We now turn to the extensive form of the policy formation stage. If a single
party holds government, policy formation is straightforward: the party acts in
unison and simply sets the vector of policy instruments q to maximize (2.4) subject
to the economic constraints (2.1) and (2.2), and taking into account equilibrium
behavior at the subsequent election stage, yet to be described.
In the event of a coalition government, the parties in government generally
disagree over the allocation of spending on local public goods, {gJ}, and rents,
{rP } — see below. We assume that each member of the coalition has unilateral
decision making power over the relevant dimension of these two variables; thus,
party P representing economic group J unilaterally sets rents for itself, rP , and
local public goods benefiting its group J, gJ, so as to maximize (2.4) subject to
the economic constraints (2.1) and (2.2), and taking into account the equilibrium
at the election stage. As we shall see, all coalition members agree about the
optimal amount of spending on the local public goods benefiting the groups not
represented in government. Hence, it is unimportant who has control over these
policy instruments, and for simplicity we assume that a coin is tossed about who
is entitled to choose them. Finally, we assume that tax revenues τ are residually
determined so as to balance the budget, once all spending decisions have been
made (assuming an interior optimum for the tax rate).
This assumption about the behavior of coalition governments can be inter-
preted as parties obtaining agenda-setting powers — say, in the form of ministerial
positions — over the policy dimensions they care about the most (such powers
could potentially be derived from bargaining over ministerial positions at a gov-
ernment formation stage, as in Laver and Shepsle, 1996). This agenda-setting
power is strong enough to allow the minister in charge to do what is unilaterally
optimal for her party. This would happen, for instance, if rejection of a policy
proposal led to a government crisis or some very costly outcome for the coalition
members, as in Diermeier and Feddersen (1998) or Persson, Roland and Tabellini
(2000).
As discussed in the next section, the implication of our assumptions about
policy formation is that coalition governments, but not single-party governments,
face a common-pool problem in setting local public goods and rents.
2.2.4. Elections and voters
We assume that citizens vote retrospectively, based on “ideology”, as well as their
economic welfare, rewarding or punishing the incumbent government according to