Transportation Research Part A 38 (2004) 593606

www.elsevier.com/locate/tra

Costbenet analyses of walking and cycling track

networks taking into account insecurity, health eects and
external costs of motorized trac
Kjartan Slensminde

Institute of Transport Economics, P.O. Box 6110, Etterstad, Oslo N-0602, Norway
Received 3 July 2003; received in revised form 22 April 2004; accepted 28 April 2004

Abstract
The study presents costbenet analyses of walking and cycling track networks in three Norwegian cities. The costbenet analyses take into account the benet of reduced insecurity and the health benets of
the improved tness the use of non-motorized transport provides. In addition to reductions in health costs,
the analyses also take into account that a change from travel by car to cycling or walking means reduced
external costs (e.g. air pollution and noise) from motorized trac and reduced parking costs. The benets
of investments in cycle networks are estimated to be at least 45 times the costs. Such investments are thus
more benecial to society than other transport investments.
The results of such complete costbenet analyses make it possible to calculate the benets to society
that are not realized because motorized trac prevents people from bicycling or walking as much as they
otherwise would prefer. These barrier costs attributable to motorized trac are estimated to be of at least
the same magnitude as air pollution costs and more than double the noise costs. Barrier costs should therefore be taken into account in the same way as other external costs, when the issue is to determine the proper
level of car taxes or to evaluate dierent kinds of restrictions on car use.
2004 Elsevier Ltd. All rights reserved.
Keywords: Costbenet analyses; Walking; Cycling; Insecurity; Health eects; Barrier costs

Tel.: +47 22 57 38 00; fax: +47 22 57 02 90.

E-mail address: ks@toi.no

0965-8564/$ - see front matter 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tra.2004.04.003

594

K. Slensminde / Transportation Research Part A 38 (2004) 593606

1. Background and a brief literature review

Costbenet analyses (CBAs) are used to assess the impacts of road investment projects in Norway and many other countries. However, such analyses are not used to assess the impacts of measures designed to improve the safety and/or mobility of pedestrians and cyclists. The reason for this
might be that important impacts such as insecurity and health eects are very dicult to make an
adequate monetary valuation of and that the results of such CBAs therefore will be uncertain. To
illustrate the limited amount of literature on this subject it may be mentioned that Elvik (2000) was
the only relevant article found in a search for literature in the databases ISI, MEDLINE and PsycINFO for dierent framings of the words insecurity, health and (bi)cycle. Nevertheless,
other articles exist that present CBAs of measures for pedestrians and cyclists, although insecurity
and health eects are not included. Elvik (2000) presents hypothetical examples of ideally designed CBAs that address all such impacts. His examples show that most important impacts
can be quantied and assigned a monetary value, and that, for example, based on reasonable
assumptions, the benets of safe crossing facilities outweigh the costs. This result is supported
by Korve and Niemeier (2002) in their CBA of an additional bicycle phase at an existing signalized
intersection. Other recently published studies related to the subject in the current paper, and therefore worth mentioning, are de Ortuzar et al. (2000), Noland and Kunreuther (1995) and Hopkinson and Wardman (1996). de Ortuzar et al. (2000) estimate the demand for a cycle-way network
and provide an overview of literature on the demand element that is an important part of input
data for the CBAs. Noland and Kunreuther (1995) discuss short-run and long-run policies for
making bicycle transportation more popular for daily commuter trips. They conclude that both
pro-bike policies and policies that are anti-auto in the long-run might be necessary for increasing the volume of bicycle transportation. Hopkinson and Wardman (1996) evaluate the demand
for new bicycle facilities based on a stated-preference route-choice study. They conclude that improved safety is more highly valued than reduced travel time and that the next stage of development would be to consider the benets of combinations of routes and facilities linked in a network.
The current study presents CBAs of walking and cycling track networks in the three Norwegian
cities of Hokksund, Hamar and Trondheim. For the rst time, CBAs take into account the fact that
a change from travel by car to cycling or walking means reduced health costs, reduced external costs
(e.g. air pollution and noise) from motorized trac and reduced parking costs. A project group
working on a National Cycling Strategy in Norway initiated the study of these more thorough
CBAs. The motivation for undertaking the study is the Norwegian Parliaments request to the
Government (Ministry of the Environment, 2001) in which the Government is asked to prepare
a National Cycling Strategy, the main goal of which is to make it safer and more attractive to choose
a bicycle as a means of transport. This Strategy must form a part of the National Transport Plan.
2. Costbenet analyses as scenarios at a strategic level
2.1. Walking and cycling track networks are not sucientother measures, like safe crossing
facilities, should also be implemented
The development of seamless walking and cycling track networks has the potential for increasing the amount of walking and cycling in Norwegian cities. However, since there is great

