Spatial autoregressive models and identification by Robert de Jong

Join us for this week's Econometrics Research Seminar, Spring Term 2026

Robert de Jong, from the Ohio State University, will present this week's seminar on Spatial autoregressive models and identification.

Abstract

This paper considers identification of general parametrized spatial autoregressive models in a spatial cross-section setting. The goal is to investigate the exact conditions needed for a parametrization of the first order spatial weight matrix to yield an identified parameter. Using an analysis of the curvature of the quasiloglikelihood function with respect to a certain spatial matrix, I arrive at various criteria for identification in general spatial models. Because the spatial weight matrix is of order $(n \times n)$ and therefore contains a number of elements that increases as the square of sample size, this is an unusual route for showing identification and presents a new methodological approach to the issue of identification in spatial models. An essential ingredient for obtaining the results of this paper is noting that the quasiloglikelihood is concave with respect to a certain $(n \times n)$ matrix. I will find improved conditions for identification for various spatial models and I demonstrate that for identification in spatial models, the usual row and column sum conditions for weight matrices are not necessary.

This seminar will be held on campus, is open to all levels of study and is also open to the public. To register your place and gain access to the webinar, please contact the seminar organisers.

This event is part of the Econometrics Research Seminar Series.