as a PDF

Distributional and Economic Impacts of Nationwide Car Road
Pricing: A CGE Analysis for Austria
Birgit Friedl*
Karl Steininger*
This version: April 9, 2004
ABSTRACT : Nationwide car road pricing schemes are discussed across Europe,
partly as extensions of the truck road pricing schemes already in existence. This
paper analyses the economic and environmental feedbacks of such nationwide car
road pricing schemes on gross production, sectoral structure, employment,
consumption and CO2 emissions by means of a passenger transport demand
oriented CGE model. As a main focus the distributional impacts across household
income groups are explored. It has been argued that the poor households would
have to take an unproportional share of the burden. We give a detailed analysis for
the case of Austria and show that, while low income households do experience a
reduction in welfare, road pricing does have a progressive effect, with a
significantly stronger welfare reduction for high income households.
Keywords: Road Pricing, Private Transport, Transport Emissions, Passenger
Transport Policy, Income Distribution.
JEL: C68, D58, H23, R48.
*
University of Graz, Department of Economics, Universitaetsstr. 15, A-8010 Graz, Austria
phone: +43 316 380- 3451, fax: +43 316 380 9520
e-mail: [email protected], [email protected]
We are thankful to Werner Gobiet, Georg Kriebernegg and Ines Omann for helpful comments.
-2-
1. Introduction
For trucks kilometre based road pricing systems have been introduced in Europe at various
levels of sophistication. From section-charging at highways in various countries, to electronic
charging at Austrian highways (2004), up to full road network charging in Switzerland
(2001). For private cars, charging in Europe has been introduced section-wise at highways on
one hand and for urban areas on the other hand, the latter mainly in the form of toll rings, e.g.
in Scandinavian countries and more recently in London. The discussion of nationwide
kilometre based charging also for private cars has taken off, however, both at the level of
various nations and at the European Commission.
What are the economic impacts of nationwide car road pricing in terms of gross production,
sectoral structure, employment, and consumption? Can we expect a significant reduction in
greenhouse gas emissions, notably of CO2 ? Who takes the primary direct burden of such
policy: is it poor or rich households that are hit harder? These are the core issues that this
paper explores in a detailed analysis for the empirical account of one European country,
Austria. A passenger transport demand focused computable general equilibrium (CGE) model
is developed for this purpose.
The present model goes beyond existing ones in several respects. First, and in some way
parallel to earlier transport policy discussion, most of the models used so far to analyse road
pricing are either limited to freight transport (e.g. De Jong et al., 2004; Steininger, 2003) or
they target at urban road pricing or toll ring pricing only (e.g. Proost and van Dender, 2001;
Mayeres et al., 1996). We focus on nationwide car road pricing. Second, while an increasing
number of papers addresses the welfare effects of a tax suitable for internalising external
transport costs (e.g. Nash et al., 2001; Jansen and Denis 1999) distributional aspects have
hardly been considered in economic transport policy models. However, in the socio-economic
literature on road pricing it is questions of distribution and equity that have been identified as
the key obstacles to the successful implementation of a road pricing system. To better
understand the distributional effects, i.e. the difference of impacts across household income
groups, we distinguish four classes of income in our model. Third, most of transport CGE
analysis is drawing from literature surveys on elasticities in transport demand. We proceed in
a different mode here. We confront the economic CGE model developed below with a pure
heuristic choice switch passenger transport demand model (switches are based on trip
characteristics purpose, distance, mode and type of region) and calibrate elasticities of
-3substituion in household demand in the CGE model using the results of this transport demand
model.
The paper proceeds in four steps as follows. In section 2 the influence of household income
on transport demand is considered, preparing for consideration of distributional impacts of a
change in (passenger) transport prices. In section 3 the small open economy passenger
transport CGE model is developed distinguishing four income groups of households. Next, in
section 4 transport and consumption data bases are merged, and a Social Accounting matrix is
constructed differentiating sufficiently the elements of private car transport. Section 5 reports
on the results when the CGE model is used for the simulation of nationwide car road pricing
in various implementation schemes in order to conclude on the economic, environmental and
distributional impacts of this policy. A final section summarises the main conclusions.
2. Transport Demand and Income Classes
While the optimal pricing problem in the transport sector has been successfully implemented
in a vast body of literature (e.g. Mayeres et al., 1996; Lindsey and Verhoef, 2001) it has been
recognized only recently that questions of distribution and equity are of major importance for
the acceptance of road pricing (see also Mayeres and Proost, 2001). The arguments can be
grouped basically along two lines, namely questions of differences in access to transport
supply (and consequently demand) already in the status quo (originating primarily from
differences in income and location), and questions of equity in outcome as a consequence of
the introduction of a road pricing scheme.
