Handbook On The External Costs of Transport

Handbook on the external
costs of transport
Version 2019 – 1.1
EUROPEAN COMMISSION
Directorate-General for Mobility and Transport
Directorate A — Policy Coordination
Unit A3 — Economic analysis and better regulation
Contact: Rolf Diemer

E-mail: MOVE-A3-SECRETARIAT@ec.europa.eu
European Commission
B-1049 Brussels
1 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

costs of transport
Version 2019 – 1.1

Europe Direct is a service to help you find answers
to your questions about the European Union.
Freephone number (*):
00 800 6 7 8 9 10 11
(*) The information given is free, as are most calls (though some operators, phone
boxes or hotels may charge you).
LEGAL NOTICE
This document has been prepared for the European Commission however it reflects the views only of
the authors, and the Commission cannot be held responsible for any use which may be made of the
information contained therein.
More information on the European Union is available on the Internet (http://www.europa.eu).
Luxembourg: Publications Office of the European Union, 2020
ISBN 978-92-76-18184-2
doi: 10.2832/51388
© European Union, 2020
Reproduction is authorised provided the source is acknowledged.

costs of transport
Version 2019 – 1.1
This report is prepared by:

Huib van Essen, Lisanne van Wijngaarden, Arno Schroten (CE Delft), Daniel Sutter, Cuno Bieler (INFRAS), Silvia
Maffii, Marco Brambilla, Davide Fiorello, Francesca Fermi, Riccardo Parolin (TRT), Kareen El Beyrouty (Ricardo)
Delft, CE Delft, January 2019
Publication code: 18.4K83.131
Client: European Commission, Directorate-General for Mobility and Transport
Publications of CE Delft are available from www.cedelft.eu
Further information on this study can be obtained from the contact person Huib van Essen (CE Delft)
© copyright, CE Delft, Delft

CE Delft
Committed to the Environment
Through its independent research and consultancy work CE Delft is helping build a sustainable world. In the
fields of energy, transport and resources our expertise is leading-edge. With our wealth of know-how on
technologies, policies and economic issues we support government agencies, NGOs and industries in pursuit of
structural change. For 40 years now, the skills and enthusiasm of CE Delft’s staff have been devoted to
achieving this mission.

Content
Glossary 11
List of tables 13
List of figures 20
Acknowledgements 22
1 Introduction 23
1.1 Background of the Handbook 23
1.2 Objective 24
1.3 Scope 24
1.4 User guide: how to use this Handbook 28
1.5 Outline of the Handbook 29
2 General methodological framework 31

2.1 Introduction 31
2.2 The concept of external costs 31
2.3 General overview of valuation methodologies 33
2.4 Value transfer approach 35
3 Accident costs 38
3.1 Introduction 38
3.2 Definition and scope 38
3.3 Total and average accident costs 41
3.4 Marginal accident costs 46
3.5 Robustness of results 49
4 Air pollution costs 51

4.1 Introduction 51
4.3 Total and average air pollution costs 51
4.4 Marginal air pollution costs for selected cases 58
5 Climate change costs 74

5.1 Introduction 74
5.3 Total and average climate change costs 75
5.4 Marginal climate change costs for selected cases 79

6 Noise costs 91
6.1 Introduction 91
6.3 Total and average noise costs 93
6.4 Marginal noise costs 98
7 Congestion costs 103

7.1 Introduction 103
7.3 Total and average road congestion costs 106
7.4 Social marginal road congestion costs 114
7.6 Congestion costs of other transport modes 121
8 Costs of well-to-tank emissions 123

8.3 Total and average costs of well-to-tank emissions 124
8.4 Marginal costs of well-to-tank emissions for selected cases 127
9 Costs of habitat damage 139

9.3 Total and average costs of habitat damage 140
10 Other external costs 146

10.1Costs of soil and water pollution 146
10.2Costs of up- and downstream emissions of vehicles and infrastructure 147
10.3External costs in sensitive areas (e.g. mountainous regions) 147
10.4Further externalities of transport 149
11 Synthesis 151
11.1Introduction 151
11.2Overview total and average external costs 151
11.3Comparison with previous studies 164
11.4Recommendations for further assessment 171
References 172
A Economic valuation of human health 186

A.1 Introduction 186
A.2 Indicators for human health 186
A.3 Recommended approaches 193
B Detailed assessment accident costs 194

B.1 Introduction 194
B.2 Detailed discussion on accident cost elements 194
B.3 Assessment of total and average accident costs 196
B.4 Assessment of marginal accident costs 204
C Detailed assessment air pollution costs 207

C.1 Introduction 207
C.2 Detailed discussion on effects of air pollution 207
C.3 Assessment of air pollution costs in the previous Handbook 208
C.4 New evidence 208
C.5 Detailed description of the methodology applied 215
D Detailed assessment climate change costs 221

D.1 Introduction 221
D.2 Detailed discussion on effects of climate change 221
D.3 Assessment of climate change costs 222
E Detailed assessment of noise costs 230

E.1 Introduction 230
E.2 Detailed discussion on effects of noise 230
E.3 Noise data 233
E.4 Noise nuisance compared between modes 234
E.5 Assessment of total and average noise costs 235
F Detailed assessment congestion costs 241

F.1 Introduction 241
F.2 Detailed discussion on effects of road congestion 241
F.3 Detailed estimated congestion costs by country - Total costs generated estimated
using the simplified approach 260
F.4 Detailed estimated congestion costs by country - Total costs borne by mode 266
F.5 Detailed estimated congestion costs by country - Costs per vkm borne on
congested roads 271
F.6 Detailed estimated congestion costs by country - Average costs on the whole
network 281
F.7 Detailed estimated congestion costs by country - Marginal costs 293
F.8 Congestion and scarcity costs for other transport modes 302
G Detailed assessment costs of habitat damage 304

G.1 Introduction 304
G.2 Detailed discussion on impacts on natural habitats (nature and landscape) 304
G.3 Assessment of costs of impacts on natural habitats (nature and landscape) 305
H Total and average costs motorways 307

H.1 Introduction 307
H.2 Total external costs of transport on motorways in the EU28 307
H.3 Average external costs of transport on motorways in the EU28 307
I EU27 values 310

I.1 Introduction 310
I.2 Total external costs of transport in the EU27 310
J Marginal cost figures 313

J.1 Introduction 313
J.2 Overview marginal external costs 313
J.3 Marginal air pollution costs for reference vehicles 322
J.4 Marginal climate change costs for reference vehicles 325
J.5 Marginal well-to-tank costs for reference vehicles 326
K Overview content Excel file 329

Glossary
Term Explanation
AG Age Group
AIS Abbreviated Injury Scale
CBA Cost Benefits Analysis
CEMT class Size class for inland waterway vessels and barges
CH 4 Methane
CRF Concentration Response Function
CO2 Carbon-dioxide
DALY Disability-Adjusted Life Year
Db (A) Decibel (A-weighted)
EBD Environmental burden of disease
EDP Ecosystem Damage Potential
EEA European Environmental Agency
EMEP European Monitoring and Evaluation Program for the long-range transmission of air pollution
ERA European Union Agency for Railways
ERF Exposure Response Function
ETS Emission Trading Scheme
DWL Dead Weight Loss
GDP Gross Domestic Product
GHG Greenhouse gas
Gt Gross tonnage
GWP Global Warming Potential
HGV Heavy Goods Vehicle
HSL High Speed Line
IAM Integrated Assessment Model (for climate change)
IMPACT Internalisation Measures and Policies for All Modes of Transport (study that delivered the
first EU wide Handbook on external cost)
IWT Inland Waterway Transport
kton Kiloton
LCV Light Commercial Vehicle
Lden Level-day-evening-night (weighted noise measure)
LRS Lower Respiratory Symptoms
LTO Landing-and –take-off
MC Motorcycle
MRAD Minor Restricted Activity Days
N2O Nitrous-oxide
NH 3 Ammonia
NMVOC Non-methane Volatile Organic Compounds
NOx Nitrogen-oxides
NSDI Noise Sensitivity Depreciation Indexes
NUTS-3 Category in the classification of territorial units for statistics indicating small regions
PAH Polycyclic Aromatic Hydrocarbon
Pax Passengers
PCE Passenger Car Equivalent
PDF Physical Damage Factor
Pkm Passenger-kilometre
PM Particulate Matter

Term Explanation
PM2.5 Particulate Matter smaller than 2.5 micro-metre
PM10 Particulate Matter smaller than 10 micro-metre
ppm Parts Per Million
ppb Parts per Billion
PPP Purchasing Power Parity
QALY Quality-Adjusted Life Years
RAD Restricted Activity Days
RG Risk Group
RoPax Roll-on-roll-of Passenger ship or ferry
RP Revealed Preference
RR Relative Risk or Relative Risk factor
SIA Secondary Inorganic Aerosols
SO2 Sulphur-dioxide
SOMO 35 Sum of ozone means over 35 ppb
SP Stated Preference
t Metric tonnes
TEU Twenty Foot Equivalent Unit (volume of a 20 foot container)
Tkm Tonne-kilometre
Vkm Vehicle-kilometre
VSL Value of Statistical Life
VOC Volatile Organic Compound
VOLY Value Of Life Year lost
WHO World Health Organisation
WLD Work Loss Days
WTT Well To Tank
WTP Willingness To Pay
YLD Years Lost due to Disability
YOLL Years of Life Lost

List of tables
Table 1 – Transport modes and vehicle types covered ................................................... 25
Table 2 – Airports and maritime ports covered ............................................................. 26
Table 3 – Level of externality of various costs categories ................................................ 33
Table 4 – Comparison of old and new EU traffic injury definitions.................................... 40
Table 5 – Correction factors to correct for underreporting of accidents ............................ 40

Table 6 – External accident cost components per casualty for the EU28 (€ 2016 ) .................. 44
Table 7 – External accident costs components per casualty for the EU28 (€2016 )................. 44
Table 8 – Total and average external accident costs for land-based modes for the EU28..... 45
Table 9 – Total and average external accident costs for aviation for 33 selected EU airports 46
Table 10 – Total and average external accident costs for maritime transport for 34 selected
EU ports.................................................................................................................. 46
Table 11 – Degree of risk internalisation for different vehicle types ................................. 48
Table 12 – Marginal external accident costs road transport for the EU28.......................... 48
Table 13 – Data sources for the emissions of air pollutants for different transport modes ... 55
Table 14 – Air pollution costs: average damage cost in €/kg emission, national averages for
transport emissions in 2016 (excl. maritime) (All effects: health effects, crop loss, biodiversity
loss, material damage) .............................................................................................. 55
Table 15 – Air pollution costs: average damage cost in €/kg emission, national averages for
maritime emissions in 2016 (all effects: health effects, crop loss, biodiversity loss, material
damage) ................................................................................................................. 56
Table 16 - Total and average air pollution costs for land-based modes for the EU28........... 57
Table 17 – Total and average air pollution costs for aviation for 33 selected EU airports ..... 57
Table 18 – Rough estimates for total and average external air pollution costs for maritime
transport for 34 selected EU ports .............................................................................. 58
Table 19 – Marginal air pollution costs road transport for selected cases.......................... 58
Table 20 - Marginal air pollution costs rail transport for selected cases ............................ 67
Table 21 – Marginal air pollution costs IWT for selected cases ........................................ 69
Table 22 – Marginal air pollution costs maritime transport for selected cases ................... 70

Table 23 – Marginal air pollution costs aviation for selected cases ................................... 72
Table 24 – Climate change avoidance costs in €/tCO2 equivalent (€2016 )............................ 78
Table 25 - Total and average climate change costs for land-based modes for the EU28....... 78
Table 26 – Total and average climate costs for aviation for selected 33 EU airports ............ 79
Table 27 – Rough estimates for total and average external climate change costs for maritime
transport for 34 selected EU ports .............................................................................. 79
Table 28 - Marginal climate change costs road transport for selected cases ...................... 80
Table 29 – Marginal climate change costs rail transport for selected cases........................ 88
Table 30 – Marginal climate change costs IWT for selected cases .................................... 88
Table 31 – Marginal climate change costs maritime transport for selected cases ............... 89
Table 32 – Marginal climate change costs aviation for selected cases............................... 89
Table 33 – Environmental price of traffic noise for the EU28 (€ 2016 /dB/person/year) .......... 95
Table 34 – Weighting factors for noise for different vehicle types .................................... 95
Table 35 - Total and average noise costs for land-based modes for the EU28 .................... 97
Table 36 – Total and average noise costs for aviation for 33 selected EU airports............... 97
Table 37 – Marginal noise costs road transport – in €-cent (2016) per pkm, tkm or vkm (data
for 2016) ................................................................................................................. 99
Table 38 - Marginal noise costs rail transport – in €-cent (2016) per pkm and tkm (data for
2016).................................................................................................................... 100
Table 39 – Marginal noise costs for aviation: data for different aircraft types .................. 100
Table 40 – Marginal noise costs for aviation, estimations based on average costs (data for
2016).................................................................................................................... 101
Table 41 - Total and average congestion costs borne by road vehicle categories in the EU28
............................................................................................................................ 111
Table 42 - Total and average congestion cost of inter-urban traffic borne by vehicle
categories on motorways network............................................................................ 111
Table 43 - Total and average congestion costs generated by road vehicle categories in the
EU28 according to the simplified approach used ......................................................... 112
Table 44 - Total and average congestion cost generated of inter-urban traffic by vehicle
categories on motorways network, according to the simplified approach used................ 113
Table 45 – Social marginal congestion costs of road transport per pkm and tkm .............. 116

Table 46 – Social marginal congestion costs of road transport per vkm........................... 117
Table 47 – Social marginal congestion costs of road transport generated per pkm and tkm
using the simplified approach .................................................................................. 119
Table 48 – Social marginal congestion costs of road transport generated per vkm using the
simplified approach ................................................................................................ 119
Table 49 – Well-to-tank air pollution costs: damage cost estimates in €/kg emission
(emissions in the year 2016, EU28 values).................................................................. 125
Table 50 – Total and average costs of well-to-tank emissions for land-based modes for the
EU28 .................................................................................................................... 126
Table 51 – Total and average well-to-tank costs for aviation for 33 selected EU airports ... 126
Table 52 – Rough estimates for total and average external well-to-tank costs for maritime
transport for 34 selected EU ports ............................................................................ 127
Table 53 – Average/marginal costs of well-to-tank emissions road transport for selected

cases .................................................................................................................... 127
Table 54 – Average/marginal costs of well-to-tank emissions rail transport for selected cases
............................................................................................................................ 136
Table 55 – Average/marginal costs of well-to-tank emissions IWT for selected cases (€-cent
per tkm)................................................................................................................ 136
Table 56 – Average/marginal costs of well-to-tank emissions maritime transport for selected

cases .................................................................................................................... 136
Table 57 – Average/marginal costs of well-to-tank emissions aviation for selected cases .. 137
Table 58 – Cost factors for costs of habitat damage EU28............................................. 141
Table 59 – Cost factors for costs of habitat damage for all countries .............................. 141
Table 60 – Cost factors for ecosystem (habitat) loss..................................................... 143
Table 61 – Total and average costs of habitat damage for land-based transport modes in the
EU28 .................................................................................................................... 144
Table 62 – Total and average habitat costs for aviation for 33 selected EU airports .......... 144
Table 63 – Mountain factors for external costs of freight transport................................ 148
Table 64 – Total external costs 2016 for road transport, rail transport and IWT per country
............................................................................................................................ 153
Table 65 – Total external costs 2016 for EU28 passenger transport by cost category and
transport mode...................................................................................................... 154

Table 66 – Total external costs 2016 for EU28 freight transport by cost category and
transport mode...................................................................................................... 155
Table 67 – Total external costs for selected EU28 (air)ports .......................................... 156
Table 68 – Indicative estimates for the most relevant total external costs for all EU28
(air)ports............................................................................................................... 157
Table 69 – Average external costs 2016 for EU-28 ....................................................... 160
Table 70 – Average external costs 2016 for EU28 passenger transport by cost category and
transport mode...................................................................................................... 160
Table 71 – Average external costs 2016 for EU28 freight transport by cost category and
transport mode...................................................................................................... 161
Table 72 – Average external costs for selected EU28 (air)ports...................................... 162
Table 73 – Average external costs 2016 for EU28 by country and transport mode (excluding
congestion)*.......................................................................................................... 163
Table 74 – Examples of disability weights................................................................... 187
Table 75 – Overview of VSL studies ........................................................................... 188

Table 76 – Overview of the VOLY values found in the literature..................................... 192
Table 77 – Overview of the literature on percentages of the VSL used for serious and slight
injuries ................................................................................................................. 198
Table 78 – Overview of human costs for new EU injury definition, EU28 aggregate value .. 200
Table 79 – Overview of values used in 2014 Handbook for the degree of risk internalisation
for marginal costs................................................................................................... 205
Table 80 – Concentration-response functions and external air pollution costs ................. 209
Table 81 – Changes compared to the NEEDS project .................................................... 214
Table 82 – Climate avoidance costs in €/tCO2 eq. (€2016 ) ............................................... 229

Table 83 – Health endpoints of noise for which the evidence is classified as at least
‘moderate quality’ by the WHO ................................................................................ 231
Table 84 – Valuation of noise annoyance in the EU28 in €2016 per person per dB, Lden........ 235
Table 85 – Noise Sensitivity Depreciation Index (NSDI) results of hedonic price studies..... 236
Table 86 – Health costs used for the EU28 (€2016/person/dB/year) based on Defra (2014)240
Table 87 - Value of time by purpose and country for short and long distance trips by car
(Euro2016/hour per person) .................................................................................... 245

Table 88 - Value of time for freight road long distance trips by country (Euro2016/hour per
tonne) .................................................................................................................. 248
Table 89 - Values of cost elasticities assumed ............................................................. 249

Table 90 - EU28 Deadweight loss per vkm of road transport on congested network ......... 251
Table 91 - EU28 Delay costs per vkm of road transport on congested network ................ 252
Table 92 - Average yearly delay cost and deadweight loss per capita depending on city
population size: TomTom sample data (EURO2016 /capita) ............................................. 254
Table 93 - Average yearly delay cost and deadweight loss per capita depending on city
population size in non-EU regions: TomTom sample data (EURO2016 /capita)................... 259
Table 94 - Passenger Car Value of Time and average occupancy factor in EU28 and non-EU
regions ................................................................................................................. 260
Table 95 - Car urban congestion costs in EU28 and non-EU regions................................ 260
Table 96 - Total delay congestion costs generated by road modes, estimated using the
simplified approach (billion Euro/year, in Euro2016 ) ...................................................... 260
Table 97 - Total deadweight loss generated by road modes, estimated using the simplified
approach (billion Euro/year, in Euro2016 ) .................................................................... 262
Table 98 - Total inter-urban congestion costs generated on motorway network, estimated

using the simplified approach (billion Euro/year, in Euro2016 ) ........................................ 264
Table 99 - Total car congestion costs (billion Euro/year, in Euro2016 )............................... 266
Table 100 - Total trucks, coaches and LCVs inter-urban congestion costs (billion Euro/year, in
Euro2016 )................................................................................................................ 267
Table 101 - Total trucks urban congestion costs (billion Euro/year, in Euro2016 )................ 268
Table 102 - Total inter-urban congestion costs on motorway network (billion Euro/year, in
Euro2016 )................................................................................................................ 269
Table 103 - Car deadweight loss per vkm on congested network (€-cent/vkm, in Euro2016 ) 271
Table 104 - Inter-urban deadweight loss per vkm borne on congested network for trucks and
coaches (€-cent/vkm, in Euro2016 ) ............................................................................. 274
Table 105 - Car delay costs per vkm borne on congested network (€-cent/vkm, in Euro2016 )
............................................................................................................................ 276
Table 106 - Inter-urban delay congestion costs per vkm borne on congested network for
trucks and coaches (Euro Cent/vkm, in Euro2016 ) ......................................................... 278
Table 107 - Car average delay congestion cost and deadweight loss generated per vkm,
estimated using the simplified approach (€-cent/vkm, in Euro2016 )................................. 281

Table 108 – Bus/coaches average delay congestion cost and deadweight loss generated per
vkm, estimated using the simplified approach (€-cent/vkm, in Euro2016 ) ......................... 283
Table 109 – Freight HGV average delay congestion cost and deadweight loss generated per
vkm, estimated using the simplified approach (€-cent/vkm, in Euro2016 ) ......................... 285
Table 110 – Light commercial vehicle average delay congestion cost and deadweight loss
generated per vkm, estimated using the simplified approach (€-cent/vkm, in Euro2016 )..... 287
Table 111 - Car average delay congestion cost and deadweight loss brone per vkm (€-
cent/vkm, in Euro2016 ) ............................................................................................. 288
Table 112 - Average inter-urban delay congestion costs and deadweight loss borne for HGV,
coaches and LCV (€-cent/vkm, in Euro2016 ) ................................................................. 289
Table 113 - Average urban delay congestion costs and deadweight loss borne for trucks (€-
cent/vkm, in Euro2016) ........................................................................................... 290
Table 114 - Average inter-urban delay congestion costs and deadweight loss borne on
motorway network (€-cent/vkm, in Euro2016 )............................................................. 291
Table 115 - Car social marginal congestion costs per vkm (€-cent/vkm, in Euro2016 ) .......... 293
Table 116 - Inter-urban social marginal congestion costs for trucks and coaches (€-cent/vkm,
in Euro2016 ) ............................................................................................................ 296
Table 117 - Inter-urban social marginal congestion costs generated for HGV, estimated using
the simplified approach (€-cent/vkm, in Euro2016 )........................................................ 298
Table 118 - Social marginal congestion costs generated for LCVs per vkm, estimated using the
simplified approach (€-cent/vkm, in Euro2016 ) ............................................................. 300
Table 119 - Marginal external costs of congestion of rail freight transport €-cent/1,000 tkm
(2016 prices) ......................................................................................................... 303
Table 120 – Literature on external costs of habitat damage due to transport activities ..... 306
Table 121 – Total external costs of transport on motorways in the EU28 (billion €) .......... 308
Table 122 – Average external costs of transport on motorways in the EU28.................... 309
Table 123 – Total external costs of transport in the EU27 (billion €) ............................... 311
Table 124 – Total external costs of transport for selected (air)ports in the EU27 (billion €) 312
Table 125 – Synthesis of marginal external costs 2016 for EU28 – passenger cars (in €-
cent/pkm) ............................................................................................................. 313
Table 126 – Synthesis of marginal external costs 2016 for EU28 – motorcycles and mopeds
(in €-cent/pkm) ...................................................................................................... 314
Table 127 – Synthesis of marginal external costs 2016 for EU28 – buses (in €-cent/pkm) .. 315

Table 128 – Synthesis of marginal external costs 2016 for EU28 – coaches (in €-cent/pkm) 315
Table 129 – Synthesis of marginal external costs 2016 for EU28 – LCVs (in €-cent/vkm) .... 316
Table 130 – Synthesis of marginal external costs 2016 for EU28 – HGVs (in €-cent/tkm) ... 317
Table 131 – Synthesis of marginal external costs 2016 for EU28 – passenger trains (in €-
cent/pkm) ............................................................................................................. 318
Table 132 – Synthesis of marginal external costs 2016 for EU28 – freight trains (in €-
cent/tkm).............................................................................................................. 319
Table 133 – Synthesis of marginal external costs 2016 for EU28 – IWT (in €-cent/tkm) ..... 321
Table 134 – Synthesis of marginal external costs 2016 for EU28 – maritime transport (in €-
cent/pkm (ferry) and €-cent/tkm (vessels)) ................................................................ 321
Table 135 – Synthesis of marginal external costs 2016 for EU28 – aviation (in €-cent/pkm) 322
Table 136 – Marginal air pollution costs (averages for metropolitan, urban and rural regions)
............................................................................................................................ 322
Table 137 – Marginal climate change costs................................................................. 325
Table 138 – Marginal well-to-tank costs..................................................................... 326

List of figures
Figure 1 – Methodology total and average accident costs............................................... 41
Figure 2 – Methodology marginal external accident costs............................................... 46
Figure 3 – Methodology total and average costs of air pollution ...................................... 52
Figure 4 – The Impact Pathway Approach for calculating air pollution costs ...................... 53
Figure 5 – Methodology total and average climate change costs ..................................... 75

Figure 6 – Methodology total and average noise costs ................................................... 94
Figure 7 – Road congestion depending on network conditions ...................................... 104
Figure 8 – Methodology for estimating urban congestion costs for delay cost and deadweight
loss approaches ..................................................................................................... 107
Figure 9 – Methodology for estimating inter-urban congestion costs for delay cost and
deadweight loss approaches .................................................................................... 108
Figure 10 – Methodology total and average costs of well-to-tank emissions.................... 124
Figure 11 – Methodology total and average costs of habitat damage ............................. 140
Figure 12 – Total external costs 2016 for EU28 (excluding congestion) ........................... 152
Figure 13 – Share of the different cost categories on total external costs 2016 for EU28... 152
Figure 14 – Share of the different transport modes on total external costs 2016 for EU28. 152
Figure 15 – Average external costs 2016 for EU28: passenger transport (excluding
congestion) ........................................................................................................... 158
Figure 16 – Average external costs 2016 for EU28: freight transport (excluding congestion)
............................................................................................................................ 159
Figure 17 – The approach for damage cost calculations: from emission to impact and damage
............................................................................................................................ 208
Figure 18 – Review of avoidance cost values found in the literature (€ 2016 /t CO2 equivalent)
............................................................................................................................ 228
Figure 19 – Comparison of results: DEFRA and Bristow et al. (2015) (€ 2016 per dB per
household) ............................................................................................................ 237
Figure 20 - The speed-flow relationship ..................................................................... 242
Figure 21 - Road congestion depending on network conditions ..................................... 242
Figure 22- Example of speed-flow functions for different road types.............................. 244

Figure 23 - The road network in TRUST ...................................................................... 257

Acknowledgements
An advance draft of this document was the subject of a one-day workshop on 10 July 2018
in Brussels. Independent experts in the field of estimation of external costs of transport
were invited to comment on the methodology and on the preliminary results. The experts
provided recommendations for the finalisation of the Handbook, which were largely taken
into account by the authors in the final version. The following experts took part in the
workshop:
— Christopher Nash, research professor on transport economics from the University of
Leeds
— Stale Navrud, Professor of School of Economics and Business at the Norwegian University
of Life Sciences
— Stefaan Proost, Prof. at Faculty of Economics and Business, University of
Leuven/Belgium
— Cristoph Lieb Senior Consultant in transport economics, ECOPLAN
— Alexandra Quandt, Swiss Federal Office of Statistics
— Martin Adler, specialist in transport and urban economics
— Panayotis Christidis, Joint Research Centre JRC C.6

1 Introduction
1.1 Background of the Handbook
Overview of the study and other deliverables
This updated Handbook on external costs of transport has been developed in the study ‘Sustainable Transport Infrastructure
Charging and Internalisation of Transport Externalities’ commissioned by the European Commission DG MOVE, by a
consortium led by CE Delft. The objective of this study is to assess the extent to which the ‘user pays’ and the ‘polluter pays’
principles are implemented in EU Member States and in other developed countries. This will allow DG MOVE to take stock of
the progress of Member States towards the goal of full internalisation of external (and infrastructure) costs of transport and
to identify options for further internalisation.
The full list of deliverables of this study are:
— Handbook on external costs – version 2019 (current report).

• This report provides an overview of the methodologies and input values that can be used to provide
state-of-the-art estimates for all main external costs of transport. Furthermore, the report and
corresponding excel file present the total, average and marginal external costs for all relevant
countries.
— Overview of transport infrastructure expenditures and costs.
• This report provides an overview of the infrastructure costs of all transport modes in all relevant
countries.
— Transport taxes and charges in Europe - An overview study of economic internalisation measures applied in
Europe.
• This study provides an overview of the structure and level of transport taxes and charges applied for
the various transport modes in the EU28 Member States (and the other relevant countries).
Furthermore, this study presents the total revenues from transport taxes and charges for the various
transport modes and countries.
— The state of play of internalisation in the European transport sector.
• This report shows the extent to which external and infrastructure costs are internalised by current
taxes and charges for all countries and transport modes.It also investigates recommended options for
further internalisation.
— Summary report.
• Providing an overview of the main findings of the other four deliverables.
In 2008 the European Commission commissioned the first Handbook on External Costs of
Transport, as part of the IMPACT study (Infras, CE Delft, ISI & University of Gdansk, 2008).
This Handbook presented the best practice on the methodology to estimate different
categories of external costs of transport. Additionally, it provided an overview of state of
the art input values (e.g. the value of time or the value of a statistical life) that can be
used to produce estimations of external costs by users of the Handbook themselves.
Finally, the Handbook presented external cost figures (mostly presented in €/vehicle
kilometre), which can be used directly by the users.
The 2008 Handbook focus was on marginal external costs of transport as a basis for the
definition of internalisation policies (in line with the marginal social cost pricing principle).
It covered all main external cost categories, including air pollution, climate change, noise,
accidents and congestion. The Handbook was based on the existing (up to 2007) scientific
and expert work, mainly carried out at the EU level and within European countries. It was

reviewed by a panel of more than thirty experts, including experts who were designated by
Member States.
In 2014 the Handbook was updated with new developments in research and policy (Ricardo-
AEA, TRT, DIW Econ & CAU, 2014). Furthermore, the scope was broadened: next to the
external costs of transport, infrastructure wear and tear costs for road and rail transport
were covered as well. In line with the 2008 Handbook, the focus of the 2014 Handbook was
on marginal external costs of transport. Next to the Handbook, an accompanying Excel file
was produced, containing country specific estimates of the main external costs of road and
rail transport.
This Handbook is an update of the 2008 and 2014 version, taking into account any new
evidence that has become available on the methods and input values (e.g. emission factors)
for estimating external costs of transport in research and policy since 2014. This version of
the Handbook does not only consider marginal external costs, as was the main focus of the
previous Handbooks, but also total and average external costs of transport in all
EU-countries, Switzerland and Norway. Furthermore, external cost figures for some
non-European countries were produced to compare them with the European figures.
1.2 Objective
The objective of this Handbook is to provide information on how to generate state-of-the-
art estimates for all main external costs of transport. This information is provided at three
levels:
— Methodological level: what are the state of the art methodologies to estimate figures
for the various external costs of transport?
— Input values: which input values (particularly at monetary terms, e.g. the value of time)
are recommended to use to estimate external costs of transport?
— Output values: which default external cost values for different transport modes (and if
meaningful, for different traffic situations) can be recommended?
In this Handbook, state of the art methodologies, input values and output values for total,
average and marginal external costs of transport are provided, both at the EU28 level as at
the level of individual countries. This is done for all transport modes and all (main) external
cost categories.
1.3 Scope
1.3.1 External cost categories

This Handbook covers all main externalities of transport:
— accidents;
— air pollution;
— climate change;
— noise;
— congestion;
— well-to-tank emissions;
— habitat damage;
— other external cost categories (e.g. soil and water pollution).

Transport infrastructure costs are not considered in this Handbook, as it is addressed in a
parallel study carried out within the broader project on internalisation of external and
infrastructure costs of transport (see the text box in Section 1).
1.3.2 Transport modes

This Handbook considers road transport, rail transport, inland waterway transport (IWT),
maritime transport and aviation. Total and average cost figures are produced for the
vehicle categories shown in Table 1. Furthermore, cost-specific differentiations of the
external cost estimates are produced when relevant (e.g. average/marginal air pollution
costs of passenger cars are differentiated to Euro class).
Table 1 – Transport modes and vehicle types covered
Road transport Rail transport IWT Maritime Aviation*

transport
− Passenger car − High speed passenger train − Inland − Freight vessel − Passenger
(HSL) vessel aircraft
− Motorcycle − Ferry
− Passenger train electric
− Bus
− Passenger train diesel
− Coach
− Freight train electric
− LCV
− Freight train diesel
− Heavy Goods Vehicle
(HGV)
* Freight aviation is not considered in this Handbook, as the data to provide reliable figures on all external cost categories is
missing.
1.3.3 Geographical coverage

For road transport, rail transport and IWT, input and output values are produced for all
EU28 countries, Norway, Switzerland, Canada, US, and Japan. For Canada and the
United States external costs are considered at the province/state level, i.e. California,
Missouri (both US), British Columbia and Alberta (both Canada) 1.
________________________________
1 Both for the US and Canada, a front runner and laggard state/province with respect to the internalisation of
external costs have been selected. For the US, California has been selected as a front runner state, among other
things because fuel and vehicle taxes are among the highest in the US and broad enabling legislation for toll
roads has been implemented. Furthermore, California is known for its progressive policies in the transport
sector (e.g. regarding electric vehicles). Missouri, on the other hand, shows relatively low fuel and vehicle taxes
as well as limited road charging legislation, suggesting a low level of internalisation. For that reason, Missouri is
selected a laggard state. According to Corporate Knights (2015), British Columbia can be regarded as the
Canadian province with the highest environmental performance for the transport sector, while Alberta is ranked
lowest. Therefore, British Columbia (front-runner) and Alberta (laggard) has been selected as Canadian
provinces in this study.

For maritime shipping and aviation, external cost figures (output values) are not provided at
the national/state level, but at the level of individual (air)ports2. The selection of (air)ports
considered in this study is given in Table 2. This selection is made based on the following
criteria:
— Airports:
1. Of all considered countries the largest airport is analysed.
2. In Canada and the US, the two largest airports are included.
3. In Europe, the five largest airports, which are not already included in the criteria
above, are also considered.
4. Only international airports (with international flights) are covered in the analysis.
— Maritime ports:
1. All 24 maritime ports considered in the study ‘Assessment of potential of maritime
and inland ports and inland waterways and of related policy measures, including
industrial policy measures’ (EY, et al., ongoing) are covered. The maritime ports
considered in this study provide a good representation of main EU ports with growth
potential up to 2030.
2. As not all countries were covered by the ports selected in Step 1, an additional set of
ten ports was included to cover the main maritime ports for all European countries
considered in this study.
3. In order to provide a good representation of the main ferry/RoPax ports as well, an

additional German port was added to the list.
4. A sample of five overseas ports in the US, Canada and Japan have been selected.
Table 2 – Airports and maritime ports covered
Country Airport(s) Maritime port(s)

Freight ports Ferry ports
Austria Wien - Schwechat
Belgium Brussels Antwerp Antwerp
Bulgaria Sofia Varna
Croatia Zagreb Pleso Rijeka Rijeka
Split Split
Cyprus Larnaka Lemessos
Czech Republic Prague Ruzyne
Denmark Copenhagen - Kastrup Arhus Arhus
Helsingør (Elsinore) Helsingør (Elsinore)
Estonia Lennart Meri Tallinn Tallinn Tallinn
Finland Helsinki - Vantaa Helsinki Helsinki
________________________________
2
This is done to be consistent with the other studies carried out within the broader study on the internalisation
of external and infrastructure costs (see text box in Section 1). Both in the study on infrastructure costs and the
study on taxes and charges the (air)port level is a more appropriate scope than the country level, as data on
infrastructure costs and taxes/charges are mainly available at the (air)port level.

Country Airport(s) Maritime port(s)
Freight ports Ferry ports
France Paris – Charles de Gaulle Calais Calais
Paris - Orly Le Havre Marseille
Marseille
Germany Frankfurt Hamburg Travemünde
Munich Bremerhaven
Greece Athens Eleftheriios Venizelos Piraeus Piraeus
Hungary Budapest Liszt Ferenc
Ireland Dublin Dublin Dublin
Italy Roma - Fiumicino Genova Genova
Trieste Trieste
Venice Venice
Latvia Riga Riga Riga
Lithuania Vilnius Klaipeida Klaipeida
Luxembourg Luxembourg
Malta Luga Marsaxxlokk
Netherlands Amsterdam - Schiphol Rotterdam Rotterdam
Poland Warsaw Chopina Gdansk Gdansk
Portugal Lisboa Sines
Romania Bucharest Henri Coandă Constantza
Slovakia Bratislava M.R. Stefanik
Slovenia Ljubljana Brink Koper
Spain Barcelona – El Prat Algeciras Algeciras
Adolfo Suarez Madrid – Barajas Barcelona Barcelona
Palma de Mallorca Bilbao Bilbao
Valencia Valencia
Sweden Stockholm - Arlanda Goteborg Goteborg
United Kingdom London - Heathrow Felixstowe
London - Gatwick
Norway Oslo - Gardermoen Oslo Oslo
Switzerland Zurich
Canada Toronto/Lester B Pearson Intl. Vancouver
Ont. Montreal
Vancouver International B.C.
United States Atlanta Hartsfield – Jackson Los Angeles
International Savannah
Los Angeles International
Japan Haneda Airport Tokyo Tokyo
1.3.4 Transport performance data

To estimate the various external cost figures (output values), several types of transport
performance data (e.g. vkms, tkms, pkms) have been used. For the purpose of this
Handbook a consistent set of transport performance data has been composed, mainly based
on EU aggregated sources (like Eurostat and COPERT). For maritime transport and aviation,
(air)port specific transport performance data (e.g. number of calls, LTOs) are collected
from port authorities and annual reports of the considered (air)ports directly.
Road transport performance data is taken from Eurostat, following the nationality principle,
i.e. transport activity is allocated to countries where the vehicle is registered. In an

alternative approach, the territorial principle, transport activity is allocated to the
countries where the activity actually takes place. For example, kilometres driven by Polish
vehicles in Germany are accounted to Poland if the nationality principle applies, and to
Germany if the territorial principle applies. The territorial principle would have been more
consistent with the scope of the external costs. However, as a detailed EU-wide data set on
road transport performance based on the territorial principle is not available, the official
Eurostat data set based on the nationality principle has been used for this study. This
choice (i.e. to apply road transport performance data based on the nationality principle)
affects the results, in particular the allocation of the noise and accident costs from road
transport to different vehicle categories.
1.3.5 Base year

All input and output values in this Handbook are presented for 2016. If some data was not
available for 2016, data for the most recent year (preferably 2015) was used.
1.3.6 Price level

All financial figures are expressed in Euro price levels of 2016. Data from sources where
price levels from other years were used, are translated to price level 2016 by using relevant
price index figures (from Eurostat). Furthermore, all financial figures are adjusted for
differences in purchase power between countries (by using Purchasing Power Standards,
PPS), in order to allow for direct comparisons between counties. This implies that all
financial figures are shown for the EU28 average price level.
1.4 User guide: how to use this Handbook

This Handbook includes guidelines for estimating external costs of transport at three levels
which differ with respect to the level of accuracy of the values produced:
— Methodological level: for each external cost category recommendations on the best
practice methodologies to estimate total/average and marginal external cost values are
provided. Using these methodologies to produce own differentiated cost figures based
on case specific input values results in the most accurate outcomes. However, this level
requires the availability of case-specific estimates of key input parameters and
evaluation models.
— Input values: for each external cost category typical European and Member State values
for key input parameters are provided. Examples of input values are the Value of Time
(in € per hour), the Value of a Life Year lost (€ per life year lost), etc. These input
values can be used to produce own output values in cases some case-specific data is
available.
— Output values: for ready estimations with limited case-specific data, total/average and
marginal external cost figures are provided for all countries and transport modes.
Where relevant, differentiations to relevant vehicle characteristics (e.g. fuel type, size
class, etc.) and traffic situation (type of road, day/night, thin/dense traffic, etc.) are
provided.
The guidelines at these three levels (methodologies, input values and output values) are
given for each external cost category in Chapters 3 to 9 in the main text of this Handbook.
These chapters all follow the same structure:
— Definition and scope of the externality considered is briefly discussed.
— Recommended methodologies, input values and output values for total/average cost
figures are presented. Input and output values are presented for the EU28.

National values are presented in the Excel Annex accompanying this Handbook (see
Annex K for more details).
— Recommended methodologies, input values and output values for marginal cost figures
are presented. Input and output values are presented for the EU28. Again, national
values are provided in the Excel Annex.
— The robustness of the recommended input and output values are discussed.
Each of the external cost chapters in the main text has its own annex (Annex B to G), which
provides:
— A more detailed discussion of the impacts of the externality (if relevant).
— A brief discussion on the methodologies and input values recommended by the previous
Handbook.
— A detailed overview of recent evidence in the literature (mainly studies published since
the previous version of the Handbook) on the methodology and input values to estimate
the external costs, including a critical assessment of this evidence. This literature
review is focussed on both total/average and marginal external costs.
— Conclusions on the best practice methodologies and input values to be recommended in
this Handbook. These recommendations are compared with the recommendations made
in the previous Handbook and any deviations are explained.
In addition to these external cost specific annexes, a more general annex on the economic
valuation of human health is provided (Annex A). This annex provides an overview of
indicators to valuate impacts on human health as well as evidence from literature on the
value of these indicators. Based on this assessment, recommended approaches to valuate
impacts on human health are provided. These recommended approaches are used in
providing input and output values for external cost categories like air pollution, noise and
accidents in a consistent way.
Finally, a synthesis of the results is presented in the last chapter of this Handbook,
comparing total, average and marginal cost values between countries and transport modes.
Furthermore, a brief comparison with previous studies (including the previous Handbooks) is
presented.
The various costs can be added up to retrieve the total external costs per transport mode,
although it should be noted that the totals per vehicle type reported in the handbook may
not necessarily sum up to the total per mode due to rounding errors. Furthermore, it should
be noted that this Handbook the scope for aviation and maritime shipping is a selection of
EU (air)ports, while for the other modes the costs are presented for the EU28. In order to
allow a cross modal comparison, also estimates for the total costs for EU28 for aviation and
maritime shipping are presented, based on extrapolation of the costs for the selected
(air)ports.
The user needs also to be aware that whilst in all cases the same issue is being measured in
principle (what economists call the external social welfare impacts), the nature of the
various cost categories is different and therefore different methodologies were used.
1.5 Outline of the Handbook

Chapter 2 of this Handbook first provides a general methodological framework for the
estimation of external costs. It defines external costs and briefly discusses the main
methodologies to estimate them. The recommended methodologies, input values and
output values for the various external cost categories are presented in Chapters 3 to 9.

In Chapter 10, some external cost categories for which no quantitative assessment can be
applied (due to a lack of scientific evidence) are discussed in qualitative way. Finally, the
main conclusions and recommendations for further research are presented in Chapter 11.

2 General methodological
framework
2.1 Introduction
In this chapter we present the general methodological framework to estimate the external
costs of transport. We start by discussing the concept of external costs in Section 2.2.
In this section we provide a general definition of external costs, explain the differences
between total, average and marginal external costs and discuss the level of externality of
different types of external costs. In Section 2.3 we provide a general overview of the
valuation methodologies that can be used to estimate external costs, discussing their main
pros and cons. Finally, in Section 2.4, we explain the procedure for transferring values from
one country to another or over time.
2.2 The concept of external costs

External costs, also known as externalities, arise when the social or economic activities of
one (group of) person(s) have an impact on another (group of) person(s) and when that
impact is not fully accounted, or compensated for, by the first (group of) person(s). In other
words, external costs of transport are generally not borne by the transport user and hence
not taken into account when they make a transport decision. Cars exhausting NO x emissions,
for example, cause damage to human health, imposing an external cost. This is because the
impact on those who suffer damage to their health is not taken into account by the driver
of the car when deciding on taking the car.
External costs of transport refer to the difference between social costs (i.e. all costs to
society due to the provision and use of transport infrastructure) and private costs of
transport (i.e. the costs directly borne by the transport user). As the market does not
provide an incentive to transport users to take external costs into account, they only take
part of the social costs into account when taking a transport decision, resulting in sub-
optimal outcomes. By internalising these costs, externalities are made part of the decision
making process of transport users. This can be done through regulation (i.e. command and
control measures) or by providing the right incentives to transport users, namely with
market based instruments (e.g. taxes, charges, emission trading, etc.). A combination of
these two basic types of instruments is possible, e.g. taxes differentiated to Euro emission
classes of vehicles.
Using market-based instruments to internalise external costs is generally regarded as an

efficient way to limit the negative side effects of transport and/or to generate income for
the government. Applying these instruments in an efficient way requires detailed and
reliable estimates of external costs. Also for other applications (e.g. use in Cost Benefit
Analyses), external cost figures are useful parameters.

Total, average and marginal external costs
Different types of external costs are distinguished in this Handbook:
— Total external costs refer to all external costs within a geographical boundary
(e.g. EU28 or a country) caused by (a specific mode of) transport. Total external costs
are usually presented in billions or millions Euros.
— Average external costs are closely related to total costs, as they express the costs per
transport performance unit3. In this study average external costs are generally
presented in €-cent/pkm, €-cent per tkm and/or €-cent/vkm. For some transport
modes/externalities alternative units are used, e.g. €-cent/LTO (aviation) or €-cent per
port call (maritime transport).
— Marginal external costs are the additional external costs occurring due to an additional
transport activity. In the short run, these costs are linked to constant infrastructure
capacity, whereas long run marginal costs do take the construction of additional traffic
infrastructure into account. This implies, for example, that short run marginal
congestion costs are, in general, higher than long run marginal congestion costs.
As short run marginal cost figures are more relevant for internalisation purposes, they
are the main focus of this Handbook. Generally, marginal external costs are presented
in the same units as average external costs (e.g. €-cent/pkm, €-cent/tkm, €-cent/vkm).
For some externalities (e.g. air pollution, climate change) average and marginal cost figures
are (approximately) equal to the size of the externality and do not depend on the density of
the traffic flow. A car entering a dense traffic flow emits the same amount of air pollutant
emissions as a car entering a thin traffic flow, assuming that all other factors are equal
(location, speed, etc.). However, for other externalities (e.g. accidents, noise, and
congestion) the costs do depend on the density of the traffic flow. For example, a car
entering a road with free flow traffic, will cause marginal external congestion costs that are
significantly lower than the average external congestion costs. However, when a car enters
the traffic flow, at the moment the capacity of the road is almost met, it will cause
marginal external congestion costs that are significantly higher than the average costs.
Whether average or marginal external costs figures should be used depends on the scope
and objective of the assessment for which the figures will be applied. For assessments on
the internalisation of external costs, marginal cost figures should be considered when
internalisation is considered from an economic efficiency point of view (marginal social cost
pricing). However, from an equity point of view, it may be more interesting to see whether
vehicles are charges at their average costs (‘average cost pricing’), ensuring that the
transport sector or vehicle categories pay for the costs they impose on society. Average and
marginal external costs may also be used for other purposes, e.g. social cost benefit
analyses, cost-effectiveness analyses of other welfare economics analyses. In these cases, it
depends on the actual scope of such analyses whether the appliance of marginal or average
cost figures is preferred. For example, for a social cost benefit analysis of the realisation of
a new road, the noise costs can best be estimated by average cost figures (as there is no
existing traffic situation). On the other hand, for a social cost benefit analysis of an
extension of a road from two to four lanes, the use of marginal cost figures is preferred (as
the change in an existing traffic situation is assessed).
________________________________
3
In other words, average costs are calculated by dividing the total costs by the total transport performance.

Level of externality
As mentioned above, external costs are the difference between social and private costs.
Table 3 provides an overview of the social and external costs for the main cost categories
considered in this Handbook. The level of externality differs between these cost categories:
— For congestion and scarcity costs only the additional costs for other transport users and
society are considered external in this Handbook. Own costs (e.g. additional travel time
or fuel costs) are private costs and hence are not considered when estimating external
congestion costs. With respect to total and average congestion costs, it should be noted
that part of the costs are borne by the same group as those cause the congestion (the
so-called club effect). For example, the total external congestion costs of passenger
cars do include costs borne by users of passenger cars who are delayed because other
passenger cars have entered the traffic flow. As marginal costs are considered from an
individual transport user perspective, club effects do not occur.
— Part of the social costs of accidents are internalised by the transport user (i.e. as they
consider their own accident risk when taking a transport decision, it may be argued that
their own accident costs are internalised) or by insurances. As for congestion, part of
the total/average external accident costs are borne by the same group of agents who
cause the accident costs.
— For environmental costs the external and social costs are the same, except for the
situation in which part of the social costs are charged for. However, as we do not
consider transport charges and taxes in this Handbook (they are considered separately
in a parallel study, see the text box in Section 1), we can assume (in general) that social
and external environmental costs are the same.
Table 3 – Level of externality of various costs categories
Cost category Social costs External costs

Congestion costs All costs for traffic users and society (delay, unreliable Additional costs imposed on all
and scarcity costs travel times and/or arriving times, additional operation other transport users and society
costs, missed economic activities) caused by high traffic excluding own additional costs.
densities given the available capacity of the
infrastructure.
Accident costs All direct and indirect costs of an accident (material Part of the social costs that is
costs, medical costs, production losses, suffering and not considered in own and
grief caused by fatalities and injuries). collective risk anticipation and
not covered by (third party)
insurance.
Environmental All damages of environmental nuisances (e.g. health Part of the social costs that is
costs costs, material damages, biosphere damages, long term not considered (paid for).
risks).
2.3 General overview of valuation methodologies

Externalities are, in general, not traded on actual markets and hence no market prices are
available for them. Therefore, alternative valuation methodologies have to be applied to
quantify external costs. Several methodologies can be used for the valuation of
externalities. The main ones are the damage cost approach, the avoidance cost approach
and the replacement cost approach. These are discussed in more detail hereafter.

Damage cost approach
The preferred option by economists to value external costs is the damage cost approach
(Botzen & Van den Bergh, 2012). This approach values all damage experienced by
individuals as a result of the existence of an externality (e.g. health impacts due to traffic
noise). As market prices are often unavailable for the damage experienced, the willingness
to pay (WTP) of individuals to (partly) avoid the damage or the willingness to accept (WTA)
the damage, is used as an indicator of individual preferences.
There are several methods available for estimating the WTP, falling broadly into two
categories:
— Stated preference (SP) methods use questionnaires or experiments where respondents
are asked to provide their WTP (or WTA) to avoid the damage of the externality.
SP methods can take two forms: contingent valuation (through use of questionnaires or
surveys, where respondents are directly asked for their WTP for a certain good) and
choice experiments (where respondents are asked to pick their most favoured
alternative from different packages, and WTP is inferred indirectly). SP methods
directly measure the WTP and they also allow the researcher to control for all external
factors, such that purely the externality considered is identified. On the other hand,
SP methods depend very much on the survey/experiment design and the level of
information, and it suffers from the fact that it involves hypothetical expenditures only.
Also avoiding strategic behaviour of respondents is a main challenge of these kind of
studies.
— Revealed preference (RP) methods deduce the monetary value of externalities from
transactions on other economic markets, e.g. the real estate market. The most
commonly used RP method is the hedonic price method, which uses price differences on
the house market to estimate the WTP for the reduction of transport noise or emissions.
The main strength of RP methods is that it relies on actual market behaviour, where
individuals’ WTP for avoiding a specific externality can be observed. However, the
results from RP studies are sensitive to the conditions of the markets observed.
Furthermore, lack of knowledge of the market actors on the damage caused by the
externality may seriously affect the reliability of the results of RP studies.
In this Handbook damage costs are applied for several external cost categories, including air
pollution, accidents and noise.
Avoidance cost approach

An alternative way to value external costs is by applying the avoidance cost approach.
In this Handbook the CO 2 price that is used to calculate the external costs of climate change
is based on this approach. The avoidance cost approach determines external cost valuation
factors (i.e. shadow prices) by determining the cost to achieve a particular policy target
(e.g. EU CO 2 reduction targets). This is done by estimating an avoidance cost function,
which provides a proxy for the supply of environmental quality. It determines how much it
would cost to supply an additional level of environmental quality (e.g. reduction of one
additional tonne of CO 2). Based on this cost curve, the minimal cost required to meet the
policy target is estimated. The assumption is that this policy target reflects collective
preferences with respect to the externality concerned and hence, that the minimum cost to
reach this target is a good proxy of the (collective) WTP to avoid the damage caused by the
externality.

The avoidance cost approach is particularly useful when the damages caused by the
externality are uncertain and/or difficult to measure. In these cases the avoidance cost
approach may provide more reliable cost values in a relatively simple way. On the other
hand, the avoidance cost approach is often criticized due to policy targets not always being
a good reflection of the individual (or collective) preferences of citizens.
Replacement cost approach

The replacement or repair cost approach estimates the value of an externality based on the
costs of replacing/repairing the adverse impacts caused by the externality. In this Handbook
the replacement cost approach have been used to estimate the costs of habitat damage.
This approach is often used to value external costs for which no reliable damage or
avoidance cost figures are available. It may result in an overestimation, as it is not always
economically efficient to repair all damage. On the other hand, the replacement cost
approach may also underestimate the actual value of external costs, as it is not always
possible to replace/repair all damage.
2.4 Value transfer approach

The input and output values estimated in this Handbook are collected from various studies
which estimate Willingness to Pay (WTP) values for the different externalities under the
particular conditions of the location and in the specific time period of the studies. To be
used across the different Member States and countries considered in this Handbook, a value
transfer procedure is applied to convert the estimated values from the ‘study site’ to
‘other sites’. The value transfer procedure provides an alternative to carrying out valuation
studies in all the different Member States and countries and could fill in any gaps where
country or regional values are not available from primary sources4. The value transfer
approach can also be used to transfer the input and output values as presented in this
Handbook to other countries or other years.
Overview value transfer approaches

Different approaches exist to undertake a value (or benefit) transfer procedure (NEEDS,
2009):
— Unit Value Transfer:
• Simple unit transfer.
• Unit Transfer with income adjustments.
— Function Transfer:
• Benefit Function Transfer.
• Meta-analysis.
The unit value transfer procedure consists of transferring the primary data from the original
location directly to the new location. This can either be a direct transfer (simple unit
transfer) or with slight adjustments, such as exchange rate, inflation and income (unit
transfer with income adjustments). The benefit function transfer approach consists of
estimating a function that establishes the relationship between the unit value and the
characteristics at the original site in order to predict the values at another site.
Although theoretically superior due to it taking more information into account, the benefit
________________________________
4
Considerations should be made here on the purpose of the analysis, see Ready & Navrud (2006).

function transfer approach tends to be more complex to apply in practice, because it
requires information on each of the characteristics at the new site. 5 For example, the
willingness to pay to reduce noise may be estimated by figures on income, population
density and age. To approximate the willingness to pay for noise reduction in a different
region, statistics will need to be collected for income, population density and age.
Another possibility is to carry out a meta-analysis of several valuation studies to estimate a
common benefit function. This approach also presents its challenges such as the lack of
relevant information and studies. For these reasons, the unit value transfer with income
adjustment is preferred for its simple, transparent and reliable results 6. This was the
approach followed in the previous Handbooks and is applied in this revision.
Recommended approach: unit value transfer with income adjustments

Transferring the unit value from the original country to the remaining Member States and
countries considered in this Handbook requires the following adjustments which control for
differences across locations:
— Differences in prices. Controlling for differences in prices is crucial to minimise errors
when transferring values across locations. The recommended approach is to use
PPP-corrected exchange rates to take into account the cost of living. If appropriate,
adjustments can also be made in line with differences in living costs between regions
within the same country.
— Differences in income. A central issue when converting values between countries is to
consider differences in income. The common approach consists of multiplying the unit
values by the ratio of income in the policy country to income in the study country as
such:
𝜀
𝐼
𝑊𝑇𝑃𝑃𝑆 = 𝑊𝑇𝑃𝑆𝑆 ( 𝐼𝑂𝑆 ) ,
𝑠𝑠
Where 𝑊𝑇𝑃𝑃𝑆 is the WTP transferred to the study site, 𝑊𝑇𝑃𝑆𝑆 is the WTP at the study
site, 𝐼𝑂𝑆 and 𝐼𝑠𝑠 are income at the other and study sites, and 𝜀 is income elasticity of
WTP. Income is defined as PPP-adjusted GDP/capita in this Handbook7 . For the income
elasticity a value of 0.8 is recommended, indicating that environmental goods can be
considered normal goods. This value of the income elasticity is based on an extensive
meta-analysis of the OECD, which concludes that the income elasticity for the WTP of
environmental and health related goods falls between 0.7 and 0.9.
— Other differences. Input and output values can be further adjusted based on the
specific characteristics of the externality. For example, accident costs should be
adjusted according to accident risk rates. The specific value transfer procedure carried
out for each type of externality is discussed in the relevant chapters.
________________________________
5 Associated to challenges such as low explanatory power due to omitted variables; extrapolating outside range of
the data; variation among individuals not the same as variation among countries; functional form choice, etc.
6 Bateman et al., 2002; OECD, 2011a; Czajkowski et al., 2017.
7
Other income indicators could be used alternatively. Czajkowski and Ščasný (2010) found that using site-specific
measures of income outperforms transfers based on GPD per capita. Ready & Navdrud (2006) provide a
discussion on the key considerations. We have chosen GDP/capita as this is a widely available indicator and used
by several studies, including the previous Handbook.

Unit value transfer is not only used to transfers input and output values between countries,
but also over time. The following adjustments have to be made for this purpose:
— Difference in price level. Consumer Price Indexes (CPI) should be used for this purpose,
which produces conservative estimates.
— Difference in income. Income levels generally increase over time. This implies that the
WTP also increases, assuming that externalities can be considered as normal economic
goods. Therefore the same correction used for the value transfer between countries can
be used. It may be argued that the income elasticity for temporal transfers is lower
than for spatial transfers, particularly as there may be diminishing marginal returns on
the WTP for health improvements in developed countries8. However, as there is no clear
evidence for this statement in the literature, we recommend to apply the same
elasticity for spatial and temporal value transfers (i.e. 0.8).
________________________________
8
As the average life expectancy of people increase over time, it may be that the WTP to increase this life
expectancy even longer may decrease. Particularly as the probab ility of additional life years with bad health
conditions grows as life expectancy increases.

3 Accident costs
3.1 Introduction
Accidents occur in all forms of traffic and result in substantial costs, consisting of two types
of components: material costs (e.g. damages to vehicles, administrative costs and medical
costs) and immaterial costs (e.g. shorter lifetimes, suffering, pain and sorrow).
Market prices can be used to calculate material costs, however, no such market prices exist
for immaterial costs. In addition, a part of the total accident costs are already internalised,
for example through insurance premiums or through accounting for risks that are well
anticipated.
In this chapter we provide an overview of the recommended approaches to value external

accident costs. In Section 3.2 we first briefly discuss the definition and scope of accident
costs. The total and average accident costs are presented in Section 3.3, and the marginal
accident costs are the topic of Section 3.4. Finally, the robustness of the accident cost
figures presented in this chapter is analysed in Section 3.5. More detailed information on
accident costs can be found in Annex B.
3.2 Definition and scope

Although there is no harmonised definition of external accident costs, we define them as
the social costs of traffic accidents that are not covered by risk oriented insurance
premiums in this edition of the Handbook. The insurance system therefore determines
the share of the accident costs that are considered internal. Any costs that are covered
by insurances are therefore not considered external to the individual (i.e. they are
internalised). Costs that are not covered by insurances are external. This approach is in line
with earlier editions of the Handbook, although it is important to acknowledge that studies
at the national level may use a different distinction. For a full discussion we refer to
Annex B.
There are five main components of accident costs:

— Human costs: This is a proxy for estimating the pain and suffering caused by traffic
accidents in monetary value. In cases of injuries it covers the victim’s pain and
suffering, in cases of fatalities it covers the victim’s loss of utility. Traffic participants
are assumed to be aware of the fact that their decision to enter the traffic may result in
an accident (they internalise this risk). Therefore, their own human costs are considered
internal to them, once they have made the decision to enter the traffic. However, they
consider the human costs of others that may result from their own transport decision as
external to them.
— Medical costs: These are the costs of the victim’s medical treatment provided by
hospitals, rehabilitation centres, general practitioners, nursing homes, etc. as well as
the costs of appliances and medicines. The medical costs cover the time period from
the moment of the accident until complete recovery from the injury or, in the case of
fatal accidents, death. In many cases a part of these costs is already internalised

through health insurance premiums.9 In this Handbook we assume 50% of the medical
costs are external.
— Administrative costs: These are the costs covering the expenses of the deployed police
force, fire service and other emergency (non-medical) services that assist at the crash
location site. In addition, costs related to the administration of justice such as legal
costs, the costs of prosecution of offenders and the costs of lawsuits and insurance are
incorporated into this category. Lastly, administrative costs related to vehicle, health or
other insurance is also included in this category. This component is assumed to be partly
internalised by traffic participants in the form of insurance. In this Handbook we assume
30% of the administrative costs are external.
— Production losses: After an accident victims are not directly capable of returning to
work, and in some cases may never return to work. These costs consists of the net
production losses due to reduced working time and the human capital replacement
costs. Not being able to carry out non-market work such as household work or
volunteering is also incorporated in this cost component. This component is assumed to
be partly internalised by traffic participants in the form of insurance. Based on (ARE,
2018), we assume that 55% of the gross production loss (as reported by SafetyCube) can
be regarded as external.
— Material damages: This consists of the monetary value of damages to vehicles,
infrastructure, freight and personal property resulting from accidents. This component
is assumed to be fully internalised by traffic participants through insurance.
— Other costs: This category covers the costs of congestion resulting from road crashes,
vehicle unavailability and funeral costs. We will not take this cost category into account
as a large part of the other costs are already incorporated in other external cost
categories investigated in this study or are not considered external.10
It is important to realise that costs related to the prevention or avoidance of crashes are not
included in accident costs. Prevention costs, e.g. police enforcement costs, are not included
because they are not a (direct) consequence of road crashes, but are intended to decrease
the number of crashes (Wijnen, et al., 2017). Furthermore, these are (partly) included in road
infrastructure costs in (CE Delft et al., Forthcoming).
Fatalities and injuries

In this study, we present the external accident costs for all five modes of transport (road, rail, aviation, IWT and
maritime). The victims of traffic accidents are classified into one of three categories: fatalities, severe injuries
and slight injuries. This is done based on the definitions in (UN ECE, 2011).
— Fatality: Any person killed immediately or dying within 30 days as a result of an injury sustained as a result
of an accident.
— Serious injury: A person who sustained an injury as a result of the accident and who was hospitalised for a
period of more than 24 hours.
________________________________
9 Please note that there are large differences in the health insurance systems across countries in the EU, e.g. in
terms of deductibles. In general we have assumed that health insurance is a way to (partly) internalise the
health costs, without taking the nuances of the different national health care systems into account.
10
For instance, congestion costs fall under ‘other costs’ but are already included as anoth er category in this
handbook. Other costs such as funeral costs may already be (partly) insured, which no longer renders them
(fully) external.

— Slight injury: A person who sustained an injury as a result of the accident but does not fall under the
definition of serious injury.
However, it is important to note that EU countries started collecting data on injuries from traffic accidents using
a new common definition in 2014. This scale, the Maximum Abbreviated Injury Scale (MAIS), is based on the most
severe injury in the Abbreviated Injury Scale (AIS) classification system commonly used by medical professionals.
The use of this scale will result in it being easier to classify accident data according to the right category, as this
is the classification system adhered to by hospitals in Europe. This also means the data between countries will
become more harmonised and consistent. The MAIS represents the most severe injury obtained by a casualty
according to the AIS. Serious traffic injuries are now classified as injuries scoring 3 or more on the medical
Maximum AIS (MAIS3+). Therefore, for traffic injury victims MAIS 1 and MAIS 2 are considered slight injuries.
Table 4 – Comparison of old and new EU traffic injury definitions
AIS scale Example Old definition

AIS 1 Minor Sprained ankle Slight injury
AIS 2 Moderate Closed fracture
AIS 3 Serious Open fracture Serious injury
AIS 4 Severe Amputation
AIS 5 Critical Ruptured liver with tissue loss
AIS 6 Maximum Unsurvivable injury Fatality
However, figures on the number of seriously injured people according to the new definition are not yet available
for all countries (and not at the same level of detail comparable to the old definition). Therefore, the
estimation of external accident costs in this Handbook will be based on accident figures according to the old
definition. However, we do propose some input values for the new definition of injuries in Annex B.3.2.
It is worth noting that the number of fatalities and injuries in official statistics only
represent reported accidents. However, for road accidents in particular, a portion of the
total accidents go unreported. Therefore, the official road accident data ought to be
corrected for these unreported accidents. The correction factors that are applied in this
study are shown in Table 5, and are based on (HEATCO, 2006), a large EU study, and a
Swiss (Ecoplan, 2002) study. One minor adjustment was made to these correction factors,
the underreporting rate for fatalities was 1.02 in HEATCO & Ecoplan, but a study by
(Ecoplan & Infras, 2014) revealed that there are no longer unreported fatalities from
accidents in Switzerland. We assume that this also holds for the rest of Europe, therefore,
the rate for fatalities has been adjusted to 1.00 (i.e. no underreporting). Previous editions
of the Handbook used the correction factors presented in Table 5 for serious and slight
injuries, but used a factor of 1.02 for fatalities.
Table 5 – Correction factors to correct for underreporting of accidents
Fatalities Serious injury Slight injury

Car, LCV, HGV, bus 1.00 1.25 2.00
Motorbike 1.00 1.55 3.20
Source: (HEATCO, 2006) & (Ecoplan, 2002).
The underreporting factors in Table 5 shows that the underreporting rate differs depending
on the vehicle type and the severity of the accident, with more vulnerable road users such

as motorcycles having higher unreported accidents than less vulnerable road users
(e.g. HGVs, buses).
Although the reporting factors are old, there are no more recent studies looking at
underreporting rates at the international level. Studies at the national level confirm that
the factors displayed in Table 5 are reasonable. For instance, studies looking at Korea show
that there are four times as many road traffic victims reported by data from insurance
agencies than there are victims reported by the police (OECD, 2016; Park, 2008).
Other studies at the national level report slightly lower underreporting rates than the
Korean values, e.g. 40-45% underreporting in Australia (Rosman, 2001) and 36%
underreporting in the UK (Mackay, 2003). One recent study even looked at vulnerable road
users in particular and revealed that only 35% of serious injuries with motorcycles are
reported, whereas only 10% of slight injuries are reported (Janstrup , et al., 2016). All in
all, it appears that even though the correction factors from HEATCO are old, there are no
indications that they are outdated.
For the other transport modes such as rail, inland waterway, maritime or aviation, we do
not use correction factors, as accidents occurring in these transport modes are much less
likely to go unnoticed.
3.3 Total and average accident costs
3.3.1 Methodology
Total and average accident costs are calculated using a top-down approach, starting with
total accidents and then allocating them to different vehicle types. Figure 1 illustrates the
corresponding methodology for calculating accident costs for road, rail, inland waterway,
maritime and aviation.
Figure 1 – Methodology total and average accident costs

The main input values that are used are the number of casualties per vehicle category and
the costs per category. For the number of casualties per vehicle category in road transport,
we use detailed statistics from the CARE database which also provides information on the
other parties involved in the accident. For the other modes, data to that level of detail is
not available. The data provided for rail accidents already excluded suicides.
The cost per casualty consists of six components, of which four (human costs, production
costs, medical costs and administrative costs) are (partly) external. By multiplying the
number of casualties by the cost per casualty and deducting the transfers from liability
insurance systems and gratification payments the total external accident costs are
estimated.
Allocation to the different vehicle categories is carried out according to damage potential
(intrinsic risk) if the accidents occur within one transport mode. This method is used in
studies such as (CE Delft & VU Amsterdam, 2004; CE Delft, INFRAS & Fraunhofer ISI, 2011).11
The method involves allocating the victims in the opposing vehicle to the other vehicle type
involved in the crash and vice versa. For instance, if a fatal accident occurs between a HGV
and a car, and the driver of the HGV sustains a slight injury whereas the driver of the car
dies, the cost of the fatality is allocated to the truck, whereas the cost of the slight injury
is allocated to the car.
For accidents between different modes, such as accidents involving a train and a car, the
casualties are allocated to the party responsible for the accident. In this study, these types
of accidents only occur between road and rail, at level crossings. For these types of
accidents it is known that they are almost always caused by the road user (Jonsson &
________________________________
11
There are two other approaches which could alternatively be used, which are elaborated in more detail in
Annex B.

Björklund, 2015) 12. These steps result in the total and average accident costs per vehicle
category in each country.
3.3.2 Input values
Accident statistics
The accident statistics that are used for road transport are taken from the EU’s Community
Road Accident Database (CARE). This highly detailed database provides information on
fatalities, serious injuries and slight injuries, in which vehicle these victims were seated,
and which other vehicle was involved in the accident at the country level. Statistics are
further split according to road type (urban, rural, motorway and unknown). Because of the
different vehicle types within this transport mode, such detailed data is needed in order to
allocate the costs to the vehicle types according to the vehicle type’s damage potential
(intrinsic risk).
The previous Handbook did not include data on the accident costs of the other transport
modes. In this Handbook we do estimate these costs, although the available data is not as
detailed as the CARE database is for road transport. For these modes, accidents are not as
frequent as for road transport, which is why we use averages over 5 years (2012-2016).
For rail transport, the accident statistics were provided by the European Union Agency for
Railways (ERA). However, the ERA does not collect figures for slight injuries, so the rail
external accident costs are only based on fatalities and serious injuries, and will therefore
be an underestimate. Suicides were excluded from the rail accident data by the ERA.
For aviation the total statistics for the EU28 are provided by European Aviation Safety
Agency, from which an accident rate per movement could be deduced. By combining this
with information on the number of flight movements per airport, the total number of
casualties could be calculated per airport. Please note that these are not the actual
casualties that occurred in that period, but a proxy for the risk. For inland waterways the
accident rate per 1,000 vkm is based on data from the Dutch Department for Waterways
and Public Works. These accidents are also a proxy for the risk of inland waterway
transport, rather than the actual accidents that occurred in the period considered.
For maritime transport, accident statistics were provided by the European Maritime Safety
Agency. These are the actual accidents that occurred over the time period considered.
Costs per casualty

The various components of the costs per casualty are largely based on SafetyCube (2017),
which estimates standard values for each of the cost components according to the methods
outlined in the international guidelines developed by (Alfaro, et al., 1994). The values
presented in (Wijnen, et al., 2017) are the social costs of accidents, not the external costs
of accidents, and therefore need to be corrected.
The only cost component in our calculation that is not based on SafetyCube is the human
cost, the largest part of accident costs. The human costs are valued based on the Value of
Statistical Life (VSL), which we base on the (OECD, 2012). A detailed discussion on the VSL
is presented in Annex A. The EU28 VSL used is € 3.6 million. To avoid double counting with
gross production loss, the consumption loss needs to be deducted from the VSL to reach
________________________________
12
Please note that if the damage potential approach would be used instead of the responsibility perspective this
would result in accident costs that are only very minorly different from the responsibility approach.

human costs for fatalities (see Section B.3.2). Consumption loss is calculated by combining
data on the consumption expenditure per capita per annum with the amount of life years
lost due to an accident (on average 42 years). This results in an EU28 average consumption
loss for a fatality of € 668,000. Therefore, the human costs of fatalities for the EU28 is
€ 2.9 million. The human costs of injuries are valued at 13% and 1% of the VSL respectively
for serious and slight injuries (HEATCO, 2006). No consumption loss is deducted from the
values for injuries.
Table 6 – External accident cost components per casualty for the EU28 (€2016 )
Human costs Production loss Medical costs Administrative Total external cost
costs per casualty
Fatalities 2,907,921 361,358 2,722 1,909 3,273,909
Serious injuries 464,844 24,055 8,380 1,312 498,591
Slight injuries 35,757 1,472 721 564 38,514
Table 7 – External accident costs components per casualty for the EU28 (€2016 )
Human costs Production loss Medical costs Administrative costs

Fatality Serious Slight Fatality Serious Slight Fatality Serious Slight Fatality Serious Slight
injury injury injury injury injury injury injury injury
EU countries
EU28 2,907,921 464,844 35,757 361,358 24,055 1,472 2,722 8,380 721 1,909 1,312 564
AT 3,202,976 532,685 40,976 393,002 26,161 1,600 2,960 9,114 784 2,076 1,427 614
BE 3,183,342 513,206 39,477 394,570 26,266 1,607 2,972 9,151 788 2,084 1,433 616
BG 1,553,981 226,042 17,388 172,290 11,469 702 1,298 3,996 344 910 626 269
HR 2,308,933 334,147 25,704 230,091 15,317 937 1,733 5,336 459 1,215 836 359
CY 1,504,105 285,078 21,929 319,468 21,266 1,301 2,406 7,409 638 1,687 1,160 499
CZ 2,789,348 406,295 31,253 236,108 15,717 962 1,778 5,476 471 1,247 858 369
DK 3,497,489 576,978 44,383 485,139 32,295 1,976 3,654 11,251 968 2,562 1,762 757
EE 2,653,497 391,365 30,105 264,696 17,620 1,078 1,994 6,139 528 1,398 961 413
FI 2,798,583 475,746 36,596 444,438 29,585 1,810 3,348 10,307 887 2,347 1,614 694
FR 2,721,569 449,900 34,608 395,712 26,342 1,612 2,981 9,177 790 2,090 1,437 618
DE 3,067,253 503,575 38,737 383,018 25,497 1,560 2,885 8,883 765 2,023 1,391 598
EL 2,026,599 328,432 25,264 296,552 19,741 1,208 2,234 6,877 592 1,566 1,077 463
HU 2,545,519 363,132 27,933 213,101 14,186 868 1,605 4,942 425 1,126 774 333
IE 4,681,432 710,688 54,668 398,560 26,531 1,623 3,002 9,243 796 2,105 1,448 622
IT 2,888,866 468,373 36,029 354,695 23,611 1,444 2,672 8,226 708 1,873 1,288 554
LV 2,091,145 314,437 24,187 244,097 16,249 994 1,839 5,661 487 1,289 887 381
LT 2,472,609 368,941 28,380 221,664 14,756 903 1,670 5,141 442 1,171 805 346
LU 6,048,974 955,627 73,510 436,719 29,071 1,779 3,289 10,128 872 2,307 1,586 682
MT 1,726,048 292,090 22,468 294,266 19,589 1,198 2,216 6,824 587 1,554 1,069 459
NL 3,144,379 506,503 38,962 400,833 26,683 1,632 3,019 9,296 800 2,117 1,456 626
PL 2,209,087 322,671 24,821 201,159 13,391 819 1,515 4,665 402 1,062 731 314
PT 2,249,642 359,065 27,620 287,703 19,152 1,172 2,167 6,672 574 1,520 1,045 449
RO 2,257,137 322,445 24,803 183,549 12,219 747 1,383 4,257 366 969 667 287
SK 2,602,350 381,986 29,384 240,873 16,034 981 1,814 5,586 481 1,272 875 376
SI 2,127,862 337,228 25,941 293,677 19,549 1,196 2,212 6,811 586 1,551 1,067 459
ES 2,690,282 427,815 32,909 325,423 21,663 1,325 2,451 7,547 650 1,719 1,182 508
SE 2,819,502 476,827 36,679 470,659 31,331 1,917 3,545 10,915 939 2,486 1,709 735
UK 2,448,105 442,196 34,015 420,407 27,986 1,712 3,167 9,750 839 2,220 1,527 656

Human costs Production loss Medical costs Administrative costs
Fatality Serious Slight Fatality Serious Slight Fatality Serious Slight Fatality Serious Slight
injury injury injury injury injury injury injury injury
Non-EU countries
NO 2,860,780 523,348 40,258 535,129 35,622 2,179 4,031 12,410 1,068 2,826 1,944 836
CH 3,860,318 707,624 54,433 554,838 36,934 2,260 4,179 12,867 1,107 2,930 2,015 866
CAN-
AL 3,487,874 453,424 34,879 419,350 27,915 1,708 3,159 9,725 837 2,215 1,523 655
CAN-
BC 3,487,874 453,424 34,879 419,350 27,915 1,708 3,159 9,725 837 2,215 1,523 655
US-
CA 3,984,276 517,956 39,843 443,577 29,528 1,806 3,341 10,287 885 2,343 1,611 693
US-
MO 3,984,276 517,956 39,843 443,577 29,528 1,806 3,341 10,287 885 2,343 1,611 693
JP 3,400,821 442,107 34,008 409,621 27,268 1,668 3,085 9,500 818 2,163 1,488 640
3.3.3 Output values

Table 8 describes the total and average external accident costs in the EU as a whole for
road and rail transport. Costs at the country level are provided in the database.
Table 8 – Total and average external accident costs for land-based modes for the EU28
Transport mode Total costs EU28 Average costs

Passenger transport Billion € €-cent per pkm €-cent per vkm
Passenger car 210.2 4.5 7.2
Motorcycle13 21.0 12.7 13.3
Bus/Coach 5.3 1.0 18.9
Total passenger road 236.5
High speed passenger train 0.1 0.1 17.3
Conventional passenger train 2.0* 0.5 52.2
Total passenger rail 2.0
Total passenger transport 238.5
Freight transport Billion € €-cent per tkm €-cent per vkm
LCV 19.8 6.0 4.1
HGV 23.0 1.3 15.5
Total freight road 42.8
Freight train 0.3 0.1 34.1
Inland Vessel 0.1 0.1 86.3
Total freight transport 43.1
Total road, rail, inland
waterway 281.7
* Total costs without highspeed passenger trains (average costs for passenger train electric: incl. high speed
trains).
________________________________
13
Please note that the costs of motorcycles does not include the costs for mopeds. Moped accidents are roughly
1% of EU fatalities and 2-3% of EU injuries. Although the CARE database has statistics available on moped
fatalities, allocation to mopeds cannot be carried out as there is no transport performance data specifically for
mopeds available.

