EC O LO G I CA L E C O N O M I CS 6 3 ( 2 00 7 ) 5 9–6 9

a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m

w w w. e l s e v i e r. c o m / l o c a t e / e c o l e c o n

ANALYSIS

Sustainability impacts of car road pricing: A computable

general equilibrium analysis for Austria

Karl W. Steininger a,b,⁎, Birgit Friedl a,b , Brigitte Gebetsroither b

a
Department of Economics, University of Graz, Universitaetsstrasse 15, A-8010 Graz, Austria
b
Wegener Center for Climate and Global Change, University of Graz, Leechgasse 25, A-8010 Graz, Austria

AR TIC LE I N FO ABS TR ACT

Article history: Nationwide car road pricing schemes are discussed across Europe. We analyse the impacts
Received 5 July 2006 of such schemes with respect to environmental, economic and social indicators of
Received in revised form sustainability, also quantifying the trade-offs among these three dimensions under
22 September 2006 different charging principles and revenue recycling options. In our analysis we employ a
Accepted 23 September 2006 computable general equilibrium (CGE) approach, develop a modelling structure for private
Available online 22 November 2006 transport and provide detailed empirical analysis for the case of Austria. Regarding the
social dimension, it has often been argued that poorer households (and commuters) would
Keywords: have to bear a disproportionate share of the road pricing burden. We find the contrary, i.e. a
Sustainable transport stronger negative policy impact on richer households, and on a small group of intensive car
Road pricing users. The choice of revenue recycling is able to ameliorate the negative social and economic
Transport emission reduction effects of road pricing, without reversing the desired positive environmental effects. For
Passenger transport policy political feasibility, questions of distributional impacts are most urgent and therefore we
Income distribution address them systematically within a quantitative framework.
© 2006 Elsevier B.V. All rights reserved.

1. Introduction While initial responses to the environmental challenge in

transport were primarily technological, there has since been a
Environmental challenges in transportation occurred repeat- focus also on reducing and re-structuring transport demand
edly in history, an early example in modern times being the (Berechman, 2002) and on using transport pricing policies
concern in the 1870s in London that, with a further growing (Hensher and Button, 2000; Viegas, 2003). However, most of
population, the horse manure (an earlier form of transport- present transport policies still mainly centre on vehicle
derived pollutant) would overwhelm the city.1 For the current emission and safety standards, annual license duties and
transport system, environmental concerns have become parking fees and thus may be classified as out-dated or, in the
increasingly important since the 1970s. Transport is now economist's jargon, “second-best” (see, e.g., Calthrop and
responsible for at least a quarter of world primary energy use Proost, 1998; Nash et al., 2001; Ubbels et al., 2002).
and for a comparable share of CO2 emissions (Berechman, Road pricing, a vehicle user charge based on distance
2002). travelled, is clearly highly suitable in addressing transport-

⁎ Corresponding author. Tel.: +43 316 380 8441; fax: +43 316 380 9520.
E-mail address: karl.steininger@uni-graz.at (K.W. Steininger).
1
The impact on daily life is captured in an example quoted by Weightman and Humphries (1985: 36): “But with all the horse traffic, there
was an awful amount of dirt on the streets, some of them were in a dreadful state. There were crossing sweepers, rather oldish men, and if
one gave them a coin they would be very pleased to sweep a path across the street in front of one.”

0921-8009/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.ecolecon.2006.09.021
60 E CO L O G I CA L E CO N O MI CS 63 ( 20 0 7 ) 5 9–6 9

related environmental problems since it allows differentiation been devoted to the social dimension. However, social ques-
of charges across location, time, vehicle type etc. (see, e.g., tions of distribution and equity are of major importance for the
Johansson and Mattsson, 1995; Button and Verhoef, 1998; acceptance of road pricing (see also Mayeres and Proost, 2001).
Sterner, 2002; Santos, 2004). It combines the ability to untie The low public acceptance for car road pricing derives from the
congestion, to reduce air pollution, and to raise revenues for perception of road pricing as intrinsically unjust (Oberholzer-
new infrastructure and other investment ( Jakobssen et al., Gee and Weck-Hannemann, 2002) and as infringing on
2000; May et al., 2002; Parry and Bento, 2002). personal freedom (Jakobssen et al., 2000). Since income con-
For trucks, kilometre based road pricing systems have been straints can be identified as the key determinant for transport
introduced in Europe at various levels of sophistication, i.e. demand reductions when travel costs increase (Jakobsson
from section-charging on highways in various countries, and et al., 2000), the burden of road pricing is likely to fall on poor
electronic charging on highways in Austria (2004) and Ger- households (see also West, 2004) and on households living in
many (2005), up to charging for use of the full road network in peripheral regions (Hammar and Jagers, in press).
Switzerland (2001). In Sweden, a mileage tax for diesel trucks This article therefore aims to not only quantify economic
was in place from 1988 up to 2004.2 For private cars, charging in and environmental impacts of road pricing, but to investigate
Europe has been introduced either section-wise for highways the effects on those groups perceived as bearing the main
or for urban areas mainly in the form of toll rings, e.g. in burden of road pricing. A passenger transport focused
Scandinavian countries and more recently in London (and computable general equilibrium (CGE) model is developed
similar as in other parts of the world, such as in Singapore). The for this purpose. To better understand the distributional
discussion of nationwide kilometre based charging also for effects of road pricing, we distinguish four classes of income
private cars has slowly taken off in various European countries in our model. Thus, the present model goes beyond existing
(see e.g. Nash et al., 2001; Ubbels et al., 2002). ones in several respects. First, and in some way parallel to
The introduction of extensive, nationwide road pricing also earlier transport policy discussion, most of the models used so
for passenger transport requires careful impact analysis. In far to analyse road pricing are either limited to freight
terms of sustainability impact assessment, the European transport (e.g. Steininger, 2003; De Jong et al., 2004) or they
Union (EU), for example, asks for “careful assessment of the target urban road pricing or toll ring pricing only (e.g. Mayeres
full effects of a policy proposal [that] must include estimates of et al., 1996; Proost and van Dender, 2001). We focus on
economic, environmental and social impacts” (EC, 2001). As nationwide car road pricing. Second, while an increasing
set out in Böhringer and Löschel (2006) in some detail, the number of papers addresses the welfare effects of a tax
quantification of trade-offs in such an impact assessment suitable for internalising external transport costs (e.g. Jansen
analysis calls for the use of numerical techniques and that one and Denis, 1999; Nash et al., 2001), distributional aspects have
of these approaches, CGE modelling, “can incorporate several hardly been considered in economic transport policy models.
key sustainability (meta-)indicators in a single micro-consis- Third, in methodological terms, our approach unites modules
tent framework, thereby allowing for a systematic quantita- from the sciences of transportation and economics to a
tive trade-off analysis between environmental quality, consistent integrated assessment of economic, environmental
economic performance and income distribution.” (Böhringer and distributional impacts. The core ingredient in this
and Löschel, 2006: 50-51). As passenger car road pricing merging is the calibration method of the economic passenger
remains a field of national policy responsibility, also within transport focused CGE model, integrating results of the pure
the EU,3 we develop a CGE model for a sustainability impact heuristic passenger mode choice model.
assessment at the national level of such policy. The paper proceeds in four steps as follows. In Section 2,
Within the different dimensions of sustainability, the the passenger transport focused CGE model is developed. In
economic one has been most broadly discussed for road Section 3, transport and consumption databases are merged, a
pricing. The initial focus was the optimal pricing in the social accounting matrix is constructed which differentiates
transport sector to combat congestion (e.g. Lindsey and sufficiently between the various cost components of private
Verhoef, 2001). Increasingly the environmental impacts of car transport, and transport elasticities of substitution are
transport have been taken up in co-determining the optimal calibrated for this model. Section 4 reports on the simulation
price (e.g. Mayeres et al., 1996).4 In our view least effort has results of nationwide car road pricing in various implemen-
tation schemes. Section 5 discusses the distributional, eco-
2
The Swedish mileage tax for diesel trucks was introduced to nomic, and environmental impacts of this policy. A final
counterbalance low diesel relative to gasoline taxes, with the section summarises the main conclusions and outlines key
former reflecting the practically identical composition of diesel
factors for increasing acceptability prior to the introduction of
and (also low-taxed) fuel oil. When Sweden joined the EU, this
mileage tax was replaced by a diesel tax (Sterner, 2002). a car road pricing system.
3
See EC (2006), Article 1, 1.e.
4
E.g., for trucks EC (2006) asks for road pricing on the basis of
external costs. Article 11 states, that “[…] the Commission shall 2. Transport demand CGE model
present, after examining all options including environment,
noise, congestion and health related costs, a generally applicable,
In order to analyse the economic, environmental and distri-
transparent, and comprehensible model for the assessment of all
butional impacts of car road pricing, we develop a CGE model
external costs to serve as the basis for future calculations of
infrastructure charges, and this model will be accompanied by an which is a standard small open economy CGE model in many
impact analysis on the internalisation of external costs for all respects, but obviously with more detailed passenger trans-
modes of transport […]”. port modelling.
 EC O LO G I CA L E C O N O M I CS 6 3 ( 2 00 7 ) 5 9–6 9 61

