test of endogeneity Durbin Wu Hausman

[Araceli R]

Dear all

I want to perform a test of endogeneity Durbin Wu Hausman. I know how

can be done

The command is: estat endogenous [var_list]

I would like to know if it can be only performed before having

fitted an instrumental variable regression.

I can´t use an iv regression because I dont have instrumental variables. I have used structural equations and I would like to use this test for endogeneity purposes. How I could perform the test? Thanks for your help in advance



Yours sincerely,

[Eric de Souza]

What do you mean by "I have used structural equations"?

[Araceli R]

Hello Eric

Thank you for your interest
When I wrote "structural equations" I mean the metodology called Structural equation modelling (SEM). It is a type of multivariate analysis, which allow you estimate two o more ecuations at the same time.. It is used usually to test causal relationships.
A type of SEM is path analysis.

[Eric de Souza]

I just wanted to make sure that is what you meant. I have never used SEMs in Stata, but if you have two or more equations you are modelling the endogeneity.
Example: if Y = C + I and C = a +b.Y +u, then you are stating that Y is endogenous in the equation for C. Assuming that I is exogenous, I will be the instrument.
Instrumental variable estimation is used when you do not want to or do not have the information to model the structure.
This is not very precise but should give you an idea of what it is about

[Araceli R]

Thank you very much Eric.
Your advice is very helpful.
Do you know any paper/book in which I could justify the method you´ve explained me?
Thank you

Araceli

[Eric de Souza]

Since I am not familiar with SEM using path diagrammes, I hesitate to give references.
The following links the two: http://davidakenny.net/cm/iv.htm
I have only glanced through it just now, but it seems to show the link between SEM and IV, at least in the early part.
The textbook references I would recommend are Wooldridge, Introductory Econometrics, and Stock and Watson, Introduction to Econometrics. See the relevant chapters on IV estimation.
You use IV estimation when you are not willing or able to specify the whole structure (also in other cases) but you suspect endogeneity