Utilisation des matrices de pondérations en économétrie spatiale. Proposition dans un contexte spatio-temporel

This paper aims to bring the reflection about using spatial weight matrix in spatial statistics and tests related to detect spatial autocorrelation as well as in the use of spatial regression model when both, time and space, dimensions are present. The development of a unique spatio-temporal weight matrix is proposed and a simple fictive example is presented. The reality of spatial data collected over time is largely different from a single cross-sectional or panel (pseudo-panel) data structure. The paper proposes a new and simple spatio-temporal approach based on a unique weight matrix.