Moran test and ivreg2

[marco lecci]

Hi,

I am running a Moran’s test of residual correlation with nearby residuals.
I know that Moran runs on regress but if i run it after ivreg2 doesn't work
this is a snipped of my code:

foreach y of numlist 2013 2018 {
sum vote_fdi`y'
*** 2SLS
ivreg2 vote_fdi`y' (opp_rate = iv_jmeansent), robust
ivreg2 vote_fdi`y' (opp_rate = iv_jmeansent) $covars, robust
}


spmatrix create contiguity W, replace
estat moran, errorlag(W)


After running the test I get this error:
estat moran not valid
r(321);

How can I solve this?
Thanks everyone

[Øyvind Snilsberg]

The estat moran help file says that the test can be run after fitting a model using regress.
Instead of ivregress you could do manual 2SLS,

regress opp_rate iv_jmeansent
predict opp_rate_hat
regress vote_fdi opp_rate_hat

[marco lecci]

That worked!! thank you so much. I now have a similar problem but with a more tangled ivreg. here is my code:
global covars = "c_ide c_antifascista pun_weak pun_strng"


*IV
est clear

foreach y1 in vote_right1948{
foreach y2 in vote_right1976{
foreach y3 in vote_right2006{
foreach x in opp_rate {
foreach z in iv_jmeansent {
ivreg2 `y1' (`x'=`z') $covars vote_fas1921, robust
est store a1
ivreg2 `y1' (`x'=`z') $covars vote_left1921, robust
est store a2
ivreg2 `y1' (`x'=`z') $covars pre_1922_opponents, robust
est store a3
ivreg2 `y2' (`x'=`z') $covars vote_fas1921, robust
est store a4
ivreg2 `y2' (`x'=`z') $covars vote_left1921, robust
est store a5
ivreg2 `y2' (`x'=`z') $covars pre_1922_opponents, robust
est store a6
ivreg2 `y3' (`x'=`z') $covars vote_fas1921, robust
est store a7
ivreg2 `y3' (`x'=`z') $covars vote_left1921, robust
est store a8
ivreg2 `y3' (`x'=`z') $covars pre_1922_opponents, robust
est store a9

esttab a1 a2 a3 a4 a5 a6 a7 a8 a9 using "ov.tex", replace label cells(b(fmt(3)) ///
se(par fmt(3)) ) stats(N) ///
keep(opp_rate vote_fas1921 vote_left1921 pre_1922_opponents)
}
}
}
}
}


I need to run the Moran test again and this is what I did for vote_right1948:
regress opp_rate iv_jmeansent
predict opp_rate_hat
regress vote_right1948 opp_rate_hat $covars vote_fas1921

But I get different results than the ivreg2. Am I missing something?

[Øyvind Snilsberg]

include covariates in both stages,

regress opp_rate iv_jmeansent $covars vote_fas1921
predict opp_rate_hat
regress vote_right1948 opp_rate_hat $covars vote_fas1921