DID + falsification test

[River Huang]

Dear All, Suppose that I have the following data for DID (difference-in-differences) analysis.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(id year d)
1 2010 0
1 2011 0
1 2012 1
1 2013 1
1 2014 1
2 2010 0
2 2011 0
2 2012 0
2 2013 0
2 2014 0
3 2010 0
3 2011 1
3 2012 1
3 2013 1
3 2014 1
4 2010 0
4 2011 0
4 2012 0
4 2013 0
4 2014 1
end

In this case, the times of treatment are different. For instance, for id=1, it is year=2012, and for id=3, it is year=2011, and so on. For falsification test (although, somehow, I have doubt), the common procedure is to redefine `d' with 1 one year before its original starting year of receiving treatment. For instance, now, for id=1, d=1 in year=2011 (and after).

• My question is that is there a general approach to redefine `d' to this end? Thanks.

[Andrew Musau]

xtset id year
bys id (year): replace d=F.d if F.d==1

[River Huang]

Dear Andrew, Many thanks for this helpful suggestion.