Balance in pre-treatment period with synthetic control

[Jackson Var]

Hi everyone,

I am constructing a Synthetic Control and have my donor pool of 58 units unaffected by the intervention.

My question is: should the mean of the donor pool units be equal to the treatment unit in the pre-intervention periods? Or should this only be the case for the units of the donor pool that have a positive weight on the synthetic control ?

Thank you,

Regards!

[Jesse Wursten]

I don't think you need equal means anywhere when using synthetic controls. The whole point of synthetic controls is that they generate an artificial control group that matches the treatment group as well as possible on the criteria you choose.