Page 1 of 1

Breusch-Godfrey and 2SLS

Posted: Thu Sep 24, 2009 5:54 pm
by startz
Does anyone happen to know if the standard Breusch-Godfrey LM test for serial correlation is valid when the residuals are from two-stage least squares. My not-too-well-informed impression is that it is not.

Re: Breusch-Godfrey and 2SLS

Posted: Fri Sep 25, 2009 3:20 pm
by EViews Gareth
Some guy called Jeff Wooldridge has a paper "A note on the Lagrange multiplier and F-statistics for two stage least squares regressions" (Econ Letters, Oct. 1990), that covers this.

Re: Breusch-Godfrey and 2SLS

Posted: Fri Sep 25, 2009 3:28 pm
by EViews Gareth
The Wooldridge paper could be seen as a generalisation of an earlier paper (1983, Econ Letters) by some guy called Richard Startz.

Re: Breusch-Godfrey and 2SLS

Posted: Sat Sep 26, 2009 2:56 pm
by startz
I'd missed that the auxiliary regression was by 2SLS.

but then, how could anyone with such great taste in literature ever go wrong?