Int Tax Public Finance

DOI 10.1007/s10797-011-9197-5

Tax avoidance versus tax evasion:

on some determinants of the shadow economy

Reinhard Neck · Jens Uwe Wächter ·

Friedrich Schneider

© Springer Science+Business Media, LLC 2011

Abstract In this paper we investigate how the possibility of (legal) tax avoidance af-
fects the extent of (illegal) tax evasion and hence the shadow economy. We formulate
a theoretical microeconomic model of household behavior in which households can
participate in the official and in the shadow economy. Using comparative statics, we
show that the complexity of the tax system affects participation in the shadow econ-
omy negatively, i.e. a more complex tax system with more possibilities of legal tax
avoidance implies, ceteris paribus, a smaller labor supply in the shadow economy. In
addition, we show that a reduction in the maximum admissible number of working
hours in the official economy increases the labor supply in the shadow economy.

Keywords Shadow economy · Tax evasion · Tax avoidance · Tax system · Labor
market regulation

JEL Classification H26 · H31 · H24

An earlier version of this paper was presented at the IIPF Congress, Uppsala, 2010. Thanks are due
to John Hudson (University of Bath) for helpful comments and to an anonymous referee and the
editor of this journal for their suggestions. The usual caveat applies.
R. Neck ()
Department of Economics, Klagenfurt University, Universitätsstrasse 65-67, 9020 Klagenfurt,
Austria
e-mail: [email protected]

J.U. Wächter
DekaBank, Frankfurt am Main, Germany

F. Schneider
Department of Economics, University of Linz, Linz, Austria
 R. Neck et al.

1 Introduction

Over the last few years, the phenomenon of the shadow economy has been paid in-
creasing attention by politicians, administrators, and social scientists. In particular,
after the recent economic crisis, the public budgets of most industrialized countries
have fallen deep into deficit, leading to calls for higher taxes or other government
revenues to compensate for the resulting increase in public debt. In such a situation,
a large shadow economy may lead to an erosion of the tax base, a decrease in tax
receipts, and thus to a further increase in the budget deficit. Moreover, in times of a
growing shadow economy, economic policy is based on misleading official indicators
(such as unemployment, labor force, income, consumption), which may encourage
policy-makers to take inappropriate decisions.
These growing concerns have led many economists to take on the challenging task
of measuring the size and growth rate of the shadow economy, to trace its main causes
and to analyze interactions between the official and unofficial economies. Thus, there
have been a number of studies measuring the size of the shadow economies and its
consequences for the official economy in several countries; see Feld and Schneider
(2010) for a recent survey and Schneider (2008) for some key contributions. In the
present paper, the main focus is on the effects of it being legally possible to avoid
income taxes (in the official labor market) on the size of the shadow economy. Sec-
tion 2 motivates the relevance of this topic by an empirical observation. In Sect. 3,
a simple theoretical model of the “complexity” of the tax system is developed while
Sect. 4 presents comparative statics results on the effects of some features of such a
tax system on the shadow economy. Finally, Sect. 5 summarizes the results.

2 An empirical observation

In an earlier study, we investigated the determinants of the shadow economy in Aus-

tria empirically; see Schneider and Neck (1993). The volume of the shadow economy
was estimated by the currency demand approach. The main focus was on the effects
of changes in tax rates and tax structures on the development of the shadow economy
over time. In line with international evidence, we found a strong influence for the
direct tax burden and a weaker one for the indirect tax burden. However, when con-
sidering the major tax reform of 1988/1989, we were initially puzzled by its effect on
the Austrian shadow economy.
The main features of this tax reform were:
– a considerable reduction in all marginal tax rates on income, e.g., from 62% to
50% (top marginal tax rate)
– a reduction in the number of income tax rates from 10 to 5
– a decrease in average tax rates by 8 to 10 percentage points
– a reduction in the number of tax exemptions and tax loopholes
– a considerable decrease in the tax rates on capital gains and profits and an overall
simplification of the system
Our expectations were that such a tax reform, with its sizeable decrease in the tax
burden on most households, should unambiguously decrease the size of the shadow
Tax avoidance versus tax evasion: on some determinants

economy. However, this was not the case; instead, the Austrian shadow economy
continued to increase (albeit more slowly than before) in the years following this tax
reform.
When looking into this state of affairs more closely, we discovered that, ceteris
paribus, the considerable reduction in the direct marginal and average tax burden
should have decreased the shadow economy. However, the negative effect of the de-
crease in personal income tax rates was offset more than twice as much by the effects
of the massive reduction in the complexity of the tax system, which in turn raised the
volume of the shadow economy. Increased regulation intensity in the labor market
contributed to increasing the shadow economy as well, compensating for the negative
effect of the reduction in income tax rates. Thus, we have to explain why there was
a negative effect caused by the complexity of the tax system (which was confirmed
by a regression analysis for Austria) and a positive one for labor market regulations
boosting the size of the shadow economy. A theoretical analysis contributing to an
explanation of these effects is provided in the following sections of this paper.

