Hey everyone,

Thank you for reading my question. It is related to a "commonly-seen" difference-in-differences question (e.g. https://www.statalist.org/forums/for...ferences-model). I understand that if Post variable is defined by year and the model includes year fixed effect, Post variable will then be dropped. My question is: if the dataset is at month-level and Post is also defined by month, but I only use year FE (for some reason), how do I interpret the coefficient of Post?

For example, my panel data is at firm-month level. The sample period is from 01/2000 to 12/2010. There is a shock in 06/2005. I want to run a DiD test on this shock. Hence, I run the following in Stata:

Code:
reghdfe y c.post##c.treat, absorb(firm year) vce(cluster firm year)
I understand ideally I would like to control for month FE, but supoose I only use year FE. In this case, Post will not be dropped. I have two questions:

(1). Should I interprete the coefficient of Post as the difference between the first half and seond half of year 2005? That is, it captures the within-year difference because of year fixed effects?

(2). How about the interaction term? Is it also a within-year difference? That is, comparing the difference between treatment group and control group in the first half of year 2005 with the difference in the second half of year 2005?



Thanks in advance for helping me!

Best,
Jacob