Estimation And Application Of The Dynamic Autoregressive Spatial Panel Model With Fixed Effects

This paper consists of two parts: Theory and empirical.In the theoretical part,the paper first studies the asymptotic properties of QML estimator of dynamic SAC panel model when n and T are large under the condition of space system stability.In this paper,one component fixed effects and two components fixed effects models are considered.For the two components fixed effects model,the transformation approach and the direct approach are mentioned to estimate.The consistency and asymptotic normality of QML estimators are verified,and the corresponding bias correction estimators are established.In the Monte Carlo simulation,the paper showes that the biased correction estimators are more effective in finite sample,while ignoring the spatial structure in the disturbance term may lead to inconsistent estimation results and statistical inference failure.Secondly,for the case that the spatial weight is exogenous and time-varying,the 2SLS estimation of the two components fixed effects dynamic SAR panel model is proposed,and the expression of the corresponding asymptotic distribution is derived.Through Monte Carlo simulation,this paper finds that the 2SLS estimator has good finite sample properties,no matter the spatial weight elements are sparse or dense,as long as the time-varying spatial weight is exogenous.In the case that n is larger than T,the finite sample properties of 2SLSEs perform better than that of QMLEs,which is more obvious reflected in the spatial lag coefficient.In the empirical part,the paper establishes a dynamic population growth model including spatial interaction terms,analyzes the influencing factors of China's population growth rate through the theoretical model constructed above,and also explains the application value of the theoretical model.