Economic Effects of VAT Reform in Germany

Discussion Paper No. 06-030
Economic Effects of
VAT Reform in Germany
Stefan Boeters, Christoph Böhringer,
Thiess Büttner and Margit Kraus

---

Discussion Paper No. 06-030
Economic Effects of
VAT Reform in Germany
Stefan Boeters, Christoph Böhringer,
Thiess Büttner and Margit Kraus
Download this ZEW Discussion Paper from our ftp server:
ftp://ftp.zew.de/pub/zew-docs/dp/dp06030.pdf
Die Discussion Papers dienen einer möglichst schnellen Verbreitung von
neueren Forschungsarbeiten des ZEW. Die Beiträge liegen in alleiniger Verantwortung
der Autoren und stellen nicht notwendigerweise die Meinung des ZEW dar.
Discussion Papers are intended to make results of ZEW research promptly available to other
economists in order to encourage discussion and suggestions for revisions. The authors are solely
responsible for the contents which do not necessarily represent the opinion of the ZEW.

---

Non-Technical Summary
Re?ecting distributional concerns, many countries apply VAT reductions to goods
which make up a larger share in the consumption of low-income households. This
paper addresses the question to what extent VAT di?erentiation can be rationalised
on distributional grounds.
We employ an applied general equilibrium (AGE) model to investigate distributional
e?ects and e?ciency implications for structural VAT reforms based on empirically
data for Germany. In our numerical simulations we compare a pure VAT reform,
where the di?erentiated VAT is replaced with a uniform rate, and scenarios in which
the additional revenues are compensated with tax reductions involving the marginal
income tax rate (MITR), the income tax allowance (ITA) or the social security
contributions (SSC).
Our main ?ndings can be summarised as follows: The abolition of the reduced VAT
rate in itself has only a small redistributive e?ect towards more inequality. Therefore,
VAT di?erentiation can hardly be considered as an e?ective means of redistribution
policy. When we compensate the abolition of reduced VAT rates with reductions
in the marginal income tax rate or cuts in social security contributions, there is
scope for signi?cant gains in overall welfare. A budget compensation scheme based
on a reduction in the income tax allowance, however, produces welfare losses, due to
the implied increase in the marginal tax burden. Policy-induced changes in macro-
economic indicators like GDP, employment, domestic capital use, or aggregate con-
sumption echo this welfare ranking of the tax instruments. While the distributional
e?ects of VAT reforms are within a relatively narrow range, the industry e?ects (in
terms of variation in industry output) are much more pronounced. This indicates
that the VAT rate di?erentiation can be viewed primarily as an industry-speci?c
subsidy rather than an instrument of redistribution. From a political economy point
of view, the sectoral implications highlight lobbying interests of adversely a?ected
sectors to work against changes of the actual VAT structure.

---

Economic E?ects of VAT Reform in Germany
Stefan Boeters?, Christoph Böhringer‡+?,
Thiess Büttner†, Margit Kraus‡
‡ZEW, Mannheim,
?CPB, Den Haag
+Department of Economics, University of Heidelberg
†Ifo Institut and Department of Economics, Munich University
April 2006
Abstract
In the tax policy debate, di?erentiation of value-added taxes is often justi?ed
by distributional concerns. Our quantitative analysis for Germany indicates
that such concerns are misplaced. We ?nd that the abolition of VAT di?er-
entiation has only negligible redistributive e?ects. Instead, reduced VAT are
found to act as industry-speci?c subsidies. Whereas the overall welfare e?ects
of pure VAT reforms are very small, a revenue-neutral introduction of a har-
monised VAT combined with reductions in the marginal income tax rates or
social security contributions turns out to produce substantial welfare gains for
all households.
Keywords: VAT, tax reforms, distribution, e?ciency, applied general equi-
librium
JEL Code: D58, H22, H24
?Corresponding author: Christoph Böhringer, Centre for European Economic Research, P.O.Box
103443, 68034 Mannheim, Germany, e-mail: boehringer@zew.de

