PUBLIC
ECONOMICS IMPORTANCE
By - Subham Mondal
I. INTRODUCTION
A
significant portion of government activity is devoted to transfer of resourced
between citizens. Some of these transfers, such as those to the poor, seem to
be consistent with traditional social welfare objectives. Others are directed
to so-called special-interest groups, such as farmers, unions, professionals
groups, or particular firms and industries. Political economy suggests at least
two reasons why politicians may choose to make such transfers. First, interest
groups may be able to enhance politicians’ chances of reelection by providing
campaign contributions or political support. Second, interest groups can
improve politicians’ financial well-being by, for example, providing bribes,
business for firms in which they have a financial interest, or future
employment opportunities.
An
important question in political economy concerns the form of transfers to
special interests. While redistribution towards the poor generally takes the
form of cash and in kind transfers, redistribution to special interests is
typically less much direct. What explains the method chosen to redistribute to
special interests?
One
perspective on this issue, often associated with the “Chicago School” of
political economy, is that political competition will ensure that the most
efficient method of redistribution available is chosen e.g. the US Government
imposed oil imports quotas, rather than made direct cash transfers to the oil
industry, reflected the superior efficiency of such quotas. The Chicago view
has the provocative served method of redistribution is inefficient, this
analysis must be missing something. The challenge for political economy,
therefore, is to provide efficiency explanations for observed transfer
mechanism.
An
alternative view, associated with the “Virginia School” of political economy,
stresses the importance of imperfect information in explaining the form of
transfers. Citizens are presumed to be poorly informed about the effects of
different policies, and this leads politicians to select inefficient “sneaky”
methods of redistribution over more transparent efficient methods. For example,
politicians will favor policies that serve to transfer resources but may be
justifiable on other, more palatable grounds so called disguised transfers
mechanisms. For example, a road may be laid out in such a way as to increase
the value of certain pieces of real estate when laying out the road in its
optimal location and making cash transfers to the owners of this real estate
would be more efficient. The idea is that voters do not know the optimal
location of the road and therefore are unable to detect the real motivation for
its location. Policies that transfer resources by changing market prices, such
as quotas or mandates, fall into this category.
While
the Virginia view is intuitively appealing, it lacks a solid analytical
foundation. The idea that voters remain rationally ignorant because the
expected benefits from becoming informed are small relative to costs. However,
Becker (1976) and Wittman (1989) question why voters should have biased beliefs
about the effects of policies and how they could be persistently fooled. It is
by no means clear that the Virginia view can be justified without making such
unreasonable assumptions.
This
paper focuses on understanding the form of transfers in an environment in which
politicians have a financial incentive to make transfers to a special interest
and have available both direct cash transfers and a disguised transfer’s
mechanism. In the model, the disguised transfer mechanism is a public project.
Introducing the project not only benefits the special interests but also, under
certain conditions, enhances the welfare of the citizens. There is asymmetric
information in the sense that the conditions are satisfied that the incumbent
politician has more information about whether these conditions are satisfied
than the citizens do. Furthermore, since the benefits of the project are
stochastic, citizens observe only a noisy signal of whether it was warranted ex
post. Thus when they observe the implementation of the project, they cannot
tell whether the politician is acting in their interest or simply making
transfers to the special interests.
II.
PREMISES
The
model suggests that if politicians are all identical and known to be so,
transfers to the special interest will be made efficient despite the
availability of a disguised transfer mechanism. Citizens will allocate
political support in such a way as to make efficient behavior in the incumbent
politician’s interest. However, if politicians differ, some being susceptible
to bribes and others not, and if politicians’ types are not perfectly
observable to citizens, then transfers to the special interest will sometimes
be made inefficiently. This reflects the fact that politicians have incentives
to build reputations. “Bad” politicians (i.e. those susceptible to bribes) will
sometimes prefer to implement the project when it is not warranted because
making direct cash transfers does greater damage to their reputations. The
paper therefore shows how a combination of asymmetric information about
policies and politicians can explain the choice of inefficient methods of
redistribution in a world in which voters are rational.
In
equilibrium, politicians set the policy at a level that maximizes the
well-being of those groups or individuals who hold political power. Since,
these agents do not care about the costs and benefits that fall on others in
society, the equilibrium level of the policy is claimed to be “inefficient.”
