Question regarding lag in a time series/panel data analyses

[jannis bauer]

Hello,

I have a question regarding lags in a panel data regression
I want to examine the effect of an event happened. I suspect that the effect of the event is realized later and so would I like to do a regression with a lag of one period.
So if I think the effect of my independent variable occurs later, do I shift the dependent variable back or forward ?

Below I wrote down both examples, which one would be correct ?

gen Revenues_lag1 = Revenues[_n-1]   gen Revenues_lag1 = Revenues[_n+1]

[Jared Greathouse]

Don't bother doing it like this, do the equivalent of

reg y L.Revenues X

[jannis bauer]

do i still have to generate the lag variable ?
Or do I just regress it like you said ?
It is also a Panel Data Analyses so it is Xtset

[Jared Greathouse]

No, Stata's time series notation (the one I used) does this for you.