Dear Statalist,

I have a doubt regarding construction of a variable. I am new to stata. I will be highly obliged if someone can solve my doubt.

I am working with panel dataset and my sample has 60000 observations over the time time period 2009-2020.

I am interested to look at the impact of a policy (reform) that was implemented in 2016. I will be using differences-in-difference estimation strategy for this purpose.

I am having some problems regarding the construction of my treatment variable and hence my post.

My treatment variables will be dummy variable for firms taking the value 1 for firms having a high average product of capital (arpk) and zero otherwise. The way I want to construct my treatment variable is as follows:

To determine whether firms had a high arpk or low arpk prior to the reform, I want to average each firm's measure of arpk over 2012-2015 (the last year prior to the policy change). I will then classify a firm as high arpk if its average is above the industry level median.

Note that the firm identifier is prowesscompanycode, the industry identifier is nic, and the year identifier is year, respectively.

I tried constructing this variable on my own but not getting the desired results. I will be highly obliged if someone could help me with this.


I hope I am able to explain my doubt. Any more information is required I will be happy to share.

Thanking you in advance,
Regards,
Sarthak.