K. Slensminde / Transportation Research Part A 38 (2004) 593606

595

uncertainty regarding the substitution of walking or cycling for driving, the analyses are presented
as scenarios or sensitivity analyses with a best estimate and a low and high estimate of future
numbers of pedestrians and cyclists. In the scenarios it is assumed that additional measures (e.g.
safer crossing facilities and safer parking facilities for bicycles) will have to be implemented to
achieve high future percentages of pedestrians and cyclists. That means that other measures are
implicitly included in the scenarios, but not directly included in the costbenet analyses of the
cycle track networks. Fortunately, is appears that providing safe crossing facilities, which may
be the most expensive measure not included in the current analyses, are benecial to society
(Elvik, 2000; Korve and Niemeier, 2002). In order to achieve the high estimate of future numbers
of pedestrians and cyclists it may be necessary not only to implement a pro-bike policy, but also
simultaneously implement an anti-auto policy (Noland and Kunreuther, 1995). However, the
results of the low estimate of future pedestrians and cyclists in the current study, and Hopkinson
and Wardman (1996), show that cycle tracks can be economically justied on the basis of benets
to current cyclists.
2.2. Costbenet analyses at a strategic level
The costbenet analyses of cycle track networks are at an overarching strategic level. That
means that the designs chosen for dierent sections of the networks are not part of the analyses,
i.e. the analyses are not suited to making decisions about the relative sizes of sections of the networks that are to be designed as dedicated tracks for pedestrians and bicycles, bicycle lanes in the
road itself and sections of mixed trac. Designing the networks could have been included in the
analyses, but this would have required detailed data about pedestrian and bicycle trac on the different sections of the networks. Such detailed data are not yet available. Therefore, the analyses
are based on average volumes of pedestrian and bicycle trac on the cycle track networks in the
three cities.

3. Data input to the costbenet analyses

3.1. Cycle track networks in Hokksund, Hamar and Trondheimlength and costs
An average construction cost per metre of NOK 7500 (1 NOK = USD 0.14 in June 2003) of
walking and cycling tracks is used in the analyses. Dierent network designs can result in higher
or lower costs per metre. However, in the cost estimate used, the Norwegian Public Roads Administration has taken into account the fact that this is a measure that may have a high cost per metre.
By implementing high, but realistic, cost estimates in the analyses, we keep open the opportunity
set in the next task, which is to choose the designs that yield the best overall solution for dierent
sections of the network. Table 1 shows the lengths and costs of construction of walking and cycle
track networks in Hokksund, Hamar and Trondheim. The dierent sizes of the networks reect
the cities sizes. In addition to the construction budget costs in Table 1, maintenance costs of NOK
35 per metre per year (Kolbenstvedt et al., 2000) and an opportunity cost of tax funding of 20% of
budget costs, are also included in the analyses.

596

K. Slensminde / Transportation Research Part A 38 (2004) 593606

Table 1
Length and costs of construction of walking and cycle track networks in Hokksund, Hamar and Trondheim (NOK
1 = USD 0.14)
Length and cost of walking cycle track networks

Hokksund

Hamar

Trondheim

Planned main network for cycling (length in km)

Remaining parts of the network (length in km)
Cost estimates for completing the main network (NOK million)

15.1
3.2
23.6

32.9
2.1
15.8

220.0
80.0
600.0

3.2. Estimates of current and future volumes of pedestrian and bicycle trac
In order to make meaningful CBAs of measures for pedestrians and cyclists without uncertainties that are too great, we need the best possible description of current and future distribution of
travel among the dierent means of transport.
The current distribution among dierent means of transport is estimated on the basis the Norwegian nationwide travel survey from 1997 to 1998 (Stangeby et al., 1999). This survey shows that
Norwegians make on average 3.2 journeys per day and provides a distribution of these daily journeys among dierent means of transport. However, we have to make assumptions based on journey lengths and the length of the cycle track networks in order to estimate the average annual
daily trac (ADT) of pedestrians and cyclists on the walking and cycling track networks. The following assumptions have been made in order to make this estimate:
 Journeys of less than 5 km by bicycle or on foot have an average length of 3 and 1 km,
respectively.
 Journeys longer than 5 km by bicycle or on foot have an average length of 6 km.
 Of the total number of kilometres travelled by bicycle or on foot in these cities 70% and 20%,
respectively, are travelled on the walking and cycle track networks.
Future distribution among dierent means of transport is estimated on the basis of the current
distribution and assumptions about percentage changes from that distribution. Our best estimate of the future distribution among dierent means of transport is based on Lodden (2002)
and the following assumptions:
(a) Walking and cycle track networks provide 20% (540% in sensitivity analyses) induced walking and cycle journeys. (Induced walking and cycle journeys mean journeys that would not
have taken place without the new networks and not journeys that today are made using other
modes of transport.)
(b) 15% (035% in sensitivity analyses) of current journeys less than 5 km made by car or public
transport are replaced by walking or cycling.
(c) Among current journeys by car or public transport that in the future are replaced by walking
or cycling, about 1/3 are replaced by walking and 2/3 by cycling.
Table 2 shows the estimated total number of km walked or bicycled in the dierent cities and
the average daily trac (ADT per km) of pedestrians and cyclists on the walking and cycling track
networks in Hokksund, Hamar and Trondheim.

