between diVerent elements of the government machine. These functions can be
carried out by institutions other than prime minister and cabinet; for example, the
Treasury and the Cabinet OYce. By deWning the core executive in functional terms,
the key question becomes, ‘‘who does what?’’
But power is contingent and relational; that is, it depends on the relative power of
other actors and events. Ministers depend on the prime minister for support in
getting funds from the Treasury. In turn, the prime ministerdepends on his ministers
to deliver the party’s electoral promises. Both ministers and prime minister depend
on the health of the American economy for a stable pound and a growing economy to
ensure the needed Wnancial resources are available. This power-dependence
approach focuses on the distribution of such resources as money and authority in
the core executive and explores the shifting patterns of dependence between
the several actors (see for example Elgie 1997 ; Rhodes 1995 ; Smith 1999 ).
The term ‘‘core executive’’ directs our attention, therefore, to two key questions:
‘‘Who does what?’’ and ‘‘Who has what resources?’’ If the answer for several policy
areas and several conXicts is that the prime minister coordinates policy, resolves
conXicts, and controls the main resources, we will indeed have prime ministerial
government.
Therational choice institutionalismapproach comes in many guises and increas-
ingly focuses on the analysis of prime ministers and cabinets. One example must
suYce: Strøm and his colleagues’ principal–agent theory of delegation and
accountability in parliamentary democracies (see also Cox 1987 ; Laver and Shepsle
1996 ; Tsebelis 2002 ).
Strøm, Mu ̈ller, and Bergman ( 2003 , chapters 3 and 23 ) conceive of parliamentary
democracy as a chain of delegation from principals to agents: from voters to their
elected representatives, from legislators to the chief executive, from the chief
executive to ministerial heads of departments, and from ministers to civil servants.
Principals and agents are in a hierarchic relationship and both act rationally to gain
exogenously given preferences. No agent is perfect. So agency loss occurs because
the actual consequences of delegation diverge from the principal’s ideal outcome.
There are two main causes of agency loss. First, there may be a conXict of interest
between the principal and the agent who may, for example, have diVerent policy
objectives. Second, there may be limited information and resources and, for
example, the principal may not know what the agent is doing. When principals
know less than agents, two problems occur, moral hazard and adverse selection.
Moral hazard arises when an agent takes actions of which a principal disapproves.
Adverse selection occurs when an agent is unwilling or unable to pursue the
principal’s interests. A principal can useex antemechanisms, such as screening
of applicants, to control adverse selection problems, andex postmechanisms, such
as contracts, to deal with moral hazard. This framework is then used to analyse, for
example, the strengths of Westminster parliamentary systems, which are said to be
coordination and eYciency.
326 r. a. w. rhodes