3 A theoretical model of the shadow economy

In this section, we describe a microeconomic model of household behavior concen-

trating on income tax and labor-supply. Starting with Allingham and Sandmo (1972),
many authors have focused on the determinants of tax evasion and participation in
the underground economy using theoretical microeconomic models.1 In such a theo-
retical analysis, it is necessary to consider relatively simple tax schedules; hence, the
literature has not paid much attention to the question of how the “complexity” of the
tax system affects the shadow economy. Even when income tax is the main focus, not
every feature of an actual tax schedule can be incorporated into the analysis. Here, we
try to capture the notion of the “complexity” of income tax as follows: a “complex”
income tax schedule allows for more possibilities of legal tax avoidance than a “sim-
ple” one by providing tax exemptions and reductions of various kinds. For example,
tax reforms in many countries have attempted to make the income tax schedule less
“complex” by reducing marginal income tax rates and simultaneously broadening the
tax base by abolishing several exemptions and loopholes in the income tax schedule.
In this context, “comprehensive income tax” can be considered to display a very low
degree of “complexity.”
Thus, if the government changes the amount of “complexity” in the income tax
schedule, it affects the choices of taxpayers between legal tax avoidance and illegal
tax evasion. These effects have been studied by Alm (1988), Cowell (1990a, 1990b)
and Slemrod (2001). Alongside different ways of modeling the costs of tax sheltering,
our analysis differs from theirs by explicitly taking the distinction between an official
and an underground labor market into account instead of regarding income as exoge-
nous. Thus, we follow the modeling framework of Isachsen and Strom (1980) but
include a progressive instead of a proportional linear income tax and the possibility
of legal tax avoidance.

1 For a survey, see, for example, Cowell (1990a).

 R. Neck et al.

We start from the partial model of the household developed in Neck et al. (1989)
and extended in Schneider and Neck (1993). All households are assumed to be iden-
tical with respect to their preferences and their endowments with time; hence we
concentrate on a representative household. This household is the only active decision-
maker considered; we assume a nonstrategic government and ignore its budget con-
straints and objective function. The only activities of the government are raising tax
revenues and punishing tax evasion. Labor is the only source of income for the house-
hold; two kinds of labor can be supplied, labor in the official sector and in the un-
official (underground) sector of the economy, denoted by the superscripts o and u,
respectively. The labor markets are assumed to be perfectly competitive; hence the
hourly wages for both kinds of labor, denoted by W o and W u respectively, are as-
sumed to be given to the household. Both kinds of labor are taken to be homogeneous;
there are different risks associated with each, but otherwise they are perfect substi-
tutes. The household can allocate its available time (measured in hours per day, T )
between the labor supply in the official economy (S o ), the labor supply in the under-
ground economy (S u ), and leisure.
Starting from a linear-progressive income tax, we assume that the household’s pre-
tax income from the official economy, W o S o , is subject to a tax of t1 W o S o −t0 , where
t1 (0 < t1 < 1) is the marginal tax rate and t0 > 0 is a minimum income guaranteed
by the government (a negative income tax). Next, this “simple” linear income tax is
converted into a “complex” tax schedule by introducing the possibility of legal tax
avoidance or sheltering, t1 (W o S o − a) − t0 , where a is the income exempt from tax-
ation. In this model, this serves to summarize, as simply as possible, the possibilities
of reducing the individual tax burden otherwise allowed by numerous complicated
regulations on exemptions from taxable income and tax deductions for the house-
hold. Under any realistic tax schedule, this amount will depend on the household’s
behavior and on the tax laws, as well as possibly on discretionary decisions made by
the government or the tax office.
To make the analysis manageable, we assume a simple function for a:

a = a1 f (e). (1)

