Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • xtivreg2 endogeneity problem

    Hi Guys,

    I am new to statistics and trying to figure out endogeneity in my panel data.

    I am using dmexogxt after xtivreg or xtivreg2 with endog comment to test endogeneity but each time I can not able to reject the null "that my explanatory variable is exogenous".

    My question is as follows:

    If I may not conclude that there is no endogeneity in the panel data, can I still use xtivreg2 - that is, is the result of xtivreg2 correct?

    Regards,

  • #2
    Hi Misc [sic].

    FIrst, dmexogxt is out-of-date code. You should probably use xtoverid instead, or just use the endog(.) or orthog(.) options of xtivreg2.

    Second, it's not clear what you mean by "can I still use xtivreg2 - that is, is the result of xtivreg2 correct?" since you haven't told us how you specified the model. Did you specify that the explanatory variable is exogenous (and test whether it needs to be treated as endogenous)? Or did you specify that the explanatory variable is endogenous (and test whether the data suggest it's OK to treat it as exogenous)? You can do either in xtiveg2.

    Assuming your other assumptions are all OK, then both specifications are "correct" in the sense that they both give you consistent estimates of the coefficient on your explanatory variable. The first is more efficient, but the results are not robust to a failure of your assumption that the explanatory variable is indeed exogenous. The second is less efficient but more robust.

    Comment

    Working...
    X