K. Slensminde / Transportation Research Part A 38 (2004) 593606

597

Table 2
Estimated total number of kilometres per day walked and cycled in the dierent cities and the average daily trac
(ADT per km) of pedestrians and cyclists at the walking and cycling track networks in Hokksund, Hamar and
Trondheim
City/mode of
transport

Hokksund
Walking
Cycling
Hamar
Walking
Cycling
Trondheim
Walking
Cycling

Number of km per day in city area

Current
estimate

ADT per km on the w/c-track network

Future estimates
Min

Best

Max

Current
estimate

Min

Future estimates
Best

Max

9777
6308

10,266
6623

12,306
11,012

15,027
16,864

129
292

136
307

163
510

199
781

20,671
13,336

21,705
14,003

26,019
23,282

31,771
35,654

126
283

132
298

158
495

193
758

143,851
152,058

151,044
159,660

180,369
228,955

219,470
321,348

131
484

137
508

164
728

200
1022

3.3. Benet components of a truly complete costbenet analysis

The CBAs include estimates of the following benet components:
Trac accidents. It is not known whether substituting walking or cycling for car and public
transport use will result in more or fewer people injured in trac accidents (Elvik et al., 1997; Elvik, 1998). A walking and cycle track network with safe crossing facilities is likely to reduce the
number of trac accidents involving pedestrians and cyclists. However, in order to avoid an overestimation of any benets, we have assumed that the number of trac accidents resulting in injury
will remain unchanged because of the new walking and cycling tracks. The sensitivity analyses include an upper estimate of increased accident cost based on current statistical accident risks and
assumptions about the number of unreported accidents.
Travel time. Cycling on a walking and cycle track could conceivably reduce travel time compared to cycling on an ordinary sidewalk. Compared to cycling in the road, travel time will probably be the same or a bit longer on a walking and cycle track. In the CBAs we have assumed that
travel times for pedestrians and cyclists remain unchanged because of the walking and cycle tracks
(Slensminde and Elvik, 2000). We assume that travel times for car drivers who do not substitute
walking or cycling for driving are reduced in cities with trac congestion. This is included in the
analysis for Trondheim as reduced congestion costs (Eriksen, 2000).
Insecurity. Insecurity felt by pedestrians and cyclists moving along a road is included in the
analyses at a cost of NOK 2 per kilometre (Stangeby, 1997; Elvik, 1998). Assuming an average
speed of 1020 km/h the cost of insecurity is about NOK 2040 per hour for cyclists. Compared
with the value assigned to travel time included in CBAs for crossing facilities of NOK 66 per hour
(Elvik, 1998; Slensminde and Elvik, 2000), the estimated cost of insecurity seems to be of reasonable relative magnitude. However, the results from Hopkinson and Wardman (1996) that
safety is more highly valued than time might indicate that the cost of insecurity in the CBAs is
a lower bound estimate. Another explanation might be that excessively high valuations of travel

