Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The car in Canada: a study of factors influencing automobile dependence in Canada’s seven largest cities,… Raad, Tamim 1998

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1998-0583.pdf [ 22.44MB ]
JSON: 831-1.0088500.json
JSON-LD: 831-1.0088500-ld.json
RDF/XML (Pretty): 831-1.0088500-rdf.xml
RDF/JSON: 831-1.0088500-rdf.json
Turtle: 831-1.0088500-turtle.txt
N-Triples: 831-1.0088500-rdf-ntriples.txt
Original Record: 831-1.0088500-source.json
Full Text

Full Text

THE CAR IN CANADA: A STUDY OF FACTORS INFLUENCING AUTOMOBILE DEPENDENCE IN CANADA'S SEVEN LARGEST CITIES, 1961-1991 i by Tamim Raad B.Comm., The University of British Columbia, 1993 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T O F T H E REQUIREMENTS F O R T H E D E G R E E O F M A S T E R O F ARTS (PLANNING) in T H E F A C U L T Y O F G R A D U A T E STUDIES School of Community and Regional Planning We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 1998 © Tamim Raad, 1998 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of f ^ ^ ^ A ^ ^X & ^ ' . ^ ( T U H A . * ^ The University of British Columbia Vancouver, Canada Date DE-6 (2/88) ABSTRACT Automobile dependence is defined as a series of convergent land use and transportation conditions in a city that leave people with few non-car options for urban travel. This dependence is compromising the environmental, social and economic health of cities in Canada. Furthermore, it appears as though automobile dependence is increasing in Canada, as are its attendant impacts. A fuller understanding of the primary relationships affecting this trend is needed if its impacts are to be adequately mitigated. However, there is little quantitative knowledge of the relative importance of factors contributing to automobile dependence in Canadian cities. A review of the literature identifies a multitude of mutually reinforcing factors that contribute to the creation of automobile dependent cities. The factors are both cause and effect and exhibit 'feedback,' which results in a cycle of intensification of the original condition. While there are many feedback relationships that contribute to automobile dependence, some may be stronger than others. Mitigating the many adverse impacts of automobile dependence requires reducing the need for both automobile ownership and automobile use by reversing these feedback relationships. This thesis identifies the relative importance of factors influencing automobile dependence in Canada's major cities through a comparative analysis of transportation, land use and population and employment distribution trends and patterns. This involves the collection and analysis of an extensive set of data from Canada's seven largest cities (Vancouver, Calgary, Edmonton, Winnipeg, Toronto, Ottawa-Hull and Montreal). To provide context and supplementary information, selected data from thirty-four additional global cities are also used. A correlation analysis of the data collected identifies the strength of correlation between factors involved in automobile dependence feedback. The data reveal commonalities between cities: those cities with higher urban densities, higher transit service provision and lower automobile infrastructure provision exhibit lower levels of car ownership and use as well as higher levels of transit use. These cities also have better utilized transit systems, have higher walking and cycling mode shares and consume less fuel. The quantitative findings are used in tandem with the qualitative findings of the literature review to identify and rank eight possible points for policy intervention in changing auto dependence feedback. Of the factors examined, metropolitan and outer area density, transit supply and C B D parking supply appear to exert the strongest relative influence on auto dependence. These are followed in importance by inner area density and car ownership, which are followed by road supply and non-motorized transport share. While the auto dependence factors ranked require further study, clarification and confirmation, they provide a preliminary basis for directing policy analysis. A policy evaluation framework is developed that enables policies prescribed in each intervention area to be assessed against a series of travel, environmental, social and economic impact criteria as well as their implementation potential. This framework can be used by policymakers to identify high leverage policies for reducing auto dependence. i n T A B L E OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iv LIST OF TABLES vii LIST OF FIGURES ix GLOSSARY OF COMMON ACRONYMS AND TERMS xi PREFACE xii ACKNOWLEDGEMENTS xiii CHAPTER 1 - INTRODUCTION 1 1.0 PURPOSE AND CONTEXT 1 7.0.0 Why Compare Transportation in Canada? 3 1.1 OTHER KEY STUDIES 3 1.1.0 Some Criticisms of CAAD 6 1.2 THESIS SCOPE 9 1.3 THESIS OBJECTIVES AND STRUCTURE 10 CHAPTER 2 - AUTO DEPENDENCE: DEFINITIONS, CAUSES, IMPACTS AND FUTURE DIRECTIONS 12 2.0 INTRODUCTION 12 2.1 PERSPECTIVES ON AUTOMOBILE DEPENDENCE 12 2.1.0 Redefining the Goal of Transportation 15 2.1.1 Measuring Auto Dependence 17 2.2 CAUSES OF AUTOMOBILE DEPENDENCE 18 2.2.0 Feedback in Transportation 23 2.2.1 Common Property Problems 25 2.3 SYMPTOMS AND EFFECTS OF AUTOMOBILE DEPENDENCE 29 2.3.0 Ecological Impacts of Auto Dependence 30 Water related impacts 30 Land related impacts 33 Resource consumption 36 Vehicle disposal 40 Airborne vehicular pollutants and emissions 42 Other environmental impacts 45 2.3.1 Social Impacts of Auto Dependence 45 Health ...45 Equity 47 Decaying urban fabric 50 Isolation 51 Dysfunctional social behaviors 52 2.3.2 Economic Impacts of Auto Dependence 52 Total costing 53 Incremental costing 56 Are the benefits worth the cost? 60 iv Summary: ecological, social and economic impacts 62 2.4 R E D U C I N G A U T O D E P E N D E N C E 63 2.4.0 The Quasi-Solution - Seeing the Trees, Missing the Forest 64 2.5 HOLISTIC DIRECTIONS 68 CHAPTER 3 - METHODOLOGY 69 3.0 INTRODUCTION 69 3.1 ORIGINS O F T H E R E S E A R C H 69 3.1.0 Rationale for Cities, Study Years and Data Items Selected 71 3.2 D A T A C O L L E C T I O N P R O C E S S 72 3.3 DEFINITIONAL ISSUES 74 3.3.0 Raw Data Definitions and Issues 74 Territorial areas 74 Population 79 Employment 79 Urbanized area 80 Parking supply 81 Road supply 82 Vehicles on register 82 Private transportation 83 Journey to work data 85 Public transport indicators 85 3.3.7 Standardized Data 87 3.3.2 Use of Data and Data Reliability. 88 3.4 M E T H O D S O F ANALYSIS 91 3.5 C O N C L U S I O N S 91 CHAPTER 4 - FINDINGS: TREND, PATTERN AND CORRELATION ANALYSIS 93 4.0 INTRODUCTION 93 4.1 O V E R V I E W O F T H E T R E N D S 1961-1991 93 4.1.0 Urban Growth Trends - Inner City Decline and Sprawl 93 4.1.1 Transportation Trends - Increasing Car Dominance and Transit Decline 97 4.2 A C L O S E R L O O K A T T H E CITIES 103 4.2.0 Urban Growth Trends 103 4.2.1 Transportation Trends 110 Car ownership and use 110 Auto infrastructure: roads and parking 113 Parking and transit 116 Energy use 117 Transit service, demand and utilization 118 The cities summarized 123 4.3 D A T A C O R R E L A T I O N ANALYSIS 125 4.3.0 Correlations of Key Factors 125 Using the correlation charts 125 Giving meaning to the correlations 127 4.3.1 Relative Strength of Factors Influencing Auto Dependence 730 Urban density 131 Transit service 135 Parking supply, transit and cars 137 Other factors of importance 138 4.4 C O N C L U S I O N S 139 CHAPTER 5 - POLICY FRAMEWORK DEVELOPMENT 142 5.0 INTRODUCTION 142 5.1 POLICY INTERVENTION: INTRODUCING VIRTUOUS F E E D B A C K 142 5.2 DECISION MAKING F R A M E W O R K 144 5.2.0 Reconciling Quantitative Findings with Criteria and Potential Policy 746 5.3 S O M E DYNAMICS O F POLICY EVALUATION 149 5.4 C O N C L U S I O N - E F F E C T I V E N E S S O F T H E F R A M E W O R K 150 CHAPTER 6 - CONCLUSIONS 153 6.0 S U M M A R Y 153 6.0.0 Auto Dependence in a nutshell 753 6.0.1 Data: The Good, the Bad and the Ugly 756 6.0.2 Findings: Crunching the Numbers 757 6.0.3 Policy Directions: From Reductionism Least a Bit More Holistic 760 6.1 A R E A S F O R F U R T H E R R E S E A R C H 161 6.2 C L O S I N G R E M A R K S 162 REFERENCES CITED 164 APPENDIX 1 - MAPS AND TERRITORIAL BOUNDARIES 178 APPENDIX 2 - CANADIAN CITIES RAW AND STANDARDIZED DATA 187 APPENDIX 3 - WORLD CITIES DATA USED FOR SPSS RUNS 230 APPENDIX 4 - CORRELATION RESULTS, CANADIAN AND WORLD CITIES 234 APPENDIX 5 - DATA SOURCES 247 vi LIST OF TABLES Table 1 - Classi f icat ion of cit ies by degree of automobi le dependence 13 Tab le 2 - Compar i son of relative price changes and transit r idership in Canada , 1980-95 20 Tab le 3 - Car costs and car use in the U.S. Pacific Nor thwest and British Co lumbia 21 Table 4 - Sources of wa te r pollution and hydrological d isrupt ions due to auto-re lated activity 31 Table 5 - S u m m a r y of road runoff const i tuents and their pr imary sources 32 Tab le 6 - Compos i t ion of the average automobi le (1990) 37 Tab le 7 - Env i ronmenta l impacts of minerals extract ion 38 Tab le 8 - Energy use of t ransportat ion related activit ies in the U.S., 1985 39 Tab le 9 - Emiss ions f rom transportat ion in the G V R D (% of total, 1991) 43 Tab le 10 - Air pollut ion and emiss ions f rom urban t ransportat ion 44 Tab le 11 - Subsid ies to t ransport in the BC Lower Main land, 1991 48 Tab le 12 - Au to ownersh ip in Ontar io, by household income, 1993 49 Tab le 13 - Motor vehic le (MV) cost categor ies (Delucchi) 55 Tab le 14 - S u m m a r y of the annual ized social costs of motor vehicle use, 1990 56 Table 15 - S tages in the research program 69 Table 16 - Global cit ies included in C A A D II 70 Tab le 17 - Researchers ' contr ibut ion to the data col lect ion process 70 Tab le 18 - Data source agenc ies 73 Tab le 19 - Canad ian city metropol i tan area definit ions (1991) 76 Tab le 20 - Canad ian inner area definit ions (1991) 78 Tab le 21 - Canad ian central business district definit ions (1991) 79 Tab le 22 - Compar i son of urban density measures 81 Tab le 23 - Data quali ty and reliability 90 Tab le 24 - Land use and private and public t ransportat ion indicators in Wor ld regions, 1990/91 102 v i i Table 25 - Change in average car ownership and car use in World cities, by region, 1981-91 103 Table 26 - Effect of density changes on developed areas in three cities between 1981 and 1991 ... 108 Table 27 - Increases in VKT/capita, population and V K T between 1981 and 1991 111 Table 28 - Outer area V K T and other transport variables in 3 Canadian and 3 U.S. cities, 1991 112 Table 29 - Change in transit service levels (VKT/capita) in seven Canadian cities, 1961-91 120 Table 30 - Change in transit demand (boardings/capita) in seven Canadian cities, 1961-91 120 Table 31 - Change in transit utilization (boardings/VKT) in seven Canadian cities, 1961-91 120 Table 32 - Land use and transit service in 7 Canadian cities, 1991 122 Table 33 - Land use and transportation characteristics of selected World cities, 1990/91 124 Table 34 - A sample of correlations between transport, land use and demographic factors in 7 Canadian cities 128 Table 35 - Sample of correlations between transport, land use and demographic factors in 41 World Cities 129 Table 36 - Influence of transit service and density on transit use in 15 World cities, 1990/91 136 Table 37 - Criteria for evaluating transportation policy 146 Table 38 - Decision making framework for potential policies impacting the variables studied 148 Table 39 - Impacts of auto dependence summarized 155 LIST OF FIGURES Figure 1 - Degrees of car dependence 14 Figure 2 - Maximizing exchange through mode choice 17 Figure 3 - Positive feedback relationships in automobile dependence 19 Figure 4 - The prisoner's dilemma 27 Figure 5 - Ecological, social and economic dimensions of auto dependence 29 Figure 6 - Motor vehicle cost categories (Litman) 57 Figure 7 - Perceived full price of travel 58 Figure 8 - The pricing problem - an economist's perspective 59 Figure 9 - Road congestion: the "capacity myth" 67 Figure 10 - Inner and outer area population in the Canadian cities, 1961-91 94 Figure 11 - Inner area share of regional population and jobs in the Canadian cities, 1961-91 95 Figure 12 - Growth in developed land vs. population in the Canadian cities, 1961-91 95 Figure 13 - Average urban densities in 46 World cities, 1990/91 96 Figure 14 - Growth in car registrations versus population in the Canadian cities, 1961-91 97 Figure 15 - Average modal split in the Canadian cities, 1961-91 98 Figure 16 - Growth in C B D parking supply versus C B D jobs in four Canadian cities, 1961-91 99 Figure 17 - Growth in car ownership, car use and transit use versus population in the average Canadian city, 1961-91 100 Figure 18 - Average vehicle occupancies in the Canadian cities, 1961-91 101 Figure 19 - Average transit share of motorized travel in 46 World cities, by region 1990/91 101 Figure 20 - Change in car ownership and use in the World cities, by region, 1981-91 102 Figure 21 - Inner and metro area populations in 7 Canadian cities, 1961-91 104 Figure 22 - Proportion of regional population living in Canadian inner areas, 1961-91 105 Figure 23 - Metro, C B D , inner and outer area densities in the Canadian cities, 1961-91 106 Figure 24 - Metro densities in the Canadian cities, 1981-1991 107 Figure 25 - Urbanized areas in Calgary, Ottawa and Winnipeg, 1961-91 108 Figure 26 - Metro density versus developed area in the G V R D , 1961-91 109 Figure 27 - Inner and outer area employment trends in the Canadian cities, 1961-91 109 Figure 28 - Change in car ownership and use in the Canadian cities, 1961-91 111 Figure 29 - C B D parking supply and employment in the Canadian cities, 1961-91 114 Figure 30 - Proportion of regional employees working in CBDs in Canadian cities, 1991 115 Figure 31 - Parking supply and modal split in Calgary, 1971-1991 116 Figure 32 - Gasoline use in seven Canadian cities and the U.S., 1991 117 Figure 33 - Transit demand and supply in 7 Canadian cities, 1961-91 118 ix Figure 34 - Transit share of motorized travel 121 Figure 35 - Correlations of factors involved in auto dependence feedback in the Canadian cities 130 Figure 36 - Correlations of factors involved in auto dependence feedback in the World cities 131 Figure 37 - Density and car use in the Canadian and U.S. cities, 1990/91 132 Figure 38 - Density and transit use in the Canadian and U.S. cities, 1990/91 133 Figure 39 - Density and transit use in Europe, Australia, Canada and the U.S., 1990/91 135 Figure 40 - C B D parking supply and C B D mode split in Canadian cities, 1991 137 Figure 41 - Possible policy intervention points in creating "virtuous" feedback 143 Figure 42 - Factors most strongly correlated with automobile dependence 159 GLOSSARY OF COMMON ACRONYMS AND TERMS ALC (see BCALC) ALR Agricultural Land Reserve ATED Access to Exchange Disadvantaged. A term coined by David Engwicht (1993) to describe those who do not have access to an automobile, where auto dominance precludes other travel options for them. BC Lower British Columbia Lower Mainland. Includes the GVRD and the Fraser, Dewdney-Alouette, Mainland Fraser Cheam, Squamish Lillooet and Sunshine Coast regional districts. BCALC British Columbia Agricultural Land Commission BCTFA British Columbia Transportation Financing Authority CAAD Cities and Automobile Dependence: An International Sourcebook (see bibliographic reference to Newman and Kenworthy, 1989). CBD Central Business District Commutershed The geographical range from which a metropolitan area draws substantial commuter traffic. CUTA Canadian Urban Transit Association GTA Greater Toronto Area. Includes the five regional municipalities of York, Durham, Peel, Halton and Metro Toronto. GVRD Greater Vancouver Regional District ha Hectares (1 ha=2.47 acres) HOV High Occupancy Vehicle ISTP The Institute for Science and Technology Policy at Murdoch University in Perth, Western Australia LOS Level of service. The degree to which traffic flows without interruptions, measured by average traffic speeds. LRT Light Rail Transit Metro Metropolitan Toronto (City of Toronto post-1998) MOTH British Columbia Ministry of Transportation and Highways MUC Montreal Urban Community NMT Non-motorized transportation (typically referring to walking and cycling, but includes other non-motorized modes). OGTA Office of the Greater Toronto Area. Provincial agency acting as a coordinating body for municipalities and regional municipalities in the GTA. Quad Quadrillion Btu (10 1 5 Btu) RMOC Regional Municipality of Ottawa-Carleton SOV Single Occupant Vehicle SOV Single Occupancy Vehicle TDM Transportation Demand Management TTC Toronto Transit Commission UniCity City of Winnipeg (metropolitan government) VKT or VKmT Vehicle Kilometres Travelled VMT Vehicle Miles Travelled VPD Vehicles Per Day xi P R E F A C E In June 1996, I had the privilege of being invited to work with Jeff Kenworthy and Peter Newman of the Institute of Science and Technology Policy (ISTP) at Murdoch University in Perth, Australia on the update of Cities and Automobile Dependence: An International Sourcebook (Newman and Kenworthy 1989a). Between June 1996 and February 1997,1 worked with a team of researchers at ISTP to help update Cities and Automobile Dependence and expand the number of cities surveyed from 32 to 46. I specifically worked with Jeff Kenworthy and Felix Laube on the collection of comparative urban data for the seven Canadian cities in the update, six of which were new additions to the study. The comparative data that appear in this thesis are drawn from this research, by permission of ISTP. The full data set, including all the international cities, will be published in the Spring of 1999 under the title An International Sourcebook of Automobile Dependence in Cities. 1960-1990 by Jeff Kenworthy and Felix Laube, with Peter Newman, Paul Barter, Tamim Raad, Chamlong Poboon and Benedicto Guia Jr. Any subsequent use of the raw or standardized data items in this thesis (Appendices 2 and 3) should cite Kenworthy and Laube et al (1999) as the data source. Where specific work has been done extending the analyses or data beyond that which is contained in the above book, this thesis should be cited as the source. xii ACKNOWLEDGEMENTS I would first of all like to thank Peter Newman, Director of the Institute of Science and Technology Policy (ISTP), who made this research possible by inviting me to spend 8 months in Perth working on C A A D II. I would also like to thank everyone of what is know as the "ISTP family" for welcoming and supporting me while I was in Perth. Thanks to Felix Laube and Paul Barter for all the ideas and friendship and to William Ross and Barrie the Cat for the hospitality, laughs and cool digs. Most of all, a special thanks to Jeff Kenworthy for all the help, encouragement and mentorship over the past two years. ISTP not only manages to produce innovative research, but also a warm and inviting community. My thesis committee has also been a tremendous help. Thanks to Professor Peter Boothroyd of the School of Community and Regional Planning (SCARP) at UBC for his clarity, his demonstrated commitment to teaching and for constantly challenging my thinking over the past five years. Thanks to Clive Rock, the GVRD's transportation guru and Adjunct Professor at SCARP for his insights and witticisms and for fielding my relentless call. Thanks also to Mark Roseland, Director of Simon Fraser University's Community Economic Development Centre for taking the time to be the external examiner and for his continuing commitment to community-based change. There are many others who kindly provided their time and expertise throughout the project including: Brendon Hemily of CUTA, Todd Limtan of the Victoria Transport Policy Institute, Clark Lim of the G V R D , Brian Hollingworth of IBI Group and Wendy Sarkissian of University of Sydney. Thanks to Karim Dossa, statistics and SPSS wizard, who generously offered many free hours of help in making sense out of a big pile of numbers, even though he didn't know me. Thanks also to Terry Sunderland and Carmen Mills of Emerald City Communications for helping me make my Macintosh map files look pretty. Many thanks to my friends, the good people at Better Environmentally Sound Transportation and to my Vancouver 'Bikeshevik' cohort (especially, the cycling dinos of Fossils Against Fossil Fuels) for their friendship, support and respite from writing. Special thanks to Gavin Davidson for sharing my perverse obsession with things transportation. Finally, thanks to my mother, father and brothers for their motivation and support (especially Samer, who let me mooch off him for so long) and to Karen Peachey for inspiring me and putting up with my paper sprawl. xiii CHAPTER 1 - INTRODUCTION 1.0 PURPOSE AND CONTEXT The purpose of this thesis is to identify the relative importance of factors that affect automobile dependence in Canadian cities through a comparative analysis of transportation patterns. Although the post-war increase in the ownership and use of automobiles has brought tremendous mobility to Canadians, it has also brought tremendous costs. Over the past five decades, public decision-making has overwhelmingly favoured and facilitated automobile use and sprawling land uses such that few viable alternatives are available to meet travel needs. This "automobile dependence" (Newman and Kenworthy 1989a) imposes considerable social, ecological and economic costs that are of significance at the local, regional and global scales.1 To date, there have been few studies that have provided comprehensive and comparable information about the complex interactions between transportation, land use and related urban development factors that influence automobile dependence at the regional scale in the Canadian urban context. Often, Canadian information on transportation trends and patterns is temporally limited, anecdotal, not comparable between cities and years, or simply not available. Canadians often depend on data from the United States to inform much of our transportation analysis as it is the country with transportation and urban development patterns most closely resembling ours. However, the transportation, land use, political and cultural realities of the two countries are still so demonstrably different that we cannot meaningfully use the American experience as a proxy for our own (for examples of such differences Frisken 1986; Goldberg and Mercer 1986; Kenworthy and Newman 1994b; Linteau 1990; Pucher 1994; Raad and Kenworthy 1998; Schimek 1996). Therefore, reliable standardized information about how the complex interrelationships of automobile dependence are manifest in the Canadian context would be valuable for urban transport policy analysis. In their landmark study, Cities and Automobile Dependence (CAAD), Peter Newman and Jeff Kenworthy analyzed trends in transportation and land use in 32 global cities in 1960, 1970 and 1980 (1989a). Although CAAD took 9 years to compile and publish, it made an invaluable contribution to the debate on automobile dependence in a global context when it was finally released. The study revealed commonalities between cities with lower levels of automobile dependence. The less car-oriented cities displayed higher urban densities, higher per capita provision and use of public transportation and lower per capita ownership of automobiles. While other key studies provided quantitative evidence of some of 1 the key variables influencing car and transit use (most notably, Pushkarev and Zupan 1977), C A A D was the first quantitative study to reveal, on a comprehensive basis, the inextricable link between sprawling land uses and increasing automobile dependence. It also provided a standardized database of urban transportation, land use and demographic data which had not previously been compiled. C A A D provides a compendium of data that is today considered one of the seminal works in the study of urban transportation and land use trends in global cities. Newman and Kenworthy (1989a) note that one of the regrets of the initial study was the inclusion of only one Canadian city. Toronto was the only Canadian city studied and attracted much interest. The data in the 1989 study indicate Toronto's transportation and land use patterns are an anomaly in the North American context. "Toronto seems to sit neatly between the land use and transportation patterns of the automobile-oriented US and Australian cities and the very public transport-oriented European cities. In this way it provides a very useful model for policy development, especially for the automobile cities" (Newman and Kenworthy 1989a, p. 11). The data show that until 1980, Toronto had much higher public transportation patronage, lower levels of car ownership and use and significantly higher urban densities than its automobile-oriented American and Australian counterparts. In their book The Myth of the North American City (1986), Michael Goldberg and John Mercer's more generalized comparison of urban realities in Canadian and American cities support the Newman and Kenworthy findings. Goldberg and Mercer supported the thesis that Canadian cities, in general, did display transportation and urban settlement patterns that are distinct in the North American context. Updated transportation data compiled by Jeff Kenworthy in the early 1990's confirmed that Toronto is indeed unique in its transportation and land use patterns (Kenworthy and Newman 1994b). Other studies have also supported the need for relevant Canadian urban transportation information. In his Master's thesis examining the implications of land use on automobile dependence in Canada, Anthony Parker says of the lack of Canadian data: "there is a particular need for better information, and quantitative data in particular, on the relationship between urban form and design characteristics and automobile use. It would be useful to have more data on travel behaviour.. .and on what factors affect this.. .an extension of Newman and Kenworthy's analysis to include Canadian cities in addition to Toronto would be valuable" (1993, p. 147). This study attempts to fill this gap in the knowledge of transportation patterns in the Canadian context. From May 1996 until February 1997,1 worked as part of a research team led by Jeff Kenworthy that was working on updating and expanding C A A D to An International Sourcebook of Automobile Dependence in Cities. 1960-1990 (Kenworthy et al. 1999). I was responsible for coordinating the A more complete discussion of definitions and measures of 'automobile dependence' is available in Chapter 2. 2 research and data collection for the six additional Canadian cities to be included in the update (Vancouver, Edmonton, Calgary, Winnipeg, Ottawa-Hull and Montreal were being added to Toronto) for four study years (1961, 1971, 1981 and 1991). The data used for each region fall into the following broad categories: population and employment distribution, size of urbanized area, transport infrastructure supply, vehicle ownership, transport energy consumption and public and private transportation usage. I have used the data compiled from the CAAD update project as the basis for this thesis. 1.0.0 Why Compare Transportation in Canada? The comparative approach to urban policy analysis can be useful for both theorists and practitioners to learn lessons from the experience of other regions. The approach allows for the testing of both dependent and independent variables between cities to yield insights into what is unique and what is commonplace (Artibise 1990; Goldberg and Mercer 1986). These tests provide a starting point for analysis of factors that account for differences. The comparative approach can contribute to an understanding of the key forces that contribute to higher or lower levels of automobile dependence. Newman and Kenworthy (1989a) showed that cities that are more automobile dependent demonstrate particular characteristics that set them apart from less automobile dependent cities. In this thesis, I attempt to identify key factors that may account for these differences and key policies that could be instituted to facilitate favourable (i.e., less auto dependent) outcomes. Comparing transportation trends and patterns amongst Canadian cities is particularly valuable because they are relatively homogenous. While lessons can be derived from comparing a city like Toronto to those in the U.S. and others globally, as much relevant information can be gleaned by comparing differences and similarities within Canada. Since there is a broader common denominator of economic, social, political and cultural forces influencing urban processes amongst these cities, there are fewer "intervening" variables. Studying the forces responsible for lower levels of auto use within a similar socio-economic, political and cultural context can help to better focus policy analysis. 