We distinguish four representative households h (reflecting

income quartiles), which differ in their respective transport
expenditures. For all consumption expenditures other than
transport, expenditure shares (but not levels) are assumed
identical across income groups. This simplification is justified
since on the aggregate level, only two consumption aggregates
differ considerably across income groups, namely expendi-
tures on housing and on transport.6
For substitution elasticities between transport and non-
transport goods, due to lack of adequate data on differences
across income groups, we assume them as uniform across
household types. This last point is important. One could think
Fig. 1 – Structure of household final demand for income group of the shift between transport and non-transport goods being
h. Note: Non-zero values denote calibrated elasticities of differently easy for poor and rich households. Using uniform
substitution: σc = transport to non-transport substitution substitution elasticities, distributional impacts of road pricing
elasticity, σT = car to public transport substitution elasticity. are driven by the different transport volumes and expenditure
shares across income groups, based on different transport
behaviour (e.g. intensity of driving). Accordingly, the same
In the production of non-passenger-transport goods 35 road pricing charge induces different reactions across income
sectors, j, are distinguished, and a nested constant elasticity of groups due to different relative increases in car user costs.
substitution (CES) structure is used. Capital Kj and labour Lj Foreign trade is modelled under the Armington assump-
serve as primary inputs, and intermediate inputs enter in a tion of product differentiation, i.e. a change in relative prices
Leontief fashion.5 In describing the production of the various of domestic and foreign goods shifts the trade balance in the
passenger transport intermediate and final consumption respective sector, the degree of which is determined by a
goods, the model structure reflects both the purpose of sectoral foreign trade elasticity.
analysis as well as the nature of data availability. In preparing As the analysis of the labour market impact of transport
for the CGE analysis we disaggregate a standard social policy is an issue, it is assumed that the labour market does
accounting matrix for Austria, constructed for the year 2000, not clear, and that unemployment is driven by classical, i.e.
in order to differentiate household expenditure across income high minimum wage, unemployment.
groups in terms of public transport and private car use. For the The model is closed by a fixed foreign trade balance at the
latter the various elements of fixed and variable (i.e. kilo- level of the reference year (“neoclassical closure”, fixed foreign
metre-dependent) costs are distinguished, all of which are savings), such that investment is savings driven and the
linked to respective sectors of supply or to government budget foreign exchange rate adjusts to achieve equilibrium.
revenues, such as car and gasoline taxes. We thus manage to
isolate a newly combined passenger-transport good of final
household consumption, which in regular final demand 3. Data set and calibration
statistics is dispersed across various sectors (refined oil,
transport equipment, distribution, finance and insurance, Due to the lack of a database combining information on
and inland transport). passenger trips (by purpose, mode, distance, and frequency)
The structure of consumption, emphasising transport and income for Austria, we econometrically merged the
consumption is given in Fig. 1. Household demand Ch of Austrian mobility survey (Herry and Sammer, 1999), the
household h is governed by a nested CES structure, with Environmental Balance of Transport (Federal Ministry of
unity elastictity of substitution among non-passenger- Agriculture, Forestry, Environment and Water Management,
transport goods Xci,h (i.e. constant expenditure shares for unpublished) and the Austrian consumption expenditure
non-passenger-transport goods), and calibrated elasticities survey (ST.AT, 2002). The challenge here was to have
of substitution between the aggregates of non-transport disaggregated data on both transport expenditures and on tran-
and transport goods Tk as well as among different sport demand in terms of quantity (passenger and vehicle
passenger-transport goods, Tph and Tuh, i.e. private and public kilometres and mode per year), distinguished by household
transport. income levels and region types (urban, semi-urban, peripheral).
As is seen in Fig. 1, fixed and variable cost components of We find that the richest income quartile households drive
car passenger transport combine in a Leontief fashion to their cars more than four times more intensively than the
private car transport Tp, i.e. the elasticity of substitution is poorest (see Fig. 2). This strong difference in mileage across
exogenously set at zero. Basically, kilometre charges (as the
most important variable cost element in the current analysis)
cannot be substituted by fixed cost (such as technical 6
Expenditures on rent, heating and lighting (relative to total
equipment or the like).
consumption expenditures) vary for the example of Austria from
30.4% for the lowest income group to 20.6% for the highest; an
opposite trend can be observed for transport: expenditures on
5 transport increase from 8.5% for low income households to 17.5%
A full list of variables and the core equations as a reference for
the reader are given in the Appendix. for high income households (ST.AT, 2002).
62 E CO L O G I CA L E CO N O MI CS 63 ( 20 0 7 ) 5 9–6 9

