Choosing a baseline and examining when the effect occurs

[Ann Ngo]

Dear Statalist experts,

I would like to post a question. Below is a part of my panel data (xtset firm_id month). There is an intervention happening in month 4, which affects UK firms. So I have "treated" = 1 for UK firms, and = 0 for non-EU firms. My goal to test the effect of this intervention on the variable "esg", specifically pinpointing when the effect began for treated firms. So I want to pick month 4 as baseline, and aim to observe the effect in each month (instead of just creating a dummy time variable = 1 for post-intervention and 0 for pre-intervention). I also include country and time fixed-effects.

Could you please help me with the relevant codes? I have tried several codes but not sure which one is correct. I understand that I can create dummy variables for each month, but that would make the interactions complex (e.g., month_dummy*treated).

Also, what is the difference if I use xtset and reghdfe?

I am hoping for your assistance. Thank you very much.

input str10 firm_id byte month str14 country float esg byte treated
"FS00008NOQ" 1 "Germany"        22.29 0
"FS00008NOQ" 2 "Germany"        22.36 0
"FS00008NOQ" 3 "Germany"        22.55 0
"FS00008NOQ" 4 "Germany"        22.58 0
"FS00008NOQ" 5 "Germany"        22.54 0
"FS00008NOQ" 6 "Germany"        22.77 0
"FS00008NOQ" 7 "Germany"        23.16 0
"FS00008WGQ" 1 "Germany"        20.03 0
"FS00008WGQ" 2 "Germany"         19.9 0
"FS00008WGQ" 3 "Germany"        20.84 0
"FS00008WGQ" 4 "Germany"        20.75 0
"FS00008WGQ" 5 "Germany"        20.77 0
"FS00008WGQ" 6 "Germany"        20.78 0
"FS00008WGQ" 7 "Germany"        19.38 0
"FS00008N1D" 1 "Austria"        22.94 0
"FS00008N1D" 2 "Austria"        22.66 0
"FS00008N1D" 3 "Austria"        23.38 0
"FS00008N1D" 4 "Austria"        23.34 0
"FS00008N1D" 5 "Austria"        23.24 0
"FS00008N1D" 6 "Austria"        23.08 0
"FS00008N1D" 7 "Austria"        22.92 0
"FS00008LRJ" 3 "France"         25.22 0
"FS00008LRJ" 4 "France"         24.98 0
"FS00008LRJ" 5 "France"          24.5 0
"FS00008LRJ" 6 "France"         24.74 0
"FS00008LRJ" 7 "France"         25.56 0
"FS0000DCP4" 1 "Italy"          23.85 0
"FS0000DCP4" 2 "Italy"          23.74 0
"FS0000DCP4" 3 "Italy"          24.83 0
"FS0000DCP4" 4 "Italy"          24.78 0
"FS0000DCP4" 5 "Italy"          24.86 0
"FS0000DCP4" 6 "Italy"          24.57 0
"FS0000DCP4" 7 "Italy"          24.24 0
"FS00009EMM" 1 "Spain"          23.57 0
"FS00009EMM" 2 "Spain"          23.46 0
"FS00009EMM" 3 "Spain"          24.57 0
"FS00009EMM" 4 "Spain"          24.55 0
"FS00009EMM" 5 "Spain"          24.56 0
"FS00009EMM" 6 "Spain"          24.34 0
"FS00009EMM" 7 "Spain"          24.52 0
"FSGBR05750" 1 "United Kingdom" 23.83 1
"FSGBR05750" 2 "United Kingdom"  23.8 1
"FSGBR05750" 3 "United Kingdom" 23.93 1
"FSGBR05750" 4 "United Kingdom" 23.55 1
"FSGBR05750" 5 "United Kingdom" 23.48 1
"FSGBR05750" 6 "United Kingdom" 23.33 1
"FSGBR05750" 7 "United Kingdom" 22.34 1
"FS0000BPTE" 1 "United Kingdom" 23.86 1
"FS0000BPTE" 2 "United Kingdom" 23.44 1
"FS0000BPTE" 3 "United Kingdom" 24.26 1
"FS0000BPTE" 4 "United Kingdom" 24.12 1
"FS0000BPTE" 5 "United Kingdom" 24.18 1
"FS0000BPTE" 6 "United Kingdom"  23.9 1
"FS0000BPTE" 7 "United Kingdom" 23.52 1
"FS0000EG0J" 1 "United Kingdom" 23.84 1
"FS0000EG0J" 2 "United Kingdom" 23.76 1
"FS0000EG0J" 3 "United Kingdom" 24.29 1
"FS0000EG0J" 4 "United Kingdom" 24.32 1
"FS0000EG0J" 5 "United Kingdom" 24.16 1
"FS0000EG0J" 6 "United Kingdom" 23.84 1
"FS0000EG0J" 7 "United Kingdom" 23.59 1
"FS0000CLMN" 1 "United Kingdom" 24.41 1
"FS0000CHAG" 1 "United Kingdom" 25.94 1
"FS0000CHAG" 2 "United Kingdom" 25.75 1
"FS0000CHAG" 3 "United Kingdom" 26.33 1
"FS0000CHAG" 4 "United Kingdom" 26.42 1
"FS0000CHAG" 5 "United Kingdom" 26.29 1
"FS0000CHAG" 6 "United Kingdom" 26.16 1
"FS0000CHAG" 7 "United Kingdom" 26.19 1
"FSGBR059ON" 1 "United Kingdom" 25.13 1
"FSGBR059ON" 2 "United Kingdom" 25.03 1
"FSGBR059ON" 3 "United Kingdom" 25.93 1
"FSGBR059ON" 4 "United Kingdom" 25.83 1
"FSGBR059ON" 5 "United Kingdom" 25.74 1
"FSGBR059ON" 6 "United Kingdom" 25.51 1
"FSGBR059ON" 7 "United Kingdom" 25.28 1
"FS0000FHDL" 1 "United Kingdom" 29.95 1
"FS0000FHDL" 2 "United Kingdom" 29.84 1
"FS0000FHDL" 3 "United Kingdom" 30.32 1
"FS0000FHDL" 4 "United Kingdom" 30.62 1
"FS0000FHDL" 5 "United Kingdom"  30.6 1
"FS0000FHDL" 6 "United Kingdom" 30.45 1
"FS0000FHDL" 7 "United Kingdom"  30.6 1
"FSUSA0A8KV" 1 "United Kingdom" 28.84 1
"FSUSA0A8KV" 2 "United Kingdom" 28.83 1
"FSUSA0A8KV" 3 "United Kingdom" 29.58 1
"FSUSA0A8KV" 4 "United Kingdom" 29.78 1
"FSUSA0A8KV" 5 "United Kingdom" 29.74 1
"FSUSA0A8KV" 6 "United Kingdom"  29.4 1
"FSUSA0A8KV" 7 "United Kingdom" 29.94 1

