I am interested in exploring the causal effect of poverty on the adoption of a number of climate-resilient agriculture practices in sub-Saharan Africa.
In exploring the causal effect of poverty (a non-random phenomenon that can't be made into an intervention), my initial idea was to create a matched group of non-poor households and implement a difference-in-difference approach on 3 waves of panel data. Using an experimental analogy, the "treatment" would be household poverty status on wave2 and the impact would be given by the difference in the mean number of agriculture practices between the "treated" and matched control group (wave3-wave2). The wave1 would only be used to prove the parallel line assumption.
What worries me is that I could not find a similar research design in the literature, apart maybe some papers on the effect of remittances (which you may argue, is still exogenous to the household ).
Does it makes sense to use a counterfactual design in the absence of an exogenous intervention?
Many thanks for your insights.