port demand model: σC = 0.275 and σT = 0.635. Thus, the CGE

simulations refer to the – in terms of economic welfare and
distributional impacts – “strongest impact”, in terms of
environmental implications “lowest impact”. If, however,
easier substitution between transport and non-transport
goods and/or between car and public transport were possible
(i.e., if larger values of σC and σT apply), negative impacts on
economic welfare would be smaller, and emission reduction
would be larger.

4. Road pricing policy simulations

We distinguish policy scenarios with respect to (a) the type of

road network charged, (b) whether there is (peak hour) time
differentiation, (c) two different charging levels per kilometre,
and (d) two different ways of revenue use. The specific
combinations and detailed information of policy scenarios
considered are stated in Table 2. The road pricing charge is
Fig. 2 – Passenger transport mileage by household and either uniform for the entire network (scenarios A-5, B-5) or is
workday across income classes, Austria 2000. doubled during peak times in urban areas to internalise
congestion externalities (scenarios C-5, D-5).8 The base rate
of 5 Euro-Cents and 10 Euro-Cents is chosen to internalise half
income groups also translates to a difference in consumption or all of estimated marginal external costs (based on average
budget shares across income groups. In accordance with external cost for diesel and petrol cars as the lower bound for
published results on significantly positive income elasticities marginal external costs) for Austria (Herry and Sedlacek, 2003).
for car ownership and car use (see, e.g. Dargay, 2001; After covering system costs (an approximate 15% of
Johansson-Stenman, 2002; Pucher and Renne, 2003; Kletzan revenues), road pricing revenues are used, a third each, for
et al., 2006), the fraction of transport expenditures in Austria road infrastructure investment, public transport improve-
increases from 9.6% for low income households to 18.6% for ment, and household refund. In scenario D-5, a much higher
high income households (ST.AT, 2002; see also Table 1). share is used for public transport improvement (5/9). We will
Furthermore, in rural areas at least 50% of the households first analyse the impacts of road pricing for one policy scenario
spend nothing on public transport while in urban areas many in more detail, before we report on the merits of the five policy
households do not even own a car (ST.AT, 2002). This initial scenarios relative to each other.
situation and its implication for the availability of reaction
options to car road pricing are important for the road pricing 4.1. Road pricing base scenario B-5
impacts we will later find.
A price change in car travel, such as it is triggered by road Implementing car road pricing on the road network nation-
pricing, induces a modal shift and a change in overall wide at a level of 5 Euro-Cent/km, without time differentiation
transport volume. In the CGE approach elasticities of substi- and applying the straight forward “third-each” revenue use
tution denote the strength of these reactions.7 The parameter rule, we find the economic impacts as stated in Tables 3 and 4
values for these transport elasticities of substitution (between (first two columns). In a comparative static analysis we
car and public transport, denoted by σT, and between total “shock” the economy in the base year 2000.
passenger transport and demand for non-transport goods, Let us first look at the overall transport, environmental and
denoted by σC in Fig. 1) were calibrated from a choice switch macroeconomic impacts (Table 3). A reduction of 6.5% in car
passenger transport demand model (Kriebernegg, unpub- vehicle kilometres (some 4 billion veh-km in absolute terms) is
lished). In this model, individual reaction patterns to car observed, with a simultaneous increase in public transport
road pricing were distinguished, namely reactions within the passenger kilometres by 6.3%.
demand structure for car passenger transport (e.g. a change of As the specific emission intensity per passenger kilometre
route or destination) and reactions in demand for public and of cars (strongly) exceeds the one of public transport,9 net
other transport (e.g. a modal shift from car to public transport).
Reactions differ with respect to trip purpose, type of region in
which the trip originates, and road pricing scenario. The 8
According to a recent survey for Europe, Proost et al. (2002)
resulting aggregate elasticities of substitution are quite low
confirm that marginal external costs vary considerably between
due to the cautious assumptions of the choice switch trans- off-peak and peak periods. They are at least twice as high during
peak hours.
9
7
An elasticity of substitution between private and public For example, in 2000 a public transport passenger kilometre on
transport with a value of 0.5, for example, means that a 1% price average was connected to 8% (CO2), 1.5% (CO) and 35% (NOx) of
rise of car travel relative to public transport induces a 0.5% change emissions of one vehicle kilometre of a car in Austria (Federal
in the modal split, here in favour of public transport. Ministry, 2006).
 EC O LO G I CA L E C O N O M I CS 6 3 ( 2 00 7 ) 5 9–6 9 63