Differences in passenger transport demand in the status quo stem from ma ny factors such as
differences in income, in location (home relative to work relative to locations of other
activities) or access to public transport. For instance, in Austria the fraction of transport
expenditures to income is 15% on average, but ranges from 9.6% for the low income
household quarter to 18.6% for high income households (ST.AT 2002; see also Table 1).
Furthermore, differences in transport demand stem from the availability of certain modes of
transport, in particular the non-availability of public transport in certain periphery regions. For
instance, in rural areas at least 50% of the households do not account any expenditures on
public transport while in urban areas many households do not even own a car (ibid).
With respect to equity, several aspects are worth noting. Viegas (2001) distinguishes between
horizontal equity or equality of opportunities in terms of comparable conditions for people
living in different parts of a country, vertical equity as a safeguard concept for those in the
worst conditions, as well as longitudinal equity as equity across generations. For the case of
-4car road pricing implementation, the main question is the incidence of road pricing, related to
both horizontal and vertical equity. Who are the groups hurt most / least and how can the
burdens be shared equally? While the research we report here has led us to prepare the data
basis for the analysis of both horizontal and vertical equity, in this paper we focus on the latter
only.
As becomes clear from the empirical examples given above, transport costs do not only differ
considerably across modes (car vs. public vs. bike vs. walking) but also across individuals.
For instance, a low income household is less likely to have a large car than a high income
household and therefore his or her fixed costs of car transport are lower. Alike variable car
costs differ due to divergent fuel efficiencies (see Table 1). In sum, the implicit price per
kilometre driven varies with income and so does the impact of road pricing.
A different strand of literature confines itself with the question of adjustment in travel demand
due to changes in transport policy (for a survey, see Goodwin, 2003). According to Jakobsson
et al. (2000), income constraints can be identified as the key determinant for consumers to
reduce their demand for transport when travel costs increase. Given this is true, the burden of
road pricing should compound for the poor rather than the rich. Put differently, this could
imply that road pricing has an regressive effect. The present paper therefore tries to verify
whether this is the case empirically.
Table 1: Transport Expenditures as % of Monthly Household Income across Income Groups, Austria
2000
monthly earnings per
household
car exp.; fixed costs
car exp.; variable costs
public transport exp.
total transport exp.
income group
1
2
3
4
less than € 1.478 less than € 2.311 less than € 3.267 more than € 3.267
5.97
2.49
1.13
9.58
11.00
3.71
0.76
15.47
12.39
3.84
0.63
16.86
14.57
3.58
0.47
18.61
Source: ST.AT (2002), own calculations
3. Transport Demand CGE Model
The computable general equilibrium model developed in this section to analyze the impacts of
car road pricing thereafter is a standard small open economy CGE model in many respects,
obviously with a detailed passenger transport modelling. We give a descriptive account of the
model here, while the core model equations are listed in the Appendix. The Appendix also
includes the list of variables as a reference for the reader.
-5Production of non-passenger-transport goods distinguishes 35 sectors, and follows a nested
CES structure, with capital and labour as primary inputs, and intermediate inputs entering in a
Leontief fashion. The production of the various passenger transport intermediate and final
consumption goods in its model structure follows the purpose of analysis as well as data
availability and is given therefore in the following section 4 on data (and in the Appendix –
Eq. (3) to (6)).
Foreign trade is modelled under the Armington assumption of product differentiation as stated
in Eq. (7) and (8). 1
As the analysis of the labour market impact of transport policy is an issue, the labour market
does not clear, but unemployment is driven by classical, i.e. minimum wage, unemployment
(Eq. 9).
Household demand is governed by a nested CES structure, with unity elastictity of
substitution among non-passenger-transport goods (i.e. constant expenditure shares for nonpassenger-transport goods), and calibrated elasticities of substitution between the aggregates
of non-transport and transport goods as well as among different passenger transport goods.
The structure of the latter is given in detail below in Section 4, in Figure 1, and in Equations
(10) to (12).
We distinguish four representative households (reflecting income quarters), substitution
elasticities are taken uniform across household types.
4. Data and Calibration
For Austria there was no data basis combining information on passenger trips (by purpose,
mode, distance, and frequency) and income. By econometrically merging the Austrian
mobility survey (Herry and Sammer, 1999), the Environmental Balance of Transport (Federal
Ministry, 1997, 2003) and the Austrian consumption expenditure survey (ST.AT 2002), we
provide such a data basis, also differentiated by four classes of household income, by three
region types (urban, central district, and peripheral), and adjusted for the different base years
of the respective data bases. The challenge here was to have disaggregated data on both
transport expenditures and on transport demand in terms of quantity (passenger and vehicle
kilometres and mode per year), distinguished by household income levels and region types.