Average costs are calculated by dividing the total costs by the transport performance data.
Motorcycles cause by far the highest average external accidents costs per pkm. The CARE
database revealed that motorcyclists are involved in a relatively high number of accidents.
This despite the fact that they drive relatively fewer kilometres with a lower occupancy
rate. Therefore, this results in a higher accident costs for motorcyclists compared to cars.
Table 9 illustrates the average external accident costs of both passenger and freight
aviation. Passenger aviation values are provided per LTO, passenger and pkm. The average
costs of freight aviation are provided per LTO, tonne and tkm. Costs at the individual
airport level are provided in the database.
Table 9 – Total and average external accident costs for aviation for 33 selected EU airports
Transport mode Total costs Average costs

Passenger aviation Million € €/LTO €/passenger €/tonne €-cent/pkm
Short haul 0.04
Medium haul 75.01 22.95 0.18 0.81 0.01
Long haul 0.001
* Costs per pax are including the complete flight (not only the half -way principle).
Table 10 presents the average external accident costs of maritime transport. For ferries,
the costs are calculated per port call, passenger and pkm. For freight maritime transport,
the average external costs are provided per port call, tonne and
tkm. Costs at the individual port level are provided in the database.
Table 10 – Total and average external accident costs for maritime transport for 34 selected EU ports

Passenger transport Million € € per port call € per million passengers
Passenger ship 3.3 26 40,996
Freight maritime transport Million € € per port call € per million tonnes
Freight ship 63.3 318 36,524
3.4 Marginal accident costs
3.4.1 Methodology
Marginal accident costs are only calculated for road transport. For all other modes of
transport the marginal accident costs are considered to be equal to the average costs.
This is because the other modes are scheduled services, this implies that the accident
risk is less dependent on the amount of traffic for these modes. Figure 2 illustrates the
methodology for calculating marginal accident costs for road transport.
Figure 2 – Methodology marginal external accident costs

The marginal accident costs represent the extra costs that adding an extra vehicle to the
traffic flow brings. The main input values for marginal accident costs are the accident risk
per vehicle type and road type, the costs per casualty and the risk elasticity. The costs per
casualty are the same as those used for the calculation of total and average costs.
Combining the accident risk with assumptions on the degree of risk internalisation, the
external costs per casualty and the risk elasticity allows us the calculate the marginal
external accident costs per vehicle category (see Annex B.4 for more details).
3.4.2 Input values

The input values used to calculate marginal external accident costs are largely comparable
to the ones used for total and average accident costs, e.g. the costs per casualty. For these
input values we refer to Section 3.3.2.
Risk elasticity
The risk elasticity represents how much the accident risk on a certain road type increases
with a 1% increase in traffic measured in vkms. It is realistically expected to vary with the
road type, road conditions and traffic intensity, although only very few sophisticated
estimates for the risk elasticity exist, and only for individual countries. For instance, a
study by (Sommer et al., 2002) looking at Switzerland found risk elasticities of -0.5, -0.25
and -0.62 for motorway, urban and other roads respectively. Rougher estimates of -0.25
irrespective of the road type were used in (Ricardo-AEA, TRT, DIW Econ & CAU, 2014;
Lindberg, 2001). All suggested risk elasticities are negative, implying that although an
increase in traffic increases the risk of an accident, the risk of an accident injury or fatality
decreases (Hesjevoll & Elvik, 2016). However, one could argue this is not necessarily true
for urban roads, where congestion and already slow traffic imply that the addition of an
extra vehicle will not lead to a significant change in the accident risk (i.e. risk elasticity of
0). This is the approach taken in (CE Delft & VU, 2014). Therefore, the recommended

approach in this study is to use a risk elasticity of 0 for urban roads, and
-0.25 for motorways and other roads.
Degree of risk internalisation

The degree of risk internalisation is important in determining the share of the human costs
that is internalised by road users. This factor differs per vehicle type, as some vehicles are
simply more vulnerable than others. Therefore, the best way to reach the degree of risk
internalisation is to calculate it from statistics from the CARE database. It is calculated by
dividing the number of fatalities inside a certain type of vehicle by the number of fatalities
in accidents involving this vehicle type (also counting victims inside other types of vehicles
involved in the accidents). This gives a good indication of a vehicles ‘vulnerability’
compared to other vehicle types. Please note that this approach implies that in cases where
there are more than one passenger in the vehicle (i.e. not only the driver) the human costs
of all these passengers are fully internalised by the (driver of the) vehicle. If a vehicle then
causes a fatal crash with another vehicle that has four passengers, the human costs of all of
the four passengers are fully external to the original vehicle. The value for the degree of
risk internalisation ranges between 0 and 1, with relatively lower values representing a
smaller share of the costs is internalised. People in passenger cars and on motorcycles are
expected to have a relatively higher share of the costs internalised (value closer to 1) than
people inside HGVs.
Table 11 – Degree of risk internalisation for different vehicle types
Vehicle type Risk internalisation factor

Passenger car 0.61
Motorcycle 0.93
Bus 0.16
Coach 0.16
LCV 0.28
Heavy Goods Vehicle 0.14
It is important to note that the risk elasticity (𝐸) and the degree of risk internalisation (𝜃)
combined lead to interesting results. If 𝜃 − 𝐸 > 1, the marginal costs are negative (Ricardo-
AEA, TRT, DIW Econ & CAU, 2014). This implies that with each vehicle entering the road the
average accident costs decreases. If 𝜃 − 𝐸 < 1, the marginal costs are positive and the
accident costs always increases with each additional vehicle. With a risk elasticity set at
-0.25 (motorways and other roads), this implies that heavy goods vehicles, busses, coaches,
LCVs, passenger cars and other vehicles all have positive marginal costs. In this case,
negative marginal costs exist for motorcyclists.
3.4.3 Output values

Table 12 describes the marginal external accident costs for road transport in the EU28.
Costs at the country level are provided in the database. The marginal external accident
costs of the other four transport modes are identical to the average external accident costs.
Table 12 – Marginal external accident costs road transport for the EU28
Vehicle type Motorway Urban road Other road

Passenger transport (€-cent per pkm)

Motorcycle -0.65 4.42 -3.21
Bus/coach 0.05 0.80 0.19
LCV (€-cent per vkm)
LCV 0.37 0.76 0.84
Freight transport (€-cent per tkm)
HGV 0.07 0.10 0.13
The interpretation of negative marginal external accident costs (i.e. for motorcyclists) is
somewhat confusing. Because traffic tends to slow down with each extra driver, the traffic
becomes safer for all other traffic participants. However, the extra road user has a higher
accident risk (compared to no accident risk if he decides not to take part in traffic).
The moment where the risk of an accident on other traffic users reduces by less than the
increase in external accident risk by the extra traffic users, negative marginal external
costs arise. This also explains why negative marginal external costs arise almost exclusively
for vulnerable road users such as motorcyclists, as they have almost fully internalised their
own risk (see Table 11). Please note that the costs presented here are marginal external
accident costs, and that even though they may occasionally turn negative, this does not
mean that marginal accident costs are negative.
3.5 Robustness of results

We have attempted to calculate the accident costs according to the most recent and high
quality evidence and methods. In this Handbook, the accident costs calculations are much
more detailed than in previous editions. Nonetheless, there are a few aspects that merit a
point of discussion regarding the robustness of the results presented in this chapter.
Firstly, the human costs are the largest component of the accident costs. These costs are in
turn highly dependent on the VSL that is used. We have conducted a detailed review of the
literature on the VSL and found the range of values is very large. In this Handbook we have
chosen to use the VSL as presented by the (OECD, 2012) as it provides the most recent high
quality evidence on the VSL to our knowledge. Nonetheless it is important to emphasize
that any estimate of the VSL remains uncertain. Use of the OECD VSL implies that the VSL is
significantly higher in this edition of the Handbook than in previous editions (see Annex A
for a full discussion), and, in turn, raises the human cost component of accident costs.
A second important uncertainty regards the percentage of the accident costs that transport
users internalise in their transport decision. For the external part of human costs, we have
made the assumption that one’s own human costs are internalised once the decision to
enter the transport is made, whereas the human costs of people in the other vehicles are
considered completely external. For the other cost components, the chosen methodology
implies that costs that are insured are fully internalised (see discussion in Annex B).
Although there is a discussion in the literature whether insurances can be seen as a way to
internalise costs, data limitations14 imply that other methodologies are not feasible.
The percentage that is internal for medical costs and administrative costs is more
uncertain, due to highly diverging values found in the literature at the country level.
When looking at the magnitude of these costs in comparison to the total external accident
________________________________
14
This is particularly relevant in terms of large differences in the structure of different insurances in different
countries, even within the EU.

costs, the sensitivity of the total costs to the percentages used is relatively small.
Varying the percentage of medical and administrative costs that is external to 100% instead
of 50% (as assumed in this Handbook) only changes the total accident costs by 0.2-4%,
depending on the severity of the casualty. For production loss, the percentage of costs that
is external (55%) is based on one value from the literature, as there is almost no literature
available on this topic. This means that 45% of production loss is covered through some form
of insurance. Because production loss is the second largest cost component the percentage
of production loss that is assumed external significantly influences the total accident costs.
If production loss is assumed to be fully internalised, accident costs would be 4-11% lower
than presented in this chapter. If production loss is assumed to be fully external, accident
costs would be 3-9% higher than presented in this chapter.
Thirdly, one recent study indicated that deducting the consumption loss from the human
costs to avoid double counting with the production loss should no longer be used in the
calculation of external accident costs (Ecoplan, 2016). This is based on another study which
concluded that there are no indications that own consumption is included in the WTP for a
statistical life for Switzerland (B,S,S. Volkswirtschaftliche Beratung AG, 2015). Up until
now, a cautious assumption was made that it was, which implied that net production loss
(= gross production loss – consumption loss) should be used to avoid double counting. This is
also the approach we have taken in this Handbook. Therefore, we have not changed the
method compared to the previous editions of the Handbook. If we were to change the
method, it would imply accident costs would increase by 20% per fatality, although the
accident cost per serious or slight injury would not change. All in all, although we admit the
(B,S,S. Volkswirtschaftliche Beratung AG, 2015) study has its merits, we believe further
research is needed to confirm whether or not one’s own consumption is taken into
consideration when the WTP for a VSL is elicited.
Last but not least the results for road transport are affected by the transport performance
data used. As explained in Section 1.3.4, in this study we use data from Eurostat, following
the nationality principle, i.e. transport activity is allocated to countries where the vehicle
is registered. The use of these data affects the results of this study, since the scope of
these data differs from the scope of the accident data, which is in line with the territorial
principle. Particularly the results for HGVs may be significantly affected at country level.
For example, in countries with a lot of transit traffic (e.g. Austria) a significant part of the
accidents should be allocated to foreign vehicles. By using transport performance data
based on the nationality principle, transport activity of these foreign vehicles is not taken
into account in the calculations.

4 Air pollution costs
4.1 Introduction
The emission of air pollutants can lead to different types of damages. Most relevant and
probably best analysed are the health effects due to air pollutants. However, other
damages are also relevant, such as building and material damages, crop losses and
biodiversity loss.
Air pollution costs are one of the external cost categories that has been analysed the most.
Since the nineties a broad range of international studies and research projects have been
conducted, particularly on European level. In the last few years, there haven’t been many
large international studies covering the entire impact pathway from emission to impact and
costs. However, epidemiological research has carried on, investigating the dose-response-
relationship between the exposure of air pollutants and the associated health risks.

The present Handbook covers the following four types of impacts caused by the emission of
transport related air pollutions:
— Health effects: The inhalation of air pollutants such as particles (PM10, PM2.5) and
nitrogen oxides (NO x) leads to a higher risk of respiratory and cardiovascular diseases
(e.g. bronchitis, asthma, lung cancer). These negative health effects lead to medical
treatment costs, production loss at work (due to illness) and, in some cases, even to
death.
— Crop losses: Ozone as a secondary air pollutant (mainly caused by the emission of NO x
and VOC) and other acidic air pollutants (e.g. SO 2, NOx) can damage agricultural crops.
As a result, an increased concentration of ozone and other substances can lead to lower
crop yields (e.g. for wheat).
— Material and building damage: Air pollutants can mainly lead to two types of damage to
buildings and other materials: a) pollution of building surfaces through particles and
dust; b) damage of building facades and materials due to corrosion processes, caused by
acidic substances (e.g. nitrogen oxides NO x or sulphur oxide SO 2).
— Biodiversity loss: Air pollutants can lead to damage to ecosystems. The most important
damages are a) the acidification of soil, precipitation and water (e.g. by NO x, SO2) and
b) the eutrophication of ecosystems (e.g. by NO x, NH 3). Damages to ecosystems can lead
to a decrease in biodiversity (flora & fauna).
4.3 Total and average air pollution costs
4.3.1 Methodology
Total and average air pollution costs are calculated by a bottom-up approach. Figure 3
illustrates the methodology used.

Figure 3 – Methodology total and average costs of air pollution
Emissions Transport Cost factors

factors performance - health costs (€/t)
Input values
- emissions - veh-km - non-health costs (€/t)

in t/veh-km
Total Emissions
- emissions in t (per
pollutant and area type)
Calculations
Total external costs of air pollution per transport mode

per country
Allocation
Average costs per pkm and tkm per mode
There are two main types of input values: the emissions and the cost factors per tonne of
pollutants.
For the emissions, there are two different approaches. For the total and average costs, the
emissions are calculated by using average emission factors per vehicle type and country
(e.g. for road transport from the COPERT database). The emission factors applied are on
the same level of differentiation as the transport data used. Total emissions are derived
from the emission factors (tonne of pollutant per vkm) and the transport performance data
(e.g. vkm), leading to a consistent set of emissions, that are in line with the emission
databases (e.g. COPERT) and the official transport statistics from the EU (Eurostat). The
resulting total emissions have been cross checked with the total emission database from the
European Monitoring and Evaluation Program under auspices of the European Environmental
Agency (EMEP/EEA). The overall results are well comparable for the main pollutants (NO x,
PM), although there are some differences (above all for NMVOC), as a result of different
transport data and emission factors from the different sources (COPERT and Eurostat vs.
EMEP/EEA). This difference cannot be avoided under the premise to take COPERT for
emission data and Eurostat for transport data as the main data sources. This issue is,
however, not relevant for any average and marginal cost factors, but only for the total
costs. In the following section, an overview of the main data sources is presented.
The second type of input value are the cost factors per pollutants. The cost factors have
been calculated in detail, based on the NEEDS approach, also taking into account the latest
results from other studies (e.g. (UBA, 2018), (Rabl, et al., 2014), (OECD, 2014)). This has
been done for the EU28 and a limited number of EU Member States in an on-going study by
CE Delft. In the study at hand, this set has been extended to all member States and also for
emissions from other sources. The following section briefly explains the methodology
followed for calculating the cost factors per pollutant.

Methodology to derive damage cost factors
The method used for damage cost estimation is the same approach as followed in the
Handbook Environmental Prices (CE Delft, 2018). This Handbook gives a damage cost
estimation for over 2,500 pollutants. It is based on a combination of two models:
— Economic damage cost estimates, as performed in NEEDS (2008).
— Lifecycle Assessment, as performed in RECIPE (2013).
Both models have been adjusted to the most recent insights. For the present project
especially the first model, the NEEDS model, is relevant. The core of the NEEDS-project is
an Impact-Pathway model (EcoSense) that estimates the relationship between emissions and
eventual impacts (see Figure 4). The Impact Pathway Approach (IPA) has been used in
several international research projects initiated by the European Commission, starting with
the original ExternE study implemented in the mid-1990s. We have adopted this model to
reflect the most recent insights on the relationship between emissions and damage.
The starting point of the quantification of the cost factors is the NEEDS (2008) results, as
they have been published in e.g. Desgauilles et al (2011) and further elaborated in Rabl et
al. (2014). Within the NEEDS model, the impact-pathway approach is followed, in which an
emission — through dispersion — results in an intake (immission) at receptor points.
Figure 4 – The Impact Pathway Approach for calculating air pollut ion costs
Source
(specification of site and technology)
Emission
(e.g. kg/yr of particulates)
Dispersion
(e.g. atmospheric dispersion model)
Increase in concentration at
receptor sites
(e.g.μg/m3 of particulates in all
affected regions)
Dose-response function
(or concentration-response function)
Impact
(e.g. cases of asthma due to ambient
concentration of particulates)
Monetary valuation
Cost
(e.g. cost of asthma)
Source: CE Delft, 2010, based on NEEDS, 2008.

Since 2009 there has been no further development of NEEDS and neither of the rival model
of CAFE-CBA (IIASA, 2014). It is also striking is that recent shadow price manuals for Ireland,
Belgium and Germany (under development) are still based on the NEEDS methodology owing
to its far greater transparency. However, one cannot simply take the NEEDS values and
apply them to air pollution because the estimation results are over a decade old and many
things have changed: background concentration levels, knowledge about impacts from
pollution and the valuation framework. For that reason, adaptations to the NEEDS
framework must be made. This is possible since we have the possession of a great deal of
modelling outcomes from the NEEDS model so that we can make required changes to reflect
more recent insights.
In total, to recalculate cost factors for air pollutants in the present study, five
‘adjustments’ (i.e. update calculations) were made to the NEEDS results.
These adjustments are broadly the same as in the Environmental Pricing Handbook
(CE Delft, 2018), but they are now applied to the EU context. These five adjustments can
be described as follows:
1. Concentration Response Functions (Step 3 in the figure above) have been adapted to the
WHO (2013) study. The taken steps are described in Annex C.
2. The population size and population structure (age cohorts) is based on the most recent
data from Eurostat.
3. The influence of the background concentration is estimated on the basis of the
relationship between damage and emissions for various emission scenarios from NEEDS
(2008). On this basis, by letting all other factors remain the same, we can estimate the
impact of a change in emissions on the harmfulness of these emissions. This harmfulness
is then the result of the change in the background concentration.
4. The valuation has been adjusted to the most recent insights with respect to valuation.
For human health we refer to Annex A. The change in valuation of ecosystems and
buildings, has been elaborated in more detail in Annex C.
5. Finally, a subdivision was made for both PM2.5 and NO2 to the population density
(people living in cities or in rural areas have different damage from pollution).
For PM2.5 a further distinction was being made to transport emissions and other sources
of emissions. For PM2.5 and NOx specific emission damages from electricity generation
have also been calculated, as this information may be relevant to estimate the damage
costs of electrical vehicles.
A detailed discussion of the adaptations is presented in Annex C.
4.3.2 Input values
Emissions
Table 13 gives an overview on the data sources used for calculating the emissions of air
pollutants for the different transport modes. For all data, 2016 was taken as the reference
year (transport data and emission factors), also for COPERT.

Table 13 – Data sources for the emissions of air pollutants for different transport modes
Transport data Emission factors (for total Emission factors

and average costs) (for selected cases)
Road Transport EU Transport in Figures, COPERT database v5 COPERT database v5

Eurostat and COPERT v5
(country data) (country data)
— Passenger transport EU Transport in Figures, COPERT database v5 COPERT database v5

Eurostat and COPERT v5,
(country data) (country data)
TRACCS database
— Freight transport Eurostat and COPERT v5,

TRACCS database
Rail Transport Eurostat, EU Transport in TREMOD (IFEU, 2017) TREMOD (IFEU, 2017)
Figures and TREMOVE
Air Transport airports (survey), Eurostat TREMOD (IFEU, 2017) EEA, EMEP Guidebook;
TREMOD
Inland Waterways EU Transport in Figures and EcoTransitWorld and EcoTransitWorld and

Eurostat TREMOD (IFEU, 2017) TREMOD (IFEU, 2017)
Maritime ports (survey) ISL Bremen ISL Bremen
Cost factors
The following tables summarises the cost factors for air pollution used for calculating the
health and other effects. Table 14 includes the cost factors per country for pollutants
emitted in road, rail and inland waterway transport. Table 15 shows the cost factors for
maritime transport.
Luxembourg has particularly high values compared to other countries. This is primarily due
to the high value of the VOLY. Islands, such as Malta, Cyprus and Ireland, tend to have
lower damage costs than countries in the mainland with comparable levels of income in
purchasing power parities. 15 Bulgaria, Latvia and Estonia tend to have lower damage costs
because of their lower income levels.
Table 14 – Air pollution costs: average damage cost in €/kg emission, national averages for transport emissions in 2016 (excl.
maritime) (All effects: health effects, crop loss, biodiversity loss, material damage)
€2016 /kg NH3 NMVOC SO2 NOx NOx PM2.5 PM2.5 PM2.5 PM10
transport transport transport transport transport average*
city° rural° metropole° city° rural°
Austria 27.8 2.3 16.2 41.4 24.3 466 151 87 30,9
Belgium 38.2 3.6 17.1 26.1 15.1 479 155 114 47,2
Bulgaria 5.6 0 4.2 10 5.9 191 61 30 5,4
Croatia 17.9 0.9 8 18.5 11.4 292 95 54 12,2
Cyprus 3.8 -0.4 7.8 8.1 4.5 n.a.** 71 17 8,2
________________________________
15
The negative value for NMVOC emissions in Cyprus is related to the fact that NO x is the main precursor of ozone
in Cyprus and that emissions of NMVOC tend to lower the ozone concentrations.

€2016 /kg NH3 NMVOC SO2 NOx NOx PM2.5 PM2.5 PM2.5 PM10
transport transport transport transport transport average*
city° rural° metropole° city° rural°
Czech Republic 27.4 1.1 11.6 24.8 14.8 361 116 72 20,1
Denmark 14.0 1.5 9.6 16.2 9.6 470 151 59 15
Estonia 10.5 0.3 5.2 5.4 3.4 n.a.** 102 35 4,9
Finland 7.0 0.4 4.6 5.3 3.5 366 118 32 5,9
France 15.4 1.5 13.9 27.2 16.2 407 131 87 24,7
Germany 28.1 1.8 16.5 36.8 21.6 448 144 93 39,6
Greece 4.8 0.3 5.9 5.1 3.1 267 86 33 8,5
Hungary 18.9 0.8 9.9 26.8 15.8 317 102 59 19
Ireland 4.1 1.7 11.8 17.6 10.1 568 183 68 17,2
Italy 21.6 1.1 12.7 25.4 15.1 409 132 79 27
Latvia 8.7 0.4 4.8 7.2 4.4 251 81 28 5,6
Lithuania 7.9 0.6 6.4 12.1 7.1 300 98 38 8
Luxembourg 60.0 6.2 29.3 66.8 38.4 n.a.** 278 191 63,9
Malta 6.4 0.4 4.3 2.3 1.4 n.a.** 72 18 5,2
Netherlands 30.0 2.8 20.2 26.5 15.3 458 148 101 47,3
Poland 14.4 0.7 8.2 14.7 8.9 282 91 52 16,1
Portugal 4.3 0.5 4.1 2.8 1.7 292 94 39 12,3
Romania 9.4 0.5 7.3 19.4 11.2 272 88 42 12
Slovakia 24.4 0.7 10.1 24.8 14.7 328 105 59 16,2
Slovenia 23.8 1.2 9.2 22.3 13.7 n.a.** 93 52 15,2
Spain 6.4 0.7 6.8 8.5 5.1 348 112 46 11,9
Sweden 10.6 0.7 5.5 9.5 6 374 120 38 10,2
United Kingdom 17.6 1.4 10 13.6 7.9 380 122 65 24,8
EU28 17.5 1.2 10.9 21.3 12.6 381 123 70 22,3
Notes:
* PM 10 cost factors can be used for the non-exhaust emission of particles PM, e.g. from brake and tyre abrasion.
** Metropole only applies to cities larger than 0.5 million inhabitants. Some countries do not have such cities hence these
damage values are hence not being reported. This is the case for Slovenia, Malta, Luxembourg, Estonia and Cyprus.
° Rural area: outside cities; metropolitan area: cities/agglomeration with more than 0.5 million inhabitants.
Table 15 – Air pollution costs: average damage cost in €/kg emission, national averages for maritime emissions
in 2016 (all effects: health effects, crop loss, biodiversity loss, material damage)
€2016 /kg NH3 NMVOC SO2 NOx PM2.5 PM10

Atlantic 0.0 0.4 3.5 3.8 7.2 4.1
Baltic 0.0 1.0 6.9 7.9 18.3 10.4
Black Sea 0.0 0.2 11.1 7.8 30.0 17.1
Mediterrenean 0.0 0.5 9.2 3.0 24.6 14.0
North Sea 0.0 2.3 10.5 10.7 34.4 19.7
4.3.3 Output values

The following tables show the resulting cost factors (output values) for the air pollution
costs per vehicle type. The tables include the total costs as well as the average costs per
vkm and per pkm or tkm.

Table 16 - Total and average air pollution costs for land-based modes for the EU28
Passenger transport Billion € €-cent/pkm €-cent/vkm
Passenger car – petrol 8.58 0.33 0.53
Passenger car – diesel 24.79 1.18 1.90
Motorcycle 1.84 1.12 1.17
Bus 1.35 0.76 14.19
Coach 2.67 0.73 14.34
Passenger train electric 0.03* 0.01 1.14
Passenger train diesel 0.52 0.80 47.0
Freight transport Billion € €-cent/tkm €-cent/vkm
LCV 15.49 4.68 3.24
LCV - petrol 0.33 1.72 1.17
LCV - diesel 15.16 4.86 3.37
HGV 13.93 0.76 9.38
Freight train electric 0.01 0.004 2.14
Freight train diesel 0.66 0.68 305.39
Total freight rail 0.67
Inland Vessel 1.93 1.29 1,869
Total road, rail, inland waterway 71.80
trains).
Table 17 – Total and average air pollution costs for aviation for 33 selected EU airports
Type of flight Billion € €-cent/pkm €-cent/pax*

Short haul (< 1,500 km) 0.27 0.30 163
Medium haul (1,500–5,000 km) 0.38 0.13 231

Long haul (> 5,000 km) 0.36 0.06 444
Total 1.01 0.10 246
Table 18 presents rough estimates for the average external air pollution costs of maritime
transport. These data are only available for freight. The average cost have been based on
the cost for reference cases presented in Section 4.4 and data on the number of port calls
for the selected ports from Eurostat. The total air pollution cost has been based on the
average cost and the number of tkms provided by DG MOVE16. The available data does not
allow an estimate of costs at the individual port level.
Table 18 – Rough estimates for total and average external air pollution costs for maritime transport for 34
selected EU ports
Transport mode Total costs (bn €) Average costs (€-cent/tkm)

Freight ship 29 0.4
4.4 Marginal air pollution costs for selected cases

For air pollution costs, the marginal costs are virtually the same as the average costs.
This is mainly because the dose-response relationships between the immissions of air
pollutants and health effects (or other damages) are nearly linear according to
epidemiological studies. Therefore, the present chapter also covers average air pollution
costs. The methodology used is the same as for the total and average costs (see Figure 3
above).
The costs for road vehicles are presented for all differentiations provided by COPERT, e.g.
different fuel types, engines or vehicle sizes, emission classes and regional areas. It needs
to be emphasized that for a modern car (after Euro 1), engine size is not a cost driver for
the air pollution costs of cars. Therefore, these costs are identical for the various engine
size classes. Any differences are the result of rounding numbers from the COPERT data.
Table 19 on marginal air pollution costs for road transport shows the costs per pkm or tkm
(except for LCV, where costs per vkm are presented due to the fact that LCV have
characteristics of freight and passenger transport). The costs per vkm for the different
vehicle categories of road transport are available in the background Excel file.
Table 19 – Marginal air pollution costs road transport for selected cases
________________________________
16 Some assumptions had to be made for calculating maritime transport performance. The Eurostat transport
volumes (i.e. tonnes) and distance matrices have been used for this purpose. By assumption, 50% of the
calculated transport performance is allocated to the origin country and 50% to the destination country
between EU Countries and EFTA and candidate countries. For the international extra-EU activity, where the
corresponding partner is outside EU28 and is not an EFTA or candidate country, 100% of transport performance is
allocated to the declaring EU MS country. These assumptions are used only for this study purposes and shall be
considered as estimates and not as official data.

Vehicle Fuel type Size Emission Metropolitan area Urban area Rural area
class
Motorway Urban Other Motorway Urban Motorway Rural
road road road road
Passenger Petrol Mini Euro 4 0.05 0.12 0.06 0.04 0.10 0.02 0.03
Cars
< 0.8 l Euro 5 0.05 0.08 0.06 0.03 0.06 0.02 0.02
Euro 6 0.05 0.09 0.06 0.03 0.06 0.02 0.02
Small Euro 0 2.91 2.69 2.99 2.88 2.63 1.76 1.82
0.8–1.4 l Euro 1 0.82 0.55 0.48 0.79 0.50 0.50 0.29
Euro 2 0.36 0.37 0.30 0.33 0.32 0.22 0.18
Euro 3 0.17 0.19 0.15 0.15 0.17 0.11 0.10
Euro 4 0.12 0.17 0.11 0.10 0.15 0.08 0.07
Euro 5 0.12 0.13 0.10 0.09 0.11 0.08 0.07
Euro 6 0.12 0.14 0.10 0.09 0.11 0.08 0.07
Medium Euro 0 4.53 3.04 3.80 4.50 2.99 2.73 2.30
1.4–2.0 l Euro 1 0.86 0.55 0.48 0.83 0.50 0.52 0.29
Euro 2 0.36 0.37 0.30 0.33 0.32 0.22 0.18
Euro 3 0.17 0.19 0.15 0.15 0.17 0.12 0.10
Euro 4 0.12 0.17 0.11 0.10 0.15 0.08 0.07
Euro 5 0.12 0.13 0.10 0.09 0.11 0.08 0.07
Euro 6 0.12 0.14 0.10 0.09 0.11 0.08 0.07
Large-SUV- Euro 0 8.16 3.77 5.65 8.12 3.71 4.87 3.39

Executive
Euro 1 0.88 0.55 0.48 0.85 0.50 0.53 0.29
> 2.0 l
Euro 2 0.36 0.37 0.30 0.33 0.32 0.22 0.18
Euro 3 0.17 0.19 0.15 0.15 0.17 0.12 0.10
Euro 4 0.12 0.17 0.11 0.10 0.15 0.08 0.07
Euro 5 0.12 0.13 0.10 0.09 0.11 0.08 0.07
Euro 6 0.12 0.14 0.10 0.09 0.11 0.08 0.07
Diesel Mini Euro 4 1.65 1.70 1.21 1.20 1.17 0.73 0.49
< 0.8 l Euro 5 0.92 1.04 0.74 0.90 0.99 0.56 0.44
Euro 6 0.76 0.86 0.61 0.75 0.82 0.47 0.37
Small Euro 0 5.81 6.83 3.84 2.47 2.79 1.45 0.99
0.8–1.4 l Euro 1 4.17 2.56 2.28 2.05 1.50 1.22 0.77
Euro 2 2.49 2.39 1.74 1.54 1.51 0.93 0.66

class
road road road road
Euro 3 2.45 1.90 1.66 1.59 1.37 0.96 0.71
Euro 4 1.65 1.70 1.21 1.20 1.17 0.73 0.49
Euro 5 0.92 1.04 0.74 0.90 0.99 0.56 0.44
Euro 6 0.76 0.86 0.61 0.75 0.82 0.47 0.37
Medium Euro 0 6.11 7.05 4.03 2.73 3.01 1.61 1.10
1.4–2.0 l Euro 1 4.21 2.56 2.29 2.06 1.50 1.22 0.77
Euro 2 2.51 2.39 1.75 1.55 1.51 0.93 0.67
Euro 3 2.47 1.90 1.66 1.60 1.37 0.97 0.71
Euro 4 1.67 1.70 1.21 1.21 1.17 0.74 0.49
Euro 5 0.93 1.04 0.74 0.90 0.99 0.56 0.44
Euro 6 0.77 0.86 0.61 0.75 0.82 0.47 0.37
Large-SUV- Euro 0 6.41 7.27 4.22 3.00 3.23 1.77 1.22

Executive
Euro 1 4.25 2.56 2.31 2.08 1.50 1.23 0.77
> 2.0 l
Euro 2 2.53 2.40 1.75 1.56 1.51 0.94 0.67
Euro 3 2.49 1.90 1.67 1.62 1.37 0.97 0.71
Euro 4 1.68 1.70 1.21 1.22 1.17 0.74 0.49
Euro 5 0.93 1.04 0.74 0.91 0.99 0.56 0.44
Euro 6 0.77 0.86 0.61 0.76 0.82 0.47 0.37
Petrol Mini n.a. 0.09 0.06 0.07 0.09 0.06 0.08 0.06
Hybrid
(PHEV) Small n.a. 0.09 0.06 0.07 0.09 0.06 0.08 0.06
Large-SUV- n.a. 0.09 0.06 0.07 0.09 0.06 0.08 0.06

Executive
LPG Small Euro 1 0.55 0.65 0.47 0.52 0.60 0.34 0.28
Bifuel
Euro 2 0.27 0.31 0.23 0.24 0.26 0.17 0.14
Euro 3 0.17 0.19 0.15 0.15 0.17 0.12 0.10
Euro 4 0.12 0.17 0.11 0.10 0.15 0.08 0.07
Euro 5 0.11 0.15 0.10 0.09 0.13 0.08 0.07
Euro 6 0.11 0.15 0.10 0.09 0.13 0.08 0.07
CNG Small Euro 4 0.12 0.17 0.11 0.10 0.15 0.09 0.08
Bifuel
Euro 5 0.11 0.14 0.10 0.09 0.12 0.08 0.07
Euro 6 0.11 0.14 0.10 0.09 0.12 0.08 0.07

class
road road road road
Electric n.a. n.a. 0.06 0.05 0.05 0.06 0.05 0.06 0.05
(BEV)
Moped Petrol 2-stroke Euro 0 7.46 7.45 7.45 3.13 3.13 2.20 2.19
< 50 cm³
Euro 1 2.41 2.40 2.40 1.30 1.29 0.92 0.92
Euro 2 1.61 1.60 1.60 0.97 0.96 0.70 0.69
Euro 3 1.20 1.20 1.20 0.76 0.76 0.53 0.53
4-stroke Euro 0 7.46 7.45 7.45 3.13 3.13 2.20 2.19

< 50 cm³
Euro 1 2.01 2.00 2.00 1.02 1.02 0.64 0.63
Euro 2 0.71 0.71 0.70 0.54 0.53 0.36 0.35
Euro 3 0.57 0.57 0.57 0.47 0.47 0.31 0.31
Motorcycle Petrol 2-stroke Euro 0 8.49 8.44 8.13 3.57 3.53 2.53 2.18
≥ 50 cm³
Euro 1 3.52 3.34 3.29 1.56 1.38 1.10 0.88
Euro 2 1.88 1.74 1.73 0.89 0.76 0.63 0.50
Euro 3 0.69 0.61 0.61 0.40 0.31 0.31 0.23
4-stroke Euro 0 2.00 1.45 1.67 1.51 0.96 0.93 0.73

< 250 cm³
Euro 1 2.08 1.45 1.73 1.59 0.96 0.98 0.76
Euro 2 1.49 0.72 1.00 1.36 0.60 0.84 0.54
Euro 3 1.27 0.67 0.86 1.15 0.55 0.70 0.45
4-stroke Euro 0 2.24 1.55 1.69 1.75 1.06 1.08 0.77

250-750
cm³ Euro 1 2.13 1.39 1.61 1.63 0.90 1.00 0.70
Euro 2 0.90 0.42 0.51 0.78 0.29 0.50 0.25
Euro 3 0.57 0.32 0.36 0.45 0.19 0.29 0.16
4-stroke Euro 0 1.95 1.32 1.28 1.46 0.83 0.94 0.53

> 750 cm³
Euro 1 1.93 1.18 1.31 1.44 0.68 0.89 0.51
Euro 2 1.54 0.47 0.64 1.41 0.34 0.87 0.32
Euro 3 0.89 0.34 0.43 0.77 0.22 0.48 0.20
Electric n.a. 0.02 0.02 0.02 0.02 0.02 0.02 0.02
Urban buses Diesel Midi <=15 t Euro 0 2.28 4.75 2.46 1.51 2.79 0.89 0.95
Euro I 1.26 2.33 1.37 0.92 1.69 0.55 0.60
Euro II 1.11 1.96 1.20 0.91 1.65 0.54 0.60
Euro III 0.83 1.87 0.93 0.65 1.55 0.39 0.45

class
road road road road
Euro IV 0.47 0.95 0.54 0.43 0.87 0.26 0.30
Euro V 0.39 1.15 0.44 0.33 1.06 0.20 0.24
Euro VI 0.03 0.13 0.05 0.03 0.13 0.02 0.03
Diesel Standard Euro 0 2.11 4.40 2.35 1.59 3.00 0.94 1.04
15-18 t
Euro I 1.30 2.53 1.43 0.96 1.80 0.57 0.63
Euro II 1.16 2.09 1.26 0.95 1.75 0.56 0.62
Euro III 0.86 1.93 0.98 0.69 1.60 0.41 0.48
Euro IV 0.50 1.00 0.57 0.46 0.92 0.28 0.32
Euro V 0.35 1.16 0.40 0.30 1.07 0.18 0.21
Euro VI 0.03 0.12 0.04 0.03 0.11 0.02 0.03
Diesel Articulated Euro 0 2.19 4.54 2.51 1.64 3.12 0.97 1.10
> 18 t
Euro I 1.37 2.64 1.52 1.00 1.88 0.59 0.67
Euro II 1.19 2.18 1.31 0.96 1.80 0.57 0.64
Euro III 0.90 1.98 1.02 0.73 1.63 0.43 0.50
Euro IV 0.51 1.07 0.62 0.47 0.99 0.28 0.34
Euro V 0.30 1.01 0.34 0.25 0.92 0.15 0.17
Euro VI 0.03 0.09 0.04 0.03 0.08 0.02 0.03
CNG CNG buses Euro I 2.13 2.13 2.13 2.10 2.10 1.24 1.24
Euro II 1.92 1.92 1.92 1.90 1.90 1.13 1.13
Euro III 1.28 1.28 1.28 1.27 1.27 0.75 0.75
EEV* 0.26 0.53 0.30 0.25 0.52 0.15 0.17
Bio- Biodiesel Euro 0 2.10 4.36 2.34 1.58 2.97 0.93 1.02
diesel buses
Euro I 1.29 2.49 1.42 0.95 1.76 0.56 0.62
Euro II 1.15 2.05 1.24 0.94 1.71 0.55 0.61
Euro III 0.85 1.90 0.97 0.68 1.56 0.40 0.46
Euro IV 0.49 0.96 0.55 0.45 0.88 0.26 0.30
Euro V 0.34 1.12 0.39 0.29 1.03 0.17 0.20
Euro VI 0.02 0.08 0.03 0.01 0.07 0.01 0.01
Electric Small n.a. 0.01 0.05 0.02 0.01 0.05 0.01 0.02
Medium n.a. 0.01 0.04 0.02 0.01 0.04 0.01 0.02
Large n.a. 0.01 0.03 0.01 0.01 0.03 0.01 0.01

class
road road road road
Coaches Diesel Standard Euro 0 1.77 3.99 2.02 1.35 2.77 0.80 0.88
<=18 t
Euro I 1.34 3.19 1.56 1.02 2.21 0.60 0.68
Euro II 1.24 2.64 1.39 1.05 2.21 0.62 0.69
Euro III 1.01 2.67 1.19 0.83 2.12 0.50 0.57
Euro IV 0.61 1.34 0.68 0.56 1.23 0.34 0.38
Euro V 0.37 1.86 0.59 0.32 1.71 0.19 0.31
Euro VI 0.04 0.21 0.07 0.03 0.19 0.02 0.04
Diesel Articulated Euro 0 1.42 3.22 1.62 1.07 2.26 0.63 0.71
> 18 t
Euro I 1.04 2.51 1.23 0.79 1.77 0.47 0.54
Euro II 0.95 2.05 1.08 0.79 1.72 0.47 0.54
Euro III 0.73 1.97 0.88 0.60 1.56 0.36 0.42
Euro IV 0.44 1.00 0.50 0.41 0.91 0.24 0.28
Euro V 0.26 1.35 0.41 0.21 1.25 0.13 0.22
Euro VI 0.02 0.14 0.04 0.02 0.13 0.01 0.03
Light commercial vehicle (€-cent per vkm)
Light Petrol Euro 0 8.31 5.84 7.05 8.26 5.75 4.96 4.20
Commercial
Vehicle Euro 1 1.32 1.20 1.02 1.27 1.12 0.80 0.61
Euro 2 0.56 0.54 0.44 0.51 0.45 0.34 0.26
Euro 3 0.39 0.33 0.29 0.36 0.30 0.25 0.19
Euro 4 0.26 0.22 0.19 0.23 0.19 0.18 0.13
Euro 5 0.27 0.16 0.18 0.17 0.13 0.14 0.10
Euro 6 0.27 0.16 0.18 0.17 0.13 0.14 0.10
Diesel Euro 1 9.18 6.22 4.93 4.91 3.94 2.91 1.82
Euro 2 9.18 6.22 4.93 4.91 3.94 2.91 1.82
Euro 3 6.65 4.66 3.68 3.79 3.13 2.26 1.45
Euro 4 4.19 3.13 2.46 2.69 2.34 1.62 1.09
Euro 5 3.98 2.69 2.58 3.96 2.65 2.38 1.54
Euro 6 3.24 2.19 2.10 3.22 2.16 1.94 1.26
Electric n.a. 0.10 0.08 0.07 0.10 0.08 0.10 0.07
HGV Diesel Euro 0 22.05 30.29 21.88 15.44 18.35 9.19 8.71

class
road road road road
Rigid Euro I 13.38 15.86 12.34 10.57 11.17 6.32 5.66

<=7,5 t
Euro II 12.35 12.87 10.96 10.35 10.78 6.20 5.58
Euro III 8.79 11.65 8.41 7.54 9.18 4.54 4.22
Euro IV 5.74 6.28 5.18 5.28 5.77 3.21 2.93
Euro V 2.84 7.85 2.96 2.37 7.20 1.49 1.59
Euro VI 0.32 1.55 0.47 0.28 1.48 0.26 0.36
Rigid Euro 0 11.55 16.51 11.64 9.07 11.58 5.39 5.26

7,5-12 t
Euro I 7.08 10.11 7.04 5.49 7.06 3.27 3.18
Euro II 6.72 8.12 6.26 5.51 6.77 3.29 3.14
Euro III 4.75 7.23 4.96 3.99 5.68 2.39 2.46
Euro IV 2.97 3.81 2.99 2.73 3.49 1.65 1.67
Euro V 1.55 4.93 1.77 1.28 4.53 0.79 0.93
Euro VI 0.20 0.70 0.26 0.18 0.66 0.14 0.18
Rigid Euro 0 6.65 9.82 6.72 5.10 6.98 3.03 3.02

12-14 t
Euro I 4.07 6.05 4.11 3.10 4.28 1.84 1.84
Euro II 3.83 4.96 3.65 3.10 4.14 1.85 1.82
Euro III 2.78 4.46 2.90 2.30 3.57 1.38 1.44
Euro IV 1.67 2.39 1.74 1.54 2.20 0.93 0.98
Euro V 0.89 2.89 1.06 0.74 2.66 0.45 0.56
Euro VI 0.11 0.38 0.14 0.10 0.35 0.08 0.10
Rigid Euro 0 7.49 12.30 7.96 5.66 8.58 3.36 3.50

14-20 t
Euro I 4.46 7.54 4.79 3.36 5.25 1.99 2.10
Euro II 4.24 6.02 4.17 3.43 5.04 2.04 2.08
Euro III 3.11 5.58 3.45 2.56 4.40 1.53 1.68
Euro IV 1.88 2.86 2.00 1.74 2.63 1.04 1.12
Euro V 1.02 3.91 1.62 0.85 3.61 0.52 0.87
Euro VI 0.13 0.51 0.18 0.11 0.48 0.08 0.12
Rigid Euro 0 3.69 6.21 3.99 2.74 4.43 1.62 1.75

20-26 t
Euro I 2.65 4.70 2.88 1.96 3.25 1.16 1.25
Euro II 2.48 3.74 2.51 1.98 3.10 1.18 1.24
Euro III 1.89 3.35 2.06 1.56 2.64 0.93 1.00

class
road road road road
Euro IV 1.13 1.75 1.20 1.05 1.60 0.63 0.67
Euro V 0.58 2.17 0.85 0.48 1.98 0.29 0.45
Euro VI 0.06 0.26 0.09 0.05 0.24 0.04 0.06
Rigid Euro 0 2.74 4.59 2.98 2.03 3.26 1.19 1.29

26-28 t
Euro I 1.96 3.43 2.15 1.44 2.38 0.85 0.92
Euro II 1.84 2.78 1.87 1.46 2.30 0.87 0.92
Euro III 1.38 2.45 1.52 1.13 1.93 0.67 0.73
Euro IV 0.82 1.29 0.87 0.76 1.18 0.46 0.48
Euro V 0.39 1.58 0.61 0.32 1.45 0.19 0.32
Euro VI 0.05 0.18 0.07 0.04 0.17 0.03 0.04
Rigid Euro 0 2.55 4.14 2.75 1.89 2.94 1.11 1.20

28-32 t
Euro I 1.86 3.14 2.02 1.37 2.21 0.81 0.88
Euro II 1.74 2.54 1.76 1.37 2.10 0.81 0.86
Euro III 1.28 2.21 1.40 1.05 1.75 0.63 0.68
Euro IV 0.76 1.19 0.82 0.71 1.09 0.42 0.45
Euro V 0.33 1.34 0.49 0.27 1.23 0.16 0.25
Euro VI 0.05 0.14 0.06 0.04 0.13 0.03 0.04
Rigid Euro 0 2.21 3.77 2.40 1.63 2.69 0.96 1.04

> 32 t
Euro I 1.62 2.89 1.76 1.18 2.01 0.70 0.76
Euro II 1.51 2.32 1.54 1.19 1.91 0.71 0.75
Euro III 1.14 2.02 1.25 0.93 1.60 0.55 0.61
Euro IV 0.68 1.08 0.73 0.63 0.99 0.38 0.40
Euro V 0.32 1.20 0.45 0.26 1.10 0.16 0.23
Euro VI 0.03 0.13 0.05 0.03 0.12 0.02 0.03
Articulated Euro 0 3.99 6.74 4.33 3.00 4.74 1.77 1.89
14-20 t Euro I 2.44 4.14 2.60 1.80 2.88 1.06 1.12
Euro II 2.27 3.33 2.27 1.81 2.77 1.07 1.11
Euro III 1.70 3.01 1.85 1.38 2.38 0.82 0.89
Euro IV 1.00 1.55 1.07 0.92 1.41 0.54 0.58
Euro V 0.51 1.95 0.76 0.42 1.79 0.25 0.39
Euro VI 0.05 0.17 0.06 0.04 0.15 0.02 0.03

class
road road road road
20-28 t Euro I 2.58 4.56 2.84 1.88 3.23 1.11 1.23
Euro II 2.37 3.68 2.43 1.86 3.05 1.10 1.19
Euro III 1.77 3.25 1.97 1.42 2.58 0.85 0.95
Euro IV 1.03 1.73 1.14 0.95 1.59 0.57 0.63
Euro V 0.52 2.01 0.76 0.43 1.84 0.26 0.40
Euro VI 0.07 0.22 0.08 0.06 0.20 0.04 0.05
28-34 t Euro I 1.62 2.89 1.77 1.17 2.03 0.69 0.77
Euro II 1.47 2.31 1.51 1.14 1.91 0.68 0.74
Euro III 1.09 2.01 1.22 0.88 1.60 0.52 0.59
Euro IV 0.63 1.09 0.71 0.58 1.00 0.35 0.39
Euro V 0.30 1.18 0.44 0.25 1.07 0.15 0.23
Euro VI 0.04 0.12 0.05 0.04 0.11 0.03 0.03
34-40 t Euro I 1.59 2.98 1.77 1.15 2.07 0.68 0.76
Euro II 1.47 2.39 1.53 1.15 1.96 0.68 0.74
Euro III 1.11 2.07 1.24 0.90 1.64 0.54 0.60
Euro IV 0.65 1.10 0.72 0.60 1.01 0.36 0.40
Euro V 0.32 1.20 0.45 0.26 1.08 0.16 0.23
Euro VI 0.03 0.12 0.04 0.03 0.11 0.02 0.03
40-50 t Euro I 1.53 2.84 1.69 1.09 1.97 0.65 0.72
Euro II 1.40 2.27 1.46 1.09 1.86 0.64 0.71
Euro III 1.04 1.93 1.17 0.85 1.54 0.50 0.57
Euro IV 0.62 1.05 0.68 0.57 0.97 0.34 0.38
Euro V 0.28 1.04 0.39 0.23 0.94 0.14 0.20
Euro VI 0.03 0.10 0.04 0.02 0.09 0.02 0.02
50-60 t Euro I 1.59 2.97 1.77 1.13 2.07 0.67 0.76
Euro II 1.45 2.39 1.52 1.12 1.95 0.66 0.73

class
road road road road
Euro III 1.09 2.04 1.21 0.88 1.62 0.52 0.59
Euro IV 0.60 1.09 0.71 0.55 1.01 0.33 0.39
Euro V 0.27 1.00 0.38 0.22 0.89 0.13 0.19
Euro VI 0.03 0.10 0.04 0.02 0.09 0.02 0.02
LNG Articulated n.a.

32 t+ 0.03 0.09 0.03 0.02 0.08 0.02 0.02
* EEV: Enhanced environmentally friendly vehicle. European emission standard for the definition of a ‘clean
vehicle’ > 3.5 t. Emission level between Euro V and Euro VI.
Table 20 - Marginal air pollution costs rail transport for selected cases
Train type Traction Emission class Metropolitan area Urban area Rural area
High speed train Electric n.a. 0.002 0.002 0.002

1)
Intercity train Electric n.a. 0.01 0.01 0.01
Diesel Equipped with EGR/SCR 0.47 0.38 0.23
Not Equipped with EGR/SC 0.70 0.67 0.40
Regional train Electric n.a. 0.02 0.02 0.02
Not Equipped with EGR/SCR 2.10 1.99 1.20
Short container Electric n.a. 0.004 0.004 0.004

freight train
(420 metres) Diesel Equipped with EGR/SCR 0.356 0.309 0.184
Short bulk Electric n.a. 0.004 0.004 0.004

freight train
(300 metres)
Long container Electric n.a. 0.004 0.004 0.004

freight train
(620 metres) Diesel Equipped with EGR/SCR 0.128 0.111 0.067
Long bulk freight Electric n.a. 0.004 0.004 0.004

train

Train type Traction Emission class Metropolitan area Urban area Rural area
(440 metres) Not Equipped with EGR/SCR 0.245 0.200 0.119
1) There is no literature on the differences between high-speed trains and ‘normal’ intercity trains in terms of PM non-exhaust
emissions. Most of the PM-emissions are caused by braking. Newer brake pads cause much less PM non-exhaust emissions
than old cast iron brake pads. High-speed trains have probably newer brake pads. Tough high-speed trains are heavier and
drive faster, they brake less because of the less winding tracks. It is not known how PM non-exhaust emissions from high-
speed trains behave compared to Intercity-trains. That’s why these are here equated with ‘normal’ Intercity trains.

Table 21 – Marginal air pollution costs IWT for selected cases
Vessel type Type of cargo Emission class Urban area Rural area
€-cent/tkm €-cent/vkm €-cent/tkm €-cent/vkm
CEMT II (350 t) Bulk CCNR 0 3.36 1,074 1.98 631
CCNR 1 2.82 899 1.66 529
CCNR 2 1.82 580 1.07 342
Average 3.25 1,039 1.91 610
Container CCNR 0 2.14 1,074 1.26 631
CCNR 1 1.79 899 1.05 529
CCNR 2 1.15 580 0.68 342
Average 2.07 1,039 1.21 610
CEMT IV Bulk CCNR 0 2.00 1,594 1.17 936

(600 t)
CCNR 1 1.67 1,335 0.98 786
CCNR 2 1.08 861 0.64 507
Average 1.84 1,470 1.08 864
CEMT Va (1,500 Bulk CCNR 0 1.82 2,912 1.07 1,711

t)
CCNR 1 1.53 2,439 0.90 1,435
CCNR 2 0.99 1,573 0.58 926
Average 1.53 2,449 0.90 1,440
Container CCNR 0 2.06 2,912 1.21 1,711
CCNR 1 1.73 2,439 1.02 1,435
CCNR 2 1.12 1,573 0.66 926
Average 1.74 2,449 1.02 1,440
Pushed convoy Bulk CCNR 0 1.48 7,799 0.87 4,582

(11,000 t)
CCNR 1 1.24 6,531 0.73 3,844
CCNR 2 0.80 4,213 0.47 2,480
Average 0.89 4,714 0.53 2,775
Container CCNR 0 1.10 7,799 0.65 4,582
CCNR 1 0.92 6,531 0.54 3,844
CCNR 2 0.60 4,213 0.35 2,480
Average 0.67 4,714 0.39 2,775

Table 22 – Marginal air pollution costs maritime transport for selected cases
Vessel type Distance at sea Tier € per port call €-cent per € per vessel-km
(km) pkm or tkm
Passenger transport
100 Tier 0 19,232 36.42 192
100 Tier 1 16,731 31.69 167
100 Tier 2 15,198 28.78 152

RoPax Ferry
(25,500 gt) 500 Tier 0 64,986 24.62 130
500 Tier 1 58,691 22.23 117
500 Tier 2 54,832 20.77 110
Freight transport
500 Tier 0 137,036 1.14 274
500 Tier 1 119,408 1.00 239
Small container 500 Tier 2 108,604 0.91 217

vessel (28,500 gt) 3000 Tier 0 370,353 0.51 123
3,000 Tier 1 365,811 0.51 122
3000 Tier 2 206,856 0.29 69
500 Tier 0 259,014 0.45 518
500 Tier 1 232,364 0.40 465
500 Tier 2 216,030 0.38 432
3000 Tier 0 868,167 0.25 289

Large container
vessel 3,000 Tier 1 858,333 0.25 286
(143,000 gt)
3000 Tier 2 514,162 0.15 171
15000 Tier 0 3,041,358 0.18 203
15,000 Tier 1 2,695,909 0.16 180
15000 Tier 2 1,751,684 0.10 117
500 Tier 0 46,763 0.62 94
500 Tier 1 41,614 0.55 83
500 Tier 2 38,459 0.51 77

Small bulk vessel
(18,000 gt)
3000 Tier 0 149,567 0.33 50
3,000 Tier 1 147,827 0.33 49
3000 Tier 2 86,922 0.19 29
500 Tier 0 115,019 0.22 230

Vessel type Distance at sea Tier € per port call €-cent per € per vessel-km
(km) pkm or tkm
500 Tier 1 102,473 0.20 205
500 Tier 2 94,783 0.18 190
3000 Tier 0 358,106 0.12 119
3,000 Tier 1 354,077 0.11 118

Large bulk vessel
(105,000 gt)
3000 Tier 2 213,057 0.07 71
15000 Tier 0 1,232,817 0.08 82
15,000 Tier 1 1,095,755 0.07 73
15000 Tier 2 721,118 0.05 48

Table 23 – Marginal air pollution costs aviation for selected cases
Type of flight Distance Emission Example of aircraft € per LTO* €-cent per pkm* € per pax*
class type
[km]
Short haul 500 Low Bombardier CRJ900 101 0.28 1.42
500 High Embraer 170 137 0.30 1.52
Medium haul 1,500 Low Airbus 320 165 0.07 1.11
1,500 High Boeing 737 185 0.11 1.58
3,000 Low Airbus 320 219 0.05 1.47
3,000 High Boeing 737 245 0.07 2.09
Long haul 5,000 Low Airbus 340 502 0.03 1.70
5,000 High Boeing 777 833 0.04 1.92
15,000 Low Airbus 340 711 0.02 2.41
15,000 High Boeing 777 1.179 0.02 2.72
* For the cost factors for air pollution costs the emissions during the LTO cycle are mainly relevant, as the cruise
emissions almost lead to no damage costs.
The marginal costs of aviation for selected cases and aircrafts cannot be directly compared
with the average costs: The marginal costs refer to very specific aircraft types, distances
and loading factors that do not match the average. E.g. for short haul flights, the average
number of passenger per flight is substantially higher than for the selected cases (since
many short haul flights are done by larger aircraft). Additionally, the average distances are
different than the one use in the selected cases.

Generally, the air pollution costs have had a long history of research and are therefore
investigated and analysed in a very detailed way. For example, the scientific knowledge on
dose-response relationships for diseases induced by air pollutants is very profound.
Hence, compared to other cost categories, the cost factors for air pollution costs can be
regarded as robust.
For the present Handbook, the most important parameters for the robustness of the results
are the quality of the emission factors and the cost factors for the different air pollutants
(damage costs per air pollutant), which are listed hereafter.
Emission factors
— For road transport, the COPERT database is the main input, which is a widely used
source and considered a reliable data source. However, it is not clear to what extent
the emissions data used fully reflect the latest findings on real world emissions, e.g.
due to degradation and/or failure of particulate filters and catalysts in older vehicles.
— For other modes, the emission factors are from different sources, mainly from TREMOD
(from the German Umweltbundesamt) and EcoTransitWorld. Both data bases are of high

quality. However, the differentiated emission factors for different emission classes for
rail, inland waterways and maritime have a higher uncertainty.
Damage cost factors

— The estimation of damage cost factors is based on the NEEDS approach and includes a
broad update of the NEEDS data. It includes on the one hand up-to-date data on
concentration response functions and the valuation of damage, and on the other hand
differentiated data per country on population size and structure (density) and
background concentration. Overall, we regard the quality of the damage cost factors
per country and air pollutant as high, although an update of the NEEDS study is
recommended.
— The cost factors reflect the cost for which the causal relation between emissions and
health impacts has been proven. However, for some potential health problems, a causal
relation is suspected, but not scientifically proven (yet). When it turns out that these
relations can be proven by ongoing research, this would result in higher cost estimates.
— An important factor for uncertainties is the valuation of immaterial damage (i.e. value
of value of life year lost VOLY). The value used is based on a meta-analysis, however,
due to results from various studies varying significantly, some uncertainty in the value
used is unavoidable.
— One of the peer reviewers commented on the way own consumption is considered.
Generally it is assumed that it is included in the WTP, but it was argued that it should
not be. We have included it in the WTP.

5 Climate change costs
5.1 Introduction
Due to the fact that the effects of climate change are global, long-term and have risk
patterns that are difficult to anticipate, identifying the costs associated with these effects
is extremely complex. Transport results in emissions of CO 2, N2O and CH 4 (methane), all of
which are greenhouse gases contributing to climate change. Therefore, identifying the
climate costs of transport is extremely important. This chapter discusses the methodology
to value the climate costs of transport.
In Section 5.2 we first briefly discuss the definition and scope of climate change costs.
The total and average climate change costs are explored in Section 5.3, and the marginal
climate change costs are the topic of Section 5.4. Finally, the robustness of the climate
change cost figures presented in this chapter is analysed in Section 0. More detailed
information on the effects of climate change and their monetary valuation can be found in
Annex D.

The emission of greenhouse gases into the atmosphere leads to global warming and climate
change. The IPCC (2013) has estimated that without concrete climate policies temperatures
may be expected to rise significantly by the end of the century. Such radical change will
have an important and largely irreversible impact on ecosystems, human health and
societies. Climate change costs are defined as the costs associated with all of the effects of
global warming, such as sea level rise, biodiversity loss, water management issues, more
and more frequent weather extremes and crop failures. For a more detailed discussion of
the effects of climate change we refer to Annex D.2.
The climate change costs are calculated for all five transport modes. For road, (diesel-
powered) rail, inland waterway and maritime transport, the global warming impacts of
transport are mainly caused by CO2, N2O and CH 4. This chapter focusses on how to calculate
the total, average and marginal costs of climate change for these transport modes.
However, for aviation there are also other aircraft emissions such as water vapour,
sulphate and soot aerosols which are harmful to the climate when emitted at high altitudes.
We slightly adapt the methodology used for other transport modes to make it suitable to
calculate the climate change costs from aviation (see the textbox in Section 5.3.1).
For maritime, it is important to note that a number of exhaust emissions (e.g. NO x and SO2)
lead to (short-term) cooling effects, which implies that maritime transport currently has a
net cooling effect on the global climate (Eyring, et al., 2009). Although it is complicated to
compare the local, short-term cooling effects to the long term global warming effects,
global warming potentials help this comparison (see textbox in Section 5.3.1).

5.3 Total and average climate change costs
5.3.1 Methodology
Total and average climate costs are calculated using a bottom-up approach. Figure 5
illustrates the corresponding methodology for road, rail, inland waterway, aviation and
maritime transport. The methodology presented in this section is identical to the
methodology used in both earlier editions of the Handbook for all transport modes.
Figure 5 – Methodology total and average climate change costs
Three input values are used: the GHG emission factors per vehicle type, vehicle
performance data and the climate change costs per tonne of CO 2 equivalent. The GHG
emissions per vehicle type can be calculated by multiplying the vehicle kilometres per
vehicle type in each country with the vehicle emission factors (in g/km) for each of the
various GHG (CO 2, N2O, CH 4 and other aircraft emissions). Using Global Warming Potentials
(GWP), the emissions of the three GHGs can be added together to achieve the total CO 2
equivalent GHG emissions (see Textbox). This is then multiplied by the climate change costs
per tonne CO2 equivalent in order to reach the total climate change costs per mode.
To reach the average climate change costs we divide the total climate change costs by the
amount of pkms or tkms driven by the vehicle type.

Comparing CO2 and non-CO2 emissions
As mentioned earlier, climate change is not just triggered by CO2 emissions. There are numerous other gases
that result in climate change, although we limit ourselves to CO 2, N2O, CH 4 and aviation emissions in this study.
Aviation emissions, such as water vapour, sulphate and soot, can be particularly damaging when emitted at
high altitudes although they may have contradicting effects. In part, these emissions result in heating effects
(soot emissions from aircraft engines, night-time contrail formation, atmospheric chemical reactions on the
basis of NO X that increases ozone concentrations) and in part they result in cooling (sulphur aerosols, day -time
contrail formation, atmospheric chemical reactions on the basis of NOx that convert methane).
As these different emission gases differ in their lifetime and their potency it can be complicated to compare
CO2 emissions to non-CO2 emissions. To allow for such comparisons, the concept of th e Global Warming
Potential (GWP) is used. The GWP is a relative measure, which compares the amount of heat trapped by a
certain mass of gas to the amount of heat trapped by a similar mass of CO 2 over a certain period of time
(e.g. 100 years). The GWP of CO 2 is standardised to 1. In the IPCC’s latest Assessment Report (IPCC, 2013) the
GWP over a 100-year time period of CH 4 and N20 are 34 and 298 respectively. This implies that the same
amount of CH 4 is 34 times more potent than the same amount of CO 2, when looking at a period of 100 years.
These factors will be used to compare CO 2 emissions with N2O and CH 4 emissions for road, rail, inland
waterways and maritime transport in this Handbook.
To account for the emissions from aviation, GWPs are used as Emission Weighting Factors (Foster, et al.,
2007). Studies shown that the EWF for aviation lies in the range of 1.3-1.4 (Lee, et al., 2009) (Azar &
Johansson, 2012). This implies that the total climate change impact from aviation is 1.3-1.4 times larger than
the impact from its CO 2 emissions alone. It is important to note that these estimates do not include the imp acts
of aviation induced cloudiness. If aviation induced cirrus is included, the uncertainty regarding EWF increases
substantially, with values ranging from 1.3 to 2.9 (with ‘best estimates’ of 1.7-2.0) (Lee, et al., 2009) (Azar &
Johansson, 2012).
An alternative methodology to determine the climate change impact from aviation uses the Radiative Forcing
Index. This index represents the ratio between the total radiative forcing from aviation at some given time to
the radiative forcing from aviation emissions of CO 2 at the same time (Forster, et al., 2006). Studies have
suggested that the RFI lies between 2 and 4, indicating that the total climate impact of aviation at a certain
point is 2-4 times larger than the impact of its CO 2 emissions alone (IPCC, 1999; Sausen, et al., 2005).
However, one of the major weaknesses of the RFI is that it does not take into account the variation in the
lifetime of different emissions. This variation is substantial, lifetimes range from just a few hours (contrails) to
10 years (aircraft induced methane reduction and its associated effects on ozone) and even up to 200 years
(CO 2). Not taking into account these differences in lifetime, and simply multiplying the current amount of
CO2 emissions from aviation by a factor 2-4, would overestimate the long-term climate impact of aviation.
In this Handbook we use emission weighting factors to value the climate change impacts of aviation, as this
methodology accounts for the differing lifetime of emissions, whereas the radiative forcing index does not.
Although we acknowledge that the uncertainties for the emission weighting factors are somewhat larger when
aviation-induced cirrus is included, we believe this is an important effect to take into account. Therefore, we
will use a factor of 2 in this study to estimate the non-CO2 climate impacts of high altitude emissions from
aviation. This implies that the total CO 2 emissions from an aircraft are multiplied by a factor 2 to reach the
total CO 2 equivalent emissions. This is the same value that was used in previous editions of the Handbook
(Infras, CE Delft, ISI & University of Gdansk, 2008; Ricardo-AEA, TRT, DIW Econ & CAU, 2014) and in (HEATCO,
2006).

5.3.2 Input values
Three types of input values are needed in order to follow the methodology outlined above.
GHG emissions per vehicle type

In case the GHG emissions per vehicle type are not yet given, we need to first multiply the
vehicle emission factors, in grams of CO 2, CH 4 and N2O per kilometre driven, with the
amount of vehicle kilometres driven by that vehicle type in each country. The sources for
the emission factors of different modes of transport that are used in this study are shown in
Table 13. The emission factors are real world emission factors from fleet averages. They are
not based on vehicle type approval tests. The transport performance data and emission
factors per country that was used is based on the sources from Table 13.
GWP of GHG emissions

To allow for comparisons between the different GHGs, the GHGs need to be made
comparable to each other. The way to compare CO2-emissions with non-CO2 emissions is to
use the concept of GWP (see Textbox in Section 5.3.1). GWP of CO2 is standardised to 1.
In the IPCC’s latest Assessment Report (IPCC, 2013; Table 8.A.1, page 731) the GWP over a
100-year time period of CH 4 and N20 are 30 and 265 respectively. In our calculations we will
use these GWP’s to be able to add the amounts of the different gases together, in order to
present the results in terms of CO 2 equivalents. Currently European Union Legislation for
Monitoring and Reporting Greenhouse Gas Emissions17 uses GWPs from the 2006 IPCC
Guidelines for National Greenhouse Gas Invenstories. For the purposes of this handbook, it
was, however decided to use the more recent 2013 IPCC estimates cited above. This can
make a difference to the impacts caused by non-CO2 greenhouse gases.
Climate change costs per tonne of CO 2 equivalent

There are two major ways that the climate change costs can be monetised: either using a
damage cost approach or an avoidance cost approach. Both methods are discussed in more
detail in Annex D.3. Like in the previous versions of this Handbook, we use the avoidance
cost approach. Damage costs have serious limitations because potentially catastrophic
effects, such as the melting of the polar ice caps in Greenland or West Antarctica or
changes in climate subsystems such as El Niño Southern Oscillation cannot be well
incorporated. The GHG emission reductions agreed in the Paris Agreement are based on
preventing temperature rises above 1.5-2 degrees Celsius. Exceeding this level is considered
to be too risky for future generations. Therefore, it makes sense to formulate climate
change costs as avoidance costs, based on the target agreed in the Paris Agreement.
Limiting temperature rise to 1.5-2 degrees Celsius roughly equates to no more than 450 ppm
(parts per million) CO 2 in the atmosphere. A wide range of literature on avoidance costs is
available. The avoidance costs used in this Handbook are based on an analysis of recent
literature which revealed that the central value for the short-and-medium-run costs (up to
2030) is € 100/tCO 2 equivalent (€2016). The central value for the long run costs (up to 2060)
________________________________
17
COMMISSION IMPLEMENTING REGULATION (EU) No 749/2014 of 30 June 2014 on structure, format, submission
processes and review of information reported by Member States pursuant to Regulation (EU) No 525/2013 of the
European Parliament and of the Council, Recital (2) https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=
CELEX:32014R0749&from=EN

is € 269/tCO 2 equivalent (€2016). Table 24 shows a low and high estimate for these time
periods, although we use the central estimates for the short-and-medium-run in this
Handbook. These values were derived by calculating the average of the low, central and
high estimates for the relevant time periods of the values from the literature, but excluding
the lowest and highest values to eliminate outliers. For a full literature review of avoidance
costs and our analysis we refer to Annex D.
Table 24 – Climate change avoidance costs in €/tCO2 equivalent (€2016 )
Low Central High

Short-and-medium-run (up to 2030) 60 100 189
Long run (from 2040 to 2060) 156 269 498
5.3.3 Output values

The following tables show the resulting cost factors (output values) for the climate costs
per transport mode and vehicle type. The tables include the total costs as well as the
average costs per vkm and per pkm or tkm. The calculations have been made based on the
cost factor of € 100 per t CO2 equivalent, the central value for short and medium run
estimations (see section above).
Table 25 - Total and average climate change costs for land-based modes for the EU28
Total costs EU28 Average costs
Bus 0.84 0.47 8.83
Coach 1.61 0.44 8.66
LCV 13.17 3.98 2.75
LCV – petrol 0.71 3.76 2.56
LCV – diesel 12.45 3.99 2.77
HGV 9.63 0.53 6.48

Total road, rail, inland 83.14

waterway
Table 26 – Total and average climate costs for aviation for selected 33 EU airports

Short haul (< 1,500 km) 2.14 2.39 1,315
Medium haul (1,500–5,000 km) 5.50 1.85 3,341
Long haul (> 5,000 km) 14.37 2.24 17,629
Total 22.01 2.14 5,383
Table 27 presents rough estimates for the average and total external climate change costs
of maritime transport. These data are only available for freight. The average cost has been
based on the cost for reference cases presented in Section 5.4 and data on the number of
port calls for the selected ports from Eurostat. The total climate change cost has been
based on the average cost and the number of tkms provided by DG MOVE18. The available
data does not allow an estimate of costs at the individual port level.
Table 27 – Rough estimates for total and average external climate change costs for maritime transport for 34
selected EU ports

Freight ship 11 0.16
5.4 Marginal climate change costs for selected cases

For climate change costs, the marginal costs are the same as the average costs. This is
because the average and marginal climate emissions per kilometre of a vehicle are equal.
This implies that an additional kg of CO 2 emitted leads to the same social (external) costs as
the average kilogram CO 2 emitted, since the CO 2 is distributed in the whole atmosphere.
Furthermore, the avoidance costs used in this Handbook are based on the entire economy
and are not significantly dependent on the emissions of the transport sector.
The average climate costs for different modes and within the modes for different vehicle
types, are calculated by multiplying the emission factors (in gram CO 2 equivalent per unit)
with the avoidance costs of CO2. These emission factors are deriven from the following
sources:
— road transport: COPERT database;
— rail transport: TREMOD database;
— inland waterways: EcoTransit World database;
— aviation: TREMOD database.
________________________________
18
See footnote 16.

to be emphasised that the Euro standard is not a cost driver for the climate costs. There are
some differences between the results for the different Euro standards though, that are the
result of the COPERT emission data. These differences are related to the improved energy
efficiency over time and impacts of emission reduction technology on fuel efficiency.
The size classes for trucks from COPERT do not match with those for the Eurostat transport
performance data used for this Handbook. The load factors for trucks have therefore been
based on an interpolation of the Eurostat data.
Annex J contains the marginal climate cost data for road vehicles for reference cases that
are defined in terms of the combination of fuel type and fuel efficiency of the vehicle
(which are the main cost drivers for climate cost).
Table 28 on marginal climate change costs for road transport shows the costs per pkm or
tkm (except for LCV, where costs per vkm are presented due to the fact that LCV have
characteristics of freight and passenger transport). The costs per vkm for the different
vehicle categories of road transport are available in the background Excel file.
Table 28 - Marginal climate change costs road transport for selected cases
Vehicle Fuel type Size Emission Motorway Urban road Other road
category class
Passenger Petrol Mini Euro 4 0.87 0.84 0.72

Cars
< 0.8 l Euro 5 0.87 0.84 0.72
Euro 6 0.87 0.84 0.72
Small Euro 0 1.25 1.53 1.06
0.8–1.4 l Euro 1 0.94 1.06 0.83
Euro 2 0.90 1.05 0.77
Euro 3 0.91 1.05 0.82
Euro 4 0.96 1.09 0.85
Euro 5 0.96 1.09 0.85
Euro 6 0.96 1.09 0.85
Medium Euro 0 1.55 1.80 1.26
1.4–2.0 l Euro 1 1.07 1.29 0.96
Euro 2 0.98 1.25 0.93
Euro 3 1.08 1.26 0.96
Euro 4 1.11 1.29 1.02
Euro 5 1.11 1.29 1.02
Euro 6 1.11 1.29 1.02
Large-SUV- Euro 0 1.79 2.21 1.50

category class
Executive Euro 1 1.36 1.66 1.21
> 2.0 l Euro 2 1.34 1.70 1.27
Euro 3 1.14 1.54 1.11
Euro 4 1.31 1.89 1.31
Euro 5 1.31 1.89 1.31
Euro 6 1.31 1.89 1.31
Diesel Mini Euro 4 0.78 0.69 0.66
< 0.8 l Euro 5 0.78 0.69 0.66
Euro 6 0.78 0.69 0.66
Small Euro 0 1.05 1.10 0.87
0.8–1.4 l Euro 1 1.05 1.10 0.87
Euro 2 1.05 1.17 0.90
Euro 3 0.97 1.11 0.88
Euro 4 0.97 1.11 0.88
Euro 5 0.97 1.11 0.88
Euro 6 0.97 1.11 0.88
Medium Euro 0 1.22 1.29 1.02
1.4–2.0 l Euro 1 1.22 1.29 1.02
Euro 2 1.23 1.33 1.04
Euro 3 1.18 1.31 1.03
Euro 4 1.18 1.31 1.03
Euro 5 1.18 1.31 1.03
Euro 6 1.18 1.31 1.03
Large-SUV- Euro 0 1.40 1.49 1.18

Executive
Euro 1 1.40 1.49 1.18
> 2.0 l
Euro 2 1.40 1.49 1.18
Euro 3 1.39 1.50 1.18
Euro 4 1.39 1.50 1.18
Euro 5 1.39 1.50 1.18
Euro 6 1.39 1.50 1.18
Petrol Mini Euro 4 0.57 0.44 0.44

Hybrid
Small Euro 4 0.57 0.44 0.44

category class
(PHEV) Large-SUV- Euro 4

Executive 0.57 0.44 0.44
LPG Small Euro 1 n.a. n.a. n.a.

Bifuel
Euro 2 n.a. n.a. n.a.
Euro 4 0.93 1.03 0.81
Euro 5 0.89 0.98 0.77
Euro 6 0.85 0.93 0.73
CNG Small Euro 4 0.81 0.89 0.69

Biofuel
Euro 5 0.78 0.85 0.67
Euro 6 0.74 0.81 0.63
Electric n.a. n.a.

(BEV) 0.00 0.00 0.00
Moped Petrol 2-stroke Euro 0 0.78 0.78 0.78

< 50 cm³
Euro 1 0.59 0.59 0.59
Euro 2 0.59 0.59 0.59
Euro 3 0.59 0.59 0.59
4-stroke Euro 0 0.78 0.78 0.78

< 50 cm³
Euro 1 0.59 0.59 0.59
Euro 2 0.59 0.59 0.59
Euro 3 0.59 0.59 0.59
Motorcycle Petrol 2-stroke Euro 0 1.18 0.90 0.84

> 50 cm³
Euro 1 1.08 0.82 0.77
Euro 2 1.06 0.80 0.75
Euro 3 1.06 0.80 0.75
4-stroke Euro 0 1.25 0.86 0.94

< 250 cm³
Euro 1 1.02 0.71 0.77
Euro 2 0.77 0.63 0.61
Euro 3 0.76 0.61 0.60
4-stroke Euro 0 1.35 1.23 1.07

250-750
cm³ Euro 1 1.34 1.12 1.01
Euro 2 1.23 1.02 0.92
Euro 3 1.21 1.00 0.90

category class
4-stroke Euro 0 1.56 1.42 1.23

> 750 cm³
Euro 1 1.36 1.38 1.12
Euro 2 1.41 1.32 1.10
Euro 3 1.39 1.32 1.09
Electric n.a. n.a. 0.00 0.00 0.00
Urban buses Diesel Midi <=15 t Euro 0 0.46 0.79 0.48
Euro I 0.38 0.62 0.39
Euro II 0.37 0.59 0.38
Euro III 0.39 0.63 0.40
Euro IV 0.39 0.57 0.40
Euro V 0.37 0.56 0.36
Euro VI 0.37 0.57 0.37
Diesel Standard Euro 0 0.42 0.80 0.46

15-18 t
Euro I 0.37 0.67 0.40
Euro II 0.37 0.64 0.40
Euro III 0.39 0.67 0.42
Euro IV 0.39 0.61 0.42
Euro V 0.36 0.60 0.38
Euro VI 0.37 0.61 0.38
Diesel Articulated Euro 0 0.43 0.82 0.48

>18 t
Euro I 0.39 0.70 0.43
Euro II 0.39 0.67 0.43
Euro III 0.40 0.70 0.44
Euro IV 0.41 0.65 0.45
Euro V 0.37 0.64 0.41
Euro VI 0.38 0.65 0.41
CNG CNG buses Euro I 0.99 0.99 0.99
Euro II 0.89 0.89 0.89
Euro III 0.74 0.74 0.74
EEV 0.39 0.67 0.42
Biodiesel Biodiesel Euro 0 0.37 0.70 0.40

buses
Euro I 0.33 0.58 0.35

category class
Euro II 0.33 0.56 0.35
Euro III 0.34 0.58 0.36
Euro IV 0.34 0.54 0.36
Euro V 0.31 0.52 0.33
Euro VI 0.32 0.54 0.33
Electric Small n.a. 0.00 0.00 0.00
Medium n.a. 0.00 0.00 0.00
Large n.a. 0.00 0.00 0.00
Coaches Diesel Standard Euro 0 0.41 0.90 0.46

<=18 t
Euro I 0.37 0.81 0.42
Euro II 0.37 0.80 0.43
Euro III 0.40 0.88 0.46
Euro IV 0.40 0.81 0.45
Euro V 0.40 0.80 0.44
Euro VI 0.40 0.82 0.46

> 18 t
Euro I 0.29 0.63 0.33
Euro II 0.29 0.62 0.33
Euro III 0.28 0.66 0.33
Euro IV 0.28 0.61 0.32
Euro V 0.30 0.61 0.34
Euro VI 0.30 0.62 0.35
Light commercial vehicles (€-cent per vkm)
Light Petrol Euro 0 2.38 3.52 2.15

Commercial
Vehicles Euro 1 2.38 3.52 2.15
Euro 2 2.37 3.47 2.12
Euro 3 2.34 3.40 2.07
Euro 4 2.34 3.38 2.06
Euro 5 1.47 1.89 1.36
Euro 6 1.47 1.89 1.36
Diesel Euro 0 2.82 2.57 1.89
Euro 1 2.82 2.57 1.89

category class
Euro 2 2.82 2.57 1.89
Euro 3 2.82 2.58 1.88
Euro 4 2.82 2.58 1.88
Euro 5 2.31 2.40 2.03
Euro 6 2.31 2.40 2.03
Electric n.a. 0.00 0.00 0.00
HGV Diesel Rigid Euro 0 4.52 5.48 4.36

<=7,5 t
Euro I 4.18 4.45 3.63
Euro II 4.05 4.17 3.51
Euro III 4.26 4.46 3.67
Euro IV 4.33 4.19 3.67
Euro V 4.30 4.03 3.56
Euro VI 4.29 4.12 3.59
Rigid Euro 0 2.32 3.22 2.33

7,5-12 t
Euro I 2.10 2.67 2.05
Euro II 2.06 2.53 1.99
Euro III 2.13 2.68 2.08
Euro IV 2.13 2.53 2.06
Euro V 2.22 2.54 1.96
Euro VI 2.23 2.59 1.98
Rigid Euro 0 1.33 1.90 1.34

12-14 t
Euro I 1.19 1.60 1.18
Euro II 1.16 1.52 1.14
Euro III 1.19 1.61 1.19
Euro IV 1.19 1.50 1.17
Euro V 1.10 1.47 1.15
Euro VI 1.11 1.50 1.16
Rigid Euro 0 1.50 2.40 1.58

14-20 t
Euro I 1.27 1.92 1.31
Euro II 1.24 1.82 1.28
Euro III 1.27 1.93 1.32

category class
Euro IV 1.25 1.76 1.29
Euro V 1.18 1.77 1.26
Euro VI 1.19 1.77 1.27
Rigid Euro 0 0.83 1.40 0.90

20-26 t
Euro I 0.73 1.18 0.78
Euro II 0.71 1.13 0.76
Euro III 0.72 1.18 0.78
Euro IV 0.71 1.10 0.76
Euro V 0.69 1.10 0.75
Euro VI 0.69 1.11 0.75
Rigid Euro 0 0.62 1.05 0.68

26-28 t
Euro I 0.54 0.89 0.59
Euro II 0.53 0.85 0.58
Euro III 0.54 0.89 0.59
Euro IV 0.54 0.84 0.58
Euro V 0.53 0.82 0.57
Euro VI 0.53 0.84 0.58
Rigid Euro 0 0.57 0.92 0.62

28-32 t
Euro I 0.51 0.81 0.55
Euro II 0.52 0.80 0.54
Euro III 0.51 0.81 0.56
Euro IV 0.51 0.77 0.55
Euro V 0.50 0.75 0.54
Euro VI 0.50 0.77 0.55
Rigid >32 t Euro 0 0.49 0.84 0.54
Euro I 0.44 0.73 0.48
Euro II 0.43 0.70 0.47
Euro III 0.44 0.73 0.48
Euro IV 0.43 0.69 0.47
Euro V 0.42 0.69 0.46
Euro VI 0.42 0.69 0.47
Articulated Euro 0 0.80 1.33 0.86

category class
14-20 t Euro I 0.69 1.09 0.73
Euro II 0.67 1.04 0.71
Euro III 0.69 1.10 0.74
Euro IV 0.68 1.02 0.72
Euro V 0.66 1.00 0.71
Euro VI 0.66 1.02 0.71

20-28 t
Euro I 0.71 1.19 0.77
Euro II 0.68 1.12 0.75
Euro III 0.70 1.17 0.77
Euro IV 0.69 1.10 0.76
Euro V 0.68 1.08 0.74
Euro VI 0.68 1.09 0.75

28-34 t
Euro I 0.44 0.75 0.49
Euro II 0.44 0.72 0.47
Euro III 0.43 0.74 0.48
Euro IV 0.43 0.70 0.48
Euro V 0.43 0.69 0.47
Euro VI 0.43 0.69 0.47

34-40 t
Euro I 0.43 0.75 0.48
Euro II 0.42 0.73 0.47
Euro III 0.42 0.76 0.48
Euro IV 0.42 0.71 0.47
Euro V 0.42 0.71 0.46
Euro VI 0.42 0.72 0.47

40-50 t
Euro I 0.40 0.71 0.45
Euro II 0.41 0.71 0.45
Euro III 0.40 0.72 0.46
Euro IV 0.40 0.68 0.45

category class
Euro V 0.40 0.68 0.45
Euro VI 0.40 0.69 0.45

50-60 t
Euro I 0.43 0.77 0.47
Euro II 0.42 0.74 0.47
Euro III 0.43 0.77 0.47
Euro IV 0.42 0.73 0.47
Euro V 0.42 0.72 0.47
Euro VI 0.43 0.73 0.47
LNG Articulated
32+ n.a. 0.21 0.36 0.23
Table 29 – Marginal climate change costs rail transport for selected cases
Train type Traction €-cent/pkm or €-cent/train-km
€-cent/tkm
Passenger transport
Intercity train Diesel 0.201 17.5
Regional train Diesel 0.735 22.8
Freight transport
Long container Diesel 0.158 118.2
Long bulk Diesel 0.087 122.5
Short container Diesel 0.074 103.2
Short bulk Diesel 0.066 105.9
Climate change costs for electric trains are zero. Emissions occur only during electricity
generation, which is covered in Chapter 8 including the cost of well-to-tank emissions.
Table 30 – Marginal climate change costs IWT for selected cases
Vessel type Type of cargo €-cent/tkm €-cent/vkm
CEMT II (350 t) Bulk 0.34 107
Container 0.21 107
CEMT IV (600 t) Bulk 0.20 159
CEMT Va (1,500 t) Bulk 0.18 290

Container 0.21 290
Pushed convoy (11,000 t) Bulk 0.15 777
Container 0.11 777
Table 31 – Marginal climate change costs maritime transport for selected cases
Vessel type Distance at sea € per port call €-cent per € per
(km) pkm or tkm vessel-km
Passenger transport
RoPax Ferry (25,500 gt) 100 5,800 10.98 58.0
500 14,598 5.53 29.2
Freight transport
Small container vessel (28,500 gt) 500 40,876 0.34 81.8
3,000 134,753 0.19 44.9
Large container vessel (143,000 gt) 500 61,795 0.11 123.6
3,000 291,770 0.08 97.3
15,000 1,395,651 0.08 93.0
Small bulk vessel (18,000 gt) 500 11,939 0.16 23.9
3,000 51,631 0.11 17.2
Large bulk vessel (105,000 gt) 500 29,092 0.06 58.2
3,000 119,549 0.04 39.8
15,000 553,747 0.04 36.9
Table 32 – Marginal climate change costs aviation for selected cases
Type of flight Distance Emission Example of aircraft type €-cent per pkm € per pax
class
(km)
Short haul 500 Low Bombardier CRJ900 2.84 14
500 High Embraer 190 3.44 17
Medium haul 1,500 Low Airbus 320 1.53 23
1,500 High Boeing 737 2.18 33
3,000 Low Airbus 320 1.43 43
3,000 High Boeing 737 2.04 61
Long haul 5,000 Low Airbus 340 1.17 58
5,000 High Boeing 777 1.32 66

15,000 Low Airbus 340 1.56 234
15,000 High Boeing 777 1.77 265

We have calculated the climate change costs according to the most recent and high quality
evidence and methods. Nonetheless, there are a few aspects that merit a point of
discussion regarding the robustness of the results presented in this chapter.
Firstly, we have used avoidance costs, rather than damage costs, to monetise the costs of
climate change. Our literature review confirmed that, the use of avoidance costs is a
superior method to the use of damage costs (see full discussion in Section D.3).
However, uncertainties will always remain. We have attempted to take away some of that
uncertainty by providing high and low case climate change costs, which can be used as a
sensitivity analysis.
Furthermore, there can be political reasons, related to distributional or competitiveness

aspects, which lead to political decisions to apply different mitigation costs in different
sectors

6 Noise costs
6.1 Introduction
Traffic noise is generally experienced as a disutility and is accompanied by significant costs.
Noise emissions from traffic pose a growing environmental problem due to the combination
of a trend towards greater urbanisation and an increase in traffic volumes. Whilst the
increase in traffic volume results in higher noise levels, the increase in urbanisation results
in a higher number of people experiencing disutility due to noise. As a result, the costs of
traffic noise are expected to grow in the future despite potential noise-reducing
improvements in vehicles, tyres and roads.
In this chapter we provide an overview of the recommended approaches to value noise

costs, as well as an overview of recommended noise cost figures. In Section 6.2 we first
briefly discuss the definition and scope of noise costs. The total and average noise costs are
explored in Section 6.3, and the marginal noise costs are the topic of Section 0. Finally, the
robustness of the noise cost figures presented in this chapter is analysed in Section 6.5.
More detailed information on noise costs can be found in Annex E.

