Treated dummies

[Mins Kim]

I am doing difference in differences analysis.

The time ranges from year 1 to 10 and the regulation started in year 5.

My question is that during the year 1~4, should I give 0 or 1 to the treated firms?

I first gave 0 to the treated firms for 1~4 years, as I thought that they were not under the regulation during that time, but I found some codings that gave 1 to the firms even before the treatment started.

Can someone clarify that?

Best,

[Clyde Schechter]

Assuming you are doing a standard classical DID analysis, you must code the treatment variable 1 for the treated firms in all years and 0 for the untreated firms in all years. You also need a prePpost variable that is 0 for all firms in years 1-4 and 1 for all firms in years 5-10. Then your regression must include all of the variables treat, pre_post, and their interaction. That is most easily done in one fell swoop by:

regression_command outcome i.treat##i.pre_post

If you are not familiar with the ## or i. operator, read -help fvvarlist- to understand what this does.

[Mins Kim]

Dear Clyde,

Thank you so much for your reply.

Now I realize that I can do DID using usual regression command.

Best,