x
B
x
L
x
G
x
ω
ω
A
ˆ
w
*
w
*
Ω
x
φ
ˆ
x
φ
x
ω
ˆ
Ω
ˆ
r
r
Figure 5:
[The following is still quite vague and can undoubtedly be tightened and improved.]
Given the uncertainty, it is critical for the legislature to design delegation in a way that
elicits the optimal amount of information from the interest group, even at some price. Figure
5 demonstrates that the legislature can do better than using the standard delegation window.
Suppose L constrains B for the most damaging information ω ∈ ¯
Ω. When the bureaucrat’s
policy choice is constrained she cannot “compensate” fully for the policy shock, thus forcing
the outcome to respond to ω and, within some range, to move closer to the group’s ideal
point. By constraining B sufficiently, for some ω the group prefers B’s informed (albeit
constrained) policy choice to the uninformed one. Thus the group now has an incentive to
reveal information that it formerly concealed.
In addition to leading to a more informed (though constrained) policy choice, revealing
ω shifts the support ¯
Ω for concealed information (see Figure 5). This changes the expecta-
tion E(ω|∅), resulting in a new locus for uninformed policy outcomes ˆ
x
φ
. The monotonic
18
Convention