Applying Callaway & Sant'Anna (2021) Staggered DiD with Cycles Instead of Years

[Armande Mahabi]

Hello,

I am applying the staggered Difference-in-Differences (DiD) approach by Callaway & Sant'Anna (2021) to study the impact of field officer changes on Savings Groups (SGs). Unlike standard applications where time is measured in years, SGs operate in cycles (each lasting 9–12 months). My supervisor raised concerns that using cycles instead of years might create methodological issues. Panel Setup:

• Panel variable: ID (unbalanced)
• Time variable: Year (2014 to 2018), but with gaps
• Delta: 1 unit

I have created a cycle variable to represent the group’s operational timeline. Main Questions:

1. Can the Callaway & Sant’Anna (2021) method be applied when the time variable represents cycles instead of calendar years?
2. Are there any known methodological adjustments needed when applying this approach in a non-yearly time framework?

I am implementing this in Stata using csdid. Any guidance on how to properly structure my model in this context would be greatly appreciated!

Thank you!

[FernandoRios]

my suggestions. Make your Cycle start , say at 10.
Use that to define when a group is is treated for the first time.
After that, apply csdid as usual

[Armande Mahabi]

Dear Fernando,

Thank you for your response. I appreciate your suggestion about redefining the cycle variable. However, I would like to better understand the reasoning behind this approach. If I start the cycle at 10, does this only serve as a relabeling, or does it address a specific issue related to the structure of the data in csdid? I am unsure how this adjustment would affect the analysis and identification strategy.
I am relatively new to this method and want to ensure I apply it correctly. Could you kindly elaborate on how this adjustment interacts with csdid’s estimation process?