the specification of sustainable urban transport strategies

This article was downloaded by: [University of Chicago Library]On: 21 November 2014, At: 13:48Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Sustainable Development &World EcologyPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/tsdw20

The specification of sustainable urban transportstrategiesAnthony May a , Simon Shepherd a & Paul Timms aa Institute for Transport Studies , The University of Leeds , Leeds, UKPublished online: 04 Mar 2011.

To cite this article: Anthony May , Simon Shepherd & Paul Timms (1999) The specification of sustainable urban transportstrategies, International Journal of Sustainable Development & World Ecology, 6:4, 293-304, DOI: 10.1080/13504509909470019

To link to this article: http://dx.doi.org/10.1080/13504509909470019

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/tsdw20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/13504509909470019

http://dx.doi.org/10.1080/13504509909470019

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Int. J. Sustain. Dev. World Ecol. 6 (1 999) 293-304

The specification of sustainable urban transport strategies

Anthony May, Simon Shepherd and Paul Timms

Institute for Transport Studies, The University of Leeds, Leeds, UK

Key words: transport, sustainability, strategies, urban, optimisation, finance

SUMMARY Transport strategies have been developed for nine European cities to achieve optimum performance in terms of a range of objective functions. The functions selected represent economic efficiency, sustainability and a combination of these. The strategies have been based on combinations of a standard set of policy instruments, including public transport infrastructure, frequency and fares, road capacity increases, low cost road pricing and parking charges. The optimisation method is described, and results presented. The reactions of city authorities to the proposed strategies are discussed and implications for transport policy outlined.

INTRODUCTION It is now generally accepted that the most appropriate transport strategy for an urban area will be an integrated one, in which a range of policy measures are combined, and synergy between them achieved. As inputs to this process, guidance is available based on experience with past integrated transport studies (May and Roberts, 1995) and with individual policy measures (IHT, 1996). The latter lists over 50 measures, under the broad headings of land use, infrastructure, management, information and pricing.

The wide range of possible policy measures makes it difficult to speci? the most appropriate combination of measures to meet agiven objective in a given urban area, and the specification will almost certainly differ if the objectives are redefined. The task of strategy development is in practice more complicated still, because each of these measures can be implemented in different locations, at different times, and at a range of scales and intensities.

Typically, integrated transport studies have involved identifying an appropriate list of

measures and testing them, in a range of combinations, using a strategic model. Professional judgement is used to determine the set of initial combinations to be tested, and the variants which are likely to perform best in meeting the specified objectives. May and Roberts (1995) describe the process and provide examples of the resulting strategies, including those for a study of Edinburgh (May et aL, 1992) in which some 70 strategic model runs were conducted in order to identlfy those strategies which performed best in terms of economic efficiency within given financial constraints.

Even with such a large number of model runs there was, of course, no guarantee that the optimum set of measures had been identified; different scales and intensities for infrastructure, management or pricing measures might well have achieved greater benefits. To tackle this problem a new methodology has been developed for streamlining the optimisation process (Fowkes et a l , 1998). This treats the relationship between the objective function (say economic efficiency)

Correspondence: A. May, Institute for Transport Studies, The University of Leeds, k e d s LS2 9JT, UK. e-rnail: [email protected]

293

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14

Sustainable urban transport policies May et al.

and the policy variables (say fare, service level) as a regression model, and uses a series of runs of the strategic model to produce a regression model from which an optimum policy specification can be estimated. Iterations of this sequence of model runs and regression are then conducted until the process is sufficiently converged. When applied to Edinburgh, this process was able to predict an economic efficiency optimum using only 24 strategic model runs, and with a net present value of benefits some 20% higher than the best strategy generated in the original study,

This approach has since been applied in an EU-funded research study of appropriate strategies in nine European cities. The project, OPTIMA (Optimisation of Policies for Transport Integration in Metropolitan Areas), has focused on strategies financed by the public sector designed to satisfy two separate objectives of economic efficiency and sustainability (May et aL, 1998). In the following sections we describe the background to the study; the study method and the selected cities and policy measures; the two objective functions; and the results and the initial reaction of the city authorities. We then draw some wider policy conclusions. Finally, we outline the ways in which reactions to the results are being addressed in subsequent research.

BACKGROUND TO THE STUDY The optimisation procedure adopted in the study, and the basis for its derivation, are more fully described elsewhere (Fowkes et aL, 1998). Briefly, the ideas behind the method emerged during the Edinburgh study referred to above (May et al., 1992). That study focused on four key policy variables, which proved central to the specification of the preferred strategies:

(1) The inclusion or otherwise of two light rail

(2) Differing levels of road capacity reduction in the city centre for environmental purposes;

(3) Differing levels of fares reductions (or increases) on the bus network;

(4) Differing levels of road pricing charge for vehicles entering the city centre.