In general, noise can be defined as unwanted sounds of varying duration, intensity or
other quality that causes physical or psychological harm to humans (CE Delft, INFRAS &
Fraunhofer ISI, 2011). In this study, we will consider noise costs for the following transport
modes: road, rail and aviation. Noise costs for inland waterway transport and maritime
transport are considered negligible or non-existent as they usually take place in sparsely
populated areas and the noise emission factors for those transport modes are relatively low
and so are not covered in this chapter.
Unit
The basic measurement index for noise is the decibel (dB). It is indexed logarithmically, reflecting the
logarithmic manner in which the human ear responds to sound pressure. Within the human range of hearing,
deep and very high tones at the same sound intensity are experienced as less noisy. To correct for this
sensitivity, a frequency weighting is applied to measurements and calculations. The most common frequency
weighting is the ‘A weighting’, dB(A).
The logarithmic nature of noise is also reflected in the relationship between noise and traffic volume. Halving or
doubling the amount of traffic results in a change of 3 dB, irrespective of the cu rrent flow. Thus an increase in
traffic volume from 50 to 100 vehicles results in the same increase in noise level as doubling transport volume
from 500 to 1,000 vehicles.
An important aspect is the time of day at which the noise takes place. In this stu dy we employ the measure Lday,
(Lden), the current legal measure for traffic noise. Lden is a weighted average of the total noise during
evening, night
day, evening and night times. One fundamental feature of Lden is that it assumes that evening- and night-time
noise is more of a nuisance than day-time noise19.
________________________________
19
Evening noise is given a penalty of 5 dB(A). Night-time noise is given a 10 dB(A) penalty.

The thresholds above which noise is considered a nuisance are somewhat arbitrary, previous
literature has employed thresholds of 50, 55 and 60 dB(A). It is important to note that the
choice of a threshold has a substantial impact on marginal noise costs. In this study we
propose to use the threshold of 50 dB(A). This threshold was chosen as it is one that is least
likely to result in an underestimation of the noise costs. However, the EEA Noise Maps which
are used as input data in this study only start from 55 dB(A), therefore we only have the
number of people exposed to noise above 55 dB(A). Ideally, the number of people exposed
to noise levels between 50 and 55 dB(A) would also be included, however, due to data
limitations of the EEA noise maps, this was not feasible in this edition of the Handbook.
Numerous studies have proposed the concept of a rail bonus, the notion that noise as a
result of rail transport is experienced as less of a nuisance than road noise. It gives rail
transport a 5 dB ‘discount’ in comparison to road noise and was widely used in noise
directives. In contrast to previous editions of the Handbook we don’t incorporate the rail
bonus in this edition of the Handbook. This is based on an extensive literature review,
which suggests that recent literature cannot support the upholding of the rail bonus
(see Annex E.4).
Effects of noise
The exposure to noise results in a number of health endpoints due to prolonged and
frequent exposure to transport noise. These health endpoints can take a multitude of
forms. Health endpoints for which significant evidence is available are ((WHO, 2011; (WHO,
2017-2018); (Defra, 2014)):
— ischaemic heart disease;
— stroke;
— dementia;
— hypertension;
— annoyance.
Grouping annoyance under the health endpoints of noise exposure is consistent with the
WHO definition of health as “a state of complete physical, mental and social well-being and
not merely the absence of disease or infirmity” (WHO, 1946). Annoyance represents the
disturbance individuals experience when they are exposed to traffic noise. It can hinder
people in performing certain activities, which may lead to a variety of negative responses,
including irritation, disappointment, anxiety, exhaustion and sleep disturbance (WHO,
2011). However, annoyance is measured in a different way to the other ‘more classical’
health impacts, and therefore we have made the somewhat arbitrary distinction to look at
annoyance separately from the other health impacts. This was decided because the
valuation of annoyance applied is very different from the valuation of the other health
endpoints.
The cost components are closely linked. For instance, sleep disturbance is classified as a
health endpoint according to (Defra, 2014), although there is likely to be significant overlap
with annoyance. These two impacts are difficult to separate. In WTP studies looking at
noise it is complicated to separate individual’s valuation for annoyance from sleep
disturbance. If one is asked about their annoyance they are inclined to also take into
account the effects of sleep disturbance. Therefore, there is an implicit risk of double
counting the valuation if both sleep disturbance and annoyance impacts are explicitly taken
into account. To avoid double counting we employ the conservative assumption that we
include both the annoyance and health costs of noise, but exclude sleep disturbance from

the health endpoints. It is possible that this leads to a small underestimate of the true costs
of noise.
For health endpoints not mentioned in the list above, e.g. breast cancer and depression,
only fragmented evidence is available. Therefore, these costs are not included in the noise
costs estimated in this study. For the same reason, productivity losses (e.g. due to loss of
concentration) and environmental impacts of traffic noise (e.g. harmful effects on wildlife)
are not covered. Finally, direct material damages as a result of vibrations are not included
in the costs of noise in this study, as the vibrations are not necessarily an effect of noise,
but rather an external effect on its own. For a more detailed discussion on the effects of
noise we refer to Annex E.
6.3 Total and average noise costs
6.3.1 Methodology
Total and average noise costs are calculated using a bottom-up approach. Figure 6
illustrates the corresponding methodology.
There are two types of input value: the number of people exposed to noise for each
transport mode, and the noise costs per person exposed. The noise classes that people are
exposed to are classified in bins, e.g. of 5 dB(A). For each noise class and transport mode,
the total number of people exposed has to be calculated. The other input value, the noise
costs per person exposed, consists of two values, an annoyance value and a health value.
The annoyance value is calculated using a WTP approach, where respondents are asked how
much they are willing to pay for changes in the noise level. The health value is based on an
environmental burden of disease method and are taken from (Defra, 2014). Summing the
health value and the annoyance value results in the total noise costs per person exposed.
These costs per person are multiplied with the number of people exposed to the
corresponding noise level. Summing these costs together gives the total external noise costs
for a transport mode (road, rail or aviation). The total costs are allocated to specific
vehicle categories (e.g. passenger cars, motorcycles, busses) based on weighting factors
based on (CE Delft, INFRAS & Fraunhofer ISI, 2011; CE Delft & VU, 2014; VROM, 2006) in
order to estimate the total costs per vehicle category. Finally, average noise costs are
estimated by dividing the total costs by total transport performance (e.g. pkm, tkm, etc.).

Figure 6 – Methodology total and average noise costs
6.3.2 Input values

Three types of input values are needed in order to follow the methodology described above.
People exposed
In order to calculate total and average noise costs for a country, the number of people
exposed to a certain noise level (in bins of 5 dB(A)) originating from a certain transport
mode is needed. This data is preferably based on national data (empirical data or specific
national model calculations). A second-best option (applied in this study to estimate total
and average noise costs for European countries) is to make use of EU-wide data that is
available from the noise maps from the EEA. Directive 2002/49/EC (EC, 2002) requires
Member States to provide data on the number of people exposed to road, rail or aviation
noise in their countries to the European Commission. This data is highly useful, although
there are also data gaps. For instance, not all data has been reported and not all cities and
urban regions are included in the scope of the noise directive. Therefore, we have carried
out corrections in order to make the data more complete. For details on the corrections,
we refer to Annex E.
Environmental prices
The environmental price of noise reflects the welfare loss that occurs with one extra
decibel of noise (CE Delft, 2018). The environmental price of noise needs to be determined
implicitly, as there is no market for noise prevention. Previous editions of the Handbook
have recommended using environmental prices based on HEATCO (2006), both for
annoyance and health endpoints. HEATCO assumes a constant valuation per dB of noise
for annoyance costs, which has recently been disputed. This Handbook therefore uses
increasing prices per dB based on the most recent insights provided by (Bristow, et al.,
2015) for annoyance costs. As for health costs, the prices according to (Defra, 2014) match
the WHO’s recommendations in their latest systematic reviews, and are therefore used in
this Handbook. For a more detailed discussion on the evidence for the environmental prices
of noise we refer to Annex E.

The recommended environmental prices for noise (EU28 averages), including both health
and annoyance costs, are shown in Table 33. Tables with the environmental price of noise
per country are provided in the database.
Table 33 – Environmental price of traffic noise for the EU28 (€ 2016 /dB/person/year)
Lden Road transport Rail transport Aviation

(db(A)) Annoyance Health Total Annoyance Health Total Annoyance Health Total
50-54 14 3 17 14 3 17 34 5 39
55–59 28 3 31 28 4 32 68 6 74
60–64 28 6 34 28 6 34 68 9 77
65–69 54 9 63 54 9 63 129 12 141
70–74 54 13 67 54 13 67 129 16 145
≥ 75 54 18 72 54 18 72 129 21 150
Weighting factors for different vehicles

To be able to allocate the total noise costs per transport mode to noise costs per vehicle
class, the total kilometres travelled by each vehicle class needs to be known.
However, noise originating from certain types of vehicles (e.g. trucks) is considered more of
a nuisance than noise from others (e.g. passenger cars). To reflect this, different weighting
factors are applied. Like in previous versions of this Handbook, it is assumed that the
weighting factors are identical for all countries. Table 34 illustrates the weighting factors
used in this study, which are based on (CE Delft, INFRAS & Fraunhofer ISI, 2011; CE Delft &
VU, 2014; VROM, 2006).
Table 34 – Weighting factors for noise for different vehicle types
Urban (50 km/h) Other roads (80 km/h or higher)
Road
Passenger car 1.0 1.0
Passenger car petrol 1.0 1.0
Passenger car diesel 1.2 1.0
Motorcycle 13.2 4.2
LCV 1.5 1.2
Bus/coach 9.8 3.3
HGV 3.5–7.5 t 9.8 3.0
HGV 7.5–16 t 13.2 4.2
HGV 16–32 t 14.9 4.8
HGV > 32 t 16.6 5.5
Rail
Passenger train 1

Freight train 4
6.3.3 Output values

Table 35 describes the total and average noise costs in the EU as a whole for road and rail
transport. Costs at the country level are provided in the database. For road transport, the
total costs originating from transport via a motorway are also provided in the database.

Table 35 - Total and average noise costs for land-based modes for the EU28

Bus 0.8 0.4 8.0
Coach 0.9 0.2 4.7
High speed passenger train 0.4 0.3 97
Passenger train electric 2.6* 0.8 106
Passenger train diesel 0.9 1.4 81
LCV 5.4 1.6 1.1
HGV 3.5–7.5 t 1.0 1.2 4.0
HGV 7.5–16 t 1.8 0.8 5.7
HGV 16–32 t 3.0 0.4 6.5
HGV > 32 t 3.2 0.4 7.2
Freight train electric 2.1 0.6 359
Freight train diesel 0.4 0.4 201
waterway 63.6
trains).
Table 36 illustrates the average costs of both passenger and freight aviation. Passenger
aviation values are provided per LTO, passenger and pkm. The average costs of freight
aviation are provided per LTO, tonne and tkm. Costs at the airport level are provided in the
annex/database.
Table 36 – Total and average noise costs for aviation for 33 selected EU airports

Passenger aviation Billion € €/LTO €/passenger €/tonne €-cent/pkm
Short haul 0.46
Medium haul 0.84 257 2.05 9.04 0.11
Long haul 0.01

6.4 Marginal noise costs
6.4.1 Methodology
Marginal noise costs differ from average noise costs for several reasons, butmainly because
local factors influence the noise level and the damage and annoyance level. There are three
main cost drivers for marginal noise costs:
— Population density: The population density close to the noise source is relevant for the
number of people exposed to the noise. The closer to an emission source people live,
the more nuisance will occur, and the higher the marginal costs will be. In the following
section we roughly distinguish between three area types (urban, suburban, rural),
representing different population densities. In general, the population density will be
highest in urban (metropolitan) areas and lowest in rural areas.
— Existing noise levels (depending on traffic volume, traffic mix and speed): If there is an
additional vehicle on an already busy urban road, the additional (marginal) noise costs
are small compared to a comparable situation along a rural road with little traffic.
The higher the existing background noise level, the lower the marginal costs of an
additional vehicle. As a proxy for the existing noise levels, three different area types
(urban, suburban, and rural) and two different traffic situations (thin or dense traffic)
are distinguished.
— Time of the day: Epidemiological studies show that the noise induced health effects
during the night are higher than during the day as a consequence of sleep disturbance.
Therefore, noise disturbances at night will lead to higher marginal costs than during the
day. In the following section, marginal noise costs are differentiated for night and day.
For road and rail transport the marginal noise costs are estimated based on the earlier
calculations of marginal costs in CE/INFRAS/ISI (2011) and INFRAS/IWW (2004). For deriving
up-to-date marginal noise costs based on the aforementioned marginal noise cost studies,
the development of the average noise costs per transport mode and vehicle type over time,
i.e. between the two older studies and the average noise costs calculated in the present
Handbook (see Section 6.3.3), has been taken into account.
Marginal noise costs for aviation depend heavily on local factors (e.g. population
density around airports), flight path, aircraft type and technology, and time of the day.
Additionally, the noise level of one single flight movement is much higher than the average
noise level. Therefore, it is very difficult to present an accurate (range of) marginal noise
cost values that could be applied for all situations. Some earlier studies estimate the
marginal noise costs of aviation (i.e. the cost per movement or per pax) to be around
30-60% of the average noise costs (per movement) for different transport situations or
aircraft types. This means those studies state that the marginal costs of an additional flight
movement are around a factor of 0.3 to 0.6 of the average noise costs (i.e. significantly
lower than the average costs). Additionally, there are different studies expressing directly
marginal noise costs for specific aircraft types, mainly based on a hedonic pricing approach
(e.g. (Pearce & B.Pearce, 2000) (CE Delft, 2003), (TRL, 2001). In the next chapter, marginal
noise costs are shown for both approaches.
6.4.2 Output values

The following tables include the marginal noise costs for road and rail transport,
differentiated by vehicle type, time of the day, traffic situation and area type.
Table 37 on marginal noise costs for road transport shows the costs per pkm or tkm (except
for LCV, where costs per vkm are presented due to the fact that LCV have characteristics of

freight and passenger transport). The costs per vkm for the different vehicle categories of
road transport are available in the background Excel file.
Table 37 – Marginal noise costs road transport – in €-cent (2016) per pkm, tkm or vkm (data for 2016)
Road Time of the Traffic situation Urban Suburban Rural

day
Passenger car Day Dense 0.5 0.03 0.004
Thin 1.1 0.07 0.009
Night Dense 0.9 0.05 0.007
Thin 2.1 0.13 0.015
Motorcycle Day Dense 7.4 0.4 0.06
Thin 18.0 1.2 0.14
Night Dense 13.5 0.8 0.11
Thin 32.7 2.1 0.24
Bus Day Dense 0.5 0.03 0.004
Thin 1.3 0.08 0.010
Night Dense 1.0 0.05 0.008
Thin 2.4 0.15 0.018
Coach Day Dense 0.3 0.02 0.002
Thin 0.7 0.04 0.005
Night Dense 0.5 0.03 0.004
Thin 1.2 0.08 0.009
LCV Day Dense 1.7 0.1 0.01
Thin 4.1 0.3 0.03
Thin 7.4 0.5 0.06
HGV average Day Dense 0.7 0.04 0.01
Thin 1.6 0.11 0.01
Thin 3.0 0.19 0.02
HGV 3.5–7.5 t Day Dense 1.5 0.08 0.01
Thin 3.6 0.23 0.03
Thin 6.5 0.42 0.05
HGV 7.5–16 t Day Dense 0.7 0.04 0.01
Thin 1.8 0.11 0.01
Thin 3.2 0.21 0.02
HGV 16–32 t Day Dense 0.6 0.03 0.00
Thin 1.3 0.09 0.01
Thin 2.4 0.16 0.02
HGV > 32 t Day Dense 0.6 0.03 0.00
Thin 1.4 0.09 0.01
Thin 2.6 0.17 0.02

Table 38 - Marginal noise costs rail transport – in €-cent (2016) per pkm and tkm (data for 2016)
Road Time of the Traffic situation Metropolitan Urban Rural

day
High speed train Day Dense 0.13 0.07 0.01
Thin 0.21 0.12 0.02
Night Dense 0.23 0.13 0.02
Thin 0.38 0.21 0.03
Conventional Day Dense 0.45 0.20 0.03
passenger train Thin 0.74 0.33 0.05
Night Dense 0.82 0.36 0.05
Thin 1.35 0.59 0.09
Freight train Day Dense 0.13 0.05 0.01
Thin 0.17 0.08 0.01
Night Dense 0.24 0.09 0.01
Thin 0.39 0.15 0.02
There are no studies available that give more differentiated data for marginal rail noise
costs (e.g. per train type or train length).
For marginal noise costs there is a number of studies that directly show the costs per
aircraft type. However, the corresponding studies are all at least 5-10 years old. Still, they
can be regarded as a good and sound basis for marginal costs of aircraft noise.
Table 39 – Marginal noise costs for aviation: data for different aircraft types
Aircraft type Marginal noise costs per LTO (€2016/LTO)
A 310 54
A 340 122
B 737-400 54
B 747-400 266
B 757 70
B 767-300 85
B 777 52
MD82 78
Source: Own calculations based on Pearce and Pearce (2000).
Alternatively, the marginal noise costs can be derived from the average costs, based on the
methodology of UIC (CE Delft, INFRAS & Fraunhofer ISI, 2011) assuming that the marginal
noise costs per movement (or per pkm or pax) are around 30 to 60% of the average noise
costs. Based on the results of the average noise costs of aviation (see Section 6.3 above),
this leads to the following range of marginal costs. The resulting costs per LTO are between
77 and 154 €2016, whereas in the table above the range is between 52 and 266 € 2016 (for most
between 52 and 122 € 2016). One reason for the slightly higher costs per LTO in the table
below might be the fact that it includes all relevant effects, health costs and annoyance
costs, whereas the data above are based on hedonic pricing studies, which do not cover the
health costs.
Table 40 – Marginal noise costs for aviation, estimations based on average costs (data for 2016)
Transport Unit Average costs Marginal costs

type
Lower boundary Upper boundary
Passenger €/LTO 257 77 154
€/passenger 2.1 0.6 1.2
Average 0.08 0.02 0.05
Short haul 0.09 0.03 0.06

€-cent/
pkm
Medium haul 0.03 0.008 0.017
Long haul 0.01 0.004 0.008

The robustness of the results presented in this chapter is largely dependent on the EEA’s
noise maps. There are a number of factors related to the noise maps that could influence
the robustness of the overall results.
Firstly, as there are regions for which data is missing in those maps, or regions that have no
reporting obligation, corrections have been carried out to take into account these aspects.
These corrections are more extensive than any other corrections that were carried out in
earlier editions of the Handbook. A full description of the corrections that have been
carried out can be found in Section E.2. Ideally, the noise maps would be complete for all
countries and all regions. However, given the current data situation, this is unfeasible.
Carrying out corrections to the data, as we have done in this Handbook, is likely to improve
the robustness of results compared to the situation with no corrections. Having a complete
EEA noise map is likely to further improve the robustness of the results.
Secondly, the latest noise maps are those for the year 2012, with data submitted up until
31/03/2017. It is possible that a substantial change in noise exposure has taken place
between 2012 and 2016. As the magnitude and direction of this change is unclear, working
with the 2012 noise maps is the best we can currently do. The robustness of results could be
significantly improved if more current noise maps were available.
Thirdly, the results for road transport are affected by the transport performance data used.
As explained in Section 1.3.4, in this study we use data from Eurostat, following the
nationality principle, i.e. transport activity is allocated to countries where the vehicle is
registered . The use of these data affects the results of this study, since the scope of these
data differs from the scope of the noie maps, which is in line with the territorial principle.
Particularly the results for HGVs may be significantly affected at country level. For
example, in countries with a lot of transit traffic (e.g. Austria) a significant part of the
noise costs should be allocated to foreign vehicles. By using transport performance data
based on the nationality principle, transport activity of these foreign vehicles is not taken
into account in the calculations.
Furthermore, the results for motorcycles should be interpreted with caution. This is
because they are highly influenced by the transport performance data we have collected for
motorcycles. Some national values show extremely high vehicle kilometres travelled by
motorcycles, e.g. Italy and Spain, which raises the question if these values might include
kilometres driven by mopeds too. Furthermore, almost half of the vehicle kilometres driven
by motorcycles at the EU28 level are driven on urban roads, where the noise weighting
factor for motorcycles is 13.2 times higher than the noise weighting factor for passenger
cars. This also significantly influences the noise costs for motorcycles.
In addition, it is important to note that the WHO recently presented new exposure response
functions. Unfortunately, within the scope of this project we could not fully develop new
cost factors based on new exposure response functions. Therefore we use the closest ready-
to-use values from the literature that best match the most recent WHO systematic reviews
2017-2018.20 Translating exposure-response functions to a valuation per dB unfortunately
fell outside of the scope of this study. However, we recommend that a future, more in
depth study into noise costs could attempt to translate the latest WHO exposure-response
values into a valuation per dB.
In addition, the noise maps only start measuring at 55 dB, with the first bin representing
55-59 dB. It is likely that the costs presented here are a lower bound estimate if a lower
bin of 50-54 dB would be included. This is because it is likely that a large group of the
population is exposed to noise within that noise bin.
An issue which builds on this is the fact that the EEA noise maps measure the number of
people exposed to certain levels of road, rail and aviation noise. It is possible that a share
of the people that are exposed to one type of noise (e.g. noise from road traffic), may also
be exposed to other types of noise (e.g. aviation). Whether there are groups of people that
are exposed to more than one noise source is currently unclear. What is further unclear is
whether or not the valuations for noise are still valid when there are multiple sources at
work. In this study we have implicitly assumed that people are only exposed to one source
of noise. However, if future studies reveal that a percentage of people are exposed to more
than one noise source, further research should be conducted on whether or not the
valuations for the different noise sources can simply be added together.
As for specific country values, the following are worth noting. For Greece, the number of
people exposed to road noise are not included in the noise maps. Therefore, EU28 average
values were used as a proxy for average noise costs in Greece. To calculate the total noise
costs, the average values were multiplied with the transport performance data for Greece.
Although this is unlikely to significantly affect the EU28 noise costs, it is an important note
of caution to keep in mind when looking at the specific Greek situation. More research
could be conducted to achieve road noise exposure data for Greece.
For aviation there was noise exposure data missing for Zagreb Plesno Airport and Ljubljana
Brink Airport. To calculate the noise costs for these airports, an average of eastern
European airports (Prague, Budapest, Warsaw, Bucharest, Riga, Sofia, Vilnius, Tallinn and
Bratislava) was first calculated. This was preferred to a selected EU airport average as
there are some airports (e.g. Heathrow and Luxembourg) that have notoriously high noise
costs, significantly affecting the average. The eastern European average noise cost was then
multiplied with the transport performance data on aviation to calculate the total noise
costs for the airports in Zagreb and Ljubljana.
________________________________
20
Please note that this is also the approach used for other cost categories.
7 Congestion costs
7.1 Introduction
Congestion is defined as a condition where vehicles are delayed when travelling.
In particular, a congestion cost arises when an additional vehicle reduces the speed of the
other vehicles of the flow and hence increases their travel time. Road congestion cost can
be defined on the basis of a speed-flow relationship in a given context21, for example at an
urban or inter-urban level. This approach cannot be expanded to other transport modes,
like rail and air, as they essentially provide scheduled services and are planned on the basis
of the allocative capacity of networks and nodes.
It is worth remarking that the road congestion costs presented in the following sections are
outputs of a model designed to estimate the overall magnitude of this externality at EU-
wide level and from which we derive representative average and marginal cost figures at
national level. Because of the high-level scale of analysis of the model used to carry out the
estimations (which for example does not incorporate network effects due to the application
of pricing policies), it is recommended to develop specific models for context-specific
evaluations.
This chapter has been organised as follows. In Section 7.2 we first briefly discuss the
definition and scope of congestion costs with respect to road and other transport modes.
The total and average road congestion costs are presented in Section 7.3 and the social
marginal congestion cost is the topic of Section 7.4. The robustness of the road congestion
cost figures presented in this chapter is analysed in Section 7.4.4. More detailed
information on road congestion costs can be found in Annex F.
It is important to note that road congestion can also have impacts on other externalities.
For instance, a variation of the level road of congestion implies a variation of the emissions
of pollutants (local and global) and road accidents, and therefore of their external costs.
These costs are handled where possible in the other chapters of the Handbook.

A variety of definitions of road congestion exist (Grant-Muller & Laird, 2007). From the
objective perspective we assumed, road congestion can be defined as the impedance that
vehicles impose on each other, as the traffic flow approaches the maximum capacity of the
network (adapted from (Goodwin, 2004)).
As far as the other transport modes are concerned, the approach to estimate the external
costs needs to be different because for scheduled services the congestion cost should, in
principle, not be an issue until a perturbation occurs and propagates through the system.
________________________________
21
Road congestion can be defined also on the basis of a bottleneck model, but it cannot be generalised as the
free-flow model to produce countrywide and Europe-wide results.
7.2.1 Road congestion costs
With respect to the road mode, the relationship between speed and flow is to be
interpreted as follows. Until the flow is significantly lower than the capacity of a link,
the vehicles can travel at free-flow speed. As the flow increases, the speed gradually
decreases.
The average travel cost borne by the road users is based on the product of the value of
time, which can be assumed constant across road users, and the average travel time.
Basically, there are two approaches to estimate road congestion costs, namely the delay
cost and the deadweight loss.
In Figure 7, the cost of travel is equal to 𝑝 0 until the free-flow condition holds. When the
flow increases, the speed reduces, the travel time increases, and consequently the average
travel cost borne by the road users increases according to the shape of the private cost
function𝐴𝐶(𝑞), until it intersects the demand curve of usage of the road link𝐷(𝑞) .
Figure 7 – Road congestion depending on network conditions

Unit cost
SMC(q)
D(q) C
DWL
B AC(q)
p2
A
p1 D
F
p0 E
O q0 q2 q1 q [vehicles/hour]
The delay cost approach defines the road congestion cost as the value of the travel time
lost relative to a free-flow situation. In Figure 7, the delay cost coincides with the rectangle
𝑝 0𝐸𝐴𝑝1.
The function 𝑆𝑀𝐶(𝑞) represents the social marginal cost function, which is equal to the
average travel cost borne by the road users 𝐴𝐶(𝑞) plus the cost of the additional travel
time, generated by the marginal vehicle that reduces the speed of all the other vehicles.
The deadweight loss approach enables to determine the economically optimal solution
(i.e., the point 𝐵 where the demand function and the social marginal cost function
intersect). According to this approach, the external cost of congestion is given by the
demand in excess with respect to 𝑞2 and the triangle 𝐴𝐵𝐶 is the so-called ‘deadweight loss’
(i.e., DWL). The road congestion cost defined in this manner is regarded as a proper basis
for transport pricing. At the same time, the delay cost reflects the total congestion cost, in
a way which is (partly) comparabale to the total external cost for the other cost
categories22. For this reason both approaches are applied and presented in this Handbook .
For a more detailed discussion on the theoretical background of road congestion we refer to
Annex F.
The calculations of delay cost and deadweight loss enables us to estimate the total external
road congestion costs of the contexts considered23. The average cost can be obtained
dividing the total cost estimated by the vkm of that context. The marginal cost is given by
the cost due to the additional vehicle that enters the flow and coincides with the segment
𝐴𝐶 in Figure 7.
7.2.2 Approaches to external congestion cost of other transport modes

For scheduled modes only, the probability that a cost arises increases with an increase of
slots. The external cost depends on the capacity and it can be identified according to two
categories, namely congestion costs and scarcity costs.
— A ‘congestion cost’ arises when one scheduled service delays another. Although the
timetables will be designed to prevent this from happening, it could be the case that at
high levels of utilisation, the presence of an additional scheduled service may lead to an
additional delay to others (i.e., ‘reactionary delay’).
— A ‘scarcity cost’ arises where the presence of a scheduled service prevents another
scheduled service from operating, or requires it to take an inferior slot. Therefore,
scarcity costs are incurred whenever a slot is reserved. In essence, scarcity costs denote
the opportunity cost to service providers for the non-availability of desired departure or
arrival times.
It is worth observing that this kind of analysis needs a large amount of information
(e.g., traffic density, mix of scheduled trains, reliability rate and average delay, etc.),
as well as complex elaborations. The necessary estimations are highly context-specific
and therefore very sensitive for traffic situations. This is an aspect that deserves due
consideration, as it significantly limits the opportunities for generalising estimations 24.
________________________________
22 The delay costs include both internal (costs borne by the same transport user who cause them) and external
delay costs. This should be considered when comparing the delay costs with the total/average costs of other
externalities.
23 It is worth reminding that the DWL implies a different approach with respect to the delay cost and the
methodology to estimate the other external costs. This depends on the nature of road congestion, as the cost
of congestion depends on marginal vehicle that enters the traffic and this situation can be represented only by
the DWL approach.
24
As an example, for 2018, preliminary results by EUROCONTROL estimates a total costs for delays in the ECAC
countries at 14.5 billion €. Such costs are caused not only by air traffic control staff shortages and capacity
issues but are also due to strikes, bad weather and technical problems.
7.3 Total and average road congestion costs
The total yearly road congestion costs per country can be calculated for both delay cost and
deadweight loss approaches developing a bottom-up approach and building on the values of
delay cost per vkm and deadweight loss per vkm estimated for representative types of
circumstances on a congested network. These representative circumstances reflect
different road types (i.e., urban roads, urban trunk roads, inter-urban roads and
motorways) and different levels of capacity occupancy (i.e., near capacity, congested and
over capacity). The values obtained have to be combined with additional input values
related to real traffic data as well as transport modelling and statistics on mobility in order
to estimate total road congestion costs.
In the following paragraph the methodology for the calculation of total and average road
congestion costs is explained, including the input and output values.
7.3.1 Methodology
The estimation of total road congestion costs, according to delay cost and deadweight loss
approaches, needs to be distinguished with respect to urban and inter-urban contexts.
The main reason is that the available information regarding the observed delay generated
by traffic is of a different nature. For the urban context, the available information consists
of congestion indexes and amounts of time losses for a sample of cities. For the inter-urban
context, the available information consists of the amount of delay for a large number of
spots localised on the European road network.
Road urban congestion costs

The methodology for the estimation of total cost of urban congestion can build on
aggregate congestion indexes by city. The information needs to be provided with respect to
the level of congestion, the road network length by road type (i.e., trunk urban road, other
urban road), the average delay per day and the total accumulated delay per year.
Building on the information above, the amount of congested network by road type and time
period can be estimated. Then, this information can be used together with the values of
deadweight loss per vkm and delay cost per vkm (differentiated by time period and road
type) to estimate the congestion cost per vehicle. Finally, in order to expand the result to
the whole city on a yearly basis, total yearly costs in each urban context can be estimated
by using the population size, the share of individuals travelling and the car share, assuming
230 work days per year25.
Based on these results, the costs estimated on the sample of cities can be applied country
by country (e.g., to the cities with at least 50,000 inhabitants) 26.
Figure 8 illustrates the methodology for estimating urban congestion costs.
________________________________
25 The set of costs obtained for an available sample of cities at country level can also be used for statistical
analyses aimed at identifying correlations between the congestion cost (e.g., cost per capita) and some known
features of the cities such as population size.
26
A simplified approach could be adopted to generalise urban congestion costs also to cities below the assumed
threshold, taking into account the population in the NUTS3 zone and the typology of NUTS3 according of the
classification urban/mixed/rural.
The estimation of average car congestion cost per vkm (pkm or tkm) at urban level is
performed as a ratio between the total congestion cost and the total vkm (pkm or tkm) on
the whole urban road network (not only the congested roads).
With reference to congestion costs for freight vehicles (i.e., HGV and LCV) at the urban
level, the methodology applied for cars could not be replicated, due to lack of information.
Therefore, in order to provide some information to also cover this aspect, a simplified
approach has been applied, based on the estimation of congestion costs for cars, the
Value of Time by road vehicle category and data on vkm at urban level for LCVs and HGVs
(estimated/collected within this Handbook).
Figure 8 – Methodology for estimating urban congestion costs for delay cost and deadweight loss approaches
Road inter-urban congestion costs

The quantification of traffic experiencing congestion on inter-urban roads can be estimated
building on two main sources. First, identifying congested spots on the inter-urban road
network, namely the spots where road traffic is delayed in the most congested peak hour 27.
Second, by means of parameters derived from the characteristics of the road network.
Assuming ranges of delays for the congested spots and the speed-flow function of the roads
where the spots are located, one can estimate the level of occupancy of each spot at peak
times, i.e. the amount of vehicles experiencing congestion in the most congested peak
hours (i.e., in terms of Passenger Cars Equivalents, or PCEs). On the basis of daily traffic
profiles assumed, the load in each hour can be estimated for different classes of road users
________________________________
27
For each spot, the amount of delay could be expressed in terms of additional time per km.
(e.g., cars, trucks and coaches). The share of demand belonging to each class can be
estimated assuming a traffic segmentation (if available, from observations or modelling
exercises) (see Box in Annex F)28.
Given the capacity of the road links and considering the sum of all vehicle types (in terms of
PCEs), the load/capacity ratios can be estimated for each road link and hour (i.e., near
capacity, congested or over capacity29)30.
After this process, the total number of vehicles incurring congestion on the inter-urban
network in an average day is obtained for the assumed classes of road users. Vkms in
congestion by vehicle type can be estimated by multiplying loads by length of congested
network on a link level. The estimation of the amount of yearly traffic can be made
assuming 230 workdays per year.
The estimation of average congestion cost per vkm (pkm or tkm) by vehicle category at
inter-urban level is performed as a ratio between the total congestion cost and the total
vkm (pkm or tkm) on the whole inter-urban road network (not only the congested roads).
With reference to congestion costs for LCVs at inter-urban level, a simplified approach has
been applied in order to provide some information to also cover this aspect. Figure 9
illustrates the methodology for estimating inter-urban congestion costs.
Figure 9 – Methodology for estimating inter-urban congestion costs for delay cost and deadweight loss
approaches
________________________________
28 See also TRansport eUropean Simulation Tool
29
The following assumptions have been applied. Near capacity: flow to cap acity ratio between 0.8 and 1.0.
Congested: flow to capacity ratio between 1 and 1.2. Over capacity: flow to capacity ratio above 1.2.
30
For example, all vehicles travelling in hours with a load/capacity ratio higher than 0.75 can be considered
experiencing congestion.
7.3.2 Input values
To estimate urban and inter-urban road congestion costs the following common input values
are needed:
— Speed-flow functions.
— Demand curve 𝐷(𝑞) based on literature cost elasticity. In particular, Littman (2011)
and Oum et.al. (1990) suggest different elasticity parameters to construct the demand
curve. They have been used by vehicle category and for peak and off-peak periods.
For cars, values have been estimated as weighted average of values by trip purpose,
considering the composition of trips in different periods.
Value of time (𝑉𝑂𝑇) for car and coach passengers by purpose (i.e., commuting, business
and leisure), for the drivers of the different vehicles and for freight road transport
(commodity). Values are needed by country and for short/long distance trips.
This Handbook uses UK Department for Transport (ARUP, 2015) for passengers and
HEATCO (2006), Comité National Routier (2016), Significance, VU University Amsterdam,
John Bates Services (2012) for freight. For freight transport (i.e. HGV and LCV) and
coaches the VOT of the driver has been considered in the analysis. The values can be
found in Annex F.2.2.
— Average vehicle occupancy/load factors for cars, buses, coaches, LCVs and HGVs are
those estimated within this Handbook by country.
With reference to the estimation of urban road congestion cost, specific inputs needed
are:
— Data on the level of congestion and road network length by road type (i.e., trunk
urban road, other urban road), average delay per day and total accumulated delay per
year (related to peak period journeys). Available from TomTom for a set of European
cities (see TOMTOM Traffic Index).
— Car mode share in a set of European cities. The information has been taken from the
EPOMM Modal Split Tool (TEMS), integrated with local sources where data from this tool
was not available.
— Population of European cities. Available from Eurostat and national statistical offices.
— Typology of NUTS3 according to the degree of urbanisation urban/mixed/rural.

The classification is provided by Eurostat.
With reference to the estimation of inter-urban road congestion cost, specific inputs
needed are:
— Localisation of the congested spots on the European inter-urban road network.

Provided by the Commission’s Joint Research Centre, it identifies spots where road
traffic is delayed in the most congested peak hour because of traffic and, for each spot,
provides the amount of delay (in terms of additional time per km).
— Road network characteristics to determine speed-flow functions of the roads where

the spots are located in order to estimate the level of occupancy of each spot in peak
time, i.e. the amount of vehicles experiencing congestion in the most congested peak
hour (i.e., in terms of PCEs31). This Handbook assumes the parameters and outputs of
the TRUST network model.
— Daily traffic profiles. Road profiles describe how traffic changes over a 24-hour period.
Representative road load profiles have to be estimated for passenger cars and trucks
during the day in different countries.
7.3.3 Output values

This section presents output values for congestion costs. Two types of costs are presented
costs borne and costs generated. Costs borne are the costs imposed on the drivers and
passengers of the vehicles. Costs generated are the costs that the drivers of the vehicles
impose on the other traffic participants. In this chapter we first present the costs borne
(Table 41 and Table 42), and subsequently present the costs generated (Table 43 and Table
44). However, in the synthesis chapter we use the costs generated, as these are the
external costs of a particular vehicle type.
Table 41 summarises the estimations of the total and average congestion costs for road
transport, with respect to delay and deadweight loss approaches. The costs in this table are
the total costs borne by each vehicle type. It should be underlined that not all vehicle
categories have been considered in each context. The congestion costs for coaches has been
estimated only at inter-urban level. The impact of cars on public transport is not estimated
at urban level. Road traffic induces deviations of public transport from scheduled services
(e.g., generating platooning) and this is a context-specific effect, which cannot be
estimated at this scale of analysis. The literature does not provide useful evidence to
generalise from context-specific applications.
The average congestion cost reported in the table per vkm (pkm or tkm) are estimated on
the basis of the traffic on the whole network (not only the congested roads).
________________________________
31
PCE coefficients are taken from the TRUST model parameters, i.e. equal to 1 for cars and 2 for HGVs.
For coaches it is assumed the same value as HGVs.
Table 41 - Total and average congestion costs borne by road vehicle categories in the EU28
Vehicle category Delay costs Deadweight loss costs
Passenger transport Total EU28 €-cent/ €-cent/vkm Total EU28 €-cent/ €-cent/vkm
[Billion €] pkm [Billion €] pkm
Passenger car 206.2 4.37 7.03 35.6 0.75 1.21
Passenger car - urban 172.6 11.82 19.03 30.0 2.06 3.31
Passenger car - inter-urban 33.6 1.03 1.66 5.5 0.17 0.27
Coach inter-urban * 2.1 0.74 14.49 0.2 0.08 1.50
Total passenger 208.3 35.8
Freight transport Total EU28 €-cent/ €-cent/vkm Total EU28 €-cent/ €-cent/vkm
[Billion €] tkm [Billion €] tkm
Light commercial vehicle 38.5 11.63 8.05 6.6 2.01 1.39
LCV – urban** 32.6 27.75 19.21 5.6 4.78 3.31
LCV - inter-urban** 5.9 2.78 1.92 1.0 0.48 0.33
Heavy Goods Vehicle (HGV) 23.8 1.30 17.72 3.8 0.21 2.81
HGV – urban** 17.6 3.81 51.94 3.1 0.67 9.11
HGV - inter-urban 6.2 0.45 6.20 0.7 0.05 0.69
Total freight 62.3 10.4
Total road transport 270.6 46.2
* Only inter-urban congestion considered for coaches.
** Simplified approach based on estimation for cars.
Table 42 summarises the estimated total congestion costs related to inter-urban traffic on
the motorways network only that is borne by the vehicles.
Table 42 - Total and average congestion cost of inter-urban traffic borne by vehicle categories on motorways
network

Passenger transport Total EU28 €-cent/pkm €-cent/vkm Total EU28 €-cent/pkm €-cent/vkm
[Billion €] [Billion €]
Passenger car - inter- 2.91 0.27 0.43 0.67 0.06 0.10
urban
Coach - inter-urban 0.20 0.15 2.87 0.03 0.02 0.49
Freight transport Total EU28 €-cent/tkm €-cent/vkm Total EU28 €-cent/tkm €-cent/vkm
LCV – inter-urban** 0.50 0.69 0.48 0.12 0.17 0.12
HGV – inter-urban 1.09 0.19 2.55 0.19 0.03 0.46
** Simplified approach based on estimation for cars.
In order to provide an estimation of the external road congestion costs based on the vehicle
categories generating congestion, a further step has been made. The total congestion costs
of passenger car, by country and by context, have been distributed amongst the other road
vehicle categories a measure representative of the share each vehicle type has in the cause
of congestion. In principle, the measure should be related to the traffic composition in a
congested situation, but this information is not readily available and strictly depends on
specific context and time of the day. Therefore, a simplified approach has been developed
by assuming that this measure could be obtained multiplying the total vkms of a given
context by the Passenger Car Equivalent coefficient (PCEs) of each vehicle type 32.
The average generated congestion costs per vkm (pkm or tkm) are then estimated as ratio
between the total costs and the traffic on the related total network (not only the congested
roads).
The following table summarises the estimated total and average congestion costs generated
by road vehicle categories in the EU28, according to the simplified approach used.
Table 43 - Total and average congestion costs generated by road vehicle categories in the EU28 according to
the simplified approach used
________________________________
32
The following assumptions have been made for PCE coefficients (taken where available from the TRUST model
parameters): equal to 1 for cars, 2 for HGVs and bus/coaches, 1.5 for LCVs.
pkm pkm
Passenger car 196.1 4.2 6.7 33.5 0.7 1.1
Urban 160.8 11.0 17.7 28.0 1.9 3.1
Inter-urban 35.3 1.1 1.7 5.5 0.2 0.3
Bus/ Coach 4.5 0.8 15.9 0.8 0.1 2.7
Urban 3.9 1.8 35.5 0.7 0.3 6.1
Inter-urban 0.5 0.2 3.1 0.1 0.0 0.5
Light commercial vehicle 55.5 16.8 11.6 9.4 2.8 2.0
Urban 46.5 39.6 27.4 8.0 6.8 4.7
Inter-urban 9.0 4.2 2.9 1.4 0.7 0.5
Heavy Goods Vehicle (HGV) 14.6 0.8 10.9 2.5 0.1 1.8
Urban 11.6 2.5 34.1 2.0 0.4 6.0
Inter-urban 3.0 0.2 3.0 0.5 0.0 0.5
Table 44 - Total and average congestion cost generated of inter-urban traffic by vehicle categories on
motorways network, according to the simplified approach used
pkm pkm
Passenger car – inter-urban 3.2 0.28 0.45 0.7 0.06 0.10
Coach – inter-urban 0.1 0.05 0.89 0.02 0.01 0.20
LCV – inter-urban 0.9 1.2 0.83 0.2 0.3 0.19
HGV – inter-urban 0.4 0.06 0.88 0.1 0.01 0.20
7.4 Social marginal road congestion costs
7.4.1 Methodology
The social marginal road costs can be calculated only when the deadweight loss approach is
applied, because it needs the estimation of the social marginal cost curve. In this respect,
it is worth noting that this is an additional input needed with respect to the delay cost
approach, which estimates the road congestion costs only relying on information regarding
the private cost curve.
Road congestion (and therefore its cost) is highly dependent on the context. Of course, it
is unrealistic to assume that all specific circumstances where congestion occurs can be
identified and measured. Therefore, the estimation needs to be developed using
representative types of circumstances. These representative circumstances can be reflected
in different road types (i.e., urban roads, urban trunk roads, inter-urban roads and
motorways) and in different levels of traffic intensity (i.e., near capacity, congested and
over capacity).
The point where the demand curve and private cost curves cross has to be calculated
(i.e. point 𝐴), as well as the point where the demand curve and the social cost curves cross
(i.e. point 𝐶). Then the difference between social and private cost curves (i.e. segment 𝐴𝐶
in Figure 7) can be calculated to obtain the estimation of the social marginal cost33.
________________________________
33 This measure is the social marginal cost at the current level of congestion and can be used to estimate the
optimal congestion price (i.e., the segment BD in Figure 7). It is also worth observing that the marginal
congestion cost can be calculated for a traffic situation, beyond the economically optimal solution, with any
congestion charge. In this respect, there are many measures governments can take in the direction to reduce a
part of the external cost of road congestion. For example a grant that is introduced to subsidise rail services to
shift part of road users to public transport. In this case, the benefit of this reduction in road traffic is the
difference between the marginal social cost and the price at the current level of traffic.
The estimation can be made differentiating peak and off-peak periods to take into account
the fact that trip purposes are not the same during different periods of the day and
therefore that values of time and elasticities of demand are also variable. As a final step,
the average daily value can be estimated based on an assumption on the amount of trips
during peak/off-peak periods.
With regards to this Handbook, for urban roads (i.e., trunk urban road and other urban
road) the estimation has been made for passenger cars only, while for inter-urban roads
(i.e., motorway, other non-urban roads), the vehicle categories considered are cars, trucks
and coaches. The differentiation between vehicle categories mainly consists in different
values of demand elasticities and different value of time per vehicle. More details on the
methodology to estimate road social marginal congestion costs are provided in Annex F.
It is important to notice that also for the marginal road congestion costs, a differentiation
between vehicle types can be made, according to the perspective of the vehicle generating
the cost. However, the estimation of the generated marginal road congestion cost by
different vehicle types has a limitation, which is related to the calculation of the cost
curves.
In general, cost curves depend much on the value of time and the marginal external
congestion cost is the difference between private cost function and marginal social cost
function (see functions 𝐴𝐶 and 𝑆𝑀𝐶 in Figure 7). If the existence of various vehicle types is
considered relevant, the calculation of the marginal social cost should consider that the
overall transport activity is made of a mix of various vehicle types. The value of time per
vehicle is highly variable across vehicle types because of different occupancy (e.g. a car vs.
a coach) and because the value of time of passengers and goods is different. Therefore, the
curve representing the marginal social costs can only be a sort of average, which will be
hardly representative of any real user.
Additionally, another relevant result of the estimation of the marginal road congestion costs
is the deadweight loss (area 𝐷𝑊𝐿 in Figure 7). Of course, for sake of consistency, the
estimation of the deadweight loss should be based on the same cost curves used to estimate
the marginal social cost. Then, also a demand curve is needed. If the curve of the marginal
social cost is a sort of average, also the demand curve should be a sort of average.
However, again, different vehicle types have different demand curves because they have
different elasticities. So, again, the representativeness of an average demand curve would
be limited.
Considering of such complexities, the estimation of marginal generated road congestion

costs in abstract terms is an exercise necessarily full of simplifying assumptions. Considering
the traffic as a whole, and so using average curves, allows to avoid the introduction of a
further element of approximation.
And because of this limitation, the approach developed basically uses a different
perspective. It considers who incurs in congestion rather than who generates congestion.
This approach allows to take into account differences in values of time (and elasticity of
demand). These differences explain why costs are different across vehicle types, even if
there is no any reference to the effect of a given type to social costs. Furthermore, this
approach does not need to compute ‘average’ cost demand curves representing all different
vehicle types, so one level of approximation is avoided.
In the light of this different approach, marginal road congestion costs incurred by a vehicle
type cannot be added to other external costs by vehicle type (if not for cars, as for this
vehicle type, generated costs are equal to incurred costs).
The social marginal congestion costs presented in Section 7.4.3 have been calculated
assuming the perspective of a vehicle incurring in a situation of road congestion. The
calculation of the marginal social costs generated has been developed according to the
sympilified methodology presented in Section 7.4.4.
7.4.2 Input values

For each road type and capacity occupancy level, the only additional input value needed to
estimate the social marginal cost (with respect to the input values already listed in Section
1.3) is the social cost curve 𝑆𝑀𝐶(𝑞) . It must be calculated as the first derivative of the
private cost curve 𝐴𝐶(𝑞).
7.4.3 Output values

Values in Table 45 and Table 46 can be used to estimate social marginal congestion costs in
specific circumstances if a more detailed analysis is not feasible. The most representative
value(s) should be picked up depending on the characteristics of the case under analysis.
Traffic situations are identified based on the volume to capacity ratio of a traffic link: it is
assumed that ‘near capacity’ is related to v/c ratios between 0.8 and 1, ‘congested’ refers
to v/c ratios between 1 and 1.2, while ‘over capacity’ is considered when v/c ratio is above
1.2.
Table 45 – Social marginal congestion costs34 of road transport per pkm and tkm
Vehicle category Traffic situation Urban area Inter-urban area
Trunk roads Other urban Motorway Other roads

roads
Passenger transport (€-cent/pkm)
Passenger car Over capacity 19.9 41.2 18.2 28.8
Congested 15.4 36.1 14.1 24.6
Near capacity 10.8 29.3 9.9 19.4
Well below capacity 0.0 0.0 0.0 0.0
Coach Over capacity n. a. n. a. 16.3 21.2
Congested n. a. n. a. 12.5 18.1
Near capacity n. a. n. a. 8.8 14.3
Well below capacity n. a. n. a. 0.0 0.0
________________________________
34
The Social Marginal congestion cost estimated in this version of the Handbook coincides with the segment AC in
Figure 7. This definition has been assumed according to: The Marginal Cost of Traffic Congestion and Road
Pricing: Evidence from a Natural Experiment in Beijing (Li, et al., 2016). The Efficient Marginal Congestion
Cost (i.e., EMCC) of the 2014 version of the Handbook (see Annex A.1) coincides with the segment BD.

roads
Freight transport (€-cent/tkm)
HGV Over capacity n. a. n. a. 9.0 11.7
Near capacity: flow to capacity ratio between 0.8 and 1.0.
Congested: flow to capacity ratio between 1.0 and 1.2.
Over capacity: flow to capacity ratio above 1.2.
Table 46 – Social marginal congestion costs of road transport per vkm

roads
Passenger transport (€-cent/ vkm)
Congested 24.8 58.2 22.6 39.6
Near capacity 17.4 47.2 15.9 31.2
Coach Over capacity n. a. n. a. 318.5 415.8
Congested n. a. n. a. 245.8 355.0
Freight transport (€-cent/ vkm)
HGV Over capacity n. a. n. a. 122.0 159.3
7.4.4 Social marginal road congestion cost generated using the simplified
approach
In order to provide values of the marginal generated road congestion costs by vehicle type
that can be compared to other marginal external costs computed in this version of the
Handbook, this section presents an estimation using the same approach adopted in (Maibach
et al. (2000) and in the 2008 version of the Handbook, i.e. taking the marginal cost of cars
and estimating those of other vehicle types by multiplying this cost by the Passenger Car
Equivalent coefficient (PCEs) of each vehicle type.
The following assumptions have been made for PCE coefficients (taken where available
from the TRUST model parameters): equal to 1 for cars, 2 for HGVs and bus/coaches, 1.5
for LCVs. Values in Table 47 and Table 48 can be used to estimate social marginal
congestion costs generated by different type of vehicles in specific circumstances if a more
detailed analysis is not feasible.
Table 47 – Social marginal congestion costs of road transport generated per pkm and tkm using the simplified
approach

roads
Passenger transport (€ cent/ pkm)
Congested 15.4 36.1 14.1 24.6
Near capacity 10.8 29.3 9.9 19.4
Bus/Coach Over capacity 3.3 6.8 3.0 4.7
Congested 2.5 5.9 2.3 4.0
Near capacity 1.8 4.8 1.6 3.2
Freight transport (€ cent/ tkm)
LCV Over capacity 69.5 143.6 63.6 100.5
Congested 53.6 126.1 49.1 85.8
Near capacity 37.8 102.2 34.5 67.5
HGV Over capacity 4.7 9.7 4.3 6.8
Congested 3.6 8.5 3.3 5.8
Table 48 – Social marginal congestion costs of road transport generated per vkm using the simplified approach

roads
Passenger transport (€ cent/ vkm)
Congested 24.8 58.2 22.6 39.6
Near capacity 17.4 47.2 15.9 31.2
Well below capacity 0 0 0 0
Bus/Coach Over capacity 64.2 132.6 58.7 92.7
Congested 49.5 116.4 45.3 79.2
Near capacity 34.9 94.4 31.9 62.3
Freight transport (€ cent/ vkm)
LCV Over capacity 48.1 99.4 44.0 69.6
Congested 37.1 87.3 34.0 59.4
Near capacity 26.1 70.8 23.9 46.7
HGV Over capacity 64.2 132.6 58.7 92.7
Congested 49.5 116.4 45.3 79.2
Near capacity 34.9 94.4 31.9 62.3

Due to the fact that road congestion costs are so highly dependent on the methodological
approach and specific for local conditions, there is a huge variation in total, average and
marginal congestion costs estimations. This large variation conceptually complicates what is
meant by road congestion costs, as there is not a unique approach.
For this reason, starting from the previous version of the Handbook, we have developed a
methodological approach that enlarges the scope of outputs available to the final user, thus
enhancing the potential usability of the outputs for further context-specific applications.
In this respect, total and average road congestion costs, have been estimated for both delay
and deadweight loss, in order to present the orders of magnitude at country and EU28 level.
The social marginal cost is the measure that can be used to estimate the optimal congestion
price.
For the above reasons, we consider the deadweight loss approach more useful for its
relevance for policy making and from the perspective of marginal cost pricing. Total and
average generated congestion costs have been estimated developing a simplified approach
to make them comparable with the other external costs estimated in this Hankbook.
7.6 Congestion costs of other transport modes
The review of existing literature for other transport modes did not reveal many new sources
(as compared to the 2014 Handbook) of congestion or scarcity costs estimates, for rail, air
or water transport that could be recommended as a best practice methodology (see also
Annex F.8).
As illustrated in Section 7.2.2, scarcity costs concern path allocation issues and should be
only considered in service planning. They correspond to the opportunity-cost of choosing
one of the incompatible services. Since the estimation of scarcity costs is very complex, it
has been suggested (see for example Quinet, 2003) to use auctions to reveal the values on
which the regulator has an information asymmetry compared to other agents. Once the
time table of the services has been designed, congestion costs may happen and they should
be taken into account in infrastructure pricing.
For rail some methodologies exist at country level to price rail networks capacity according
to type of service, time and path. However, there is no straightforward evidence that the
actual charges reflect the scarcity of slots.
The congestion costs of a rail network can be estimated starting from the information on
the actual reactionary delays of trains, multiplied by the number of affected passengers and
by a suitable average value of time. In the rail industry, there are not many costs borne
by the infrastructure manager and operators because of the congestion, excluding the
compensation paid to the users for the delays suffered. However, to include these
compensations would represent double counting.
Marginal cost estimates for rail freight congestion are available by Christidis and Brons
(2016)35. The average value for EU is equal to €-cent 43.20 per 1.000 tkm (2016 prices).
Table 119 in Annex F.8 shows the values for EU Member States. For US, Austin (2015)
estimates rail freight congestion costs to be equal to €-cent 0.0-14.0 per 1.000 tkm (2016
prices).
The urban public transport congestion costs received little consideration. Prud’homme
et al. (2012) argue that, in principle, users shifting from cars to bus, tram or subway,
increasing the crowding on urban public transport vehicles, causing a comfort loss to all
other passengers already travelling. This comfort loss due to increased crowding from modal
shift, which can be considered another form of congestion cost, is an externality leading to
a suboptimal usage of the tram, bus or subway36. That could be corrected by means of a
congestion tax (e.g., subway toll). Prud’homme et al. (2012) elaborated a congestion cost
curve for crowding of public transport for the Paris subway on data of 2009. The estimated
willingness to pay for a non-congested travel emerged to be equal to € 1.43 per trip.
It is also worth noting that, according to de Palma et al. (2017), urban public transport
congestion may be as important as road congestion37. For example, Borjesson et al. (2017)
found that in Stockholm the marginal cost of an extra passenger bus can be higher than the
marginal external cost of a car during the peak period. Similar results were found for Paris
in Kilani et al. (2014).
________________________________
35 Freight marginal average external costs of transport per country_v3.xls.
36 This consideration assumes a short run period approach in which the number of vehicles can be varied, but the
infrastructure cannot. If vehicles are added, for example in the peak period to offset the comfort loss, the cost
for expanding the vehicles is not an external cost, but an internal cost of the operator.
37
For example, in large metropolitan areas (e.g., Paris and London), there are more trips by public than by
private transport.
For air transport, congestion can be associated with a lack of sufficient capacity to
accommodate the required demand of aircraft movements for landings and take-offs.
Like for rail transport, in airports the airlines do not enter the system randomly, with
aircraft movements scheduled in advance (i.e., in time slots). However, any perturbation
introduced by exogenous factors causes cascade-effects congestion and accumulation of
delays during the next periods.
Congestion costs of airports can be estimated starting from the information on actual delays
of flights, observed over a certain time period. That allows us to calculate the total time
lost by passengers and airlines to be translated into monetary terms.
For air passengers, once that delay time per flight is available, it is necessary to estimate
the number of passengers affected. Ideally, one should have the exact number of
passengers on each flight, so that the number of minutes of delays can be translated into
total passengers’ time38. If this is not the case, it is possible to approximate the number of
passengers knowing at least the capacity of the aircraft serving a route. If the type of
aircraft is known, one can rely on the average number of passengers that each model of
aircraft may accommodate. Otherwise, standard values of 60-80% can be taken. The value
of time for passengers is generally derived from the literature (e.g., (Wardman, et al.,
2012) and if possible this should be differentiated with respect to ticket classes (i.e., first,
business and economy).
For airlines congestion involves extra costs because of the degradation of the airline
product. First, direct costs that refer to additional fuel consumption, time at gate, taxiing
time to enter the runway, en-route flight (e.g., circling) and maintenance. Second, but
much more difficult to estimate are the indirect costs borne in terms of lost revenues,
compensation to passengers and opportunity costs. Clearly, if passengers’ cost of time lost
has already been considered, the airlines’ compensation payments would represent double
counting 39.
As far as inland waterway and maritime transport are concerned, no illustrative

quantification of marginal congestion costs could be identified (see Annex Y for findings
from 2008 Handbook, as reported in 2014 version). Regarding scarcity costs for inland
navigation, GRACE estimates values between €-cent 38.0 to €-cent 50.0/TEU-km at Kaub
and €-cent 65.0 to €1.25/TEU-km at Duisburg. Christidis and Brons (2016) indicate that
congestion costs of freight transport for both inland waterways and short sea shipping can
be assumed to be negligible for EU Member States.
________________________________
38 Eurocontrol gathers data that allow delays costs for passengers to be calculated (Eurocontrol, 2017).
Eurocontrol publishes data in CODA (i.e., all-causes delay and cancellations to air transport) reports, where
delay causes are grouped with respect to primary causes (i.e., airline operations, airport operations, en -route,
governmental, weather and miscellaneous.) and reactionary delay. Data refers both to single airports (i.e.,
ranking the most affected) and pairs of connected airports (i.e., ranking the most delayed links).
39
Passengers cost of delay is a dominating cost for airlines, although generally poorly quantified or supported by
incomplete evidence (Cook & Tanner, 2015).
8 Costs of well-to-tank emissions
8.1 Introduction
All cost categories discussed in the chapters before cover direct effects of the transport
operation process. However, there is a broad range of other up- and downstream processes
directly related to transport that also lead to negative external effects. Taking a life-cycle
oriented view on transport, the energy production (well-to-tank), the vehicle and
infrastructure production, maintenance and disposal all lead to the emission of air
pollutants, greenhouse gases, toxic substances and other negative environmental impacts.
The by far most relevant effects are the emissions due to energy production, often also
called well-to-tank emissions. These costs are directly linked to the transport activity and
can be calculated on a profound basis. Therefore, the present chapter focuses on the costs
of well to tank emissions. Any other indirect costs of other up- and downstream processes
are covered in Chapter 10 on ‘other external costs’.
The cost of well-to-tank emissions (= costs of energy production) includes the production of
all different type of energy sources which leads to emissions and other externalities.
The extraction of energy sources, the processing (e.g. refining or electricity production),
the transport and transmission, the building of energy plants and other infrastructures: all
these processes lead to emission of air pollutants, greenhouse gases and other substances.
The emissions during the production of energy sources are very relevant in terms of total
external costs. Mainly for electricity driven transport modes, the effects of energy
production are very relevant since the energy use is virtually emission-free.40

This chapter focuses the cost related to energy production, i.e. the well-to-tank emissions.
The energy production, e.g. electricity production of fossil fuel production, lead to a broad
range of negative environmental impacts: emission on air pollutants, greenhouse gases,
emission of toxic substances, land use and environmental risks. The present study covers
the emission of greenhouse gases (CO 2, CH4, and N2O) and air pollutants (PM2.5, PM10, NOx,
SO2, NMVOC) related to all energy production (‘precombustion’) processes: extraction,
processing, transport and transmission. For these effects, there is a well-developed basis
for monetization.
Different studies on external costs of transport cover different external cost categories.
The most recent studies generally cover the air pollution costs and the climate change
costs. Negative impacts on ecosystems, land use or nuclear power risks are not covered for
methodological reasons (high uncertainty and monetization not (yet) feasible). Chapter 10
on ‘other external costs’ further elaborates on some of those topics.
________________________________
40
For electric cars, WTT costs are presented in the marginal cost section for selected cases. However, no total
and average costs are calculated due to lack of transport performance data.
8.3 Total and average costs of well-to-tank emissions
8.3.1 Methodology
The methodology for calculating the cost of well-to-tank emissions is the same as for the
cost of air pollution (see Chapter 4) and the climate change costs (see Chapter 5).
The valuation is done with the same cost factors. However, it has to be taken into account
that it is very often not known at what place (meaning: at which population density) the
emissions occur. Therefore, the valuation has to be made with an average cost factor or a
cost factor for ‘unknown sources’.
Figure 10 – Methodology total and average costs of well-to-tank emissions
Emissions Transport Cost factors

factors performance - air pollution costs (€/t)
- WTT emissions & energy us - climate change costs
Input values
in t/veh-km, t/ - veh-km (€/t)

kWh or t/l fuel - energy use
Total WTT Emissions

- emissions in t (per pollutant)
Calculations
Total external costs of WTT emissions per transport mode

per country
Allocation
Average costs per pkm and tkm per mode
8.3.2 Input values

There are two types of input values for calculating the cost of well-to-tank emissions:
The emission factors for the emissions of energy related upstream emissions (i.e. well-to-
tank emissions) and the damage cost factors for monetizing the emissions.
The emission factors include the emission of greenhouse gases and air pollutants emitted
during the process of energy production. Generally, the average energy consumption
(electricity consumption, diesel or petrol consumption) is the central basis for deriving
specific emission factors for the well-to-tank emissions.
The main data source for emission factors of well-to-tank emissions are the emission
database TREMOD (IFEU, 2017) as well as the European research data from the JEC well-to-
wheel analysis (2014). For electric driven vehicles, the emission factors of the country
specific electricity mix have been used.
The cost factors for monetizing the emissions are directly taken from the two corresponding
cost categories: the air pollution costs (see Chapter 4) and the climate change costs (see
Chapter 5). For climate change costs, the same shadow price per tonne of CO2 is applied.
For the air pollution costs, the shadow prices for the emission of air pollutants from
unknown sources (EU average) are applied.
Table 49 – Well-to-tank air pollution costs: damage cost estimates in €/kg emission (emissions in the year 2016, EU28 values)
€2016 /kg NOX NMVOC SO2 PM2.5 (exhaust) PM10 (non- exhaust)
Austria 21.9 2.3 16.2 26.8 30.9
Belgium 16.2 3.6 17.1 34.6 47.2
Bulgaria 5.7 0 4.2 7.1 5.4
Croatia 12.2 0.9 8 16.3 8.2
Cyprus 5.4 -0.4 7.8 10.9 20.1
Czech Republic 14.5 1.1 11.6 22.6 39.6
Denmark 10.3 1.5 9.6 14 15
Estonia 3.2 0.3 5.2 5.9 4.9
Finland 4 0.4 4.6 4.8 11.9
France 17.3 1.5 13.9 25.1 5.9
Germany 20.2 1.8 16.5 37.7 24.7
Greece 3 0.3 5.9 7.8 24.8
Hungary 15.3 0.8 9.9 20.4 8.5
Ireland 9.8 1.7 11.8 13.7 12.2
Italy 14.1 1.1 12.7 21.1 19
Latvia 4.8 0.4 4.8 5.7 17.2
Lithuania 7.5 0.6 6.4 7.8 27
Luxembourg 35.5 6.2 29.3 63.9 8
Malta 1.7 0.4 4.3 6.2 63.9
Netherlands 14.4 2.8 20.2 37.4 5.6
Poland 8 0.7 8.2 16.3 5.2
Portugal 1.4 0.5 4.1 5.2 47.3
Romania 9.3 0.5 7.3 12.5 16.1
Slovakia 13.8 0.7 10.1 18.4 12.3
Slovenia 13.1 1.2 9.2 16.1 12
Spain 4.9 0.7 6.8 9.9 10.2
Sweden 6.9 0.7 5.5 6.3 15.2
United Kingdom 7.2 1.4 10 18.3 16.2
EU28 10.9 1.2 10.9 19.4 22.3
8.3.3 Output values

Table 50 shows the output values for the well to tank emissions.
Table 50 – Total and average costs of well-to-tank emissions for land-based modes for the EU28

Passenger transport Billion € €-cent/pkm €-cent/vkm
Bus 0.30 0.17 3.12
Coach 0.53 0.15 2.85
Passenger train electric 2.70* 0.80 106.5
Freight transport Billion € €-cent/tkm €-cent/vkm
LCV 3.79 1.15 0.79
LCV – petrol 0.22 1.18 0.81
LCV – diesel 3.57 1.14 0.79
HGV 3.71 0.20 2.50
Freight train electric 0.50 0.16 86.5
31.23
waterway
trains).
The costs due to greenhouse gas emissions from well-to-tank contribute to about 60–65% of
the well-to-tank costs. For road transport, for example, the share of climate change costs is
62%, the share of air pollution costs 38%.
Table 51 – Total and average well-to-tank costs for aviation for 33 selected EU airports
Short haul (< 1,500 km) 0.95 1.06 583
Medium haul (1,500–5,000 km) 2.10 0.70 1,276
Long haul (> 5,000 km) 5.83 0.91 7,151
Total 8.88 0.86 2,171
* Costs per pax are including the complete flight (not only the half -way principle). The WTT costs for aviation
have been calculated based on specific emission factors for WTT emissions (greenhouse gases and air
pollutants) from TREMOD (IFEU 2017).
Table 52 presents rough estimates for the average and total external well-to-tank costs of
maritime transport. These data are only available for freight. The average cost have been
based on the cost for reference cases presented in Section 8.4 and data on the number of
port calls for the selected ports from Eurostat. The total well-to-tank cost has been based
on the average cost and the number of tkms provided by DG MOVE41. The available data
does not allow an estimation of costs at the individual port level.
Table 52 – Rough estimates for total and average external well-to-tank costs for maritime transport for 34
selected EU ports

Freight ship 4 0.06
8.4 Marginal costs of well-to-tank emissions for selected cases

Like for climate change costs, the marginal costs of well-to-tank emissions, are virtually the
same as the average costs.
to be emphasised that, like for climate costs, the Euro standard is not a cost driver. There
are some differences between the results for the different Euro standards though, that are
the result of the COPERT emission data. These differences are related to the improved
energy efficiency over time and impacts of emission reduction technology on fuel
efficiency.
The size classes for trucks from COPERT do not match with those for the Eurostat transport
performance data used for this Handbook. The load factors for trucks have therefore been
based on interpolation of the Eurostat data.
Annex J contains the marginal WTT cost data for road vehicles for reference cases that are
defined in terms of the combination of fuel type and fuel efficiency of the vehicle (which
are, just like for climate cost, the main cost drivers for WTT cost).
The following table on marginal costs of well-to-tank emissions for road transport shows the
costs per pkm or tkm (except for LCV, where costs per vkm are presented due to the fact
that LCV have characteristics of freight and passenger transport). The costs per vkm for the
different vehicle categories of road transport are available in the background Excel file.
Table 53 – Average/marginal costs of well-to-tank emissions road transport for selected cases
category class
Passenger Petrol Mini Euro 4 0.33 0.32 0.27

Cars
< 0.8 l Euro 5 0.33 0.32 0.27
Euro 6 0.33 0.32 0.27
________________________________
41
See footnote 16.
category class
Small Euro 0 0.47 0.58 0.40
0.8–1.4 l Euro 1 0.35 0.40 0.31
Euro 2 0.34 0.39 0.29
Euro 3 0.34 0.39 0.31
Euro 4 0.36 0.41 0.32
Euro 5 0.36 0.41 0.32
Euro 6 0.36 0.41 0.32
Medium Euro 0 0.58 0.68 0.47
1.4–2.0 l Euro 1 0.40 0.49 0.36
Euro 2 0.37 0.47 0.35
Euro 3 0.41 0.47 0.36
Euro 4 0.42 0.49 0.38
Euro 5 0.42 0.49 0.38
Euro 6 0.42 0.49 0.38
Large-SUV- Euro 0 0.67 0.83 0.57

Executive
Euro 1 0.51 0.63 0.46
> 2.0 l
Euro 2 0.51 0.64 0.48
Euro 3 0.43 0.58 0.42
Euro 4 0.49 0.71 0.49
Euro 5 0.50 0.71 0.49
Euro 6 0.50 0.71 0.49
Diesel Mini Euro 4 0.18 0.16 0.15
< 0.8 l Euro 5 0.18 0.16 0.15
Euro 6 0.18 0.16 0.15
Small Euro 0 0.25 0.26 0.20
0.8–1.4 l Euro 1 0.25 0.26 0.20
Euro 2 0.25 0.27 0.21
Euro 3 0.23 0.26 0.21
Euro 4 0.23 0.26 0.21
Euro 5 0.23 0.26 0.21
Euro 6 0.23 0.26 0.21
Medium Euro 0 0.29 0.30 0.24
category class
1.4–2.0 l Euro 1 0.29 0.30 0.24
Euro 2 0.29 0.31 0.24
Euro 3 0.28 0.30 0.24
Euro 4 0.28 0.30 0.24
Euro 5 0.28 0.30 0.24
Euro 6 0.28 0.30 0.24
Large-SUV- Euro 0 0.32 0.35 0.28

Executive
Euro 1 0.32 0.35 0.28
> 2.0 l
Euro 2 0.33 0.35 0.28
Euro 3 0.32 0.35 0.27
Euro 4 0.32 0.35 0.27
Euro 5 0.32 0.35 0.27
Euro 6 0.32 0.35 0.27
Petrol Mini Euro 4 0.20 0.15 0.15

Hybrid
Small Euro 4 0.20 0.15 0.15
(PHEV)
Large-SUV- Euro 4
Executive 0.20 0.15 0.15
LPG Small Euro 1 n.a. n.a. n.a.

Bifuel
Euro 4 0.20 0.22 0.17
Euro 5 0.19 0.21 0.17
Euro 6 0.18 0.20 0.16
CNG Small Euro 4 0.17 0.19 0.15

Bifuel
Euro 5 0.17 0.18 0.14
Euro 6 0.16 0.17 0.14
Electric n.a. n.a.
(BEV) 0.83 0.83 0.83
Moped Petrol 2-stroke Euro 0 0.29 0.29 0.29

< 50 cm³
Euro 1 0.22 0.22 0.22
Euro 2 0.22 0.22 0.22
Euro 3 0.22 0.22 0.22
category class
4-stroke Euro 0 0.29 0.29 0.29

< 50 cm³
Euro 1 0.22 0.22 0.22
Euro 2 0.22 0.22 0.22
Euro 3 0.22 0.22 0.22
Motorcycle Petrol 2-stroke Euro 0 0.44 0.34 0.32

> 50 cm³
Euro 1 0.41 0.31 0.29
Euro 2 0.40 0.30 0.28
Euro 3 0.40 0.30 0.28
4-stroke Euro 0 0.47 0.33 0.35

< 250 cm³
Euro 1 0.39 0.27 0.29
Euro 2 0.29 0.24 0.23
Euro 3 0.29 0.23 0.23
4-stroke Euro 0 0.51 0.47 0.40

250-750
cm³ Euro 1 0.50 0.42 0.38
Euro 2 0.47 0.38 0.35
Euro 3 0.46 0.38 0.34
4-stroke Euro 0 0.59 0.54 0.46

> 750 cm³
Euro 1 0.51 0.52 0.42
Euro 2 0.53 0.50 0.41
Euro 3 0.53 0.50 0.41
Electric n.a. n.a. 0.16 0.16 0.16
Urban buses Diesel Midi <=15 t Euro 0 0.11 0.18 0.11
Euro I 0.09 0.14 0.09
Euro II 0.09 0.14 0.09
Euro III 0.09 0.15 0.09
Euro IV 0.09 0.13 0.09
Euro V 0.09 0.13 0.08
Euro VI 0.09 0.13 0.09
Diesel Standard Euro 0 0.10 0.19 0.11

15-18 t
Euro I 0.09 0.15 0.09
Euro II 0.09 0.15 0.09
Euro III 0.09 0.16 0.10
category class
Euro IV 0.09 0.14 0.10
Euro V 0.08 0.14 0.09
Euro VI 0.09 0.14 0.09

> 18 t
Euro I 0.09 0.16 0.10
Euro II 0.09 0.16 0.10
Euro III 0.09 0.16 0.10
Euro IV 0.09 0.15 0.10
Euro V 0.09 0.15 0.09
Euro VI 0.09 0.15 0.10
CNG CNG buses Euro I 0.21 0.21 0.21
Euro II 0.19 0.19 0.19
Euro III 0.16 0.16 0.16
EEV 0.08 0.14 0.09
Biodiesel Biodiesel Euro 0 0.03 0.05 0.03

buses
Euro I 0.02 0.04 0.02
Euro II 0.02 0.04 0.02
Euro III 0.02 0.04 0.03
Euro IV 0.02 0.04 0.03
Euro V 0.02 0.04 0.02
Euro VI 0.02 0.04 0.02
Electric Small n.a. 0.54 0.54 0.54
Medium n.a. 0.63 0.63 0.63
Large n.a. 0.64 0.64 0.64
Coaches Diesel Standard Euro 0 0.09 0.21 0.11

<=18 t
Euro I 0.09 0.19 0.10
Euro II 0.09 0.19 0.10
Euro III 0.09 0.20 0.11
Euro IV 0.09 0.19 0.11
Euro V 0.09 0.19 0.10
Euro VI 0.09 0.19 0.11
category class
> 18 t Euro I 0.07 0.15 0.08
Euro II 0.07 0.14 0.08
Euro III 0.07 0.15 0.08
Euro IV 0.07 0.14 0.08
Euro V 0.07 0.14 0.08
Euro VI 0.07 0.15 0.08
Light Petrol Euro 0 0.90 1.33 0.81

Commercial
Vehicles Euro 1 0.90 1.33 0.81
Euro 2 0.89 1.31 0.80
Euro 3 0.88 1.28 0.78
Euro 4 0.88 1.27 0.78
Euro 5 0.55 0.71 0.51
Euro 6 0.55 0.71 0.51
Diesel Euro 0 0.73 0.66 0.49
Euro 1 0.73 0.66 0.49
Euro 2 0.73 0.66 0.49
Euro 3 0.73 0.67 0.49
Euro 4 0.73 0.67 0.49
Euro 5 0.60 0.62 0.52
Euro 6 0.60 0.62 0.52
Electric n.a. 3.81 3.81 3.81
HGV Diesel Rigid Euro 0 1.05 1.28 1.02

<=7,5 t
Euro I 0.97 1.04 0.85
Euro II 0.94 0.97 0.82
Euro III 0.99 1.04 0.85
Euro IV 1.01 0.98 0.86
Euro V 1.00 0.94 0.83
Euro VI 1.00 0.96 0.84
Rigid Euro 0 0.54 0.75 0.54
category class
7,5-12 t Euro I 0.49 0.62 0.48
Euro II 0.48 0.59 0.46
Euro III 0.50 0.62 0.48
Euro IV 0.50 0.59 0.48
Euro V 0.52 0.59 0.46
Euro VI 0.52 0.60 0.46
Rigid Euro 0 0.31 0.44 0.31

12-14 t
Euro I 0.28 0.37 0.27
Euro II 0.27 0.35 0.27
Euro III 0.28 0.37 0.28
Euro IV 0.28 0.35 0.27
Euro V 0.26 0.34 0.27
Euro VI 0.26 0.35 0.27
Rigid Euro 0 0.35 0.56 0.37

14-20 t
Euro I 0.30 0.45 0.31
Euro II 0.29 0.42 0.30
Euro III 0.30 0.45 0.31
Euro IV 0.29 0.41 0.30
Euro V 0.27 0.41 0.29
Euro VI 0.28 0.41 0.30
Rigid Euro 0 0.19 0.33 0.21

20-26 t
Euro I 0.17 0.27 0.18
Euro II 0.17 0.26 0.18
Euro III 0.17 0.28 0.18
Euro IV 0.17 0.26 0.18
Euro V 0.16 0.26 0.17
Euro VI 0.16 0.26 0.18
Rigid Euro 0 0.14 0.24 0.16

26-28 t
Euro I 0.13 0.21 0.14
Euro II 0.12 0.20 0.13
Euro III 0.13 0.21 0.14
Euro IV 0.13 0.20 0.14
category class
Euro V 0.12 0.19 0.13
Euro VI 0.12 0.20 0.13
Rigid Euro 0 0.13 0.22 0.15

28-32 t
Euro I 0.12 0.19 0.13
Euro II 0.12 0.19 0.13
Euro III 0.12 0.19 0.13
Euro IV 0.12 0.18 0.13
Euro V 0.12 0.18 0.13
Euro VI 0.12 0.18 0.13
Rigid Euro 0 0.11 0.20 0.13

> 32 t
Euro I 0.10 0.17 0.11
Euro II 0.10 0.16 0.11
Euro III 0.10 0.17 0.11
Euro IV 0.10 0.16 0.11
Euro V 0.10 0.16 0.11
Euro VI 0.10 0.16 0.11

14-20 t
Euro I 0.16 0.25 0.17
Euro II 0.16 0.24 0.17
Euro III 0.16 0.26 0.17
Euro IV 0.16 0.24 0.17
Euro V 0.15 0.23 0.16
Euro VI 0.15 0.24 0.17

20-28 t
Euro I 0.16 0.28 0.18
Euro II 0.16 0.26 0.17
Euro III 0.16 0.27 0.18
Euro IV 0.16 0.26 0.18
Euro V 0.16 0.25 0.17
Euro VI 0.16 0.25 0.17

28-34 t
Euro I 0.10 0.17 0.11
category class
Euro II 0.10 0.17 0.11
Euro III 0.10 0.17 0.11
Euro IV 0.10 0.16 0.11
Euro V 0.10 0.16 0.11
Euro VI 0.10 0.16 0.11

34-40 t
Euro I 0.10 0.18 0.11
Euro II 0.10 0.17 0.11
Euro III 0.10 0.18 0.11
Euro IV 0.10 0.17 0.11
Euro V 0.10 0.17 0.11
Euro VI 0.10 0.17 0.11

40-50 t
Euro I 0.09 0.17 0.11
Euro II 0.10 0.17 0.10
Euro III 0.09 0.17 0.11
Euro IV 0.09 0.16 0.10
Euro V 0.09 0.16 0.10
Euro VI 0.09 0.16 0.10

50-60 t
Euro I 0.10 0.18 0.11
Euro II 0.10 0.17 0.11
Euro III 0.10 0.18 0.11
Euro IV 0.10 0.17 0.11
Euro V 0.10 0.17 0.11
Euro VI 0.10 0.17 0.11
LNG Articulated n.a. 0.03 0.05 0.04

32+
Table 54 – Average/marginal costs of well-to-tank emissions rail transport for selected cases
Train type Traction €-cent/pkm or €-cent/train-km
€-cent/tkm
Passenger transport
High-speed train Electricity 0.39 117.4
Intercity train Electricity 0.73 117.4
Intercity train Diesel 0.18 16.1
Regional train Electricity 0.89 97.9
Regional train Eiesel 0.26 8.0
Freight transport
Long container Electricity 0.11 151.3
Long bulk Electricity 0.10 156.8
Short container Electricity 0.26 132.0
Short bulk Electricity 0.18 135.4
Long container Diesel 0.03 42.0
Long bulk Diesel 0.03 43.5
Short container Diesel 0.07 36.7
Short bulk Diesel 0.05 37.6
Table 55 – Average/marginal costs of well-to-tank emissions IWT for selected cases (€-cent per tkm)
Vessel type Type of cargo €-cent/tkm €-cent/vkm
CEMT II (350 t) Bulk 0.15 46.82
Container 0.09 46.82
CEMT IV (600 t) Bulk 0.09 69.49
CEMT Va (1,500 t) Bulk 0.08 126.96
Pushed convoy (11,000 t) Bulk 0.06 340.02
Table 56 – Average/marginal costs of well-to-tank emissions maritime transport for selected cases
Vessel type Distance at sea € per trip €-cent per € per

Passenger transport
RoPax Ferry (25,500 gt) 100 2,190 4.15 22
Vessel type Distance at sea € per trip €-cent per € per
500 5,513 2.09 11
Small container vessel (28,500 gt) 500 15,437 0.13 31
3,000 50,891 0.07 17
Large container vessel (143,000 gt) 500 23,338 0.04 47
3,000 110,190 0.03 37
15,000 527,084 0.03 35
Small bulk vessel (18,000 gt) 500 4,509 0.06 9
3,000 19,499 0.04 6
Large bulk vessel (105,000 gt) 500 10,987 0.02 22
3,000 45,149 0.01 15
15,000 209,129 0.01 14
Table 57 – Average/marginal costs of well-to-tank emissions aviation for selected cases
Type of flight Distance Emission Example of aircraft type €-cent per pkm € per pax
class
(km)
Short haul 500 Low Bombardier CRJ900 1.3 6
500 High Embraer 190 1.4 7
Medium haul 1,500 Low Airbus 320 0.5 8
1,500 High Boeing 737 0.7 11
3,000 Low Airbus 320 0.6 17
3,000 High Boeing 737 0.8 24
Long haul 5,000 Low Airbus 340 0.5 25
5,000 High Boeing 777 0.6 28
15,000 Low Airbus 340 0.8 121
15,000 High Boeing 777 0.9 137
The well-to-tank costs due to the emission of air pollutants and greenhouse gases are quite
well investigated. However, the emissions are occurring always indirectly and the exact
place of the emissions is generally not known. Therefore, the emission levels always have
some uncertainty. The cost factors also have some uncertainty due to the number of people
exposed and country where the emissions occur generally being unknown. Overall, the
quality of the results for well-to-tank costs is acceptable, but not as high as for the costs
due to direct emissions (tank-to-wheel emissions).
A further conceptual consideration is that part of the well-to-tank greenhouse gas emissions
will be regulated under the EU Emissions Trading System (EU ETS), which means a carbon
price is applied. As the EU ETS is a capped system, reducing well-to-tank emissions does not
necessarily lead to equal economy-wide emission reductions in the short term.
9 Costs of habitat damage
9.1 Introduction
Transport has different effects on nature, landscape and natural habitats. The main effects
reported in literature are habitat loss (ecosystem loss), habitat fragmentation and negative
effects on ecosystems due to the emission of air pollutants (e.g. biodiversity loss).
However, the negative effects of transport on nature and landscape are covered in limited
external costs studies.
In this chapter, the scope of the cost category is presented (see Section 9.2) and then the
total and average costs of nature and landscape are described (see Section 9.3) — first the
methodological approach, then the results.