598

K. Slensminde / Transportation Research Part A 38 (2004) 593606

time are used in Norwegian CBAs. Excessively high estimates of travel time savings are shown to
be a result both because embedding eects are present (Slensminde, 2003), and because inconsistent choices are not handled properly (Slensminde, 2001, 2002) in the Norwegian Stated
Choice time-value studies.
School bus transport. School children are oered transport to and from school if the route they
would take to school is classied as too dangerous to walk or cycle along. We have assumed that
50% of these children will not need transport if walking and cycle track networks (with safe crossing facilities) are constructed. Information from the municipalities indicates that the reduction in
the number of school children oered transport to and from school could be about 78 in Hokksund, 34 in Hamar and 120 in Trondheim. Based on an estimated cost of NOK 3.90 per child/km
(price adjusted from Engebretsen and Hagen, 1996) the cost per child of transport to and from
school is calculated to be NOK 4680 per year.
Less severe diseases and ailments and less short-term absence. As a benet of extra physical activity (walking and cycling) we have assumed that short-term absence from work is reduced by 1 percentage point (from 5% to 4%) (Elvik, 1998). With an average wage cost estimated at NOK 250,000
per year, this means an economic saving of NOK 2500 per year per person employed who becomes
more physically active. Twenty ve percent of all journeys are assumed to be trips to or from work.
In order not to overestimate this benet we have assumed that 50% of new pedestrians and cyclists
will see improvements in their health due to the additional walking or cycling.
Severe diseases and ailments and long-term absence/disability. Moderate amounts of daily physical activity (e.g., 30 min walking or cycling) reduce the risk of premature mortality in general (US
Department of Health and Human Services, 1996). In order not to overestimate this benet we
have only included risk reductions related to four types of severe diseases or ailments in the current costbenet analyses. The four types included are the diseases or ailments for which the Norwegian National Council on Nutrition and Physical Activity (NCNPA, 2000) has estimated the
costs to the society in the form of medical costs, treatment costs and potential productivity loss.
The four types of diseases are cancer (ve dierent types), high blood pressure, type-2 diabetes
and musculoskeletal ailments. In addition we have estimated costs due to welfare loss for people
suering from these diseases or ailments. The welfare loss is estimated to be 60% of the total costs.
This is the same magnitude as for welfare loss for people injured in trac accidents that the Norwegian Public Roads Administration uses in its CBAs of other road investments (PRA, 1995). In
order not to overestimate the benets of reduction in severe diseases and ailments we have assumed that 50% of new pedestrians and cyclists will enjoy better health due to the additional walking and cycling. In the costbenet analysis an economic saving of NOK 7300 per person per year
who becomes moderately more physical active is included.
External costs of road transport. Included in the external costs are CO2-emissions, local emissions to air, noise, congestion and infrastructure costs. These are from Eriksen (2000) and are
price-adjusted to NOK 1.36 and NOK 9.03 per kilometre for cars and buses, respectively, in major cities (Trondheim). For smaller cities (Hokksund and Hamar) the external costs are NOK 0.40
and NOK 4.57 per kilometre for cars and buses, respectively.
The Norwegian Public Roads Administration and The Ministry of Transport and Communication in their CBAs of transport investments use Eriksens values for external costs. In an attempt to make the results from the CBAs of cycling and walking track networks comparable
with results from CBAs of other transport investments we used the same values.

K. Slensminde / Transportation Research Part A 38 (2004) 593606

599

In order not to overestimate the accident costs these are excluded from external costs of road
transport. The reason is that we have assumed that the number of injury accidents is not aected
by a substitution of walking or bicycling for car or public transport use.
Parking costs. Parking costs are estimated on the basis of rental prices companies pay for parking spaces in the dierent cities. Although most companies probably have lower parking costs
than Europarks rental prices, these prices are the ones actually paid by the companies for parking
space rental for their employees (and customers?). These parking costs are therefore judged to be a
realistic estimate of companies marginal cost of parking. In the analysis we have not included any
reduction in the need of parking spaces for customers. Trips to and from work by car replaced by
walking or cycling were assumed to reduce parking costs for businesses in Trondheim, Hamar and
Hokksund by NOK 1165, NOK 560 and NOK 325 per month, respectively.
The reason for including the four latter components is to do costbenet analyses that are as
complete as possible. By complete we mean that the most essential components are included.
However, because of uncertainties in the valuation of the dierent components, it is not claimed
that the components are included with accurate estimates. These more complete costbenet analyses will nevertheless provide an insight into the magnitude and scale of the most important components that ought to be included in analyses of the economic consequences to society of
measures for pedestrians and cyclists.
3.4. Reduced insecurity is the only change in the generalized travel costs
Analyses of the demand for travel usually assume that this demand is a function of generalized
travel costs. The concept of generalized travel costs traditionally includes the sum of actual travel
costs and costs of travel time. However, a journey often involves other costs than pure cash outlays and time consumption. Such costs may for example be related to distaste and inconvenience
related to travelling.
In our analyses we assume that some travellers will begin travelling as pedestrians and cyclists
as a result of a new walking and cycling track. However, this new (induced and substituted) walking and cycling activity is not attributable to assumptions about reductions in statistical accident
risk, but on the travellers subjective experience that it has become safer and less inconvenient to
walk or cycle. Although peoples subjective perception of insecurity may be inuenced by the
information they have of statistical accident risk, it is their subjective perception of insecurity that
inuences their choice of transport mode, i.e. it is reduced insecurity that enters into the generalized travel costs and not the statistical accident risk.
Compared to todays situation, we assume that travel costs and travel time for dierent modes
of transport remain unchanged because of a new walking and cycle track. These costs are therefore not changed in the generalized travel cost in the costbenet analyses.
4. Results of the costbenet analyses
4.1. Costbenet analyses based on the best estimate of future pedestrian and bicycle trac
An investment projects calculated protability to society is denoted as its net benet. If the net
benet is positive, the project is assumed to be benecial to society (given the premises on which

