Received 23 July 2016; accepted 22 August 2016; published 25 August 2016

1. Introduction
Tax revenue sharing seeks to distribute revenue for central and subnational governments from a given tax base. Some of the advantages of tax revenue sharing recognized in the literature include: 1) Tax harmonization between the central and subnational governments to reduce inefficiency costs from tax competition; 2) Rao [1] argues that tax revenue sharing leads to welfare gains by ensuring greater subnational fiscal autonomy (by ensuring subnational governments a source of revenue), and 3) Revenue sharing is a tool to deal with vertical fiscal imbalances.
Tax revenue sharing is commonly implemented in many developed and developing countries through the use of formulas for the allocation of tax revenue between the central and subnational governments. Rao [1] classifies revenue-sharing systems in two types: in the first type, the central and sub-national governments share a tax base and different levels of governments can determine tax rates. In the second type, one level of government could collect tax revenue that is shared among different levels of governments1. In spite of the empirical relevance of this policy, there is little formal research on the optimal design of formulas for subnational revenue sharing2. Such analysis is worthwhile because knowing the properties of different forms of subnational revenue sharing systems would help us to identify potential advantages and shortcomings of feasible policy options. In this paper we seek to contribute to fill this gap in the literature and we analyze the optimal design of subnational tax revenue sharing. To do so, we study a model with a benevolent social planner in which subnational governments provide local public goods with inter-regional spillovers. The central government selects the nationwide tax rate on income to determine the optimal size of tax revenue and its shares to be allocated to subnational governments.
The structure of the paper is as follows: in Section 2, we develop optimal formulas for tax revenue sharing for an economy with local public goods that are imperfect inter-regional substitutes. Section 3 contains the analysis for an economy with perfect substitutes. Section 4 concludes.
2. Optimal Tax Revenue Sharing for Local Public Goods that Are Imperfect Inter-Regional Substitutes
. To see that our preference relation for
and
characterizes local public goods that are imperfect inter-re- gional substitutes, note that the elasticity of substitution between
and
is positive but finite since
where
. In the case of perfect inter-regio-
nal substitutes, the elasticity of substitution is infinite (see Section 3).
We consider a benevolent social planner ruling the central government that seeks to maximize the nation’s so- cial welfare
where
are the populations of districts
and
. The problem of policy design for the central government is to maximize
by choosing the size of the pro- portional income tax
the budget B, and the formula for sharing revenue in districts
and
,
to finance local public goods
. The budget constraint of the central government is
where
is the central government’s budget. The distribution of tax revenue in the economy is determined by formulas for revenue sharing such that the budget constraints of subnational governments are given by
and
where
is the share of the budget allocated to finance the local public good of district
. For simplicity of the analysis, districts
and
only supply, respectively,
and
. In this section, we analyze the case of local public goods that are imperfect substitutes, which are defined by a finite elasticity of substitution, while for the case of local public goods that are perfect substitutes (analyzed in Section 3) the elasticity of substitution is infinite6. Formally, the problem of tax revenue sharing for the central government when local public goods are imperfect substitutes is:
(1)
(2)
where
is the indirect utility function of the representative individual of district
on feasible local public goods (a similar expression is given for
).
Proposition 1. The optimal budget
and formulas for distribution of tax revenue
for local public goods that are imperfect inter-regional substitutes of districts
and
are given by:
(3)
With
and
(4)
(5)
Proof
The first order conditions for the government’s problem are:
(6)
(7)
Rearrange terms to obtain conditions (3), (4) and (5).
Proposition 2. The optimal budget
and formulas for distribution of tax revenue
satisfy the following properties:
2.1) The implied distribution of local public goods,
is Pareto efficient.
2.2)
is a constant:
.
2.3) The share of tax revenue allocated in district
, also satisfies the following:
a)
as
and ![]()
b) ![]()
c) ![]()
d)
, that is
is non decreasing with
, and ![]()
e)
, that is
is non decreasing with
, and ![]()
Proof
See the Appendix.
