1、Creates centroid variables*genid(newvarname):Creates unique id variable for database.dtashp2dta using CHN_adm1,database (chinaprovince) coordinates(coord) genid(id) gencentroids(c)*绘制2016年中國GDP分布圖*spmap:Visualization of spatial data*clnumber(#):number of classes*id(idvar):base map polygon identifier
2、(识别符,声明变量名,一般以字母或下划线开头,包含数字、字母、下划线)*_2016GDP:变量*coord:之前创建的坐标系数据集spmap _2016GDP using coord, id(id) clnumber(5)*更改变量名rename x_c longituderename y_c latitude*生成距离矩阵*spmat:用于定义与管理空间权重矩阵*Spatial-weighting matrices are stored in spatial-weighting matrix objects (spmat objects). *spmat objects contain ad
3、ditional information about the data used in constructing spatial-weighting matrices. *spmat objects are used in fitting spatial models; see spreg (if installed) and spivreg (if installed).*idistance:(产生距离矩阵)create an spmat object containing an inverse-distance matrix W *或 contiguity:create an spmat
4、object containing a contiguity matrix W*idistance_jingdu:命名名称为“idistance_jingdu”的距離矩陣*longitude:使用经度*latitude:使用纬度*id(id):使用id*dfunction(function, miles):(设置计算距离方法)specify the distance function. *function may be one of euclidean (default), dhaversine, rhaversine, or the Minkowski distance of order p
5、, where p is an integer greater than or equal to 1.*normalize(row):(行标准化)specifies one of the three available normalization techniques: row, minmax, and spectral. *In a row-normalized matrix, each element in row i is divided by the sum of row is elements. *In a minmax-normalized matrix, each element
6、 is divided by the minimum of the largest row sum and column sum of the matrix. *In a spectral-normalized matrix, each element is divided by the modulus of the largest eigenvalue of the matrix.spmat idistance idistance_jingdu longitude latitude, id(id) dfunction(euclidean) normalize(row) *保存stata可读文
7、件idistance_jingdu.spmatspmat save idistance_jingdu using idistance_jingdu.spmat*将刚刚保存的idistance_jingdu.spmat文件转化为txt文件spmat export idistance_jingdu using idistance_jingdu.txt *生成相邻矩阵spmat contiguity contiguity_jingdu using coord, id(id) normalize(row)spmat save contiguity_jingdu using contiguity_jin
8、gdu.spmatspmat export contiguity_jingdu using contiguity_jingdu.txt*计算Morans I*安装spatwmat*spatwmat:用于定义空间权重矩阵imports or generates the spatial weights matrices required by spatgsa, spatlsa, spatdiag, and spatreg. *As an option, spatwmat also generates the eigenvalues matrix required by spatreg.*name(
9、W):读取空间权重矩阵W使用生成的空间权重矩阵W*xcoord:x坐标*ycoord:y坐标*band(0 8):宽窗介绍 *band(numlist) is required if option using filename is not specified. *It specifies the lower and upper bounds of the distance band within which location pairs must be considered neighbors (i.e., spatially contiguous) *and, therefore, ass
10、igned a nonzero spatial weight.*binary:requests that a binary weights matrix be generated. To this aim, all nonzero spatial weights are set to 1.spatwmat, name(W) xcoord(longitude) ycoord(latitude) band(0 8)*安装绘制Morans I工具:splagvar*splagvar - Generates spatially lagged variables, constructs the Mora
11、n scatter plot, *and calculates global Morans I statistics.使用变量_2016GDP*wname(W):使用空间权重矩阵W *indicate the name of the spatial weights matrix to be used*wfrom(Stata):indicate source of the spatial weights matrix *wfrom(Stata | Mata) indicates whether the spatial weights matrix is a Stata matrix loaded