600

K. Slensminde / Transportation Research Part A 38 (2004) 593606

the calculations are based). The net benet is the present value of the benet minus the cost components. The net benet/cost ratio indicates the benet per cost unit. If net benet is larger than
zero, the net benet/cost ratio will also be larger than zero. If this is the case, the investment project is judged to improve overall welfare. The higher the net benet/cost ratio, the more benecial
a project is to society per unit of money invested.
In addition to the specic assumptions outlined above, the results of the costbenet analyses
presented in Table 3 are based on a discount rate of 5% and a 25-year lifetime of the projects. As is
evident, the investment in walking and cycle track networks in these three Norwegian cities seems
to be highly benecial to society indeed.
In the costbenet analyses presented in Table 3, reduced costs related to severe diseases and
ailments constitute approximately two-thirds of the total benet in Hokksund and Hamar and
approximately half of the total benet in Trondheim. The benet components relative contributions to total benets are shown in Fig. 1. Fig. 1 shows that considerable contributions come from
reduced parking costs, reduced costs due to less short-term absence and reduced external costs of
transport. Reduced costs because of less need for transporting school children constitutes only
about 1% of the total benet.
Reduced insecurity, the only component that changes in the generalized travel costs, constitutes
about 11% of total benet in Hokksund, 4% in Hamar and 20% in Trondheim. These dierences

Table 3
Benets and costs (based on best estimates of future pedestrian and bicycle trac) of investments in walking and cycling
track networks in Hokksund, Hamar and Trondheim
Benet and cost components
Benets of walking and cycling tracks (present value)
Accidents (assumed no change)
Travel time (assumed no change)
Reduced insecurity for current pedestrians
Reduced insecurity for current cyclists
Reduced insecurity for new future pedestrians
Reduced insecurity for new future cyclists
Reduced costs for transporting school children
Reduced costs related to less severe diseases and
ailments and less short-term absence
Reduced costs related to severe diseases and ailments
Reduced external costs of motorized road transport
Reduced parking costs for employers
TOTAL BENEFIT
Costs of walking and cycle tracks (present value)
Capital costs
Maintenance costs
Tax-cost factor, 20% of budget costs
TOTAL COSTS
Net benet/cost ratio
Unit: NOK million (NOK 1 = USD 0.14).

Hokksund

Hamar

Trondheim

0
0
4.2
9.5
0.5
3.5
2.6
16.7

0
0
2.7
6.1
0.4
2.3
1.1
35.4

0
0
107.6
398.2
13.7
100.7
3.6
269.2

97.7
9.4
9.5
153.7

206.6
20.0
34.6
309.1

1572.4
124.4
433.4
3023.3

23.6
1.6
5.0
30.2

15.8
1.0
3.4
20.1

600.0
39.5
127.9
767.4

4.09

14.34

2.94

K. Slensminde / Transportation Research Part A 38 (2004) 593606

100 %
90 %

Reduced parking costs for

employers

80 %
70 %

Reduced external costs of

motorized transport

60 %
50 %
40 %

Reduced costs due to

severe diseases and long
term absence
Reduced cost due to less
severe diseases and short
term absence
Reduced costs of trans
porting school children

30 %
20 %

601

Reduced insecurity costs

10 %
0%
Hokksund

Hamar

Trondheim

Fig. 1. Benet components as percentages of total benets of investments in walking and cycling track networks in
Hokksund, Hamar and Trondheim.

reect the fact that Hamars network of walking and cycling tracks was closest to being complete,
and Trondheims was furthest from being complete (cf. Table 1). In addition, Trondheim has a
relatively larger volume of bicycle trac today that benets from reduced insecurity than the
other cities.
4.2. Sensitivity analyses
The sensitivity analyses included the eects of uncertainties relating to future pedestrian and
bicycle trac, the future number of accidents, the cost estimates ( 50% and +100%), and the discount rate (3% and 8%). Future pedestrian and bicycle trac is included in the sensitivity analyses
with a minimum estimate (assumed 5% induced walking and cycling journeys and 0% motorized
journeys replaced by walking or cycling) and a maximum estimate (assumed 40% induced walking
and cycling journeys and 35% motorized journeys replaced by walking or cycling). The volumes of
future pedestrian and bicycle trac for the minimum and maximum estimates are shown in Table
2. Table 4 shows the results from the costbenet analyses based on minimum and maximum estimates of future pedestrian and bicycle trac.
In order to investigate the eect of the possibility of more future pedestrian or cycling accidents
the sensitivity analyses include an upper estimate of increased accident cost based on current statistical accident risks, assumptions regarding the number of unreported accidents and ocial accident costs (PRA 1995). The analyses based on the best estimate of future trac with dierent
transport modes and current statistical accident risks estimate the increased accident costs (present value) at a magnitude of NOK 30 million for Hokksund, NOK 60 million for Hamar and
NOK 500 million for Trondheim. If unreported accidents for dierent modes are included in
the analyses the corresponding increase in accident costs for the three cities is estimated to be
NOK 50 million, NOK 100 million and NOK 800 million, respectively. This upper estimate of
increased accident costs constitutes approximately 30% of the total benet of the walking and cycling track networks presented in Table 3.
The high estimate used for costs and low estimate used for future walking and cycling in the
sensitivity analyses are extreme values that are very unlikely to occur in combination. The