The detailed regression approach and the full data base is given in Gebetsroither et al. (2004).
We find that the richest income quarter households drive their cars more than fourfold than
1
Equation references throughout this paper refer to the Appendix.
-6the poorest. Across regions car verhicle kilometers are high in both central and peripheral
districts.
Preparing for the CGE analysis we disaggregate a standard Social Accounting Matrix for
Austria, constructed for the year 2000, to differentiate household expenditure for public
transport and for private car use (distinguishing for the latter the various elements of fixed
respectively variable costs, linking all cost components to respective sectors of supply or
government budget revenues, such as car and gasoline taxes).
Thus we isolate a newly combined passenger transport good of final household consumption,
which in regular final demand statistics is dispersed across various sector.
The sectors contributing to this passenger transport good Tp are refined oil, transport
equipment, distribution, finance and insurance, and inland transport. This contribution is
differentiated across modes, variable (i.e. kilometre dependent) and fixed cost components,
and household income groups.
Figure 1: Structure of Household Final Demand for income class h
Consumption Ch of household h
0.17
consumption of non-transport goods i, Xh C
passenger transport demand Th
0.4
public
transport Th u
car Th p
0
fixed costs Tpf
variable costs Tpv
Note: Non-zero values denote calibrated elasticities of substitution
The Social Accounting is thus augmented by four categories of households who differ in their
transport expenditures. For all other consumption expenditures, we assume that the structure
is identical for all household income groups, differing only in absolute levels. This
simplification is justified since on aggregate level, only two consumption aggregates differ
considerably across different groups of income, namely expenditures on living and on
-7transport2 , with only the latter being the focus of the policy analysed in this paper. Figure 1
gives the structure of household demand.
The parameter values for elasticities of substitution were calibrated from a choice switch
passenger transport demand model (Kriebernegg, 2004). In this model, the following
individual reaction patterns to car road pricing were distinguished: (a) reactions within the
demand structure for car passenger transport (change of route, change of travel time, car
pooling, merging more purposes per trip, change of destination, and trip elimination) and (b)
reactions in demand for public and other transport (modal shift from car to public transport or
to cycling or walking; but also newly arising transport demand due to a more favourable
environment for non-car trips). Based on this model transport projections for one road pricing
policy scenario (modest road pricing tax charged on the total road network) the following
parameters were calibrated within the CGE model described in section 3: elasticities of
substitution between car and public transport, and between total passenger transport and
demand for non-transport goods. Elasticities were taken to be uniform across household
income groups at this stage of the research.
The resulting elasticities of substitution (see Figure 1 for their values) are quite low which is
the result of a cautious transport demand model (lower bound transport demand change).
Thus, the CGE simulations refer to the – in terms of economic and distributional impacts “strongest impact”. If easier substitution between transport and non-transport goods and/or
between car and public transport were possible, negative impacts on economic welfare as well
as on output would be smaller. On the other hand we will focus on road pricing revenue use
below. It is evident, that with easier substitution revenues ceteris paribus turn out lower, as
will economic and distributional effects achieved by their use.
As is seen in Figure 1, fixed and variable cost components of car passenger transport combine
in a Leontief fashion to private car transport Tp , i.e. elasticity of substitution is exogenously
set at zero. Basically, kilometer charges (as the most important variable cost element in the
current analysis) cannot be substituted by fixed cost technical equipment or the like.
2
Expenditures on living, heating and lighting (relative to total consumption expenditures) vary from 30.4% for
the lowest income group to 20.6% for the highest; an opposite trend can be observerd for transport: expenditures
from transport decrease from 8.5% for low income households to 17.5% for high income households (ST.AT
2002).
-8-
5. Road Pricing Policy Simulations
In road pricing 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. Road pricing
revenues are first used for the coverage of system costs; a share of 15% is assumed to be
devoted for that end. For the remaining road pricing revenue use we define the following: (i)
for road infrastructure investment, public transport improvement, and household refund (a
third each) and, alternatively, (ii) a much higher share used for public transport improvement
(5/9), household refund as before, and only the remaining for road infrastructure investment.
Table 2: Policy scenarios
scenario label
network charged
time differentiation
charging level
A-5
B-5
urban: full network
rest of Austria:
full network
primary road network
none
none
5 Cent/km
5 Cent/km
C-5
C-10
D-5
full network
full network
full network
7-9a.m. and 7-9a.m. and
4-6p.m.