transport emission decline as a result of this modal shift and Table 1 – Transport expenditures across income groups,
overall transport volume reduction, with net emission reduc- Austria 2000
tions reported for CO2, CO and NOx in Table 3.10 Monthly earnings Income group
Using the foreign price level as numeraire, GDP increases. A per household
1 2 3 4
new service, the environment, is now paid for, which raises the
overall price level relative to abroad. The relevant measure Less Less Less More
therefore is GDP in purchasing power parity (PPP) terms, which than than than than
we find to decline (by 0.3%). While this decrease in physical € 1478 € 2311 € 3267 € 3267
production lowers indirect tax revenues and exerts a declining Transport expenditures as % of monthly household income
pressure on employment, the latter is outweighed by the Car exp.; fixed costs 5.97 11.00 12.39 14.57
employment increase due to a sectoral shift in production. Car exp.; variable costs 2.49 3.71 3.84 3.58
More specifically, the sectors of construction, textiles and public Public transport exp. 1.13 0.76 0.63 0.47
Total transport exp. 9.58 15.47 16.86 18.61
transport gain at the cost of the sectors machinery, transport
Variable transport expenditure in Euro per km
equipment and all types of energy supply. In total, public
Car 0.050 0.058 0.061 0.046
revenues decline, but lost public tax revenues are more than
compensated for by the (semi-public) net revenues from car Source: ST.AT (2002), Federal Ministry of Agriculture, Forestry,
Environment and Water Management (unpublished), own
road pricing (which account for 1.7 billion Euros in this scenario).
calculations.
We find a considerable variation in car road pricing impact
across household income groups (see Table 4, first two
columns). Due to the introduction of road pricing, private car 4.2. Impacts across road pricing policy scenarios
transport expenditure rises most for the poor by far, increas-
ing by almost a fifth. Nevertheless, the original pre-policy When we also simulate the other policy scenarios in this
mileage level of this group is so low, that overall welfare comparative static analysis, we obtain the results shown in
reduction, as measured by the Hicksian equivalent variation,11 the further columns in Table 3. The most significant impacts
is smallest for this group. Public transport increases degres- arise when the charge is raised to 10 cents/km (scenario C-10):
sively across income groups, due to the pre-policy situation road pricing revenues rise to a level of 5.7 billion Euros (3.5
that the share of public transport decreases with income. billion Euros net of system costs and household refunds).
In order to quantify the net welfare benefit of this policy Nationwide car vehicle kilometre reduction comes close to
scenario at the aggregate level, we differentiate the benefit of 15%, transport CO2 emissions decrease by 1.5 million tons
congestion and other external costs reduction and the cost of (which corresponds to 2.5% of overall Austrian CO2 emissions).
consumption reduction (visible in its direction also in PPP GDP GDP in purchasing power parity terms declines by 1%. Net
decline). Based on average external costs per kilometre, as welfare benefits are at least 644 million Euros, including 170
quantified in Herry and Sedlacek (2003), as approximation for million Euros congestion reduction benefits.
marginal external transport costs,12 we get a value of welfare In scenario D-5, the elasticity of substitution between
benefit of 329 million Euros. However, since external transport private and public transport is raised to 0.9 in order to model
costs will rather rise progressively with transport volume than that a larger share of revenues is directed towards public
linearly as we assume, our benefit quantification can be con- transport (i.e. service improvement). The resulting rise in the
sidered a very conservative one (for more details see use of public transport is significant.
Steininger et al., 2005). Across policy scenarios we find the relevance of two
opposing effects in the labour market. When road pricing
10
We employ a simple emission modeling here, using emission revenue use, e.g. for secondary road infrastructure mainte-
coefficients for private and public transport respectively for all nance, increases (e.g. from A-5 to B-5), labour demand (and
relevant air and greenhouse gas transport emissions (including employment) increases due the high labour intensity of this
those given in Table 3). As car road pricing in the policy scenarios
use.13 However, road pricing revenue use also increases prices
usually discussed (and in the ones we implemented in this
of consumer relevant goods (e.g. construction or electronics),
analysis) is based on the charging principles of distance and/or
time of use (but not of airborne emission intensity) this simplifica- which increases wage rate demands and reduces employ-
tion appears justified. ment. For high road pricing scenarios (C-10), the latter effect
11
The equivalent variation gives the amount of income neces- dominates, which then reduces employment.
sary to compensate an individual (in the pre-policy situation) in Analysing the results across household income groups (see
order to be able to claim equality with the post-policy utility level Table 4), we find a rise in car transport cost by up to almost
(see, e.g. Just et al., 2004). Thus, in the present analysis, the lowest
35% for the poorest group (C-10), but also by up to some 25% for
income households would be willing to pay a fraction of 0.56% of
the other income groups. Variable costs of car transport in
their income to avoid the implementation of road pricing (B-5
scenario, Table 4). As we only look at the user costs of road pricing scenario C-10 roughly triple (with a higher factor for the
across income groups, this Hicksian welfare impact of road richest and poorest, a lower one for the other groups).
pricing is negative. Consumption reduction impacts, which are the income
12
Herry and Sedlacek (2003) take account of the following group specific costs of the road pricing instrument, remain
external costs categories: infrastructure costs, external accident
costs, environmental costs (noise, local pollutants, climate
13
effects), each differentiated by type of street and user, and net But see the sensitivity analysis section in the Appendix for the
of public revenues raised, e.g. from taxes on insurance, vehicle relevance of the assumption on the type of revenue use for the
registration and fuels. net direction of the labour market impact.
64 E CO L O G I CA L E CO N O MI CS 63 ( 20 0 7 ) 5 9–6 9

Table 2 – Policy scenarios

Scenario A-5 B-5 C-5 C-10 D-5

Network charged Urban: full network Full network Full network Full network Full network
rest of Austria:
primary road
network
Time differentiation None None 7–9 a.m. and 7–9 a.m. and 7–9 a.m. and
4 –6 p.m. + 100% 4 –6 p.m. + 100% 4 –6 p.m. + 100%
Charging level 5 cents/km 5 cents/km 5 cents/km 10 cents/km 5 cents/km
Revenue use 1/3 each: road infrastructure, public transport, household refund 1/9 road infrastructure,
5/9 public transport,
1/3 household refund

progressive with rising road pricing charge level across declines, but – contrary to public discussion – with a
households (Table 4). Even though total car transport expen- significantly stronger impact on rich households. This pro-
ditures in scenario C-10 rise by a quarter for the rich, their gressive impact of car road pricing occurs because poor house-
consumption decline is confined to below 4%, and even lower holds spend a smaller share of their income on transport, and
for the other income groups. This consumption impact, which use more intensively public transport. Both factors diminish
is negative across all scenarios, has to be counterbalanced the relative burden of car road pricing on poor households,
with the welfare benefits of an improved environmental and the latter factor also eases the modal switch to public
situation, lower accident costs or the reduction in time spent transport for such households.
in transport, to get the overall net welfare benefit depicted in
Table 3 at the national aggregate level.
Overall, we thus find that purchasing power parity GDP 5. Discussion
slightly declines, and due to reduced transport volumes envi-
ronmental benefits strongly increase as the scope of applica- Despite the modest impact of the road pricing scenarios
tion is widened (i.e. from primary roads only as in scenario A-5 considered, two further points need to be discussed regard-
to total road network) and as the road pricing rate is raised. ing the social consequences of road pricing. The first is usage of
Further, in our burden analysis we find consumption welfare road pricing revenues and the second the impact on severely