602

K. Slensminde / Transportation Research Part A 38 (2004) 593606

Table 4
Results of sensitivity analyses
Benet and cost components

Hokksund

Hamar

Trondheim

Min

Max

Min

Max

Min

Max

Benets of walking and cycle tracks (present value)

Accidents (assumed no change)
Travel time (assumed no change)
Reduced insecurity for current pedestrians
Reduced insecurity for current cyclists
Reduced insecurity for new future pedestrians
Reduced insecurity for new future cyclists
Reduced costs for transporting school children
Reduced costs related to less severe diseases and
ailments and less short-term absence
Reduced costs related to severe diseases and ailments
Reduced external costs of motorized road transport
Reduced parking costs for employers
TOTAL BENEFIT

0
0
4.2
9.5
0.1
0.2
2.6
1.4

0
0
4.2
9.5
1.1
7.9
2.6
37.2

0
0
2.7
6.1
0.1
0.2
1.1
3.0

0
0
2.7
6.1
0.7
5.1
1.1
78.6

0
0
107.6
398.2
2.7
10.0
3.6
29.8

0
0
107.6
398.2
28.3
221.7
3.6
588.5

8.2
0
0
26.2

217.1
22.0
22.1
323.7

17.3
0
0
30.5

458.9
46.6
80.6
680.5

173.9
0
0
725.8

3437.1
290.4
1011.1
6086.6

Costs of walking and cycle tracks (present value)

Capital costs
Maintenance costs
Tax-cost factor, 20% of budget costs
TOTAL COSTS

23.6
1.6
5.0
30.2

23.6
1.6
5.0
30.2

15.8
1.0
3.4
20.1

15.8
1.0
3.4
20.1

600.0
39.5
127.9
767.4

600.0
39.5
127.9
767.4

Net benet/cost ratio

0.13

9.71

0.51

32.78

0.05

6.93

Benets and costs (based on minimum and maximum estimates of future pedestrian and bicycle trac) of investments
in walking and cycling track networks in Hokksund, Hamar and Trondheim. Unit: NOK million (NOK 1 = USD 0.14).

sensitivity analyses therefore put disproportionate weight on extreme values. Despite the fact that
each of the components of the costbenet analyses is subject to great uncertainties and that the
minimum estimate of future pedestrian and bicycle trac in the sensitivity analyses shows benets
less than costs, we conclude that:
(a) Isolated uncertainties regarding the future number of accidents, cost estimates and the discount rate do not inuence the conclusion about protability to society.
(b) Uncertainties in the estimates of future pedestrian and bicycle trac inuence the magnitude
of the net benet/cost ratio, but in any case, this ratio is probably higher than zero. Irrespective of this uncertainty, investments in walking and cycling track networks are therefore likely
to improve welfare.
5. Estimates of the barrier costs related to motorized trac
The costbenet analyses of walking and cycling tracks include estimates of the most important
benet components. The results of such more complete CBAs make it possible to calculate the
benets to society that are unrealized because motorized trac in Norwegian cities today prevents