By
assumption, politicians are constrained to use a given tax rule and can
redistribute only by choosing different levels of the policy. At an
equilibrium, any change in the level of the policy will reduce the well-being
of the politically influential. The equilibrium utility allocation is therefore
on the (second-best) Pareto frontier. This paper is that it assumes that politicians
have two different methods of redistribution available.
In
this paper, concern about reputation is key to explaining the inefficiency.
The Model
We
employ an agency style model of political competition of the sort pioneered by
Barro (1973) and Ferejohn (1986) and further developed by Austen-Smith and
Banks (1989) and Banks and Sundaram (1993). We consider a two-period model. In
the first period, an incumbent politician must decide whether or not to
introduce a public project. The incumbent also has the ability to make direct
cash transfers from the citizens to the special interest, so that there is no
need to use the project as a transfer device. At the end of the first period,
an election is held. The incumbent faces a randomly drawn challenger. The
political power is held by the citizens, who alone determine the outcome of the
election. In the second period, the winner of the election simply selects a
cash transfer to the special interest.
Citizens and the Special
Interest
A
single representative citizen receives income yc at the beginning of both periods. The citizen gets
utility from consumption and public projects. His utility per period is given
by
,
where
t denotes taxes and B represents the
benefits from public projects. His sole decision is whether to reelect the
incumbent at the end of the first period.
Along
with the citizens, there is a special interest that derives income indirectly
from public projects (e.g., the special interest might be a firm that supplies
publicity provided goods). The special interest may also receive income
directly through government transfers. The special interest’s income in each
period is given by
s , where R denotes the
income derived from public expenditures and T is the direct cash transfer.
Policies
In
each period, the politician holding office chooses a cash transfer T >
0 to the special interest. In the first period, the incumbent must also decide
whether or not to implement a public project. The project costs an amount C and
is financed by taxation of the citizen. It provides income R, for the special
interest. The benefit the project provides to the citizen is uncertain. It may
produce BH or BL units of benefits, BH > BL > 0. The probability that the project will produce
high benefits (i.e., that B = BH)
is denoted
.
This probability can take on one of two values,
0 or
1. The incumbent is
assumed to observe the value of
prior to deciding whether to implement the
project.
The
ex ante probability that the project is likely to yield high benefits (i.e.,
that
=
1) is π. The expected net
gain to the citizen from the project when the probability that it produces high
benefits is
is denoted
(
)
that is,
(
)
=
BH + (1 –
.
We
make the following key assumption concerning the efficiency of the project.
Assumption
1. (i)
(
1) > 0 and (ii)
(
0) <
.
When
=
1, the project yields a
positive expected net gain to the citizen. When
=
0, the citizen would
actually be better off with a tax financed transfer of Rs, to the
special interest than with the €implementation of the project. Thus when
=
1, the project is
efficient, when
=
0, introducing the
project is an inefficient way to make a transfer of Rs to the
special interest.
Politicians
Politicians come in two types:
“good” (i= g) and “bad” (i = b). Both types of politicians receive zero utility
when not in office and discount the future according to the discount rate
.
When a good politician is in office, his payoff depends positively on the
utility gain his decisions generate for the citizen. Thus a good politician’s
utility per period is
,
where
g (. ) is some smooth, increasing function.
A bad politician cares not only
about the utility he generates for the citizen, but also about the income
received by the special interest. A bad politician is susceptible to bribes and
other nonmonetary rewards offered by the special interest. The more income the
special interest receives, the greater the reward given to the politician.
Thus, when a bad politician is in power, his utility per period is
(
),
where
( .
) is smooth, increasing in both arguments, and strictly concave.
There
are two assumptions concerning a bad politician’s preferences. The first is
that, from a bad politician’s viewpoint, the gain to the special interest
resulting from introducing the project when
0 is more than sufficient to offset
the loss to the citizen.
Assumption
2:
b
(
(
0
),
s
)
>
b
( 0,
0 ).
For
the second assumption, let T*(
)
denote the direct cash transfer that would maximize a bad politician’s utility
per period when the pretransfer utility gain for the citizen is
and the pretransfer income for the special
interest is
; that is,
T*(
)
= argmax
b (
+ T ).
