Fixed effects with xtabond2

[krishantha Ainsworth]

Dear Statalist,

May I know is there a way to add fixed effects in xtabond2 different to the method

xi: xtabond2 x y z i.firmid, gmm (x y , lag(2 2)) iv(z)  two small

I just want to know is their a similar approach to

xtreg  x y z, fe

in xtabond2 other than using i.x notation.
This is because i.x notation creates a separate variable in the variable list for each firm, but

xtreg  x y z, fe

does not create such variable and efficient.

[David Roodman]

Hi Krishantha,
No, xtabond2 does not have a feature for partialling out fixed effects or any other variables.
--David

[Joro Kolev]

The command -xtabond2- by David Roodman already controls for / eliminates the fixed effects. Why would you want to add them?

[krishantha Ainsworth]

Dear David and Joro,

Thank you so much for your replies. I want to estimate the equation (2) (page 2) of

Flannery, M. J., & Hankins, K. W. (2013). Estimating dynamic panel models in corporate finance. Journal of Corporate Finance, 19, 1-19.

Could you tell me what Stata code should I use in xtabond2 to include the firm fixed effects mentioned in that equation ?

[Joro Kolev]

If you

xtset firm_id

and you follow with -xtabond2- this would eliminate the firm fixed effects with first difference transformation, and you do not worry about the fixed effects -- the firm fixed effects are controlled for/eliminated just as if you follow the -xtset firm_id- with -xtreg, fe-.

[krishantha Ainsworth]

Hi Joro,

Thank you so much for the answer.