csdid omits cohort in the estimation

[Marco Errico]

Dear all,

I am trying to use csdid command. My treatment is staggered over time across firms.
My firms are identified with -cik- varibale
The treatment is implemented over three years (2009-2020-2011)- I.e. some firms receive on 2009, other 2009 and last cohort on 2011.
I identified when the firm has been treated with the following code:

egen gvar=csgvar(treatclean), tvar(calendar_yr) ivar(cik)

Then I run the estimation as follow

csdid ESGScore , t(calendar_yr) g(gvar) i(cik)

Where ESGScore identfies my outcome of interest


The issue is the outpout. I have an estimate for 2009, but 2010 and 2011 are omitted.
I'm wondering what could be the driver?

This below the output when I run

estat all

after csdid command

I tag FernandoRios that might help.

Below an example of my dataset

* Example generated by -dataex-. For more info, type help dataex
clear
input double ESGScore long cik double calendar_yr float(treatclean1 treatclean2 treatclean3 gvar)
               .   20 2000 0 0 0    0
               .   20 2001 0 0 0    0
               .   20 2002 0 0 0    0
               .   20 2003 0 0 0    0
               .   20 2004 0 0 0    0
               .   20 2005 0 0 0    0
               .   20 2006 0 0 0    0
               .   20 2007 0 0 0    0
               .   20 2008 0 0 0    0
               .   20 2009 0 0 0    0
               . 1750 2000 0 0 0 2011
               . 1750 2001 0 0 0 2011
               . 1750 2002 0 0 0 2011
               . 1750 2003 0 0 0 2011
               . 1750 2004 0 0 0 2011
               . 1750 2005 0 0 0 2011
               . 1750 2006 0 0 0 2011
               . 1750 2007 0 0 0 2011
               . 1750 2008 0 0 0 2011
               . 1750 2009 0 0 0 2011
               . 1750 2010 0 0 0 2011
               . 1750 2011 0 0 1 2011
               . 1750 2012 0 0 1 2011
               . 1750 2013 0 0 1 2011
               . 1750 2014 0 0 1 2011
               . 1750 2015 0 0 1 2011
24.8875342412315 1750 2016 0 0 1 2011
25.0066277591702 1750 2017 0 0 1 2011
23.8684421018747 1750 2018 0 0 1 2011
27.7347922458799 1750 2019 0 0 1 2011
24.6719089687439 1750 2020 0 0 1 2011
               . 1800 2000 0 0 0 2009
               . 1800 2001 0 0 0 2009
39.8570323185223 1800 2002 0 0 0 2009
40.6709249900856 1800 2003 0 0 0 2009
27.3570272742741 1800 2004 0 0 0 2009
69.2289921627812 1800 2005 0 0 0 2009
57.0094851221679 1800 2006 0 0 0 2009
64.6477562824319 1800 2007 0 0 0 2009
73.3858614355101 1800 2008 0 0 0 2009
77.1656746514662 1800 2009 1 0 0 2009
81.8918738745576 1800 2010 1 0 0 2009
83.6335988366222 1800 2011 1 0 0 2009
80.8784665162317 1800 2012 1 0 0 2009
73.8408140658563 1800 2013 1 0 0 2009
68.1007655933921 1800 2014 1 0 0 2009
75.5692216335206 1800 2015 1 0 0 2009
79.4495985961115 1800 2016 1 0 0 2009
82.2468782407938 1800 2017 1 0 0 2009
77.9907365272183 1800 2018 1 0 0 2009
81.4002450976869 1800 2019 1 0 0 2009
70.6894083238113 1800 2020 1 0 0 2009
               . 1923 2000 0 0 0    0
               . 1923 2001 0 0 0    0
               . 1923 2002 0 0 0    0
               . 1923 2003 0 0 0    0
               . 1923 2004 0 0 0    0
               . 1923 2005 0 0 0    0
               . 1923 2006 0 0 0    0
               . 1923 2007 0 0 0    0
               . 1923 2008 0 0 0    0
               . 1923 2009 0 0 0    0
               . 1923 2010 0 0 0    0
               . 1961 2011 0 0 0 2011
               . 1961 2012 0 0 0 2011
               . 2062 2000 0 0 0    0
               . 2062 2001 0 0 0    0
               . 2062 2002 0 0 0    0
               . 2062 2003 0 0 0    0
               . 2062 2004 0 0 0    0
               . 2098 2000 0 0 0 2011
               . 2098 2001 0 0 0 2011
               . 2098 2002 0 0 0 2011
               . 2098 2003 0 0 0 2011
               . 2098 2004 0 0 0 2011
               . 2098 2005 0 0 0 2011
               . 2098 2006 0 0 0 2011
               . 2098 2007 0 0 0 2011
               . 2098 2008 0 0 0 2011
               . 2098 2009 0 0 0 2011
               . 2098 2010 0 0 0 2011
               . 2098 2011 0 0 1 2011
               . 2098 2012 0 0 1 2011
               . 2098 2013 0 0 1 2011
               . 2098 2014 0 0 1 2011
               . 2098 2015 0 0 1 2011
               . 2098 2016 0 0 1 2011
               . 2098 2017 0 0 1 2011
               . 2098 2018 0 0 1 2011
               . 2098 2019 0 0 1 2011
25.9223976667432 2098 2020 0 0 1 2011
               . 2135 2001 0 0 0    0
29.8184301731622 2135 2002 0 0 0    0
34.0214333100352 2135 2003 0 0 0    0
30.3390743471523 2135 2004 0 0 0    0
41.6717460153372 2135 2005 0 0 0    0
 53.879438568687 2135 2006 0 0 0    0
24.2843001639474 2135 2007 0 0 0    0
49.6341419871165 2135 2008 0 0 0    0
28.9786856496672 2135 2009 0 0 0    0
end
format %ty calendar_yr

[FernandoRios]

Hi there
try method(reg)
since you have no controls it will be faster and more reliable with the same numerical results

[Marco Errico]

Hi Fernando,

Thanks. Unfortunately I do have controls and here I just show you for simplicity basic with only dependant variable.

Do you have any suggestion?

[FernandoRios]

In that case, 1) try dripw. Its more stable that drimp, the current default
2) it may be that once you add controls you lose more data.
3) if the problem persists, please send me an email with a replicable example, and i will take a look.
Fernando