Retaining base variable coefficient in regression output

[Ashani Abayasekara]

Hi all,

I'm running DiD regressions where I compare yearly event-time dummies to a base time period. My code is as follows:

xtreg lnwage i.year treat05-treat09 treat11-treat21, fe i(id) r

where treat05-treat09 and treat11-treat21 refer to the event time dummies which are compared against treat10 (omitted variable).

The issue is that I want the coefficient and standard error of treat10 (both 0 in this case) to be reported in the regression output as well, so that when I use subsequent commands to graph the effects, I can demonstrate a 0 value and 0 confidence interval for treat10, the base year. How can I change the code such that the output for treat10 is also recorded in the regression output, while still retaining it as the base variable?

Thank you.

[Maxence Morlet]

If treat10 is the reference period, you cannot get a point estimate nor a standard error for it by definition. Coefficients for the other treat dummies are calculated in relation to it, so it must be omitted. Coefficients will be relative to the base year.