黃河泉 发表于 2017-1-8 16:46 
请精确一点,我不知道你在问哪里的问题?
不好意思,第一句是在程序前面的介绍里面
1. 2-step FD-GMM for dynamic threshold panel
i. Can allow for threshold effect in the intercept as well
ii. averaging involves some randomness in the estimates
iii. choice of IV can alter the estimates. See procedure setiv
第二句是在后面
x = c~Tq~d; @ set regressors @
// tt= ones(n,1).*.eye(t); tt= tt[.,t0+1:t]; // time dummies. SET _con =0
// x = x~tt; with dummies we need more IV's as they are already part of IVs
q = c;
xx= ones(nt,1)~tq; @ variables to be used as IV: FIRST COLUMN MUST BE 1's@
"threshold variable = c ";
"qn~ns~jm~jn~t0 : " qn~ns~jmy~jn~t0;
我是在tq那里出现的问题,tq是那个Tobin’s Q吗?