It's not uncommon for the statistical significance of independent variables to change when using the iv() option in system-GMM estimations like xtabond2 or xtdpdgmm. This is due to the instrumental variables helping to address endogeneity issues, which can lead to more accurate coefficient estimates and improved significance levels. Using instrumental variables can help control for omitted variable bias and measurement error, resulting in more reliable results.

Thank you for your help. I'm new to system-GMM, and am trying to figure out many things. Do you see something wrong in the following command? In the command , "smei4" shows up both as an independent variable and as an instrumental variable in iv(). Is this command problematic? Do I need to lag the instrumental variable like "iv(smei4,lag(1 1) model(level))"? Is it also legitimate to use it like "iv(smei4,lag(0 1) model(level))" or is it problematic again?

Thank you for your kind help.

Best,

Taka

"xtdpdgmm gdpgrow smei4 inflation gfcfgrow hfcegrow tradeopen l.lrgdpopc if id~=10,gmm(gdpgrow inflation gfcf > grow hfcegrow , lag(2 2) collapse model(diff)) gmm(gdpgrow inflation gfcfgrow hfcegrow , lag(1 1) diff colla > pse model(level)) iv(smei4,model(level)) one vce(cl id) small"

We used conventional statistical methods in analysis of findings rather than any particular approaches. Perhaps it would be good for you to contact advocates of particular approaches to answer your questions.