[lilis husna]

thanks Jared Greathouse, I think it's clear enough about the model. So, can I take coefficient of interacting variabel as DiD coefficient?

[Jared Greathouse]

Yep

[lilis husna]

Jared Greathouse i got different result when I run these two models

 
 reghdfe pxikc tarifi ftai A sizet, absorb(CT IK)

and

 
 reghdfe pxikc tarifi ftai A sizet, absorb(country#year firm#hs)

which one is correct?

[Jared Greathouse]

In this business, very few things (well, many things) aren't so much "incorrect" as they are, rather, unwise, in a given situation.

Your model includes two sets of fixed effects in the first one, and it includes one way interactions (why?) betwixt country and time and firm and whatever HS means here in the second case, so no wonder the effects will be different. Which one is "wise" depends on the situation. I don't know how to even interpret 1 way interactions, so I can't comment on whether this makes sense here or not. So, much of this isn't for me to decide, it's between you and whoever else knows enough about the research question.