STATA空间面板模型命令演示 - 第5页 - Stata专版

41楼

发表于 2019-7-19 09:53:33

花芽子发表于 2019-7-16 15:07
楼主你好，用这个spregdpd做LM/LR /WALD检验时，发现没有生成，得到的结果没有
*** Groupwise Pan ...

. help spregsemxt

. sysuse spregsemxt.dta, clear

. spregsemxt y x1 x2 , nc(7) wmfile(SPWxt) mfx(lin) test

==============================================================================
*** Binary (0/1) Weight Matrix: 49x49 - NC=7 NT=7 (Non Normalized)
==============================================================================

initial:    log likelihood =  -187.4251
rescale:    log likelihood =  -187.4251
rescale eq: log likelihood =  -187.4251
Iteration 0: log likelihood =  -187.4251
Iteration 1: log likelihood =  -187.3013
Iteration 2: log likelihood = -187.29949
Iteration 3: log likelihood = -187.29949
==============================================================================
* MLE Spatial Error Panel Normal Model (SEM)
==============================================================================
  y = x1 + x2
------------------------------------------------------------------------------
  Sample Size    =       49 | Cross Sections Number =          7
  Wald Test       = 144.2997 | P-Value > Chi2(2)    =    0.0000
  F-Test          =    72.1499 | P-Value > F(2 , 40)    =    0.0000
(Buse 1973) R2    =    0.7583 | Raw Moments R2       =    0.9560
(Buse 1973) R2 Adj =    0.7099 | Raw Moments R2 Adj    =    0.9473
  Root MSE (Sigma)  =    9.0116 | Log Likelihood Function = -187.2995
------------------------------------------------------------------------------
- R2h= 0.5523 R2h Adj= 0.4628  F-Test = 28.38 P-Value > F(2 , 40)  0.0000
- R2v= 0.5647 R2v Adj= 0.4776  F-Test = 29.83 P-Value > F(2 , 40)  0.0000
------------------------------------------------------------------------------
         |    Coef. Std. Err.    z P>|z|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
y          |
      x1 |  -.2848547 .1037195 -2.75 0.006 -.4881411 -.0815682
      x2 |  -1.593681 .3232377 -4.93 0.000 -2.227216 -.9601471
   _cons | 67.78676 5.045627 13.43 0.000    57.89751    77.676
-------------+----------------------------------------------------------------
   /Lambda | .009203 .0232009    0.40 0.692    -.03627 .0546759
   /Sigma | 11.06048 1.117298    9.90 0.000    8.870613 13.25034
------------------------------------------------------------------------------
LR Test SEM vs. OLS (Lambda=0): 0.1573 P-Value > Chi2(1) 0.6916
Acceptable Range for Lambda:    -0.5201 < Lambda < 0.3115
------------------------------------------------------------------------------

==============================================================================
* Panel Model Selection Diagnostic Criteria
==============================================================================
- Log Likelihood Function                LLF          = -187.2995
---------------------------------------------------------------------------
- Akaike Information Criterion             (1974) AIC    =    74.9280
- Akaike Information Criterion             (1973) Log AIC =    4.3165
---------------------------------------------------------------------------
- Schwarz Criterion                      (1978) SC    =    84.1292
- Schwarz Criterion                      (1978) Log SC  =    4.4324
---------------------------------------------------------------------------
- Amemiya Prediction Criterion             (1969) FPE    =    86.1804
- Hannan-Quinn Criterion                   (1979) HQ    =    78.2941
- Rice Criterion                         (1984) Rice =    75.5428
- Shibata Criterion                      (1981) Shibata =    74.4101
- Craven-Wahba Generalized Cross Validation (1979) GCV    =    75.2215
------------------------------------------------------------------------------

==============================================================================
*** Spatial Panel Aautocorrelation Tests
==============================================================================
  Ho: Error has No Spatial AutoCorrelation
  Ha: Error has Spatial AutoCorrelation

- GLOBAL Moran MI          = 0.1131    P-Value > Z( 1.177) 0.2393
- GLOBAL Geary GC          = 0.8362    P-Value > Z(-1.161) 0.2456
- GLOBAL Getis-Ords GO    =  -0.3230    P-Value > Z(-1.177) 0.2393
------------------------------------------------------------------------------
- Moran MI Error Test       = 0.5410    P-Value > Z(4.938) 0.5885
------------------------------------------------------------------------------
- LM Error (Burridge)       = 0.5962    P-Value > Chi2(1)    0.4400
- LM Error (Robust)       = 1.2481    P-Value > Chi2(1)    0.2639
------------------------------------------------------------------------------
  Ho: Spatial Lagged Dependent Variable has No Spatial AutoCorrelation
  Ha: Spatial Lagged Dependent Variable has Spatial AutoCorrelation

