European Economic Review 44 (2000) 1633}1657

Tax evasion, "scal competition and economic

integration
Helmuth Cremer *, Firouz Gahvari

IDEI and GREMAQ, University of Toulouse and Institut Universitaire de France, Place Anatole France,
F-31048 Toulouse Cedex, France

Department of Economics, University of Illinois at Urbana-Champaign, Champaign, IL 61820, USA

Received 1 December 1996; accepted 1 February 1999

Abstract

This paper examines the implications of tax evasion for "scal competition and tax
harmonization policies in an economic union. First, for symmetric countries, it proves
that the equilibrium values of the tax and audit rates are less than optimal. Tax
harmonization alone will also lead to a less than optimal audit rate. Second, for
asymmetric countries, the paper shows that integration may turn an honest country into
an evading one. It argues that tax harmonization alone may be a bad policy in that it can
make both countries worse o!. It may also cause an otherwise honest country to turn to
evasion.  2000 Elsevier Science B.V. All rights reserved.

JEL classixcation: H20; H26; H87; F15

Keywords: Tax evasion; Economic integration; Fiscal competition; Tax harmonization

1. Introduction

An important policy aspect of the recent trends towards international economic

integration, both in Europe and North America, is its "scal implications.

* Corresponding author.

0014-2921/00/$ - see front matter  2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 1 4 - 2 9 2 1 ( 9 9 ) 0 0 0 2 5 - 2
1634 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

A central concern has been that tax competition would lead to a lowering of tax
revenues and suboptimal levels of expenditures on public goods and social
policies. The solution to this problem has then been sought in some form of tax
coordination among countries. The purpose of this paper is to take a fresh look
at the tax competition problem, and the e!ectiveness of "scal coordination
policies, in the presence of tax evasion.
Economic integration enhances mobility of goods, capital and people. This
increases the potential for movements of tax bases among nations. Conse-
quently, member states may attempt to expand their individual tax base by
engaging in tax competition. It is this process that is often expected to lead to
less than optimal tax rates and public expenditure levels. It is also naturally
believed that one can prevent such ine$ciencies by following certain tax har-
monization policies. The results mirror earlier such results in the "scal federal-
ism literature. It has long been recognized in that literature that mobility of tax
base between lower level jurisdictions, creates a potential for an e$ciency loss
due to non-cooperative tax setting, and which can be corrected through policy
coordination.
The literature has thus far ignored the problem of tax evasion. This is
a serious omission as tax evasion has important implications for tax competition
and "scal coordination. In the presence of tax evasion, tax revenues are not
determined solely by the di!erent countries' &legislated' tax rates but also by
their enforcement policies. This observation implies that the audit strategy
becomes an added policy tool for tax competition. To avert this ine$ciency, the
two countries must thus harmonize their tax rates and audit strategies. This is an
important policy implication which has gone unnoticed.
Whether or not full coordination would take place is of course an open
question. However, it is important to bear in mind that coordinating in audit
strategies may prove more problematic than harmonizing tax policies. It is true
that full coordination would generally bene"t all parties. Nevertheless such
a policy may never take hold. Any attempt at harmonizing audit strategies
would face a particular di$culty. Legislated tax rates are publicly observable. As
such, adherence to agreed upon tax policies can easily be checked and veri"ed
by all countries concerned. This is clearly not the case for audit strategies. The

 Mintz and Tulkens (1986) have pointed out that tax competition may also result in &excessive'
tax rates. However, it is the less-than-optimal-tax-rates result which has received the greatest share
of attention. See, e.g., Sinn (1994) and Edwards and Keen (1996).
 The recent literature on tax competition and tax harmonization also include Keen (1989), Sinn
(1990) and Kanbur and Keen (1993). On the earlier literature on "scal federalism, see, among others,
Stigler (1957), Oates (1972), Zodrow and Mieszkowski (1986), Wilson (1986) and Wildasin (1989).
 In a recent paper, Xiwen and Wilson (1996) study the implications of tax evasion for the
structure of optimal taxes on capital incomes from home and abroad.
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1635

government of one country can hardly observe the enforcement e!orts of the
other countries. Most signi"cantly, it will be too di$cult, if not impossible, to
verify such e!orts. The number of persons audited or the amount of money
spent on auditing do not necessarily translate into e!orts. This is more a matter
of the organization of the tax administration (i.e. the structure of the auditors'
incentives and the like). Such provisions are too complex to be spelled out in
detail or written into contracts between countries. Consequently, "scal competi-
tion in audit strategies may remain e!ective even in the face of tax harmoniz-
ation policies.
The experiences of a number of federalist states in Europe support our
contention. There are instances of enforcement policies being used as instru-
ments for "scal competition, particularly when competition in tax rates are
banned. Lenk et al. (1998) argue that the richer German states (LaK nder) tend to
be more lax in their tax enforcement policies. This is to compensate for (federally
mandated) tax rates they consider too high. They document a complaint "led by
federal authorities (Bundesrechnungshof) against the (rich) state of Hessen on
the grounds that its auditing e!orts have declined. Similarly, in Belgium, the
French speaking media have repeatedly accused the tax enforcement agencies in
the (richer) Flemish region of tolerating (and even encouraging) tax evasion.
In order to bring out the complexities that the presence of tax evasion may
lead to, we introduce a particular type of evasion (indirect tax evasion) into the
simplest model of economic integration. The model consists of two countries,
a taxed private good and a public good. The tax is collected from "rms and used
to "nance the provision of the public good. National governments maximize
their residents' welfare. The countries may be of two types: &honest' or &evading'.
The two types are alike in all aspects but one. In an honest-type country there is
no tax evasion; in an evading-type there is.
Within this framework, we "rst determine the optimal government policy in
both country types when their economies are closed. We show that the evading-
type country will have a lower level of public goods. This arises because tax
evasion increases the marginal cost of public funds in the economy. Next, we
study the properties of the non-cooperative equilibria when the borders open.
We examine this "rst when the two countries are of similar types. We show that,
with two honest countries, economic integration results in less than optimal tax
rate and public good provision in both. The ine$ciency can be averted if the two
countries adopt a speci"ed tax rate. When the two countries are evading, eco-
nomic integration results in less than optimal values for their tax and audit rates.
Moreover, while coordination on audit strategies may prove di$cult, the poten-
tial for "scal competition will not be deterred by a policy of tax harmonization