Assumption
3 (i) T*(
)
(0,
yc
C) and (ii)
b
*(
)
b (
)
<
b *(0, 0)
Part
i says that over the relevant range, a bad politician wishes to make some
transfers to the special interest but does not want to bankrupt the citizen)
Part ii implies that the loss in utility resulting from forgoing the optimal
direct cash transfer is always less than the discounted value of the maximal
utility obtainable when (
)
= (0, 0).
The Information
Structure
The
citizen’s decision whether to reelect the incumbent politician at the end of
the first period is complicated by imperfect information. First, there is
“policy uncertainty” the citizen is unable to observe the realization of the
random variable
.
Thus only the incumbent knows whether the project is or is not in the citizen’s
interest. The idea is that the results of the study are observed only by the
incumbent, and there is no way of credibly conveying them to the citizen.
Naturally, the citizen can observe the level of benefits generated by the
project, but this is not a perfectly revealing signal. Even when
=
1 and the project is
warranted, it may fail to produce high benefits.
The
citizen also faces “politician uncertainty” he cannot directly observe whether
politicians are good or bad. The citizen is not completely uninformed for, when
he first encounters a politician, he does observe some signal of his type. This
signal allows the citizen to form an initial estimate of the likelihood that
the politician is good. Let
1
(0,
1) denote the citizen’s estimate of the probability that the incumbent is god
at the beginning of the first period. Thus
1 is the incumbent’s initial reputation. The incumbent is
assumed to be aware of his initial reputation. Let
c
denote
the challenger’s initial reputation, that is, the citizen’s estimate of the
probability that the challenger is good. This is not known by the citizen or
the incumbent until the end of the first period, when the challenger is
selected. We assume that
c is drawn from some
cumulative distribution function G(
).
This cumulative distribution function is assumed to be smooth and increasing
and to satisfy the property that G(0) = 0.
The Game and the
Definition of Equilibrium
This
two-period model defines a game among the incumbent, challenger, and citizen.
At the beginning of the game, nature chooses the type of the incumbent (i
{b,
g}). Nature then chooses the probability that the project will produce high
benefit (
0,
1}). This choice is
observed solely by the incumbent The citizen knows only that the probability
that
=
1
is π.
The incumbent must then choose a transfer to the special interest and decide
whether or not to implement the project. Formally, the incumbent can be thought
of as choosing a project decision transfer pair (
{P, N } X
+, where D = P (N) means
that the project is (is not) implemented. If the project is implemented, nature
chooses the benefits it produces for the citizen. These benefits are BH
with probability
and
BL with probability
The
incumbent’s first-period record is the triple (D, T, B).
The
election is held at the end of the first period. Nature chooses the type of the
challenger (i
{b,
g}), and the citizen observes some noisy information about his type. In
particular, the noise is such that his estimate of the probability that the
challenger is good is
c, where
c is drawn from the
cumulative distribution function G (
).
Knowing
c and the incumbent’s
first-period record, the citizen must decide whom to elect. Once the election
is over, the winning politician makes a transfer decision and the game ends.
Strategies
A
strategy for the incumbent has two components. The first is a rule that
specifies a project and transfer decision in the first period for each type the
incumbent might be and each realization of
The
second component is a rule that specifies a transfer decision in the second
period should the incumbent be reelected. Since the game ends at the end of the
second period, this second-period decision depends only on the incumbent’s
type.
A
strategy for the challenger is simply a rule that specifies the transfer he
will make should he be elected. Again this decision will simply depend on the
type. A strategy for the citizen is a rule that specifies the probability that
he will re-elect the incumbent. This rule will depend on the incumbent’s first
period record (D, T, B) and the initial reputation of the challenger
c.
A
perfect Bayesian equilibrium of this game consists of a strategy for the
incumbent, a strategy for the challenger, and a strategy and beliefs for the
citizen that satisfy four properties.
(i)
The
citizen’s beliefs are consistent with the incumbent’s strategy in the sense
that they are generated by Bayes updating where possible.
(ii)
The
citizen’s strategy is optimal given these beliefs and the strategies of the
incumbent and challenger.
(iii)
The
incumbent’s strategy is optimal given the citizen’s beliefs and strategy and
the challenger’s strategy.
(iv)
The
challenger’s strategy is optimal.
Inefficient Transfers
The
task of this section is to solve for the equilibrium of the game and to analyze
the incumbent’s equilibrium policy choices. Equilibrium will be solved for by
backward induction.
