure of a spatial econometric model for a given dataset, through a multiple testing procedure that allows choosing between a generalized version of the model and a nested version derived from the general one by imposing restrictions on the parameters. The proposal can be used to test the heterogeneity of the model, but also to test the presence of specific components, such as spatial effects, dynamic effects or external regressors. The theoretical properties of the testing procedure are derived in the high dimensional setup, where the number of spatial units grows to infinity with the sample size. A simulation study and an application to real data give empirical evidence of the testing procedure performance in presence of finite samples.