Sven Schreiber schrieb:
Also, I don't see any remark in the output on whether I used a HAC
covariance matrix estimator or not. (This actually also seems to apply
to plain OLS, was this the case already in earlier versions? All this
is with 1.7.0 on windows.)
To add to the list of small tsls quirks, it seems to me that the printed
information about the specification is not complete. Consider the command
tsls y1 0 y2 x1 ; x1 x2
This means that x1 is its own instrument and thus is considered to be
but y2 is endogenous and needs to be instrumented.
However, gretl's output just says:
and thus the status of y2 and x1 is not clear from the output alone,
which should be the case IMHO.