Second-Period Behavior
of Politicians
Suppose
that the incumbent is in power in the second period. If he is good, he will
make no cash transfers to the special interest and his second-period utility
will be
g (0). If he is bad, he
will make a cash transfer T0 = T*(0, 0) and obtain a utility level
b* (0, 0.). If the
challenger is in power in the second period, he will follow exactly the same
strategy as the incumbent. If good, he will make no transfer, if bad; he will
make a transfer T0.
The Citizen’s behavior
The
citizen will be better off with a good politician in power in the second period
than with a bad one. He will therefore elect that politician who he believes is
most likely to be good. Thus if
(D,T,B)
denotes the citizen’s estimate of the probability that the incumbent is good
when his first-period record is (D,T,B) and if the challenger’s reputation is
c
, the
citizen will reelect the incumbent if and only if a (D,T,B) >
c. Since the challenger’s
reputation is a random draw from the cumulative distribution function G(
),
the probability that the incumbent will be reelected is simply G (
(D,T,B)).
The Incumbent’s
First-Period Behavior and the Citizen’s Beliefs
Suppose
that the probability that the project will yield high benefits is
.
If the incumbent is good and selects a transfer T, his expected payoff will be
Vg (N,T,
)
= vg (
T)
+
G(
(N,
T, 0))
g (0) (2)
If
he does not implement the project and
Vg (P,T,
) = vg
(
(
)
– T) +
G(α(P,T,BH)) (3)
+ (1
-
)
G (α (P,T,BL))] vg (0)
If
he does. If the incumbent is bad, his expected payoff will be
Vb(N,T,
) =
b (
T,T)
+
G(α(N,T,0)
b*(0,0) (4)
If
the project is not implemented and
Vb (P, T,
)
=
b
(
(
)
T, Rs + T) +
[
G(α(P,T,BH))
(5)
+ (1
-
)
G (α (P,T,BL))}
b*(0, 0)
If
it is.
The first-period strategy that
maximizes the incumbent’s expected payoff will obviously depend on the
citizen’s beliefs. In this game, as in others, there exist equilibria than
depend on rather unnatural out-of-equilibrium beliefs. Thus there exist equilibria
in which the incumbent (whether good or bad) always makes cash transfers to the
special interest.
We shall focus on equilibria in
which the citizen’s beliefs (on and off the equilibrium path) satisfy that ,
ceteris paribus, a first-period record with lower cash transfers cannot result
in more pessimistic beliefs about the incumbent. The citizen has monotonic beliefs if, for any paid of
first-period records (D,T,B) and (D,T’,B) such that T’ > T, α(D,T,B)
α(D,T,B).
We shall refer to an equilibrium with this property as an equilibrium with monotonic beliefs (EMB).
A good incumbent will never choose
to make cash transfers to the special interest in the first period. Making such
transfers lowers his first-period utility and, if the citizen has monotonic beliefs,
reduces his probability of reelection. If the citizen observes the incumbent
making a cash transfer, he will conclude that he is bad and vote him out of
office. This implies that if a bad incumbent does choose to make cash transfers
in the first period, he might choose those actions that maximize his
first-period utility. When
=
0, this means not
implementing the project and choosing the cash transfer T0. When
=
1, this implies
undertaking the project
and selecting the
cash transfer T1. =
T*(
(
1),Rs). We
therefore have the following result.
In an EMB, a good incumbent chooses
(P, 0) or (N, 0). A bad incumbent chooses (P, 0), (N, 0), or (P,T1)
when
=
1 and (P, 0), (N,0), or
(N,T0) when
=
0.
We let
gP.(
)
(
gN.(
))
denote the probability that a good incumbent will choose (not) to implement the
project when the probability that it yields high benefits is
.
A good incumbent will be said to behave
efficiently if he always implements (does not implement) the project when
=
1
(
=
0.), that is, if
gP. (
1) = 1 and
gN. (
0) = 1. Similarly, we let
bP. (
)
denote the probability that a bad incumbent chooses (P, 0) when the probability
that the project will yield high benefits is
,
bN (
)
denote the probability that he chooses (N,0), and
bU (
)
denote the probability that he acts in an unconstrained manner. Acting in an
unconstrained manner involves choosing (P,T1) when
=
1 and (N, T0)
when
=
0. Again, we shall say
that a bad incumbent behaves efficiently
if he always chooses (not) to implement the project when
=
1
(
=
0), that is if
bP, (
1)
+
bU, (
1) = 1 and
bN(
0) +
bU (
0)
= 1. A
bad incumbent make inefficient transfers
to the special interest if he ever implements the project when
=
0,
that
is, if
bP (
0)
> 0.
