Economics of Governance (2007) 8:17–36

DOI 10.1007/s10101-006-0012-1

O R I G I NA L PA P E R

Tax competition between unitary and federal countries

Lisa Grazzini · Alessandro Petretto

Received: 20 October 2003 / Accepted: 1 July 2005 /

Published online: 2 June 2006
© Springer-Verlag 2006

Abstract We analyse taxation of capital in a two-country model, where one country

is unitary while the other one is federal, consisting of two identical regions. Both
national and regional governments levy a tax on capital. The countries play a nonco-
operative game between them, with the government of the federal country acting as
a Stackelberg leader with respect to its regional governments. We show under what
circumstances, at equilibrium, the federal country sets its tax rate inefficiently low,
while the unitary country sets it inefficiently high.

Keywords Tax competition · Mobility of factors of production · Fiscal federalism

JEL Classification H23 · H77 · H87

1 Introduction

Tax competition in federations may give rise to two kinds of tax externalities: hori-
zontal and vertical. Horizontal externality arises when governments at the same level
tax a base, like capital investments, which is mobile across borders. In this case, tax
rates are chosen strategically by governments in order to attract more capital invest-
ments inside the country. Since each government disregards the benefit (harm) it does
others by increasing (reducing) its own tax rate, tax rates tend to be set too low. On
the contrary, vertical externality arises when different levels of government tax the
same base. Since neither level of government takes into account the loss suffered
by the other level of government from the shrinking of the common tax base, tax
rates tend to be set too high, at the noncooperative equilibrium. Accordingly, both
horizontal and vertical externalities distort state taxation, but in opposite directions

We are deeply grateful to three anonymous referees for their comments, and especially to one of
them for extremely helpful suggestions. We also wish to thank participants at ESEM 2004 for fruitful
discussions.

L. Grazzini (B) · A. Petretto

Department of Economics, University of Florence, Via delle Pandette 9, 50127 Florence, Italy
e-mail: lisa.grazzini@unifi.it,alessandro.petretto@unifi.it.
18 Lisa Grazzini, Alessandro Petretto

(Kessler et al. 2002). Thus, when both externalities are considered simultaneously,
one interesting question to raise is whether tax rates tend to be too high or too low,
at equilibrium.
The present approach attempts to answer such a question taking into account not
only the strategic behaviour of countries, but also their institutional structure. In par-
ticular, we are interested in analysing whether too high or too low equilibrium tax
rates may depend on the fact that, when considering a federation of countries, some
of them have a unitary structure while others are themselves of a federal type, i.e. they
are composed by regions whose governments can choose independent fiscal policies.
The European Union, for instance, is a federation of countries with different institu-
tional structures: some of them have a predominantly unitary structure, while others
have a federal structure.
To this end, we propose a model representing a federation in which one country
is unitary while the other has a federal structure, being divided into two identical
regions. Both state and regional governments are benevolent, and tax capital income
according to the source-based principle. The countries play a noncooperative game
between them, with the government of the federal country acting as a Stackelberg
leader with respect to its regional governments (i.e. a sequential game), which in turn
play a noncooperative game between them. Thus, two sources of horizontal exter-
nalities arise: first, between regions inside the federal country, and second, between
countries, at a national level. Further, a vertical externality arises between regional
and national governments inside the federal country. Our main result shows that the
standard finding in the horizontal tax competition literature, according to which two
unitary countries competing for a mobile capital tax base tend to choose tax rates
which are too low, may change when countries are asymmetric in their institutional
structure, i.e. one of the two countries has a federal structure. In this set-up, we show
that, from a social point of view, the federal country may still set an inefficiently low
tax rate, while the unitary country may instead choose an inefficiently high tax rate, at
equilibrium. Roughly speaking, our result is linked to the fact that while the federal
country, as usual, disregards the effects of its fiscal policy on the policy decided by the
unitary country, the latter does not only disregards the reaction of the federal national
fiscal policy to its own, but also fails to take into account the effects of its fiscal policy
on the ones chosen by regional governments in the federal country, and how these
in their turn affect the federal national fiscal policy. Accordingly, overtaxation in the
unitary country occurs as a somewhat surprising result which is due to vertical effects
which appear to be, in some sense paradoxically, more relevant for the level of the
unitary country tax rate than for that of the federal country.
In the economic literature, horizontal and vertical externalities have been stud-
ied by two separate fields of research.1 However, some recent contributions have

1 Horizontal competition has been more extensively studied, see Gordon (1983), Wilson (1986),
Zodrow and Mieszokowski (1986) and Wildasin (1989). For a paper using a political economy ap-
proach, see Persson and Tabellini (1992). More recent papers are those by Baldwin and Krugman
(2004) and Wilson and Wildasin (2004). Clear and comprehensive surveys are Wilson (1999), Well-
isch (2000, Ch. 4) and Cremer and Pestieau (2004). Vertical competition has been more recently
examined. Among those papers assuming benevolent governments see, for instance, Johnson (1988),
Dahlby (1996), Boadway et al. (1998) and Dahlby and Wilson (2003). Cassing and Hillman (1982),
Wrede (2000) and Flowers (1988) use instead a Leviathan approach. Sato (2003) examines a set-up
where fiscal decisions made by governments are conditioned by rent-seeking groups. Keen (1998)
reviews the literature on vertical fiscal externalities, both in the case of a government acting as
benevolent dictator and as malevolent Leviathan.
Tax competition between unitary and federal countries 19

concentrated on the simultaneous working of both externalities, given that both effects
are important in federal frameworks, and their interaction may be crucial for under-
standing whether tax rates tend to be too high or too low at equilibrium. To analyse
this point, Kessler et al. (2002) propose a model with a federal country composed
by identical states, and since horizontal and vertical externalities point in opposite
directions, they study under what circumstances one dominates the other.2 They show
that state taxes are too high or too low at equilibrium, depending on the relative
elasticity of the supply of savings and the demand for capital, and on the extent to
which the states tax rents. The interplay between horizontal and vertical externalities
is also at the heart of a paper by Janeba and Wilson (2004). They show that decen-
tralization, and so more vertical tax competition, may have a counterdistortionary
role offsetting the inefficiencies due to horizontal tax competition, in terms of public
goods underprovision. Such a counterdistortionary role is analysed also by Flochel
and Madies (2002) in a Leviathan setting. They show that when tax competition is
more intense, public subsidies are more efficiently provided at a federal level rather
than at a local level. The interplay between horizontal and vertical externalities is
also analysed by Wrede (1996) who shows that uncoordinated Leviathans will not
generally find themselves on the downward-sloping side of the Laffer curve for total
tax revenue.
This paper addresses an issue which has not, to the best of our knowledge, been
considered in the existing literature. The interaction between horizontal and vertical
externalities is analysed in a set-up where both the strategic behaviour of regional
and national governments and the institutional structure of countries is taken into
account. The literature cited above typically examines both externalities in the frame-
work of a single federal country divided into regions.3 Our model adds to such a
framework a second country which has a unitary structure in order to study not only
the interplay between horizontal and vertical externalities within a federal country,
but also the effects of such an interplay on the strategic interaction between countries
with a different institutional structure. It seems to us that such an issue may become
even more interesting, for instance, with respect to the expected enlargement of the
European Union, which will lead to the coexistence of many countries differing in
their institutional structure within the same federal set-up.
The plan of the paper is as follows. Section 2 describes the model, and section 3
analyses the solution of the game between regions inside the federal country and
between countries. Section 4 examines the question of whether national tax rates are
too high or too low at equilibrium. Section 5 discusses our results, while section 6
contains some concluding remarks.