 See Edwards and Keen (1996) who, in the context of a model of economic integration, synthesize
this view with the opposing one of governments as revenue maximizers.
1636 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

alone. Given a speci"ed tax rate, the countries will engage in "scal competition
by cutting their audit probabilities. In e!ect, they will follow a policy which
implicitly encourages tax evasion.
We also examine the rami"cations of integrating the economies of two
countries that are of di!erent types. The outcome in this case may be radically
di!erent from that for similar countries. Whereas integration makes two honest
countries necessarily worse o!, this will not be the case if only one of the countries
is honest. With di!erent types, it is in fact possible for the honest country to
become better o! with integration (as compared to the closed border case)
despite "scal competition. The paper derives a su$cient condition for this case.
An interesting feature of our analysis is that the characteristics of the two
countries are endogenous. That is, evasion in a country is a policy decision
linked to its audit technology. With the opening up of the markets, the calculus
of costs and bene"ts associated with pursuing an honest versus an evading
policy changes. This gives rise to the following important question. Is it possible
for a country to be honest when its borders are closed but become evading as
a result of integration? Interestingly, the answer to the question is in the
a$rmative. This possibility points out to a certain rami"cation of integration
that has hitherto gone unnoticed. However, we will also argue that when an
initially honest country becomes better o! as a result of integration, it will
remain honest.
Finally, we will show that simple tax harmonization policies, which ignore the
potential for tax competition through audit strategies, may hurt both countries.
They may also bene"t the evading country and cause an otherwise honest
country to turn to evasion.

2. The model

Assume two countries, 1 and 2, lie on the interval [!1, 1] with a border
between them at the origin. The countries are of equal size; each with a popula-
tion which is uniformly distributed over the space it occupies. The population
size is normalized at one. There is one private good and one public good in each
country. Consumers have quasi-linear preferences. They will purchase the pri-
vate good if its cost is less than its reservation price; otherwise they will consume
none. The reservation price is the same in both countries and su$ciently high to
ensure that each person will purchase the good. Within each country, consumers
are able to buy the good at their place of residence.

 This is Kanbur and Keen's (1993) setting with two di!erences. We introduce a public good into
their model. We also replace their assumption regarding governments' objective. Instead of being
Leviathans, governments are assumed to maximize the welfare of their citizens.
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1637

Production technologies are linear and identical in both countries. Firms

behave competitively producing the private good at constant marginal and
average costs. There is a tax on consumption of the private good. The tax, levied
at t per unit, is collected from the "rms. Tax revenues are used to "nance the
provision of the public good, y. Each country will provide the public good to its
own residents only. The objective of a country's government is to maximize the
welfare of its residents.
Prior to economic integration, there is no trade between the two countries.
Upon integration, citizens of each country will be able to purchase the private
good from the foreign as well as the home country. To do so, they must travel to
the border (but not any further) and purchase the good from foreign "rms
located there. These consumers face no tari! from the home country; nor will
they get a tax rebate from the foreign country. Traveling to any destination and
back entails a cost of d per unit of distance. There is no residential mobility
within or across countries.
The countries may be of two types: &h' (honest) or &e' (evading). The two types
are alike in all aspects but one. In country h there is no tax evasion, but in
country e there is. We will discuss the question of what makes a country honest
or evading later in Section 6. Initially, we assume that both countries are of the
same type.
Competitive behavior implies that in an h-type country the consumer price of
the private good, pF, must be equal to its unit cost, c, plus the per unit tax, tF.
That is,
pF"c#tF. (1)
In an e-type country, on the other hand, "rms engage in tax evasion so
that a particular "rm's taxes depend on its reported sales. The "rm may
thus attempt to evade taxes by reporting only a proportion, a, of its sales.
One can easily show that under this circumstance, a "rm's expected pro"t, n, is
given by
n"[ pC!c!g(1!a)!(a#(1!a)bq)tC]x, (2)
where g(1!a) is the resource cost of concealment per unit of output, with
g'0 and g'0, b (04b41) is the fraction of "rms audited by the tax admin-
istration, (q!1) is the penalty rate applied to the amount of tax evaded (to be
paid in addition to the taxes due), x is the "rm's output, pC the consumer price
and tC the legislated tax rate.

 In Section 6, we will allow for audit technologies to di!er in the two countries.
 This model is based on Cremer and Gahvari (1992,1993).
1638 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

The "rm chooses a to maximize n. The "rst-order condition, assuming

x'0, is
g(1!a)"(1!bq)tC. (3)
We assume bq(1, for otherwise the "rms report honestly and the tax evasion
problem does not arise. Let
h,(a#(1!a)bq)tC (4)
denote the "rm's expected tax payment per unit of output. Then from Eq. (2), the
market equilibrium occurs at
pC"c#g#h, (5)
where both g and h are evaluated at the "rm's optimal value of a. This is the
analogue of the &pF"c#tF' result above in the absence of evasion.

