Dear Statalist Users,
I have observational data and would like to estimate the causal effect of a historic political reform on current household-level outcomes.
As my treatment is a continous or rather a fractional variable (measuring the historic proportion of the population being affected by the reform), I cannot rely on the convential treatment effect estimators.
Moreover, I assume that there might be some selection bias as the reform was implemented during a severe drought.
One possibility would certainly be to generate a binary treatment variable so that I could use doubly robust methods that would allow me to specify the outcome and treatment process.
However, I would lose a lot of information in doing so and there is no theory-based benchmark by which I could generate this variable.
Another option would be to rely on a IV-strategy. However, I could not find a suitable instrument for the treatment.
What alternative ways would there be to estimate a causal effect in this case?
Any help or advice is greatly appreciated.
---
Jule Beck
Stata 17.0
I have observational data and would like to estimate the causal effect of a historic political reform on current household-level outcomes.
As my treatment is a continous or rather a fractional variable (measuring the historic proportion of the population being affected by the reform), I cannot rely on the convential treatment effect estimators.
Moreover, I assume that there might be some selection bias as the reform was implemented during a severe drought.
One possibility would certainly be to generate a binary treatment variable so that I could use doubly robust methods that would allow me to specify the outcome and treatment process.
However, I would lose a lot of information in doing so and there is no theory-based benchmark by which I could generate this variable.
Another option would be to rely on a IV-strategy. However, I could not find a suitable instrument for the treatment.
What alternative ways would there be to estimate a causal effect in this case?
Any help or advice is greatly appreciated.
---
Jule Beck
Stata 17.0
