question on placebo tests using synth

[alice tuberoni]

Good morning,

My question is about placebo tests using synth command.

To do a synthetic control method on a policy intervention in the year 1991 on country1, I run the following command:

synth depvar depvar(1960(5)1990) predictor1 predictor2 predictor3 predictor4, trunit(1) trperiod(1991) nested allopt unitnames(country) figure

This is working quite well for me. I know I may use more precise commands but I would like to stick with synth. However, I want to conduct a placebo test in time. What code may I use? I was thinking of just changing the year of trperiod(1991) to a previous one, like 1981. However this resulted in a big placebo effect, so I don't know what my error is.

Thank you in advance for the help!

[Carlo Lazzaro]

Alice:
welcome to this forum.
If by "big placebo effect" you mean that the trajectories of the treated nation and its synthetic control (SC) counterpart remakably drift apart during 1981-1990, you may want to reconsider the way your SC was defined.
That said, have you already perfomed an "in-space placebo" study? If that were the case, what is the post/pre RMSPE ratio for the treated nation and the base case controls?