Economics ResourcesTaxes on Savings - GRUBER File.pdf.docx

Economics Resources/Taxes on Savings - GRUBER File.pdf

__MACOSX/Economics Resources/._Taxes on Savings -
GRUBER File.pdf
Economics Resources/Laffer - Ihori.pdf
Laffer paradox, Leviathan, and political contest

Author(s): Toshihiro Ihori and C.C. Yang
Source: Public Choice, Vol. 151, No. 1/2 (April 2012), pp. 137-
148
Published by: Springer
Stable URL: https://www.jstor.org/stable/41406919
Accessed: 18-11-2018 17:29 UTC
JSTOR is a not-for-profit service that helps scholars,
researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information
technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected]
Your use of the JSTOR archive indicates your acceptance of the
Terms & Conditions of Use, available at
https://about.jstor.org/terms
Springer is collaborating with JSTOR to digitize, preserve and
extend access to Public Choice
This content downloaded from 140.211.65.60 on Sun, 18 Nov
2018 17:29:19 UTC
All use subject to https://about.jstor.org/terms
Public Choice (2012) 151:137-148
DO! 10. 1007/sl 1 127-010-9737-z
Laffer paradox, Leviathan, and political contest

Toshihiro Ihori • C.C. Yang
Received: 17 January 2010 / Accepted: 20 October 2010 /
Published online: 3 November 2010
€> Springer Science+Business Media« LLC 2010
Abstract This paper considers a political contest model wherein
self-interested politicians
seek rents from the public budget, while general voters make
political efforts to protest
against politicians* rent seeking directly (for example, through
voting in referendums such
as the passage of Proposition 13) or indirectly (for example,
through donating money to
organized groups such as the National Taxpayer Union). We
show that the political contest
may ironically lead to the Laffer paradox; that is, rent-seeking
politicians may intend to
set the tax rate higher than the revenue-maximizing rate. For
taming Leviathans, political
protests may not be as effective as competition among
governments.
Keywords Laffer paradox • Leviathan • Political contest •
Revenue-maximizing rate
JEL Classification D72 F20 H41 H71
1 Introduction
Leviathan-type governments without constitutional constraints
impose taxes at a rate that
maximizes the tax revenue. This rate is higher than the rate that
maximizes social welfare
in the standard framework wherein social welfare depends on
useful public goods but not

T. Ihori (El)
Department of Economics, University of Tokyo, Hongo, Tokyo
1 13-0033, Japan
e-mail: ihori @e.u-tokyo.ac.jp
C.C. Yang
Institute of Economics, Academia Sinica, Nankang, Taipei 1
15, Taiwan
e-mail: [email protected]
C.C. Yang
Department of Public Finance, National Chengchi University,
Taipei, Taiwan
C.C. Yang
Department of Public Finance, Feng Chia University, Taichung,
Taiwan
Ô Springer
2018 17:29:19 UTC
138 Public Choice (2012) 151:137-148
on wasteful spending or rent. It is widely known that a strong
constraint can be set by the
constitution to limit the tax rate. However, other constraints are
required when constitutional
constraints are not available. Plausible candidates include
competition among governments
(as an exit) and pressure by voters (as a voice).

Institutional competition among governments may take the
form of an "exit" such
as tax competition, which is popular in theory and practice.1
Brennan and Buchanan
(1980), among others, showed that because of institutional
competition among governments,
revenue-seeking governments in a federation will end up on the
upward-sloping part of the
Laffer curve. On the contrary, Apolte (2001) indicated that
such a taming effect can only
be expected if a certain rule of competition among several
decentralized governments is ap-
plied. He suggested that federalism is not necessarily a
substitute for constitutional limits to
Leviathans.
In addition to institutional competition, it is important to
examine the role of political
pressure by general voters because the amount of rent seeking
is usually affected by the
voters' "voice," as pointed out by Hoyt (1999).2 See also
Cheikobossian (2008), Edwards
and Keen (1996), and Besley and Smart (2007).
Suppose that there are two types of public spending: wasteful
spending and useful spend-
ing. A rent-seeking government would prefer to increase the
share of wasteful spending by
conducting its political activities. On the other hand, the voters
also have an incentive to
perform their political activities or make efforts to reduce the
share of wasteful spending
and increase that of useful spending. The actual distribution of
tax revenue between use-
ful and wasteful spending is determined as the outcome of

political contests between the
rent-seeking government and the voters.
In this paper, we consider a simple formulation of a political
contest. In our approach,
the rent-seeking politicians and the general voters engage in a
political contest in terms
of resources. The greater the amount of political effort by the
voters (rent seekers), the
greater is the share of useful spending (wasteful spending) at
the given level of total tax
revenue. This political contest can result in a compromise. In
reality, voters make some
political efforts to influence budgetary outcomes through
voting, writing articles, lobbies,
and protests, while politicians make such efforts through
campaigns, logrolling, bribery,
and corruption.
Buchanan (1980) suggested a property right perspective on rent
seeking wherein rent-
seeking activities may be viewed as attempts to redefine
property rights. Our political con-
test model adopts this approach. More specifically, the voters
may have property rights over
the tax revenue collected nominally. However, these rights are
not secure, since they can be
altered or reallocated as a result of theft or rent seeking by the
politicians. Offense creates
a demand for defense, and hence, as first pointed out by
Wenders (1987), rent seeking self-
generates rent defending. Instead of remaining idle and
awaiting the outcome of politicians'
rent seeking, the voters may intend to protest against such
activities. As a parallel to cam-
paigns, logrolling, bribery, and corruption by politicians

aiming to exploit budgetary rents,
voting, writing articles, lobbies, and protests by voters against
exploitation by politicians
can be observed in the real world.
A natural conjecture about the outcome of a political contest is
that the equilibrium tax
rate will be set on the upward slope of the Laffer curve since
the political effort by general
voters imposes some degree of political constraints on rent-
seeking behavior. Contrary to
1 Tiebout (1956) and Hirsschman (1970) are two classical
works on the "exit" issue.
2 For research on rent-seeking, see Congleton et al. (2008).
Ф Springer
2018 17:29:19 UTC
Public Choice (2012) 151:137-148 139
this conjecture, we show that the rent-seeking politicians may
intend to set the tax rate
higher than the revenue-maximizing rate. This is mainly
because an increase in the tax rate
will engender a negative income effect on the political efforts
of voters.
The inclusion of a political contest leads to two main effects if
the tax rate is raised. First,

the corresponding increase in the tax revenue, if any, will
stimulate both the rent-seeking be-
havior of politicians and the rent-reducing behavior of voters in
the political contest. How-
ever, this tax revenue effect is nil at the revenue-maximizing
tax rate because the tax revenue
will not marginally change at this rate. Second, an increase in
the tax rate at the revenue-
maximizing point will undermine the political efforts of voters
by reducing their disposable
income. This negative income effect is beneficial to the rent
seeker in the political contest as
it, other things being equal, raises the relative share of tax
revenue allocated to him/her. The
second effect dominates the first effect at the top of the Laffer
curve, thereby leading to the
Laffer paradox. The main message conveyed by our paper is
that the "voice" of the general
public may not be as effective as competition among
governments at curbing politicians'
rent seeking.
We also consider an extended model in which politicians
exhibit neither completely self-
interested nor completely benevolent behavior. We show that if
the degree of a politician's
rent seeking is not very high, the Laffer paradox does not
occur. It occurs only if the degree
of politicians' rent seeking exceeds some threshold.
Shughart and Tollison (1991) and Wrede (1996, 1999), among
others, showed that in the

case of tax source sharing, revenue-seeking governments in a
federation will end up on the
downward-sloping part of the Laffer curve.3 In the present
framework, we assume away tax
source sharing but incorporate the cost of obtaining rent.
Interestingly, politicians still intend
to set the tax rate at a level higher than the revenue-
maximizing rate.
The rest of the paper is organized as follows. Section 2
presents the analytical framework.
Section 3 considers the political contest model where the
politician is a rent seeker, while
Sect. 4 examines a more general version of the model in which
the politician maximizes the
weighted sum of his/her rent and the welfare of voters. Finally,
Sect. 5 concludes the paper.
2 Basic model
2. 1 Analytical framework
We develop a simple budgetary model in which the rent-
seeking politicians (RPs) and the
general voters (VTs) interact in a small open economy.
The government not only provides useful public goods G but
also engages in wasteful
spending S. Public good G is beneficial to the voters, whereas
wasteful spending S is ben-
eficial to the rent-seeking politicians. Following the tradition
of Leviathan models of gov-
ernment, as in Brennan and Buchanan (1980) and others,
politicians prefer wasteful public

spending (5), which provides them with opportunities to
enhance their personal welfare.
The relative price of public and private goods is set to unity for
simplicity. Let r denote
the tax rate, У, the total income, and г У, the total tax revenue.
The government budget
constraint is given as follows:
С+5=гУ, (1)
3See also Anderson et al. (1989).
Ф Springer
2018 17:29:19 UTC
140 Public Choice (2012) 151:137-148
where S denotes the gross wasteful spending or gross rent of
RP. With regard to the budget
constraint of politicians, we have
5=S + a, (2)
where S represents the net wasteful spending or net rent of RP,
and a political spending or
efforts by RP. The objective of the representative RP is to
maximize S.
On the other hand, the social welfare, W, which reflects VTs
preferences over public
goods G and private consumption c, is given by

W = u(G) + h(cl (3)
where u(G) denotes utility from public consumption G, with u!
> 0 and u" < 0, and h(c)
refers to utility from private consumption с with b! > 0 and h"
< 0.
VTs are consumers and investors in the economic sphere. They
engage in private in-
vestment &, which has the productive effect of raising income,
and thereby, tax revenue.
Moreover, к may be regarded as the various efforts made to
increase private income, such
as physical investment, human investment, or labor supply. We
assume that Y is dependent
on private investment by the private sector with Y = £ f(k) =
nf(k ), where / is the per
capita income and n is the number of general voters. The
function / is assumed to satisfy the
standard condition: /' > 0, and f" < 0. Henceforth, we assume n
= 1 for simplicity; this
implies that the free-rider problem does not exist among VTs,
which provides them with the
best scenario for dealing with RPs. However, the main result of
our paper will qualitatively
hold even if we allow for the case wherein n > 1 (see Sect.
3.4).
VTs also make political efforts e . These political efforts may
be direct, for example,
through voting in referendums such as the passage of
Proposition 13,4 or indirect, for ex-
ample, through donating money to organized groups such as the
National Taxpayer Union.5

The budget constraint of each voter is given as
c + e + k = ('-r)f(k). (4)
For simplicity, investment is assumed to produce output
instantaneously. Therefore, we may
use the static model.
2.2 Pure rent-seeking model
We first consider the pure rent-seeking model as a benchmark.
Without any political contest,
a = e = 0 and G = 0. RP is assumed to maximize S simply by
choosing r. The timing of
the game is as follows. First, RP chooses r to maximize S .
Then, VT determines к and c.
The first-order condition with respect to к for VT is
(l-r)/'(*)=l. (5)
4 Proposition 13 endorsed by California voters to limit
property tax burdens is a renowned example.
5The National Taxpayer Union in the USA is "a nonprofit,
nonpartisan citizen group whose members work
every day for lower taxes and smaller government at all
levels." There are many other similar oiganized
groups, including the California Taxpayers' Association ("a
watchdog group founded in 1926 to protect tax-
payers from unnecessary taxes and to promote efficient, quality
government services'*) and World Taxpayers
Associations ("working together for lower taxes, less waste,
accountable government and taxpayer rights all
over the world"). The quotations here appear in the websites of
the respective organizations.