---

1
Introduction
Consumption taxation through value-added taxes (VAT) is usually considered as
a relatively e?cient way of raising public funds. Theoretical analysis points to the
neutrality of VAT with respect to intertemporal consumption decisions, whereas in-
come taxes tend to distort the trade-o? between consumption and savings. On the
other hand, a uniform VAT is often criticised on the basis of its allegedly regres-
sive distributional e?ects. Re?ecting distributional concerns, many countries apply
VAT reductions to speci?c goods, which make up a larger share in the consump-
tion of low-income households. In the EU, all countries but one use reduced VAT
rates for speci?c consumption commodities. Especially in the old EU member states
VAT reductions on food, water, medication, and public transport are quite common
(European Commission, 2005).
This paper addresses the question to what extent VAT di?erentiation can be ra-
tionalised on distributional grounds. VAT di?erentiation is an indirect instrument
of distribution as it is not associated with the individual ability to pay of di?erent
consumers. There are more direct instruments of distributive policy such as income
taxation or monetary transfers. Thus, from an applied policy perspective, we must
be concerned with how large the redistributive e?ects of VAT di?erentiation are
in practice, and whether or not alternative policy instruments are more e?ective
as redistributive devices. Answers to these concerns cannot be given by abstract
theoretical considerations. They depend on the precise type of products favoured
by VAT reductions and the demand and supply conditions on the respective mar-
kets, which are determined by household preferences, production technologies, factor
endowments, and the market structure.1
In this paper, we employ an applied general equilibrium (AGE) approach to in-
vestigate e?ciency and distributional impacts for structural VAT reforms based on
1 In the public ?nance literature a number of reasons are mentioned why VAT di?erentiation
might be justi?ed under e?ciency considerations: (i) administrative and compliance costs (Keen
and Mintz, 2004), (ii) the existence of shadow markets, (iii) di?erences in price elasticities of
goods, or (iv) complementarity of consumption goods with untaxed leisure activities. However,
these reasons are either di?cult to ascertain on empirical grounds (due to the lack of data) or
irrelevant in policy practice. (As a prime example VAT reductions are applied to goods with
inelastic demand such as food, which is contrary to optimal taxation reasoning.) More recently,
VAT reductions have also been proposed as a measure to stimulate employment in labour intensive
service industries.
1

---

empirically observed data for Germany. The AGE approach provides a comprehen-
sive framework for studying the e?ects of policy interference on all markets of an
economy, rigorously based on microeconomic theory. The simultaneous considera-
tion of the origin and spending of the agents’ income makes it possible to address
both economy-wide e?ciency as well as distributional impacts of policy regulation.
This has made AGE models a standard tool for the quantitative analysis in many
policy domains including ?scal, trade and environmental policy.
The strand of AGE literature that is directed to the analysis of VAT reforms is
relatively small compared to other public ?nance issues such as income taxation or
pension reform: Ballard et al. (1987) analyse VAT in the USA as a possibility to
increase the dynamic e?ciency of the tax system. Hamilton and Whalley (1989) use
a static model to explore special intricacies of the interaction of federal and provin-
cial taxes in Canada. Gottfried and Wiegard (1991) focus on the implementation
of the VAT and compare two di?erent institutional settings, tax exemption vs. zero
rating, for the German economy. Dixon and Rimmer (1999) use a dynamic model
for Australia to investigate VAT reforms with a special focus on the induced interna-
tional trade e?ects. In a more recent paper, Åvitsland and Aasness (2004) combine
a dynamic AGE and a microsimulation model to assess VAT reform scenarios for
Norway.
In Germany, the VAT has a standard and a reduced rate. The latter applies mostly
to food, public transport, and print-media products. We use our AGE model to
simulate variants of a revenue-neutral abolition of the reduced VAT rate. The results
of the simulations con?rm doubts about the e?ectiveness of reduced VAT rates as a
redistributive instrument and point to welfare gains from uniform taxation. These
welfare gains are boosted if taxes other than VAT are included in the tax reform.
Among alternative sources of revenue which keep the overall budget constant (tax
recycling instruments), revenue-neutral reductions in marginal income tax rates and
— in particular — cuts in the social security contributions provide larger welfare
gains. At the sectoral level, the reduced VAT rate works mainly as a subsidy to the
respective ?nal-goods producers and their intermediate-input suppliers.
The remainder of this paper is organised as follows. Section 2 gives an overview
of the model structure and parametrisation. Section 3 provides the results of the
scenario simulations. Section 4 concludes.
2

---

2
Model and Parametrisation
For our simulation analysis we draw on a standard AGE model which has been re-
?ned to address central issues of VAT reforms.2 Speci?c extensions include the disag-
gregation of the household sector into income terciles, where each tercile has a special
income composition and consumption structure. For the empirical parametrisation of
the model, various data sources are used including the German Input-Output Table
for 1997, the production-consumption transition matrix — the so-called “Z-matrix”,
and the German Income and Expenditure Survey (EVS) have been combined to
form a consistent benchmark dataset.
In the following, we ?rst summarise the basic features of our AGE model (Sec-
tion 2.1). A detailed description of the household representation follows in Section
2.2. Finally, we discuss data and calibration issues (Section 2.3). A comprehensive
algebraic summary of the model is provided in the appendix.
2.1
Basic Model Structure
Firms and factors of production
The AGE model underlying our VAT reform analysis for Germany features 69 pro-
duction sectors. In each sector, output is produced from intermediate inputs, capital,
and labour of two skill types (high skilled and low skilled). Production possibilities
are characterised through nested constant-elasticity-of-substitution (CES) produc-
tion functions, which describe the trade-o? between various inputs. Perfect compe-
tition implies that there are no pure pro?ts. The primary factors labour and capital
are remunerated according to their respective marginal productivities. Cost min-
imisation by ?rms yields demand functions for production inputs at the sectoral
level.
The domestic labour market is characterised through frictions and equilibrium un-
employment. We make use of a wage-curve relationship in which the rate of unem-
ployment is linked to the degree of progressivity of the income tax due to an implicit
wage-bargaining mechanism (Koskela and Vilmunen, 1996). Capital is fully mobile
2 Speci?cally adapted re?nements of the standard model have been applied recently to the
climate policy debate (Böhringer and Lange, 2005) and labour market policies (Böhringer, Boeters,
and Feil, 2005).
3