Here, e is the “effort” the household makes to take advantage of the possibilities of tax
avoidance. This may consist of actual activities undertaken by the household (learn-
ing the tax laws, etc.), but may also include outlays for advice and any other monetary
costs the household incurs to secure tax avoidance. Both kinds of tax avoidance activ-
ities entail opportunity costs and hence affect the household’s utility negatively; they
may also affect the household’s budget constraints in connection with its allocation
of time and income. In order to keep our model as simple as possible for deriving
comparative statics results, we simplify the decision problem of the household as
follows: we summarize all inconveniences to the household due to (legal) tax avoid-
ance activities by the effort variable e, irrespective of whether they take the form of
time-consuming activities reducing the household’s utility from leisure or require ex-
penditure for the purchase of services, which also decreases its utility. This variable e
is measured in terms of monetary units; it expresses opportunity costs and enters the
household’s utility function negatively. In a more elaborate model, one might distin-
guish between effort in terms of the time spent by the household to identify possible
Tax avoidance versus tax evasion: on some determinants

loopholes on the one hand and monetary expenditure for tax consultants, etc. on the
other. Then there would be two effort variables, one in terms of time and the other in
terms of monetary units. Such modeling would complicate the analysis considerably
as it would necessitate taking the household’s consumption decision into account in
addition to its labor supply decision. Instead, here we assume that as far as time-
consuming activities of the household are involved in its efforts, we assume them to
take place in its leisure time.
f (.) can be interpreted as the household’s “tax avoidance production function”; it
relates its effort to the extent the household can avoid tax given by a1 . Denoting par-
tial derivatives by subscripts, we assume positive but diminishing returns for effort:
fe > 0, fee < 0.
The parameter a1 , which is determined by the government or tax laws, trans-
lates the results of the household’s efforts into the actual amount of tax avoidance.
It summarizes the opportunities provided by the government or tax laws to reduce the
household’s tax payments and can be interpreted as a measure of the “complexity” of
the tax system: When a1 = 0, we are back to the “simple” linear income tax sched-
ule without any exemptions. The greater a1 is, the more the household will be able
to avoid taxes legally by applying some amount of effort for this purpose, and the
greater the income exempt from taxation will be, ceteris paribus. Needless to say, this
is a rather crude model for an income tax schedule with exemptions, but we think it
captures the essential idea of what is meant by a “complex” tax schedule. The amount
of tax avoidance secured by the household’s effort e is now t1 a1 f (e); as this cannot
exceed the amount of taxes to be paid (net of the lump-sum transfer t0 ) t1 W o S o , we
assume that W o S o ≥ a1 f (e) holds for the following analysis.
Alongside the possibility of engaging in (legal) tax avoidance activities, the house-
hold can also obtain income from (illegal) participation in the shadow economy. The
household may supply labor in both the official and the underground economy. In-
come from shadow economy activities is not, of course, subject to taxation unless
discovered by the tax authorities, in which case it is subject to a penalty imposed by
the government. We assume a fixed penalty tax rate t2 ≤ 1 for working in the un-
derground economy, i.e., not more than the income generated by shadow economy
activities is confiscated when detected. This is equivalent to assuming that in the case
of detection, the household is subject to income taxes for the hidden income plus a
fine which does not reduce its legally obtained income.
Households are examined by tax authorities according to a random scheme.
The probability of being detected for receiving income in the shadow economy is
p, 0 < p < 1; both p and t2 are determined by the government. The income obtained
in the shadow economy W u S u is a random variable in this model; only if the house-
hold’s income from the shadow economy is not detected does it accrue fully to the
household.
When the household works in the shadow economy, it may find itself in one of two
possible situations: not being detected (situation 1) or being detected (situation 2). As
[W o S o − a1 f (e)](1 − t1 ) is the household’s taxed after-tax income, the disposable
income of the household in situation i, i = 1, 2, is given by
 
y1 = W o S o − a1 f (e) (1 − t1 ) + a1 f (e) + t0 + W u S u , (2)
 o o 
y2 = W S − a1 f (e) (1 − t1 ) + a1 f (e) + t0 + W u S u (1 − t2 ), (3)
 R. Neck et al.

respectively. For notational convenience, we write


0 if y = y1 (non-detection),
t2 = (4)
t2 if y = y2 (detection).