[George Ford]

g post = month>4
reghdfe esg c.treated#c.post#b4.month , absorb(firm_id month)

[Ann Ngo]

g post = month>4
reghdfe esg c.treated#c.post#b4.month , absorb(firm_id month)

Thank you for your reply.

When I ran the code, the first 4 months were omitted. Is this how it works? How can I find the effect before the intervention? Because the firms might know and react to the intervention before it happened. I would appreciate if you could help me further with this matter, please.

[George Ford]

Yes. You said you aren't interested in the pre period coefficients.

[Ann Ngo]

Yes. You said you aren't interested in the pre period coefficients.

Could you please guide me on how to examine the coefficients of pre-period time as well? Because there might happen that the firms are aware of the intervention before it took effect, so they might react in advance. In general, can I observe the coefficients for all 7 months?
Thank you very much for your support. I am quite struggling with this issue.

[George Ford]

 
 reghdfe esg c.treated#b1.month , absorb(firm_id month)

[Ann Ngo]

reghdfe esg c.treated#b1.month , absorb(firm_id month)

Thank you very much for your assistance. May I clarify this a bit further? As I expect that the coefficients of other months would be relative to month 4 (e.g., pre month 4 esg's maybe expected to have negative coefficients and post would have positive coefficients), I aim to omit month 4 from the result. How should I modify the codes?

[George Ford]

This sets 4 as the base

reghdfe esg c.treated#b4.month , absorb(firm_id month)

Not much reason to exclude month 4 since you have monthly fixed effects. But to do so,

 reghdfe esg c.treated#b1.month if month!=4, absorb(firm_id month)

Not that the coefficients are differences from the base (here month 1). For you, I think this makes the most sense:

 reghdfe esg c.treated#b4.month , absorb(firm_id month)

[Ann Ngo]

Thank you soooo much for your patience to help me with this matter. I have learned a lot from the codes. Your support is invaluable.