Table 3 – Transport, environmental and macroeconomic impacts across road pricing policy scenarios
Reference level Policy Policy Policy Policy Policy
(year 2000) Scenario Scenario Scenario Scenario Scenario
B-5 A-5 C-5 C-10 D-5

Transport variables
Road pricing rate 0 0.05 €/km 0.05 €/km 0.05 €/km 0.10 €/km 0.05 €/km
Revenue use (beyond system costs) for 1:1:1 1:1:1 1:1:1 1:1:1 higher share
road infrastructure, public transport, public transport
household refund
Revenues from road pricing (mn Euro) 0 2949 1915 3073 5720 3066
Car vehicle kilometres (mn veh-km) 63,068 − 6.48% −5.12% − 6.73% −14.44% −6.96%
Public transport (mn pass-km) 21,613 + 6.29% +4.56% + 6.55% +14.77% +12.32%

Environmental variables
CO2 emissions pass. transport (1000 t) 12,395 − 722 −569 − 744 −1,581 −798
CO emissions pass. transport (1000 t) 189.9 − 12.1 −9.5 − 12.5 −26.7 −13.0
NOx emissions pass. transport (1000 t) 38.0 − 2.0 −1.6 − 2.1 −4.4 −1.9

Macroeconomic variables
Welfare change (mn Euro) ( lower bound) 329 273 399 644
GDP (mn Euro) 204,616 + 1.37% +0.87% + 1.43% +2.51% +1.39%
GDP in PPP (mn Euro) 204,616 − 0.34% −0.27% + 0.35% −0.96% −0.41%
Number of employees + 1364 −833 + 1454 −9194 +2103
Unemployment rate 5.84% 5.80% 5.86% 5.80% 6.12% 5.78%
Price of capital 0.07% 0.05% 0.07% 0.15% 0.09%

Budgetary effects (mn Euro)

Due to change in revenues from direct taxes + 57 +12 + 60 −63 +81
Due to change in revenues from indirect taxes − 285 −192 − 297 −601 −430
Due to change in labour market expenditures + 14 −8 + 14 −91 +21
Change in government demand − 424 −338 − 441 −1,087 −655
Revenues from road pricing (semi-public) 1671 1085 1742 3489 1737
 EC O LO G I CA L E C O N O M I CS 6 3 ( 2 00 7 ) 5 9–6 9 65

Table 4 – Distributional impacts across road pricing policy scenarios and household income groups
Transport expenditure impacts

Scenario B-5 Scenario A-5 Scenario C-5 Scenario C-10 Scenario D-5

Car Public Car Public Car Public Car Public Car Public
transport transport transport transport transport

Expenditures Expenditures Expenditures Expenditures Expenditures

Income
b € 1478 +19.3% +8.6% + 11.2% + 6.2% + 20.2% + 9.0% + 34.6% + 19.8% +19.1% +16.0%
b € 2311 +14.0% +6.1% + 8.1% + 4.4% + 14.7% + 6.4% + 25.3% + 14.5% +14.3% +12.0%
b € 3267 +12.4% +5.4% + 7.2% + 3.9% + 13.0% + 5.7% + 22.5% + 12.9% +12.8% +10.7%
N € 3267 +13.5% +5.8% + 7.9% + 4.2% + 14.1% + 6.0% + 24.3% + 13.6% +14.0% +11.7%

Consumption impacts

B-5 A-5 C-5 C-10 D-5

Consumption change relative to reference scenario [%]

Income
b € 1478 −0.56% − 0.35% − 0.58% − 0.98% − 0.59%
b € 2311 −1.41% − 0.92% − 1.47% − 2.83% − 1.47%
b € 3267 −1.46% − 0.95% − 1.52% − 2.98% − 1.52%
N € 3267 −1.95% − 1.28% − 2.03% − 3.96% − 2.03%

impaired groups lacking possibilities of modal shift due to labour force increases GDP in PPP terms by 2%. Feedback
settlement structures and income (referred to as captives). effects on the transport sector include a smaller overall
reduction in vehicle mileage of a mere 5.4% (instead of 6.5%
5.1. Revenue use earlier) relative to the reference level. There is also a feedback
on the direction of distributional impact. With the rise in
Testing for alternative uses of road pricing revenue, we find labour supply, the relative scarcity of capital increases, which
significant changes in impacts. First, to obtain greater social induces a capital price rise by 3.5%. As owners of capital are
redistribution, we only give household refunds (the total of primarily the rich households, under this revenue use option,
which remains constant at 1/3 of net road pricing revenue) to car road pricing no longer has a progressive distributional
below-median income households. This has only a marginal impact overall.
impact on vehicle mileage (and thus environmental impacts),
public transport passenger kilometres, and macroeconomic 5.2. Captives
indicators. But as the refund is in sum significant, distribution
of welfare impacts across income groups changes substan- Hammar and Jagers (in press), for the instrument of a fuel tax
tially. The consumption of the poorest household group now in Sweden, find that in general households’ perception is that
increases by 0.7%, that of the second-poorest declines by 0.7% car transport users should contribute their share in green-
modestly, while consumption of the richer households house gas emission reduction, and therefore fuel taxes to be
decreases by up to 2.3%, which is 0.4%-points more than in raised. However, they find this overall preference for justice to
B-5. It has to be noted, however, that the practical implemen- be dominated by self-interest for the sub-group of intensive
tation of such a strong redistribution may be difficult and car users. Our approach allows us to quantify how intensive
connected to substantial control costs in many countries. car users are hit by road pricing, and how these costs relate to
It is evident that another often discussed mode of redis- the average household costs.
tributive revenue recycling, setting road pricing charges for In our distributional analysis so far we distinguished be-
poor and peripheral households to zero, is not only of exten- tween four income groups. This allows us to draw general
sive administrative demand, but obviously eliminates the core conclusions concerning the direction of distributional impact.
idea of the instrument. Refunding for distributional matters However, within these income groups themselves, impacts
needs to be completely independent of car mileage driven, are also subject to a within-group distribution. When we want
with a basis for example solely in terms of income and/or to take a look at those households potentially hit hardest, we
residence location in order to sustain the intended incentives. have to further disaggregate within the income groups. The
Second, when – instead of the use options discussed so far – strongest impact will arise with those households that are
road pricing revenues (net of system costs) are used fully for simultaneously characterised by high mileage, low income,
reducing payroll fringe costs, we find a change in macroeco- and peripheral living location (with a low supply level of
nomic impacts relative to the earlier result of scenario B-5. alternative transport options). We select households that have
Road pricing revenues are sufficient to reduce payroll fringe below-median income, drive more than 15,000 km a year, and
costs by 4% of net wages, which – in the medium term – causes are sited in a peripheral region. We refer to them as “captives”,
unemployment to decline by 2%. Employment of a former idle and our database identifies this group as accounting for 1.8%
66 E CO L O G I CA L E CO N O MI CS 63 ( 20 0 7 ) 5 9–6 9

