Hello everyone,

I have a large dataset of forest pixels with different control variables such as altitude, slope, distance to nearest river etc. I would like to match the observations from my treatment area with the control area, to later estimate the effect of a policy trying to mitigate deforestation via diff-in-diff.
However, both areas (treatment and control) have large sig. differences in their baseline values and the decision of program implementation does not really depend on those but the deforestation rate does. Therefore, I think that PSM is not the right approach here.
When estimating the pscore I obviously get very large values near 1 for the treated and very low ones near 0 for the control area, so the common support is very limited (from treated 24.000 off support and only 6000 on support, while all controls are on support)
Therefore, I am looking for a matching method where I can match my treatment with control observations based on the covariates to implement a diff-in-diff afterwards.
I was searching for literature on the topic but got quite confused with the variety of matching methods out there.

Do you know useful literature on the topic which gives an overview over different matching methods and their applicability?
Can you suggest a matching method which is more applicable for my situation given I want to conduct a diff in diff analysis?

Thanks!

Best,
David