The different negative effects of transport on nature and landscape can be described as the
following:
— Habitat loss: Transport infrastructure requires land and/or natural surfaces. Therefore,
transport infrastructure also leads to a loss of natural ecosystems, which are natural
habitats of plants and animals. The land use of transport therefore leads to a loss of
habitats (ecosystems), which has a negative effect on biodiversity. Habitat loss is
occurring during the building phase of transport infrastructure, but will last over the
whole lifetime of the infrastructure.
— Habitat fragmentation: Transport infrastructure can also have additional fragmentation
and separation effects for animals. These fragmentation effects can negatively affect
the natural habitats of certain species and lead to adverse effects for species and
consequently on biodiversity. Habitat fragmentation due to transport infrastructure is
a consequence of the infrastructure itself plus the transport demand on the
infrastructure. The main negative effects are caused by large and broad main
infrastructures such as motorways and high-speed rail lines. Large wildlife mammals
such as deer, rabbit, badger, etc. as well as smaller animals such as amphibians are
negatively affected by habitat fragmentation.
— Habitat degradation due to emissions: Habitat degradation can also occur via the
emission of air pollutants of other toxic substances (e.g. heavy metals, PAH).
These effects again lead to biodiversity loss and therefore external costs.
The biodiversity loss due to air pollution is already covered in the air pollution chapter
(4), where all adverse impacts of air pollution are included. The negative effects of the
emission of toxic substances are covered in Chapter 10.
There are limited studies covering the external costs of habitat damage due to transport
activities. A brief overview on the relevant literature is shown in Annex G. The most
detailed bottom-up calculations of the cost of habitat damage have been made by the
European research project NEEDS (2006), INFRAS, Ecoplan (2018) and UBA (2018), based on
NEEDS.
9.3 Total and average costs of habitat damage
9.3.1 Methodology
Total and average costs of habitat damage are calculated based on the infrastructure
network length (or area) and average cost factors for habitat loss and habitat
fragmentation. Figure 6 illustrates the corresponding methodology.
Figure 11 – Methodology total and average costs of habitat damage
Network length / area Cost factors

Input values
- road length (motorways, - Habitat loss (covered area)

others) €/km or €/km2
- rail length, waterways - Habitat fragmentation,
- airport area, # of flights €/km
Calculations
Total external costs of habitat damage per transport

mode per country
Allocation of total habitat costs to vehicle categories

Allocation
based on: transport demand, average vehicle length
Total and average habitat costs per vehicle category in

each country
9.3.2 Input values
Network length (network area)

The following data sources for transport infrastructure network length (or infrastructure
area) are being used:
— Road Transport: EU Transport in Figures.
— Rail Transport: EU Transport in Figures.
— Air Transport: airports (survey); and EU Transport in Figures.
— Inland Waterways: EU Transport in Figures and Eurostat.
— Maritime: ports (survey).
Cost factors
The following table summarizes the cost factors for habitat loss and habitat fragmentation
for the EU28 average. The cost factors are derived from the latest bottom-up calculations
for the Swiss study on external costs of transport (INF (INFRAS en Ecoplan, 2018).
Table 58 – Cost factors for costs of habitat damage EU28
Cost in €2016 per km and year Road Rail Aviation Inland

€/(km *a) €/(km*a) €/(km2 *a) waterways
€/(km*a)
Motorways Other roads High-speed Other railways
Habitat loss 78,900 1,900 57,500 8,200 437,500 6,600
Habitat fragmentation 14,600 2,200 27,000 5,900 0 0
Total habitat damage 93,500 4,100 84,500 14,100 437,500 6,600
Source: Own calculations based on INFRAS, Ecoplan 2018 (External effects of transport in Switzerland 2015) .
Based on the EU28 average values, cost factors for all countries have been calculated (Table
59), based on the same value transfer approach used in the whole Handbook (GDP/cap PPP
adjusted).
Table 59 – Cost factors for costs of habitat damage for all countries
Cost in €2016 per km and year Road €/(km *a) Rail €/(km*a) Aviation Inland
€/(km2 *a) waterways
Motorways Other roads High-speed Other railways €/(km*a)
EU Aggregate (EU28) 93,500 4,100 84,500 14,100 437,500 6,600
Austria 99,700 4,400 90,200 15,000 466,800 7,100
Belgium 93,900 4,100 84,900 14,100 439,200 6,700
Bulgaria 52,900 2,300 47,800 8,000 247,400 3,800
Croatia 56,100 2,500 50,700 8,500 262,700 4,000
Cyprus 69,500 3,100 62,900 10,500 325,500 4,900
Czech Republic 65,400 2,900 59,100 9,900 306,100 4,600
Denmark 125,300 5,500 113,300 18,900 586,500 8,900
Estonia 74,900 3,300 67,800 11,300 350,700 5,300
Finland 125,100 5,500 113,100 18,900 585,500 8,900
France 111,100 4,900 100,500 16,700 520,000 7,900
Germany 120,500 5,300 108,900 18,200 563,800 8,600
Greece 67,500 3,000 61,000 10,200 315,700 4,800
Hungary 55,400 2,400 50,100 8,300 259,100 3,900
Ireland 88,600 3,900 80,100 13,300 414,400 6,300
Italy 74,700 3,300 67,500 11,300 349,400 5,300
Latvia 66,000 2,900 59,600 9,900 308,700 4,700
Lithuania 66,400 2,900 60,000 10,000 310,600 4,700
Luxembourg 98,600 4,300 89,100 14,900 461,400 7,000
Cost in €2016 per km and year Road €/(km *a) Rail €/(km*a) Aviation Inland
€/(km2 *a) waterways
Motorways Other roads High-speed Other railways €/(km*a)
Malta 76,900 3,400 69,500 11,600 359,800 5,500
Netherlands 103,000 4,500 93,100 15,500 482,000 7,300
Poland 64,200 2,800 58,100 9,700 300,600 4,600
Portugal 55,000 2,400 49,700 8,300 257,500 3,900
Romania 48,000 2,100 43,400 7,200 224,600 3,400
Slovakia 69,400 3,000 62,700 10,500 324,700 4,900
Slovenia 66,900 2,900 60,500 10,100 313,000 4,800
Spain 72,100 3,200 65,200 10,900 337,400 5,100
Sweden 142,100 6,200 128,500 21,400 665,000 10,100
United Kingdom 99,000 4,300 89,500 14,900 463,400 7,000
Norway 134,500 5,900 121,600 20,300 629,700 9,600
Switzerland 153,900 6,800 139,100 23,200 720,300 10,900
Canada 105,300 4,600 95,200 15,900 492,900 7,500
United States 110,200 4,800 99,600 16,600 515,500 7,800
Japan 103,400 4,500 93,400 15,600 483,700 7,300
Source: Own calculations based on INFRAS, Ecoplan 2018 (External effects of transport in Switzerland 2015) .
An alternative source for cost factors for habitat loss is the European research study NEEDS
(2006), where cost factors for biodiversity loss due to ecosystem (habitat) loss have been
reported. The cost factors are based on a restoration cost approach and presented for
different ecosystem types.
Table 60 – Cost factors for ecosystem (habitat) loss
Land/ecosystem type €2015/m2
Built-up land 0
Intensive arable 0
Integrated arable 0.2
Organic arable 0.5
Organic orchards 4.8
Intensive pasture 0.4
Less intensive pasture 0.9
Organic pasture 2.3
Forests 3.2
Forest edge 10.0
Average (all ecosystems) 1.8
Source: Own calculations based on NEEDS 2006.
The following results on the external costs of habitat damage have been based on the cost
factors presented in Table 58 and Table 59 (based on INFRAS, Ecoplan 2018).
9.3.3 Output values

The following tables show the resulting cost factors (output values) for the cost of habitat
damage per vehicle type, based on the cost factors in Table 59 (based on INFRAS, Ecoplan
2018). The tables include the total costs as well as the average costs per vkm and the
average costs per pkm and tkm.
Table 61 – Total and average costs of habitat damage for land-based transport modes in the EU28
Bus 0.2 0.10 1.9
Coach 0.4 0.11 2.2
High speed passenger train 0.7 0.62 185
Passenger train electric 1.4* 0.57 75
Passenger train diesel 0.5 0.84 49
LCV 4.4 1.35 0.9
HGV 3.6 0.19 2.4
Freight train electric 0.8 0.24 134
Freight train diesel 0.2 0.25 111
Total road, rail & inland 39.1

waterway
trains).
Table 62 – Total and average habitat costs for aviation for 33 selected EU airports
Type of flight Billion € €-cent/pkm €-cent/pax
Passenger aviation
Short haul (< 1,500 km) 0.024 0.027 14.9
Medium haul (1,500–5,000 km) 0.021 0.007 12.3
Long haul (> 5,000 km) 0.005 0.0008 6.5
Total passenger aviation 0.050 0.007 12.2
Freight aviation Billion € €-cent/tkm
Freight aviation total 0.006 n.a.
Concerning the marginal costs of habitat damage, the marginal costs of habitat loss are
virtually zero (only if infrastructure capacity has to be enhanced due to high demand, there
are additional marginal costs). The marginal costs of habitat fragmentation, however, can
be assumed to be substantial, and in some cases almost as high as the average cost of
habitat fragmentation, since the traffic on a road really hinders animals to pass.
It is not possible to make a generally applicable estimation of the marginal costs of habitat
damage. However, the marginal costs will be between zero (as a minimum estimation) and
the average costs of habitat fragmentation (as a maximum estimation).
10 Other external costs
The goal of this chapter is to present the further negative externalities of transport.
The chapter focuses on other external cost categories that are not quantified or discussed
in the previous chapters. Many sub-categories are already mentioned in the previous
chapters such as damage costs due to toxic emissions or cost of downstream emissions
(see Chapter 8).
This chapter focuses on a qualitative description and brief discussion of other external costs
of transport and gives some selective quantitative references. All cost categories with
quantitative results and recommendations of cost factors are covered in the detailed
chapters above.
10.1 Costs of soil and water pollution

Transport activities lead to certain negative impacts on soil and water:
— Pollution of soil (and water) with heavy metals. Different processes lead to the emissions
of heavy metals: abrasion of brakes (road, rail), abrasion of tyres and rail track (road,
rail), abrasion of the overhead line (rail) as well as the fuel combustion. There are limited
studies monetizing the external cost of the emission of heavy metals in transport. In
Switzerland, these costs are quantified, but are below 1% of the total external costs of
transport. They are most relevant for motorised road transport and rail transport
(INFRAS en Ecoplan, 2018). A similar approach has been applied in the last UCI study on
external costs of transport in Europe (CE Delft, INFRAS & Fraunhofer ISI, 2011).
— Pollution of soil (and water) with organic toxic substances. The burning of fuels also
leads to the emission of organic toxic substances (persistent organic pollutants POP).
However, the corresponding emissions are relatively low. One of the few studies
covering the cost of organic toxic emissions of transport is INFRAS, Ecoplan (2018).
In water transport (maritime transport, inland waterways) the use of antifouling agents
for ship paint is another source of emissions of toxic substances, generally organic
pollutants, metal-organic compounds or heavy metals (e.g. zinc).
Specific cost factors for the emission of toxic pollutants such as heavy metals and
organic compounds are recommended in the new Dutch Shadow Prices Handbook
(CE Delft, 2018) or in the European research study EXIOPOL (2011).
— Pollution of soil (and water) due to waste and ballast water. In maritime transport and
inland waterways, waste water and ballast water leads to the pollution of seawater,
lakes or rivers. The emissions mainly occur in ports and can lead to substantial water
pollution. There are several methods for treating ballast and other waste water, which
all lead to substantial costs. Hence, the cost of waste and ballast water could be
quantified by using a restoration cost approach. The external cost of waste and ballast
water are discussed in many studies (e.g. (Hayman, et al., 2000), (Trozzi, 2003) and
where quantified (which is rare) this is done by the means of the restoration costs
(e.g. (JRC, 2009).
— Pollution of soil (and water) due to oil spills and oil risks. A specific type of water
pollution is the cost of oil spills or accidents related to the extraction of oil. Such
uncontrolled oil emissions lead to substantial pollution of the sea. Above all, the large
accidents with oil drillings (e.g. Deepwater Horizon) have led to high environmental
costs. Many studies mention those effects and give indicative data on the external costs
(e.g. (Navrud, et al., 2016); (Farrow & Larson, 2012); (Bigano, et al., 2009); (VTPI,
2017). Those cost factors are however mainly from single events or accidents or refer to
the costs of preventative measures (Navrud, et al., 2016).
10.2 Costs of up- and downstream emissions of vehicles and infrastructure

The upstream costs of energy production are covered in Chapter 8. However, there are
different other up- and downstream processes directly related to transport that also lead to
negative external effects. Taking a life-cycle oriented view on transport, the following
processes lead to the emission of air pollutants, greenhouse gases, toxic substances and
other negative environmental impacts:
— Vehicles: An important input factor for transport are the vehicles (e.g. cars, trucks,
planes, rolling stock, etc.). The production, maintenance and disposal of vehicles causes
the emission of air pollutants, greenhouse gases and other pollutants. Hence, during the
whole life cycle of the vehicles, negative environmental impacts occur.
— Infrastructure: Similar to vehicles, transport infrastructure also leads to negative
environmental effects during its life cycle. Emission of greenhouse gases, air pollutants
and other substances are caused by the construction, maintenance and disposal of
transport infrastructures.
— Energy production (other effects): The energy production (fossil fuels, electricity) do
not only lead to the emission of air pollutants and greenhouse gases (covered in Chapter
8), but also to ecosystem damage and loss. For example, electricity production by wind
power stations, hydroelectric plants or large solar energy plants can cause substantial
ecosystem damage.
The life-cycle emissions of air pollutants and greenhouse gases for vehicles and
infrastructure are covered in a number of external cost studies (e.g. (UBA, 2018), (CE Delft,
INFRAS & Fraunhofer ISI, 2011), (INFRAS en Ecoplan, 2018)) . However, the ecosystem
damage due to energy production is covered only in selected studies, generally focussing on
very concrete cases and ecosystems (e.g. based on a willingness-to-pay study). Generalised
cost factors or shadow prices for ecosystem damage due to wind, solar or hydro plants are
not available yet.
10.3 External costs in sensitive areas (e.g. mountainous regions)

Different studies have proved that certain types of external costs are higher in sensitive
areas, such as mountain areas, than in non-sensitive areas. An important study analysing
these impacts was the European research study GRACE (2006), which identified so called
mountain-factors describing the differences in external costs between mountain areas and
non-mountain areas. As a result, the Eurovignette directive allows mark-ups to HGV tolls in
sensitive areas.
Recently, two studies have validated and updated the mountain factors, based on the
methodological approach of the GRACE (2006) study (EUSALP, 2017), (CEREMA, 2018). In the
study from EUSALP (2017), all cost drivers that influence the different environmental costs
are reassessed, considering the latest research results. Additionally, possible additional cost
drivers are examined as well as additional cost categories (accident costs, costs for nature
and landscape). Finally, new mountain factors for Alpine regions have been suggested.
The following section summarises briefly the methodology and results from the EUSALP
(2017) study.
The analysis to derive cost factors (mountain factors) follows the approach along the
‘impact pathway approach’, as the main methodology to assess environmental costs based
on a damage cost approach. The methodological approach used in the present study is the
same as in the GRACE study (2006), which is based on cost drivers and ‘cost differential
factors’ (mountain factors) along the impact-pathway. The following elements of the
impact-pathway can influence the external costs in mountain regions:
— Emissions: higher emission level e.g. due to gradients and altitude.
— Concentration: higher concentration of air pollutants e.g. due to topographical and

meteorological conditions.
— Impacts: different impacts based on the dose(concentration)-response evidence, e.g. due
to other population density or other risk factors.
— Damage cost: different cost factors for damage costs, i.e. due to country-specific
monetization factors, specific prices, etc.
In summary, the mountain factors for all levels of the impact-pathway approach result in an
overall factor for the cost per transport performance (vkm) in mountain regions in
comparison to non-mountain regions (or a country average). The result of the analysis is a
‘mountain factor’ for a certain category of external costs (e.g. air pollution costs, noise
costs, accident costs) and transport mode (road, rail).
The EUSALP (2017) study analysed the following cost categories: air pollution costs, noise
costs, nature and landscape costs (habitat damage), accident costs and climate change
costs. The EUSALP study focuses on rail and road freight transport. The analysis is based on
a corridor approach, meaning that the cost factors derived apply to whole corridors and not
only specific infrastructures.
The following table summarises the main results of the EUSALP (2017) study, showing the
mountain factors for the different external cost categories. Additionally, the values of the
GRACE study (2006) are also represented as a comparison. It is important to state that the
different mountain factors do not say anything about the absolute level of external costs,
but only represent the factors between external costs in mountainous and external costs in
non-mountainous areas.
Table 63 – Mountain factors for external costs of freight transport
Cost category EUSALP (2017) GRACE (2006)
Road transport Rail transport Road transport Rail transport
Air pollution 4.2 2.6 5.25 3.5
Noise 4.1 3.0 5.0 4.15
Nature & landscape 1.3 1.4 n.a.* n.a.*
Accidents 3.9 n.a. n.a. n.a.
Source: EUSALP (2017). n.a.: not available/no data available.
* For visual intrusion, the GRACE study suggested a factor of 10.7 for road transport and 5.3 for rail transport.
The results of the analysis of external costs of transport in mountain areas can be
summarised as following (cited from EUSALP 2017):
— Air pollution costs: The main cost driver for the air pollution costs in the Alpine Region
are the higher immissions due to inversion (factor 4.4). Other cost drivers are the higher
emissions due to the higher gradients and the altitude (e.g. higher emissions due to
steeper roads/rails and higher exhaust emissions in higher altitudes). The resulting
mountain factor for air pollution is slightly lower than in the GRACE (2006) study, which
is mainly as a result of the lower factor for population density (based on a more detailed
GIS based analysis), which outweighs the slightly higher value for the immission
(concentration). However, overall the air pollution costs in mountain areas are
substantially higher than in non-mountain areas.
— Noise costs: For noise costs, the main cost driver in mountain areas are the higher
immissions due to topographical and meteorological conditions (inversion, amphitheatre
effect). Other relevant factors are the gradients (higher noise emissions due to steeper
roads/rails) and the population density. The resulting mountain factor for noise costs
are slightly lower than in the GRACE (2006) study (again mainly as a result of the lower
factor for population density).
— Costs of habitat damage (nature and landscape): For nature and landscape, a mountain
factor has been derived for the first time in the EUSALP (2017) study. Based on detailed
results of the Swiss study on external costs of transport (INFRAS en Ecoplan, 2018),
significantly higher costs for habitat loss and habitat fragmentation in mountain areas
compared to non-mountain areas have been derived due to more diverse and more
valuable ecosystems. The resulting mountain factors are 1.3 for road (motorways) and
1.4 for rail transport.
— Accident costs: For accident costs, there is also evidence for higher costs in mountain
areas, mainly due to higher infrastructure investments to keep the accident rate as low
as possible. For the first time, a mountain factor has been derived for accident costs in
the EUSALP (2017) study. The calculation is based on an avoidance cost approach taking
into account additional infrastructure safety measures on roads in Alpine corridors.
The resulting factor for accident cost in mountain areas is around 5.
— Climate change costs: For climate change, a mountain factor has been checked in the
EUSALP (2017) study. The conclusion was that no mountain factor cannot be derived for
methodological reasons (it is a global issue with global effects).
10.4 Further externalities of transport

There are several other negative externalities of transport that can be mentioned, but that
are only partially covered in literature:
— Separation costs in urban areas: large transport infrastructures in urban areas (mainly
motorways and large rail fields) lead to separation effects and time losses for
pedestrians.
— Land use and ecosystem damage for upstream processes: Different upstream processes
of transport can lead to ecosystem damage and/or land use, e.g. the electricity
generation or the exploitation of mineral oil products. The external costs of those effects
are not in the main focus of the Handbook, although they are relevant. Additionally,
there are existing some selected studies on external costs associated with those
processes (e.g. Mattmann et al. 2016), but no well accepted cost factors, e.g. for
different land use or ecosystem damage due to different types of electricity generation.
— Cost of nuclear risks: Another type of upstream cost from energy production is the risks
from nuclear power plants. The risk of a nuclear incident has a very low probability but a
potentially very high damage potential (a ‘Damocles risk’). Therefore, it is very difficult to
quantify the corresponding external costs. In addition to the costs of potential nuclear
incidents, the disposal of nuclear waste is linked to costs (the costs of the disposal, often
not covered in the electricity production costs; and also the risk of the disposal). The
nuclear risks are a relevant externality, although it can hardly be quantified. Some older
studies cover the costs due to nuclear power risks, however, the basis for monetizing
those external effects is very old and no broadly accepted new studies on the cost of
nuclear power risks have been conducted. Hence, this Handbook does not include cost
factors for the nuclear power risks (in line with previous versions of the Handbook).
11 Synthesis
11.1 Introduction
This chapter contains the overall results on average and total external costs. First the total
external costs are discussed, then the average external costs. All figures refer to the EU28
member states for the year 2016. The congestion costs are not shown in certain graphs
because they cannot be shown for all vehicle types. However, the sum of the congestion
costs is included in the analysis.
11.2 Overview total and average external costs
11.2.1 Total external costs

Figure 12 presents the total external costs of transport for EU28 by transport mode and cost
category for 2016. The total external costs for road, rail, inland waterway transport,
aviation and maritime (excluding congestion costs, because they are not calculated for all
modes) amount to € 716 billion, which corresponds to 4.8% of the total GDP in EU28. The
congestion costs amount to another € 271 billion for 2016 (delay costs generated by road
transport modes)42. The total external costs including congestion costs sum up to € 987
billon (6.6% of the GDP).
For aviation and maritime transport, the detailed calculation of the external costs has only
been done for for a set of selected airports and ports. The total external costs for EU28 for
aviation and maritime have only been roughly estimated. Table 67 shows the results for the
external costs for the selected airports and ports, Table 68 includes the indicative figures
(estimations) for EU28 for aviation and maritime transport43. For the selected 33 EU airports
the external costs amount to € 33 billion, for the selected 34 EU ports the costs amount to
€ 44 billion.
The most important cost category is accident costs equating to 29% of the total costs,
followed by the congestion costs (27%) (see Figure 13). Climate change and air pollution
costs both contribute to 14% of the total costs, noise costs to 7% and habitat damage to 4%
of the total costs. Well to tank emission costs due to energy production and distribution
lead to 5% of the costs.
Road transport is the predominant mode that causes by far the most external costs (83% of
the total costs incl. aviation and maritime; 97.5% excl. aviation and maritime). Maritime
transport causes 10%, aviation 5%, rail transport 1.8% and inland waterways 0.3% of the
costs (see Figure 14). 69% of the total costs are due to passenger transport, 31% of the costs
are caused by freight transport (including LCVs).
________________________________
42 Please notice, that part of these delay costs are internalised and hence that they are only partly external. The
size of the external part of the delay costs is unknown.
43
Not for all external costs categories an estimation of the total EU28 costs for aviation and maritime can be
calculated. For those (small) cost categories without EU28 estimation, the results for the selected airports and
ports have been used.
Figure 12 – Total external costs 2016 for EU28 (excluding congestion)
Bi l lion € per year
Habitat damage Well-to-Tank

400
350 Noise Climate
300 Air Pollution Accidents
250
200
150
100
50
0
Car Bus/Coach MC Rail Aviation° LDV HGV Rail IWT Aviation° Maritime°
Passenger transport Freight Transport
° Data for aviation and maritime: rough estimations for EU28.
Figure 13 – Share of the different cost categories on total external costs 2016 for EU28
5% 4% Accidents
Air Pollution
29%
Climate
27%
Noise
Congestion
14%
7% Well-to-Tank
14%
Habitat damage
Including data for aviation and maritime: rough estimations for EU28.
Figure 14 – Share of the different transport modes on total external costs 2016 for EU28
5% 1.8% 0.3%
Road
10%
Maritime°
Aviation°
Rail
83%
IWT
* Data for aviation and maritime: rough estimations for EU28.
The total external costs per country are shown for road, rail and IWT in Table 6444.
This table also shows the share of these costs in the national GDP. This share range from
3.4% in Norway to over 7% in Portugal and Luxembourg.
Table 64 – Total external costs 2016 for road transport, rail transport and IWT per country
Country Road (bn €) Rail (bn €) IWT (bn €) Total (bn €) % of GDP
EU 28 820.4 17.87 2.90 841.1 5.7%
Austria 18.3 0.85 0.044 19.2 5.9%
Belgium 26.4 0.42 0.183 27.0 7.0%
Bulgaria 6.5 0.12 0.047 6.6 6.5%
Croatia 5.0 0.07 0.015 5.1 6.9%
Cyprus 1.1 - - 1.1 5.1%
Czech Republic 13.6 0.40 0.004 14.0 5.2%
Denmark 8.2 0.18 - 8.4 4.1%
Estonia 1.5 0.04 0.014 1.5 5.3%
Finland 7.4 0.23 0.073 7.7 4.4%
France 109.1 1.76 0.181 111.0 5.5%
Germany 165.7 5.37 1.228 172.3 5.8%
Greece 12.8 0.06 - 12.8 6.0%
Hungary 11.1 0.43 0.037 11.5 6.0%
Ireland 14.3 0.06 - 14.4 5.7%
Italy 115.0 2.20 0.009 117.2 6.8%
Latvia 2.3 0.18 - 2.5 6.7%
Lithuania 3.9 0.12 - 4.0 6.3%
Luxembourg 3.2 0.03 0.009 3.3 7.5%
Malta 0.4 - - 0.4 3.6%
Netherlands 29.6 0.35 0.848 30.8 4.9%
Poland 40.2 1.28 0.018 41.5 5.5%
Portugal 16.8 0.18 - 16.9 7.2%
Romania 21.2 0.46 0.171 21.8 6.5%
Slovakia 5.4 0.33 0.012 5.7 4.7%
Slovenia 2.7 0.05 - 2.7 5.5%
Spain 64.3 0.83 - 65.1 5.2%
Sweden 15.3 0.46 - 15.8 4.5%
United Kingdom 99.4 1.42 0.009 100.8 4.9%
Norway 7.4 0.17 - 7.6 3.4%
Switzerland 15.3 0.76 0.001 16.1 4.1%
________________________________
44
As not all ports and airports are included, they have to be excluded when comparing different countries.
Otherwise small countries where e.g. all air traffic takes places in one airport would see a higher cost as share
of GDP only because all its aviation activity is accounted for.
Table 65 – Total external costs 2016 for EU28 passenger transport by cost category and transport mode
Passenger Transport
Road Rail
Pass car - Pass car - Pass car - Bus Coach MC High-speed Electric pax convent (non Diesel tot pax
petrol diesel total Train high speed)
Cost category bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a
Accidents 210.2 5.3 21.0 0.06 2.0
Air Pollution 8.6 24.8 33.4 1.4 2.7 1.8 0.002 0.03 0.52
Climate 32.0 23.5 55.6 0.8 1.6 1.5 0.00 0.00 0.22
Noise 13.8 12.4 26.2 0.8 0.9 14.8 0.4 2.6 0.9
Congestion * 196.1 4.5
Well-to-Tank 10.4 7.7 18.1 0.3 0.5 0.8 0.3 2.7 0.1
Habitat damage 14.1 11.8 25.9 0.2 0.4 0.5 0.7 1.4 0.5
Total 565.4 19.3 40.5 1.4 11.0
Total per mode 625.2 12.5
Total as % of EU28 GDP 4.2% 0.1%
* Congestion in terms of delay cost generated by the various vehicle categories.
Table 66 – Total external costs 2016 for EU28 freight transport by cost category and transport mode
Freight Transport
Road Rail IWT
LCV-petrol LCV-diesel LCV-total HGV - total Electric freight Diesel freight Inland vessel
Cost category bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a
Accidents 19.8 23.0 0.3 0.1
Air Pollution 0.3 15.2 15.5 13.9 0.01 0.7 1.9
Climate 0.7 12.5 13.2 9.6 0.00 0.2 0.4
Noise 5.4 9.1 2.1 0.4
Congestion* 55.5 14.6
Well-to-Tank 0.2 3.6 3.8 3.7 0.5 0.1 0.2
Habitat damage 0.2 4.2 4.4 3.6 0.8 0.2 0.3
Total 117.6 77.5 5.4 2.9
Total per mode 195.1 5.4 2.9
Total as % of EU28 GDP 1.31% 0.04% 0.02%
* Congestion in terms of delay cost generated by the various vehicle categories.
For aviation and maritime transport the total costs have been calculated only for selected
airports and ports. The total and average costs for aviation and maritime shipping for the
selected (air)ports are shown in the following table.
Table 67 – Total external costs for selected EU28 (air)ports
Aviation Maritime
Passenger Freight Total
Short- Medium- Long- Short- Medium- Long- Aviation Maritime ship

passenger passenger passenger belly belly belly
freight freight freight
Cost category bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a bn €/a
Accidents 0.1 0.1 0.1
Air Pollution 0.24 0.32 0.30 0.03 0.06 0.06 1.0 29.1
Climate 1.9 4.9 12.3 0.21 0.63 2.06 22.0 10.6
Noise 0.8 0.8 n/a
Congestion** n/a n/a n/a
Well-to-Tank 0.86 1.84 4.93 0.09 0.26 0.90 8.9 3.9
Habitat damage 0.050 0.006 0.056 n/a
Total selected 28.6* 4.3 32.9 43.6

(air)ports
Total for 0.2% 0.03% 0.3%

selected
(air)ports as %
of EU28 GDP
* Noise and accident costs have been allocated to passenger transport.
In addition, the total for the selected (air)ports have been extrapolated to retrieve
estimates for EU28, for the most important cost categories. It should be noted that these
estimates are relatively rough, as they assume that the the transport to/from the selected
(air)ports (in terms of aircraft, ships and distances) are representative for the entire EU.
Table 68 – Indicative estimates for the most relevant total external costs for all EU28 (air)ports
Aviation45 Maritime
Cost category bn €/a bn €/a
Air Pollution n/a 65
Climate 33 24*
Well-to-Tank 13 9
Other cost categories** 2 0.1
Total EU28 48 98
Total for EU28 as % of 0.3% 0.7%

EU28 GDP
* These total climate costs for maritime shipping have been based on the transport performance, in order to be
consistent with the other transport modes and cost categories. The 2018 DG MOVE pocketbook reports
167.2 million tonnes CO 2 equivalent by navigation, out of which 20.3 is domestic navigation (which includes
inland waterways). According to these data the climate cost of maritime and inland waterways together are
16.7 bn €. As the climate costs of inland navigation are 0.4 bn € (see Table 66), the costs of maritime shipping
amount 16.3 bn €, according to these data. However, these data are based on bunkered fuels which is not a
appropriate proxy for the fuel consumed for th e international maritime shipping to/from EU ports.
** For the (small) cost categories for which no estimations of EU28 was available, the results for the selected
airports and ports have been used.
11.2.2 Average external costs

The average external costs of transport are expressed in Euro cent per pkm and tkm.
Looking at passenger transport (see Figure 15), passenger cars cause external costs of
€-cent 7.8 per pkm without congestion and 12.0 €-cent per pkm including congestion.
The average costs of passenger rail transport amount to €-cent 2.8 per pkm, which is 2.8
times lower than the costs for the road sector (without congestion). Average costs for rail
transport differ a lot between electric trains and diesel trains. Due to significantly higher
climate change and air pollution costs, the average costs of diesel trains are €-cent 3.9 per
pkm, whereas the costs of electric trains only amount to €-cent 2.6 per pkm (average of all
electric trains). The cost of high speed rail is even lower, i.e. €-cent 1.3 per pkm. A second
reason for this difference (apart from the higher emission factors) is the fact that passenger
diesel trains have lower load factors (number of passengers per vehicle) than electric
trains. Motorcycles cause by far the highest average external costs per pkm, which is a
result of their high accident and noise costs (plus their low occupancy rate).
The average costs of air transport are around €-cent 3.4 per pkm, which is only about 20%
higher than average rail costs. However, the result for air transport is an average, including
data for short, medium and also long haul flights to and from European airports.
The average costs between these distance classes differ from €-cent 4.3 per pkm for short
haul flights, to €-cent 2.8 per pkm for medium haul and €-cent 3.2 per pkm for long haul.
When comparing aviation and rail for the same distance classes, external costs of aviation
________________________________
45 The total climate change costs of aviation for the total EU28 have been estimated roughly the following:
the total CO 2 emission of aviation in EU28 have been 163.7 Mio. t CO 2 eq. in 2016. Under the assumption that for
aviative greenhouse gas emissions an emission weighting factor (EWF) of 2.0 is applicaple to the total CO 2
emissions and the climate cost factor of 100 €/t CO 2 eq., the total climate costs are around 32.7 bn €/a.
The WTT costs have been based on the Climate costs
(short haul flights: 4.3 €-cent/pkm) are 3 times higher than rail (high speed rail:
1.3 €-cent/pkm).
Main cost drivers for the external costs of aviation are the share of the LTO cycle of the
total flight (which is higher for short haul flights), the size and fuel use of the aircrafts and
the load factor. For road transport, the predominant cost categories are accidents and
emissions (climate change, air pollution). For air transport, climate change costs are the
main category. Please note that for aviation and maritime, the EU average costs are
averages for the selected EU-(air)ports that may not be representative for all EU (air)ports.
Figure 15 – Average external costs 2016 for EU28: passenger transport (excluding congestion)
€-ct/pkm
Habitat damage
30
Well-to-Tank
25 Noise
Climate
20
Air Pollution
15 Accidents
10
0
Car Bus/Coach MC Rail Aviation*
Passenger transport
* Aviation: average for selected EU28 airports.
For freight transport (see Figure 16), the average costs for rail transport are 1.3 €-cent per
tkm. The costs for inland waterways are slightly higher (€-cent 1.9 per tkm) than for rail.
The average costs for road freight transport (HGVs) are €-cent 3.4 per tkm (without
congestion) which is 2.6 times higher than for rail. Including congestion, the average costs
for road freight transport are 4.2 €-cent per tkm (3.2 times higher than for rail freight
transport). It may be surprising that the noise costs for rail are higher per tkm than for a
HGV. This is due to the data used from the noise maps. There are separate maps for road
and rail transport, which reveal that fewer people experience noise nuisance per vkm on
the road, compared to the number of people that experience noise nuisance per vkm on the
railway tracks. For air cargo freight transport, no external costs have been calculated due
to lack of data.
Light commercial vehicles (LCV) are used both for freight and passenger transport.
Therefore, a comparison with other passenger or freight modes cannot be easily made.
The derivation of average costs per tkm or pkm is not feasible as it is not known which part
of the transport performance (vkm) is freight or passenger transport. Therefore, the results
for LCV are presented in €-cent per vkm (see Table 69).
Figure 16 – Average external costs 2016 for EU28: freight transport (excluding congestion)
€-ct/tkm
Habitat damage
4.0
Well-to-Tank
3.5
Noise
3.0
Climate
2.5 Air Pollution
2.0 Accidents
1.5
1.0
0.5
0.0
HGV Rail IWT Maritime
Freight Transport
* Maritime: average for selected EU28 ports.
Table 69 – Average external costs 2016 for EU-28
Car Bus/Coach° MC Rail Aviation°* LCV HGV Rail IWT
Cost category €-cent/pkm €-cent/pkm €-cent/pkm €-cent/pkm €-cent/pkm €-cent/vkm €-cent/tkm €-cent/tkm €-cent/tkm
Accidents 4.5 1.0 12.7 0.5 0.02 4.1 1.3 0.1 0.1
Air Pollution 0.7 0.7 1.1 0.12 0.2 3.4 0.8 0.2 1.3
Climate 1.2 0.5 0.9 0.05 2.2 2.8 0.5 0.06 0.3
Noise 0.6 0.3 9.0 0.9 0.2 1.1 0.5 0.6 n.a.
Congestion** 4.2 0.8 0.0 0.0 0.0 11.6 0.8 0.0 0.0
Well-to-Tank 0.4 0.2 0.5 0.7 0.9 0.8 0.2 0.2 0.1
Habitat damage 0.5 0.1 0.3 0.6 0.01 0.9 0.2 0.2 0.2
Total 12.0 3.6 24.5 2.8 3.4 24.7 4.2 1.3 1.9
° Bus/coach: average for bus and coach. Aviation: average for the different distance classes.
* For aviation, the EU average costs are averages for the selected EU airports that may not be representative for all EU airports.
** Congestion in terms of delay cost.
Table 70 – Average external costs 2016 for EU28 passenger transport by cost category and transport mode
Passenger Transport
Road Rail
Pass car - Pass car - Pass car - Bus Coach MC High speed Electric pax Diesel pax
petrol diesel total Train tot
Cost category €-cent/pkm €-cent/pkm €-cent/pkm €-cent/pkm €-cent/pkm €-cent/pkm €-cent/pkm €-cent/pkm €-cent/pkm
Accidents 4.5 4.5 4.5 1.0 1.0 12.7 0.1 0.5 0.5
Air Pollution 0.3 1.2 0.7 0.8 0.7 1.1 0.0 0.01 0.80
Climate 1.2 1.1 1.2 0.5 0.4 0.9 0.0 0.0 0.3
Noise 0.5 0.6 0.6 0.4 0.2 9.0 0.3 0.8 1.4
Congestion** 4.2 4.2 4.2 0.8 0.8
Well-to-Tank 0.4 0.4 0.4 0.2 0.1 0.5 0.3 0.8 0.1
Habitat damage 0.5 0.6 0.5 0.1 0.1 0.3 0.6 0.6 0.8
Total 11.6 12.4 12.0 3.7 3.5 24.5 1.3 2.6 3.9

Table 71 – Average external costs 2016 for EU28 freight transport by cost category and transport mode
Freight Transport
Road Rail IWT
LCV-petrol LCV-diesel HGV - total Electric freight Diesel freight Inland vessel
Cost category €-cent/vkm €-cent/vkm €-cent/tkm €-cent/tkm €-cent/tkm €-cent/tkm
Accidents 4.1 4.1 1.3 0.1 0.1 0.1
Air Pollution 1.2 3.4 0.8 0.0 0.7 1.3
Climate 2.6 2.8 0.5 0.0 0.2 0.3
Noise 1.1 1.1 0.5 0.6 0.4 n/a
Congestion** 11.6 11.6 0.8
Well-to-Tank 0.8 0.8 0.2 0.2 0.1 0.1
Habitat damage 0.9 0.9 0.2 0.2 0.2 0.2
Total 22.3 24.7 4.2 1.1 1.8 1.9
Table 72 – Average external costs for selected EU28 (air)ports
Aviation passenger Maritime

Short haul Medium haul Long haul Maritime ship
Cost category €-cent/pkm €-cent/pkm €-cent/pkm €-cent/tkm
Accidents 0.04 0.01 0.00 0.001
Air Pollution 0.30 0.13 0.06 0.44
Climate 2.39 1.85 2.24 0.16
Noise 0.46 0.11 0.01 n/a
Congestion n/a n/a n/a n/a
Well-to-Tank 1.06 0.70 0.91 0.06
Habitat damage 0.03 0.01 0.00 n/a
Total 4.26 2.81 3.22 0.66
11.2.3 Average external costs per country
The results per country are presented in Table 73. It has to be noted that the accuracy level
of the disaggregated results per country is in general considerably lower than at the
aggregate EU level.
The results per country can differ for many different reasons. Some of the most important
reasons for different average costs are differences in:
— GDP per capita (PPP adjusted);
— load factors (for all transport modes);
— vehicle stock (share of efficient, low-emission vehicles);
— share of diesel and electric trains;
— electricity mix for rail;
— population density (mainly for noise and air pollution cost);
— accident risk.
Table 73 – Average external costs 2016 for EU28 by country and transport mode (excluding congestion)*
Car Bus MC Rail high Rail Rail diesel LDV HGV Rail IWT°
speed** electric
Country €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/
pkm pkm pkm pkm pkm pkm vkm tkm tkm tkm
EU28 7.8 2.9 24.5 1.3 2.6 3.9 13.1 3.4 1.3 1.9
Aggregate
Austria 12.8 3.8 69.3 2.9 8.4 16.3 4.3 3.2 2.5
Belgium 10.9 4.2 33.4 2.6 2.8 13.9 22.5 5.7 1.6 1.8
Bulgaria 5.9 2.2 43.3 5.4 5.3 8.6 2.6 1.1 0.8
Croatia 10.2 2.8 17.9 3.8 5.4 11.0 3.0 1.0 1.7
Cyprus 6.0 2.0 21.1 0.0 0.0 10.5 3.8 0.0 0.0
Czech
8.0 3.1 16.8 2.9 5.1 14.2 4.4 1.2 11.7
Republic
Denmark 5.8 2.2 21.7 2.6 2.1 9.7 4.4 0.9 0.0
Estonia 6.7 3.2 13.4 3.6 2.7 8.9 2.0 1.0 1.0
Finland 5.2 1.7 11.8 2.9 3.3 10.6 2.7 1.3 56.4
France 6.5 2.8 20.7 0.9 1.4 2.5 11.2 3.7 1.5 2.1
Germany 9.8 3.6 40.4 1.6 3.5 7.1 19.3 4.4 1.9 2.2
Greece 4.6 1.5 16.3 8.8 3.8 10.7 2.4 1.9 0.0
Hungary 9.1 2.6 21.4 4.3 5.9 18.2 3.5 0.9 2.0
Ireland 5.9 2.4 14.9 5.8 2.5 10.1 2.7 4.1 0.0
Italy 7.9 2.6 21.8 1.9 3.0 13.4 13.0 4.5 3.0 15.2
163 4.K83 - Handbook on the external costs of transport – January 2019

Car Bus MC Rail high Rail Rail diesel LDV HGV Rail IWT°
speed** electric
Country €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/ €-cent/
pkm pkm pkm pkm pkm pkm vkm tkm tkm tkm
Latvia 7.8 2.9 82.9 2.0 2.2 12.3 2.7 0.6 0.0
Lithuania 7.1 2.5 14.0 5.9 5.4 11.1 2.4 0.6 0.0
Luxembourg 11.8 4.6 30.6 5.6 26.9 15.9 3.2 3.7 3.7
Malta 8.8 2.6 36.4 0.0 0.0 19.4 2.3 0.0 0.0
Netherlands 8.6 4.3 23.1 1.7 1.6 2.2 16.2 3.3 1.1 1.7
Poland 7.8 2.7 23.1 5.0 4.1 8.6 2.5 1.0 20.1
Portugal 6.6 2.4 28.2 2.9 3.8 11.8 2.6 1.9 0.0
Romania 10.6 4.2 77.3 6.3 6.2 35.8 3.3 1.1 1.3
Slovakia 8.5 3.4 38.7 5.6 8.1 9.8 3.2 1.7 1.6
Slovenia 5.4 1.8 28.0 2.4 5.6 7.2 2.5 0.7 0.0
Spain 8.0 2.7 22.9 1.7 2.8 2.1 19.2 2.6 1.4 0.0
Sweden 5.4 1.8 27.7 1.9 6.5 9.5 2.7 1.6 0.0
United
6.2 4.0 30.0 1.0 1.5 2.3 9.9 3.0 0.8 5.5
Kingdom
Norway 4.4 2.7 21.4 2.8 2.8 10.6 2.8 2.4 0.0
Switzerland 9.5 3.7 56.1 2.5 3.4 21.8 6.7 4.0 1.9
* Results are presented for all countries and modes where transport acitivty is reported. Empty cells mean no
activity data.
** For rail high speed, activity is considered only in countries where high speed rail infrastructure exists (and not
for countries with only high speed service), in order to be consistent with the other reports (infrastructure
costs).
° For some countries, the reported transport performance of IWT is low, altough the inland waterway network is
still substantially large, leading to high average habitat damage costs (mainly Finland, but also Poland, Italy,
Czech Republic).
11.3 Comparison with previous studies
11.3.1 Previous Handbooks

Although the method used to calculate the average costs of the EU28 member states in this
study is largely the same as in the previous 2014 Handbook, a direct comparison is not easy.
For instance, the costs in the first edition of the Handbook were presented for the price
level in 2000. The second edition of the Handbook (2014 edition) presented costs at the
2010 price level. This Handbook is presented using 2016 prices. On average EU prices
increased by 35% and 8% since the first and second edition of the Handbook respectively.
Furthermore, the current emission factors come from various sources, e.g. COPERT v5 for
road transport and HBEFA 3.3 was used where COPERT had gaps. In the previous Handbook,

the emission factors came from TREMOVE, which is based on an older version of COPERT.
The damage cost rates for the pollutants taken into account (NO x, SO2, NMVOC and PM) and
the climate cost rate are also different from the last Handbook. In addition, transport
performances have changed, which also has an impact on average costs. Therefore, a
comparison between the previous Handbook and the current one is not easy.
When comparing the results, the reasons for the changes can be very different.
Accidents
In the first edition of the Handbook no average road accident costs were presented, only
marginal costs were provided. Costs were only presented at the country level and not
presented for an EU average in the first Handbook. The marginal costs were shown to range
from negative to positive, depending on the degree to which the average accident risk is
internalised by the transport users. The 2014 Handbook also only presented marginal costs,
and did not have average costs available. The marginal costs from the 2014 Handbook were
available for an EU average, and were all positive in the range of 0.1-0.3 €-cent per vkm
for passenger cars, 0.2-1.9 €-cent per vkm for motorcycles and 0.8-1.2 €-cent per vkm
for HGVs. This edition of the Handbook provides EU28 marginal costs ranging from
0.25-1.41 €-cent per pkm for passenger cars, -3.21-4.42 €-cent per pkm for motorcycles,
0.05-0.80 €-cent per pkm for buses and coaches, 0.37-0.84 €-cent per vkm for LCVs and
0.07-0.13 €-cent per tkm for HGVs. These figures reveal marginal costs are only negative for
vulnerable road users such as motorcycles.
This edition of the Handbook goes further and also presents average costs of 7.2, 13.3 and
15.5 €-cent per vkm for passenger cars, motorcycles and HGV respectively. In addition, this
Handbook also provides average costs for LCVs (4.1 €-cent per vkm) and buses and coaches
(18.9 €-cent per vkm).
Average accident costs for rail transport were estimated to be € 0.08-0.30 per vkm in the
first Handbook. In the second Handbook costs were € 0.20 per 1,000 tkm for freight trains
and € 0.60 per 1,000 pkm for passenger trains, taken from (CE Delft, INFRAS & Fraunhofer
ISI, 2011). For a comparison with those values please see Section 11.3.2. This Handbook
presents costs of € 0.18 per vkm for high speed trains (0.06 €-cent per pkm), and € 0.52 per
vkm for all passenger trains (0.47 €-cent per pkm), roughly in line with the costs from the
first Handbook.
The accident costs for aviation (expressed per LTO) ranged from € 12-209 in the first
Handbook. In the second Handbook costs were € 0.50 per 1,000 pkm for passenger aviation,
taken from (CE Delft, INFRAS & Fraunhofer ISI, 2011). For a comparison with those values
please see Section 11.3.2. Costs per LTO range from € 12-46 in this Handbook, providing a
narrower range than the values from the first Handbook. Expressed per pkm the costs
ranged from 0.001 to 0.04 €-cent per pkm, depending on whether a flight is short-,
medium- or long haul.
Fundamental differences in the cost estimates between earlier editions of the Handbook
and the current edition can largely be explained because of the higher valuation of the
fatalities, serious injuries and slight injuries (higher VSL), an overall increase in price level
(8-35% depending on which edition of the Handbook the figures are compared to), which are
to a certain extent compensated by a reduction in the accident rate. Particularly for road
accidents the latter has been noticeable, it has more than halved since the base year from
the first edition of the Handbook (2000) and declined by 19% since the base year from the
second edition of the Handbook (2010).

Air pollution
Over 370 different emission and finally cost factors are reported for passenger cars in the
actual Handbook, which is a lot more than in the last Handbook. It is therefore difficult to
compare all of them individually. The main reasons for the differences between the
versions of the Handbook is the different/new sources for emission factors. For road, with
COPERT, a completely new data source has been taken. Also for the other cost categories,
more recent and very often more differentiated sources for emission factors have been
applied. Another important reason is the fact that the present Handbook now includes all
relevant damage effects (health, crop loss, biodiversity loss, material/building damage),
whereas previous Hanbooks only focused on health effects.
The differences in the cost factors used for air pollution costs is not so large, as the new
factors are based on NEEDS (updated values). An important factor for the update of the cost
factors per air pollutant is the new Value Of Life Year lost (VOLY), which is substantially
(roughly 50%) higher than in the last Handbook. Therefore, the cost factors are substantially
higher, which often overcompensates the slightly lower emission factors in the present
Handbook.
When looking at the different vehicle types, it can be said that diesel vehicles have slightly
higher air pollution cost rates than in the last study. The cost rates for petrol vehicles are
significantly higher than in the last study. In the case of LCV, the cost rates are higher for
both petrol and diesel vehicles. The cost rates for buses and coaches in the new study are
slightly lower than in the old study. The new cost rates per tkm for HGVs are also generally
lower than in the last study.
The marginal external air pollution costs for rail are significantly lower than in the last
Handbook mainly because of the different emissions data used.
For aviation, the new Handbook shows higher average external air pollution costs than the
previous study. For inland waterways, the average air pollution costs, are difficult to
compare, due to a very different set of vessel types. In general, they are slightly lower than
in the previous study.
Climate change
For most road vehicle types the average costs of climate change are slightly higher in the
present Handbook than in the previous version. The main reason for this is the higher CO 2
cost factor (100 € per tonne of CO2 eq. in the present study, compared to € 90 per tonne of
CO2 eq. in the previous Handbook).
For aviation, the new Handbook shows significantly higher average external climate costs
than the previous study. The different aircraft types are more differentiated in the new
version of the manual, but nevertheless it is clear that the average costs are higher.
Noise
No average road noise costs were presented at the country level in the first edition of the
Handbook. The second edition of the Handbook only presents illustrative average noise
costs for German motorways. These were 0.15, 0.61, 0.18, 0.48 and 0.44-0.61 €-cent per
vkm for cars, motorcycles, LCVs, buses and HGVs respectively. This Handbook’s values for
German roads are 0.5, 3.6, 0.7, 3.6 and 2.6-4.5 €-cent per vkm respectively. These values
are roughly a factor of 3-5 higher, a difference which can largely be explained because of a

better correction for missing data, and new insights from the literature (e.g. a higher
valuation of noise annoyance and the inclusion of new health endpoints). This higher
valuation of noise annoyance is in part due to the fact that annoyance is no longer valued as
being linear, rather, higher valuation is given at higher noise levels. New health endpoints
that are incorporated in this Handbook are strokes and dementia.
No average rail noise costs were presented in either the first or second edition of the
Handbook. This is a novel aspect in this Handbook.
No average noise costs were presented for aviation in the first edition of the Handbook,
although both this edition and the second edition presented average noise costs per LTO
at the airport level. A comparison can be drawn between the airports that overlap.
For instance, at London Heathrow the average noise costs per LTO were € 652 per LTO in
the second edition of the Handbook. In this edition, the costs are € 1,549 per LTO.
Similar increases in costs can be witnessed for a number of the other major airports in
Europe, e.g. Amsterdam (from € 39 to € 118), Paris Charles de Gaulle (from € 111 to € 256)
and Frankfurt (from € 180 to € 376). In general, the trend holds for all airports in Europe,
and can be explained byby better corrections for missing data, and a higher valuation of
noise annoyance and health.
Congestion
In the previous Handbook road congestion costs were firstly estimated for UK using the
FORGE model. This model was used to quantify road congestion by estimating speed-flow
relations of UK road networks, specific to areas and road types. The FORGE model
distinguished between motorways and other roads and the results for ‘conurbations’ (other
large cities, except for London) were used as a proxy for typical metropolitan areas.
Given the level of congestion in different areas and road types, marginals cost of congestion
were estimated for the UK. Values for EU Member States were then derived by means of
value transfer.
In this edition of the Handbook, the methodology for congestion makes reference to
different conditions (i.e., road types and level of congestion) and the level of congestion is
not based on modelled results for one country only, but on observed measures for several
cities and the whole European main road network. Furthermore, a different source has been
used as reference for the Value of Time.
Importantly, the methodology adopted in the previous Handbook did not lead to the
estimation of total and average congestion costs. The social marginal congestion costs
estimated in this Handbook are also different from those of the previous version. The values
estimated in this version are for segment AC in Figure 7, according to the definition in (Li,
et al., 2016), those estimated in the previous one (i.e., the Efficient Marginal Congestion
Cost, EMCC) are for segment BD. Moreover, the methodolgy developed within this Handbook
calculates the incurred social marginal congestion cost and then estimates the generated
social margial congestion cost through a simplified approach to allow for comparability with
the other external costs. Figures and trends on congestion costs are compared with respect
to the 2011 version of the Handbook, as reported in Section 11.3.2.
In comparison with the Study on urban mobility prepared for DG Move Branningan et al.
(2016) this Handbook applies in principle the same methodological approach.
Nevertheless, different data sources are used for some input values and/or the same
information in a more recent release. In particular, the Value of Time plays a key role for

the magnitude of the estimation of total costs in terms of delay and deadweight loss; i.e.
the value of Time used as reference in this Handbook is larger than the previous
datasource, therefore it contributes to estimate an increased value of total congestion costs
(i.e. about 50% for passenger cars) in terms of delay cost and deadweight loss with respect
to the study mentioned above.
Well-to-tank emissions
The data base for well-to-tank emissions are completely new. One main reason for
differences to the previous Handbook is the slightly higher CO2 cost factor. Also, the cost
factors for air pollution are higher (mainly due to a higher VOLY).
Looking at the different vehicle types, the cost factors for passenger cars and LCV are very
similar to the previous Handbook. For trucks, the costs are higher in the present Handbook.
So, for road transport the higher cost factors are compensated by the lower emission
factors.
For rail, aviation and inland waterways the costs of well-to-tank emissions in the present
Handbook are considerably higher than in the previous one. Here, both emission and (more
important) cost factors are higher in the present Handbook.
11.3.2 Overview study ‘External costs of transport in Europe’ (2011)

The (CE Delft, INFRAS & Fraunhofer ISI, 2011) study was presented for the year 2008.
Accidents
The total costs of accidents presented in this Handbook are very much in line with those
presented in the (CE Delft, INFRAS & Fraunhofer ISI, 2011) study. For motorcycles total
accident costs fell from € 22 billion to € 21 billion. A similar decline was visible in the total
accident costs for buses and coaches (from € 6.8 to € 5.3 billion). For the other road
transport vehicle types, there was an increase in total accident costs. For HGVs the costs
increased from € 20 to € 23 billion and for LCVs the costs increased from € 19 to
€ 20 billion. The biggest increase was visible for the total costs of accidents for passenger
cars. Costs increased from € 157 billion to € 210 billion.
In general, the trends observed for total road accident costs were echoed by the average
road accident costs. A decline was visible for motorcycles (from € 157 to € 127 per
1,000 pkm) and buses and coaches (from € 12.3 to € 9.8 per 1,000 pkm). The largest
increase was visible for passenger cars, where average costs increased from € 32 to € 45 per
1,000 pkm. The average costs for HGVs also increased from € 10 to € 13 per 1,000 tkm.
There are three major factors that explain the differences in costs between this Handbook
and the (CE Delft, INFRAS & Fraunhofer ISI, 2011) study. Firstly, a higher valuation was used
for the cost of an injury or fatality. The largest parameter is the VSL, which is used as an
input for the human costs. Because of the higher valuation of the VSL, the costs of a
fatality, serious and slight injury were increased by 98%, 130% and 133% respectively,
compared to the valuation in the (CE Delft, INFRAS & Fraunhofer ISI, 2011) study.
This effect should lead to higher costs in this Handbook in comparison with the (CE Delft,
INFRAS & Fraunhofer ISI, 2011) study. Secondly, the studies look at a different price level.
The inflation that took place between 2008 and 2016 results in 2016 prices being on average
12% higher than in 2008. This effect should also result in costs in this Handbook being higher

than in the (CE Delft, INFRAS & Fraunhofer ISI, 2011) study. Thirdly, statistics show a
decline in accidents in the period between 2008 and 2016 (for instance, the number of road
fatalities has declined by 35% between 2008 and 2016).
For railway transport, the total costs of passenger rail and freight rail transport were
€ 0.2 billion and € 0.07 billion respectively in CE Delft, Infras & Fraunhofer ISI, (2011).
In this current Handbook, the total costs of passenger rail transport are € 2.1 billion and
€ 0.3 billion for freight rail transport. Similarly, average costs have also increased, from
€ 0.6 to € 4.7 per 1,000 pkm for passenger rail, and from € 0.2 to € 0.7 per 1,000 tkm for
freight rail. This is a significant increase for both average and total costs which cannot
solely be explained by the increased valuation and inflation (mentioned above). In this
Handbook we use data from the European Railway Agency, from which we conclude that
there are 737 fatalities and 556 serious injuries in the EU28. The CE Delft, Infras &
Fraunhofer ISI (2011) used data from the UIC, where there were 114 fatalities and
612 serious injuries. These differences in input data explain the increased costs.
For aviation the total accident costs mentioned in CE Delft, Infras & Fraunhofer ISI (2011)
amounted to € 223 million, and the average accident costs was € 0.5 per 1,000 pkm. In this
Handbook the total costs of aviation amount to € 75 million. The difference in these
numbers can be explained by the change in scope of this study. This Handbook only looks at
33 European airports, whereas CE Delft, Infras & Fraunhofer ISI (2011) look at all airports in
the EU. For average costs the values in this Handbook are in line with CE Delft, Infras &
Fraunhofer ISI (2011) at € 0.40, € 0.10 and € 0.01 per 1,000 pkm for short, medium and
long-haul respectively.
For the two other transport modes (inland waterway transport and maritime transport)
accident costs were not calculated in the CE Delft, Infras & Fraunhofer ISI (2011) study,
which implies no comparison can be carried out.
Air pollution
The total air pollution costs for road, rail and air transport are very similar than in the last
UIC study from 2011. For LCV, the costs are substantially higher, mainly due to better
transport activity data and more recent emission factors. For inland waterways, the total
air pollutant costs are almost double as high resulting from more recent transport acticity
data and emission factors.
The average air pollution costs are on a similar level than in the UIC study from 2011.
The main reasons for differences are the completely new set of emission factors, the new
Value Of Life Year lost (VOLY), which is substantially (roughly 50%) higher than in the UIC
study.
Climate change
The UIC study analysed the external climate costs using a low and a high scenario.
The difference between the two scenarios was the climate cost factor which at that time
ranged between € 42 and € 146. In the current study, a climate cost factor of € 100 is used.
The two studies cannot be compared with each other on this point. In general however, it
can be said that today’s average external climate costs lie between the high and low
scenarios of the previous UIC study.

Noise
The total costs of noise presented in this Handbook are substantially higher than those
presented in the CE Delft, Infras & Fraunhofer ISI (2011) study. Noise costs increased most
for motorcycles and mopeds (a factor 7, from € 2.1 billion to € 14.8 billion) and for
passenger cars (a factor 3, from € 8.2 billion to € 26.2 billion). Total costs more than
doubled for LCVs (from € 2.1 billion to € 5.4 billion) and HGVs (from € 3.5 million to
€ 9.1 billion). The total noise costs for buses and coaches increased from € 0.9 billion to
€ 1.6 billion.
In general, the trends observed for total road noise costs were echoed by the average road
noise costs. Average costs increased by a factor 6 (from € 14.4 to € 89.7 per 1,000 pkm) and
4 (from € 1.7 to € 5.5 per 1,000 pkm) for motorcycles and passenger cars respectively.
Increases in average costs of a factor of 3 and 2 were observed for HGVs and buses/coaches
respectively, mirroring the increases in total costs.
There are three major factors, apart from the different price levels (price year 2016 vs.
price year 2008), that explain the differences in costs between this Handbook and the
CE Delft, Infras & Fraunhofer ISI (2011) study. Firstly, a higher valuation was used for the
cost of noise, in particular for the noise annoyance. New in this Handbook is the increased
valuation with higher noise levels, this results in substantially higher costs per person
annoyed in each noise bin. This explains part of the higher noise costs in this Handbook
compared to CE Delft, Infras & Fraunhofer ISI (2011) study. Secondly, this study uses the
noise maps from 2012, which are more recent than the ones used in CE Delft, Infras &
Fraunhofer ISI (2011). It is likely that the increased urbanisation that took place over this
period implies that more people are now exposed to (higher) noise levels, which results in
higher noise costs in this Handbook compared to CE Delft, Infras & Fraunhofer ISI (2011).
Thirdly, the noise maps are incomplete as there are agglomerations and countries with a
reporting obligation that may not provide data. Furthermore, only some agglomerations
have a reporting obligation, for agglomerations without a reporting obligation noise costs
are not taken into account. Previous studies have attempted to correct the data for this
effect, but this edition of the Handbook carries out a better and more detailed correction.
This is a further reason for the noise costs being higher in this Handbook than in CE Delft,
Infras & Fraunhofer ISI (2011).
For railway noise the total costs of passenger rail and freight rail transport were
€ 0.5 billion each in CE Delft, Infras & Fraunhofer ISI (2011). In this current Handbook, the
total noise costs of passenger rail transport are € 3.9 billion and € 2.5 billion for freight rail
transport. Similarly, average costs have also increased, from € 1.2 per 1,000 pkm to € 3.2,
€ 8.0 and € 13.8 per 1,000 pkm for high speed, electric and diesel passenger trains
respectively. For freight rail costs increased form € 1.0 to € 6.5 and € 4.5 per 1,000 tkm for
electric and diesel freight trains respectively. This increase is roughly in line with the
increase in total costs. The costs have thus increased by a factor of 6 to 9 in between the
studies, which can be explained by the fact that the previous study incorporated a rail
bonus, whereas the current Handbook does not. As a result, higher rail costs prevail.
For aviation the total noise costs mentioned in CE Delft, Infras & Fraunhofer ISI (2011)
amounted to € 0.5 billion, and the average noise costs were € 1.0 per 1,000 pkm. In this
Handbook the total noise costs of aviation amount to € 0.84 billion. The increase in total
costs is relatively small compared to the other modes. This can be explained by the change
in scope of this study. This Handbook only looks at total noise from 33 European airports,
whereas (CE Delft, Infras & Fraunhofer ISI (2011) look at all airports in the EU. For average
costs the values in this Handbook are differentiated to distance, with short, medium and

long haul noise costs being € 4.6, € 1.10 and € 0.14 per 1,000 pkm respectively.
This increase can be explained by the increase in the valuation of noise, and the improved
corrections made to the noise maps.
For the two other transport modes (inland waterway transport and maritime transport) no
noise costs were calculated in either study, as the noise exposure is considered negligible or
non-existent as it usually takes place in sparsely populated areas. Furthermore, data on
noise exposure from these modes is not available.
Congestion
The methodology and the measures of road congestion costs presented in this Handbook are
similar to those applied in (CE Delft, Infras & Fraunhofer ISI (2011).
Furthermore, also the total costs of congestion presented in this Handbook are largely in
line with the range of costs presented in the 2011 version of the Handbook (CE Delft, Infras
and Fraunhofer ISI, 2011). For passenger cars both delay cost and deadweight loss are about
20%-25% higher than the maximum (i.e., € 196 billion versus € 161 billion for delay cost and
€ 33.5 billion versus € 26 billion for DWL, respectively). Total costs are about 45% lower
than the minimum value for HGVs (€ 14.6 million in this version of the Handbook versus
€ 26.7 billion for delay cost and € 2.5 billion in this version of the Handbook versus € 4.3
billion for DWL), while for LCVs the total cost estimated in this Handbook are doubled with
respect to the maximum reported in the 2011 version (€ 55.5 billion versus € 27.6 billion for
delay cost and € 9.4 billion versus € 4.5 billion for DWL). The use of the different Value of
Time can explain part of the different estimations.
11.4 Recommendations for further assessment

This Handbook provides a state of the art overview on the external costs of transport.
However there are various topics that require further research. For each cost category, the
main uncertainties have been discussed in the sections on the robustness of the results.
All these can be interpreted as issues for further research. In addition there are some more
general issues for further research. Overall the main issues for further research are:
— There appear to be significant inconsistencies between the available data sets on
emissions factors (e.g. from COPERT), transport performance data (from Eurostat) and
national emission reporting. This requires further harmonisation.
— Data on congestion and scarcity for non-road transport modes.
— Data on other external costs (e.g. nuclear risks).
— Level of internalisation by insurances for accident costs and immaterial damages for
slight and serious injuries.
— Valuation of accidents for new categorisation of injuries (six categories of the MAIS
scale instead of the ‘serious’ and ‘slight’ injuries).
— CO2 avoidance cost for meeting the Paris agreement.
— The noise maps do not cover all infrastructure and locations.
— The data basis for the marginal accident and marginal noise costs.
— The air pollutant emission factors: to what extent do they fully reflect real world
emissions, including potential degradation or removal of emission reduction equipment
like particulate filters.

References
Ackerman, F. & C. M., 2012. Climate damages in the FUND model: A disaggregated analysis.
Ecological Economics, 77(May), pp. 219-224.
Ackerman, F. & Stanton, E. A., 2012. Climate risks and carbon prices: Revising the social
cost of carbon. Economics.
Adelphi et al., 2015. A new climate for peace: Taking action on climate and fragility risks,
Berlin: Adelphi.
AEA, 2005. Methodology for the Cost-Benefit analysis for CAFE. Vol. 2: Health Impact
Assessment, Didcot: AEA, Technology Environment (AEA).
Aldy, J. E. & Viscusi, W. K., 2004. Age variation in workers' value of a statistical life.
Harvard Law School - Discussion Paper 468.
Alfaro, J., Chapuis, M. & Fabre, F., 1994. Socio-economic costs of road accidents : Final
report of action COSTS 313, Brussels: Commission of the European Community (EC).
Anastasopoulos et al., 2011. The economic value of quiet areas - Final report, London:
URS/Scott Wilson.
Andersson et al., 2010. Property prices and exposure to multiple noise sources: Hedonic
regression with road and railway noise. Environmental and resource economics, pp. 73-89.
Andersson et al., 2013. Benefit measures for noise abatement: Calculations for road and rail
traffic noise. European Transport Research Review, pp. 135-148.
Andersson, H. e. & Treich, N., 2009. The Value of a Statistical Life, Toulouse: LERNA,
University of Toulouse.
Andersson, H., Schlärdh, J. & Ögren, M., 2015. Traffic noise effects of property prices:
Hedonic estimates based on multiple noise indicators, Stockholm: Center for Transport
Studies.
Anthoff, D., 2007. Report on marginal external damage costs inventory of greenhouse gas
emissions, Hamburg: Hamburg University.
Anthoff, D., Rose, S., Tol, R. S. & Waldhoff, S., 2011. The time evolution of the social cost
of carbon : An application of FUND. Economics, Issue Online, pp. 1-22.
Anthoff, D. & Tol, R. S., 2007. On international equity weights and national decision
making on climate change, s.l.: The Fondazione Eni Enrico Mattei Note di Lavoro (FEEM).
ARE, 2018. Externe Kosten und Nutzen des Verkehrs in der Schweiz: Strassen-, Schienen-,
Luft- und Schiffsverkehr 2015, Bern: Bundesamt für Raumentwicklung (ARE).
ARUP, 2015. Provision of market research for value of travel time savings and reliability,
Non-Technical Summary Report, London: UK Department for Transport.
Azar, C. & Johansson, D. J. A., 2012. Valuing the non-CO2 climate impacts of aviation.
Climatic Change, 111(3-4), pp. 559-579.
B,S,S. Volkswirtschaftliche Beratung AG, 2015. Monetarisierung des statistischen Lebens im
Strassenverkehr, Basel: B,S,S. Volkswirtschaftliche Beratung AG.
Babisch, W., 2014. Cardiovascular burden of diseae from environmental noise evidence,
uncertainties and public health implications, s.l.: s.n.
Baker, E. & Barron, R., 2013. Technological change and the marginal cost of abatement.
Encyclopedia of Energy, Natural Resource and Environmental Economics, pp. 117-122.
Barker, T. & Scrieciu, S. S., 2010. Modeling low climate stabilization with E3MG: Towards a
'New Economics' approach to simulating energy-envrironment-economy system dynamics.
Energy Journal, 31(Special issue), pp. 137-164.
Basner, M. & McGuire, S., 2018. WHO Environmental Noise Guidelines for the European
Region: A systematic review on environmental noise and effects on sleep. International
Journal of Environmental Research and Public Health, 15(3).
Bateman et al., 2002. Economic valuation with stated preference techniques: A manual,
s.l.: Edward Elgar.

Bateman, I., Day, B., Lake, I. & Lovett, A., 2001. The effect of road traffic on residential
property values: A literature review and hedonic pricing study, Edinburgh: Scottish
Executive Development Department.
Beimer, W. & Maennig, W., 2017. Noise effects and real estate prices: A simultaneous
analysis of different noise sources. Transportation Research Part D, pp. 282-286.
Bennet, C., 2015. Failure to act on climate change means an even bigger refugee crisis. The
Guardian, 7 September.
Beresniak, A. et al., 2015. Validation of the Underlying Assumptions of the Quality-Adjusted
Life Years Outcome: Results from the ECHOUTCOME European Project. PharmacoEconomics,
33(1), pp. 61-69.
BfU, 2015. Nichtberufsunfälle in der Schweiz, Bern: Beratungsstelle für Unfallverhütung
(BfU.
Bigano, A., G.L. Cassinelli, F. & Navrud, S., 2009. 2009. The External Cost of European
Crude Oil Imports. Fondazione Eni E. Mattei (FEEM) Working Papers n.13 .
Bijlenga, D., Bonsel, G. J. & Birnie, E., 2011. Eliciting willingness to pay in obstetrics:
Comparing a direct and an indirect valuation method for complex health outcomes. Health
Economics, 20(11), pp. 1392-1406.
Blincoe et al., 2014. The economic and societal impact of motor vehicle crashes 2010,
Washington D.C.: National Highway Administration.
Blincoe, L., 2015. The economic and societal impact of motor vehicle crashes 2010,
Washington, DC: National Highway Traffic Safety Administration.
Bobinac, A., Van Exel , N., Rutten, F. & Brouwer, W., 2013. Valuing QALY gains by applying
a societal perspective. Health Economics Letters, 22(10), pp. 1272-1281.
Bodin et al., 2015. Annoyance, sleep and concentration problems due to combined traffic
noise and the benefit of quiet side. International Journal of Environmental Research and
Public Health.
Börjesson , M., Fung, C. & Proost, S., 2017. Optimal prices and frequencies for buses in
Stockholm,. Economics of Transportation, 9(C), pp. 20-36.
Botzen, W. & Van den Bergh, J., 2012. How sensitive is Nordhaus to Weitzman? Climate
policy in DICE with an alternative damage function. Economic Letters, 117(1), pp. 372-374.
Brandt, S. & Maennig, W., 2011. Road noise exposure and residential property prices:
Evidence from Hamburg. Transportation Research Part D.
Brink, C. & Grinsven, H. v., 2011. Costs and benefits of nitrogen in the environment. In: The
European Nitrogen Assessment : Sources, effects and perspectives. Cambridge: Cambridge
University Press, p. Chapter 22 .
Bristow, A. L., Wardman, M. & Chintakayala, V. P. K., 2015. International meta-analysis of
stated preference studies of transportation noise nuisance. Transportation, 42(1), pp. 71-
100.
Brown, L. & van Kamp, I., 2017. WHO Environmental Noise Guidelines for the European
Region: A systematic review of transport noise interventions and their impacts on health.
International Journal of Environmental Research and Public Health, 14(8).
Bruitparif, ORS Ile-de-France & WHO, 2013. Health impact of noise in the Paris
agglomeration: Assessment of healthy life years lost, s.l.: Bruitparif.
Burke et al., 2016. Opportunities for advances in climate change economics. Science, pp.
292-293.
Carlsson, F., Daruvala , D. & Jaldell, H., 2010. Preferences for lives, injuries and age: A
stated preference survey. Accident Analysis and Prevention, 42(6), pp. 1814-1821.
CE Delft & VU Amsterdam, 2004. The Price of transport : Overview of the social costs of
transport, 2004: CE Delft.
CE Delft & VU, 2014. Externe en infrastructuur kosten van verkeer, Delft: CE Delft.
CE Delft et al., Forthcoming. Transport infrastructure expenditures and costs, Delft: CE
Delft.

CE Delft, INFRAS & Fraunhofer ISI, 2011. External costs of transport in Europe, Delft: CE
Delft.
CE Delft, 2003. External Costs of Aviation, Berlin: German Environment Agency
(Umweltbundesamt).
CE Delft, 2010. Handboek Schaduwprijzen - Waardering en weging van emissies en
milieueffecten, 2010: CE Delft.
CE Delft, 2016. The cost of road traffic accidents in the Netherlands, Delft: CE Delft.
CE Delft, 2017. Handboek Milieuprijzen 2017, Delft: CE Delft.
CE Delft, 2018. Environmental Prices Handbook 2017 : Methods and numbers for valuation
of environmental impacts, Delft: CE Delft.
CEREMA, 2018. External costs of Transport in Alpine Corridors. Ongoing study., s.l.: s.n.
Chanel, O. & Luchini, S., 2014. Monetary values for risk of death from air pollution exposure
: A context dependent scenario with a control for intra-familian altruism. Journal of
Environment Economics and Policy, 3(1), pp. 67-91.
Christidis, P. & Brons, M., 2016. External costs of freight transport in European Union
Member States, s.l.: European Commission, Joint Research Centre (JRC).
Christidis, P. & Ibáñez , N., 2012. Measuring road congestion, Sevilla: JRC, Institute for
Prospective Technological Studies.
Clark, C. & Paunovic, K., 2018. WHO Environmental Noise Guidelines for the European
Region: A systematic review on environmental noise and cognition. International Journal of
Environmental Research and Public Health, 7(15).
Comité National Routier, 2016. Comparative study of employment and pay conditions of
international lorry drivers in Europe, s.l.: Comité National Routier.
Commissariat général à la stratégie et à la prospective, 2013. Éléments pour une révision
de la valeur de la vie humaine, Paris: Commissariat général à la stratégie et à la
prospective.
Concawe et al., 2012. Evaluating the Value of a Life Year (VOLY), Brussels: Concawe.
CONCAWE, 2006. Analysis of the CAFE cost benefit analysis, Brussels: CONCAWE.
Cook, A. & Tanner, G., 2015. European airline delay cost reference values, Updated and
extended values, Version 4.1, London: University of Westminster.
Corporate Knights, 2014. 2014 Green Provinces & States : Introduction and results for the
2014 Green Provinces and States Report Card. [Online]
Available at: http://www.corporateknights.com/reports/sustainable-provinces-
states/green-crown-goes-14017173/
[Accessed 2018].
CPB; PBL, 2016. WLO-klimaatscenario's en de waardering van CO2-uitstoot in MKBA's, Den
Haag: CPB ;PBL.
Czajkowski, M., Ahtiainen, H., Artell, J. & Meyerhoff, J., 2017. Choosing a functional form
for an international benefit transfer : evidence from a nine-country valuation experiment.
Ecological Economics, 134(April), pp. 104-113.
Czajkowski, M. & Ščasný, M., 2010. Study on benefit transfer in an international setting.
How to improve welfare estimates in the case of the countries' income heterogeneity?.
Ecological Economics, 69(12), pp. 2409-2416.
D., A., 2015. Pricing Freight Transport to Account for External Costs, Working Paper Series,
s.l.: Congressional Budget Office, .
D.P., V. V., M., I., d. E. M. & Stehfest E., v. V. J., 2010. Low stabilzation scenarios and
implications for major world regions from an integrated assessment perspective. Energy
Journal, 31(Special issue 1), pp. 165-192.
Day, B., Bateman, I. & Lake, I., 2007. Beyond implicit prices: Recovering theoretically
consistent and transferable values for noise avoidance from a hedonic property price model.
Environmental and resource economics, pp. 211-232.

De Blaeij, A., Florax, R., Rietveld, P. & Verhoef, E., 2003. The value of a statistical life in
road safety : Stated preference methodologies and empirical estimates for the Netherlands.
Accident Analysis and Prevention, Volume 35, pp. 973-986.
De Palma, A., Lindsey, R. & Monchambert, G., 2017. The economics of crowding in rail
transit. Journal of Urban Economics, 101 (2017) 106–122., 101(C), pp. 106-122.
DECC, 2009. Carbon valuation in UK policy appraisal: A revised approach, London:
Department of Energy & Climate Change (DECC).
DECC, 2015. Updated short-term traded carbon values used for UK policy appraisal,
London: Department of Energy & Climate Change (DECC).
Defra, 2005. The social cost of carbon review : Methodological approaches for using SCC
estimates in policy assessment, London: Department for Environment, Food and Rural
Affairs.
Defra, 2007. An Economic Analysis to inform the Air Quality Strategy : Ipdated Third
Report of the Interdepartmental Group on Costs and Benefits, London: Department of
Environment, Food and Rural Affairs (Defra).
Defra, 2014. Environmental noise : Valuing impacts on: sleep disturbance, annoyance,
hypertension, productivity and quiet, London: Defra.
Dekkers, J. & van der Straaten, W., 2008. Monetary valuation of aircraft noise: A hedonic
analysis around Amsterdam airport, s.l.: s.n.
Dekker, T., Brouwer, R., Hofkes, M. & Moeltner, K., 2011. The effect of risk context on the
Value of a Statistical Life: A Bayesian meta-model. Environmental Resourse Economics,
49(4), pp. 597-624.
Desaigues, B. et al., 2011. Economic valuation of air pollution mortality: A 9-country
contingent valuation survey of the value of a life year (VOLY). Ecological Indicators, 11(3),
pp. 902-910.
DfT, 2012. A valuation of road accidents and casualties in Great Britain : Methodology
note, London: Department for Transport (DfT).
Doll, C., 2002. Unification of Accounts and Marginal Costs for Transport Efficiency (UNITE)
Deliverable 7: User Cost and Benefit Cast Studies. ., Leeds (UK): ITS, University of Leeds.
Doucouliagos, H., Stanley, T. D. & Viscusi, W. K., 2014. Publication selection and the
income elasticity of the value of a statistical life. Journal of Health Economics, pp. 67-76.
Dreger, S., Meyer, N., Fromme, H. & Bolte, G., 2015. Environmental noise and incident
mental health problems: A propsective cohort study among school children in Germany.
Environmental Research, 143(PtA), pp. 49-54.
Duarte, C. & Tamez, C., 2009. Does noise have a stationary impact on residential values?.
Journal of European Real Estate Research, pp. 259-279.
Dunkerley, F., Rohr, C. & Daly, A., 2014. Road traffic demand elasticities : A rapid
evidence assessment, Cambridge (UK): RAND Europe.
Dutilleux, G., 2012. Anthropogenic outdoor sound and wildlife: It's not just bioacoustics!.
Proceedings Acoustics, pp. 2301-2306.
EC, 2002. Directive 2002/49/EC of the EP. and of the Council of 25 June 2002 relating to
the assessment and management of environmental noise-Declaration by the Commission in
the Conciliation Committee(...)relating to the aassessment and management of
environment. Official Journal of the European Communities, L189(18.7.2002), pp. 12-25.
EC, 2005. ExternE: Externalities of Energy Methodology 2005 Update, Luxembourg:
European Commission (EC).
EC, 2009a. Impact Assessmnet Guidelines, Brussels: European Commission.
EC, 2009b. Part III: Annexes to impact assessment guidelines, Brussels: European
Commission.
EC, 2016. The EU Emissions Trading Scheme, Brussels: European Commission.
ECMT, 1998. Efficient transport for Europe : Policies for Internalisation of External Costs,
Paris: OECD, European Conference of Ministers of Transport (ECMT).
Ecofys, 2014. Subsidies and costs of EU energy : Final, Annex 3, s.l.: Ecofys.