2 The model

We study a two-period and two-country model. Specifically, consider a world economy

composed of two countries, Home and Abroad, labelled H and A, respectively. Coun-
try A has a unitary system while country H has a federal structure, i.e. it is divided into
two identical regions, m and n . Both state and regional governments have the power
2 The same kind of question is also analysed by Keen and Kotsogiannis (2003), but in a model where
governments behave as revenue-maximising Leviathans.
3 An exception is the paper by Janeba and Wilson (2004) which analyse a set-up with two federal
countries.
20 Lisa Grazzini, Alessandro Petretto

to levy a per unit tax on capital income, which is taxed according to the source based
principle. Capital is assumed perfectly mobile while agents are immobile.4 Further, we
assume that national and regional governments use tax revenue to finance lump-sum
transfers to their citizens.
Events in the model unfold as follows. First, both countries choose their tax rate
playing a Nash game between them, and with the federal country acting as a Stac-
kelberg leader with respect to its regions. Second, both regions in the federal country
choose their tax rate playing a Nash game between them, and acting as Stackelberg
followers with respect to their national government.5 Finally, agents in both countries
make their consumption and investment decisions.6
To simplify the analysis, we suppose that there is a single consumer in each region
of country H, so that there are only two consumers in both countries. In particular,
the two countries are inhabited by individuals with the same preferences and initial
endowment. More precisely, in a region/country i, i = m, n, A, each agent owns a fixed
endowment E of first period income.7 His preferences are described by the following
utility function:
Ui = U(Ci1 ) + Ci2 , i = m, n, A, (1)
where U(·) is a well-behaved utility function, and Ci1 and Ci2 denote consumption in
the first and second period, respectively. In the first period, each agent decides how
much to save and where to save. His budget constraint is given by
E = Ci1 + km n A
i + ki + ki , i = m, n, A, (2)
where km
i , kni , kA
denote individual savings, with the subscript referring to the
i
region/country i, i = m, n, A, where an agent lives, and the upperscript referring
to the region/country i, i = m, n, A, where he chooses to save. Accordingly, individual
savings are given by Si ≡ E − Ci1 = km i + ki + ki , i = m, n, A. In the second period,
n A

each agent receives principal and interest on his savings, plus a lump-sum transfer from
both the regional and the national government in the case of an agent living in country
H, only from the national government in the case of an agent living in country A. Thus,
the second period budget constraint for an agent i living in country H is given by
Ci2 = [1 + (rm − tm − T H )]km n n H n
i + [1 + (r − t − T )]ki
+[1 + (rA − T A )]kA i H
i +g +G , i = m, n, (3)

4 Some recent contributions analyse tax competition problems and fiscal federalism issues by con-
sidering both capital and labour mobility, and also other specific aspects of economic integration, such
as migration. See, for instance, Wellisch (2000, Ch. 4 and 5), Kessler et al. (2002) and Bretschger and
Hettich (2002).
5 See also Goodspeed (2002) and Kessler et al. (2002) for models in which the national government
acts as a Stackelberg leader, while local governments act as followers with respect to the central
government, but play a Nash game between them. Kessler et al. (2002) also consider the case when
each local government plays Nash relative to all other governments, local and central. This case is
analysed in Keen (1998) too.
6 To describe the timing of the game we follow Persson and Tabellini (1992), as this timing ensures
that no credibility problems arise with respect to the decisions made by the private sector when capital
income taxation is chosen. In particular, the timing we have supposed can be enforced by designing
political institutions in such a way that policy decisions are made by elected legislators. However, the
analysis of a previous stage of the game when elections take place is beyond the aim of this paper.
See the paper by Persson and Tabellini for such an analysis.
7 The subscript i is also used to denote an agent living in region/country i, i = m, n, A.
Tax competition between unitary and federal countries 21

while the second period budget constraint for an agent i living in country A is given
by
2
CA = [1 + (rm − tm − T H )]km n n H n
A + [1 + (r − t − T )]kA
+[1 + (rA − T A )]kA A
A+G , (4)
where ri , i = m, n, A, denotes the gross remuneration to savings in region m, n, and
country A; tm , tn , T H , T A denote the tax rate in region m, n, country H and A, respec-
tively; gi , i = m, n, denotes the regional per capita lump-sum transfer while GH and
GA denote the national per capita lump-sum transfer in country H and A, respectively.
In both countries, the same consumption good is produced by using the same tech-
nology, which uses capital as the sole input. Specifically, in each country, the production
function is defined as
f (Ki ), i = m, n, A, (5)
where Ki = kim + kin + 2kiA , i = m, n, A, denotes total savings in each region m and
n, and in country A, 8 and f is increasing, strictly concave, and at least three times
continuously differentiable. Furthermore, we suppose that the market is perfectly
competitive, and accordingly firms’ profit maximising behaviour implies the following
familiar condition on marginal factor productivity:9

f (Ki ) = ri , i = m, n, A. (6)
This condition can be used to obtain the demand for capital:
Ki = K(ri ), i = m, n, A. (7)

Rents arising in region/state i, ≡ i −f f (Ki ) (Ki )Ki ,
i = m, n, A, are assumed to
be fully taxed at regional level, in case of country H, and at national level, in case
of country A. 10 Further, as in Persson and Tabellini (1992), we assume that, in each
region/country, savings are taxed in order to finance a lump-sum transfer to its resi-
dents. 11 This simplification allows us to concentrate our attention only on the effects
of different countries’ institutional features on tax policies, removing the effects on
citizens’ welfare through public good provision by national and regional governments.
12 This would instead require to take into account the different marginal valuation

that consumers attach to national and regional public goods. Our set-up can thus be