4-6p.m.
+100%
+100%
5 Cent/km 10 Cent/km
7-9a.m. and 4-6p.m.
+100%
5 Cent/km
1/9 road infrastructure,
1/3 each: road infrastructure, public transport, household refund
revenue use
5/9 public transport,
1/3 household refund
Note: The "primary road network" includes highways & "federal" long distance roads (Landesstrassen B)
5-Cent scenario
We will first analyse the impacts of road pricing for one policy scenario in more detail, before
we report on the merits of the five policy scenarios relative to each other.
Implementing car road pricing on the road network nationwide at a level of 5 Cent/km,
without time differentiation and applying the straight forward third-each revenue use rule
(policy scenario B-5), we find the economic impacts as stated in Table 3. In a comparative
static analysis we “shock” the economy in the base year 2000.
-9-
Table 3: Transport and macroeconomic results for policy scenario B-5
Reference level
(year 2000)
Policy Scenario B-5
change
absolute
absolute
in%
TRANSPORT VARIABLES
road pricing rate
revenues from road pricing (mn Euro)
car vehicle kilometers (mn veh-km)
public transport (mn passenger-km)
CO2 emissions pass. transport (1000 t)
0
0
0.05 €/km
3,007
60,744
21,613
11,568
60,148
22,362
-4.63%
+3.46%
-533
MACROECONOMIC VARIABLES
GDP (mn Euro)
GDP in PPP (mn Euro)
number of employees
unemployment rate (national def.)
price of capital
204,616
204,616
+1.42%
-0.29%
+3,636
5.84%
5.73%
+0.07%
BUDGETARY EFFECTS (mn Euro)
due to change in revenues from direct taxes
due to change in revenues from indirect taxes
+91
-299
due to change in labor market expenditures
change in government demand
+36
-359
revenues from road pricing (semi-public)
1,704
A reduction of 4.6% in car vehicle kilometres (some 600 million veh-km in absolute terms) is
observed, with a simultaneous increase in public transport passenger kilometres by 3.5%. As
we use the foreign price level as numeraire, GDP increases. A new service, the environment,
is now paid for, which raises the overall price level relative to abroad. Therefore we need to
analyse GDP in purchasing power parity terms, where it declines (by 0.3%). While this
decrease in physical production lowers indirect tax revenues and exerts a declining pressure
on employment, the latter is outweighed by the employment increase due to a sectoral shift in
production. We will closely look at the sensitivity of the labour market effect below. In total,
public revenues decline, but lost public tax revenues, are more than compensated by the
(semi-public) net revenues from car road pricing (which account for 1.7 billion Euro in this
scenario).
- 10 Table 4: Transport and welfare results across income groups for policy scenario B-5
less than € 1.478 less than € 2.311 less than € 3.267 more than € 3.267
TRANSPORT VARIABLES (levels)
car vehicle kilometers (mn veh-km)
public transport (mn passenger-km)
car expenditures
public transport expenditures
TRANSPORT VARIABLES (% change)
car vehicle kilometers (mn veh-km)
public transport (mn passenger-km)
car transport expenditures
public transport expenditures
CHANGE IN WELFARE
5,909
4,386
1,298
148
12,012
4,217
3,386
155
15,636
5,737
4,835
168
26,591
7,999
7,513
173
-5.16%
+5.14%
+22.85%
+5.25%
-4.47%
+3.27%
+16.19%
+3.37%
-4.17%
+2.82%
+14.38%
+2.92%
-4.85%
+2.84%
+15.67%
+2.94%
-0.56%
-1.41%
-1.45%
-1.94%
We find a quite differentiated car road pricing impact across household income groups (see
Table 4). The relative reduction in vehicle kilometres is highest for the poorest and the richest
households. This is parallel to the lowest pre-policy variable costs per kilometre, that these
two groups experience, albeit out of different reasons. For the poor households it is small cars,
for the rich it is newer and more expensive (and therefore efficient) cars that reduce pre-policy
variable costs below the economy-wide average, while larger family cars primarily show up
in the middle income range.
Due to the introduction of road pricing, private car transport expenditure rises by far most for
the poor, increasing by almost a quarter. Nevertheless, the original pre-policy vehicle
kilometres level of this group is so low, that overall welfare reduction (Hicksian welfare
index) is smallest for this group.
Public transport increases degressively across income groups, due to the pre-policy strong
decline – with rising income – in the share of public transport in overall mobility.
