Hi all,

I am currently working on a dissertation to assess the state level of tax impact on marijuana uses with Tax (dummy) being independent variable, Unemployment, Per Capita Income and Numbers of people in the state with high school education and above as my controlled variables and State GDP as IV. I conducted a System GMM on my regression and the result was as follows:

One Step System GMM
  • 1 lag on my dependent variable ln marijuana has a AR(2) of 0.001 while Hansen Test being 0.394
  • 2 lags on my dependent variable ln marijuana has a AR(2) of 0.013 while Hansen Test being 0.001
  • 3 lags on my dependent variable ln marijuana has a AR(2) of . while Hansen Test being 0.528
Two Step System GMM
  • 1 lag on my dependent variable ln marijuana has a AR(2) of 0.001 while Hansen Test being 0.394
  • 2 lags on my dependent variable ln marijuana has a AR(2) of 0.008 while Hansen Test being 0.132
  • 3 lags on my dependent variable ln marijuana has a AR(2) of . while Hansen Test being 0.809
So does it mean that I have to stop at 2 lags dependent as my AR(2) Pr > z become . with additional lag? I also understand that Hansen Test is important in addressing the research question on if its meaningful, is there anyway I can improve my model by including more variable?