Marginsplot change "reference" year

[Tony Liu]

Hi, I am using the function marginplots to plot margins for treated (toxic1=1) vs control (toxic1=0) over years from 2000 to 2020. Currently, the default of the margins command is to use 2000 (the beginning of my sample) as the reference year, therefore have missing margins for 2000 and shows up in the marginsplot like this (note the missing element in year 2000).

But what I want is to have the reference year to be 2007 (i.e. no margins estimate for 2007 and missing element in marginsplot for 2007). How can I accomplish that? The following is the current data and code I used.

reghdfe log_debt i.year##toxic1, absorb(i.insee)
margins i.year#toxic1 if year>=2000
marginsplot,xlabel(,angle(vertical)) title("")

* Example generated by -dataex-. For more info, type help dataex
clear
input float(log_debt year toxic1) double insee
 12.32517 2013 1 33063
11.971442 2013 0 67482
11.811516 2013 0 92026
  12.9581 2014 0 69123
11.982915 2012 1 93051
 12.87717 2013 0 69123
12.342184 2012 0 34172
 11.03804 2020 0 92060
11.149103 2012 0 92026
12.198172 2014 0 93048
11.988816 2013 1 13005
 12.49912 2013 1 95018
12.316158 2014 0 67482
12.220356 2019 0 67482
12.054286 2020 0 97415
  12.2634 2015 1 68224
12.331805 2012 1 95018
 8.369261 2017 0 92040
  12.3976 2020 0 67482
 8.067302 2020 0 92040
12.200433 2019 0 34172
12.093436 2013 0 93048
12.510007 2020 1 33063
12.162685 2019 1 92063
12.558186 2014 1 33063
11.632102 2019 1 94076
  11.9771 2012 1 93070
 11.20678 2019 0 59606
12.117535 2012 1  6004
11.174264 2014 0 94002
12.309046 2014 1 97209
11.358448 2019 0 82121
  11.9289 2019 1 93005
 11.23069 2017 0 83137
11.032966 2015 1  3185
11.422158 2015 0 92023
12.904583 2020 0 69123
11.855364 2016 0 92025
11.028974 2019 0 11262
 11.28666 2013 1 79191
11.509714 2013 0 72181
13.131512 2019 1  6088
 11.03228 2014 0 92023
12.415907 2012 1 76351
11.982006 2013 1 63113
11.100784 2013 0 90010
11.074895 2017 0 97416
11.827296 2013 0 92025
10.659093 2016 0 97408
10.677785 2013 0 84031
12.482695 2014 1 76351
10.324215 2013 0 74012
 11.48608 2013 1 94033
11.240614 2012 1 94033
11.190235 2013 1 78440
 10.05102 2013 0 13103
 11.36747 2019 0 64445
11.268244 2013 0 19031
11.792292 2015 1 93006
 11.50492 2014 0 93010
 11.39246 2014 1 78440
12.500934 2017 1 76351
 11.47896 2012 1 93005
11.203295 2017 0 92060
10.603282 2013 0 93027
 11.35412 2012 0 72181
11.961753 2016 0 92050
13.118523 2016 1  6088
11.067586 2014 0 56260
 11.67479 2015 1 93066
11.961278 2013 1 68224
 10.60875 2014 1 44162
10.240132 2018 0 92049
11.998708 2020 1 93005
13.174577 2013 1  6088
12.090904 2019 1 63113
12.201494 2012 1 33063
11.533978 2019 1 28085
10.052868 2019 0 94042
10.851552 2013 0 92023
11.557556 2017 0 34003
  11.4003 2014 0 19031
 11.59707 2013 1 93005
10.617874 2015 1 97302
12.411562 2015 0 67482
10.781545 2014 1 95280
 11.46871 2012 1 91174
 10.62157 2019 1 94038
 11.06794 2013 0 68066
 11.54898 2014 1 94033
11.914995 2012 1 63113
10.256122 2013 1 44162
12.205373 2019 1  6004
 11.34796 2016 1 28085
11.603178 2013 1 94041
11.094548 2017 0 64445
 10.25087 2020 1 94019
 10.81221 2013 0 94080
 9.188096 2019 0  6012
10.920685 2018 1  6083
end

[Ken Chui]

I'm not sure if that'd be an improvement, but try using "ib2007.year" in the line of the regression command.