Difference in difference problem.


I want to examine the effects of a legislative amendment on limited liability companies. I am using a binary variable as my dependent variable, and are measuring the effect of the amendment by using a ratio of gross profit before and after the ”treatment”. I have panel data containing information from 2009-2012, and the legislative amendment was introduced in mid 2011. My initial thought was to use a binary logit function with a difference in difference estimator as the independent variable estimating the effect of the amendment (of course controlling for company size, costs etc.)

By using the D2. function STATA calculates the difference in say (2010 – 2011) - (2011-2012). My first question is: if the variable turns out to be statistical significant, does the command take into account that the difference might be caused by difference between the control groups in the first time period (before the treatment)? How can i control for this initial difference?

Thanks for any advice on this matter!