An issue of interest for the design of a policy of tax revenue sharing is the relationship of
with the dis-
trict’s population. Our analysis suggests that
as
. This is relevant for policy design
since empirical evidence suggests the use of linear formulas between
and
(see [10] ). However, our analysis suggests that the effect of an increase in the population of district
over the share of tax revenue to be
allocated to district
is contingent to the level of
. For sufficiently low values of
(for
)
an increase in
should lead to an increase in
. However, for sufficiently high levels of
(for
) an increase in
should lead to a fall in
. The explanation of this result is straightforward: from the optimality condition
where
is the net social marginal benefit for residents of district
of increasing
where ![]()
is the gross social marginal benefit for residents of district
of increasing
(an increase in
leads to a higher level of the local public good supplied by district
which increases the wellbeing of residents of that district). Moreover, a higher
means that
falls which leads to a lower supply of
and the ben-
efits for residents of district
from consuming
fall by
. Hence the net benefit of a marginal in- crease in
is
. If
then
hence an increase in ![]()
leads to a net increase in the social marginal benefits of residents of district
which leads to a higher share of
tax revenue allocated to this district (the opposite occurs if
).
3. Optimal Tax Revenue Sharing for Local Public Goods that Are Perfect Inter-Regional Substitutes
![]()
In this section, we analyze the case for formulas for subnational tax revenue sharing when local public goods are perfect inter-regional substitutes which means that the elasticity of substitution between
and
is infinite. To distinguish our notation from our previous section, we denote
as the equilibrium level of the budget and
as the allocation formula of tax revenue for district
. For this case,
is the indirect utility for a resident in dis-
trict
where
are positive constants and
and
7. For this economy, the problem of policy design is8:
(8)
(9)
Proposition 3. The optimal budget
and formulas for distribution of tax revenue
for local public goods that are perfect inter-regional substitutes in districts
and
are given by:
(10)
Recall
and
(11)
And
(12)
Proof
The first order conditions for the government’s problem are:
(13)
(14)
Rearrange terms to obtain conditions (10), (11) and (12).
Proposition 4. The optimal budget
and formulas for distribution of subnational tax revenue
satisfy the following properties:
4.1) The implied distribution of local public goods,
is Pareto efficient.
4.2)
is a constant: ![]()
4.3) The share of tax revenue allocated in district
, satisfies:
a)
and ![]()
b)
and
.
c)
and ![]()
d) ![]()
e)
if
with ![]()
f)
if
with ![]()
Proof
See the Appendix.
Proposition 4 shows some comparative static results on the determinants of formulas for revenue sharing for local public goods that are perfect inter-regional substitutes. In particular,
is non decreasing with the population of the district, however, this relationship is not linear since
. In addition, the distribution of tax revenue for district
is increasing on the spillovers of the public good provided by district
, and decreasing in the spillovers of the local public good supplied by district
, (see propositions 4.3.b and 4.3c). These last results are very intuitive: a higher level of
increases the nationwide social marginal benefits of
(by the effect of the spillovers of
over district
) which leads to an increase in the tax revenue allocated to district
. A similar effect occurs with an increase in
which makes more attractive to allocate more resources in district
and reduces
. The distribution of income should not be considered in the design of formulas for lpg’s that are perfect substitutes since
(see condition 4.3c). As in the previous section, the marginal social benefits and costs of a change in
do not depend on
and
(see condition 11). Proposition 4 also shows that
has an ambiguous relationship with the distribution of
4. Concluding Remarks
In this paper, we study the optimal design of formulas for subnational revenue sharing when local governments provide public goods with spillovers in two cases of interest: when local public goods are perfect and imperfect inter-regional substitutes. Even though, tax revenue sharing policy is implemented in many developed and developing countries there is little formal research on the design of formulas for subnational revenue sharing. In this paper we seek to contribute to fill this gap. Our main contribution is to distinguish the determinants for optimal formulas for subnational revenue sharing when local governments provide local public goods with spillovers that are perfect and imperfect inter-regional substitutes. This analysis has the potential to provide relevant information for policy makers on the determinants and some properties of formulas for subnational tax revenue sharing.
Although the paper provides insights for the design of revenue sharing systems, it does not consider political institutions (such as electoral competition, the interaction between the executive and legislative bodies), the role of special interest groups and other issues raised by political economy models that might be central in shaping incentives of policy makers. Analysis on revenue sharing systems should also incorporate the role of mobility of households and firms, and adopt a systemic view of the optimal composition of the tax structure and spending of subnational governments and the central government. Future research on this topic should address these issues.
Acknowledgements
Raul Alberto Ponce Rodriguez thanks Consejo Nacional de Ciencia y Tecnología (CONACyT) for the financial support of the program “estancias sabáticas en el extranjero 2015” that helped to develop this research.
