This paper compares PA models of delegation and discretion with an incomplete contracts-theoretical (‘ICT’) approach to the design and operation of European institutions. In my understanding of PA models, they must always come with a set of strict, unavoidable, and also quite heroic assumptions regarding (a) the actors’ hyper-rationality and impeccable computational capabilities, (b) the immutable identity of the principal and the agent, (c) the clarity of the actors’ preferences and risk postures, and ultimately (d) the observability and enforceability of complete contracts. One way to address the heroism of each and every one of these assumptions would, of course, be to try relaxing them one by one and see what happens. Another, more promising and faster, way, is to drop them all together, and straightforwardly compare the merits of PA analysis to those of ICT. I do so by starting from the start - i.e. by examining whether the institutional architecture created in the 1950s obeyed a PA or an ICT logic. I find clear support for the latter hypothesis.