3. Closed borders

In a closed economy, the government determines the level of public good, y,

and its "scal instrument(s). This will be studied for h-type and e-type economies
separately.
(i) The h-type country: The government determines yF and tF in such a way as to
maximize the utility of a representative consumer in h. Formally, this is written
as

max v!(c#tF)#u(yF), (6a)

WF RF
subject to tF5yF, (6b)
where v denotes the reservation price of the private good and u( . ) is assumed to
be an increasing and concave function of its argument. The "rst-order condi-
tions are:
jFA"u(yFA)"1, (7)
where j is the Lagrangian multiplier associated with the government's budget
constraint, and a superscript hc on a variable denotes its equilibrium value
(when the borders are closed and the country is honest). It follows from (7) that
the marginal cost of public funds in this case is one. The tax is essentially
a lump-sum tax with no welfare loss associated with it. Eq. (7) also determines
the optimal level of yF. The required tax rate is then determined from the
government's budget constraint to be tFA"yFA.
(ii) The e-type country: The objective of the government of e is also to maximize
the utility of its residents. The policy instruments now include the audit
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1639

probability b, in addition to the level of public good provision, yC, and the tax
rate, tC. Formally, one can write the problem of country e as

max v!(c#g#h)#u(yC), (8a)

WC RC @
subject to h!d(b)5yC, (8b)
where d(b) denotes the audit cost with d(b)'0. The "rst-order conditions are:
1
jCA"u(yCA)"
(1!bc)tCA d(b)
1# !
(1!a)g (1!a)tCAq
1
" '1, (9a)
(1!bq)tCA
1!
(a#(1!a)bq)g
yCA"hCA!d(b), (9b)
where a superscript ec on a variable denotes its equilibrium value (when the
borders are closed and the country is evading). For ease in notation, the ec
superscripts on b and a have been omitted. The marginal cost of public funds
now exceeds one: Tax evasion associates a welfare loss to a tax which would
otherwise be lump-sum (as it was for the h-type country). It also follows from
(9a) and the concavity of u( . ), that yCA(yFA. Tax evasion increases the resource
cost of providing one unit of public relative to private good. Consequently, at
the optimum, the amount of the public good provided must decline. These
results are summarized as

Proposition 1. In a closed economy, tax evasion increases the marginal cost of

public funds and reduces the level of public good provision.

 In deriving these equations, and similar "rst-order conditions that appear in the paper, we are
making use of Eq. (3) and Lemma A.1 which is given in the Appendix.
 Cowell and Gordon (1988) study the question of tax evasion and public good provision. They
are mainly concerned with the relationship between tax rates and tax evasion and do not address
this particular point. Our result is more reminiscent of Atkinson and Stern's (1974) where they study
the e!ect of distortionary taxation on the level of public good provision (as well as on Samuelson's
rule). They prove that in general it is possible for distortionary taxation to increase, as well as
decrease, the optimal level of public good provision. They further show that a su$cient condition for
a reduction is that the value of the marginal cost of public funds be greater under distortionary
taxation than under lump-sum taxes. This condition holds in our setup where the marginal cost of
public funds is greater than one with tax evasion and equal to one without.
1640 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

Note also that, in the absence of any other di!erences between honest and
evading countries, maximum utility attained in h must be greater than that in e.
This follows as tF"hCA!d(b) and yF"yCA is a feasible policy for h.

4. Open borders

With opening of the borders, the citizens of a given country may "nd it
advantageous to shop from the other country. Whether a particular individual
would do that or not, depends on how far he lives from the border, the
transportation cost and the price di!erential between the two countries. It is
easy to see that in our setup (p !p )/d residents of country 2 shop from country
 
1, where subscripts 1 and 2 refer to the countries.
When the borders are closed, the &tax base' (number of taxpayers) is the
population size and is thus "xed. When borders open, the tax base becomes
endogenous varying with the size of the price di!erential between the two
countries. The government of each country will then be able to a!ect it by the
choice of its tax rate. This introduces a strategic interaction between the
countries. We shall follow the literature on "scal competition and model this
interaction using the Nash equilibrium concept.
In selecting their tax rates, the governments must also take into account the
additional consumer surplus that residents of the high price country will reap by
making their purchases in the low price country. To calculate this, suppose
p 'p . Then a citizen of country 2 who resides at the border and buys from
 
country 1 will have an additional surplus of (p !p ). This extra surplus
 
diminishes as one moves away from the border and dissipates completely at
a distance of (p !p )/d. Given the uniform distribution of people, there will
 
then be additional consumers' surplus of (p !p )/2d accorded to the citizens
 
of country 2 (the high price country). For the purpose of welfare aggregation, we
assume that governments use a utilitarian framework.

5. Identical types

This section studies the implications of two similar countries integrating their
economies.
(i) Union of two h-type countries: Assume country 2 is the high price country.
The government of 2 determines the level of public good provision and its tax

 Negative numbers are allowed. A negative out#ow of persons from country 2, who shop in
country 1, indicates a positive in#ow of shoppers into country 2.
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1641

rate, treating the tax rate of the other country as "xed. Formally, this is written
as
(p !p )
max v!(c#t )#   #u(y ), (10a)
 2d 
W R


p !p
subject to t 1!   5y . (10b)
 d 

Similarly, one can write the problem of the government of country 1 as

max v!(c#t )#u(y ), (11a)

 
W R


p !p
subject to t 1#   5y . (11b)
 d 

When the two countries are of the same type, one can easily prove that there
must exist a symmetric equilibrium such that t "t and y "y . Writing the
   
two sets of "rst-order conditions and simplifying results in
d
jFM"u(yFM)" '1, (12)
d!tFM
where the superscript ho on a variable denotes its equilibrium value (when
countries are of h-type and borders are open).
Eqs. (7) and (12) show that tax competition increases the marginal cost of
public funds; it associates a welfare loss to an otherwise lump-sum tax. More
interestingly, comparison with the corresponding result derived for an h-type
country when the economy was closed, implies that yFM(yFA and tFM(tFA.
Observe that when the two countries are of the same type, the closed-
economy solution corresponds to the optimal solution under integration. Con-
sequently, economic union implies a less than optimal tax rate and public good
provision. These "ndings are in accordance with the standard results in models
of tax competition. They arise as each country attempts to &lure' the other
country's taxpayers by cutting its own tax rate. This type of &ine$ciency' is easily
surmountable if the two countries were to set their tax rate at tFA (optimum tax
under the closed economy). We shall refer to this as a &tax harmonization' policy
and adopt the following terminology:

Dexnition 1. The countries are said to &harmonize' a policy instrument if they set
its value at a common speci"ed level.

