Exactly the same testing you do for climate science. Both are types of science where you get only one effective shot at predicting a given event because once an event passes you can no longer predict it.

Which is harder is hard to tell. Climate science has to deal with far more variables (current models still measure their grids in terms of miles, AFAIK) and the data is radically inferior to economic data. On the other hand, climate science does not (yet!) have to contend with people who take the predictions and actively use them to further confound the system.

(I say "yet" because if climate science becomes good enough to use it for engineering, all long-term predictions go out the window once nations start actively engineering outcomes. Note this isn't a criticism of climate science, it would in fact be strong validation, if the interventions worked as predicted. It's just that predictions predicated on the lack of intervention would no longer be valid.)