- LM Lag (Anselin)          = 0.1395    P-Value > Chi2(1)    0.7088
- LM Lag (Robust)          = 0.7914    P-Value > Chi2(1)    0.3737
------------------------------------------------------------------------------
  Ho: No General Spatial AutoCorrelation
  Ha: General Spatial AutoCorrelation

- LM SAC (LMErr+LMLag_R)    = 1.3876    P-Value > Chi2(2)    0.4997
- LM SAC (LMLag+LMErr_R)    = 1.3876    P-Value > Chi2(2)    0.4997
------------------------------------------------------------------------------

==============================================================================
*** Panel Heteroscedasticity Tests
==============================================================================
  Ho: Panel Homoscedasticity - Ha: Panel Heteroscedasticity

- Engle LM ARCH Test AR(1): E2 = E2_1 = 0.3297 P-Value > Chi2(1)  0.5658
------------------------------------------------------------------------------
- Hall-Pagan LM Test: E2 = Yh       = 0.0376 P-Value > Chi2(1)  0.8462
- Hall-Pagan LM Test: E2 = Yh2       = 0.0066 P-Value > Chi2(1)  0.9355
- Hall-Pagan LM Test: E2 = LYh2    = 0.0105 P-Value > Chi2(1)  0.9184
------------------------------------------------------------------------------
- Harvey LM Test: LogE2 = X       = 4.7766 P-Value > Chi2(2)  0.0918
- Wald Test:       LogE2 = X       =  11.7859 P-Value > Chi2(1)  0.0006
- Glejser LM Test:    |E| = X       = 8.4592 P-Value > Chi2(2)  0.0146
- Breusch-Godfrey Test:  E = E_1 X    =  10.3725 P-Value > Chi2(1)  0.0013
------------------------------------------------------------------------------
- Machado-Santos-Silva Test: Ev=Yh Yh2  = 0.0942 P-Value > Chi2(2)  0.9540
- Machado-Santos-Silva Test: Ev=X    = 7.1616 P-Value > Chi2(2)  0.0279
------------------------------------------------------------------------------
- White Test - Koenker(R2): E2 = X    = 9.3655 P-Value > Chi2(2)  0.0093
- White Test - B-P-G (SSR): E2 = X    =  13.4664 P-Value > Chi2(2)  0.0012
------------------------------------------------------------------------------
- White Test - Koenker(R2): E2 = X X2 =  10.6604 P-Value > Chi2(4)  0.0307
- White Test - B-P-G (SSR): E2 = X X2 =  15.3285 P-Value > Chi2(4)  0.0041
------------------------------------------------------------------------------
- White Test - Koenker(R2): E2 = X X2 XX=  24.9508 P-Value > Chi2(5)  0.0001
- White Test - B-P-G (SSR): E2 = X X2 XX=  35.8762 P-Value > Chi2(5)  0.0000
------------------------------------------------------------------------------
- Cook-Weisberg LM Test: E2/S2n = Yh = 0.0541 P-Value > Chi2(1)  0.8161
- Cook-Weisberg LM Test: E2/S2n = X    =  13.4664 P-Value > Chi2(2)  0.0012
------------------------------------------------------------------------------
*** Single Variable Tests (E2/Sig2):
- Cook-Weisberg LM Test: x1             = 4.4590 P-Value > Chi2(1) 0.0347
- Cook-Weisberg LM Test: x2             = 2.3833 P-Value > Chi2(1) 0.1226
------------------------------------------------------------------------------
*** Single Variable Tests:
- King LM Test: x1                      = 0.5462 P-Value > Chi2(1) 0.4599
- King LM Test: x2                      = 2.8806 P-Value > Chi2(1) 0.0897
------------------------------------------------------------------------------

==============================================================================
* Panel Groupwise Heteroscedasticity Tests
==============================================================================
  Ho: Panel Homoscedasticity - Ha: Panel Groupwise Heteroscedasticity

- Lagrange Multiplier LM Test    = 7.3373    P-Value > Chi2(6) 0.2908
- Likelihood Ratio LR Test       = 7.1253    P-Value > Chi2(6) 0.3094
- Wald Test                      =  12.4812    P-Value > Chi2(7) 0.0858
------------------------------------------------------------------------------