Impacts across road pricing policy scenarios
When we also simulate the other policy scenarios in this comparative static analysis, we get
the results as reported in Table 5. Most significant impacts arise, when the charging level is
raised to 10 Cent per kilometre (scenario C-10): road pricing revenues rise to a level of 6
billion Euro, with 3.4 billion net of system costs and household refunds, nationwide car
vehicle kilometre reduction comes close to 9%, transport CO2 emissions decrease by 1 million
tons. GDP in purchasing power parity terms declines by 0.6%.
In scenario D-5 we acknowledge the fact that a larger share of revenues is used to foster
public transport (i.e. service improvement) in raising the elasticity of substitution between
private and public transport to 0.6. Public transport rise thus is significant.
- 11 Table 5: Transport and macroeconomic results across road pricing policy scenarios
Reference
level
Policy
Scenario B-5
Policy
Scenario A-5
Policy
Policy
Policy
Scenario C-5 Scenario C-10 Scenario D-5
0
0.05 €/km
nationwide
no
1:1:1
0.05 €/km
highways
no
1:1:1
0.05 €/km
nationwide
yes (+100%)
1:1:1
0.05 €/km
nationwide
yes (+100%)
1:1:1
0
3,007
1,956
3,180
6,092
0.05 €/km
nationwide
yes (+100%)
higher share
public
transport
3,173
60,744
21,613
11,568
-4.64%
+3.50%
-533
-3.07%
+2.30%
-354
-4.88%
+3.64%
-562
-8.87%
+6.59%
-1,022
-5.06%
+7.91%
-556
204,616
204,616
+1.42%
-0.29%
+3,636
5.73%
+0.07%
+0.92%
-0.22%
+2,233
5.77%
+0.04%
+1.50%
-0.31%
+3,878
5.72%
+0.07%
+2.90%
-0.59%
+8,460
5.59%
+0.14%
+1.43%
-0.40%
+2,521
5.76%
+0.09%
+91
-299
+36
-359
+1,704
+57
-195
+22
-238
+1,108
+97
-316
+38
-378
+1,802
+203
-601
+84
-684
+3,452
+88
-471
+25
-691
+1,798
(year 2000)
TRANSPORT VARIABLES
road pricing rate
network where road pricing is applied
higher rate in peak periods (7-9, 16-18)
revenue use (beyond system costs) for
road infrastructure, public transport,
household refund
revenues from road pricing (mn Euro)
car vehicle kilometers (mn veh-km)
public transport (mn passenger-km)
CO2 emissions pass. transport (1000 t)
MACROECONOMIC VARIABLES
GDP (mn Euro)
GDP in PPP (mn Euro)
number of employees
unemployment rate (national def.)
price of capital
5.84%
BUDGETARY EFFECTS (mn Euro)
due to Change in revenues from direct taxes
due to Change in revenues from indirect taxes
due to Change in labor market expenditures
change in government demand
revenues from road pricing (semi-public)
Analysing the results across household income groups (see Table 6), we find a rise in car
transport cost by up to almost 50% for the poorest group (C-10), but also by up to some 30%
for the other income groups. Variable costs of car transport in scenario C-10 roughly triple
(with a higher factor for the richest and poorest, a lower one for the other groups).
Table 6: Trans port expenditure impacts across road pricing policy scenarios and household income
groups
Scenario B-5
car
public
expenditures
income
< € 1.478
< € 2.311
< € 3.267
> € 3.267
+22.85%
+16.19%
+14.38%
+15.67%
+5.25%
+3.37%
+2.92%
+2.94%
Scenario A-5
car
public
expenditures
+14.76%
+10.47%
+9.29%
+10.14%
+3.48%
+2.24%
+1.94%
+1.96%
Scenario C-5
car
public
expenditures
+24.18%
+17.14%
+15.22%
+16.58%
+5.53%
+3.56%
+3.08%
+3.10%
Scenario C-10
car
public
expenditures
+47.17% +10.07%
+33.35% +6.46%
+29.65% +5.62%
+32.12% +5.54%
Scenario D-5
car
public
expenditures
+23.43%
+16.91%
+15.07%
+16.46%
+10.74%
+7.70%
+6.85%
+7.31%
- 12 Table 7: Welfare impacts across road pricing policy scenarios and household income groups
B-5
A-5
C-5
C-10
D-5
welfare change relative to reference scenario
income
< € 1.478
< € 2.311
< € 3.267
> € 3.267
-0.56%
-1.41%
-1.45%
-1.94%
-0.36%
-0.91%
-0.93%
-1.26%
-0.60%
-1.49%
-1.53%
-2.05%
-1.20%
-2.91%
-2.99%
-3.96%
-0.60%
-1.50%
-1.54%
-2.07%
Welfare reduction impacts remain progressive with rising road pricing charge level across
households (Table 7). Even though total car transport expenditures in scenario C-10 rise by a
third, welfare decline is confined below 4% for the rich, and even lower for the other income
groups. In this welfare measure no account is taken for the welfare benefit of an improved
environmental situation, which most likely benefits the poor overproportionally, and thus
would enhance the progressive impact of car road pricing.
