主办单位：

中国人民大学应用经济学院

主讲嘉宾：

Paul Elhorst

Professor of Spatial Econometrics at the University of Groningen, the Netherlands

参编期刊：

Editor-in-Chief:

Spatial Economic Analysis

Editorial Board member:

Journal of Regional Science

Regional Science & Urban Economics

Journal of Geographical Systems

Geographical Analysis

Journal of Spatial Econometrics

Letters in Spatial and Resource Sciences

会议主持人：中国人民大学应用经济学院 卢昂荻

会议信息：

时间：2021年11月9日 16:00-18:00

线上地点：Zoom 会议926 5003 9057 密码2021

Zoom link:

https://zoom.us/j/92650039057?pwd=V2lMbGxRV3NmWDVVQmRZU0pKWmxuUT09

主讲人简介:

Paul Elhorst is Professor of Spatial Econometrics at the University of Groningen, the Netherlands, and the author of the book “Spatial econometrics: from cross-sectional data to spatial panels” that was published in 2014. This book is also translated in Chinese.

He is Editor-in-Chief of the RSA journal Spatial Economic Analysis, as well as the Editorial Board member of the Journal of Regional Science, Regional Science & Urban Economics, Journal of Geographical Systems, Geographical Analysis, Journal of Spatial Econometrics, and Letters in Spatial and Resource Sciences.

His research interests include spatial econometrics, both theoretical and empirical, software, regional labor market analysis, economic growth, research productivity, military spending, contagion, FDI, fiscal policy interaction and transport economics.

He has written more than 100 papers in refereed journals, both in English and Dutch, and supervised 7 PhD theses. In 2007, Paul Elhorst was awarded the Martin Beckmann Prize for the best paper in Papers in Regional Science.

讲座简介:

Title:

Parameterizing Spatial Weight Matrices in Spatial Econometric Models

Abstract:

Spatial econometric models allow for interactions among units of observations through the specification of spatial weight matrices. Although each variable can have its own spatial weight matrix, practitioners generally adopt one common specification for all of them. This paper breaks this practice by parameterizing the spatial weight matrix with a distance decay parameter and by estimating this parameter for each individual variable. This setup is implemented in a model not only containing spatial interactions in the dependent variable but also in each of the independent variables, which is a prerequisite for obtaining indirect spillover effects that can take any empirical value and are the main focus of empirical studies. We consider and compare the performance of negative exponential and inverse distance decay forms, and of two types of normalization, row-normalization and normalization of the weight matrix by its largest eigenvalue. We adapt the assumptions under which the parameters are identified and their maximum likelihood estimates consistent and asymptotically normal if $N$ is large and $T$ small. We also empirically investigate the asymptotic properties of the proposed estimator in a Monte Carlo simulation experiment and present and discuss the results of an empirical application.

直播观看地址：

https://m.inmuu.com/v1/live/news/1393718