六步学会用MATLAB做空间计量回归详细步骤.docx

1、六步学会用MATLAB做空间计量回归详细步骤1.excel与MATLAB链接：Excel：选项加载项COM加载项转到没有勾选项 2. MATLAB安装目录中寻找toolboxexlink点击,启用宏 E:MATLABtoolboxexlink然后，Excel中就出现MATLAB工具（注意Excel中的数据：）3.启动matlab（1） 点击start MATLAB（2） senddata to matlab ，并对变量矩阵变量进行命名（注意：选取变量为数值，不包括各变量）(data表中数据进行命名)（空间权重进行命名）（3） 导入MATLAB中的两个矩阵变量就可以看见4.将elhorst和jp

2、lv7两个程序文件夹复制到MATLAB安装目录的toolbox文件夹5.设置路径：6.输入程序，得出结果T=30; N=46; W=normw(W1); y=A(:,3); x=A(:,4,6); xconstant=ones(N*T,1);nobs K=size(x);results=ols(y,xconstant x);vnames=strvcat(logcit,intercept,logp,logy);prt_reg(results,vnames,1);sige=results.sige*(nobs-K)/nobs);loglikols=-nobs/2*log(2*pi*sige)-1/

3、(2*sige)*results.resid*results.resid% The (robust)LM tests developed by ElhorstLMsarsem_panel(results,W,y,xconstant x); % (Robust) LM tests解释每一行分别表示：该面板数据的时期数为30（T=30），该面板数据有30个地区（N=30），将空间权重矩阵标准化（W=normw(w1)），将名为A（以矩阵形式出现在MATLABA中）的变量的第3列数据定义为被解释变量y,将名为A的变量的第4、5、6列数据定义为解释变量矩阵x，定义一个有N*T行，1列的全1矩阵，该矩阵

4、名为：xconstant，（ones即为全1矩阵）说明解释变量矩阵x的大小：有nobs行，K列。（size为描述矩阵的大小）。附录：静态面板空间计量经济学一、OLS静态面板编程1、普通面板编程T=30; N=46; W=normw(W1); y=A(:,3); x=A(:,4,6); xconstant=ones(N*T,1);nobs K=size(x);results=ols(y,xconstant x);vnames=strvcat(logcit,intercept,logp,logy);prt_reg(results,vnames,1);sige=results.sige*(nobs-

5、K)/nobs);loglikols=-nobs/2*log(2*pi*sige)-1/(2*sige)*results.resid*results.resid% The (robust)LM tests developed by ElhorstLMsarsem_panel(results,W,y,xconstant x); % (Robust) LM tests2、空间固定OLS （spatial-fixed effects）T=30; N=46; W=normw(W1); y=A(:,3); x=A(:,4,6); xconstant=ones(N*T,1);nobs K=size(x);

6、model=1;ywith,xwith,meanny,meannx,meanty,meantx=demean(y,x,N,T,model);results=ols(ywith,xwith);vnames=strvcat(logcit,logp,logy); % should be changed if x is changedprt_reg(results,vnames);sfe=meanny-meannx*results.beta; % including the constant termyme = y - mean(y);et=ones(T,1);error=y-kron(et,sfe)

7、-x*results.beta;rsqr1 = error*error;rsqr2 = yme*yme;FE_rsqr2 = 1.0 - rsqr1/rsqr2 % r-squared including fixed effectssige=results.sige*(nobs-K)/nobs);logliksfe=-nobs/2*log(2*pi*sige)-1/(2*sige)*results.resid*results.residLMsarsem_panel(results,W,ywith,xwith); % (Robust) LM tests3、时期固定OLS（time-period

8、fixed effects）T=30; N=46; W=normw(W1); y=A(:,3); x=A(:,4,6); xconstant=ones(N*T,1);nobs K=size(x);model=2;ywith,xwith,meanny,meannx,meanty,meantx=demean(y,x,N,T,model);results=ols(ywith,xwith);vnames=strvcat(logcit,logp,logy); % should be changed if x is changedprt_reg(results,vnames);tfe=meanty-mea

9、ntx*results.beta; % including the constant termyme = y - mean(y);en=ones(N,1);error=y-kron(tfe,en)-x*results.beta;rsqr1 = error*error;rsqr2 = yme*yme;FE_rsqr2 = 1.0 - rsqr1/rsqr2 % r-squared including fixed effectssige=results.sige*(nobs-K)/nobs);logliktfe=-nobs/2*log(2*pi*sige)-1/(2*sige)*results.r

