SURPRICE (Sustainable mobility through road user charging) ; Swiss contribution: Equity effects of congestion charges and intra-individual variation in preferences

• Forschungsstelle: ASTRA SBT
• Projektnummer: ASTRA2010/007
• Projekttitel Englisch: SURPRICE (Sustainable mobility through road user charging) ; Swiss contribution: Equity effects of congestion charges and intra-individual variation in preferences

Schlüsselwörter (Englisch)

Congestion pricing, equity, distributional impacts, case study, acceptability, preference variation, intra-individual preference variation

Kurzbeschreibung (Englisch)

If there is a large intra-individual variability in mode preferences and values of time, consumers' gains and losses when introducing congestion charges will be more evenly distributed in the population than if preferences were constant within the individual. This implies that disregarding intra-individual preferences will over-estimate the negative equity effects of road pricing.

We estimate a mode and destination choice model using the the six week travel diary Mobidrive, where a number of households recorded all their daily trips continuously over six weeks. Since the diary extends over a number of weeks, it is possible to estimate within-individual variation (over days and trips) of preferences for such factors as travel time, travel cost and mode specific constants.

We also estimate route choice effects and -models based on the Danish AKTA GPS-dataset. This followed routes with and without road pricing for a 2x12 week period.

Next, we apply the estimated models. We use the models to evaluate the consumer costs and benefits of a simple hypothetical congestion charging system.

Projektbeschreibung (Englisch)

When assessing equity effects, i.e. the distribution of gains and losses, of congestion pricing, a critical piece of information is personal preferences, such as value of time or the preferences for driving. This indicates the value to be placed on travel time savings due to congestion relief or the cost to adapt the travel behavior, and hence also the willingness to pay for charging. In the simplest of assessments, preferences are assumed to be equal for all individuals in an affected population. A more sophisticated approach is to assume that preferences vary between individuals, usually that the value of time depends exclusively on income.

Yet there is good reason to believe that the value of time varies not only among individuals, but also on different occasions for the same individual dependent on the context and purpose of the trip. Each of these kinds of variation may have important implications for estimates of both the overall costs and benefits of congestion pricing, and for the equitable distribution of those costs and benefits.

This study aims to explore how the equity and efficiency effects from congestion charging systems depend on intra-individual variation in preferences. The basic hypothesis is that preferences and therefore travel choices vary within individuals (over days and trips), and that they are only partially dependent on income. This hypothesis is supported by evidence from Stockholm. Also, from the latest Value of Time Studies in Sweden and Denmark, and the Danish AKTA experiment it is apparent that there is very large unexplained variation in value of time even after controlling for income. The value of time is as much dependent on trip-related variables such as distance and purpose, as it is on income.

If there is a large intra-individual variability in mode preferences and values of time, consumers' gains and losses when introducing congestion charges will be more evenly distributed in the population than if preferences were constant within the individual. This implies that disregarding intra-individual preferences will over-estimate the negative equity effects of road pricing.

We estimate a mode and destination choice model using the the six week travel diary Mobidrive, where a number of households recorded all their daily trips continuously over six weeks. Since the diary extends over a number of weeks weeks, it is possible to estimate within-individual variation (over days and trips) of preferences for such factors as travel time, travel cost and mode specific constants.

We also estimate route choice effects and -models based on the Danish AKTA GPS-dataset. This followed routes with and without road pricing for a 2x12 week period.

Next, we apply the estimated models. We use the models to evaluate the consumer costs and benefits of a simple hypothetical congestion charging system. We will assume realistic charges and travel time reductions for car trips going to the city centre. Using the estimated models, we will compute the overall consumer surplus for each individual and aggregate over all individuals for a number of weeks. We compare this outcome with a situation where there is no intra-individual preference variation. For both sets of assumptions, we compare the with-charges case to the case of no charges.

The project is part of a comprehensive CTS project program, which will provide many opportunities to exchange experiences and knowledge between the participants during the project.

Erwartete Erkenntnisse/ Nutzen, Nutzniesser (Englisch)

It is widely recognized that congestion pricing could be an effective measure to solve environmental and congestion problems in urban areas—a reform that normally would also generate a net welfare surplus. Despite this, implementation of congestion pricing has been very problematic. One reason for a low political and public social acceptability could be the widely spread assumption that congestion pricing has negative equity effects.

Recently, results from the congestion charging system in Stockholm and the recently undertaken Value of Time Study in Sweden have indicated that travel preferences, such as the value of time and mode choice preferences, vary not only among individuals, but also between different trips for the same individual; and that income is not the only variable that has an impact on these preferences.