If the incumbent’s
first-period record in equilibrium is (P,0,BH), then Bayes’s rule
implies that
(P,0,BH)
=
(6)
The numerator is the probability
that a good incumbent would generate this record, and the denominator is the
probability that either type of incumbent would generate it. If the citizen
observes the records (P,T1, BH), (P,T1, BL),
or (N,T0, 0),
will
equal zero.
III. Propositions
Proposition
1:
Under assumptions 1-3, there exists some
such that, in any EMB, at least one type of
incumbent behaves inefficiently if the incumbent’s initial reputation
,
exceeds
Proof. Define
from the equality
(7)
That
follows from assumption 3). Here the first step
involves showing that in any EMB in which both types of incumbent behave
efficiently, a bad incumbent does not make positive cash transfers if
.
If both types of incumbent behave
efficiently, then when
, a good incumbent selects (N,)) and a bad
incumbent selects (N,0) or(N,T0); when
,
a good incumbent selects (P,0) and a bad incumbent selects (P,0) or (P,T1).
Thus the citizen’s beliefs must be such that
and
If in equilibrium a bad incumbent
chooses cash transfers, then he must choose either (N,T0) when
or (P,T1) when
. Consider the first possibility. Since
(N,T0,
0) must equal zero, his payoff from choosing (N,T0) is
(0,0).
But the payoff from selecting (N, 0) is
(0, 0)
Since
(7)
implies that this payoff exceeds
(0,0). This equilibrium cannot involve his
choosing (N,T0). The possibility of his choosing (P,T1)
is similarly eliminated.
Suppose that there existed such an
equilibrium. Then
and the equilibrium payoff to a bad
incumbent from choosing (N,0) when
is
(0,0) +
(0,0).
Assumption 2 implies that this is less than
(0,0),
which is the payoff from choosing (P,0)
a contradiction.
If a bad incumbent’s initial
reputation is high, he is unwilling to lose it by making cash transfers to the
special interest. It follows that equilibrium cannot involve such transfers. In
any efficient equilibrium, therefore, there can be no reputational penalty for
simply implementing the project (i.e.,
(P,0,BH)
=
(P,0,BL) =
(N,0,0)).
However, if this were the case, a bad incumbent would always implement the
project irrespective of
,
which is inefficient. Thus equilibrium cannot be efficient.
Assumption
4. (i)
(ii)
Part I states that the utility gain
of generating the citizen an expected utility increase of
.exceeds
the discounted value of one period of ego rent. Part ii states a similar
condition for the utility loss from generating the citizen an expected utility
gain of
.
Proposition 2. Under assumptions 1-3,
there exists
(0,
1) such that a bad incumbent’s always choosing (P, 0) and a good incumbent’s behavior
efficiently is an EMB if the incumbent’s initial reputation exceeds
.
Moreover, if assumption 4 is satisfied, this is the unique EMB.
Proof: If a good incumbent
behaves efficiently and a bad incumbent always chooses
the citizen’s beliefs along the equilibrium
path given by:
Now define the function
as follows:
Note that
is continuous and increasing in both its
arguments and that
Let
be the smallest value of
such that
and
Observe that assumption 1, 2 and 3
and the properties of
guarantee that such a value exists and is an
element of (0, 1).
We now demonstrate that, for
,
a bad incumbent’s always choosing (P, 0) and a good incumbent’s behaving
efficiently is an EMB with out-of-equilibrium beliefs given by
for all T > 0. We first check that a good
incumbent behaves efficiently. It is clear that he will choose (N, 0) when
,
since this is his one – period optimum
and it gets him re-elected with probability one. When
,
the definition of
implies that payoff from ( P , 0 ).
exceeds the payoff from selecting ( N, 0 ),
Next we check that a bad incumbent
always wants to choose (P, 0). When
,
the payoff from choosing (P, 0) is
Since
is
increasing in
,
this exceeds the payoff from choosing
when
Thus it must exceed the payoff from choosing
which
is independent of
The
definition of
guarantees that the payoff from choosing
exceeds
the
payoff from
To complete the proof, we must show that this is the unique EMB under
assumption 4 if
As noted in the text, assumption 4 implies
that a good incumbent always behaves efficiently. This implies that the citizen’s
beliefs must satisfy
Moreover, if in equilibrium a bad
incumbent chooses
or
,
the citizen’s beliefs at these first-period records must be that the incumbent
is bad. Since the relative benefit of selecting
has not decreased, it follows from our earlier
argument that the bad incumbent’s choice of
is the only possible outcome.
