European governments have adopted policy instruments for regulatory appraisal, oversight, ex-post evaluation, and simplification in the context of the so-called ‘smart regulation agenda’. In this article we compare the two most important instruments, that is, regulatory impact assessment (RIA) and the standard cost model (SCM). We answer the following questions: What are the economic rationales that, at least in principle, should make the SCM and RIA work? What are the learning models that, yet again in principle, allow the two instruments to produce effects? The RIA economic rationale is grounded in welfare economics. The SCM economics is rudimentary: one can hardly make an economic case for the SCM. With regard to learning models, RIA draws on rational-synoptic models, whilst the SCM is inspired by experience-based learning. We then discuss economic rationales and learning models jointly, thus explaining the different implementation patterns of the two instruments and exposing the ambiguities in the relationship among instruments, ideas, and behavioural change.