The household is assumed to have a quasi-cardinal utility function U following the

von Neumann–Morgenstern axioms on decisions under uncertainty and depending
positively on income and negatively on labor supplied in the official and the shadow
economy and on the household’s effort to obtain tax avoidance. The latter is consid-
ered to be independent of the two kinds of labor supply, as it need not involve actual
work by the household but may be delegated to a tax consultant. By including both
S o and S u instead of leisure as arguments of the utility function, differential disu-
tilities of both types of labor can be incorporated. This makes sense because many
people feel uneasy about cheating the government, as shown by empirical studies of
tax morale (see, for instance, Torgler and Schneider 2007); hence, underground labor
may be expected to cause more disutility than working in the official economy. Thus,
we have
 
U = U y, S o , S u , e (5)
with the usual concavity assumptions:

Uy > 0, US o < 0, US u < 0, Ue < 0, Uyy < 0,

(6)
US o S o < 0, US u S u < 0, Uee < 0.

Moreover, we assume additive separability of the utility function:

UyS o = UyS u = Uye = US o S u = US o e = US u e = 0. (7)

This is also a rather restrictive assumption, but we do not see an a priori reason for
a particular sign for any of these expressions; moreover, without this assumption it
is not possible to obtain results which can be interpreted in economic terms. The
household maximizes its von Neumann–Morgenstern expected utility function
   
E[U ] = (1 − p)U y1 , S o , S u , e + pU y2 , S o , S u , e (8)

with respect to S o , S u and e, subject to

So + Su ≤ T , (9)

S o ≤ X, T >X > 0, (10)

S o , S u , e ≥ 0. (11)
Equation (9) defines an upper limit for the total supply of labor in both sectors, while
(10) represents an upper limit for labor in the official economy. X is the maximum
admissible number of hours of work as prescribed by collective agreements, law, or
similar labor market regulations. Working fewer than X hours is possible by part-time
work agreements. The effort variable does not enter the constraints here because we
Tax avoidance versus tax evasion: on some determinants

assume that time-consuming activities for tax avoidance are converted into monetary
units and are regarded as being equivalent to monetary expenditures for the same
purpose.

4 The complexity of the tax system and the shadow economy

4.1 The Household’s optimization problem

The Lagrangian of the household’s problem described in the previous section is

       
L = (1 − p)U y1 , S o , S u , e + pU y2 , S o , S u , e − λ S o + S u − T − μ S o − X .
(12)
This constrained optimization problem can be solved using the Kuhn-Tucker condi-
tions. The first-order necessary conditions are given by

LS o = E[Uy ]W o (1 − t1 ) + E[US o ] − λ − μ ≤ 0, S o ≥ 0, LS o S o = 0, (13)

 
LS u = E Uy (1 − t2 ) W u + E[US u ] − λ ≤ 0, S u ≥ 0, LS u S u = 0, (14)
Le = E[Uy ]a1 t1 fe + E[Ue ] ≤ 0, e ≥ 0, Le e = 0, (15)
Lλ = −S − S + T ≥ 0,
o u
λ ≥ 0, Lλ λ = 0, (16)
Lμ = −S + X ≥ 0,
o
μ ≥ 0, Lμ μ = 0, (17)

where we use the notation (4) and

   
E[Uy ] = (1 − p)Uy y1 , S o , S u , e + pUy y2 , S o , S u , e (18)

and analogous for E[US o ], E[US u ] and E[Ue ].

From the first-order conditions (13)–(17), we can determine optimal values of the
decision variables as functions of the model parameters. As the functional form of
the utility function is not specified, the solutions can be given in implicit form only:
 
S̄ o = S o W o , W u , t0 , t1 , t2 , a1 , p, X, T , (19)
 
S̄ u = S u W o , W u , t0 , t1 , t2 , a1 , p, X, T , (20)
 o 
ē = e W , W , t0 , t1 , t2 , a1 , p, X, T .
u
(21)

4.2 Comparative statics: effects of working time

Next, we investigate the comparative statics effects of variations in the parameters on

the optimal values of the decision variables, holding the values of the other param-
eters constant. To do so, we have to take the total derivative of the first-order con-
ditions. The resulting system of equations can be written in matrix form, using the
(5 × 5) bordered Hessian of the partial second-order derivatives of the Lagrangian
denoted by H. Its determinant is given by

|H| = −E[Uyy ]a12 t12 fe2 − E[Uy ]a1 t1 fee − E[Uee ] > 0. (22)
 R. Neck et al.