测试过了，确实 spregdpd 没提供这个检验，我以为两个面板异质性检验是一样的。
这里只有 spregsemxt 提供了Panel Groupwise Heteroscedasticity Tests

这个不是你的错误，好像我去查询过网站应该是叫 idea 的网站。作者是埃及的经济学家。
这几个空间面板估计命令都是他写的。他的 spregdpd 确实没提供你列举的这个统计检验。

不是你的错误啊，是程序的作者没提供这个检验。只要你做空间动态面板估计，如果不需要这个检验可以不考虑，如果需要，那没办法作者的程序命令没提供这个检验，你只能缺省这个检验。

最近忙其他的事情去了，今天才回复，不好意思。

42楼

tulipsliu

发表于 2019-7-19 09:53:46

花芽子发表于 2019-7-16 15:07
楼主你好，用这个spregdpd做LM/LR /WALD检验时，发现没有生成，得到的结果没有
*** Groupwise Pan ...

. help spregsemxt

. sysuse spregsemxt.dta, clear

. spregsemxt y x1 x2 , nc(7) wmfile(SPWxt) mfx(lin) test

==============================================================================
*** Binary (0/1) Weight Matrix: 49x49 - NC=7 NT=7 (Non Normalized)
==============================================================================

initial:    log likelihood =  -187.4251
rescale:    log likelihood =  -187.4251
rescale eq: log likelihood =  -187.4251
Iteration 0: log likelihood =  -187.4251
Iteration 1: log likelihood =  -187.3013
Iteration 2: log likelihood = -187.29949
Iteration 3: log likelihood = -187.29949
==============================================================================
* MLE Spatial Error Panel Normal Model (SEM)
==============================================================================
  y = x1 + x2
------------------------------------------------------------------------------
  Sample Size    =       49 | Cross Sections Number =          7
  Wald Test       = 144.2997 | P-Value > Chi2(2)    =    0.0000
  F-Test          =    72.1499 | P-Value > F(2 , 40)    =    0.0000
(Buse 1973) R2    =    0.7583 | Raw Moments R2       =    0.9560
(Buse 1973) R2 Adj =    0.7099 | Raw Moments R2 Adj    =    0.9473
  Root MSE (Sigma)  =    9.0116 | Log Likelihood Function = -187.2995
------------------------------------------------------------------------------
- R2h= 0.5523 R2h Adj= 0.4628  F-Test = 28.38 P-Value > F(2 , 40)  0.0000
- R2v= 0.5647 R2v Adj= 0.4776  F-Test = 29.83 P-Value > F(2 , 40)  0.0000
------------------------------------------------------------------------------
         |    Coef. Std. Err.    z P>|z|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
y          |
      x1 |  -.2848547 .1037195 -2.75 0.006 -.4881411 -.0815682
      x2 |  -1.593681 .3232377 -4.93 0.000 -2.227216 -.9601471
   _cons | 67.78676 5.045627 13.43 0.000    57.89751    77.676
-------------+----------------------------------------------------------------
   /Lambda | .009203 .0232009    0.40 0.692    -.03627 .0546759
   /Sigma | 11.06048 1.117298    9.90 0.000    8.870613 13.25034
------------------------------------------------------------------------------
LR Test SEM vs. OLS (Lambda=0): 0.1573 P-Value > Chi2(1) 0.6916
Acceptable Range for Lambda:    -0.5201 < Lambda < 0.3115
------------------------------------------------------------------------------

==============================================================================
* Panel Model Selection Diagnostic Criteria
==============================================================================
- Log Likelihood Function                LLF          = -187.2995
---------------------------------------------------------------------------
- Akaike Information Criterion             (1974) AIC    =    74.9280
- Akaike Information Criterion             (1973) Log AIC =    4.3165
---------------------------------------------------------------------------
- Schwarz Criterion                      (1978) SC    =    84.1292
- Schwarz Criterion                      (1978) Log SC  =    4.4324
---------------------------------------------------------------------------
- Amemiya Prediction Criterion             (1969) FPE    =    86.1804
- Hannan-Quinn Criterion                   (1979) HQ    =    78.2941
- Rice Criterion                         (1984) Rice =    75.5428
- Shibata Criterion                      (1981) Shibata =    74.4101
- Craven-Wahba Generalized Cross Validation (1979) GCV    =    75.2215
------------------------------------------------------------------------------

==============================================================================
*** Spatial Panel Aautocorrelation Tests
==============================================================================
  Ho: Error has No Spatial AutoCorrelation
  Ha: Error has Spatial AutoCorrelation