Table 5 – Impacts on intensive car use households in peripheral regions with below-median income (“captives”)
Scenario B-5 Sensitivity analysis considering captives

Car Public transport Overall consumption Car Public transport Overall consumption

Expenditures Expenditures

Income [change in %]
b € 1.478 19.3% 8.6% −0.6%
non-captives 19.5% 8.7% −0.5%
captives ⁎ 12.1% 3.6% −6.1%
b € 2.311 14.0% 6.1% −1.4%
non-captives 14.0% 6.2% −1.3%
captives ⁎ 12.8% 4.6% −3.6%
b € 3.267 12.4% 5.4% −1.5% 12.4% 5.4% −1.5%
N € 3.267 13.5% 5.8% −2.0% 13.5% 5.8% −2.0%

⁎ Captives are defined as households in peripheral regions with below-median income and an annual car mileage N 15,000 vehicle-km.

of Austrian households (0.3%-points of which are from the income group have low variable costs per kilometre, and
lowest income quartile, 1.5%-points from the second). In therefore their reduction in mileage is higher than for the
Table 5 we show the impact of scenario B-5 on these captives. medium income households. This makes both groups less
We find a significantly stronger impact on the households so vulnerable to the implementation of car road pricing than
defined within the respective income groups. For the poorest other income groups. On the other hand, pre-policy car
income quartile as a whole, the average consumption reduc- transport demand levels strongly increase with income. This
tion is 0.56%, while for captives within this quartile it is 6.1%. works clearly progressive in car road pricing implementation,
In the second-poorest income quartile the reduction is 3.6% and tends to dominate the former effect.
for captives (versus a quartile average of 1.4%). Investigating the burden on different income groups
should be a starting point to be further addressed by future
research. It may well be that many of the “really poor”
6. Conclusions households are categorised not within the lowest income
group (which surely includes many young single households,
In our analysis of car road pricing we explored the use of a students, single retired, etc.), but within the second lowest
transport oriented computable general equilibrium model for income group, which may include households with single
sustainability impact assessment. As in the literature so far income, but more household members.
the coverage of social impacts is rare, we put a main emphasis Also, we focus on the distribution of costs of car road
on quantifying distributional implications of car road pricing pricing policies only. While we quantify overall benefits, such
and potential trade-offs with the environmental and econom- as health benefits or benefits from reduced congestion, we do
ic objectives. not analyse their distribution across income groups. For a
Based on an in-depth analysis of the effects of environ- final evaluation of distributional impacts this would be
mental transport policies, West (2004) concludes in a – partial necessary. Of these health and reduced congestion benefits,
equilibrium – joint analysis of vehicle choice and mileage that a pre-analytic assumption could be that the former might
(i) demand elasticities of vehicle miles driven decline with more strongly benefit the poor (since they are less able to flee
income levels, and that thus (ii) distance-dependent policies, environmental burdens), the latter the rich (who both drive
such as a gasoline tax or road pricing, are regressive only more and have higher time costs).
across higher income households but progressive across low However, questions of fairness are not the only ones
income households. The higher degree of price-responsive- relevant for increasing the acceptance of road pricing prior to
ness of lowest income households produces this result. introduction. Obviously, one natural extension here is proper
We find, however, in a general equilibrium model, and thus communication of the purpose and effects of road pricing, i.e.
within a more integrative assessment acknowledging also road pricing should be anchored as a justifiable environmental
overall feedback effects, that road pricing has a progressive policy, not only as an instrument for revenue raising or as a
impact across all income ranges, at least if consumption based measure primarily addressing congestion, and it should be
welfare effects are chosen as an indicator for fairness. This discussed relative to other policy options (see, e.g. Odeck and
result is steered by two effects which work in opposite Bråthen, 1997; Oberholzer-Gee and Weck-Hannemann, 2002;
directions. On the one hand, the strongest percentage reduc- Hammar and Jagers, in press). Also, the design of compensation
tions in car kilometres can be observed for the lowest (and the is essential, in order to meliorate the effects on certain groups,
highest) income households: The lowest income household particularly on those identified as “road pricing victims” (or
group experiences a high reduction since they usually drive captives). As argued before, the characteristics used to identify
small cars (with relatively low variable costs per kilometre these groups need to be investigated more intensively (income
driven) and a price increase of 5 Euro-cent implies a doubling of and household location being only two of many possible
their variable costs (see Table 1). Households in the highest indicators).
 EC O LO G I CA L E C O N O M I CS 6 3 ( 2 00 7 ) 5 9–6 9 67