First, we determine the comparative statics effect of the restriction of working hours
on the household’s decisions. Using Cramer’s rule, we obtain

∂S o 1  
= −E[Uyy ]a12 t12 fe2 − E[Uy ]a1 t1 fee − E[Uee ] = 1, (23)
∂X |H|
∂S u 1  
= E[Uyy ]a12 t12 fe2 + E[Uy ]a1 t1 fee + E[Uee ] = −1. (24)
∂X |H|

This means that we get corner solutions for the optimum amount of labor supply in
the official and in the shadow economy. The household uses its entire available time
for producing income by working in the official or the shadow economy. This implies
that it is working in the official economy up to the upper limit. If the upper limit X
is reduced, holding the entire available time T constant, the household substitutes
unofficial for official work 1:1, and the total labor supply remains constant.
For an analysis of further effects, we have to assume that the restriction (9) on the
household’s time budget is not binding; otherwise, the signs of virtually all compar-
ative statics effects are undetermined due to the divergence of the substitution effect,
the income effect and the portfolio effect. Under this additional assumption, the max-
imization problem is reduced to a problem under one constraint only. We now can
reduce the model to a (4 × 4) system with Hessian H1 , whose determinant is given
by
 2
|H1 | = a12 t12 fe2 W u p(1 − p)t22 Uyy (y1 )Uyy (y2 )
    2 
+ E[Uy ]a1 t1 fee + E[Uee ] E Uyy (1 − t2 )2 W u + E[US u S u ]
+ E[US u S u ]E[Uyy ]a12 t12 fe2 > 0. (25)

The effect of the upper limit of working hours in the official economy on official and
unofficial labor and on tax avoidance now becomes
∂S o 1  2 2 2  u 2
= a t f W p(1 − p)t22 Uyy (y1 )Uyy (y2 )
∂X |H1 | 1 1 e
    2 
+ E[Uy ]a1 t1 fee + E[Uee ] E Uyy (1 − t2 )2 W u + E[US u S u ]

+ E[US u S u ]E[Uyy ]a12 t12 fe2 = 1, (26)
∂S u 1     
= −E Uyy (1 − t2 )2 W o W u (1 − t1 ) E[Uy ]a1 t1 fee + E[Uee ] < 0,
∂X |H1 |
(27)
∂e 1   2
= −W o W u (1 − t1 )a1 t1 fe p(1 − p)t22 Uyy (y1 )Uyy (y2 )
∂X |H1 |
 
− E[Uyy ]W o (1 − t1 )a1 t1 fe E US u S u < 0. (28)

As in (23), increasing the admissible working hours results in a proportionate in-

crease in the official labor supply. The unofficial labor supply again decreases but we
Tax avoidance versus tax evasion: on some determinants

cannot say whether the decrease compensates precisely for the effect on official la-
bor. The reason for this is the household’s preference for working in the official over
the unofficial economy due to the risk of being detected for working in the shadow
economy and losing (part of) its shadow economy income. The negative effect of
the upper limit to working hours on the tax avoidance effort can be interpreted to
mean that longer official working hours reduce the household’s incentive to engage
in tax avoidance due to the higher productivity of official labor as compared to the
necessary tax avoidance effort.

4.3 Comparative statics: effects of the tax schedule

Next, consider the comparative statics effect of guaranteed minimal income:

∂S o 1
= {0} = 0, (29)
∂t0 |H1 |
∂S u 1     
= −E Uyy (1 − t2 ) W u E[Uy ]a1 t1 fee + E[Uee ] < 0, (30)
∂t0 |H1 |
∂e 1   2
= −a1 t1 fe W u p(1 − p)t22 Uyy (y1 )Uyy (y2 )
∂t0 |H1 |

− E[US u S u ]E[Uyy ]a1 t1 fe < 0. (31)

Increasing the minimal income has no effect on the official labor supply because of
the household’s tendency to use all available time for work in the official economy.
The unofficial labor supply and tax avoidance effort are, however, affected negatively
by the increase in guaranteed income, higher guaranteed income allowing the house-
hold to reduce these unwanted activities.
The effects of the marginal tax rate are given by

∂S o 1
= {0} = 0, (32)
∂t1 |H1 |
∂S u 1     
= E[Uy ]E Uyy (1 − t2 ) W u a1 t1 a1 fe2 − (a1 f − W o S o )fee
∂t1 |H1 |
    
