I ran the xtabond2 command using Stata 11 and got all the results needed for different models. I saved all the commands in a do file.
I am trying to estimate the same models using the same commands and data set, but using Stata 13. Stata 13 omits the most important variable due to collinearity although it was not omitted when using Stata 11.

What might be the reason behind this?

Your help is highly appreciated