Dear members,

I am struggling in modeling a panel data with lag variables in Stata. I would like some help on what should I use on my panel data.
I tried the three equations below, theoretically they seem to be the same equation to me, but why the results are all different?
I read some panel VAR materials and papers, but I have trouble finding the right answer.
Which one is correct to use or can I use all of them?

1) pvar y1 y2 y3 y4, lag(1) instl(1/5) gmmstyle overid exog(x1 x2 x3) fod vce(cluster id)
2) pvar D.y1 D.y2 D.y3 D.y4, instl(1/5) gmmstyle overid exog(D.x1 D.x2 D.x3) fod vce(cluster id)
3) pvar D.y1 D.y2 D.y3 D.y4, lag(1) instl(1/5) gmmstyle overid exog(D.x1 D.x2 D.x3) fod vce(cluster id)