Springer
2018 17:29:19 UTC
Public Choice (2012) 151:137-148 141
VT's responses to r can be summarized by the functions
k = k( x) and
c = c( r).
It is clear that the total tax revenue, xY , also becomes a
function of r.
The optimal condition with respect to r for RP is given as
dS d(xY)
-j- dx = dx - - = °. (6) -j- dx dx
Since the total tax revenue is used solely for rent (or wasteful
spending), it is optimal for
RP to choose the tax rate that maximizes the total tax revenue.
We denote the revenue-
maximizing tax rate by Хм .
2.3 Pure benevolent model
We consider the pure benevolent model as the other
benchmark. A benevolent RP chooses

G and X to maximize
W = M(G) + /i[(l-r)/-*].
The timing of the game is as follows. First, RP chooses G and r
to maximize W. Then VT
chooses к to maximize W at the given G and r.
In the second stage of the game, the first-order condition with
respect to к is the same as
that in Sect. 2.2; that is,
(l-r)/'=l. (5)
As a result, VT's response functions for к and с are the same as
those in Sect. 2.2.
In the first stage of the game, the first-order condition with
respect to r (and hence G) is
= u'd-^- dx + *'[(1 - r )/' - 1]^ dx - tí f = 0. (7) dx dx dx
The second term reduces to zero owing to the first-order
condition with respect to k' that
is, (5). The third term represents the (negative) income effect
of raising tax on consumption.
Thus, at the optimum level,
- - > 0 since h f J > 0. dx > since J
In other words, the optimal tax rate set by the benevolent RP is
less than the revenue-
maximizing tax rate, г м. This is the standard result since the
benevolent RP considers the
marginal cost (negative income effect) of raising r on private
consumption с as well as the
marginal benefit from raising r on the provision of public good
G.

Э Political contest approach
Section 2 considers two extreme governments, namely, pure
rent seeking and pure benevo-
lent. These two extremes correspond to two broad types of
governments that are based on
the doctrine of self-interest and the doctrine of the common
good, respectively. In both the
Springer
2018 17:29:19 UTC
6See Nitzan (1994), Garfìnkel and Skaperdas (2006), and
Konrad (2007) for surveys of relevant literature.
7This form of the contest success function is widely employed
in studies on conflict/contest. See Konrad
(2007, Sect. 2.3) for its justifications. We discuss a more
general formulation in Sect. 3.4.
8 As noted by Garfìnkel and Skaperdas (2006), this property is
analytically convenient like the Cobb-Douglas
form in the case of production functions in neoclassical
economics, and this may be a reason for its popularity
among applications.
â Springer
142 Public Choice (2012) 151:137-148

models, RP is only allowed to choose tax rate r, and VT is only
allowed to choose invest-
ment k. In particular, RP's rent seeking is constrained only by
his/her ability to extract tax
revenues from VT through taxation. However, it is obvious that
VT makes political efforts to
influence budgetary outcomes through voting, writing articles,
lobbies, and protests. Like-
wise, it is obvious that RP makes such efforts through
campaigns, logrolling, bribery, and
corruption.
We now incorporate political efforts by RP and VT into the
pure model. The timing of
the game is as follows:
Stage I: RP determines the tax rate and his/her political effort.
Stage П: VT decides his/her investment, private consumption,
and his/her political effort.
Stage III: The political contest determines the actual
distribution of tax revenue between
useful and wasteful projects.
This formulation is a natural extension of the pure rent-seeking
model (Sect. 2.2) and the
pure benevolent model (Sect. 2.3) wherein RP is allowed to
choose a apart from tax rate r,
while VT is allowed to choose e apart from investment k. The
variable a represents RP's
political efforts to seek rents from the government budget,
while the variable e represents
VT's political efforts to oppose RP's rent seeking.
3.1 Stage III

RP's political efforts to exploit budget rents and VT's political
efforts to oppose RP's ex-
ploitation trigger a conflict or contest between RP and VT. The
conflict/contest involved
is presumably complicated, but a key factor used to determine
the "output" of the con-
flict/contest is the "inputs" expended by players. Following the
seminal work of Tullock
(1980) and the ensuing literature,6 we adopt the "production
function" approach to the con-
flict/contest and assume that the outcome of the political
conflict/contest is a function of the
relative share of the political spending of players. Specifically,
RP's gross gain S is deter-
mined by
5=-^-гУ, a -te (8.1) a -te
whereas VT's gross gain G is determined by
G = -?-tY. (8.2)
a + e
The outcome of the political conflict/contest between RP and
VT is summarized by contest
success functions (8.1) and (8.2).7 These functions show that
an increase in a at the given e
results in an increased distribution of the "pie" r Y in favor of
RP but against VT and vice
versa. Moreover, the functions exhibit the property of
homogeneity of degree zero such that
the same proportional increase or decrease in a and e leaves the
conflict/contest outcome
unchanged.8

2018 17:29:19 UTC
Public Choice (2012) 151:137-148 143
The contest success function usually yields the probability of
winning or losing. This for-
mulation may then be justified if both RP and VT are
concerned with the expected division
of tax revenue. Alternatively, following Long and Vousden
(1987) and others, the contest
success function may be given a nonprobabilistic
interpretation: players expend resources
competing for a share of divisible rent rather than the entire
indivisible rent and, therefore,
the relative share of tax revenue is allocated according to the
relative share of players' polit-
ical efforts.
Note that (8.1) can be rewritten as
S = - - tY - a. (8.1)'
a- he
Thus, RP's net gain or rent S is given by the difference between
S and a. If a = 0, then
5= 5 = 0 according to (8. 1 ) and (8. 1 )'.
Moreover, note that an increase in e reduces с at a given level
of k. The gain in G for VT
is at the expense of private consumption. The central trade-off

faced by agents in the conflict
literature is between producing goods and exploiting what
others have produced (Garfinkel
and Skaperdas 2006). The main trade-off faced by VT in our
model is between investments
for producing goods, which can be used in the private or public
sector, and protests against
RP's rent seeking in the public sector.
3.2 Stage II
Next, the representative household (VT) maximizes W by
choosing his/her investment, con-
sumption, and political effort, taking RP's political effort and
the tax rate as given, and antic-
ipating the political contest constraint (8.2). Then, for the first-
order conditions with respect
to e and к , we have
u'Ge = tí and (9)
u'GxYTf' + - r) - 1] = 0, (10)
where G< = ä and g'y -
From the optimizing behavior of voters, we obtain the response
functions for e, к , and c.
In general, e, к, and с are formulated as functions of r and a.
However, since we are mainly
interested in the effect of г on e , that is, |^, at r = zM> it is
appropriate to separate the effect
of r Y on e from that of т on e. At r = z^, xY is fixed with
respect to r in the first-order
effect sense, such that the value of is the same between the two
formulations. Then, we

have
е = е(г,гУ,д), (11.1)
£ = £(г,гГ,я), and (11.2)
c = c(r, г Y, a). (11.3)
With regard to the partial derivatives of (1 1.1), (1 1.2), (1 1.3)
with respect to r, we have
= '[-u'Gxyf' A + h'f']h"(f - 1), (12.1) 3r A
Як 1
= ±[u"(Ge)2 1 + u'Gee+h"K-u'GtYf' + h'f ), and (12.2)
Зт A
& Springer
2018 17:29:19 UTC
144 Public Choice (2012) 151:137-148
^- = (/' - ')kx - eX9 respectively, (12.3)
ox
where
A m [u"(Ge)2 + m'G„ + h")[u'GrYTf" + h'f"( 1 - r) + A"[/'(l - г)
- 1 ](/' - 1)]
+ h"(f'-ì)[u"GeGtrTf'-h"[f'ì-T)-ì]],

Gee rr = - ,2аг л , and Л > 0 if the second-order condition is
satisfied. Note that these deriva-
rr ( a+e)¿
tives are meaningful only at the point of г = Тд/ . At this point,
= / + r/'fcT = 0.
Using (10), (12.1) reduces to
= !-/,"(/' -I)2. (12.1)'
ox Л r
Since h" < 0, we have ex = < 0 (unless /' = 1). In addition,
(12.2) reduces to
^ = 7[«"(G,)2 A + M'G„ + h")-(f- 1). (12.2)' ox A T
The sign of kT = II depends on the sign of /' - 1 . On the other
hand, since = / +
r/'fcT = 0 at r = tm, кт is negative at this point. Considering
(12.2)', it follows that /' > 1,
and hence, ex < 0 at that point.
Note that the sign of ex is generally ambiguous at т Ф xM. In
other words, an increase
in r normally reduces the disposable income, which is the
negative income effect. On the
other hand, an increase in r raises the total tax revenue if r <
rM, which stimulates political
effort e; this may be called the tax revenue effect. If the
positive tax revenue effect dominates
the negative income effect, an increase in r would simulate
political effort e. However, the
total tax revenue effect is absent at the revenue-maximized
point гд/, and hence, an increase
in г undermines political effort e .