With respect to transport volume, the stronger car reduction for the poorest and richest holds
across policy scenarios. The divergence across household income groups, however, is at most
1.5% (policy C-10, richest to second richest).
Overall, we thus find GDP in terms of purchasing power parity slightly declining, and
environmental benefits strongly increasing with the scope of application (i.e. highways only
versus all road network) and road pricing rate. Further we find pure economic welfare
declining, but - contrary to public discussion - with a significantly stronger impact on rich
households. This progressive impact of car road pricing occurs because of both poor
households spending a smaller share of their income on transport pre-policy and those
households larger use of public transport pre-policy. Both diminishes their relative burden of
car road pricing, and the latter also eases their modal switch to public transport.
Let us finally consider the crucial question how sensitive these results are to a change in
model parameters. Sensitivity analysis has to be carried out primarily with respect to three
areas:
-
elasticity of substitution between priva te car and public transport
-
elasticity of substitution between passenger transport and other goods
-
use of road pricing revenues.
In the following we report the results of sensitivity analysis using one road pricing scenario
only, B-5, for matters of representation.
- 13 Table 8: Sensitivity analysis for road pricing policy scenario B -5, transport and macroeconomic results
sensitivity analysis
reference
scenario
low subst.
strong
transport social reother goods distribution
scenario
B-5
low subst.
car-public
high subst.
car-public
0.4
0.17
1:1:1:1
0.2
0.17
1:1:1:1
0.8
0.17
1:1:1:1
0.4
0.13
1:1:1:1
0.4
0.17
1:1:0:0
0
60,744
21,613
11,568
3,007
-4.63%
+3.46%
-533
3,012
-4.47%
-0.45%
-540
2,997
-4.97%
+11.71%
-520
3,028
-4.47%
-0.45%
-540
3,007
-4.66%
+3.70%
-535
204,616
204,616
+1.42%
-0.29%
+3,636
5.73%
+0.07%
+1.43%
-0.28%
+4,211
5.71%
+0.07%
+1.40%
-0.32%
+2,423
5.77%
+0.07%
+1.45%
-0.27%
+4,839
5.69%
+0.07%
+1.42%
-0.29%
+3,511
5.73%
+0.07%
-359
+1,704
-334
+1,707
-410
+1,698
-320
+1,716
-362
+1,704
(year 2000)
PARAMETER VALUES
elastiicty of substitution car - public transport
elasticity of substitution transport - other consumption goods
refund to households (share of income groups h1:h2:h3:h4)
TRANSPORT VARIABLES
road pricing revenues (mn Euro)
car vehicle kilometers (mn veh-km)
public transport (mn passenger-km)
CO2 emissions pass. transport (1000 t)
MACROECONOMIC VARIABLES
GDP (mn Euro)
GDP in PPP (mn Euro)
number of employees
unemployment rate (national def.)
price of capital
5.84%
BUDGETARY EFFECTS (mn Euro)
change in government demand
revenues from road pricing (semi-public)
Results are reported in Table 8. We find, that cutting elasticity of substitution between private
car and public transport in half implies hardly any difference in macroeconomic impacts
relative to those found earlier, but implies a significant change in public transport demand
(which now declines). Setting this elasticity to a higher value does have a more pronounced
impact on both private car and public transport demand.
We further find, that model results are not strongly sensitive to changing the elasticity of
substitution between transport and other goods in consumption; results hardly change at all.
When using household refunds of road pricing revenues for distributional purposes (see Table
9 and last column in Table 8), we find for the example of refunding only to households below
median income, that significant changes in impact distribution of road pricing can be
achieved.