Various combinations of these were tested,

lines and a new radial road link;

together with a common set of other policy measures, including urban traffic control, bus priorities, residential traffic calming and changes in public transport service levels, which had previously been identified as essential to any preferred strategy. Even with this reduction in the complexity of the policy space being tested, it rapidly became clear that it would not have been possible readily to identify the combination of the policy variables which best met a particular objective. Instead, a process was adopted, within the computing resources available, of keeping all but one policy variable fixed and testing a range of values for the other variable; fixing it at its best value and repeating the process. For the three continuous variables (road capacity, fares and road pricing) it became clear that the relationship between the objective being considered and the value of the variable was either monotonic throughout the range of values tested (with a clear highest value at either the upper or lower end of the range) or continuous, with a clear maximum within the range of values. This led to the concept of a multidimensional policy response surface, bounded by the highest and lowest acceptable values for each policy, which could be differentiated to identify that maximum value within that policy space.

Such a procedure should, ideally, be able to reinforce the standard model-based analysis of transport policy options based solely on professional judgement. In particular, the use of mathematical optimisation techniques should enable better combinations of measures, achieving higher performance against the defined objectives, to be identified. It should also enable the number of model runs required to be significantly reduced. As noted above, both of these goals were achieved in the application in Edinburgh (Fowkes et aL, 1998).

It was this procedure which was further developed in a subsequent study funded by the UK Engineering and Physical Sciences Research Council, following a review which identified no successful attempts to address the same problem.

STUDY METHOD The resulting optimisation procedure is shown in Figure 1. It starts with the definition of the objective functions, as described in the next

294 International Journal of Sustainable Development and World Ecology

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14

Sustainable urban transport policies

Specify calibrate regression model

(Step 4)

May et al.

-

section. It then specifies the policy variables, the acceptable range for each, and their capital and operating costs. The next step is to conduct transport model runs to test a 'do-minimum' and an initial set of combinations of transport policy variables for a horizon year. The number of combinations in this set is the minimum number required to start up the optimisation process. The actual combinations are chosen using an orthogonal design so that as many different types of combination of measure as possible are tested (subject to the limit on the overall number of initial runs).

Using the objective function values for these initial runs, a multiple regression is carried out, which aims to explain the (objective function) results in the form of an equation. The variables in this equation are the values of the measures. Following the experience in Edinburgh, the form of the equation was specified as having:

( 1 ) Dummy values for inclusion of specific infrastructure projects;

(2) Linear and square terms for each of the continuous variables (e.g. fares);

(3) Interaction terms represented by products of the variables (e.g. fare x frequency),

This form of equation provides an approximate representation of the transport model results. However, the iteration procedure described below ensures that the shape of the surface is only important close to the optimum value. The curve defined by the equation will have a maximum value either within the range of feasible values of each of the policy variables or else at the lowest or highest value that has been specified. This maximum value of the curve gives an estimate of the set of values of the transport policy variables which give the highest value of the objective function, i.e. an estimate of the optimum set of measures within the ranges specified.

The transport model is next run to determine the true value of the objective function for this predicted optimum combination. The true value is likely to differ significantly from the prediction at this stage, because the prediction is based on only the minimum number of policy runs. To improve on the estimate, the model run for the predicted optimum combination, and runs for other combinations close to the estimated optimum, are added to the set of model runs.

Define objective, key indicator

(Step 1) . Specify policy measures

(Step 2) . I 1 1 Initial runs of strategic model I

Estimate optimal policy

Step 5

~ ~~

Run approximation to predicted optimum and additional runs

around this

(Step 6)

1'

4' Optimum defined as in step 5

Figure 1 The optimisation process

International Journal of Sustainable Development and World Ecology 295

Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14

Sustainable urban transpwt policies May et al.

Table 1 City chracteristics

fillinburgh Mmqrside Vienna Eisenstrutt Tmm0 Oslo Helsinki Tmho S d m t o

Population (k) 420 1440 1540 10 57 919 891 1454 157 Density/ha 29.9 22.2 37.9 2.4 0.3 1.7 12.0 23.7 26.2 Car ownership per person 0.32 0.27 0.32 0.66 0.38 0.44 0.32 0.63 0.40 Trips by car (%) 51* 78* 37 56 54 62 47 67* 40

*Motorised trips only

Then, using the results of the new transport model runs as well as the initial runs, a new regression estimate is made, leading to a new estimated optimum. Further transport model runs are then carried out to calculate the objective function for this new estimated optimum. This procedure (involving transport model runs and statistical regressions) carries on iteratively until the user is convinced that a true optimum has actually been achieved (Shepherd et aL, 1997). This process is relatively simple to execute, but requires skill in interpretation. It is perfectly possible for it to focus on a local optimum; equally there may be discontinuities in the response surface within the policy space. One purpose of Project OPTIMA was to test the method using a range of transport models, in a number of cities, and operated by different research groups.

