| From the 1970 s,spatial econometrics has been found and rapid developed,and now it is an important branch of econometrics.Different from the traditional econometrics,spatial econometrics considers the spatial correlation.Spatial agglomeration and spatial spillover effects seem to be effective link with independent individuals.If spatial correlation is not taken into account,the model will be biased.Therefore,domestic and foreign experts and scholars in spatial econometrics have widespread concerning and research.This paper focuses on the spatial dynamic panel data model,through the spatial spillover,direct effect and indirect effect and impulse response function,and applies it to practical research.In this paper,the theory of spatial static data model and spatial dynamic panel data model are analyzed.Specification errors and its correction of the general dynamic panel data model are promoted and innovated based on the analysis of space time dynamic model which used the Monte Carlo simulation to analysize the error correction degree of the BCLSDV estimation method in small sample;According to the main characteristics of China's spatial data,the three spatial dynamic panel data model is applied to innovative research.The paper reviews the spatial static data model which includes the spatial cross sectional data model and the spatial static panel data model.The spatial cross section data model is used to analyze the model setting,estimation,spatial correlation,spatial weight matrix and spatial spillover.The spatial static panel data model is used to analyze the classification of the model,the maximum likelihood estimation of the panel data spatial lag regression model and panel data spatial error model,and the Hauseman test.In the analysis of the spatial dynamic panel data model,this paper analyzed theoretically LSDV estimation and error correction of BCLSDV estimation of the time-space simultaneous model and the time-space dynamic model with spatial fixed effects spatial;In the time-space dynamic model,Monte Carlo simulation experiments are analysized small samples of deviation degree of BCLSDV estimation method.Simulation experiments generate two sets of small sample data(T(28)10 ?N(28)36 ? T(28)30 ?N(28)144)and analyze them.Through the simulation experiment,it is found that the two sets of experimental results are very similar: through the Bias index,we can see that after using the BCLSDV method to estimate,the parameters can be corrected very well.However,the RMSE index can be seen using the BCLSDV method of the correction effect is mainly reflected in the time lag coefficients,and for the space-time lag coefficient,exogenous variable coefficient and spatial lag coefficient,effect of correction is not obvious.Similar conclusions are obtained even when replacing the weight matrix.Compared T(28)30 ?N(28)144 with T(28)10 ? N(28)36,the coefficients of Bias and RMSE are smaller when T(28)30 ? N(28)144,it shows that the error decreases as the sample capacity increases.The paper establishes a general spatial panel data model and analyzes the impact of R&D,industrial agglomeration on China's technological innovation.In this paper,the BCLSDV method is used to estimate the model.Model results indicate the technological innovation have the time lag effect,spatial spillover effect and the lagging of the spatial spillover effects,and R&D has significant influence on regional technological innovation,but industrial agglomeration effects on regional technological innovation is not significant.The indirect effect of R&D and industrial agglomeration on technological innovation is greater than the direct effect,and has a positive impact.Based on the analysis of the spatial dynamic panel data model theory,aiming at the main characteristics of China's spatial data,the spatial dynamic panel data model and spatial panel GVAR model and spatial panel Sp VAR model are used to research on the application of innovation.First,extended Griliches-Jaffe production function,including human capital and industrial agglomeration and other factors,the knowledge production function is rewritten to analyze influence of 2000-2014 R&D,industrial agglomeration on technology innovation in china.By the method of BCLSDV,Spatial dynamic panel data model with spatial fixed effect,space and time fixed effects,and first-order differential panel data model are estimated.The spatial spillover effect of China's technological innovation has been decomposited.The model results show that the technical innovation has time lag effect,spatial spillover effect and the lag of spatial spillover effects;R&D has significant influence on regional technological innovation;The effect of industrial agglomeration on regional technology innovation is not significant;The indirect effect of the R&D and industrial agglomeration on technological innovation is greater than the direct effect,and have a positive impact.Second,the initial spatial dynamic panel GVAR model is mainly used in international economic relations.The space connection matrix can be expressed by the distance between countries,economic and trade links.This paper applies GVAR model to the industry research in China for the first time.It connects all industries through input-output table,which is also the innovation of this paper.By establishing the dynamic panel GVAR model,this paper analyzes the impact of international trade and technology shocks on the labor market of China's manufacturing industry.The empirical results show that in the short term,for specific industry,the positive impact of trade and technology will make the industry average wage increased.The average wage of capital and technology intensive industries increases greatly,the smaller the average wage increase of labor intensive industry.For specific industry,the positive impact of trade and technology will make labor-intensive industries per capita assets decreased,such as the food processing industry,textile industry,textile and garment industry and wood processing industry,while the oil processing industry,non-metallic mineral products industry,metal products industry,general equipment manufacturing industry,transportation industry,instrument industry and other capital intensive industry per capita assets increased.The impact of industry specific technology shocks on average wages and per capita assets is longer than the impact of trade shocks.Third,the spatial dynamic panel Sp VAR model is evolved by GVAR model,this paper analyzes differential GMM estimation of the spatial dynamic panel Sp VAR model.The results of Monte Carlo simulation program show that the differential GMM estimators are still biased,but when the samples are from T=10 N=36,to T=20,N=64,and then T=30,N=81,Bias and RMSE index are smaller,indicating differential GMM estimation error with the increase of the sample will be more and more small.Since Sp VAR is an unstructured model,this paper only applies it to panel unit root,panel cointegration,Grainger causality test and impulse response analysis.Through the establishment of spatial dynamic panel Sp VAR model,the paper analyzes the relationship between R&D investment and technological innovation.According to Beijing,Shanghai,Jiangsu,Zhejiang,Guangdong and Hubei,the impulse response function analysis,draw the conclusion: the positive impact of technological innovation variables,will have a positive impact on local technological innovation and have a negative impact on innovation variables in other regions;the positive impact of technological innovation on local R & D investment variables have a positive effect in the first period,but from the beginning of second period,all parts of the R & D investment variables have a negative impact,to finally converge to 0;The positive impact of R & D investment variables will have a positive effect on local R & D investment variables and technology innovation variables.In this paper,the differential GMM estimation of the dynamic panel Sp VAR is applied to the research of our country,and the model and method are seldom applied in the published literatures.In summary,this paper introduces the spatial cross section data model,spatial static panel data model,spatial dynamic panel data model and estimation method.It analyses the spatial dynamic panel data model in the setting of errors and correction,and through Monte Carlo simulation experiments show that the correction effect;and the general spatial dynamic panel the data model and the spatial panel GVAR model and spatial panel Sp VAR model is used to analyze the reality of our country economy,through direct and indirect effects,and impulse response function to reflect the spatial spillover features,which is value of the spatial econometrics difference from the traditional econometrics. |
PDF Full Text Request