 A su$cient condition for this, in addition to the stated conditions on u( . ), is u(0)'1.

1642 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

The following proposition summarizes our results for the union of two h-type
countries.

Proposition 2. Integrating the economies of two h-type countries results in less than
optimal tax rate and public good provision in both. The inezciency can be averted
by a policy of tax harmonization.

(ii) Union of two e-type countries: Governments now have one more strategic
variable, namely, the audit rate. The problem of the higher price country,
country 2, is given by
(g #h !g !h )
max v!(c#g #h )#     #u(y ), (13a)
  2d 
W R@


g #h !g !h
subject to h 1!     !d(b )5y . (13b)
 d  

The problem of the government of country 1 may also be formulated in a similar

fashion. From the "rst-order conditions of these problems, one can easily show
that the symmetric Nash equilibrium (in t and b) satis"es
1
u(yCM)"
(1!bq)tCM d(b) hCM
1# ! !
(1!a)g (1!a)tCMq d
1
" , (14a)
(1!bq)tCM hCM
1! !
(a#(1!a)bq)g d
yCM"hCM!d(b), (14b)
where the superscript eo on a variable denotes its equilibrium value (when
countries are of e-type and borders are open). For ease in notation, the eo
superscripts on b and a have been omitted.
To determine the impact of integration, one must then compare this equilib-
rium with that under closed borders as characterized by Eqs. (9a) and (9b). To
make this comparison, recall that the closed-economy solution is optimal in the
sense that it maximizes the welfare (utility) of a representative person when the
countries are treated symmetrically. We can then write welfare, thus de"ned, as
a function of policy parameters, t and b. Hence, from Eqs. (8a) and (8b) while

 Of course, one cannot compare the two sets of "rst-order conditions directly. While they have
similar algebraic expressions, the values of the variables appearing in them di!er. (Recall that tax
and audit rates are determined endogenously.)
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1643

making use of (3) and (4),

=(t, b),v!(c#g(t, b)#h(t, b))#u(h(t, b)!d(b)). (15)
In the Appendix, we prove that
*=
(tCM, bCM)'0, (16a)
*t
*=
(tCM, bCM)'0. (16b)
*b
These inequalities give rise to:

Proposition 3. Integrating the economies of two e-type countries results in an

equilibrium for the tax and audit rates (tCM, bCM), such that (i) marginal increases in
tCM and bCM are welfare enhancing, and (ii) tCM and bCM are smaller than their optimal
values if W is strictly concave and *=/*t*b'0.

The condition *=/*t*b'0 in Proposition 3 requires that the marginal

social bene"t of auditing will be higher the greater is the tax rate. This appears to
be a reasonable condition.
Intuitively, suboptimality of both instruments may be explained in terms of
"scal externalities. A marginal increase in tax and audit e!orts in one country
confers bene"ts on the other (by reducing the number of their cross-border
shoppers). The externalities are ignored in setting the tax and audit rates. To
avert this ine$ciency, the two countries must harmonize their tax rates and
audit strategies, setting them at their optimal levels. This is an important policy
implication which has thus far gone unnoticed.
Whether or not full cooperation would take place is of course an open
question. However, it is important to bear in mind that cooperation in audit
strategies may prove more problematic than cooperation on pure tax policies. It
is true that, with two similar countries, full cooperation would bene"t them
both. Nevertheless such a policy may never take hold. A di$culty arises as the
member countries attempt to harmonize their audit strategies. Legislated tax
rates are publicly observable. As such, adherence to agreed upon tax policies can
easily be checked and veri"ed by all countries concerned. This is clearly not the
case for audit strategies. The government of one country can hardly observe the
enforcement e!orts of the other countries. Most importantly, it will be too
di$cult, if not impossible, to verify such e!orts. In particular, the number of

 Formally, inequalities (A.6a) and (A.6b), which appear in the Appendix, re#ect these externali-
ties.
1644 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

persons audited or the amount of money spent on auditing do not necessarily

translate into e!orts. This is more a matter of the organization of the tax
administration (i.e. the structure of the auditors' incentives and the like). Such
provisions are too complex to be spelled out in detail or written into contracts
between countries.
An implication of the above discussion is that "scal competition in audit
strategies may remain e!ective even in the face of tax harmonization policies. To
shed light on this, we next investigate the e!ects of a &partial' harmonization
policy in which the countries agree on setting their tax rates only. We prove in
the Appendix:

Proposition 4. Integrating the economies of two e-type countries, creates a poten-

tial for xscal competition which will not be deterred by a policy of tax harmoniz-
ation alone. In particular, given any harmonized tax rate, tK , and assuming W to be
strictly concave, the corresponding Nash equilibrium value for the audit probability,
b (tK ), is less than optimal.
L
The message of Proposition 4 is that harmonization of tax policies alone is
not su$cient to prevent "scal competition. Audit strategies open up another
avenue for "scal competition. Given any harmonized tax rate, "scal competition
will lead to a Nash equilibrium value for b which is smaller than its second-best
value (that which maximizes = conditional on t"tK ). In particular, if (tH, bH)
maximizes =, "xing t at tH, implies that b (tH) will be below bH. That is, in
L
attempt to lower their e!ective tax rates, countries will tacitly encourage tax
evasion! This is a crucial point which has thus far been neglected in the literature
on tax competition and tax harmonization.