3.3 Stage I
Here, the rent-seeking RP maximizes 5 by choosing his/her
political effort a and tax rate r,
anticipating the political contest outcome (8.1) and VT's
response functions. The first-order
condition with respect to a reduces to
( e-aea)xY = (a- he)2, (13)
The left-hand side of (13) represents the marginal benefit of
increasing a and the right-hand
side indicates the marginal cost of increasing a for RP.
At the same time, the effect of the tax rate on S at r = хм can
be written as
3S _dxY[ a v aexY 1 v aeT
Эх _dxY[ dz [ a + a e v (û aexY + e)2 J 1 v (a- aeT be)2' ( }
where denotes the derivative of total tax revenue with respect
to г and exY = 377. By
definition, = 0 at r = xM* and hence, the first term of (14)
reduces to zero at this point.
On the other hand, since ex < 0, the second term of (14) is
positive, and hence, > 0 at
X = хм- In other words, ~¡r < 0 at RP's optimal choice of r,
which implies that the optimal
level of X set by RP is higher than xM. We can call this the
Laffer paradox.
Springer

2018 17:29:19 UTC
Public Choice (2012) 151:137-148 145
3.4 Some remarks
First, let us consider the case of n > 1 and compare the
cooperative and non-cooperative
solutions. Suppose that VTs behave noncooperatively in
choosing e at Stage II. (8.2) is now
replaced by
G = ei+e-' r Y,
a +e,+ <?_,
where e_, = e¡. Each VT chooses e¡ to maximize
Wi = u { a+* + e-T [/,Л) + fjikj^' j + Л(0 " t)f(ki) ~ ki ~ e,)
by taking e-i as given. Then the first-order condition with
respect to e¡ gives
u'Ge = h' (9)'
where Ge = with e = e¡ for all i's under the symmetric
assumption. Note that Y =
f(k) if n = 1 but Y = nf(k) if n > 1.
In the cooperative case where VTs maximize £ W¡, the first-
order condition becomes
nu'Ge = h' . (9)"
Since VTs internalize the spillover effect of each member's

political effort at the coopera-
tive solution, the total marginal benefit of e is the sum of each
member's marginal benefit,
which is expressed in the left-hand side of (9)". Comparing the
two first-order conditions (9)'
and (9)", it is clear that the equilibrium level of e (VT's
political effort) at the noncooperative
solution is less than that of the cooperative solution.
Nevertheless, we can still show that in the noncooperative case,
Cx < 0 at the revenue-
maximizing point. This is because the first-order condition for
each VT in the noncoopera-
tive case is qualitatively the same as that in the cooperative
case as long as n < oo. If n goes
to infinity, then the non-cooperative solution implies that e = 0
and the equilibrium reduces
to the pure rent-seeking model of Sect. 2.2.
Second, our seemingly paradoxical outcome holds in more
general formulations of the
political contest as long as r is set before VT determines e'
therefore, an increase in г may
reduce e at r = xM. For example, consider the following
setting, which is more general
than (8.2):
3G 3G л

G = G(a,e,rY), Ga = - < 0, Ge = - > 0, л
da oe
3G л 3Ge л
rY=dxÝ> " = "э7 <
It can be shown that the Laffer paradox still occurs under this
formulation.
Ф Springer
2018 17:29:19 UTC
146 Public Choice (2012) 151:137-148
4 Degree of rent seeking
4.1 Analytical framework
In the basic framework presented in Sect. 2, we follow the
Weberian tradition and suppose
that politicians are distinct from voters.9 Specifically, it has
been assumed that politicians
adopt politics as a vocation and strive to make it their sole
source of income. In this section,
we relax this assumption.
We now consider that politicians themselves are identical to
voters, except that politi-
cians use their political influence to seek rents once they are in
power. Alternatively, to be

elected or reelected, rent-seeking politicians must also pay
attention to voter welfare. In any
case, politicians may exhibit neither completely self-interested
nor completely benevolent
behavior.
Suppose that many types of politicians or governments exist.
The types of governments
or politicians may be represented by their degree of rent
seeking, L. If the politicians are
only concerned with rent seeking, as in the pure rent-seeking
model developed in Sect. 2.2,
the degree of rent seeking is the highest and it is normalized as
unity. On the other hand, if
the politicians are purely benevolent and seek to maximize the
social welfare of voters, as
in Sect. 2.3, rent seeking is absent and its degree is normalized
as zero.
In general, the degree of rent seeking, L, is given between 0
and 1. This formulation of
0 < L < 1 is an interesting combination of pure Leviathan and
pure benevolent models. We
allow politicians to choose a besides r and allow voters to
choose e besides k. We consider
L to be exogenously given in our model. It can be perceived
that the actual L in a society
emerges from the electoral systems or political institutions of
the society; evidently, different
resulting Ls reflect the different qualities of these systems and
institutions.
Specifically, once a type of politician, L, is selected, the
objective of RP, E, is given as

E = LS + (1-L)W. (15)
Keen (1995) and Edwards and Keen (1996) use a similar
formulation.
4.2 Analytical result
Suppose that a type of politician, L, is exogenously given. The
objective of RP, Z, is given
as (15). Then the effect of the tax rate on E is given by
ЭЕ dS „ rdW
ã7 = iS7 + (1-"ã7' „ (l<*
where
3 S ЭгКГ a aety "I aet
3 дт~ S Эт [a + a e T (a aety + e)*'~ "I (a+<?)2' aet (17)
dW 'dxY' a aeTy "1 aex 1 ( dxY'
dW эг " c|"ãr[^T7 'dxY' a + ry(^TTj2J+ aeTy "1 (ï+ê)i' aex 1 +
Ас(Сг+СгУ"9Г) ( dxY' (18)
9For a discussion on the Weberian tradition of modeling
politicians, see Merlo (2006).
Springer
2018 17:29:19 UTC
Public Choice (2012) 151:137-148 147
ст = 1^, and cXY = ^7. Hence, at r = Тм , we have

ЭЕ ae г Г aer Л
ЭЕ э7 = -Lry bŽ? ae г + 0 - L)rcrK Г aer + hcCt[ Л (19)
Since ex < 0, the first term of (19) is positive. Examining the
sign of cT = f£ at r = xM
yields
cr = (/'-D*r-*r <0. (20)
With cr < 0, the second term of (19) is negative. If the second
term dominates the first term,
an increase in r at a given level of r Y would reduce E. In this
case, the Laffer paradox does
not occur.
From (19), it is clear that is increasing with L. Let us define L,
which satisfies = 0
at r = хм . Then > 0 at r = тм if and only if L > L. In other
words, the Laffer paradox
will occur if L > L and not occur otherwise. Note that if L = 0,
then < 0 at r = тм must
be true. It then follows that 0 < L < 1 .
In this general version of the rent-seeking model, the Laffer
paradox does not necessarily
occur, since the paradoxical possibility also depends on the
level of L. If the degree of a
politician's rent-seeking is higher, and his/her L is greater than
L, the Laffer paradox is
more likely to occur, and vice versa.
5 Concluding comments

Pure benevolent governments impose a tax rate at a level lower
than the revenue-maximizing
tax rate. On the other hand, pure Leviathan-type governments
impose taxes at the level that
maximizes the tax revenue. It is now widely recognized that
competition among govern-
ments can serve as an appropriate substitute for constitutional
constraints on the power of
politicians.
Instead of institutional competition, we have examined the role
of political protests as
limits to Leviathans. More specifically, we consider a political
contest model wherein self-
interested politicians seek rents from public budgets, while
general voters make political
efforts to protest against politicians' rent seeking directly (for
example, through voting in
referendums such as the passage of Proposition 13) or
indirectly (for example, through do-
nating money to organized groups such as the National
Taxpayer Union). It is shown that
ironically, the Laffer paradox can occur in the political contest
between rent-seeking politi-
cians and general voters. Therefore, we provide an example
where "voice" can increase,
rather than decrease, the tax rate.
We have explored the possibility that political protests may not
limit the power of politi-
cians. We do not claim that the Laffer paradox always occurs in
a political contest model. If
the degree of a politician rent seeking is low, the Laffer
paradox is less likely to occur. Our
model is admittedly highly stylized, and it abstracts from
several possible complications in

the real world. In particular, we focus on the conflict/contest
between voters and politicians
but ignore their heterogeneity. This excludes possible conflicts
among voters (for exam-
ple, various individuals or interest groups competing for
budgets, as addressed in Becker
1983) and among politicians themselves (for example,
politicians pursuing their own career
and personal interests and disagreeing over the distribution of
budgets as revealed in Baron
and Ferejohn 1989). Nevertheless, we hope that this paper has
highlighted the limitation of
"voice" in constraining the power of politicians and served as a
meaningful attempt toward
attaining a relatively complete solution for containing
Leviathans.
Springer
2018 17:29:19 UTC
148 Public Choice (2012) 151:137-148
Acknowledgements Earlier versions of the paper were
presented at the Public Choice Society meetings
in Las Vegas, European Public Choice Society meetings in
Athens, and Australasian Public Choice Confer-
ence in Melbourne in 2009. We would like to thank the
participants, referees, and the editor for their useful
comments.
References

Anderson, G. M., Shughart, W. F. II, Robert, D., & Tollison, R.
D. (1989). Political entry barriers and tax
incidence: The political economy of sales and excise taxes.
Public Finance/Finances Publiques , 44,
8-18.
Apolte, T. (2001). How tame will Leviathan become in
institutional competition? Public Choice , /07, 359-
381.
Baron, D., & Ferejohn, J. (1989). Bargaining in legislatures.
American Political Science Review , 83, 1181-
1206.
Becker, G. (1983). A theory of competition among pressure
groups for political influence. Quarterly Journal
of Economics y 98 , 371-400.
Besley, T., & Smart, M. (2007). Fiscal restraints and voter
welfare. Journal of Public Economics , 9/, 755-
773.
Brennan, G., & Buchanan, J. M. (1980). The power to tax:
analytical foundations of a fiscal constitution.
Cambridge: Cambridge University Press.
Buchanan, J. M. (1980). Rent seeking and profit seeking. In J.
M. Buchanan, R. D. Tollison, & G. Tullock
(Eds.), Toward a theory of the rent-seeking society (pp. 3-15).
College Station: Texas A&M University
Press.
Cheikobossian, G. (2008). Heterogeneous groups and rent-
seeking for public goods. European Journal of
Political Economy , 24, 133-150.