− E Uyy (1 − t2 ) W u a1 f − W o S o E[Uee ] , (33)
∂e 1   u 2  
= − W a1 t1 fe a1 f − W o S o p(1 − p)t22 Uyy (y1 )Uyy (y2 )
∂t1 |H1 |
  2
− E Uyy (1 − t2 )2 W u E[Uy ]a1 fe
   
− E[US u S u ] E[Uyy ]a1 t1 fe a1 f − W o S o + E[Uy ]a1 fe > 0. (34)

The marginal tax rate has no effect on the official supply of labor. In contrast to the
results of Isachsen and Strom (1980) and Neck et al. (1989), in this model the sign of
the effect of the marginal tax rate on labor in the unofficial economy is not uniquely
determined without further assumptions. Under the (not too restrictive) assumption of
a1 f < W o S o , meaning that it is not possible for the entire legal income to be exempt
 R. Neck et al.

from taxation, we get (34), i.e., a positive influence of the marginal tax rate on the
tax avoidance effort. This means that an increasing marginal tax rate makes efforts to
reduce the household’s tax obligations more profitable, ceteris paribus.

4.4 Comparative statics: effects of tax complexity

Another aspect of the tax system under consideration is the amount of complexity
built into it. The comparative statics of the complexity parameter a1 are:

∂S o 1
= {0} = 0, (35)
∂a1 |H1 |
∂S u 1   
= E[Uy ]a1 t12 f12 E Uyy (1 − t2 ) W u
∂a1 |H1 |
    
− E[Uy ]a1 t1 fee + E[Uee ] E Uyy (1 − t2 ) W u t1 f < 0, (36)
∂e 1   u 2 2
= − W a1 t1 fe fp(1 − p)t22 Uyy (y1 )Uyy (y2 )
∂a1 |H1 |
   
− E[Uy ]t1 fe E[US u S u ] + E Uyy (1 − t2 )2 (W u )2

− E[US u S u ]E[Uyy ]a1 t12 fe f . (37)

Again, there is no effect on the supply of labor in the official economy. The sign of
the effect on the tax avoidance effort is not uniquely determined. Equation (36) shows
that the effect of the complexity of the tax system on the shadow economy is negative,
as was already shown by Schneider and Neck (1993, p. 350). This means that a more
complex tax system (a tax schedule which allows more exemptions) implies, ceteris
paribus, a smaller labor supply in the shadow economy. An economic interpretation
of this result is that a more complex tax system makes taxpayers’ efforts to avoid tax-
ation legally more profitable. It decreases the household’s inclination for working in
the underground economy, as the reduced tax burden makes tax evasion (with the risk
of being caught and punished) relatively less attractive than tax avoidance. Broaden-
ing the income tax base and removing tax exemptions can therefore increase the size
of the shadow economy, ceteris paribus. Or, to put it differently, illegal tax evasion
is substituted for the decreased possibilities for engaging in legal tax avoidance. This
result provides a theoretical argument for the empirically observed fact that a tax re-
form reducing the complexity of the income tax schedule may result in an increase of
the shadow economy. Instead of engaging in efforts for legal tax avoidance, taxpayers
are motivated to increase their engagement in illegal tax evasion.

4.5 Comparative statics: effects of government policies against tax evasion

The shadow economy’s advantage over the official economy, namely that its income
is not reduced by taxation, has to be balanced against its disadvantage of being subject
to the risk of being discovered and punished. The penalty for working in the shadow
economy and the probability of being detected are crucial factors determining this
Tax avoidance versus tax evasion: on some determinants

risk. Let us first consider the effect of the penalty tax rate t2 :

∂S o 1
= {0} = 0, (38)
∂t2 |H1 |
∂S u 1   u 2 u 2 2 2
= − W S a1 t1 fe p(1 − p)t2 Uyy (y1 )Uyy (y2 )
∂t2 |H1 |
 2  
+ pUyy (y2 )(1 − t2 ) W u S u E[Uy ]a1 t1 fee + E[Uee ]
 
+ pUy (y2 )W u E[Uyy ]a12 t12 fe2 + E[Uy ]a1 t1 fee + E[Uee ] , (39)
∂e 1  u 3 u
= W S a1 t1 fe p(1 − p)t2 Uyy (y1 )Uyy (y2 )
∂t2 |H1 |
+ E[US u S u ]pUyy (y2 )a1 t1 fe W u S u
  2 
− E Uyy (1 − t2 ) W u a1 t1 fe pUy (y2 ) > 0. (40)