8 For example, total savings in region m, K m , is equal to km + km + 2km , i.e. the sum of savings in
m n A
region m by agents living in regions m, n, and the unitary country A.
9 Derivatives are denoted by a prime for functions of one argument.
10 See Kessler et al. (2002) for an analogous assumption.
11 Notice also that we assume that rent taxation is not sufficient to entirely finance lump-sum trans-
fers. Further, notice that nothing would change if rents were not taxed in both countries, or rents
were taxed at a national level also in country H. In any case, rents are a component of second period
income for consumers, either directly because they are earned by consumers, i.e. they would appear
on the RHS of (3), or indirectly via a regional lump-sum transfer, i.e. they would appear on the RHS
of (3) via (8), or via a national lump-sum transfer, i.e. they would appear on the RHS of (3) via (9).
See Kotsogiannis and Makris (2002) for a discussion on the role of the allocation of rents between
different levels of governments, with respect to production efficiency.
12 Taxes are used only to pursue strategic purposes, such as shifting of tax burden and terms of
trade effects, and not to supply public goods. This set-up corresponds, in intertemporal terms, to the
standard textbook excess burden exercise where tax revenue is given back to agents. However, notice
that our results could change if tax revenue were used to finance public goods provision.
22 Lisa Grazzini, Alessandro Petretto

interpreted as equivalent to the case where national and regional public goods are
perfect substitutes, and further the marginal valuation of public spending is equal to
unity. 13
Accordingly, the regional budget constraints obtain as
ti Ki (ri ) + i (ri ) = gi , i = m, n (8)
while the national budget constraints obtain as
T H (Km (rm ) + Kn (rn )) = 2GH , (9)
in country H, and
T A KA (rA ) + A (rA ) = 2GA , (10)
in country A.
In each region m and n of country H (country A), consumer solves his optimization
problem by maximising his utility function (1) subject to the first and second period
budget constraint (2) and ( 3), ((2) and (4)). It is easy to check that the first order
conditions imply that
MRSCi1 ,Ci2 = 1 + ρ, i = m, n, A, (11)
where ρ denotes the net return to savings, which is different from the cost of capital
for firms because of the presence of savings taxation. Further, the assumption of per-
fect mobility of capital implies that arbitrage by capital investors insures that, in each
region/country, an identical net return ρ will prevail:
ρ = rm − τ m = rn − τ n = rA − T A , (12)
where τ i = ti + T H , i = m, n, is the consolidated tax rate in country H. In other words,
with perfect mobility of capital, investors of both countries share a common capital
market, and this implies the equalization on the net returns to savings across coun-
tries. Thus, by (11), the individual first and second period demand functions obtain
as Ci1 (ρ) and Ci2 (ρ), i = m, n, A, respectively, while supply function obtains as Si (ρ),
i = m, n, A. The common rate of return ρ is determined on the international capital
market in the following way. Assuming full employment of capital permits to obtain
the market clearing condition
Km (ρ + τ m ) + Kn (ρ + τ n ) + KA (ρ + T A ) = (ρ), (13)
where (ρ) ≡ Sm + Sn + 2SA is total savings, i.e. total supply of capital, with   (·) 0.
The net return ρ is thus the solution to the above equation, which means that it is a
function of tm , tn , T H , T A , i.e. ρ = ρ(tm ,tn ,T H ,T A ). Differentiating (13) with respect
to ti , i = m, n, and ρ yields

∂ρ Ki
=    , i = m, n, (14)
∂ti   − Km + Kn + KA
which implies
∂ρ
−1< < 0, i = m, n. (15)
∂ti
13 Keen and Kotsogiannis (2003) use as a benchmark case a context where federal and state spend-
ing are perfect substitutes, so that consumers are indifferent between public expenditures at the two
levels. See their paper for a discussion of this point.
Tax competition between unitary and federal countries 23

Similarly, by differentiating (13) with respect to T H and ρ yields

 
∂ρ Km + Kn  ∂ρ
=     = , i = m, n, (16)
∂T H 
 − K +K +K
m n A ∂ti
i=m,n

which implies
∂ρ
−1< < 0, (17)
∂T H
and by differentiating (13) with respect to T A and ρ yields

∂ρ KA
=    , (18)
∂T A   − Km + Kn + KA
which implies
∂ρ
−1< < 0. (19)
∂T A
Accordingly, in region i, i = m, n, we also obtain that
∂ri ∂ri ∂ri ∂ri
> 0, < 0, > 0, < 0, i, j = m, n, i  = j. (20)
∂ti ∂tj ∂T H ∂T A
Equivalent expressions obtain for country A. Further, we also obtain that the fol-
lowing inequalities hold for the demand for capital in region i, i = m, n (equivalent
expressions obtain for country A):
∂Ki ∂Ki ∂Ki ∂Ki
< 0, > 0, < 0, > 0, i, j = m, n, i  = j. (21)
∂ti ∂tj ∂T H ∂T A

3 The solution of the game

In this section, we solve the game described above by backward induction. In the pre-
vious section, we had already solved the third stage of the game, when consumption
and investment decisions are made. We now turn to the solution of the second and
first stage of the game.

3.1 Nash game between regions

In stage two of the game, each region i, i = m, n, composing country H has to decide
the tax rate to levy on each unit of capital located in its jurisdiction. In the noncoop-
erative game between these regions, each of the two governments behaves as a Nash
player, i.e. takes as given the tax rate in the other region. Furthermore, each gov-
ernment acts as a Stackelberg follower with respect to the national government, and
thus it also takes as given the national tax rate, T H . By the solution to the consumer
maximization problem in stage 1 and by the regional government budget constraint
(8), the indirect utility function of agent i obtains as
Vi = U(E − Si (ρ)) + (1 + ρ) Si (ρ) + i (ρ + τ i ) + ti Ki (ρ + τ i )
TH  i
+ K (ρ + τ i ), i = m, n, (22)
2 i=m,n
24 Lisa Grazzini, Alessandro Petretto

where the reader must keep in mind that ρ depends on tm , tn , T H , T A , i.e. ρ =

ρ(tm , tn , T H ,
T A ). For expositional convenience, consider the case of region m. Noticing that
i = −Ki , the first order condition with respect to tm , after some manipulations,14
obtains as follows
    
∂ρ ∂ρ m T H  ∂Ki
m m m ∂K
Sm (ρ) − K 1+ m + K +t + = 0. (23)
∂tm ∂t ∂tm 2 i=m,n ∂tm

