t*****e
发帖数: 224 | 1 ☆─────────────────────────────────────☆
jjwwjj (jjwwjj) 于 (Tue Feb 20 03:01:42 2007) 提到:
哪里有比较好的notes介绍总结比较GMM，ML, QML的有效性的？
现在学了一堆公式，但是对它们的各自的特性，何时最优，何时无偏，何时等价等等却
是很糊涂。
☆─────────────────────────────────────☆
tuc (中年无肥) 于 (Tue Feb 20 06:25:50 2007) 提到:
They are not comparable. They are for different purposes.
ML makes assumptions about the underlying data generating processes, so it
runs a much higher risk of specification errors. However once you estimate
with ML, you will get an est |
|