1.1 OTHER KEY STUDIES The comprehensiveness of the parameters and the rigorous methodology employed in the surveying and compilation process makes the data set an important and unique contribution to the understanding of urban transportation in Canada. Although other studies have recorded some of the various relationships between transportation, urban activity and land use, none have been able to achieve an acceptably accurate level of inter-city comparability. Key problems with these various studies include: • the lack of trend data (most studies report on one specific year); 3 • some studies have missing parameters key to the understanding of the inter-relationships of urban activity; and • poor data comparability due to imprecise definitions of land areas, poor matching of populations, urban areas and parameters, inaccurate or unclear reporting on parameters and soft estimations that all render the validity of comparisons questionable. The earliest of these studies to offer a comprehensive survey of transportation and land use trends in Canada was ND Lea's 1966 study, Urban Transportation Developments in Eleven Canadian Metropolitan Areas (N.D. Lea and Associates 1966). This study provided a series of indicators regarding population, urban structure, transit service, motor vehicle traffic and transportation facilities in 11 Canadian metro areas. It also complemented these with corresponding demographic, economic and infrastructure indicators. However, from a comparative perspective, the study has several flaws and omissions that make standardization and meaningful analysis difficult. For example: • definitions of urbanized areas are unclear or poorly defined; • often, it is not clear whether other data correspond to the stated "urbanized area"; • the "urbanized area" and associated populations seem understated. One possible reason for this could be that only areas with population densities over 1000 people per square mile were included. However, many large tracts of industrial and institutional land that are urban in nature would likely have lower "population" densities; • fuel consumption figures lack consistency between municipalities and are based on very soft estimations; • there are no region-wide total vehicle travel (vehicle miles travelled, or VMT) estimates for the entire urban areas. Only traffic volumes for arterial roads with over 10,000 vehicles per day of traffic are counted; • road network data lack completeness and are therefore not comparable; and • while urban structure data corresponds to 1931, 1951 and 1961, the population, transit and motor vehicle data correspond to 1945 and 1965. In an age of extremely rapid motorization and suburbanization, such inconsistencies make accurate comparisons difficult. Nonetheless, the study was a formidable achievement, particularly given the difficulty in accessing such data at that time. It provided a picture of the general relationships between sprawling land use, motorization and transit decline. Later studies also attempted to develop a series of indicators on transportation and land use patterns with mixed success. In 1979, Transport Canada performed an extensive study of many aspects of urban transportation in Canada with The Role of the Automobile Study (Transport Canada 1979). Although extensive data were provided, they were often sourced from Statistics Canada publications and were often not disaggregated to the metropolitan level (i.e., they were either provincial or national figures). Of course, such reporting can do little for city-specific policy making and can only communicate the most macroscopic trends. Where transportation data were provided on a metropolitan level, no historical trends were offered, nor were they standardized or linked to other comparable land use, transit, population or demographic data. 4 Two studies in the 1990's presented a series of raw and standardized indicators to describe the urban structure, population and employment and transportation supply and demand characteristics of Canadian cities. The IBI Group report for the Canadian Mortgage and Housing Commission (CMHC), Urban Travel and Sustainable Development: The Canadian Experience, detailed travel patterns for ten Canadian cities (IBI Group 1993). Then, the Transportation Association of Canada (TAC) produced Urban Transportation Indicators in Eight Canadian Urban Areas, a survey of transport patterns in Canada's largest cities in 1991 (TAC 1996). Since comprehensive data on transportation and land use in Canadian cities are rare, these studies both provided a wealth of needed data. However, both also had some methodological challenges that affected the reliability and comparative value of some of the data. The IBI report, for example, uses retail gasoline sales to estimate several other transportation activity, energy use and pollution parameters. Since vehicle kilometres travelled (VKT), energy consumption and pollution data in the IBI study are solely a function of (and vary directly with) gasoline sales, they will simply measure retail gasoline sales, expressed in different units (i.e., they will co-vary). Many factors other than gasoline consumption explain VKT, energy use and pollution. For example, local peculiarities such as climate, topography, fleet vehicular efficiency and age and average traffic speed all have an impact on fuel consumption (Kenworthy 1986; NDRC 1997). One city's fleet may have a higher proportion of older, heavier vehicles that consume more gas and pollute more on a per kilometre basis. Another city may have lower vehicle occupancies or fewer hills resulting in less gas consumed per kilometre. Winnipeg's winter cold-starts consume more gas and pollute more than Vancouver's. Theoretically, gasoline sales can provide a ballpark estimate of other parameters, however, most of the data necessary to do this (such as city-specific vehicular fleet fuel efficiencies) are not widely available. Therefore, such estimations are of marginal use for comparative purposes. The best way to achieve comparability is to obtain independent estimations of each parameter in each city. Also, the range of parameters surveyed does not give a full enough picture of the urban activity patterns that characterize Canadian cities. For example, there are no parameters measuring road and parking infrastructure provision, job and population distribution or the relative proportion of passenger travel by car or transit. The TAC study that followed IBI's report (which was completed with the assistance of IBI) was more thorough and provided a more comprehensive picture of transportation and urban structure. It even provided highly valuable and difficult to obtain transportation cost and finance data outlining consolidated capital and operating expenses for each urban area. However, although great pains were taken to explicitly define parameters in the survey, the survey responses reveal inconsistencies in reporting between municipalities. This likely reflects poor data availability and different data reporting conventions at the municipal level. 5 For example, although the study year was 1991, years for source data varied from 1987 to 1994. In fact, of the eight cities surveyed only Toronto and Ottawa were able to consistently provide 1991 data. Other key parameters were missing for many cities (such as freeway and arterial vehicle kilometres), while others were poorly reported and/or vague (such as transit usage data) or not requested (such as transit supply and total VKT). TAC is attempting to address many of these issues for future updates to the study. However, the inconsistencies and omissions make it difficult to benchmark the data and derive reliable performance indicators for the time being. There are several methodological issues common to both the IBI and TAC reports that merit attention. Firstly, none of the surveys subsequent to the 1966 ND Lea study provided trend data on the evolution of transportation in Canadian cities. Understanding how land use, urban activity and transportation interact over time provides valuable lessons about the forces that contribute to or mitigate against auto dependence. Secondly, both the IBI and TAC reports excluded Hull, Quebec from their surveys. Hull is an important part of the contiguous and functional area of metropolitan Ottawa. It represents 25% of the region's population base (see Ottawa-Hull data table) and is a signification trip generator in the region. Thirdly, both reports provide values for "urbanized area" that are internally inconsistent and defined differently on a region-by-region basis. The TAC recognizes this, stating that noticeably different conventions for defining the urbanized area were used in each region's reporting (TAC 1996, p. 7). Most often, these differences overstate the size of the urbanized land as the estimates include much non-urban land (e.g., water, regional parks, farmland, etc.). The degree to which this is overstated varies depending on the amount of non-urban land included in each administrative boundary. This overstatement results in a distortion of varying degrees in the true values of urban densities2. Urbanized land figures are used to derive other performance indicators and also form the basis for the comparison of transportation data amongst cities. As much of the discourse in transportation policy revolves around its relationship with land use, and particularly compact development, an accurate assessment of urban density in Canada is needed for appropriate policy formulation. 1.1.0 Some Criticisms of CAAD Although some criticisms have been levied on the comparative work done by Newman and Kenworthy (see Brindle 1993; Gomez-Ibaiiez 1991; Gordon and Richardson 1989; Lave 1992), most of this critique focusses on issues of interpretation rather than methodology. Despite these criticisms, the approach taken in the compilation of the data is regarded as methodologically sound and the data set is 2 The issue of inaccuracies in calculating urban densities will be discussed in greater detail in Chapter 3: Methodology. 6 regarded as an excellent compendium of comparative urban transportation and land use data (Gomez-Ibafiez 1991). Nonetheless, some of the key methodological and interpretative criticisms warrant some attention. Gordon and Richardson (1989) offer several criticisms of Newman and Kenworthy's prelude to CAAD, their APA journal article "Gasoline Consumption and Cities: A comparison of U.S. cities with a global survey" (1989b). While many items of contention are those found in the normal course of debate on policy analysis, two in particular are of methodological significance and warrant attention. Firstly, Gordon and Richardson assert that urban structure and the lack of transit service alone cannot explain differentials in transportation characteristics. Other factors such as lifestyle and travel behaviour need to be considered. While this assertion is true, the CAAD data is not meant to be exhaustive, but rather point to some key variables that influence travel characteristics. The data are not meant to imply singular causality between car use and urban structure and transit provision, but rather to provide a picture of some of the key influences in travel behavior. Indeed, while Newman and Kenworthy's qualitative analysis of the data suggests strong prescriptive measures that are open to debate, the real value in the data set is the amassing of a standardized set of data that can be complemented by the policy analyst with other independent data and qualitative information. Gordon and Richardson's more relevant criticism is that "[Newman and Kenworthy] are pre-occupied by work trips" (Gordon and Richardson 1989, p. 343). While the criticism is overstated, many studies have confirmed that non-work car trips represent the bulk of regional travel in most North American cities (see, for example Altshuler 1980; Calgary 1993; Cervero 1989; GVRD 1993a). The exclusion of non-work modal share and trip length data stems mostly from a lack of availability of "all trips" numbers, particularly in earlier years. Most cities compile trip length and modal split data primarily for the journey to work. In most of the Canadian cities in this study, complementary data on "all trips" were also collected where possible. However, these data were only occasionally available. The inclusion of joumey-to-work data is only a problem if the analysis of the data depends on it to the exclusion of other information. Again, complementary quantitative and qualitative information is needed for meaningful interpretation. Furthermore, these data only represent a small portion of indicators used and they are still useful as measures of work-trip car use and as anecdotal evidence of general car dependency. Brindle (1992) criticized CAAD for using data from the much denser Metro Toronto (which contains only half the region's population), while leaving the more auto-dependent suburban areas of the Greater Toronto Area (GTA) out of the analysis. Brindle claims Toronto does not stand as a model of good transit-oriented planning, but rather is a "paradigm lost." Brindle claims that the Metro definition of Toronto is used to overstate its performance, and that when the outlying areas of the GTA are considered, Toronto is no better, or even worse, than many U.S. cities in automobile orientation. Brindle's assertion 7 that a truly complete analysis of Toronto needs to include the entire GTA is a valid one. As mentioned section below (Territorial Areas), in order to ensure 'apples to apples' comparisons, data need to be collected for the entire functional urban regions, not simply on the basis of arbitrary planning boundaries. This study therefore uses all areas within the functional urban region.3 However, Brindle's assertion that Toronto would be in the league of its more auto-dependent U.S. neighbours when the entire GTA was considered and then refuted by subsequent analysis of Toronto (Kenworthy and Newman 1994b). Although the GTA did show marginally higher automobile use, it was still very low by North American and Australian standards. Furthermore, Kenworthy and Newman they confirmed the link between higher density land use patterns and high transit use, even at that much larger regional scale. Steiner (1994, p. 37) criticized CAAD, and other similarly regionally-oriented studies, for using "grossly aggregate data" and "a narrow definition of urban form that considered the density of both employment and housing but omitted the type of land uses and their spatial distribution within the region." This criticism does have some validity. Sub-regional spatial distributions of employment and population are provided within CAAD for the CBD, central (inner) city and outer area. While more detailed observations of the interaction of land use and transportation at the sub-regional scale (e.g., in polycentric regions, or at the neighbourhood level) are both necessary and laudable as complementary research, this is beyond both the scope and practicality of CAAD's research. The objective of CAAD (and this thesis) is not to conclusively solve the problem of auto dependence by collecting every possible piece of relevant data and establish absolute causality. Rather it is to describe general transportation trends and patterns so that macroscopic inter-urban comparisons of aggregate city performance can be made. Sub-regional data and studies are important complements to this effort. However, given the time required to assemble the data and the difficulties in standardizing them for meaningful comparison, such a task would be truly monumental. Finally, Gomez-Ibanez (1991), Lave (1992) and Steiner (1994) have argued that CAAD is fundamentally flawed because it ignores the contribution of regional wealth and other economic factors in creating high levels of auto ownership and use. Higher incomes are argued to necessarily lead to high car ownership and use (Gomez-Ibanez 1991; Lave 1992). The impact of regional wealth on auto use is a key variable that would help to shed light on contributory factors to auto dependence. While there has been little proof of Lave's and Gomez-Ibanez's assertions, they have been held as conventional wisdom. However, subsequent studies (Aschauer and Campbell 1991; Kenworthy et al. 1997; Laube 1998) have debunked the myth that auto ownership and use are necessary by-products of increasing wealth. 3 See section 3.3 (Territorial areas) in Chapter 3 for a complete definition of "urbanized area" used for each region. 8 Kenworthy et al. (1997) showed that there is no strong correlation between auto use and wealth and that amongst many developing and developed cities, gross regional product (GRP) actually begins to decline in the most automobile-oriented regions. That is, after certain levels of ownership and use, there are substantial diseconomies associated with further auto dependence. 1.2 THESIS SCOPE This thesis attempts to describe some of the general transportation trends and patterns in Canadian cities such that some key factors contributing to automobile dependence can be identified. An extensive review of the literature examining causes and impacts of automobile dependence will inform the analysis of the data in this thesis. There is also a large body of literature that examines policy prescriptions that address these impacts. Prescribing policy based on the data analysis would be a substantial project unto itself if it were to be meaningful. Therefore, rather than prescribe policy based on the data interpretation, this thesis will simply lay out a framework for policy analysis that reconciles the findings with possible points for intervention. The scope of this thesis is also defined by the following: • the data analysis will concentrate on the major Canadian cities, however data from the larger global cities sample will be used to contextualize the results and provide a larger sample size for correlations; • the data and analysis in this thesis is limited to the four study years of 1961, 1971, 1981 and 1991. While there have been notable transportation developments in Canadian cities since 1991, it is difficult to collect current data that is comprehensive and comparable between cities. As a result, I will limit discuss of post-1991 developments to occasional references where appropriate. The thesis is therefore more an exercise in 'learning from history' than a description of current performance; • the transportation data will primarily consist of private transportation (automobile) and transit data. This limitation is due mainly to data availability. Only limited modal split data indicating non-motorized travel (walking and cycling, or NMT) are available and only for work trips. These data, plus some additional modal split data for some cities, will be assumed to be representative of basic relative conditions for pedestrians and cyclists; • sub-regional comparisons and data will be limited to the CBD and inner area. Some complementary data may be drawn on from other quantitative studies. While additional information would useful, it would be too time consuming to collect it for the purpose at hand; • by its nature, this thesis focusses on the quantitative aspects of urban transportation and urban morphology, however, many qualitative aspects (e.g., culture, physical design, etc.) also influence travel choices. This thesis recognizes this and will draw on these perspectives where possible. However, more detailed qualitative analyses are beyond the scope of this thesis; and 9 • this thesis will orient discussion around the goal of making cities "less automobile dependent" rather than making them "sustainable." My review of the literature indicates that there is substantial debate as to what constitutes "sustainability" in general. Furthermore, there is not enough consensus as to what constitutes "sustainable transportation" to make such a concept operationally meaningful. I will simply define automobile dependence in Chapter 2 and assume reducing it is a means to better transportation. 1.3 THESIS OBJECTIVES AND STRUCTURE As mentioned at the beginning of this chapter, the purpose of this thesis is to identify the relative importance of factors that can reduce automobile dependence in Canadian cities through a comparative analysis of transportation patterns. Towards this purpose, the objectives of the thesis are: 1. to discuss the value of comparative urban policy analysis and establish the need for such an analysis of transportation in the Canadian context; 2. to conduct a thorough review of the literature to define automobile dependence and to identify its major causes and impacts; 3. to detail the methodology employed for collecting meaningful and reliable comparative urban transportation data in Canada; 4. to analyze the data collected and identify the relative importance of factors influencing auto use in Canada; and 5. to develop a framework for directing policy based on the data results. This thesis is divided into five additional chapters, as follows. C h a p t e r 2 consists of a major review of the literature. It provides an operational definition of automobile dependence to frame the analysis in the thesis. It then surveys the major literature to identify the key factors governing urban travel patterns in general and automobile ownership and overuse in particular. The social, economic and ecological implications of automobile dependence are identified and discussed and a critique on conventional policy is offered. The methodology employed in the collection of the data use in this thesis is given in C h a p t e r 3. Area and data definitions are also provided and discussion of data quality and limitations provides important background information for interpreting the data. City maps and the data survey sheets can be found in A p p e n d i x 1. The data are analyzed in C h a p t e r 4 and key findings of Canadian transportation trends and patterns are presented and discussed. The relative importance of key factors making regions more, or less, auto dependent are identified. Comprehensive raw and standardized data tables for each of the Canadian cities are in A p p e n d i x 2, a summary "master sheet" of data for all the world cities referred to in the Chapter are in A p p e n d i x 3 and the detailed correlation tables are in A p p e n d i x 4. 10 Chapter 5 sets out criteria and a framework for prescribing policy based on the literature as reviewed in Chapter 2 and the relative importance of factors influencing auto dependence identified in Chapter 4. Chapter 6 concludes with a summary of the thesis findings, directions for policy analysis and some suggestions for further research. 11 CHAPTER2 - AUTO DEPENDENCE: DEFINITIONS, CAUSES, IMPACTS AND FUTURE DIRECTIONS 2.0 INTRODUCTION An understanding of automobile dependence, including its causes and its consequences, is required to develop policy that is appropriate and effective. There is a broad and extensive range of literature highlighting the many complex relationships and implications of the car's role in urban transportation. In an attempt to contextualize the thesis data and make recommendations based on them, this chapter identifies and characterizes some of the key issues surrounding the use and overuse of automobiles through a review and synthesis of the literature. This chapter provides the following: 1. a working definition of 'automobile dependence' and discussion of the goal of transportation (what is the problem?); 2. a discussion of the causes of automobile dependence (what causes the problem?); 3. an assessment of the range of ecological, social and economic implications of auto dependence (why should we care about the problem?); and 4. a critique of some conventional reductionist approaches in solving transportation problems (why many certain types of solutions do not work). 2.1 PERSPECTIVES ON AUTOMOBILE DEPENDENCE The term 'automobile dependence' was coined and popularized by Peter Newman and Jeff Kenworthy in their 1989 book Cities and Automobile Dependence and describes a series of convergent land use and transportation conditions in cities that leave people few non-car options for urban travel. In examining transportation and land use patterns, Newman and Kenworthy found that while some cities displayed a robust mix of travel modes (car use, transit, cycling and walking) others displayed a much stronger orientation towards the automobile. Some of the more automobile-oriented cities are effectively mono-modal, with as much as 93% of trips being made by car. These most 'automobile dependent' cities display low density, dispersed and uniformly zoned land uses and high priority for car use. Robert Cervero's observations of transportation and land use patterns in America's suburban centres supports this finding: "the low density, single use, and non-integrated character of many suburban office-commercial centers and corridors has compelled many workers to become dependent on their automobiles for accessing work and circulating within projects" (1989, p. 3). Increasingly development on the periphery of Canada's cities demonstrate these same qualities (Perl and Pucher 1995; Raad and Kenworthy 1998). These land use and infrastructure characteristics effectively preclude the availability 12 or viability of non-car modes. Cities with such limited transportation choice are dependent on automobiles to meet urban travel needs. High levels of car ownership and use result. Newman and Kenworthy (1989a) categorized the cities of their study into five classes from very high levels of auto dependence (Class I) to very low levels of auto dependence (Class V). These city classes appear in Table 1 below. One of the key defining features of these cities is an inverse relationship between urban densities and automobile dependence. Table 1 - Classification of cities by d egree of automobile dependence Class 1 Class II Class III Class IV Class V Very High High Moderate Low Very Low Auto Auto Auto Auto Auto Dependence - Dependence - Dependence - Dependence - Dependence -almost no role for minor though important role for transit, walking, transit, walking, transit, walking, significant role for transit, walking, cycling equal with cycling more cycling; very high transit, walking, cycling; moderate cars; low gas use important than gas use cycling; high gas gas use cars; very low gas use use Phoenix Washington Toronto Amsterdam Munich Houston Melbourne New York Frankfurt Singapore Denver Boston Copenhagen West Berlin Paris Detroit Chicago Hamburg Vienna Hong Kong Perth San Francisco Zurich London Tokyo Adelaide Sydney Brussels Stockholm Los Angeles Brisbane Source: (Newman and Kenworthy 1989a) The regional scale dimensions of automobile dependence have important implications for individual transportation choice. The degree to which one is dependent on the car for transportation varies with the degree to which non-car transportation options exist (car need) and the access some one has to an automobile (car availability). Figure 1 below describes how captive the individual may be to the automobile given the relative need for and availability of a car. Many factors, such as proximity to destinations, availability of alternative modes and age, may affect need, while income is the primary determinant of availability. Individuals who may need a car to access employment because they cannot afford more central housing, but also cannot afford a car, would be "transport poor." Those with high car need and high car access will be completely dependent on the car for transport. For example, an affluent CBD-employed business person who lives in an ex-urban area with no commuter rail will likely use their car for most trips. People at the other end of the spectrum have low car need because of location, greater transportation choice, or other factors, and therefore are either "car free" or lucky enough to have the "luxury" of a car. Need, of course, is relative. Many people who drive frequently would claim that they "need" their car. However, virtually every large city in the world has a public transportation system. Likewise, 13 most people have the ability to walk or cycle. An appropriate gauge for "need" would be whether the person would frequently forfeit the opportunity to meet basic economic and social necessities in the absence of a car (i.e., they would not walk, bike or take transit even if possible). For example, in Perth, Australia, some bus routes run at frequencies of only 2-3 times per day and there is no bus service after 8 p.m. on Sundays. Few alternatives to the car exist for longer distance journeys in these cases. Figure 1 - Degrees of car dependence Q UJ UI z < o No Car. Always Needed No Car Sometimes Needed No Car. Car Not Needed Car Sometimes Available. Always Needed Car Sometimes Available Sometimes Needed Car Sometimes Available Car Not Needed Ci r Available. Always Needed Car Available Sometimes Needed Car Available Car Not Needed % C A R A V A I L A B I L I T Y Source: (Adaptedfrom Pharoah 1996) If one is poor and/or disabled, they simply would not be able to make the journey. Similarly, a homemaker in a single-car, low-density subdivision whose partner is at work will have few opportunities to take advantage of education, social or shopping opportunities and will therefore be isolated. Sometimes, transit service does not exist, and other times, it does not go where or when one wishes to travel. Other times, ability or distances are prohibitive for cycling or walking. Clearly, these are cases where a car is "needed." It is useful to take Pharoah's typology for individual levels of auto dependence and apply Newman and Kenworthy's metropolitan-scale classification to it. Clearly, certain cities would fall neatly into certain categories. Detroit's citizens might waver between "transport poor" and "car dependence." In Zurich, where car ownership is moderately high, but need and use is low, citizens have cars "as 14 luxury." Many in cities such as Amsterdam, Tokyo and Hong Kong may be lucky enough to live "car free" and still meet all their travel needs through transit, walking and cycling. 