Each of the terms in this expression can be given a simple interpretation. The first
term describes the negative impact on the net remuneration to individual savings fol-
lowing an infinitesimal rise in tm . However, in equilibrium all agents bear a decrease
in the net return to their savings. Accordingly, this term can be interpreted in terms
of a horizontal externality: the government of region m does not take into account the
harm borne by citizens of the other region and of country A when the international
net remuneration to savings decreases in response to an increase in tm . The second
term is negative by (15), and it describes the reduction in rent tax revenue deriving
from an increase in the cost of capital which follows an increase in the regional tax
rate. Also this term can be interpreted in terms of a horizontal externality. Indeed, the
government of region m does not take into account the benefits accruing to citizens of
the other region and country A when capital employed within their borders increases,
and also rent tax revenue increases, in response to an increase in tm . The third term
represents the sum of the direct and the indirect effect on regional tax revenue of the
tax increase. As usual, the direct effect is positive while the indirect effect is negative,
in the case of a positive tax. More precisely, the latter describes the positive horizontal
externality, in terms of capital flight, which benefits region n and country A when
region m increases its tax rate. Since region m, when it increases its tax rate, does not
take into account such an externality, it perceives this indirect effect in a negative way,
i.e. as a deadweight loss, that creates a disincentive to redistribution.15 Finally, the
fourth term represents the impact on country H tax revenue deriving from a change
in the national tax base due to an infinitesimal rise in tm , i.e. a vertical externality from
the regional to the national level (a tax base effect). More precisely, an increase in tm
leads to a decrease in savings in region m while it leads to an increase in savings in
region n.16

14 An application of the envelope theorem is behind (23) and the following first order conditions
(27) and (33).
15 This point can also be described by rearranging the terms in (14), namely

∂Km ∂ρ  

= m   − Kn − KA < 0,
∂t m ∂t

where it is shown that the reduction in Km due to an increase in tm equals the rise in savings in region
n and country A.
16 By differentiating (13) with respect to tm , and by using ( 14), it is easy to check that

⎡ ⎤
 
∂Kn 
⎣   − Km − KA
=K m
− 1⎦
∂tm  
  − Km + Kn + KA


which is positive by (21).

Tax competition between unitary and federal countries 25

Condition (23) defines region m’s reaction function:

tm = tm tn , T H , T A . (24)

An expression similar to (23) holds for region n so that region n ’s reaction function
obtains as

tn = tn tm , T H , T A . (25)

A Nash equilibrium of the game played by the two regions is given by the solution to
the system of the two above reaction functions. In what follows, we concentrate our
attention on the case of a symmetric equilibrium, i.e. ti = t, i = m, n.

3.2 Nash game between countries

We now turn to a description of the first stage of the game, in which the governments of
the unitary and the federal country choose their fiscal policy. More precisely, national
governments play a noncooperative game between them: each country chooses its tax
rate, given the tax rate chosen in the other country, and further, the federal country
acts as a Stackelberg leader with respect to the fiscal decisions of its regions.
Let us firstly consider country H. The national government chooses T H in order to
maximise a social welfare function which is given by the sum of the two consumers’
indirect utility function (22) in regions m and n, subject to the public budget constraint
(9). Accordingly, its objective function obtains as
   
WH = 2 U(E − S(ρ)) + (1 + ρ)S(ρ) + i (ρ + τ ) + τ Ki (ρ + τ ), (26)
i=m,n i=m,n
 
where τ = t + T H , and further the reader must keep in mind that ti = ti tj , T H , T A ,
i, j = m, n, i  = j, and thus ρ = ρ(tm (·) , tn (·) , T H , T A ). Accordingly, the first order
condition with respect to T H , after some manipulations, is given by
  ∂ti
  dKi   
∂ti

i i
2S(ρ)α + 1+ K + τ − K α + 1 + = 0,
i=m,n
∂T H i=m,n
dT H i=m,n ∂T H
(27)
where
∂ρ  ∂ρ ∂ti
α≡ + , (28)
∂T H
i=m,n
∂ti ∂T H

and
  
dKi i ∂ti
= K α + 1 + . (29)
dT H ∂T H
Firstly notice that, at this stage of the game, α describes the effect of an infinitesi-
mal increase of T H on the net remuneration to savings, according to the previously
specified function, ρ = ρ(tm (·), tn (·) , T H , T A ). In particular, an infinitesimal change
in T H affects the net remuneration ρ in two ways: directly, i.e. ∂ρ/∂T H , and indirectly
via the change in the regional tax rates, i.e. (∂ρ/∂ti )(∂ti /∂T H ), i = m, n. Secondly,
the expression for α can be simplified by noticing that ∂ρ/∂tm = ∂ρ/∂tn = ∂ρ/∂t and
∂tm /∂T H = ∂tn /∂T H = ∂t/∂T H , i = m, n, at a symmetric equilibrium of the game
26 Lisa Grazzini, Alessandro Petretto

played by the two regions in the second stage. More precisely, (28) can be rewritten
as follows
 
∂ρ ∂t
α≡2 1+ , (30)
∂t ∂T H
where we have used the fact that ∂ρ/∂T H = 2(∂ρ/∂t) by (16), at a symmetric equilib-
rium of the second stage. Since ∂ρ/∂t < 0 by (15), the sign of α depends on the sign
taken by ∂t/∂T H , which can be both positive and negative. 17 However, we can show
that, under plausible assumptions, 1 + ∂t/∂T H > 0.18 Thus, by (30), α < 0, which
means that we are considering the case when an infinitesimal change in T H negatively
affects the net remuneration to savings.
We are now in a position to interpret the terms in the first order condition of country
H maximization problem. To this aim, equation (27) can be rewritten as follows
    dKi    
∂t i ∂t
2S(ρ)α + 1 + K + τ − α + 1 + Ki = 0.
∂T H i=m,n i=m,n
dT H ∂T H
i=m,n
(31)
The first term of (31) represents the negative impact of an infinitesimal increase in
T H on the net remuneration to individual savings. As we already pointed out when
we discussed equation (23), this term can be interpreted in terms of a horizontal
externality. The second term is given by the sum of the direct effects on national
and regional tax revenue of the domestic national tax increase. The direct effect on
national tax revenue is always positive, while the direct effect on regional tax revenue
is positive (negative) when country H and regions i, i = m, n, tax rates are strategic
complements (substitutes). This term describes the increase (reduction) in regional
tax revenue due to an increase (decrease) in the regional tax rate in response to an
increase in the domestic national tax rate, i.e. a vertical externality – a revenue effect
(Goodspeed 2000). However, notice that the sum of the two direct effects is always
positive since 1 + (∂t/∂T H ) > 0. The third term is given by the sum of the indirect
effects on national and regional tax revenue of the domestic national tax increase:
both effects are negative since dKi /dT H < 0 (in the case of T H > 0 and t > 0). 19 This
means that an increase in T H , and accordingly a decrease in the net remuneration to
savings, at the margin, leads to a decrease in savings in regions m and n, and so to a
decrease both in national and regional tax revenue. Thus, the term representing the

