18
his party’s request.
27
For instance, C
e
will be high in the case of a pro-market
president who pays his party by increasing tariffs on agriculture. Second, C
e
also
varies as a function of presidents and their parties’ divergent constituencies and
reelection concerns. For example, C
e
will be big in the case of a president who
reciprocates his party through electoral fraud, which will risk his good reputation and
place in history.
X
p
= the party’s potential benefits, or costs, from presidential coattails. By presidential
coattails, I mean the benefits or costs for the party derived from the executive’s
performance in office. Presidential coattails do not presuppose any ideological bias.
That is, the party may benefit from its president’s performance even if it contradicts
the party’s principles.
28
The party’s valuation of coattails may diverge from the
president’s assessment. For instance, when the president asks his party to approve
a tax increase, these actors may disagree about the short and long term electoral
effects of this policy. A high expected value for X
p
would be the party supporting a
war that is later won. An example of a low, even negative, X
p
in the perception of
parties would be a failed economic reform.
Y
e
= the president’s costs of not paying back; these are related to the party’s control and
monitoring means. They come from two sources. First, are those derived from the
party’s capacity for direct retaliation. Y
e
will increase to the degree that the party’s
threat is more credible. For example, the party may threaten the president, implicit or
explicitly, with impeachment.
29
A second source of Y
e
costs are exogenous to the
27
Even though B
p
and C
e
(similarly to B
e
and C
p
) refer to the same policy, these terms differ on how each actor
values them as a function of their relative resources and policy preferences.
28
This is the case, for instance, of leftist parties in Latin America benefiting from market reforms implemented by
their presidents (Stokes, 2001).
29
Note that retaliation is not stated here in terms of repeated games with imperfect information—implying that
behavior at time t affect actors’ choices in t+1. This implies that every iteration is independent from the previous
one. Building a reputation of paying back thus is not a concern for presidents in this one-shot game. Future