I am examining the effects of Covid on the computer-use wage premium in the USA. I have individual-level panel data and have run the following regression:

eststo clear
eststo: xtreg lwage c.Computer##b15.Date, fe vce(robust)
matrix list e(b)
coefplot est1, vertical drop(_cons) keep (*.Date#c.Computer) yline(0)

lwage --> log of hourly wages

c.Computer --> continuous variable for the importance of computer-use in an occupation

b15.Date --> date (month and year), with the 15th month in the sample (March 2020) set as the excluded category when the Covid-19 shock "hit"

I want to end up with diff-in-diff graphs as shown in the attachments "Example 1" and "Example 2", where the effects are all normalised relative to March 2020, the excluded month. I was wondering what the best way to produce these sorts of graphs in Stata is. Currently, I am using Coefplot, but that doesn't seem to achieve the results seen in the examples where the excluded month when the "treatment" (Covid-19 shock) takes place is set at zero, with a point. Instead, it appears that month 15 is excluded from the graph completely.

My current results are also attached. 

Many thanks in advance,

Charles

3 attachments