17 Strategic substitutability or complementarity between federal and state taxes has been investi-
gated both theoretically and empirically. From a theoretical perspective, Kessler et al. (2002), Keen
and Kotsogiannis (2003) point out under what circumstances the two taxes are strategic substitutes or
complements. From an empirical perspective, results are mixed. For instance, by using a sample of 13
OECD countries for the period 1975–1984, Goodspeed (2000) shows that higher national income tax
rates and lower poverty rates lead to lower local income tax rate. On the contrary, Besley and Rosen
(1998) find a positive reaction of states to a change in US federal gas and cigarette taxes.
18 The proof follows the same type of arguments as Keen and Kotsogiannis (2003, 2004), and it is

available upon request to the authors. Notice that 1 + ∂tH > 0 means that country H and regions i,
∂T
i = m, n, tax rates are either strategic complements, i.e. ∂t/∂T H > 0, or are moderately strategic sub-
stitutes, i.e. −1 < ∂t/∂T H < 0. By remembering that ∂τ /∂T H = 1 + ∂t/∂T H , Keen and Kotsogiannis
point out that the latter case implies that an increase in country H tax rate may lead to a decrease
in the equilibrium regional tax rate, but the amount of such a decrease cannot be so large since the
consolidated tax rate increases.
19 It can easily be checked by using (16) and (30) into (29).
Tax competition between unitary and federal countries 27

effects on national tax revenue describes the positive horizontal externality, in terms of
capital flight, which benefits country A when country H increases its tax rate. The term
representing the effects on regional tax revenue describes instead a negative vertical
externality, i.e. a tax base effect. Finally, the fourth term describes the reduction in rent
tax revenue deriving from an increase in the cost of capital which follows an increase
in the national tax rate. As we already pointed out, also this term can be interpreted in
terms of a horizontal externality. Before analysing the case of country A, notice that,
in our set-up, the national government of country H fully internalizes both vertical
externalities described above since it cares about the residents in both country H’s
regions, and it sees through the regional budget constraints when it chooses its own
fiscal policy (Kessler et al. 2002, Janeba and Wilson 2004, Wilson 1999).
Similarly, the government of country A chooses T A in order to maximise a social
welfare function which is given by the sum of the two consumer’s indirect utility
functions subject to the public budget constraint (10). Thus, its objective function is
 
WA = 2 U(E − SA (ρ)) + (1 + ρ)SA (ρ) + T A KA (ρ + T A ) + A (ρ + T A ), (32)

where the reader must keep in mind that for country A’s government the international
net remuneration to savings depends on tm , tn , T H , T A , i.e. ρ = ρ(tm , tn , T H , T A ). By
maximising (32) with respect to T A , after some manipulations, we obtain the following
first order condition
 
∂ρ A A ∂K
A
A ∂ρ
2 A SA (ρ) + K + T −K + 1 = 0, (33)
∂T ∂T A ∂T A
where
 
∂KA ∂ρ
= KA + 1 ,
∂T A ∂T A
which is negative by (19). As usual, the terms in the above expression can be given
a simple interpretation. The first term in (33) describes the negative impact of an
infinitesimal increase in T A on the net remuneration to individual savings and, as we
already pointed out, gives rise to a horizontal externality. The second term represents
the positive direct effect on national tax revenue of the national tax increase, while
the third term describes the negative indirect effect on national tax revenue of the
national tax increase. As usual, the latter effect represents a horizontal externality: an
unilateral increase in the tax rate leads to a capital flight towards country H. Finally,
the fourth term describes the reduction in rent tax revenue deriving from an increase
in the cost of capital which follows an increase in the national tax rate. As we already
pointed out, also this term can be interpreted in terms of a horizontal externality.
By solving equation (27), we obtain country H’s reaction function, namely

T H = T H tm tn , T H , T A , tn tm , T H , T A , T A . (34)

Similarly, by solving equation (33), we obtain country A’s reaction function, namely

T A = T A tm , tn , T H . (35)

A Nash equilibrium of the game played by the two countries is given by the solution
to the system of the two above reaction functions.
28 Lisa Grazzini, Alessandro Petretto

4 Equilibrium analysis: overtaxation or undertaxation?

In this section, we raise the question whether the two countries H and A choose too
high or too low tax rates, at equilibrium. In particular, we are interested in checking
whether the answer to the previous question is affected by the fact that country H has
a federal structure while country A is a unitary state.
To investigate this point, consider country H and country A social welfare func-
tions which obtain at the equilibrium of the first stage of the game described in the
previous section. More precisely, country H social welfare function obtains as in (26)
and country A social welfare function as in (32), but where the reader must keep in
mind that the net remuneration to savings, at equilibrium, is now given by

ρ = ρ tm tn , T H , T A , tn tm , T H , T A , T H tm tn , T H , T A ,

tn tm , T H , T A , T A , T A tm , tn , T H , (36)

where we have used (34) and (35). Accordingly, national tax rate is inefficiently  too
∂Wi 
high (low), at equilibrium, i.e. there is overtaxation (undertaxation), when ∂T i  <
Nash
(>)0, i = H, A.20

Let us now consider country H, and focus our attention on the most interesting
and empirically relevant set-up when national tax rates are strategic complements, i.e.
∂T A /∂T H > 0. By substituting (36) into (26), we obtain the social welfare function
for country H, at equilibrium, which can be differentiated with respect to T H , giving
rise to
  dKi
∂WH  dρ dρ  i
 = 2S(ρ) H + τ − K, (37)
∂T H
Nash dT dT H
i=m,n
dT H i=m,n

where
dρ ∂ρ ∂t ∂ρ ∂ρ ∂T A
≡ 2 + 2 + (38)
dT H ∂t ∂T H ∂T H ∂T A ∂T H
and
  
dKi dρ ∂t
= Ki + 1 + .
dT H dT H ∂T H
Since we evaluate (37) at the Nash equilibrium, we use (27 ) into (37), which
accordingly obtains as
    

∂WH  dρ i i
= −α 2S(ρ) + τ K − K = βψ, (39)
∂T H Nash dT H i=m,n i=m,n

where
dρ ∂ρ ∂ρ ∂T A
β≡ − α = + , (40)
dT H ∂T H ∂T A ∂T H

20 Overtaxation and undertaxation are intended only in the sense that a national tax increase would
rise or decrease welfare when regional taxes were to remain unchanged. See also Kessler et al. (2002).
Tax competition between unitary and federal countries 29

by using (30) and (38), and

 
ψ ≡ 2S(ρ) + τ Ki − Ki . (41)
i=m,n i=m,n

We are now in a position to state the following proposition.21

Proposition 1 In country H, there is undertaxation if national tax rates are strategic

complements.

Proof See the Appendix. 



