同样的数据做跑固定效应，可以得出结果，不知道是什么原因造成的，求指点求帮忙。
Fitting negative binomial (constant dispersion) model:


Iteration 16:  log likelihood = -12513.111  (not concave)
Iteration 17:  log likelihood =  -12513.11  
Iteration 18:  log likelihood =  -12513.11  (backed up)
Iteration 19:  log likelihood =  -12513.11  (backed up)
Iteration 20:  log likelihood =  -12513.11  (not concave)
Iteration 21:  log likelihood =  -12513.11  (not concave)
Iteration 22:  log likelihood =  -12513.11  (not concave)
Iteration 23:  log likelihood =  -12513.11  (not concave)
Iteration 24:  log likelihood =  -12513.11  (not concave)
Iteration 25:  log likelihood =  -12513.11  (not concave)
Iteration 26:  log likelihood =  -12513.11  (not concave)
Iteration 27:  log likelihood =  -12513.11  (not concave)
如此循环下去了