拥抱大海的鱼 发表于 2012-7-22 07:56 
楼主,请教工具变量估计可靠吗?有时发现即使内生性检验以及工具变量有效性检验都通过了,也就是从计量的 ...
Theoretically speaking, you can never make sure your instruments are valid. Also, DID is estimating the average treatment effect, while IV is estimating local treatment effects. Without homogeneity assumption, they are even incomparable. You may choose the identification for which assumptions are more likely to hold.