Endogeneity test

[Carmen Blanco]

Dear colleagues,

I have a regression analysis for panel data. The model takes the following form:
Y=x1+x2+x3

And I'd like to know if the x2 variable is endogenous.

Please, can someone tell me about which test to run and how?

Thanks in advance

C.B.

[Mostafa Harakeh]

Hello Carmen,

First I need to inform you that I am sill a PhD student so don't take my suggestions for granted.

To test if x2 is endogenous, you first need to have an exogenous instrument. A conventional practice in finance takes the lagged value as an instrument (call the instrument: z).

Then you run an IV regression using this command: ivregress 2sls Y x1 (x2=z) x3, robust

Then type: estat endogenous

This will perform the Durbin-Wu-Hausman test of endogeneity. H0 is that the variable is exogenous; therefore, if you reject H0 then x2 is endogenous.

I hope this helps.

I would like to know if there is another test for endogeneity without the need for an exogenous instrument.

Best Wishes,

Mustafa