The contextual approach which guided the development of the OPTIMA proposal is described in Timms (1999). Nine cities were selected which were reasonably representative of urban transport conditions in Europe. It was important that these cities had transport models which had previously been used in strategic studies, and which were accepted as credible models by the city authorities. Given this, the project was able to focus on the use of the mathematical optimisation method to identify optimal strategies in the nine cities, using the cities’ own accepted analytical techniques, and on comparison of these optima between cities.

The OPTIMA project team involved research groups from five European countries, and each was invited to identify two cities, for which it had access to a transport model, and whose city authority was interested in participating in the study. One country, Finland, only selected one city, Helsinki. Overall, the nine were selected to represent a wide range of conditions (Table 1). Three are large in population terms, three

medium and three small. Three have much lower population density than the others. Car ownership varies widely, with much higher levels in Eisenstadt and Torino. Car use is highest as a percentage of trips in Oslo, Merseyside, Torino, Eisenstadt and Tromsra.

An inventory of measures in use, planned or already rejected was conducted in the nine cities. Based upon this inventory and discussion with the city authorities, a set of common measures was selected for use in the optimisation process. Table 2 shows these measures and the maximum ranges considered (some cities used narrower ranges where it was felt that the maximum range was simply infeasible). The criteria for selection of measures were that the measures:

Were common to all nine case study cities (either already used or planned);

Could be modelled by all the nine city- specific transportation models;

Were likely to be used or planned in a large number of cities throughout Europe;

Were (or arguably should be) controlled by the city authorities.

All measures were defined relative to a pre- specified ‘do-minimum’ strategy for the city concerned for a horizon year taken, typically, as 2010. The changes relative to the ‘do minimum’ for road capacity, public transport frequency, fares, road pricing and parking charges were directly comparable across cities. The public transport infrastructure investments were less directly comparable, and were based on asking each city to identify their preferred infrastructure project(s) (in addition to any included in the ‘do- minimum’) requiring high investment (typically rail-based) and medium investment (typically bus- based), Inevitably the resulting projects depended on city authorities’ interpretations of this brief.


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14


Table 2 Measures tested

Measure Minimum value Maximum value

High public transport infrastructure investment 0 1 (dummy) Medium public transport infrastructure investment 0 1 (dummy) Increased/decreased road capacity (whole city) -20% +20% Increased/decreased public transport frequency (whole city) -50% +loo% Increased/decrceased public transport fares (whole city) -100% +loo% Road pricing charge to enter city centre 10.0 ecus 0 Increased/decreased parking charges (city centre) -100% +500%

Following consultation, the only changes in road capacity included were low-cost ones using traffic management and intelligent transport systems. Estimates were obtained from each city of the implementation and operating costs of each of the measures.

OBJECTIVE FUNCTIONS Based on initial discussions with the city authorities two policy objectives, economic efficiency and sustainability, were chosen as the basis for optimisation in OPTIMA.

The Economic Efficiency Function (EEF) reflects the cities’ objectives of overall efficiency of the transport system, and involves a cost-benefit analysis of the tested policy following procedures in the UK’s Common Appraisal Framework (MVA et ad, 1994), The optirnisation with regard to this function is to find the policy with the best Net Present Value (NPV) of social benefits and costs. Apart from direct capital and operating costs, the only resource costs reflected in NPV were travel times, which were costed by using country-specific values of time. No attempt was made to place values on environmental or accident externalities, since not all countries had accepted values for these, and not all city models predicted the relevant variables.

Conventional discounting approaches were adopted, using the single horizon year model run to estimate the stream of costs and benefits, relative to the do minimum, over a 30-year period from the present day, and discounting these back to the present day, using a discount rate specified by each country, which varied from 6% to 9%. Concern was expressed by city authorities that those strategies which maximised NPV might not be affordable. To reflect this, an additional term

was included to represent the shadow price of public funds in excess of those required for the ‘do-minimum’ strategy. After consultation, a shadow price of 1.25 ecu per ecu spent was specified. No similar shadow price was attached to policies which reduced public expenditure, in order to avoid favouring policies which concentrated on generating revenue.

