THE COST OF URBAN CONGESTION IN CANADA

Transport Canada Environmental Affairs

March 2006 (revised July 2007)

TABLE OF CONTENTS 1. Introduction................................................................................................................ 1 2. Introduction to Congestion ....................................................................................... 2 3. Past Efforts to Measure the Costs of Congestion in Canada ................................. 3 4. Nature and Causes of Urban Road Congestion ...................................................... 4 5. Definition of Congestion............................................................................................ 5

Introduction to the Definition of Congestion.................................................................. 5 Engineering Approach to a Definition of Congestion .................................................... 6 Economic Issues: Congestion Externalities .................................................................... 6 Economic Issues: Economically Efficient Pricing.......................................................... 6

6. Approaches to Measuring Recurrent Congestion................................................... 7 Texas Transportation Institute ........................................................................................ 7 Ministère des Transports du Québec............................................................................... 8 Transport Canada ............................................................................................................ 8

7. Data Sources and Coverage ...................................................................................... 9 Data Sources ................................................................................................................... 9 Definitions of Expressways and Arterials....................................................................... 9 Coverage ....................................................................................................................... 10

8. Components and Method ........................................................................................ 10 Delay Costs (Time Lost During Congestion) ............................................................... 11 Fuel Costs (Fuel Wasted Due to Congested Conditions).............................................. 12 Greenhouse Gas Emission Costs (Emissions Due to Congested Conditions) .............. 14

9. Congestion Indicators Used in the Study............................................................... 15 10. Main Findings on the Costs of Congestion in Urban Canada ........................... 16

Total Congestion Costs ................................................................................................. 16 Congestion Costs Attributable to Delay ....................................................................... 16 Cost of Wasted Fuel due to Congestion........................................................................ 17 Cost of GHG Emissions due to Congestion.................................................................. 18 Comparison with Montréal and Vancouver studies...................................................... 19 Comparison with American studies .............................................................................. 20

11. Recommendations for Future Research .............................................................. 20 12. Conclusions............................................................................................................. 21 13. References............................................................................................................... 24

- i -

EXECUTIVE SUMMARY

Transport Canada’s The Cost of Urban Congestion in Canada research study developed congestion indicators for the nine largest urban areas in Canada. From east to west, the urban areas were: Québec City, Montréal, Ottawa-Gatineau, Toronto, Hamilton, Winnipeg, Calgary, Edmonton and Vancouver. The study focused on recurrent peak period congestion, that is, the congestion that is associated with the regular, daily build-up of traffic during the morning and afternoon commuter peak periods. The research developed common measures of congestion, using data from each urban area’s travel demand models. Although these models each produce the same outputs (i.e. simulations of vehicle trips), there were structural and methodological differences among them. These differences are important enough to preclude a simple comparison of congestion across all nine urban areas. However, the congestion indicators and methods developed in this research provide useful tools to aid individual urban and provincial transportation authorities in the planning of their own urban areas and, if applied over time, could be particularly useful in tracking trends.

The analysis of congestion has two general approaches: ‘engineering’ and ‘economic.’ This study considered both but – by design – it focused on the engineering approach to define and to develop the congestion indicators and to estimate the social costs of congestion associated with extra time wasted due to congestion and the cost associated with greenhouse gas (GHG) emissions. This focus on the direct and physical characteristics of congestion based on engineering principles that link vehicle flow/traffic speed to road capacity (measured as vehicles per hour). This approach differs from the economic approach, which estimates the cost of congestion as the “deadweight loss” or loss to society associated with excessive road use – due to the absence of proper pricing of the road infrastructure use that reflects the social cost of congestion, including the environmental and external costs of congestion. The economic approach recognizes that there is an “optimal” amount of delay (i.e. an economically efficient level of congestion) caused by impeding users, and that some congestion is already internalized for some users.

The study estimates that the total annual cost of congestion (in 2002 dollars) ranges from $2.3 billion to $3.7 billion for the major urban areas in Canada. More than 90 percent of this cost represents the value of the time lost to auto travellers (drivers and their passengers) in congestion. The remainder represents the value of fuel consumed (around 7-8 percent) and GHGs emitted under congestion conditions (around 2-3 percent).

However, these estimates of congestion costs are conservative, because limited data precluded the inclusion of costs of non-recurrent congestion (i.e. congestion caused by random events, such as bad weather, accidents and stalled vehicles), freight traffic, congestion during the non-peak hours and other congestion-related costs such as noise and stress. More data are required to better understand these additional costs.

Some Canadian authorities have attempted to measure the cost of congestion for specific urban areas. However, methods and data have varied. This study provides the first

- ii -

systematic analysis of urban congestion across Canada. In this respect, it represents a major contribution to our understanding of urban congestion in Canada.1

1 The study was conducted by the consultant team of Delcan (prime consultant), iTRANS and ADEC.

- 1 -

THE COST OF URBAN CONGESTION IN CANADA 1. Introduction This paper provides a summary of Transport Canada’s research study on The Cost of Urban Congestion in Canada. The content is drawn mainly from the findings and conclusions of the full-length study. The intention is to provide the reader with a brief, but comprehensive understanding of congestion in urban Canada.

The purpose of the study of The Cost of Urban Congestion in Canada was:

• to develop indicators and measures of congestion; • to estimate the associated impacts of congestion and their costs; and • to apply these indicators and measures to the nine largest urban areas in Canada

(from east to west, the urban areas were: Québec City, Montréal, Ottawa-Gatineau, Toronto, Hamilton, Winnipeg, Calgary, Edmonton and Vancouver).

The nine urban areas represented just over half (51 percent) of Canada’s population in 2003.2

The study focused on recurrent peak period congestion, which is the congestion, associated with the regular, daily build-up of traffic during the morning and afternoon commuter peak periods.

The study recognizes that there exist two approaches to analyze congestion: the ‘engineering’ and ‘economic’ approaches. The study emphasizes the engineering approach which focuses on the direct and physical characteristics of congestion based on engineering principles that link vehicle flow/traffic speed and density to road capacity. (The economic approach estimates the cost of congestion as the “deadweight loss” or loss to society associated with the excessive road use that arises from the absence of proper pricing of the road infrastructure.)

This paper is organized as follows:

• Section 2 introduces congestion; • Section 3 is a short overview of past efforts made for individual cities to measure

the cost of urban congestion within their own region; • Section 4 describes the two main types of congestion (recurrent and non-

recurrent); • Section 5 defines congestion according to the engineering principles that link

traffic flow to road capacity. Section 5 also discusses the economic interpretation of congestion as an externality and the implication of this interpretation on evaluating congestion costs;

• Section 6 discusses the sources of data used by the study; • Section 7 describes the three main components of congestion costs considered in

the study: time loss, fuel wasted and greenhouse gas emissions; • Section 8 presents the congestion indicators developed; • Section 9 presents the estimates of the costs of congestion for components

covered by the study;

2 Annual Demographic Statistics, Statistics Canada, 91-213-XPB.

- 2 -

• Section 10 summarizes the main recommendations made by the study in terms of congestion-related data and modeling needs. Finally,

• Section 11 presents the conclusions and considers the expected growth of traffic congestion in Canada based on major economic drivers.

