如题。我使用stata进行空间面板tobit分析。
使用sptobitsemxt命令分析空间误差模型时,出现matrix operation not found。
搜索了论坛里,貌似也有同学出现了类似的情况。
我的数据结构是一个包含全国31个省市5年的数据集。
我的命令如下:
sptobitsemxt te_cm gra urb ip inv fd ipr c inc , nc(31) wmfile(W.dta) ll(0)
结果如下:
. sptobitsemxt te_cm gra urb ip inv fd ipr c inc , nc(31) wmfile(W.dta) ll(0)
==============================================================================
*** Binary (0/1) Weight Matrix: 155x155 - NC=31 NT=5 (Non Normalized)
==============================================================================
*** te_cm Lower Limit = 0
*** te_cm Left- Censored Observations = 0
*** te_cm Left-UnCensored Observations = 155
------------------------------------------------------------
initial: log likelihood = -12.831656
rescale: log likelihood = -12.831656
rescale eq: log likelihood = -12.831656
Iteration 0: log likelihood = -12.831656
Iteration 1: log likelihood = -11.467842 (not concave)
Iteration 2: log likelihood = -11.043374 (not concave)
Iteration 3: log likelihood = -10.983994 (not concave)
Iteration 4: log likelihood = -10.944418 (not concave)
Iteration 5: log likelihood = -10.929872
Iteration 6: log likelihood = -10.704988 (not concave)
Iteration 7: log likelihood = -10.692386
Iteration 8: log likelihood = -10.673735
Iteration 9: log likelihood = -10.668479
Iteration 10: log likelihood = -10.668474
Iteration 11: log likelihood = -10.668474
==============================================================================
* Tobit MLE Spatial Error Panel Normal Model (SEM)
==============================================================================
te_cm = gra + urb + ip + inv + fd + ipr + c + inc
------------------------------------------------------------------------------
Sample Size = 155 | Cross Sections Number = 31
Wald Test = 86.6387 | P-Value > Chi2(8) = 0.0000
F-Test = 10.8298 | P-Value > F(8 , 116) = 0.0000
(Buse 1973) R2 = 0.3724 | Raw Moments R2 = 0.8252
(Buse 1973) R2 Adj = 0.1668 | Raw Moments R2 Adj = 0.7680
Root MSE (Sigma) = 0.2687 | Log Likelihood Function = -10.6685
------------------------------------------------------------------------------
- R2h= 0.1977 R2h Adj=-0.0652 F-Test = 4.50 P-Value > F(8 , 116) 0.0001
- R2v= 0.1999 R2v Adj=-0.0622 F-Test = 4.56 P-Value > F(8 , 116) 0.0001
------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
te_cm |
gra | -.0454749 .0269217 -1.69 0.091 -.0982404 .0072906
urb | .042601 .1883429 0.23 0.821 -.3265443 .4117462
ip | -.0081308 .0253562 -0.32 0.748 -.0578281 .0415666
inv | .0825376 .0410264 2.01 0.044 .0021272 .1629479
fd | -.1156577 .1535322 -0.75 0.451 -.4165753 .1852598
ipr | .326603 .1652947 1.98 0.048 .0026314 .6505746
c | .1046859 .0597849 1.75 0.080 -.0124903 .2218622
inc | -.0587263 .0363569 -1.62 0.106 -.1299846 .012532
_cons | -2.192343 .7289925 -3.01 0.003 -3.621142 -.7635443
-------------+----------------------------------------------------------------
/Lambda | -.0114965 .0063832 -1.80 0.072 -.0240073 .0010142
/Sigma | .2591376 .0147183 17.61 0.000 .2302904 .2879849
------------------------------------------------------------------------------
LR Test SEM vs. OLS (Lambda=0): 3.2439 P-Value > Chi2(1) 0.0717
Acceptable Range for Lambda: -0.3692 < Lambda < 0.1894
------------------------------------------------------------------------------
matrix operation not found
r(501);
end of do-file
r(501);
但是,做空间滞后模型(sptobitsarxt:Tobit MLE Spatial Lag Panel Regression)没有问题,所以应该不是数据集的问题。
求教该如何修改命令呢?(已附上 回归数据,空间权重矩阵请见上一个帖子)