In order to gain more insights from our results, let us rewrite equation ( 39) as
follows
  
∂WH   
i i
= βτ K + β 2S(ρ) − K . (42)
∂T H  Nash i=m,n i=m,n

Each of the terms in this expression can now be given a simple interpretation. In
the previous section, we have already analysed the first term that now we define the
capital flight effect. The second term is named the terms of trade effect and shows how
a country tax affects the international net remuneration to savings. More specifically,
a capital importing (exporting) country pays (receives) a lower net remuneration for
(from) its capital import (export) following an increase of the tax rate in any of the
two countries. This means that a capital importing country has an incentive at the
margin to increase its tax rate for improving its terms of trade, and conversely for a
capital exporting country. The simultaneous interplay of these two types of horizontal
externality is at the basis of the explanation of our results.
Since β < 0 when national tax rates are strategic complements (see the proof of
Proposition 1 in the Appendix), in equation (42), the capital flight effect is positive,
and
 the terms of trade effect is positive when country H imports capital (2S(ρ) −
K i < 0), while it is negative when country H exports capital (2S(ρ)− i
i=m,n i=m,n K
> 0 ), with the sum of these two terms being positive. Accordingly, when country H
chooses its tax rate, it perceives the capital flight as larger than it occurs since it does
not take into account the effects of its choice of T H on the tax rate chosen in country
A. For example, country H disregards the fact that an increase in T H also leads to an
increase in T A , and thus the capital flight is not so high as it is perceived by country
H, when choosing its optimal fiscal policy. Thus, the capital flight effect would lead
country H to set an inefficiently too low tax rate.
This tendency is even reinforced when country H imports capital via the terms of
trade effect. In this case, country H would like to pay a lower net remuneration for its
capital import, and thus it has an incentive at the margin to increase T H . However,
country H disregards the fact that an increase in T H also leads to an increase in T A ,
with a negative impact on the net remuneration to savings. Accordingly, the terms of
trade effect would lead country H to choose an inefficiently too low tax rate. When
country H imports capital, both the capital flight and the terms of trade effect are
positive, and so country H undertaxes. The same kind of reasoning, mutatis mutandis,
applies when country H is instead a capital exporting country.

21 It can also be proved that strategic substitutability between national tax rates may lead to under-
taxation or overtaxation depending on the value taken by ∂T A /∂T H .
30 Lisa Grazzini, Alessandro Petretto

We now turn to the analysis of country A. By substituting (36) into (32), we obtain
country A social welfare function, at equilibrium, which can be differentiated with
respect to T A , giving rise to

∂WA  dρ A dK
A dρ
 = 2 S(ρ) + T − KA A , (43)
∂T Nash
A dT A dT A dT
where
 
dρ ∂ρ ∂t ∂ρ ∂T H ∂t ∂T H ∂ρ
≡2 + 2 + + (44)
dT A ∂t ∂T A ∂T H ∂t ∂T A ∂T A ∂T A

and
 
dKA A dρ
= K 1 + .
dT A dT A
Since we evaluate (43) at the Nash equilibrium, we use (33 ) into (43), which accord-
ingly obtains as
  
∂WA  dρ ∂ρ A A A

= − 2S(ρ) + T K − K = δγ (45)
∂T A  Nash dT A ∂T A
where
 
dρ ∂ρ ∂ρ ∂t ∂ρ ∂T H ∂t ∂T H
δ≡ − =2 + 2 + (46)
dT A ∂T A ∂t ∂T A ∂T H ∂t ∂T A ∂T A

and
γ ≡ 2S(ρ) + T A KA − KA . (47)
Expression (46) for δ is fairly complicated since it takes into account how an infini-
tesimal change in T A affects the net remuneration to savings (36) not only directly,
but also via a series of indirect effects. Indeed, a change in T A also affects country H
regional and national tax rates, whose variations in their turn affect the net remuner-
ation to savings. Thus, to be able to sign δ, we need to make assumptions not only on
the strategic relationships between national and regional tax rates inside the federal
country, but also between the national tax rate of the unitary country and the regional
tax rate of the federal country. Furthermore, before we proceed to present our result
on country A, we propose the following definition aimed at characterizing a particular
case that can take place within our framework.

Definition 1 National tax rates are moderately strategic complements if ∂T H /∂T A ∈

(0, π), where π ≡ −2(∂T H /∂t)(∂t/∂T A ).

We are now in a position to state the following proposition.22

Proposition 2 In country A, when ∂T H /∂t > 0 and ∂t/∂T A < 0, there is overtaxation
if national tax rates are moderately strategic complements.

22 It can also be proved that overtaxation arises if national tax rates are strategic substitutes while
undertaxation arises if national tax rates are strategic complements, but ∂t/∂T A > 0.
Tax competition between unitary and federal countries 31

Proof See the Appendix. 



The results presented in Proposition 2 can be explained by rewriting equation (45)

as follows


∂WA 
 = δT A KA + δ 2S(ρ) − KA . (48)
∂T A
Nash

As usual, the first term describes the capital flight effect, and the second term rep-
resents the terms of trade effect. Under the assumptions that there is strategic com-
plementarity between national and regional tax rates, in the federal country, i.e.
∂T H /∂t > 0, and there is instead a strategic substitutability between regional tax rates
of the federal country and national tax rate of the unitary country, i.e. ∂t/∂T A < 0,
the proof of Proposition 2 (see the Appendix) shows that δ > 0 if national tax rates
are moderately strategic complements.23 Accordingly, in equation (48), the capital
flight effect is negative, and the terms of trade effect is negative when country A
imports capital (2S(ρ) − KA < 0), while it is positive when country A exports capital
(2S(ρ) − KA > 0), with the sum of these two terms being negative.
Since national tax rates are supposed moderately strategic complements, when
country A chooses its tax rate, it perceives the capital flight as smaller than it is
since it does not take into account the effects of its choice of T A on the tax rates
chosen in country H both at regional and national level. For example, country A
disregards the fact that an increase in T A , on the one hand, leads to an increase in T H
(∂T H /∂T A > 0), but on the other hand, it also leads to a reduction in t (∂t/∂T A < 0),
which in its turn leads to a reduction in T H (∂T H /∂t > 0). Accordingly, the tax wedge
between the two countries (T A vs. t + T H ) may be higher than that perceived by
country A when choosing its tax rate, and thus also the capital flight may be higher
than that perceived. Thus, the capital flight effect would lead country A to set an
inefficiently high tax rate.
This tendency is reinforced when country A imports capital (both terms in (48) are
negative), while it is partly reduced when country A exports capital (the first term
in (48) is negative, and the second is positive, but their sum is negative). Take for
example the case when country A imports capital. Then, country A would like to pay
a lower net remuneration for its capital import, and thus it has an incentive at the
margin to increase T A . However, country A does not perceive that an increase in
T A , on the one hand,
 leads to an increase in H
 T , with a negative impact on the net
remuneration ρ (∂ρ/∂T )(∂T /∂T ) < 0 , but on the other hand, also leads to a
H H A

reduction not only in t, but also in T H via t, with a positive impact on the net remuner-
ation ρ (2(∂ρ/∂t)(∂t/∂T A ) > 0, and 2(∂ρ/∂T H )(∂T H /∂t)(∂t/∂T A ) > 0, respectively)
. In other words, an increase in T A causes not only a negative indirect effect on the
net remuneration ρ, but also positive indirect effects, which offset the former (δ > 0).
Since such indirect effects are not taken into account, the terms of trade effect is
incorrectly anticipated, and when country A imports capital, it would lead such a
country to choose an inefficiently high tax rate. The same kind of reasoning, mutatis
mutandis, applies when country H exports capital.