Congleton, R. D., Hillman, A. L., & Konrad, K. A. (Eds.)
(2008). Forty years of rent-seeking research.
Heidelberg: Springer.
Edwards, J., & Keen, M. (1996). Tax competition and
Leviathan. European Economic Review , 40 , 1 13-134.
Garfìnkel, M. R., & Skaperdas, S. (2006). Economics of
conflict: an overview. In T. Sandler & K. Hartley
(Eds.), Handbook of defense economics (Vol. 2). Amsterdam:
Elsevier.
Hirsschman, A. O. (1970). Exit, Voice and Loyalty: Responses
to decline in firms, organization, and states.
Cambridge: Harvard University Press.
Hoyt, W. H. (1999). Leviathan, local government expenditures,
and capitalization. Regional Science and
Urban Economics , 29, 155-171.
Keen, M. (1995). Pursuing Leviathan: fiscal federalism and
international tax competition. In IIPF 51st
Congress , Lisbon.
Konrad, K. A. (2007). Strategy in contests - an introduction,
book manuscript. Berlin: WZB.
Long, N. V., & Vousden, N. (1987). Risk-averse rent seeking
with shared rents. Economic Journal , 97, 971-
985.
Merlo, A. (2006). Whither political economy? theories, facts
and issues. In R. Blundell, W. K. Newey, &
T. Persson (Eds.), Advances in economics and econometrics
(pp. 381-421). Cambridge: Cambridge

University Press.
Nitzan, S. (1994). Modelling rent-seeking contests. European
Journal of Political Economy , /0, 41-60.
Shughart, W. F. II, & Tollison, R. D. (1991). Fiscal federalism
and the Laffer curve. Economia Delle Scelte
Pubbliche , /, 21-28.
Tiebout, С. M. (1956). A pure theory of local expenditure.
Journal of Political Economy , 54, 416-424.
Tullock, G. (1980). Efficient rent-seeking. In J. M. Buchanan,
R. D. Tollison, & G. Tullock (Eds.), Toward a
theory of the rent-seeking society (pp. 97-1 12). College
Station: Texas A&M University Press.
Wenders, J. T. (1987). On perfect rent dissipation. American
Economic Review , 77, 456-459.
Wrede, M. (1996). Vertical and horizontal tax competition: will
uncoordinated Leviathan end up on the wrong
side of the Laffer curve? Finanzarchiv, 53 , 461-479.
Wrede, M. (1999). Tragedy of the fiscal common: fiscal stock
externalities in a Leviathan model of federalism.
Public Choice, 101 , 177-193.
Ф Springer
2018 17:29:19 UTC
Contentsp. [137]p. 138p. 139p. 140p. 141p. 142p. 143p. 144p.
145p. 146p. 147p. 148Issue Table of ContentsPublic Choice,
Vol. 151, No. 1/2 (April 2012) pp. 1-407Front MatterAn
evaluation of EU regional policy. Do structural actions crowd

out public spending? [pp. 1-21]Bargaining unexplained [pp. 23-
41]Incumbent positioning, ideological heterogeneity and
mobilization in U.S. House elections [pp. 43-61]Voter
uncertainty and failure of Duverger's law: an empirical analysis
[pp. 63-90]Inequity and risk aversion in sequential public good
games [pp. 91-119]Coalition incentives for political budget
cycles [pp. 121-136]Laffer paradox, Leviathan, and political
contest [pp. 137-148]Bureaucrats and short-term politics [pp.
149-163]Fiscal decentralization and natural hazard risks [pp.
165-183]Islam and democracy [pp. 185-192]Social identity and
voting behavior [pp. 193-214]Do ideological and political
motives really matter in the public choice of local services
management? Evidence from urban water services in Spain [pp.
215-228]þÿ�þ�ÿ��T��h��e��
��e��c��o��n��o��m��i��c��
��e��f��f��e��c��t��s��
��o��f��
��f��e��d��e��r��a��l��i��s��
m�� a��n��d��
��d��e��c��e��n��t��r��a��l��i
��z��a��t��i��o��n��a��
��c��r��o��s��s��-
��c��o��u��n��t��r��y��
��a��s��s��e��s��s��m��e��n��
�t�� [��p��p��.��
��2��2��9��-
��2��5��4��]Determinants of government size:
evidence from China [pp. 255-
270]þÿ�þ�ÿ��C��h��i��n��a��'��s��
��e��v��o��l��u��t��i��o��n��
��t��o��w��a��r��d�� a��n��
��a��u��t��h��o��r��i��t��a��r
��i��a��n��
��m��a��r��k��e��t��
��e��c��o��n��o��m��y��a��
� ��p��r��e��d��a��t��o��r��-

��p��r��e��y��
��e��v��o��l��u��t��i��o��n��
a��r��y�� m��o��d��e��l��
��w��i��t��h��
��i��n��t��e��l��l��i��g��e��n
��t�� d��e��s��i��g��n��
��[��p��p��.�� 2��7��1��-
��2��8��7��]European monetary policy and the
ECB rotation model: Voting power of the core versus the
periphery [pp. 289-323]Do elections affect the composition of
fiscal policy in developed, established democracies? [pp. 325-
362]Do the IMF and the World Bank influence voting in the UN
General Assembly? [pp. 363-397]BOOK REVIEWSReview:
untitled [pp. 399-401]Review: untitled [pp. 403-404]Review:
untitled [pp. 405-407]Back Matter
__MACOSX/Economics Resources/._Laffer - Ihori.pdf
Economics Resources/Structure and Coherence in the Political
Economy of Public Finance - Winer File.pdf

__MACOSX/Economics Resources/._Structure and Coherence
in the Political Economy of Public Finance - Winer File.pdf

Economics Resources/Fiscal Competition - Wildasin File.pdf

__MACOSX/Economics Resources/._Fiscal Competition -
Wildasin File.pdf
Economics Resources/Positive Principles of Taxation -
HOLCOMBE File.pdf

__MACOSX/Economics Resources/._Positive Principles of
Taxation - HOLCOMBE File.pdf
Economics Resources/Federalism - Kessler.pdf
Communication in Federal Politics: Universalism, Policy
Uniformity, and the Optimal

Allocation of Fiscal Authority
Author(s): Anke S. Kessler
Source: Journal of Political Economy , Vol. 122, No. 4 (August
2014), pp. 766-805
Published by: The University of Chicago Press
Stable URL: https://www.jstor.org/stable/10.1086/676404
JSTOR is a not-for-profit service that helps scholars,
researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information
technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected]
Your use of the JSTOR archive indicates your acceptance of the
Terms & Conditions of Use, available at
https://about.jstor.org/terms
The University of Chicago Press is collaborating with JSTOR
to digitize, preserve and extend access
to Journal of Political Economy
This content downloaded from
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
https://www.jstor.org/stable/10.1086/676404
Communication in Federal Politics:

Universalism, Policy Uniformity, and
the Optimal Allocation of Fiscal Authority
Anke S. Kessler
Simon Fraser University
The paper presents a positive model of policy formation in
federal leg-
islatures when delegates engage in the strategic exchange of
policy-
relevant information. Depending on the type of policy under
consider-
I. I
The
Daro
Torst
Instit
and s
with
dian
Coun
usual
[ Journa
© 2014
ation,communicationbetweendelegatesgenerallysuffersfromabias
that
makes truthful communication difficult and sometimes

impossible. This
generates inefficient federal policy choices that are often
endogenously
characterized by overspending, universalism, and uniformity.
Building
on these findings, I develop a theory of fiscal ðde-
Þcentralization, which
revisits the work of Oates in a world of incomplete information
and
strategic communication. Empirical results from a cross section
of US
municipalities are consistent with the predicted pattern of
spending.
ntroduction
majority of countries are federally organized; multiple tiers of
gov-
ernment fulfill a variety of functions in revenue raising,
taxation, and
public expenditure. Among those, there has been a recent trend
toward
I am indebted to Philip Reny ðthe editorÞ for his detailed and
thoughtful comments on
an earlier draft and two anonymous referees for valuable
suggestions. I also wish to thank
n Acemoglu, Philippe Aghion, Elhanan Helpman, Kevin
Milligan, Krishna Pendakur,
en Persson, Guido Tabellini, as well as various participants of
the 2012 International
ute of Public Finance congress, the 2008 Canadian Public
Economic Group Meeting,
eminars at the Universities of Basel, Cologne, and Munich for

helpful discussions,
special thanks belonging to Tim Besley and Ken Shepsle. I am
grateful to the Cana-
Institute for Advanced Research and the Social Science and
Humanities Research
cil for financial support and to Ross Hickey for invaluable
research assistance. The
disclaimer applies.
l of Political Economy, 2014, vol. 122, no. 4]
by The University of Chicago. All rights reserved. 0022-
3808/2014/12204-0003$10.00
766
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
reallocating fiscal responsibilities from central governments to
regional
or local governments. Examples of countries that decentralized
include
the United States, the United Kingdom with its newly created
regional
communication in federal politics 767
legislatures in Scotland and Wales, a number of Latin American
coun-
tries ðFaguet 2004Þ, as well as Indonesia and Pakistan.1
Decentralization
is also one of the World Bank’s stated policy objectives for

developing
countries.2 At the same time, the debate among member
countries of the
European Union has evolved around “subsidiarity” versus
centralization,
that is, which functions should remain with the regions and
which can
sensibly be assumed by the European Parliament and the
Council.
In popular discussions decentralization is often seen as
preferable, fos-
tering efficiency through intergovernmental competition and
account-
ability through local say over service provision. The economic
literature
building on Oates’s ð1972Þ famous decentralization theorem
promotes a
more balanced view by emphasizing a trade-off between local
policy de-
cisions that are better tailored to the needs of the local
population ð“closer
to the people”Þ and the obvious advantages of centralization if
policies
exhibit large economies of scale or spillover effects across
jurisdictions.
While Oates’s perspective has shaped the theory of fiscal
federalism for
decades, it has recently come under criticism on the grounds
that an es-
sential aspect of policy formation is missing, namely, an
understanding
of how political actors on various levels of government are
incentivized
and interact.
This paper addresses fiscal federalism from a political economy

