互助问答第75期：动态空间计量模型的估计问题

今日问题1

 您好！使用动态空间计量模型对问题进行估计时，同时使用的是矩估计的方法，如何实现那个AR（1）和AR(2)的检验呢？相应的代码是什么？

今日解答1

AR(1)，AR(2)是用来检验动态面板，如差分或系统GMM是否存在扰动项一阶和二阶自相关的。动态空间计量模型的使用未见推广的这些检验。空间计量中有SAR模型，即空间自相关(回归)模型。动态空间计量，动态空间面板模型的构建与检验仍然处在前沿的研究阶段，Elhost（2010）对Arrelano和bond的差分GMM估计量进行扩展，使其包括一个内生交互效应，且同时发现这种估计量仍然可能存在严重的偏误，特别是Wyt的参数的估计。

Kukenova andMonteiro（2009）、Jacobs et al（2009）同时研究了动态面板数据模型（Yt-1，Wyt-1），并扩展了Blundell and Bond（1998）的系统GMM估计，使之包含内生交互效应（Wyt），前者同时研究了内生解释变量Zt，后者研究了误差项的空间自相关Wt。他们发现系统GMM估计量要优于差分GMM，且减小了参数估计。

动态空间面板模型的stata程序如下：
clear *清空
ssc install spregdpd(or spregdhp) *安装spregdpd或spregdhp动态空间面板模板
help spregdpd(or spregdhp)  *参阅帮助文件（有案例） 

help输出：

Title+-------------------------------------------------------------

spregdpd: Spatial Panel Arellano-Bond Linear Dynamic Regression:Lag & Durbin Mo

> dels

Syntax+------------------------------------------------------------

     spregdpd depvarindepvars [weight] , nc(#) wmfile(weight_file)

    [ model(sar|sdm)run(xtabond|xtdhp|xtdpd|xtdpdsys) be fe re

     lmspac lmhet lmnormdiag tests stand inv inv2 mfx(lin, log)

     noconstant

      predict(new_var)resid(new_var) inst(vars) diff(vars) endog(vars)

     pre(vars)

      dgmmiv(varlist) collzero tolog twostep level(#) vce(vcetype) ]

Model Options +-----------------------------------------------------

     1- model(sar)  MLE Spatial Panel Lag Model (SAR)

     2- model(sdm)  MLE Spatial Panel Durbin Model (SDM)

Run Options+-------------------------------------------------------

1- run(xtdhp)      [NEW]Han-Philips (2010) Linear Dynamic Panel Regression

2- run(xtabond)    [xtabond]Arellano-Bond Linear Dynamic Panel Regression

3- run(xtdpd)      [xtdpd]Arellano-Bond (1991) Linear Dynamic Panel Regression

4- run(xtdpdsys)  [xtdpdsys] Arellano-Bover/Blundell-Bond (1995, 1998)

                             System Linear Dynamic Panel Regression

Options+-----------------------------------------------------------

  *  nc(#)         Number of Cross Sections Units

   Time series observationsmust be Balanced in each Cross Section

 

  wmfile(weight_file) OpenCROSS SECTION weight matrix file.

…..

tests            display ALLlmh, lmn, lmsp, diag tests

 twostep          two-step estimate run(xtdpd)

  dgmmiv(varlist)         GMM Instruments for DifferencedEquation run(xtdpd)

  inst(varlist)           Additional Instrumental Variablesrun(xtabond)

                      Dependent Variable Lag length is lag(1)

  diff(varlist)           Already Differenced ExogenousVariables run(xtabond)

  endog(varlist)          Endogenous Variables run(xtabond,xtdpdsys)

  pre(varlist)            Predetermined Variablesrun(xtabond, xtdpdsys)

    vce(vcetype) ols,robust, cluster, bootstrap, jackknife, hc2, hc3

  level(#)         confidence intervals level; default islevel(95)

Spatial Panel Aautocorrelation Tests+------------------------------

lmspac Spatial Panel Aautocorrelation Tests:

        * Ho: Error has NoSpatial AutoCorrelation

          Ha: Error has    Spatial AutoCorrelation
*** Spatial Econometrics Regression Models:
--------------------------------------------------------------------------------
*** (1) Spatial Panel Data Regression Models:
spregxt      Spatial Panel Regression Econometric Models:Stata Module Toolkit
gs2slsxt     Generalized Spatial Panel 2SLS Regression
gs2slsarxt   Generalized Spatial Panel Autoregressive 2SLS Regression
spglsxt      Spatial Panel Autoregressive Generalized LeastSquares Regression
spgmmxt      Spatial Panel Autoregressive GeneralizedMethod of Moments Regression
spmstarxt    (m-STAR) Spatial Lag Panel Models
spmstardxt   (m-STAR) Spatial Durbin Panel Models
spmstardhxt  (m-STAR) Spatial Durbin MultiplicativeHeteroscedasticity Panel Models
spmstarhxt   (m-STAR) Spatial Lag Multiplicative HeteroscedasticityPanel Models
spregdhp     Spatial Panel Han-Philips Linear DynamicRegression: Lag & Durbin Models
spregdpd     Spatial Panel Arellano-Bond Linear DynamicRegression: Lag & Durbin Models
spregfext    Spatial Panel Fixed Effects Regression: Lag & DurbinModels
spregrext    Spatial Panel Random Effects Regression: Lag &Durbin Models
spregsacxt   MLE Spatial AutoCorrelation Panel Regression (SAC)
spregsarxt   MLE Spatial Lag Panel Regression (SAR)
spregsdmxt   MLE Spatial Durbin Panel Regression (SDM)
spregsemxt   MLE Spatial Error Panel Regression (SEM)
--------------------------------------------------------------------------------
*** (2) Spatial Cross Section Regression Models:
spregcs      Spatial Cross Section Regression EconometricModels: Stata Module Toolkit
gs2sls       Generalized Spatial 2SLS Cross SectionsRegression
gs2slsar     Generalized Spatial Autoregressive 2SLS CrossSections Regression
gs3sls       Generalized Spatial 3SLS Cross SectionsRegression
gs3slsar     Generalized Spatial Autoregressive 3SLS CrossSections Regression
spautoreg    Spatial Cross Section Regression Models
spgmm        Spatial Autoregressive GMM CrossSections Regression
spmstar      (m-STAR) Spatial Lag Cross Sections Models
spmstard     (m-STAR) Spatial Durbin Cross Sections Models
spmstardh    (m-STAR) Spatial Durbin MultiplicativeHeteroscedasticity Cross Sections Models
spmstarh     (m-STAR) Spatial Lag MultiplicativeHeteroscedasticity Cross Sections Models
spregsac     MLE Spatial AutoCorrelation Cross SectionsRegression (SAC)
spregsar     MLE Spatial Lag Cross Sections Regression(SAR)
spregsdm     MLE Spatial Durbin Cross Sections Regression(SDM)
spregsem     MLE Spatial Error Cross Sections Regression(SEM)
--------------------------------------------------------------------------------
*** (3) Tobit Spatial Regression Models:
*** (3-1) Tobit Spatial Panel Data Regression Models:
sptobitgmmxt Tobit Spatial GMM Panel Regression
sptobitmstarxtTobit (m-STAR) Spatial Lag Panel Models
sptobitmstardxtTobit (m-STAR) Spatial Durbin Panel Models
sptobitmstardhxtTobit (m-STAR) Spatial Durbin Multiplicative HeteroscedasticityPanel Models
sptobitmstarhxtTobit (m-STAR) Spatial Lag Multiplicative HeteroscedasticityPanel Models
sptobitsacxt Tobit MLE Spatial AutoCorrelation (SAC) Panel Regression
sptobitsarxt Tobit MLE Spatial Lag Panel Regression
sptobitsdmxt Tobit MLE Spatial Panel Durbin Regression
sptobitsemxt Tobit MLE Spatial Error Panel Regression
spxttobit    Tobit Spatial Panel Autoregressive GLS Regression
--------------------------------------------------------------
*** (3-2) Tobit Spatial Cross Section Regression Models:
sptobitgmm   Tobit Spatial GMM Cross Sections Regression
sptobitmstar Tobit (m-STAR) Spatial Lag Cross Sections Models
sptobitmstardTobit (m-STAR) Spatial Durbin Cross Sections Models
sptobitmstardhTobit (m-STAR) Spatial Durbin Multiplicative HeteroscedasticityCross Sections
sptobitmstarhTobit (m-STAR) Spatial Lag Multiplicative Heteroscedasticity CrossSections
sptobitsac   Tobit MLE AutoCorrelation (SAC) Cross Sections Regression
sptobitsar   Tobit MLE Spatial Lag Cross Sections Regression
sptobitsdm   Tobit MLE Spatial Durbin Cross Sections Regression
sptobitsem   Tobit MLE Spatial Error Cross Sections Regression
--------------------------------------------------------------------------------
*** (4) Spatial Weight Matrix:
spcs2xt      Convert Cross Section to Panel Spatial WeightMatrix
spweight     Cross Section and Panel Spatial Weight Matrix
spweightcs   Cross Section Spatial Weight Matrix
spweightxt   Panel Spatial Weight Matrix

José-AntonioMonteiro & Madina Kukenova, 2009. "Spatial Dynamic PanelModel and System GMM: A Monte Carlo Investigation," IRENE Working Papers 09-01, IRENE Institute of EconomicResearch.

学术指导：张晓峒老师

本期解答人：谢杰老师

编辑：李光勤

统筹：易仰楠 李丹丹

技术：知我者 赵雅轩 郭凯

往期回顾

互助问答第74期:评估政策影响没有对照组数据问题

互助问答第73期:空间面板问题

互助问答第72期:倾向值匹配问题

互助问答第71期:中介效应检验问题

长按二维码识别关注我们

如果您在计量学习和实证研究中遇到问题，请及时发到邮箱szlw58@126.com，专业委员会有30多名编辑都会看，您的问题会得到及时关注！请您将问题描述清楚，任何有助于把问题描述清楚的细节都能使我们更方便地回答您的问题，提问细则参见：实证研究互助平台最新通知

如果您想成为问题解答者，在帮助他人过程中巩固自己的知识，请发邮件至szlw58@126.com（优先）或给本公众号留言或加微信793481976给群主留言，我们诚挚欢迎热心的学者和学生。具体招募信息请参见：实证研究互助平台志愿者团队招募公告

鲜活的事例更有助于提高您的研究水平，呆板的教科书让人生厌。如果您喜欢，请提出您的问题，也请转发推广！

（欢迎转发，欢迎分享；转载请注明出处，引用和合作请留言。本文作者拥有所有版权，原创文章最早发表于“论文导向计量实证社区”。任何侵权行为将面临追责！）