Table 9: Sensitivity analysis for road pricing policy scenario B -5, welfare impacts across household
income groups
scenario B-5
public
car
transport
expenditures
income
< € 1.478
< € 2.311
< € 3.267
> € 3.267
+22.85%
+16.19%
+14.38%
+15.67%
+5.25%
+3.37%
+2.92%
+2.94%
welfare
change
-0.56%
-1.41%
-1.45%
-1.94%
sensitivity analysis
strong social redistribution
public
car
transport
welfare
expenditures
change
+24.38%
+17.06%
+13.77%
+15.24%
+6.56%
+4.15%
+2.37%
+2.56%
+0.68%
-0.67%
-1.97%
-2.30%
- 14 In testing the sensitivity of labour market results with the model, we find that these react
strongest to revenue use, most importantly to the way road infrastructure investments are
carried out. The larger the share that is devoted to road maintenance, which is more labour
intensive, the stronger the net overall increase in labour demand. If road infrastructure
investment, on the other hand, is such that it meets the average production structure of the
overall construction sector, overall employment decreases in all scenarios reported above.
Taking account of the poor quality of much of the secondary road network (managed by local
and provincial authorities), it seems likely that a significant share of revenues is devoted to
their repair. Model results presented above use a 50% share for maintenance within road
infrastructure investment.
6. Conclusions
The highest impact in terms of percentage reductions in car kilometres can be observed for the
lowest and the highest income group. The lowest income group experiences a high reduction
since they usually drive small cars (quite low variable costs per kilometre driven) and an price
increase by € 0.05 implies a doubling of their variable costs. On the other hand, also
households in the highest income group have low variable costs per kilometre, but high total
costs (because they drive a lot), and therefore their reduction in mileage is higher than for the
medium income households.
The welfare impact across household income groups is first driven by the somewhat different
relative vehicle kilometre reduction. It is dominated, however, by the second determinant,
which is pre-policy car transport demand levels. In the pre-policy status quo these rise so
over-proportionally with income, that car road pricing has a progressive incidence.
While the current analysis expands transport CGE analysis beyond so far available research
results along the three lines stated in the introduction, there are at least two further issues of
development for future analysis, both leading to more regionally differentiated modelling.
First, in the policy scenarios we included time differentiation in urban areas, while the impact
on the national level is nevertheless hardly visible. Congestion in agglomeration areas,
obviously, nevertheless declines significantly. As a fraction of nationwide vehicle kilometres,
the kilometres driven in urban areas account for 18%, of which one third (and therefore only
6% of national vehicle kilometres) occur during the four hour peak time as specified.
Therefore we hardly see the congestion pricing impact at national level results, but would do
so in a regionally differentiated analysis for urban areas.
- 15 Second, the change in transport demand in the transport demand model occurs mainly in
urban areas, and less for longer distance trips. Again this reflects the “upper bound
expenditure impact” characteristic of the current analysis. The impact on overall national
transport demand is thus quite low, but much higher in urban areas (where people switch from
driving their cars to public transport, walking, cycling). A sensitivity analysis with the
transport demand model such that public transport distances increase by 50% as a
consequence of car road pricing shows, however, that the impact on further vehicle kilometre
reduction by this increased modal shift is only 0.5% at the national level (for scenario B-5, in
this case). Also here, a more detailed regional account of impacts can give conclusions that
are more regionally differentiated.
- 16 References
De Jong, G., H. Gunn, W. Walker (2004), National and international freight transport models:
An overview and ideas for future development, Transport Reviews 24 (1): 103-124.
Federal Ministry of Agriculture, Forestry, Environment and Water Management (1997, 2003),
Umweltbilanz Verkehr, Wien.
Gebetsroither, Brigitte, Jürgen Weissenbacher, Birgit Friedl, Karl Steininger (2004), Austrian
Passenger Transport by Income Groups, mimeo, University of Graz, Department of
Economics.
Goodwin, Phil (2003), How easy is it to change behaviour?, in: ECMT, Fifty Years of
Transport Policy, ECMT, Paris: 49-73.
Herry, Max and Gerd Sammer (1999), Mobilitätserhebung österreichischer Haushalte 1995,
Rohdaten (Mobility Survey of Austrian Households 1995), Wie n.
Jakobsson, C., S. Fujii and T. Gärling (2000), Determinants of private car users’ acceptance
of road pricing, Transport Policy 7, 153-158.
Jansen, Heinz and Cécile Denis (1999), A welfare cost assessment of various policy measures
to reduce pollutant emissions from passenger road vehicles, Transportation Research
D 4: 379-396
Kriebernegg (2004), Austrian transport demand model for car road pricing analysis, mimeo,
Technical University of Graz, Institute of Road and Transport Systems.
Lindsey, Robin and Erik Verhoef (2001), Traffic Congestion and Congestion Pricing, in:
Button, Kenneth and David A. Hensher, Handbook of Transport Systems and Traffic
Control, Pergamon, Amsterdam: 77-105.
Mayeres, Inge and Stef Proost (2001), Marginal tax reform, externalities and income
distribution, Journal of Public Economics 79, 343-363.