2.1.0 Redefining the Goal of Transportation According to many transportation critics, one of the fundamental reasons transportation outcomes tend towards auto dependence in North America lies in how we define 'transportation.' Traditionally, planning has viewed the goal of transportation as mobility: the maximization of the free movement as measured by the speed and volume of traffic (Altshuler 1979). The automobile is often regarded as a perfect solution to challenges to mobility as it offers drivers control, comfort, freedom and convenience in going anywhere at anytime. In promoting mobility, the efficient and quick movement of traffic has become a singular objective of traffic planners in many cities. This free movement is traditionally measured by level of service (LOS) on roadways, or the degree to which traffic flows without interruption (i.e., average travel speeds) (Ewing 1993). The higher the LOS and the higher the vehicle ownership levels, the greater the level of mobility. Therefore, as a means of promoting mobility, public policy has focussed on increasing LOS and personal car ownership. However, many authors have also argued that there is a profound irony in the fact that the quest for total mobility has resulted in the increasing immobility of many, often including those able to drive (Altshuler 1979; Ewing 1993; Freund and Martin 1993; Lowe 1990; Newman and Kenworthy 1989a; Schwartz 1971; Whitelegg 1993; Zielinski 1994). Maintaining high LOS (i.e., preventing congestion) has proven illusive (Freund and Martin 1993; Hart and Spivak 1993; Newman and Kenworthy 1988b) and car ownership will likely never be universal. While some continue to be unable to meet their basic needs (e.g., the young, old, poor and disable unable to afford cars), others are simply stuck in traffic. As will be discussed later, the pursuit of high levels of mobility through private motorized transportation also results in substantial social, economic and ecological costs to the individual and society. Transportation is a means of facilitating social and economic interaction for urbanites, not an end in itself. Lewis Mumford (1953) and Jane Jacobs (1961) warn that unfettered mobility is actually antithetical to the purpose of the city in that it interrupts the basic city functions of proximity for contact and production as well as multiplicity of choice. In The Death and Life of Great American Cities Jane Jacobs states: " Good transportation and communication are not only among the most difficult things to achieve; they are also a basic necessity.. .But multiplicity of choice and intensive city trading depend also on immense concentrations of people, and on intricate minglings of uses and complex interweaving paths" (1961, p. 339-340). In The Highway and the City. Mumford writes: "The paradoxical result of this concentration on motorcars is a curbing of freedom of movement, a removal of alternate choices of transportation, the steady reduction of the speed of local travel, and the total defeat of the city itself as a 15 place that offers the maximum possibilities for face-to-face meeting, social cooperation, and transactions of every kind" (1953, p.222). The single-minded pursuit of individual mobility can actually serve to undermine the basic urban functions it aims to enhance. Little has changed since then. In his critique of the time and space dimensions of auto use, John Whitelegg argues for a paradigm that reflects the still-illusive values espoused by Jacobs and Mumford four decades ago. "The ability to make contact with places and other people is the central organizing feature of human activity and that it is ease of access to other people and facilities that determines the success of a transportation system, rather than the means or the speed of transport. It is relatively easy to increase the speed at which people move around, much harder to introduce changes that enable us to spend less time gaining access to the facilities that we need" (Whitelegg 1993, p. 131). The systemic biases towards auto dependence are still entrenched today. A more holistic way of defining transportation to provide a more appropriate framework for transportation policy formulation is needed. Focussing on access rather than mobility, it is argued, would result in more equitable, efficient and socially responsible satisfaction of travel needs (Altshuler 1979; Engwicht 1993; Ewing 1993; Litman 1995; Whitelegg 1993; Zielinski 1994). Rather than maximizing movement and speed for its own sake, access-based transportation seeks to maximize the contact that is the very reason for 'clustering' human activity in cities. Access requires proximity in destinations, but also choice in reaching them. This contrasts with the mobility-focussed, auto-centric paradigm that deprives people of access because it is mono-modal and assumes people can travel great distances. By promoting access, it is possible to reduce the "car need" of individuals and cities (see Figure 1 above) that is responsible for varying degrees of transport poverty or car dependence. David Engwicht (1993) provides a passionate and persuasive argument for a transportation system that access-based rather than mobility-based. Engwicht argues that goal of a city, and therefore the transportation system, is to facilitate ease of "exchange." The greater the diversity and proximity of land uses and the greater the number of options for accessing them, the more likely the opportunities for planned and spontaneous exchange. Figure 2 below demonstrates the tradeoffs between movement and exchange. Cities with exchange friendly transport offer more opportunities for social and economic exchange for similar amounts of "movement" and therefore offer greater access. Movement (through automobility) is good up to a certain extent, however it offers diminishing marginal benefits to exchange. Auto-dependent cities begin to limit exchange since much of the potential exchange space (e.g., shops, homes, parks, and paths) is actually occupied by movement space (e.g., roads, parking lots and freeways). The mere presence of this vast amount of movement space, and the landforms that accompany it, effectively precludes exchange-friendly transport. It also creates what Engwicht terms the "access-to-exchange disadvantaged" 16 (ATED) (Engwicht 1993). The ATED are those who do not have access to an automobile, where auto dominance precludes other travel options. In more automobile dependent cities, where movement is favoured over exchange, the gap between those who are ATED and those who are not is larger than in non-car dependent cities. Figure 2 - Maximizing exchange through mode choice Movement — • Source: (Adapted from Engwicht 1993) Lewis Mumford provides perhaps the most cogent articulation of the goal of transportation. "The purpose of transportation is to bring people or good to places where they are needed, and to concentrate the greatest variety of goods and people within a limited area, in order to widen the possibility of choice without making it necessary to travel. A good transportation system minimizes unnecessary transportation; and in any event, it offers a change of speed and mode to fit a diversity of human purposes" (Mumford 1953, p. 236). 2.1.1 Measuring Auto Dependence The degree to which a city is automobile dependent (or, conversely, is accessible) can be measured as a function of the spatial distribution of destinations, the availability of choices in getting there and the level of car use. While some detailed performance measures of accessibility have been proposed and used (Barter 1998; Laube 1998; Pirie 1979), there are no widely accepted conventions for measuring it (Handy 1994). There are, however, basic proxy indicators of accessibility that can provide anecdotal evidence of relative levels of car dependence. For example, urban activity densities can provide basic information about the proximity of people and activities and the levels of "exchange" possible. Modal splits, transit service levels and average trip 17 lengths can provide information about the practical range of non-car options available to urban residents in accessing these. While these convey information about level of "car need" and, by extension, accessibility, other indicators such as vehicle kilometers travelled (VKT), parking and road supply and motor vehicles on register provide a gauge of the priority and need for private transport to meet travel needs. Of course, all these individual measures should be analyzed with consideration to some of the more qualitative variables that influence access such as the quality of urban design. Newman and Kenworthy (Newman and Kenworthy 1989a) used "cluster analysis" techniques to assess levels of auto dependence based on a variety of land use and transportation data collected for their study. They then assigned each city a composite score ranking relative levels of auto dependence. For the purposes of this thesis, I will simply highlight the relationships between the various parameters identified as being good "proxies" for measuring car dependence and access. 2.2 CAUSES OF AUTOMOBILE DEPENDENCE The literature identifies a range of causes of automobile dependence such as road building, urban sprawl and the decline of transit, as well as economic, demographic and lifestyle factors. In reality, there is no singular cause of automobile dependence. Rather, there are multiple contributing factors, many of which are both cause and consequence. Furthermore, these display "positive feedback" which reinforces the various contributing problems making it seemingly intractable. Figure 3 below highlights some of the basic factors influencing increasing levels of car use.4 Changes in transportation technology provided the initial catalyst for widespread car ownership and use. The widespread availability of automobiles, particularly in wealthy countries in the post-war period, provided the mobility necessary to travel greater distances at relatively high speeds (Illich 1974; Mumford 1953). Convergent factors such as immigration, the baby boom and changing lifestyle preferences increased demands to open up new tracts of land to accommodate population increases. The availability of the automobile afforded planners the opportunity to accommodate much more dispersed settlement patterns than previously possible. With automobiles available, road building made suburbanization possible. Low-density suburbs proliferated throughout the developed world, but particularly in North America. Manufacturing and retail activities were able to free themselves from the locational constraints of rail and streetcar lines. Roads offered a new accessibility option to residences and industry. While roads make these low-density development patterns possible, they also make automobiles a necessity as transit services are unable to support themselves. Transit servicing costs vary inversely with 4 Of course, this diagram of influences can be placed in the context of more macroscopic social and economic dynamics. 18 density (Altshuler 1979; Newman and Kenworthy 1989a; Pucher 1988). Furthermore, the curvilinear roads characteristic of many suburbs makes these areas difficult to access and service by transit. Single use zoning means that travel distances are usually long. Increased subsidies are therefore required to service transit in low-density areas. Consistent and high subsidies usually result in pressures to curtail or eliminate transit services in low-density areas. These pressures have been particularly acute in Canada in recent years (Pucher 1998). Figure 3 - Positive feedback relationships in automobile dependence Subsidies Road building "Hard" services Utilities Home Ownership 1 L. X ^ Road Building Freeways HOV facilities Capacity improvements Urban Sprawl Low density land uses Single-use zoning Malls/big box stores Public Space Wide roads Traffic domination Low enviro. quality Less walk & cycle Culture Lifestyle Advertising Demographics ' - - L . Traffic/Congestion Vehicle growth outpaces capacity growth Transit Decreased viability Less service Higher subsidies Car Ownership & Use More trips Increased VKT Increased ownership Parking Free or subsidized Needed end-of-trip Building costs rise Subsidies Direct to user Borne by society Not perceived Note: Plus (+) indicates increased effect Minus (-) indicates decreased effect Not all suburbs are created equal. For example, Canadian outer areas (i.e., suburbs) are more than twice as dense as outer areas in the United States. The result has been much higher levels of transit use (and much lower levels of car use) in Canadian versus U.S. cities (Raad and Kenworthy 1998). The more methodical, planned suburbanization of Metro Toronto, with a linear street network, bus-rail integration and planned transit catchment areas, has resulted in high transit ridership even in Metro's suburbs (Frisken 1991). However, more recent transit-hostile land use decisions outside of municipalities outlying Metro have resulted in negligible levels of transit usage in the newest generation of suburban 19 office parks. Perl and Pucher (1995) report that up to 25% of workers arrived by bus in Toronto's first generation of office sites, whereas the figure is now down to nearly zero in the newest sites. The location and surrounding uses of these sites (usually, vast ground-level parking) precludes any viable use of transit. Monetary and nonmonetary subsidies for infrastructure, and directly to individuals, also influence development patterns in substantial ways (Frisken 1994a; Litman 1995; OECD 1994; Peat Marwick Stevenson & Kellogg 1993). In North America, governments subsidies (in the form of tax relief, cheap loans and grants) at the state, provincial and federal levels encouraged residential sprawl (Goldberg and Mercer 1986). The construction of roads, trunk sewer and water lines and utilities, and the provision of higher cost public services (such as schools and hospitals) to service low-density land uses, encourages sprawl by opening up tracts of seemingly "cheap" land for development. This sprawl, in turn, requires automobile use for access. Similarly, there are a whole host of financial costs (such as road maintenance and construction, traffic enforcement and free or subsidized parking) as well as non-financial costs (such as congestion, environmental and social impacts) not borne directly by the driver (Litman 1995).5 Since the driver does not perceive many of these costs, they merely serve to encourage a cycle of inefficient location and increased driving. Table 2 - Comparison of relative price changes and transit ridership in Canada. 1980-95 CPI -- Auto Urban Transit all Goods Operating Transit Fares Ridership Costs 1980-1990 +77.8% +108.4% +113.1% +16.3% 1990-1995 +11.7% +12.2% +34.5% -11.4% Source: (Pucher 1998) The low and dropping real and perceived costs of driving, combined with the rising transit operating costs and reduced transit subsidies, has resulted in dropping transit patronage and increasing car use. Table 2 above highlights some of the basic relationships between relative transit and auto costs and transit ridership levels. While transit fares rose only slightly greater than auto operating costs between 1980-1990, they rose substantially greater than the CPI and auto costs between 1990-1995. Furthermore, the RMOC (1995) reports that automobile purchasing costs and gasoline prices have dropped by 11 and 12%, respectively (inflation adjusted) since 1981. Pucher (1994; 1998) and Perl and Pucher (1995) attribute much of the decline in transit ridership in Canada to decreasing service and the growing gap in between transit ticket price and car operating costs. 5 The issue of subsidies to driving will be discussed in greater detail in section 2.3.2 below. 20 Durning (1996) describes some of the other relationships between relative car costs and car use in British Columbia, Washington, Oregon and Idaho in Table 3 below. While there are many other intervening factors that influence car use, the table below charts some clear relationships between car use, gasoline price and insurance costs. Litman (1997b) reports that the underpricing and cross-subsidies insurance premiums encourages driving and is inequitable. This flat price structure for insurance may also encourage inefficient locational decisions. Table 3 - Car costs and car use in the U.S. Pacific Northwest and British Columbia Gasoline Vehicle Travel Net Fuel Tax Average Consumption per per Capita (US$/gallon) Insurance Capita (miles) Premium (gallons) (US$/year) British Columbia 294 -6,000 0.59 -$700 Washington 462 8,880 0.32 $588 Oregon 457 9,540 0.42 $535 Idaho 466 10,230 0.33 $402 Adapted from: (Durning 1996) Some institutional practices also provide financial incentives for sprawl and increased driving. For example, bank-lending practices for home mortgages applied equally to prospective buyers, regardless of location, effectively discriminate against inner city home ownership. Currently mortgage assessments do not take into account locational choices that require substantially lower expenditures on car ownership and use. Goldstein (1996) estimates that accounting for lower transportation costs in certain locations (such as lower car ownership and operating needs) can provide an additional margin of affordability of up to US$40,000 on a house and lead to a higher demand for location efficient housing. The increased levels of driving, both in terms of the number of trips and vehicle kilometres travelled (VKT), leads to congestion conditions in the absence of increased LOS (i.e., road capacity). The phenomenal growth in vehicle ownership and use in Canada has generated a substantial demand for automobile infrastructure. New roads, freeways and capacity enhancement measures (such as HOV lanes, light sychronization and left-turn bays) have been traditional tactics to improve vehicle flow and speeds, ostensibly with the longer-term goal to reduce congestion. However, this strategy of trying to "relieve" congestion though comprehensive capacity additions and enhancements is widely recognized to be a futile task (Freund and Martin 1993; Gordon 1991; Lowe 1990; Renner 1988). There is a direct relationship between road provision levels and VKT (Kenworthy and Newman 1994a; Newman and Kenworthy 1989a). Furthermore, free flowing traffic has a strong association with sprawl and higher levels of driving and energy consumption on a regional scale (Newman and Kenworthy 1988a; Newman and Kenworthy 1988b). 21 More recently, empirical research has found substantial evidence that road transportation improvements actually derive greater demand for automobile travel (Goodwin 1996; Hansen 1995; Johnston and Ceerla 1996; SACTRA 1994; Williams et al. 1991). In other words, these improvements do not actually relieve traffic congestion, they "generate" or "induce" additional amounts of traffic in the short and long term. Since there is a substantial 'latent demand' for automobile travel in most urban areas (constrained in many cases by congestion), any short-term improvements in congestion conditions are eroded by new (induced) travel over the long term. In an analysis of the aggregate effects of road improvement projects, Hanson (1995) estimates that every 1% increase in lane miles induces a 0.9% increase (i.e., a 90% net growth) in VKT within 5 years. Where travel demand is strong, the effects of generated traffic will ensure that predicted congestion relief will be illusive. Furthermore, additional capacity catalyzes longer term changes in land use, public transport viability and parking demand which will have further generative effects on road traffic well into the future (Goodwin 1996; Newman and Kenworthy 1988b; SACTRA 1994; Williams et al. 1991). However, most transportation models do not consider the long run implications of generated traffic and do not generally incorporate generated traffic costs, particularly "external" ones, into economic analysis of projects (Litman 1997a; Williams et al. 1991). The effects of generated traffic are not limited to large road projects. The cumulative effect of many minor efficiency improvements (e.g., left turn bays, one way streets, signal enhancements) can also serve to induce more traffic if their cumulative effect is to raise the overall LOS for automobiles relative to other modes. Bill Curtis of the Sierra Legal Defence Fund captures the problem of induced traffic quite well. He likens building extra capacity for automobiles to "drilling holes in the bottom of a leaky boat to let the water out" (Berger 1993, p. 153). There is a growing body of evidence shows that reductions in road capacity can actually induce a net reduction in total traffic. Such measures can include the conversion of lanes for transit or parking use or the complete pedestrianization of roads. A major study recently completed for London Transport and the Department of Environment in the UK concluded that, taking into account traffic diversions to other roads, cases of capacity reductions have resulted in net traffic reductions of 25-50% original levels (Bates et al. 1998). Such evidence provides strong support to the theory that reducing auto capacity will not necessarily increase traffic congestion and may increase accessibility. Free or subsidized parking also provides an incentive for automobile use (Litman 1998a; Shoup 1996; Shoup 1997). Most cities have ample free or underpriced parking at roadsides, shopping malls and places of employment. These are subsidized or provided by retailers, employers or municipalities. Occasionally, residential parking will be incorporated into the cost of the dwelling. 22 Ample parking is institutionalized. Most municipalities have minimum parking requirements, which dictate a (usually generous) ratio of parking to square footage or number of building occupants/users. These minimum parking standards are usually a response to political pressures to provide places to park cars. Donald Shoup calls minimum parking requirements "a fertility drug for cars" that in some places has become "the arbiter of urban form" (1997, p.7). Shoup indicates that high parking requirements raise housing costs, reduce urban density and reduce land values. Often, costs for providing the spaces are not borne directly by users. The massive subsidies therefore encourage the overuse of cars. Furthermore, planners usually set minimum parking requirements to meet the peak demand for free parking, thereby deriving and inflating demand for parking at other times. Again, the ample supply of automobile infrastructure generates demand for car use, encouraging sprawl and reducing transit viability in the long run. The various by-products of the cycle of auto dependence serve to undermine non-motorized travel (e.g., walking and cycling) and transit. For example, sprawling land uses undermine the financial viability of public transportation by driving up the costs of providing service while simultaneously decreasing the size of the population catchment from which it can draw. Lower ridership leads to less service, which leads to increasing car ownership and use, and so on, through the cycle. Similarly, longer distances and more dispersed land uses make cycling and walking to shops, schools, work and entertainment less viable. Wider streets, priority to cars, traffic domination and safety concerns undermine the 'public realm' making these environments even less attractive, even hostile to non-drivers. The prevalence of malls and 'big box' stores oriented towards serving motorists further weakens the urban fabric. Again, this feeds into a cycle of increasing car use, congestion, more road building, more sprawl to get away from it all, and less walking, cycling and transit use. 2.2.0 Feedback in Transportation The 'cycle of dependence' described above depends on 'positive feedback.' Positive feedback is a well-recognized phenomenon in environmental and social problems in general (Berry 1977; Dubos 1970; Ehrenfeld 1978; Hardin 1968; Hardin 1985; Schwartz 1971) as well as in transportation behaviour and planning specifically (Altshuler 1979; Engwicht 1993; Freund and Martin 1993; Illich 1974; Litman 1995; Mumford 1953; Newman and Kenworthy 1989a; Pushkarev and Zupan 1977). Jacobs (1961) describes positive feedback as a process whereby an action leads to a reaction which in turn intensifies the condition responsible for the initial action. The need for repeating the initial action is amplified and a cycle is set in motion. Positive feedback loops can have positive consequences or negative consequences. In the case of urban transportation in the developing nations context, most cities are caught in positive feedback of socially negative consequence as is demonstrated in Figure 3 above. There is no single 23 feedback loop responsible for the downward spiral, but rather a web of mutually reinforcing feedback loops that, left alone, create an seemingly intractable cycle of dependence on automobiles. Jane Jacobs describes the cycle of "erosion of cities by automobile": Erosion of cities by automobiles entails so familiar a series of events that these hardly need describing. The erosion proceeds as a kind of nibbling, small nibbles at first, but eventually hefty bites. Because of vehicular congestion, a street is widened here, another is straightened there, a wide avenue is converted to one way flow, staggered-signal systems are installed for faster movement, a bridge is double-decked as its capacity is reached, an expressway is cut through yonder, and finally whole webs of expressways. More and more land goes into parking, to accommodate the ever increasing numbers of vehicles while they idle. No one step in this process is, in itself, crucial. But cumulatively the effect is enormous. And each step, while not crucial in itself, is crucial in the sense that it not only adds its own bit to the total change, but actually accelerates the process(Jacobs 1961, p. 349-350). Jacobs goes on to describe a series of attendant reactions that set in motion a series of complicating feedbacks and self-defeating palliatives with respect to transit use, pedestrians presence and intensity of districts. There are a variety of other perspectives on positive feedback in transportation. For example, Freund and Martin focus on road building, using the "black hole theory" of highway building to describe the problem: highway congestion begets added capacity, which causes sprawl and auto dependent spaces which leads to more car use and congestion, which leads to calls for renewed rounds of road building (1993, p. 20). Pushkarev and Zupan (1977) focus on land use as both "cause and consequence" of auto dependence and the decline of transit, walking and urban vitality. David Engwicht takes a broader view, incorporating many of the feedback loops shown in Figure 3 in a description of what he calls "burning down the house to stay warm" (1993, p. 55). He also implicates the death of the comer store, fear of crime and the atomization of society in the perpetuation of this cycle. Litman (1995) touches on many of the aspects of feedback described in Figure 3, including induced traffic, however he frames his analysis in terms of behavioural queues provided in the way transportation costs and benefits are distributed (i.e., whether the users pays). Reversing the problem is a matter of reversing the positive feedback such that it yields positive outcomes. Jacobs claims the "attrition of automobiles by cities" [emphasis added] can serve to reverse the erosion of cities by cars (1961, p. 359). For example, one way of doing this is to renew districts with vitality such that congestion develops and using an automobile is inconvenient and other modes more attractive. Over time, district activity can intensify further for locational reasons. Engwicht (1993) suggests a broad range of measures that move a city towards exchange-friendly transportation. More exchange opportunities are afforded individuals for the same amount of movement (Figure 2 above) because the feedback effects do not undermine the gains to the same extent. Litman (1995) prescribes a 24 realignment of the distribution of cost and benefits such that positive or negative (action suppressing) feedback yield more positive social outcomes. While Susan Handy offers many of the same land use and accessibility remedies suggested by others for reversing the cycle of dependence, she warns that, as a result of entrenched practice and patterns, "we may, in fact, be beyond redemption" (1993, p. 40). 