Hi. I am trying to use the teffects to implement matching using Nearest Neighbor Matching. The stata syntax is:

teffects nnmatch (ovar omvarlist) (tvar) if in weight , stat options

I have almost 20 variables in the model and I don't want to use all of them for matching. If I put all variables in omvarlist, I am assuming the method will use all variables to find nearest neighbors. Is there any way to use some covariates for matching (step 1) and others for the regression (step 2). Appreciate any suggestions.

Also, on a related note, do you know if NNM prunes/imputes any observations from the data to find matches.

Thanks
Rajat