when the people are distanced from their representatives, they can judge their actions without being implicated in them themselves.6
There are two ways in which accountability is conceived consistent with the principles of the delegation model. The first, borrowed from economic models of employer/employee relations, is the principal-agent model (IIa) (Pollack 1997). Political officials (including elected representatives) are understood to be “employees” (“agents”) of their constituents or their executive superiors (“principals”). In principal-agent models, the principal’s preferences are taken as given, and the relationship is judged a failure if the agent deviates from them. In this understanding, accountability is ensured when agents have incentives to do what the principals want them to do. Agents are seen as instruments of the principal’s will, and they ought to act as the principal himself would if he held the position (Maravall, 1999:155).
Yet the relations of representative/constituent and officeholder/public differ in important respects from the employee/employer relation. First, once appointed or elected, officeholders make decisions that govern members of the public; employees remain the subordinates of their employers. The power relations in the case of politics are more complex (Rousseau 1762/1988, I.6). Second, unlike most employees, when a politician is given a job, he or she is generally given a great deal of power and considerable discretion as to how to use it. Often, officeholders are entrusted with power on account of their superior expertise or judgment. They ought to exercise their discretion in performing their duties.7 Principal-agent models are inappropriate, for example, when calling judges to account. Power is certainly delegated to judges, but
6 Hamilton, et. al. (1787/1982) , nos. 10 and 49. See Grant and Grant (1981), pp. 35-6.
7 For an interesting discussion of “two logics of delegation,” agency and fiduciary, see Majone (2001).