2.2.1 Common Property Problems Many of the behavioural responses in transportation that elicit positive feedback are rooted in problems of common property. That is, where there is common property (e.g., public air, water and land) there is often no direct responsibility for the consequences of an action and the benefits are high and the costs diffuse. In his essay, Tragedy of the Commons. Garrett Hardin popularized the notion that the unregulated pursuit of self-interest in the management of common property (e.g., public air, water, land) leads to collective ruin (1968). In Filters Against Folly. Hardin builds on his critique of the unmanaged commons by identifying the "distributional paths" of benefits and costs (profits and losses) are the types of responsibilities that accompany each (1985). In p r i v a t i s m , both profits and losses in the use of common property accrue to the individual. This is said to involve intrinsic responsibility because there is a direct negative or positive impact as a result of one's actions. In S o c i a l i s m , profits and losses are differentially distributed by bureaucrats amongst the group that owns the common property (this is different from privatism in that the actor and the acted upon is the community). C o m m o n i s m privatizes all profits while all losses are indiscriminately distributed to the population as a whole, or are "commonized." These "unmanaged commons" are characterized by what Hardin terms negative responsibility: it pays for the individual to make the wrong decision. While Hardin recognizes that no distributional path is best in all circumstances, combinations of both socialism and privatism are preferable to unfettered commonism. Currently, transportation decisions can be characterized as practicing a form of commonism: benefits accrue mostly the driver (internalized), while costs are commonized to society (externalized).6 Many of the diffuse costs of transportation decisions do not enter into the feedback equation and responsibility is not assumed because there is no direct adverse consequence to the action (i.e., responsibility is negative). Therefore, from a social perspective, the omission of these "common" factors from the feedback equation results in feedback that is 'incomplete.' Transportation behaviour in Bangkok, Thailand provides an outstanding example of incomplete and positive feedback. Bangkok has some of the world's worst transportation-induced urban air quality (Faiz 1993). However, it is not uncommon for people to avoid walking for even short trips of 4-5 blocks 6 1 discuss "internal" and "external" costs in detail in section below. 25 because the air is so noxious. Instead, many will get in their cars and drive, thereby making their own contribution to the problem. From an individual perspective, feedback is incomplete. The individual does not perceive their additional contribution to air pollution and therefore does not adjust their behaviour. There is 'negative responsibility' because each person is rewarded with clean(er) air at the expense of others. At the societal level, the aggregate effect of such decisions leads to positive feedback: more pollution leads to more driving, leads to more pollution, and so on. Similar scenarios are played out in many other cities and in other circumstances to varying extremes. For example, public opinion studies in the GVRD indicate that although people recognize the main contribution they can make to improve air quality is to drive less, 9 out of 10 people still use their cars all or most of the time. Furthermore, although most say they would take public transit if it was convenient, most are unsure of just how convenient or inconvenient it is, and 38% of people have not even tried public transportation in the recent past (GVRD 1995b). Although transit may well be inconvenient for some, it is still a viable and convenient option in many Canadian cities. But clearly, many will not sacrifice their own total mobility for the benefits of one sacrificed trip divided by 1.8 million Vancouverites. The marginal impact off the individual's decision to drive is perceived as negligible and there is a 'negative' responsibility. Lack of cooperation also creates a major challenge in dealing with common property problems where there exists positive feedback and rewards for self-interested action. Political scientists and economists describe the decision-making processes that characterize the "Tragedy of the Commons" in the field of 'game theory' with a decision simulation called The Prisoner's Dilemma. The Prisoner's Dilemma (Figure 4a below) is a classic problem of conflict and cooperation that is used to predict economic, political, military and personal responses. In its simplest and most popular form, each of the two players (e.g., two captured criminals) has a choice of cooperating with the other or defecting to minimize their potential sentence. Depending on the two players' decision, each receives payoff (e.g., number of years in jail) according to a payoff matrix. When both players cooperate by refusing to fink on the other they are both rewarded at an equal, intermediate level (reward, R), such as a light sentence of 2 months. When only one player defects, they receive the highest level of payoff of no sentence (temptation, T), while the other player gets the fool's punishment of 10 months (fooled, F). When both players defect they each receive a high prison sentence. The dilemma demonstrates the difficulty of achieving cooperative behaviour when rewards are available for the successful defector. The ultimate result of failing to cooperate is a lose-lose situation for both. 26 Figure 4 - The prisoner's dilemma a) Generic Dilemma DO. i_ a> c o (A R = Reward T = Temptation F = Fooled P = Punishment Prisoner A Co-operate Defect Co-operate R=2, R=2 (win, win) F=10, T=0 (lose, win) Defect T=0, F=10 (win, lose) P=8, P=8 (lose, lose) b) Big Box Dilemma ra c No Big Box co Big Box Vancouver No Big Box R=6, R=6 (win, win) T=10, F=0 (win, lose) Big Box F=0, T=10 (lose, win) P=-1, P=-1 (lose, lose) Multi-player individual and municipal decision-making matrices often exhibit the similar outcomes. In the Prisoner's Dilemma model, Cooperate/Defect can be replaced with Walk/Drive, or a whole host of other individual decisions. At the municipal level, Cooperate/Defect can be replaced Big Box/No Big Box - in other words, the decision of whether to allow the construction of a big box store (see "Big Box Dilemma" in Figure 4b).7 It is well recognized that big box stores have tremendous 7 Big box stores are also referred to a 'megastores' or 'superstores' (e.g., Costco, Real Canadian Superstore, Home Depot). They are typically located in industrial areas that have been rezoned commercial. They have a large amount of floorspace are usually able to offer relatively cheap prices because they can engage in bulk buying and selling and can accommodate a large amount of on-site storage. They are typically car oriented establishments surrounded by large ground-levels parking lots and are reported to each generate up to 1 million car trips a year (Seelig 1998). 27 potential negative consequences for cities. While they do offer cheap goods, they offer relatively low marginal gains to property tax bases, they undermine local business, they destroy the urban fabric and they are almost completely automobile dependent. However, if the City of Vancouver does not allow Price-Costco to set up in its boundaries, the City of Bumaby almost surely will. So there is a 'temptation' for each city to pre-empt the other in order to capture a bigger piece of the proverbial 'pie.' If both begin a cycle of attrition to win the most of this type of development, it will almost certainly have adverse effects for both in terms of transportation, the loss of local small-scale business, decreased tax revenues, urban vitality and so on. Again, the individual actor (the municipality) perceives high personal gain and diffuse losses, and so acts with self-interest to the detriment of all. Completing feedback loops and facilitating cooperation amongst individuals and organizations can help overcome common property problems. Distributing seemingly indivisible costs and benefits such that they are perceived in the feedback equation helps develop intrinsic responsibility. Since it is difficult for people to perceive the incremental pollution that may result from an action, building in other feedback responses, for example through pricing, may provide appropriate behavioural queues.8 Reducing "temptation" responses in common property management such that individuals work together for the common good requires "mutual coercion, mutually agreed upon" (Hardin 1968, p. 1247). The lack of cooperation and coordination between municipalities in dealing with the challenges of transportation and land use has been identified as a major aggravation to sprawl and automobile dependence (Freilich and White 1994; Frisken 1991; Frisken 1994b; Frisken et al. 1996; Goldberg and Mercer 1986; Linteau 1990; Mumford 1953; Yago 1983). In a comparative assessment of transportation and planning in the U.S. and Canada, Raad and Kenworthy (1998) argue that institutional structures aimed at achieving cooperation and coordination between municipalities was an important factor in controlling automobile dependence in Canada relative to its neighbour. However, the systematic deterioration of means to manage matters of common interest among municipalities is leading to increased sprawl and car use in Canadian cities. Municipalities increasingly face "temptations" to appropriate many of the benefits of sprawl, while commonizing its many costs to the public at large. Clearly, a means of managing common property and indivisable costs and benefits (Franklin 1990) is necessary in order to rein in the positive feedback, privatization of benefits and commonization of costs that characterize auto dependence. 8 See section below for more discussion marginal cost pricing of externalities as a means of modifying transportation behaviour. 28 2.3 SYMPTOMS AND EFFECTS OF AUTOMOBILE DEPENDENCE Transportation impacts our lives in many ways. The range of urban transportation options available can make getting around the city easy or difficult. The transportation choices we make as individuals and as a society are a key determinant of whether our urban environments are ecologically sensitive, equitable, lively and efficient. However, those transportation choices we have made, and increasingly make, in Canada are compromising the well being of our cities. Automobile dependence characterizes major Canadian cities (Perl and Pucher 1995; Raad and Kenworthy 1998). There are few options available for urban dwellers, particularly outside the inner cities, for meeting their accessibility needs. The result is very high levels of urban automobile ownership and use, with tremendous ecological, social and economic impacts. However, the "problem" of transportation is invariably expressed in narrow terms, particularly amongst policy-makers. For example, transportation policy typically seeks to address singular and compartmentalized problems such as congestion, air pollution and, to a lesser extent, sprawl (for example, see Calgary 1995b; GVRD 1995a; Reynolds 1971; Transport Canada 1979). However, many of these are merely "symptomatic" of larger problems (Altshuler 1979; Schwartz 1971), and myriad other problems go unnoticed or ignored. Figure 5 - Ecological, social and economic dimensions of auto dependence Toad kill ^^*^^^ublic realm . . . . . . . . social dysfunctidf habitat loss f \ severance isolation Ecolog ica l / traffic domination Soc ia l oil spjJtGl equity .. , .. S , \ noise ^ vehicle disposal \ . . urban run/ff \ m I I \sprawl/ housing atfordabilih energy use ^efource depletion infrastructure insurance opportunity cost free parkinc E c o n o m i c ^armaintenance^ Many impacts of auto dependence are inter-related. While many impacts fall neatly into one of the 3 dimensions of auto dependence, many clearly have dual or multiple dimensions. The various symptoms of auto use are extensive, complex, inter-related and, often, mutually dependent (Figure 5 above). If policy is to effectively address the many complex ecological, social and 29 economic dimensions of automobile dependence, then these various symptoms and their relationships first need to be identified. The following sections provide an analysis of these ecological, social and economic dimensions of auto dependence identified in the review of the literature, with a special focus on the Canadian context.9 While the list of impacts discussed is comprehensive, it is by no means exhaustive. It is also import to note that there are many ways in which one "problem" impacts many areas. For example, air pollution occurs throughout a vehicle's life cycle (from resource extraction to disposal) and effects water, air and soil quality. Similarly, sprawl has distinct environmental, social and economic elements. Some symptoms of auto dependence may be discussed in several of the subsections, however, the discussion will touch on different dimensions of the symptom and a such will not "double count." 2.3.0 Ecological Impacts of Auto Dependence There are a whole host of ecological problems associated with the ownership and use of private automobiles. The following sub-sections touch on six broad areas of ecological impacts: water-related, land-related, resource consumption, vehicle disposal, air pollution and emissions, and "other" impacts. Many of the impacts discussed are not well recognized in the general literature, nor are they traditionally directly attributed to automobile use. I have attempted to take a "life cycle" approach to identifying the impacts so that their full extent can be revealed. Water related impacts Water-related impacts are perhaps the least recognized of the environmental consequences associated with automobile dependence. Impacts on water range from disruptions of natural hydrology systems, to contamination of waterways from road runoff, to spill-related impacts of oil extraction and transport. Of all land uses, sprawling urbanization has the greatest adverse impact on water hydrology and quality (MacKenzie 1987). Impacts include modifications to the local climate, increased erosion and sedimentation, increased precipitation and flooding potential, effects on groundwater recharge rates, and reduced water quality. These effects result from the paving over of soils which previously served to absorb rain and snow and keep water cycles and flows in balance with ecological needs. Paving causes a channeling of higher volume and higher speed water that leads to the above mentioned impacts. Also of importance with respect to water-related impacts is the contamination of water associated with this runoff. 9 The literature regarding the impact of transportation in Canada is limited and not well-recognized in the larger body of transportation literature. However, I have attempted to draw on the Canadian experience wherever possible in order to enhance the relevance of future policy discussions. 30 The United States Environmental Protection Agency suggests that urban runoff may rival agriculture as the worst contributor of non-point pollution, and may be a far more serious polluter in many areas (Hall and Anderson 1988). However, the general public does not perceive that urban runoff problem to be a major concern (McConnell 1991). A significant portion of urban runoff pollution stems from automobile-related activities (NDRC 1997; Newman 1994; OECD 1995). Table 4 below details some of the sources of water pollution and hydrological disruptions associated with vehicle use. Most of the sources are non-point, except where they originate from oil and gas related industry and storage facilities. Motor vehicles, roads, auto-related industry and parking facilities all release contaminants that are then washed into stormwater drains, soils and water tables with each successive rainfall. For example, in the U.S. each year, a total of 1.4 billion gallons of lubricating oils are used. Of that total, 600 million gallons are burned or spilled in leaks and 180 million gallons are disposed of improperly (poured onto the ground or into sewers). Furthermore, an estimated 46 per cent of all cars on the road leak hazardous fluids, including - transmission fluid, crankcase oil, hydraulic fluid, and antifreeze onto the roadways, which are then washed down storm sewers and into soils (Bein, Litman, and Johnson 1994). Table 4 - Sources of water pollution and hydrological disruptions due to auto-related activity Water Pollution Hydrological Impacts • Leaks of hazardous fluids • Increased impervious surfaces • Road de-icing (salt) damage • Concentrated runoff • Pavement and vehicle wear • Loss of wetlands • Leaking underground storage tanks • Shoreline modifications • Air pollution settlement • Increased water temperature • Asphalt leachate • Construction disruptions of riparian zones Source: (Bein, Litman, and Johnson 1994) In British Columbia, urban runoff is considered a potentially larger source of toxic contaminants than sewage discharge (BC Environment and Environment Canada 1993). In the GVRD, urban runoff accounts for 32% of all wastewater discharges into the Fraser River Estuary and Boundary Bay (Environment Canada and BC Environment 1992). Although this problem is acute and persistent, urban runoff has not been monitored and there are few regulations controlling discharges in British Columbia. Table 5 below indicates the many constituent pollutants of road runoff and their many sources. The primary sources are directly related to various aspects of automobile operation and the infrastructure dedicated to it. The sheer scale and complexity of the problem precludes any one effective technological solution to solve it. Atmospheric deposition of pollutants is another source of contamination (namely, hydrocarbons) to aquatic environments (UNEP 1993). These contaminants eventually find their way from water sources 31 into the ecosystems as they are assimilated in marine life, agricultural products and drinking water (Miller 1993). Table 5 - Summary of road runoff co nstituents and their primary sources C o n s t i tuents P r i n i a r v S o u r c e Particles pavement wear, vehicles, atmospheric deposition, highway maintenance Nitrogen, phosphorous atmosphere, road fertilizer Lead leaded gas, tire wear, lubricating oil, grease, bearing wear, road paints Zinc tire wear, motor oil, grease Iron autobody rust, steel highway structures, automobile parts Copper metal plating, bearing wear, engine wear, brake lining, fungicides, insecticides Cadmium tire wear, insecticides Chromium metal plating, moving engine parts, brake lining wear Nickel gas, diesel, lubricating oil, metal plating, bushing wear, brake linings, asphalt Manganese moving engine parts Bromide auto exhaust Cyanide anti-cake compound for de-icing salts Sodium, Calcium de-icing salts, grease Chloride de-icing salts Sulphate roadway beds, fuel, de-icing salts Petroleum spills, leaks, engine blow-by, antifreeze and hydraulic fluids, asphalt leachate PCB's pcb catalyst in synthetic tires, atmospheric deposition Rubber tire wear Source: (Bein, Litman, and Johnson 1994) Finally, the exploration and transport of oil result in significant water-related impacts.10 For example, the exploration of oil and gas results in substantial leaking of petrochemical products and by-products from improperly sealed wells and pipelines (Miller 1993; NDRC 1997; Schwartz 1971). In addition, accidental spills resulting from the marine transport routinely dump vast amounts of oil into the sea. Between, 1972 and 1992 there were 437 significant oil spills (i.e., over one tonne) in British Columbia's Georgia Straight, mostly in ecologically sensitive areas (BC Environment and Environment Canada 1993).11 On a global basis, nearly 2.9 million barrels are spilled into the sea every year (Lowe 1990).12 Similar water impacts can be traced for the many other aspects of life cycle of cars from the mining of materials through to disposal. For example, mill tailings from mines leak heavy metals and acids into water and soils as does the seepage of effluents from the production of lubricants and petrochemical products used for automobile operation. 1 1 The number of spills decreased dramatically after new tanker technologies, stiff penalties and improved harbour traffic management was introduced. However, the frequency of accidents increased steadily since. BC Environment (1993) suggests this may be due to increased tanker traffic. 1 2 This is over 10 times the volume of the Exxon Valdez oil spill off the coast of Alaska in 1989. 32 Lowe also estimates that an additional six times this amount of oil is dumped into the oceans through the routine flushing of carrier tanks, road runoff and other petroleum by-products. Often, chemical dispersants used to break up oil spills do more harm than good (Schwartz 1971). While the chemicals themselves are harmful to marine life, the breakup of oil into tiny particles makes it easier for the oil to be assimilated. Again, all of the pollutants of these activities enter the cycle of marine and terrestrial life impacting the entire food chain. The complexity of treating these by-products of auto use and their many residual impacts precludes effective action. Land related impacts The consumption of semi-rural and rural lands for urban uses is noted as a serious environmental problem in most urban areas in the world (BCALC 1996a; OECD 1995; Wackernagel and Rees 1996). This is particularly in the case in North America and Australia where low-density patterns of peripheral urban development prevail (Altshuler 1979; BCALC 1993; Durning 1996; Freund and Martin 1993; Newman and Kenworthy 1996; OECD 1995). A review of the literature indicates that impacts of urban expansion fall into three general categories: • the consumption and loss of land used for food production and vital ecological functions; • secondary impacts associated with the loss of non-urban land and urban encroachment; and • the aggravation of auto-related impacts resulting from sprawling, low-density development. Loss of lands serving agricultural and ecological functions Urban growth is particularly problematic because many of the world's cities are located on, or in close proximity to, the earth's most fertile and ecologically productive lands such as low-lying tidal zones, river basins and the like. This is certainly the case for Canada's major urban centres. For example, the BC Lower Mainland region's agricultural productivity is the highest in Canada and twice the global average (Wackernagel and Rees 1996). Similarly, the most highly arable farmlands in Canada are situated just on the periphery of the country's major urban centres (Statistics Canada 1994). As urban encroachment consumes these lands, large portions are covered with impervious surfaces (roads, sidewalks, parking lots, buildings) while much of the rest is converted to uses such as lawns that are of marginal ecological or productive value. Urban growth in Canada in the post-war period has been characterized for the most part by urban sprawl: new low-density development on previously rural land on the periphery of the urban envelope (IBI Group 1993; Linteau 1990). The development of the Vancouver region is an excellent case in point. The following accounting of land use trends and impacts is drawn mostly from experience in the Lower Mainland, however the trends are similar, and perhaps more acute, in other regions in Canada. 33 Between 1961 and 1981 the GVRD's urbanized area grew at twice the rate of population growth (see Vancouver data table in Appendix 2). Today, the GVRD stands as the lowest density major urban region in Canada (Raad and Kenworthy 1998). While the inner area population of the GVRD has remained relatively stable, most of this new growth since 1961 has been accommodated on previously rural and agricultural land. By the early 1970's, prime agricultural land in the Lower Mainland was being converted to urban uses at a rate of approximately 6000 hectares a year, mostly within the Lower Mainland of BC (BCALC 1996b). In response to this massive loss of arable land, the Province of BC enacted the Agricultural Land Commission Act in 1973. The Commission created a bank of land known as the Agricultural Land Reserve and regulated the subdivision and development of agricultural lands within it. However, lands are still sub-divideable by application to the Commission requesting an 'exclusion' from the ALR. Much land has still been lost to development from the ALR since, particularly in the Lower Mainland. Since 1974, over 15,000 hectares have been converted from the ALR to urban uses (BCALC 1996b). Countless other hectares of rural land not registered in the ALR have also been converted. Of the land 'excluded' from the ALR since its inception, it is estimated that 60-65% was rated as "prime" or "prime dominant"13 agricultural land (Lew 1997). This loss of highly productive land from the ALR is not being replaced with similarly capable land. For example, for every 3.5 hectares of "prime" or "prime dominant" land removed from the ALR in 1994 only one hectare of land with similar agricultural capability was included in the reserve (BCALC 1997). Estimates from elsewhere in Canada confirm similar patterns. In Ontario, over 17,000 ha was lost to non-farm uses between 1981 and 1986 alone (Harcourt 1993) and substantial losses have continued since (Swainson 1998). Manitoba continues to lose an average of 1,215 ha of farmland each year due to subdivision on Winnipeg's periphery (Manitoba Environment 1995). Overall, it estimated that that up to 60% of urban growth in Canada's major urban centres has been onto high quality farmland (Zielinski 1994). There are many other ways in which a loss of agricultural capability affects Canadians: • reduced food security (Harcourt 1993; Wackernagel and Rees 1996). In 1990, 60% of BC's food supply came from BC producers (BCALC 1993). Assuming land productivity remains the same, an increasing population and a shrinking arable land base means that BC must import increasing amounts of food to meet it's local needs. Meanwhile, BC's existing out-of-province suppliers of agricultural products are facing even more severe pressures on their agricultural land base. California, BC's largest supplier, is losing over 20,000 ha of prime farmland to urban use every year (Harcourt 1993). Such trends threaten to reduce food self-The Canadian Land Inventory classification system for agriculture rates agricultural land according to its soil quality and agricultural capability. "Prime" and "prime dominant" are the most capable, followed by "prime subordinate" and "secondary." 