2. Introduction to Congestion “Congestion” is commonly cited as a major and growing urban ill. Many Canadian drivers encounter congestion and traffic jams every day, although both the severity and the frequency of the jams vary from area to area and from day to day. Urban, regional and provincial transportation authorities aim to manage the problem of congestion through a variety of measures.

But what exactly is meant by congestion? How can it be quantified? What does it cost individuals, firms and society as a whole? Individual drivers do not have a common view and understanding of what congestion means. For some, it could mean traffic jams with complete stops and long delays. For others, it could mean stop-start driving, moving slowly or, more broadly, travelling at less than the speed limit.

This research into the cost of urban congestion in Canada addresses two major topics:

• Why is it important to understand congestion? and

• How do we comprehensively measure congestion?

Both topics must be understood if urban and provincial transportation authorities are to develop potential solutions to congestion.

• Why is it important to understand congestion?

o Research in the United States (U.S.) and Europe has demonstrated that an understanding of congestion (what it means, how it can be measured) is fundamental to urban authorities wanting to develop solutions to congestion. This study represents only a first step, albeit an important one, by suggesting a method that Canadian urban authorities could use to develop their own congestion estimates. The method suggested could also be useful in tracking congestion trends, if data were collected regularly.

o It follows that an understanding of the costs of congestion – that is, the manifestation of congestion’s impacts on travellers and on society at large – allows potential solutions to be developed in the full context of urban goals, such as improved quality of life, increased productivity, etc.

o The interest in measuring congestion and its costs does not imply that the only ‘solution’ is for urban authorities to build more roads and highways. The development of congestion indicators could significantly improve decision-making and could contribute to the development of a more sustainable transportation system, particularly in the urban sector. Measures of congestion may provide a basis for understanding how congestion can be incorporated explicitly into urban authorities’ assessments of their transportation plans, including transit, transportation demand management schemes or other alternatives to driving as well as

- 3 -

road building. The end result could be a transportation system that operates efficiently and that meets mobility needs without jeopardizing quality of life.

• How do we comprehensively measure congestion?

o A key product of this study is the development of a method that transportation authorities could use to make consistent assessments of the congestion reduction benefits of alternative transportation proposals. As described below, the method quantifies the time lost by travellers under congested conditions, as well as the additional fuel consumed and greenhouse gas emissions generated due to congestion.

o Different approaches can be used to quantify congestion. The Texas Transportation Institute’s well-known “Urban Mobility Report” focuses on comparing congestion in different urban areas in the United States. This comparison is possible because common sets of data are available nation-wide. However, a corresponding nation-wide set of data does not exist for Canada; and the U.S. approach has been criticized by some as masking underlying crucial differences among the urban areas. Accordingly, the approach adapted for this research makes uses of existing, local data and acknowledges their differences. This allows only some comparisons to be made; however, the approach provides a multi-perspective profile of congestion as well as tools that urban authorities can apply directly to their own analyses and plans. The approach also took into account the specific needs of the Canadian transportation community; notably, the consideration of congestion’s impacts on transit.

3. Past Efforts to Measure the Costs of Congestion in Canada Governments at all levels recognize that addressing congestion is an important part of implementing urban transportation plans and meeting national, provincial and local transportation sustainability goals. The ability to measure (quantify) and value congestion is a fundamental first step in being able to address it.

Canada has no national estimate of congestion. In the past, some isolated attempts to measure the cost of congestion in specific urban areas have been made, but no regular measurements of congestion take place.

The following three studies are of interest:

• A 1987 study estimated the cost of congestion to the trucking of goods in what is now the City of Toronto and in neighbouring Peel Region (which has one of the country’s highest concentration of goods-generating activities) to be of the order of $2 billion each year;3

• In 1996, Transport Canada estimated congestion costs in Greater Vancouver to be $1.2 billion per year: $0.7 billion for the movement of people in automobiles, and $0.5 billion for the movement of urban goods. The estimate was based upon the

3 Municipality of Metropolitan Toronto. 1987. Metropolitan Toronto Goods Movement Study: Technical

Report. Toronto.

- 4 -

region’s travel demand model, which took into account an estimate of the value of travellers’ time, vehicle operating costs and fuel costs. Truck operating costs were derived from the BC Trucking Association. A 1999 estimate, using newly acquired data on truck travel characteristics, updated these costs to $0.8 - $1.5 billion (including both autos and goods movement).

• By far the most comprehensive Canadian analyses of urban congestion and its costs have been conducted by the ministère des Transports du Québec (MTQ) for the Montréal region. Based on MTQ’s travel demand model, an annual congestion cost of $0.5 billion (1997$) was estimated for 1993 traffic patterns. In 2001, this was updated to $0.9 billion for 1998 ($0.8 billion for automobiles and $0.1 billion for trucks). Thus, between 1993 and 1998, the vehicle-hours of delay have increased by 54%. These model-based studies provided a prototype for the current research. They defined congestion and methods to measure congestion separately for auto and truck travel, developed costs that reflected congestion impacts such as time, vehicle operations, fuel and emissions, described how congestion varied in different parts of the urban area and, finally, described how Montréal’s congestion compared with that of similar-sized cities in the United States.4

The three studies showed that the cost of congestion is substantial. Congestion costs are measured in billions of dollars and may be increasing significantly. There are, however, inconsistencies in the methods and data upon which these studies were based, making it impossible to generalize from the findings.

4. Nature and Causes of Urban Road Congestion When measuring congestion, it is important to distinguish between the two main types of congestion, recurrent congestion and non-recurrent (incident) congestion. An understanding of recurrent congestion is necessary before non-recurrent congestion can be analyzed. Accordingly, as noted in the Introduction, the Cost of Urban Congestion in Canada study focuses on recurrent congestion.

Recurrent congestion occurs mainly when there are too many vehicles wanting to use the road at the same time. Recurrent congestion typically occurs during weekday morning and afternoon peak periods, when most people go to work and return home at around the same time. In large urban areas, the peak periods can range from 6:00 to 9:30 a.m. and from 3:30 to 7:00 p.m. In smaller urban areas, the peaks may have a shorter duration (one or two hours).

Of interest is the growing recurrent congestion that occurs during off-peak periods (i.e. during other weekday hours, and even on weekends). This reflects, in large part, a rapid growth in off-peak travel (off-peak travel is growing faster than peak-period travel in some areas). The result is that certain sections of expressways in Toronto and Montréal are regularly congested between the peak periods, essentially operating under all-day-peak period conditions.

4 Gourvil, L. and F. Joubert. 2004. Évaluation de la congestion routière dans la région de Montréal.

(“Estimation of roadway congestion in Montréal.”) ADEC Consultants and the ministère des Transports du Québec. Montréal: Ministère des Transports du Québec.

