0
$\begingroup$

I am interested in using the Synthetic Control Method to measure the effects of a certain policy. I am afraid that the synthetic control will not be accurate because the underlying variables that make up the synthetic are dependent on the policy. Imagine a situation where a policy is applied to a single town it is likely that the synthetic control contains aspects of the neighboring towns which will indirectly be affected by the policy. One way around this would be to not consider the neighboring towns in the creation of the synthetic? A bit confused about the role of dependent variables in the construction of a synthetic control.

$\endgroup$

1 Answer 1

1
$\begingroup$

For the synthetic cohort approach, you are trying to find weights for the untreated regions that make their weighted sum behave like the treated region before the policy happens. In other words, the weights cannot be affected by the policy since you are not using the post-policy data to find them. If you are worried about changes that can happen in anticipation of the policy, you can start the clock ticking when the policy is announced, rather than when it kicks in.

If you are worried about spillovers to neighboring or linked regions after the policy starts, you can exclude them from the set of potential donor states to be re-weighted. This will help preserve the SUTVA.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.