For cross-sectional data, classical assumptions include random sampling. However, based on random sampling, "serial correlation" in the error terms is not a problem at all. This becomes interesting with TS data and Panel ones. if yo consider the correlation between the errors and the regressors, then this becomes a mispecification problem.

[此贴子已经被作者于2006-4-26 2:47:20编辑过]

多谢指点！还想再问下如果是面板数据的误差项间（非误差与回归元）自相关问题又如何处理？或者是面板数据中某一（非全部）纵列（时间序列）存在误差项间的自相关也不用处理吗