34 sufficiency and increase vulnerability to vagaries of the global food market (i.e., price and quality uncertainties); • higher "ecological footprint" as a result of increased demand for imported food and commodities. Increasing imports of food requires energy for transport and distribution; • higher food costs; • loss of farm capital investment as land owners anticipate the sale and conversion of land for urban uses; • increased competition for surface and ground water resources (BCALC 1993); and • reduced viability of agricultural services sector as farms fall below a 'critical mass' (BCALC 1993). Despite losses of land since the ALR's creation in 1974, the ALR has proven to be successful relative to other provinces and American states in providing some level of protection for agricultural land. From the high of 6000 ha in losses of arable land in the years before the ALR, and the thousands of hectares lost in exclusions from the ALR since, conversions from the land reserve have diminished in the Lower Mainland to only 15.6 ha in 1994 and 357 ha in 1995 (BCALC 1997). Although BC's ALR is beginning to stabilize and show results, few other provinces, with the exception of Quebec, have agricultural land preservation regimes as comprehensive or successful. Notwithstanding the stabilization of agricultural stock in the Lower Mainland, pressures still exist to convert other non-urban land currently excluded from the ALR to urban uses. For example, substantial pressure has been building to develop the Bums Bog in Delta, the largest undeveloped urban landmass in Canada. Only 5% of Burns Bog is held in the Delta Nature Reserve, while most of the rest is privately owned. Proposals ranging from residential subdivisions to industrial complexes to golf courses have been brought forward for large portions of the site. This 4000 ha wetland serves many important ecological functions. It is habitat for a substantial and diverse population of flora and fauna. It is also an important stopover for migratory birds. Peatlands like Burns Bogs serve important regulatory functions in the environment. The sphagnum moss found in the bog can hold up to 30 times its weight in water (BBCS 1997), assisting in floodpeak reduction. Finally, Burns Bog and other peatlands are also important greenhouse gas sinks, providing long term storage for methane and C 0 2 which contribute to global warming (BBCS 1997; Miller 1993). It is estimated that peatlands hold up to twice as much undecomposed organic carbon than forests (BBCS 1997). Secondary impacts associated with land loss to urbanization There are many secondary impacts associated with this loss of land and concomitant low-density development. The following is a selected inventory of some of the secondary impacts: • ecological impacts from land coverage with impervious surfaces (i.e., roads, buildings, etc.). Alan Thein Durning estimates that just 15 percent of impervious coverage on watershed land is required to dramatically alter waterflow regimes and ecosystem balance (1996). For example, coho salmon are seldom found in streams when coverage exceeds this amount as 35 delicate food chains are disrupted and only the most hardy plants and insects remain (Durning 1996; Miller 1993). Large portions of the GVRD are built in the Fraser River catchment area and upon wetlands in both Richmond and Delta; • increased water consumption due to watering of large lawns (Durning 1996); • increased energy consumption and costs for heating and cooling of low density building (Durning 1996; Newman 1991; Pagani 1997). Newman (1991) estimates that heating higher density developments can use as little as 50% of the energy for heating as dispersed housing. Wackernagel and Rees (1996) estimate that choosing medium density housing over low-density suburban housing, combined with driving a compact rather than standard-size car, can reduce a household's housing and transport "ecological footprint" by a factor of three. • higher material and energy needs for the construction of infrastructure and buildings in low density development (Altshuler 1979; Miller 1993; Newman, 1989 #29 and 1996; Pagani 1997; Pushkarev and Zupan 1977; Wackernagel and Rees 1996). For example, road infrastructure in low-density developments is underutilized. Newman and Kenworthy (1991) estimate that up to three quarters of streets in North American cities are minor and local streets, while Cervero (1991) and Kenworthy (1986) have found that only 15-28% of travel occurs on these roads. In Ottawa, 71% of road network accommodates only 26% of vehicle kilometrage (RMOC 1994). Provision of sewers, utilities, poles, pipes and other infrastructure all have similarly high material and energy requirements per household in low density developments (Miller 1993); and • poor recycling rates in low density areas due to high collection costs (Newman 1991). Impacts due to increased auto use and dependence Many studies have shown that car use (Durning 1996; IBI Group 1993; Newman and Kenworthy 1989a; Pushkarev and Zupan 1977) and car ownership (GVRD 1996; JPINT 1996; Newman and Kenworthy 1989a; OECD 1995; Winnipeg 1995) exhibit an inverse relationship to urban density. As density decreases and land uses become segregated, transit provision becomes less viable and invariably falls. Automobiles are then required to meet accessibility needs and private automobile ownership increases. Lower densities also force longer and more frequent trips (BTS 1990?; Lowe 1990; Newman and Kenworthy 1996; Renner 1988; Yago 1983). Therefore, the environmental, social and economic impacts associated with increased motor vehicle operation and ownership are all aggravated by falling urban densities. Resource consumption Meeting the resource needs for automobile production and use through the entire cycle (from resource extraction to assembly) results in substantial degradation to the environment (see Freund and Martin 1993; Greenpeace 1992; Miller 1993). Generally, the impacts include: • degradation of landscapes and loss of ecologically productive lands due to mining; • pollution in the process of material extraction and processing; • energy consumption and pollution in transporting materials for manufacture; • energy consumption and pollution in the production automobiles; • disposal problems of spent automobiles. 36 Table 6 below details the breakdown of materials needed to manufacture an automobile. Many of the inputs in car production require the mining and processing of primary resources, while others depend on manufacturing processes (rubber and plastics, for example). Greenpeace (1992) estimates that 10% bf plastics production of industrialized countries is attributed to automobiles production. Metal requirements in cars require substantial mining of coal, iron ore, limestone, bauxite, copper, platinum zinc and lead. In 1990, the U.S. motor vehicle industry's national share of material consumption in the nation for various inputs was: 13% of steel, 16% of aluminum, 69% of lead, 36% of iron, 36% of platinum, and 58% of natural and synthetic rubber (Freund and Martin 1993). Ores are all non-renewable resources whose stocks are finite (Goodland, Daly, and El Serafy 1992; Wackernagel and Rees 1996; WCED 1987) and the known reserves of many of these are expected to be depleted within 25 to 100 years current rates of consumption (Freund and Martin 1993; Jacobs 1991). Table 6 - Composition of the average automobile (1990) Material Weight Weight (kg) (% of vehicle) Low carbon steel 530 46.3 Alloy steel 91 7.9 Cast iron 91 7.9 Aluminum 136 11.9 Copper, brass 6 0.6 Zinc 4 0.3 Lead 8 0.7 Other metals 16 1.4 Rubber 58 5.0 Glass 32 2.8 Plastics 105 9.2 Other non-metals 68 6.0 Totals 1145 100 Source: (Greenpeace 1992). From Henstock, 1988. Design for Recyclability. Institute of Metals: London. Furthermore, the mining, processing and refining of ore resources necessary for automobile production generate substantial environmental impacts. Abandoned mines not decommissioned properly can also leave a legacy of environmental problems for areas far removed from the source, well into the future (Manitoba Environment 1995). Table 7 below highlights some of the potential impacts of mining activities. With Canadian production of automobiles typically 15% of U.S. production levels (see Renner 1988), per capita consumption of natural resources devoted to automobile production is even higher in Canada than the United States.14 This estimate assumes car inputs and Canadian fleet production require the same average inputs as in the U.S.. Using estimated 1995 population of 263,437,000 (U.S.) and 28,537,000 (Canada), Canada's population is roughly 1 1 % that of the U.S. 37 The automobile manufacturing process follows the materials extraction process with substantial pollution and energy consumption of its own (Miller 1993; Moriguchi, Kondo, and Shimizu 1993). The US EPA estimates that transportation manufacturing is the fourth largest source of toxic chemical releases into the environment in the USA (in TJNEP 1993).15 From materials extraction, to material transport, to manufacture, to the showroom, the automobile acquires much 'embodied' energy. The OECD (1995) estimates that one quarter of the energy consumption in the life cycle of a car occurs before it leaves the showroom. Table 7 - Environmental impacts of minerals extraction Activity Potential Impacts Excavation and • destruction of plant and animal habitat, human settlements and other Ore Removal surface features (open pit) • land subsidence (underground) • increased erosion • waste generation (overburden) • changes in river regime and ecology due to siltation and flow modification • acid drainage and heavy metals contamination of lakes, streams and groundwater Ore • waste generation (tailings) Concentration • organic chemical contamination from tailings • acid drainage Smelting/Refining • air pollution (including sulfur dioxide, arsenic, lead, cadmium and other toxic substances) • waste generation (slag) • impacts related to producing energy for smelter operation (depends on power source) Mine • abandoned equipment, plant and buildings Abandonment • release of methane from mine • leaching of pollutants Source: (UNEP 1993; Young 1992) The actual use of automobiles is the next stage in the life cycle where considerable resources are consumed and concomitant environmental impacts are experienced. The provision of infrastructure, such as roads and parking, to service the automobile requires the mining of gravel (to produce concrete), the production of asphalt (which requires oil), the manufacture of construction vehicles (with similar impacts as noted above for cars) and the consumption of non-renewable energy. Al l of these are material and energy intensive as well as polluting. Gordon (1991) estimated that one-third of all transportation energy This estimate does not include toxic chemicals released from other industries that supply the auto-manufacturing complex and for servicing auto use (e.g., machinery, plastics, petroleum, metals, chemicals, electrical, etc.). 38 (or 14% of all energy) in the US is consumed in these ancillary activities. Table 8 below demonstrates the relative energy intensity of various transportation-related activities. Table 8 - Energy use of transportatio n related activities in the U.S.. 1985 Activity . Energy (Quads) Automobile use3 6\8 Light truck/van 4.1 Infrastructure repair 1.7 Infrastructure construction 4.8 Producing, refining and distributing fuel 3.4 Transit bus use3 0.07 Source: (Gordon 1991) Note: a. 1987 data Of course, the use of internal combustion engines for transport also necessitates direct consumption of fossil fuels. This is perhaps the most recognized resource consumed by transportation activities. Transportation is 97% dependent on petroleum as its source of energy (Gordon 1991). In Canada, 68% of all transportation energy used is gasoline, primarily for the operation of private vehicles (Gordon 1991).16 In all OECD countries, petroleum consumption is rising by an average of 1.5% annually (OECD 1995). While there are important implications of fuel consumption in terms of pollution (see section below), there are other significant implications in terms of resource scarcity and geo-political and economic instability (Fleay 1995). Oil and gas are fossil fuels and, as a result, cannot be regenerated within a human-time scale. They are therefore considered finite, non-renewable resources. While the finite-nature of oil resources is not contested, the relevant supply is (The Economist 1997). Over the past few decades, the reserve supply of oil has been constantly revised upward to account for new discoveries. However, much research has pointed to a sharp decline in the frequency, size and economical viability of new oil discoveries. Fleay (1995) estimates that at current rates of exploitation, reserves of oil will be depleted within 50 years. While it has been argued that new discoveries may continue to extend this depletion deadline (The Economist 1997), these resources nonetheless almost certainly face exhaustion in the foreseeable future at current consumption rates. Continued dependency on oil brings economic and geo-political uncertainty. Most OECD countries are currently net importers of oil and therefore various sources are not secure. In recent years, this uncertainty has resulted in economic instability (Kenworthy et al. 1997) as well as wars (Greenpeace 1992). Fleay (1995) and Hart and Spivak (1993) argue that the real concern in the medium term is not Private vehicles primarily use gasoline. Diesel and LPG fuel is primarily used for commercial transport (trucks, buses, trains, taxis and ferries). Therefore, gasoline consumption can almost entirely be attributed to private vehicle 39 that all oil stocks will be depleted, but that the surplus extractive capacity will disappear.17 When the 'consumption' curve meets the 'supply' curve (expected within 25 years), oil-consuming nations will be 'captive' markets. The perpetuation of geo-political and economic uncertainty is likely to be aggravated as this production surplus is depleted and dependence on imported oil increases. Comparative evaluations of various transportation modes demonstrate the high resource intensity of automobile production and use. Pendakur, Badami and Lin (1995) describe costs and benefits of non-motorized travel in terms of "bicycle equivalents." They show that the single occupant vehicle has 48 times the material requirements, 20 times the space requirements and 60 times the energy consumption18 of the bicycle or bicyclist on a per unit basis. The much higher use and weight loads of automobiles on roadways have considerable implications in terms of secondary materials and energy consumption. Whitelegg (1993) also shows that the SOV has inordinately high space requirements. Whitelegg estimates pedestrian, cyclist, rail transit and SOV space requirements at 0.8, 3, 1.5-4.6 and 60 M 2 per person, respectively. While per unit resource costs are extremely high for automobiles, their disproportionate share of use in absolute terms makes these figures even more concerning. Clearly, motorized transportation requires the consumption of substantial resources for the manufacture and servicing of the automobile. Vehicle disposal The disposal of vehicles and vehicle parts has substantial environmental impacts. These impacts include space consumed for dumps, toxic leachates from automobile parts and residual fluids, and impacts due to accidents. Automobile dumping represents a serious problem globally (Ginley 1994; Greenpeace 1992). Currently, in the United States, approximately 10 million cars are retired every year. Of this amount, approximately 71% of an automobile's gross vehicle weight is recycled (AAMA 1997).19 This translates into the rough equivalent of 3 million cars, by gross vehicle weight, still being disposed of in landfills annually. In places such as British Columbia, where per capita vehicle ownership approaches U.S. levels (Raad and Kenworthy 1998) and where 60% of the province's landfills will reach capacity by 2000 (BC Environment and Environment Canada 1993), such waste is of serious concern. Environmental pollutants resulting from vehicle and vehicle component disposal are also a problem. Leachates from metals, batteries and plastics (Beaumont 1993; Greenpeace 1992) elevate concentrations of lead, zinc, cadmium and other heavy metals and toxins. The disposal of tires also use. In 1986, this consumption amounted to more than 28 billion litres of gasoline out of 42 billion litres of petroleum. 1 7 They examine the famed "Hubbert Curve" which traces oil discovery and production levels since the early 1900s. The Hubbert Curve indicates oil discoveries have peaked and that we are now in an era of declining reserves. 1 8 Furthermore, the energy consumed by the bicyclist is human energy and therefore non-polluting. 40 results in toxic leachates entering the soil and water as well as the threat of potentially toxic emissions from tire fires. These all pose a threat to local human and ecosystem health. Day et al (1993) show that leachates from tires can be lethally toxic to aquatic biota.20 Tire stockpile fires emit a large amount of semi-volatile organic compounds as well as zinc and lead (Lemieux and Ryan 1993). In 1990, a dump of 14 million tires in Hagersville, Ontario burned for over two weeks resulting in the release of high levels of toxic organic contaminants to the air, soil and water runoff (Environment Canada 1991; Steer et al. 1995)21. Some argue that recycling automobile parts "saves energy and conserves resources" (AAMA 1997).22 While this may be true from a limited perspective, it is not true from a more holistic one. First, the recycling of automobile componentry (namely, metals) involves yet further energy intensive and polluting processes to reintroduce them for commercial uses. Second, substantial amounts of virgin material are still required in addition to the recycled material to make complete products. Third, many recycled materials are only suitable for lower-grade uses and repeated recycling can undermine the structural integrity of certain metals. Finally, in the same way that pollution abatement technologies and improvements in vehicle efficiency can be overcome by growth in VKT (Freund and Martin 1993; Hart and Spivak 1993; Lowe 1990; Renner 1988), gains from the more efficient utilization of resources can be overwhelmed by increasing growth in auto sizes and ownership levels. For example, since the early 1990s, the trends in new vehicle ownership have tended towards larger vehicles such as small trucks and sport utility vehicles (Gordon 1991; Renner 1988), which require greater resources in production and use. If developing nations were to develop an appetite for automobile similar to levels found in Canada, any gains from efficiency would quickly be overwhelmed by total increase in material demand for new cars. For example, in 1990, China's had about 1.6 million buses and cars available for passenger use, or about two vehicles for every 1000 people (Hook 1998). If car ownership were to rise to Canadian levels, China's vehicle fleet would grow by over six hundred million vehicles, roughly equivalent to the entire current global car fleet. Although the recycling of vehicles and the more efficient use of resources is desirable, they offer no panacea for the impacts incurred in through production, use and disposal. In order to truly realize the efficiency benefits that recycling offers (i.e., lower consumption of energy and resources), efficiency 1 9 75% of the average vehicle is recovered, however 5% of total retired vehicles are not recycled at all. 2 0 Used tires demonstrated the highest levels of toxicity. 2 1 Over 12.6 million tires were consumed and there was substantial ground and surface water contamination (Environment Canada 1991). 22 Our Common Future (WCED 1987) also calls for increases in efficiency as a means of achieving "sustainability." 41 savings must be "captured" (Wackernagel and Rees 1996) through stabilized or reduced car ownership levels. Says Jane Jacobs of the squandered efficiencies offered by cars: "Automobiles are hardly the inherent destroyer of cities...[they were] potentially an excellent instrument for abetting city intensity, and at the same time for liberating cities from one of [the horse-and-buggy's] noxious liabilities... We went awry by replacing, in effect, each horse on the crowded city streets with half a dozen or so mechanized vehicles, instead of using each mechanized vehicle to replace half a dozen horses " (1961, 343). Airborne vehicular pollutants and emissions The operation of vehicles results in vehicular emissions that harm the environment in many ways both at the local and global levels. Some pollutants are directly due to the internal combustion process, others are due to other aspects of vehicle operation (such as air conditioner operation) and others are the result of the reactions between vehicle emissions in the atmosphere. Emissions and pollutants that result from vehicle use that are of local concern23 include particulates (PM10, PM 2.5), sulfur dioxide (S02), carbon monoxide (CO), oxides of nitrogen (NOx), volatile organic compounds (VOCs) (also known as hydrocarbons, or HCs), tropospheric (ground level) ozone, total suspended particulates (TSPs) and lead (though lead is no longer a major auto tailpipe pollutant in North America). Local scale air pollutants have implications for air, water and soil pollution and, therefore, human and ecosystem health (French 1990). Emissions of global significance are greenhouse gases (GHGs) which include carbon dioxide (C02), nitrous oxide (N20), methane (CH4) and chlorofluorocarbons (CFCs), all of which contribute to global warming (BC Environment 1995b; IPPC 1990). In addition to being a GHG, CFCs also contribute to ozone depletion (Bovard-Concord and ARA 1994; Vancouver 1990). Automobile air conditioners are the largest single source of CFCs in British Columbia (BC Environment 1995a) and account for 23% of all CFC releases in Canada (Environment Canada 1993). Table 9 below shows the proportion of selected emissions in the GVRD that are attributable to mobile sources,24 with a breakdown of the road motor vehicle share. Over 76% of total emissions in the GVRD come from private vehicles alone. Similar shares of motor vehicle pollution levels can be found in other North American and OECD cities (Gordon 1991; OECD 1995). Table 10 below outlines the causes of various vehicular emissions, their environmental and health implications, and the spatial scale at which the implications are experienced25. Most of the pollutants result from the internal combustion Some local pollutants also have global significance. For example, ground-level ozone (formed by the reaction of NMHCs and oxides of nitrogen) also aggravates, and is aggravated by, global warming. 2 4 All transportation including motor vehicles, trains, aircraft, marine vessels, off-road equipment. 2 5 Health implications will be discussed in subsequent sections. 42 process. The environmental implications of these are well documented in the literature. They include smog, acid rain, ozone depletion and the enhanced greenhouse effect and many associated secondary impacts. Table 9 - Emissions from transportation in the GVRD (% of total. 1991) Transportation Contaminant Emissions Source NO x VOC S O x CO PM Total Light duty vehicles 41% 47% 6% 91% 6% 76% Heavy duty vehicles 16% 2% 8% 2% 9% 4% Other transport sources 25% 4% 18% 4% 5% 6% Total transport 82% 53% 32% 97% 20% 86% Source: (ARA and BOVARD-CONCORD 1994) Many of the impacts confound one another and preclude, or are aggravated by, technical fixes (Gordon 1991; Lowe 1990; Schwartz 1971). While technologies serve to reduce certain emissions, the primary determinant of absolute motor vehicle pollution levels in a city is the amount of vehicle kilometers travelled (Gordon 1991; Newman and Kenworthy 1988a). Newman and Kenworthy (1988a) show that those cities with the freest flowing traffic actually have the highest gasoline consumption. Ironically, Many of these cities also have some of the strictest emissions standards in the world. The higher fuel consumption and emission levels result from longer and more frequent trips as well as induced traffic effects, which far outstrip any gains from increased fuel efficiency or temporary congestion relief. Just as previous gains in efficiency were squandered by VKT increases in the 1970s and 1980s, so too will future advances. ORTEE (ORTEE 1992) estimates that despite improvements in vehicular fuel efficiencies, transportation emissions of S0 2 , PM, and C 0 2 are all expected to rise 60, 50 and 29 percent, respectively between 1988 and 2005 in Ontario. While NOx, CO and VOC are all expected to decline between 1988 and 2000, they are expected to begin rising again after 2000. 43 » 8.1 Q E % *: , Q . ° CO CO Q. c o "-4—» CD o Q to c 03 L -c CO J D E o CO c o 'to to 'E a> X5 c c o c o o Q < i o 10 c t CO s °-£ E > c UJ o g s = co o 81 UJ Q-0) c — o CO CD 2 CO C — ° 2 CO CO )US ral X) ' > an CO co" 'c JD — CO CL 0) cz cu Q. o CO (0 TJ 3 C CO CO CO to" CO CD c CO o f~ CO o p CJ o 2 c c JD CO a ° CO CD CU > LT 2 O ^ , f2 ? 3 g CB C o cu JO cu cu CO i I CO 3 <; o cu CD N ? CO CL 8 2 82 co o o co OJ c CL — .!= 3 O i l i f f £ T5 2 > 3 3 CU 2 JD te > O P CO C c C C O <= . - ° E — co cj CO CO C CD _ CJ CO 2 TO O CL O 00 8E. to .y o cu OJ co cu XI 'x o c cu OJ o cu E CO tz CO c m ° 3<2 c 8 I . - JD > cu u, a> co $ x> co 3) — cu ~ co .<2 E QJ TJ P x-•C Q. £ S E S 2 | CO c CO • -CD CO •S <0 CO CU •e.s CU CO 3 2 0 ) T J CD u) — m CD 3 g co c CO g g o w t ' « cu CM S -32, s-g g,|cj 1 8-i I CO 5- cu CD. CO JZ Ci | O) 2 CD o r £ <= co cu ^ £ J D O § I 'S'S co c „ e o 2 co o "m co > , 2 2 CO cu o E £ .E E p • s o y or 8§ 8 o c _ « W CD </) CD - 8S > cu = 5. eg (i OJ c c cu 8 !§^^ a to 5 ~ c CU J 'co I1 c E 2 i f f ' 08 o aj co 5 JB s c c a> « co" •2 o cu 25 E 0 £ co 9 oJ OJ S o . . co 0 M 1 g 2 J= . 0 c c" 'co 'co <* 8,1 3 2 *- TD O c JD CO £ c I1 ^ C CO 5 . 2 E — "eq -2 o 7 3 co 92. • -8 9>'co co JZ CO CO o •g "o 0 'o co co E -  2.S l i . cu >. cu co" CO CU OJ c c CO JZ 5 gj i § cu c o N o o c OJ j r CL CO o CL o cu cu co_ o i2 3 o cu a. o o N CU O JD o CL CO > aj .3 8 1 "5,0 CO C — m ° 2 8'g»o o HI — T3 C CO TD JD i_ O E cu, i i 2 c?1 co 2 •C CD § 1 JD ^ E 8 o E a3 O CO o cu •o Q-aj CO ? t OJ CO CL CU 83 52 3 iS ' t : O CO I— 0-QJ T3 £ 0 cu -a C CO 2 2 =^ co OJ ;= a c _^ 3 5 E o> p OJ-£ OJ si 1 o — &i cu cu S 2 •Q. CD 8 i as to <o ^ 1 o £ 1 2 cu o CO [#I § E S.1 co c 8 a, I — OJ *= .« g i co £ _ CU CO nj E TJ J l CD. O o u S 2 ^ to CO O JD s-S I i l ls CO CO CJ _ I J : O — 8 S o » So or z O OJ •c o a> E 8 s I- o co 3 o ® s c E 2 Q.I co o CL CU E 3 . . XJ >< 2 2 c S«E 2 E O CD cu co' 'o o c iS c CO XJ JZ c o CO "o fe § i i - -4—' co ^ CO OJ| c I cu £ CO cu ~, > §1 "i — o co E « CD C CD cu XJ c CU 3 as E 2 inc es q CO O CO CD ntia repr 05 P co" CL cu 0 ° > CU .£ to c s E E °» 2 E cu c "55 >* co co 5 =5 CO CO CD fO .CO co * — c CD XJ 3 c O - CO 2 "co c tter CD L . O ultu pa E nc 2 OJ CO IS > . _ E CD co in cli sea even CO OJ i_ c cu »- CO CL CO o> E _ ±3 <o E 00 >> CO CO CD XJ O o c o O CO CD CO O •2 xi c CO X J w CO 3 a> £ — CO 12-2 80 S 2>o • -iS co — u CO CD CD CD £ O U ^ 2 2 CD CL CO CO O —• so 8° co co 8 CD JZ co ±= CO _ 2 O c CO 2 o •i= eg t i c g c o 2 8.2 2 1 t « S-2 3 > 2 2 "to 3 o OJ c ^ co 8 g x> 2 c .-t± CO XJ .is "CO .!= CD 2 O o O O o § 2 O 2 Si CD p 3 *-JD .CO CD p Qi O > OJ OJ O E E E E E o k_ CD CO -7 fo 0 JQ gS 2 2 o OJ OJ r-co ^« CO •4-* CO o OJ c 'OJ CO co" co CD o o CO CD XJ 'x o CO i o CD > 3 O O c CO CD > is-£ 8 0 1 CN CO S cu ? £ cu c cj — CO CO, S c 0- co §3 £0 CJ) CL CN ^ OJ £ OJ o o S m 1 cu JD CO OJ OJ CO 0- § LU *= CJ -1 ro 2 OJ o OJ *i 3 CO CD o LU s 8 OJ ™ •<- XJ §1 3.1 CD O OJ JD ° > CD = CO 2 OJ "I •£ TO 2 LL -g CD < O . . D S o o CO z ^r Other environmental impacts There are several other significant environmental impacts associated with automobile dependence that warrant mention. These include: • heritage and architectural and loss and damage. Buildings, monuments and heritage sites incur substantial structural and surface damage due to pollution and vibration (Gratz 1993; Miller 1993; Newman, Kenworthy, and Vintilla 1995). • noise pollution. Road traffic is regarded as the most common source of unwanted noise (Morton-Williams et al. 1978 in OECD 1995). The Canada Mortgage and Housing Corporation (CMHC) maximum acceptable outdoor noise level is 55 decibels.26 However, traffic noise in medium and large cities routinely exceeds this level (OECD 1995). The effects of noise are for the most part subjective (Miller 1993). Noise results in a loss of environmental amenity and psychological well-being. This, in turn, can result in health, sleep and productivity losses. Excessive transportation noise is cited as a major factor in the decision not to walk to a destination (Energy Probe 1989). Traffic noise can is also a major disturbance to Wildlife. Reijnen et al. (1997) indicate that traffic noise is the most critical factor in reduced wildlife densities and bird breeding in broad zones adjacent to busy roads. Substantial reductions in engine and transmission noise emissions are unlikely to significantly mitigate total noise due to increasing vehicle volumes, increased stop-starts and because a considerable portion of the noise attributable to driving is due to tire contact with the road surface (OECD 1995). • wildlife deaths. Every year in North America alone, millions of large mammals and countless lower order species are killed by motor vehicle collisions. The 1991 U.S. "road-kill" total just for deer is conservatively estimated at half a million (Romin and Bissonette 1996). • habitat disruptions and loss due to roads. Disruptions to wildlife habitat are substantial. For example, large carnivores are often 'keystone' species on which ecosystem balance depends. However, roads are a major threat to carnivores, particular endangered species in recovery, because of road barrier effects, vehicle collisions and increased accessibility to poachers (Noss et al. 1996). Fragmentation of habitat threatens many species that depend on a large range. Reeder al. (1996) examined fragmentation in over 30,000 ha of Rocky Mountain habitat and found that fragmentation from roads was 1.5-2 times worse than forest clearcuts in terms of converting interior habitat into edge habitat. As mention in sectioned above, road construction and urbanization also result in substantial wetland loss and disruption. Wetlands play vital ecological roles in terms of diverse species habitat, shoreline stabilization, groundwater recharge, food/nutrient production, and toxin/pathogen retention (De Santo and Flieger 1995). 2.3.1 Social Impacts of Auto Dependence Health Auto dependence impacts the human health through: fatalities and injuries attributable to collisions involving motor vehicle; increased sickness and death due to pollution; and a more sedentary lifestyle which results in increased risks of illness and a loss of productivity. 2 6 See Barron Kennedy Lyzun & Associates. 1991. LRT SYSTEM NOISE STUDY: Sound Level Measurements Made along the Existing SkyTrain Guideway. Prepared for BC Transit: October 21, 1991. 