In equilibrium, a bad incumbent knows that if
he chooses to make direct cash transfers, his type will be revealed and he will
be voted out of office. An alternative way of transferring extra income to the
special interest is to undertake the project when
=
.
While the citizen understands the bad incumbent’s incentives, he cannot
perfectly infer from a record of introducing the project that an incumbent is
bad. The reason is that a good incumbent undertakes the project when
=
and the citizen cannot observe the realization
of
.
The citizen is unable to commit to
a voting strategy ex ante. If the citizen could commit, transfers would be made
efficiency.
Assumption
5, (i)
and
(ii)
Part I simply reverse part I of
assumption 4, and part ii says that a bad incumbent would also be willing to
forgo the gains from unconstrained behavior when
=
to stay in office.
PROPOSITION 3.
Under
assumption 1-3 and 5, there exists
(0,1)
such that both types of incumbent always choosing (N,0) is an EMB if the
incumbent’s initial reputation exceeds
Proof. Define
by
the equality.
G (
)
= max
(8)
That
(0,
1) follows from assumption 5. We claim that if
I >
both
types of incumbent choosing (N,0) is an EMB supported by the out-of-equilibrium
beliefs
(D,T,B)
= 0 for all (D,T,B)
(N,0,0).
If both types of incumbent always
choose (N,0), then
(N,0,0)
=
I . Thus the payoffs to
the two types of incumbents from choosing (N,0) are
(0) +
(
I)
(0) and
(0,0)
+
G(
I)
(0,0). The maximum payoffs that the two types
of incumbents could get if they deviated are
(
1)) and
*
(
1)), Rs),
respectively. Equation (8) therefore guarantees that deviation is not
worthwhile.
PROPOSITION 4. The optimal reelection
rule induces the incumbent to behave efficiently.
Proof: Let {(
(
0),
(
0)), (
(
1),
(
1))} be the incumbent’s
first- period choices induced by the optimal reelection rule. If the incumbent
were not behaving efficiently, then there are two possibilities. The first is
that
(
0) =
(
1) =
,
so that the project is overprovided. The second is that
(
0
) =
,
so that the project is overprovided. We shall rule out each of these two
possibilities.
Suppose that first
(
0) =
(
1) =
.
We may assume without loss of generality that
(
0) =
(
1).
If
(
0)
(
1), it must be the case
that
(
0),
(
0))
=
(
*(0, 0).
Consequently, if
),
the incumbent could be induced to always select (
by setting
),
0) equal to zero. Let
denote
common value of the cash transfer. Assumption 1 and the fact that the incumbent
is induced to select
when
imply
that
*(0, 0)
)
*(0,0),
where
. Now select any
and
such that
*
*
The citizen can induce the
incumbent to select
when
and
when
by
promising to reelect him with probability
if
he chooses (N, T), probability
if he chooses,
and
probability zero otherwise. Since
,
this dominates the incumbent’s choice of
in
both states. We conclude therefore that the optimal reelection rule cannot be
such as to induce the incumbent to underprovide the project.
Now suppose that
. Note first that assumptions 2 and 3 imply
that the citizen can induce the incumbent to select (P, 0) in each state by
simply promising not to reelect him if he does anything else. Thus we may
assume with no loss of generality that
.
For each
let
T (
. Then,
By part ii of assumption 1, for sufficiently small
.
)
.
For such an
,
the citizen can induce the incumbent to select (N, T (
)
when
and (P, 0) when
by
promising to reelect him with probability one if he chooses (N, T (
)
or (P, 0) and probability zero otherwise. Since , this dominates the incumbent’s choice of (P,
0) in both states. Thus the optimal reelection rule cannot be such as to induce
the politician to overprovide the project.
IV. Application
It is clear that there be both
policy and politician uncertainty. Without the latter, the incumbent would have
no reason to worry about his reputation. The writers in the Virginia focus on
the role of imperfect information about the effects of policies in generating
inefficiencies. This raises the question of whether the assumption of
politician uncertainty is superfluous.