---

across sectors, and the domestic capital market is perfectly competitive. At the in-
ternational level, domestic and foreign capital are treated as imperfect substitutes
to account for less than perfect international capital mobility. The calibration of the
respective parameters is discussed in Section 2.3.3.
sectoral output
?Y
intermediate Inputs
KLE aggregate
?KLE
?M
KE aggregate
high
low
from different sectors
skilled
skilled
labour
labour
?KE
capital
energy
?E
different energy carriers
Figure 1: Production structure for a representative sector
In Figure 1 we adopt the following notation:
?Y
:=
elasticity of substitution between the aggregate of intermediate pro-
duction inputs and the input composite of labour, capital and en-
ergy,
?KLE
:=
elasticity of substitution between the capital-energy aggregate and
(skilled as well as unskilled) labour,
?M
:=
elasticity of substitution between intermediate inputs entering the
sectoral composite of intermediate inputs,
?KE
:=
elasticity of substitution between capital and aggregate energy,
?E
:=
elasticity of substitution between di?erent energy carriers entering
the aggregate energy input.
4

---

Foreign trade
Domestically produced goods are converted through a constant-elasticity-of-trans-
formation function into goods destined for the domestic market and the export
market, respectively. Export and import prices in foreign currency are considered
as exogenous (small-open-economy assumption). Analogously to the export side,
we adopt the Armington assumption of product heterogeneity for imports. A CES
function characterises the trade-o? between imported and domestically produced
varieties of the same good. The Armington good enters intermediate and ?nal de-
mand. Foreign closure of the model is warranted through the balance-of-payments
constraint.
Government budget
Given our focus on VAT reform, the model emphasises the role of consumption
taxation. The VAT captures di?erences across consumption categories with three
levels of the tax rate (full rate, reduced rate, and tax exempt goods). Furthermore,
we account for the indirect impact of value-added taxation in the production of goods
which are tax exempt. Besides the VAT, direct taxes and social security contributions
of households are di?erentiated by household types. Social security contributions are
assumed to be proportional to labour income while income taxation takes the form
of a linear progressive schedule (tax allowance combined with a constant marginal
tax rate). Finally, the model contains sectoral output taxes and subsidies as well as
import and export levies.
Private households
We distinguish three representative households capturing the lower, middle, and up-
per tercile of the income distribution. Each household takes a labour-leisure decision
and chooses between di?erent consumption goods. Details about the characteristics
of the disaggregated households are provided in the following section.
2.2
Representation of the Household Sector
2.2.1
Household Disaggregation
The private household sector is disaggregated into three households representing,
respectively, the lower, middle and upper income tercile of the households according
to the German Income and Expenditure Survey (EVS). The EVS is a representa-
tive household survey by the German Federal Statistical O?ce. The 1998 sample
5

---

comprises 62.000 households. The ?rst part of the survey reports data on household
structure, housing situation, ?nancial and tangible assets as well as debt. The sec-
ond part contains income and expenditure items adapted to the classi?cation of the
input-output accounts.
Households are grouped into the three income terciles according to their “equivalent
household income”. Household income is divided by the respective number of house-
hold members in order to compare households of di?erent sizes. We use the square
root of the household size as equivalence scale to compute the respective number
of household members, thereby re?ecting economies of scale due to ?xed costs in
household consumption.3 The income and expenditure values of the three ?ctitious
representative households are then set to the arithmetic mean of the respective in-
come class.
Table 1 summarises basic characteristics of the household types. Disposable income
— the sum of the rows “consumption” and “savings” — varies substantially across the
three terciles. Taking the ?rst tercile as the basis of comparison, disposable income
of the second tercile is higher by roughly one half, whereas the disposable income
of the third tercile is three times as high. Less than two thirds of gross income (or
not even more than one third as in the case of the ?rst tercile) are made up of
factor income. The residual income consists mainly of transfer payments, pensions
and private credit (with “savings” meaning gross savings). The income tax schedule
is progressive as can be seen from the average and marginal tax rates4; in addition,
we report the implicit tax allowances associated with a linear progressive income
tax scheme. Average social security contributions (SSC) are decreasing in income
due to an assessment threshold for the base of SSC.
Labour supply of the representative households is split into skilled and unskilled
labour by summing up skill-speci?c incomes of all individual households in the re-
spective tercile. We assume a uniform wage per skill type which amounts to an
e?ciency weighting of individual working hours. Furthermore, the unemployment
rate is assumed to be uniform across households (but di?erent for the skill types)
and is calculated by summing up employed and unemployed persons in the terciles.
We count the registered unemployed as involuntary unemployed while the unem-
ployed that are not registered are classi?ed as voluntary unemployed.
Table 2 reports the consumption shares of the household terciles by VAT categories.
3 Cf. e.g. Biewen (2000) or Atkinson et al. (1995, 18?.) for alternative scales.
4 The percentage numbers are given relative to gross factor income.
6

---