Ecoplan & Infras, 2014. Externe Effekte des Verkehrs - Monetarisierung von Umwelt-,
Unfall- und Gesundheidseffekten. Studie im Auftrag des Bundesamtes für
Raumentwicklung, Bern ; Zürich: Ecoplan.
Ecoplan, 2002. Unfallkosten im Strassen- und Schienenverkehr der Schweiz 1998, Altdorf:
Ecoplan.
Ecoplan, 2016. Empfehlungen zur Festlegung der Zahlungsbereitschaft für die
Verminderung des Unfall- und Gesundheitsriskos (value of statistical life), Bern: Ecoplan.
Edenhofer, O. et al., 2010. The economics of low stabilization : Model comparison of
mitigation strategies and costs. Energy Journal, 32(Special Issue 1), pp. 11-48.
EEA, 2010. Good practice guide on noise exposure and potential health effects,
Copenhagen: European Environmental Agency.
EEA, 2014. Noise in Europe 2014, Copenhagen: European Environmental Agency.
EEA, 2018. file EEA: END_DF4_DF8_Results_2012_170331. [Online]
Available at: https://www.eea.europa.eu/data-and-maps/data/data-on-noise-exposure-4
Elmenhorst et al. , 2014. Nocturnal air, road, and rail traffic noise and daytime cognitive
performance and annoyance. Journal of the Accoustical Society of America.
Elmenhorst et al., 2012. Examining nocturnal railway noise and aircraft noise in the field:
Sleep, psychomotor performance and annoyance. The Science of the Total Environment.
ERSO, 2006. Cost-benefit analysis, Brussels: European Road Safety Observatory (ERSO).
Eurocontrol, 2017. CODA Digest 2016, All-Causes Delay and Cancellations to Air Transport
in Europe – 2016, Brussels: Eurocontrol.
European Commission, 1994. Cost 313 Socio-economic costs of road accidents: Final report,
Luxembourg: European Commission .
European Commission, 2002. Position paper on dose response relationships between
transportation noise and annoyance, Luxembourg: European Commission.
Eurostat, 2018. Distribution of population by degree of urbanisation, dwelling type and
income group - EU-SILC survey. [Online].
EUSALP, 2017. External costs in mountain areas. on behalf of EUSALP (EU Strategy for the
Alpine Region), Zurich: INFRAS and Herry Consult.
EXIOPOL , 2011. A new environmental accounting framework using externality data and
input-output tools for policy analysis. Cluster II – Externality Estimates: Extension and
Development (FEEM).. [Online]
Available at: Online: http://www.feem-project.net/exiopol/adkajdkaj/EXIOPOL
[Accessed 2018].
ExternE, 2005. Externalities of Energy, Methodology 2005 update, Luxembourg: European
Commission.
EY ; BG ; TNO ; Fraunhofer; Mcrit; University of Craiova, ongoing. Assessment of potential
of maritime and inland ports and inland waterways and of related policy measures, Lot 1:
Support measures for the implementation of the TEN-T Core network related to sea ports,
inland ports and inland waterway transport, Paris: s.n.
Eyring, V. et al., 2009. Transport impacts on atmosphere and climate: Shipping.
Atmospheric Environment, 44(37), pp. 4735-4771.
Farrow, S. & Larson, D., 2012. News and Social Cost: The Case of Oil Spills and Distant
Viewers. Journal of Benefit-Cost Analysis, De Gruyter,, 3(4), pp. 1-24.
Fiorello, D. & Zani, L., 2016. EU survey on issues related to transport and mobility, Seville:
Joint Research Centre (JRC), Institute for Prospective Technological Studies.
Forster, P. M. d. F., Shine, K. P. & Stuber, N., 2006. It is premature to include non-CO2
effects of aviation in emission trading schemes. Atmospheric Environment, 40(6), pp. 1117-
1121.
Foster, P., Shine, K. P. & Stuber, N., 2007. Corrigendum to 'It is premature to include non-
CO2 effects of aviation in emission trading schemes'. Atmospheric Environment, 41(18), p.
3941.

Fraunhofer ISI & Infras, 2018. Methodenkonvention 3.0 - Sachverständigengutachten AP 1:
Unweltkosten durch Lärm, Karlsruhe: Fraunhofer ISI.
Fridstrom, L., 2011. A framework for assessing the marginal external cost of road use and
its implications for insurance ratemaking, Paris: OECD.
Friedrich & Quinet, 2011. External costs of transport in Europe: A Handbook of Transport
Economics, Celtenham: Edward Elgar Publishing.
Friedrich, R., 2008. Note on the choice of values of marginal external costs of greenhouse
gas emissions: Internal discussion paper in project NEEDS, Stuttgart: University of Stuttgart.
Gallash et al., 2016. Road and rail traffic noise induce comparable extra-aural effects as
revealed during a short-term memory test. Noise Health, pp. 206-213.
Gambhir et al., 2015. The contribution of non-CO2 greenhouse gas mitigation to achieving
long-term temperature goals: AVOID 2 WCP2B Report, London: Grantham Institute.
Getzner, M. & Zak, D., 2012. Health impacts of noise pollution around airports: Economic
valuation and transferability. In: Environmental health - Emerging issues and practice.
s.l.:InTech, pp. 247-272.
Gillingham, K. et al., 2015. Modelling uncertainty in climate change: A multi-model
comparison, Cambridge, MA: National Bureau of Economic Research.
Global Humanitarian Forum, 2009. Human Impact Report: Climate Change, Geneva: Global
Humanitarian Forum.
Goodbody, 2002. Cost benefit parameters and application rules for transport project
appraisal, Dublin (IE): Goodbody Economic Consultants.
Goodwin, P., 2004. The economic costs of road traffic congestion, London: University
College London (UCL), The Rail Freight Group.
GRACE, 2006. Generalisation of Research on Accounts and Cost Estimation (GRACE) :
Deliverable 2: Marginal cost case studies for road and rail transport, ITS, Leeds, Leeds: ITS.
Grant-Muller, S. & Laird, J., 2007. Costs of Congestion : Literature Based Review of
Methodologies and Analytical Approaches, Edinburgh: Scottish Executive.
Greiser & Greiser, 2009. Risikofaktor nächtliger Fluglärm - Abschlussbericht über eine Fall-
Kontroll-Studie zu kardiovaskulären und Psychischen Erkrankungen im Umfeld des
Flughafens Köln-Bonn, Musweiler: Epi.Consult GmbH.
Greiser, E. & Greiser, C., 2015. Umgebungslärm und Gesundheit am Beispiel Bremen,
Musweiler: Epi.Consult GmbH.
Guski et al., 2017. WHO environmental noise guidelines for the European region: A
systematic review on environmental noise and annoyance. International Journal of
Environmental Research and Public Health.
Guski, R. & Schreckenberg, D., 2015. NORAH Noise-related annoyance, cognition and
health. Verkehrslärmwirkungen im Flughafenumfeld. Band 7: Gesamtbetrachtung des
Forschungsprojekts NORAH, Kelsterbach: Umwelthaus gGmbH.
Hayman, B., Dogliani, M., Kvale, I. & Fet, A., 2000. Technologies for reduced environmental
impact from ships: Ship building, maintenance and dismantling aspects. TRESHIP. . In:
Proceedings, ENSUS 2000 Marine Science and Technology for Environmental Sustainability
Department of Marine Technology and Sc. Newcastle upon Tyn: University of Newcastle
upon Tyn.
Health & Safety Laboratory, 2011. Quantifying the links between environmental noise
related hypertnesion and health effects, Buxton: Health & Safety Laboratory.
HEATCO, 2006. Developing Harmonised European Approaches for Transport Costing and
Project Assessment (HEATCO). Deliverable D5: Proposal for harmonised guidelines,
Stuttgart: IER University of Stuttgart.
HEIMTSA, 2011. Monetary values for heath end-points used in the HEIMTSA / INTARESE
Common Case Study : D 4.1.2 of the HEIMSTA project, s.l.: s.n.
Hensher, D. A., Rose , J. M., Dios Ortúzar , J. d. & Rizzi, L. I., 2009. Estimating the
willingness to pay for and value of risk reduction for car occupants in the road environment.
Transportation Research Part A, 43(7), pp. 692-707.

Hesjevoll, I. S. & Elvik, R., 2016. Effect of traffic volume on road safety, s.l.: s.n.
Hirth, R. et al., 2000. Willingness to pay for a quality adjusted life year: In search of a
standard. Medical Decision Making, 20(3), pp. 323-342.
Holland, M., 2014. Cost-benefit Analysis of Final Policy Scenarios for the EU Clean Air
Package Version 2 Corresponding to IIASA TSAP Report 11, s.l.: EMRC.
Holland, M., 2014. Cost-benefit Analysis of Final Policy Scenarios for the EU Clean Air
Package Version 2 Corresponding to IIASA TSAP Report 11, Version 1, sl: EMRC.
Hurley, P. J., Physick, W. L. & Luhar, A. K., 2005. TAPM: a practical approach to prognostic
meteorological and air pollution modelling. Environmental Modelling & Software, 20(6), pp.
737-752.
IAWG, 2013. Technical Support Document: Technical update of the social cost of carbon for
regulatory impact analysis under executive order 12866, s.l.: Interagency Working Group on
the Social Cost of Carbon.
IEA, 2008. Energy Technology Perspectives - Scenarios & Strategies to 2050, Paris:
OECD/IEA.
IEA, 2010. Energy Technology Perspectives - Scenarios & Strategies to 2050, Paris:
OECD/IEA.
IEA, 2017a. Energy Techology Perspectives 2017, Paris: IEA.
IEA, 2017b. World Energy Outlook 2017, Paris: International Energy Agency.
IER, 2004. New elements for the assessment of external costs from energy technologies:
Final report to the European Commission, DG RTD, Stuttgart: Insitute for Energy Economics
and the Rational Use of Energy (IER).
IFEU, 2017. Energieverbrauch und Schadstoffemissionen des motorisierten Verkehrs in
Deutschland 1960-2035“ (TREMOD), Datenbank mit Daten für 2016, Dessau:
Umweltbundesamtes Deutschland.
IGCB, 2010. Noise & Health : Valuing the human health impacts of environmental noise
exposure, London: The Interdepartmental Group on Costs and Benefits Noise Subject Group
(IGCB(N)).
IIASA, 2014. A flexibility mechanism for complying with national emission ceilings for air
pollutants. TSAP report #15, Laxenburg: International Institute for Applied Systems Analysis
(IIASA).
InDeV, 2016. In-Depth understanding of accident causation for Vulnerable road users
(InDeV) : Review of European Accident Cost Calculation Methods : With regard to
vulnerable road users, s.l.: European Commission (EC).
Infras & IWW, 2004. External costs of transport: Update study, Karlsruhe & Zürich: Infras &
IWW.
INFRAS en Ecoplan, 2018. External Effects of Transport 2015 in Switzerland (‘Externe
Effekte des Verkehrs 2015’). Update study of the calculations of environment, accident and
health costs of road, rail, air and water transport 2010-2015, Bern: Swiss Federal Office for
Spatial Development.
Infras, CE Delft, ISI & University of Gdansk, 2008. Handbook on estimation of external costs
in the transport sector ; Produced within the study Internalisation Measures and Policies
for All external Cost of Transport (IMPACT), Delft: CE Delft.
Institute for Health Metrics and Evaluation, 2013. The global burden of disease: Generating
evidence, guiding policy, Seattle, Washington: IHME.
Intraplan & Planco, 2014. Grundsätzliche Überprüfung und Weiterentwicklung der Nutzen-
Kosten-Analyse im Bewertungsverfahren des Bundesverkehrswegeplanes, Munich: Intraplan.
IPCC, 1999. Aviation and the global atmosphere, Cambridge: IPCC.
IPCC, 2013. Fifth Assessment Report: Working Group 1, Cambridge: The Intergovernmental
Panel on Climate Change (IPCC) .
iRAP, 2015. Methodology Fact Sheet #13 - Economic analysis, s.l.: International Road
Assessment Programme (iRAP).

Isacs et al., 2016. Choosing a monetary value of greenhouse gases in assessment tools: A
comprehensive reveiw. Journal of Cleaner Production, pp. 37-48.
Istamto et al., 2014. Multi-country willingness to pay study on road-traffic environmental
health effects: Are people willing and able to provide a number?. Environmental Health.
Jakeman, G. & Fisher, B. S., 2006. Benefits of multi-gas mitigation: An application of the
global trade and environment model (GTEM). The Energy Journal Special Edition 3, pp. 323-
342.
Janstrup , K. et al., 2016. Understanding traffic crash under-reporting : Linking police and
medical records to individual and crash characteristics. Journal of traffic injury prevention,
17(6), pp. 580-584.
Jonsson, L. & Björklund, G., 2015. Marginal cost estimation for level crossing accidents :
Evidence from the Swedish railways 2000-2012, Stockholm: Centre for Transport Studies.
JRC, 2009. External costs of transportation, case study : maritime transport, Luxembourg:
Office for Official Publications of the European Communities.
JRC, 2014. Well-to-Tank Report Version 4.a, JEC Well-to-Wheels analysis of future
automotive fuels and powertrains in the European context, Luxembourg: European
Commission Joint Research Centre (JRC), Institute for Energy and Transport.
Keall, M. D., Guria, J., Howden-Chapman, P. & Baker, M. G., 2011. Estimation of the social
costs of home injury: A comparison with estimates for road injury. Accident Analysis and
Prevention, 43(3), pp. 998-1002.
Kemfert et al., 2006. Economic impact assessment of climate change: A multi-gas
investigation with WIAGEM-GTAPEL-ICM. The Energy Journal Special Edition 3, pp. 441-460.
Kempen, E. et al., 2018. WHO Environmental Noise Guidelines for the European Region: A
systematic review on environmental noise and cardiovascular and metabolic effects: A
summary. International Journal of Environmental Research and Public Health, 22(15).
Kilani, M., Proost, S. & Van der Loo, S., 2014. Road pricing and public transport pricing
reform in Paris : Complements or substitutes?. Economics of Transportation, 3(2), pp. 175-
187.
Kitous, A., Criqui, P., Bellevrat, E. & Chateau, B., 2010. Transformation patterns of the
worldwide energy system : Scenarios for the century with the POLES model. Energy Journal,
32(Special Issue 1), pp. 49-82.
Kruitwagen, S., Udo, J. & Janssen, L., 2006. Noise annoyance in the Netherlands: An
empirical valuation study, s.l.: s.n.
Kuik, O. et al., 2008. Report on the monetary valuation of energy related impacts on land
use, D.3.2. CASES Cost Assessment of Sustainable Energy Systems, updated july 2008, s.l.:
s.n.
Kuik, O., L. B. & R. S. T., 2009. Marginal abatement costs of greenhouse gas emissions : A
meta-analysis. Energy Policy, 37(4), pp. 1395-1403.
Kurosawa, A., 2006. Multigas Mitigation: An economic analysis using GRAPE model. The
Energy Journal Special Edition 3, pp. 235-250.
Lambert, J., Poisson, F. & Champelovier, P., 2001. Valuing benefits of a road traffic noise
abatement programme: A contingent valuation survey. Rome, s.n.
Lee, D. S. et al., 2009. Aviation and global climate change in the 21st century. Atmospheric
Environment, 43(22-23), pp. 3520-3537.
Leimbach, M. et al., 2010. Technological change and international trade: Insights from
REMIND-R. Energy Journal, 31(Special Issue 1), pp. 109-136.
Lercher, p. et al., 2010. The effects of railway noise on sleep medication intake: Results
from the ALPNAP-study. Noise Health, 12(47), pp. 110-119.
Lijesen, M. et al., 2010. How much noise reduction at airports?. Transportation Research
Part D: Transport and Environment, 15(1), pp. 51-59.
Lindberg et al., 2006. Marginal cost case studies for road and rail transport - Deliverable
D.3, s.l.: s.n.

Lindberg, G., 2001. Traffic insurance and accident externality charges. Journal of Transport
Economics and Policy, 35(3), pp. 399-416.
Lindhjem, H., Navrud, S. & Braathen, N. A., 2010. Valuing lives saved from environmental,
transport and health policies : A meta-analysis of stated preference studies, Paris: OECD.
Li, S., Purevjav, A.-O. & Yang, J., 2016. The Marginal Cost of Traffic Congestion and Road
Pricing : Evidence from a Natural Experiment in Beijing, s.l.: s.n.
M.J., G. et al., 2013. ReCiPe 2008, A life cycle impact assessment method which comprises
harmonised category indicators at the midpoint and the endpoint level; First edition
Report I: Characterisation. Update may 2013, Bilthoven: RIVM.
M.J., N., G., R. & P., D., 2017. WHO Environmental Noise Guidelines for the European
Region: A systematic review on environmental noise and adverse birth outcomes.
International Journal of Environmental Research and Public Health, 14(10).
Mackay, M., 2003. National differences in European mass accident data bases. In:
Proceedings of the Joint Session on Injury Scaling Issues, IRCOBI Annual Conference, Lisbon,
September 2003. Lisbon: s.n.
Magné, B., Kypreos, S. & Turton, H., 2010. Technology options for low stabilisation
pathways with MERGE. Energy Journal, 31(Special Issue 1), pp. 83-108.
Maibach et al., 2000. External costs of transport, accident, environmental and congestion
costs in Western Europe, s.l.: Infras, IWW & Universität Karlsruhe.
Marten et al., 2013. Improving the assessment and valuation of climate change impacts for
policy and regulatory analysis. Climatic Change, pp. 433-438.
Mattmann, M., Logar, I. & Brouwer, R., 2016. Hydropower externalities : a meta analysis.
Energy Economics, Volume 57, pp. 66-77.
Meersman, et al., 2006. Generalisation of Research on Accounts and Cost Estimation
(GRACE) – Deliverable D4: Marginal Cost Case Studies for Air and Water Transport,
Antwerp: University of Antwerp.
Michael, J. et al., 2009. Long-Term Ozone Exposure and Mortality. The New England Journal
of Medicine, 360(11), pp. 1085-1095.
Miedema, H. M. E. & Oudshoorn, C. G. M., 2001. Annoyance from transportation noise:
Relationships with exposure metrics DNL and DENL and their confidence intervals.
Environmental Health Perspectives, pp. 409-416.
Ministry of Transport, 2013. The social cost of road crashes and injuries : 2013 update,
Wellington: Ministry of Transport.
Moore, F. C. & Diaz, D. B., 2015. Temperature impacts on economic growth warrant
stringent mitigation policy. Nature Climate Change, Volume 5, pp. 12-131.
Morris, J. & Hammitt, J. K., 2001. Using life expectancy to communicate benefits of health
care programs in contingent valuation studies. Medical Decision Making, pp. 468-478.
Mouter, N. & Chorus, C., 2015. Value of Time : A citizen perspective. Transportation
Research Part A: Policy and Practice, 91(September), pp. 317-329.
Nateco; Econcept, 2004. External costs of transport due to impacts on nature and
landscape (first botton-up calculation), s.l.: s.n.
Navrud, 2002. The state of the art on economic valuation of noise, Brussels: EC DG
Environment.
Navrud, S., Lindhjem , H. & Magnussen, K., 2016. Valuing Marine Ecosystem Services Loss
from Oil Spills for Use in Cost-Benefit Analysis of Preventive Measures. In: P. A. Nunes, P.
Kumar, L. E. Svensson & A. Markandya, eds. Handbook on the Economics and Management
for Sustainable Oceans. Cheltenham: Edward Elgar Publishing.
Navrud, S. et al., 2005. HEATCO Deliverable 4 - Economic values for key impacts valued in
the stated preference surveys, s.l.: s.n.
NEEDS, 2006a. Assessment of Biodiversity Losses, NEEDS deliverable D.4.2.-R.S. 1b/WP4,
priority 6.1: Sustainable Energy Systems and, more specifically Sub-priority 6.1.3.2.5:
Socio-economic tools and concepts for energy stragegy. [Online]

Available at: http://www.needs-project.org/docs/results/RS1b/RS1b_D4.2.pdf
[Accessed 2017].
NEEDS, 2006b. Deliverable 6.7: Final report on the monetary valuation of mortality and
morbidity risks from air pollution, Brussels: European Commission.
NEEDS, 2007. Final report on casual links between pollutants and health impacts.
Deliverable RS 1b D 3.7. : A set of concentration-response functions. (...),Sub-priority
6.1.3.2.5: Socio-economic tools and concepts for energy strategy, Brussels: European
Commission.
NEEDS, 2008. Deliverable No 1.1 RS 3a Report on the procedure and data to generate
averaged/aggregated data, Brussels: European Commission.
NEEDS, 2009. New Energy Externalities Developments for Sustainability, Deliverable 2.1 -
Value Transfer Techniques and Expected Uncertainties, s.l.: s.n.
Nellthorp, J., Bristow, A. & Day, B., 2007. Intorducing willingness-to-pay for noise changes
into transport appraisal: An application of benefit transfer. Transport reviews.
Nelson, J. P., 2008. Hedonic property value studies of transportation noise : aircraft and
road traffic. In: A. Baranzini, J. Ramirez, C. Schaerer & P. Thalmann, eds. Hedonic methods
in housing market economics : Pricing Environmental Amenities and Segregation.
s.l.:Springer Nature Switzerland AG, pp. 55-82.
NewExt, 2004. Final Report : New Elements for the Assessment of External Costs from
Energy Technologies, Stuttgart: Institute of Energy Economics and Rational Energy Use
(IER).
NICE, 2018. NICE Glossary. [Online].
OECD ; ECMT, 2007. Managing Urban Traffic Congestion, Paris: OECD/ECMT.
OECD, 2012. Mortality Risk Valuation in Environment, Health and Transport Policies, Paris:
OECD.
OECD, 2014. The costs of air pollution : Health impact of road transport, Paris: OECD.
OECD, 2016. Road infrastructure : inclusive development and traffic safety in Korea, Paris:
OECD.
Österreichischer Verkehrssicherheitsfonds & BMVIT, 2012. Unfallskostenrechnung Strasse
2012, Wien: Österreichischer Verkehrssicherheitsfonds.
Oum, T. H., Waters, W. & Yong, J. S., 1990. A survey of recent estimates of price
elasticities of demand for transport, Washington D.C.: The World Bank Group.
Park, S.-k., 2008. Road accidents in Korea. IATSS Research (International Association of
Traffic and Safety Sciences), 32(2), pp. 6-128.
Pearce, D. & B.Pearce, 2000. Valuing the external costs of aviation, London: The Stationery
Office.
Persson, U., Hjalte, K., Nilsson, K. & Norinder, A., 2000. Värdet av att minska risken för
vägtrafikskador - Beräkning av riskvärden för dödliga, genomsnittligt svara och lindriga
skador med Contingeng Valuation metoden, Lund: Institutionen för Teknik och Samhälle -
Lunds Tekniska Högskola.
Pigou, A., 1920. Wealth and Welfare. London: MacMillan.
Polinder, S. et al., 2016. The economic burden of injury : Health care and productivity costs
of injuries in the Netherlands. Accident Analysis & Prevention, Volume 93, pp. 92-100.
Popp, C., 2006. Lärmbelastung, ökonomische Folgen und Handlungsoptionen im Verkehr,
Hamburg: Lärmkontor GmbH.
Prud’homme, R., Koning , M., Lenormand, L. & Fehr, A., 2012. Public transport congestion
costs: The case of the Paris subway, Transport Policy 21 (2012)101-109.. Transport Policy,
Volume 21, pp. 101-109.
Quinet et al., 2013. L'évaluation socioéconomique des investissements publics, Paris:
Commisariat général à la stratégie et à la prospective.
Quinet, E., 2003. Short term adjustments in rail activity: The limited role of infrastructure
charges. Transport Policy , 10(1), pp. 73-79.

R., G., D., S. & R.., S., 2017. WHO Environmental Noise Guidelines for the European Region
: A systematic review on environmental noise and annoyance. International Journal of
Environmental Research and Public Health, 14(12).
Rabl, A., 1999. Rabl, A., 1999. Air pollution and buildings : An estimation of damage costs in
France. Environment Impact Assessment Review, Volume 19, pp. 361-385..
https://www.sciencedirect.com/science/journal/01959255, 19(4), pp. 361-385.
Rabl, A., Spadaro, J. & Holland, M., 2014. How Much Is Clean Air Worth?: Calculating the
Benefits of Pollution Control, Cambridge: Cambridge University Press.
Ready, R. & Navrud, S., 2006. International benefit transfer: Methods and validity tests..
Ecological Economics, 60(2), pp. 429-434.
Ready, R. et al., 2004. Benefit Transfer in Europe: How Reliable Are Transfers between
Countries?. Environmental and Resource Economics, 29(1), pp. 67-82.
Revesz, R. L. et al., 2014. Global warming: Improving economic models of climate change".
Nature, 508(April), pp. 173-175.
Riahi, K. et al., 2015. Locked into Copenhagen pledges : Implications of short-term emission
targets for the cost and feasibility of long-term climate goals. Technological Forecasting
and Social Change: Part A, 90(Part A), pp. 8-23.
Ricardo ; TRT, 2017. Study on urban mobility : Assessing and improving the accessibility of
urban areas, Final report and policy proposals, Brussels: European Commission, DG Move.
Ricardo-AEA, TRT, DIW Econ & CAU, 2014. Update of the Handbook on External Costs of
Transport, London: Ricardo-AEA.
RIVM, 2006. Zor voor gezondheid: Volksgezondheid toekomst verkenning, Bilthoven:
National Institute for Public Health and the Environment Netherlands (RIVM).
RIVM, 2014. Op weg naar maatschappelijke kosten-batenanalyses voor preventie en zorg:
Themarapport volksgezondheid toekomst verkenning 2014, Bilthoven: National Institute for
Public Health and the Environment Netherlands (RIVM).
Rosman, D. L., 2001. The Western Australian road injury database (1987-1996): Ten years of
linked police, hospital and death records of road crashes and injuries. Accident : Analysis
and Prevention, 33(1), pp. 81-88.
RVZ, 2006. Zinnige en duurzame zorg, Zoetermeer: Raad voor de Volksgezondheid en Zorg
(RVZ).
Salomon et al., 2012. Common values in assessing health outcomes from disease and injury:
Disability weights measurement study for the Global Burden of Disease Study 2010. Lancet,
pp. 2129-2143.
Sassi, F., 2006. Calculating QALYs, comparing QALY and DALY calculations. Health Policy
and Planning, pp. 402-409.
Sausen, R. et al., 2005. Aviation radiative forcing in 2000: An update on IPCC (1999).
Meteorologische Zeitschrift, 14(4), pp. 555-561.
Schade, W. et al., 2006. COMPETE Final Report: Analysis of the contribution of transport
policies to the competitiveness of the EU economy and comparison with the United States
(COMPETE), s.l.: Karlsruhe.
Schipper, Y. J. J., 1996. European Regional Science Association: 36th European Congress.
Zurich, s.n.
Seidler, A. et al., 2015. NORAH Noise-related annoyance, cognition and health -
Verkehrslärmwirkungen im Flughafenumfeld. Band 6: Sekundärdatenbasierte
Fallkontrollstudie met vertiefender Befragung, Kelsterbach: Gemeinnützige Umwelthaus
GmbH (Hsg.)..
Significance, VU University Amsterdam & John Bates Services, 2012. Values of time and
reliability in passenger and freight transport in The Netherlands, s.l.: Significance, VU
University Amsterdam & John Bates Services.
Sliwinska-Kowalska, M. & Zaborowski, K., 2017. WHO Environmental Noise Guidelines for
the European Region: A systematic review on environmental noise and permanent hearing

loss and tinnitus. International Journal of Environmental Research and Public Health,
14(10).
Smith, S. & Braathen, N. A., 2015. Monetary carbon values in policy appraisal: An overview
of current practice and key issues, s.l.: OECD publishing.
Sommer et al., 2002. Unification of accounts and marginal costs for trasnport efficiency
(UNITE) Deliverable 9: Accident Cost Case Study 8a: Marginal External Accident Costs in
Switzerland, Leeds: ITS University of Leeds.
Sommer, H., Brügger, O., Lieb, C. & Niemann, S., 2007. Volkwirtschaftliche Kosten der
Nichtberufsunfälle in der Schweiz : Strassenverkehr, Sport, Haus und Freizeit, Bern:
Beratungsstelle für Unfallverhütung (BFU).
Sørensen, M. et al., 2013. Long-term exposure to road traffic noise and incident diabetes: A
cohort study. Environmental Health Perspectives, 121(2), pp. 217-222.
Sørensen, M., Ketzel, M., Overvad, K. & Tjønneland, A., 2014. Exposure to road traffic and
railway noise and postmenopausal breast cancer: A cohort study. International Journal of
Cancer, 134(11), pp. 2691-2698.
Stern, N., 2006. The economics of climate change: The Stern Review, Cambridge:
Cambridge University Press.
Swoboda, A., Nega, T. & Timm, M., 2015. Hedonic analysis over time and space: The case
of house prices and traffic noise. Journal of regional science, pp. 644-670.
SWOV, 2014. Kosten van verkeersongevallen in internationaal perspectief, Den Haag:
Stichting Wetenschappelijk Onderzoek Verkeersveiligheid.
Theebe, M. A., 2004. Planes, trains and automobiles: The impact of traffic noise on house
prices. Journal of Real Estate Finance and Economics, pp. 2009-234.
TML, 2010. Internalisering van externe kosten van transport in Vlaanderen, Mechelen:
MIRA.
Tol, R., 2006. Multi-gas emission reduction for Climate Change Policy: An application of
FUND. The Energy Journal Special Edition 3, pp. 235-250.
Tol, R., 2008. The social cost of carbon: Trends, outliers and catastrophes. Economics,
2(Online), pp. 1-24.
Tol, R. S., 2013. Targets for global climate policy : An overview. Journal of Economic
Dynamics and Control, 37(5), pp. 911-928.
Trafikverket, 2015. Samhällsekonomiska principer och kalkylvärden för transportsektom:
ASEK 5.2 Kapitel 20 - English summary of the ASEK Guidelines, Borlänge: Trafikverket.
Trawén, A., Hjalte, K., Norinder, A. & Persson, Ulf , U., 1999. Skattning av riskvärden och
dödsfallsekvivalenter i vägtrafiken med hjälp av conjoint analysis, risk-risk trade off och
standard gamble metoderna - En piloundersökning, Lund: Institutionen för Teknik och
Samhälle - Lunds Tekniska Högskola.
TRL, 2001. A study on the cost of transport in the European Union in order to estimate and
assess the marginal costs of the use of transport. TRL on behalf of DG TREN., s.l.: s.n.
Trozzi, C., 2003. Environmental impact of port activities 1th International Scientific
Symposium Environment and Transport 19-20 June 2003, Avignon (F).: s.n.
UBA & IER, 2014. Best practice kostensätze für luftschadstoffe, verkehr, strom- und
wärmeerzeugung, Dessau-Rosslau: UBA.
UBA, 2012. Ökonomische Bewertung von Umweltschäden : Methodenkonvention23.0 zur
Schätzung von Umweltkosten, Berlin: Umweltbundesamt Deutschland (UBA) (German
Environment Agency)..
UBA, 2018. Methodenkonvention 3.0 zur Schätzung von Umweltkosten, Berlin:
Umweltbundesamt Deutschland (UBA) (German Environment Agency)..
Udo, J., Janssen, L. H. & Kruitwagen, S., 2006. Stilte heeft zijn 'prijs'. ESB, pp. 14-16.
UN ECE, 2011. Statistics of road traffic accidents in Europe and North America, Geneva: UN
Economic Commission for Europe (ECE).

UNITE, 2003. UNIfication of accounts and marginal costs for Transport Efficiency (UNITE) :
Deliverable 11: Environmental marginal cost case studies, Stuttgart: IER University of
Stuttgart.
US Department of Transportation, 2015. Memo : Guidance on treatment of the economic
value of a statistical life in US Department of Transportation Analyses - 2015 adjustment,
Washington DC: US Department of Transportation.
Van den Bijgaart, I., Gerlagh, R., Korsten, L. & Liski, M., 2013. A simple formula for the
social cost of carbon, Milan: Fondazione Eni Enrico Mattei (FEEM).
Van Grinven, H. et al., 2013. Costs and benefits of nitrogen for Europe and implications for
mitigation. Environmental Science & Technology, 47(8), pp. 3571-3579.
van Wee, B. & Rietveld, P., 2013. Using value of statistical life for the ex-ante evaluation of
transport policy options: A discussion based on ethical theory. Transportation, pp. 295-314.
Verme et al., 2016. The welfare of Syrian refugees : Evidence from Jordan and Lebanon,
Washington D.C.: The World Bank Group.
Vickrey, W., 1963. Pricing in urban and suburban transport. The American Economic
Review, 53(2), pp. 452-465.
Victorian Competition & Efficiency Commission, 2006. Making the Right Choices : Options
for Managing Transport Congestion, A draft report for further consultation and input,
Melbourne: State of Victoria.
Viscusi, W. K., 2010. The heterogeneity of the value of statistical life: Introduction and
overview. Journal of Risk and Uncertainty, pp. 1-13.
Viscusi, W. K. & Masterman, C. J., 2017. Income elasticities and global vlaues of a
statistical life. Journal of Benefit-Cost Analysis, pp. 226-250.
VMM, 2013. Milieurapport Vlaanderen MIRA : Themabeschrijving zwevend stof, Mechelen:
Vlaamse Milieumaatschappij (VMM).
Vogt-Schilb, A. & Hallegatte, S., 2014. Marginal abatement cost curves and the optimal
timing of mitigation measures. Energy Policy, pp. 645-653.
VROM, 2006. Reken- en meetvoorschrift wegverkeerslawaai, Den Haag: Ministerie van
VROM.
VTPI, 2017. Transportation Cost and Benefit Analysis: Techniques, Estimates and
Implications. [Online]
Available at: http://www.vtpi.org/tca/
[Accessed 2018].
VU, 2005. De waardering van interne en externe veiligheid in het verkeer, Amsterdam:
Vrije Universiteit (VU).
Waldhoff, S., Anthoff, D., Rose, S. & Tol, R. S. J., 2014. The marginal damage costs of
different greenhouse gases: An application of FUND. Economics, 8(online), pp. 1-33.
Wardman, M., Chintakayala, P., Jong, G. d. & Ferrer, D., 2012. European Wide Meta-
Analysis of Values of Travel Time, Final Report to the European Investment Bank, Leeds:
University of Leeds.
Watkiss, P., Downing, T., Handley, C. & Butterfield, R., 2005. The impacts and costs of
climate change, Harwell: AEA Technology Environment.
Watkiss, P., Holland, M., Hurley, F. & Pye, S., 2006. Damage Costs for Air Pollution Final
Report to Defra, Didcot: AEA Technology.
WHO, 1946. Preamble to the Consitution of the WHO as adopted by the International
Health Conference, New York, 19 June-22 July 1946, New York: World Health Organisation
(WHO).
WHO, 2005. Air quality guidelines for particulate matter, ozone, nitrogen dioxide and
sulfur dioxide : Global update 2005, summery of risk assement, Geneva: World Health
Organization (WHO).
WHO, 2011. Burden of disease from environmental noise : Quantification of healthy life
years lost in Europe, Copenhagen: WHO.

WHO, 2013. Health risks of air pollution in Europe – HRAPIE project. Recommendations for
concentration–response functions for cost–benefit analysis of particulate matter, ozone and
nitrogen dioxide, Geneva: World Health Organization (WHO).
WHO, 2014. WHO Expert Meeting: Methods and tools for assessing the health risks of air
pollution at local, national and international level, meeting report 12-13 May , Bonn: WHO.
WHO, 2017-2018. WHO in: Systematic Reviews 2017-2018. s.l.:BioMed Central.
WHO, 2018. Health Statistics and Information Systems. [Online]
Available at: http://www.who.int/healthinfo/en/
[Accessed 2018].
Wijnen, W. & Stipdonk, H., 2016. Social costs of road crashes: An international analysis.
Accident Analysis and Prevention, 94(September), pp. 97-106.
Wijnen, W. et al., 2017. Crash cost estimates for European countries, deliverable 3.2 of
the H2020 project SafetyCube., Loughborough: Loughborough University, SafetyCube.
Winke, T., 2017. The impact of aircraft noise on apartment prices: A differences-in-
differences hedonic approach for Frankfurt, Germany. Journal of Economic Geography,
17(6), pp. 1283-1300.
World Bank, 2005. Valuation of accident reduction, Washington D.C.: The World Bank
Group.

A Economic valuation of human
health
A.1 Introduction
Human health can be valued in a number of ways, which all slightly differ from each other.
This can result in an overwhelming amount of terminology, and seeing the interlinkages
between all of them can be quite confusing. In this chapter, we elaborate on the different
ways in which human health can be valued, and how the terminologies can be converted.
The valuation of human health is particularly relevant for the chapters on accidents
(Chapter 3, Annex B), air pollution (Chapter 4, Annex C) and noise (Chapter 6, Annex E).
A.2 Indicators for human health

There is a large variety in indicators for human health. This section provides an overview of
the main ways in which health can be quantified and valued. We distinguish two types of
indicators: the first type reflects changes in health in terms of years (e.g. YOLL, YLD, DALY
and QALY), the second type provides a valuation for an (un)healthy life(year) (e.g. VSL and
VOLY).
A.2.1 Quantification of health in terms of years
Years of life lost

The Years of Life Lost (YOLL) is the amount of years of premature mortality as a result of a
health condition in the population (WHO, 2018). It can be defined by the number of deaths
due to a health condition, multiplied by the standard life expectancy at the age at which
death occurs.
𝑌𝐿𝐿 = 𝑁 × 𝐿
Where:
N = number of deaths due to an illness
L = standard life expectance at the age of death in years
Years lost due to disability

The Years Lost due to Disability (YLD) represent the amount of healthy years lost for people
living with the health condition or its consequences (WHO, 2018). To calculate YLD, the
concept of Disability Weights (DW) is used. The DW reflects the severity of the disease on a
scale from 0 (perfect health) to 1 (dead), based on individuals’ perceptions of the impact of
the disease on people’s lives (Institute for Health Metrics and Evaluation, 2013). YLD is
calculated by multiplying the number of prevalent cases by the disability weight (Institute
for Health Metrics and Evaluation, 2013). This is an update of the previous definition of
YLD, which was based on incidence, rather than prevalence.

𝑌𝐿𝐷 = 𝑃 × 𝐷𝑊
Where:
P = number of prevalent cases
DW = disability weight of illness
Table 74 illustrates the most recent disability weights associated with a number of illnesses.
The same disability weights are used for everyone living a year in a specified health state
(WHO, 2018). The most recent disability weights are determined using data gathered from
thousands of respondents from all over the world, rather than expert judgement.
Table 74 – Examples of disability weights
Disease Disability weight

Cancer: diagnosis and primary therapy 0.294
Cancer: terminal phase: without medication 0.519
Stroke: long-term consequences, moderate 0.076
Acute myocardial infarction: days 1-2 0.422
Source: Based on (Salomon et al., 2012).
Disability-adjusted life year

One Disability-Adjusted Life Year (DALY) can be defined as one lost year of ‘healthy’ life,
and is the sum of YOLL and YLD due to a health condition. The sum of DALYs across a
population is the burden of disease, and can be thought of as a measure of the g ap between
the current health status and the ideal health status of the population (Institute for Health
Metrics and Evaluation, 2013).
𝐷𝐴𝐿𝑌 = 𝑌𝐿𝐿+ 𝑌𝐿𝐷
Quality-adjusted life year

Quality-Adjusted Life Years (QALY) have the underlying assumption that health is a function
of the length and quality of life. One QALY equals one year of life in perfect health.
By multiplying the utility weight46 associated with a given health state by the years lived in
that health state, the QALYs can be calculated (NICE, 2018). E.g. if medical treatment
improves a patient’s utility from 0.9 to 1, and the person has 10 life years left, then the
health gains are equal to one QALY: 0.1 QALY per year over 10 years. The utility value is
based on individual perception of their health status on well-being. In general, the
relationship between a QALY and a DALY can be summarised as follows, 1 DALY is equal to
1.087 QALY (Sassi, 2006; CE Delft, 2017). Based on (CE Delft, 2017) conservative estimates
show that a VOLY (A.2.3) is roughly equal to a DALY, although there are some indications
that a DALY should be valued higher than a VOLY. 47 Therefore, one can conclude that a
QALY is also roughly equal to a VOLY.
A large study conducted on behalf of the European Commission investigated whether the
theoretical assumptions underlying the QALY could be validated by an experiment eliciting
respondents’ preferences (Beresniak, et al., 2015). The results suggested that most critical
________________________________
46
The utility weight and disability weight are inversely related concepts. For instance, if the disability weight is
0.33, then the utility weight is 1 - 0.33 = 0.67.
47
Please see (CE Delft, 2017) for a full discussion.

assumptions underlying the QALY could not be validated. For instance, responses indicated
that individuals are not risk neutral, that willingness to gain or lose life-years is not
constant over time, that life-years and quality of life are independent of each other and
that the quality of life cannot be measured in consistent intervals (Beresniak, et al., 2015).
Other weaknesses of QALYs are that they can only measure the effects of long term care to
a limited extent, and that mortality is not part of the QALY, which complicates its use in
quantifying external effects.
A.2.2 Valuation of health - Value of a statistical life
Definition
The Value of a Statistical Life (VSL) is also known as the Value of a Prevented Fatality
(VPF). There are two main ways in which the VSL can be calculated: labour market studies
or willingness-to-pay (WTP) studies. In general, US estimates of the VSL are based on wage
risk studies, whereas it is more common in Europe to calculate the VSL based on WTP. The
VSL is the amount of money that a community of people are willing to pay to lower the risk
of an anonymous instantaneous premature death within that community. It can be
calculated by dividing the amount people are willing to pay by the change in mortality risk.
Labour market studies or wage risk studies in particular identify the amount of financial
compensation needed to accept a job with a higher mortality rate. This serves as the basis
for the calculation of the VSL. In general, a VSL that is calculated based on wage risk
studies is much higher than a VSL calculated based on WTP studies. For instance the current
recommended VSL is € 6.9 million ($ 9.2 million) in the U.S. and € 4.1 million in Canada
(partly based on WTP and partly on labour market studies). European VSLs, based on WTP
methods, are more commonly found in between € 1-2 million. There are indications that
labour market studies tend to overestimate the VSL because of the cognitive bias in
individuals to overestimate small risks and underestimate large risks.
Literature
For a good overview of different VSL studies up to 2007 we refer to (Andersson & Treich,
2009). For a more recent overview of VSL values we refer to (Ecoplan, 2016) (and Table 75
below).
Table 75 – Overview of VSL studies
Study Country Calculated for VSL Source

Non-European multi-country studies
OECD All OECD countries All areas $ 3 million Meta-analysis
(€3.6 million)
European VSL studies
ExternE (2005) Europe Air pollution € 1 million Own study
CE Delft (2008) Europe Traffic accidents € 1.5 million HEATCO
(HEIMTSA, 2011) Europe Health € 1.65 million Alberini (2006)
CE Delft (2011) Europe Traffic accidents € 1.67 million UNITE
Ricardo et al (2014) Europe Traffic accidents € 1.7 million HEATCO
on streets
Ricardo et al (2014) Europe Air pollution € 1.65 million HEIMTSA
WHO HEAT (2014) Europe Health benefits € 3.4 million OECD

Study Country Calculated for VSL Source
Single country VSL studies
Department for Ireland Traffic accidents € 1.3 million Charty et al. (1999)
Transport (2007) on streets
Abellan Perpinan et al. Spain Traffic accidents € 1.3 million Own study
(2011) on streets
(Österreichischer Austria Traffic accidents € 2.32 million HEATCO
Verkehrssicherheitsfonds on streets
& BMVIT, 2012)
Sachstandpapier Luft Germany Air pollution € 1.65 million HEIMTSA
(2012)
(DfT, 2012) United Kingdom Traffic accidents £ 1.1 million Charty et al. (1999)
on streets (~ € 1.4 million)
Norwegian Ministry of Norway All areas 30 million NOK Own study and
Finance (2012) (~ € 4.0 million) OECD
(Commissariat général à France All areas € 3.0 million OECD
la stratégie et à la
prospective, 2013))
(Intraplan & Planco, Germany Traffic accidents € 1.3 million HEATCO
2014) on streets
(SWOV, 2014) Netherlands Traffic accidents € 2.0 million Own study
on streets
(Trafikverket, 2015) Sweden Traffic accidents 22.3 million SEK Own study
on streets (~ € 2.4 million)
Source: Based on (Ecoplan, 2016).
Table 75 presents an overview of the main studies that have recently been published.
The table consists of values from country-specific studies, as well as broader European
studies. The most recent, high quality meta-analysis was conducted by the (OECD, 2012).
The (OECD, 2012) study is the largest meta–analysis of stated preference VSL studies to date
and uses 261 VSL values from 28 studies conducted in OECD countries. The recommended
VSL for OECD countries is $ 3 million (range: $ 1.5-4.5 million, 2005 prices), and the
recommended EU27 value is $ 3.6 million (2005 dollars), with a range of $ 1.8-5.4 million.
This EU27 value is based on 16 European studies in the OECD meta-analysis. However,
because this value is elicited from a narrower base, it is arguably more uncertain. For those
reasons (Ecoplan, 2016) recommend using the OECD country value of $ 3 million and
adjusting it to individual country situations. If we use the OECD value it translates to an
EU28 VSL of € 3.6 million (2016 prices). The OECD study was published before the previous
edition of the Handbook, yet its values were not used. This was because “the OECD meta-
analysis only reports an EU-wide VSL figure and does not include values for each Member
State”. Although this is true, the OECD study does provide guidance on how country specific
values for the VSL can be calculated.
As Table 75 shows, studies have calculated the VSL values for different circumstances, such
as a life lost through air pollution, a traffic accident, during sports or household work.
Some scholars have argued that a life lost under one circumstance (e.g. traffic accident)
should not be valued the same as a life lost under a different circumstance. For instance,
(B,S,S. Volkswirtschaftliche Beratung AG, 2015) conducted a stated preference study with
3000 respondents and identified a VSL of 4.5 million CHF, 30.6 million CHF and 10.5 million
CHF for lives lost in road traffic accidents, rail traffic accidents and lives lost due to noise
or air pollution respectively. In the past, a VSL identified in one context was frequently
used in a completely different context using benefits transfer methods, for instance (Keall,

et al., 2011) use the same VSL to value road traffic accidents and accidents that occur at
home. A study conducted by (Dekker, et al., 2011) carried out a meta-analysis of 26
international stated preference studies looking at the VSL, and found that the road safety
VSL should be multiplied by a factor 1.8 before being applicable in the context of air
pollution. This roughly echoes the findings from (B,S,S. Volkswirtschaftliche Beratung AG,
2015). The underlying reason for the different VSL values is due to different risk contexts,
such as risk perception and the population at risk. Contrastingly, the meta-analysis carried
out by the (OECD, 2012), revealed that the VSL from transport related risk is larger than the
VSL from environmental risks, although this difference was not significant. Other studies
since, such as (Ecoplan, 2016), build on the OECD results, and therefore recommend the use
of the same VSL for lives lost under any circumstance. This is also the recommendation in
this Handbook.
It is important to note that the VSL is not constant over one’s lifetime (Aldy & Viscusi,
2004); (Carlsson, et al., 2010). Its value peaks around the age of 30-50, after which it starts
to decline. There are a number of factors that contribute to this phenomenon. A higher
education and a higher income positively contribute to a higher valuation of the VSL, both
of which tend to rise (to some extent) with age. In addition, adults are valued most highly
because of the important role they have in the family and society, in terms of caring for
children and the elderly. (Aldy & Viscusi, 2004) show that the VSL of a 60-year-old is roughly
half of the VSL of a 30 or 40-year-old. This echoes the trend in life expectancy. The life
expectancy of a 60-year-old is half of the life expectancy of a 30-year-old. Another issue
related to age and the VSL is how to value the lives of children. Carlsson, et al., (2010)
show that the VSL of children is valued 1.4 times higher than the VSL of 35 to 45-year-old
adults, and 3.3 times higher than the VSL of 65 to 75-year-olds. Studies such as (Ricardo-
AEA, TRT, DIW Econ & CAU, 2014; HEIMTSA, 2011; UBA & IER, 2014) recommend adjusting
the VSL by a factor 1.5 to reach the VSL for children. In Norway, the VSL of children is
valued at twice the regular VSL.
In this Handbook, we use the VSL based on the OECD meta-analysis, with a value of
€ 3.6 million for the EU28. We do not differentiate it to different age groups in this study,
although we do recommend the use of an age differentiated VSL if detailed data on e.g. the
age of accident victims is available.
A.2.3 Valuation of health - Value of a life year
Definition
The Value of a Life Year (VOLY) is sometimes known as the Value of One Year Lost.
The VOLY is the amount of money that people are willing to pay for one year of additional
life expectancy (CONCAWE, 2006). Arguably, the VOLY is the correct metric to use in
circumstances of non-instantaneous death (IER, 2004) (EC, 2005). In theory the VOLY is
related to the VSL, as the VSL can be seen as a discounted sum of annual VOLY-values
(CE Delft, 2018).
𝑇
𝑎𝑃𝑖
𝑉𝑆𝐿 = 𝑉𝑂𝐿𝑌 ∑
(1 + 𝑟)𝑖−𝑎
𝑖=𝑎

Where:
𝑇 = the maximum life expectancy (110 years)
𝑎 = average age of the person whose VOLY we are interested in (40 years)
𝑎𝑃𝑖 = probability that a person of age a will reach age i
𝑟 = discount rate
The precise relationship between the VOLY and the VSL is hard to pinpoint as they measure
fundamentally different things. The VSL attaches a monetary value to an instant death,
whereas the VOLY measures the value attached to one additional year of life expectancy.
The VOLY is a proxy for the valuation at the end of one’s life, whereas the VSL represents
the valuation in the middle of your life. It is therefore not surprising that the literature
does not reveal a fixed relationship between the VSL and the VOLY. The relationship is not
linear in the sense that one cannot take the average life expectancy and multiply it by the
VOLY to reach the VSL. As both the VSL and the VOLY values are based on meta-analysis of
studies looking at the individual measure in question, it would be unwise to link the two
together as they measure fundamentally different things.
Literature
The literature has suggested a range of different values for the VOLY, of which an overview
is provided in Table 76. VOLYs are presented in original prices as well as 2016 prices, where
corrections for inflation and growth in GDP per capita were made.
Although the VOLY can be directly derived from the VSL as mentioned in the section above,
there is a debate on whether or not that is appropriate in certain contexts. For valuing the
life-years of instantaneous deaths, using a VOLY derived from the VSL is logical, as the VSL
is elicited based on changes in mortality risk. For non-instantaneous mortality, using a VOLY
directly derived from the VSL is arguably comparing apples and oranges. For instance, the
NEEDS project argues that VOLYs derived from the VSL overestimate the WTP to reduce
health impacts of air pollution and noise. Therefore, the NEEDS VOLY frames the WTP
questions to elicit the value of a change in life expectancy, rather than a change in
mortality rate. For non-instantaneous deaths (e.g. deaths as a result of air pollution),
this is a better method because air pollution affects the life expectancy, rather than the
mortality rate. The NEEDS project recommended a VOLY of € 40,000, with € 25,000-
€ 100,000 (2006 prices) as confidence intervals (NEEDS, 2006b; Desaigues, et al., 2011).
This value is based on surveys conducted in France, Spain, the UK, Denmark, Germany,
Switzerland, Czech Republic, Hungary and Poland, with a total sample size of 1,463.
The previous edition of the Handbook (Ricardo-AEA, TRT, DIW Econ & CAU, 2014) uses
(and updates) the NEEDS value (€ 43,000 in 2010 prices, range € 27,000-€ 130,000).

Table 76 – Overview of the VOLY values found in the literature
Study Year Country/ VOLY in original Price VOLY in 2016 Source

region prices level prices48
NewExt 2004 Europe € 50,000 2004 € 77,800 Own study
ExternE 2005 Europe € 50,000 2005 € 73,700 NewExt
NEEDS 2006 Europe € 40,000 2006 € 55,200 Own study
CE Delft 2008 Europe € 50,000 2002 € 84,900 NewExt,
ExternE,
NEEDS
European 2009 Europe € 50,000-100,000 2009 € 64,800-129,600 Range of
Commission studies
CE Delft 2010 Netherlands € 55,000 2008 € 88,500 Range of
studies
HEIMTSA 2011 Europe € 90,000 2010 € 110,700 Own research
(€ 60,000-220,000) (€ 73,800-
270,600)
CE Delft 2011 Europe € 44,800 2008 € 55,700 HEATCO
Concawe et al. 2012 Europe €,9.000-13,000 2012 € 10,100-14,600 NEEDS
Quinet et al. 2013 France € 115,000 2010 € 141,400 OECD
Holland 2014 Europe € 135,000 (mean) 2005 € 199,100 NewExt
€ 58,000 (median) (mean)
€ 85,500
(median)
Chanel & Luchini 2014 France € 140,000 2001 € 250,400 Own research
Ricardo-AEA et al. 2014 Europe € 43,000 2010 € 52,900 NEEDS
CE Delft 2017 Netherlands € 70,000 2015 € 71,900 Range of
(€ 50,000- (€ 51,400- studies
€ 100,000) 102,700)
Similar values are presented in (ExternE, 2005), where a value of € 50.000 per VOLY (2005
prices) is recommended. (HEIMTSA, 2011) advises values that are higher, in the range of
€ 60,000-€ 90,000 (2010 prices). The VOLY-value used by the European Commission for
health damages as a result of environmental pollution in the impact assessment reports is
€ 50,000-€ 100,000. Likewise, VOLY values are recommended in (CE Delft, 2017) ranging
from € 50,000-€ 100,000, with a central value of € 70,000 (2015 prices). The French
government recommends a VOLY of € 115,000 (2010 prices) for use in cost benefit analysis
(Quinet et al., 2013). Slightly higher values are presented in (Holland, 2014), where the
median and mean VOLY values are € 58,000 and € 135,000 respectively (2005 prices).
However, it should be noted that the values mentioned in (Holland, 2014) are based on
(NewExt, 2004), where the willingness to pay was elicited from questions framed around
changes in mortality rate, rather than changes in life expectancy. This is not appropriate if
the VOLY is used for non-instantaneous deaths.
Other literature has presented us with values both near the lower end of the confidence
interval, as well as near the higher end. It has, for example, been argued that the
distribution of VOLY-values elicited from stated preference studies is broadly spread and
skewed, which implies that using the mean or median values is not a robust approach.
A ‘maximised societal revenue’ value is arguably better (Concawe et al., 2012).
________________________________
48
Adjustments were made for inflation and growth in GDP per capita, using an income elasticity of 0.8.

This approach proposes a simple flat fee, where only those who express a WTP that is
higher than or equal to the fee would need to pay. The flat fee is chosen so as to maximise
the revenue from the survey population. The total revenue of the fee becomes the VOLY.
The major advantage of this method is that it takes into account the full distribution of
expressed WTP values and is less sensitive to extremes. This method results in VOLY-values
between € 9,000 and €13,000 (2012 prices), considerably lower than other values.
Recent criticism of stated preference studies revolves around the fact that estimates are an
underestimation of the true value if they only take the effect on one’s own life expectancy
into account (Mouter & Chorus, 2015). In France, a WTP study revealed that many people
want cleaner air, not only for themselves but also for friends and family (Chanel & Luchini,
2014). If the effects on loved ones is also taken into account, the VOLY could increase to
€ 140,000 in 2001 prices (Chanel & Luchini, 2014).
Furthermore, studies have shown that the method with which the study is administered
affects the height of the valuation. Questionnaires administered via websites lead to
3-5 times lower values than questionnaires that are administered face-to-face (Istamto et
al., 2014). Furthermore, in general discrete choice experiments lead to a higher VOLY-value
than WTP studies (such as NEEDS) (Bijlenga, et al., 2011).
Taking all of the above into consideration, our literature review has revealed that an EU28
VOLY of € 70,000 (2016 prices) is not unreasonable. Excluding outliers we can see two peaks
in the distribution of VOLY values. The first peak around € 65,000 and the second peak
around € 130,000. As we know that the distribution of values for the valuation of health is
skewed, with median values lower than the mean values, we propose the use of € 70,000
(2016 prices) as a central EU28 VOLY. We will adjust this EU28 VOLY to national VOLYs, the
same way we do for the VSL.
A.3 Recommended approaches

In this Handbook we apply an EU28 average VSL of € 3.6 million (2016 prices), which we
differentiate to the country level, following (OECD, 2012). The EU28 VOLY is € 70,000 (2016
prices). We differentiate both values to the individual country level. For a full list of VSL
and VOLY values for each country we refer to the database.

B Detailed assessment accident
costs
B.1 Introduction
This annex presents more information on how to calculate accident costs, supplementing
Chapter 3. First we discuss the components of accident costs in more detail in Section B.2.
Section B.3 explores the total and average accident costs in more detail, describing the
approach in the previous Handbook and examining new evidence and providing updated
values. Lastly, Section B.4 provides an updated assessment of marginal external accident
costs.
B.2 Detailed discussion on accident cost elements

Accidents occur in all forms of traffic, both within and between traffic modes, as well as
between the same or different vehicle types. In recent years, considerable attention has
been placed on reducing the number of accidents in the EU. This is particularly true for
road accidents, where the European Commission adopted the ambitious Road Safety
Programme. Although the prevalence of accidents has generally fallen in recent years, the
costs of accidents still constitute a substantial part of the total external costs of transport.
Although there is no harmonised definition of external accident costs, we define them as

the social costs of traffic accidents that are not covered by risk oriented insurance
premiums in this edition of the Handbook. The insurance system therefore determines the
share of the accident costs that are considered internal. Economic theory suggests that for
true internalisation of external costs to happen, the marginal costs should be paid for by
the causer of those external costs. However, most insurance policies are not, or are only
slightly based on the driven kilometres. Furthermore, insurance policies are only somewhat
based on risky driving, as the insurance premiums only increase after an accident has
happened. Therefore, arguably insurances can only be seen as a way to partly internalise
the external costs (VU, 2005). Some studies at the national level therefore argue that costs
that are insured, should not be considered fully internalised (CE Delft & VU, 2014).
However, to fully assess which part of the insured costs are internal and external, a lot of
detailed data on national insurance schemes is needed, which is currently unavailable at
the European level. Therefore, this Handbook considers costs that are insured as fully
internalised.
There are five main components of accident costs:

— Human costs: This is a proxy for estimating the pain and suffering caused by traffic
accidents in monetary value. Traffic participants are assumed to internalise their own
human costs, but they consider the human costs of others as external.
— Medical costs: These are the costs of the victim’s medical treatment provided by
hospitals, rehabilitation centres, general practitioners, nursing homes, etc. as well as
the costs of appliances and medicines. This Handbook assumes 50% of the medical costs
are external.
— Administrative costs: These are the costs covering the expenses of the deployed police
force, fire service and other emergency (non-medical) services that assist at the crash
location site. This Handbook assumes 30% of the administrative costs are external.

— Production losses: These costs represent the lost output per casualty, due to reduced
working time (hospital stay and revalidation) and the human capital replacement costs.
This component is assumed to be partly internalised by traffic participants through
insurance.
— Material damages: This consists of the monetary value of damages to vehicles,
infrastructure, freight and personal property resulting from accidents. This component
is assumed to be fully internalised by traffic participants through insurance.
— Other costs: This category covers all other costs not incorporated into any of the above
categories. It includes, amongst others, the costs of congestion, vehicle unavailability
and funeral costs. These costs are assumed to be fully internalised by traffic
participants.
There are three main ways to allocate the total external accident costs (the sum of the
aforementioned components) to the different vehicle categories. This can be done based on
the responsibility perspective, the monitoring perspective or the damage potential
perspective.
— Monitoring: The monitoring perspective involves allocating the casualties of an accident
to the vehicle type they were using when the accident occurred. In case of an accident
between a car and a truck where two individuals seated in the car die and the truck
driver sustains an injury, the two fatalities would be allocated to the car and the injury
would be allocated to the truck. This is the classic way in which accident statistics are
reported.
— Responsibility: The responsibility perspective involves allocating the costs of the
accident to the party responsibly for causing the accident. This method would imply
that if the truck in the above example was responsible for causing the accident, both
fatalities and the injury would be allocated to the truck. No costs would be allocated to
the passenger car. This is arguably the fairest way to allocate the accident costs.
Unfortunately, accident statistics at the EU level do not contain information on
responsibility, although some national databases do (e.g. Germany).
— Damage potential: This perspective involves allocating all victims in a certain vehicle to
the other vehicle involved in the accident. This approach is favoured over the other two
methods for two reasons. Firstly, the accident statistics with differentiations on
responsibility are not available for all countries within the scope of this study.
Secondly, as argued by (CE Delft & VU Amsterdam, 2004) the ‘responsibility’ for an
accident in a moral and causative sense does not only lie with the party ‘in error’, but
may also lie with the party that, legally speaking, did not commit an error at al.
After all, certain activities undertaken by society are accompanied by a certain intrinsic
risk, even if no ‘error’ was made. Thus, even if vehicle drivers comply to traffic
regulations, there is still the mutual danger to which drivers continually expose each
other. The heavier and faster a vehicle, the bigger its damage potential, as it exposes
other road users to greater danger. In our example, the costs of the two fatalities that
fell in the car would be allocated to the truck, and the injury in the truck would be
allocated to the car.
This study employs the intrinsic risk perspective for accidents occurring within road
transport. For accidents between different modes, such as accidents involving a train and a
car, the casualties are allocated according to the responsibility perspective. In this study,
these types of accidents only occur between road and rail, at level crossings. For these
types of accidents it is known that they are almost always caused by the road user (Jonsson
& Björklund, 2015).

B.3 Assessment of total and average accident costs
B.3.1 Recommended approach previous Handbook

The first Handbook calculated average external accident costs using a top-down approach
by multiplying accident statistics per vehicle type by the unit cost per accident and a factor
representing what percentage of the unit costs were considered external. The unit cost per
accident was based on cost allocation to different vehicle categories from a causation
perspective. The percentage of the costs that are considered external was based on
information from the insurance system. The 2014 Handbook only looked at marginal costs.
The previous Handbooks use the VSL as a proxy for the human costs. Fatalities were valued
at the full VSL. Serious injuries were valued at 13% of the VSL, and slight injuries were
valued at 1% of the VSL.
Direct and indirect economic costs for fatalities were valued at 10% of the VSL. These costs
were considered fully external. Furthermore, both previous editions of the Handbook did
not calculate the total and average costs of IWT accidents and maritime accidents.
B.3.2 New evidence

In general, the major new source of literature since the last edition of the Handbook is the
SafetyCube study (Wijnen, et al., 2017). SafetyCube collected information on the national
methods to estimate accident costs in 31 European countries to be able to provide crash
cost estimates at the country level. For each of the cost components, the method of
calculation was checked against the recommended international guidelines created in
(Alfaro, et al., 1994). Value transfer was conducted to fill in gaps (for countries that did not
measure certain cost components) or to estimate comparable values according to the
international guidelines (for countries that did not apply the recommended methods).
According to international guidelines for estimating costs of road crashes (Alfaro, et al.,
1994) (Wijnen, et al., 2017) there are three types of methods that can be used to estimate
the costs of road crashes. Each of the components of accident costs outlined above can be
estimated using one of these three methodologies, which are described below.
— Restitution costs (RC) approach: This approach identifies the costs of the resources
that would be needed to restore accident victims and their families and friends to the
situation where the accident hadn’t happened. In general, these costs can be
interpreted as the direct costs from a crash, such as the costs of medical treatment and
vehicle repair (ERSO, 2006). This is the appropriate method to estimate medical costs,
material damages and administrative costs according to international guidelines.
— Human capital (HC) approach: This approach measures the value for society of the
productive capacities that are lost in road accidents. This is usually based on the added
value that a person produces for society. This is the appropriate method to estimate
production losses according to international guidelines.
— Willingness to pay (WTP) approach: This method estimates costs based on the amount
that individuals are willing to pay for a risk reduction. Based on WTP studies, the value
of a statistical life (VSL) can be estimated, which can be used to calculate human costs
(or risk value) of fatalities. As there is no market price for immaterial impacts such as
the risk value, this method is the appropriate method for calculating the risk value.
As we assume that material damages and other costs are fully internalised through relevant
insurances (see Chapter 3) we will not discuss them in detail in this section.

Human costs
Using a willingness-to-pay (WTP) approach is generally acknowledged to be the most
theoretically sound approach in estimating human costs. It is both recommended in
international guidelines for accidents (Wijnen, et al., 2017; Alfaro, et al., 1994), as well as
considered good practice in European studies (HEATCO, 2006) and national studies (Wijnen
& Stipdonk, 2016) investigating accidents. In WTP studies respondents are asked how much
they are willing to pay for a reduction in the risk of being in a fatal traffic accident.
Fatalities
Two approaches can be used to determine the willingness to pay for a reduction in the
risk of getting killed in an accident, the ‘stated preference’ (SP) method or ‘revealed
preference’ (RP) method. The results of either study can be used to derive the VSL, the
value attached to a (prevented) fatality. The difference between RP and SP studies is that
RP values risk reductions on the basis of actual behaviour, such as through the purchasing
decisions of safety provisions (e.g. airbags, seatbelts). In SP methods, questionnaires are
presented to individuals, in which they are asked directly or indirectly how much they are
willing to pay for safety, from which a VSL can be deduced (see Annex A).
The most recent literature, e.g. (Wijnen, et al., 2017) recommends the SP method as the
most appropriate and scientifically sound approach to estimate the VSL in the context of
road safety. One of the main advantages of the SP method is that it is not dependent on
information on actual consumer behaviour, and therefore can be applied more broadly than
RP. As consumers are usually not fully aware of the risk reduction resulting from safety
devices, SP provides an edge over RP. This is because the former allows for the provision of
this information, helping respondents to understand (small) risk reductions in the correct
manner. These above reasons explain why SP methods are most commonly used in the
context of road safety in Europe. In regions outside of Europe, particularly in North-
America, RP methods are more commonly used (Lindhjem, et al., 2010). A more detailed
discussion on the advantages and disadvantages of SP can be found in (De Blaeij, et al.,
2003).
It is important to note that literature on the VSL takes a cautious approach and assumes the
VSL that elicited from people’s WTP for safety includes both human costs and the future
consumption that is lost when people abruptly lose their life. Therefore, to avoid double
counting with gross production loss (gross production loss consists of net production loss and
consumption loss), the consumption loss needs to be subtracted from the VSL to reach
human costs, or net production loss (instead of gross production loss) should be used.49
In this study we subtract the consumption loss from the VSL to avoid this double counting.
The consumption loss of fatalities that we use in our calculations is based on the number of
life years lost and annual (market) consumption from Eurostat.50
________________________________
49 See section on production loss in Section B.3.2 for more information.
50 Eurostat data reveals that the average age of someone sustaining an injury in a road accident is roughly 40.
In the EU28 a person aged 40 in 2016 is expected to live for another 42 years. The average consumption
expenditure per capita in the EU was € 15,900 in 2016. Therefore, the consumption loss of a fatality in the
EU28 can be calculated by multiplying the annual consumption loss with the number of life years lost
(€ 15,900 * 42 years = € 667,800). This consumption loss needs to be deducted from the EU28 VSL to reach the
EU28 human costs of a fatality (€ 3.6 - € 0.7 = € 2.9 million). This exercise is carried out with consumption
expenditures differentiated to the national level, resulting in human costs at the national level.

Although the international guidelines recommend the use of WTP to calculate human costs,
it is important to note that there are numerous other approaches (e.g. (World Bank, 2005)).
Such approaches are, for instance, based on the financial compensation that is awarded to
(the relatives of) road casualties in courts or by law (statutory values), the public
expenditures on improving (road) safety or the premiums people pay for life insurance.
Although there are countries that use such approaches to estimate human costs, these
methodologies have severe limitations. The main limitation is the fact that they are not
based on the valuations of the road users themselves, which conflicts with economic
welfare theory. This is the reason why the WTP method is recommended in international
guidelines (Wijnen, et al., 2017; Alfaro, et al., 1994), and why it is used in the majority of
studies into the cost of accidents, such as (Wijnen & Stipdonk, 2016; HEATCO, 2006; Infras,
CE Delft, ISI & University of Gdansk, 2008; Ricardo-AEA, TRT, DIW Econ & CAU, 2014).
In this Handbook we follow the international guidelines and use the WTP approach to
estimate the VSL, which is an input for the human costs of fatalities. We deduct the
consumption loss from the VSL to reach the human costs of fatalities (see paragraph on
production loss). In line with earlier editions of the Handbook, we assume that drivers
completely internalise their own (potential) human costs, but that the human costs of the
other traffic participants are 100% external to them (see Chapter 3).
Injuries
The WTP approach can not only be applied to the human costs of fatalities, but also to
calculate the human costs of injuries. However, it must be noted that because of the
diversity in severity and duration of the consequences of injuries, the human costs of
injuries can be very diverse (CE Delft, 2016; InDeV, 2016). A simple way to express the
human costs of injuries is as a percentage of the VSL. This was first developed in (HEATCO,
2006) and (UNITE, 2003), where human costs were assumed to be 13% of the VSL for serious
injuries and 1% of the VSL for slight injuries.
For an overview of the percentages of VSL that the literature recommends for different
injury categories, we refer to the table below, which is based and builds on (Sommer, et
al., 2007). It is important to note that direct comparisons between most studies are not
ideal, because many studies differ in their definition of a ‘serious’ or ‘slight’ injury.
The table shows more detailed injury categories than we use in this Handbook.
Table 77 – Overview of the literature on percentages of the VSL used for serious and slight injuries
Study Country/ Serious Serious injury Average Slight injuries

Region injury with with serious
lasting temporary injury
impairment impairment
Finland official Finland 45.7% 0.5% 0.1%
Finland official Finland 45.7% 0.5% 0.1%
Sweden official Sweden 15.4% 0.7%

UK official UK 11.4% 0.9%
Norway official Norway 55.2% 16.7% 2.9%
(Jones-Lee 1995) UK 15.1–87.5% 5.5–23.2%
(ECMT, 1998) Europe 13% 1%
(Trawén, et al., 1999) Sweden 13.3–40.4% 0.5–32.1%
(Persson, et al., 2000) Sweden 40% 11% 16% 1.5%
(Evans 2001) UK 11% 0.9%

Study Country/ Serious Serious injury Average Slight injuries
Region injury with with serious
lasting temporary injury
impairment impairment
(Persson 2001) Sweden 40.4% 13.3% 0.9–1.8%
(UNITE 2001) Europe 32.0% 9% 13% 1%
(Ecoplan, 2002) Switzerlan 32.0% 9% 1%
d
(Goodbody, 2002) Ireland 13.9% 1%
(HEATCO, 2006) Europe 13% 1%
(Sommer, et al., 2007) Switzerlan 32% 3.5–15% 1%
d
(Hensher, et al., 2009) Australia 3–5% 0.9–1.2% 0.26–0.32%
(Institute of Transport Norway 51.1% 15.4% 1.8%
Economics, 2010)
(Carlsson, et al., 2010) Sweden 28.6%
(DfT, 2012) UK 13.9% 1%
(Österreichischer Austria 13% 1%
Verkehrssicherheitsfonds &
BMVIT, 2012)
(Commissariat général à la France 15% 2%
stratégie et à la prospective,
2013)
(Ministry of Transport, 2013) New- 10% 0.4%
Zealand
(Intraplan & Planco, 2014) Germany 13% 1%
(SWOV, 2014) Netherlan 12%
ds
(Ricardo-AEA, TRT, DIW Econ & Europe 13% 1%
CAU, 2014)
(BfU, 2015) Switzerlan 37.3% 3.5–16.3% 0.4%
d
(B,S,S. Volkswirtschaftliche Switzerlan 33.6% 0.5–3.3% 0.03%
Beratung AG, 2015) d
(iRAP, 2015) Worldwide 25%
(Trafikverket, 2015) Sweden 16.6% 0.65%
(Wijnen & Stipdonk, 2016) Worldwide 10-16% 0.9–1.6%
Source: Based on (Ecoplan, 2016). (Sommer, et al., 2007).
Human costs on the MAIS scale

As mentioned in the textbox in Section 3.2, the EU decided that data on traffic injuries would be collected using
a new common definition in 2014. The previous definition distinguished between serious and slight injuries.
Since 2014, accident statistics use the MAIS definition, which has 6 categories. MAIS 1 and MAIS 2 roughly
correspond to the old definition of a slight injury. MASI 3, MAIS 4 and MAIS 5 roughly correspond to the old
definition of a serious injury. MAIS 6 is defined as an unsurvivable injury, and therefore cor responds to the old
definition of a fatality.
There is almost no literature available on valuing the human costs of injuries according to the new EU
definition. One study that we found looking at the valuation of MAIS categories is (Blincoe, 2015). This study
presents an average valuation of quality of life lost to crash victims for all 6 MAIS categories. However, this
method is based on QALYs lost through traffic injuries, rather than on the WTP method to approximate the

human costs, which is the recommended approach in this Handbook. If one expresses the costs for those
categories as a percentage of a fatality (see Table 78), the percentages are shown not to differ significantly
from the percentages recommended by the (US Department of Transportation, 2015) for the AIS 51 (last column
of Table 78).
To approximate values for the human costs of the traffic injuries based on the new MAIS classification, we
express the costs of all other injury categories as a percentage of MAIS 6 (fatality) based on (Blincoe, 2015).
We can then use these percentages of the fatality value as a proxy for the other injury categories and apply it to
our valuation of the human costs of a fatality (with consumption loss already deducted). The table below shows
the human costs of the MAIS injury categories according to this method . Please note that this is a very rough
first estimate, which certainly deserved further research in the future.
Table 78 – Overview of human costs for new EU injury definition, EU28 aggregate value
Injury category Human costs from Costs as a Human costs Recommended

(Blincoe, 2015) percentage of a based on WTP fraction of VSL
using QALY (in fatality (€2016 ) recommended based
$2010 ) on AIS-scale
MAIS 1 23,241 0.3% 8,724 0.3%
MAIS 2 340,872 4.4% 127,949 4.7%
MAIS 3 805,697 10.4% 302,424 10.5%
MAIS 4 2,037,483 26.3% 764,783 26.6%
MAIS 5 4,578,525 59.1% 1,718,581 59.3%
MAIS 6 (Fatality) 7,747,082 100.0% 2,907,921 100.0%
A second way to measure the human costs of injuries is based on Disability Adjusted Life
Years (DALYs). DALYs are a measure for impact of injuries on quality of life that combines
impact on mortality (fatalities) and morbidity (injuries). An estimate of the monetary value
of a DALY can be made based on the willingness to pay for quality of life (see for example
(Bobinac, et al., 2013)) or based on the VSL (see for example (Hirth, et al., 2000)).
Studies show that the monetary value of a DALY in the Netherlands can be found in the
range of € 60,000-€ 80,000 (RIVM, 2014; RVZ, 2006), although there have also been studies
suggesting the value exceeds € 100,000 (RIVM, 2006). The major advantage of the DALY
approach is that it takes into account the impact of injuries on quality of life, whereas the
WTP studies looking at injuries often use a very limited number of broad injury categories
from which the valuation of injuries is elicited. However, to date, the only studies that
have used DALYs to calculate the human cost of injuries have been conducted in the US
(Blincoe et al., 2014).
For the reasons mentioned above, this Handbook values the human costs of injuries using
the WTP method at 13% of the VSL for serious injuries and 1% of the VSL for slight injuries.
Although the consumption pattern of an injured person may change for the time that they
are injured, it does not completely disappear (unlike for fatalities). Therefore, we do not
subtract any consumption loss from the human costs for injured people.
________________________________
51
Please note the subtle difference between the AIS and the MAIS.

Medical costs
Common international practice uses the restitution cost method to estimate these costs.
This method uses the actual costs of medical treatment, such as the costs of ambulance
trips, overnight hospital stays and non-hospital treatment. This bottom-up approach to
medical costs implies that for each aspect of the medical costs (e.g. ambulance trip,
hospital day, rehabilitation centre day, treatment by GP, etc.), data on both the number,
and the costs of this aspect is needed (Wijnen, et al., 2017; InDeV, 2016).
In principle, this is relatively straightforward. However, in terms of data requirements this

method is very demanding. Data is needed on the number of ambulance and helicopter
trips, the length of the stay in hospital, the number of polyclinic visits, the number of visits
to GPs and physiotherapists, the number of accident victims admitted to a rehabilitation
centre, nursing home or receiving home care, etc. Ideally, this information should be
available per injury severity category. Furthermore, cost estimates of each of these aspects
of treatment should be available. In some countries, models have been developed to
calculate the medical costs of injuries, which include most of these data (e.g. (Polinder, et
al., 2016) for the Netherlands).
In (Wijnen, et al., 2017) information on the social medical costs of road crashes in 31
European countries was collected. Subsequently, a standard value for the medical costs is
calculated according to the recommended method from the international guidelines.
These standard values are € 5,430 for fatalities, € 16,719 for serious injuries and € 1,439 for
slight injuries (2015 prices). These costs are the total medical costs. However, part of the
medical costs are internalised through health insurance. Unfortunately, there is no EU-wide
data available on the percentage of health care costs that are covered by insurance. In the
Netherlands, 65% of the medical costs is insured and therefore internalised, whereas 35% is
external (CE Delft & VU, 2014). In Switzerland medical costs are 90% external (Ecoplan &
Infras, 2014). Due to lack of more detailed EU data on this and the fact that medical costs
are a relatively minor cost component, we assume that 50% of the medical costs are
internalised, resulting in 50% that is external. To calculate medical costs at the country
level we update the costs provided in (Wijnen, et al., 2017) to the 2016 level and apply a
PPP correction per country.
Administrative costs
There are four major component of administrative costs: police and fire service costs,
insurance costs and legal costs.
The police and fire service costs can be calculated using a bottom-up methodology or a
top-down methodology. The bottom-up approach multiplies the costs per unit (e.g. police
costs per hour or fire service costs per crash) by the time spent on a crash or the number of
crashes. The top-down approach estimates the share of costs related to transport accidents
of the total police or fire service costs. This is done using information on police or fire
service time spent on transport accidents. Both methodologies are theoretically sound.
The administrative costs for vehicle insurance or health insurance relate to the cost of
personnel handling claims, including overhead costs. These costs need to be attributed to
transport accidents, because no insurance would be needed if there were no road
accidents. The administrative insurance costs can be deduced from insurance branch
statistics. Alternatively, the insurance costs can be calculated from the costs that are
specifically related to handling insurance claims from traffic accidents. This would be done
based on the number of claims and the costs per claim.

Lastly, the legal costs can be estimated using a top-down or a bottom-up approach.
A bottom-up approach uses cost per unit (e.g. prosecuted offenders, lawsuit, prisoner) and
multiplies it by the number of units. In a top-down approach the proportion of total legal
costs (costs of prosecution, lawsuits and imprisonment) is determined based on certain
variables (e.g. prosecuted people due to a traffic accidents as a proportion of total number
of prosecuted people).
Wijnen, et al., (2017) collected information on the total administrative costs of road
crashes in 31 European countries. Based on the recommended methods from international
guidelines (Alfaro, et al., 1994), a standard value for the administrative costs was presented
for fatalities (€ 6,346), serious injuries (€ 4,364) and slight injuries (€ 1,876) (2015 prices).
These costs are the total administrative costs. In this study we look at external
administrative costs (i.e. the part that is not internalised through insurance).
Unfortunately, there is no EU-wide data available on the percentage of administrative
accident costs that are covered by insurance. In the Netherlands, 84% of the administrative
cost is insured, whereas only 16% is external (CE Delft & VU, 2014). In Switzerland
administrative costs are 44% external (ARE, 2018). Due to lack of more detailed EU data on
this and the fact that administrative costs are a relatively minor cost component, we
assume that 70% of the administrative costs are internalised, resulting in 30% that is
external. To calculate administrative costs at the country level we update the (Wijnen, et
al., 2017) costs to the 2016 level and apply a PPP correction per country.
Production losses
International guidelines recommend the human capital approach as the appropriate method
to calculate the production loss. This implies the market production loss of a casualty can
be calculated by multiplying the valuation of production per person per unit of time, by the
time that the person is unable to work due to the accident. For fatalities the period of time
the person is unable to work due to an accident is the remaining number of productive
years until retirement. For injured casualties, the remaining period may vary from a few
days of absence at work, to all remaining productive years until retirement in the case of a
person becoming permanently disabled as a result of the accident.
One major methodological discussion which arises in estimating production losses is whether
actual or potential production losses should be used. It is important to note that potential
production losses are commonly used to measure production losses (Wijnen, et al., 2017;
Wijnen & Stipdonk, 2016). This implies that the capacity to produce is what is valued,
regardless of what someone actually produces. This approach is considered a better
measure, because it is not sensitive to macro-economic conditions (e.g. labour markets or
unemployment rates). Gross national/domestic product per capita, or income are
frequently used as an indication for production losses.
Production losses also encompass non-market production loss, e.g. voluntary work. This loss
can be estimated using a shadow price that reflects the value of time (usually based on
wages) and information on the time spent on household work, child care, etc. However,
adding non-market production losses to potential production losses implies there is a risk of
double counting production losses. This is because unemployed people may be more
involved in non-market production. It is not usually common practice to explicitly include
non-market production losses in these costs.

