Question on synthetic control methods in Stata (multiple treated units)

[Fred Liu]

Hi all,
New user here, so apologies if this is the wrong forum etc.

I'm analysing the effects of new financial regulation on firms in Malaysia, and I'm thinking about using synthetic control methods as a way of generating a counterfactual to proxy as "untreated" Malaysian firms. However, I don't really want to use the single treated unit case that the Stata synth module seems to be designed for, since if I aggregate individual firm characteristics up to a country level, I'm essentially left with 12 treated observations and it feels like I'm throwing away too much data.

It seems that there have been a few recent papers who outline methodologies to essentially take synthetic control methods and apply it to multiple treated units (e.g. Acemoglu et al, 2016 https://economics.mit.edu/files/11926 ; Krief et al., 2015 http://onlinelibrary.wiley.com/doi/10.1002/hec.3258/pdf etc. ). Therefore I'm wondering if anybody knows if there are any Stata modules that apply these newer adaptations of synthetic control methods? Or if anybody knows how else I can apply these methods within Stata?

Thanks in advance for any help on this!

[Phil Bromiley]

You did not get a quick answer. Normally, you'll get a more helpful answer if you follow the FAQ on asking questions - provide Stata code in code delimiters, Stata output, and sample data. It helps if your code eliminates extraneous stuff so it only has what is needed to demonstrate the problem.

I assume you've entered findit synth in the Stata command line - this will show you what is available. You might check the websites of the authors of the papers you are referencing to see if they have software. You might Google synth Stata. Otherwise, you may have to program it yourself.

[Shruthi Venkatesh]

There's a package called synth_runner that implements synthetic controls with multiple treated units:

http://econweb.umd.edu/~galiani/files/synth_runner.pdf