45 Injuries and deaths resulting from motor vehicles are so commonplace that their magnitude is often forgotten. In Ontario, over 1,200 people are killed and 120,000 are injured annually in collisions involving motor vehicles (Zielinski 1994). In British Columbia, the statistics are equally grave. In 1995, 493 people were killed and 47,472 were injured (ICBC 1997). While total collisions have been in decline in British Columbia in recent years, pedestrian and cyclist collisions with motor vehicles have been on the rise. OECD (1995) indicates that although many countries with high traffic volumes have low fatalities when expressed in terms of deaths per vehicle kilometre, pedestrian traffic fatalities per capita are lower in countries with less vehicle travel.27 Furthermore, they indicate that while cyclists are 9 times more likely to be killed than a car driver, car drivers are 13 times more likely to be involved with a traffic fatality.28 Therefore, rather than encourage modes that pose the largest threat to the general public, policy should focus on encouraging modes that pose the least threat to other road users (Hillman 1992 in OECD 1995). These less threatening modes include cycling and walking. Air pollution is another source of morbidity and mortality related to auto use. Table 10 above highlights some of the health impacts associated with various pollutants. The health impacts range from direct illness and death from pollutants to lowered immunity, which results in indirect illness and death. The old, young and those with pre-existing medical conditions are particularly vulnerable. There has been extensive research establishing a strong causal link between air quality and health. For example, Delfino et al (1994) found a positive relationship between photochemical smog levels and hospital admissions for respiratory illnesses. A broader based study of 16 Canadian cities by Burnett et al (Burnett et al. 1997) found a similarly strong positive relationship with photochemical smog and respiratory hospitalizations as well as between particulate matter and CO concentrations and hospitalization. Proximity to traffic and therefore exposure to higher levels of ground-level pollutants can result in chronic respiratory ailments (van Vliet et al. 1997). Finally, transportation-induced stress can have significant impacts on the health, quality of life and employment productivity of an individual. Raymond Novaco, a psychologist at the University of Californ ia, has done the most extensive research in this area (for example, see Novaco 1989; Novaco 1992; Stokols and Novaco 1981). Novaco's work focuses on measuring the dimensions of physical travel impedance.29 High levels of impedance are associated with high blood pressure, low tolerance for 2 7 This applies to developed (OECD) countries. Litman (1997b) also reports that accident rates and fatality risk highly correlate to distance travelled. 2 8 This indicates that although a motorist may be less likely to die from an accident, they are more likely to cause a fatality. 2 9 Physical impedance measures the distance and time spent on a journey, as well as the number of roads and freeways travelled on a trip. 46 frustration, family tensions, negative moods and illness. The symptoms are especially acute in women. Many of the effects on health and wellbeing are realized over time and repeated exposure to high levels of physical impedance reinforce and aggravate them. Furthermore, these impacts spill over to employers and are manifest in illness-related absence from work, high employee turnover and reduced productivity and morale. Equity Transportation infrastructure planning and funding in Canada favours automobiles disproportionately over transit and non-motorized modes. Few transportation users, with the exception of pedestrians and cyclists, actually pay an amount close to the full cost of their transportation choice.30 This results in an inequitable distribution of transportation costs and benefits between users. In most developed cities, automobile users generally receive the highest subsidies of any transportation system user. A heavy bias towards subsidized automobile travel for leads to 'irrational' consumer choices and an aggravation of auto dependence. The lack of funding for viable alternatives means that those unable to afford automobiles enjoy lower levels of accessibility to services and economic opportunities (Altshuler 1979; Litman 1997c; Yago 1983). In some cities (such as Detroit and Houston), this bias altogether eliminates transportation choice, effectively forcing the use of cars despite affordability (Newman and Kenworthy 1989). In these situations, the users least able to afford transportation services or are forced to spend higher proportions of their disposable income to meet their basic access needs. The imbalance between subsidies for automobiles, transit and non-motorized transportation is well documented and quantified in the literature (Altshuler 1979; Delucchi 1996; Kenworthy et al. 1997; Litman 1995; Litman 1998a; MacKenzie, Dower, and Chen 1992; Miller 1993; Yago 1983).31 Todd Litman (1997c), has done perhaps the clearest work in the area of defining transportation costs as well as their distribution and equity implications. Litman identifies three types of equity well known to economists as being relevant to transportation: 1. Horizontal Equity - equity between individuals who have comparable wealth and ability to pay. 2. Vertical Equity with Regard to Income and Social Class - focusses on the allocation of costs between different income and social classes. 3. Vertical Equity with Regard to Mobility Need and Ability - focusses on whether an individual is relatively transportation disadvantaged. 3 0 Transportation decision involve two broad categories of costs: internal (those imposed and paid for directly by the individual) and external (those imposed by the individual, but paid for by society at-large). These will be discussed at length in section 2.3.2. 3 1 This section is primarily concerned with the distribution of transportation costs and benefits. The actual economic costs and benefits of various modes will be discussed in section 2.3.2. 47 Horizontal inequities result between users of the same mode as well as between users of different modes, regardless of income. In the case of auto use, many costs are fixed and shared amongst users, regardless of distance travelled or size of the vehicle. Those who drive less or drive smaller vehicles, for example, effectively "cross-subsidize" those who drive more. The same phenomenon occurs with transit systems that have fixed or semi-fixed fare structures.32 Suburban bus riders who make longer trips on buses with relatively low loads are effectively cross subsidized by urban riders on denser routes paying the same fare. Recent attempts at fare reform in Vancouver have alleviated these inequity to a certain extent (Bohn 1997), however suburban and longer-distance travellers still underpay. The more substantial horizontal inequities lie in the degree to which the various modes are subsidized vis a vis one another. For example, cyclists and pedestrians pay almost all of the costs of their transportation out-of-pocket. Meanwhile, all of the motorized modes have 'external' costs not paid for by the user, but shared by society at-large (Bohn 1997; Litman 1998a; MacKenzie, Dower, and Chen 1992; Peat Marwick Stevenson & Kellogg 1993). Litman (1995) notes that while the average cyclist tends to overpay for transportation infrastructure, motorists in similar socioeconomic circumstances will tend to underpay. The GVRD estimates that motorists received $2.7 billion in subsidies33 in 1991, while transit received $360 million and non-motorized transportation received just $2 million (see Table 11 below). Furthermore, the report estimates that cars accounted for 76, 87, 96, 98 and 99 percent of time, social, infrastructure, sprawl and parking costs, respectively. In terms of horizontal equity, motorists, transit users and pedestrians and cyclists of similar socioeconomic standing all impose different degrees of external costs. Table 11 - Subsidies to transport in th e BC Lower Mainland. 1991 Total Subsidy Subsidy Subsidy % of Subsidy per capita per pass, km total cost (millions) Automobile $2,654 $1,507.00 $0.15 Transit $360 $204.00 $0.24 Non-motorized $2 $1.13 Source: Adapted from (Peat Marwick Stevenson & Kellogg 1993) While the imbalances in the allocation of subsidies between modes leads to horizontal equity, they also lead to vertical inequities with respect to income and class. Non-drivers tend to earn less money and therefore spend a higher proportion of their disposable income on transportation (Altshuler 3 2 Some exceptions are made on some transit systems for "youth" and "seniors" fares. 3 3 Many external costs attributable to automobile use and ownership are not included in this estimate. Despite this, automobile transportation accounts for 85% of all non-operating costs for transport in the GVRD. 23% 37% 8% 48 1979; Blumenberg and Ong 1997; Haines 1978; Schrecker 1996; Yago 1983). Typically those who earn less and have minimal access to a car (the transit and non-motorized "captive," including the young, disabled, elderly, poor and women) are most greatly impacted (CUTA 1991). In this way, the distribution of transportation costs is quite regressive. Not only do cyclists, pedestrians and transit users pay a disproportionate share of costs, they pay even more as a portion of their income. Table 12 - Auto ownership in Ontario, by household income. 1993 Automobiles Household % Owning % Owning Income Range One Two + Under $10,000 39.6 7.0 $10,000-14,999 40.8 5.0 $15,000-19,999 55.1 6.3 $20,000-24,999 63.2 9.1 $25,000-29,999 61.9 13.7 $30,000-34,999 60.9 15.7 $35,000-44,999 63.0 21.0 $70,000 & over 46.2 46.9 Source: (RMOC 1995) While auto operating costs have increased 12.2% between 1990 and 1995, transit fares have risen over 34.5% (Pucher 1998). This persistent underpricing of auto use also leads to longer term changes in urban structure that preclude other transport options and exacerbate auto dependence (Newman and Kenworthy 1989a). With transit and non-motorized modes not viable or unavailable, many households effectively require an automobile to access basic services and economic opportunities. The Regional Municipality of Ottawa-Carleton (1995) estimates motor vehicle operating costs at approximately $3,000 (excluding capital costs and depreciation) per household per year. Table 12 above shows that automobile ownership is still very high amongst lower income households. Obviously, auto ownership for these households is a considerable financial hardship. Vertical inequities extend beyond hard financial costs. One way this is manifest is in the disproportionate cost of environmental impacts borne by those on lower incomes. Urban property values are generally inversely related to air quality, noise and traffic volumes (Schrecker 1996), therefore those who drive less (and make less) are also more likely to be subjected to higher levels of these external costs. Impacts are also disproportionately distributed through systemic biases towards motorists (generally, higher income earners) in transportation system design. For example, Coffin and Morall (1995) attribute substantial difficulties for the elderly in crossing roads in Calgary to poorly designed crosswalks, insufficient crossing times on signals and the barrier effects of traffic. Also, transit system scheduling provides the highest level of service for peak period CBD inbound and outbound trips. This 49 favours commuters who tend to have higher incomes and secure employment. Meanwhile, lower income individuals (who have lower labour force participation rates), the disabled and the elderly, whose trips are generally non-CBD focussed, all tend experience less convenient and less frequent transit services (Altshuler 1979). Finally, vertical inequities with regard to mobility need result because those marginalized by exclusive transportation planning have difficulty accessing the employment, services and social opportunities necessary to live productive lives. Those most typically affected include the young, poor, disabled, elderly and women, who all have low car ownership rates (Altshuler 1979; Calgary 1994a; Engwicht 1993; Schrecker 1996; Yago 1983). These groups have been called the transport disadvantaged (Litman 1997c), the transport deprived (Altshuler 1979) and the access-to-exchange disadvantaged (Engwicht 1993). Women are especially disadvantaged with respect to access and mobility (Mensah 1995; Schrecker 1996). Schrecker indicates that women, particularly single mothers, have particularly demanding transport needs between child-care responsibilities and employment, however, they are especially likely to be transit captive. Altshuler (1979) asserts that transport deprivation is a cause of unemployment and poverty since securing affordable housing often means locating in areas with low access to services and economic opportunity. Recent research from highly auto-dependent regions in the U.S. has found that welfare recipients are confined to labour market areas that are one quarter the size of labour-market areas available to the general population due to poor housing location, poor transit service and low car ownership (Blumenberg and Ong 1997). Philp (1997) reports that the lack of access to, and affordability of, public transit in Toronto increases the hardships imposed on the homeless and actually aggravates homelessness itself. The homeless in Toronto spend two hours a day walking in order to secure basic shelter and food, with foot problems being a critical health issue. Furthermore, the large portion of time spent securing basic needs and the lack of access to transit means that there is simply not the time or ability to access the more advanced health and social services necessary to make a permanent move from the streets. Decaying urban fabric "Traffic arteries, along with parking lots, gas stations and drive-ins, are powerful and insistent instruments of city destruction. To accommodate them, city streets are broken down into looses sprawls, incoherent and vacuous for anyone afoot. Downtowns and other neighbourhoods that are marvels of close-grained intricacy and compact mutual support are casually disembowelled...City character is blurred until every place becomes more like every other place, all adding up to Noplace" (Jacobs 1961 p.338). 50 Domination of traffic and spaces given over to the car for roads and parking reduces the quality and amount of space dedicated as 'public realm' for human exchange and interaction (Appleyard 1981; Engwicht 1993; Kunstler 1993; Newman, Kenworthy, and Vintilla 1995; Yago 1983). The evisceration of neighbourhoods by freeways and major road projects to service car use has been most acute in the United States, where inner cities were blighted by a retreat to the suburbs (Leavitt 1970). However, neighbourhoods in several Canadian cities such as Montreal, Ottawa, Toronto and Calgary have also experienced similar damage accompanying major road projects. These projects usually cut through poorer neighbourhoods and bring visual intrusion, pollution, noise and unsafe streets (Appleyard 1981; Engwicht 1993). In Canada, the high costs of freeway projects brought major protest movements in most Canadian cities in the 1960s and 1970s. For example, communities mobilized against the Spadina Expressway in Toronto and the Strathcona Freeway in Vancouver, preventing their construction (Newbury 1989; Nowlan and Nowlan 1970; Pendakur 1972). Few central city freeway projects have been completed in Canada since. However, major road expansions still routinely occur within central cities in Canada, as do major freeway projects on the urban periphery. The expansion of Pacific Boulevard in central city Vancouver and the construction of Highway 407 on the outskirts of Toronto are examples. The amount of urban space and the intensity of auto traffic on roads reduces the amount of social space available in a city and constrains what David Engwicht calls "access to exchange" opportunities (Engwicht 1993). Wide streets and ample parking lots consumes land that could otherwise be used for socially productive purposes. Not only do cars consume "exchange" space, they have a "zone of influence" that increases with the speed and volume of traffic, reducing the effectiveness of the exchange space that remains (Engwicht 1993). In Donald Appleyard's studies of traffic on residential streets, he found a strong inverse relationship between traffic volumes and the amount of social interactions on the street, particularly amongst the young and the elderly (Appleyard 1981). Essentially, higher volumes of traffic on the street forced a continuing rollback in residents' perception of their home territory range, thereby reducing social exchange. Isolation Many authors have commented on the impact of segregated land uses on the isolation and alienation for those who lack the mobility that the automobile offers (Jacobs 1961; Newman and Kenworthy 1989a; Whitelegg 1993; Yago 1983). As previously mentioned, those who are transportation disadvantaged are relatively immobile and are only able to access services and social opportunities within walking distance. Within many suburban subdivisions this leaves little opportunity for human interaction 51 within the community. Furthermore, large amounts of time spent commuting means fewer opportunities for social exchange within families (Whitelegg 1993). Fear of crime and assault is another concern associated with transportation, particularly for women. Walking to, and waiting at, public transportation stops is a major concern where public visibility is low. Low density and isolated developments ensure wait and walk times are long, thereby increasing perceived vulnerability. Rosenbloom and Bums (Rosenbloom and Bums 1994) show that safety is a major influencing factor in womens' decisions to drive alone rather than take transit. Dysfunctional social behaviors One emerging social concern that has received only cursory academic attention is the relationship between anger and aggression and their relationship to driving. As congestion increases, so too does driver frustration (Novaco 1991). One manifestation of this frustration is a phenomenon called "road rage" - behaviours ranging from vehicle obstruction, to obscene gestures, to physical assault. Of course, the latter is of greatest concern. One survey in the UK found that 90% of drivers had experienced road rage incidents, while 1% of drivers claim to have been physically assaulted by other motorists (Joint 1995). Many of the assaultive behaviours exhibited with road rage are traceable to "disinhibitory" factors unique to driving an automobile (Novaco 1991). These include mass media imagery popularizing automobile machismo, the anonymity of highways, the protection offered by cars and the opportunity to escape quickly. Novaco indicates that, combined with these aggression disinhibitors, higher blood pressure, increases in negative moods and lower tolerance for frustration, all conspire to trigger driver aggression (Novaco 1990; Novaco 1991). 2.3.2 Economic Impacts of Auto Dependence Typically, individuals are only aware of a limited range of the costs of driving such as vehicle price, fuel, repairs, insurance, registration and parking. For example, the Canadian Automobile Association estimates the annual out-of-pocket cost of owning and operating the average vehicle in Canada to be over $7,300 annually (CAA 1997). However, there are many more monetary and nonmonetary costs borne both by drivers themselves and by society at-large. Furthermore, many of the costs that are bome directly by motorists are not perceived as immediate and therefore do not influence the decision to drive (Litman 1998a). This failure to account for the full cost of transportation, as well as the lack of clarity and efficiency in transportation pricing, means that the magnitude of the costs of driving is underestimated. This skews transportation decision-making. On one hand 'irrational' decisions are made regarding individual transportation choice. On the other hand, public policy is misdirected. Unless the full cost 52 dimensions of automobile ownership and use are recognized and incorporated into decision-making processes at the individual and societal levels, the problems of auto dependence cannot be fully addressed. The monetary costs of auto use include those paid for out of pocket by drivers (such as fuel, repairs, insurance and fees) as well as those financed by society (such as road construction and maintenance, parking, congestion and highway services). However, there are also a wide range of costs that have distinct "economic" dimensions that are not typically ascribed dollar values. These include many of the social and environmental costs described in some detail in the preceding sections. Many authors have attempted to comprehensively quantify the full range of financial, social and environmental costs of transportation (Delucchi 1996; IBI Group 1995; Litman 1995; MacKenzie, Dower, and Chen 1992; Miller 1993; Peat Marwick Stevenson & Kellogg 1993). Also, much work has also been done examining the costs on a sectoral basis (e.g., congestion, agriculture, accident loss, sprawl, air pollution and the like), though they are too many to enumerate here. Although most of the authors listed offer a range of perspectives on the categorization of costs,34 there is a great deal of congruence in the identification of various cost elements. Two authors, Todd Litman and Mark Delucchi, provide the clearest and most comprehensive identification, estimation and categorization of transportation costs. They also offer insight into two different approaches to, and applications of, transportation costing. Total costing Delucchi is primarily concerned with simply identifying and calculating the aggregate costs associated with motor vehicle use. Although Delucchi's estimates indicate motor vehicle use costs more than most people realize, he offers no evaluation of which transportation mode is 'better' or any judgement of whether these costs exceed the benefits (1996). Rather, he is interested in identifying the 'opportunity cost' of motor vehicle use (that is, what society as a whole gives up, or would otherwise save, as a result of auto use). In this respect, Delucchi's "social cost analysis" informs general discourse on transportation decisions and offers a framework for analyzing costs. 3 4 For example, Ketcham and Komanoff (1992) classify costs as direct costs borne by users, direct costs bome by non-users, externality costs borne by users and externality costs borne by non-users (in Murphy and Delucchi 1996). Miller and Moffet (1993) categorize transportation costs as personal, government subsidies, societal and unqualified and MacKenzie et al. (1992) categorize them simply as market costs and external costs. 53 Table 13 below provides Delucchi's classifications of a wide range of transportation costs of motor vehicle use according to how explicitly they are priced and allocated in the economy. Delucchi's framework is useful in that it determines whether the cost is monetary or nonmonetary, who the payer is and, where prices exist, whether prices are explicit or implicit. Accordingly, he identifies costs that are: • p e r s o n a l n o n m o n e t a r y costs i n f l i c t e d u p o n onese l f (generally, these are not fully recognized and therefore inefficiently incurred); • p r i v a t e sec tor g o o d s a n d serv ices (generally, these are the most efficiently allocated of all the costs, as they are borne directly by the users, however many are not perceived explicitly); • " b u n d l e d " p r i v a t e - s e c t o r g o o d s (these costs, such as condominium parkades, are large costs that are inefficiently allocated because they are priced implicitly in the cost of a package of other goods, such as condominiums); • p u b l i c i n f r a s t r u c t u r e (these are incurred by government and are price inefficiently or simply not priced); and • m o n e t a r y a n d n o n m o n e t a r y ex terna l i t i e s (which are rarely priced). 54 Table 13 - Motor vehicle (MV) cost categories (Delucchi) Personal Nonmonetary costs of MV's (unpriced) Explicitly priced private-sector MV goods and services "Bundled" private-sector goods (implicitly priced) Public infrastructure and services for MV use Monetary externalities (unpriced) Nonmonetary externalities (unpriced) Nonmonetary M o n e t a r y Nonresidential offstreet parking included in the price of goods and services or offered as employee benefit Home garages and other residential parking included in the price of housing Roads provided or paid for by the private sector and recovered in the price of structures Nonmonetary Uncompen-sated personal travel time Accidental pain and suffering and death upon self Noise inflicted on self Personal time spend working on MVs Air pollution inflicted on self Usually included in GNP accounts: Purchase of MVs Fuel, lube oil, except costs due to travel delay Maintenance, repair, washing, renting, storage and towing Finance charges on purchases of MV Parts, tires, tubes and accessories Automobile insurance Accident costs paid by insurance, lost productivity, medical and legal services, victim restitution Parking away from residence Usually not included in GNP accounts: Compensated time of travellers Overhead expenses of business fleets Accident costs paid by responsible party Vehicle inspection by private garages Legal services, security devices due to MV-related crime Public highway construction and maintenance, including on-street parking Municipal off-street parking not priced at marginal cost Highway patrol Environmental regulation, protection and cleanup, including landfills and sewerage treatments plants Energy and technology R&D Costs of travel delay imposed by others, including fuel oil, maintenance and compensated travel time Probabilistic loss of GNP due to sudden changes in oil prices Accident costs not paid for by responsible party: productivity, medical, legal, property Price effect of using fuels for MVs: increased payments to other countries for oil used in other sectors Losses for MV thefts and robberies Police protection, court and prison system Military expenditures to secure oil supply Fire protection MV related costs of other agencies Air pollution inflicted on others: effects on human health, crops, materials and visibility Accidents: pain and suffering and death not paid for by responsible party Extra uncompensated time due to delay Global warming due to fuel-cycle emissions Noise inflicted on others Price effect of using fuels for MVs: loss of consumer surplus Water pollution: health and environmental effect of leaking storage and waste sites, spills and road runoff Pain, suffering and inconvenience costs due to MV crime Not estimate here: Land use damage, species loss Socially divisive effects of roads Vibration damages Aesthetic impacts Source: (Delucchi 1996) Table 14 below presents low and high estimates Delucchi has made of the various costs associated with motor vehicles. Delucchi's estimates indicate that the total social cost of motor vehicle transportation may be as high as US$3 trillion annually in the U.S. and that per vehicle costs are in the range of US$9,900-15,000. Using Delucchi's data, I have estimated the total social subsidy (costs shared by society, including externalities) to be roughly US$4,300-8,400 annually, while the direct monetary subsidy (excluding externalities) to be in the range of $880-2,100 per vehicle. This estimate seems 55 consistent with Peat Marwick et al. (1993) figures which put the total social cost of motor vehicles in the BC Lower Mainland to be C$2.64 billion annually, or C$2,590 per vehicle in 1991.35 The main utility in Delucchi's work is in the thoroughness of identifying 'opportunity costs,' the precision of classifying them and in the detailed estimates of many of the costs.36 Many of the costs previously hidden or unrecognized are now made plain for analysis. Table 14 - Summary of the annualized social costs of motor vehicle use. 1990 COST FOR U.S. COST PER VEHICLE COST ITEM (Billion $/year) ($/year) Low High Low High 1) Personal nonmonetary costs 411 601 2,180 3,189 2) Private-sector 947 1,067 5,020 5,659 3) Bundled 71 223 337 1,181 4) Public infrastructure 125 207 662 1,099 5) Monetary externalities 80 147 423 780 6) Nonmonetary externalities 246 593 1,305 3,145 Total social cost 1,880 2,839 9,967 15,054 Subtotal: Monetary costs (2,3,4,5) 1,222 1,645 6,482 8,720 Subtotal: Payments by MV users 109 173 580 918 Total social subsidy3 824 1,599 4,367 8,477 Total monetary subsidy3 166 400 882 2,143 Source: Adapted from (Delucchi 1996) Note: a. To calculate these, I subtracted 'payments by MV users' and 'private sector' costs from the total social costs and the monetary cost subtotals. None of the other costs were subtracted because they are not directly paid for by users (i.e., internalized) and are "socialized." Bundled prices are more directly paid for by MV users, by these costs are still shared by non-MV users to a great extent. Incremental costing Todd Litman's approach to assessing, evaluating and interpreting transport costs differs from that of Delucchi in that Litman's primary concern is rooting out the inefficiencies endemic in current transportation pricing schemes. While Delucchi inventories the total costs of motor vehicle transportation, Litman examines the marginal costs of many modes in order to determine the extent to which various modes are 'priced' and how this pricing influences transportation equity and efficiency and land uses. Litman's basic thesis is that many transportation costs are either ignored, subsidized or not perceived as immediate, particularly for motor vehicle users. Auto use is therefore 'underpriced' or priced ineffectively, leading to inefficient transportation choices by individuals and decision-makers (Litman 1995). 