If there is important within-individual variation in preferences that is disregarded in equity analysis, the negative equity effects of congestion charges are probably wildly over-estimated. In that case, the same individual could belong to those that gain on some occasions, and to those who lose on other occasions. This is of particular relevance in European cities with congested road networks and efficient public transport, making travel choices highly variable from day to day. Moreover, it might be that other preferences besides value of time, such as basic preferences for different modes, vary somewhat independently from income; this could also mitigate equity effects.

To verify and quantify the role of intra-individual variation in preferences is highly important since it might either brush aside or reaffirm the conception that congestion charges have negative equity effects.

This has mostly been described above. The most important point to stress is the partners' (perhaps primarily CTS and DTU:s) continuing work as expert advisors to policy-makers regarding congestion charges in many contexts - on national, European and global levels. This is a guarantee both for the applicability of the questions and the results and their subsequent dissemination.

Methoden (Englisch)

The project comprises four sub-groups, based at KTH, CTS (Stockholm), ETH (Zürich), EPFL (Lausanne) and DTU (Copenhagen). Each research group has unique and important scientific skills and resources that all are crucial for the accomplishment of this project. Stockholm has world leading experience of acceptability and equity effects of road pricing, Zürich has a unique data set of high quality and documentation, Lausanne is world leading in travel demand modelling, and DTU has a unique GPS-based panel data set on route choice effects and is international leading in the field of route choice modelling. This means that all parts have a unique role and that without true collaboration, this project cannot be carried out.

CTS, Stockholm: Since the introduction and evaluation of congestion charges, researchers in Stockholm have been world leading in acceptability and equity effects of road pricing. CTS has an excellent track record as to applying RUC research results to applied policy - in terms of lectures, seminars, advice etc. This is also where the issue of preference variation's impact on equity of road pricing has come up. The hypothesis that preferences varies within individuals cannot be tested in Stockholm/Sweden, since there are no mulity-day travel survey in Stockholm.

ETH, Zürich has collected a unique multi-day travel survey of high quality and documentation that are necessary for this project. They are also experts in discrete choice modelling and have extensively experience on road pricing. There are both abstract as well as realistic scenarios ready in their simulation environment for the experiments.

EPFL, Lausanne. Unique and word leading in software development for discrete choice modeling, econometrics and Dynamic Traffic Management Systems.

DTU, Copenhagen. Has carried out the largest known GPS-based experiment (AKTA) on road user charging, and has carried out several research projects and public sector advises concerning road user charging. Has developed route choice models that makes it possible to model within and between personal variation, and has applied these in several projects for road pricing evaluating in Copenhagen, in Denmark and in the European Trans-Tools model (for the European Commission). The Danish TRIP project on road pricing lead to a monograph on road pricing published by Springer in 2008.

In recent years, these four groups have had transitory contact with each other, given their similar research interests, and collaborations have previously been discussed given our compatible and complementary interests and capabilities. This proposal is a direct result of such interactions, and the hope is that such interactions will be deepened by engaging collectively in the project described.

Our cooperation will continue in the form of joint authorship of scientific papers and presentation of results at international conferences.

Spezielle Geräte und Installationen (Englisch)

None.

Allgemeiner Stand der Forschung (Englisch)

Several scholars have argued that those with high incomes gain most from congestion charges, since they are assumed to have a higher value of time, and hence more often feel that the time gain is worth the charge (Richardson, 1974; Evans, 1992; Arnott et al., 1994; Small, 1983). Other scholars argued that the situation may be different in typical European cities, where citizens have access to and actually patronise public transport and "slow modes", in particular during peak hours (Armelius, 2004). Since those still driving usually have high incomes, they pay more tax (Glazer and Niskanen, 2000; Foster, 1974).

Small and Verhoef (1999) show that if there is heterogeneity only in values of time, then pricing on serial links makes the users with the lowest values of time suffer the greatest average welfare losses, or enjoy the smallest gains. This pattern changes, however, if there are parallel routes without road pricing: then, the users with the trade-off value of time, being indifferent between the two routes, suffer most or gain least. Ignoring heterogeneity also causes great underestimation of total consumer welfare benefits through self-selection.

In summary, the different conclusions regarding the equity effects on road pricing largely rely on what assumptions are made about the values of time of different income groups, and how that determines travel behavior through self-selection. Self-selection will tend to ‘‘sort’’ trips such that those with highest value will stay on the priced road (and enjoy time benefits). The essential point is that it is trips that are "sorted", not individuals: it is the confusion of trips and individuals that has obscured earlier results and interpretations of equity and acceptability effects.

Up to now, earlier studies on equity effects of road pricing have disregarded the possibility that preferences might vary between different trips for the the same individual. If different individuals belong to the groups that benefit most and least at different occasions, the cost and benefits will be more evenly distributed in the population. Eliasson (2009) put forward the theory that the high acceptance of the congestion charges in Stockholm could be due to the fact that individuals belong to different value of time segments on different days, or different journeys, decreasing the negative effects on equity.

