Page 1 of 1

estimate and estimate(model=....

Posted: Wed Jan 27, 2010 2:12 am
by airways1
The model is specified follow,

linreg(define=longeqn) y
# constant x
system(model=m2)
variables x y z
lags 1 to 2
det d_x1 constant
ect longeqn
end(system)

estimate ;*<<----------OK
estimate(model=m2) 1 100 ;*<<----------- NOT OK; (insteaded the ect model is estimated, the var model did.)

i need to use the second instruction, since i have different model in a loop. I am not sure if this is a winrats bug, but i need a solution. Please help. thanks.

Re: estimate and estimate(model=....

Posted: Wed Jan 27, 2010 5:17 pm
by moderator
The ECT doesn't "follow" the MODEL data type, so you can only have model using ECT at a time. How are you using the two models? It only matters if you're using ESTIMATE on both, and the OLS model can be estimated with SUR(MODEL=...) instead.

Re: estimate and estimate(model=....

Posted: Thu Jan 28, 2010 4:20 am
by airways1
system(model=m1)
.
ect longeqn1
end

system(model=m2)
.
ect longeqn2
end(system)
.
.
.
(some orther code)
.
estimate(model=m1);<<--generating VAR(m1) result,not VAR with ect result.
;if i type "estimate" without model name, it give me VAR(m2). But I need the result of
VAR(m1)'s ect result.
any suggestion?

Re: estimate and estimate(model=....

Posted: Thu Jan 28, 2010 7:33 am
by TomDoan
Is there any reason you can't do

Code: Select all

system(model=m1)
...
end(system)
estimate
*
system(model=m2)
...
end(system)
estimate

Re: estimate and estimate(model=....

Posted: Thu Jan 28, 2010 7:00 pm
by airways1
reason is simple,check out the following function:

function mdl
type model mdl
.
.
estimate(model=mdl)
.
end

now whether I call this function with either ect type or not , the result is not ect result.

Re: estimate and estimate(model=....

Posted: Thu Jan 28, 2010 9:20 pm
by TomDoan
You'll have to post the whole program; I can't really give you a workaround without it.