Mayerers, Inge, Sara Ochelen and Stef Proost (1996), The marginal external costs of urban
transport, Transportation Research Part D 1(2), 111-130.
Nash, Chris, Tom Sansom, Ben Still (2001), Modifying transport prices to internalise
externalities: evidence from European case studies, Regional and Urban Economics
31: 413-431.
Proost, Stef and Kurt Van Dender (2001), The welfare impacts of alternative policies to
address atmospheric pollution in urban road transport, Regional Science and Urban
Economics 31: 383-411.
ST.AT (2002), Vebrauchsausgaben 1999/2000 – Sozialstatistische Ergebnisse der
Konsumerhebung (Austrian Consumer Expenditures 1999-2000), Statistik Austria
(ST.AT), Vienna.
Steininge r, Karl (2003), The foreign trade and sectoral impact of truck road pricing for crossborder trade: a CGE analysis for a small open economy, Environmental and Resource
Economics.23(2): 213-253.
Viegas, José M. (2001), Making urban road pricing acceptable and effective: searching for
quality and equity in urban mobility, Transport Policy 8, 289-294.
- 17 APPENDIX
Variables
Factor demand
L
K
total labour demand
total capital demand
Production
Xj
Kj
Lj
Hj
Aj,aij
δj
σj
gross production of sector j
capital input in sector j
labour input in sector j
factor aggregate in sector j
Leontieff-input -output -coefficients in sector j
CES-distribution parameter in sector j
elasticity of substitution in production between labour and capital in sector j
Foreign trade
EXj
Mj
Pj
P jW
EX0, M 0
εj
export of sector j
import of sector j
production price of goods aggregate X in sector j
world market price of goods aggregate M in sector j
export and import quantities in sector j in the reference year
foreign trade price elasticity of demand in sector j
Labour Market
w
nominal wage rate
wlow
lower bound on the real wage rate
pp
u
Paasche index of the aggregate price level
rate of unemployment
Transport
Tp
Private car passenger transport
Tpf
Private car passenger transport production fixed input
Tpfv
Private car passenger transport production variable input (directly kilometre dependent)
Tu
Public passenger transport
Apf, A pv , Apfi, A pv iLeontieff-input -output -coefficients in private car passenger transport
Akmp
kilometre input coefficient in private car passenger transport
Au i
Leontieff-input -output -coefficients in public transport
kmp
vehicle kilometres driven in private car transport
Consumption
Ch
Xh c
Th
δh C
δh T
δh,iX
σh C
σh T
σh X
Total Consumption of household type h
Consumption of non-transport goods of household h
Transport consumption of household h
CES-distribution parameter in consumption for household h
CES-distribution parameter in transport consumption for household h
CES-distribution parameter in non-transport consumption for household h
elasticity of substitution between transport and non-transport demand for household h
elasticity of substitution between private car transport and public transport demand for household h
elasticity of substitution between non-transport goods in household h consumption
- 18 List of Core Model Equations
Production
X
(1)
j
= min (H
(
(σ j −1) / σ j
(2) H j = δ j L j
(3)
T
p
(4)
T
pf
(5)
T
pv
(6)
T
u
(
= min T
A j , X ij a ij
j
pf
A
pf
,T
(X
i
A ipf
= min
(X
i
A ipv , km
(X
for j = 1, ....35
)
(σ j −1) / σ j σ j /( σ j −1)
+ (1 − δ j )K j
= min
= min
)
i
pv
A
for j = 1, ....35
)
pv
)
p
A kmp
)
A ui )
Foreign Trade
0
w
(7) EX j = EX J ( Pj / Pj )
(8) M j = M 0j ( Pj / Pjw )
Σj
for j = 1, ....35
Σj
for j = 1, …35
Labour Market
(9)
w
≥ wlow
pP
⊥u
Household Demand
(10)
(11)
Ch = δ hC X h
+ (1 − δ hC )Th

X
X
X
X
c
X
c (σ ´h −1) / σ h  σ h /(σ h −1)

X h = [∑  δ h ,i X h ,i
]


i
C
C
c (σ´ h −1) / σ h
(σ hC −1) / σ C
h



with
 T p (σ´Th −1) / σ Th + (1 − δ T )T u (σ Th −1) / σ Th 
(12) Th = δ h Th

h
h


C
σC
h /(σ h −1)
for h=h1 ,....h4
∑ (δ ) = 1
X
h ,i
for h=h1 ,....h4
i
σ hT /(σ Th −1)
for h=h1 ,....h4