It is important to distinguish between gross and net production loss. This difference exists
because individuals that are killed in an accident cannot consume goods and services
anymore, and injured people may temporarily consume less as a result of their injuries.
Net production loss can therefore be defined as gross production loss minus consumption
loss. It is common practice to use gross production loss as a measure for production loss
(Wijnen & Stipdonk, 2016). Therefore, if gross production loss is used in calculating
accident costs, consumption loss should be deducted from the VSL used to calculate the
human costs. This should be done to avoid double counting the consumption loss. 52
This habit stems from a cautious approach to the WTP value of the VSL. Since it is not
entirely known if individuals take into consideration the fact that if a life ends early their
consumption is also reduced, it was always cautiously assumed that individuals did take this
into account. A recent Swiss study (B,S,S. Volkswirtschaftliche Beratung AG, 2015) showed
that people do not take into account their consumption when reporting their WTP for a VSL.
This is an interesting conclusion from the first study of its kind, which suggests there is no
need for the deduction of the consumption loss from the VSL, as there is no double counting
of consumption loss in the two cost components (human costs and production loss).
The conclusions from the (B,S,S. Volkswirtschaftliche Beratung AG, 2015) study highlight the
need for further research to confirm these findings in a broader (international) context.
In this study, we deduct the consumption loss from the VSL, but highlight the importance of
further research. If we would not do this, the accident costs per fatality would be roughly
20% higher. It would make no difference in the costs for serious and slight injuries, as we
assume that these casualties do not suffer from consumption loss, although their spending
habits may change.
In this Handbook, we use gross production loss and base the values on (Wijnen, et al.,
2017), where a standard value for the production loss was calculated based on data from
31 European countries. The standard value for production losses of € 655,376, € 43,627 and
€ 2,669 for fatalities, serious injuries and slight injuries respectively (2015 prices) (Wijnen,
et al., 2017). The production loss is considered partly internalised. In a study looking at
external costs in Switzerland around 55% of the production loss was considered external
(Ecoplan & Infras, 2014). This percentage is also used in this Handbook. To calculate costs
of production loss at the country level we update the costs from (Wijnen, et al., 2017) costs
to the 2016 level and apply a PPP correction.
B.3.3 Conclusions
The different cost components of accident costs are each calculated using the approach
suggested by international guidelines.
— Human costs: These costs are calculated based on the WTP value of the VSL. The EU28
VSL is € 3.6 million and is based on the (OECD, 2012) and differentiated to the country
level (see Annex A). To reach the human costs, the consumption loss is deducted from
the VSL. The consumption loss is based on annual market consumption and life years
lost through the accident. Drivers consider all human costs of individuals inside their
vehicle as fully internal, but the human costs of individuals in other vehicles as fully
external.
— Medical costs: These costs are based on values presented in (Wijnen, et al., 2017),
which are calculated using the restitution costs method. 50% of the medical costs is
assumed to not be covered by insurance and is therefore considered external.
— Administrative costs: These costs are based on values presented in (Wijnen, et al.,
2017). 50% of the medical costs is assumed to not be covered by insurance and is
therefore considered external.
________________________________
52
See section on human costs in Section B.3.2 for more information.

— Production loss: These costs are estimated using the human capital approach and are
based on values presented in (Wijnen, et al., 2017). 55% of the costs of production loss
are considered external.
— Material damage: These costs are considered fully internalised.
— Other costs: These costs are considered fully internalised.
Compared to earlier editions of the Handbook, this Handbook includes a more detailed
breakdown of the non-human costs of accidents. In earlier editions they were all grouped
and assumed to be 10% of the VSL. However, this edition of the Handbook breaks them
down into their individual components and calculates them based on the recommended
international guidelines.
B.4 Assessment of marginal accident costs

By entering the traffic the driver of the vehicle exposes himself/herself to the average
accident risk. However, each additional vehicle may change the accident risk of other
transport users. For instance, each additional vehicle means vehicles are more likely to
encounter one another, and therefore have a higher accident risk. Simultaneously, extra
vehicles lead to busier roads, which leads to more careful driving or lower vehicle speed.
This could lower the accident risk. Therefore, the marginal accident costs may be
substantially different from average accident costs (higher or lower), depending on the
type of road (motorway, urban or other) and the traffic flow (dense or thin).
Marginal accident costs are only calculated for road transport. Rail, aviation, inland
waterway and maritime marginal accident costs are assumed to be equal to the average
accident costs, as these modes are scheduled services. This implies that the size of the
traffic flow is not a determinant in the accident costs. This is in contrast to road transport
where the accident risk is highly dependent on how busy a certain road is.
B.4.1 Recommended approach previous Handbook

Earlier editions of the Handbook calculate marginal external accident costs based on a
bottom-up approach. This approach multiplied four factors with each other (traffic volume,
risk elasticity, unit cost per accident, and percentage of cost that is considered external).
The risk elasticity can be defined as the risk of an additional accident at the actual level of
traffic volume.
In the second edition of the Handbook (2014) the aforementioned approach was formalised
in the following formula.
𝑀𝐶𝑖𝑣 = 𝑟𝑖𝑣 (𝑎 + 𝑏 + 𝑐)(1 + 𝐸𝑖𝑣 ) − 𝜃 𝑣 𝑟𝑖𝑣 (𝑎 + 𝑏)

With
𝑋𝑖𝑣
𝑟𝑖𝑣 =
𝑄𝑣𝑖
And
𝜕𝑟𝑖𝑣 𝑄𝑖𝑣
𝐸𝑖𝑣 =
𝜕𝑄𝑣𝑖 𝑟𝑖𝑣
Where:
𝑟𝑖𝑣 = The accident risk for each vehicle type (v) and road type (i)
𝑋𝑖𝑣 = The number of fatality or injury cases for each vehicle type (v) and road type (i)
𝑄𝑣𝑖 = The number of vehicle kilometres for each vehicle type (v) and road type (i)

𝑎= The costs due to an accident for the person exposed to the risk
𝑏= The costs for the relatives and friends of the person exposed to the risk
𝑐= The costs of the accident to the rest of society (production loss, material damages,
administrative costs, medical costs)
𝐸= The risk elasticity, which reflects how much a 1% increase in traffic (measured in
vkms) increases the accident risk in percent
𝜃𝑣 = The share of the accident costs that is internal for each vehicle category
(Social) Marginal costs can be expressed as a function of the costs per accident, the
accident risk and the risk elasticity. The latter two terms may be dependent on the vehicle
and road type, as certain vehicle or road types may be more or less dangerous than others.
To reach external marginal costs, the share of the costs that are internal still needs to be
deducted from the social marginal costs, this is done in the second half of the formula.
The risk elasticity (𝐸 ) that was used in the 2014 Handbook was a conservative estimate of
-0.25, irrespective of the vehicle and road type. This was based on (Lindberg, 2001; TML,
2010). The degree of risk internalisation (𝜃) that was used in the 2014 Handbook differs per
vehicle type, see table below.
Table 79 – Overview of values used in 2014 Handbook for the degree of risk internalisation for marginal costs
Vehicle type Degree of risk internalisation Source

Cars 0.76 (Lindberg et al., 2006)
LCVs and HGV 0.22 (Lindberg et al., 2006)
Buses 0.16 (Lindberg et al., 2006)
Motorcycles 0.8 (Fridstrom, 2011)
B.4.2 New evidence

Very little research has been conducted on marginal accident costs. As a consequence there
is little to no new literature on these costs. We therefore use the same formula as
identified in the 2014 edition of the Handbook.
For the risk elasticity it was common practice to use the same risk elasticity parameter for
each of the road types in the previous Handbooks. The value was -0.25, a value which was
not differentiated according to vehicle type or road type (Lindberg, 2001; TML, 2010).
This implies an increase in vehicle kilometres of 1% has the same effect on the average
accident rate, regardless if this is because of trucks or motorcycles, and regardless of
whether it occurs on motorways, urban roads or other roads. Although some authors have
argued that the elasticities should be corrected for each vehicle type’s traffic share,
e.g. (Fridstrom, 2011), such studies have only been carried out at the national level and
cannot be generalised to other countries. Similarly, a Swiss study by (Sommer et al., 2002)
showed that risk elasticities can be differentiated to road types. For this national study, risk
elasticities were found to be -0.5, -0.25 and -0.62 for motorway, urban and other roads
respectively. All risk elasticities used in the previous Handbook are negative, implying that
an increase in traffic (measured in vehicle kilometres) leads to reduction in the accident
risk. This is caused by two opposing effects. Firstly, an increase in traffic exposes more
drivers to the possibility of an accident, which should increase the accident rate. However,
average traffic speed also decreases with more traffic, which reduces the possibility of an
accident. The results from the study by (Sommer et al., 2002; Lindberg, 2001) confirm that
the latter effect is larger than the former, implying an increase in traffic leads to a
reduction in the accident risk. Arguably in locations where the speed is already slow

(i.e. urban roads or congested areas), the latter effect is not likely to outweigh the former.
This was argued in (CE Delft & VU, 2014) who use a risk elasticity of 0 for these roads,
implying that an extra vehicle does not significantly change the accident risk. In this study
we use the risk elasticity of -0.25 for motorways and other roads conform (Lindberg, 2001;
TML, 2010; Ricardo-AEA, TRT, DIW Econ & CAU, 2014), and a risk elasticity of 0 for urban
roads (conform (CE Delft & VU, 2014)).
For the human costs, (𝑎 + 𝑏) in the equation presented above, we base our calculations on
a different VSL than the 2014 Handbook. The new VSL values are significantly higher than
those presented in the 2014 Handbook. A more detailed discussion on the VSL can be found
in Annex A.
The degree of risk internalisation can be based on the literature (as was done in the
previous 2014 Handbook, see Table 78), but can also be calculated if detailed data on
accidents are available. The latter is the approach taken in this Handbook. This approach
was developed by (Lindberg, 2001) and assumes that a part of the accident risk is
internalised, which can be estimated by dividing the number of fatalities inside a certain
vehicle type by the number of fatalities in accidents involving this vehicle type (and
therefore also counting victims inside other types of vehicles involved in the accidents).
Based on the CARE database, such detailed data is available, and is used to calculate the
degree of risk internalisation in this Handbook, which results in the internalisation factors
as shown in Table 11.
B.4.3 Conclusions
This Handbook uses the approach as outlined by the 2014 Handbook, but updates the input
values that were used in the 2014 Handbook. The VSL used is taken from the (OECD, 2012)
(EU28 value of € 3.6 million in 2016 prices) and differentiated to the country level (see
Annex A). The other cost components are taken from (Wijnen, et al., 2017). The risk
elasticity is set at -0.25 for motorways and other roads, which is identical to the value
presented in the previous Handbook and (Lindberg, 2001; TML, 2010), and set at 0 for urban
roads, conform (CE Delft & VU, 2014). This value is set to be the same, regardless of vehicle
type. The degree of risk internalisation is differentiated to vehicle type, calculated from
the CARE database. This accident risk per vehicle type is also calculated based on the CARE
database and the transport performance data.

C Detailed assessment air pollution
costs
C.1 Introduction
This annex presents more information on how air pollution costs are calculated, giving
additional information to Chapter 4.
C.2 Detailed discussion on effects of air pollution

The present Handbook covers the following four types of impacts caused by the emission of
transport related air pollutions:
— Health effects: The inhalation of air pollutants such as particles (PM10 , PM2.5 ), nitrogen
oxides (NOx) and others lead to a higher risk of respiratory diseases (e.g. bronchitis,
asthma, lung cancer) and cardiovascular diseases. These negative health effects lead to
medical treatment costs, production loss at work (due to illness) and partially even to
death.
— Crop losses: Ozone as secondary air pollutant (mainly caused by the emission of NO x and
VOC) and other acidic air pollutants (e.g. SO2 , NOx) can lead to damage of agricultural
crops. As a consequence, an increased concentration of ozone and other substances can
lead to lower crop yield (e.g. for wheat).
— Material and building damage: Air pollutants can mainly lead to two types of damage of
buildings and other materials: a) pollution of building surfaces through particles and
dust; b) damage of building facades and materials due to corrosion processes, caused by
acidic substances (e.g. nitrogen oxides NOx or sulphur oxide SO2 ).
— Biodiversity loss: Air pollutants can lead to damage of ecosystems. The most important
damages are a) the acidification of soil, precipitation and water (e.g. by NO x, SO2 ) and b)
the eutrophication of ecosystems (e.g. by NOx, NH3 ). Damages at ecosystems can lead to
a decrease in biodiversity (fauna, flora).
The overall framework adopted for the estimation of the damage cost of pollutants is
schematically summarized in the following figure.

Figure 17 – The approach for damage cost calculations: from emission to impact and damage
Intervention Impacts
Measure Emission Concentration Damage €
(endpoints)
Nuisance, resources
A given activity leads to emissions. In the case of transport emissions, these emissions are
primarily emissions to the air (a certain amount of tyre wear can end up as emissions to
soils or water, but these have not been taken into account in this Handbook).
These emissions are subsequently transported through the atmosphere to other regions
where they are added to existing emission concentrations. This concentration then leads to
changes in ‘endpoints’ relevant to human welfare. These changes can be monetarily valued
by quantifying the amount of damage caused at the endpoints. The entire chain from
emissions, nuisance and resources through to damage in monetary terms is the subject of
the present Handbook.
C.3 Assessment of air pollution costs in the previous Handbook

The methodological approach applied in the last Handbook has been the same as in the
present Handbook, using cost factors per pollutant as input values. The main source for the
cost factors in the previous Handbook has been NERI (2010) plus the EU-project CAFE CBA.
C.4 New evidence

The following section gives an overview on the latest literature on different aspects for the
calculation of the air pollution costs.
C.4.1 General overview

The updates in the Concentration Response functions have been set up by comparing
the NEEDS outcome on the Concentration Response functions with the WHO (2013)
recommended values and approaches. This is not straightforward as both studies report
different units. Whereas the NEEDS study reports CRF functions, expressed in µg/m3, works
the WHO with Relative Risks (RR).
Most epidemiological studies report their results in terms of relative risk RR, defined as the
ratio of the incidence observed at two different exposure levels. The RR can therefore be
interpreted as the increase in percentages in the relative risk in the reported impact due to
an increase in exposure levels of 10/µg/m3. To quantify damages, one needs to translate
this RR in terms of a concentration response function, also called exposure response
function (Rabl, et al., 2014). For this one needs to know the existing risk on these incidents.
So for an RR of 1.046 per 10/µg/m 3 for Working Days Loss due to PM2.5 lung diseases, one
needs to understand how often the population already is suffering from these diseases.
The CRF can then be regarded as the product of the baseline and the Delta RR.

We have started from the table of health impacts in NEEDS as can be seen in the following
Table.
Table 80 – Concentration-response functions and external air pollution costs

Core Endpoints
phys. External
Impact per Monet Val costs per
CRF person per per case or person per
Pollutan RGF AGF [1/(µg/m3) µg per m3 per YOLL µg per m3
t Risk group (RG) value Age Groupe (AG) value ] [1/(µg/m3)] unit [Euro] [1/(µg/m3)]
primary and SIA < 2.5, i.e. Particle < 2.5µm
Life expectancy reduction - YOLLchronic PM2.5 all 1.000 Total 1 6.51E-04 6.51E-04 YOLL 40,000 2.60E+01
netto Restricted activity days (netRADs) PM2.5 all 1.000 MIX 1 9.59E-03 9.59E-03 days 130 1.25E+00
Work loss days (WLD) PM2.5 all 1.000 Adults_15_to_64_years 0.672 2.07E-02 1.39E-02 days 295 4.10E+00
Minor restricted activity days (MRAD) PM2.5 all 1.000 Adults_18_to_64_years 0.64 5.77E-02 3.69E-02 days 38 1.40E+00
primary and SIA < 10, i.e. Particle < 10µm
Increased mortality risk (infants) PM10 infants 0.002 Total 0.009 4.00E-03 6.84E-08 cases 3,000,000 2.05E-01
New cases of chronic bronchitis PM10 all 1.000 Adults_27andAbove 0.7 2.65E-05 1.86E-05 cases 200,000 3.71E+00
Respiratory hospital admissions PM10 all 1.000 Total 1 7.03E-06 7.03E-06 cases 2,000 1.41E-02
Cardiac hospital admissions PM10 all 1.000 Total 1 4.34E-06 4.34E-06 cases 2,000 8.68E-03
Children meeting
PEACE criteria - EU
Medication use / bronchodilator use PM10 average 0.200 Children_5_to_14 0.112 1.80E-02 4.03E-04 cases 1 4.03E-04
Medication use / bronchodilator use PM10 asthmatics 0.045 Adults_20andAbove 0.798 9.12E-02 3.27E-03 cases 1 3.27E-03
Lower respiratory symptoms (adult) PM10 symptomatic_adults 0.300 Adults 0.83 1.30E-01 3.24E-02 days 38 1.23E+00
Lower respiratory symptoms (child) PM10 all 1.000 Children_5_to_14_years 0.112 1.86E-01 2.08E-02 days 38 7.92E-01
Ozone [µg/m3] - from SOMO35
Increased mortality risk SOMO35 Baseline_mortality 0.0099 Total (YOLL = 0.75a/case) 1 3.00E-04 2.23E-06 YOLL 60,000 1.34E-01
Respiratory hospital admissions SOMO35 all 1.000 Elderly_65andAbove 0.158 1.25E-05 1.98E-06 cases 2,000 3.95E-03
MRAD SOMO35 all 1.000 Adults_18_to_64_years 0.64 1.15E-02 7.36E-03 days 38 2.80E-01
Medication use / bronchodilator use SOMO35 asthmatics 0.045 Adults_20andAbove 0.798 7.30E-02 2.62E-03 cases 1 2.62E-03
LRS excluding cough SOMO35 all 1.000 Children_5_to_14_years 0.112 1.60E-02 1.79E-03 days 38 6.81E-02
Cough days SOMO35 all 1.000 Children_5_to_14_years 0.112 9.30E-02 1.04E-02 days 38 3.96E-01
Abbreviations: Risk Group, RG: group within the general population with a handicap; RGF value: share of RG within the
general population; Age group, AG: groups distinguished by different age cohorts; AG value: share of
different age cohorts; CRF: concentration-response function; YOLL: Years of Life Lost; RAD: Restricted
Activity Days; SIA: Secondary Inorganic Aerosols; SOMO35: sum of ozone means over 35 ppb; WLD: Work
Loss Days; MRAD: Minor Restricted Activity Days; LRS: lower respiratory symptoms. Table constructed for
the whole of Europe.
Source: NEEDS (2008a), based on NEEDS (2007).
Below for various impact groups the relevance of these CRFs for the present Handbook is
discussed in light of the recent WHO (2013) update. First, the mortality impacts are
discussed and then the morbidity impacts are identified.
C.4.2 Mortality impacts

Mortality impacts occur because of PM2.5, NO2, and ozone pollution (also called SOMO-35,
Sum Of Means Over 35ppb, e.g. the excess of max daily 8-hour averages over 35 ppb which
is about 70 µg/m 3).
All-cause mortality PM2.5

The HRAPIE experts recommended estimation of the impact of long-term (annual average)
exposure to PM2.5 on all-cause (natural) mortality in adult populations (age 30+ years) for
cost– effectiveness analysis (Group A). A linear ERF, with an RR of 1.062 (95% CI = 1.040,
1.083) per 10 μg/m3, has been recommended — even though some recent evidence has
suggested a RR of 1.066. We observe that these RRs are practically similar to the used RR of

1.06 in the NEEDS project. As the Iref is probably nowadays slightly lower due to better
health in population due to healthier lifestyles. Therefore, our conclusion is that this value
will not be altered compared to the NEEDS estimates.
All-cause mortality SOMO 35

The NEEDS project only includes acute mortality (e.g. heart attack) with an RR of 1.003 per
10 µg/m3 compared to the normal change of having a heart attack (which was established
as 1% of population). The valuation of acute mortality is 50% higher than for chronic
impacts. WHO (2013 and 2014) provide insights that there are also chronical components
included in ozone pollution. For a population of 30 years old or older, the WHO (2013)
recommends adopting a relative risk factor (RR) of 1.014 per 10 μg/m 3 in the summer
months (April–September) for 8-hours concentration higher than 35ppb. As explained in
Jerrett et al. (2009) (Michael, et al., 2009), this may increase the CRF a factor of 9
compared to the acute impact. This is not only due to the higher RR, but also due to taking
a different incidence rate. However, the precise impact is very uncertain. In our model we
proposed to use a factor of 3.5 as a lower bound and a factor of 9 as an upper bound, so
that the average factor through which the NEEDS outcomes need to be multiplied is
equivalent to a factor of 6. Therefore, it is proposed to include the mortality impacts by
calculating them as a factor of 6 higher compared to NEEDS (2008) and by keeping the
incidence rate the same (% of population with a heart attack).
Mortality NO 2
The REVIHAAP project (WHO, 2013) reports that since 2004 a growing number of studies
have been published identifying short- and long-term correlations between NO 2 and
mortality and morbidity that come on top of the impacts of NO 2 on PM formation and of NO2
on acute mortality due to ozone formation. There is thus a third category that is not
associated with particulate matter formation or ozone formation and that has here been
added to the theme of acidification. These have not yet been included in the NEEDS
project.
At the time of the NEEDS project these impacts were not included because the team was
unable to identify sufficient studies that properly quantified these epidemiological impacts
(NEEDS, 2007). Today (2016) the situation has changed and the WHO (2013) recommends
adopting a higher CRF for NO 2 than was previously used. The HRAPIE experts (WHO, 2013)
recommend including the long-term mortality impacts (all-cause and cardiovascular) of NO2
and advise adopting a linear CRF for NO 2 for all-cause mortality, translating to an RR of
1.055 per 10 μg/m 3 (WHO, 2013). In this context the WHO (2014) notes that when employing
this RR-value in multi-emission studies due care should be taken to avoid double-counting
with respect to the impact of NO 2 on PM formation, which they state can be as much as
33%.
To make this double-counting explicit, we examined the contribution of NO 2 to the RR-value

for PM formation. For PM, NEEDS (2007) uses an overall RR for premature mortality of
1.06 per 10 μg/m 3. The relative contribution of NO 2 to PM formation can be derived from
the characterisation factors.
For characterising NO2 with respect to PM formation, ReCiPe takes a value of 0.22.
This means that 22% of the RR increase can be attributed to impacts already taken into

account under the theme of PM formation, equal to an RR of 1.013 per 10 μg/m 3.53
Assuming, in line with WHO (2014), a linear CRF for NO2-values over the 20 μg/m 3 threshold,
it can be concluded that the additional NO 2 RR-value must be 1.042 per 10 μg/m3 for
pollution in areas above the threshold level. This implies that the chronic health damage
attributable to NO 2 should be a factor 3 higher than assumed in NEEDS, based on its
contribution to PM formation.
To this factor, two additional corrections should be made:

1. The mortality applies only to people older than 30 years.
2. The mortality applies only to population living in areas with an annual mean
concentration of pollution above 20 μg/m 3.
C.4.3 Morbidity impacts

For the morbidity impacts we have consulted WHO (2014) and WHO (2013). Below we will
discuss first the morbidity impacts of particulate matter, ozone pollution and NO 2.
Morbidity impacts of PM2.5 and PM 10
Cardiac hospital admissions

The value in Rabl et al. (2014) has been taken. This is taken from Hurley et al. (2005) and
based on a RR of 1.006 per 10/µg/m3 PM10. Calculated to PM2.5 we use the factor 1.6 as in
(CE Delft, 2017), which implies that this would translate itself to a RR of 1.0096 per
10 µg/m3 PM2.5. This in turn is more or less equivalent to the recommended value of 10091
from the WHO. Therefore, our conclusion is that this value will not be altered
compared to the NEEDS estimates.
PM2.5: Net restricted activity days

The analysis in WHO (2014) is based on the same sources as NEEDS (2008) and Rabl et al.
(2016). We use the routine in the EcoSense model where the Restricted Activity Days have
been netted by subtracting the working days loss, the minor restricted activity days and the
hospital admissions due to PM2.5 pollution from the RR from WHO. We have followed this
routine and have used the values from the EcoSense model. Therefore, our conclusion is
that this value will be taken from the EcoSense model.
PM2.5: Minor restricted activity days (MRAD)

This category has not been included in WHO (2014) separately but is added to the net
restricted activity days. We follow NEEDS as the valuation of both days differs and our aim
is to include this differentiation in our calculations. Therefore, our conclusion is that this
value will be taken from the EcoSense model.
PM2.5: Working days loss

The approach and data in the NEEDS (2008) project are the same as in WHO (2014,
background paper 6). Therefore, our conclusion is that this value will not be altered
compared to the NEEDS estimates.
________________________________
53
This estimate is feasible because in ReCiPe PM formation is considered only in terms of its impacts on the
endpoint ‘human health’.

Respiratory hospital admissions
The WHO (2014) reports a RR of 1.019 for the whole population on the basis of a meta-
analysis. This is slightly lower than the RR that has been used in the NEEDS project, which
would be around 1.022 recalculated on the basis of the factor between PM 2.5 and PM10.
Since these values only differ slightly we have decided not to update this estimate.
Therefore, our conclusion is to update the NEEDS estimate with the estimate from the
WHO (2014).
Medication use and lower respiratory symptoms because of asthma

These categories relate to the costs of medication and disutility for asthmatic people from
additional coughing days. The additional medication use is valued at 1 €/day and the
disutility is valued at 38 Euro/day. The recent WHO (2014) update advises to only take
impacts on children (age 5-19) into account, reporting a RR of 1.028 for children with
asthma. In Europe, on average, 4.5% of the children suffer from asthma. Taking the
incidence rate of 17% of the days that they suffer from asthma, the ERF becomes:
0.17*(1.028-1)=0.00476 days. Our conclusion is to follow the WHO (2014) approach and
only use medication use and lower respiratory symptoms for asthmatic children. The
costs have been based on Ready et al. (2004), as quoted in Rabl et al., (2014) where we
assumed that every fourth cough day for children leads to an additional visit to the doctor.
The medical costs are then calculated as €11/day.
New cases of chronic bronchitis and COPD for adults

WHO (2014) advices to use an RR differentiated between children and adults. The RR for
adults is 1.117 and for children 1.08. There is quite some discussion on the basic incidence
rate (see e.g. (Hurley, et al., 2005)), but the WHO proposes to use an incidence rate of
18.6% for children and 0.39% of adults. The NEEDS project used an RR of 1.07 per 10/µg/m3
and an incidence rate of 0.378%. This implies that the new RR is about 70% higher.
We therefore used thus a 70% higher ERF in our modelling. In addition, WHO (2014) advices
to use this factor for all population older than 18, whereas NEEDS used this impact only for
those 27 and older. Therefore, our conclusion is that the NEEDS underestimates the
recent WHO Guidelines. We have therefore updated our estimates using a 70% higher
estimate. One should notice that the WHO classifies this information with a ‘B’ label
indicating that these impacts are more uncertain than other impacts. We have also decided
not to include potential new cases of chronic bronchitis for children (also labelled as ‘B’, as
the unit in which this indicator is not an endpoint in the NEEDS modelling effort).
Morbidity impacts of ozone (SOMO-35)
Hospital admissions
WHO (2014) reports hospital admissions from ozone both for respiratory and cardiac
diseases. NEEDS (2008) has only used respiratory diseases. The RR used in NEEDS for
respiratory diseases is very similar to the one proposed in WHO (2014). Therefore, our
conclusion is to follow WHO and extend this category by including cardiac hospital
admissions.

Minor restricted activity days
The background studies and assumed RR is the same for NEEDS (2008) and WHO (2014).
Therefore, our conclusion is that this value will not be altered compared to the NEEDS
estimates.
Medication use, lower respiratory symptoms and cough days

These impacts have not been included in WHO (2014). We propose here to follow WHO
(2014) and not include these symptoms in the cost calculations.
Morbidity impacts of NO 2
Morbidity impacts of NO 2 have not been included in the NEEDS project as scientific evidence
was not yet overwhelming as to the chronic impacts from NO 2 pollution. WHO (2013)
recommends including these in cost-benefit analyses.
Prevalence of bronchitis in asthmatic children

For calculation of the impacts of bronchitis, we follow the same routine as in the impacts of
PM2.5 on bronchitis and medication use for asthmatic children (see point F above), assuming
a European average of 4.5% of children are being asthmatic. The additional costs of NO 2
pollution is very small.
Hospital admissions respiratory problems.

We follow the same routine as in Hurley et al. (2015) where the estimated baseline of
hospital admissions related to respiratory problems is 617 per 100.000 inhabitants.
The CRF and estimated baseline rates can be linked to provide an impact function:
Annual rate of attributable emergency respiratory hospital admissions
= background incidence rate (617/100,000) × change per 10 µg/m3 NO2 (1.8%)
= 7.03 (95% CI 3.83, 10.30) per 10 µg/m3 PM10 per 100,000 people (all ages)
Also here, the additional impact of NO 2 on hospital admissions is very small and does not
influence the final results.
Outcome health effects

The following table presents the adopted changes from the NEEDS project for the EU
population. All cells in green (CRF functions) and orange (population) are adaptations from
the original NEEDS project.

Table 81 – Changes compared to the NEEDS project
risk Age
group RGF Group CRF
Core Endpoints pollutant (RG) value (AG) AGF value [1/ug/m3] unit
Primary and SIA < 2.5 i.e. Particle < 2,5 um
Life expectancy reduction - YOLLchronic PM2.5 all 1 Total 1 6,51E-04 YOLL
netto Restricted activity days (netRADs) PM2.5 all 1 MIX 1 9,59E-03 days
Work loss days (WLD) PM2.5 all 1 Beroepsbevolking
0,4131472 2,07E-02 days
Minor restricted activity days (MRAD) PM2.5 all 1 Adults_18_to_64_years
0,6232605 5,77E-02 days
Primary and SIA < 10 i.e. Particle < 10 um
Increased mortality risk (infants) PM10 infants 0,0019 Total 0,0102755 4,00E-03 cases
New cases of chronic bronchitis PM10 all 1 Adults_18andAboves
0,812034 4,51E-05 cases
respiratory hospital admissions PM10 all 1 Total 1 7,03E-06 cases
cardiac hospital admissions PM10 all 1 Total 1 4,34E-06 cases
Children
with
severe
medication use/bronchodilator use PM10 astma 0,045 Children_5_to_14
0,1046751 4,76E-03 cases
medication use/bronchodilator use PM10 asthmatics 0,045 Adults_20andAboves
0,7907585 0,00E+00 cases
lower respiratory symptoms (adult) PM10 symptomatic_adults 0,3 Adults 0,812034 0,00E+00 days
lower respiratory symptoms (child) PM10 all 1 Children_5_to_14
0,1046751 0,00E+00 days
Ozone [ug/m3] - from SOMO35
Increased mortality risk SOMO35 baseline_mortality0,0099 Total (YOLL = 0,75a/case)
1 3,00E-04 YOLL
respiratory hospital admissions SOMO35 all 1 Elderly_65andAbove
0,1887735 1,25E-05 cases
MRAD SOMO35 all 1 Adults_18_to_64_years
0,6232605 1,54E-02 days
medication use/bronchodilator use SOMO35 asthmatics 0,045 Adults_20andAboves
0,7907585 7,30E-02 cases
LRS excluding cough SOMO35 all 1 Children_5_to_14
0,1046751 1,60E-02 days
Cough days SOMO35 all 1 Children_5_to_14
0,1046751 9,30E-02 days
NO2 [ug/m3] -
Increased mortality risk NO2 all 0,28 Adults 30+ 0,6690976 4,41E-04 YOLL
Prevalence of bronchitis in asthmatich children
NO2 all 0,045 Children_5_to_19
0,1578638 5,25E-03 cases
Hospital admissions due to respiratory diseases
NO2 all 1 Total 1 1,11E-05 cases
Abbreviations: Risk Group, RG: group within the general population with a handicap; RGF value: share of RG within the
general population; Age group, AG: groups distinguished by different age cohorts; AG value: share of
different age cohorts; CRF: concentration-response function; YOLL: Years of Life Lost; RAD: Restricted
Activity Days; SIA: Secondary Inorganic Aerosols; SOMO35: sum of ozone means over 35 ppb; WLD: Work
Loss Days; MRAD: Minor Restricted Activity Days; LRS: lower respiratory symptoms.
Source: Adjusted from NEEDS (2008a), based on NEEDS (2007) with own recalculations of the green and orange
cells.
C.4.4 Impacts on building and materials

The impacts on buildings and materials have been approached in the same way as in
(CE Delft, 2017). For the impacts on building and materials we have opted for the high
prices scenario, because it is only in this scenario that all impacts of air pollution on
buildings and materials have been quantified.
In the Handbook the following four cost categories have been adopted:
1. Corrosion due to acidification. As in (CE Delft, 2010), the corrosive impacts of acidifying
emissions on metals, building stone and paint are based on NEEDS (2008a). NEEDS itself
derives its prices from maintenance costs per square meter for a number of different
materials. These prices have not been adjusted to the slightly higher density of

buildings in 2015 compared with 2000, because we assume this has been offset by use of
less corrosion-sensitive materials in buildings (including renovations).
2. Particulate pollution. The impacts of particulate pollution are based on Defra (2006),
who in turn derive their calculations from Rabl (1999), who analysed expenditure on
restoration of pollution-soiled buildings in fifteen French cities. Applying a regression
analysis, Rabl estimated damage costs, defined a CRF-function and calculated damage
costs as € 0.21/kg PM10 in 1998 prices. This value has been taken as the basis for EU28
restoration costs, correcting for inflation and higher populations resulting in an estimate
of € 0.3 for 1 kg PM10 in the EU28. It should be noted however that this value holds only
for primary particles, because this is the fraction containing soot. For secondary
particulates, eventual damage has been set to zero. Given a ratio of ½ for PM2.5/PM10,
this means the value for PM2.5 is € 0.15/kg PM2.5.
3. Corrosion impacts on cultural heritage. In line with the British and Belgian Handbooks,
impacts on cultural heritage have not been valued using a central value, as the
uncertainties are too great. VMM (2013b) states that these are about the same as the
restoration costs under category (1). For Paris, Rabl (1999) calculates these to be 62% of
the combined restoration costs under (2) and (3). This is in line with the approach
adopted in VMM (2013). We have therefore taken this as the upper damage value.
4. Impacts on paint and plastics. For the costs of damage to paint and plastics due to
ozone, we adopted the values reported in Watkiss, et al. (2006), who state that paint
damage is unlikely to have any major impact as average ozone concentrations are
generally too low. According to Watkiss, et al., evidence of such impacts are deriven
from mainly from US studies carried out in the late ’60s. For damage to rubber
materials empirical evidence does exist however. For the UK, a central value of
£ 85 million/year has been estimated, with a range from £ 35 million to 189 million
(1997 data). If this is compared with total 1997 UK emissions — 2,032 kilo-tonne — this is
a modest sum. Since then there has been a further decline in the use of natural rubber,
which has been largely superseded by synthetic materials. Given these facts, we opted
for a central value based on the CRF-function from the literature underpinning Watkiss
et al. (2006), giving a damage figure of € 0.1/kg NMVOC.
For the differentiation of PM2.5 emissions we have assumed that the national average is valid
for cities and can be doubled in the case of metropoles while no damage can be expected in
rural areas.
C.5 Detailed description of the methodology applied

A short overview on the methodology applied to derive cost factors for air pollutants is
given in Section 4.3.1. A more detailed description of the methodology is given in the
following section.
Since 2009 there has been no further development of NEEDS and neither of the rival model
CAFE-CBA (IIASA, 2014). It is also striking is that recent shadow price manuals for Ireland,
Belgium and Germany (under development) are still based on the NEEDS methodology owing
to its far greater transparency. However, one cannot simply take the NEEDS values and
apply them to air pollution because the estimation results are over a decade old and many
things have changed: background concentration levels, knowledge about impacts from
pollution and the valuation framework. For that reason, adaptations to the NEEDS
framework must be made. This is possible since we have the possession of a great deal of
modelling outcomes from the NEEDS model so that we can make required changes to reflect
more recent insights.

To recalculate cost factors for air pollutants in the present study, five ‘adjustments’ (i.e.
update calculations) were made to the NEEDS results. These adjustments are broadly the
same as in the Environmental Pricing Handbook (CE Delft, 2018) but they are now applied to
the EU context. These five adjustments can be described as follows:
1. Concentration Response Functions have been adapted to the WHO (2013) study.
The taken steps taken are described in more detail below.
2. The population size and population structure (age cohorts) is based on the most recent
data from Eurostat.
3. The influence of the background concentration is estimated on the basis of the
relationship between damage and emissions for various emission scenarios from NEEDS
(2008). On this basis, by letting all other factors remain the same, we can estimate the
impact of a change in emissions on the harmfulness of these emissions. This harmfulness
is then the result of the change in the background concentration.
4. The valuation has been adjusted to the most recent insights with respect to cost of the
physical and health impacts. For human health we refer to Annex A. For the change in
valuation of ecosystems and buildings, see Section C.4.4.
5. Finally, a subdivision was made for both PM2.5 and NO2 to the population density (people
living in cities or in rural areas have different damage from pollution). For PM 2.5 a
further distinction was made to transport emissions and other sources of emissions.
For PM2.5 and NOx specific emission damages from electricity generation have also been
calculated, as this information may be relevant to estimate the damage costs of
electrical vehicles.
In the following sub-chapters, the adaptations made in those five areas are described in
more detail.
C.5.1 Changes in concentration response functions

The NEEDS project was largely based on health impact information as it was present in the
WHO (2005) study on the harmful impacts of air pollution. In 2013 and 2014, the WHO
presented a major update of the health impacts of air pollution. In the present study all the
CRFs used in the NEEDS project were case-by-case individually checked and discussions held
on whether they still reflect the latest scientific understanding. On this basis the CRFs for
especially ozone pollution (> 35 ppb) and NOx were adjusted upwards or newly impact
factors have been included. In addition, a few categories of PM 2.5 pollution were revised
(e.g. impact on asthmatic part of the population from PM2.5 pollution).
In Section C.4 a detailed account of the changes in RR that have been adopted compared to
the NEEDS estimate are given. In total we have updated about 7 of the 18 CRF functions in
NEEDS and have introduced four new CRF functions of impacts that are reported in the WHO
(2013) but have not been taken into account in the NEEDS estimates. The result is an up-to-
date and precise calculation of the impacts of air pollution on human health.
The CRFs for PM2.5 have also been applied to PM10 taking into account the fraction in PM10
that is being PM2.5. This relationship between PM10 and PM2.5 emissions is based on country-
specific emissions of both pollutants as reported by Eurostat (2016 values have been taken).
We have assumed that within the EU 28% of the population is living in areas with annual NO 2
concentrations larger than 20/µg/m3.
As for the CRFs of biodiversity impacts and crop losses we have not taken new information
into consideration compared to the NEEDS project. As for impacts on building materials, we
have followed the treatment as (CE Delft, 2018).

C.5.2 Changes in population
The size and age cohorts of the population matters for estimating the damage costs of air
pollution, especially for morbidity, since some impacts (e.g. cardiovascular diseases) only
affect elderly people and other impacts (e.g. asthma) only impact on younger people.
Therefore, we have adjusted the age cohorts in the NEEDS study with the demographic
statistics from Eurostat for the EU28.
C.5.3 Change in emission concentrations

Parts of the NEEDS model, such as the dispersion and atmospheric-chemistry models, could
not be explicitly unpacked by us. However, because there are numerous NEEDS modelling
runs available for estimating emission reduction scenarios, the underlying model structure
can to a certain extent be derived. It was opted to proceed from the 2010 and 2020
emission scenarios in the NEEDS Excel tool (as used in the Ecosense dispersion model).
Actual 2016 EU27 emissions (e.g. EU28 excluding Croatia) were then used to scale to the
difference between the 2010 and 2020 values. These results were put to and discussed with
atmospheric-chemistry experts and explanations for a rise or fall in damage costs per kg
pollutant elaborated. In this way an adjustment was made for the lower background
pollutant levels in 2015 and their influence on damage estimates. It proved that this was
particularly important for the amount of ammonia in the atmosphere. NH 3, NOx and SO2 all
react to form secondary particulates, but in the case of NO x the relationship is linear, while
for NH3 it is quadratic. Thus, as long as NH 3 do not decrease twice as fast as NO x, an
additional emission of NO x and SO2 will cause more damage because of the available
ammonia in the air.as there will be relatively more atmospheric NH 3 for the NOx and SO2 to
react with. This is the main reason that lower emissions of NO x and SO2, if unaccompanied
by an equal decline in NH 3 emissions, lead to higher damage costs per kg emission for these
pollutants. This impact is included in our estimates because of the basis in the NEEDS
modelling results.
C.5.4 Change in valuation: human health

The change in valuation of human health impacts includes the choice of the VOLY.
That issue is covered in detail in Annex A.
C.5.5 Differentiation towards source and location of pollution

The values at the level of EU28 represent average values for average emissions in the year
2015. These have been differentiated in three different ways:
— towards average values for individual countries;
— towards emissions specifically applying for the transport sector and location of pollution
(cities/rural);
— towards emissions applying for electricity generation (relevant for electrical vehicles).
This differentiation has been done by observing ratios in the NEEDS model between damage
costs of EU28 compared to the national averages, and by observing ratios in the literature
between the various sources of exhaust emissions. This yields insights into the likely
damage costs per country for transport emissions.
We, as authors and researchers, fully acknowledge that such an approach where ratios are
being used is less preferred than a new modelling effort in which the impact-pathway of
emissions through the environment is being modelled for different countries and different
heights of stack. However, this is very labour intensive trajectory that has only been

established in very large pan-European research programs, like ExternE, CASES, Newext,
CAFÉ-CBA and NEEDS. Overall over 10 million Euros of research funds have been used in
these programs resulting in the EcoSense model through which MonteCarlo simulations of
the trajectory of emissions have been made in order to estimate a damage cost per country
and per type of emission. Such an effort is not feasible in the timeline and budget of the
present project. Therefore, we have to use ratios from these bigger projects in order to
estimate the likely relationship between the calculated average EU28 damage costs and the
damage cost per type of emission per country. We also observe that such a ‘value transfer’
approach has been used more frequently in the literature (see e.g. (HEATCO, 2006); (UBA,
2012); (CE Delft, 2018). The key here is to be transparent about the modifications that have
been made to the general EU figures.
Below we will elaborate on the empirical basis of our modifications.
Differentiation towards countries

Within the NEEDS project, an Excel tool was developed. From this, we have calculated the
difference between the individual country estimate of damage costs and the EU27 average,
as was reported in the NEEDS background documentation of the Ecosense model resulting in
the Excel tool that was put online in 2008. We have used the information from the unknown
height of release, damage costs in the year of release, based on average meteorology
(assuming equivalent damage from secondary particles as to primary particles)
corresponding to emissions from all sectors. We have found information from the EU27
average and a value per individual MS. This results in a ratio for emissions of NO x, NH 3,
NMVOC, SO 2, PM2.5 and PMcoarse (e.g. PM with a diameter larger than 2.5 micrometer).
For the corresponding value of PM10 we have assumed that this is the sum of the share of
PM2.5 in the emissions of PM10 of that particular country plus the damage of PMcoarse. For the
share of PM2.5 in PM10 we have used information on the national emissions of PM2.5 and PM10
for the year 2015 in Eurostat.
Differentiation for transport emissions and location

For transport, we have used the information from Heatco (2006) that provides YOLL
estimates for transport related impacts of emissions of PM2.5. The relative risk of PM2.5
emissions in Heatco is the same as applied in our study for mortality (which explains over
70% of the damage costs of PM2.5), while the impacts on morbidity are only slightly
different. We have used this information and applied the VOLY to the YOLL estimates.
Heatco (2006) does not provide values for Romania, Croatia and Bulgaria. For Bulgaria and
Romania we have taken the average from the YOLL values of two nearby countries: Greece
and Hungary. For Croatia we have taken the YOLL as an average of Austria and Italy.
As HEATCO differentiates between the emissions from a metropole region (e.g. cities with
> 0.5 million of inhabitants) and emissions outside built areas, we use this differentiation as
well. In order to obtain an estimate for small and medium sized cities, we took the
relationship between metropole emissions and small and medium-sized cities from a
previous version from the IMPACT Handbook (Infras, CE Delft, ISI & University of Gdansk,
2008). This concludes that the impacts on small- and medium-sized cities are about 1/3 of
the impact of the metropole cities.

For NOx a differentiation between cities and rural sources of NO x emissions has been
calculated. We have taken here the assumption that 80% 54 of people living in a city is
exposed to annual NO 2 values larger than 20/µg/m3 while only 10% of people living in the
countryside are not exposed to annual NO 2 concentrations larger than 20/µg/m3 (this will be
mostly people living nearby motorways). These values have then be used to calculate a
specific value for emissions of NO x located in city or rural areas for stacks up to 100 meter
by adjusting the RGF (Risk Group Factor) in the NEEDS modelling result (see Annex B).
Although there are indications that NO2 emitted at ground level may be more dangerous
than NO2 emitted from higher stacks, we do not have information that would make it
possible to differentiate between both sources.
Differentiation for electricity emissions

For electricity emissions we have used the ratio between the average emissions (unknown
height of release) and the electricity emissions in the NEEDS project. This ratio differs per
country where more densely populated countries tend to have a higher relative impact from
electricity emissions than more sparsely populated countries. For Finland and Sweden, the
damage from electricity generation is in general similar to the damage from lower stacks in
rural areas.
C.5.6 Changes in valuation: biodiversity, crop losses and materials

To finalise and complete, the approach for the valuation (deriving cost factors) for the
other non-health effects (biodiversity loss, crop loss and material damage) is also briefly
described.
In CE Delft (2010), the value adopted for biodiversity was the average value of an EDP 55 per
m2 per annum of €2004 0.4706, based on Kuik et al. (2008). This value is the average value
from a meta-analysis encompassing a number of European countries. However, the median
value in this study is € 2004 0.0604, a factor 8 lower. This implies that the overall distribution
of values comprises relatively many high values (Kuik, et al., 2008). In a study on the
external costs of energy production, Ecofys (2014) takes Kuik’s median value rather than
the average. Generally speaking, in meta-analyses more value is attached to the median
than to the average.
On the other hand, an earlier study by NEEDS (2006) arrived at a PDF-value of € 0.45/€ 0.49
per PDF/m2, the same as the average value in Kuik et al. (2008).56 NEEDS (2006) uses the
restoration-cost approach. The Éclaire project (Holland, 2014); (IIASA, 2014) investigated
the economic value of air pollution impacts on ecosystem services, with biodiversity valued
using WTP (as with Kuik), restoration costs and revealed preferences (costs of legislation).
This project indicates that WTP-based values are conceptually the most robust, but that
data availability may be a problem. In that case, use can be made of restoration costs.
Restoration costs can also be used to validate WTP-values. Holland (2014); IIASA (2014)
report that restoration costs represent a minimum value for biodiversity, because even
________________________________
54 This is a best guess but the number can be calculated more precisely on the basis of Eurostat statistics if needed
55 Ecosystem Damage Potential, which is a slightly different measure, but (Kuik, et al., 2008) state that for all
practical applications EDP and PDF can be considered identical.
56
In (CE Delft, 2010) it was reported that the average value of NEEDS (2006) was € 0.45, which is an EU-average.
In NEEDS (2006) it is stated that minimum restoration costs in Germany are € 0.49/PDF/m2. The figure of € 0.45
is a conversion from the German price level (using purchasing power parities) to an average European price
level.

after recovery genetic information may still be lost, for example. Rabl (1999) raises the
value of NEEDS (2006) by a factor of 2 to capture the true damage. Brink and Grinsven
(2011) works with a range, multiplying the value of NEEDS (2006) by an (arbitrary) factor
of 5 to obtain an upper bound and taking the value of NEEDS (2006) as a lower bound.
This approach was also adopted by Grinsven et al. (2013). Holland (2014); IIASA (2014),
too, state that restoration costs represent the best possible estimate.
In our study we have used the Kuik et al. (2008) estimate of the WTP since this and the
restoration costs yield similar values.
Damage to agricultural crops has been added to the valuation of ecosystems. For the
valuation itself the same method was employed as in NEEDS (2008), adjusting prices to
present-day levels in the markets concerned. New prices have been based on average
European producer prices for the EU28 as reported by FAO. Prices in USD2014 were converted
to EUR 2014 using the average 2014 exchange rate and then converted to € 2015 using the
general Harmonized Index of Consumer Prices (HICP) . These prices were then weighted by
consumption of the crop concerned to determine the average price rise between 2000 and
2015. Finally, 18% VAT was added.
In Annex C.4.4 the changes in valuation of the impacts on materials and buildings are
already briefly described.

D Detailed assessment climate
change costs
D.1 Introduction
This annex supplements Chapter 5, presenting background information on the calculation of
climate change costs. First, we discuss the effects of climate change in Section D.2. Section
D.3 explores the different ways in which the costs of climate change can be assessed,
delving into detail on the differences in methodology between damage costs and avoidance
costs. It also describes the recommended approach in the previous Handbook, before
examining new evidence on the climate change costs.
D.2 Detailed discussion on effects of climate change

When greenhouse gases are emitted into the atmosphere this leads to global warming.
The (IPCC, 2013) has estimated that without concrete climate policy temperatures may be
expected to rise by up to 6°C by the end of the century. Other physical effects are changes
in precipitation patterns, resulting in different levels of average and extreme precipitation
and changes in the occurrence of extreme weather events. Such radical change will have an
important and largely irreversible impact on ecosystems, human health and societies.
These impacts of climate change will not be spread equally across the globe, but rather be
much worse in developing countries where there are fewer opportunities for adaptation
(Global Humanitarian Forum, 2009).
The IPCC’s fifth Assessment Report presents the current state of knowledge regarding the
effects of climate change (IPCC, 2013). The main costs resulting from the effects of climate
change are presented below:
— Sea level rise: An increase in temperature will result in the melting of (parts of) the
polar ice caps and other snow-covered surfaces, which in turn will lead to rising sea
levels. As a result, land that is currently used for living or agriculture will be lost, and
more effort will need to be dedicated to the protection of coastal areas.
— Crop failure: An increase in temperature and changes in average and extreme
precipitation can negatively affect agriculture in certain areas (whereas it may have a
positive impact in other areas). These changes may manifest itself on a relatively short
timescale and lead to important socioeconomic adaptation problems where famines
could occur more frequently and which may trigger migration.
— Health costs: Temperature increases may lead to more hospital admissions and
mortality as a result of heat stress. At the same time, an increase in temperature can
lead to a reduction in mortality from extreme cold weather events. Changes in
temperature can also lead to a larger dispersion of a number of disease that are carried
by parasites or insects, such as malaria. In addition, an increase in and/or worsening of
extreme weather events, such as hurricanes and heatwaves can lead to an increase in
the number of fatalities from such events.
— Damage to buildings and materials: Climate change may lead to an increase in the
frequency of extreme weather events, such as hurricanes, as well as a potential
intensification of such effects. Such weather extremes can severely damage buildings,
materials and infrastructure.

— Water management issues: Water shortages and droughts in some areas may be
aggravated due to climate change, e.g. due to further salinification of ecosystems.
Other areas may have more water available than anticipated based on historical levels.
This will be accompanied by water management problems.
— Impacts on ecosystems and biodiversity: Changes in the climate may have negative
effects on flora and fauna. Animals, and plants in particular, can only adapt to such
changes in a limited manner. Migration and the extinction of certain species will be a
likely consequence of climate change.
Each of these effects is likely to trigger human responses, which may have important
socioeconomic consequences. For instance, it has been argued that one of the contributing
factors to the Syrian civil war was anthropogenic climate change (Bennet, 2015; Verme et
al., 2016; Adelphi et al., 2015). These effects are even harder to quantify than the direct
physical costs outlined above, but are extremely important in accurately identifying climate
change costs.
D.3 Assessment of climate change costs

In general, there are two major ways in which climate change costs can be calculated: using
damage costs or avoidance costs. The damage cost approach values each of the individual
effects of climate change and sums these together. The avoidance cost approach centres
around the costs of avoiding the effects of climate change up to a desired extent
(e.g. specified in a policy target).
D.3.1 Damage costs

Damage costs are an evaluation of the total costs of climate change under the assumption
that no efforts are taken to reduce the pace of climate change. They are calculated using
the impact pathway approach which relies on detailed modelling to assess the physical
impacts of climate change. These physical impacts are then combined with estimates of
their economic impacts (Watkiss, et al., 2005; Defra, 2005). Put differently, damage costs
are the total present value of future costs and benefits related to the emission of CO2.
These damage costs are calculated with the help of climate-economic models, which merge
assumptions regarding climate effects with assumptions regarding future developments of
(the division of) income. For instance, the damage costs of sea level rise can be expressed
as the costs of land loss. Agricultural impacts can be expressed as costs or benefits to
producers and consumers, and changes in water runoff might be expressed as new flood
damage estimates. From a scientific point of view, it is desirable that all external effects
are monetised, although this is likely to be extremely complicated as certain effects
(e.g. climate feedbacks) are not yet fully understood.
Damage costs increase over time, resulting in the fact that damage costs in 2050 will be
higher than damage costs in 2020. This is because of the long lifetime of carbon in the
atmosphere, the decreasing discount rates used for future emissions due to uncertainties
about future economic development and the non-linearity in the impacts of CO 2 emissions
(Defra, 2005).

There are a number of major sources of uncertainty and weaknesses in the estimates of the
damage costs of climate change (CE Delft, 2018); (Watkiss, et al., 2005); (Burke et al.,
2016); (Botzen & Van den Bergh, 2012):
— Certain important cost categories are missing in the estimates: Economic valuation,
especially in the area of climate change is often controversial. This is a consequence of
the lack of knowledge about the physical impacts caused by global warming. Some
effects are quite certain and have been proven by detailed modelling, whereas other
possible effects, which may potentially be extremely damaging, are not yet fully
understood. They are therefore either only partially or not at all included in these
models. Classic examples of effects that are difficult to model are biodiversity losses,
the potential effects of climate change on future economic growth, political instability,
violent conflicts, prolonged flooding and behavioural change as a result of climate
change, such as climate-induced migration. Furthermore, each integrated assessment
model (IAM) that is used to calculate the damage costs differs in the effects that they
include in the model. Examples of IAM’s that are frequently used are FUND, DICE or
PAGE. Therefore, the cost estimates are highly sensitive to the IAM used. These IAMS
often rely on old studies and are in need of an update.
— Uncertainty regarding the effects of climate change: There are large uncertainties
regarding the magnitude of climate change and its consequences on temperatures, sea
level rise and extreme weather events. Potentially catastrophic effects, such as the
melting of the polar ice caps in Greenland or West Antarctica or changes in climate
subsystems such as El Niño Southern Oscillation are not, or only partly, incorporated in
the analyses.
— Uncertainty regarding the vulnerability of society: It has been argued that IAMs
underestimate the vulnerability of societies and economies to historical temperature
fluctuations, and that these models are not calibrated for higher temperature variations
(Revesz, et al., 2014; Marten et al., 2013). In addition, weather variability is much more
important for agricultural yield (and therefore security of food supply) than average
weather (Revesz, et al., 2014). This is an aspect which is not yet fully taken into
account in most IAMs.
— Uncertainties regarding the social discount rate that should be used: As the effects of
climate change occur at different points in the future, discounting should be applied.
Discounting is the process of valuing future costs and benefits in terms of their value
today (present value). There is a lot of discussion regarding the discount rate that
should be used (see e.g. Stern vs. Nordhaus). Available damage cost estimates of
climate change vary by orders of magnitude depending on the discount rate.
Defra (2005) show that damage costs can increase by a factor of 5 if a pure rate of time
preference of 1% is used, as opposed to a 0% rate. The different discount rates that can
be used reflect theoretical valuation problems related to equity, irreversibility and
uncertainty. For equity both intergenerational and intra-generational equity should be
considered.
— Risk aversion and loss aversion are not taken into account properly: In general people
are risk averse and loss averse. Risk aversion implies that when exposed to a situation
with an unknown payoff, there is a preference for a more predictable lower expected
payoff. Loss aversion implies that individuals do not value gains and losses in the same
manner. Translating this to the climate implies that humans should prefer a more costly
route that leads to lower warming with more certainty than a less costly route where
there is more uncertainty about the extent of the warming. Similarly, as humans are
loss averse they will need to be overcompensated for losses in the climate as we know

it. Most studies do not take these psychological aspects into account at all in their
climate change cost calculations (or if they do then only to a very limited extent).
— Equity weighting: Although climate change is a global problem, it will give rise to
different impacts in different countries, which each have different levels of
development. To account for these income differences equity weighting may be
applied. As the marginal utility of consumption is declining with additional consumption,
a richer person will obtain less utility from extra money available than a poorer person.
Equity weighting implies that climate change impacts are weighted differently based on
which region of the world the impacts take place, taking into account differences in the
marginal utility of consumption. Studies have shown that the use of equity weights
could increase damage costs by a factor of 10 (Defra, 2005; Anthoff & Tol, 2007).
In general there are two approaches to equity weighting (Friedrich, 2008):
• World average weights: Using world average weights entails adjusting regional
monetary values to a world average income. The value of climate damages in
Europe would be lowered, whereas the value of climate damages outside of Europe
will increase. As most of the climate change damages will occur outside of Europe,
this equity weighting approach will result in higher damage costs than damage costs
that do not use equity weighting.
• Regional/EU weights: Using regional/EU weights implies that the damages are
valued by the monetary value from the region in which GHG are emitted. This would
imply that European values should be applied to all damages caused by European
GHG emissions. This approach can be justified from an ethical point of view as
Europe would pay for (the risks of) the damages they are causing, and would result
in higher damage costs than an approach where equity weighting is not considered.
— The assessment of the baseline scenario: The state of the baseline scenario with global
long-term economic development, technological development and the associated
greenhouse gas emissions, is important to determine the marginal external costs of
additional CO 2 emissions. Some IAMs do not take into account the effect that climate
change has on labour productivity and general productivity. These, in turn, affect
economic growth, and lead to an intensification of the welfare loss through the interest
rate effect. Therefore, uncertainty regarding the baseline scenario is a source of
uncertainty in the estimates of climate change damage costs.
— Adaptation measures: Adaptation measures have the potential to lower damage costs,
however, they are currently barely taken into account in the calculation of damage
costs.
D.3.2 Avoidance costs

Avoidance costs centre around the marginal costs of achieving a certain level of
environmental quality (usually determined by policy). The method is based on a
cost-effectiveness analysis, which determines the least-cost option to achieve a required
level of GHG emission reduction. This approach has been applied and recommended in
several studies (UNITE & ExternE).
Crucial to the use of avoidance costs is the level of the emissions target that is set.
To achieve accurate avoidance costs, this level has to reflect society’s target level of CO2
reductions. Several targets are relevant at the EU level:
— A reduction of CO 2 emissions of 20% by 2020 compared to 1990 levels. This is the EU
target that has been detailed with concrete measures (e.g. the EU ETS, Effort Sharing
Decision, etc.).

— The policy target that the EU has agreed upon for 2030 (at least 40% reduction by 2030
compared to 1990 levels).Concrete measures have also recently been agreed for 2030
(e.g. revision of the EU ETS, Effort Sharing Regulation, etc.)
— The target mentioned in the Paris Agreement (below 1.5 or 2 degrees warming), which
implies a target of 80-95% reduction in CO 2 emissions by 2050 compared to 1990 levels.
From a welfare economics perspective the avoidance cost approach is not a first-best
solution (Defra, 2005) as it does not directly measure and value all impacts of climate
change. However, if the emissions target adequately reflects the preferences of society and
can therefore be used to determine society’s WTP for a certain abatement level, the
avoidance cost approach can be seen as a theoretically sound alternative. Ideally, to
indicate the social desirability of the target, the targets should be confirmed by binding
policies.
As humans are known to be risk averse, risk aversion should be taken into account in the
assessment of climate change costs. The precautionary principle57 was introduced in climate
policy to take this risk aversion into account. After all, there are certain risks of climate
change that could create very high damages in the long run (e.g. methane outbursts, loss or
reversal of the Gulf Stream). However, there are currently no methodologies that include
marginal risk aversion in the assessment of damage costs. Therefore, the only way to
include risk aversion and the precautionary principle in the assessment is through using an
avoidance based approach, assuming that the political decision of the reduction target does
take these unknown, but important impacts into account.
There are a number of important aspects in determining the avoidance cost estimates:
— The choice of target level: Avoidance cost estimates are highly sensitive to the choice
of target level. A stronger emission reduction target will imply higher costs than a lower
emission reduction target. However, the target level is also sensitive to the system to
which the target is applied. For instance, whether the target holds for all sectors or
specific sectors (e.g. transport-only) will affect the cost estimate. Similarly, costs
may differ depending on whether the target set is valid at the national level or the
supranational level. If the emissions target chosen does not reflect the preferences of
society, this methodology will not be optimal from a welfare economics perspective.
— Estimation of options for mitigation: Avoidance costs rely on the accurate prediction of
technological progress and technological breakthroughs. Such progress or efficiency
gains can reduce the costs of the technologies used in the baseline scenario, as well as
those of the additional measures. A factor that is of particular interest here is the
future development of the energy prices. With low energy prices, abatement options
will be very expensive, whereas with higher energy prices abatement options will be
cheaper.
— Estimation of baseline scenario: Avoidance costs are highly reliant on assumptions

regarding the development of GHG emissions. These mainly depend on the development
of the world economy and fuel prices, which are in turn influenced by policies (to be)
taken. Higher baseline GHG emissions imply more abatement policies need to be taken,
resulting in higher avoidance costs. Similarly, lower baseline GHG emissions, for
example due to lower global economic growth, will result in lower avoidance costs.
________________________________
57
Where there are threats of serious or irreversible damage, lack of full scientific certainty shall not be used as a
reason for postponing cost-effective measures to prevent environmental degradation – Rio Declaration on
Environment and Development.

— Different GHGs: Some measures that prevent emissions are GHG specific, therefore,
their costs are also GHG specific. Thus, the costs of avoiding CH 4 or N2O emissions
cannot always be deduced from the behaviour of those GHG in the atmosphere
compared to CO 2, for instance by comparing the GWP of the different gases.
For instance, the costs of reducing a tonne of CH 4 may not be proportional to the costs
of reducing 34 tonnes of CO 2.58 Avoidance costs are usually expressed per tonne of CO2
equivalent, rather than calculating separate avoidance costs for each gas. Numerous
studies have shown that CO 2-specific avoidance costs are lower than the avoidance costs
for CH 4 and N2O (Jakeman & Fisher, 2006; Kemfert et al., 2006; Kurosawa, 2006; Tol,
2006; Gambhir et al., 2015).
— Risk and loss aversion: This is an import issue for both avoidance and damage costs.
For avoidance costs it can only be taken into account when one assumes the political
decision underlying the reduction target takes risk and loss aversion into account.
For more detail please see Section D.3.1.
— Equity weighting: This is an important issue for both avoidance and damage costs, for
more detail please see Section D.3.1.
EU ETS
In principle, an emissions cap-and-trade scheme, such as the EU Emission Trading Scheme, can be seen as a
variant of the avoidance cost approach. However, emission trading prices only reflect social costs when the
allocated maximum emissions equal the optimal level from a welfare economics point of view (Smith & Braathen,
2015). In the EU ETS there is a surplus of allowances (EC, 2016), such that the socially optimal emissions level
cannot be reached. This will change in the future because of the Market Stability Reserve. Furthermore, the use
of the current emissions trading prices as a proxy for avoidance costs is limited to the
CO2 emissions only in the markets that are included in the EU ETS (Smith & Braathen, 2015). As there are sectors
that are not included in the EU ETS, the use of EU ETS trading prices is not likely to be representative for CO 2
emissions in sectors outside the EU ETS or for other pollutants (CH 4 or N2O). Because of these reasons, this
variant of avoidance costs will not be explored further here.
D.3.3 Recommended approach previous Handbook

In the first edition of the Handbook (Infras, CE Delft, ISI & University of Gdansk, 2008) the
use of avoidance costs was advised for the short term (up to 2020). For the long term
(2030-2050) the use of damage costs was recommended, as no global long term climate
goals had yet been established.
In the second edition of the Handbook (Ricardo-AEA, TRT, DIW Econ & CAU, 2014) the use of
avoidance costs was exclusively advocated. The target level of emissions was 450 ppm CO 2
equivalents, which corresponds to 2˚C of warming. A climate change costs per tonne of
carbon of € 90 was used.
________________________________
58
The GWP of CH 4 is 34, observing over a 100 year period.

D.3.4 New evidence climate change costs
Damage costs - Values

Since the early literature on damage costs carried out in NEEDS (Anthoff, 2007) or the Stern
Review (Stern, 2006) there have been numerous studies that have examined the damage
cost approach (Tol, 2008; Botzen & Van den Bergh, 2012; Van den Bijgaart, et al., 2013;
Waldhoff, et al., 2014; Ackerman & , 2012; IAWG, 2013; Moore & Diaz, 2015; Gillingham, et
al., 2015; Tol, 2013).
A meta-analysis conducted by Tol incorporating 211 studies revealed that there is an

enormous spread in the results of damage costs (Tol, 2008). Values range from less than
€ 1/tCO2 to more than € 500/tCO 2, with an average damage cost of € 5/tCO 259 (Tol, 2008).
However, he notes that the chances of the average damage cost exceeding € 20/tCO 2 are
less than 1%. Therefore, Tol argues, the damage costs presented in the influential Stern
Review ($85/tCO 2) (Stern, 2006) are an outlier. Many researchers have argued that Tol’s
claims are premature as damage costs are characterised by very high uncertainty (Botzen &
Van den Bergh, 2012; Van den Bijgaart, et al., 2013). (Van den Bijgaart, et al., 2013) have
attempted to take away most of that uncertainty, by modelling damage costs using only the
most crucial parameters and Monte Carlo methods. They conclude that average damage
costs amount to € 37/tCO 2, median damage costs are € 17/tCO 2 and the chances of damage
costs exceeding € 100/tCO 2 are 8% (Van den Bijgaart, et al., 2013), all of which are
significantly higher than Tol’s estimates.
In general, recent studies find higher damage costs than the early study conducted by
(Anthoff, 2007). Recent literature that use the same IAM (FUND) results in values that are
1.3-3.4 times higher than those provided by (Anthoff, 2007) (e.g. (Anthoff, et al., 2011;
Waldhoff, et al., 2014)). Studies that don’t (only) rely on FUND find values that are
between 1.8-6.9 and 2.7-17 times higher for current (< 2025) and future (± 2050) emissions
respectively (Ackerman & Stanton, 2012; IAWG, 2013).60 Only one recent study found a
lower damage cost value than Anthoff (2007), although the costs were not calculated using
one IAM, but rather as an average of the results of three IAMs (Gillingham, et al., 2015).
Anthoff’s (2007b) damage costs for emissions in the near future are negative when using
high discount rates (>1%), which is likely to be as a result of the CO 2-fertilisation effect.
However, similar recent studies (Moore & Diaz, 2015; IAWG, 2013) dispute this.
Their findings suggest these costs are positive. Lastly, when comparing Anthoff (2007)’s
values to the meta-analysis carried out by (Tol, 2013) minor differences (factor 1.7) are
observed for mean damage costs, but strong differences (factor 17) are observed for median
costs.
Avoidance costs - Values

The first edition of the Handbook used avoidance costs to calculate climate change costs for
just the short-term (Infras, CE Delft, ISI & University of Gdansk, 2008). However, since then
multiple studies on the external costs of transport have switched to using avoidance costs
for both the short-term and the long-term costs (Ricardo-AEA, TRT, DIW Econ & CAU, 2014;
CE Delft, INFRAS & Fraunhofer ISI, 2011) (CE Delft, 2018). These three studies all base their
________________________________
59
For CO 2 that is currently being emitted, using a 3% discount rate.
60
Using a low pure rate of time preference (between 0 and 1%).

avoidance costs on the same paper by (Kuik, et al., 2009), which is the most recent
meta-analysis study of avoidance costs. It consists of 26 studies with 31 avoidance costs
estimates. However, most of the studies included in their analysis present avoidance costs
for relatively high ppm targets. For instance, 19 of the 31 avoidance costs estimates
included in their analysis investigated a target of 650 ppm or higher. Only one study
included looked at a target of 450 ppm. Kuik et al. therefore extrapolates based on the cost
estimates for higher ppm targets, to ultimately present avoidance costs for 450 ppm.
Arguably, the Kuik et al.’s (2009) extrapolated avoidance cost estimate for 450 ppm is ‘out
of sample’. However, if we look at Kuik et al.’s values only seem to somewhat stand out for
2025. For 2050 the cost estimates appear to be in line with the rest of the literature.
The main new studies that have been published since previous editions of the Handbook are
(DECC, 2009), (DECC, 2015), (CPB; PBL, 2016), (IEA, 2008) & (IEA, 2010), (IEA, 2017b) &
(IEA, 2017a) (Riahi, et al., 2015) and a range of studies conducted by researchers at the
Potsdam Institut für Klimafolgenforschung ( (Edenhofer, et al., 2010), (Kitous, et al., 2010),
(Magné, et al., 2010), (Leimbach, et al., 2010), (Barker & Scrieciu, 2010) and (D.P., et al.,
2010)). The figure shows the low (triangle), central (round) and high (diamond) value from
the various studies. In general, the values for 2025 from Kuik et al lie slightly higher than
the rest of the literature values, whereas the CPB-PBL (2016) values for both 2030 and 2050
are extremely high compared to the rest of the literature.
Figure 18 – Review of avoidance cost values found in the literature (€ 2016 /t CO2 equivalent)
The avoidance costs used in this Handbook are based on an average of the values found in
the literature, grouped according to the short-and-medium-term (up to 2030) and the long
term (2040-2060). The values for both these time periods were reached by calculating the
average of all central estimates, excluding the lowest and the highest one to eliminate
outliers. This provides us with a central avoidance cost value of € 100 in short-and-medium-
term. This process was repeated for the low and high estimates, which are € 60 and € 189 in
the short-and-medium-term respectively. Table 82 also shows the avoidance costs for the
long term. In the long term the climate costs were shown to have a central estimate of
€ 269/tCO 2 equivalent.

Table 82 – Climate avoidance costs in €/tCO2 eq. (€2016 )
Low Central High
Short and medium run € 60 € 100 € 189
Long run € 156 € 269 € 498
It is important to note that excluding 2030 from the cut-off point between the short-and-
medium-run vs. the long run would not significantly change the short-and-medium-run
avoidance cost estimates. This cut-off is currently set at 2030, but if one were to set it at
2025, the low, central and high estimates for the short-and-medium, run would change to
€ 59, € 97 and remain at € 189 respectively, all very minor changes.
D.3.5 Conclusions
Despite significant advancements in the field of damage cost calculation we use avoidance
costs to calculate climate change costs. This brings us in line with the previous edition of
the Handbook. Numerous studies have advocated the use of avoidance costs rather than
damage costs (CE Delft, INFRAS & Fraunhofer ISI, 2011; DECC, 2009; DECC, 2015; Isacs et
al., 2016; Smith & Braathen, 2015) (CE Delft, 2018). The major weakness of the damage
cost approach is the fact that all climate damages need to be fully understood and
quantified. Although many of the climate damages are somewhat understood, there are
certain feedbacks and potentially extreme events that are not yet fully understood.
Furthermore, the spread of results from different studies assessing climate costs based on
avoidance costs is significantly smaller than for studies using damage costs. This seems to
suggest that there is more certainty around the avoidance costs estimates than the damage
costs estimates.
The biggest argument for the use of avoidance costs rather than damage costs is the fact
that countries have signed up to the Paris Agreement. Since the EU ratified the Paris
Agreement on the 5th of October 2016, committing itself to limit the increase in global
average temperatures to well below 2 degrees above pre-industrial levels (this can be
roughly translated to 450 ppm CO2 equivalents), we have reason to believe this is a credible
long-term reduction target, which is representative of the interests of society, and which
we can use as a target to base the avoidance cost on. Furthermore, if the Paris Agreement
is to be retained, then the marginal cost of extra CO 2 from transport will need to be offset
by reduced emissions elsewhere. This implies that it is not extra damage from global
warming, but extra avoidance in other sectors. Therefore, there will be no additional
damage costs, only avoidance costs of CO 2.
The avoidance costs used in this Handbook are based on the analysis presented in Section
D.3.4. Therefore, the costs are not based on a single study, but rather on multiple studies,
resulting in a central climate changes cost estimate of € 100 per tonne CO2 equivalent for
the short-and-medium-term.

E Detailed assessment of noise
costs
E.1 Introduction
This annex presents more information on how noise costs are calculated, supplementing
Chapter 6. First we discuss the effects of noise on health and annoyance in more detail in
Section E.2. Section E.4 elaborates how noise from different sources is experienced in
different manners. Lastly, Section E.5 explores the total and average noise costs in more
detail, describing the approach in the previous Handbook, examining new evidence and
providing updated values for annoyance and health costs.
E.2 Detailed discussion on effects of noise

Two major impacts are usually considered when assessing noise: annoyance and health
endpoints. Some studies consider annoyance effects as health related effects (Defra, 2014)
(IGCB, 2010). These studies follow recommendations from (WHO, 2011) where physical,
mental and social well-being are considered part of health. Other studies explicitly
differentiate between annoyance and health related effects of noise (Bristow, et al., 2015)
(Nelson, 2008). In this study, we take into account both the costs of annoyance and the
costs of health separately, although they are closely linked. Although noise may also result
in direct material damages as a result of noise vibrations, such damages are not included in
the noise costs mentioned in this paper, because of a lack of data. This is in line with the
scope of the previous versions of this Handbook.
Annoyance represents the disturbance individuals experience when they are exposed to
traffic noise. It can hinder people in performing certain activities, which may lead to a
variety of negative responses, including irritation, disappointment, anxiety and exhaustion
(WHO, 2011). In addition, general wellbeing may be affected. This may be particularly
prevalent during a sunny day spent outside.
Health endpoints can take a multitude of forms. Recently, the WHO commissioned a series
of systematic reviews for the health effects of environmental noise (R., et al., 2017) (M.J.,
et al., 2017) (Kempen, et al., 2018) (Clark & Paunovic, 2018) (Sliwinska-Kowalska &
Zaborowski, 2017) (Basner & McGuire, 2018) (Brown & van Kamp, 2017). Their aim was to
conduct extensive meta-analyses in order to classify the evidence base for the health
effects according to four definitions:
High quality: Further research is very unlikely to change our confidence in the
estimate of the effects.
Moderate quality: Further research is likely to have an important impact on our
confidence in the estimate of the effect and may change the
estimate.
Low quality: Further research is very likely to have an important impact on our
confidence in the estimate of the effect and is likely to change the
estimate.
Very low quality: Any estimate of the effect is uncertain.

Looking purely at the health endpoints for which the WHO classifies evidence as ‘high
quality’ implies that only ischaemic heart disease as a result of road noise should be taken
into account. If we also take into account those health endpoints for which evidence is
classified as ‘moderate quality’ the list significantly expands (see Table 83).
Table 83 – Health endpoints of noise for which the evidence is classified as at least ‘moderate quality’ by the
WHO
Road Rail Aviation

Ischaemic heart disease Diabetes Stroke
Stroke Certain reduced cognitive abilities Certain reduced cognitive abilities
Diabetes Sleep disturbance Obesity
Sleep disturbance Sleep disturbance
A brief summary of a few of the main health endpoints of traffic noise included in the 2018
WHO Systematic Reviews are presented below:
— Cardiovascular diseases and hypertension: Multiple studies have confirmed the link
between cardiovascular disease and traffic noise (Seidler, et al., 2015) (Ecoplan & Infras,
2014) (WHO, 2011). Furthermore, there is increasing evidence that traffic noise increases
the risk of hypertension (Health & Safety Laboratory, 2011), which not only contributes
to increased risk of cardiovascular disease, but also increases the risk of stroke and
dementia (Defra, 2014).
— Sleep disturbance: Sleep disturbance as a result of traffic noise raises stress levels,
reduces wellbeing due to insomnia and tiredness, increases irritability and reduces social
contacts (Fraunhofer ISI & Infras, 2018). Furthermore, evidence suggests a link between
diabetes and exposure to traffic noise through sleep disturbance (Sørensen, et al., 2013)
(Babisch, 2014). Changes in sleeping patterns have been linked to changes in eating
habits, which have been associated with an increased risk of type 2 diabetes and obesity.
— Cognitive impairment: Chronic and acute exposure to aircraft noise seems to affect the
long-term memory and reading capabilities of children (Guski & Schreckenberg, 2015).
Similar effects have also been identified for road noise (Dreger, et al., 2015).
In addition, studies have reported a number of other health endpoints that are not
supported by the latest WHO Systematic Reviews:
— Breast cancer: Three cohort studies have shown significant correlations between breast
cancer in women and traffic noise, but only for specific ranges of noise nuisance and
types of breast cancer (Seidler, et al., 2015) (Greiser & Greiser, 2015) (Sørensen, et al.,
2014). For standardised noise measures and all types of breast cancers, no significant
correlation was found.
— Depression: There are currently very few recent studies exploring the link between
depression and traffic noise. Clearly significant findings from (Seidler, et al., 2015)
suggest a strong correlation between the prevalence of unipolar depression and
exposure to traffic noise. (Greiser & Greiser, 2009) find similar significant findings, but
only for women.

— Tinnitus: Tinnitus, the hearing of sound when no external sound is present, was
included in the health effects of noise by the WHO in 2011, but not in the most recent
WHO Systematic Review. Tinnitus is usually caused by chronic exposure to noise levels
above 85 dB(A), a noise level not usually reached by traffic noise. Therefore, we
propose not to include these aural noise effects in calculating the costs of noise.
Overall, there appears to be weaker evidence for these latter three health endpoints of
traffic noise. In addition, they are not supported by the WHO in their latest Systematic
Reviews. For this purpose, we will not include the costs of breast cancer in women,
depression or tinnitus in the costs of noise. Nonetheless, it is important to note that the
WHO evaluated the quality of the evidence according to very strict guidelines valid for
clinical medicine.
In this study, we take into account both the costs of annoyance and the costs of health
separately, although they are closely linked. For instance, sleep disturbance is classified as
a health impact according to (Defra, 2014) although there is likely to be significant overlap
with annoyance. These two impacts are difficult to separate. In WTP studies looking at
noise it is complicated to separate individual’s valuation for annoyance from sleep
disturbance. If one is asked about their annoyance they are inclined to also take into
account the effects of sleep disturbance. Therefore, there is an implicit risk of double
counting valuation if both sleep disturbance and annoyance impacts are explicitly taken
into account. To avoid double counting we employ the conservative assumption that sleep
disturbance is excluded from the health endpoints and is only considered in the annoyance
endpoint. The health endpoints that we do include in this study are ischaemic heart
disease, stroke and dementia (the latter two both indirectly through hypertension).
In addition to the costs of annoyance and costs of health, there are a number of other
effects as a result of exposure to noise, including disturbance of quiet areas, effects on
ecosystems and effects of restricted land use. Due to the novel nature of research and
complexity of their valuation, none of these effects are incorporated in this study.
— Disturbance of quiet areas: There is evidence of a peaceful environment, or ‘quiet areas’,
being negatively impacted by noise (Anastasopoulos et al., 2011). Noise reduces the
quality of the quiet areas, such as parks, experienced by recreational users. As these
areas are classic public goods without a market price, valuing them is difficult. Lack of
appropriate valuation of these quiet areas prevents the negative effects due to noise
from being incorporated in this study.
— Effects on ecosystems: A growing number of studies have indicated that (man-made)

noise influences the wellbeing of animals. Noise may result in behavioural change in
animals, leading to reduced reproductive success and species density (Dutilleux, 2012).
Due to the novel nature of research in this area, it is not yet possible to quantify these
effects (Dutilleux, 2012).
— Cordon sanitaires: An additional impact of transport noise is the restricted land use
possibilities in areas around airports and some (rail) roads. In many countries
governments establish ‘cordon sanitaires’ around large noise sources like airports.
In these cordon sanitaires land use is restricted, and building new houses is prohibited.
These restrictions in land use change, result in welfare losses and hence should be taken

into account by estimating the external costs of noise. However, due to lack of available
data on this issue, we will not estimate these costs in this study.
Lastly, individual and collective behavioural responses to noise exposure, such as avoidance
measures (e.g. the closing of windows, choice of residential area, good noise isolation and
moving house) or adaptation, are not included in the valuation of noise in this study.
The reasons for this are twofold. Firstly, these behavioural responses can be extremely
difficult to measure, and secondly, because their importance is deemed an order of
magnitude less important than the costs of health and annoyance (Fraunhofer ISI & Infras,
2018).
For a list of the impacts of noise that are included in the valuation in this Handbook we
refer to Section 6.2 and Section E.5.3.
E.3 Noise data

The data used to calculate the noise costs from transport is based on data from the EEA.
The number of people exposed to road and rail noise is based on data provided by
Member States under the frame of the Environmental Noise Directive that are compiled by
EEA (EEA, 2018). The data for road and rail are divided into:
— noise in agglomerations: It concerns agglomeration defined in the Environmental Noise
Directive (END) scope (> 100,000 inhabitants);
— noise on majors roads (> 3 million vehicles/year) and majors railways (> 30,000 train
passages per year).
The compiled data, however, are not complete as not all data have been reported and not
all cities and urban regions are included in the scope of the noise directive. The following
corrections have been applied to complete unreported data on agglomerations to be
reported:
— For countries with partly incomplete reporting of agglomerations, the number of people
exposed to the different noise-bands have been estimated by extrapolating the reported
number to the to-be reported number, linearly with the number of inhabitants
(reported and to-be reported).
— For countries that have no data at all (Greece) the relative exposures per noise band of
a neighbouring country (Bulgaria) have been applied on the number of people to be
reported.
For major roads and railways the following corrections have been applied:
— For countries with partly unreported kilometres road or railway, the number of people
exposed to the different noise-bands have been extrapolated from the reported
kilometres road or rail to the to-be reported kilometres road or rail (linearly with the
kilometres road or railway).
— For countries that have no data at all (Greece, Estonia) the exposed people per
kilometre road or rail of a neighbouring country (Bulgaria, Lithuania) have been applied
on the kilometres road or rail to be reported.
To also include people exposed to road and rail noise in areas that are not within the scope
of the Environmental Noise Directive the following procedure has been applied:
— Data from Eurostat (Eurostat, 2018) on the number of people living in cities (areas with
a centre with a density > 1,500 inhabitants/km 2) and people living in towns and suburbs
(> 300 inhabitants/km 2) were extracted.

