Hello,
I am using the synthetic control method for a project and am unsure of how to include lags of the outcome variable as a predictor, since I’ve understood that if I use lags of the outcome variable there is no need for covariates to be included in the model. Should I construct the lagged variables manually or does Stata do it for me? hts is my main variable where unit 72 is treated and the others are not. My data is only made up from exports from one country to the US, where hts is export values of different units (commodity groups). Which code looks the most accurate?
Synth2 hts hts(10) hts(20) hts(30), trunit(72) trperiod(40) xperiod(1(1)39) fig
or
Synth2 hts lag1_hts lag2_hts lag3_hts, trunit(72) trperiod(40) figure
where lag1_hts, lag2_hts and lag3_hts are constructed manually by generating a new lagged variable. Does Stata understand to only do lagged values of my outcome variable hts72, even if lag1_hts, lag2_hts and lag3_hts are made for all my observations.
I am using the synthetic control method for a project and am unsure of how to include lags of the outcome variable as a predictor, since I’ve understood that if I use lags of the outcome variable there is no need for covariates to be included in the model. Should I construct the lagged variables manually or does Stata do it for me? hts is my main variable where unit 72 is treated and the others are not. My data is only made up from exports from one country to the US, where hts is export values of different units (commodity groups). Which code looks the most accurate?
Synth2 hts hts(10) hts(20) hts(30), trunit(72) trperiod(40) xperiod(1(1)39) fig
or
Synth2 hts lag1_hts lag2_hts lag3_hts, trunit(72) trperiod(40) figure
where lag1_hts, lag2_hts and lag3_hts are constructed manually by generating a new lagged variable. Does Stata understand to only do lagged values of my outcome variable hts72, even if lag1_hts, lag2_hts and lag3_hts are made for all my observations.
Comment