At the time of the election, if
both incumbent and challenger are bad, the citizen will be indifferent as to
which one wins. Thus, if
(D,T,B)
denotes the citizen’s reelection rule (i.e., the probability that the incumbent
is reelected when his first-period record is (D,T,B), any specification of
(.) is consistent with optimizing behavior on
the part of the citizen.
The reelection rule employed by the
citizen does influence the incumbent’s first-period choices and hence the
citizen’s ex ante payoff. The reelection rule used in this equilibrium as the optimal reelection rule.
If there were no policy uncertainty
(i.e., the citizen could observe the realization of
),
then, under assumption 3, the incumbent could be induced to choose (N,0) when
=
0 and (P,0) when
=
1 by a reelection rule
that promises not to reelect him if he does anything else. If the reelection
rule promises to reelect the incumbent if and only if he selects (N,0) or
(P,0). Then, under assumption 2, the incumbent will choose to implement the
project when
=
0 , thereby making
inefficient transfers
The ‘public project’ in our model
has four key features. First, it indirectly benefits a special interest.
Second, it may or may not benefit the rest of society. Third, citizens have
less information about whether it will benefit them than politicians do.
Fourth, citizens cannot perfectly observe whether its implementation was in
their interest even ex post because its outcome is stochastic. The logic of our
argument suggests that any policy that shares these four features may be used
to redistribute even when cash transfers are both feasible and more efficient.
The first feature implies that the policy can be used to transfer resources to
special interests. The remaining features imply that the reputational penalty
for using the policy to make transfers may be less than that for making direct
cash transfers. By the second feature, even good politicians will implement the
policy under some conditions, and by the third and fourth features, the
citizens cannot observe whether these conditions are satisfied. Almost all
public expenditure projects have these four features.
Subsidy or regulatory policies that
purport to be in the public interest sometimes have these features. The
assumption of heterogeneity in politicians’ tastes is critical to the
inefficiency result. The model does not suggest why politicians should be of
two different types. It would certainly be more satisfying theoretically to
model the process by which individuals become politicians, we do not believe
that the assumption of politician heterogeneity is unreasonable. The model is
trying to capture here is differences in honesty and integrity.
A second criticism concerns the
limited number of policy instruments available to the politician. In reality,
there may exist instruments that a good politician could use to separate
himself from a bad politician when he introduces a project.
In reality, however, citizens are
likely to be highly uncertain (both ex ante and ex post) about the extent to
which a special interest gains from a particular public policy. Thus a bad
politician might choose to implement a project and impose no taxes, denying
that the special interest was gaining significantly. Provided that there is
some probability that this action would also be taken by a good politician, the
reputation penalty for doing it will be smaller than that for making cash
transfers. A more fundamental criticism concerns the limited notion of
political competition implicit in this type of model. In particular, the role
played by challenger is entirely passive. Thus while individual voters would
have no incentive to invest resources to find out whether a particular policy
was in the public interest, it would pay the challenger to find out this
information and inform the voters. In the context of our model, the challenger
would find out the realization of
and expose the incumbent if he had introduced
the project when
.
This would make the reputational penalty for implementing the project identical
to that from using cash transfers. If the voters believed what the challenger
told them, then the challenger would have an incentive to inform the voters
that the project was unwarranted whenever the incumbent introduced it. Thus
there is no reason why the voters should believe the information provided by
the challenger. Thus the reputational penalty for undertaking the project would
still be less than from direct cash transfers even if the challenger could
provide information.
V. Conclusion
We have analyzed the form of
transfers in a model of political competition in which politicians have
financial incentives to make transfers to a special interest and voters are
imperfectly informed. When there is asymmetric information about both the
effects of policy and the predispositions of politicians, inefficient methods
of redistribution may be employed. This reflects the fact that even politicians
who do not pander to special interest will sometimes introduce projects. In
characterizing the common features of such policies, we have refined and
clarified Tullock’s notion of a disguised transfer mechanism. However, we have
also pointed out that our model suggests that politician uncertainty is
necessary to explain the use of such mechanism. The mere existence of disguised
transfer mechanisms does not undermine the Chicago view. This analysis suggests
that the key to understanding the use of disguised transfers mechanisms is to
recognize that politicians are concerned with protecting their reputations.