Appendix
Proposition 2. The optimal budget
and formulas for distribution of tax revenue
satisfy the following properties:
2.1) The implied distribution of local public goods,
is Pareto efficient.
2.2)
is a constant:
.
2.3) The share of tax revenue allocated in district
, also satisfies the following:
a)
as
and ![]()
b) ![]()
c) ![]()
d)
, that is
is non decreasing with
, and ![]()
e)
, that is
is non decreasing with
, and ![]()
Proof
2.1) Efficiency: local public goods
are Pareto efficient. Since
,
.
2.2)
is a constant:
. This follows trivially by the fact
are constants. Mo-
reover,
and ![]()
2.3)
satisfies the following:
a)
as
and ![]()
From condition (4),
. Since
, the term
. Moreover from
, it is simple to see that
.
b) ![]()
From condition (4),
.
c)
.
From condition (4),
.
d)
is non decreasing on
and
.
From (4),
and
since
.
e)
is non decreasing on
and
.
From (4),
and
since
.
Proposition 4. The optimal budget
and formulas for distribution of subnational tax revenue
satisfy the following properties:
4.1) The implied distribution of local public goods,
is Pareto efficient.
4.2)
is a constant: ![]()
4.3) The share of tax revenue allocated in district
, satisfies:
a)
and
.
b)
and
.
c)
and
.
d)
.
e)
if
with
.
f)
if
with
.
Proof
4.1) Efficiency: local public goods
and
are Pareto efficient since
and
feasible
.
4.2)
is a constant:
. Since
are constants. Moreover,
and ![]()
4.3) The share of tax revenue for district
,
, also satisfies the following:
a)
and
.
From (11),
since
![]()
.
It also holds that
.
b)
and ![]()
Note that
and
![]()
![]()
.
c)
and
.
Define
From proposition 4.3b
and
.
d)
.
From condition (11), ![]()
e)
if
with ![]()
It is satisfied that ![]()
.
Moreover, it is simple to show
.
f)
if
with
.
From (11),
as
and
.
![]()
Submit or recommend next manuscript to SCIRP and we will provide best service for you:
Accepting pre-submission inquiries through Email, Facebook, LinkedIn, Twitter, etc.
A wide selection of journals (inclusive of 9 subjects, more than 200 journals)
Providing 24-hour high-quality service
User-friendly online submission system
Fair and swift peer-review system
Efficient typesetting and proofreading procedure
Display of the result of downloads and visits, as well as the number of cited articles
Maximum dissemination of your research work
Submit your manuscript at: http://papersubmission.scirp.org/
NOTES
![]()
*Corresponding author.
![]()
1It is common that, under this type of revenue sharing, the central government is responsible for collecting tax revenue but subnational governments are in charge of the spending choices and (or) the administration of public programs.
2There is a large literature on the optimal design and effects of intergovernmental transfers, see (among many others) [2] - [5] , and more recently [6] . There has been also interest on revenue assignments, see [7] . However, to the best of our knowledge, there has been little interest on developing a formal theory of the optimal design of formulas for subnational revenue sharing.
3Spillovers of local public goods lead to some degree of inter-regional substitutability among lpg’s produced in different regions. Moreover, in real life, the mobility of households and firms makes lpg’s more substitutable. Hence, it is relevant to conduct an analysis of the distinction between lpg’s that are perfect inter-regional substitutes (which is the limiting case) and imperfect substitutes in the design of formulas for subnational tax revenue sharing. However, for simplicity of the analysis, in this paper we ignore inter-regional mobility. We leave such analysis for future research.
4In Section 2, we provide a formal distinction between local public goods that are perfect and imperfect inter-regional substitutes. A real life example of a perfect substitute-local public good with inter-regional spillovers could be productive local infrastructure (such as local bridges, roads etc.). To see this, consider a firm (similar examples can be found for the case of households) that sells products in all districts (regions). If a bridge in some district has a constant marginal rate of technical substitution with respect a bridge provided in some other district then these local public goods show spillovers and are considered as perfect inter-regional substitutes. For the case of imperfect substitutes, the marginal rate of technical substitution is not constant.
5Our choice of the utility function is for simplicity of the analysis. Moreover, this type of utility function is common in the literature see [8] and [9] , among many others.
6To distinguish local public goods that are perfect vs imperfect inter-regional substitutes, we use the elasticity of substitution
where
. Use the household’s budget constraint and the government’s budget con-
and ![]()