6. Di4erent types

We now turn to examining the integration of countries 1 and 2 when they can
be of di!erent types. This raises the interesting question of what determines
a country's type. It may be that type is an exogenous characteristic of a country
re#ecting the prevailing &attitude' of a country's residents. For example, suppose
the residents (including the owners of the "rms) in a country derive such

 Proposition 4 concerns the implication of tax harmonization at any given rate. However, if the
governments of the two countries decide to cooperate on a tax policy, they will attempt to choose the
&best' harmonized rate. The best is of course conditional on the fact that they cannot harmonize their
audit probabilities. This raises the following interesting question. How does the optimal harmonized
tax rate in this case di!er from the optimal harmonized tax rate when audit probabilities too can be
harmonized? Unfortunately, we have been unable to draw any unambiguous conclusions regarding
this comparison.
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1645

disutility from evasion, no matter how small, that they will refrain from it. Such
a country will always be of the honest type. Alternatively, it may be that type is
a result of a policy decision re#ecting not a country's attitude but its audit and
concealment technologies. This is a less restrictive assumption and the one
which we will adopt.
To make our points in the simplest possible way, we consider the extreme case
of one country with evasion and the other with none. To model this situation, it
will be su$cient to assume that there is some "xed level of investment in audit
technology which, if undertaken, allows a country to carry out audits at zero
marginal cost. The cost of this "xed investment varies between countries
resulting in their being either honest or evading. The cost is "nanced by
a lump sum tax on the country's residents.
To minimize the number of cases considered, we make a further simplifying
assumption. Thus assume one of the counties, say country 2, faces a prohibitively
high xxed investment cost. This country will always be evading. Country 1, on the
other hand, faces a "xed investment cost, F, that is low enough to make it choose
to be of type h when its economy is closed. The "rst interesting question that arises
in this environment is this. Will it be possible that "scal competition induces
country 1, which was honest before integration, to become evading after?
To answer this question, one has to characterize equilibria under a policy of
honesty and evasion for country 1 and examine the possibility that the latter will
result in a higher level of utility for it. If country 1 becomes evading, the
equilibrium under integration will be identical to the case we studied previously
when examining the union of two e-type countries; see (14a) and (14b). In this case,
superscript ee will be used to denote the equilibrium values (in either country).
The second possible outcome is when country 1 "nds it optimal to remain
honest even after integration. To characterize the equilibrium in this case, one
can no longer appeal to symmetry. In writing country i's (i"1, 2) best reply
function, one has to distinguish between three possible cases: p 'p , p "p
   
and p (p . We start by characterizing the equilibrium for p 'p .
   
The government of 1 would now determine y and t while treating the tax
 
rate and the audit probability of country 2 as "xed. Formally, this is written as
(t !g !h )
max v!(c#t )#    #u(y )!F, (17a)
 2d 
W R


t !g !h
subject to t 1!    5y . (17b)
 d 
The problem of the government of 2, on the other hand, is

 This assumes that audit technologies continue to di!er between the two countries even after
economic union. This seems to be a reasonable assumption, at least in the &short run'.
1646 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

max v!(c#g #h )#(y ), (18a)

  
W R @


t !g !h
subject to h 1#    !d(b )5y . (18b)
 d  

The "rst-order conditions of each problem determine one country's

policy parameters as a function of the other country's. Simplifying the
resulting equations, while making use of the equality versions of (17b) and (18b),
yields

d!t #g #h
u(y )"    , (19a)
 d!2t #g #h
  


t !g !h
y "t 1!    , (19b)
  d

1
u(y )" , (19c)




t !g !h (1!b q)t h
1#    1!   ! 
d (a#(1!a)b q)g d


1
u(y )" , (19d)




t !g !h (1!b q)t h d(b )
1#    1#   ! ! 
d (1!a)g d (1!a)t q



t !g !h
y "h 1#    !d(b ). (19e)
  d 

Eqs. (19a)}(19e) determine the equilibrium values for y , t , y , t , b on the

    
assumption that in equilibrium p 'p . Similarly, one can write a set of
 
equations characterizing equilibrium under the assumption of p 4p . How-
 
ever, at this level of generality, it is not possible to determine if the problem has
an equilibrium with p 'p or p 4p . Indeed, there may even be more than
   
one equilibrium or none at all. Assuming a unique equilibrium, we will denote
the values of the variables by superscript he.
To examine the possibility that country 1 becomes evading, we denote the
maximum attainable utility by ; and determine if it is possible for ;CC';FC,
 
given that ;CA(;FA. Interestingly, the answer to this question is yes, as demon-
 
strated by the following numerical example.
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1647

Table 1

Closed borders Economic integration Economic integration

1 honest; 2 evading 1 honest; 2 evading 1 evading; 2 evading

t 1.0 0.9534 0.9350


b 0.0 0.0000 0.0530

h 1.0 0.9534 0.9045

p 1.0 0.9534 0.9302

y 1.0 0.9523 0.8482

; 1.0!F 0.9977!F 0.9052

t 0.9935 0.9351 0.9350

b 0.0560 0.0531 0.0530

h 0.9635 0.9047 0.9045

p 0.9888 0.9303 0.9302

y 0.9008 0.8494 0.8482

; 0.9067 0.9064 0.9052


6.1. Numerical example C1

Assume a logarithmic utility for the consumption of public good. That is,
u(y)"ln(y). (20)
Further assume that the audit costs are quadratic and given by
d"20b, (21)
while the resource cost of evasion is
g"(1!a)#0.3(1!a). (22)
Finally, set the penalty rate at q"10, transportation cost at d"20, and
normalize c"0 and v"2.
Using the expressions in Section 3, we "rst calculate the equilibria for
countries 1 (honest) and 2 (evading) when the borders are closed; see column 1 of
Table 1. Next, we calculate the Nash equilibrium with open borders under the
assumption that country 1 remains honest while country 2 is evading. One can
do this by following the steps described earlier in this section. It turns out that
there is one equilibrium here with pFC'pFC; see column 2 of Table 1. Finally, we
 
calculate the Nash equilibrium for the union of two evading countries (see last
column of Table 1).
The reported "gures in Table 1 show that, when the borders are closed,
country 1 will opt for honesty if F(1.0!0.9067"0.0933. Similarly, when

 For simplicity, we ignore cases that arise with equality signs.