Furthermore, revenue spending can have a significant particularly to the way road infrastructure investments are
impact on the effects of road pricing. In particular, if households carried out. The larger the share that is devoted to road
are compensated in a lump-sum fashion for higher transport maintenance, which is more labour intensive, the stronger the
expenditures (as is assumed in policy scenarios A to D), small net overall increase in labour demand. If road infrastructure
welfare burdens for the poor come at the price of GDP investment, on the other hand, is assumed to follow the average
reductions and ambiguous employment impacts (whereas if production structure of the overall construction sector, employ-
revenues are spent such as to lower labour taxes, the effects of ment decreases in all scenarios. Model results presented in the
road pricing on GDP and employment are clearly positive). main text assume a 50% share for maintenance within road
Thus, we can confirm that revenue spending is an important infrastructure investment, since much of the secondary road
element in the appropriate design of a road pricing scheme, in network is of poor quality and requires maintenance.
addition to the “sales arguments” put forward by others
(Calthrop and Proost, 1998; Jakobssen et al., 2000; Oberholzer- A.2. Variables
Gee and Weck-Hannemann, 2002; Proost et al., 2002). Factor demand
Finally, in discussing the benefits of introducing nation- L total labour demand
wide road pricing, fuel taxes are often seen as a preferable K total capital demand
substitute. While fuel taxes are easier in administration, a road
Production
pricing scheme has mainly three advantages. First, it allows for
Xj gross production of sector j
regional and peak-time differentiation, possibly even in short- Kj capital input in sector j
term response to weather related emission dispersion condi- Lj labour input in sector j
tions. Second, while the range of potential fuel tax increments Hj factor aggregate in sector j
is limited due to potential tax avoidance caused by refilling in Aj, aij Leontief–input–output-coefficients in sector j
neighbouring countries (see Calthrop and Proost, 1998; or δ CES-distribution parameter in sector j
σj elasticity of substitution in production between
Ubbels et al., 2002, for a formal analysis; for an empirical
labour and capital in sector j
application, see De Borger et al., 2004), road pricing allows for
national independence to set (car) road pricing rates, by which Transport
also all users (domestic and foreign) within one's territory are Tp Private car passenger transport
charged. Third, the visibility of full user costs per kilometre T pf Private car passenger transport production fixed
travelled is higher for a permanent cost counter, like the road input
pricing on board unit (OBU), than for a fuel tax. T pv Private car passenger transport production variable
input (mileage dependent)
Tu Public passenger transport
Apf, Apv Leontief–input–output-coefficients in private car
Acknowledgments passenger transport
Akmp kilometre input-coefficient in private car passenger
We are thankful to Werner Gobiet, Georg Kriebernegg, Ines transport
Omann, to the participants of the annual meeting of the Aui Leontief–input–output-coefficients in public transport
“Ausschuss für Umwelt- und Ressourcenökonomie” of the kmp vehicle kilometres driven in private car transport

“Verein für Socialpolitik” in St. Gallen, the plenary session at

Foreign trade
the EcoMod conference “Input–Output and General Equilibri- EXj export of sector j
um” at Brussels and the workshop “The economic modelling Mj import of sector j
and integrated assessment of climate change” at the DIW Berlin, Pj production price of goods aggregate X in sector j
and to two referees of this journal for very helpful comments. PjW world market price of goods aggregate X in sector j
EXj0, Mj0 export and import quantities in sector j in the
reference year
εj foreign trade price elasticity of demand in sector j
Appendix A. Supplementary material
Labour market
A.1. Sensitivity analysis of transport elasticities of w nominal wage rate
P
substitution and transport sector investment wlow lower bound on the real wage rate
pp Paasche index of the aggregate price level
Cutting the elasticity of substitution between private car and u rate of unemployment

public transport in half implies hardly any difference in macro-

Consumption
economic impacts relative to those found in section 4 of the
Ch Total consumption of household type h
main text, but implies a significant change in public transport Xch Consumption of non-transport goods of household h
demand (which now declines). On the other hand, raising Th Transport consumption of household h
elasticity leads to a more pronounced impact on both private car δch CES-distribution parameter in consumption for
and public transport demand, and thus environmental impacts. household h
We further find that model results are not strongly sensitive δTh CES-distribution parameter in transport demand for
household h
to changing the elasticity of substitution between transport
δXh,i CES-distribution parameter in non-transport
and other goods in consumption; results hardly change at all. consumption for household h
In testing the sensitivity of labour market results with the
(continued ononnext
(continued page))
nextpage
model, we find that these react strongest to revenue use,
68 E CO L O G I CA L E CO N O MI CS 63 ( 20 0 7 ) 5 9–6 9

Appendix A.2 (continued ) De Borger, B., Courcelle, C., Swyson, D., 2004. Optimal pricing of
σC elasticity of substitution between transport and non- transport externalities in an international environment:
transport consumption some empirical results based on numerical optimization
σT elasticity of substitution between car transport and model. Regional Science and Urban Economics 34, 163–201.
public transport demand De Jong, G., Gunn, H., Walker, W., 2004. National and international
σX elasticity of substitution between non-transport freight transport models: an overview and ideas for future
goods in household h consumption development. Transport Reviews 24 (1), 103–124.
EC (European Commission), 2001. A sustainable Europe for a better
world: a European Union strategy for sustainable development.
Commission's Proposal to the Gothenburg European Council,
COM (2001)264 Final, Brussels, 15 May 2001.
A.3. List of core model equations
EC, 2006. Directive 2006/38/EC amending Directive 1999/62/EC on
Production the charging of heavy goods vehicles for the use of certain


Xj ¼ min Hj =Aj ; Xij =aij for i, j = 1,….35 (1) infrastructures. Brussels, 17 May 2006.
Federal Ministry of Agriculture, Forestry, Environment and Water
 rj =ðrj −1Þ
ðrj −1Þ=rj
 ðrj −1Þ=rj Management, unpublished manuscript. Umweltbilanz Ver-
Hj ¼ dj Lj þ 1−dj Kj for i, j = 1,….35 (2) kehr. Vienna.
Hammar, H., Jagers, S.C., in press. What is a fair CO2 tax increase?
Transport On fair emission reductions in the transport sector, Ecological
 
TP ¼ min Tpf =Apf ; Tpv =Apv (3) Economics.
Hensher, D.A., Button, K.J., 2000. Introduction. In: Hensher, D.A.,
 
Tpf ¼ min Xi =Apf i = 1,….35 (4) Button, K.J. (Eds.), Handbook of Transport Modelling. Perga-
mon, Amsterdam, pp. 1–10.
 