The Economic Efficiency Function (EEF) is defined as:

EEF = B - I t 0.25PVF if PVF < 0 = B - I ifPVF2O

(1)

where B, the present value of net benefits over a 30-year period relative to the ‘do minimum’, is given by:

30 1 i=l (I+r)’

B=C- x(f+u)

I is the present value of the cost of infrastructure investment, compared to the ‘do-minimum’ scenario;

f is the net financial benefit to transport suppliers and government in the modelled target year, compared to the ‘do-minimum’ scenario, taking into account both revenue and operating costs;

r is the annual (country specific) discount rate;

u is the net benefit to transport users in the target year, compared with the ‘do-minimum’ scenario, calculated as described above; and

PVF, the Present Value of Finance, is the net present value of the stream of costs and revenues relative to the ‘do minimum’, defined as:

PVF =-I+C- 30 x f (3) i=* (I+r)’


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14


While the Economic Efficiency Function represented a relatively uncontroversial application of standard cost-benefit analysis, the specification of the Sustainability Objective Function was novel. A full report of the construction of SOF is given by Minken (1998), who describes the approach as an adaptation of the work of Chichilnisky (1996) and Heal (1999) for the specific requirements of OPTIMA. Essentially, the SOF approach is to create a perspective that could be termed ‘dictatorship of the future’, whereby the welfare of future generations totally outweighs the welfare of the current generation. This perspective should be seen as in opposition to the ‘dictatorship of the present’ perspective which underlies the EEF approach. Ideally, SOF should reflect attributes of the transport system of relevance to future generations such as resource depletion, loss of life, degradation of the local environment, land consumption, ecological impacts and global warming. Furthermore it would also assess these for a series of horizon years. However, such an approach was not feasible in OPTIMA due to limitations of the modelling tools. Instead, a consciously simplified approach was adopted which was able to be implemented with all the models, and which would be readily understood by the city authorities. In it sustainability was based on the depletion of fossil fuels, as measured by the level of fuel consumption, and only one horizon year was modelled. The level of fuel consumption was assumed to act as a proxy for all other attributes of sustainability, as listed above. Since fuel consumption is directly related to CO, emissions, and closely correlated with atmospheric pollutants and accidents, such an assumption seemed to be reasonable for the purposes of OPTIMA, although future research should allow for improvements to be made to the definition of SOF.

Fuel consumption was assigned a shadow price, of 5 ecu per ecu of direct cost, to reflect the wider implications for sustainability, and an additional large penalty was imposed to constrain fuel consumption not to exceed that in the ‘do- minimum’ strategy. The level of shadow price, 5 ecus per ecu, is dependent upon the range of issues included in the definition of sustainability and is a subject for further research. On the other hand, the large penalty was simply a figure that

was large enough to ensure that no recommended transport policy would lead to a higher level of fuel consumption than in the ‘do-minimum’. Any large figure that achieves this objective would be appropriate. Costs and benefits were only assessed for the horizon year, in order to focus solely on future generations. However, some tests were also made with a weighted combination of EEF and SOF to reflect the current implications of an emphasis on a sustainable future.

The Sustainability Objective Function (SOF) is defined as:

SOF=b-y-z (4) (if fuel consumption exceeds ‘do-minimum 7 SOF=b-y (othenuise)

where b is the benefit in the horizon year to travellers, operators and governments, given by:

b = f t u ( 5 )

y is the ‘weak penalty’ on fuel consumption in the target year (calculated by multiplying the fuel consumption cost by 4 to give a shadow price of 5 ecu/ecu); and

z is the ‘strong penalty’ on fuel consumption in the target year (a large value taken as 1000 Mecu, which ensures that no package of measures can be selected if it increases fuel consumption from the ‘do-minimum’) .

RESULTS Table 3 summarises the optimal strategies for the economic efficiency objective function. Taking the policy variables in turn, several issues arise.

New public transport infrastructure is only advocated for UK cities, where past investment has been lower, and then only at the medium, bus-based, level. This may, of course, be because the other cities advocated over-designed, expensive schemes, but it is noteworthy that this has happened in all the five continental European cities involved. This is not, of course, to say that public transport infrastructure projects will never be worthwhile. It may well be that other, lower cost additions to the ‘do minimum’ strategy could have been justified.


