Dear all,

I run in stata my model (full sample) with system GMM to deal with endogeneity issue. When I divide in last section of my paper my sample into sub-samples according to an index thresholds, it give me 7 countires than 10 countires than 32 ....what is the appropriate technique for those small samples shoold i choose, cause I cannot apply for GMM which number instruments exceed groups and the results of coefficients and tests are ommitted. Shoold I apply for 2sls ? Or 3sls? or ? Thanks.