Hello,
I am working on a comparative case-study using the synthetic control method (traditionally implemented by synth package).
However, I find the synth_runner (package) more handy and easy to implement unlike synth (package).
But, when I use synth_runner, the reference line for the policy/treatment implementation is plotted one period (say year) prior to the actual assigned date.
For instance, see the following codes:
. ssc install synth, all
. cap ado uninstall synth_runner //in-case already installed
. net install synth_runner, from(https://raw.github.com/bquistorff/synth_runner/master/) replace
. sysuse synth_smoking, clear
. tsset state year
1. The following code uses synth (the traditional package/code) and at least assigns the reference line to the actual treatment period. This makes sense to me because there reference period aught to be the actual treatment period.
. synth cigsale beer lnincome(1980&1985) retprice cigsale(1988) cigsale(1980) cigsale(1975), trunit(3) trperiod(1989) fig
2. The following codes uses synth_runner and ideally the treatment period is 1989 but the reference line (the vertical red solid line in the second fig.) is at 1988 (not 1989).
. synth_runner cigsale beer(1984(1)1988) lnincome(1972(1)1988) retprice age15to24 cigsale(1988) cigsale(1980) cigsale(1975), trunit(3) trperiod(1989) gen_vars
. single_treatment_graphs, trlinediff(-1) raw_gname(cigsale1_raw) effects_gname(cigsale1_effects) effects_ylabels(-30(10)30) effects_ymax(35) effects_ymin(-35)
. effect_graphs , trlinediff(-1) effect_gname(cigsale1_effect) tc_gname(cigsale1_tc)
Now, my question is, after using the synth_runner, is it appropriate to manually change the reference line to reflect the actual treatment date?
Would that alter the treatment effect?
Thank you.
I am working on a comparative case-study using the synthetic control method (traditionally implemented by synth package).
However, I find the synth_runner (package) more handy and easy to implement unlike synth (package).
But, when I use synth_runner, the reference line for the policy/treatment implementation is plotted one period (say year) prior to the actual assigned date.
For instance, see the following codes:
. ssc install synth, all
. cap ado uninstall synth_runner //in-case already installed
. net install synth_runner, from(https://raw.github.com/bquistorff/synth_runner/master/) replace
. sysuse synth_smoking, clear
. tsset state year
1. The following code uses synth (the traditional package/code) and at least assigns the reference line to the actual treatment period. This makes sense to me because there reference period aught to be the actual treatment period.
. synth cigsale beer lnincome(1980&1985) retprice cigsale(1988) cigsale(1980) cigsale(1975), trunit(3) trperiod(1989) fig
2. The following codes uses synth_runner and ideally the treatment period is 1989 but the reference line (the vertical red solid line in the second fig.) is at 1988 (not 1989).
. synth_runner cigsale beer(1984(1)1988) lnincome(1972(1)1988) retprice age15to24 cigsale(1988) cigsale(1980) cigsale(1975), trunit(3) trperiod(1989) gen_vars
. single_treatment_graphs, trlinediff(-1) raw_gname(cigsale1_raw) effects_gname(cigsale1_effects) effects_ylabels(-30(10)30) effects_ymax(35) effects_ymin(-35)
. effect_graphs , trlinediff(-1) effect_gname(cigsale1_effect) tc_gname(cigsale1_tc)
Now, my question is, after using the synth_runner, is it appropriate to manually change the reference line to reflect the actual treatment date?
Would that alter the treatment effect?
Thank you.