Previous literature has also usually assumed that heterogeneity in preferences depend exclusively on income. In Stockholm and in other European cities with congested road networks and efficient public transport, it might be that not only values of time, but also preferences for different modes, are only somewhat dependent on income as a factor in mode choice. Also, from the latest Value of Time Studies in Sweden and Denmark, it is apparent that there is very large random variation in value of time, and that these values of time are more dependent on trip-related variables, than on income (Börjesson and Eliasson 2010). The Danish AKTA experiment (Nielsen and Vuk, 2008) also revealed a strong within-person variation of Value of Times with regard to route choices with and without road user charging.

In summary, there are clear indications of potentially important intra-individual preferences distributions, but for now this remains an unconfirmed anecdote. Using two long-panel data sets - the Swiss Mobidrive diary and the Danish AKTA route choice data - it is possible to investigate this question more conclusively.

Projektziele (Englisch)

This study aims to explore how the equity and efficiency effects from congestion charging systems depend on intra-individual variation in preferences. The basic hypothesis is that preferences and therefore travel choices vary within individuals (over days and trips), and that they are only partially dependent on income. This hypothesis is supported by evidence from Stockholm. Also, from the latest Value of Time Studies in Sweden and Denmark, and the Danish AKTA experiment it is apparent that there is very large unexplained variation in value of time even after controlling for income. The value of time is as much dependent on trip-related variables such as distance and purpose, as it is on income.

Forschungsplan (Englisch)

The first stage in the project will be an exploratory analysis of the intra-individual variability of travel choices. This will give a first insights as to how much travel choices actually do vary over days and trips for an individual. The work builds on two data sources; the Swiss Mobidrive travel diary, on which mode and destination choice models can be estimated, and the Danish GPS-based AKTA data, on which route choice models can be estimated. Both data sets contains observations of each individual or car over several weeks.

The next step in the project is software development. There are currently no software packages designed for estimation of discrete choice models that incorporate multi-level panel effects. We therefore need to add this feature to the existing software package Biogeme (Bierlaire, 2003). Biogeme is a software package that is widely used in research in the context of discrete choice models.

In the third step mode and destination choice models are estimated on the data from the Mobidrive survey, and route choice models based on the AKTA data. The estimation will reveal the intra-individual distributions of preferences for level-of-service attributes, different modes and types of destinations. The choice of attributes for which we model multi-level panel effects is chosen to support the case study simulation in the next step. We are, for instance, particularly interested in variability in preferences for the different modes and values of time for trips entering or leaving the city centre.

In the fourth step we will carry out two case study simulations.

1) The simulation environment for mode and destination choice experiments can be either MATSim (www.matsim.org) or MiniStadt. There are both abstract as well as realistic scenarios ready for Switzerland and the Greater Zürich area. Using the model estimated in the previous step, we evaluate a hypothetical congestion charging system, in terms of net benefits and equity effects.

2) The simulation environment for route choice experiments will the the Copenhagen model developed by DTU Transport. This model already contain a number of different road pricing scenarios (Rich & Nielsen, 2007). The Traffic Analyst software is used for the route choice simulations. This software contains a great flexibility with regard to ability to model within and between individual taste variation, and also with regard to linkage between value of time and income and other socioeconomic variables.

Both cases compare the outcome of the evaluation where the preferences are taken to be fixed within individuals with the outcome where preferences are taken to be variable within individuals.

The project is part of a comprehensive CTS project program, where participants learn from each other in several ways and meet in workshops during the project.

The research results will be disseminated through the publication of three scientific papers, as well as through presentations at international conferences in the field of transport economics.

D1. Paper on the general results of intra-individual preference estimation in mode/destination choice models

D2. Paper on the general results with regard to route choice effects

D3-4. 1-2 papers on the specific results from the simulation case studies, analyzing effects of taking intra-individual preference variation into account in mode/destination choice and route choice, respectively.

The findings in all of these publications will presented at top international conferences and published in scientific journals.

One of the main dissemination channels is also the steady stream of seminars and lectures given by the partners directly to policy-makers. For example, researchers at the Centre for Transport Studies have acted as expert advisors to policy-makers from e.g. Norway, Italy, Netherlands, France, Indonesia, Thailand, Hungary, United States, United Kingdom and Finland etc. These contacts have been at various levels, from the civil servant level to direct communication with e.g. ministers, senators and members of national parliaments. The type of dissemination has varied from lectures and seminars to specific advice regarding e.g. strategies for implementation or scheme design. In these activities, results from several earlier research projects are used, and the results form the present project should be no exception. In our experience, these direct contacts with policy-makers, where results and conclusions from multiple research projects can be amalgamated to advice tailor-made for the policy-makers' specific context, is the most efficient dissemination channel imaginable.