The penalty tax rate has no effect on the official labor supply and an ambiguous one
on the unofficial labor supply. The effect on the tax avoidance effort is positive: the
higher expected cost of illegal tax evasion increases the differential return to legal tax
avoidance, ceteris paribus. The higher expected income loss due to the higher penalty
tax is compensated for by greater investment in exploiting the legal possibilities of
avoiding taxes.
Next, consider the effect of the probability of being detected:

∂S o 1
= {0} = 0, (41)
∂p |H1 |
∂S u 1    
= Uy (y1 ) − Uy (y2 )(1 − t2 ) W u E[Uy ]a1 t1 fee + E[Uee ]
∂p |H1 |
 
+ a12 t12 fe2 t2 W u (1 − p)Uy (y2 )Uyy (y1 ) + pUy (y1 )Uyy (y2 ) , (42)
∂e 1     2
= E[US u S u ]a1 t1 fe Uy (y1 ) − Uy (y2 ) − a1 t1 fe t2 W u
∂p |H1 |
 
× (1 − p)Uy (y2 )Uyy (y1 ) + pUy (y1 )Uyy (y2 )(1 − t2 ) > 0. (43)

We get the same qualitative results as for the penalty tax rate, with an analogous inter-
pretation. In both cases, government policies aimed at intensifying its battle against
the shadow economy will result in increased efforts by taxpayers to avoid taxation by
legal means. Therefore, such policies may be successful at reducing illegal behavior
but need not be successful if their sole aim is to increase government revenue.

4.6 Comparative statics: effects of wage rates

Finally, let us consider the effects of wage rates in the official and the unofficial
economy on the supply of labor in both sectors and on the tax avoidance effort. For
 R. Neck et al.

the wage rate in the official economy, we get

∂S o 1
= {0} = 0, (44)
∂W o |H1 |
∂S u 1      
= − E[Uy ]a1 t1 fee + E[Uee ] E Uyy (1 − t2 ) W u S o (1 − t1 ) < 0,
∂W o |H1 |
(45)
∂e 1   u 2 o
= − W S (1 − t1 )a1 t1 fe p(1 − p)t22 Uyy (y1 )Uyy (y2 )
∂W o |H1 |

− E[US u S u ]E[Uyy ]a1 t1 fe S o (1 − t1 ) < 0. (46)

The household uses all hours allowed for work, hence a change in the wage rate has no
effect on its supply of labor in the official economy. Its supply of labor in the shadow
economy depends negatively on the wage rate in the official economy due to the
income effect of changes in the wage rate. A higher wage rate in the official economy
leads ceteris paribus (in particular, with fixed working hours) to higher income in
the official economy, making it less attractive for the household to engage in the
unofficial economy. As we saw in (30), higher income from the minimum income
transfer reduces the underground labor supply, and the same is true for higher income
due to a higher wage rate. A similar effect works for tax avoidance activities: due to
its increasing marginal disutility, it becomes less attractive as legal pre-tax income
increases.
The wage rate in the unofficial economy has the following effects:
∂S o 1
= {0} = 0, (47)
∂W u |H1 |
∂S u 1 
= −W u S u a12 t12 fe2 p(1 − p)t22 Uyy (y1 )Uyy (y2 )
∂W u |H1 |
   
− E Uyy (1 − t2 )2 W u S u E[Uy ]a1 t1 fee + E[Uee ]
  
− E Uy (1 − t2 ) E[Uyy ]a12 t12 fe2 + E[Uy ]a1 t1 fee + E[Uee ] , (48)
∂e 1       
= E Uy (1 − t2 ) W u − E[US u S u ]S u E Uyy (1 − t2 ) a1 t1 fe < 0.
∂W u |H1 |
(49)

There is again no effect on the supply of labor in the official economy while the
effect in the shadow economy is ambiguous, mirroring the ambiguity of the wage
rate effect on the labor supply in general due to the different signs of income and
substitution effect. Tax avoidance efforts are negatively affected by the wage rate in
the shadow economy. This reflects the substitution of higher expected income from
shadow economy activities for less income from legal tax exemptions.

4.7 Summary of comparative statics effects

The comparative statics effects are summarized in Table 1.

