Dear,

I have used two commands, ivreg2 and ivregress, to estimate the same model. errors are assumed to be hetoroskedastic and autocorrelated.

However, the results of the two command is different. Can anyone help me solve this problem?

Sincerely yours,
Trung

HTML Code:
. qui ivreg2 ipolicy F.pigap F.pigapcris L.pigap L.pigapcris ygap ygapcris (L.ipolicy = l(2/2)..ipolicy l(2/2)..pigap l(1/
> 1).ygap), ///
>          gmm2s robust bw(4) kernel(bartlett) first

.          predict res0, res
(2 missing values generated)

. estimate store ivreg2

. 
. qui ivregress gmm ipolicy F.pigap F.pigapcris L.pigap L.pigapcris ygap ygapcris (L.ipolicy = l(2/2)..ipolicy l(2/2)..pig
> ap l(1/1).ygap) , ///
>         vce(hac bartlett 4) first

. estimate store ivregress

. esttab ivreg2 ivregress

--------------------------------------------
                      (1)             (2)   
                  ipolicy         ipolicy   
--------------------------------------------
L.ipolicy           0.913***        0.911***
                  (23.80)         (24.63)   

F.pigap             0.530***        0.535***
                   (4.88)          (5.27)   

F.pigapcris        -0.235          -0.241   
                  (-1.61)         (-1.70)   

L.pigap            -0.379**        -0.382** 
                  (-2.88)         (-3.19)   

L.pigapcris         0.229           0.237   
                   (1.22)          (1.34)   

ygap              -0.0467          -0.117   
                  (-0.00)         (-0.01)   

ygapcris            13.59           13.28   
                   (1.33)          (1.32)   

_cons               0.877           0.892   
                   (1.63)          (1.66)   
--------------------------------------------
N                      65              65   
--------------------------------------------
t statistics in parentheses
* p<0.05, ** p<0.01, *** p<0.001