- GLOBAL Moran MI          = 0.1131    P-Value > Z( 1.177) 0.2393
- GLOBAL Geary GC          = 0.8362    P-Value > Z(-1.161) 0.2456
- GLOBAL Getis-Ords GO    =  -0.3230    P-Value > Z(-1.177) 0.2393
------------------------------------------------------------------------------
- Moran MI Error Test       = 0.5410    P-Value > Z(4.938) 0.5885
------------------------------------------------------------------------------
- LM Error (Burridge)       = 0.5962    P-Value > Chi2(1)    0.4400
- LM Error (Robust)       = 1.2481    P-Value > Chi2(1)    0.2639
------------------------------------------------------------------------------
  Ho: Spatial Lagged Dependent Variable has No Spatial AutoCorrelation
  Ha: Spatial Lagged Dependent Variable has Spatial AutoCorrelation

- LM Lag (Anselin)          = 0.1395    P-Value > Chi2(1)    0.7088
- LM Lag (Robust)          = 0.7914    P-Value > Chi2(1)    0.3737
------------------------------------------------------------------------------
  Ho: No General Spatial AutoCorrelation
  Ha: General Spatial AutoCorrelation

- LM SAC (LMErr+LMLag_R)    = 1.3876    P-Value > Chi2(2)    0.4997
- LM SAC (LMLag+LMErr_R)    = 1.3876    P-Value > Chi2(2)    0.4997
------------------------------------------------------------------------------

==============================================================================
*** Panel Heteroscedasticity Tests
==============================================================================
  Ho: Panel Homoscedasticity - Ha: Panel Heteroscedasticity

- Engle LM ARCH Test AR(1): E2 = E2_1 = 0.3297 P-Value > Chi2(1)  0.5658
------------------------------------------------------------------------------
- Hall-Pagan LM Test: E2 = Yh       = 0.0376 P-Value > Chi2(1)  0.8462
- Hall-Pagan LM Test: E2 = Yh2       = 0.0066 P-Value > Chi2(1)  0.9355
- Hall-Pagan LM Test: E2 = LYh2    = 0.0105 P-Value > Chi2(1)  0.9184
------------------------------------------------------------------------------
- Harvey LM Test: LogE2 = X       = 4.7766 P-Value > Chi2(2)  0.0918
- Wald Test:       LogE2 = X       =  11.7859 P-Value > Chi2(1)  0.0006
- Glejser LM Test:    |E| = X       = 8.4592 P-Value > Chi2(2)  0.0146
- Breusch-Godfrey Test:  E = E_1 X    =  10.3725 P-Value > Chi2(1)  0.0013
------------------------------------------------------------------------------
- Machado-Santos-Silva Test: Ev=Yh Yh2  = 0.0942 P-Value > Chi2(2)  0.9540
- Machado-Santos-Silva Test: Ev=X    = 7.1616 P-Value > Chi2(2)  0.0279
------------------------------------------------------------------------------
- White Test - Koenker(R2): E2 = X    = 9.3655 P-Value > Chi2(2)  0.0093
- White Test - B-P-G (SSR): E2 = X    =  13.4664 P-Value > Chi2(2)  0.0012
------------------------------------------------------------------------------
- White Test - Koenker(R2): E2 = X X2 =  10.6604 P-Value > Chi2(4)  0.0307
- White Test - B-P-G (SSR): E2 = X X2 =  15.3285 P-Value > Chi2(4)  0.0041
------------------------------------------------------------------------------
- White Test - Koenker(R2): E2 = X X2 XX=  24.9508 P-Value > Chi2(5)  0.0001
- White Test - B-P-G (SSR): E2 = X X2 XX=  35.8762 P-Value > Chi2(5)  0.0000
------------------------------------------------------------------------------
- Cook-Weisberg LM Test: E2/S2n = Yh = 0.0541 P-Value > Chi2(1)  0.8161
- Cook-Weisberg LM Test: E2/S2n = X    =  13.4664 P-Value > Chi2(2)  0.0012
------------------------------------------------------------------------------
*** Single Variable Tests (E2/Sig2):
- Cook-Weisberg LM Test: x1             = 4.4590 P-Value > Chi2(1) 0.0347
- Cook-Weisberg LM Test: x2             = 2.3833 P-Value > Chi2(1) 0.1226
------------------------------------------------------------------------------
*** Single Variable Tests:
- King LM Test: x1                      = 0.5462 P-Value > Chi2(1) 0.4599
- King LM Test: x2                      = 2.8806 P-Value > Chi2(1) 0.0897
------------------------------------------------------------------------------

