search

synth-inference / synthdid

Synthetic difference in differences
https://synth-inference.github.io/synthdid
BSD 3-Clause "New" or "Revised" License
261 stars 96 forks source link

Replication code for Table 1 #111

Closed elizabeth-richardson closed 9 months ago

elizabeth-richardson commented 9 months ago

Thanks for a great paper and package. I've found the replication code for the simulations, but would like to also see the code for the cigarette tax results that you've reported in table 1. Is this available?

davidahirshberg commented 9 months ago

It is! Take a look here: https://synth-inference.github.io/synthdid

elizabeth-richardson commented 9 months ago

Thank you for prompt reply! Is there code for the other (non-SDID) parts of the table? (I'm specifically asking because I haven't found a package that will give me a standard error for a synth control estimate).

davidahirshberg commented 9 months ago

Yes for SC and DID. We don't include the Matrix Completion or DIFP estimators there.

Keep in mind that while we report a standard error for these things, they're based on very strong assumptions. There's just not much you can do without them when you have only one treated unit.

davidahirshberg commented 9 months ago

And our package is giving us these standard errors for the SC and DID estimates.

elizabeth-richardson commented 9 months ago

Ah! I see. Very helpful. Thanks for your help.