- 5 -

Non-recurrent congestion is the other main source of traffic congestion. Non-recurrent congestion is associated with random conditions or special and unique events, such as traffic incidents (disabled vehicles), accidents, work zones, irregular facility maintenance operations (e.g., seasonal street cleaners), adverse weather and special events. Because of the random nature of this type of congestion, non-recurrent congestion is more difficult to predict and address. The impact of non-recurrent congestion is significant, in that the reliability and predictability of travel time is of utmost importance to the public, to the goods-generating industries and to the economy in general. Variability in travel time leads to costly uncertainty for commuters and, in particular, for goods transporters who must meet fixed delivery schedules. Low variability in travel times can be even more important to the public than the actual duration of the trip.

The ability to identify and measure different types of congestion, including the non-recurrent type, is key to developing appropriate mitigation solutions and policy responses. The remainder of this paper focuses on recurrent congestion, so it is worth noting here some additional considerations regarding non-recurrent congestion. In the U.S., a preliminary assessment by the Federal Highway Administration estimates that roughly half of the congestion experienced by Americans is due to non-recurrent congestion. The lack of data on incidents, in combination with the scarcity of travel time data, forces some agencies to find proxy solutions to measure non-recurrent congestion. For example, the Washington State Department of Transport estimates that non-recurrent delay generally comprises between 30 and 50 percent of all peak-period delay.

The lack of coherent incident sets that could be correlated to travel time makes the estimation of non-recurrent congestion somewhat cumbersome. In some areas, where data are readily available, the challenge lies in linking the appropriate data together and selectively filling gaps. Annual traffic count programs, traffic studies and police accident reports often contain a plethora of traffic information that could be made available. These data are not, however, usually compiled in a format that is usable for congestion analysis, and different data sets may not be consistent with each other. The study recognizes the importance of measuring non-recurrent congestion to fully reflect the cost of all types of congestion. However, given the noted challenges in terms of data availability, it was premature to include non-recurrent congestion in the study. In this regard, Transport Canada has recently completed research that examines different methods that could be used to estimate non-recurrent congestion using existing data.5

5. Definition of Congestion

Introduction to the Definition of Congestion

Individually, each person has its own definition of congestion and what it refers to in terms of travelling conditions. Drivers recognize that congestion leads to costs in terms of time delays and wasted fuel. Other impacts, such as increased stress level or noise, could also be important to some individuals or groups.

The social costs of congestion are the economic costs that congestion imposes on all drivers and on society as a whole. The social costs include the loss of productive time and increases in fuel consumption, vehicle operating and related costs, incident risk, air

5 See Kriger, D. 2005.

- 6 -

pollution and GHG emissions. As noted, this study is an attempt to assess the social cost of congestion.

Engineering Approach to a Definition of Congestion

Transport Canada recognizes that the literature presents a number of definitions of congestion. After consultation with urban and provincial transportation experts The Cost of Urban Congestion in Canada study adopted the following definition:

“Congestion is the inconvenience that travellers impose on each other while using their vehicles and attempting to use the road network at the same time, because of the relationship that exists between traffic density and speed (with due consideration of capacity).”

This definition defines congestion in terms of the physical characteristics of the road. The definition is consistent with the theoretical engineering principles that link vehicle flow (actual throughput, measured as vehicles per hour) to road capacity (available capacity for throughput, measured as vehicles per hour).

Traffic streams are described by three variables: density k (vehicles per lane per kilometre), speed v (km/h) and flow q (vehicles per lane per hour). Traffic flow is the product of traffic density (vehicles/km) and speed (km/h), so these three variables are related by the equation q = kv. All things being equal, as more vehicles enter the same road, traffic density increases, travel speed falls and travel time increases. This is the fundamental traffic flow principle on which the engineering approach bases its definition of congestion.

Economic Issues: Congestion Externalities

From an economic point of view, traffic congestion occurs when the cost of travel is increased by the presence of other vehicles. By definition, externalities refer to costs (or benefits) that are not market-priced and which accrue to third parties as a result of actions taken by individuals. Congestion externalities arise because the presence of additional road users increases travel times for other vehicles.

The externality of congestion differs in some respects from most other examples of externalities. In contrast to environmental and noise externalities, congestion externality costs are in part internal to the transport sector as a whole, so they are considered an intra-sectoral externality. Motorists recognize the increased time costs they face when delayed by road congestion, but they do not recognize that they impose delay costs on other motorists on the road. Motorists do not perceive themselves as a cause of congestion; motorists perceive themselves as the victims of congestion. In a sense, the externality is internalized in the aggregate, a form of collective internalization unlike most other externalities.

Economic Issues: Economically Efficient Pricing

Economically efficient pricing requires that market prices reflect social costs as well as private costs. Measures designed to bring about efficient pricing require estimates for all elements of social marginal cost, including external costs.

From an economic efficiency perspective, the crucial issue is to distinguish between the external element of the cost of congestion and the total or average cost of congestion.

- 7 -

The total cost of congestion is borne by the sum of the users themselves. Under congestion conditions, however, each additional user inflicts costs on other users. The external element, the costs inflicted on others, is the relevant part for pricing purposes. Additional users of transport infrastructure impose costs on other transport users as well as on themselves through congestion.

From an economic standpoint, the cost of congestion has a precise meaning only if the cost refers to an optimal situation based on a determined level-of-service objective accompanied by the full economic price of this level of service. In other words, the observed congestion is compared to an optimal traffic level. This approach measures the congestion that would be eliminated by efficient congestion pricing, where such efficient pricing includes the marginal external costs associated with congestion (the costs of delays, wasted fuel, additional vehicle operations, accidents and environmental damage imposed by road users on other road users). The congestion and associated costs that would not be eliminated by efficient congestion pricing would be “efficient” in the sense of being internalized and accepted by users. Finally, in working towards congestion reduction, it is important to differentiate between “external” congestion and “internal” or “potentially-internalized” congestion.

6. Approaches to Measuring Recurrent Congestion Texas Transportation Institute

As noted, the Texas Transportation Institute’s (TTI) “Urban Mobility Report” has presented annual congestion measures for major cities in the U.S. over the last several years. Congestion is considered to occur for all traffic delay beyond “free-flow”6 or unimpeded conditions.

The TTI method relies on traffic data recorded in the U.S. Federal Highway Administration’s Highway Performance Monitoring System7 for specific sections of highways and arterials in U.S. urban areas. The data are expanded statistically to represent other roads. The method estimates speeds and travel times for each section and each time period based on theoretical speed-flow relationships and geometric characteristics. The estimates are compared with free-flow times.

Some have expressed concern that the TTI method suggests that free-flow speed is the desired objective; meaning in turn that the appropriate infrastructure is needed to meet this objective. However, such levels of capacity are neither environmentally sustainable nor economically efficient. This is because:

a) Maximum traffic flow (i.e., capacity flow) occurs only at speeds lower than free-flow (as shown in standard speed-flow relationships);

6 Free-flow is the condition of traffic when there is no interference from other vehicles on the road; consequently there is only one vehicle on the road or other vehicles are far enough away so as not to be a constraint for the driver. Drivers can travel freely at the speed they wish. 7 The Federal Highway Administration’s Highway Performance Monitoring System (HPMS) is a database of operational performance measurements for the transportation system in the US. The HPMS is a comprehensive database built from annual reports submitted by State departments of Transportation to the Federal Highway Administration as a minimum requirement to apply for federal funding.

