我用空间计量的matlab程序跑出来的结果是这样。无法解释,因为按照一般教科书理论t值与系数符合应该相同。论坛上有过类似的帖子,大多持造假的看法。
Pooled model with spatially lagged dependent variable and spatial fixed effects
Dependent Variable = lnAMT(t)
R-squared = 0.9976
corr-squared = 0.9890
sigma^2 = 28.4927
Nobs,Nvar,#FE = 841, 6, 34
log-likelihood = -2603.2223
# of iterations = 1
min and max rho = -1.0000, 1.0000
total time in secs = 0.2380
time for optimiz = 0.0670
time for lndet = 0.0370
time for t-stats = 0.0100
No lndet approximation used
***************************************************************
Variable Coefficient Asymptot t-stat z-probability
lnAMT(t-1) 1.012922 128.942601 0.000000
W*lnAMT(t-1) -0.153384 -5.898976 0.000000
CITY -2.614549 -0.438251 0.661204
lnIRRG 2.300923 2.852373 0.004339
LnPIND -1.379316 1.325403 0.185038
W*dep.var. 0.159781 5.760241 0.000000