xtabond2 - system GMM - high p-values and Hansen-test - how can i get less instruments and make my model more specific

[Lorenz Franz]

Hello together,

for a bachelor thesis I am trying to analyze the effect of entrepreneurial education on economic growth. For this I wanted to use a GMM model with the xtabond2 command.

My dependent variable: ln(GDPperCapita)
explanatory variable: BasicSchoolEntrepreneurialEducation, PostSchoolEntrepreneurialEducation
control variables: R&D_Expenditure, GrossCapitalFormation, Unemployment, GovernmentExpenditure, PopulationGrowth.

Im trying to analyze developed and underdeveloped countries separated and I have a problem for the model for the underdeveloped countries.

My command is the following:

xtabond2 lnGDPperCapita L.lnGDPperCapita Basicschoolentrepreneurialedu Postschoolentrepreneurialeduc R_D GrossCapital Unemploymenttotaloftotal Government Populationgrowthannual, gmm(L.lnGDPperCapita,laglimit(1 1) collapse)  iv(Basicschoolentrepreneurialedu Postschoolentrepreneurialeduc  R_D GrossCapital Unemploymenttotaloftotal Government Populationgrowthannual)  nodiffsargan twostep orthogonal small robust

I was testing to reduce the p-values of my variable and of the Hansen test by changing the laglimit, but unfortunately my results are still insignificant:

Can someone help me to adapt my command, so my variables get significant and the Hansen-test gets lower as well.

Many thanks in advance