Syndicate content

Add new comment

Submitted by Nicolas on

Hi !
Thanks for the presentation of the package. I have a question on the usage of ritest with continuous variables.

I am using it in the context of panel data ; the scale of observation is municipality (N=30 000) and the time dimension is years (T=26).

I regress the average level of income by municipality on several weather conditions and I would like to check the robustness of the results by using Randomization Inference.

I therefore tried to estimate it with: ritest mean_temp _b[mean_temp], reps(1000) strata(municipality_id) cluster(region) : xtreg income mean_temperature i.year, fe cluster(region) and it returned:

mean_temp doesnt seem to be constant within clusters

I nevertheless want to allow for correlations at the regional level (or even higher) because two close municipalities cannot be considered as having independent weather conditions, ie: I would like ritest to assign to each commune in one cluster the same year of observation.

Is there a simple way of doing it using your package or should I create my own permutation programme?

Thanks a lot for your help,