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14


Table 3 OPTIMA. EEF - optimal measures

Infrastructure investment No No No * * No No NO NO

* * * No NO NO Infrastructure investment Yes Yes No high

medium Road capacity t20%' t5% t10%t +lo%+ t20%t t20%t t20%' +lo%' t l O % t

Road pricing (ecus) 1.6 Ot O+ Of O+ 1.2 0' Ot 1 .o

PT frequency t85% +60% +loo%+ +loo%+ -35% -26% -30% 0% +50%+ FT fares -60% -100%' t31% -100%t -50% -70% t25% -25% -50%

Parking charges - -100%' +226% t149% 0% 0% +500%' -50%

*Indicates that the measure was not tested -, indicates that the value of the measure was irrelevant at the optimum; tindicates a boundary value of the measure. See Table 2 for definition of measures

In all cities, low cost increases in road capacity through, for example, application of Intelligent Transport Systems and conventional traffic management, were advocated. This is at first surprising, since such increases might be expected to increase car use. However, it appears that their combination with measures to improve public transport and to increase the costs of car use allows them to be used, cost effectively, to improve the efficiency of the vehicle movements which continue to use the road. Conversely, no city included low cost reductions in road capacity as a way of controlling car use; in all cases these were found to be inefficient.

In all but one case, the optimal strategy involved improvements in public transport, either by increasing frequency, or by reducing fares, or both. Fares reductions were more common, arising in seven cities, while frequency increases were proposed in five. In two cases fares reductions were accompanied by a frequency reduction, while in one case both frequency and fare increased. These examples indicate that, where frequency is high, or fares low, it may be more efficient to reallocate the way in which resources are used to attract patronage. Only in one case, Helsinki, was a reduction in the service offered, by reducing frequency and increasing fares, proposed. This inevitably attracted considerable concern; it appeared on examination that the Helsinki public transport system may well be over-subsidised in terms solely of the efficiency objective.

Finally, all but three cities included an increase in car use costs. In three cases this involved road pricing, typically with a partially offsetting reduction in parking costs. In three cities parking

charges were increased substantially. Unfortunately some of the models were not particularly well able to distinguish between these two measures, and it is thus difficult to assess their relative merits.

In all cases the preferred strategy involved a combination of policy measures, and the process of optimisation demonstrated clearly the importance of the interaction between these policy variables. It is difficult to claim that true synergy was arising, since many of the policy measures were mutually reinforcing rather than achieving a benefit greater than the sum of the parts. However, the importance of determining the correct levels of these policy variables together cannot be overstated.

Attempts were made to explain the recommended strategies in terms of the characteristics of the cities. Inevitably, with only nine data points, no attempt could be made to generalise the policy recommendations. However, it was clear, as might be expected, that the more congested cities were likely to involve more substantial increases in car costs, with those least congested experiencing no increase or, in the case of Merseyside, a reduction. Similarly, those with the highest levels of public transport provision or lowest fares were least likely to identify further public transport improvements.

One important feature of the results is that in many cases the optimal value within the policy space is a boundary value. This is particularly true of road capacity, where the maximum increase was recommended in all but one city. The same occurred for public transport frequency in three cities. In all of these cases higher values might


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14

Sustainable urban transpmt policies May et al.

Table 4 OPTIMA. SOF - optimal measures

Infrastructure investment high

Infrastructure investment medium

Road capacity PT frequency PT fares Road pricing (ecus) Parking charges

Yes No Yes * * Yes

No Yes No * * *

+20%' t20%+ tl% +lo%' +20%t t20%t +100%' t59% tIOO%t +loo%' -28% -20% -100%' -100%t t l % -100%' -100%t -100%t

2.8 0' Ot Ot 2.5 7.0 - -loo%+ t250% t149% -loo%+ -100%'

Helkrinhi Ta'no Salerno

No Yes Yes

No No No

0% t l O % ' t10%' 0% -30% t50%t

-100%' -50% -100%' Ot 0' 2.0

t92% t500%' -100%'

*Indicates that the measure was not tested; -, indicates that the value of the measure was irrelevant at the optimum: 'indicates a boundary value of the measure. See Table 2 for definition of measures

have been considered, but it is probable that additional costs or practical infeasibility would constrain any significant further increase. Zero fares and zero parking charges were each proposed in two cities; here the boundary is an absolute one, but one which brings with it changes in operating costs which were not fully modelled.

Table 4 summarises the recommendations for the SOF objective function. By comparison with the EEF optima, several attributes of the SOF optima can be identified. In all but one city where they were tested, new public transport investments were justified, typically at the high level, This can be explained straightforwardly by the exclusion of capital costs from the SOF objective functions, which focuses solely on future costs and benefits.

In all but one case, low cost road capacity increases were still advocated, although in Vienna they were lower. This was a surprising result, but needs to be considered alongside the recommendations for greater improvements in public transport and higher car costs. Once these are in place, improving the efficiency of the road system for remaining traffic appears still to be justified in terms of the benefits gained.

