Hello Brant! First of all, a big thank you to you and your co-author Pedro for this package and your work on DiD with multiple time periods. I am an econ grad student and I will be writing my master's thesis this semester using your approach - difference-in-differences with multiple time periods. I will be looking at firearm purchase delay laws and how they affect handgun-related suicides in the United States.
However, I have two questions:
I am a bit confused when thinking about the correct regression equation when using this model. How would my regression equation differ from a normal DiD regression equation? For example: ln(HandgunSuicides)_st = \alpha + \beta_1(Delay_s x Post_st) + \beta X_s + + \lambda_s + \mu_t + \epsilon_st
Is it problematic that I only have 1 treatment state per cohort? I.e., I have 4 'treatment' states that each change treatment status at different points in time (2004,2007,2012,2014). I have a large group of 'never-treated' units (around 15-20 states).
Hello Brant! First of all, a big thank you to you and your co-author Pedro for this package and your work on DiD with multiple time periods. I am an econ grad student and I will be writing my master's thesis this semester using your approach - difference-in-differences with multiple time periods. I will be looking at firearm purchase delay laws and how they affect handgun-related suicides in the United States.
However, I have two questions:
Looking forward to hearing from you.
Thank you very much again!
Best regards,
Luka Parisi Tilburg University