— The number of people in cities has been compared with the number or people in the
agglomeration to be reported. For countries where the number of people in cities is
higher than in agglomerations, the same level of exposure is assumed for the unreported
people in cities as for people in the reported agglomerations.
— The number of people in towns and suburbs has been compared with the number of
people in agglomerations minus the people in cities. For countries where the number of
people in towns and suburbs is higher, it was assumed that unreported people in towns
and suburbs were exposed to 3 dB lower noise levels than people in the reported
agglomerations. This was done to account for the differences in noise exposure between
cities and suburbs.
— (A part of) the population exposed in unreported cities, towns and suburbs may already
be covered by the reporting under major roads and railways. To correct for this, it has
been assumed that the same coverage of unreported cities, towns and suburbs for major
roads applies to reported agglomerations under major roads. So if the exposed people to
major roads in agglomeration is 40% of the exposed people in agglomerations, it is
assumed that 40% of the estimated unreported exposures is also covered by the
reporting under major roads. This means the estimated exposure in unreported cities,
towns and suburbs is corrected by subtracting the exposed people with 40% of the
exposed people. The number of subtracted people, however, cannot be higher than the
number of exposed people under major roads.
E.4 Noise nuisance compared between modes

Noise originating from different sources, but at the same volume, is not experienced in the
same manner. This was shown by Table 82, where evidence for the health endpoints of
noise is not homogenous across transport modes. In addition, different levels of annoyance
result from rail, road and aviation noise at the same volume. In previous editions of this
Handbook, the concept of the rail bonus was upheld to account for the lower level of
annoyance for rail noise in comparison to road noise. The rail bonus gives rail transport a
5 dB ‘discount’ in comparison to road noise, implying rail noise at 55 dB is considered
equally annoying as road noise at 60 dB.
However, some recent studies have indicated that relationships between noise level and
reported annoyance are stronger for railway noise than for road noise (Guski et al., 2017),
or that this relationship holds below or above certain noise levels (Bodin et al., 2015)
(Lercher, et al., 2010). Furthermore, some studies found night-time railway noise to be
more annoying than road noise (Elmenhorst et al., 2012) (Elmenhorst et al. , 2014).
Other studies, using physical indicators for annoyance such as heart rate, systolic blood
pressure and stress biomarkers, have not shown a different response to rail noise, compared
to road noise (Gallash et al., 2016). As the literature is not unanimous, evidence makes it
hard to continue to support the rail bonus. Contradicting evidence from hedonic pricing
studies (see Section E.5.2) also complicates upholding the rail bonus. Therefore, based on
the aforementioned literature, we have decided to no longer include a rail bonus in the
costs of annoyance from noise in this Handbook.

E.5 Assessment of total and average noise costs
E.5.1 Recommended approach previous Handbook
Annoyance costs
Earlier editions of the Handbook valued annoyance based on the SP results of HEATCO
(Navrud, et al., 2005). HEATCO’s values are based on a general shadow price of noise
nuisance of € 25 per dB per household, which was then translated to the different national
circumstances. The 2014 edition of the Handbook also used HEATCO’s values, and merely
corrected them for inflation.
Health costs
For health costs, earlier editions of the Handbook were based on UNITE (2003). The health
endpoints covered were: hypertension, angina pectoris and myocardial infarction. To
evaluate the economic costs of these health endpoints a VOLY of € 40,300 (price level 2002)
was used. The cost of illness approach was applied to estimate the medical costs.
E.5.2 New evidence annoyance costs
Stated Preference (SP)

Since HEATCO, there has only been one extensive meta-analysis of SP studies in this area,
carried out by (Bristow, et al., 2015). Their study is the first meta-analysis and most
extensive review to date of SP studies of transportation noise nuisance. Their meta-analysis
is based on a compiled data set of 258 values from 49 studies and 23 countries, and spans a
period of more than 40 years. In contrast, the most extensive meta-analysis of hedonic
pricing includes 53 noise valuations. A novel aspect in their study is the presentation of
higher monetary valuations for higher levels of noise, something which was previously
hinted at by (Kruitwagen, et al., 2006). Moving from 55 to 56 dB is no longer valued the
same as the move from 75 to 76 dB. This is in stark contrast with HEATCO, which used a
constant valuation per dB. The increased valuation with increased noise levels is in line with
the valuation used in several European countries (e.g. Denmark, Sweden and the UK).
Bristow et al.’s (2015) study provides annoyance values for road and aviation noise, but not
for rail noise.
Table 84 illustrates the values from (Bristow, et al., 2015)’s study adapted to the average
income for the EU28, expressed as the WTP for a welfare loss (increase in noise level) per
person per dB.
Table 84 – Valuation of noise annoyance in the EU28 in €2016 per person per dB, L den
50-55 dB 55–64 dB > 65 dB
Road 14 28 54
Aviation 34 68 129

Revealed Preference (RP)
There have been a few studies investigating noise annoyance through RP since the last
Handbook was published. Table 84 presents an overview of the Noise Sensitivity
Depreciation Indexes (NSDI) from a vast amount of literature, both old and new, and shows
that newer studies don’t find vastly different results. The NSDI ranges from 0.08 to 2.22.
However (Bateman et al., 2002) and (Navrud, 2002) both suggest that the average NSDI is
probably found at the bottom of that range (0.55). More recent studies agree with them.
Assuming an average price of € 230,000, an average household size of 2.2 people, a discount
rate of 5% and a duration of 10 years, the NSDI of 0.55 corresponds to a WTP of € 75 per
person per dB per year. This value matches well with the values that (Bristow, et al., 2015)
provide for higher noise levels.
Whether or not the valuation of noise increases more steeply with higher levels of noise has
also been investigated by the hedonic price literature. Both (Theebe, 2004) and (Udo, et
al., 2006) have explicitly taken a non-linear relationship into consideration. (Theebe, 2004)
find this effect holds for noise levels above 65 dB, whereas (Udo, et al., 2006) witness this
effect for the entire range of noise levels.
Table 85 – Noise Sensitivity Depreciation Index (NSDI) results of hedonic price studies
Study NSDI Road NDSI Rail NSDI Aviation

(Bateman, et al., 2001) 0.20 0.25
(Bateman et al., 2002) 0.55 (0.08–2.22)
(Navrud, 2002) 0.08–2.22
(Theebe, 2004) 0.0–0.5 0.0–0.5 0.0–0.5
(Popp, 2006) 0.48 0.48 0.48
(Udo, et al., 2006) 1.7 (1.1–1.9) 1.7 (1.1–1.9)
(Day, et al., 2007) 0.18–0.55 0.67
(Nellthorp, et al., 2007) 0.2–1.07 0.67
(Dekkers & van der Straaten, 2008) 0.16 0.77
(Nelson, 2008) 0.4–0.6 0.7–0.9
(Duarte & Tamez, 2009) 0.08 0.08
(Andersson et al., 2010) 1.35–2.9 0.08–4.09
(Lijesen, et al., 2010) 0.8
(Brandt & Maennig, 2011) 0.23 0.11 0.13
(Getzner & Zak, 2012) 0.85 (0.5–1.3)
(Andersson et al., 2013) 1.15–2.9 0.08–4.09
(Ecoplan & Infras, 2014) 0.21 0.24
(Andersson, et al., 2015) 0.3 1.3
(Swoboda, et al., 2015) 0.5
(Beimer & Maennig, 2017) 0.61 0.65 1.27
(Winke, 2017) 1.7
Furthermore, the studies presented in Table 85 support the acoustic literature (e.g.
(Miedema & Oudshoorn, 2001) which suggests that people consider noise from aircraft
more of a nuisance than road noise. However, the results are not harmonised in terms of
justifying the rail bonus (see Section E.4). The results from (Andersson et al., 2010) and
(Andersson et al., 2013) suggest road noise is more of a nuisance than rail noise, which is in
line with the acoustic literature, and supports the existence of a rail bonus. However, the
studies by (Day, et al., 2007) and (Dekkers & van der Straaten, 2008) find higher NSDI values

for noise originating from rail than road noise. These conflicting results are one of the
reasons for not incorporating the rail bonus in this edition of the Handbook (see Section
E.4).
Environmental burden of disease (EBD)

Although the EBD method seems more appropriate to value health effects, it has also been
used to value annoyance, e.g. by (Defra, 2014). (Defra, 2014) uses dose-response functions
to identify the proportion of people that are highly annoyed at a certain noise level.
This number is then multiplied with the (WHO, 2011) disability weights (see Annex A.2),
where noise annoyance has a central value of 0.02 (ranging from 0.01–0.12). The resulting
number is then multiplied by the health value in the UK. This was set at £ 60,000 per
Disability-Adjusted Life Year (DALY).
Comparing (Defra, 2014) to the SP results from (Bristow, et al., 2015), shows that the EBD
method results in considerably lower valuation than the SP method (see Figure 19). This can
(partially) be explained through the fact that the EBD method uses a conservative approach
for multiple reasons. Firstly, it only incorporates nuisance by highly annoyed persons (%HA).
Annoyed people (%A) are not taken into account using this valuation methodology. It is
currently not possible to correct for leaving out the less severe forms of annoyance, as
(WHO, 2011) does not provide any disability weights that could be used to make these
adjustments. Secondly, the WHO recommended disability weights of annoyance of
0.02 (range spanning 0.01-0.12) is conservative. The sensitivity range for the disability
weights reflect the low and high range from the literature, whereas the point estimate of
0.02 reflects medical experts’ judgement based on a relative weighting compared to other
health endpoints. This implies the values provided by DEFRA could be up to 6 times higher if
a different disability weight was chosen. It is clear that there is large uncertainty regarding
the correct disability weight, but that the disability weight of 0.02 is likely to be an
underestimate of the health costs. Lastly, the dose-response function for %HA for aviation
noise that was used by DEFRA is likely to underestimate the %HA from aviation noise.
This was revealed by an EEA report (EEA, 2010), which noted that recent studies indicated
that the %HA from aviation noise at a given exposure was higher than that predicted by the
function in the EC Position paper (European Commission, 2002). Therefore, the costs
provided by DEFRA for aviation noise annoyance are likely to be an underestimate.
Figure 19 – Comparison of results: DEFRA and Bristow et al. (2015) (€2016 per dB per household)

Aviation
400
350
300
250
€ 2016
200
150
100
50
0
45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79
dB
DEFRA Bristow et al.
The major similarity between (Defra, 2014) and (Bristow, et al., 2015) (and some RP
studies) is that findings suggest that noise nuisance is not linear, and is considered worse at
higher noise levels. Furthermore, in line with the acoustic literature, the EBD findings
support the hypothesis that noise nuisance from different types of transport is valued
differently (e.g. aviation noise is worse than road noise).
Conclusion: Annoyance costs

The literature is not unanimous on which of the three methods, SP, RP or EBD, is the best.
In previous editions of the Handbook, HEATCO’s values were used, which were based on the
SP method. At the time, these were the most recent, highest quality insights.
Although there has been extensive work in this field since the publication of earlier editions
of this Handbook, the meta-analysis of SP studies by Bristow et al. (2015) is the major
revolutionary piece in this literature. Therefore, in this Handbook the values by (Bristow, et
al., 2015) are used. These nuisance values are also based on the SP method, but their main
advantage is that they increase with the noise level. This brings them in line with the most
recent insights in literature, but also with the valuation figures used in practice in a range
of European countries (e.g. Denmark, UK, Sweden). RP results are notoriously difficult to
generalise, due to large differences between housing markets and an NSDI that is highly
dependent on underlying assumptions. With our research spanning a large geographical
area, we believe the use of SP is more appropriate in the valuation of noise than RP.
Compared to EBD results, (Bristow, Wardman, & Chintakayala, 2015)’s results incorporate a
larger share of the nuisance, are less likely to be a clear underestimate and are less
sensitive to certain assumptions (e.g. disability weight). Therefore, we opted to use SP
values in this study.
Although (Bristow, et al., 2015) only present annoyance values for road and aviation noise,
we recommend applying their annoyance values for road and rail noise (see Section E.4 for
a discussion of the rail bonus). We use a threshold value for annoyance of 50 dB(A), in line
with recommendations from previous editions of the Handbook. Although it is widely known
that noise nuisance also exists at lower levels of noise, e.g. (WHO, 2011) (EEA, 2010), it is

currently unclear in how far the valuation studies present reliable values for lower levels of
noise.
E.5.3 New evidence health costs

One can distinguish between two types of health effects of noise: the effects on the person
himself (pain, discomfort, etc.), which can be valued using VOLYs (Value Of Life Year) or
DALYs (Disability-Adjusted Life Year), and the effects on the rest of society (e.g. medical
costs). The main method to calculate health costs uses the environmental burden of disease
(EBD).
Costs to the person himself

There are two recent studies which have calculated health costs of noise to the person
himself since the publication of the most recent edition of the Handbook: (Defra, 2014) and
(Fraunhofer ISI & Infras, 2018). The studies differ in which health effects they take into
consideration. As the WHO is the world-leading authority on health, we aim to align the
health effects incorporated in this Handbook as much as possible with their latest
publications. The latest WHO Systematic Reviews revealed a number of health effects from
traffic noise for which evidence was considered at least of moderate quality (Section E.2).
Between (Defra, 2014) and (Fraunhofer ISI & Infras, 2018), the health effects that Defra
(2014) takes into consideration match the WHO findings best, although it leaves out the
diabetes, obesity and cognition costs of noise (for which moderate quality evidence was
found for certain transport modes). (Fraunhofer ISI & Infras, 2018) do not take into account
the costs from noise exposure related strokes (although this is recommended by the WHO),
but do take into account the costs from breast cancer and depression (which is not
recommended by the WHO). Overall, we conclude that the study conducted by Defra (2014)
aligns best with the WHO’s most recent recommendations. Therefore, the findings from
Defra will be used as a basis for calculating health related costs of noise to the person
himself.
We make one minor modification to the health costs as calculated by Defra. Defra takes
into account annoyance and sleep disturbance separately, although arguably there is some
overlap between the two. Most of the annoyance will take place because of the sleep
disturbance. Therefore, if individuals are well informed of the health effects of noise we
can expect them to take sleep disturbance costs into account in their WTP/WTA values for
annoyance shown in SP studies. HEATCO (2006) assumes that costs of sleep deprivation are
already part of the annoyance costs of noise, and therefore do not need to be taken into
account in the valuation of health costs. Previous editions of the Handbook also followed
this approach. This method is deemed plausible and therefore also followed this study.
Medical costs
The medical costs of noise exposure are not included in Defra (2014). Under medical costs,
we also consider the costs of productivity losses due to illness, e.g. working days lost and
days lost due to reintegration into work. Therefore, we base our calculations of the medical
costs on (HEATCO, 2006), where the medical costs due to noise were calculated to be 8% of
the VOLY. We therefore use this rule of thumb to estimate the medical costs of noise and
apply it to all the countries that we are considering.

Conclusion: Health costs
For health costs the values are based on Defra (2014), but adjusted to also incorporate the
medical costs. The final values used in this Handbook are presented in Table 86.
Table 86 – Health costs used for the EU28 (€2016/person/dB/year) based on Defra (2014)
Transport mode < 55 dB 55-59 dB 60-64 dB 65-69 dB 70–74 dB > 75 dB

Road 3 3 6 9 13 18
Rail 3 4 6 9 13 18
Aviation 5 6 9 12 16 21
E.5.4 Conclusions
Since the previous editions of the Handbook (Infras, CE Delft, ISI & University of Gdansk,
2008) (Ricardo-AEA, TRT, DIW Econ & CAU, 2014), substantial research has been conducted
on the external costs of noise. For annoyance costs, these new insights have resulted in a
new set of cost values that will be used in this Handbook, based on (Bristow, et al., 2015).
A particularly novel aspect of these values is that they increase with the noise level,
reflecting the increased annoyance with higher levels of noise. For the health costs, a lot of
research has focussed on identifying the correct health endpoints that are affected by
noise. Unfortunately, there is no health cost study that incorporates all health effects for
which evidence is considered at least ‘moderate quality’ by the WHO’s latest systematic
reviews. However, the costs by (Defra, 2014) best match the WHO’s findings. Therefore,
the values from (Defra, 2014) will be used and adapted to each of the countries considered
in this study.

F Detailed assessment congestion
costs
F.1 Introduction
This annex presents in detail the methodology used to estimate the road congestion costs
(Annex F.2). Detailed results for road congestion costs per Member State are presented in
Annex F.3 to F.7. Annex F.8 presents the findings on congestion and scarcity costs for other
modes of transport.
F.2 Detailed discussion on effects of road congestion

The methodology is based on the model summarised in Figure 7 in Section 7.3.1, as it
provides a sound theoretical background. Nevertheless, it should be taken into account that
this model is a simplification.
The model refers to a single link to which a specific speed-flow relationship applies,
whereas in the real world, links belong to networks made of links with different features.
Furthermore, in the model there is one demand function whereas in the real world there
are several user categories with different preferences and willingness to pay. Since the
different categories share the use of the network, the definition of the socially optimum
demand level is not straightforward. For those reasons, cost estimates should necessarily be
considered as approximations.
The approach to estimate the road congestion cost postulates a relationship between the
speed 𝑣 and the flow 𝑞 on a road link (i.e., the quantity of vehicles passing through a
transversal road section of the link in a unit of time).
The speed-flow relationship 𝜈 = 𝜈(𝑞) (see Figure 20) is to be interpreted as follows. If the
flow takes any value lower or equal to 𝑞0 , the vehicles travel at free-flow speed 𝑣0. As the
flow increases above the free-flow, say 𝐹(𝑞0 ; 𝑣0 ), the speed gradually decreases to points B
or A. When the flow approaches the capacity of the link, say k at point C, the speed of the
flow reduces to the lower limit 𝑣𝑙 due to a local blockage61.
________________________________
61 According to road traffic theory, if the speed reduces significantly, less cars can pass through the local
blockage, therefore the flow should decrease and the curve should return back towards the vertical axis, in
form of a horizontally U-shape function. For the purpose of this model, the calculations have been developed
only considering a traffic flow situation represented in the upper part of the speed -flow relationship (i.e.,
above point C).

Figure 20 - The speed-flow relationship
F
v [km/hour]
v0 B
v2 v(q)
A
v1
vl C
O q0 q2 q1 k
q [vehicles/hour]
In Figure 21, the function 𝐴𝐶(𝑞), representing the average travel cost borne by the road
users, results from the product of the value of time, which is assumed to be constant across
road users and the average travel time. If the flow takes any value lower or equal to 𝑞0, the
cost of travel is equal to 𝑝 0 and it corresponds to the cost of travel time, at free-flow
speed. When 𝑞 increases, the speed reduces, the travel time increases, and consequently
𝐴𝐶(𝑞) increases.
Figure 21 - Road congestion depending on network conditions

Unit cost
SMC(q)
D(q) C
DWL
B AC(q)
p2
A
p1 D
F
p0 E
O q0 q2 q1 q [vehicles/hour]

The straight line D (q) is an inverse of a function representing the demand of usage of a road
link, depending on the unit cost, expressed in monetary terms. At the point where functions
D (q) and AC(q) intersect, an equilibrium is reached (i.e., point A(q1 ; p1 )) and the
willingness to pay of the user is equal to the private cost borne.
The function 𝑆𝑀𝐶(𝑞) is the social marginal cost function, which is equal to the private cost
𝐴𝐶(𝑞) plus the cost of the additional travel time, generated by the marginal vehicle that
reduces the speed of all the other vehicles. The driver of the marginal vehicle experiences
only his own travel time and not the effect of his decision on the travel times of the other
drivers. Therefore, if an additional vehicle enters the flow, the total social cost is increased
by:
𝑑𝑆𝐶 𝑑𝐴𝐶 (𝑞 )
= 𝐴𝐶(𝑞 ) + 𝑞 ∙
𝑑𝑞 𝑑𝑞
In the equation above the first term is the ‘private cost’ of a single vehicle, while the
second term is the external cost borne by the other users already in the flow due to the
additional vehicle.
Including this external cost element, which is not considered by the single user, the social
marginal cost curve is the relevant decision base. The point 𝐵 (𝑞2; 𝑝2 ), where the functions
𝐷(𝑞) and 𝑆𝑀𝐶(𝑞) intersect, represents the optimal solution. Beyond this point, any
additional vehicle generates a social cost higher than the social benefit62.
F.2.1 Definition of the private costs curve

The private cost curve is assumed to reflect the monetary cost of driving time, namely the
travel time. The time cost of driving increases as the average speed falls, i.e. as the traffic
flow increases. A non-linear speed-flow function is applied to represent this relationship,
namely as follows:
𝑇 = 𝑇0 ∙ (1 + 𝑃𝑎𝑟𝐴 ∙ 𝑟 𝑃𝑎𝑟𝐵 )
Where:
— 𝑇 is the actual travel time;
— 𝑇0 is the travel time in free flow conditions;
— 𝑟 is the flow/capacity ratio;
— 𝑃𝑎𝑟𝐴 and 𝑃𝑎𝑟𝐴 are parameters of the function.
Different speed-flow functions have been implemented in order to differentiate between

the road types (i.e., trunk urban road, other urban road, motorway and other non-urban
roads) (see Figure 22).
________________________________
62
To stimulate the user decisions to follow the SMC(q) function, the difference between average private costs and
marginal social costs (i.e., the segment DB) has to be charged.

Figure 22- Example of speed-flow functions for different road types
F.2.2 Values of time

National values of time for cars and occupancy factors assumed within this Hanbook have
been used to estimate the private (average) cost curve 𝐴𝐶, in order to take into account
that the delay is suffered by all individuals travelling in cars experiencing congestion and
not only drivers.
National values of time for car and coach passengers have been estimated from the UK’s
Department for Transport (ARUP, 2015). Data for UK have been used to estimate values for
the other countries based on GDP per capita (PPP adjusted) by country. Values of time for
coach passengers are about 48% lower than the value of time of long distance car
passengers. For coaches, the VOT of the driver has beed considered on the basis of the data
available from Comité National Routier (2016) by country. The values of time applied for
trucks in terms of commodity transported have been quantified building on those reported
in deliverable 5 of the project HEATCO (2006) and Significance, VU University Amsterdam &
John Bates Services (2012). Moreover, also the VOT of the driver has been considered in the
analysis on the basis of the data available from Comité National Routier (2016) by country.
More specifically, the values for road modes have been used (tonne per hour), assuming the
average load factor by country for heavy duty vehicles (e.g. about 13.6 tonnes per vehicle
for HGV on average in EU28 countries and 0.7 tonnes per vehicle for LCV) and the VOT of
the driver has been added to the estimation in order to the measure the VOT per vehicle.
For LCV it has been assumed that the VOT of the driver is lower than for HGV, assuming a
lower cost of labour due to lower skills requested. The GDP deflator has been applied to
update all the values to Euro 2016.

Table 87 - Value of time by purpose and country for short and long distance trips by car (Euro2016/hour per
person)

Country Short distance* (urban) Long distance* (inter-urban)
Commuting - Personal Commuting - Personal

business business
Austria 16.9 7.8 19.8 7.8
Belgium 15.6 7.2 21.2 7.2
Bulgaria 6.5 3.0 8.5 3.0
Cyprus 11.0 5.1 12.5 5.1
Croatia 8.0 3.7 9.6 3.7
Czech Republic 11.6 5.4 14.0 5.4
Denmark 16.4 7.6 20.7 7.6
Estonia 10.0 4.6 12.1 4.6
Finland 14.5 6.7 20.6 6.7
France 13.8 6.4 15.7 6.4
Germany 16.4 7.6 20.0 7.6
Greece 9.0 4.1 11.4 4.1
Hungary 9.0 4.1 11.7 4.1
Ireland 24.2 11.2 28.3 11.2
Italy 12.8 5.9 16.7 5.9
Latvia 8.6 4.0 10.6 4.0
Lithuania 10.0 4.6 12.1 4.6
Luxembourg 34.2 15.8 38.9 15.8
Malta 12.5 5.8 15.3 5.8
Netherlands 16.9 7.8 22.2 7.8
Poland 9.1 4.2 10.4 4.2
Portugal 10.3 4.8 12.4 4.8
Romania 7.7 3.6 8.8 3.6
Slovakia 10.2 4.7 13.4 4.7
Slovenia 11.0 5.1 13.2 5.1
Spain 12.1 5.6 15.0 5.6
Sweden 16.4 7.6 19.6 7.6
United Kingdom 14.3 6.6 17.3 6.6
Norway 19.7 9.1 24.0 9.1
Switzerland 21.4 9.9 26.2 9.9
Short distance: less than 32 km.

Long distance: more than 32 km.
Source: TRT elaboration on UK Department for Transport (ARUP, 2015).

Table 88 - Value of time for freight road long distance trips by country (Euro2016/hour per tonne)
Country Value per tonne Value per HGV/coach driver

Austria 2.1 22.5
Belgium 1.1 33.4
Bulgaria 0.7 8.0
Cyprus 1.5 13.2
Croatia 0.8 10.6
Czech Republic 1.4 10.2
Denmark 2.0 21.9
Estonia 1.1 13.3
Finland 2.6 19.3
France 1.6 29.8
Germany 2.0 20.9
Greece 1.9 11.9
Hungary 1.4 9.6
Ireland 1.4 32.3
Italy 1.4 28.1
Latvia 1.1 11.4
Lithuania 1.1 8.9
Luxembourg 2.7 28.2
Malta 1.7 19.5
Netherlands 1.7 22.5
Poland 1.4 10.0
Portugal 1.8 13.2
Romania 0.8 9.0
Slovakia 1.2 11.3
Slovenia 1.8 13.1
Spain 1.8 19.5
Sweden 2.4 21.8
United Kingdom 1.9 19.0
Norway 4.2 26.2
Switzerland 1.3 28.5
Source: TRT elaboration on HEATCO project, Significance, VU University Amsterdam, John Bates Services (2012) and Comité
National Routier (2016)
F.2.3 Definition of marginal cost curve

The social cost curve has been derived from the private (average) cost curve, using the first
derivative of the expression reported above, namely:
𝑆𝑀𝐶 = 𝐶0 + 𝐶0 ∙ 𝑃𝑎𝑟𝐴 ∙ 𝑃𝑎𝑟𝐵 ∙ 𝑟 (𝑃𝑎𝑟𝐵 −1)
where 𝐶0 = 𝑇𝑜 ∙ 𝑉𝑂𝑇.

F.2.4 Definition of demand curve
The demand curve has been estimated using demand elasticity parameters 63 starting from
the initial demand (i.e., the level of demand corresponding to the assumed level of
capacity occupancy).
Given the speed flow curves, the load/capacity ratio producing the reported delay could be
calculated separately for each road type during peak and off-peak periods. Then, the
following function has been used to estimate the flow/capacity ratios for alternative values
of cost (i.e., of travel time):
𝑟 = 𝑚 ∙ 𝐴𝐶 + 𝑘
Where:
— 𝑟 is the flow/capacity ratio;
— 𝐴𝐶 is the average cost of driving (i.e., time cost of driving only);
— 𝑚 is the cost elasticity;
— 𝑘 is a constant parameter defining the position of the demand curve.
Cost elasticity parameters 𝑚 have been defined in literature, in particular Littman (2011)
and Oum et al. (1990). Different elasticity parameters have been used for peak and
off-peak periods64. Values have been estimated as weighted average of values by trip
purpose, considering the composition of trips in different periods. Values used are reported
in Table 89.
Table 89 - Values of cost elasticities assumed
Mode Elasticity urban demand Elasticity inter-urban demand

Car – Commuting/business trips -0.49 -0.56
Car – Personal trips -0.58 -0.67
Truck n. a. -0.30
Coaches n. a. -0.30
Source: TRT elaboration on literature data.
F.2.5 Calculation of the optimal demand level

In order to estimate the deadweight loss, point B of Figure 21 is also required. To identify
this point, where the demand curve and the social cost curves cross, an iterative process
has been applied. The value of the load/capacity ratio 𝑟 ∗ such that the social marginal costs
curve provides a marginal cost corresponding to the same load/capacity ratio according to
the demand curve.
________________________________
63 Demand elasticity parameters are defined here as a measure of the relationship between a change in the
transport demand (i.e. the number of trips by car) and a change in the related cost (in this case the private cost
per trip). As an example, a value of the elasticity parameter of -0.5 means that an increase of cost by 20% is
reflected in a decrease of transport demand by -10%.
64
It is assumed that during peak period about 49% of trips is commuting and the residual is for other purposes,
instead during off-peak the commuting trips are about 22%.

F.2.6 Estimation of total deadweight loss
Using the elements obtained in the steps above, the value of deadweight loss 65. is obtained
as the area of the triangle ABC in Figure 21 using the following equation:
(𝑆𝐶1 − 𝑃𝐶1 )
𝐷𝑊𝐿∗ = (𝑟1 − 𝑟 ∗ ) ∙
2
Where:
— 𝑟 ∗is the optimal load/capacity ratio;
— 𝑟1 is the load/capacity ratio in the assumed congestion conditions;
— 𝑆𝐶1 is the social cost in the assumed congestion conditions;
— 𝑃𝐶1 is the private cost in the assumed congestion conditions.
F.2.7 Estimation of deadweight loss per vkm

The value of deadweight loss DWL∗ is related to one vehicle travelling for 1 km in
congestion. Based on assumed road capacity by road type and load (i.e., near capacity,
congested, over-capacity), the deadweight loss per vkm 𝐷𝑊𝐿 has been estimated as
follows:
𝐿𝑟
𝐷𝑊𝐿 = 𝐷𝑊𝐿∗ ∙
𝑉𝑟
Where:
— 𝐿𝑟 is the capacity of the road (i.e., by road type 𝑟);
— 𝑉𝑟 is the vkm of the road (i.e., by road type 𝑟) in the assumed congestion conditions
(i.e., near capacity, congested, over-capacity).
𝑉𝑟 = 𝐿𝑟 ∙ 𝑟1
Where:
𝑟1 is the flow-capacity ratio (v/c) in the assumed congestion conditions.
The estimation has been made differentiating peak and off-peak periods; then, the average
daily value has been estimated based on an assumption of the amount of trips during
peak/off-peak periods.
Table 90 presents the differentiation of the deadweight loss per vkm on congested network
with respect to context and road type, and the level of utilisation of the capacity of the
road link. Traffic situations are identified based on the volume to capacity ratio of a traffic
link. It is assumed that ‘near capacity’ is related to v/c ratios between 0.8 and 1,
‘congested’ refers to v/c ratios between 0.8 and 1.0, while ‘over capacity’ is considered
when v/c ratio is above 1.2.
________________________________
65 As general remark, it is worth noting that the total external cost of road congestion coincides with the area FCA
and that the area FBA is the part of the total external cost which is already accepted by the users (they have a
willingness to pay higher than the cost borne). Therefore, the area FBA is neither mentioned in the literature,
nor commonly used in developing empirical applications to estimate road congestion cost, which is estimated in
terms of delay cost and deadweight loss, as done in this version of the handbook.

Table 90 - EU28 Deadweight loss per vkm of road transport on congested network
Trunk roads Other urban roads Motorway Other roads
Passenger transport (€-cent/vkm)
Congested 2.6 6.5 2.5 4.7
Coach Over capacity n.a. n.a. 28.4 35.3
Congested n.a. n.a. 19.6 28.8
Near capacity n.a. n.a. 10.9 19.6
Freight transport (€-cent/vkm)
HGV Over capacity n.a. n.a. 10.9 13.5
F.2.8 Estimation of delay cost per vkm

Delay congestion costs per vkm IC are estimated by applying values of travel time and the
average vehicle occupancy factors to the amount of delay estimated from the application of
speed-flow functions.
𝐼𝐶 = (𝑇 − 𝑇0 ) ∙ 𝑉𝑂𝑇 ∙ 𝑂𝐹
Where:
— 𝑇 is the actual travel time in the assumed congestion conditions;
— 𝑇0 is the travel time in free flow conditions;
— 𝑉𝑂𝑇 is the value of travel time;
— 𝑂𝐹 is the occupancy factor.
The estimation has been made differentiating peak and off-peak periods; then, the average
daily value has been estimated based on the assumption of the amount of trips during
peak/off-peak periods.
Table 91 presents the differentiation of the delay costs per vkm on congested network with
respect to context and road type, and level of utilisation of the capacity of the road link.
Traffic situations are identified based on the volume to capacity ratio of a traffic link: it is
assumed that ‘near capacity’ is related to v/c ratios between 0.8 and 1.0, while ‘over
capacity’ is considered when v/c ratio is above 1.2.

Table 91 - EU28 Delay costs per vkm of road transport on congested network
Trunk roads Other urban roads Motorway Other roads
Passenger transport (€-cent/vkm)
Congested 12.4 38.8 11.3 24.7
Near capacity 5.1 17.8 4.6 11.1
Coach Over capacity n.a. n.a. 254.8 427.7
Freight transport (€-cent/vkm)
HGV Over capacity n.a. n.a. 97.6 163.9
F.2.9 Methodology to estimate yearly road congestion costs

Marginal congestion costs per vkm and delay cost per vkm have been used to estimate the
respective yearly road congestion costs in the EU. It should be noted that this cost is an
estimation of the monetary equivalent of additional travel time rather than a true financial
cost borne by individuals or companies. With a few exceptions (e.g., higher fuel
consumption, clients of taxicabs paying a higher charge because of longer travel time due
to congestion, companies delivering freight that have to use more vehicles — and therefore
drivers — to complete consignment in due time), travel time wasted in congestion does not
entail any monetary expenditure or missing revenues.
Also, interpreting the estimation in monetary terms of congestion cost as an economic

benefit that individuals could enjoy if congestion were removed is actually incorrect.
As pointed out by Goodwin (2004): “The implied annual dividend [...] to be distributed to
each family is a fiction. It is calculated by comparing the time spent in traffic now, with
the reduced time that would apply if the same volume of traffic was all travelling at free
flow speed, and then giving all these notional time savings the same cash value that we
currently apply to the odd minutes saved by transport improvements. But this could never
exist in the real world — not for reasons of practical difficulty, but because it is internally
inconsistent. If all traffic flowed at free flow speed, we can be quite certain there would
be more of it, at least part of the time saved would be spent on further travel, and
further changes would be triggered whose value is an unexplored quality. It is apparently a
precise answer to a phantom question”.
The estimation has been made separately for urban and inter-urban congestion given the
different type of information available.

F.2.10 Estimation of yearly road urban congestion cost
Yearly urban congestion costs were estimated building on the information provided by
TomTom data on about 212 cities in Europe. Data is provided on the level of congestion and
road network length by road type (i.e., trunk urban road, other urban road), average delay
per day (based on two 30 minute peak period journeys per day) and total accumulated
delay per year (related to peak period journeys).
The two road types (i.e., urban trunk roads and other urban roads) are differentiated
according to the data already available from the TomTom dataset. The purpose of this
differentiation is to obtain different speed-flow functions. Peak and off-peak periods are
separated to take into account that trip purposes are not the same during different periods
of the day and therefore that values of time and elasticities of demand are also variable.
This segmentation is not directly available in the TomTom dataset so the data has been
elaborated to derive wasted time in congestion separated for peak/off peak periods crossed
with urban trunk roads/other urban roads.
The amount of delay on urban trunk roads (and other urban roads) during off-peak periods
has been estimated in two steps. First, given the ratio between the congestion index and
the average congestion index during peak periods the daily amount of delay on urban trunk
roads (other urban roads) has been estimated. Second, the part of this daily delay occurring
during off-peak periods has been estimated building on the shares of trips in peak and
off-peak (based on travel surveys statistics). Basically, given these shares and given the
amount of delay during peak periods the amount of delay in off-peak time has been
computed to sum the two components to reproduce the total daily delay (on urban trunk
roads and other urban roads respectively).
Building on the information above, the amount of congested network by road type and time
period 𝐶𝑁𝑟𝑝 has been estimated:
𝐷𝑇𝑟𝑝
𝐶𝑁𝑟𝑝 =
𝑇𝑐𝑟 − 𝑇0𝑟
Where:
— 𝐷𝑇𝑟𝑝 is the average delay per day by road type 𝑟 and time period 𝑝;
— 𝑇𝑐𝑟 is the actual travel time in the assumed congestion conditions 𝑐;
— 𝑇0𝑟 is the travel time in free flow conditions by road type 𝑟.
The most representative congestion conditions for each city (based on TomTom data) have
been used to estimate total cost per vehicle, by multiplying the related deadweight loss and
delay cost per car vkm (i.e., 𝐸𝐶𝑝𝑟 and 𝐼𝐶𝑝𝑟, differentiated by time period and road type) by
the related estimated congested network (during peak and off-peak periods).
Finally, total yearly costs in the urban context (i.e., 𝐼𝐶 and 𝐸𝐶) have been estimated using
the population size (i.e., 𝑃), the share of individuals travelling (i.e., 𝑠ℎ𝑇𝑃) and the car
share (i.e., 𝑠ℎ𝐶𝑎𝑟), assuming 230 work days per year66.
𝐼𝐶 = 𝑃 ∙ 𝑠ℎ𝑇𝑃 ∙ 𝑠ℎ𝐶𝑎𝑟 ∙ ∑ 𝐼𝐶𝑝𝑟 ∙ 230

𝑝𝑟
________________________________
66
This value is consistent with assumptions reported for the estimation of the TomTom congestion index.

𝐸𝐶 = 𝑃 ∙ 𝑠ℎ𝑇𝑃 ∙ 𝑠ℎ𝐶𝑎𝑟 ∙ ∑ 𝐸𝐶𝑝𝑟 ∙ 230
𝑝𝑟
Car share data has been defined based on information reported in the EPOMM Modal Split
Tool (TEMS)67 integrated with local sources where data from this tool was not available.
The share of population travelling is estimated as 76% of population living in the city,
according to AUDIMOB data for 2013.
This methodology provided urban congestion costs related to passenger cars for the
European cities included in TomTom data. The generalisation of results to all urban areas
for each country was made as explained below.
F.2.11 Methodology for cities with more than 50,000 inhabitants

First of all, in order to obtain a set of data comparable among different cities in different
countries, marginal costs have been re-estimated using an average value of time and an
average occupancy factor for all EU. This set of costs has been used for a statistical analysis
aimed at identifying correlations between congestion cost per capita and population of the
cities by class.
It was found that the higher the population size class the city belongs to, the higher the
deadweight congestion cost per vehicle (see Table 92).
Table 92 - Average yearly delay cost and deadweight loss per capita depending on city population size:
TomTom sample data (EURO2016 /capita)
City population size Average delay cost per capita Average Deadweight lossper capita
More than 5 million 898.5 165.0
2 to 5 million 806.6 145.5
1 to 2 million 754.1 134.4
500,000 to 1 million 734.4 130.2
250,000 to 500,000 724.6 128.2
100,000 to 250,000 719.3 127.0
Source: TRT estimation on TomTom data.
Since we are estimating congestion costs in cities, we considered not only national
differences of values of time, but also regional differences in terms of average income per
capita. The average values have been scaled to consider (i) the national values of time (see
Table 86) and (ii) the ratio between regional and national GDP per capita. Therefore, the
estimation of urban congestion cost has been made assuming that the opportunity cost of
time depends on local features and particularly economic activity.
These values of congestion cost per capita by NUTS3 region have then been applied to all
related cities with at least 50,000 inhabitants. The list of cities has been compiled based on
information collected from different sources, e.g. the website City Population as well as
national statistical offices. The list includes about 1,275 cities in 30 European countries
(i.e., EU28 plus Switzerland and Norway).
________________________________
67
See also http://www.epomm.eu/tems/

F.2.12 Methodology for cities with less than 50,000 inhabitants
It is reasonable to expect that some congestion occurs also in cities with less than 50,000
inhabitants, however extending the dataset to cities larger than e.g. 15,000 inhabitants
would have been too complex given the number of urban areas of this size in Europe.
A simplified approach was adopted to generalise urban congestion costs also to cities below
the threshold of 50,000 inhabitants.
The simplified approach consisted of estimating the number of additional urban areas to
consider in each NUTS3 zone. Two elements have been used for this estimation. First, the
total amount of population in the NUTS3 zone compared to the amount of population in the
cities with more than 50,000 inhabitants located in the same zone. Intuitively if these cities
explain a large share of total population of the NUTS3 it is likely that only a few or even no
cities between 15,000 and 50,000 inhabitants exist in that zone. Vice-versa, if the cities
above 50,000 inhabitants explain only a limited share of total population, a higher number
of smaller cities can be expected.
The second element was the typology of NUTS3 according to the classification
urban/mixed/rural. In rural areas, cities tend to be smaller and so a lower number of
urban areas between 15,000 and 50,000 inhabitants can be expected for a given share of
population not explained by the cities above 50,000 inhabitants. NUTS3 population was
extracted from the Eurostat database. The classification of NUTS3 regions in three
categories: predominantly urban, predominantly rural, mixed is also provided by Eurostat.
The cities with at least 50,000 inhabitants have been associated to the NUTS3 zone they
belong. Then, for each NUTS3 region, the sum of the population living in cities with at least
50,000 inhabitants has been compared to the total population of the region. Depending on
the share of population living in the city/cities with more than 50,000 inhabitants in each
NUTS3 region and the category of the region itself, different rules have been applied to
estimate how many additional urban areas should be considered for the generalisation of
urban congestion cost.
Using the average congestion costs per capita related to passenger cars in a NUTS3 region,
total urban congestion cost by NUTS3 region was quantified. An additional assumption made
has been that in cities with less than 50,000 inhabitants, congestion occurs only in the peak
period of the day.
F.2.13 Estimation of yearly marginal congestion cost for inter-urban roads

The quantification of traffic experiencing congestion on inter-urban roads has been carried
out building on two main sources. One source was a map of the congested spots on the
European inter-urban road network provided by JRC Sevilla. This map identified spots where
road traffic is delayed in the most congested peak hours because of traffic and, for each
spot, provided the amount of delay (in terms of additional time per km). The map was
drawn using real traffic data for the year 2013.
The map was very helpful to identify where congestion occurs and the range of its severity,
however this source alone did not allow to quantify the amount of demand involved in
congestion. This further element could be estimated by means of parameters used in the
TRUST network model. TRUST is a transport network model covering the whole of Europe
developed by TRT (see the following Box for details).
In TRUST the road network is classified into different link types (e.g., motorways, dual
carriageways roads, etc.). Each link type is associated to specific features; in particular

speed-flow functions. Speed flow functions link traffic on one road to the time required to
travel onto the road itself. They differ according to road types, for instance urban roads
free flow speed is disturbed already for relatively low level of traffic whereas on motorways
speed is maintained longer but is then more rapidly reduced when traffic approaches
capacity.
The TRUST model

TRUST (TRansport eUropean Simulation Tool) is a transport network model developed by TRT in the MEPLAN
software environment. TRUST is a transport network model for the assignment of Origin -Destination matrices at
the NUTS3 level of detail for passenger and freight demand. TRUST covers the whole Europe, including Accession
and Neighbouring countries.
Road as well non-road transport modes are dealt with in TRUST. The road network includes all the relevant links between the
NUTS3 regions, i.e. motorways, primary roads as well as roads of regional and sub-regional interest. Also ferry connections
(i.e., ro-ro services) between European regions are explicitly modelled with their travel time and fare. Road network links are
separated in different classes, each with specific features in term of capacity, free-flow speed and toll.
The main output of TRUST is the load on road network links in terms of vehicles per day and on non-road links in terms of
either trips or tonnes per day. The model is calibrated to reproduce tonnes-km and passengers-km by country consistent to
the statistics reported in the Eurostat Transport in Figures pocketbook net of intra-NUTS3 demand (available from ETISplus),
which is not assigned to the network. Using load as an input parameter and emissions factors the model also provides
emissions by link for NO x , PM and CO. Figure 23 shows the road network modelled in TRUST.

Figure 23 - The road network in TRUST
The TRUST model uses the most recent data made available by the ETISplus 68 project. Apart the features of the network links
(speed, capacity, etc.), the main parameters used in TRUST are:
− transport costs by mode;
− speed-flow functions;
− values of travel time;
− average fuel consumption and emission factors (for road modes).
The TRUST model has been successfully applied for the assessment of the Eurovignette directive on behalf of the
European Commission.
________________________________
68
ETISplus provides a set of data including networks, matrices, etc. to serve for the development of transport
modelling at the European level.

Using the range of delay69 reported on the map of the congested spots and the speed-flow
function of the roads where the spots are located, it was possible to estimate the level of
occupancy of each spot in peak time, i.e. the amount of vehicles (in terms of Passenger
Cars equivalent Units – PCEs) experiencing congestion in the most congested peak hour.
Traffic on road links includes several vehicle types: cars, trucks, etc. Since the congestion
cost associated with each type is different, a segmentation of the estimated loads has been
estimated. Two main classes of road users have been considered, namely: cars and trucks.
The share of demand belonging to each class has been estimated making reference to the
segmentation of traffic on each link assigned by the TRUST model. Indirectly, coaches have
also been considered in the analysis based on the assumption that the share of interurban
coach vkm with respect to car and truck vkm (available as data by country) could be used to
estimate the related traffic on the congested sport.
The most congested peak hour is when motorists experience the highest delays, however
there are other peak and off-peak periods when some congestion occurs. In order to
estimate the overall cost of inter-urban congestion also delays outside the most congested
peak hours should be estimated. This task was addressed using representative road load
profiles for passenger cars and trucks during the day in different countries (i.e. Italy,
France, United Kingdom and the Netherlands).
Road profiles describe how traffic changes over a 24-hour period. Of course in principle
each road has its own profile but the distribution of traffic during the day is very similar for
different roads and also in different countries (although peak time can be slightly different
according to local habits about e.g. working time).
Using these profiles and using the estimated load in the most congested peak time, it has
been possible to estimate the load in each hour. Given the capacity of the road (taken from
TRUST) and considering the sum of all vehicle types (in terms of Passenger Cars equivalent
Units – PCEs) the load/capacity ratios have been estimated for each link and hour.
All vehicles travelling in hours with a load/capacity ratio higher than 1 have been
considered experiencing congestion (congested with load/capacity ratio up to 1.2 and over
capacity with load/capacity ratio above 1.2); furthermore, also vehicles travelling in hours
with a load/capacity ratio between 0.75 and 1 have also been considered experiencing
congestion (near capacity).
After this process, the total number of vehicles incurring congestion on the inter-urban
European networks in an average day has been obtained for each vehicle type (i.e., car and
truck). Vkms in congestion by vehicle type have been estimated multiplying loads by length
of congested network on a link level. The estimation of the amount of yearly traffic is made
assuming 230 work days per year.
In order to estimate the amount of coaches experiencing congestion on inter-urban roads,

the amount of vkm has been estimated on the basis of assumptions on the share of coaches
traffic with respect to car traffic by country (taken from the ASTRA model).
F.2.14 Estimation of yearly congestion cost for other road transport categories
With reference to congestion costs for trucks at the urban level, due to lack of information,
the methodology applied for cars cannot be replicated. Therefore, in order to provide some
________________________________
69
Data has been used in terms of classes of delay instead of punctual values due to some discrepancies occurring
when joining the TRUST network with the JRC network, which are not perfectly matching.

information to also cover this aspect, a simplified approach has been applied, based on the
estimation of congestion costs for cars, Value of Times by mode and data on vkm at urban
level for LCVs and HGVs (estimated/collected within this Handbook).
A similar simplified approach is also applied for LCVs at inter-urban level. The estimation has
been based on the value of congestion costs for cars, the Value of Time and data on vkm at
inter-urban level (estimated/collected within this Handbook).
F.2.15 International comparisons of urban congestion costs

The methodology for the estimation of urban congestion costs has been applied for
international comparison to the following non-EU regions: British Columbia and Alberta for
Canada and California and Missouri for US, respectively. The methodology was built on
TomTom data on urban congestion of 13 cities of the above mentioned regions, following
the same steps applied for the European context.
Although the sample of cities was limited in this case, a statistical analysis has been
performed to estimate the correlation between congestion cost per capita and city
population size (see Table 93). These values of road congestion cost per capita have then
been applied to all related cities with at least 20,000 inhabitants, multiplying the values by
the population data 70.
The estimation of urban congestion costs for Japan followed the same approach, although
TomTom data was not available. Another congestion index, built on top of the traffic layer
within Google Maps, has been used for the city of Tokyo. The index has been adapted to the
TomTom format following the comparison of the data of European cities available in both
Google Maps and TomTom format.
The trend of the correlation between congestion cost per capita and city population size of
North-America has been used to estimate the values for Japan (e.g. average delay cost per
capita for cities between 1 to 2 million inhabitants is about 43% less than the average cost
for cities with more than 5 million inhabitants). These values of road congestion cost per
capita have then been applied to all related cities above 500,000 inhabitants, multiplying
the values by the population data 71.
Table 93 - Average yearly delay cost and deadweight loss per capita depending on city population size in
non-EU regions: TomTom sample data (EURO2016 /capita)
City population size Average delay cost per capita Average Deadweight loss per capita
More than 5 million 1,918.7 275.5
2 to 5 million 1,396.7 228.5
1 to 2 million 1,098.4 201.7
500,000 to 1 million 986.5 191.6
250,000 to 500,000 930.6 186.6
100,000 to 250,000 900.8 183.9
Source: TRT estimation on TomTom data.
________________________________
70
For Canada:”Population and dwelling counts highlight tables, 2016 census”. Statistics Canada For US, the 2010
US Census.
71
For Japan: Population data reported by Prefectural Governement.

Table 94 reports the input parameters used for Europe and the non-EU regions under
analysis in terms of Value of Time and average car occupancy factor.
Table 94 - Passenger Car Value of Time and average occupancy factor in EU28 and non-EU regions72
Context Short distance Average car occupancy factor

Commuting - business Personal
Euro 2016 /hour per person Persons/vehicle
EU28 13.3 6.1 1.7
Canada 7.7 7.0 1.5
US 11.3 10.8 1.7
Japan 12.3 10.3 1.5
Table 95 reports total urban congestion costs and urban congestion costs per capita for cars
in Europe and in the non-EU regions under analysis.
Table 95 - Car urban congestion costs in EU28 and non-EU regions
Context Total costs Average costs per capita

Delay costs Deadweight loss Delay costs Deadweight loss
costs costs
Billion € €/year per capita
EU28 172 30.0 337.2 58.6
Canada* 5 0.9 767.7 149.2
US* 64 10.9 1,840.6 314.3
Japan* 22 4.7 493.0 105.5
* Canada: Estimations for Alberta and British Columbia.
* USA: Estimations for California and Missouri.
* Japan: Estimations for cities above 500,000 inhabitants.
F.3 Detailed estimated congestion costs by country - Total costs generated

estimated using the simplified approach
Table 96 - Total delay congestion costs generated by road modes, estimated using the simplified approach
(billion Euro/year, in Euro 2016 )
Country Urban Inter-urban
Car LCVs Trucks Coaches Car LCVs Trucks Coaches

(HGV) (HGV)
EU28 160.775 46.462 11.561 3.947 35.276 9.022 3.034 0.528
Austria 1.893 0.736 0.138 0.040 0.618 0.211 0.037 0.013
________________________________
72
For Canada:”The cost of urban congestion in Canada, Transport Canada Environmental Affairs 2006”. For US:
the National Household Travel Survey (NHTS) and US department of transportation. For Japan: “Valuation of
travel time saving With revealed preference data In Japan: further analysis.”

Belgium 4.639 0.927 0.164 0.178 2.648 0.527 0.142 0.040
Bulgaria 1.278 0.088 0.126 0.055 0.078 0.006 0.008 0.002
Croatia 0.744 0.274 0.062 0.013 0.059 0.019 0.004 0.001
Cyprus 0.263 0.124 0.008 0.013 - - - -
Czech Republic 2.120 0.591 0.333 0.072 0.265 0.076 0.051 0.012
Denmark 1.374 0.525 0.098 0.035 0.614 0.248 0.049 0.012
Estonia 0.223 0.025 0.022 0.009 0.050 0.005 0.005 0.002
Finland 1.763 0.313 0.057 0.047 0.135 0.024 0.010 0.002
France 17.196 5.731 0.588 0.374 9.481 3.836 0.436 0.127
Germany 30.970 4.601 2.711 0.417 4.790 0.340 0.366 0.064
Greece 3.225 0.639 0.133 0.095 0.419 0.101 0.021 0.013
Hungary 1.319 0.544 0.230 0.084 0.273 0.096 0.061 0.019
Ireland 3.516 3.825 0.055 0.225 0.982 0.399 0.034 0.013
Italy 19.032 5.799 0.886 0.683 4.010 0.981 0.244 0.059
Latvia 0.417 0.068 0.082 0.012 0.024 0.003 0.004 0.001
Lithuania 0.612 0.169 0.123 0.017 0.030 0.005 0.012 0.000
Luxembourg 0.314 0.388 0.048 0.015 0.203 0.250 0.050 0.004
Malta 0.062 0.011 0.020 0.002 - - - -
Netherlands 6.091 3.129 0.350 0.052 1.234 0.267 0.119 0.004
Poland 7.066 1.477 2.042 0.311 1.966 0.384 0.705 0.057
Portugal 3.231 2.980 0.125 0.044 0.929 0.153 0.054 0.005
Romania 4.332 0.852 0.590 0.257 0.533 0.112 0.078 0.013
Slovakia 0.631 0.170 0.156 0.025 0.163 0.045 0.049 0.005
Slovenia 0.248 0.068 0.037 0.005 0.091 0.022 0.012 0.002
Spain 13.469 3.976 0.802 0.259 2.818 0.132 0.243 0.037
Sweden 4.743 1.192 0.208 0.113 0.243 0.061 0.023 0.003
United Kingdom 30.002 7.239 1.368 0.494 2.622 0.719 0.218 0.019
Norway 1.516 0.672 0.072 0.029 0.210 0.093 0.024 0.002
Switzerland 1.882 0.251 0.091 0.021 1.069 0.123 0.044 0.012

Table 97 - Total deadweight loss generated by road modes, estimated using the simplified approach (billion
Euro/year, in Euro 2016 )
Country Urban Inter-urban

(HGV) (HGV)
EU28 28.026 7.988 2.020 0.676 5.518 1.439 0.453 0.082
Austria 0.341 0.132 0.025 0.007 0.099 0.034 0.006 0.002
Belgium 0.729 0.146 0.026 0.028 0.455 0.091 0.025 0.007
Bulgaria 0.222 0.015 0.022 0.010 0.008 0.001 0.001 0.000
Croatia 0.131 0.048 0.011 0.002 0.008 0.003 0.001 0.000
Cyprus 0.046 0.022 0.001 0.002 - - - -
Czech Republic 0.378 0.105 0.059 0.013 0.033 0.009 0.006 0.001
Denmark 0.242 0.093 0.017 0.006 0.120 0.048 0.010 0.002
Estonia 0.040 0.004 0.004 0.002 0.005 0.001 0.000 0.000
Finland 0.315 0.056 0.010 0.008 0.023 0.004 0.002 0.000
France 3.039 1.013 0.104 0.066 1.556 0.628 0.073 0.021
Germany 5.490 0.816 0.481 0.074 0.711 0.051 0.055 0.009
Greece 0.581 0.115 0.024 0.017 0.085 0.021 0.004 0.003
Hungary 0.232 0.096 0.040 0.015 0.034 0.012 0.008 0.003
Ireland 0.454 0.493 0.007 0.029 0.111 0.047 0.005 0.002
Italy 3.406 1.038 0.158 0.122 0.679 0.165 0.042 0.011
Latvia 0.073 0.012 0.014 0.002 0.003 0.000 0.000 0.000
Lithuania 0.107 0.029 0.022 0.003 0.004 0.001 0.002 0.000
Luxembourg 0.053 0.066 0.008 0.003 0.034 0.042 0.009 0.001
Malta 0.010 0.002 0.003 0.000 - - - -
Netherlands 1.069 0.549 0.061 0.009 0.223 0.048 0.022 0.001
Poland 1.250 0.261 0.361 0.055 0.222 0.043 0.080 0.006
Portugal 0.574 0.529 0.022 0.008 0.142 0.025 0.009 0.001
Romania 0.630 0.124 0.086 0.037 0.049 0.011 0.007 0.001
Slovakia 0.110 0.030 0.027 0.004 0.020 0.006 0.006 0.001
Slovenia 0.043 0.012 0.006 0.001 0.015 0.004 0.002 0.000
Spain 2.439 0.720 0.145 0.047 0.423 0.021 0.040 0.006
Sweden 0.837 0.210 0.037 0.020 0.043 0.011 0.004 0.000
United Kingdom 5.186 1.251 0.236 0.085 0.414 0.113 0.035 0.003

Norway 0.264 0.117 0.013 0.005 0.039 0.017 0.005 0.000
Switzerland 0.321 0.043 0.016 0.004 0.187 0.022 0.008 0.002

Table 98 - Total inter-urban congestion costs generated on motorway network, estimated using the simplified
approach (billion Euro/year, in Euro 2016 )
Country Deadweight loss Delay cost

(HGV) (HGV)
EU28 0.724 0.196 0.085 0.015 3.182 0.861 0.375 0.066
Austria 0.014 0.005 0.001 0.000 0.063 0.022 0.005 0.001
Belgium 0.069 0.014 0.006 0.001 0.299 0.060 0.024 0.004
Bulgaria 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000
Croatia 0.001 0.000 0.000 0.000 0.004 0.001 0.000 0.000
Cyprus - - - - - - - -
Czech Republic 0.004 0.001 0.001 0.000 0.018 0.005 0.003 0.001
Denmark 0.011 0.003 0.001 0.000 0.048 0.015 0.004 0.001
Estonia - - - - - - - -
Finland 0.004 0.001 0.000 0.000 0.019 0.003 0.001 0.000
France 0.192 0.073 0.013 0.003 0.848 0.321 0.057 0.014
Germany 0.119 0.010 0.010 0.001 0.526 0.044 0.044 0.004
Greece 0.010 0.005 0.001 0.001 0.044 0.023 0.005 0.004
Hungary 0.010 0.003 0.003 0.001 0.045 0.014 0.012 0.005
Ireland 0.013 0.009 0.002 0.000 0.056 0.037 0.008 0.002
Italy 0.077 0.016 0.007 0.003 0.337 0.071 0.031 0.015
Latvia - - - - - - - -
Lithuania 0.003 0.000 0.001 0.000 0.011 0.002 0.006 0.000
Luxembourg 0.010 0.013 0.004 0.000 0.045 0.056 0.016 0.001
Malta - - - - - - - -
Netherlands 0.054 0.011 0.007 0.000 0.231 0.047 0.032 0.001
Poland 0.007 0.001 0.003 0.000 0.032 0.005 0.014 0.000
Portugal 0.021 0.009 0.004 0.000 0.091 0.038 0.018 0.002
Romania 0.002 0.001 0.001 0.000 0.008 0.004 0.003 0.001
Slovakia 0.001 0.000 0.000 0.000 0.006 0.002 0.002 0.000
Slovenia 0.005 0.001 0.001 0.000 0.020 0.005 0.003 0.000
Spain 0.049 0.007 0.015 0.002 0.220 0.031 0.067 0.010
Sweden 0.011 0.003 0.001 0.000 0.050 0.012 0.003 0.001
United Kingdom 0.037 0.010 0.004 0.000 0.158 0.041 0.016 0.001

Norway 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000
Switzerland 0.045 0.006 0.003 0.001 0.190 0.024 0.011 0.002

F.4 Detailed estimated congestion costs by country - Total costs borne by
mode
Table 99 - Total car congestion costs (billion Euro/year, in Euro 2016 )
Country Deadweight loss Delay cost Deadweight Delay cost Deadweight loss Delay cost
loss
Total (urban and inter-urban) Urban Inter-urban
EU28 35.553 206.158 30.006 172.574 5.547 33.584
Austria 0.475 2.671 0.371 2.059 0.104 0.612
Belgium 1.186 7.268 0.762 4.854 0.423 2.414
Bulgaria 0.245 1.438 0.237 1.361 0.008 0.077
Croatia 0.164 0.939 0.155 0.875 0.009 0.064
Cyprus 0.057 0.322 0.057 0.322 - -
Czech Republic 0.492 2.855 0.453 2.540 0.038 0.314
Denmark 0.395 2.144 0.262 1.490 0.133 0.655
Estonia 0.049 0.294 0.043 0.238 0.006 0.055
Finland 0.348 1.938 0.327 1.832 0.021 0.106
France 4.616 25.863 3.078 17.417 1.538 8.446
Germany 6.230 35.624 5.555 31.339 0.675 4.285
Greece 0.721 3.921 0.630 3.499 0.091 0.422
Hungary 0.318 1.880 0.280 1.592 0.038 0.288
Ireland 0.769 6.074 0.648 5.022 0.121 1.052
Italy 4.228 23.633 3.549 19.831 0.680 3.803
Latvia 0.080 0.467 0.078 0.443 0.003 0.024
Lithuania 0.131 0.764 0.126 0.722 0.005 0.042
Luxembourg 0.130 0.773 0.079 0.464 0.051 0.309
Malta 0.010 0.063 0.010 0.063 - -
Netherlands 1.458 8.276 1.223 6.969 0.235 1.307
Poland 1.624 10.023 1.353 7.650 0.271 2.373
Portugal 0.861 4.934 0.718 4.043 0.142 0.891
Romania 0.812 5.784 0.755 5.189 0.057 0.594
Slovakia 0.154 0.946 0.128 0.736 0.026 0.210
Slovenia 0.061 0.357 0.044 0.255 0.018 0.102
Spain 3.001 16.716 2.631 14.530 0.370 2.187

Country Deadweight loss Delay cost Deadweight Delay cost Deadweight loss Delay cost
loss
Sweden 0.927 5.231 0.883 5.007 0.043 0.224
United Kingdom 6.012 34.960 5.572 32.232 0.440 2.728
Norway 0.311 1.745 0.271 1.555 0.040 0.190
Switzerland 0.524 2.972 0.338 1.979 0.186 0.993
Canada* 0.965 4.965 0.965 4.965 - -
US* 10.906 63.873 10.906 63.873 - -
Japan* 4.715 22.022 4.715 22.022 - -
* Canada: Estimations for Alberta and British Columbia.
* USA: Estimations for California and Missouri.
* Japan: Estimations for cities above 500,000 inhabitants.
Table 100 - Total trucks, coaches and LCVs inter-urban congestion costs (billion Euro/year, in Euro 2016 )

Trucks Coaches LCVs* Trucks Coaches LCVs*
(HGV) (HGV)
EU28 0.696 0.218 1.031 6.234 2.112 5.930
Austria 0.014 0.005 0.018 0.108 0.055 0.103
Belgium 0.056 0.030 0.068 0.470 0.086 0.388
Bulgaria 0.001 0.001 0.000 0.007 0.008 0.002
Croatia 0.001 0.000 0.001 0.008 0.003 0.008
Cyprus - - - - - -
Czech Republic 0.003 0.006 0.003 0.028 0.039 0.023
Denmark 0.009 0.004 0.034 0.071 0.028 0.169
Estonia 0.000 0.000 0.000 0.000 0.004 0.002
Finland 0.006 0.000 0.002 0.048 0.005 0.011
France 0.176 0.030 0.534 1.956 0.538 2.939
Germany 0.106 0.017 0.028 0.910 0.188 0.177
Greece 0.004 0.010 0.009 0.027 0.064 0.040
Hungary 0.008 0.006 0.004 0.058 0.068 0.034
Ireland 0.011 0.009 0.023 0.113 0.074 0.189
Italy 0.058 0.030 0.129 0.395 0.369 0.726
Latvia 0.001 0.000 0.000 0.006 0.001 0.002
Lithuania 0.001 0.000 0.000 0.003 0.000 0.002
Luxembourg 0.005 0.004 0.026 0.028 0.015 0.154
Malta - - - - - -
Netherlands 0.022 0.010 0.027 0.153 0.014 0.149
Poland 0.051 0.011 0.019 0.460 0.114 0.166
Portugal 0.017 0.006 0.012 0.145 0.039 0.067
Romania 0.005 0.003 0.003 0.048 0.062 0.032
Slovakia 0.003 0.002 0.002 0.026 0.011 0.017
Slovenia 0.001 0.001 0.002 0.004 0.009 0.011

Trucks Coaches LCVs* Trucks Coaches LCVs*
(HGV) (HGV)
Spain 0.086 0.023 0.011 0.741 0.241 0.060
Sweden 0.008 0.001 0.007 0.060 0.011 0.034
United Kingdom 0.047 0.010 0.068 0.362 0.065 0.423
Norway 0.007 0.001 0.013 0.070 0.008 0.061
Switzerland 0.016 0.005 0.012 0.169 0.020 0.065
* Estimated with a simplified approach based on estimated value of congestion costs for passenger cars.
Table 101 - Total trucks urban congestion costs (billion Euro/year, in Euro 2016 )
HGVs LCVs* HGVs LCVs*
EU28 3.090 5.614 17.609 32.562
Austria 0.045 0.090 0.251 0.498
Belgium 0.039 0.127 0.246 0.808
Bulgaria 0.026 0.007 0.148 0.038
Croatia 0.013 0.025 0.075 0.143
Cyprus 0.002 0.013 0.011 0.074
Czech Republic 0.065 0.038 0.363 0.212
Denmark 0.028 0.067 0.161 0.383
Estonia 0.005 0.002 0.027 0.012
Finland 0.026 0.036 0.146 0.203
France 0.208 0.936 1.179 5.293
Germany 0.788 0.517 4.444 2.917
Greece 0.046 0.060 0.258 0.335
Hungary 0.060 0.043 0.343 0.244
Ireland 0.009 0.327 0.068 2.532
Italy 0.242 0.933 1.353 5.215
Latvia 0.018 0.005 0.105 0.031
Lithuania 0.022 0.012 0.128 0.070
Luxembourg 0.014 0.037 0.082 0.218
Malta 0.004 0.001 0.025 0.007
Netherlands 0.093 0.372 0.532 2.121
Poland 0.467 0.107 2.640 0.606
Portugal 0.045 0.370 0.256 2.082

Romania 0.080 0.043 0.547 0.294
Slovakia 0.031 0.012 0.179 0.067
Slovenia 0.012 0.006 0.067 0.037
Spain 0.267 0.453 1.476 2.500
Sweden 0.078 0.143 0.440 0.809
United Kingdom 0.356 0.832 2.059 4.812
Norway 0.032 0.096 0.184 0.549
Switzerland 0.017 0.028 0.102 0.163
Table 102 - Total inter-urban congestion costs on motorway network (billion Euro/year, in Euro 2016 )
Car Trucks Coaches LCVs* Car Trucks Coaches LCVs*

(HGV) (HGV)
EU28 0.670 0.194 0.034 0.122 2.913 1.086 0.203 0.498
Austria 0.013 0.005 0.001 0.002 0.051 0.026 0.005 0.009
Belgium 0.060 0.017 0.002 0.010 0.244 0.094 0.010 0.040
Bulgaria 0.000 0.000 0.000 0.000 0.002 0.000 0.000 0.000
Croatia 0.001 0.000 0.000 0.000 0.004 0.001 0.000 0.001
Cyprus - - - - - - - -
Czech Republic 0.005 0.000 0.000 0.000 0.020 0.002 0.003 0.001
Denmark 0.011 0.002 0.000 0.002 0.046 0.010 0.002 0.009
Estonia - - - - - - - -
Finland 0.003 0.001 0.000 0.000 0.014 0.007 0.001 0.001
France 0.162 0.058 0.009 0.052 0.658 0.318 0.050 0.214
Germany 0.104 0.027 0.004 0.005 0.424 0.151 0.022 0.021
Greece 0.012 0.001 0.001 0.002 0.049 0.007 0.009 0.010
Hungary 0.009 0.005 0.002 0.001 0.037 0.025 0.010 0.004
Ireland 0.015 0.003 0.001 0.004 0.062 0.018 0.005 0.018
Italy 0.072 0.014 0.006 0.012 0.295 0.077 0.034 0.048
Latvia - - - - - - - -
Lithuania 0.004 0.000 0.000 0.000 0.015 0.002 0.000 0.001


(HGV) (HGV)
Luxembourg 0.015 0.004 0.001 0.008 0.062 0.020 0.003 0.032
Malta - - - - - - - -
Netherlands 0.052 0.014 0.000 0.006 0.211 0.075 0.003 0.022
Poland 0.008 0.003 0.000 0.000 0.032 0.016 0.002 0.002
Portugal 0.025 0.004 0.001 0.005 0.102 0.023 0.005 0.019
Romania 0.002 0.000 0.000 0.000 0.010 0.002 0.001 0.001
Slovakia 0.002 0.000 0.000 0.000 0.008 0.001 0.000 0.001
Slovenia 0.005 0.000 0.000 0.001 0.023 0.002 0.002 0.003
Spain 0.042 0.024 0.004 0.003 0.165 0.127 0.022 0.014
Sweden 0.009 0.004 0.000 0.001 0.038 0.021 0.002 0.006
United Kingdom 0.037 0.007 0.001 0.005 0.150 0.041 0.004 0.022
Norway 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000
Switzerland 0.046 0.003 0.001 0.003 0.190 0.019 0.004 0.014

F.5 Detailed estimated congestion costs by country - Costs per vkm borne on congested roads
Table 103 - Car deadweight loss per vkm on congested network (€-cent/vkm, in Euro 2016 )
Country Urban area Inter-urban area
Trunk road Other urban road Motorways Other road
Over Congested Near Over Congested Near Over Congested Near Over Congested Near
capacity capacity capacity capacity capacity capacity capacity capacity
EU28 3.76 2.61 1.47 7.65 6.54 4.72 3.66 2.54 1.45 5.73 4.67 3.21
Austria 4.18 2.90 1.64 8.49 7.26 5.24 4.06 2.82 1.61 6.36 5.19 3.56
Belgium 4.35 3.01 1.70 8.84 7.55 5.45 4.26 2.96 1.68 6.67 5.44 3.73
Bulgaria 2.41 1.67 0.94 4.90 4.19 3.02 2.31 1.61 0.91 3.62 2.95 2.03
Croatia 2.90 2.01 1.14 5.90 5.04 3.64 2.78 1.93 1.10 4.35 3.55 2.44
Cyprus 3.11 2.16 1.22 6.32 5.41 3.90 2.98 2.07 1.18 4.67 3.81 2.61
Czech Republic 4.13 2.86 1.62 8.39 7.17 5.17 4.03 2.80 1.59 6.31 5.15 3.53
Denmark 3.90 2.70 1.52 7.92 6.77 4.88 3.74 2.60 1.48 5.86 4.78 3.28
Estonia 3.24 2.24 1.27 6.58 5.63 4.06 3.19 2.22 1.26 5.00 4.08 2.80
Finland 3.74 2.59 1.46 7.61 6.50 4.69 3.64 2.53 1.44 5.70 4.65 3.19
France 3.58 2.48 1.40 7.28 6.22 4.49 3.45 2.40 1.36 5.40 4.41 3.02
Germany 3.77 2.61 1.47 7.66 6.55 4.72 3.68 2.56 1.46 5.77 4.70 3.23
Greece 2.99 2.07 1.17 6.08 5.20 3.75 2.94 2.04 1.16 4.61 3.76 2.58
Hungary 3.26 2.26 1.28 6.63 5.67 4.09 3.09 2.15 1.22 4.84 3.95 2.71
Ireland 7.39 5.12 2.89 15.01 12.83 9.26 7.51 5.22 2.97 11.77 9.60 6.59

Italy 3.59 2.48 1.40 7.29 6.23 4.49 3.47 2.41 1.37 5.44 4.43 3.04
Latvia 3.29 2.28 1.29 6.69 5.72 4.13 3.09 2.15 1.22 4.84 3.95 2.71
Lithuania 3.58 2.48 1.40 7.27 6.22 4.49 3.36 2.33 1.33 5.26 4.29 2.94
Luxembourg 8.21 5.69 3.21 16.68 14.26 10.29 7.93 5.51 3.13 12.42 10.13 6.95
Malta 3.41 2.36 1.33 6.92 5.92 4.27 3.26 2.27 1.29 5.11 4.17 2.86
Netherlands 4.47 3.10 1.75 9.08 7.76 5.60 4.24 2.95 1.68 6.65 5.42 3.72
Poland 3.34 2.31 1.31 6.79 5.80 4.18 3.28 2.28 1.30 5.13 4.19 2.87
Portugal 3.28 2.27 1.28 6.66 5.69 4.11 3.16 2.20 1.25 4.95 4.04 2.77
Romania 2.99 2.07 1.17 6.08 5.20 3.75 3.44 2.39 1.36 5.39 4.39 3.01
Slovakia 4.09 2.83 1.60 8.31 7.11 5.13 3.89 2.71 1.54 6.10 4.98 3.41
Slovenia 3.16 2.19 1.24 6.43 5.50 3.97 3.02 2.10 1.19 4.73 3.86 2.65
Spain 3.66 2.54 1.43 7.44 6.36 4.59 3.59 2.50 1.42 5.63 4.59 3.15
Sweden 4.08 2.82 1.60 8.29 7.08 5.11 3.98 2.76 1.57 6.23 5.08 3.49
United Kingdom 3.66 2.54 1.43 7.44 6.36 4.59 3.58 2.49 1.41 5.61 4.57 3.14
Norway 4.93 3.42 1.93 10.02 8.57 6.18 4.43 3.08 1.75 6.94 5.66 3.88
Switzerland 5.04 3.49 1.97 10.24 8.76 6.32 4.80 3.34 1.90 7.53 6.14 4.21

Canada73 2.83 1.96 1.11 5.76 4.92 3.55
United States74 4.88 3.38 1.91 9.91 8.47 6.11
Japan75 4.57 3.16 1.79 9.28 7.93 5.72
________________________________
73
Estimations for Alberta and British Columbia.
74 Estimations for California and Missouri.
75
Estimations based on the city of Tokyo.

Table 104 - Inter-urban deadweight loss per vkm borne on congested network for trucks and coaches (€-cent/vkm, in Euro 2016 )
Country Trucks Coaches
Motorways Other road Motorways Other road
EU28 10.87 7.50 4.17 13.52 11.05 7.51 28.36 19.57 10.89 35.30 28.83 19.60
Austria 13.93 9.61 5.35 17.33 14.16 9.62 35.13 24.23 13.48 43.71 35.70 24.27
Belgium 12.55 8.66 4.82 15.61 12.75 8.67 32.39 22.34 12.43 40.30 32.92 22.38
Bulgaria 5.20 3.58 1.99 6.47 5.28 3.59 13.41 9.25 5.15 16.69 13.63 9.26
Croatia 5.87 4.05 2.25 7.30 5.96 4.05 16.52 11.40 6.34 20.56 16.80 11.42
Cyprus 7.28 5.02 2.79 9.06 7.40 5.03 22.85 15.76 8.77 28.44 23.23 15.79
Czech Republic 7.50 5.18 2.88 9.34 7.63 5.18 24.17 16.68 9.28 30.08 24.57 16.70
Denmark 11.56 7.98 4.44 14.39 11.75 7.99 34.09 23.51 13.09 42.42 34.65 23.55
Estonia 7.77 5.36 2.98 9.67 7.90 5.37 20.68 14.27 7.94 25.73 21.02 14.29
Finland 18.27 12.61 7.02 22.74 18.57 12.63 30.12 20.78 11.56 37.48 30.62 20.81
France 13.99 9.65 5.37 17.40 14.21 9.66 28.70 19.80 11.02 35.72 29.18 19.83
Germany 12.22 8.43 4.69 15.20 12.42 8.44 33.99 23.45 13.05 42.30 34.55 23.49
Greece 10.80 7.45 4.15 13.44 10.98 7.46 18.60 12.83 7.14 23.15 18.91 12.85
Hungary 7.82 5.39 3.00 9.73 7.94 5.40 18.60 12.83 7.14 23.15 18.91 12.85
Ireland 13.37 9.22 5.13 16.63 13.58 9.23 50.33 34.72 19.32 62.63 51.15 34.77
Italy 10.42 7.19 4.00 12.97 10.60 7.20 26.63 18.37 10.22 33.14 27.06 18.40
Latvia 7.58 5.23 2.91 9.43 7.70 5.23 17.75 12.25 6.81 22.09 18.04 12.26

Lithuania 6.04 4.17 2.32 7.52 6.14 4.17 20.77 14.33 7.97 25.85 21.11 14.35
Luxembourg 18.98 13.09 7.28 23.61 19.29 13.11 70.91 48.92 27.22 88.24 72.08 48.99
Malta 8.19 5.65 3.14 10.19 8.32 5.66 25.97 17.91 9.97 32.31 26.39 17.94
Netherlands 11.57 7.98 4.44 14.40 11.76 7.99 35.13 24.23 13.48 43.71 35.70 24.27
Poland 8.02 5.53 3.08 9.98 8.15 5.54 18.79 12.96 7.21 23.38 19.10 12.98
Portugal 10.55 7.28 4.05 13.13 10.72 7.29 21.34 14.72 8.19 26.56 21.69 14.74
Romania 5.46 3.77 2.10 6.79 5.55 3.77 16.05 11.07 6.16 19.98 16.32 11.09
Slovakia 7.84 5.41 3.01 9.76 7.97 5.42 21.15 14.59 8.12 26.32 21.50 14.61
Slovenia 11.05 7.62 4.24 13.75 11.23 7.63 22.76 15.70 8.74 28.32 23.13 15.72
Spain 12.57 8.67 4.83 15.64 12.78 8.68 25.21 17.39 9.68 31.37 25.62 17.42
Sweden 16.33 11.27 6.27 20.32 16.60 11.28 33.99 23.45 13.05 42.30 34.55 23.49
United Kingdom 10.28 7.09 3.95 12.79 10.45 7.10 29.65 20.45 11.38 36.90 30.14 20.48
Norway 22.52 15.53 8.64 28.02 22.89 15.56 40.89 28.20 15.70 50.88 41.56 28.25
Switzerland 10.51 7.25 4.03 13.08 10.68 7.26 44.38 30.61 17.04 55.23 45.11 30.66

Table 105 - Car delay costs per vkm borne on congested network (€-cent/vkm, in Euro 2016 )
Over capacity Congested Near capacity Over capacity Congested Near capacity Over capacity Congested Near capacity Over capacity Congested Near capacity
EU28 25.7 12.4 5.1 73.4 38.8 17.8 23.5 11.3 4.6 47.7 24.7 11.1
Austria 28.5 13.7 5.6 81.6 43.1 19.7 26.1 12.6 5.2 53.0 27.5 12.3
Belgium 29.7 14.3 5.9 84.9 44.8 20.5 27.3 13.2 5.4 55.5 28.8 12.9
Bulgaria 16.5 7.9 3.2 47.1 24.9 11.4 14.8 7.2 2.9 30.1 15.6 7.0
Croatia 19.8 9.6 3.9 56.7 29.9 13.7 17.8 8.6 3.5 36.2 18.8 8.4
Cyprus 21.2 10.2 4.2 60.7 32.1 14.7 19.1 9.2 3.8 38.8 20.1 9.0
Czech Republic 28.2 13.6 5.6 80.6 42.6 19.5 25.9 12.5 5.1 52.5 27.3 12.2
Denmark 26.6 12.8 5.2 76.0 40.2 18.4 24.0 11.6 4.7 48.7 25.3 11.3
Estonia 22.1 10.7 4.4 63.2 33.4 15.3 20.5 9.9 4.0 41.6 21.6 9.7
Finland 25.5 12.3 5.0 73.0 38.6 17.7 23.3 11.3 4.6 47.4 24.6 11.0
France 24.4 11.8 4.8 69.9 36.9 16.9 22.1 10.7 4.4 45.0 23.3 10.4
Germany 25.7 12.4 5.1 73.6 38.9 17.8 23.6 11.4 4.7 48.0 24.9 11.1
Greece 20.4 9.8 4.0 58.4 30.9 14.1 18.9 9.1 3.7 38.4 19.9 8.9
Hungary 22.2 10.7 4.4 63.6 33.6 15.4 19.9 9.6 3.9 40.3 20.9 9.4
Ireland 50.4 24.3 10.0 144.1 76.1 34.9 48.2 23.3 9.5 98.0 50.8 22.8
Italy 24.5 11.8 4.8 70.0 37.0 16.9 22.3 10.7 4.4 45.2 23.5 10.5
Latvia 22.5 10.8 4.4 64.3 34.0 15.5 19.8 9.6 3.9 40.3 20.9 9.4

Lithuania 24.4 11.8 4.8 69.8 36.9 16.9 21.6 10.4 4.3 43.8 22.7 10.2
Luxembourg 56.0 27.0 11.1 160.2 84.6 38.8 50.9 24.5 10.1 103.4 53.6 24.0
Malta 23.2 11.2 4.6 66.5 35.1 16.1 20.9 10.1 4.1 42.5 22.0 9.9
Netherlands 30.5 14.7 6.0 87.2 46.1 21.1 27.2 13.1 5.4 55.3 28.7 12.9
Poland 22.8 11.0 4.5 65.2 34.4 15.8 21.0 10.1 4.2 42.7 22.2 9.9
Portugal 22.3 10.8 4.4 63.9 33.8 15.5 20.3 9.8 4.0 41.2 21.4 9.6
Romania 20.4 9.8 4.0 58.4 30.8 14.1 22.1 10.6 4.4 44.8 23.3 10.4
Slovakia 27.9 13.5 5.5 79.8 42.2 19.3 25.0 12.1 4.9 50.8 26.3 11.8
Slovenia 21.6 10.4 4.3 61.8 32.6 14.9 19.4 9.3 3.8 39.3 20.4 9.1
Spain 25.0 12.0 4.9 71.4 37.7 17.3 23.1 11.1 4.6 46.9 24.3 10.9
Sweden 27.8 13.4 5.5 79.6 42.0 19.3 25.5 12.3 5.0 51.9 26.9 12.1
United Kingdom 25.0 12.1 4.9 71.5 37.8 17.3 23.0 11.1 4.5 46.7 24.2 10.8
Norway 33.6 16.2 6.6 96.3 50.9 23.3 28.4 13.7 5.6 57.7 29.9 13.4
Switzerland 34.4 16.6 6.8 98.4 52.0 23.8 30.9 14.9 6.1 62.7 32.5 14.6
Canada76 19.3 9.3 3.8 55.2 29.2 13.4
United States77 33.2 16.0 6.6 95.1 50.2 23.0
________________________________
76 Estimations for Alberta and British Columbia.
77
Estimations for California and Missouri.