1648 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

borders open, country 1 will achieve a utility level of 0.9977!F, if honest, and
0.9052, if evading. It follows that if F(0.0925, the country will be honest both
before and after economic integration. On the other hand, if 0.0925(F
(0.0933, then country 1 will be honest before integration but evading after. This
is summarized as:

Proposition 5. Integrating the economies of an honest and an evading country may

induce the honest country to also become evading.

While Proposition 5 invokes a possibility only, this is a most interesting

phenomenon. It points out to a certain rami"cation of integration that has
hitherto gone unnoticed.

6.2. Integration, welfare and honesty

The previous subsection has demonstrated the possibility of an honest coun-

try turning evading. It would be interesting to investigate under what circum-
stance this cannot occur. One such case is particularly telling. Consider country
1 which opts for honesty when its borders are closed, so that

;FA';CA. (23)
 
Now suppose integration increases country 1's welfare even further (assuming it
remains honest). That is, suppose

;FC';FA. (24)
 
One can then show that under this circumstance, country 1 can never do any
better by becoming evading. It will thus remain honest.
To see this, note that from (23) and (24),

;FC';CA. (25)
 
On the other hand, if country 1 becomes evading as a result of integration, it will
end up with a utility level less than that under evasion with closed borders.
(Recall that integrating the economies of two evading countries makes both
worse o! as compared to their positions under closed borders). That is,

;CC(;CA. (26)
 
Inequalities (25) and (26) then imply that ;FC';CC: Country 1 will remain
 
honest after integration. This argument is summarized as

Proposition 6. Consider country 1 for which ;FA';CA holds so that it chooses to

 
be honest when its economy is closed. If by remaining honest, the country's welfare
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1649

increases over its closed-border level (;FC';FA), the country will remain honest
 
(;FC';CC).
 
What happens to a country's welfare is closely linked to what happens to the
country's prices. In particular, note that country 1 faces not only negative but
also positive consequences as a result of integration. The negative consequences
are the familiar ones emanating from "scal competition. On the positive side,
country 1's comparative advantage in audit technology allows it, for a given
e!ective tax rate in both countries, to sell the private good at a lower price
(compared to the evading country whose price has to cover the concealment
cost). In this way, country 1 will gain some of the taxpayers of country 2. The
end result is complicated and depends on how price competitive country 2 can
become through "scal competition. The following proposition stipulates a su$-
cient condition for country 1 to become better o! as a result of integration by
remaining honest. If this occurs, country 1 will remain honest as it can
never do any better by becoming evading. The proposition is proved in the
Appendix.

Proposition 7. Consider country 1 for which ;FA';CA holds so that it chooses to

 
be honest when its economy is closed. Assume the country remains honest after it
forms a union with country 2. If the equilibrium price of the private good in country
2 does not fall below the price that prevailed in country 1 before integration
(pFC5pFA), then (a) by remaining honest, country 1's welfare increases over its
 
closed-border level (;FC';FA); and (b) it will remain honest (;FC';CC).
   
To see the intuition behind this result, recall that "scal competition arises as
each country attempts to increase its tax base by cutting its e!ective tax and
lowering its price. Now suppose country 2 (the evading country) will never "nd
it optimal to lower its price below the closed-border price of country 1 (the
honest country). As a result, there will be no reason for country 1 to deviate from
its closed-border price, unless such a move will increase its welfare even further.
Under this circumstance, however, as Proposition 6 shows, country 1 will not
become evading.
Recall that the prices of country 2 will have to cover its concealment cost.
Consequently, whether or not country 2 will "nd it optimal to lower its price in
the way that was suggested by Proposition 7, is closely linked to its concealment
cost. The condition is equivalent to one in which country 2 will not "nd it
optimal to undercut (e!ectively) the tax that prevailed in country 1 (when its
border was closed) by a margin that exceeds country 2's concealment cost.
As a "nal observation, note that the numerical example of the previous
subsection (with pFC"0.9303 and pFA"1) does not satisfy the su$cient condi-
 
tion of Proposition 7. For the purpose of illustration, we next present another
numerical which does. The example is identical to the previous one except in the
1650 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

speci"cation of the resource cost of evasion, which now takes the following
form


0.5(1!a) if 0((1!a)4,
g"  (27)
3(1!a)!2(1!a)#0.5 otherwise.

The equilibrium solutions before and after integration are reported in columns
1 and 2 of Table 2. In this case, pFC!pFA"1.1242!1.0'0, with the welfare of
 
country 1 increasing from 1.0!F to 1.0074!F as a result of integration.
One implication of these numbers, and Proposition 7, is worth emphasizing.
We have argued before that lack of public observability of audit strategies
implies that member countries will not be able to harmonize all aspects of their
"scal policies. Now, even if one ignores this problem, Proposition 7 suggests that
some countries may not have the incentive to want to coordinate their "scal
policies. When the member countries are of the same type, they will all su!er as
a result of "scal competition. Consequently, all members will have an incentive
to coordinate their policies. Our example indicates that such incentives may in
fact be lacking when the countries are dissimilar.

6.3. Tax harmonization

Next, we examine the rami"cations of a tax harmonization policy. We have

argued before that when the two countries are of the same type, such a policy is
not su$cient to eradicate the ine$ciency of "scal competition. However, if the
countries are unable to coordinate their audit strategies, they may consider
harmonizing their tax rates only. This will be more complicated when the
countries are of di!erent types. To investigate this, assume that the countries
agree to set their respective (legislated) tax rates, say, at the optimal tax rate for
country 2 under closed borders.
Given this restriction, one can easily "nd the new Nash equilibrium under
integration. Country 1, if it remains honest, has no optimization problem. With
its tax rate set, it will spend whatever tax revenue it gets on the provision of the
public good. Country 2, on the other hand, will continue to optimize as in
Section 4 but not over t. The results for this case are summarized in the third
column of Table 2.