10、esid*results.residLMsarsem_panel(results,W,ywith,xwith); % (Robust) LM tests4、空间与时间双固定模型T=30; N=46; W=normw(W1); y=A(:,3); x=A(:,4,6); xconstant=ones(N*T,1);nobs K=size(x);model=3;ywith,xwith,meanny,meannx,meanty,meantx=demean(y,x,N,T,model);results=ols(ywith,xwith);vnames=strvcat(logcit,logp,logy);

11、 % should be changed if x is changedprt_reg(results,vnames)en=ones(N,1);et=ones(T,1);intercept=mean(y)-mean(x)*results.beta; sfe=meanny-meannx*results.beta-kron(en,intercept);tfe=meanty-meantx*results.beta-kron(et,intercept);yme = y - mean(y);ent=ones(N*T,1);error=y-kron(tfe,en)-kron(et,sfe)-x*resul

12、ts.beta-kron(ent,intercept);rsqr1 = error*error;rsqr2 = yme*yme;FE_rsqr2 = 1.0 - rsqr1/rsqr2 % r-squared including fixed effectssige=results.sige*(nobs-K)/nobs);loglikstfe=-nobs/2*log(2*pi*sige)-1/(2*sige)*results.resid*results.residLMsarsem_panel(results,W,ywith,xwith); % (Robust) LM tests二、静态面板SAR

13、模型1、无固定效应（No fixed effects）T=30; N=46; W=normw(W1);y=A(:,3);x=A(:,4,6); for t=1:T t1=(t-1)*N+1;t2=t*N; wx(t1:t2,:)=W*x(t1:t2,:);endxconstant=ones(N*T,1);nobs K=size(x);info.lflag=0; info.model=0;info.fe=0; results=sar_panel_FE(y,xconstant x,W,T,info); vnames=strvcat(logcit,intercept,logp,logy);prt_s

14、pnew(results,vnames,1)% Print out effects estimatesspat_model=0;direct_indirect_effects_estimates(results,W,spat_model);panel_effects_sar(results,vnames,W);2、空间固定效应（Spatial fixed effects）T=30; N=46; W=normw(W1);y=A(:,3);x=A(:,4,6); for t=1:T t1=(t-1)*N+1;t2=t*N; wx(t1:t2,:)=W*x(t1:t2,:);endxconstant

15、=ones(N*T,1);nobs K=size(x);info.lflag=0;info.model=1;info.fe=0; results=sar_panel_FE(y,x,W,T,info); vnames=strvcat(logcit,logp,logy);prt_spnew(results,vnames,1)% Print out effects estimatesspat_model=0;direct_indirect_effects_estimates(results,W,spat_model);panel_effects_sar(results,vnames,W);3、时点固

16、定效应（Time period fixed effects）T=30; N=46; W=normw(W1);y=A(:,3);x=A(:,4,6); for t=1:T t1=(t-1)*N+1;t2=t*N; wx(t1:t2,:)=W*x(t1:t2,:);endxconstant=ones(N*T,1);nobs K=size(x);info.lflag=0; % required for exact resultsinfo.model=2;info.fe=0; % Do not print intercept and fixed effects; use info.fe=1 to tu

17、rn onresults=sar_panel_FE(y,x,W,T,info); vnames=strvcat(logcit,logp,logy);prt_spnew(results,vnames,1)% Print out effects estimatesspat_model=0;direct_indirect_effects_estimates(results,W,spat_model);panel_effects_sar(results,vnames,W);4、双固定效应（Spatial and time period fixed effects）T=30; N=46; W=normw

18、(W1);y=A(:,3);x=A(:,4,6); for t=1:T t1=(t-1)*N+1;t2=t*N; wx(t1:t2,:)=W*x(t1:t2,:);endxconstant=ones(N*T,1);nobs K=size(x);info.lflag=0; % required for exact resultsinfo.model=3;info.fe=0; % Do not print intercept and fixed effects; use info.fe=1 to turn onresults=sar_panel_FE(y,x,W,T,info); vnames=s

19、trvcat(logcit,logp,logy);prt_spnew(results,vnames,1)% Print out effects estimatesspat_model=0;direct_indirect_effects_estimates(results,W,spat_model);panel_effects_sar(results,vnames,W);三、静态面板SDM模型1、无固定效应（No fixed effects）T=30; N=46; W=normw(W1);y=A(:,3);x=A(:,4,6); for t=1:T t1=(t-1)*N+1;t2=t*N; wx

20、(t1:t2,:)=W*x(t1:t2,:);endxconstant=ones(N*T,1);nobs K=size(x);info.lflag=0; info.model=0;info.fe=0; results=sar_panel_FE(y,xconstant x wx,W,T,info); vnames=strvcat(logcit,intercept,logp,logy,W*logp,W*logy);prt_spnew(results,vnames,1)% Print out effects estimatesspat_model=1;direct_indirect_effects_

21、estimates(results,W,spat_model);panel_effects_sdm(results,vnames,W);2、空间固定效应（Spatial fixed effects）T=30; N=46; W=normw(W1);y=A(:,3);x=A(:,4,6); for t=1:T t1=(t-1)*N+1;t2=t*N; wx(t1:t2,:)=W*x(t1:t2,:);endxconstant=ones(N*T,1);nobs K=size(x);info.lflag=0; % required for exact resultsinfo.model=1;info.

