Regressor Endogeneity Test with ARMA terms
Posted: Sat Sep 01, 2012 11:00 pm
Hi all,
These forums have been fantastic - thankyou for all the responses and posts it has helped tremendously.
If you could please help with the below question it would be greatly appreciated!
I am using eviews 7
I am attempting to run a 2SLS regression with quantity the dependant variable, price the endogenous explanatory variable and a bunch of exogenous explanatory variables.
My estimation has significant autocorrelation so I have ARMA tems included: AR(1), AR(2) and AR(3).
This has proven to give a solid result in terms of t-statistics of the various terms, the R-squared and the durbin watson.
I now wish to run a regressor endogeneity test to show that price is indeed endogenous with quantity. However using the "view -> IV diagnostics & tests -> regressor endogeneity test" I get an error term that states that ARMA terms are not allowed in this procedure. However how am I supposed to compare the J statistics as per the Hausman test with the restricted and unrestricted J statistics without the full specification of the model? i.e. including the ARMA terms?
If I run the regressor endogeneity test without the ARMA specifications the test works and is unable to reject the null of a exogonous price variable however I do not believe I am reviewing the true specifications of my model by excluding the ARMA terms?
Any help would be GREATLY appreciated.
Thanks,
Michael
These forums have been fantastic - thankyou for all the responses and posts it has helped tremendously.
If you could please help with the below question it would be greatly appreciated!
I am using eviews 7
I am attempting to run a 2SLS regression with quantity the dependant variable, price the endogenous explanatory variable and a bunch of exogenous explanatory variables.
My estimation has significant autocorrelation so I have ARMA tems included: AR(1), AR(2) and AR(3).
This has proven to give a solid result in terms of t-statistics of the various terms, the R-squared and the durbin watson.
I now wish to run a regressor endogeneity test to show that price is indeed endogenous with quantity. However using the "view -> IV diagnostics & tests -> regressor endogeneity test" I get an error term that states that ARMA terms are not allowed in this procedure. However how am I supposed to compare the J statistics as per the Hausman test with the restricted and unrestricted J statistics without the full specification of the model? i.e. including the ARMA terms?
If I run the regressor endogeneity test without the ARMA specifications the test works and is unable to reject the null of a exogonous price variable however I do not believe I am reviewing the true specifications of my model by excluding the ARMA terms?
Any help would be GREATLY appreciated.
Thanks,
Michael