 This function is, as stipulated, an increasing and convex function of 1!a. It is also continuous
over (0,1), and di!erentiable everywhere over (0,1) except at 1!a"1/3. In the di!erent scenarios
considered below, this lack of di!erentiability does not matter as the "rms will always choose a value
for 1!a greater than this &critical value'.
 Of course, this is not to suggest that this, or the closed-border solution for country 1, is the
&optimal cooperative' policy.
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1651

Table 2

Closed Economic Harmonizing Harmonizing

borders integration taxes taxes
1 honest; 2 evading 1 honest; 2 evading 1 honest; 2 evading 1 evading; 2 evading

t 1.0 0.9528 1.2658 1.2658


b 0.0 0.0000 0.0000 0.0440

h 1.0 0.9528 1.2658 0.9455

p 1.0 0.9528 1.2658 1.1541

y 1.0 0.9609 1.2588 0.9069

; 1.0!F 1.0074!F 0.9646!F 0.7481

t 1.2658 1.2193 1.2658 1.2658

b 0.0467 0.0449 0.0441 0.0440

h 0.9649 0.9199 0.9462 0.9455

p 1.1696 1.1242 1.1547 1.1541

y 0.9213 0.8718 0.9127 0.9069

; 0.7485 0.7393 0.7539 0.7481


Compare economic union with and without tax harmonization. The numbers
suggest that tax harmonization hurts country 1 while bene"ting country 2.
That country 1 becomes worse o! is not surprising. Tax harmonization deprives
it from its sole instrument for "scal competition. On the other hand, country 2,
while constrained from competition in tax rates, can alter its e!ective tax
rate through its audit policy. Note also this does not mean that country 2 always
bene"ts; it will depend on the level of the harmonized tax. In this example,
2 bene"ts as the tax is set at its unrestricted choice level. Harmonizing the
tax at a lower rate can make both countries worse o!. For instance, repeating the
example for a harmonized tax rate of one, the optimal tax rate for country
1 under closed borders, results in ;FC"0.9983!F and ;FC"0.7237.
 
Finally, a most interesting implication of tax harmonization is that country 1,
which up to this point has been honest, may "nd it advantageous to become
evading! The last column in Table 2 reports the equilibrium values of the
variables that result in the two countries, assuming both are evading under the
tax harmonization policy. From column 1, we know that since country 1 is
honest under closed borders, F(0.2515. Now compare maximum welfare levels
in country 1 under honesty and evasion (from columns 3 and 4) given tax
harmonization. We see immediately that for 0.2165(F(0.2515, country 1 will
turn to evasion as a result of tax harmonization. Recall that, in the absence of

 That harmonizing on one policy instrument, while retaining national discretion over others,
may be welfare reducing has also been shown by Fuest (1995) in another context.
1652 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

tax harmonization, this country would remain honest even after economic
integration (assuming that it was honest initially i.e. if F(0.2515). We have

Proposition 8. Tax harmonization may hurt both countries. It may also benext the
evading country and cause an otherwise honest country to turn to evasion.

7. Conclusion

This paper has studied the implications of tax evasion for tax competition and
"scal coordination policies in an economic union. Its starting point has been
that of a closed economy. The conclusion there is that tax evasion increases the
marginal cost of public funds and reduces the level of public good provision.
Second, the paper has considered integrating the economies of two symmetric
countries. Tax evasion now leads to "scal competition in tax rates and audit
probabilities, and both instruments will be set at less than their optimal values.
Moreover, while coordinating audit strategies may prove di$cult, the potential
for "scal competition would not be eliminated by a policy of tax harmonization
alone. Barred from competing in legislated tax rates, the member states lower
their e!ective tax rates by cutting their audit probabilities.
Finally, the paper has examined integration of countries which are of di!erent
types. The characteristics of the countries were assumed endogenous in that
evasion in a country is a policy decision linked to its audit technology. The
important lesson here is that it will be possible for a country to be honest when
its borders are closed but become evading as a result of integration. However, it
will also be possible that, despite "scal competition, an honest country would
become better o! with integration (as compared to the closed border case).
Under this circumstance, the country would remain honest. The paper has
derived a su$cient condition for this case. This possibility is particularly
important in that it suggests di!erent countries may view tax competition
di!erently and thus have di!erent incentives when they negotiate common "scal
policies. Our last conclusion concerns implications of tax harmonization. Such
a policy may reduce welfare in both countries; it may also bene"t the evading
country and cause an otherwise honest country to turn to evasion. While these
"ndings describe only possibilities, they are potentially very important. They
point out to certain rami"cations of integration which have hitherto gone
unnoticed. They must be investigated more fully.

Acknowledgements

We thank two anonymous referees and Franc7 ois Bourguignon for valuable
comments and advice.
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1653

Appendix

Lemma A.1.

*a !(1!bq)
" (0, (A.1a)
*tC g
*h (1!bq)tC
"(a#(1!a)bq)! J0, (A.1b)
*tC g
*pC
"a#(1!a)bq'0, (A.1c)
*tC
*a tCq
" '0, (A.1d)
*b g
*h (1!bq)tCq
"(1!a)tCq# '0, (A.1e)
*b g
*pC
"(1!a)tCq'0. (A.1f )
*b
Proof. Di!erentiate Eqs. (3)}(5) in the text with respect to tC and b, and simplify.

Proof of inequalities (16a) and (16b). De"ne

(g #h !g !h )
X (t , b , t , b )"v!(c#g #h )#    
       2d

 
g #h !g !h
#u h 1!     !d (b ) . (A.2)
 d 

by substituting for y from the equality version of (13b) into (13a). This rewrites

the objective function of country 2 as a function of its, and the other country's,
policy instruments while ensuring that the government's budget constraint is
always in balance. From (15) and (A.2), it immediately follows that 
=(t, b),X (t, b, t, b). (A.3)


 Alternatively, one may de"ne X similarly to X . One can then also see that
 
=(t,b),X (t, b, t, b).