3 5 The Peat Marwick et at estimate is less exhaustive and includes fewer external cost items than Delucchi's and are therefore lower. 3 6 Some of the nonmonetary externalities I have identified in the previous sections are not identified or costed in Delucchi's framework. 56 In order to make transportation more efficient and equitable, Litman presents and categorizes transportation costs according to how they influence modal choice. Three categories of transportation costs are identified in Litman's framework: internal and external, variable and fixed and market and non-market. These categories and some constituent costs are identified in Figure 6 below. Internal (user) costs are those borne directly by the user. External (social) costs are those uncompensated costs that are borne by society at-large. External costs are similar in nature to what Garrett Hardin refers to as "commonized" costs (1985), or costs shared by everyone, regardless of their contribution to that cost. Figure 6 - Motor vehicle cost categories (Litman) f Us Variable Fixed Internal Fuel (User) Short term parki Vehicle maintenance Partly) User time & stress I User accident risk ExternaX Road maintenance (Social) \ T r a f f j c | a w enforcement insurance disbursements Congestion delays Environmental impacts Uncompensated accident risk Source: Adapted from (Litman 1995) Note: Bold italicized items=non-market costs hide purchase Verfisje registration Insurance payments Long-term parking facilities Vehicle maintenance (partly) Road construction Free" or subsidized king Traffic planning Street lighting Land use impacts Social inequity The diffuse nature of external cost and individual imposes on society means that it is not factored into the individual's cost calculation (Baumol and Oates 1988). For example, most drivers do not consider the external costs such as noise, pollution and congestion that their driving imposes on others. This practice primarily stems mostly from the fact that motorists do not pay directly for these costs, but also from the fact that the impacts they experience as a result of their own actions are diffuse and shared by many (often millions) of others. Variable costs (such as fuel and parking) are those that change according to the level of use, whereas fixed costs (such as insurance and vehicle purchases) are 'sunk' costs that do not vary with use. Variable costs offer immediate feedback to user behaviour, whereas fixed costs have already been incurred and therefore do not inhibit driving. In fact, these "sunk" costs may further encourage driving (Hart and Spivak 1993). For example, a person may conclude that since they have already spent $20,000 on a vehicle and insurance, they will get full value for their cash outlay by driving as often as they like. Finally, market costs are those which involve a monetary transaction (either explicitly or implicitly), while non-market costs are those costs which generally go unpriced (indicated in bold in Figure 6). Users do not directly pay for many of the market and non-market costs they impose, which effectively constitute a subsidy. This makes car use artificially cheap. 57 In order to address the equity and efficiency dimensions of transportation pricing and reduce the amount of motor vehicle use, two-pronged approach is required. First, the costs of driving need to reflect the full cost of driving. Underpricing results in 'irrational' individual decisions at the consumer level and policies skewed toward auto dependence at the institutional level. The omission of market subsidies and social and environmental costs from the transportation cost equation distorts the true costs of automobile use. Secondly, the pricing structure for automobile use needs to be restructured such that it precisely aligns the perceived costs of travel with actual costs. Currently, travellers consider only a narrow range of internal variable costs (Figure 6) in their decision to drive or take the bus. The equation in Figure 7 below is a basic representation of the perceived costs of transportation to the user. Like the transit user, the automobile user presently perceives the full cost of transport to be out of pocket expenses (ticket or gas) and the value of time (V*T). These variable costs are often all that is considered in the modal choice cost equation. Not surprisingly, the automobile often seems to be cheaper. Fixed costs such as insurance, repairs, purchase price and external costs not considered, and therefore, are not part of the decision-making criteria for most users. Figure 7 - Perceived full price of travel pp = F + V*T where: PP = perceived full price of travel F = fare (bus) or out of pocket variable expenses of travel (auto) T = travel time (including waiting time) V = value of time (what one is willing to pay for an hours time) Ideally, the optimal pricing structure would have all costs which are external and fixed converted to costs which are internal and variable (see arrows in Figure 6). Among these costs, those which are non-market, or unpriced, should be assigned a price equal to their social marginal cost (i.e., incorporated into the full price equation). This is an often talked about (but seldom practiced) principle in transportation economics called "marginal cost pricing."37 Figure 8a represents the pricing problem in a simplified manner. Since drivers are only aware of the marginal private costs of using their car, they consume Q* trips. However, because the social marginal costs far outstrip the variable ones paid by the driver, she or he remains blind to the social cost that they exact. The economist solves this problem in Figure 8b by pricing trips at their social marginal cost. This effectively "rationalizes" the pricing of transport by changing travel demand, while passing on See Frankena (1979) for some early 'textbook' discussions of transportation economics that consider social marginal costs. 58 the remaining social costs through a "congestion tax" type of measure38. This model, of course, assumes that these costs are recoverable through money (for environmental degradation and urban sprawl, for example) and also assumes that people will actually act rationally and consume at Q i . Consuming at Q 2 , while only paying the outlined congestion tax, may still result in a social loss, albeit of a lesser amount than at Q* in Figure 8a. Figure 8 - The pricing problem - an economist's perspective a) Status quo - subsidies and hidden costs Cost of a Trip Social Marginal Cost Number of Trips Q* = Private Trips b) The pricing approach - marginal cost pricing Cost of a Trip Social Marginal Cost /P Q, = Socially Optimal # of Trips Q 2 = Private Solution Social Costs Private Marginal Cost Number of Trips These taxes for external costs imposed on society should be redirected towards mitigative efforts (or compensation) for those who bear the costs (usually society as a whole). 59 External costs currently account for 32% the cost of driving, while internal fixed costs account for 23% (Litman 1998a). An optimal pricing solution would see these 55% of external and fixed costs converted to internal-variable costs. While it may not be practical or possible to convert all of these costs, there are a variety of measures that could see many fixed costs converted to variable ones, with good results on travel demand (see, for example, Calgary 1994b; Litman 1995; Litman 1997b; OECD 1995; Shoup 1996, as well as the wide body of literature on TDM). Litman (1998a) estimates that the elimination of subsidies for motor vehicle use and the implementation of marginal cost pricing can reduce private motor vehicle travel by 30-50%. While some direct transportation costs may substantially increase for some motorists, Litman asserts these will be more than offset by savings in vehicle ownership expenses, housing costs, taxes, healthcare and environmental degradation. Are the benefits worth the cost? One argument forwarded by those justifying diseconomies in external costs of private motor vehicle transportation is that its external benefits (or positive externalities) outweigh the external costs. For example, it is often argued that automobiles offer substantial benefits to users in terms of increased mobility or that road investment and motor vehicle maintenance expenditures generate substantial economic activity. While Mark Delucchi does not perform cost-benefit analysis in his full social costing of transportation, he does acknowledge motorized transportation to have "social" (read, "external") benefits stating "motor-vehicle use provides enormous social benefit and, in our view, probably exceeds the social cost" (Delucchi 1996 p. 9). However, many authors have criticized this claim, showing that many of the benefits cited are, in fact, not "external," that external benefits rarely exist in transport and that many of the benefits attributed to auto use can otherwise be achieved. The argument of whether benefits such as increased access and economic spin-offs are indeed "external" rests on the definition of what an external effect is. The basic features of an external effect that they result an unintended consequence of an activity that is shared by society at large (see Rothengatter 1994; Verhoef 1994). However, Rothengatter (1994), Litman (1995) and OECD (1995) indicate that most of these benefits facilitate, or result in, market transactions which allow external benefits to be internalized by individuals over the long term. Benefits are essentially competed away (Litman 1995). Rothengatter shows that road transport subsidies involve the creation of consumer or producer surpluses (e.g., by lowering consumer or producer costs). This benefit, however, is one that is internalized and not realized by society at-large. Consumers and producers will continue to extract these surpluses and "internalize" them until no more benefits are available. Rothengatter concludes that "most of the effects mentioned such as the improvement of economic efficiency or development of new 60 consumption/production structures are basically not external but normal consumer's or producer's surpluses induced by market interactions" and that "the number and the relevance of positive externalities is low" (Rothengatter 1994 p. 321). As mentioned in Chapter 2 (section 2.2.1), this phenomenon of internalizing benefits and externalizing costs is a well-known social and ecological dilemma. Marginal social cost pricing essentially converts commonized transportation costs to privatized ones in Hardin's framework. The application of intrinsic responsibility through these market instruments allows transportation users to respond directly to gains and losses by adjusting transportation choices. Litman (1998b) also argues that the important questions with respect to assessing auto benefits is not whether there are benefits to auto use, but whether: • you (the individual or society) would benefit if your neighbour drove? • you would benefit if your neighbour drove morel and • any of the benefits of driving could be accrued by using different modes? Litman argues that driving is rarely inherently good for society and that the benefits of others driving more rarely extend beyond those accrued to the individual. Most of the benefits are capture by individuals and businesses, while the external costs are shared by all (and some more than others). Furthermore, many of the benefits enjoyed from driving can often be met (and exceeded) by utilizing other modes. For example, British Columbia Treasury data indicate that the economic development benefits of transit exceed those of auto spending: every million dollars of transit spending yields 21 full-time jobs, while one million dollars spent on autos yields just 7 full-time jobs (1998a). Aschauer and Campbell (1991) also find that investment in transit in the U.S. has greater potential as an economic stimulant than does highway spending (in Kenworthy et al. 1997). Much of the spending that is dedicated to automobiles and their use simply transfers money (capital) elsewhere in the economy that could be used for other productive investment. Or worse, these monies are often transferred outside the local and national economy removing any longer-run domestic economic benefits. Pricing correctly, Litman argues, will provide users with appropriate feedback when driving is reduced and eventually correct these inefficient transfers. Finally, it has also been shown that transportation infrastructure and modal choices have much more profound impacts that extend beyond sectoral economic development. Research completed in 1997 for a World Bank commissioned report the relationships between land use, transportation and regional productivity, found weak overall correlations between auto dependence and gross regional product (GRP) (Kenworthy et al. 1997).39 This study built on earlier research that produced only anecdotal evidence that auto dependence may impede regional economic productivity (Newman, Kenworthy, and Vintilla 1995). GRP is defined as the gross domestic product (GDP) contribution of the functional urban region. 61 However, Kenworthy et al (1997) were able to examine a larger sample of 37 cities in a more rigorous fashion and identify broad global patterns. They found that excessive car use and ownership does not necessarily confer substantial economic benefit (or what some would call an "external" benefit). Rather, they found that amongst developed cities, regional productivity actually declines after certain levels of car ownership and use and that external costs (such as energy depletion, sprawl, emissions and transport deaths) actually grow. That is, there are potentially cfoeconomies associated with excessively high levels of car use. Key reasons for these diseconomies include the high cost of servicing suburban sprawl40, the cost of deaths and injuries, time lost in congestion, higher transport expenditures and inefficient housing patterns. The excessive personal and societal spending on transportation also ties up capital that could otherwise be used for economic development. Walter Hook (1994) attributes much of Japan's economic success over the past four decades to its low levels of car ownership and use. Not only was energy saved and urban systems made more efficient, but monies saved from car ownership and operation created a larger pool of potential investment capital critical for economic development. Today, Japan has some of the highest levels of NMT use in the developed world (Hook 1994). Summary: ecological, social and economic impacts There is a multitude of ecological, social and economic impacts associated with automobile use. Many of these impacts display tremendous complexity in terms of their multi-dimensionality, their scope and their mutually reinforcing relationships. While a wide range of impacts is recognized in a broad survey of the literature examining the implications of automobile dependence, few references provide a comprehensive inventory of these impacts. This problem is particularly acute in the government literature where policy prescriptions are formulated based on a limited understanding of the full, life-cycle costs of automobile dependence. Since documents typically assess only a narrow range of the most explicit impacts, the policies prescribed will necessarily be incomplete. It is quite clear that addressing these many "symptoms" of automobile dependence individually is a complex, and perhaps futile task. In assessing the impacts inventoried and discussed above, they seem to fall in to two broad categories amenable to clear policy analysis. Those which are "fixed" and those which are "variable." Fixed impacts are those that are incurred regardless of vehicle kilometers driven (i.e., by virtue of ownership). Examples of fixed impacts include those incurred in the vehicle production process, those Sprawling cities necessitate higher costs for the provision and servicing of fixed infrastructure such as roads, sewers, cables and electrical wires. They generally also have lower per capita utilization of public facilities such as schools and hospitals, and higher capital and operating costs for the provision of distance-sensitive public services such as transit, fire, police, ambulance, garbage collection, snow plowing, road de-icing and the like. See Calgary 62 associated with minimal fixed infrastructure and services to accommodate cars and those associated with automobile disposal. Variable impacts fall or rise (not necessarily proportionately) with the number of cars and their level of use. Examples of variable impacts include local pollution, health costs, accidents, maintenance, infrastructure and the like, which are incurred after vehicle purchase. This dichotomy provides a useful starting point for policy analysis since it addresses the complexity of transportation symptoms with a great deal of simplicity and clarity: by addressing the root source of auto-related impacts. Those policies which reduce vehicle ownership will necessarily reduce the fixed costs, and any variable costs that vehicle would otherwise have produced throughout its life cycle. Those policies that reduce the number of kilometres driven will necessarily reduce the preponderance of variable impacts associated with driving. 2.4 REDUCING AUTO DEPENDENCE In the previous sections, I have reviewed some of the conventional approaches to transportation planning, presented some of the key relationships between various causes of auto dependence and discussed some of the resultant impacts. I have argued that transportation planning is primarily rooted in a paradigm that focusses on mobility over access, effectively deriving demand for automobile use. A series of mutually dependent and mutually reinforcing positive feedback relationships lead to the aggravation and seeming intractability of auto dependence. The impact of this over dependence on cars has profound implications for the ecological, social and economic wellness at a local and global scale. These are basically rooted in the need to own and use cars. However, many conventional approaches designed to deal with auto dependence tend to focus on narrow efficiency-focussed goals and technological fixes that often aggravate auto dependence rather than reduce ownership and use. These approaches are reductionist in nature isolating singular causes and impacts such that they are amenable to easy treatment. By merely "nibbling at the margins" (Hart and Spivak 1993), these technology and efficiency-oriented palliatives ignore the tremendous complexity of the issue at hand and tend away from more holistic approaches. Scholars such as David Ehrenfeld (1978), Ursulu Franklin (1990), Lewis Mumford (1934) and Rene Dubos (1970) have all argued that technology and efficiency alone are no panacea for problems wrought by technology. They merely offer refuge from treating root problems and often make them worse. Traditionally, transportation policy has focussed on resolving the problems of congestion and air pollution as most the visible problems associated with car use (Freund and Martin 1993; Gordon 1991; (1995c), Miller and Moffet (1993), MacKenzie et al. (1992), Newman and Kenworthy (1989a) and Altshuler (1979) for some discussion of the impact of sprawl on infrastructure capital and servicing costs. 63 Lowe 1990; Renner 1988; Schwartz 1971). Even today, much transportation policy in Canada and the U.S. focusses on treating congestion and air pollution as the primary problems requiring treatment (BC Environment 1995c; BC Transit 1996; BCTFA 1996; Gordon 1991; ORTEE 1992). More effective strategies for dealing with the problem are effectively forgone since policy efforts are directed at micro-and meso-level policy rather than macroscopic, holistic policies which consider a wider range of criteria. Defining problems in such narrow terms means the prescriptions for affecting them will necessarily be incomplete. 2.4.0 The Quasi-Solution - Seeing the Trees, Missing the Forest Author Eugene Swartz characterizes incomplete solutions as quasi-solutions. In his book, Overskill: The Decline of Technology in Modern Civilization. Swartz provides a critique of technology-oriented solutions and their inability to provide a long-term fix for problems. Swartz asserts that a technological solution is always a quasi-solution because it gives rise to a residue of unsolved problems. He outlined three sources of this residue: 1. the incompleteness of the technological solution; 2. the augmentation of the original problem; and, 3. secondary effect Furthermore, residual problems result in the creation of future generations of problems which drive the endless cycle of positive feedback technological solution-seeking (Schwartz 1971). The first residue problem, incompleteness, is mainly an extension of the original problem. Many technologies, for example, are not 100% efficient. The problem is therefore never completely solved, require further iterations of the process of technological refinement to perfect the process. Reducing vehicle emissions is an example such an iterative process that does not seem to resolve itself. The drive for completeness creates new problems through augmentation and secondary effects. Augmentation occurs when the initial problem is aggravated, or a higher-level problem is created, that requires the development of a new technology to address it. Secondary effects are other foreseen or unforeseen effects that result from the treatment of the original problem. For example, the quasi-solution of developing higher performance fuels to reduce pollution has reduced some pollutants, but increased others, such as hydrocarbons and carbon monoxide (augmentation). Meanwhile, the quasi-solution to the local pollution problem, emission control devices, has resulted in higher C 0 2 emissions and the secondary effect of global warming. Schwartz argues that the residue of a multitude of quasi-solutions becomes so compounded that technological solutions become increasingly difficult. The difficulty is attributable to the increasing complexity of the problem, the dynamics of technology, decreased resources, increased costs and the inertia of political institutions. New problems multiply at such a rapid rate such that real solutions can no 64 longer be found to address them each individually. Schwartz's argues that most of what we actually perceive as problems in transportation (as well as in other areas) are actually symptoms of larger, more significant problem. This assessment is also shared by Alan Altshuler (1980) in his seminal critique of transportation policy. Transportation policy developments in the United States (that in some respects mirror Canada's experience) demonstrate how the application of quasi-solutions fail to address the root problem, aggravate the original problem and spawn new ones. In the early 1960's, residents in cities such as Los Angeles and San Francisco began to lobby heavily to have the congestion and air pollution impacts of the automobile mitigated. The response of policy-makers was two-fold: the first was to require that vehicle manufacturers develop more fuel efficient vehicles (cleaner engines and catalytic converters) and oil companies develop cleaner burning fuels; the second was to increase capacity and roadway efficiency to allow for freer flowing traffic. However, both strategies failed to either decrease emission or reduce congestion. The first California air pollution law in 1960 was followed by the federal Motor Vehicle Air Pollution and Control Act in 1965. Both of these set out guidelines and for vehicle efficiency and, in subsequent amendments, required the use of catalytic converters to reduce the amount of certain tailpipe emissions. The Clean Air Act of 1970 and the Arab oil embargo of 1973 prompted yet more stringent regulations on tailpipe emissions and fuel quality to be in place by 1975. The result was that, from 1975 to 1987, the fuel economy of new cars in America increased nearly twofold - an impressive technological achievement. However, these gains in fuel efficiency were completely erased by the increased fleet sizes and vehicle miles travelled (VMT) of these, now more "environmentally friendly," cars (Freund and Martin 1993; Gordon 1991). The cheaper capital and operating costs offered by newer, more efficient cars further increased the demand for auto trips. The sheer increase in vehicle miles travelled from increased demand caused a net increase of air-borne pollutants (Renner 1988). Furthermore, like most technologies, pollution control devices are not perfect. At low speeds, and in cold conditions, catalytic converters experience dramatically lower efficiency. They also do not age very well. Unless pollution checks are in place, a catalytic converter can be inoperable or inefficient for the life of the car without being noticed. When they do operate properly, catalytic converters only eliminate some pollutants, but create others, namely greenhouse gases such as C0 2 . These newer greenhouse gas emissions are now of great global consequence. With increasing automobile use came congestion. From 1970 to 1987 the number of automobiles in the U.S. increased at a rate of 2.4% per year, while highways financing increased at a rate of 15% (Gordon 1991). Despite the added capacity, VKT has increased dramatically since the early 1960s (Lowe 65 1990; Newman and Kenworthy 1996; Newman and Kenworthy 1989a), while congestion continues and has worsened in many cases (Lowe 1990; Renner 1988). Many transportation policies pursued today in Canada can also be categorized as quasi-solutions. For example, a strategy of encouraging HOV use through the provision of HOV lanes41 suffers from incompleteness and may actually result in substantial augmentation and secondary effects. Conventional wisdom has it that HOV is desirable because it gets more people into fewer cars for commuting purposes. It therefore is purported to reduce the demand for auto travel and reduce emissions (BCTFA 1996; ORTEE 1995). However, it has been shown that HOV facilities may actually derive greater demand for auto travel in the long run by inducing travel (Johnston and Ceerla 1996; Vuchic et al. 1995; Vuchic et al. 1997). HOV provision essentially amounts to a freeing of existing capacity (and, in some cases, the creation of new capacity), thereby encouraging longer distance travel and sprawling land uses. Ultimately, these magnify the ecological, social, economic impacts highlighted in section 2.3 above. Although HOV conversions are traditionally characterized as TDM measures, their effect is to increase supply and they therefore derive greater demand for trips. HOV provisions are viewed as demand measures because they serve to reduce the demand for SOV trips. However, because "encouraging" HOV measures necessitates additional infrastructure and leads to less traffic in mixed-use lanes, it effectively results in a greater supply of road for SOVs. In the short run, this trip reduction and greater supply of capacity will no doubt lower the trip times for HOVs, while reducing pollution and congestion on other roads. However, the initial euphoria of such a strategy will eventually be dampened as the sobering reality of an auto-focused, supply-oriented solution takes hold. There is a substantial latent demand for automobile travel in most urban centres (Litman 1995). For every car-pool,42 at least two additional SOVs will be eliminated, providing cleaner air and less congested streets. However, as the congestion on the roads is reduced and a free-flow of traffic is restored, this newly created capacity for cars will merely act to satisfy an existing latent demand for SOV trips. Failing restrictions or a dampening of demand on new SOV use, this vacated capacity will attract new users and those previously deterred by congestion. Furthermore, a modal switch from those using "greener" modes (e.g., public transit, bicycles, walking, and even carpools themselves) now attracted to SOVs may compound this problem. Cars will continue to fill the free flow until congestion again is a deterrent and a new congestion equilibrium is reached. However, this new equilibrium operates at a 4 1 HOV lanes can be provided by either constructing new lanes, converting general purpose lanes to HOV use only, or opening curb parking lanes for HOV use. 4 2 Assuming 3+ occupancy. 