The project will be led by Jonas Eliasson, with support from Maria Börjesson, both of whom are based at the Centre for Transport Studies at the Royal Institute of Technology (KTH), Stockholm, Sweden. Other major research teams are at Institut für Verkehrsplanung und Transportsysteme (team led by Kay Axhausen), Transport and Mobility Laboratory, at Ecole Polytechnique Fédérale de Lausanne Switzerland (team led by Michel Bierlaire) and DTU Transport (team led by Otto Anker Nielsen).

Several SURPRICE proposals have been submitted by the Centre for Transport Studies, Stockholm (CTS). Of those that are selected for funding, the project teams will actively collaborate, making expertise available across disciplines, facilitating events to foster increased dialog between the separate participants, and to share

administrative resources. By creating a structure for cooperation in advance, we will benefit from the close relation of subject matter between the various projects included throughout the lifetime of the projects. Each of the CTS projects includes this attachment as a supplement to the main proposal.

To supplement the project leader of each participating project, CTS will allocate a shared program office, responsible for cross-project collaborative activities, dissemination of research results, administrative reporting,

and financial management.

Collaborative activities will include:

• A kick-off event in Autumn 2010, where the project leaders and key collaborators will meet to present their project plans, allowing ample time for discussion and feedback from other project teams. The event will also allow personal contacts to be made across project teams, so that future feedback can be more easily obtained.

• An online collaboration platform for the CTS-SURPRICE projects, hosted by CTS, where members of the project teams can freely post current thoughts or results in an open forum, and where other project teams can freely comment. This platform will also include a blog, visible to the public, financers, and other stakeholders.

• A joint summary paper will be co-authored by select members of each project team and submitted to a scientific journal.

All four parts have a unique role and contribution to in this project, which are explained in section 4h.

Within each CTS-Surprice project, the collaborators contribute as given in the main text. The motivation behind the crossproject collaboration activities mentioned above is that there are contributions that individual experts working on one project could lend to other projects; and that these kinds of contributions should be encouraged, to the benefit of all of these projects.

As an example, there are several projects that concern "acceptability" as a key concept, but which take very different methodological approaches. International experts are participating in each of these projects, but their expertise could be even better utilized over several projects, by soliciting their participation in seminars, followup correspondence, and joint dissemination of results. This participation needs to be tempered by the demands of the respective projects in themselves, but we expect the net benefit to each project to be positive.

The communication within the project will be based on:

- research visits and joint workshops

- group-blog on dedicated web-page for all CTS-Surprice projects

- e-mail discussions

As to cross-project communication in the CTS-Surprice project package, the main means of communication across projects will be through personal contacts obtained at the kick-off seminar, and through the use of the collaboration tools, where even those who did not meet personally can begin to exchange feedback. Coordination of the cross-project events will be handled by the shared program office, and communication concerning the planning and participation in those events will go through them.

Umsetzung und Anwendungen (Englisch)

It has often been suggested that the high acceptance of the congestion charges in Stockholm could be due to the fact that individuals belong to different value of time segments on different days, or different journeys, decreasing the negative effects on equity. Yet, for now this has remained an unconfirmed anecdote. This study will provide a conclusive answer as to whether the anecdote is true.

Both with regard to modelling in Sweden, Denmark, and possible in the European TransTools model, this may lead to changes in the model paradigms, which may lead to the conclusion that road congestion charging might be (even) more favorable, than what is shown by the present evaluation methods.

With regard to the specific design of road charging systems, the project will lead to a more precise forecasting tools for the behavioral responses and traffic impacts. This may lead to better designed systems with regard to the spatial and temporal variation of the charge levels

The project will also result in advanced model development to enhance existing software.

A likely outcome of this project is the conclusion that the assumed negative equity effects of congestion pricing are overestimated in European cities where there is efficient public transport as an alternative.

If verified, this finding could largely bolster arguments in favor of congestion pricing, and thereby increase the political and public social acceptability for congestion pricing, since these negative equity effects are seen to be one of the most serious objections of congestion pricing.

Hence, if congestion charging systems have less negative effect on equity that previously have been anticipated, this could have profound implications for strategic transport planning in European cities. Many European cities are now considering introducing congestion charging. Equity questions then immediately come to the forefront: policy-makers need to know what characteristics of a congestion charging system are crucial for avoiding negative equity effects. The outcome of this project will support this process.

Intra-personal variation in preferences, in value of time in particular, is also of importance for equity effects in CBA in general, and not only in the context of congestion charges. If mode choices vary highly within individuals, negative equity effects of using different values of time for different travel modes are less important. This project will therefor also contribute to a broader decision making context.

Berichtsnummer (Deutsch)

1500

Berichtsnummer (Englisch)

1500