23 We refer to the relation between regional tax rates of the federal country and national tax rate
of the unitary country as a “mixed vertical and horizontal” relation since it takes into account both
the horizontal competition between countries and a sort of vertical competition between regional
governments of country H and national government of country A.
32 Lisa Grazzini, Alessandro Petretto

Finally, before discussing our results, it is useful to consider as a benchmark case

the framework when both countries are unitary. Indeed, this set-up corresponds to
the standard one analysed in the horizontal tax competition literature. In this respect,
we can state the following24

Corollary 1 When countries H and A are unitary, there is undertaxation in both coun-
tries if national tax rates are strategic complements.

Proof See the Appendix. 



In order to analyse the results in Corollary 1, let us focus again on equations (42)
and (48), which are now identical since in a symmetric equilibrium, identical countries
choose the same tax rate. Countries are neither importers nor exporters of capital, and
no terms of trade effect occur: countries cannot improve their income from abroad by
manipulating the terms of trade. Further, the proof of Corollary 1 (see the Appendix)
shows that β = δ < 0, which implies that the capital flight is positive. Indeed, both
countries perceive the capital flight as larger than it is since they disregards the fact
that the other country will react to their own tax rate by moving its own tax rate in the
same direction, thus reducing the actual capital flight with respect to that perceived.
Accordingly, both countries undertax.

5 Discussion of results

When national tax rates are strategic complements, and both countries are unitary
(and large enough to make the net return on capital endogenous), Corollary 1 shows
that both of them undertax, i.e. they choose inefficiently low tax rates at equilibrium.
This result is consistent with common intuition, and it describes the set-up usually
examined in the horizontal tax competition literature. Intuitively, a fiscal externality
occurs: countries fear capital flight abroad when taxes are too high, thus reducing
the domestic tax base, and hence tax revenue, but conversely, augmenting foreign tax
revenues. However, since countries are symmetric, they choose the same tax rate in
equilibrium, the conjectured capital flight does not occur, but both taxes are set at
inefficiently low levels.
When countries are asymmetric in their institutional structure, i.e. one of the
two has a federal structure, we provide sufficient conditions under which the pre-
vious result may not hold any more. Indeed, we show that the federal country still
undertaxes (Proposition 1), while the unitary country may now choose to overtax
(Proposition 2). This happens when strategic complementarity between national tax
rates is sufficiently low, and there is also a strategic complementarity between na-
tional and regional tax rates, in the federal country, and a strategic substitutability
between regional tax rates of the federal country and the national tax rate of the
unitary country. Since countries are now asymmetric, the capital flight effect de-
scribed above is not the only externality at work: another type of externality – a
pecuniary one – takes place.25 Countries can either import or export capital, and if
they are large enough, they can manipulate their terms of trade, i.e. they have an
incentive at the margin to decrease or increase the international net remuneration at

24 Overtaxation would instead arise in both countries if national tax rates were strategic substitutes.
25 See DePater and Myers (1994).
Tax competition between unitary and federal countries 33

which capital is borrowed or lent. Accordingly, depending on its sign, the terms of
trade effect may reinforce or partly offset the capital flight effect, that is negative for
the unitary country (see (48)), and positive for the federal country (see (42)).
Thus, both countries choose inefficient levels of taxation since they do not correctly
perceive the capital flight and the terms of trade effect, as they neglect to take into
account all the reactions to their fiscal policy. On the one hand, the federal country,
as usual, disregards the effects of its fiscal policy on the policy decided by the uni-
tary country: the capital flight effect is overestimated and the terms of trade effect
is underestimated (overestimated) when country H imports (exports) capital, so that
the interplay of these two externalities leads to an equilibrium tax rate which is chosen
at an inefficiently low level. On the other hand, the unitary country does not only dis-
regard the reaction of the federal national fiscal policy to its own, but also fails to take
into account the effects of its fiscal policy on the ones chosen by regional governments
in the federal country, and how these in their turn affect the federal national fiscal
policy. Accordingly, the capital flight effect is underestimated while the terms of trade
effect is overestimated (underestimated) when country A imports (exports) capital,
so that the interplay of these two externalities leads to an equilibrium tax rate which
is chosen at an inefficiently high level.
In this respect, pushing our model a bit further would enable us to apply our discus-
sion to the case of the expected enlargement of the European Union. For instance, our
set-up would apply when a unitary country entering the EU chooses its optimal tax
rate without taking into account the effects of its choice not only on the national fiscal
policy of a federal country like Germany, but also on the fiscal policy of the länders,
and on the way the latter affects German national fiscal policy. On the other hand,
Germany would neglect the effects of its national tax rates on the entering country’s
fiscal policy. Due to these behaviours, both countries would not correctly anticipate
the capital flight and the terms of trade effects. More specifically, if the horizontal
relationship between national fiscal policies were of a (sufficiently low) strategic com-
plementarity type, as the vertical relationship inside Germany between regional and
national tax policies, while the “mixed horizontal and vertical” relationship between
the entering country and the länders’ fiscal policies were of a strategic substitutability
type, our model would predict overtaxation in the entering country and undertaxation
in Germany. Accordingly, overtaxation in the unitary country occurs as a somewhat
surprising result which is due to vertical effects which appear to be, in some sense
paradoxically, more relevant for the level of the unitary country tax rate than for that
of the federal country.

6 Concluding remarks

In this paper, we have analysed strategic interaction between two countries of a

federation where one country is unitary while the other is itself of a federal type, i.e.
composed by two identical regions. In our set-up, horizontal tax competition arises not
only between countries but also between regions within the federal country, together
with a vertical competition between national and regional governments.
The question we raise is whether the two countries choose inefficiently high or low
national tax rates, at equilibrium, depending on the fact that one country has a federal
structure while the other one is instead a unitary state. In this respect, our main result
is obtained when national tax rates present a sufficiently low strategic complementar-
34 Lisa Grazzini, Alessandro Petretto

ity. In this case, we show that the standard finding according to which two countries
competing on acquiring a mobile capital tax base tend both to undertax capital may
not hold any more when one of the two countries is of a federal type. Sufficient con-
ditions are provided under which the federal country continues to undertax capital
income, while the unitary country overtaxes it.
Finally, notice that our analysis has been cast into a particular model of strategic
interaction between countries. Firstly, the countries are different as regards their insti-
tutional structure, but they are identical in all other respects. Thus, problems related to
different population size, preferences for equity, and social composition are not taken
into account. Secondly, taxes are simply used to finance uniform lump-sum transfers
to individuals, and more complicated income redistributive policies or public goods
provision are left aside. We feel, however, that such issues are open fields for further
research.