22、fe=0; % Do not print intercept and fixed effects; use info.fe=1 to turn onresults=sar_panel_FE(y,x wx,W,T,info); vnames=strvcat(logcit,logp,logy,W*logp,W*logy);prt_spnew(results,vnames,1)% Print out effects estimatesspat_model=1;direct_indirect_effects_estimates(results,W,spat_model);panel_effects_s

23、dm(results,vnames,W);3、时点固定效应（Time period fixed effects）T=30; N=46; W=normw(W1);y=A(:,3);x=A(:,4,6); for t=1:T t1=(t-1)*N+1;t2=t*N; wx(t1:t2,:)=W*x(t1:t2,:);endxconstant=ones(N*T,1);nobs K=size(x);info.lflag=0; % required for exact resultsinfo.model=2;info.fe=0; % Do not print intercept and fixed ef

24、fects; use info.fe=1 to turn on% New routines to calculate effects estimatesresults=sar_panel_FE(y,x wx,W,T,info); vnames=strvcat(logcit,logp,logy,W*logp,W*logy);% Print out coefficient estimatesprt_spnew(results,vnames,1)% Print out effects estimatesspat_model=1;direct_indirect_effects_estimates(re

25、sults,W,spat_model);panel_effects_sdm(results,vnames,W)4、双固定效应（Spatial and time period fixed effects）T=30; N=46; W=normw(W1);y=A(:,3);x=A(:,4,6); for t=1:T t1=(t-1)*N+1;t2=t*N; wx(t1:t2,:)=W*x(t1:t2,:);endxconstant=ones(N*T,1);nobs K=size(x);info.bc=0;info.lflag=0; % required for exact resultsinfo.m

26、odel=3;info.fe=0; % Do not print intercept and fixed effects; use info.fe=1 to turn onresults=sar_panel_FE(y,x wx,W,T,info); vnames=strvcat(logcit,logp,logy,W*logp,W*logy);prt_spnew(results,vnames,1)% Print out effects estimatesspat_model=1;direct_indirect_effects_estimates(results,W,spat_model);pan

27、el_effects_sdm(results,vnames,W)wald test spatial lag% Wald test for spatial Durbin model against spatial lag modelbtemp=results.parm;varcov=results.cov;Rafg=zeros(K,2*K+2);for k=1:K Rafg(k,K+k)=1; % R(1,3)=0 and R(2,4)=0;endWald_spatial_lag=(Rafg*btemp)*inv(Rafg*varcov*Rafg)*Rafg*btempprob_spatial_

28、lag=1-chis_cdf (Wald_spatial_lag, K)wald test spatial error% Wald test spatial Durbin model against spatial error modelR=zeros(K,1);for k=1:K R(k)=btemp(2*K+1)*btemp(k)+btemp(K+k); % k changed in 1, 7/12/2010% R(1)=btemp(5)*btemp(1)+btemp(3);% R(2)=btemp(5)*btemp(2)+btemp(4);endRafg=zeros(K,2*K+2);f

29、or k=1:K Rafg(k,k) =btemp(2*K+1); % k changed in 1, 7/12/2010 Rafg(k,K+k) =1; Rafg(k,2*K+1)=btemp(k);% Rafg(1,1)=btemp(5);Rafg(1,3)=1;Rafg(1,5)=btemp(1);% Rafg(2,2)=btemp(5);Rafg(2,4)=1;Rafg(2,5)=btemp(2);end Wald_spatial_error=R*inv(Rafg*varcov*Rafg)*Rprob_spatial_error=1-chis_cdf (Wald_spatial_err

30、or,K)LR test spatial lagresultssar=sar_panel_FE(y,x,W,T,info); LR_spatial_lag=-2*(resultssar.lik-results.lik)prob_spatial_lag=1-chis_cdf (LR_spatial_lag,K) LR test spatial errorresultssem=sem_panel_FE(y,x,W,T,info); LR_spatial_error=-2*(resultssem.lik-results.lik)prob_spatial_error=1-chis_cdf (LR_spatial_error,K) 5、空间随机效应与时点固定效应模型T=30;