- 8 -

b) Capacity expansion is “lumpy” and expensive. The life cycle of an investment typically involves an initial period with excess capacity and free-flow speed conditions. The initial period is followed by traffic growth and increasing congestion until additional expansion is warranted. This life cycle means that the average condition will involve some congestion;

c) Economically efficient congestion pricing would eliminate only the external part of the congestion costs, leaving a substantial amount of congestion; and

d) Drivers expect a certain level of quality of service from the road network, but the level expected depends upon the real and perceived cost of road use in congested conditions. Most drivers may accept a certain level of congestion as long as any given trip could be completed safely, within a reasonable and predictable time and with minimum interruption.

Ministère des Transports du Québec

The Ministère des Transports du Québec (MTQ) implements an alternative approach to the measurement of recurrent congestion in its Montréal studies. The approach attempts to measure the social cost of recurring congestion for the Montréal region.

The MTQ approach suggests that users expect and accept congestion in peak hours on their normal routes, but that a “threshold” exists beyond which congestion becomes unacceptable. The choice of threshold is based upon observed traffic and speed data and also takes into account local perceptions of congestion. Accordingly, the threshold provides a more realistic basis against which congestion can be measured than free-flow. For practical estimation purposes, the study adapts a threshold of 60 percent of the free-flow speed– i.e. 60 km/h for highways or expressways with free-flow speeds of 100km/h and 30 km/h for urban arterials with free-flow speeds of 50 km/h. The thresholds and the observations (on selected links) are applied to the entire region, using MTQ’s travel demand model. The study then estimates congestion delays and costs by comparing speeds on the Montréal network with these thresholds.

Transport Canada

The research that is summarized in this paper builds upon the MTQ work in Montréal and estimated congestion using the threshold approach. However, whereas the MTQ approach provided detailed analyses within the Montréal region, in the interests of providing a systematic analytical basis across the nine urban areas, the Transport Canada research adapted a region-wide perspective for each area. The study uses data output from travel demand models that had been calibrated for use in each of the nine urban areas. The models allow speeds and travel times to be synthesized on the urban networks from traffic volume and origin-destination data. The speeds and travel times could then be compared to free-flow or to any chosen threshold speeds. As data accumulate, it will be possible to use the models to make comparisons over time.

The models provide a comprehensive framework for analyzing congestion across all nine urban areas. However, their structures, base data and network depictions are not consistent. For example, except for Edmonton and Calgary, they only model a single peak; and only some models simulate truck movements (Montréal, Calgary, Edmonton and Vancouver). In addition, the functions used to represent speed as a function of

- 9 -

volume (i.e., volume-delay curves) have very different formulations from one urban area to another, and consequently may produce estimates of speed that differ significantly.8

An important outcome is that – due to the sensitivity of the estimates to the pre-determined threshold - the study examines three threshold values: 50 percent, 60 percent and 70 percent of free-flow speeds for both expressways and arterials. This recognizes that although free-flow conditions can be fixed, the percentage of free-flow speed that represents the threshold varies according to local conditions (quantitatively) and local perceptions (qualitatively). There is no single ‘acceptable threshold’ for all municipalities or for a given network.

7. Data Sources and Coverage Data Sources

The key source of data was the outputs from the travel demand models in each urban area. A consultative approach was developed to collect the data and to agree on measurements and indicators. Seven technical meetings were held for the nine urban areas with the relevant urban and provincial authorities. This consultative work brought about a common definition of expressways and arterials. It also identified differences in data definition, gaps in the available data and differences in modelling methods. With the cooperation of the urban and provincial stakeholders, it was possible to collect the relevant data and to develop a consistent set of measurements for the study’s research.

Definitions of Expressways and Arterials

For the purposes of the study, the following definitions of expressways and arterials were applied:

An expressway is a facility that serves only through traffic, with no access to abutting properties. Access is controlled and, generally, limited to ramp locations. Expressways feature uninterrupted traffic flow, i.e. there are no signalized or stop-controlled at-grade intersections. Such facilities typically have two or more lanes for the exclusive use of traffic in each direction and a divided cross section, and generally have posted speed limits of 90 km/h or higher.9 On street parking or stopping is prohibited at all times.

An arterial10 is a signalized facility that primarily serves through traffic and provides access to abutting properties as a secondary function (although intense roadside development can generate friction to traffic, thus impeding speeds along the arterial). Arterials have interrupted traffic flow, with signalized intersections spaced at distances of up to 3 kilometres or less. Arterials can have multiple lanes

8 The regional transportation models used in this study are static in nature. That is, they do not provide a dynamic reflection of traffic but rather estimate average properties, such as road speed, over the whole of a peak period. It is very hard to calibrate such models in order to come up with reliable estimates of the speeds that occur in highly congested situations, in particular where there are queues. 9 The majority of expressway links in the (urban) centre of the Island of Montréal have posted speed limits

of 70 km/h. Technically, they cannot be classified as expressways; however, in fact, under free-flow conditions, vehicles tend to travel at ‘expressway’ speeds (i.e., well above the posted limit).

10 These definitions were drawn from the Highway Capacity Manual (2000).

- 10 -

and can have a divided or undivided cross section. Posted speed limits typically range between 50 km/h and 90 km/h.

Coverage

The study’s congestion indicators and estimates of traffic congestion were for:

• Expressways and arterials only. Collectors and local streets are, by definition, assumed not to have pervasive congestion.11 (Collectors and local streets often lack any reference traffic counts, may be treated in the abstract (e.g. one link can represent several parallel local streets) or may not be included in model networks at all;

• Delays for passenger vehicles (i.e., autos) only. (As truck traffic data were available for only a small number of urban areas, and as none of the models accounted for other vehicles in the traffic mix (such as taxis, emergency vehicles, and light-duty vehicles used for commercial activity), these vehicle types were are excluded from the scope of the study); and

• Peak-period congestion only. (The urban models simulate travel in one or the other peak-period, with the exception of the Calgary and Edmonton models, which simulate both peak-periods as well as a mid-day period.)

8. Components and Method

The Cost of Urban Congestion in Canada study concentrated on three components:

•

•

•

•

•

•

•

•

•

Delay costs (time lost during congestion);

Fuel costs (fuel wasted due to congested conditions); and

An imputed cost for green house gas (GHG) emissions due to traffic congestion.

All the costs were calculated on the basis of personal passenger vehicle use. The focus on these three components was a function of the study purpose and of the availability of consistent and reliable information for all nine urban areas. A unit price value was determined for each of the three main components of congestion.

Other potential components of the cost of congestion include:

Productivity loss;

Change in accident frequency;

Change in accident characteristics;

Increase in air pollutants;

Increase in vehicle operating costs; and

Increase in noise nuisance.

Most of these effects are externalities, meaning that they are not integrated in the travel market. There is no direct market price attached to them and they are not the object of a

11 MTQ’s costs of congestion research in Montréal found that 95 percent of the region’s congestion occurred on expressways and arterials.