All cities included improvements in public transport frequencies and/or fares, and in most cases these improvements were greater than for the eficiency optimum. Again, there was a greater emphasis on changes in fares (which were reduced below EEF levels in six cities) than in frequency (which was increased in four). One apparent anomaly was Torino, where the frequency was reduced, but this reflected the substantial investment in new rail-based infrastructure.

Six cities imposed more intensive charges on car use than in the EEF optimum. In most cases these involved higher road pricing charges but, as noted above, the models did not in many cases allow the relative merits of road pricing and parking charges to be differentiated. As with the economic optima, there was clear

evidence that the interaction between measures was important in achieving the optimal solution. There was much less difference between the cities in the recommended solutions than with the economic optima. Only in Merseyside, where economic decline has led to low levels of congestion, were increased charges not imposed on cars, and only in Helsinki were road capacity increases not advocated. Otherwise the main differences were in intensity of impact, and these could largely be related to the existing levels of public transport provision and of congestion.

More of the optimal measures than for EEF involved boundary values either of high road and public transport capacity or of zero fares and charges. The comments above apply to these also.

Table 5 summarises selected performance indicators for the two optima for the nine cities. The change in car trips from the 'do-minimum', in the modelled year, provides a simple indicator of the extent to which the proposed strategies achieve benefits by transferring travel to more efficient and sustainable modes (or reduce overall travel). The EEF, SOF and PVF indicators, as defined above, provide a comparative indication of the extent to which strategies achieve economic efficiency and sustainability (as defined) and require expenditure of additional public funds. In all cases, following the definitions above, these


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14

Sustaina6le urban transpwt policies May et al.

Table 5

Indiwtor Opimum Edinburgh Mersty.de Viennti Eisenstnrlt Tromrg Oslo Hekinhi Torino S n h ~ o

OPTIMA: performance of the optima

Change in car trips EEF -17 -5 -10 -9 -1 -1 +6 -12 -5 from ‘do min’ (%) SOF -25 -5 -20 -9 -11 -22 -28 -14 -10

EFF(mil1ionecu) EEF t1847 t2963 t1294 t20 t37 +1230 t341 t1675 +I67 SOF t1012 t2722 -2100 +20 +16 -2146 -915 -1958 +132

SOF(mil1ion ecu) EEF +266 t352 +444 +2 t17 t227 -1012 t230 +18 SOF t295 +407 t745 +2 +20 +526 t240 +270 +23

PVF (million ecu) EEF t5 -2361 t127 -1 -2 t29 +999 t940 -58 SOF -1230 -2604 -7077 -1 -17 +1874 -2815 -4169 -176

values are relative to the specified ‘do-minimum’ strategy.

In all but one case the economic optimum strategy reduces car use by comparison with the ‘do-minimum’, but the reductions are, in most cases small, with only three of 10% or more. Since the ‘do-minimum’ level will typically be 20% or more above the present day, this indicates that car use would still increase above today’s levels, raising questions about current traffic reduction targets. The one exception is Helsinki, where the reduction of bus frequencies and increase in fares, on efficiency grounds, would lead to an increase in car use. The sustainability optimum involves greater reductions in car use than the efficiency optimum in all but two cities. In four of these the reductions are of 20% or more from the ‘do-minimum’, suggesting reductions to below today’s levels.

The optimum values of EEF and SOF are generally, as might be expected, well related to city size. The sustainability optimum is usually highly suboptimal in terms of EEF; for Edinburgh and Tromsa it is around half the optimal value (1012 vs 1847; 16 vs 37) while for Vienna, Oslo, Helsinki and Torino it has a negative EEF, reflecting a loss of economic efficiency relative to the ‘do-minimum’. Conversely, in all but three cities the SOF value for the economic optimum is within 20% of the optimal SOF value, suggesting that a weighted optimum would be closer to that specified for the efficiency optimum. This was confirmed in subsequent sensitivity tests.

The PVF values are of particular relevance. In all but two cases (Merseyside and Salerno) the economic optimum has a positive, or close to

zero, value of PVF, indicating that increased revenue from fares and charges is sufficient, when discounted to the present day, to meet the discounted costs (relative to the ‘do minimum’) of the strategy. Thus, while initial capital costs may be a problem, a strategy which retained revenues within the transport budget would enable them to be paid off, With the one exception of Oslo, the same is not true of the sustainability optima, which would require a very considerable net financial outlay.

