Two of the theories which have attracted EU analysts' attention over the past twenty years are (1) principal-agent models (e.g. Pollack 2003, Dehousse 2008, Héritier 2009), and (2) transaction cost economics (e.g. Majone 1996 and 2009, Franchino 2007, Karagiannis 2007). In this paper I explore the lessons those two theories can offer in terms of suggestions for a better union. Principal-agent models turn out to be of rather limited utility, because the lessons they teach are either already learned ("limit information asymmetries" or "screen between different types of agents") or difficult to implement ("be perfectly rational", "create high-powered incentives"). Transaction cost economics, on the other hand, produce an important insight: to achieve deeper integration we need to be patient until two conditions are met simultaneously, namely (a) the willingness to transact, and (b) the need to do so on an exclusive basis. I illustrate my propositions with examples from the historical development of the EU and Mercosur.