- 11 -

conscious exchange between parties. Although this research focused on the immediate and direct implications of congestion, it is recognized that at a certain level of severity, congestion costs could explain the location and relocation of land-based activities (i.e., industries, businesses, retail establishments, etc.) and the choices made by individuals who decide to relocate their homes or even to change jobs or schools.

Delay Costs (Time Lost During Congestion)

Transport Canada’s 1993 approach to the valuation of travel time was used to estimate delay costs.12 Because Transport Canada has been involved mostly in interurban travel issues, the approach’s methodological basis is interurban rather than urban. On the other hand, these data provide the only uniform, nation-wide source of time values.

The 1993 approach develops time values for business and for non-business trip purposes. The Cost of Urban Congestion in Canada study adapted the approach to develop work / work-related and non-work trip purposes. Based on a review of the literature, the values were updated to reflect the base year for each urban area’s travel demand model. The data were then updated to a common year (2002). The 2002 time values are summarized in Table 1 for work / work-related and for non-work travel.

Table 1. Factors Used to Calculate Costs of Delay (2002 $)

Urban Area Base Year

% Work / Work-Related * % Non-Work* $/hr – Work /

Work-Related $/hr – Non-Work

Vancouver 2003 48% 52% $29.72 $9.26

Edmonton 2000 31% 69% $25.48 $7.84

Calgary 2001 37% 63% $28.57 $8.79

Winnipeg 1992 88% 12% $24.71 $7.63

Hamilton 2001 36% 64% $29.64 $9.14

Toronto 2001 55% 45% $30.86 $9.50

Ottawa-Gatineau 1995 43% 57% $31.35 $9.67

Montréal 1998 70% 30% $27.32 $8.48

Québec City 2001 58% 42% $25.96 $8.15*Trip purpose factors were provided by respective urban authorities (Edmonton, Calgary, Montréal, Québec City) or derived by the consultant from their model (Winnipeg) or travel survey (Vancouver, Toronto, Hamilton, Ottawa-Gatineau). Table 1 also shows the breakdown of trips between work / work-related trips and non-work trips. The proportions of both work / work-related trips and non-work trips are not constant from one urban area to another, and this is serious problem raised by the study. A more thorough analysis would be needed to clarify the problem, but it can be related to the following facts:

12 Transport Canada. 1993. Value of Passenger Time Savings. Report TP11788.

- 12 -

•

•

•

the treatment and definition work-related trips differ from one region to another;

data were sometimes available only for a 24-hour period rather than for the peak period (during which we find most of the home-to-work and work-to-home trips associated with work-related trips); and

some estimates were available only for all modes (rather than just the automobile alone, although the automobile-based trips accounted for auto-persons [driver and passenger]).

Fuel Costs (Fuel Wasted Due to Congested Conditions)

Due to a lack of data suitable for estimating the amount of fuel wasted by congestion, the study applied a simplified fuel calculation without distinguishing between vehicle types: i.e. all personal vehicles in the model were treated as a single class of light-duty gasoline-powered vehicles. Variations in fuel consumption by average speed and by facility type (expressways and arterials) were taken into account.

The method used in the study was based on the Virginia Tech microscopic energy and emission model (VT-Micro) and the Comprehensive Modal Emissions Model (CMEM). The VT-Micro model is a microscopic fuel consumption and emission model developed at Virginia Polytechnic Institute. CMEM is a modal emission model and vehicle emission data set developed at University of California Riverside.

Although network micro-simulation models provide more precise estimates of fuel consumption and emissions, such models generally are available only for selected corridors in a given urban area; and some urban areas do not yet have these models in any event. Accordingly, the methods used for this study are approximations.

Fuel consumption rates vary according to the vehicle fleet. As uniform data on the composition of vehicle fleets do not exist for the nine Canadian urban areas, the study used the rates shown in Table 2 for a ‘representative’ vehicle fleet in California. This representative fleet includes high-emitters, which was considered to be an acceptable compromise for the purposes of this research.

The approximate average speeds shown in Table 2 represent the boundaries of a speed range (e.g. the value for Arterial Level of Service (LoS) A represents fuel rates at speeds of 45 km/h or higher, the value for Freeway LoS D represents 85 km/h to 94 km/h, and the value for Arterial LoS E represents 20 km/h or less). The data presented in Table 2 are consistent with findings elsewhere that optimal fuel consumption on freeways (expressways) is achieved between 85 km/h and 105 km/h and that fuel consumption tends to increase as average speeds drop (i.e., they increase with increased congestion).

- 13 -

Table 2. Fuel Consumption Rates by Speed Category (Light Duty Gasoline Vehicles – Representative California Fleet, including High-Emitters)

Driving Cycle Approximate Average Speed (km/h)

Fuel (ml/veh-km)

Arterial LOS A-B 45 81.98 Arterial LOS C-D 30 95.92 Arterial LOS E-F 20 141.59 Freeway High-Speed 105 65.22 Freeway LOS A-C 95 63.79 Freeway LOS D 85 62.81 Freeway LOS E 50 71.84 Freeway LOS F 30 98.28 Freeway LOS G 20 113.13 Source: Hellinga, B. and Chan, T. 2002. Issues Related to Quantifying the Environmental Impacts of Transportation Strategies Using GPS Data. Paper presented at the Canadian Institute of Transportation Engineers Annual Conference, Ottawa. May 2002.

The study’s estimates of the cost of fuel wasted took into account the following:

•

•

•

unit prices (fuel prices) were obtained for each urban area (because fuel costs vary from one region to another in Canada);

diesel and gasoline unit prices were distinguished (because diesel and gasoline prices differ); and

taxes were excluded from the fuel prices (because Transport Canada was interested in the social costs of congestion, and not in private costs).

The study used average prices for all unleaded fuels and did not differentiate between gasoline prices by octane level.

Table 3 lists the gasoline unit price used for each urban area according to the base year of the traffic model (again reflecting the assumed depiction of gasoline vehicles in the models).

The total value of excess fuel consumption due to congestion was calculated by multiplying the total quantity of estimated fuel consumption by its unit value, without taxes.

- 14 -

Table 3. Regular Unleaded Gasoline Prices for Each Urban Area

Urban area Base year

Regular Unleaded Gasoline – excluding taxes

Regular Unleaded Gasoline – excluding taxes (2002 $)13

Vancouver 1996 30.41 ¢/litre 38.65 ¢/litre

Edmonton 2000 40.74 ¢/litre 41.22 ¢/litre

Calgary 2001 41.64 ¢/litre 42.09 ¢/litre

Winnipeg 1992 27.95 ¢/litre 37.95 ¢/litre

Toronto 2001 37.72 ¢/litre 38.20 ¢/litre

Hamilton 2001 36.33 ¢/litre 36.30 ¢/litre

Ottawa-Gatineau 1995 25.91 ¢/litre 37.30 ¢/litre

Montréal 1998 21.86 ¢/litre 35.41 ¢/litre

Québec City 1996 27.74 ¢/litre 37.49 ¢/litre Source: M.J. Ervin & Associates

Greenhouse Gas Emission Costs (Emissions Due to Congested Conditions)

The Cost of Urban Congestion in Canada study estimated only the costs of GHG emissions. The study did not estimate the costs of other air pollutants.