K. Slensminde / Transportation Research Part A 38 (2004) 593606

603

people from bicycling or walking as much as they otherwise would prefer. This unrealized benet
to the society is an estimate of the barrier costs caused by motorized trac.
Motorized trac prevents a natural volume of walking and cycling for occurring in urban
areas. By a natural volume of walking and cycling we mean the volume of walking and
cycling that would take place if people were able to choose their mode of transport according
to their preferences in a situation where motorized trac did not cause insecurity and other
inconveniences.
Table 5 presents the calculated average barrier costs for Hokksund, Hamar and Trondheim.
The barrier costs (benet losses) are calculated by taking the total benet from the best estimate
of future pedestrian and bicycle trac in Table 3 and subtracting the total benet from the minimum estimate of future pedestrian and bicycle trac from the sensitivity analyses in Table 4. In
order to estimate the external barrier cost and avoid double counting, the benets of reduced insecurity for pedestrians and cyclists are excluded from this calculation.
Including the barrier costs calculated in Table 5 in total external costs from motorized trac
(Eriksen, 2000), also including accident costs, will increase the external cost for buses by approximately 33%. For cars (using petrol as fuel) inclusion of the barrier cost will increase the total
external cost by about 43% for Hokksund and Hamar, and by about 31% for Trondheim. If
one wishes to include the barrier costs as part of the basis for determining car taxes, the reduction
in external costs that constitute about 47% of total benets (cf. Table 3), should be excluded from
the estimate of barrier costs, to avoid double counting.
Our best estimate of future pedestrian and bicycle trac corresponds to an increase in the
share of bicycle journeys from 5% today to 9% in the future in Hokksund and Hamar. In Trondheim the increase in bicycle trac will be from 9% today to 13% in the future. If the natural
Table 5
Calculated average barrier costs related to motorized road trac in Hokksund, Hamar and Trondheim
Barrier costs calculated as benet loss (dierent units)

Hokksund

Hamar

Trondheim

Benet loss due to unrealized benet of a natural volume

of pedestrian and bicycle trac (NOK, present value)
Benet loss, NOK per year (annuity)
Benet loss, NOK per day

123,773,667

276,192,952

2,195,788,978

8,782,046
24,060

19,596,569
53,689

155,796,624
426,840

7.98

8.42

9.60

3.74

3.95

4.33

0.73

0.77

1.33

0.15

0.15

0.27

0.26

0.27

0.47

1.46

1.54

3.20

Benet loss, NOK per journey unrealized

pedestrian and bicycle trac
Benet loss, NOK per km unrealized
pedestrian and bicycle trac
Benet loss, NOK per motorized journey (all passenger transport
journeys added 20% freight transport journeys)
Benet loss, NOK per motorized person/km
(assumed an average of 5 km per motorized journey)
Benet loss, cars (NOK per vehicle/km,
assuming an occupancy of 1.77 persons per car)
Benet loss, buses (NOK per vehicle/km,
assuming an occupancy of 1012 passengers per bus)

Calculations based on best estimate of future pedestrian and bicycle trac (NOK 1 = USD 0.14).

604

K. Slensminde / Transportation Research Part A 38 (2004) 593606

volume of pedestrian and bicycle trac is larger than we have assumed in our best estimate, the
barrier costs presented in Table 5 are underestimates. If, for example, the natural volume of
pedestrian and bicycle trac corresponds to a future share of bicycle trac of about 13% in
Hokksund and Hamar, and 18% in Trondheim, the barrier cost will be more than doubled compared with the estimates in Table 5.

6. Conclusions and further research

The CBAs presented are based on high, though realistic cost estimates, and low benet estimates in order to prevent overestimates. The analyses are therefore judged to produce down-toearth, conservative estimates of the protability to society of building walking and cycling track
networks in Norwegian cities. CBAs conducted in this way constitute an adequate platform for
politicians and other decision makers for addressing to the overall question of setting priorities
and spending scarce public funds. With regard to the applicability of the analyses, protability
to society and perspectives with respect to setting priorities among transport investments, the
following conclusions are drawn:
(a) Best estimates of future pedestrian and bicycle trac leave no doubt that building walking
and cycling track networks in Hokksund, Hamar and Trondheim is benecial to society.
Net benet/cost ratios in these cities are approximately 4, 14 and 3, respectively.
(b) By implementing high, though realistic, cost estimates in their strategic analyses, planners of
walking and cycling track networks have a large set of possibilities when in the next stage a
choice must be made of the designs that provide the best overall solutions for the dierent
sections of the network.
(c) Compared to the relatively low net benet/cost ratios for other transport investments (cf. e.g.
The National Transport Plan 20022011 from the Ministry of Transport and Communications, 2000), investment in walking and cycle tracks in Norwegian cities is a chance for the
transport sector to make investments yielding considerably higher protability to society than
has been seen for a long time.
(d) Barrier cost is a large external cost related to motorized trac. It is therefore important
to take barrier cost into account in the same way as other external costs, when the issue is
to determine the proper level of car taxes or to evaluate dierent kind of restrictions on
car use.
The CBAs of walking and cycling track networks are based on a limited knowledge of many of
the benets and preliminary cost estimates included. In order to reduce uncertainties resulting
from this limited knowledge, the following points are suggested as necessary further research:
(a) It will be essential to perform new valuation studies to improve valuations of accidents, travel
time and insecurity. The relationship between statistical risk reected in road safety statistics
and the more subjectively felt insecurity is one topic that should be investigated in such valuation studies.

