HELP: Using Data on Different Spatial Scales for Econometric Regressions

*Background: Using a Production Function in form of Autoregressive Regression Model for Regional Perfomance*


This question is more of a theoretical nature. I have dataset that comprises all data on municipality level (n=290). However, due to data restrictions one variable, "self employment rate" is on county level (n=24).

I was wondering if anyone had this issue before? I would rather not drop the variable and obviously do not want to aggregate all data on county level.

I thought of using the self-empl rate for counties as a proxy for each municipality weighed with their number of employed population:

log(self-empl(county) * employed pop (munic))

However, I don't know if this is a scientific way to do it, and if so, I didn't find any literature using similar approaches.

Thank you.