GHG emissions were calculated as a direct function of the wasted gasoline fuel that was consumed under congested conditions.

The cost analysis necessitated the application of a monetary value for CO2 and other equivalent GHG emissions determined from a review of the literature. Nonetheless, monetary values for GHG emissions are difficult to assess in the absence of a market for CO2 emission reduction. Some estimates, especially those based on the potential impact of climate change and related environmental damage are too uncertain for broad application. Eventually, the implementation of the Kyoto agreement may facilitate the establishment of a real international market for these gases and, in turn, more reliable monetary values.

In the meantime, to cope with the range of air emission values, some sources propose adopting a median value. The median, unlike the mean, is not affected by extreme values. For example, Kevin Bell (1994) compiled pollutant values from 37 research studies and from governmental agencies in the United States. These values, summarized in Table 4, were used in the Montréal congestion studies (Section 3 and Section 5). The current study translated Bell’s values were into Canadian dollars and updated the values to reflect 2002 prices.

13 Most recent common year available.

- 15 -

Table 4. Air Emission Median Values

CO2 CO HC NOx SOx PM

1990 US$/short ton $20 $907 $3,300 $4,209 $1,793 $2,496

1990 C$/short ton $23.34 $1,058.26 $3,850.35 $4,910.95 $2,092.03 $2,912.27

1998 C$/short ton $27.19 $1,232.88 $4,485.66 $5,721.26 $2,437.21 $3,392.79

1998 C$/ton $29.97 $1,359.00 $4,944.54 $6,306.54 $2,686.54 $3,739.87Adapted from Convergence Research, 1994 in Litman, 1995. The U.S. dollar was worth C$1.16667 in 1990 and the 1990 Canadian dollar was worth C$1.165 in 1998. A metric tonne is equivalent to 1.1023 short tonnes.

9. Congestion Indicators Used in the Study Congestion measurement is complex and multi-dimensional. Accordingly, the study required a range of indicators that together would provide a broad and meaningful means of measuring and expressing the level of congestion.

After a literature review of potential congestion indicators, four indicators of auto congestion were selected. The selection was based on relevance to the Canadian situation, data availability, data quality, practicality and replicability. It was also important that the indicators could be applied immediately by transportation authorities in their analyses.

The four congestion indicators were:

• Travel delay;

• Wasted fuel;

• Roadway congestion is an indicator developed by the TTI which measures the relative importance – i.e., pervasiveness - of links with high volumes: that is, links on which daily traffic exceeds 14,000 vehicles/lane on expressways and 5,500 vehicles/lane on arterials; and

• Travel rate is an indicator which measures the additional time required to travel on a road link during a peak hour, relative to non-congested conditions: it is related to travel delay but is expressed as a ratio (for example, a ratio of 1.20 indicates that it takes 20% more time to make the trip during the peak hour than it would at free-flow speeds).

It is important to note that costs were applied only to the first two (travel delay and wasted fuel from which, in turn, GHG emissions were estimated). Costs are not calculated for the roadway congestion and travel rate indicators: these are developed only to enhance the multi-dimensional profile of congestion.

An additional indicator was developed for transit congestion (that is, the travel delay imposed on bus and rail transit that operates in mixed traffic). The need to develop an indicator for transit congestion was identified after the study began through the technical consultation meetings held with urban and provincial authorities. However, although the transit indicator addresses this need, the study found that its applicability was limited due to a lack of suitable data. For example, the extent to which transit priority measures

- 16 -

(which could mitigate congestion suffered by transit vehicles) are taken into account in the models was not clear. Additional research is required.

10. Main Findings on the Costs of Congestion in Urban Canada Total Congestion Costs

Table 5 summarizes the total costs of congestion (in 2002 dollars). The total costs shown are the sum of the time value of delay, the costs of wasted fuel and the costs of the resultant GHG emissions.

Table 5. Total Costs of Congestion (2002 $ million)

Urban Area Base Year at 50%

Threshold at 60%

Threshold at 70%

Threshold

Vancouver 2003 $402.8 $516.8 $628.7

Edmonton 2000 $49.4 $62.1 $74.1

Calgary 2001 $94.6 $112.4 $121.4

Winnipeg 1992 $48.4 $77.2 $104.0

Hamilton 2001 $6.6 $11.3 $16.9

Toronto 2001 $889.6 $1,267.3 $1,631.7

Ottawa-Gatineau 1995 $39.6 $61.5 $88.6

Montréal 1998 $701.9 $854.0 $986.9

Québec City 2001 $37.5 $52.3 $68.4

Total, all available urban areas $2,270.2 $3,015.0 $3,720.6 The total annual cost of congestion (in 2002 dollars) for the nine urban areas ranges from about $2.3 billion (50 percent threshold) to about $3.7 billion (70 percent threshold). For an urban area the size of the Greater Toronto Area, the total cost of congestion for traffic levels in 2001 was estimated to range from about $890 million (50 percent threshold) to about $1.6 billion (70 percent threshold). For an urban area the size of Edmonton, the total cost of congestion for traffic levels in 2000 was estimated to range from about $49 million (50 percent threshold) to about $74 million (70 percent threshold).

Congestion Costs Attributable to Delay

Table 6 lists congestion costs attributable to delay (in 2002 dollars), expressed as a percent. Equating the proportions to the costs shown in Table 5 yields a range for the costs of delay of between about $2.0 billion (50 percent threshold) and about $3.4 billion (70 percent threshold). The table does not include Calgary and Edmonton because these cities were able to provide data only for delay, and not for wasted fuel or GHG emissions.

- 17 -

Table 6. Proportion of Congestion Costs Attributable to Delay (2002 $ millions)

Urban Area Year at 50%

Threshold at 60%

Threshold at 70%

Threshold

Vancouver 2003 92.4% 91.9% 92.7%

Edmonton 2000 -- -- --

Calgary 2001 -- -- --

Winnipeg 1992 88.1% 88.0% 90.2%

Hamilton 2001 79.4% 85.4% 90.1%

Toronto 2001 87.4% 90.4% 92.4%

Ottawa-Gatineau 1995 84.4% 87.0% 90.4%

Montréal 1998 92.3% 93.1% 93.9%

Québec City 2001 87.4% 87.0% 88.5%

Total, all available urban areas 90.6% 91.8% 93.0%

Table 6 shows that the time value of delay represents by far the largest component of the total cost of congestion, usually 90 percent or more. The percentages shown in Table 6 are similar to the findings of the MTQ studies that used different methods to estimate congestion costs and the percentage of the total attributable to the time value of delay.

Cost of Wasted Fuel due to Congestion

Table 7 presents the estimates for the total amount (litres) and cost of wasted fuel due to congestion14.

The total annual cost of wasted fuel due to congestion for the seven urban areas for which data are available ranges from about $176 million (50 percent threshold) to about $213 million (70 percent threshold).

14 Results are not available for Edmonton and Calgary as these cities were unable to provide specific figures regarding wasted fuel and greenhouse gas emissions.