Japan78 31.1 15.0 6.1 89.0 47.0 21.5
Table 106 - Inter-urban delay congestion costs per vkm borne on congested network for trucks and coaches (Euro Cent/vkm, in Euro 2016 )
Congested Congested Near Congested Congested Near

Over capacity Near capacity Over capacity Over capacity Near capacity Over capacity
capacity capacity
EU28 97.6 47.1 19.3 163.9 85.0 38.1 254.8 122.9 50.3 427.7 221.9 99.4
Austria 125.1 60.3 24.7 210.0 108.9 48.8 315.5 152.2 62.3 529.6 274.7 123.0
Belgium 112.7 54.4 22.3 189.2 98.1 43.9 290.9 140.3 57.5 488.3 253.3 113.4
Bulgaria 46.7 22.5 9.2 78.4 40.6 18.2 120.4 58.1 23.8 202.2 104.9 47.0
Croatia 52.7 25.4 10.4 88.5 45.9 20.6 148.4 71.6 29.3 249.2 129.2 57.9
Cyprus 65.4 31.5 12.9 109.8 56.9 25.5 205.3 99.0 40.5 344.5 178.7 80.0
Czech Republic 67.4 32.5 13.3 113.1 58.7 26.3 217.1 104.7 42.9 364.5 189.1 84.7
________________________________
78


capacity capacity
Denmark 103.9 50.1 20.5 174.3 90.4 40.5 306.2 147.7 60.5 514.0 266.6 119.4
Estonia 69.8 33.7 13.8 117.1 60.8 27.2 185.8 89.6 36.7 311.8 161.7 72.4
Finland 164.1 79.2 32.4 275.5 142.9 64.0 270.6 130.5 53.4 454.2 235.6 105.5
France 125.6 60.6 24.8 210.9 109.4 49.0 257.8 124.3 50.9 432.8 224.5 100.5
Germany 109.7 52.9 21.7 184.2 95.6 42.8 305.3 147.3 60.3 512.5 265.9 119.1
Greece 97.0 46.8 19.2 162.9 84.5 37.8 167.1 80.6 33.0 280.5 145.5 65.2
Hungary 70.2 33.9 13.9 117.8 61.1 27.4 167.1 80.6 33.0 280.5 145.5 65.2
Ireland 120.1 57.9 23.7 201.5 104.5 46.8 452.1 218.0 89.3 758.9 393.6 176.3
Italy 93.6 45.2 18.5 157.2 81.5 36.5 239.2 115.3 47.2 401.5 208.3 93.3
Latvia 68.1 32.8 13.4 114.2 59.3 26.5 159.5 76.9 31.5 267.7 138.8 62.2
Lithuania 54.3 26.2 10.7 91.1 47.3 21.2 186.6 90.0 36.9 313.2 162.5 72.8
Luxembourg 170.5 82.2 33.7 286.1 148.4 66.5 637.0 307.2 125.8 1069.2 554.6 248.4
Malta 73.5 35.5 14.5 123.4 64.0 28.7 233.3 112.5 46.1 391.5 203.1 91.0
Netherlands 103.9 50.1 20.5 174.4 90.5 40.5 315.5 152.2 62.3 529.6 274.7 123.0
Poland 72.1 34.7 14.2 120.9 62.7 28.1 168.8 81.4 33.3 283.3 147.0 65.8
Portugal 94.8 45.7 18.7 159.1 82.5 36.9 191.7 92.4 37.9 321.8 166.9 74.7
Romania 49.0 23.6 9.7 82.3 42.7 19.1 144.2 69.5 28.5 242.0 125.6 56.2
Slovakia 70.4 34.0 13.9 118.2 61.3 27.5 190.0 91.6 37.5 318.9 165.4 74.1


capacity capacity
Slovenia 99.2 47.9 19.6 166.6 86.4 38.7 204.4 98.6 40.4 343.1 178.0 79.7
Spain 112.9 54.4 22.3 189.5 98.3 44.0 226.5 109.2 44.7 380.1 197.2 88.3
Sweden 146.7 70.8 29.0 246.3 127.7 57.2 305.3 147.3 60.3 512.5 265.9 119.1
United Kingdom 92.3 44.5 18.2 155.0 80.4 36.0 266.3 128.4 52.6 447.1 231.9 103.9
Norway 202.3 97.5 40.0 339.5 176.1 78.9 367.3 177.1 72.5 616.5 319.8 143.2
Switzerland 94.4 45.5 18.6 158.4 82.2 36.8 398.6 192.2 78.7 669.2 347.1 155.5

F.6 Detailed estimated congestion costs by country - Average costs on the
whole network
Table 107 - Car average 79 delay congestion cost and deadweight loss generated per vkm, estimated using the
simplified approach (€-cent/vkm, in Euro 2016 )
Country Deadweight Delay cost Deadweight loss Delay cost Deadweight loss Delay cost
loss
EU28 1.14 6.69 3.09 17.73 0.27 1.74
Austria 1.10 6.45 3.05 17.47 0.29 1.83
Belgium 0.90 5.12 2.38 13.23 0.28 1.78
Bulgaria 1.50 9.26 3.86 24.55 0.76 4.43
Croatia 0.65 3.81 1.42 8.16 0.04 0.39
Cyprus 0.74 4.26 2.39 13.54 0.06 0.44
Czech Republic 1.20 6.78 2.72 15.41 0.00 0.00
Denmark 0.94 5.47 2.18 12.20 0.13 1.01
Estonia 1.02 5.62 1.96 11.10 0.52 2.67
Finland 0.54 3.23 1.35 7.52 0.10 0.91
France 0.71 4.01 1.90 10.64 0.08 0.44
Germany 1.04 6.02 2.27 12.83 0.50 3.06
Greece 1.07 6.15 3.50 19.74 0.17 1.13
Hungary 1.08 5.92 2.15 11.91 0.25 1.21
Ireland 1.00 5.96 2.40 13.64 0.20 1.60
Italy 1.59 12.67 4.26 33.02 0.45 3.95
Latvia 0.96 5.42 3.90 21.78 0.20 1.19
Lithuania 0.89 5.19 2.47 14.04 0.05 0.43
Luxembourg 0.71 4.14 1.81 10.37 0.04 0.32
Malta 1.90 11.28 4.85 28.56 0.97 5.83
Netherlands 0.64 3.98 1.28 7.96 0.00 0.00
Poland 1.23 6.97 4.75 27.06 0.27 1.49
Portugal 1.28 7.88 2.78 15.73 0.32 2.82
Romania 1.35 7.87 3.10 17.46 0.41 2.70
________________________________
79
Based on the whole network (not only congested network).

Slovakia 1.77 12.69 3.73 25.67 0.23 2.48
Slovenia 0.75 4.61 1.60 9.20 0.20 1.57
Spain 0.35 2.08 0.90 5.23 0.13 0.79
Sweden 1.44 8.22 3.52 19.42 0.33 2.19
United Kingdom 1.34 7.57 3.63 20.58 0.10 0.57
Norway 1.40 8.18 3.27 18.94 0.17 1.09
Switzerland 0.88 5.01 2.19 12.58 0.17 0.94

Table 108 – Bus/coaches average80 delay congestion cost and deadweight loss generated per vkm, estimated
using the simplified approach (€-cent/vkm, in Euro 2016 )
loss
EU28 2.69 15.89 6.09 35.52 0.48 3.09
Austria 2.57 15.20 6.02 35.20 0.48 3.13
Belgium 1.74 9.94 4.76 26.47 0.54 3.36
Bulgaria 4.42 27.68 7.71 49.11 1.61 9.42
Croatia 1.48 8.62 2.84 16.33 0.05 0.51
Cyprus 1.45 8.30 4.78 27.07 0.12 0.82
Czech Republic 3.05 17.29 5.44 30.83 0.00 0.00
Denmark 1.69 9.94 4.35 24.41 0.26 2.14
Estonia 2.34 12.94 3.91 22.20 1.10 5.69
Finland 1.07 6.45 2.70 15.04 0.20 1.83
France 2.21 12.37 3.80 21.27 0.15 0.88
Germany 2.40 13.80 4.53 25.65 0.97 5.84
Greece 2.45 14.20 7.00 39.48 0.40 2.76
Hungary 1.81 9.89 4.29 23.83 0.40 1.91
Ireland 1.89 11.21 4.79 27.29 0.42 3.06
Italy 5.45 42.36 8.52 66.03 0.74 6.01
Latvia 2.50 13.97 7.80 43.57 0.28 1.57
Lithuania 1.79 10.39 4.94 28.09 0.09 0.86
Luxembourg 2.04 11.69 3.62 20.75 0.09 0.62
Malta 5.53 32.83 9.69 57.11 1.99 12.14
Netherlands 1.28 7.96 2.55 15.92 0.00 0.00
Poland 3.92 22.25 9.50 54.13 0.50 2.71
Portugal 3.11 18.64 5.56 31.45 0.64 5.75
Romania 2.80 15.78 6.20 34.92 0.48 2.77
Slovakia 4.23 29.56 7.47 51.34 0.31 3.24
Slovenia 1.76 10.70 3.19 18.41 0.45 3.66
Spain 0.69 4.07 1.80 10.47 0.25 1.52
________________________________
80

Sweden 2.22 12.40 7.03 38.84 0.36 2.16
United Kingdom 3.97 22.52 7.26 41.16 0.20 1.14
Norway 4.19 24.32 6.55 37.89 0.37 2.34
Switzerland 2.50 14.30 4.38 25.15 0.35 1.91

Table 109 – Freight HGV average81 delay congestion cost and deadweight loss generated per vkm, estimated
using the simplified approach (€-cent/vkm, in Euro 2016 )
loss
EU28 1.84 10.85 5.96 34.10 0.45 3.02
Austria 1.83 10.81 5.89 33.65 0.46 3.13
Belgium 1.77 10.06 4.76 26.47 0.49 3.04
Bulgaria 2.28 13.75 7.71 49.11 1.32 7.51
Croatia 1.03 6.09 2.84 16.33 0.06 0.58
Cyprus 1.51 8.64 4.78 27.07 0.11 0.74
Czech Republic 1.90 10.79 5.44 30.83 0.00 0.00
Denmark 1.73 10.13 4.35 24.41 0.26 2.10
Estonia 1.95 10.67 3.91 22.20 1.03 5.25
Finland 0.97 5.83 2.70 15.04 0.16 1.49
France 0.89 5.00 3.80 21.27 0.16 0.93
Germany 1.74 10.09 4.53 25.65 0.93 5.55
Greece 2.26 13.00 7.00 39.48 0.33 2.18
Hungary 1.77 9.66 4.29 23.83 0.42 2.03
Ireland 1.73 10.43 4.79 27.29 0.41 3.14
Italy 1.41 10.68 8.52 66.03 0.62 4.53
Latvia 1.48 8.33 7.80 43.57 0.36 2.11
Lithuania 1.63 9.46 4.94 28.09 0.08 0.70
Luxembourg 0.94 5.45 3.62 20.75 0.09 0.62
Malta 3.01 17.50 9.69 57.11 1.84 10.51
Netherlands 1.28 7.96 2.55 15.92 0.00 0.00
Poland 1.55 8.70 9.50 54.13 0.47 2.51
Portugal 2.31 14.39 5.56 31.45 0.64 5.60
Romania 1.35 7.68 6.20 34.92 0.47 2.74
Slovakia 2.84 20.34 7.47 51.34 0.34 3.64
Slovenia 1.41 8.73 3.19 18.41 0.41 3.29
Spain 0.71 4.13 1.80 10.47 0.24 1.42
________________________________
81
Based on the whole network (not only congested n etwork).

Sweden 1.38 7.77 7.03 38.84 0.35 2.13
United Kingdom 1.61 9.15 7.26 41.16 0.20 1.15
Norway 1.92 11.21 6.55 37.89 0.33 2.07
Switzerland 1.19 6.73 4.38 25.15 0.39 2.13

Table 110 – Light commercial vehicle average 82 delay congestion cost and deadweight loss generated per vkm,
estimated using the simplified approach (€-cent/vkm, in Euro 2016 )
loss
EU28 1.97 11.61 4.71 27.42 0.47 2.92
Austria 1.97 11.62 4.68 27.24 0.50 3.13
Belgium 1.41 8.05 3.57 19.85 0.42 2.62
Bulgaria 2.25 13.87 5.79 36.83 1.14 6.62
Croatia 0.77 4.57 2.13 12.24 0.04 0.43
Cyprus 1.19 6.84 3.59 20.30 0.09 0.65
Czech Republic 1.43 8.09 4.08 23.12 0.00 0.00
Denmark 1.39 8.06 3.27 18.31 0.19 1.51
Estonia 1.56 8.58 2.93 16.65 0.82 4.23
Finland 0.80 4.75 2.02 11.28 0.13 1.23
France 1.07 6.01 2.85 15.95 0.11 0.66
Germany 1.47 8.54 3.40 19.24 0.76 4.67
Greece 2.34 13.36 5.25 29.61 0.24 1.58
Hungary 1.33 7.25 3.22 17.87 0.31 1.52
Ireland 1.62 9.68 3.60 20.47 0.30 2.42
Italy 3.50 27.34 6.39 49.52 0.61 5.16
Latvia 1.69 9.55 5.85 32.68 0.31 1.84
Lithuania 1.34 7.75 3.70 21.07 0.06 0.58
Luxembourg 1.42 8.17 2.72 15.56 0.06 0.47
Malta 2.85 16.90 7.27 42.84 1.45 8.71
Netherlands 0.96 5.97 1.91 11.94 0.00 0.00
Poland 3.10 17.62 7.12 40.60 0.42 2.31
Portugal 1.99 12.19 4.17 23.59 0.48 4.26
Romania 3.17 17.89 4.65 26.19 0.41 2.49
Slovakia 2.13 15.25 5.60 38.51 0.26 2.73
Slovenia 1.11 6.82 2.39 13.80 0.29 2.35
Spain 0.56 3.27 1.35 7.85 0.19 1.16
________________________________
82

Sweden 3.54 19.61 5.27 29.13 0.29 1.80
United Kingdom 2.00 11.36 5.45 30.87 0.15 0.85
Norway 1.98 11.56 4.91 28.42 0.26 1.66
Switzerland 1.32 7.52 3.29 18.86 0.26 1.41
Table 111 - Car average 83 delay congestion cost and deadweight loss brone per vkm
(€-cent/vkm, in Euro 2016 )
loss
EU28 1.21 7.03 3.31 19.03 0.27 1.66
Austria 0.97 5.45 2.59 14.39 0.30 1.76
Belgium 1.51 9.23 4.04 25.69 0.71 4.03
Bulgaria 0.69 4.04 1.51 8.69 0.04 0.38
Croatia 0.87 4.98 2.81 15.92 0.07 0.48
Cyprus 1.47 8.31 3.33 18.88 - -
Czech Republic 1.13 6.54 2.61 14.63 0.15 1.20
Denmark 1.12 6.06 2.12 12.03 0.58 2.85
Estonia 0.58 3.46 1.44 8.03 0.11 1.01
Finland 0.74 4.09 1.98 11.05 0.07 0.34
France 1.04 5.83 2.30 12.99 0.50 2.73
Germany 1.07 6.13 3.54 19.97 0.16 1.01
Greece 1.17 6.37 2.33 12.93 0.26 1.23
Hungary 1.19 7.04 2.89 16.46 0.22 1.69
Ireland 2.17 17.11 6.08 47.15 0.49 4.24
Italy 0.99 5.56 4.06 22.70 0.20 1.13
Latvia 0.95 5.51 2.63 14.94 0.05 0.43
Lithuania 0.85 4.93 2.14 12.25 0.05 0.44
Luxembourg 2.83 16.86 7.15 42.16 1.46 8.88
Malta 0.64 4.02 1.29 8.04 - -
Netherlands 1.39 7.87 5.43 30.96 0.28 1.58
Poland 1.42 8.75 3.01 17.03 0.39 3.41
________________________________
83

Portugal 1.63 9.33 3.88 21.85 0.41 2.59
Romania 2.12 15.08 4.47 30.75 0.26 2.77
Slovakia 0.89 5.49 1.86 10.73 0.25 2.02
Slovenia 0.38 2.20 0.92 5.37 0.15 0.89
Spain 1.51 8.44 3.79 20.95 0.29 1.70
Sweden 1.41 7.94 3.83 21.72 0.10 0.52
United Kingdom 1.51 8.77 3.52 20.35 0.18 1.14
Norway 0.90 5.07 2.25 12.91 0.18 0.85
Switzerland 0.92 5.25 2.05 11.98 0.46 2.48
Table 112 - Average 84 inter-urban delay congestion costs and deadweight loss borne for HGV, coaches and LCV
Trucks (HGV) Coaches LCVs* Trucks (HGV) Coaches LCVs*
EU28 0.69 1.50 0.33 6.20 14.49 1.92
Austria 1.13 1.49 0.22 8.87 17.93 1.28
Belgium 2.97 12.76 0.85 24.90 36.24 4.87
Bulgaria 0.04 0.17 0.02 0.46 2.59 0.15
Croatia 0.17 0.28 0.04 1.45 3.32 0.27
Cyprus - - - - - -
Czech Republic 0.12 1.23 0.06 1.14 8.41 0.46
Denmark 0.92 4.41 0.58 7.59 30.46 2.88
Estonia 0.02 0.57 0.06 0.13 7.54 0.54
Finland 0.54 0.30 0.06 4.47 3.39 0.31
France 2.23 1.50 0.65 24.90 26.71 3.58
Germany 0.63 0.97 0.13 5.41 11.01 0.82
Greece 0.34 1.41 0.13 2.65 9.18 0.61
Hungary 0.43 1.07 0.11 2.99 12.07 0.85
Ireland 1.48 4.21 0.29 15.07 33.21 2.45
Italy 0.50 0.81 0.24 3.43 10.15 1.36
Latvia 0.10 0.26 0.03 0.91 2.88 0.25
Lithuania 0.03 2.75 0.03 0.16 3.36 0.21
________________________________
84

Luxembourg 0.95 23.52 0.89 5.81 95.66 5.39
Malta - - - - - -
Netherlands 0.47 10.99 0.23 3.24 15.66 1.29
Poland 0.40 2.24 0.21 3.65 22.65 1.84
Portugal 0.85 3.34 0.19 7.31 20.73 1.10
Romania 0.22 0.73 0.08 2.26 14.99 0.78
Slovakia 0.18 2.13 0.11 1.70 12.82 0.88
Slovenia 0.07 0.65 0.10 0.46 8.10 0.59
Spain 0.76 1.36 0.15 6.51 13.99 0.82
Sweden 0.40 0.29 0.09 2.98 5.35 0.48
United Kingdom 0.44 1.63 0.16 3.43 11.00 0.98
Norway 0.63 0.72 0.20 6.10 9.06 0.93
Switzerland 1.73 7.00 0.39 18.16 27.51 2.05
Table 113 - Average urban delay congestion costs and deadweight loss borne for trucks (€-cent/vkm, in
Euro2016)
EU28 9.11 3.31 51.94 19.21
Austria 8.67 2.42 48.14 13.42
Belgium 11.61 5.04 73.88 32.11
Bulgaria 3.32 0.93 19.08 5.32
Croatia 5.79 1.88 32.80 10.62
Cyprus 7.95 2.43 45.04 13.79
Czech Republic 4.74 1.17 26.58 6.58
Denmark 6.40 2.14 36.31 12.13
Estonia 3.42 1.02 19.07 5.68
Finland 9.69 1.85 54.19 10.35
France 9.09 3.14 51.42 17.77
Germany 11.47 3.33 64.71 18.77
Greece 8.34 1.69 46.32 9.37
Hungary 7.13 1.61 40.61 9.17
Ireland 10.56 4.23 81.87 32.77

Italy 11.91 5.26 66.56 29.39
Latvia 6.28 1.68 35.73 9.57
Lithuania 3.76 1.13 21.52 6.45
Luxembourg 16.71 4.10 98.47 24.13
Malta 3.16 1.33 19.72 8.31
Netherlands 14.45 4.83 82.36 27.52
Poland 7.19 1.71 40.67 9.68
Portugal 12.65 3.25 71.21 18.29
Romania 6.93 1.93 47.66 13.28
Slovakia 3.66 0.95 21.09 5.46
Slovenia 3.30 0.73 19.19 4.22
Spain 12.95 3.32 71.50 18.32
Sweden 15.36 3.70 87.03 20.96
United Kingdom 9.86 3.26 57.04 18.89
Norway 11.16 2.68 64.06 15.41
Switzerland 4.37 1.90 25.57 11.12
* Estimated with a simplified approach based on estimated valued for passenger cars.
Table 114 - Average inter-urban delay congestion costs and deadweight loss borne on motorway network

(HGV) (HGV)
EU28 0.10 0.46 0.49 0.12 0.42 2.55 2.87 0.48
Austria 0.08 0.69 0.61 0.06 0.31 3.78 3.49 0.23
Belgium 0.24 1.67 2.03 0.29 0.97 9.21 11.70 1.17
Bulgaria 0.01 0.01 0.02 0.00 0.03 0.04 0.13 0.02
Croatia 0.02 0.05 0.08 0.01 0.06 0.27 0.48 0.04
Cyprus - - - - - - - -
Czech Republic 0.06 0.06 0.39 0.02 0.26 0.31 2.35 0.10
Denmark 0.17 0.65 1.43 0.16 0.69 3.55 8.33 0.66
Estonia - - - - - - - -
Finland 0.05 0.96 0.81 0.04 0.20 5.31 4.69 0.18
France 0.18 1.88 1.15 0.23 0.72 10.35 6.67 0.93

Germany 0.06 0.35 0.68 0.05 0.23 1.94 3.92 0.20
Greece 0.14 0.28 0.41 0.08 0.57 1.53 2.41 0.34
Hungary 0.23 0.83 0.90 0.11 0.91 4.43 5.05 0.45
Ireland 0.22 0.66 0.82 0.14 0.87 3.56 4.66 0.58
Italy 0.07 0.29 0.25 0.08 0.28 1.63 1.45 0.33
Latvia - - - - - - - -
Lithuania 0.11 0.04 0.45 0.05 0.43 0.21 2.58 0.21
Luxembourg 1.05 1.43 10.49 0.64 4.24 7.75 60.30 2.59
Malta - - - -
Netherlands 0.12 0.44 0.87 0.09 0.48 2.43 4.97 0.38
Poland 0.24 0.39 1.49 0.13 0.94 2.11 8.42 0.50
Portugal 0.16 0.26 0.60 0.10 0.64 1.45 3.53 0.43
Romania 0.05 0.04 0.10 0.02 0.19 0.23 0.59 0.07
Slovakia 0.06 0.03 0.37 0.03 0.27 0.19 2.18 0.12
Slovenia 0.10 0.06 0.69 0.07 0.42 0.36 4.06 0.28
Spain 0.07 0.26 0.28 0.06 0.28 1.41 1.59 0.25
Sweden 0.10 1.49 1.66 0.09 0.38 8.15 9.43 0.34
United Kingdom 0.05 0.20 0.53 0.04 0.21 1.12 3.09 0.18
Norway 0.00 0.03 0.08 0.00 0.02 0.18 0.48 0.02
Switzerland 0.24 0.64 2.05 0.21 1.00 3.60 12.01 0.86

F.7 Detailed estimated congestion costs by country - Marginal costs
Table 115 - Car social marginal congestion costs per vkm (€-cent/vkm, in Euro 2016 )
EU28 32.1 24.8 17.4 66.3 58.2 47.2 29.4 22.6 15.9 46.4 39.6 31.2
Austria 35.6 27.5 19.4 73.6 64.6 52.4 32.6 25.2 17.7 51.5 44.0 34.6
Belgium 37.1 28.6 20.1 76.6 67.3 54.5 34.2 26.4 18.6 54.0 46.1 36.3
Bulgaria 20.6 15.9 11.2 42.5 37.3 30.2 18.5 14.3 10.1 29.3 25.0 19.7
Croatia 24.8 19.1 13.4 51.2 44.9 36.4 22.3 17.2 12.1 35.2 30.1 23.7
Cyprus 26.5 20.5 14.4 54.8 48.1 39.0 23.9 18.4 13.0 37.8 32.2 25.4
Czech Republic 35.2 27.2 19.1 72.7 63.9 51.8 32.3 25.0 17.6 51.1 43.6 34.3
Denmark 33.2 25.6 18.0 68.6 60.2 48.9 30.0 23.1 16.3 47.4 40.5 31.8
Estonia 27.6 21.3 15.0 57.1 50.1 40.6 25.6 19.8 13.9 40.5 34.6 27.2
Finland 31.9 24.6 17.3 65.9 57.9 46.9 29.2 22.5 15.9 46.1 39.4 31.0
France 30.5 23.6 16.6 63.1 55.4 44.9 27.7 21.4 15.0 43.7 37.3 29.4
Germany 32.1 24.8 17.5 66.4 58.3 47.3 29.5 22.8 16.0 46.7 39.8 31.4
Greece 25.5 19.7 13.9 52.7 46.3 37.5 23.6 18.2 12.8 37.3 31.8 25.1
Hungary 27.8 21.5 15.1 57.5 50.4 40.9 24.8 19.1 13.5 39.2 33.5 26.3

Ireland 63.0 48.6 34.2 130.1 114.2 92.6 60.3 46.5 32.7 95.2 81.3 64.0
Italy 30.6 23.6 16.6 63.2 55.5 45.0 27.8 21.5 15.1 44.0 37.5 29.6
Latvia 28.1 21.7 15.3 58.0 50.9 41.3 24.8 19.1 13.5 39.2 33.5 26.3
Lithuania 30.5 23.5 16.6 63.1 55.4 44.9 27.0 20.8 14.6 42.6 36.3 28.6
Luxembourg 70.0 54.0 38.0 144.6 127.0 103.0 63.6 49.1 34.6 100.5 85.8 67.6
Malta 29.0 22.4 15.8 60.0 52.7 42.7 26.2 20.2 14.2 41.3 35.3 27.8
Netherlands 38.1 29.4 20.7 78.7 69.1 56.0 34.1 26.3 18.5 53.8 45.9 36.2
Poland 28.5 22.0 15.5 58.8 51.6 41.9 26.3 20.3 14.3 41.5 35.5 27.9
Portugal 27.9 21.6 15.2 57.7 50.7 41.1 25.3 19.6 13.8 40.0 34.2 26.9
Romania 25.5 19.7 13.9 52.7 46.3 37.5 27.6 21.3 15.0 43.6 37.2 29.3
Slovakia 34.9 26.9 18.9 72.1 63.3 51.3 31.2 24.1 17.0 49.4 42.1 33.2
Slovenia 27.0 20.8 14.7 55.7 48.9 39.7 24.2 18.7 13.2 38.3 32.7 25.7
Spain 31.2 24.1 17.0 64.5 56.6 45.9 28.8 22.3 15.7 45.6 38.9 30.6
Sweden 34.8 26.8 18.9 71.8 63.1 51.1 31.9 24.6 17.3 50.4 43.1 33.9
United Kingdom 31.2 24.1 17.0 64.5 56.7 45.9 28.7 22.2 15.6 45.4 38.7 30.5
Norway 42.1 32.4 22.8 86.9 76.3 61.9 35.5 27.4 19.3 56.1 47.9 37.7
Switzerland 43.0 33.2 23.3 88.8 77.9 63.2 38.6 29.8 20.9 60.9 52.0 40.9

Canada85 24.1 18.6 13.1 49.8 43.7 35.5
United States86 41.5 32.0 22.6 85.8 75.3 61.1
Japan87 38.9 30.0 21.1 80.4 70.5 57.2
________________________________
85
Estimations for Alberta and British Columbia.
86 Estimations for California and Missouri.
87

Table 116 - Inter-urban social marginal congestion costs for trucks and coaches (€-cent/vkm, in Euro 2016 )
EU28 122.0 94.2 66.3 159.3 136.0 107.1 318.5 245.8 173.0 415.8 355.0 279.4
Austria 156.4 120.7 84.9 204.2 174.3 137.2 394.4 304.3 214.2 514.9 439.6 346.0
Belgium 140.9 108.7 76.5 183.9 157.0 123.6 363.7 280.6 197.5 474.8 405.3 319.1
Bulgaria 58.3 45.0 31.7 76.2 65.0 51.2 150.6 116.2 81.8 196.6 167.8 132.1
Croatia 65.9 50.8 35.8 86.0 73.4 57.8 185.5 143.2 100.8 242.2 206.8 162.8
Cyprus 81.7 63.1 44.4 106.7 91.1 71.7 256.6 198.0 139.4 335.0 286.0 225.1
Czech Republic 84.3 65.0 45.8 110.0 93.9 73.9 271.4 209.4 147.4 354.4 302.5 238.1
Denmark 129.8 100.2 70.5 169.5 144.7 113.9 382.7 295.3 207.9 499.7 426.6 335.8
Estonia 87.2 67.3 47.4 113.9 97.2 76.5 232.2 179.2 126.1 303.1 258.8 203.7
Finland 205.2 158.3 111.5 267.9 228.7 180.0 338.2 261.0 183.7 441.6 377.0 296.7
France 157.0 121.2 85.3 205.0 175.0 137.8 322.3 248.7 175.1 420.8 359.2 282.8
Germany 137.2 105.9 74.5 179.1 152.9 120.4 381.7 294.5 207.3 498.3 425.4 334.9
Greece 121.3 93.6 65.9 158.4 135.2 106.4 208.9 161.2 113.5 272.7 232.8 183.3
Hungary 87.8 67.7 47.7 114.6 97.8 77.0 208.9 161.2 113.5 272.7 232.8 183.3
Ireland 150.1 115.8 81.5 195.9 167.3 131.7 565.1 436.0 307.0 737.8 629.8 495.8
Italy 117.0 90.3 63.6 152.8 130.5 102.7 299.0 230.7 162.4 390.3 333.2 262.3
Latvia 85.1 65.6 46.2 111.1 94.8 74.6 199.3 153.8 108.3 260.2 222.2 174.9

Lithuania 67.8 52.4 36.9 88.6 75.6 59.5 233.3 180.0 126.7 304.5 260.0 204.7
Luxembourg 213.1 164.4 115.7 278.2 237.5 187.0 796.2 614.4 432.5 1039.5 887.4 698.6
Malta 91.9 70.9 49.9 120.0 102.4 80.6 291.6 225.0 158.4 380.7 325.0 255.8
Netherlands 129.9 100.2 70.6 169.6 144.8 114.0 394.4 304.3 214.2 514.9 439.6 346.0
Poland 90.1 69.5 48.9 117.6 100.4 79.0 211.0 162.8 114.6 275.5 235.2 185.1
Portugal 118.4 91.4 64.3 154.6 132.0 103.9 239.6 184.9 130.2 312.8 267.1 210.2
Romania 61.3 47.3 33.3 80.0 68.3 53.8 180.2 139.1 97.9 235.3 200.9 158.1
Slovakia 88.0 67.9 47.8 114.9 98.1 77.2 237.5 183.2 129.0 310.1 264.7 208.4
Slovenia 124.0 95.7 67.4 161.9 138.2 108.8 255.5 197.2 138.8 333.6 284.8 224.2
Spain 141.1 108.9 76.7 184.3 157.3 123.8 283.1 218.4 153.8 369.6 315.5 248.4
Sweden 183.4 141.5 99.6 239.4 204.4 160.9 381.7 294.5 207.3 498.3 425.4 334.9
United Kingdom 115.4 89.0 62.7 150.7 128.6 101.2 332.9 256.9 180.8 434.6 371.0 292.1
Norway 252.8 195.1 137.3 330.1 281.8 221.8 459.1 354.2 249.4 599.4 511.7 402.8
Switzerland 118.0 91.0 64.1 154.0 131.5 103.5 498.3 384.5 270.7 650.6 555.4 437.2

Table 117 - Inter-urban social marginal congestion costs generated for HGV, estimated using the simplified approach (€-cent/vkm, in Euro 2016 )
EU28 64.2 49.5 34.9 132.6 116.4 94.4 58.7 45.3 31.9 92.7 79.2 62.3
Austria 71.3 55.0 38.7 147.3 129.3 104.8 65.2 50.3 35.4 103.0 87.9 69.2
Belgium 74.2 57.2 40.3 153.2 134.5 109.1 68.3 52.7 37.1 108.0 92.2 72.5
Bulgaria 41.1 31.7 22.3 85.0 74.6 60.5 37.1 28.6 20.1 58.6 50.0 39.4
Croatia 49.5 38.2 26.9 102.3 89.8 72.8 44.6 34.4 24.2 70.5 60.2 47.4
Cyprus 53.1 40.9 28.8 109.7 96.3 78.1 47.8 36.9 26.0 75.5 64.5 50.8
Czech Republic 70.4 54.3 38.2 145.5 127.7 103.6 64.7 49.9 35.1 102.2 87.2 68.7
Denmark 66.4 51.3 36.1 137.3 120.5 97.7 60.0 46.3 32.6 94.8 80.9 63.7
Estonia 55.3 42.6 30.0 114.2 100.2 81.3 51.2 39.5 27.8 81.0 69.1 54.4
Finland 63.8 49.2 34.7 131.9 115.8 93.9 58.4 45.0 31.7 92.2 78.7 62.0
France 61.1 47.1 33.2 126.2 110.8 89.8 55.3 42.7 30.1 87.4 74.6 58.8
Germany 64.3 49.6 34.9 132.8 116.6 94.5 59.1 45.6 32.1 93.3 79.7 62.7
Greece 51.0 39.4 27.7 105.5 92.6 75.1 47.2 36.4 25.6 74.6 63.7 50.1
Hungary 55.6 42.9 30.2 114.9 100.9 81.8 49.6 38.3 27.0 78.4 66.9 52.7
Ireland 125.9 97.2 68.4 260.3 228.4 185.3 120.6 93.0 65.5 190.5 162.6 128.0
Italy 61.2 47.2 33.2 126.4 110.9 90.0 55.7 43.0 30.2 88.0 75.1 59.1
Latvia 56.2 43.3 30.5 116.1 101.9 82.6 49.6 38.3 27.0 78.4 66.9 52.7

Lithuania 61.0 47.1 33.2 126.1 110.7 89.8 53.9 41.6 29.3 85.2 72.7 57.2
Luxembourg 140.0 108.0 76.0 289.3 253.9 205.9 127.3 98.2 69.1 201.0 171.6 135.1
Malta 58.1 44.8 31.5 120.0 105.3 85.4 52.3 40.4 28.4 82.6 70.5 55.5
Netherlands 76.2 58.8 41.4 157.5 138.2 112.1 68.1 52.6 37.0 107.6 91.9 72.3
Poland 56.9 43.9 30.9 117.7 103.3 83.8 52.6 40.6 28.6 83.1 70.9 55.8
Portugal 55.9 43.1 30.3 115.4 101.3 82.2 50.7 39.1 27.5 80.1 68.4 53.8
Romania 51.0 39.4 27.7 105.4 92.5 75.0 55.2 42.6 30.0 87.2 74.4 58.6
Slovakia 69.8 53.8 37.9 144.2 126.5 102.6 62.5 48.2 33.9 98.7 84.3 66.3
Slovenia 54.0 41.6 29.3 111.5 97.9 79.4 48.4 37.4 26.3 76.5 65.3 51.4
Spain 62.4 48.2 33.9 129.0 113.2 91.8 57.7 44.5 31.3 91.1 77.8 61.2
Sweden 69.5 53.6 37.8 143.7 126.1 102.3 63.9 49.3 34.7 100.9 86.1 67.8
United Kingdom 62.5 48.2 33.9 129.1 113.3 91.9 57.4 44.3 31.2 90.7 77.5 61.0
Norway 84.1 64.9 45.7 173.8 152.6 123.7 71.1 54.8 38.6 112.2 95.8 75.4
Switzerland 85.9 66.3 46.7 177.6 155.9 126.4 77.1 59.5 41.9 121.8 104.0 81.9

Table 118 - Social marginal congestion costs generated for LCVs per vkm, estimated using the simplified approach (€-cent/vkm, in Euro 2016 )
EU28 48.1 37.1 26.1 99.4 87.3 70.8 44.0 34.0 23.9 69.6 59.4 46.7
Austria 53.5 41.2 29.0 110.5 97.0 78.6 48.9 37.7 26.6 77.3 66.0 51.9
Belgium 55.6 42.9 30.2 114.9 100.9 81.8 51.3 39.5 27.8 81.0 69.1 54.4
Bulgaria 30.8 23.8 16.8 63.7 56.0 45.4 27.8 21.5 15.1 43.9 37.5 29.5
Croatia 37.1 28.7 20.2 76.7 67.4 54.6 33.5 25.8 18.2 52.9 45.1 35.5
Cyprus 39.8 30.7 21.6 82.2 72.2 58.5 35.9 27.7 19.5 56.6 48.4 38.1
Czech Republic 52.8 40.7 28.7 109.1 95.8 77.7 48.5 37.4 26.3 76.6 65.4 51.5
Denmark 49.8 38.4 27.1 103.0 90.4 73.3 45.0 34.7 24.4 71.1 60.7 47.8
Estonia 41.4 32.0 22.5 85.6 75.2 61.0 38.4 29.7 20.9 60.7 51.8 40.8
Finland 47.9 36.9 26.0 98.9 86.8 70.4 43.8 33.8 23.8 69.2 59.0 46.5
France 45.8 35.3 24.9 94.7 83.1 67.4 41.5 32.0 22.5 65.6 56.0 44.1
Germany 48.2 37.2 26.2 99.6 87.4 70.9 44.3 34.2 24.1 70.0 59.8 47.0
Greece 38.3 29.5 20.8 79.1 69.4 56.3 35.4 27.3 19.2 55.9 47.8 37.6
Hungary 41.7 32.2 22.7 86.2 75.6 61.3 37.2 28.7 20.2 58.8 50.2 39.5
Ireland 94.5 72.9 51.3 195.2 171.3 138.9 90.4 69.8 49.1 142.9 122.0 96.0
Italy 45.9 35.4 24.9 94.8 83.2 67.5 41.8 32.2 22.7 66.0 56.3 44.3
Latvia 42.1 32.5 22.9 87.0 76.4 62.0 37.2 28.7 20.2 58.8 50.2 39.5

Lithuania 45.8 35.3 24.9 94.6 83.0 67.3 40.4 31.2 22.0 63.9 54.5 42.9
Luxembourg 105.0 81.0 57.0 217.0 190.4 154.4 95.4 73.6 51.8 150.8 128.7 101.3
Malta 43.6 33.6 23.7 90.0 79.0 64.1 39.2 30.3 21.3 62.0 52.9 41.7
Netherlands 57.2 44.1 31.0 118.1 103.7 84.1 51.1 39.4 27.8 80.7 68.9 54.2
Poland 42.7 33.0 23.2 88.2 77.5 62.8 39.4 30.4 21.4 62.3 53.2 41.9
Portugal 41.9 32.3 22.8 86.6 76.0 61.6 38.0 29.3 20.7 60.1 51.3 40.4
Romania 38.3 29.5 20.8 79.0 69.4 56.3 41.4 31.9 22.5 65.4 55.8 43.9
Slovakia 52.3 40.4 28.4 108.1 94.9 77.0 46.9 36.2 25.5 74.0 63.2 49.8
Slovenia 40.5 31.2 22.0 83.6 73.4 59.5 36.3 28.0 19.7 57.4 49.0 38.6
Spain 46.8 36.1 25.4 96.7 84.9 68.9 43.3 33.4 23.5 68.3 58.3 45.9
Sweden 52.1 40.2 28.3 107.8 94.6 76.7 47.9 37.0 26.0 75.7 64.6 50.8
United Kingdom 46.9 36.2 25.5 96.8 85.0 68.9 43.1 33.2 23.4 68.0 58.1 45.7
Norway 63.1 48.7 34.3 130.4 114.4 92.8 53.3 41.1 28.9 84.2 71.9 56.6
Switzerland 64.5 49.7 35.0 133.2 116.9 94.8 57.8 44.6 31.4 91.4 78.0 61.4

F.8 Congestion and scarcity costs for other transport modes
This annex gathers the main findings from the literature review conducted on congestion
costs of the other transport modes.
For rail Rotoli et al. (2016) reviewed different analytical and optimisation procedures to
evaluate rail networks capacity. The level of detail, data availability and complexity has
been linked with the evaluation of utilised capacity, and in turn, with the probability of
expected ‘reactionary delay’. The methodologies considered were the UIC’s Analytical
Method (Code 405R) and Compression Method (Code 406), the Capacity Utilisation Index
(CUI) Method and the STRELE Formula (i.e., Method of Schwanhäußer). The analysis
suggests that the methodologies considered are a useful way to estimate the congestion and
the trade-offs between capacity of a railway network, delays and related costs.
Vromans et al. (2006) and Haith et al. (2014) suggested an alternative methodology to CUI,
based on the theory that the level of ‘reactionary delay’ can be determined by the
minimum gaps that exist between trains.
Brunel et al. (2013) analysed the relationship between traffic density (i.e., number of trains
per hour), reliability rate and average delay in order to assess rail congestion. The analysis
focussed on 42 lines of the French rail network estimating to what extent the delay of a
train (i.e., direct delay) impacts on another train (i.e., indirect delay). Estimations were
diversified for 9 different categories to account for the type of train (i.e., high speed,
intercity and regional) and traffic density (i.e., low, medium and high). However, the
authors did not develop estimations of the marginal costs in monetary terms.
Jansson and Lang (2013) developed a methodology to evaluate the external delay costs in
rail transport. In the application for passengers transport in Sweden, the authors estimated,
how the marginal cost-based charges (initially limited to external costs for wear and tear,
maintenance, emissions, etc.) would change if delays due to additional departures were
also taken into account. For example, if an additional departure of a commuter train leads
to a delay of two minutes in the network shared with high speed trains, the authors
estimated the marginal external cost effect of this delay to correspond to a 25% increase in
the commuter train fare for this additional journey, and a 5% increase in the fares for high
speed trains.
Maritime shipping: By considering cargo handling and port logistics (stevedoring) costs and
wait time records at several international ports of the 1970s, the UNITE project (Doll,
2002)concludes that there are no external congestion costs in seaport operations.
The analysis of EU and US ports in the COMPETE project (Schade, et al., 2006)
however,clearly show that capacity in particular in North American ports is approaching its
limits and that congestion at cargo handling and storage facilities is a priority issue. The
GRACE D4 report (Meersman, et al., 2006) estimates the additional (marginal) crew costs of
a vessel having to wait to call at a port at € 185 per hour. However, as ports usually do not
keep records of vessel waiting times the calculation of price relevant marginal external
congestion costs in maritime transport is not easy to carry out.
Inland navigation: COMPETE results suggest that European countries do not face any
capacity problems in their inland waterway networks. However, the GRACE case studies
found a number of local bottlenecks at locks, although they largely depend on local
conditions. Delay times range between zero and 160 minutes, in the latter case passage

costs per ship are found to increase by € 50 if demand increases by 1%. Besides lock
capacity, the availability of sufficiently deep water levels to operate all vessel types is a
problem, particularly in summer time. Based on the Low Water Surcharge, which has to be
paid on the river Rhine when water levels fall below a certain value.
Table 119 - Marginal external costs of congestion of rail freight transport €-cent/1,000 tkm (2016 prices)
Country Value
Austria 35.74
Belgium 35.89
Bulgaria 65.25
Cyprus n.a.
Czech Republic 43.56
Denmark 32.01
Estonia 63.54
Finland 36.30
France 33.64
Germany 32.96
Greece 42.49
Hungary 53.37
Ireland 41.30
Italy 31.60
Latvia 62.60
Lithuania 51.76
Luxembourg 46.23
Malta n.a.
Netherlands 32.87
Poland 47.47
Portugal 34.30
Romania 120.82
Slovakia 58.62
Slovenia 49.44
Spain 40.40
Sweden 37.31
United Kingdom 37.01
EU average 43.20
Source: (Christidis & Brons, 2016).

G Detailed assessment costs of
habitat damage
G.1 Introduction
Transport has different effects on nature and landscape or natural habitats. The main
effects reported in literature are habitat loss (ecosystem loss), habitat fragmentation and
negative effects on ecosystems due to the emissions of air pollutants (e.g. biodiversity
loss). The following chapter summarizes the most important impacts of transport on natural
habitats (i.e. nature and landscape). Section G.2 gives an overview on the latest literature,
Section G.3 draws conclusions on how to proceed in the present Handbook.
G.2 Detailed discussion on impacts on natural habitats (nature and landscape)

The different negative effects of transport on nature and landscape can be described as the
following:
— Habitat loss: Transport infrastructure requires land and/or natural surfaces. Therefore,
transport infrastructure also leads to a loss of natural ecosystems, which are natural
habitats of plants and animals. The land use of transport therefore leads to a loss of
habitats (ecosystems), which has a negative effect on biodiversity. Habitat loss is
occurring during the building phase of a transport infrastructure, but will last over the
whole lifetime of a transport infrastructure.
— Habitat fragmentation: Transport infrastructure can also have additional fragmentation

and separation effects for animals. These fragmentation effects can negatively affect the
natural habitats of certain species and lead to adverse effects for species and
consequently on biodiversity. Habitat fragmentation due to transport infrastructure is
aconsequence of the infrastructure itself plus the transport demand on the
infrastructure. The main negative effects are caused by large and broad main
infrastructures such as motorways and high-speed rail lines. Large wildlife mammals such
as deer, rabbit, badger, etc. as well as smaller animals such as amphibians are negatively
affected of habitat fragmentation.
— Habitat degradation due to emissions: Habitat degradation can also occur via the
emission of air pollutants of other toxic substances (e.g. heavy metals, PAH) or road salt.
These effects again lead to biodiversity loss and therefore external costs. The biodiversity
loss due to air pollution is already covered in the air pollution Chapter (4), where all
adverse impacts of air pollution are included. The negative effects of the emission of
toxic substances are covered in a separate chapter (‘other external costs’).
— Visual intrusion (landscape scenery): Transport infrastructure often also have a negative
impact on the landscape and its scenery. This is not a negative effect for the
environment, but is anthropogenic and negatively affecting inhabitants or visitors of a

region (landscape). This effect is generally not quantified since it is very much dependent
on the specific situation.
— Invasive plants: Invasive plants are appearing very often along transport infrastructure
and are sometimes even spread thanks to transport infrastructure.
— Light emmissions: The emission (and immission) of light of transport vehicles during the
night can negatively impact natural ecosystems (e.g. natural fauna).
In external costs literature, the first effect — habitat loss — has been discussed in
several studies. Also the habitat fragmentation has been analysed in certain studies.
Habitat degradation has been an issue in several studies when focussing on the emission of
air pollutants and toxic substances. Other effects have not been quantitatively analysed and
reported in the literature.
The corresponding Chapter (9) is therefore focussing on the habitat damage, i.e. the first of
the three effects described above.
G.3 Assessment of costs of impacts on natural habitats (nature and landscape)
G.3.1 Recommended approach previous Handbook

The cost of impacts on nature and landscape has not been covered in the last Handbook.
G.3.2 New evidence

There are only few studies covering the external costs of habitat damage due to transport
activities. A brief overview on the relevant literature is given in the following table.

Table 120 – Literature on external costs of habitat damage due to transport activities
Literature Title of Study and main focus Effects covered Results, type of cost factors
INFRAS, IWW 2004 (UIC) External costs of transport in Habitat loss: unsealing costs, Total and average costs:
Europe restoration costs € and €/pkm, €/tkm
(Nateco; Econcept, 2004) External costs of transport Habitat loss & fragmentation: Total and average costs:
due to impacts on nature and restoration cost approached; CHF and CHF/pkm, CHF/tkm
landscape (first bottom-up bottom-up calculation based
calculation) on aerial photo analysis
(NEEDS, 2006a) Assessment of Biodiversity Habitat loss: restoration costs Cost factors per m2 and
Losses; ecosystem type
Report of the EU-research
project NEEDS
DG MOVE 2008 1st Handbook on estimation Costs of nature and landscape: Only qualitative information
of external costs of transport only short overview on the and some selective cost
(Infras, CE Delft, ISI & relevant effects and data factors from NEEDS and
University of Gdansk, INFRAS, IWW 2004
2008)
(CE Delft, INFRAS & External costs of transport in Habitat loss: unsealing costs, Total and average costs:
Fraunhofer ISI, 2011) Europe for 2008 restoration costs € and €/pkm, €/tkm
(UBA, 2018) Methodenkonvention 3.0 on Habitat loss & fragmentation, Cost factors per vehicle type:
the estimation of based on (INFRAS en Ecoplan, €/vkm and €/pkm, €/tkm
environmental costs 2018)
(INFRAS en Ecoplan, External costs of transport in Habitat loss & fragmentation Total costs and average
2018) Switzerland 2015 (update costs: CHF/pkm, CHF/tkm
study, bottom-up calculation)
G.3.3 Conclusions
The most detailed bottom-up calculations of the cost of habitat damage have been made by
the European research project NEEDS (2006) and in the latest Swiss study by INFRAS,
Ecoplan (2018). Additionally, the most recent UBA study (UBA, 2018) also covers up-to-date
cost factors for the negative effects on habitats, mainly also based on INFRAS, Ecoplan
(2018).
In the main Chapter 9, the calculation of cost factors for nature and landscape focusses on
habitat damage (habitat loss and fragmentation) and is mainly based on the most recent
bottom-up study on external costs of transport in Switzerland (INFRAS en Ecoplan, 2018).

H Total and average costs
motorways
H.1 Introduction
This section presents the total and average external costs for motorways only.
H.2 Total external costs of transport on motorways in the EU28

The total external costs of transport on motorways are shown in Table 121.
H.3 Average external costs of transport on motorways in the EU28

The average external costs of transport on motorways are shown in Table 122.

Table 121 – Total external costs of transport on motorways in the EU28 (billion €)
Vehicle Accidents Air Climate Noise Congestion WTT Habitat Total Total
category pollution change Delay costs Deadweight emissions damage (congestion (congestion
loss costs based on based on
delay costs deadweight
genereated) loss costs
genereated)
Passenger car 6.08 12.99 7.7 4.24 9 56.71 54.21
Passenger car –
1.55 7.08 4 2.31 5.1
petrol 13.5 3.2 0.7
Passenger car –
4.58 5.87 3.7 1.92 3.9
diesel
Motorcycle 1 0.31 0.28 1.1 n/a n/a 0.16 0.13 2.98
Bus 0.02 0.03 0 0.01 0.01 1.75 1.67
0.2 0.1 0.02
Coach 0.31 0.43 0.3 0.14 0.2
LCV 3.01
1.9 3.15 1.5 0.9 0.2 0.9 1.3 12.66 11.96
HGV 1.93
5.5 2.63 2.3 0.4 0.1 1.04 1.56 15.36 15.06

Table 122 – Average external costs of transport on motorways in the EU28
Vehicle category Accidents Air pollution Climate Noise Congestion WTT Habitat Total Total
change Delay costs Deadweight emissions damage (congestion (congestion
loss costs based on based on
delay costs) deadweight
loss costs)
Passenger car 0.55 1.17 0.69 0.38 0.81 5.10 4.88
Passenger car –
0.26 1.2 0.68 0.39 0.20
petrol 1.22 0.29 0.06
Passenger car –
0.88 1.13 0.71 0.37 0.20
diesel
Motorcycle 3.57 1.12 1.01 3.88 n/a n/a 0.57 0.48 10.86
Bus 0.28 0.38 0.17 0.14 0.17 1.18 1.12
0.14 0.07 0.01
Coach 0.22 0.31 0.22 0.1 0.14 1.19 1.14
LCV 2.89 3.03
1.81 1.46 0.94 0.21 0.87 1.25 12.51 11.78

HGV 0.33 0.45
0.96 0.3-0.9 0.07 0.02 0.18 0.27 2.55-3.15 2.55-3.10

I EU27 values
I.1 Introduction
This section presents the total and average external costs for road, rail and inland
waterways at the EU27 level (excluding the UK).
I.2 Total external costs of transport in the EU27

The total external costs of transport in the EU27 are shown in Table 123 and Table 124.

Table 123 – Total external costs of transport in the EU27 (billion €)
Vehicle Accidents Air pollution Climate Noise Congestion WTT Habitat Total Total
category change Delay Deadweight loss emissions damage (congestion (congestion
costs costs based on delay based on
costs) deadweight loss
costs)
Road
Passenger car 30.9
48.3 23.7 15.9 24.3 491.9 356.4
Passenger car
185.4 7.9 27.4 12.3 163.4 27.9 9.0 13.1
– petrol
Passenger car
23.1 20.8 11.3 6.9 11.2
– diesel
Motorcycle 19.1 1.8 1.4 14.4 n/a n/a 0.8 0.5 38.0
Bus 1.3 0.8 0.7 0.3 0.2
4.4 4.0 0.7
Coach 2.5 1.4 0.8 0.5 0.4 17.2 13.9
LCV 17.3 14.1 11.6 4.9 47.5 8.1 3.3 4.1 102.8 63.4
HGV 21.0 13.2 8.6 8.5 13.0 2.2 3.4 3.4 71.1 60.3
Rail
High speed
0.062 0.002 0 0.341 n/a n/a 0.307 0.662
passenger train 1.4
Conventional
electric 0.027 0 2.431 n/a n/a 2.402 1.338
passenger train
1.932 10.0
Conventional
diesel 0.444 0.173 0.700 n/a n/a 0.060 0.448
passenger train
Electric freight
0.012 0 2.063 n/a n/a 0.497 0.773
train
0.270 5.1
Diesel freight
0.622 0.208 0.335 n/a n/a 0.115 0.209
train
IWT
IWT Vessel 0.089 1.927 0.395 n/a n/a n/a 0.197 0.285 2.9

Table 124 – Total external costs of transport for selected (air)ports in the EU27 (billion €)
Accidents Air pollution Climate change Noise Congestion WTT emissions Habitat damage Total
Aviation
Selected EU27 airports 0.92
0.07 18.53 0.47 n/a 7.43 0.05 27.47
Maritime
Selected EU27 ports 0.07 n/a n/a n/a n/a n/a n/a n/a

J Marginal cost figures
J.1 Introduction
This section presents an overview of the marginal cost figures for selected cases for all
modes in Section J.2.
Sections J.3, J.4 and J.5 present additional marginal cost figures for the categories air
pollution, climate change and well-to-tank for the reference vehicles that have been
defined for road. These reference vehicles have been used for the overview in Sections 4.4,
5.4 and 8.4. They have been based on the main cost drivers for climate and WTT cost which
are the combination of fuel type and fuel efficiency. These have been combined with the
main cost drivers for air pollution, which are the Euro standard and road type (for cars and
LCVs).
J.2 Overview marginal external costs

This section provides a synthesis of the marginal external costs for the selected reference
cases. With reference to road congestion, the values of social marginal congestion cost
generated by vehicle type have been used (see Section 7.4.4).
J.2.1 Road
Table 125 – Synthesis of marginal external costs 2016 for EU28 – passenger cars (in €-cent/pkm)
Reference vehicle Accident Air Climate Noise Congestion WTT Habitat

pollution Change
Marginal costs for dense metropolitan traffic during the day
2016 Euro 6 fuel efficient petrol car: 141 0.26

0.14 1.00
g/km
2016 Euro 6 fuel inefficient petrol car: 256 0.48

0.14 1.81
g/km
0.0
2000 Euro 3 fuel efficient petrol car: 176 0.43
0.19 1.26
g/km
2000 Euro 3 fuel inefficient petrol car: 243 0.63

0.19 1.74
g/km
1.41 0.5 41.2
2016 Euro 6 fuel efficient diesel car: 126 0.14
0.86 0.88
g/km
2016 Euro 6 fuel inefficient diesel car: 169 0.19

0.86 1.18
g/km
0.0
1.90 1.02
g/km

1.90 1.34
g/km

pollution Change
LPG Euro 6 0.19 0.83 0.18 0.0
CNG Euro 6 0.15 1.36 0.29 0.0
Full electric Euro 6 0.05 0.00 0.83 0.0
PHEV Euro 6 – petrol 0.06 0.22 0.08 0.0
Marginal costs for dense traffic on rural motorways during the day
2016 Euro 6 fuel efficient car: 141 g/km 0.08 0.86 0.23
2016 Euro 6 fuel inefficient car: 256 g/km 0.08 1.56 0.41
0.0
2000 Euro 3 fuel efficient car: 176 g/km 0.11 1.08 0.37
2000 Euro 3 fuel inefficient car: 243 g/km 0.11 1.49 0.54

0.46 0.79
g/km

0.46 1.06
g/km
0.25 0.04 18.2 0.0
0.95 0.93
g/km

0.95 1.21
g/km
LPG Euro 6 0.12 0.75 0.16 0.0
CNG Euro 6 0.08 1.24 0.27 0.0
Full electric Euro 6 0.05 0.00 0.83 0.0
PHEV Euro 6 – petrol 0.06 0.29 0.10 0.0
Table 126 – Synthesis of marginal external costs 2016 for EU28 – motorcycles and mopeds (in €-cent/pkm)

pollution Change
Fuel efficient Euro 3 motorcycle: 100 g/km 0.32 0.92 0.33
Fuel inefficient Euro 3 motorcycle: 128 0.32 1.27 0.46

g/km
4.42 7.4 n/a 0.0
Moped 1.20 0.44 0.16
Electric motorcycle 0.02 0.00 0.15
Fuel efficient Euro 3 motorcycle: 100 g/km 0.29 1.11 0.40

0.0
-0.65 0.06 n/a
Fuel inefficient Euro 3 motorcycle: 128 0.29 1.34 0.48
g/km

Moped 0.53 0.44 0.16
Electric motorcycle 0.02 0.00 0.15
Table 127 – Synthesis of marginal external costs 2016 for EU28 – buses (in €-cent/pkm)
Reference vehicle Accident Air pollution Climate Change Noise Congestion WTT Habitat
Fuel efficient Euro 6 bus:

0.12 0.76 0.18
954 g/km
Fuel inefficient Euro 6 bus: 1,155 g/km 0.12 0.93 0.20
Fuel efficient Euro 3 bus:

1.93 0.76 0.18
954 g/km 0.80 0.5 6.8 0.0
CNG Euro 6 0.53 0.06 0.01
Full electric bus 0.04 0.00 0.63
Fuel efficient Euro 6 bus: 954 g/km 0.02 0.46 0.11
Fuel efficient Euro 3 bus: 954 g/km 0.41 0.46 0.11

0.05 0.04 3.0 0.0
CNG Euro 6 0.15 0.06 0.01
Full electric bus 0.01 0.00 0.63
Table 128 – Synthesis of marginal external costs 2016 for EU28 – coaches (in €-cent/pkm)
Fuel efficient Euro VI coach: 583 g/km 0.21 0.56 0.11
Fuel inefficient Euro VI coach: 742 g/km 0.17 0.75 0.13

0.80 0.3 6.8 0.0
Fuel inefficient Euro VI coach: 742 g/km 0.02 0.34 0.06 0.0
0.05 0.02 3.0

Table 129 – Synthesis of marginal external costs 2016 for EU28 – LCVs (in €-cent/vkm)

pollution Change
2016 fuel efficient Euro 6 LCV: 137 0.48

0.16 1.65
g/km
2016 fuel inefficient Euro 6 LCV: 233 0.81

0.16 2.80
g/km
0.0
0.33 2.82
g/km

0.33 3.74
g/km
2016 fuel efficient Euro 6 LCV 135 0.76 1.7 99.4 0.29
2.19 1.44
g/km

2.19 1.85
g/km
0.0
4.66 2.00
g/km
2000 fuel inefficient Euro 3 LCV:245 0.62

4.66 2.61
g/km
Full electric 0.08 0.00 2.37 0.0

0.14 1.28
g/km

0.14 2.18
g/km
0.0
0.25 1.94
g/km

0.25 2.57
g/km
2016 fuel efficient Euro 6 LCV 135 0.37 0.01 44.0 0.28
1.94 1.39
g/km

1.94 1.78
g/km
0.0
2.26 2.18
g/km
2000 fuel inefficient Euro 3 LCV:245 0.67

2.26 2.84
g/km
Full electric 0.10 0.00 2.37 0.0

Table 130 – Synthesis of marginal external costs 2016 for EU28 – HGVs (in €-cent/tkm)

pollution Change
Fuel efficient HGV: 370 g/km Euro VI - diesel 0.34 1.00 0.22
(3.5-7.5 t)
Low fuel efficiency Euro VI - diesel (3.5-7.5 t) 0.34 1.39 0.32

1.5
Fuel efficient HGV: 370 g/km Euro III - diesel 2.59 1.00 0.22
(3.5-7.5 t)
Low fuel efficiency Euro III - diesel (3.5-7.5 t) 2.59 1.39 0.32
(7.5-16 t)
Fuel inefficient HGVs: 716 g/km Euro VI - 0.14 0.83 0.19

diesel (7.5-16 t)
0.7
(7.5-16 t)
Fuel inefficient HGVs: 716 g/km Euro III - 1.61 0.83 0.19
diesel (7.5-16 t)

diesel (16-32 t) 0.10 9.7 0.0

diesel (16-32 t)
0.6
diesel (16-32 t)
diesel (16-32 t)
(> 32 t)
Fuel inefficient HGVs: 1,033 g/km Euro VI - 0.11 0.83 0.19

diesel (>32t)
Fuel efficient HGV: 848 g/km Euro III - diesel 1.69 0.66 0.6 0.15
(> 32 t)
Fuel inefficient HGVs: 1,033 g/km Euro III - 1.69 0.83 0.19
diesel (> 32 t)
LNG Euro 6 (> 32 t) 0.09 0.72 0.04
(3.5-7.5 t)
Low fuel efficiency Euro VI - diesel (3.5-7.5 t) 0.07 0.06 1.14 0.01 4.3 0.27 0.0
(3.5-7.5 t)

pollution Change
Low fuel efficiency Euro III - diesel (3.5-7.5 t) 1.01 1.14 0.27
(7.5-16 t)

diesel (7.5-16 t)
0.01
(7.5-16 t)
diesel (7.5-16 t)

diesel (16-32 t)

diesel (16-32 t)
0.005
diesel (16-32 t)
diesel (16-32 t)
(> 32 t)
Fuel inefficient HGVs: 1,033 g/km Euro VI - 0.03 0.49 0.11

diesel (> 32 t)
Fuel efficient HGV: 848 g/km Euro III - diesel 0.44 0.41 0.005 0.10
(> 32 t)
Fuel inefficient HGVs: 1,033 g/km Euro III - 0.44 0.49 0.11
diesel (> 32 t)
LNG Euro 6 (> 32 t) 0.02 0.42 0.03
J.2.2 Rail
Table 131 – Synthesis of marginal external costs 2016 for EU28 – passenger trains (in €-cent/pkm)
Marginal costs on dense metropolitan railways during the day
High speed trains 0.1 0.002 0 0.13 0.39 0.0
Electric intercity 0.01 0 0.45 0.73 0.0
Diesel intercity with EGR/SCR 0.47 0.201 0.45 n/a 0.18

0.3 0.0
Diesel intercity without EGR/SCR 0.70 0.201 0.45 0.18
Electric regional train 0.02 0 0.45 0.89 0.0

Diesel regional train with EGR/SCR 1.52 0.735 0.45 0.26
0.0
Diesel regional train without EGR/SCR 2.10 0.735 0.45 0.26
Marginal costs on dense rural railways during the day
High speed trains 0.1 0.002 0 0.01 0.39 0.0
Electric intercity 0.01 0 0.03 0.73 0.0
Diesel intercity with EGR/SCR 0.23 0.201 0.03 0.18

0.0
Diesel intercity without EGR/SCR 0.40 0.201 0.03 n/a 0.18
0.3
Electric regional train 0.02 0 0.03 0.89 0.0
Diesel regional train with EGR/SCR 0.71 0.735 0.03 0.26

0.0
Diesel regional train without EGR/SCR 1.20 0.735 0.03 0.26
Table 132 – Synthesis of marginal external costs 2016 for EU28 – freight trains (in €-cent/tkm)
Marginal costs on dense metropolitan railways during the day
Electric long container 0.004 0 0.13 0.11
Electric long bulk 0.004 0 0.13 0.10
Electric short container 0.004 0 0.13 0.26
Electric short bulk 0.004 0 0.13 0.18
Diesel long container with EGR/SCR 0.13 0.158 0.13 0.03
Diesel long container without EGR/SCR 0.28 0.158 0.13 0.03

0.1 n/a 0.0
Diesel long bulk with EGR/SCR 0.11 0.087 0.13 0.03
Diesel long bulk without EGR/SCR 0.25 0.087 0.13 0.03
Diesel short container with EGR/SCR 0.36 0.074 0.13 0.07
Diesel short container without EGR/SCR 0.78 0.074 0.13 0.07
Diesel short bulk with EGR/SCR 0.24 0.066 0.13 0.05
Diesel short bulk without EGR/SCR 0.52 0.066 0.13 0.05
Marginal costs on dense rural railways during the day
Electric long container 0.002 0 0.001 0.11
Electric long bulk 0.002 0 0.001 0.10
Electric short container 0.002 0 0.001 0.26
Electric short bulk 0.1 0.002 0 0.001 n/a 0.18 0.0
Diesel long container with EGR/SCR 0.07 0.071 0.001 0.03
Diesel long container without EGR/SCR 0.14 0.071 0.001 0.03
Diesel long bulk with EGR/SCR 0.06 0.067 0.001 0.03

Diesel long bulk without EGR/SCR 0.12 0.067 0.001 0.03
Diesel short container with EGR/SCR 0.18 0.087 0.001 0.07
Diesel short container without EGR/SCR 0.38 0.087 0.001 0.07
Diesel short bulk with EGR/SCR 0.12 0.084 0.001 0.05
Diesel short bulk without EGR/SCR 0.25 0.084 0.001 0.05

J.2.3 IWT
Table 133 – Synthesis of marginal external costs 2016 for EU28 – IWT (in €-cent/tkm)
CEMT II bulk – CCR-0 3.36 0.34 0.15
CEMT II bulk – CCR-1 2.82 0.34 0.15
CEMT II bulk – CCR-2 1.82 0.34 0.15
CEMT II bulk – Average 3.25 0.34 0.15
CEMT II container – CCR-0 2.14 0.21 0.09
CEMT II container – Average 2.07 0.21 0.09
CEMT IV bulk – CCR-0 2.00 0.20 0.09
CEMT IV bulk – CCR-1 1.67 0.20 0.09
CEMT IV bulk – CCR-2 1.08 0.20 0.09
CEMT IV bulk – Average 1.84 0.20 0.09

0.1 n/a n/a 0.0
CEMT Va bulk – CCR-0 1.82 0.18 0.08
CEMT Va bulk – CCR-1 1.53 0.18 0.08
CEMT Va bulk – CCR-2 0.99 0.18 0.07
CEMT Va bulk – Average 1.53 0.18 0.07
CEMT Va container – CCR-0 2.06 0.21 0.09
CEMT Va container – Average 1.74 0.21 0.09
Pushed convoy bulk – CCR-0 1.48 0.15 0.06
Pushed convoy bulk – Average 0.89 0.15 0.06
J.2.4 Maritime
Table 134 – Synthesis of marginal external costs 2016 for EU28 – maritime transport (in €-cent/pkm (ferry) and €-cent/tkm
(vessels))
Ferry (RoPax) – Tier 0 24.62 5.53 2.09

n/a n/a n/a n/a
Ferry (RoPax) – Tier 1 22.23 5.53 2.09

Ferry (RoPax) – Tier 2 20.77 5.53 2.09
Small container vessel–Tier 0 0.51 0.19 0.07
Large container vessel–Tier 0 0.25 0.08 0.03
Small bulk vessel – Tier 0 0.33 0.11 0.04
Large bulk vessel – Tier 0 0.12 0.04 0.01
J.2.5 Aviation
Table 135 – Synthesis of marginal external costs 2016 for EU28 – aviation (in €-cent/pkm)
Bombardier CRJ900 0.14 0.28 2.84 1.3 0.0
Embraer 170 (ERJ-170-100) 0.30 3.44 1.4
Airbus A320-232 0.03 0.07 1.53 0.6 0.0

0.05 n/a
Boeing 737-700 0.11 2.18 0.8
Airbus A340-300 0.03 0.03 1.36 0.8 0.0
Boeing 777-300 ER 0.04 1.54 0.9
J.3 Marginal air pollution costs for reference vehicles

Table 136 – Marginal air pollution costs (averages for metropolitan, urban and rural regions)
Vehicle Fuel Fuel efficiency (average real- COPERT category Motor- Urban Other Average
category type world CO 2 emissions) ways roads roads all roads
2016 fuel efficient car:

Petrol medium Euro 6 0.08 0.14 0.09 0.10
141 g/km
Passenger
Petrol
car 2016 fuel inefficient car:
256 g/km


176 g/km
2000 fuel inefficient car:

243 g/km

Diesel medium Euro 6 0.47 0.86 0.49 0.61
126 g/km

169 g/km
Diesel
147 g/km

192 g/km
LPG 169 g/km Petrol medium Euro 6 0.08 0.15 0.08 0.10
CNG 136 g/km Petrol medium Euro 6 0.08 0.14 0.08 0.10
Full
0 g/km Euro 6 0.06 0.05 0.05 0.05
electric
PHEV -
133 g/km Petrol medium Euro 6 0.08 0.06 0.06 0.07
petrol
Fuel efficient motorcycle:

Euro 3 0.29 0.32 0.26 0.29
100 g/km
Motorcycle Petrol
Fuel inefficient motorcycle: 128
Euro 3 0.29 0.32 0.26 0.29
g/km
Moped Petrol 70 g/km Moped 4 stroke 0.53 1.20 0.86 0.86
Motorcycle Electric 0 g/km 0.02 0.02 0.02 0.02
Fuel efficient bus: 954 g/km 0.02 0.12 0.04 0.06

Diesel Fuel inefficient bus:
0.41 1.93 0.73 1.02
1,155 g/km
Bus (18 t)
CNG 1,007 g/km Euro 6 0.15 0.53 0.23 0.31
Electric 0 g/km 0.01 0.04 0.02 0.02
Fuel efficient coach:

0.02 0.21 0.05 0.10
583 g/km
Coach Diesel
Fuel inefficient coach:
0.50 2.67 0.88 1.35
742 g/km
2016 fuel efficient LCV:

Euro 6 petrol 0.14 0.16 0.14 0.15
137 g/km
LCV Petrol
2016 fuel inefficient LCV:
Euro 6 petrol 0.14 0.16 0.14 0.15
233 g/km


Euro 3 petrol 0.25 0.33 0.24 0.27
216 g/km

Euro 3 petrol 0.25 0.33 0.24 0.27
286 g/km

Euro 6 diesel 1.94 2.19 1.68 1.94
135 g/km

Euro 6 diesel 1.94 2.19 1.68 1.94
173 g/km
Diesel
Euro 3 diesel 2.26 4.66 2.57 3.16
188 g/km

Euro 3 diesel 2.26 4.66 2.57 3.16
245 g/km
Electric full electric 0 g/km 0.10 0.08 0.07 0.08
Fuel efficient HGV: 370 g/km 0.26 1.55 0.42 0.74

HGV
Diesel
3.5 t-7.5 t Fuel inefficient HGV:
4.54 11.65 6.32 7.50
450 g/km
Fuel efficient HGV:

0.08 0.38 0.12 0.19
596 g/km
HGV
Diesel
7.5 t-16 t Fuel inefficient HGVs:
1.38 4.46 2.17 2.67
716 g/km

HGV
Diesel
16 t-32 t Fuel inefficient HGVs:
0.67 2.45 1.13 1.42
875 g/km

HGV
Diesel
> 32 t Fuel inefficient HGVs:
0.54 2.07 0.92 1.18
(truck 1,033 g/km
trailer)
LNG 900 g/km 0.02 0.09 0.03 0.04

J.4 Marginal climate change costs for reference vehicles
Table 137 – Marginal climate change costs
Vehicle category Fuel type Fuel efficiency (average type Motor- Urban Other Average all
approval CO 2 emissions) ways roads roads roads
2016 fuel efficient car: 99 g/km 0.87 1.01 0.80 0.90

1.58 1.84 1.45 1.63
180 g/km
Petrol

1.51 1.77 1.35 1.54
233 g/km

1.08 1.20 0.94 1.07
119 g/km
Passenger car Diesel

1.23 1.36 1.07 1.22
176 g/km
LPG 119 g/km 0.77 0.84 0.66 0.76
CNG 196 g/km 1.26 1.39 1.08 1.25
Full
0 g/km 0.00 0.00 0.00 0.00
electric
PHEV -
39 g/km 0.29 0.22 0.23 0.25
petrol

1.11 0.92 0.82 0.95
100 g/km
Motorcycle Petrol
Fuel inefficient motorcycle:
1.34 1.27 1.05 1.22
128 g/km
Moped Petrol 46 g/km 0.44 0.44 0.44 0.44
Motorcycle Electric 0 g/km 0.00 0.00 0.00 0.00

Diesel
Fuel inefficient bus: 1,155 g/km 0.54 0.93 0.58 0.68
Bus (18 t)
CNG 1,007 g/km 0.06 0.06 0.06 0.06
Electric 0 g/km 0.00 0.00 0.00 0.00
Fuel efficient coach: 583 g/km 0.23 0.45 0.25 0.31

Coach Diesel Fuel inefficient coach:
0.27 0.60 0.31 0.39
742 g/km

Vehicle category Fuel type Fuel efficiency (average type Motor- Urban Other Average all
approval CO 2 emissions) ways roads roads roads

Petrol 1.28 1.65 1.18 1.37
105 g/km

2.18 2.80 2.01 2.33
179 g/km

1.94 2.82 1.71 2.16
198 g/km

2.57 3.74 2.27 2.86
262 g/km
LCV 2016 fuel efficient LCV:

1.39 1.44 1.22 1.35
104 g/km

1.78 1.85 1.56 1.73
133 g/km
Diesel
2.18 2.00 1.46 1.88
172 g/km

2.84 2.61 1.90 2.45
225 g/km
Freigh transport (€-cent per tkm)
Diesel Fuel efficient HGV: 370 g/km 4.49 4.49 4.49 4.49
HGV 3.5 t-7.5 t
Fuel inefficient HGV: 450 g/km 5.15 6.25 4.97 5.46

HGV 7.5 t-16 t Diesel
Fuel inefficient HGVs: 716 g/km 1.61 2.30 1.62 1.85

HGV 16 t-32 t Diesel

Diesel
HGV > 32 t Fuel inefficient HGVs:
0.49 0.83 0.53 0.62
(truck trailer) 1,033 g/km
LNG 900 g/km 0.42 0.72 0.47 0.54
J.5 Marginal well-to-tank costs for reference vehicles

Table 138 – Marginal well-to-tank costs
Vehicle category Fuel type Fuel efficiency (average real-world CO 2 Motor- Urban Other Average all
emissions in g/km) ways roads roads roads
Passenger car Petrol 2016 fuel efficient car: 99 g/km 0.33 0.38 0.30 0.34


0.60 0.69 0.55 0.61
180 g/km

0.57 0.67 0.51 0.58
233 g/km

0.25 0.28 0.22 0.25
119 g/km
Diesel

0.29 0.32 0.25 0.28
176 g/km
LPG 119 g/km 0.16 0.18 0.14 0.16
CNG 196 g/km 0.27 0.30 0.23 0.27
Full
0 g/km 0.84 0.84 0.84 0.84
electric
PHEV -
39 g/km 0.10 0.08 0.08 0.09
petrol

0.42 0.35 0.31 0.36
100 g/km
Motorcycle Petrol
Fuel inefficient motorcycle:
0.51 0.48 0.40 0.46
128 g/km
Moped Petrol 46 g/km 0.17 0.17 0.17 0.17
Motorcycle Electric 0 g/vkm 0.10 0.10 0.10 0.10

Diesel
Fuel inefficient bus: 1,155 g/km 0.13 0.20 0.13 0.15
Bus (18 t)
CNG 1,007 g/km 0.01 0.01 0.01 0.01
Electric 0 g/km 0.63 0.63 0.63 0.63
Fuel efficient coach: 583 g/km 0.07 0.13 0.08 0.09

Coach Diesel
Fuel inefficient coach: 742 g/km 0.08 0.16 0.09 0.11

0.48 0.62 0.45 0.52
105 g/km

LCV Petrol 0.82 1.06 0.76 0.88
179 g/km

0.73 1.06 0.65 0.81
198 g/km


0.97 1.41 0.86 1.08
262 g/km

0.36 0.37 0.31 0.35
104 g/km

0.46 0.48 0.40 0.45
133 g/km
Diesel
0.56 0.52 0.38 0.49
172 g/km

0.73 0.67 0.49 0.63
225 g/km
Freigh transport (€-cent per tkm)

HGV 3.5 t-7.5 t Diesel
Fuel inefficient HGV: 450 g/km 1.13 1.18 0.97 1.09

HGV 7.5 t-16 t Diesel

HGV 16 t-32 t Diesel

Diesel
HGV > 32 t Fuel inefficient HGVs:
0.12 0.21 0.13 0.15
(truck trailer) 1,033 g/km
LNG 900 g/km 0.06 0.11 0.07 0.08

K Overview content Excel file
Two Excel annexes are attached to the report.
The file “Complete overview of country data” contains:

— Total and average costs figures per country:
• for road, rail and IWT;
• for all external cost categories;
• total costs figures are expressed in billion €;
• average figures in €/pkm (passenger transport), €/vkm (all modes), €/tkm (freight
transport, except for LCVs).
— Total and average cost figures per (air)port:
• for aviation and maritime transport;
• for all relevant external cost categories;
• total costs figures are expressed in million €;
• additionally, average figures in €/LTO & €/passenger & €/tonne for aviation and
€/port call & €/million tonnes for maritime transport are presented;
— Marginal costs for:
• Accidents, per country, in vkm/pkm/tkm
• Congestion, per country, in vkm
— Input values per country:
• for all external cost categories;
• the same units and differentiations as for the EU28 figures are applied.
— All tables and figures from the synthesis Chapter (Chapter 11).
The file “Marginal costs air pollution, climate, WTT, noise” contains marginal costs for:
— Noise, for road and rail in the EU28;
— Reference cases for climate change, air pollution and WTT for all modes in the EU28;
— Selected cases for climate change, air pollution and WTT for road transport in the EU28.
All annexes can be downloaded at the following link

https://transport.ec.europa.eu/transport-themes/sustainable-transport/internalisation-
transport-external-costs_en

GETTING IN TOUCH WITH THE EU
In person
All over the European Union there are hundreds of Europe Direct
information centres. You can find the address of the centre nearest you
at: https://europa.eu/european-union/contact_en
On the phone or by email
Europe Direct is a service that answers your questions about the

European Union. You can contact this service:
– by freephone: 00 800 6 7 8 9 10 11 (certain operators may charge

for these calls),
– at the following standard number: +32 22999696, or
– by email via: https://europa.eu/european-union/contact_en
FINDING INFORMATION ABOUT THE EU
Online
Information about the European Union in all the official languages of

the EU is available on the Europa website at:
https://europa.eu/european-union/index_en
EU publications
You can download or order free and priced EU publications from:

https://publications.europa.eu/en/publications. Multiple copies of free
publications may be obtained by contacting Europe Direct or your local
information centre (see https://europa.eu/european-union/contact_en
).
EU law and related documents
For access to legal information from the EU, including all EU law since
1952 in all the official language versions, go to EUR-Lex at: http://eur-
lex.europa.eu
Open data from the EU

The EU Open Data Portal ( http://data.europa.eu/euodp/en ) provides
access to datasets from the EU. Data can be downloaded and reused for
free, for both commercial and non-commercial purposes.

MI-03-20-275-EN-N
ISBN: 978-92-76-18184-2 doi: 10.2832/51388

Handbook On The External Costs of Transport

Uploaded by

Copyright:

Available Formats

Handbook On The External Costs of Transport

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Handbook On The External Costs of Transport

Uploaded by

Copyright:

Available Formats

Handbook on the external

Directorate A — Policy Coordination

Unit A3 — Economic analysis and better regulation

Contact: Rolf Diemer

1 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

3 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

Freephone number (*):

More information on the European Union is available on the Internet (http://www.europa.eu).

Luxembourg: Publications Office of the European Union, 2020

© European Union, 2020

Reproduction is authorised provided the source is acknowledged.

4 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

This report is prepared by:

Delft, CE Delft, January 2019

Publication code: 18.4K83.131

Client: European Commission, Directorate-General for Mobility and Transport

Publications of CE Delft are available from www.cedelft.eu

© copyright, CE Delft, Delft

5 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

6 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

2 General methodological framework 31

4 Air pollution costs 51

5 Climate change costs 74

7 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

7 Congestion costs 103

8 Costs of well-to-tank emissions 123

9 Costs of habitat damage 139

10 Other external costs 146

A Economic valuation of human health 186

B Detailed assessment accident costs 194

8 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

C Detailed assessment air pollution costs 207

D Detailed assessment climate change costs 221

E Detailed assessment of noise costs 230

F Detailed assessment congestion costs 241

G Detailed assessment costs of habitat damage 304

H Total and average costs motorways 307

I EU27 values 310

9 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

J Marginal cost figures 313

K Overview content Excel file 329

10 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

11 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

12 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

Table 2 – Airports and maritime ports covered ............................................................. 26

Table 3 – Level of externality of various costs categories ................................................ 33

Table 4 – Comparison of old and new EU traffic injury definitions.................................... 40

Table 5 – Correction factors to correct for underreporting of accidents ............................ 40

Table 11 – Degree of risk internalisation for different vehicle types ................................. 48

13 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

Table 24 – Climate change avoidance costs in €/tCO2 equivalent (€2016 )............................ 78

14 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

Table 53 – Average/marginal costs of well-to-tank emissions road transport for selected

Table 56 – Average/marginal costs of well-to-tank emissions maritime transport for selected

Table 58 – Cost factors for costs of habitat damage EU28............................................. 141

Table 60 – Cost factors for ecosystem (habitat) loss..................................................... 143

Table 63 – Mountain factors for external costs of freight transport................................ 148

15 4.K83 - Handbook on the external costs of transport – January 2019 – V1.1

Table 69 – Average external costs 2016 for EU-28 ....................................................... 160