How can I calculate RMSPE post-intervention/pre-intervention in Synthetic Control Method?

[Miguel Ángel Ortiz]

Hello everyone. Mi name is Miguel Ángel and currently I am working with the synthetic control method from Abadie, Gardeazabal, Diamond and Hainmueller. The statistical software package is available at http://www.stanford.edu/~jhain/software.htm#Synth. I have run my synthetic unit and the placebo effects (the graph with every countries when are used as treatment unit). At this point, everything is ok.

However, I would like to calculate the robustness test proposed by authors above mentioned and that consists in calculating the ratio of RMSPE post/pre treatment, but I have not, or I do not know how to calculate the "post-intervention RMSPE". Someone can help me? I would be very grateful, because that is the last part of my master thesis and I need to know how to finish.

Thanks for all,

Best Regards,

Miguel

[Nick Cox]

You posted this yesterday and no-one answered. In the old Statalist, we had explicit advice: one repost is allowed, but no more. That's still good advice.

[Rich Goldstein]

more importantly, you need to ask why no one is answering your question; in my case, it's because I don't understand the question; see section 17 of the FAQ and the following Stata blog post: http://blog.stata.com/?s=question (and it would be helpful if that link was in the FAQ)

[Miguel Ángel Ortiz]

Ok sorry, I will try tonight to rewrite in the well-sense my question. However, after reading the answer, I have to say: I did not know how to formulate a question here, I am new. It is not necessary to be disagreeable with me, only because I didn't know something, I am refeering to Nick Cox.

[Nick Cox]

Sorry that you think I am being disagreeable. I was informing you of an old policy that still is relevant, but not expecting that you knew about it. I think it's true that your question would have been improved by reading the Advice in the FAQ.

[Sho Shigen]

Hello everyone. Mi name is Miguel Ángel and currently I am working with the synthetic control method from Abadie, Gardeazabal, Diamond and Hainmueller. The statistical software package is available at http://www.stanford.edu/~jhain/software.htm#Synth. I have run my synthetic unit and the placebo effects (the graph with every countries when are used as treatment unit). At this point, everything is ok.

However, I would like to calculate the robustness test proposed by authors above mentioned and that consists in calculating the ratio of RMSPE post/pre treatment, but I have not, or I do not know how to calculate the "post-intervention RMSPE". Someone can help me? I would be very grateful, because that is the last part of my master thesis and I need to know how to finish.

Thanks for all,

Best Regards,

Miguel

I found a way to calculate the ratio of RMSPE post/pre treatment. But it is kind of slow.
1.use synth and calculate the pre RMSPE.
synth rate rate(1) rate(2) rate(3) rate(4) rate(5) rate(6) rate(7), trunit(101) trperiod(8)xperiod(1(1)7) nested fig keep(nameyoulike)

then Stata will create a nameyoulike.dta, which contains the real outcome and Synthetic outcome of treated observation.

2.use
then open the file nameyoulike.dta, and calculate the post RMSPE.
like this

use nameyoulike.dta, clear
gen effect= _Y_treated - _Y_synthetic
gen e2= effect^2
sum e2 if _time==8 | _time==9

then you could copy the value to the excel and calculate the ratio of RMSPE post/pre treatment.
for loop is kind of useful in this case