- 18 -

Table 7. Annual Costs of Wasted Fuel (2002 $ millions)

at 50% Threshold

at 60% Threshold

at 70% Threshold

Urban Area) Litres,

millions $

millions Litres,

millions $

millions Litres,

millions $

millions

Vancouver 65.4 $25.3 89.0 $34.4 98.3 $38.0

Edmonton -- -- -- -- -- --

Calgary -- -- -- -- -- --

Winnipeg 12.6 $4.8 20.1 $7.6 22.2 $8.4

Hamilton 3.0 $1.1 3.7 $1.4 3.8 $1.4

Toronto 241.3 $92.2 263.8 $100.8 267.8 $102.3

Ottawa-Gatineau 13.5 $5.1 17.6 $6.6 18.7 $7.0

Montréal 124.0 $43.9 135.3 $47.9 138.7 $49.1

Québec City 10.4 $3.9 15.0 $5.6 17.3 $6.5

Total, avail. urban areas 470.2 $176.2 544.4 $204.2 566.8 $212.7* Information not available for Edmonton or Calgary.

The total annual cost of wasted fuel due to congestion accounted for about 7.8 percent of the total cost of congestion (50 percent threshold), about 6.8 percent (60 percent threshold) or about 5.7 percent (70 percent threshold).

Cost of GHG Emissions due to Congestion

Table 8 presents the estimates for the annual amount (tonnes) and cost of GHG emissions due to congestion. As noted, no results are available for Edmonton and Calgary as these cities were unable to provide specific figures regarding GHG emissions.

The values are based on the estimated 1998 median rate of $29.97 per tonne of CO2. To ensure consistency with other costs, the 1998 median rate was inflated to a 2002 value of $32.82 per tonne of CO2

15.

The total annual cost of GHG emissions due to congestion for the seven urban areas for which data are available ranges from about $38 million (50 percent threshold) to about $46 million (70 percent threshold).

15 This estimation is based on a 9.5 percent inflation rate for the 4 year period. See Statistics Canada, Consumer Price Index, Historical Summary ,62-001-XPB and 62-010-XIB.

- 19 -

Table 8. Annual Costs of Greenhouse Gas Emissions from Delay (2002 $ millions)

at 50% Threshold

at 60% Threshold

at 70% Threshold Urban Area

Tonnes $, millions Tonnes $, millions Tonnes $, millions

Vancouver 161,521 $5.3 219,713 $7.2 242,791 $8.0

Edmonton * -- -- -- -- -- --

Calgary * -- -- -- -- -- --

Winnipeg 31,003 $1.0 49,569 $1.6 54,686 $1.8

Hamilton 7,512 $0.2 9,196 $0.3 9,271 $0.3

Toronto 595,709 $19.6 651,318 $21.4 661,226 $21.7

Ottawa-Gatineau 33,447 $1.1 43,385 $1.4 46,107 $1.5

Montréal 306,100 $10.0 334,100 $11.0 342,500 $11.2

Québec City 25,600 $0.8 36,900 $1.2 42,700 $1.4

Total, avail. urban areas 1,160,800 $38.1 1,344,200 $44.1 1,399,300 $45.9* Information not available for Edmonton or Calgary.

The total annual cost of GHG emissions due to congestion accounted for about 1.7 percent of the total cost of congestion (50 percent threshold), about 1.5 percent (60 percent threshold) or about 1.2 percent (70 percent threshold).

Comparison with Montréal and Vancouver studies

There is little information with which to compare these costs. Data are either not available or methodological differences limit the comparisons that can be made between this study and earlier studies.

The aforementioned MTQ work for Montréal is an exception. The methodological similarity between this study and the MTQ study allows comparison. This study’s total congestion costs of $854 million (in 2002 dollars) for 1998 traffic conditions (60 percent threshold) are approximately 7 percent higher than the $779 million calculated by the MTQ study for 1998 traffic conditions at the 60 percent threshold. The difference can be attributed to differences in the way inflation was handled, the use of a slightly different methodology to distinguish between work and non-work trips, differences in the methods used for calculating fuel consumption,16 and the MTQ study’s inclusion of vehicle operating costs and the costs of pollutant emissions.

If the costs of the time lost in delay are compared, the results of the two studies are very similar. This study estimated Montréal’s time lost in delay costs for 1998 traffic

16 The study conducted by Transports Québec estimated 60.9 million litres of fuel at the 60%

threshold – less than half the 135.3 million litres estimated in this research at the same threshold. This variation is due to methodological differences between the two studies and to the fact that the Montreal vehicle fleet has a much higher proportion of compacts than the California fleet used in the study under consideration here.

- 20 -

conditions to be $795 million (2002 dollars). The corresponding MTQ result was $704 million (1998 dollars). The difference of 13 percent is largely attributable to time unit values. Consequently, the results of the current research appear reasonable.

The results can also be compared with Transport Canada’s 1996 Greater Vancouver estimate (see Section 3), which estimated total congestion costs for auto travellers to be about $0.7 billion. The total congestion costs of about $403 million to about $629 million for Vancouver (under 2003 traffic conditions) estimated by this study are lower the estimates made in 1996. The differences may be due to methodological differences (free-flow was used as a threshold in the BC study and the unit time values also were different).

Comparison with American studies

It was beyond the scope of this study to compare the congestion estimates obtained in this research for Canadian urban areas with congestion estimates available for American urban areas. It is difficult to make such comparisons, partly because the methodological approaches differ.

The MTQ study did, however, compare the results for Montréal with the results obtained for a sample of similarly sized urban areas in the U.S. (cities with populations between 2 and 4 million, such as Boston, Dallas, St. Louis and Seattle). The MTQ study found that Montréal’s congestion levels were lower than those of the American cities and explained the difference in terms of the higher urban population densities, higher rates of public transit use and lower car ownership rates in Montréal.

11. Recommendations for Future Research

The Cost of Urban Congestion in Canada study made a number of recommendations for possible future research to improve estimates of traffic congestion in Canada.

Critical considerations in congestion estimation include a lack of data, a lack of common definitions, and considerable variation among models in terms of the models’ structure and coverage. These problems made it impossible, for example, to undertake a simple comparison of the nine urban areas.

Several of the study’s recommendations were directed at urban and provincial authorities. These recommendations are designed to improve data availability and quality and to improve the models required for the estimation of congestion costs:

• Urban and provincial authorities should develop common definitions for network link categorizations, should be able to model traffic beyond the peak period, and should incorporate freight and commercial vehicle traffic into the models;

• In the short term, urban and provincial authorities should be encouraged to conduct regular origin-destination surveys and to expand the surveys’ vehicle coverage to include freight and commercial vehicle traffic; and

• In the longer term, urban and provincial authorities would benefit from collecting ‘real-time’ data on speed and volumes through the use of geographical positioning

- 21 -

system technologies. Authorities would also benefit from improving network micro-simulation models17.