per-
spective. It has two purposes. First, I develop a model of
legislative be-
havior that allows for communication as a key factor in
legislative deci-
sion making on the federal ðcentralÞ level.3 In a second step, I
am then
1 In the United States, the most prominent explicit
decentralization initiative was to re-
turn responsibility over major welfare programs to the
individual states through the Per-
sonal Responsibility and Work Opportunity Reconciliation Act
of 1996, under which fund-
ing for state-run welfare programs switched from open-ended
matching grants to fixed
block grants and simultaneously increased the discretion of
states to make decisions re-
garding their own welfare expenditures. Although the Obama
administration has moved
toward more centralized provision of some aspects of
government like health insurance,
federalism is still important when it comes to policies such as
the stimulus package and aid
to states, which has totaled $2.4 trillion over the past 3 years
ðsource: Congressional Budget
OfficeÞ.
2 In 2002, around 30 World Bank projects had decentralization
components. A total of
US$500 million was given in the form of various loans to
countries such as Mexico, Ar-

gentina, Brazil, India, Mexico. and Pakistan. See
http://www1.worldbank.org/publicsector
/decentralization/http://www1.worldbank.org/publicsector/decen
tralization/.
3 Communication in legislatures has received surprisingly little
attention in the literature.
Austen-Smith ð1990Þ analyzes the informational role of debate
in federal legislatures in a
series of examples with majority voting over fixed or
endogenous agendas. The author shows
that communication allows legislators who would otherwise
reveal their information through
proposals to share this information prior to the agenda-setting
stage. Gilligan and Krehbiel
ð1989Þ, Epstein ð1998Þ, and Krishna and Morgan ð2001Þ study
legislatures in which informed
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
able to compare federal decisions to local decisions in order to
gain in-
sights into the optimal degree of ðde-Þcentralization and the
determi-
nants of fiscal authority.
768 journal of political economy
The analytical framework I employ conceptually formalizes the
process
of federal policy formation through consultation with interested

parties
ðlocal officialsÞ or, alternatively, a policy debate among local
delegates in
a federal assembly. The aim is to capture the stylized elements
of a coun-
try with multiple layers of government and to study which
factors deter-
mine the incentives of local representatives to truthfully reveal
locally dis-
persed information on their constituents’ preferences over
public policy
decisions. The economy is divided into a given number of
ðgeographicalÞ
districts, which are defined by common tastes over a local
public project
ðgoodÞ with spillover effects on other districts. Local
preferences are local
knowledge. Under federal fiscal authority, districts ðthrough a
represen-
tativeÞ can communicate their preferences to the central
government,
which implements whichever policy it considers best given the
transmit-
ted information.4 Federally controlled public policies are
funded through
general taxation, which implies that costs are underestimated at
the local
level ceteris paribus. As I show, this implies that local
representatives have
an incentive to overstate the local benefit, on average, in order
to seek
federal ratification. Moreover, this tendency unambiguously
increases in
the number of districts and is reinforced for projects whose
benefits are
locally concentrated ð“distributivepolitics”Þ. Forpolicies that

havea public
good character, however, this tendency is mitigated. Intuitively,
delegates
will have an incentive to understate the local benefit
ðrespectively, over-
state the local costÞ for projects that are costly for their
constituents but—
because of their positive and large spillovers—are nevertheless
likely to
receive federal approval. Irrespective of the type of policy
under consid-
eration, the federal legislature recognizes the resulting
communication
bias, implying that meaningful transmission of information
becomes dif-
ficult or even impossible. In the former case, the equilibrium is
charac-
terized by universalism: every interested district is assured a
project. In the
latter case, federal policies suffer from uniformity of provision:
although it is
commonly known that local preferences ðgenericallyÞ differ,
those differ-
ences fail to be taken into account when the policy decision is
made. The
committees can communicate information through policy
proposals ðsubmitting billsÞ to an
uninformed floor. The objective of these papers is to understand
how different rules for
adding amendments affect equilibrium informational efficiency.
In an abstract model out-
4 In other words, the central government cannot commit to
ðmessage-contingentÞ pol-
icies that are different from what maximizes its own objective.
Communication therefore

takes the form of cheap talk ðCrawford and Sobel 1982Þ.
side the political economy context, Dessein ð2002Þ studies
communication vs. delegation in a
model that is similar to mine but has only two actors, a
principal and an agent. The agent has
private information that is relevant to the decision, which the
principal can either make
herself ðafter communicating with the agentÞ or delegate.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
failure is rooted in the strategic considerations between the
federal gov-
ernment and local officials ðbetween local representatives in a
federal as-
sembly, respectivelyÞ, which entirely prevents the credible
transmission of
any information. Thus, my results provide a theoretical
foundation for
two stylized facts about federal politics that have been
frequently invoked
as assumptions in the theoretical literature, namely, that the
legislative
process is characterized by universalism and that federal
policies are uni-
form, that is, do not take local circumstances into account.
On the basis of the above findings, I next compare the

equilibrium
policy under centralization with the policies that would have
been chosen
under decentralized decision making. The results give some
understand-
ing of what determines which level of government should have
author-
ity over a policy. The reasoning here, although conceptually
different
from Oates’s original argument, is similar in its conclusions:
centraliza-
tion is better at internalizing externalities but worse at
accommodating
the needs of the local population. In this sense, the model
provides a
foundation for Oates’s celebrated decentralization theorem that
is de-
rived from an explicit model of government behavior: policy
uniformity is
not assumed from the outset but is a direct consequence of the
impossi-
bility of credibly communicating information about local tastes
to higher
levels of government in equilibrium. Because the problem is
more severe
the larger the number of districts, however, I also find a
congestion effect,
which takes the form of diseconomies of scale in
communication. As a
result, centralization is better only if there are not too many
districts, ce-
teris paribus.
The theoretical model has several implications regarding the
rela-

tionship between the size of the legislature and both the level
and the
composition of public spending. In the final part of the paper, I
show
that the data are consistent with these predictions. In particular,
I doc-
ument that the number of legislators has a strong positive effect
on the
extent of government spending, while at the same time it has a
signifi-
cant negative effect on the relative share of targetable
expenditures in
the budget. This second main prediction of the theory, namely,
that a
larger legislative body should be associated with a shift in the
composi-
tion of the government budget away from pork toward relatively
more
public goods, ceteris paribus, is unique to the model and, to the
best of
my knowledge, has never been empirically documented before.
Using
an instrumental variable strategy, I show that these findings are
robust to
a possible endogeneity of legislative size.
Related literature.—This paper stands at the intersection of
several
strands of research in political economy. First, the paper
contributes to
the literature on legislative behavior and organization by
providing a ra-
tionale for “universalism” in Congress. This empirically
relevant feature
of legislative decision making generates what is sometimes
referred to as

��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
“the law of 1/n” ðWeingast, Shepsle, and Johnson 1981Þ, which
postulates
that ðthe inefficiency ofÞ government spending is increasing in
the size
of the legislature.5 Weingast et al. and, more recently, Primo
and Snyder
ð2008Þ formally derive this law in models that are very similar
to the pres-
ent framework but assume universalism from the outset and do
not con-
sider the possibility of delegates in the federal assembly
communicating
privately held information.6
Second, the empirical part of the analysis fits well into the
burgeoning
literature that examines the effects of political institutions on
outcomes
in politics and economic policy. In particular, Bradbury and
Crain ð2001Þ,
Gilligan and Matsusaka ð2001Þ, Baqir ð2002Þ, Perotti and
Kontopoulos
ð2002Þ, and Egger and Koethenbuerger ð2010Þ empirically
study the law
of 1/n and find support for the predicted positive relationship
between

government spending and legislature size. My results confirm
this find-
ing but go beyond these studies in that I am able to relate the
size of the
legislature to the composition of government spending as well.
For this
reason, the paper is also related to Persson, Roland, and
Tabellini ð2000Þ,
Lizzeri and Persico ð2001Þ, Milesi-Ferretti, Perotti, and
Rostagno ð2002Þ,
and Persson and Tabellini ð2004Þ, which study how electoral
rules and
the types of democratic political institutions matter for the level
and the
composition of government spending.7 The empirical work in
this liter-
ature mostly uses cross-country comparisons, however, which
are prone to
suffer from an omitted variables problem. Those are less likely
to arise if
one exploits within-country variation as I do in the present
paper.
Finally, the paper contributes to the literature on fiscal
federalism. In
particular, one key element in the classic theory of fiscal
federalism ðOates
1972Þ is that a central government will provide a single,
uniform level of
public output in all jurisdictions, a presumption that has come
under crit-
icism ðSeabright 1996; Lockwood 2002; Besley and Coate
2003Þ.8 In con-
5 There have been several earlier papers that explain the norm
of universalism. They rely

on a simple expected value comparison between the rewards to
legislators in the coalition
as the whole against the uncertainties associated with minimal
winning coalitions. As noted
by Niou and Ordeshook ð1985Þ, however, such a comparison
fails to model directly the
decision processes within a legislature and, thus, fails to
explain why it is in no member’s
self-interest to defect from such a norm.
6 Although powerful, the logic in Weingast et al. ð1981Þ and
Primo and Snyder ð2008Þ is
unsatisfactory from a theoretical point of view because the
legislature is assumed to adopt
the “norm” of universalism, unanimously passing omnibus bills
that ensure every legislator
the projects he or she desires, even though all members would
collectively be better off
agreeing to a lower level of overall spending.
7 See Besley and Case ð2003Þ for an overview of results with
an emphasis on US in-
stitutions.
8 Empirically, it is not difficult to document cases in which a
central government pro-
vides public goods in a discriminatory manner across regions,
although a tendency toward
“equal treatment” arguably remains. On the theoretical side, it
is unclear why a central
government does not differentiate between districts.

��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
trast,policyuniformityatthefederallevelendogenouslyarisesinthep
resent
model precisely in those circumstances in which centralization
is Pareto
preferred to decentralization. Although previous contributions
to the lit-
erature on fiscal federalism have looked at informational
asymmetries be-
tween a central government and regional governments ðsee, in
particular,
Klibanoff and Morduch 1995; Klibanoff and Poitevin 2013Þ, I
am not
aware of any similar result. To my knowledge, the only other
paper that
explicitly addresses policy uniformity is by Harstad ð2007Þ and
is quite
different in focus. The author considers a model in which two
politically
autonomous regions negotiate an agreement over local public
goods with
spillovers under asymmetric information. Although policy
uniformity is
not an equilibrium phenomenon as in the present framework,
Harstad
shows that a mutual commitment to policy harmonization
ðuniform pol-
iciesÞ may be advantageous in interregional negotiations

because it re-
duces delay in bargaining.
The paper proceeds as follows. Section II introduces the basic
frame-
work, derives the equilibrium conditions, and shows how the
information
transmitted in equilibrium depends on various parameters of the
model.
The theoretical model is taken to the data in Section III. Section
IV gives a
brief discussion of the findings and presents conclusions. All
proofs are
relegated to the Appendix.
II. The Model
A. The Basic Framework
Consider an economy divided into n constituencies or districts
indexed
by i ∈ f1, . . . , ng. Each district is composed of a continuum of
homo-
geneous households with exogenous income y and mass unity.
For ex-
positional purposes, it is convenient to think of these districts
as being
geographically distinct, and I will often refer to them as
regions. How-
ever, it is equally possible to interpret them as broadly defined
con-
stituencies, which are separated along observable demographic
or eco-
nomic ðrather than geographicÞ lines and share a common
objective with
regard to the policy under consideration. There are n 1 1 goods
in the
economy, one private consumption good x and n public projects

gi ∈ f0,
1g, one for each district. If the latter are geographic entities
ðregions,
states, municipalitiesÞ, the assumption is that policies are
targeted to a
particular locality and, as such, have the natural interpretation
of public
projects in infrastructure, recreation, urban renewal, or the
environment.
Otherwise, the projects can be thought of as entitlement
programs tar-
geted to the respective constituency ðfamilies with children,
sayÞ. Public
spending on projects is financed by nondistortionary local
income taxes
ti ∈ R, which may differ across districts.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
Within each district, a household gains a private benefit vi > 0
from
the locally realized public project. In addition, there are
spillovers from
projects in other districts j ≠i, denoted by g . The cost of
realizing proj-
ij
ect gi is ci ∈ ð0, yÞ. All costs and benefits are measured in
terms of

the private consumption commodity. The utility of a household
in re-
gion i from the consumption vector ðg1, . . . , gn, xiÞ is thus
equal to
ui 5 vi gi 1 o
j ≠i
gij gj 1 xi;
and the Pareto-optimal allocation of public projects is
characterized by
g *i ðviÞ 5
1 if vi 1 ji ≥ ci
0 otherwise;
�
where ji ; ojgji measure the aggregate spillover of the public
project
associated with district i. The parameter measuring the private
benefit vi
from their own project gi is private information of the
individuals be-
longing to the respective constituency.9 People outside the
constituency,
including a central authority, know only that the vi’s are
independently
distributed according to some smooth distribution function
FiðviÞ over
the full support Vi 5 ½0; �vi�. All other variables and
parameters of the
economy are common knowledge. To eliminate trivial cases, I
will assume
in the remainder that ji ∈ ½ci 2 �vi; ci� for all projects gi, i 5

1, . . . , n.
Since people within a district are identical, we can represent
each
district by a single agent i 5 f1, . . . , ng who acts on behalf of
all citizens.
This agent’s payoff from a policy vector fðgi; tiÞgi51; : : : ;n is
ui 5 vigi 1 o
j ≠i
gji gj 1 y 2 ti;
which also captures the aggregate surplus generated in district i.
Under
local authority, the decision over project gi lies with the local
district, and
the agent should simply be thought of as an elected policy
maker who
directly determines local policy. In this case, each district is
financially
responsible for its own ðand only its ownÞ project; that is, the
ðlocalÞ bud-
get constraint reads ti 5 cigi. The agent chooses gi to maximize
uLi 5 vi gi 1 o
j ≠i
gji gj 1 y 2 ci gi; ð1Þ
taking the public good supply in all other districts j ≠ i as given.
In equi-
librium,
9 I will assume that vi is “soft” information; i.e., agents cannot
certify or “prove” the value

of vi to others, even if they would like to.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
g Li ðviÞ 5
1 if vi ≥ ci
0 otherwise;
�
i 5 1; : : :; n;
where the superscript L stands for local authority. Because each
district’s
government takes only the benefit received by its own citizens
into ac-
count, the policy decisions are Pareto suboptimal:10 there is
underpro-
vision of public projects if the externality is positive, ji > 0, and
over-
provision if the externality is negative, ji < 0. Moreover, the
size of the
distortion, measured by the expected difference between optimal
and
actual expected surplus,
DLi 5
��Eci2ji
ci

ðvi 1 ji 2 ciÞFiðviÞ
�� > 0;
is increasing in the degree of spillovers jjij.
If the decision over project gi lies with a central ðfederalÞ
authority, the
district representative may be thought of as a regional delegate
to the
federal assembly, an appointed local public official, or a
lobbyist who
advances the constituency’s interests in the central government.
Total
federal expenditures oicigi are still funded by taxes on local
residents ti,
i 5 1, . . . , n, but since funds are often shared at the federal
level, the link
between a district’s tax bill and the implementation of its
project will
generally not be perfect; that is, local tax revenues could be
raised in-
dependently from local project realization. To fix ideas, I will
assume that
there is some arbitrary but exogenous sharing rule ti 5 o
n
j51sijcjgj, where
sij, with o
n
i51sij 5 1, denotes district i’s share in the funding of project gj.
The central authority’s objective is to choose policies ðg1, . . . ,
gnÞ to max-
imize total surplus:

uC 5 o
i
ui 5 o
i
�
vigi 1 jigi 1 y 2 o
n
j51
sijcjgj
�
: ð2Þ
There are two interpretations to this benevolent objective
function.
The first is that the central government is a policy maker whose
constit-
uency consists of the entire economy: while local policy makers
care only
about their own ðregionalÞ district, the center cares about the
welfare of
10 The presumption that politically autonomous jurisdictions do
not coordinate their
policies is standard in the literature and appears natural in many
circumstances. If inter-
jurisdictional contracts are enforceable, however, it is also
conceivable that regions engage

in Coasian bargaining. So far, only a few papers have studied
this possibility of Pareto-
improving contractual arrangements under decentralization. See
Kessler, Lülfesmann, and
Myers ð2009Þ for a model with efficient bargaining and Harstad
ð2007Þ for an analysis of
bargaining under asymmetric information.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
all citizens. There is an alternative and perhaps more
compelling inter-
pretation, however. Suppose that all districts choose one agent
as a del-
egate to send to the national legislature. Since the federal
legislature is
an enduring institution in which delegates interact repeatedly
through
communicating, bargaining, and voting over a fairly long period
of time,
one can expect regional representatives to overcome
inefficiencies caused
by majoritarian decision rules ðminimum winning coalitionsÞ
and nego-
tiate their way to the Pareto frontier. This is what Besley and
Coate ð2003Þ
have called a “cooperative” legislature. In this view, the above

objective
represents a utilitarian bargaining solution that could be
mutually ac-
ceptable under the veil of ignorance; that is, legislators—prior
to receiv-
ing private information—agree to implement ex post a policy
vector that
maximizes the joint surplus.11 As we will see shortly, however,
this ðex anteÞ
efficient objective does not translate into ex post efficiency
since credible
communication is not always feasible once delegates receive
private in-
formation on their local project. Before I analyze this case,
however, it is
instructive to study the case without communication as a
benchmark.
Thus, suppose that the federal government does not know the
re-
alization of the vector of district-specific preference parameters
v 5
ðv1; : : :; vnÞ. Given prior beliefs FiðviÞ, federal policies will
satisfy
ðg1; : : :; gnÞ ∈ arg maxE
�
o
i
�
vigi 1 jigi 2 o
j
sijcjg j

��
or, for all vi ∈ Vi and i 5 1, . . . , n,
g EEi ðviÞ 5 1 ⇔E½vi� 1 ji ≥ ci; ð3Þ
where E½�� is the unconditional expectation operator and the
superscript
EE stands for the ex ante efficient decision, which naturally is
the optimal
policy for an uninformed federal authority that maximizes total
surplus.
From ð3Þ, the lack of information for the central government
yields
uniform policies gi, at least up to observable differences:
consider any
two districts i and j whose projects are ex ante
indistinguishable, that is,
ðci; ji; ViÞ 5 ðcj; jj; VjÞ. Even though local benefits will
generally differ,
11 Some would argue that creating a forum for communication
to debate policies and
negotiate mutually beneficial agreements is one of the main
purposes of a national as-
sembly. Evidence backs this view: in the US House, for
instance, minimum winning coa-
litions are the exception rather than the rule. In the European
Union, the number of
representatives is relatively small, which makes it likely that
they will exploit the benefits of
cooperation. Also, many decisions require unanimity, which

may force legislators to co-
operate. Also note that if the districts are sufficiently
symmetric, implementing a joint
surplus-maximizing objective under the veil of ignorance would
not require side “pay-
ments” ðlogrollingÞ.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
vi ≠ vj, a federal authority without further information must
treat them
equally; that is, either g EEi ðviÞ 5 g EEj ðvjÞ ; 1 or g EEi ðviÞ
5 g EEj ðvjÞ ; 0. More-
over, the fact that federal policy is insensitive toward local
preferences
under incomplete information implies that decentralized policy
making
may be preferable. Indeed, because a local authority is “closer
to the
people,” it is easy to see that it will often be strictly better if
spillovers are
very small, jjij → 0. For very large spillovers, in contrast, the
inefficiency
due to the lack of coordination under local authority dominates
and cen-
tral authority is the desirable mode of governance. Obviously,
these con-

clusions mirror Oates’s ð1972Þ classic arguments in favor of or
against de-
centralization within a framework of locally dispersed private
information,
taking the informational disadvantage of the federal government
as given
ðsee also Oates 1999Þ.
While alluding to the lack of information on the part of the
central
government to explain policy uniformity is appealing at first
glance, it
remains unsatisfactory on second thought. If the federal
government
does not know regional preferences and if this is what prevents
it from
adapting policies that are better suited for the local
constituencies, why
does it not ask local public officials? More generally, what
prevents re-
gions from communicating their preferences to the federal
government?
Indeed, is communication not what a federal assembly with
regional del-
egates is all about? I therefore next turn to the main part of the
analysis,
which endogenizes the lack of information at the federal level
by allow-
ing for communication between local representatives and the
federal
government ðrespectively, a debate among members in the
legislatureÞ.
B. Legislative Communication
Now suppose that authority over spending and taxation rests
with the

