On Wed, 21 Mar 2012, Daniel Bencik wrote:
one more little thing. When I try to estimate
vecm 7 1 AllHigh AllLow --quiet
everything goes fine even on small samples of, say, 100 observations.
However, when I try to add one exogenous variables, like
vecm 7 1 AllHigh AllLow; d_AllOpen
it only works on the full sample of 999 observations. When I
restrict the sample to 746 observations, I get an
"Insufficient degrees of freedom for regression" dialog.
For us to make sense of this you'd have to show us your data,
for d_AllOpen in particular. There's no relevant inherent
limitation affecting the vecm command. Have missing values
crept into your data? That could happen if you define
d_AllOpen with one sub-sample in force, then switch to another
sub-sample.
Allin Cottrell