Tax avoidance versus tax evasion: on some determinants

Table 1 Results of comparative

statics analysis Change of parameter Effect on
So Su e

X + − −
t0 0 − −
t1 0 ? +
a1 0 − ?
t2 0 ? +
p 0 ? +
Notes: +: positive effect; Wo 0 − −
−: negative effect; 0: no effect; Wu 0 ? −
?: ambiguous effect

The lack of comparative statics influences of all parameters but X on S o reflects

the assumption that the overall time constraint is not binding, therefore, the household
will always work up to the maximum number of hours allowed for. Thus, under the
conditions of this model, liberalizing working hours will be an effective instrument
if a government wants to increase work in the official economy at the expense of
shadow economy activities without losing revenues from additional tax avoidance
activities. On the other hand, changing tax rates or increasing efforts to detect shadow
economy activities do not necessarily contribute to this goal. The model can provide
some explanations for the observed effect of the shadow economy increasing after a
tax reform reducing the complexity of the income tax schedule and after increasing
labor market regulations.

5 Summary and future research areas

In this paper, we analyzed the effects of some aspects of the tax system, especially
the possibility of avoiding taxes legally, and labor market regulation on the size of the
shadow economy, emphasizing the negative effect of the complexity of the tax system
on the extent of the shadow economy: reducing the complexity of the income tax
schedule was shown to result in an increase in the shadow economy, ceteris paribus.
This provides a theoretical argument for an empirically observed phenomenon. When
the tax system and its structure are significantly changed, as was the case in Austria
in 1988/1989, for example, one would expect that a massive decrease in the direct
tax burden would lead to a decline in the shadow economy as well. However, the
direct and indirect tax burden is not the only important factor influencing the shadow
economy; it is also affected by the complexity of the tax system and the burden of
labor market regulations. The theoretical results in this study suggest that both factors
may offset a lower direct tax burden, showing that merely lowering the tax burden is
not sufficient to bring about a decline in the shadow economy. Such considerations
may be relevant for tax policy design when introducing a comprehensive income tax
reform or similar.
In general, our results should be seen as a first step in studying the complicated
interactions between changing tax structures and their effects on the shadow econ-
omy. Obviously, our model is simple and neglects many features of real world tax
 R. Neck et al.

systems. However, it may serve as a starting point for further research. A next step
will consist in studying the influence of changing tax systems on the shadow econ-
omy and their consequences for the official economy in a general equilibrium model
comprising two sectors (an official and an unofficial sector) and in a model allowing
for a more active (strategic) role on the part of the government. General qualitative
results cannot be expected from such generalizations, however; one would have to
resort to numerical analyses of the effects of parameter changes.

References

Allingham, M. G., & Sandmo, A. (1972). Income tax evasion: a theoretical analysis. Journal of Public
Economics, 1, 323–338.
Alm, J. (1988). Compliance costs and the tax avoidance tax evasion decision. Public Finance Quarterly,
16, 31–66.
Cowell, F. A. (1990a). Cheating the government: the economics of evasion. Cambridge: MIT Press.
Cowell, F. A. (1990b). Tax sheltering and the cost of evasion. Oxford Economic Papers, 42, 231–243.
Feld, L. P., & Schneider, F. (2010). Survey on the shadow economy and undeclared earnings in OECD
countries. German Economic Review, 11, 109–149.
Isachsen, A., & Strom, S. (1980). The hidden economy: the labor market and tax evasion. The Scandina-
vian Journal of Economics, 82, 304–311.
Neck, R., Schneider, F., & Hofreither, M. F. (1989). The consequences of progressive income taxation for
the shadow economy: some theoretical considerations. In D. Bös & B. Felderer (Eds.), The political
economy of progressive taxation (pp. 149–176). Heidelberg: Springer.
Schneider, F. (Ed.) (2008). The economics of the hidden economy, 2 vols. Cheltenham: Edward Elgar.
Schneider, F., & Neck, R. (1993). The development of the shadow economy under changing tax systems
and structures: some theoretical and empirical results for Austria. FinanzArchiv, 50, 344–369.
Slemrod, J. (2001). A general model of the behavioral response to taxation. International Tax and Public
Finance, 8, 119–128.
Torgler, B., & Schneider, F. (2007). What shapes attitudes toward paying taxes? Evidence from multicul-
tural European countries. Social Science Quarterly, 88, 443–470.