A Random Matrix Model of Regional Economic Disparity Based on Spatial Panel Data Analysis

This paper adopts a random matrix approach to model spatial panel data and applies the analysis of regional economic disparity. A clustering method based on variable influence response structural relationship is proposed define distance, cluster from so that within class have same relationship, give specific algorithm for determining classification number in process relationship. The methods with presence linear, nonlinear, functional relationships are considered separately aggregate classify terms parameters. Using modeling this paper, empirical better than estimation effect case no classification; test also show there strong cross-sectional correlation between financial development different regions, treatment without treatment. study finds significant imbalance development, phenomenon exclusion serious, dynamic relatively less developed regions regions. In addition influencing factors, we should improve efficiency investment utilization, play geographical advantages, keep industrial structure rationalized, develop special economy according local conditions, full as well development.