Testing of Endogeneity in Panel Fixed Effect

[kofi mintah oware]

Dear Colleague

My data is panel and have performed FE with robust standard errors to correct heteroscedasticity in the error term. However I found auto correlation present after performing Pesaran test. I read that robust standard errors can correct the presence of both

heteroscedasticity

and auto correlation. I will like suggestion if I am on the right track. In addition , how do I test for the presence of endogeneity in a Panel FE.
Thank you

[Carlo Lazzaro]

Kof.:
welocme to this forum.
yes, you're on the right track, since robust/clustered standard errors in -xtreg-account for heteroskedasticity and/or autocorrelation.
As far as testing for endogeneity is concerned, you might be interested in the following Stata thread (starring: Jeff Wooldridge): https://www.statalist.org/forums/for...est-panel-data

[kofi mintah oware]

Thank you for your quick response

[Pedro Antonio]

When write xtivreg y x2 (x1=z), fe endog(x1) in order to test for x1 endogeneity, Stata 14.2 replies,

. xtivreg y x2 (x1=z), fe endog(x1)
option endog() not allowed
r(198);

Could anybody explain to me where the error is?

Thanks in advance,

[Joro Kolev]

Pedro, the option -endog() does not exist in -xtivreg-.

It exists in the user contributed command -xtivreg2-.

Type from command line

findit xtivreg2

and follow the instructions to install it.

Then

xtivreg2 y x2 (x1=z), fe endog(x1)

[Pedro Antonio]

Thanks a lot Joro: it works!