1654 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657

Di!erentiating (A.3) partially with respect to t and b then yields

*= *X *X
" #  , (A.4a)
*t *t *t
 
*= *X *X
" #  . (A.4b)
*b *b *b
 
Now, at the symmetric Nash equilibrium (tCM, bCM), we have

*X
 (tCM, bCM, tCM, bCM)"0, (A.5a)
*t

*X
 (tCM, bCM, tCM, bCM)"0. (A.5b)
*b

While from di!erentiation of (A.2) and Lemma A.1,

*X
 (tCM, bCM, tCM, bCM)"(hCM/d)[aCM#(1!aCM)bCMq]u'0, (A.6a)
*t

*X
 (tCM, bCM, tCM, bCM)"(hCM/d) (1!aCM)tCMqu'0. (A.6b)
*b

Substituting from (A.5a), (A.5b) and (A.6a), (A.6b) in (A.4a) and (A.4b) completes
the proof.

Proof of Proposition 4. Fix t at tK and let countries compete in b. From (A.4b), one
then has

*= *X *X
(tK , b (tK ))"  (tK , b (tK ), tK , b (tK ))#  (tK , b (tK ), tK , b (tK )). (A.7)
*b L *b L L *b L L
 
But from the de"nition of Nash equilibrium, Eq. (A.2) and Lemma A.1, it follows
that

*X
 (tK , b (tK ), tK , b (tK ))"0, (A.8a)
*b L L

*X
 (tK , b (tK ), tK , b (tK ))"(hK /d)(1!a( )tK qu'0. (A.8b)
*b L L

 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1655

Consequently,

*=
(tK , b (tK ))'0. (A.9)
*b L

Inequality (A.9), together with the strict concavity assumption on =, proves the
proposition.

Proof of Proposition 7. There are two possible cases which we will consider
separately.

(i) pFC4pFC. We will then have

 


pFC!pFC
;FC"v!(c#tFC)#u tFC 1#   !F
   d


pFC!pFA
'v!(c#tFA)#u tFA 1#   !F
  d


pFC!pFA
"v!(c#tFA)#u(tFA)!F#u tFA 1#   !u(tFA)
   d 


pFC!pFA
";FA#u tFA 1#   !u(tFA). (A.10)
  d 

From (A.10), we have

pFC!pFA50N;FC';FA.
   

(ii) pFC'pFC. In this case,

 


pFC!pFC (pFC!pFC)
;FC"v!(c#tFC)#u tFC 1!   #   !F
   d 2d


pFA!pFC (pFA!pFC)
'v!(c#tFA)#u tFA 1!   #   !F
  d 2d
1656 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657


pFC!pFA
"v!(c#tFA)#u(tFA)!F#u tFA 1#  
   d

(pFA!pFC)
!u(tFA)#  
 2d


pFC!pFA (pFA!pFC)
";FA#u tFA 1#   !u(tFA)#   . (A.11)
  d  2d

This time, from (A.11), we have

pFC!pFA50N;FC';FA.
   
This proves part (a); part (b) follows from (a) and Proposition 6.

References

Atkinson, A.B., Stern, N.H., 1974. Pigou, taxation and public goods. Review of Economic Studies 41,
119}128.
Cowell, F.A., Gordon, P.F., 1988. Unwillingness to pay: Tax evasion and public good provision.
Journal of Public Economics 36, 305}321.
Cremer, H., Gahvari, F., 1992. Tax evasion and the structure of indirect taxes and audit probabilit-
ies. Public Finance/Finances Publiques, supplement 47, 351}365.
Cremer, H., Gahvari, F., 1993. Tax evasion and optimal commodity taxation. Journal of Public
Economics 50, 261}275.
Edwards, J., Keen, M., 1996. Tax competition and the Leviathan. European Economic Review 40,
113}134.
Fuest, C., 1995. Interjurisdictional competition and public expenditure: Is tax co-ordination
counterproductive? Finanzarchiv 52, 478}496.
Kanbur, R., Keen, M., 1993. Jeux sans frontières: Tax competition and tax coordination when
countries di!er in size. American Economic Review 83, 877}892.
Keen, M.J., 1989. Pareto-improving indirect tax harmonisation. European Economic Review 33,
1}12.
Lenk, T., Fuge, H., Schneider, F., 1998. ZuruK ck zu mehr FoK deralismus: Ein Vorschlag zur Neugestal-
tung des Finanzausgleichs in der BRD unter besonderer BeruK cksichtigung der oK konomischen
Theorie der Politik. Arbeitspapier: 9805, Institut fuK r Volkswirtschaftslehre, Johannes Kepler
UniversitaK t Linz.
Mintz, J., Tulkens, H., 1986. Commodity tax competition between member states of a federation:
Equilibrium and e$ciency. Journal of Public Economics 29, 133}172.
Oates, W.E., 1972. Fiscal Federalism. Harcourt Brace Jovanovich, New York.
Sinn, H.W., 1990. Tax competition and tax harmonization in Europe. European Economic Review
34, 489}504.
Sinn, H.W., 1994. How much Europe? Subsidiarity, centralization and "scal competition. Scottish
Journal of Political Economy 41, 85}107.
Stigler, G.J., 1957. The tenable range of functions of local government. Joint Economic Committee,
Federal Expenditure Policy for Economic Growth and Stability, Reprinted In: Phelps, E.S.,
(Ed.), Private Wants and Public Needs, Revised edition, 1965, New York, Norton, pp. 167}176.
 H. Cremer, F. Gahvari / European Economic Review 44 (2000) 1633}1657 1657

Wildasin, D.E., 1989. Interjurisdictional capital mobility: Fiscal externality and a corrective subsidy.
Journal of Urban Economics 25, 193}212.
Wilson, J., 1986. A theory of inter-regional tax competition. Journal of Urban Economics 19,
296}315.
Xiwen, F., Wilson, J., 1996. Tax evasion and the optimal tax treatment of foreign-score income.
Department of Economics, Indiana University, Bloomington, IN.
Zodrow, G.R., Mieszkowski, P.M., 1986. Pigou, property taxation and the under-provision of local
public goods. Journal of Urban Economics 19, 356}370.