federal government but information can flow between the
central gov-
ernment and the districts ðlocal officialsÞ in the sense that the
latter can
communicate their local benefit vi to the former. In doing so,
however,
they have to take into account the fact that the center—upon
having
received the information communicated by the districts and
possibly
updated its prior on v—will implement its most preferred
policy. There
are two interpretations of this communication consistent with
the two
views of a central authority laid out above. First, if the central
authority is
a federal government that is distinct from the local
representatives, one
could imagine the center consulting regional representatives and
offi-
cials on the project before making a decision. Alternatively, the
central
authority may simply be a federal assembly, which itself is
composed of
regional delegates. In this interpretation, the information
transmission
stage can be seen as formalizing delegates communicating with
each
other—a policy debate. The constraint they operate under,
however, is
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��

that, at the end of the day, the assembly will pass a bill that
maximizes
joint surplus.12
Formally, communication and policy selection under the central
au-
thority are described by the following three-stage game. In the
first stage,
nature chooses the vector of local benefits v according to
FiðviÞ, and each
district representative learns the vi ∈ Vi for its local project gi.
The sec-
ond stage is a communication stage in which the central
government
consults with local officials or, alternatively, the delegates in
the federal
assembly engage in a political debate. At a very general level,
this com-
munication can be formalized by the exchange of messages mi.
Upon
observing mi, the central authority forms new beliefs miðmi;
�Þ over vi ∈ Vi.
Given m 5 ðm1; : : :; mnÞ, the central authority then
implements a policy
vector ðg1, . . . , gnÞ that maximizes expected social surplus.
Note that be-
cause the federal government always chooses its most preferred
policy,
conditional on beliefs m, the only thing communication may

achieve is
to change m. Any communication is therefore “cheap talk” and
could in
principle be quite complicated ðthe exchange of messages could
be con-
ditional and repeatedÞ. Under our assumptions that ðaÞ private
infor-
mation is not correlated across districts and ðbÞ as far as a
single district
representative is concerned, the decision on her own local
project is in-
dependent of what happens in other districts, however, it is easy
to see
that the cheap talk game for each district can be analyzed
separately.
Moreover, there is no loss of generality restricting attention to a
single
message mi that is transmitted from the local representative to
the central
authority, who then decides on gi given its updated beliefs
miðmiÞ.
1. Equilibrium Analysis
Cheap talk games with a single sender have been studied
extensively in
the seminal contribution of Crawford and Sobel ð1982Þ, who
consider a
generic version of the game in which a better-informed sender
can send
arbitrary messages to a receiver who eventually makes an
irreversible de-
cision that affects the well-being of both. The authors show that
the
Bayesian Nash equilibria of the game will be characterized by a
partition

12 One might argue that allowing the federal government to
design a mechanism to
elicit the private information of the districts—as opposed to the
communication outlined
above—would be a natural next step. A mechanism ðcompete
contractÞ requires full com-
mitment, however, and by definition, budget authority rests with
the federal government in
a centralized regime. Importantly, this includes the right to
“renege” on promises made,
especially when the result would be a Pareto improvement. By
assumption, the only way the
federal government can commit not to implement its most
preferred policy is to decen-
tralize, i.e., formally place projects under local authority. By
the same token, I also do not
consider the possibility of regions “cooperating” under
decentralization. Consistent with
the central government’s inability to commit, the analysis also
implicitly assumes that a local
policy maker cannot commit to any policy that does not
maximize the respective objective
function. See also the discussion following theorem 1.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��

of the parameter space ðin our case, ViÞ into connected sets
ðintervalsÞ. In
each of these equilibria, the sender optimally and truthfully
announces
to which interval the realized state of nature belongs given the
receiver’s
decision rule, and the receiver maximizes her expected utility,
given her
updated beliefs based on the correct but coarse information
about the
state of nature she received from the sender.
Before I analyze equilibria with meaningful communication,
however,
it is important to note that there always exists an equilibrium in
which no
information is transmitted. Given that the sender’s message is
unrelated
to his private information, the receiver rationally does not
update her be-
liefs and picks the optimal action on the basis of her prior.13
Conversely,
since the receiver “ignores” the message sent, any message is
consistent
with an equilibrium.
Of course, there may be other equilibria with finer partitions in
which
informative communication is feasible. Since there is one
decision made
for every message sent, the number of actions taken in
equilibrium is
limited by the number of elements of the partition
ðcorresponding to in-

tervals of ViÞ. The converse is also true, however: if there are
two mes-
sages that trigger identical actions, then we can combine them
into a
single message without changing the equilibrium outcome.
Since there
are at most two decisions in the present model, we can thus
without loss
of generality assume that the message space is f0, 1g; a strategy
for dis-
trict i is then a mapping mi : Vi → f0; 1g while a strategy for
the federal
government is g Ci : f0; 1g → f0; 1g. Because preferences
satisfy single cross-
ing, the finest partition of Vi thus has two intervals, with mi 5 0
if vi
belongs to one interval and mi 5 1 if vi belongs to the other
interval.
Moreover, any message sent by agents with private information
on the
partition containing higher ðrespectively, lowerÞ values of vi
must trigger
gi 5 1 ðrespectively, gi 5 0Þ; otherwise, it would be optimal for
local rep-
resentatives with extreme values of vi to change their strategy
ðlieÞ.
Lemma 1. Communication is limited under centralization. In
par-
ticular, for each district i, there are at most two types of
Bayesian Nash
equilibria ðup to labeling differencesÞ:
a. Communication is completely uninformative. In this
equilibrium,
the local representative sends a message miðviÞ 5 miðv0iÞ for

all vi,
v
0
i ∈ Vi, and the centralized policy consequently satisfies
g Ci ðmiðviÞÞ 5 g EEi ðviÞ ∀ vi ∈ Vi:
13 For instance, the sender could truthfully announce that the
realized state belongs to
the entire parameter space.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
b. Communication is informative but incomplete. The
correspond-
ing equilibrium is characterized by a cutoff value ~vi ∈
ð0;�viÞ such
that
miðviÞ 5 1 if vi ∈ ½
~vi; �vi�
0 otherwise
�

and
g Ci ðmiðviÞÞ 5 miðviÞ ∀ vi ∈ Vi:
Proof. The proof follows directly from Crawford and Sobel
ð1982Þ and
the above discussion. QED
The equilibrium in part b, which I refer to as the informative
commu-
nication equilibrium below, is illustrated in figure 1.
Since giðmiðviÞÞ 5 miðviÞ, communication in this type of
equilibrium
has a simple structure. The local representatives makes a
“recommen-
dation” as to whether or not their policy project should be
realized, and
the federal government follows the recommendation. In
equilibrium, all
proposals are thus rubber-stamped.
Corollary. Whenever communication between the federal gov-
ernment and a representative of district i 5 1, . . . , n is
informative, the
equilibrium policy is characterized by universalism; that is,
each represen-
tative asking for a project is assured its approval ðregardless of
whether
the project is socially desirable or notÞ.
Lemma 1 shows that whenever the informative communication
equi-

librium exists, it is not unique. So why should we focus on this
equilib-
rium; that is, is it a natural candidate for equilibrium selection?
The an-
swer is given in lemma 2.
Lemma 2. The equilibrium with informative communication ex
ante
Pareto dominates the equilibrium in which no information is
transmitted.
In what follows, I will assume that agents coordinate on the
Pareto-
superior equilibrium, provided that it exists. It remains to
analyze when
this is the case. To this end, consider an informative
communication
equilibrium and assume that the representative of district i
follows his
or her equilibrium strategy as prescribed in lemma 1, that is,
sends a
FIG. 1.—Informative communication
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
message miðviÞ 5 1 for values vi ≥ ~vi and miðviÞ 5 0 for
values vi < ~vi, re-
spectively. For the federal authority to follow its own
equilibrium strategy
of rubber-stamping the district’s proposal given beliefs m, we

therefore
must have
Evi≥~vi fvi 1 ji 2 cig ≥ 0 and Evi<~vi fvi 1 ji 2 cig ≤ 0;
where the first ðsecondÞ inequality ensures that the government
opti-
mally chooses g Ci ðmiÞ 5 1 ðrespectively, g Ci ðmiÞ 5 0Þ after
inferring vi ≥ ~vi
ðrespectively, vi < ~viÞ upon hearing the message mi 5 1
ðrespectively, mi 5
0Þ. Next, consider the representative of the district. Since talk
is “cheap”
ðsending messages is costlessÞ, for the communication strategy
to be op-
timal given the government’s prescribed equilibrium response
of rubber-
stamping, it must be the case that the representative prefers gi 5
1 to
gi 5 0 whenever vi ≥ ~vi and gi 5 0 to gi 5 1 otherwise. Since vi
is distrib-
uted with full support over the interval Vi, a representative with
prefer-
ence parameter ~vi must be indifferent between both outcomes.
Denote
by si 5 sii a district’s cost share in its own local project. From
ð1Þ, when ci
is replaced with sici, whatever the ðexpectedÞ values of gj, j ≠
i, the differ-
ence between a district’s payoff between gi 5 1 and gi 5 0 is vi
2 sici.
Indifference at vi 5 ~vi thus requires
~vi 5 sici: ð4Þ

Hence, we have the following lemma.
Lemma 3. In equilibrium, there is informative communication
be-
tween the central government and district i if and only if
Efvi 1 ji 2 cijvi ≥ sicig ≥ 0 ð5Þ
and
Efvi 1 ji 2 cijvi < sicig ≤ 0: ð6Þ
In summary, we can characterize the information that is
transmitted
through a political debate and the subsequent course of action
under
centralization as follows. First, communication is imperfect in
general,
which implies that centralized decisions are never efficient, as
would be
the case under perfect information. Second, whenever
informative com-
munication is feasible, the central authority in effect rubber-
stamps local
proposals or, put differently, the central legislature operates
with univer-
salistic criteria: every district interested in carrying out a
project is as-
sured approval. However, the federal government still can—and
generally
will—discriminate among districts: in equilibrium, given the
communi-
cated information, the decision rule giðmiÞ depends on the
realization of

��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
the local benefit vi. Third, whether or not meaningful
communication is
in fact feasible depends on the characteristics of the public
project ðji; ciÞ
as well as on the fiscal cost sharing rule s .14 The following
subsection stud-
i
ies these relations in greater detail.
2. How Much Information Flows? Local-Bias Effect
and Common-Pool Effect
Lemma 3 has shown that the credibility of the shared
information and,
by extension, the centralized equilibrium policy decision is
determined
by two exogenous variables: the extent of the spillovers and the
rule gov-
erning cost sharing. To develop some understanding of the
mechanisms
underlying ðcredibleÞ information sharing, consider si 5 1; that
is, the
district “pays the piper” even though the federal government
“calls the

tune.” Informative communication then requires ~v 5 ci; that is,
the dis-
trict proposes to implement its local project under federal
authority
whenever it would have implemented the project itself under
local au-
thority. Were the federal authority always to follow the local
recommen-
dation, the outcome under both governance structures would be
iden-
tical. If spillovers are significant, however, the federal
government will
sometimes optimally “overrule” the district, effectively
undermining any
informative communication.
To see this, suppose first that spillovers are negative. Then, the
federal
government may want to scrap the project even though the local
repre-
sentative is in favor. Because federal approval is gained less
often than is
desired by his constituency, the local representative’s incentives
to truth-
fully communicate vi are diminished. In particular, the desire to
coun-
teract the federal reluctance to realize the project by overstating
local
benefits vi may become sufficiently strong as to render any
meaningful
communication infeasible. Formally, this happens for ðin
absolute termsÞ
large values of ji that do not satisfy condition ð5Þ. Similarly, if
spillovers
are positive, the federal government will want to realize the
project more

often than is desired by the local population. Again, informative
com-
munication will become infeasible because at some point the
incentive
of the local representative to counteract federal activism by
understat-
ing local benefits vi is too pronounced: condition ð6Þ is
violated for suf-
ficiently large ðpositiveÞ values of ji. The federal government
then ratio-
nally ignores the local representative’s information and realizes
the project
against the expressed will of the local population.
Because the motive to overstate ðrespectively, understateÞ the
local
value of the project primarily depends on the discrepancy
between the
14 It is easy to construct examples to assess how pervasive
informed communication ðand,
as a result, universalismÞ is in equilibrium. If vi is uniformly
distributed on ½0; �vi�, e.g., con-
ditions ð5Þ and ð6Þ reduce to ji ≤ cið1 2 12 siÞ ≤ ji 1 12 �vi.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
private value and the social value of the policy, we may call this
effect
the local-bias effect. Also note that because of the simple
structure of the

model, the local-bias effect operates only through a threshold
value of j
i
above ðrespectively, belowÞ which the informative equilibrium
ceases to
exist. It has no influence on the equilibrium policy as long as
informa-
tion is transmitted.
In practice, of course, a project that is decided on at the federal
level
is almost certain to be funded at the federal level as well. In
what follows,
I will therefore focus on the empirically more relevant case in
which
projects that are under federal jurisdiction are federally funded
as well.
Assuming a balanced budget, this de facto means that a
jurisdiction will
bear a share of the project cost equal to its share of the federal
tax base.
For example, irrespective of any ðobservable or unobservableÞ
differ-
ences in the tax bases across districts, cost sharing will
necessarily take
place whenever the federal budget is at least partly financed by
a uni-
form tax instrument such as a federal income tax or a federal
consump-
tion tax. In the simplest case of a central budget financed by a
uniform
tax on identical tax bases, for instance, we would have si 5 1=n.
More

generally, I will make the following assumption.15
Assumption 1. The nature of federal taxation is such that costs
are
de facto shared on the federal level. Moreover, the district tax
shares
decline in the number of districts. For all districts i 5 1, . . . , n,
we have
0 < siðnÞ < 1 and siðnÞ > siðn 1 1Þ ∈ ð0; 1Þ:
Public projects thus have “pork barrel” features: they largely
benefit a
single district at a cost to all others. These types of projects
capture the
important elements of many real-life policies and are commonly
as-
sumed in the literature on legislative politics ðe.g., Weingast et
al. 1981;
Grossman and Helpman 2005Þ. Recall from the previous
sections that
how funds are raised and costs shared under federal authority
had no
impact on federal policies in the benchmark cases of perfect
informa-
tion and imperfect information without communication,
respectively.
Once we allow for communication between the federal
government and
15 To see how the local tax burden ti varies with the decision of
the local project gi, we can
differentiate the federal budget to obtain dti=dgi 5 ci=y ∈ ð0;

1Þ, so siðnÞ ∈ ð0; 1Þ follows
directly from previous assumptions. Naturally, one can also
think of situations in which
siðnÞ is negative or exceeds one. For example, the federal
government may promise a
greater federal share in future local spending if the district
agrees to a project that has high
positive spillovers and low local returns. Similarly, there may
be ðimplicitÞ penalties in-
volved if districts push for projects with large negative
spillovers. But this requires project-
ðdistrict-Þ specific subsidies or penalties that dominate a
district’s share in the federal tax
base that naturally results from uniform federal taxes such as
income, payroll, or con-
sumption taxes. In either case, allowing for si < 0 or si > 1
would not alter the results
significantly.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
local officials ðamong local delegates in a federal assembly,
respectivelyÞ,
this is no longer true: in conjunction with the lack of
information on the
part of the federal government, cost sharing now creates a
common-pool

problem. Project costs are not entirely borne by the local
population,
which leads local representatives to underestimate the cost of
the proj-
ect, ceteris paribus. This in turn creates an incentive to
overstate the lo-
cal benefits vis-à-vis the federal authority, again affecting the
transmis-
sion of credible information.16
How the common-pool effect and the local-bias effect work
together is
summarized in theorem 1.
Theorem 1. Consider any informative communication
equilibrium.
For parameter values ji < ð1 2 siÞci, the federal policy gi is
characterized
by overprovision, and this tendency will grow with the number
of dis-
tricts. For values ji > ð1 2 siÞci, the federal policy gi is
characterized by
underprovision. This tendency is mitigated as the number of
districts
increases to some optimal n*, which minimizes the loss of
information
and grows again thereafter. If ji 5 ð1 2 siÞci, communication is
perfect
and the federal policy decision gi is socially optimal.
The theorem states how informative the communication between
a
central and a local government on the local policy depends on
the ex-

tent of the policy’s spillovers and on how costs are shared at the
federal
level. To understand the result intuitively, suppose first that ji <
ð1 2 siÞci.
The informative communication equilibrium ðassuming it
existsÞ is then
characterized by overprovision since the project is realized
whenever
vi ≥ sici but should be realized only for values vi ≥ ci 2 ji >
sici. Adding
additional districts implies s 0i < si and clearly makes matters
worse: an
already overprovided public project is prone to be even more
overpro-
vided as the cost share of the constituency declines and the
incentive to
overstate its value increases further. Hence, the combination of
imper-
fect communication of privately held information and a
common-pool
problem endogenously generates diseconomies of
ðorganizationalÞ scale:
the more districts there are, the more difficult it becomes to
truthfully
communicate a project’s true benefits in the political process,
and the
more distorted the resulting policy decision will be. The
consequences
are best seen in the limit case in which si → 0. From ð4Þ, if
informative
communication is feasible at all, we must have g ICi ðviÞ 5 1
almost always
under federal authority. Intuitively, since the local district’s
share of the
cost is almost nil, it has a strong incentive to overstate local
benefits in

order to persuade the federal government to realize the project.
But we
know that the latter always approves whenever it listens to the
former,
which is necessarily the case if communication is meaningful.
For values
16 In a more general formulation with variable project size, this
effect holds whenever
locally earmarked expenditures grow more rapidly with project
scale than local taxes.
��140.211.65.60 on Sun, 18 Nov 2018
17:38:07 UTC��
ji > ð1 2 siÞci, on the other hand, the local-bias effect is
sufficiently strong
that the informative communication equilibrium is characterized
by un-
derprovision: the project is realized for values v ≥ s c > c 2 j .
The local-
i i i i i
bias effect and the common-pool effect then work in opposite
directions,
and which effect dominates will depend on the extent of the
local bias
ðjiÞ and on the degree of cost sharing ðthe number of districts
nÞ. In
particular, additional districts now have the benefit of

counteracting the
underprovision problem: more information will flow and the
decision
will be more efficient as we increase the number of districts
from n to
n 1 1 ðignoring the integer problemÞ. Eventually, though, the
common-
pool problem will dominate, assuming si → 0 as n grows
sufficiently large.
In such a situation, there obviously is an optimal organizational
size n*
that balances the common-pool effect with the local-bias effect
and thus
minimizes the loss of information under federal authority.
Theorem 1 also points to a case in which federally chosen
policies are
efficient, namely, if ji 5 ð1 2 siÞci or si 5 1 2 ji=ci. Note that
this value of
the cost share parameter corresponds to what is commonly
known as
the Clarke-Groves mechanism, which induces truth telling in
dominant
strategies. As explained earlier ðsee n. 12Þ, the problem with
setting the
cost shares optimally is that this requires full commitment and
complete
contracting on the part of all parties involved. In contrast, we
adopt the
incomplete contracting approach by assuming that by its very
nature,
the central authority cannot commit to any policy that does not
ex post
maximize its objective ðrecall that the Clarke-Groves
mechanism is not
balanced and, hence, not renegotiation-proofÞ. It is difficult to

envision
the federal government writing binding contracts on policies it
has budget
authority over in the absence of an external enforcement
mechanism.
Also, a government may not be able to fully bind its successor.
An entirely
different, but perhaps equally compelling, argument is that
optimally leg-
islated tax shares will necessarily be both project specific and
district spe-
cific. As local needs differ and potential projects change
frequently ðas
reflected in annual budgetsÞ, the tax code must vary across
districts and
over time, which could give rise to considerable transaction
costs: non-
uniform taxation will result in inefficient reallocation of the tax
base, and
frequent changes create uncertainty hindering investment.
C. Assigning Fiscal Authority
I now briefly address how the local-bias effect and the common-
pool
effect together translate into the overall efficiency of policies
chosen on
a federal level and then compare the outcome under federal
authority
with that under local authority. Note that the comparison is not
trivial: as
we saw earlier, for instance, communication under central
authority im-
proves as the local bias shrinks. Thus we would expect
centralization to

Economics ResourcesTaxes on Savings - GRUBER File.pdf.docx

Economics ResourcesTaxes on Savings - GRUBER File.pdf.docx

Recommended

Recommended

More Related Content

Similar to Economics ResourcesTaxes on Savings - GRUBER File.pdf.docx

Similar to Economics ResourcesTaxes on Savings - GRUBER File.pdf.docx (10)

More from tidwellveronique

More from tidwellveronique (20)

Recently uploaded

Recently uploaded (20)

Economics ResourcesTaxes on Savings - GRUBER File.pdf.docx