K. Slensminde / Transportation Research Part A 38 (2004) 593606

605

(b) More information on the relationship between physical activity and the incidence and costs of
dierent diseases and ailments will probably be available in the near future. This will make it
possible to include more reliable cost estimates and include more types of severe diseases and
ailments without double counting.
(c) It is not known whether substituting walking or cycling for car and public transport use will
result in more or fewer people injured in trac accidents. This issue is complicated by the fact
that accidents involving single cyclists are underreported in Norwegian accident statistics.
Obtaining more accurate information on the number of accidents involving cyclists should
therefore have high priority in order to estimate correctly the costs of more bicycle use in
the future.
(d) In Norway there is limited information on both current and projected future numbers of
pedestrians and cyclists. Related to this issue is a need for more information about risk
and insecurity related to cycling on dedicated tracks versus cycling on bicycle lanes in the
roads and how dierent designs of cycle routes inuence the volume of bicycle use.

Acknowledgments
The author is grateful to Rune Elvik for valuable comments on previous versions of this paper.
This work has been funded by the Norwegian Agency for Health and Social Welfare, the Norwegian Public Roads Administration and Institute of Transport Economics in Oslo.

References
de Ortuzar, J.D., Iacobelli, A., Valeze, C., 2000. Estimating demand for a cycle-way network. Transportation Research
Part A 34, 353373.
Elvik, R., 1998. Opplegg for konsekvensanalyser av tiltak for gaende og syklendeForprosjekt. TI notat 1103/1998,
Institute of Transport Economics, Oslo.
Elvik, R., 2000. Which are the relevant costs and benets of road safety measures designed for pedestrians and cyclists?.
Accident Analysis and Prevention 32 (1), 3745.
Elvik, R., Mysen, A.B., Vaa, T., 1997. Trakksikkerhetshandbok. Institute of Transport Economics, Oslo.
Engebretsen, ., Hagen, K.E., 1996. Omfanget av skoleskyss og kostnader ved alternative skyssgrenser i barneogungdomsskolen. TI-rapport 333/1996, Institute of Transport Economics, Oslo.
Eriksen, K.S., 2000. Calculating external costs of transportation in Norway. European Journal of Transport and
Infrastructure Research, 925.
Hopkinson, P., Wardman, M., 1996. Evaluating the demand for new cycle facilities. Transport Policy 3 (4), 241249.
Kolbenstvedt, M., Solheim, T., Amundsen, A., 2000. Miljhandboken. Trakk og miljtiltak i byer og tettsteder.
Institute of Transport Economics, Oslo.
Korve, M.J., Niemeier, D.A., 2002. Benet-cost analysis of added bicycle phase at existing signalized intersection.
Journal of Transportation Engineering 128 (1), 4048.
Lodden, U.B., 2002. Sykkelpotensialet i norske byer og tettsteder. TI rapport 561/2002, Institute of Transport
Economics, Oslo.
Ministry of the Environment, 2001. Stortingsmelding nr. 24 (20002001) Regjeringens miljpolitikk og rikets tilstand,
Oslo.
Ministry of Transport and Communications, 2000. Stortingsmelding nr. 46 (19992000) Nasjonal transportplan 2002
2011, Oslo.

606

K. Slensminde / Transportation Research Part A 38 (2004) 593606

NCNPA, 2000. Fysisk aktivitet og helseAnbefalinger. Rapport nr 2/2000, The National Council on Nutrition and
Physical Activity, Oslo.
Noland, R.B., Kunreuther, H., 1995. Short-run and long-run policies for increasing bicycle transportation for daily
commuter trips. Transport Policy 2 (1), 6779.
PRA, 1995. Handbok 140. Konsekvensanalyser. Del I. Prinsipper og metodegrunnlag, Public Roads Administration,
Oslo.
Stangeby, I., 1997. Attitudes towards walking and cycling instead of using a car. TI report 370/1997, Institute of
Transport Economics, Oslo.
Stangeby, I., Haukeland, J.V., Skogli, A., 1999. Reisevaner i Norge 1998. TI report 418/1999, Institute of Transport
Economics, Oslo.
Slensminde, K., 2001. Inconsistent choices in Stated Choice data. Use of the logit scaling approach to handle resulting
variance increases. Transportation 28, 269296.
Slensminde, K., 2002. The impact of choice inconsistencies in Stated Choice studies. Environmental and Resource
Economics 23, 403420.
Slensminde, K., 2003. Embedding eects in valuation of non-market goods. Transport Policy 10, 5972.
Slensminde, K., Elvik, R., 2000. Prioriteringsverkty for gang- og sykkeltiltakPremisser og veiledning. TI rapport
479/2000, Institute of Transport Economics, Oslo.
US Department of Health and Human Services, 1996. Physical Activity and Health: A Report of the Surgeon General.
Atlanta, GA, USA.