Tpv ¼ min Xi =Apv ; kmp =Akmp Herry, M., Sammer, G., 1999. Mobilitätserhebung österreichischer
i = 1,….35 (5)
Haushalte 1995, Rohdaten (Mobility Survey of Austrian
Tu ¼ minðXi =Au Þ i = 1,….35 (6) Households 1995). Vienna.
Herry, M., Sedlacek, N., 2003. Österreichische Wegekostenrech-
nung für die Straße 2000. Austrian Ministry of Transport,
Foreign trade
 ej
Innovation and Technology, Vienna.
EXj ¼ EX0j Pw j =Pj for j = 1,….35 (7) Jakobsson, C., Fujii, S., Gärling, T., 2000. Determinants of private car
 ej users’ acceptance of road pricing. Transport Policy 7, 153–158.
Mj ¼ M0j Pw j =Pj Jansen, H., Denis, C., 1999. A welfare cost assessment of various
for j = 1, …35 (8)
Labour market policy measures to reduce pollutant emissions from passenger
road vehicles. Transportation Research. Part D, Transport and
w P
z wlow ⊥u (9) Environment 4, 379–396.
pP
Johansson, B., Mattsson, L.-G. (Eds.), 1995. Road Pricing: Empirical
Household demand Assessment and Policy. Kluwer, Boston.
 rC =ðrC −1Þ Johansson-Stenman, O., 2002. Estimating individual driving dis-
ðrC −1Þ=rC
 ðrC −1Þ=rC
Ch ¼ dCh Xch þ 1−dCh Th for h = h1,.…h4 (10) tance by car and public transport use in Sweden. Applied
Economics 34, 959–967.
" #rX =ðrX −1Þ
X ðrX −1Þ=rX
 Just, R.E., Hueth, D.L., Schmitz, A., 2004. The Welfare Economics of
Xch ¼ dXh;i Xch;i Public Policy: A Practical Approach to Project and Policy
i
X  Evaluation. Edward Elgar, Cheltenham.
with dXh;i ¼ 1 for h = h1,.…h4 (11) Kletzan, D., Köppl, A., Kratena, K., Schleicher, S., Wüger, M., 2006.
i
Household demand Towards sustainable consumption: economic modelling of
!rT =ðrT −1Þ mobility and heating for Austria. Ecological Economics 57,
p ðrT −1Þ=rT
 uðrT −1Þ=rT for h = h1,…h4 (12) 608–626.
Th ¼ dTh Th þ 1−dTh Th
Kriebernegg, G., unpublished manuscript. Austrian transport
demand model for car road pricing analysis. Technical
University of Graz, Institute of Road and Transport Systems.
Lindsey, R., Verhoef, E., 2001. Traffic congestion and congestion
REFERENCES pricing. In: Button, K., Kenneth, Hensher, D.A. (Eds.), Handbook
of Transport Systems and Traffic Control. Pergamon, Amster-
dam, pp. 77–105.
Berechman, J., 2002. Transport investment and economic devel- May, A.D., Liu, R., Shepherd, S.P., Sumalee, A., 2002. The impact of
opment: is there a link? In: ECMT (Ed.), Transport and Economic cordon design on the performance of road pricing schemes.
Development. Round Table, 119. ECMT, Paris, pp. 103–138. Transport Policy 9, 209–220.
Böhringer, C., Löschel, A., 2006. Computable General Equilibrium Mayeres, I., Proost, S., 2001. Marginal tax reform, externalities and
Models for Sustainability Impact Assessment: Status quo and income distribution. Journal of Public Economics 79, 343–363.
Prospects. Ecological Economics 60 (1), 49–64. Mayerers, I., Ochelen, S., Proost, S., 1996. The marginal external
Button, K.J., Verhoef, E.T. (Eds.), 1998. Road Pricing, Traffic costs of urban transport. Transportation Research. Part D,
Congestion and the Environment: Issues of Efficiency and Transport and Environment 1 (2), 111–130.
Social Acceptability. Edward Elgar, Cheltenham. Nash, C., Sansom, T., Still, B., 2001. Modifying transport prices to
Calthrop, E., Proost, S., 1998. Road transport externalitities. internalise externalities: evidence from European case studies.
Interaction between theory and praxis. Environmental and Regional and Urban Economics 31, 413–431.
Resource Economics 11 (3–4), 335–348. Oberholzer-Gee, F., Weck-Hannemann, H., 2002. Pricing road use:
Dargay, J.M., 2001. The effect of income on car ownership: evidence politico-economic and fairness considerations. Transportation
of asymmetry. Transportation Research. Part A 35, 807–821. Research. Part D, Transport and Environment 7, 357–371.
 EC O LO G I CA L E C O N O M I CS 6 3 ( 2 00 7 ) 5 9–6 9 69

Odeck, J., Bråthen, S., 1997. On public attitudes toward imple- Steininger, K., 2003. The foreign trade and sectoral impact of truck
mentation of toll roads—the case of the Oslo toll ring. road pricing for cross-border trade: a CGE analysis for a small
Transport Policy 4 (2), 73–83. open economy. Environmental and Resource Economics 23 (2),
Parry, I.W.H., Bento, A., 2002. Estimating the welfare effect of 213–253.
congestion taxes: the critical importance of other distortions Steininger, K.W., Gobiet, W., Binder, C., Friedl, B., Gebetsroither, B.,
within the transport system. Journal of Urban Economics 51, Niederl, A., Omann, I., Seebauer, S., 2005. Technologien und
339–365. Wirkungen von Pkw-Road Pricing im Vergleich. Report No 1-
Proost, S., Van Dender, K., 2001. The welfare impacts of 2005, Wegener Center for Climate and Global Change. Univer-
alternative policies to address atmospheric pollution in urban sity of Graz, Graz.
road transport. Regional Science and Urban Economics 31, Sterner, T., 2002. Environmental road pricing. In: Sterner, T. Policy
383 – 411. Instruments for Environmental and Natural Resource Man-
Proost, S., Van Dender, K., Courcelle, C., De Borger, B., Persion, J., agement. RFF Press, Washington, pp. 228–238.
Sharp, D., Vickerman, R., Gibbons, E., O'Mahony, M., Heaney, Q., Ubbels, B., Rietveld, P., Peeters, P., 2002. Environmental effects of a
Van den Bergh, J., Verhoef, E., 2002. How large is the gap kilometre charge in road transport: an investigation for the
between present and efficient transport prices in Europe? Netherlands. Transportation Research. Part D, Transport and
Transport Policy 9, 41–57. Environment 7, 255–264.
Pucher, J., Renne, J.L., 2003. Socioeconomics of urban travel: Viegas, J.M., 2003. Transport policy development in Europe 1950–
evidence from the 2001 NHTS. Transportation Quarterly 57 (3), 2020. In: ECMT (Ed.), Fifty Years of Transport Policy. ECMT,
49–77. Paris.
Santos, G. (Ed.), 2004. Road pricing: theory and practice. Special Weightman, G., Humphries, S., 1985. The Making of Modern
issue. Research in Transportation Economics 9, 3–308. London: 1815–1914. Sidgwick & Jackson, London.
ST.AT, 2002. Vebrauchsausgaben 1999/2000 — Sozialstatistische West, S.E., 2004. Distributional effects of alternative vehicle
Ergebnisse der Konsumerhebung (Austrian Consumer Expen- pollution control policies. Journal of Public Economics 88,
ditures 1999–2000). Statistik Austria (ST.AT), Vienna. 735–757.