Choice theoretic foundation for libertarian paternalism: Reconciling the behavioral and libertarian approaches to welfare

In this paper, we generalize the standard revealed preference theory in line with the idea of libertarian paternalism; roughly, x is said to be welfare superior to y (denoted x≿_ℓP y) if x is chosen when y is available, and x is more normative than y according to a possibly endogenous normative criterion (i.e. a criterion that depends on a positive model). We find testable conditions on choice behavior that are equivalent to ≿_ℓP being an order. The resulting framework is applicable to many positive models of bounded rationality, with intuitive welfare implications, while maintaining a natural libertarian principle. In addition, our framework can reconcile the often-conflicting libertarian and behavioral approaches to welfare.