Appendix

Proposition 1 In country H, there is undertaxation if national tax rates are strategic

complements.

Proof Rewrite (27) as follows

    
  ∂t
i i
2S(ρ) + τ K − K α = −τ 1 + Ki . (49)
i=m,n i=m,n
∂T H i=m,n

By using (41), equation (49) obtains as

  
∂t
ψα = − 1 + τ Ki > 0, (50)
∂T H i=m,n
 
since 1 + ∂t/∂T H > 0, and Ki < 0. Since α < 0, (50) implies that ψ < 0. Thus, from
H
(39), ∂W
∂T H
 ≷ 0 ⇔ β ≶ 0. By using (17) and (19) into (40), it is easy to show that
Nash
β < 0 if national tax rates are strategic complements, i.e. ∂T A /∂T H > 0. 


Proposition 2 In country A, when ∂T H /∂t > 0 and ∂t/∂T A < 0 , there is overtaxation
if national tax rates are moderately strategic complements.

Proof Rewrite equation (33) as follows

∂ρ
2S(ρ) + T A KA − KA = −T A KA > 0. (51)
∂T A
By using (47), the LHS of (51) obtains as

∂ρ ∂ρ
2S(ρ) + T A KA − KA =γ > 0,
∂T A ∂T A

∂WA 
which implies that γ < 0 since we use (19). Accordingly, from (45), 
∂T A Nash
≷0⇔
δ≶ 0. By using (15) and (17) into (46), when ∂T H /∂t > 0, it is easy to show that δ > 0
if ∂t/∂T A < 0 and ∂T H /∂T A ∈ (0, π). 

Tax competition between unitary and federal countries 35

Corollary 1 When countries H and A are unitary, there is undertaxation in both coun-
tries if national tax rates are strategic complements.
Proof We only provide a sketch of the proof, which can be easily derived by using
the same reasonings underpinning the proof of Propositions 1 and  2. In particu-
H
lar, by using ti = 0, i = m, n, it is easy to check firstly that ∂W 
∂T H Nash 
= βψ, with
A
β ≡ (∂ρ/∂T A )(∂T A /∂T H ), and ψ < 0 as in (41), and secondly that ∂W∂T A
 = δγ ,
Nash
with δ ≡ (∂ρ/∂T H )(∂T H /∂T A ), and γ < 0 as in (47). Since the two countries are
symmetric, it follows that T H = T A , at equilibrium, and thus ψ = γ and β = δ. 


References

Baldwin RE, Krugman P (2004) Agglomeration, integration and tax harmonization. Eur Econ Rev
48, 1–23
Besley TJ, Rosen HS (1998) Vertical externalities in tax setting: evidence from gasoline and cigarettes.
J Public Econ 70, 383–398
Boadway R, Marchand M, Vigneault M (1998) The consequences of overlapping tax bases for redis-
tribution and public spending in a federation. J Public Econ 68, 453–478
Bretschger L, Hettich F (2002) Globalisation, capital mobility and tax competition: theory and evi-
dence for OECD countries. Eur J Polit Econ 18, 695–716
Cassing JH, Hillman AL (1982) State-federal resource tax rivalry: the Queensland railway and the
federal export tax. Econ Rec 235–241
Cremer H, Pestieau P 2004 Factor mobility and redistribution. In: Henderson, J.V., Thisse, J.F. (eds.)
Handbook of regional and urban economics, vol 4. Amsterdam: North Holland
Dahlby B (1996) Fiscal externalities and the design of intergovernmental grants. Int Tax Public Finan
3, 397–412
Dahlby B, Wilson LS (2003) Vertical fiscal externalities in a federation. J Public Econ 87, 917–930
DePater JA, Myers GM (1994) Strategic capital tax competition: a pecuniary externality and a
corrective device. J Urban Econ 36, 66–78
Flochel L, Madies T (2002) Interjurisdictional tax competition in a federal system of overlapping
revenue maximizing governments. Int Tax Public Finan 9, 121–141
Flowers MR (1988) Shared tax sources in a Leviathan model of federalism. Public Finan Q 16, 67–77
Goodspeed TJ (2000) Tax structure in a federation. J Public Econ. 75, 493–506
Goodspeed TJ (2002) Tax competition and tax structure in open federal economies: evidence from
OECD countries with implication for the European Union. Eur Econ Rev 46, 357–374
Gordon H (1983) An optimal taxation approach to Fiscal Federalism. Q J Econ 100, 1–27
Janeba E, Wilson JD (2004) Decentralization and international tax competition. J Public Econ (in
press)
Johnson WR (1988) Income redistribution in a federal system. Am Econ Rev 78, 570–573
Keen MJ (1998) Vertical tax externalities in the theory of fiscal federalism. Int Monet Fund Staff Pap
45(3), 454–485
Keen MJ, Kotsogiannis C (2002) Does federalism lead to excessively high taxes? Am Econ Rev 92(1)
363–370
Keen MJ, Kotsogiannis C (2003) Leviathan and capital tax competition in federations. J Public Econ
Theory 5(2), 177–199
Keen MJ, Kotsogiannis C (2004) Tax competition in federations and the welfare concequences of
decentralization. J Urban Econ 56, 397–407
Kessler AS, Lülfesmann C, Myers G (2002) Redistribution, fiscal competition and the politics of
economic integration. Rev Econ Stud 69, 899–923
Kotsogiannis C, Makris M (2002) On production efficiency in federal systems. Econ Lett 76, 281–287
Persson T, Tabellini G (1992) The politics of 1992: fiscal policy and European integration. Rev Econ
Stud 59, 689–701
Sato M (2003) Tax competition, rent-seeking and fiscal decentralization. Eur Econ Rev 47, 19–40
Wellisch D (2000) Public finance in a federal state. Cambridge: Cambridge University Press,
Wildasin DE (1989) Interjurisdictional capital mobility: fiscal externality and a corrective subsidy.
J Urban Econ 25(2), 193–212
36 Lisa Grazzini, Alessandro Petretto

Wilson JD (1986) A theory of interregional tax competition. J Urban Econ 19(3), 296–315
Wilson JD (1999) Theories of tax competition. Natl Tax J 2, 269–304
Wilson JD, Wildasin DE (2004) Capital tax competition: bane or bone? J Public Econ 88, 1065–1091
Wrede M (1996) Vertical and horizontal tax competition: will uncoordinated Leviathans end up on
the wrong side of the Laffer curve? Finanzarchiv 53, 461–479
Wrede M (2000) Shared tax sources and public expenditures. Int Tax Public Finan 7, 163–175
Zodrow G, Mieszokowski P (1986) Pigou, Tiebout, property taxation and the underprovision of local
public goods. J Urban Econ 19(3), 356–370