CONSULTATION Based on these results, consultations were held with officials in each of the nine cities, who were invited to assess them against a set of criteria which focused on issues of feasibility and acceptability. Inevitably there was some overlap between the concerns under these two headings. The officials were also invited to suggest alternative strategies which they would wish to have tested, and the opportunity was taken to discuss these results. None of the alternatives proposed performed better than the predicted optima.

By far the most frequent concern was with financial feasibility. As noted above, this was only a serious problem for the economic optimum in Merseyside and Salerno, but all city authorities except Vienna (and Oslo) expressed concern with the financial cost of the sustainability optima. It is clear that pursuit of the most sustainable strategies will imply substantial financial outlay in most cities, and that there is a need to try to find slightly suboptimal strategies which are significantly more affordable.


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14

Sustainable urban transpmt policies May et al.

In a few cases, city authorities expressed doubts about the feasibility of the measures tested, and this was reinforced by the tendency to include the upper- or lower-bound measures in the optimal strategy. Specific concerns included the higher levels of road capacity increase, which were considered in some cases only to be achievable by new road construction and potentially to cause environmental damage; public transport service reductions, which would result in increased loadings; and zero fares and zero parking charges, which would both result in major changes in operating practices and costs.

In the UK and Italy examples were identified of the need for new legislation to enable optimal strategies to be implemented. These concern ability to introduce road pricing and to control private parking (for which legislation would in practice be needed in all countries), changes in the UK bus deregulation regime to permit city authorities to influence service levels and fares more directly, and changes in the Italian anti- inflation legislation, which currently requires public transport operators to increase fares and reduce subsidies. These are important conclusions, and imply that legislative changes should be sought to facilitate optimal strategies.

Several cities expressed concern over the public acceptability of certain measures. It is important to stress that these views are based on officials’ judgements rather than on public consultation. The main concerns related to road capacity increases and, as might be expected, reduced services, increased fares, road pricing and increased parking charges. Not surprisingly the deterioration in public transport in Helsinki was considered particularly unacceptable. Where strategies are fullyjustified, it will be important to present the arguments clearly and allay the fears of the public. Where a strategy involves both positive and negative measures, the latter need to be preceded, where possible, by the former.

City officials’ assessments of political acceptability were inevitably influenced by their views of feasibility and public acceptability, as reported above. However, Vienna commented that the SOF optimum was more acceptable than the EEF, since it accorded more closely with their overall approach. Some cities expressed doubts about the objective functions used. The most frequent concern was with impacts on the local environment

and safety; some would also have preferred agreater emphasis on accessibility and land use. Some city officials would also like to have seen more inclusion of measures to improve conditions for cyclists, pedestrians and disabled travellers.

CONCLUSIONS The most important conclusion to be drawn from all the nine cities is that the optimal strategies involve a combination of measures, and rely on the interaction between them in defining an optimal strategy. There is no single best measure for any city, and there is certainly no best solution for European cities more generally. The OPTIMA study has, however, generated recommendations which should be broadly relevant to urban areas throughout the EU:

Strategies should be based on Combinations of measures, and should draw fully on the interaction between such measures;

Economically efficient and sustainable measures can be expected to include low- cost improvements to road capacity, improvements in public transport levels or reductions in fares, and increases in the cost of car use;

Strategies aimed at achieving sustainability (at least as defined in the project) will require more intensive use of public transport improvements and charges for car use;

Public transport infrastructure projects may well not be justifiable in terms of economic efficiency, and it will therefore be particularly important to concentrate on low-cost solutions;

Low-cost increases in road capacity are justified as part of a strategy in which car use is controlled by increased charges, and the capacity increases improve the efficiency of the remaining traffic stream;

There does not appear to be a strong case for using low-cost reductions in road capacity as a means of reducing car use;

The level of public transport improvements will depend on current levels of service


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14


and fares, and in some cases the most efficient solutions may involve changing the balance between service level and fares without necessarily increasing costs;

The scale of increase in costs of car use will depend, in part, on current levels of congestion; the study suggests that road pricing and parking charge increases are broadly interchangeable, but this needs assessing in more detail;

Revenue from car use charges may well be sufficient to meet the capital and operating costs of the remainder of the strategy; however, to achieve this, revenues must be made available for financing these other strategy elements;

Legislation will be needed to enable implementation of road pricing and to control parking charges; in the UK and Italy there is also a case for changing legislation to permit economically more efficient public transport strategies;

Public acceptability will be a significant barrier with those measures which reduce service levels or increase costs; this implies the need for effective public relations campaigns, and carefully designed implementation programmes;

Availability of finance will be a major barrier to implementation of many sustainability-optimal strategies, and further work is needed to investigate the extent to which financial costs can be reduced by strategies which are slightly sub-optimal.

