ivreg2 estimated, but results not displayed

[Mina Wu]

Hi,

I am trying to implement gmm using ivreg2 command. I first label the endogenous and exogenous variables, as well as instrument. Afterwards, I run the ivreg2 regression, by groups (industry_2d). The commands are below:

* Defining vectors of exogenous, endogenous, and instrumental variables.
global exoreg /*
*/ lnk ln_K_l1 ln_M_l1 /*
*/ km_l1 k2_l1 m2_l1 k2m_l1 km2_l1 k3_l1 m3_l1
global endoreg lnm lnl
global instr ln_L_b_l1 ln_M_l2

sort industry_2d

statsby _b _se size=e(N), by(industry_2d) saving(tmp_WLP_GO_M_1, replace): /*
*/ ivreg2 lnrev $exoreg ($endoreg = $instr), gmm cluster(id)

However, I obtain the following results(no results, but the commands seem to be applied well):

.         statsby _b _se   size=e(N), by(industry_2d) saving(tmp_WLP_GO_M_1, replace): /* 
>          */ ivreg2 lnrev $exoreg ($endoreg = $instr), gmm cluster(id)
(running ivreg2 on estimation sample)

      command:  ivreg2 lnrev lnk ln_K_l1 ln_M_l1 km_l1 k2_l1 m2_l1 k2m_l1 km2_l1 k3_l1
                    m3_l1 (lnm lnl = ln_L_b_l1 ln_M_l2), gmm cluster(id)
    _eq2_size:  e(N)
           by:  industry_2d

Statsby groups
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50

Does anyone know what I am doing wrong?
Many thanks in advance.

[River Huang]

Did you have a look at the file "tmp_WLP_GO_M_1.dta"?

[Mina Wu]

I did. And contains the following variables, per industry:

indus~2d | _b_lnm | _b_lnl | _b_lnk | _b_ln_K~1 | _b_ln_M~1 | _b_km_l1 | _b_k2_l1 | _b_m2_l1 | _b_k2m_l1 | _b_km2_l1 | _b_k3_l1 | _b_m3_l1 | _b_cons |
_se_lnm | _se_lnl | _se_lnk | _se~K_l1 | _se~M_l1 | _s~km_l1 | _s~k2_l1 | _se_m2~1 | _s~2m_l1 | _se_km2~ | _se_k3~1 | _se_m3~1 | _se_cons | _eq2_s~e |

[River Huang]

Roughly, it works in the following example.

use http://fmwww.bc.edu/ec-p/data/hayashi/griliches76.dta, clear

statsby _b _se size=e(N), by(smsa) saving(smsa.dta, replace): ///
ivreg2 lw s expr tenure rns smsa i.year (iq=med kww age mrt), gmm2s robust

use smsa.dta, clear

Thus, my first suggestion is to use "gmm2s" instead of "gmm". If it still does not work, please remove the options "gmm cluster(id)", and re-run to see if it generates the desired results.

[Mina Wu]

River Huang, thanks!