Abstract Considering local spatial correlation and time series correlation in panel data analysis framework, we propose a series-spatial correlation panel data model. In order to overcome operational difficulties of maximum likelihood estimation method (ML), we propose equal weight GMM and optimal weight GMM methods to estimate variance coefficients and series correlation coefficient, and on this basis, we propose a feasible generalized least squares estimation (FGLS) method to estimate model coefficients, which solves the parameter estimation problem of econometric empirical study under simultaneous spatial and series correlation.
|