Hi, I’m using csdid with repeated cross-sectional data (Peru, 2008–2016) to estimate the effect of an education reform.
Only public schools are treated, and all of them start being treated in 2014. Private schools are never treated.
I defined:
I don’t use ivar(), since I don’t have consistent unit IDs over time.
Where treatment = 1 for public schools and 0 for private schools (controls)
My goal is to produce an event-study graph.
One issue I’m facing is that when using estat event, the year before treatment (2013) is not omitted — I expected it to be the reference period (t = -1).
Is this definition of gvar valid for csdid, or is there a better way to do this without a true panel structure?
Thanks in advance!!!!!!
Only public schools are treated, and all of them start being treated in 2014. Private schools are never treated.
I defined:
Code:
gen g = . replace g = 2014 if treatment == 1 replace g = 0 if treatment == 0
Where treatment = 1 for public schools and 0 for private schools (controls)
My goal is to produce an event-study graph.
One issue I’m facing is that when using estat event, the year before treatment (2013) is not omitted — I expected it to be the reference period (t = -1).
Is this definition of gvar valid for csdid, or is there a better way to do this without a true panel structure?
Thanks in advance!!!!!!
