Conner Mullally <[email protected]>:
Yes, gamma hat is biased and inconsistent. You have two endogenous
variables, X and XZ. Put another way if E(X'e) is not equal to zero
then E(Z'X'e) is not either; if plim(X'e) is not equal to zero then
plim(Z'X'e) is not either. So you need some excluded instruments for
IV, or another approach.
On Thu, Jan 15, 2009 at 7:31 PM, Conner Mullally
<[email protected]> wrote:
> Hello statalisters,
>
> I am having a "duh" moment with what I think should be an
> easy question. Suppose you have the following model:
>
> y=beta*x + alpha*z + gamma*x*z + error term
>
> Further suppose that while z is independent of the error
> term, x is not. z is also independent of x. We want to
> estimate the marginal effect of z on y, holding x constant
> (at whatever value). Will the estimated marginal effect be
> biased due to the endogeneity of x? That is, will our
> estimate of the parameter gamma be biased?
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
