On Tue, 17 Aug 2010, artur bala wrote:
I'm trying to do the CUSUM test through separated regressions
but the result differs slightly froms gretl's cusum command. I
can't see where the trick is... Any suggestion?
The gretl code is at
http://gretl.cvs.sourceforge.net/viewvc/gretl/gretl/lib/src/compare.c?vie...
(see the function cusum_compute()).
The algorithm is described by William Greene, Econometric
Analysis, 5e, pp 135-6. Computing the scaled residual is slightly
tricky; see Greene's equation 7-19.
Allin Cottrell