12、 in memory or a Mata file located in the working directory. *If the spatial weights matrix had been created using spwmatrix it should exist as a Stata matrix or as a Mata file.*moran(_2016GDP):计算变量_2016GDP的Morans I值*plot(_2016GDP):构建变量_2016GDPMoran散点图splagvar _2016GDP, wname(W) wfrom(Stata) moran(_2
13、016GDP) plot(_2016GDP)*使用距离矩阵计算空间计量模型cd D:软件学习软件资料statastata指导书籍命令陈强高级计量经济学及stata应用(第二版)全部数据*使用product.dta数据集(陈强的高级计量经济学及其stata应用P594)*将数据集product.dta存入当前工作路径use product.dta , clear*创建新变量,对原有部分变量取对数gen lngsp=log(gsp)gen lnpcap=log(pcap)gen lnpc=log(pc)gen lnemp=log(emp)*将空间权重矩阵usaww.spat存入当前工作路径spma
14、t use usaww using usaww.spmat*使用聚类稳健的标准误估计随机效应的SDM模型xsmle lngsp lnpcap lnpc lnemp unemp,wmat(usaww) model(sdm)robust nolog*使用选择项durbin(lnemp),不选择不显著的变量,使用聚类稳健的标准误估计随机效应的SDM模型xsmle lngsp lnpcap lnpc lnemp unemp,wmat(usaww) model(sdm) durbin(lnemp) robust nolog noeffects*使用选择项durbin(lnemp),不选择不显著的变量,使
15、用聚类稳健的标准误估计固定效应的SDM模型xsmle lngsp lnpcap lnpc lnemp unemp,wmat(usaww) model(sdm) durbin(lnemp) robust nolog noeffects fe*存储随机效应和固定效应结果qui xsmle lngsp lnpcap lnpc lnemp unemp,wmat(usaww) model(sdm) durbin(lnemp) r2 nolog noeffects reest sto requi xsmle lngsp lnpcap lnpc lnemp unemp,wmat(usaww) model(s
16、dm) durbin(lnemp) r2 nolog noeffects feest sto fe*esttab:将保存的结果汇总到一张表格中 *b(fmt):specify format for point estimates *beta(fmt):display beta coefficients instead of point ests *se(fmt):display standard errors instead of t statistics *star( * 0.1 * 0.05 * 0.01):标记不同显著性水平对应的P值 *r2|ar2|pr2(fmt):display (
17、adjusted, pseudo) R-squared *p(fmt):display p-values instead of t statistics *label:make use of variable labels *title(string):specify a title for the tableesttab fe re , b se r2 star( * 0.1 * 0.05 * 0.01)*hausman检验 *进行hausman检验前,回归中没有使用稳健标准误(没用“r”), *是因为传统的豪斯曼检验建立在同方差的前提下 *constant:include estimate
18、d intercepts in comparison; default is to exclude *df(#):use # degrees of freedom *sigmamore:base both (co)variance matrices on disturbance variance estimate from efficient estimator *sigmaless:base both (co)variance matrices on disturbance variance estimate from consistent estimatorhausman fe re *有
19、时我们还会得到负的chi2值,即chi20,表明模型不能满足Hausman检验的渐近假设。产生这些情况的原因可能有多种, *但我认为一个主要的原因是我们的模型设定有问题,导致Hausman 检验的基本假设得不到满足。 *这时,我们最好先对模型的设定进行分析,看看是否有遗漏变量的问题,或者某些变量是非平稳的等等。 *在确定模型的设定没有问题的情况下再进行Hausman 检验,如果仍然拒绝原假设或是出现上面的问题, *那么我们就认为随机效应模型的基本假设(个体效应与解释变量不相关)得不到满足。 *此时,需要采用工具变量法或是使用固定效应模型。 *连玉君(论文(2014):Hausman检验统计量有
20、效性的Monte Carlo模拟分析) *研究了hausman检验统计量的小样本性质,结果表明, *内生性问题(解释变量与个体效应相关)是导致hausman统计量出现负值的主要原因, *进一步分析表明,修正后的hausman统计量,以及过度识别检验方法能够很好地克服上述缺陷, *且具有很好的有限样本性质。 *陈强(高级计量经济学及其stata应用P153)介绍工具变量法与豪斯曼的stata命令及实例qui reg lw iq s expr tenure rns smsaest sto olsqui ivregress 2sls lw s expr tenure rns smsa (iq=med
21、 kww)est sto ivhausman iv o ls, constant sigmamore*由于传统的豪斯曼检验在异方差的情形下不成立,故进行异方差稳健的DWH检验,estat endogenous *使用ivreg2检验选择的工具变量是否为内生解释变量ivreg2 lw s expr tenure rns smsa (iq=med kww),r endog(iq) *endog(iq)表示检验变量iq是否为内生变量*若果存在异方差,则GMM比2SLS更有效率,故进行最优GMM估计ivregress gmm lw s expr tenure rns smsa (iq=med kww)*进行过度识别检验estat overid *若P值不显著,则认为所有工具变量均为外生。*接下来考虑迭代GMMivregress gmm lw s expr tenure rns smsa (iq=med kww),igmm *(迭代GMM与两步GMM的系数估计值相差无几)
copyright@ 2008-2023 冰点文库 网站版权所有
经营许可证编号:鄂ICP备19020893号-2