The principal-agent framework (PAF) emerged in private sector analyses of human interaction that are more complex than the simple games studied in classical game theory (Gintis, 2000). The problem studied in the PAF is the following: If people interact longer than a one-shot game and if one party hires another as an expert of some kind, then how do they behave? This question is relevant to many private sector relationships that go beyond the simple adaptation inoceanicmarkets, including insurance, lawyers, agriculture (sharecropping), managing directors and health care (Rasmusen, 1994). How about public sector interaction? The application of the PAF to the public sector has not resulted in the same impressive list of new findings as with the private sector analysis using the PAF. The PAF comes up against the following perennial difficulties in approaching the public sector, such as:

• Who is the principal? The population at large or the elected politicians? • What is maximised? The welfare of society or the private utility of the

actors? • Who is the agent? An entire organisation or single individuals?