FURTHER RESEARCH Project OPTIMA has demonstrated that the optimisation process can be applied effectively in a range of cities and with different transport models. The consultation with cities identified a number of concerns over its use which were addressed in a subsequent project, FATIMA, which is being reported separately. In particular, FATIMA has addressed:

(1) The need for a shadow price on revenue generation;

(2) The need to include costs for air pollution

(3) The need to weight the impacts of EEF and SOF;

(4) Doubts over the feasibility, and costs, of some of the boundary values for individual policy variables.

FATIMA also considered the impacts of budget constraints, and the objective functions which might be appropriate were the private sector to be involved in financing and operating elements of the transport strategy.

The methodology can be used to consider a wider range of policy variables, and greater spatial and temporal variation in their implementation. It may also be possible to extend it, with land- use/ transport interaction models, to identify optimal combined land-use and transport strategies. Clearly, the more complex the problem becomes, the more sophisticated the optimisation algorithm will need to be to cope with this complexity. Research into developing an improved algorithm which does not depend upon regression functions is being carried out in the follow-up EU project SAM1 (Strategic Assessment Methodology for the Interaction of CTP instruments).

There are clearly improvements which can be made to the specification of the objective functions, particularly once models are able to generate more relevant performance indicators, and further research on these is envisaged.

Finally, the method is inevitably dependent on the transport models used. Project OPTIMA was specifically designed to take the models currently in use in the nine cities, and the differences in results will to some extent be dependent on differences in the models. It will be of interest both to conduct similar research with the same model in different cities and to assess the effects of model specification and evaluation procedures on the optimal specification in any one city.

and accidents;

ACKNOWLEDGEMENTS The research reported in this paper has been financed by Directorate General VII of the European Commission under the EU’s Fourth Research Framework. The research is co-


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14


ordinated by the Institute for Transport Studies a t the University of Leeds. The other partners are the Institute of Traffic Planning and Traffic Engineering of the Technical University of Vienna; the Technical Research Centre of Finland;

the Centre for the Study of Transport Systems, Turin; Turin Transport; and the Institute of Transport Economics, Oslo. T h e authors are grateful to colleagues in all of these organisations for their major contributions to the project.

REFERENCES Chichilnisky, G. (1996). An axiomatic approach to

sustainable development. Social Choice and Welfare,

Department of Environment, Transport and the Regions (1997). Road accidents, Great Britain, 1996. (London: HMSO)

Fowkes, A.S., Bristow, A.L., Bonsall, P.W. and May, A.D. (1998). A shortcut method for optimisation of strategic transport models. Transportation Research, 32A( 2)

Heal, G. (1999). ValuingtheFuture: Economic Themy and Sustainability To be published by Columbia University Press, New York

Institution of Highways and Transportation (1996). Guidelines fw developing urban transport strategies. (London: IHT)

May, A.D., Rand, L., Timms, P.M. and Toffolo, S. (1997). OPTIMA, Optimisation of Policies for Transport Integration in Metropolitan Areas: a review of the results applied to nine European cities. Paper presented to the 25thEuropean Transport Forum, 1-5 September

May, A.D., Roberts, M. and Mason, P.T. (1992). The development of transport strategies for Edinburgh. Proc Inst Civil Engineers, 95

1 3 ~

May, A.D. and Roberts, M. (1995). The design of integrated transport strategies. TranspmtPolicy, 2(2)

May, A.D., Shepherd, S.P. and Timms, P.M. (1998) The specification of optimal urban transport strategies. Paper to 8th Wmld Confeenceon Transpmt Research, Antwerp

Minken, H. (1998). A sustainability objective function for local transport policy evaluation. Paper presented to 8th World Confeence on Transport Research, Antwerp

The MVAConsultancy, Oscar Faber TPAand Institute for Transport Studies (1994). A common appraisal framuwk. (London: HMSO)

Shepherd, S.P., Emberger, G., Johansen, K and Jarvi- Nykanen, T. (1997). OPTIMA: Optimisation of Policies for Transport Integration in Metropolitan Areas: a review of the method applied to nine European cities. Paper presented to the 25th European Transport F m m , 1-5 September

Timrns, P.M. (1999). The interface between transport policy-making and science: a 'two-person' model. Innmation, 12( 1)


Dow

nloa

ded

by [

Uni

vers

ity o

f C

hica

go L

ibra

ry]

at 1

3:48

21

Nov

embe

r 20

14

the specification of sustainable urban transport strategies

Documents