The study made two additional recommendations for future research:

• Non-recurrent congestion costs are also critical to the development of a comprehensive picture of urban congestion in Canada. A better understanding of the magnitude and importance of non-recurrent congestion costs requires the collection of traffic data from major incidents. The measurement of non-recurrent congestion requires the availability of real-time speed-delay data that cover in each urban area a sufficiently representative number of roads, origins/destinations and time periods; as well as the availability of passenger vehicle fleets that are equipped with GPS (e.g., taxis); and

• Congestion costs data obtained from the engineering approach should be complemented with congestion costs data obtained from an economic approach. The insights obtained from approaching congestion costs from two points of view will increase our understanding of congestion costs and our understanding of the most appropriate steps that could be taken to address traffic congestion.

12. Conclusions The Cost of Urban Congestion in Canada study provides the first systematic analysis of urban congestion across Canada. In this respect, it represents a major contribution to our understanding of urban congestion in Canada. The study was designed to achieve a better understanding of the nature and extent of congestion in Canada, and to develop a consistent approach to estimating congestion costs. Only recurrent congestion costs were considered. Recurrent congestion is the congestion associated with the regular build-up of traffic during the morning and afternoon commuter peak periods. The study used data from Canada’s nine largest urban areas: Québec City, Montréal, Ottawa-Gatineau, Toronto, Hamilton, Winnipeg, Calgary, Edmonton and Vancouver. Data were limited to expressways and arterials, to passenger vehicles (autos) and to peak-period congestion.

The study found that urban recurrent congestion costs Canada’s nine largest urban areas between $2.3 billion and $3.7 billion in 2002 dollar values. More than 90 percent of these costs are due to time lost in traffic by drivers and passengers; 7-8 percent is attributable to increased fuel consumption; and 2-3 percent is from increased GHG emissions.

The estimates of congestion costs are conservative. They exclude:

• vehicle operating costs (other than fuel);

• the costs associated with increased air pollutant emissions;

• costs associated with the increase in noise nuisance; and

• congestion costs borne by the freight transportation sector;

17 Network micro-simulation models are computer models where the movements of individual vehicles travelling around road networks are represented. These models, presumably, provide a better, and 'purer', representation of actual driver behaviour and network performance. They are becoming increasingly popular for the evaluation and development of road traffic management and control systems.

- 22 -

• the costs of off-peak congestion which is known to occur in many urban areas;

• the congestion costs of non-recurrent congestion (i.e. congestion caused by random events, such as bad weather, accidents, stalled vehicles and other incidents);

As each urban area has a different definition of when congestion begins and ends and as each urban area collects different data using different methods, it is difficult to make accurate comparisons among the urban areas. The findings are influenced by a lack of data and by variation among the models available in the nine urban areas. Accordingly, the study recommends that urban and provincial authorities should collect more and better congestion-related data using common congestion indicators and that these authorities should develop common modelling approaches for implementing congestion measurements.

Over time, the collection of congestion indicators will be very useful in tracking congestion trends. The importance of this is underlined by Canada’s expected continued growth in population, urbanization, gross domestic product, car ownership and car use. The Cost of Urban Congestion in Canada study is an important first step. The study created a benchmark for future congestion costs studies and suggested a method that Canadian municipalities could use to develop their own congestion estimates.

As for future congestion cost studies, it is useful to note that several factors influence recurrent road congestion; notably, the rapid growth in population, urbanization, and car ownership and use. In the decade from 1993 to 2003, Canada’s population grew by 2.7 million, or 9.3 percent. The urbanized proportion of the population grew from 77 percent to 80 percent. Growth was faster in the largest metropolitan areas, with cities of over 100,000 population experiencing growth of nearly 16 percent during the decade. Growth was concentrated in four large urban regions: the extended Golden Horseshoe,18 Montréal and adjacent region, the Lower Mainland and southern Vancouver Island, and the Calgary-Edmonton corridor. By 2003, about 65 percent of Canada’s population, or about 20 million people, lived in the nation’s 27 Census Metropolitan Areas.

Growth was particularly rapid in the metropolitan areas of Vancouver, at 23 percent, and Calgary, at 27 percent. By far the largest absolute increase in population took place in the metropolitan area of Toronto, where population grew by over 900,000, or 21.5 percent, in the decade from 1993 to 2003.19

The number of motor vehicles in use has increased faster than the total population: the number of cars and trucks (combined) rose by 13 percent from 1993 to 2003. Statistics reporting trends for total national vehicle use suggest that total vehicle-kilometres grew at least as fast as the growth in the number of vehicles.

18 This region consists of the urban centres of Oshawa, Toronto, Hamilton and St. Catharines-Niagara, plus Kitchener, Guelph and Barrie. It accounted for 59 percent of Ontario's population and 22 percent of the nation's population in 2001. Almost one-half of Canada's total population growth occurred there. 19 Annual Demographic Statistics, Statistics Canada 91-213-XIB2004000. Information on the Census is also available at Statistics Canada website (www.statcan.ca).

- 23 -

Truck traffic has grown even faster than private vehicle traffic under the unprecedented stimuli of deregulation, North American Free Trade and continued changes in logistics management (e.g. “just-in-time” delivery).

Thus, it can be expected that congestion will be influenced by the expected continued growth in population (0.75 percent per annum to 2020) and its increased urbanization (to 80 percent), the even faster anticipated growth rates in gross domestic product (2 percent per year) and – if recent trends continue – car ownership and use.

- 24 -

13. References

BELL, K. (1994), Valuing Emissions from Germiston Generating Project, Convergence Research (Seattle). GOURVIL, L. AND JOUBERT, F. (2004), Évaluation de la congestion routière dans la région de Montréal (“Estimation of roadway congestion in Montréal.”), Ministère des transports du Québec. HELLINGA, B. AND CHAN, T. (2002), Issues Related to Quantifying the Environmental Impacts of Transportation Strategies Using GPS Data, Paper presented at the Canadian Institute of Transportation Engineers Annual Conference, Ottawa, May 2002. KRIGER, D., JOUBERT, F. , BAKER, M. AND JOUBERT, G. (2005). The Costs of Urban Congestion in Canada, study conducted for Transport Canada. KRIGER, D. (2005), Methods to Estimate Non-Recurrent Congestion in Canada, study conducted for Transport Canada. MUNICIPALITY OF METROPOLITAN TORONTO. 1987. Metropolitan Toronto Goods Movement Study: Technical Report. Toronto. THE CANADIAN VEHICLE SURVEY (1999-2005): A survey funded by Transport Canada and undertaken by Statistics Canada, 53F0004XIE. SCHRANK, D. AND LOMAX , T. (2004). The 2004 Urban Mobility Report, Texas Transportation Institute. STATISTICS CANADA, Annual Demographic Statistics, 91-213-XPB. STATISTICS CANADA ,Consumer Price Index, Historical Summary, 62-001-XPB and 62-010-XIB. TRANSPORT CANADA (1993). Value of Passenger Time Savings, Report TP11788. TRANSPORTATION RESEARCH BOARD (2000), Committee on Highway Capacity and Quality of Service, Highway Capacity Manuals. WATERS II, MARCH W.G. (1992), The Value of Time Savings for the Economic Evaluation of Highway Investments in British Columbia. WATERS II, W.G. (2004), Chapter 4: Congestion and the Valuation of Travel Time Savings: Draft. Transport Canada.