Hi,
I am trying to run an instrumental regression, but using the hausman test returns negative chi values even when using options sigmamore and sigmaless.
Normal Hausman :
chi2(8) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= -64.17
Sigmamore :
chi2(8) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= -1.70
Sigmaless :
chi2(8) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= -0.71
I have run xtoverid after a random-effects instrumental variable regression (xtivreg, re)
. xtoverid
Test of overidentifying restrictions:
Cross-section time-series model: xtivreg g2sls
Sargan-Hansen statistic 0.000 (equation exactly identified)
Does this mean I should use a fixed-effects model?
Or because I am using Instrumental Variables, is this a test of the validity of the instrument?
This is the part that confuses me, as a lot of the past examples I have looked at on this site are not using instrumental variables.
Thanks
I am trying to run an instrumental regression, but using the hausman test returns negative chi values even when using options sigmamore and sigmaless.
Normal Hausman :
chi2(8) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= -64.17
Sigmamore :
chi2(8) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= -1.70
Sigmaless :
chi2(8) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= -0.71
I have run xtoverid after a random-effects instrumental variable regression (xtivreg, re)
. xtoverid
Test of overidentifying restrictions:
Cross-section time-series model: xtivreg g2sls
Sargan-Hansen statistic 0.000 (equation exactly identified)
Does this mean I should use a fixed-effects model?
Or because I am using Instrumental Variables, is this a test of the validity of the instrument?
This is the part that confuses me, as a lot of the past examples I have looked at on this site are not using instrumental variables.
Thanks
