In an ideal experiment, one would randomly grant payday loans to a group of applicants and randomly deny the loans as well as close substitutes to a similar group of applicants. One would then track indicators of financial stress over time across the two groups.
This is to answer the question of whether payday lending helps customers or hurts them. The idea is to use an experiment, as opposed to a priori theory or some parameterized model.
I favor more of this approach in economics.
Thanks to the indispensable Timothy Taylor for the pointer.
[UPDATE: a commenter points to a blog post by Paige Marga Skiba that describes a paper that looks at people near the borderline between approval and non-approval for a loan. Those who were not approved were better off. I have not read the paper, but the approach seems sensible.]