==============================================================================
* Panel Groupwise Heteroscedasticity Tests
==============================================================================
  Ho: Panel Homoscedasticity - Ha: Panel Groupwise Heteroscedasticity

- Lagrange Multiplier LM Test    = 7.3373    P-Value > Chi2(6) 0.2908
- Likelihood Ratio LR Test       = 7.1253    P-Value > Chi2(6) 0.3094
- Wald Test                      =  12.4812    P-Value > Chi2(7) 0.0858
------------------------------------------------------------------------------

测试过了，确实 spregdpd 没提供这个检验，我以为两个面板异质性检验是一样的。
这里只有 spregsemxt 提供了Panel Groupwise Heteroscedasticity Tests

这个不是你的错误，好像我去查询过网站应该是叫 idea 的网站。作者是埃及的经济学家。
这几个空间面板估计命令都是他写的。他的 spregdpd 确实没提供你列举的这个统计检验。

不是你的错误啊，是程序的作者没提供这个检验。只要你做空间动态面板估计，如果不需要这个检验可以不考虑，如果需要，那没办法作者的程序命令没提供这个检验，你只能缺省这个检验。

最近忙其他的事情去了，今天才回复，不好意思。

43楼

2754439263 发表于 2019-7-19 18:00:50 来自手机

tulipsliu 发表于 2014-10-22 11:47
近几年遇到许多朋友，都是做空间计量的。这里有一个命令，用系统的数据演示一下他们的结果。
命令为 spreg ...

收藏了～～

44楼

那世故衣 发表于 2019-7-21 20:48:47

楼主你好，我要做LM检验确定使用SAR或SEM模型，但同样的面板数据，在spregsemxt和spregdpd做出来LM检验结果不一样，不知道该相信哪一个了。。请您赐教，感谢~

45楼

花芽子 发表于 2019-7-23 22:41:45

tulipsliu 发表于 2019-7-19 09:53
. help spregsemxt

. sysuse spregsemxt.dta, clear

谢谢你啊非常的热心，我尝试很多次，对比了确实没有。

46楼

TobetheBest 发表于 2019-9-19 09:30:32

左手、微笑向暖发表于 2017-3-10 21:28
楼主，您好!我有一个问题想请教您，就是我有32*32的权重矩阵，数据是2003年-2014年32个国家的面板数据，样本 ...

需要把长面板转换成宽面板，才能继续

47楼

花芽子 发表于 2019-12-13 11:39:38

tulipsliu 发表于 2014-11-25 13:00
是一个STATA 命令。你通过 Help 菜单下，输入整个命令。点 search all。
最后会找到，在线安装。
会有整 ...

你好，我又打扰你啦，暑假工作没做完，现在还有些疑问，看到好多文章说事在非空间面板模型上进行LM检验，结果出来会有传统的LM检验和稳健的Lm检验（那个埃及人Elhorsts写的空间计量经济学从横截面数据到空间面板书79-81页）。但是非空间模型不就是面板数据模型吗。这个的命令是什么，为啥得出lLM-Error和R-LM Error、LM Lag、R- LM Lag。我知道目前 spregdpd可以做sar的Lm检验，spregsemxt做SEM的lm检验，但是这两个命令是在SAR和sem模型上进行的，也不是非空间模型啊。这思路整的我好乱。帮忙指导一下，先谢谢了

48楼

tulipsliu

发表于 2019-12-14 22:45:23

花芽子发表于 2019-12-13 11:39
你好，我又打扰你啦，暑假工作没做完，现在还有些疑问，看到好多文章说事在非空间面板模型上进行LM检验， ...

你这样问不对；
LM拉格朗日乘子检验就是为了检验空间面板模型的；

我给你举例：在  lesage 的MATLAB 程序里检验空间横切面模型用的好像不是LM 检验，用的是，我很久没看MATLAB 程序了；我上传人大经济论坛 Paul elhorst 的MATLAB 程序的，这个人是芬兰计量经济学家，欧洲人。不是埃及人。

STATA 的命令里确实是埃及人的。

还有你问的问题不成立。应该非空间面板模型是没有LM 检验的。这个LM 检验是在 MATLAB 程序里。是专门用来区别  SAR 和 SEM 模型的。

模型估计后，实际上如果你用了 MATLAB 的话还有一个豪斯曼检验。  这些你应该多看计量经济学里的面板模型的估计，而且也要看空间面板的估计。

而区别于MATLAB

这个命令还可以做 GMM 估计，它可以用过度识别检验 sargan test

这些你可以多思考多摸索