Page 1 of 1

Unsolvable serial correlation

Posted: Wed Apr 15, 2015 4:34 am
by Ties
Hello,

Currently I am testing my equation "tr_w_ar_1985q4_2013q4" (which is the Taylor rule) for serial correlation using the Breusch-Godfrey serial LM test (View > Residual Diagnostics > Serial Correlation LM test). I notice that serial correlation is present, but when correcting with an AR(1) model I cannot solve for serial correlation.

The model is already corrected for heteroskedasticity with the white robust standard errors. I have attached my workfile to this message.

What can I do to solve for serial correlation?


Kind regards,
Ties

Re: Unsolvable serial correlation

Posted: Wed Apr 15, 2015 7:08 am
by startz
Perhaps include an AR(2) term as well?

Re: Unsolvable serial correlation

Posted: Wed Apr 15, 2015 7:09 am
by Ties
Yep, that was the solution. Thank you!