66 higher state: the increase in total capacity means there are now even more vehicles congesting and polluting and that some of this new car travel may been switched from "greener" modes. Figure 9 - Road congestion: the "capacity myth" Time per F, F 2 < \ Vehicle km Bottleneck Congestion \ Maximum Flow Low Congestion Moderate Congestion F, = Flow of Vehicles F 2 = Flow with Additional Capacity VPH Flow of Vehicles Per Lane Per Hour Adapted from (Frankena 1979) Figure 9 above demonstrates the phenomenon I will term the "capacity myth." F] refers to the flow of vehicles for a road of a given capacity. Initially, there is little congestion and a relatively unconstrained flow of traffic. As the lane approaches capacity, cars begin to slow down, thereby creating moderate congestion. Once maximum flow is reached, gridlock sets in and cars move at very low rates of speed, if at all. Position F 2 describes the flow of vehicles once transport supply measures, such as new road and HOV lane infrastructure construction, are implemented. These measures serve to push the curve outward from Fi to F 2 , effectively increasing an artery's vehicle flow capacity. Here, the same phenomenon of congestion equilibrium sets in, but with a greater number of vehicles and vehicle trips. Similar phenomena have been documented in the construction of new capacity throughout North America and elsewhere (see previous discussion of induced traffic in section 2.2 above). In the longer run, HOV policies may have the effect of encouraging sprawling land uses and creating "future latent demand" for car travel (Hart and Spivak 1993). HOV strategies suffer from incompleteness. They focus on mitigating a narrow range of problems such as vehicle emissions, low vehicle occupancies and high peak period demand for car travel. However, in supplying easier automobile access, HOV strategies lead to augmentation of the original problems it sought to address (emissions, occupancy and peak demand). Furthermore, other secondary 67 effects such as sprawl, increased off-peak demand for car travel and other related impacts of auto dependence may ensue. 2.5 HOLISTIC DIRECTIONS A more holistic approach to reducing auto dependence is necessary. Nibbling at the margins offers no panacea and often makes matters worse. In order to affect substantive change, our prescriptions for addressing auto dependence must address the fundamental root causes. Quasi-solutions at best offer temporary reprieve from the symptoms of car dependence and are most often self-defeating in the long run. More holistic solutions address the fundamental problems of auto dependence identified in this chapter. They: • promote exchange and access over mobility and speed; • reduce car ownership and use; • provide a counterweight to the positive feedback relationships that feed auto dependence; they either provide 'negative feedback' (suppress unwanted action) or set in motion desirable positive feedback loops; and • engender intrinsic responsibility or mutual cooperation in using and managing common property. Of course, any measures that meet these criteria must also be considered within a wider context of time required for implementation, cost effectiveness and political/public acceptability. Chapter 5 sets out a more detailed framework for evaluating measures based on the policy directions suggested in Chapter 4. 68 CHAPTER 3 - METHODOLOGY 3.0 INTRODUCTION This analysis in this thesis uses comparative data on transportation and land use patterns in seven major Canadian urban regions for four years (1961, 1971, 1981 and 1991). The data used for each region fall into the following broad categories: population and employment distribution, developed area, transport infrastructure supply, vehicle ownership, transport energy consumption and public and private transportation usage. This chapter provides an overview of the methodology employed in the data collection process and a discussion of some of the issues relevant to data interpretation. It builds on the discussion in Chapter 1 of the need for comparative transportation data for Canadian cities and provides a rationale for the selection of data items, cities studied and years surveyed. I follow with a description of the data surveyed and of the collection process. The indicators collected from the survey are then defined and issues relevant to their collection and interpretation are discussed. Finally, I end with some notes regarding the use and interpretation of the data and an assessment of their reliability. 3.1 ORIGINS OF THE RESEARCH The data collection methodology used in this thesis was initially developed by Peter Newman and Jeff Kenworthy in their landmark 1989 study, Cities and Automobile Dependence (CAAD). This initial study was based on a sample of 32 global cities (13 European, 10 American, 5 Australian, 3 wealthy Asian and 1 Canadian) and covered 3 study years (1960, 1970 and 1980). In early 1996,1 was invited to work on assisting with the collection of the Canadian cities data for this update to CAAD (CAAD II) titled An International Sourcebook of Automobile Dependence in Cities. 1960-1990 (Kenworthy et al. 1999). Table 15 - Stages in the research program Research Activity Location Time Frame 1. General survey of relevant literature Australia May-August 1996 2. Learning data collection and processing techniques Australia July-August 1996 3. Collection of transportation and land use data Australia August 1996-January 1997 4. Collection of missing data items Canada March 1997-May 1997 5. Detailed literature review Canada May 1997-August 1997 6. Analysis and thesis writing Canada July 1997-April 1998 I worked with Jeff Kenworthy and a team of PhD students at the Institute for Science and Technology Policy (ISTP) at Murdoch University in Perth, Australia from June 1996 until February 1997 69 on CAAD II. The project also involved several months of follow-up work in Vancouver to finalize and correct various data items (see Table 15- Stages of the Research Program). Each student was charged with responsibility for a particular geographical area (e.g., Asia, Europe and North America), however the collection and processing of data was very much a collaborative effort. Table 16 lists all the cities of study by region. Table 17 indicates the researcher(s) who made the primary contribution for the collection of data. Table 16 - Global cities included in CAAD II American Australian Canadian European Wealthy Asian Developing Asian cities cities cities cities Cities cities Boston Adelaide Calgary Amsterdam Hong Kong Bangkok Chicago Brisbane Edmonton Brussels Singapore Jakarta Denver Canberra Montreal Copenhagen Tokyo Kuala Lumpur Detroit Melbourne Ottawa-Hull Frankfurt Manila Houston Perth Toronto Hamburg Seoul Los Angeles Sydney Vancouver London Surabaya New York Winnipeg Munich Phoenix Paris Portland Stockholm Sacramento Vienna San Diego Zurich San Francisco Washington Table 17 - Researchers' contribution to the data collection process City/City Group Main data collector(s) Australian cities Jeff Kenworthy US cities Kenworthy, Felix Laube and Tamim Raad Canadian cities Calgary Kenworthy, Laube and Raad Edmonton Kenworthy, Laube and Raad Montreal Kenworthy, Laube and Raad Ottawa-Hull Raad and Laube Toronto Kenworthy, Laube and Raad Vancouver Kenworthy, Laube and Raad Winnipeg Raad European cities Laube and Kenworthy Wealthy Asian cities Hong Kong, Tokyo Kenworthy Singapore Kenworthy and Paul Barter Developing Asia Bangkok Chamlong Poboon Beijing Hu Gang Jakarta, Kuala Lumpur, Seoul, Surabaya Barter Manila Jun Guia and Barter The data collection process was at various stages of completion when I began work on this project. The data collection for Toronto was almost complete. Vancouver, Calgary and Montreal had 70 much of the 1991 data collected, so most of the work in these cities involved obtaining the 1961, 1971 and 1981 data. Edmonton required significant work for all years. Winnipeg and Ottawa-Hull were added when I joined the project, so these required all data for all four study years. 3.1.0 Rationale for Cities, Study Years and Data Items Selected Seven Canadian metropolitan areas were surveyed in this study: Vancouver, Edmonton, Calgary, Winnipeg, Toronto, Ottawa-Hull and Montreal. Section 3.3 below provides a detailed description of how the metropolitan areas, inner cities and CBDs of each of the regions were defined for survey purposes.43 The seven regions were chosen mainly because they are Canada's major population centres. From a comparative perspective, the selection also illuminates differences in the urban transportation experience within Canada. The cities represent a range of urban density levels, public transit services (bus only or with combinations of bus, busway, LRT, ALRT, subway and commuter rail) and private transportation infrastructure supply levels (e.g., freeways, parking, etc.). Initially, Winnipeg and Ottawa-Hull were not included in the study. However, they were added because they are unique in the context of other large Canadian cities: both have bus-based systems (Ottawa has a segregated busway) with no urban rail provision. The years used in the survey are 1961, 1971, 1981 and 1991.44 For the most part, most data items were obtained for these study years. Where data were not available for these study years, data within 1-2 years of them were obtained and the appropriate population base was used to standardize them. Otherwise, best-guess approximations were made in concert with local planning staff. The base year of 1961 was chosen as it marks the beginning of the period of rapid motorization and suburbanization in Canadian cities and is the earliest date for which any of the data required in the survey are widely and accurately available. Data for subsequent years trace the evolution of urban transportation and land use in the face of continuing motorization and suburbanization. Each of the years of the study was selected to correspond to census years to enhance data availability and the usefulness of results for planning purposes. Data for many other cities in the CAAD study (the American cities, for example) used 1960, 1970, 1980 and 1990 data to correspond with their respective census years. The one-year difference is not highly significant because of the large time frame necessary to see meaningful changes in urban form and transport. Therefore, there is still a high degree of Throughout the thesis the use of the city name alone refers to the entire metropolitan area as defined in Table 19. For example, the use of 'Vancouver' refers to the GVRD, 'Toronto' to the GTA and 'Ottawa' to Ottawa-Hull (or RMOC, MRCCO and CUO combined). Where reference to a specific municipality within these regions bearing the same name is made, it will be qualified with reference to the City of Vancouver, the City of Toronto, etc. 4 4 The only exception is the City of Winnipeg. As comprehensive database of transportation, land use and demographic data already existed for 1962, 1971, 1981 and 1992, these were used as the study years for Winnipeg. 71 comparability between various countries. The 10-year spread between survey years is used also because of the time lag necessary to see substantive changes in urban form and infrastructure. Although it would be desirable to have an updated set of 1996 data for Canadian cities, it would not be realistic to begin collecting and collating comprehensive data until at least mid-1998 as census data take from 1-3 years to be processed and fully available. The data items in the study were selected to describe the basic relationships between urban land use patterns and transportation use, supply and transportation efficiency over time. The 'raw' data items needed to describe these basic relationships, but also needed to be universally collected and reported to allow for standardization and comparison. To track the movement and dispersal of population and jobs each region was divided into three distinct sectors: the CBD, the inner city, and the outer area. The inner city includes the CBD and describes the pre-World War II (pre-automobile) city. The outer area is simply the remainder of the metropolitan area. Urbanized area, population and jobs data were collected for each of these three sectors. Parking and road length data provide a basic picture of infrastructure supply dedicated to car use. Meanwhile, motor vehicle registrations, vehicle travel, energy use and trip lengths data are all standard transportation planning indicators measuring the use of private automobiles in urban areas. Modal split data measure the balance between car, transit and non-motorized travel in a region. Finally, the public transport data used are widely indicators used to describe the levels of use and supply of transit in a city. While the raw data provide information regarding the absolute levels of car dependency, standardizing these raw data (e.g., per person, hectare, kilometre, etc.) provide insights into transport and land use efficiency, intensity of use and supply. Standardizing also allows for comparative assessment between regions. 3.2 DATA COLLECTION PROCESS The data collection process involved 3 steps: 1) The selection of parameters and distribution of the survey forms. As mentioned earlier, the parameters were selected to provide insights into the basic relationships between urban form and transportation service, infrastructure and use. These parameters were the same as those used in CAAD, with some refinement of certain parameters. The surveys were distributed to officials at relevant planning agencies in each city along with an explicit definition of each parameter requested. Typically, the initial surveys were sent to a regional planning body or municipality in each city as well as the regional transit operator(s). Table 18 below shows the primary and secondary agencies that supplied the requested information. Appendix 5 (Data Sources) provides detailed source agency information. 72 2) Follow up work to collect missing data items. Only in rare cases, such as Winnipeg, was high quality and relatively complete data immediately available. Substantial additional follow-up work with other departments with the agencies, as well as contact with other agencies (such as Statistics Canada, provincial ministries, central city municipalities and CUT A) was required to complete the data set. Visits to each city, faxing and telephone calls were required to follow-up with each agency. 3) Crosschecking and confirming of data to ensure consistency and reliability. On occasion, data supplied by agencies were clearly in error. However, intense scrutiny of each data item supplied usually ferreted out errors. Data were crosschecked so that dramatic trends or inconsistencies were identified and confirmed. For example, transportation data were compared with one another to ensure they corroborated one another. Very high vehicle ownership levels and low transit use would be inconsistent with low VKT. Such inconsistencies are investigated and confirmed with local agencies. If aberrant were found, they were either confirmed or corrected with local planning staff. Table 18 - Data source agencies Data type Primary data source Secondary data source POPULATION/AREA Total pop. • Cities of Edmonton, Calgary, and Winnipeg • GVRD, Metro, OGTA, RMOC and CUM • Statistics Canada Urbanized area • Cities of Edmonton, Calgary, and Winnipeg • GVRD, Metro, OGTA, RMOC and CUM • Detailed planometer measurements CBD/lnner City pop. • Central city governments (all cities) • Statistics Canada CBD/lnner City area • Central city governments (all cities) • Detailed planometer measurements EMPLOYMENT Region • Cities of Edmonton, Calgary, and Winnipeg • GVRD, Metro, OGTA, RMOC and CUM • Statistics Canada CBD/lnner city • Central city governments (all cities) • Statistics Canada Parking • Central city governments (all cities) Road Network • Cities of Edmonton, Calgary, and Winnipeg • GVRD, Metro, OGTA, RMOC and CUM • BC Ministry of Municipal Affairs Vehicles Registered • Cities of Edmonton, Calgary, and Winnipeg • BC MoTH, Metro, RMOC and CUM • Statistics Canada • TAC (1996); JPINT (1996) PRIVATE TRANSPORT • Cities of Edmonton, Calgary, and Winnipeg • GVRD, Metro, OGTA, RMOC and CUM • ND Lea (1966) • Quebec Ministry of Transport Private Energy • GVRD, Metro, OGTA, RMOC and CUM • City of Calgary • Kent Marketing, Inc. Mode split/Trip Length • Cities of Edmonton, Calgary, and Winnipeg • GVRD, Metro, OGTA, RMOC and CUM TRANSIT • City/Regional transit agency in each city • RMOC in Ottawa • Detailed cross-checking with CUTA • RMOC provided Hull data 73 There was rarely "one stop shopping" in collating the data. This process at a minimum required contact with several departments in one regional agency. In most cities, it was necessary to liaise with several departments and agencies at the local, regional and provincial level, as well as others. 3.3 DEFINITIONAL ISSUES There are two forms in which that data in this study appear: "raw" and "standardized". These data are located in Appendix 2. The raw data are basically the unadjusted absolute measures for each parameter. The standardized data provide indicators derived from the raw data. These measure the efficiency of public and private transportation, the intensity of urban activity and infrastructure use and the per capita performance of transportation and land use. So long as the raw data inputs are collected on a consistent basis for each city, standardizing the data allows for accurate comparison between metropolitan areas. Since the purpose of the thesis is to come to a better understanding of the factors accounting for differences in auto dependence in Canadian cities, I will rely primarily on the standardized data for the analysis. This is not to understate the value of the raw data. They are the base from which the standardized data are derived. Also, from a 'sustainability' perspective they provide important information regarding absolute changes in urban characteristics and auto dependence. For this reason, I will infuse the analysis with information regarding the absolute changes as well. However, for comparative purpose the analysis will revolve primarily around the standardized data. To ensure that these comparisons are accurate and meaningful, it is important to be precise in defining the parameters used, the methodology employed in collecting them and any difficulties encountered with specific items. 3.3.0 Raw Data Definitions and Issues The raw data were chosen to represent the broad range of factors that affect the evolution of urban transportation. The parameters were chosen to allow for the analysis of the relationships between land use, demographic change, and public and private transport supply and use on a regional scale. Below are definitions of each data item as well as (where appropriate) discussion of specific problems and difficulties encountered in their collection and processing. Territorial areas Data were collected for three geographical areas for each city: the metropolitan area, inner area and central business district (CBD). A fourth area, the outer area, was derived from the data provided. Appendix 1 contains maps outlining the relationship of these geographical areas to one another. For 74 accurate comparisons, raw data items correspond to these geographical definitions in every case. In rare cases where raw data items correspond to a different geographical regions or year, the appropriate (corresponding) population and land area data are used to standardize them. Only the Canadian cities in the study are described below. Specific definitions for data from the other 40 cities in CAAD can be found in Kenworthy and Laube, et al. (1998). Metropolitan area The metropolitan area is defined as the functional urban area of a city, or what is commonly referred to as the 'city-region.' The city region can be defined as agglomeration of urban units that share a common social, environmental and economic destiny (Golden et al. 1996; Jacobs 1969; Sancton 1994). The individual municipalities of a metropolitan area display economic and spatial interconnectedness, typically manifest in a common 'commutershed.' The urbanized area is generally contiguous, with occasional "satellite" communities on the periphery. The metropolitan areas, as defined here, may or may not have a single corresponding administrative unit. How closely this definition is followed depends to a large extent on data availability. In the case of the Canadian cities included in the study, it was possible to get fairly comprehensive data for the entire metro regions as defined. Some cities, such as Edmonton, have smaller satellite communities with poor data availability (e.g., St. Albert). In cases such as these, these areas were simply omitted. Data reliability remains high as these omissions are generally small in population and size and the remaining data items are kept internally consistent. In several cases (most notably Toronto, Montreal and Ottawa), the regional governing bodies lack territorial comprehensiveness. Metro in Toronto and MUC in Montreal each have roughly half their regions' populations. Meanwhile, RMOC is territorially comprehensive on the Ontario side, but does not include the Hull on the Quebec side that accounts for roughly one third of the region. Therefore data from outlying areas that are essentially part of the functional urban area, but are not included in a single regional jurisdictional unit, had to be sought out and included. Table 19 below defines the functional metropolitan areas of the seven Canadian cities, as used in this study. Maps of the metro regions are found in Appendix 1. The City of Calgary encompasses the entire functional Calgary region. Its boundaries have been and still are progressively expanded through annexations to include new development on the city's periphery. Calgary's 1991 population was 710,677. Edmonton's functional region is somewhat larger than the City of Edmonton. However, the City of Edmonton contains the majority of the region's population. The satellite communities of St. Albert, Fort Saskatchewan and Sherwood Park roughly total 90,000 people. Due to extraordinary difficulty in obtaining data for these cities, the City of Edmonton, 75 with a 1991 population of 614, 665, was chosen to represent the region. The travel patterns of these satellite communities are not expected to impact the overall metropolitan averages significantly. In Montreal, the Montreal Urban Community (MUC) is the official regional planning agency for the area. However, even at its inception, it lacked territorial comprehensiveness. Today, less than half the region's population is located in MUC. The Quebec Ministry of Transportation, however, defines the total region based on its functional area and commuter shed. This area, Region de Montreal (RM), includes Laval and all or part of eleven municipalities to the north and south of Montreal Island and has a 1991 population of 3,119,570. Table 19 - Canadian city metropolitan area definitions (1991) City Metro area definition Calgary City of Calgary Edmonton City of Edmonton Montreal Region de Montreal(RM). This includes the Montreal Urban Community (MUC) plus Laval and all or part of eleven surrounding suburban Regional Municipalities. Ottawa-Hull Regional Municipality of Ottawa-Carleton (RMOC) on the Ontario side and bodies Municipalite Regionale de Comte Collines-de-rOutaouais (MRCCO) and the Communaute Urbaine de I'Outaouais (CUO) on the Quebec side. Toronto The Municipality of Metropolitan Toronto (Metro) for most data. Where the Greater Toronto Area (GTA) is used in 1981 and 1991, it includes Metro as well as the outlying Regional Municipalities of Durham, Halton, Peel and York. Vancouver The Greater Vancouver Regional District (GVRD). Winnipeg City of Winnipeg (UniCity). Ottawa consists of two distinct parts, a larger area in the Province of Ontario called the Regional Municipality of Ottawa-Carleton (RMOC), and population about one third its size in the Province of Quebec with the two administrative bodies, Municipalite Regionale de Comte Collines-de-1'Outaouais (MRCCO) and the Communaute Urbaine de I'Outaouais (CUO). This is a large area in size that corresponds roughly to the National Capital Region. Much of the land within the RMOC, MRCCO and CUO are rural; the actual urbanized area of these jurisdictions is small and mostly contiguous. The 1991 population of Ottawa-Hull was 907,919. The Municipality of Metropolitan Toronto (Metro; 1991 population 2.2 million) collects high quality data for Toronto. Upon inception in 1954, Metro was territorially comprehensive, but has been significantly outgrown since the 1970s by peripheral urban development. A more appropriate definition for the region is what is now known as the Greater Toronto Area (GTA), with a 1991 population of 4,235,756. The GTA includes the Regional Municipalities of Durham, Halton, Peel and York in addition to Metro. Since there has been no single jurisdictional unit corresponding to this definition, data 76 collection has been difficult and has required the collation of data from multiple agencies. Some data are available for the GTA for 1981 and 1991, however earlier years are much patchier. For the purposes of this study, 1981 and 1991 Toronto data used will generally refer to Metro Toronto, unless otherwise noted. Additional 1981 and 1991 data for the GTA are used where available to supplement these data and provide a broader region-wide perspective. Only Metro data are used for 1961 and 1971, however this is not deemed to be a problem for comparative purposes as the functional regional area corresponded closely to Metro in this period45. Furthermore, when standardized indicators are calculated, data items are matched with their geographical equivalent (e.g., Metro population is divided by Metro urbanized area to determine densities), thereby ensuring 'per unit' consistency and 'apples to apples' comparisons. On January 1, 1998, the Province of Ontario amalgamated Metro Toronto and all its constituent municipalities into one "megacity" now called the City of Toronto. The structure of the remaining regions and municipalities outside Metro remains largely unchanged. To eliminate confusion throughout the thesis, all references to cities and regions in the Toronto region will be in pre-1998 nomenclature, unless otherwise noted46. Most of Vancouver's functional region is included in the Greater Vancouver Regional District (GVRD; 1991 population of 1.5 million), although it could be argued that some satellite communities such as Abbotsford and Chilliwack are part of the commutershed and should be included. The GVRD, however, is considered to be territorially comprehensive and has expanded to include newly developed suburban precincts throughout the years. The GVRD keeps regular, consistent and reliable data. In Winnipeg, the City of Winnipeg (UniCity; 1992 population 641,850) is used to define the region. This is includes almost all of the functional urban region and excellent data are readily available. Winnipeg had excellent data availability for 1962, 1971, 1981 and 1992. Since Winnipeg had excellent data availability for these years, 1962 and 1992 were used as the study years for that city in lieu of 1961 and 1991 for all the other cities). Since few dramatic changes occur in urban transportation and land use patterns within the span of one year (especially in low-growth cities), these were taken as a good approximation for 1961 and 1991. Inner area The inner area (or, alternatively, inner city) is defined as the part of the metropolitan area that was contiguously developed by the 1940s (that is, the pre-automobile era). Fewer complementary data items 4 5 Metro represented 77, 72, 63 and 54% of the GTA's population in 1961, 1971, 1981 and 1991, respectively (see Toronto in Appendix 2, Data Tables). In 1961 and 1971, outlying population centres were not considered part of the functional regional area, and commuting patterns there reflected this. 4 6 For example, if a reference is made to the larger, post-January 1, 1998 City of Toronto, it will be referred as post-1998 Toronto, post-amalgamation Toronto, or the Megacity. 77 are required for this area, so complete data were easy to obtain in most cases. Furthermore, because the area definitions were highly precise (down to the intersecting streets) and the area usually fell within the jurisdiction of the central-city municipality, data reliability is quite high. Table 20 - Canadian inner area definitions (1991) City Administrative Unit Names Calgary Community District Altadore, Banff Trail, Bankview, Bridgeland/Riverside, Britannia, Cambrian Heights, Capitol Hill, Canadian Forces Base Currie, Chinatown, Cliff Bungalow, Connaught, Crescent Heights, Downtown East, Downtown West, Eau Claire, Elbow Park, Elboya, Hilton, Highland Park, Highwood, Hillhurst, Hounsfield Heights/Briar Hill, Inglewood, Killamey/Glengarry, Lincoln Park Redevelopment, Mayland Heights, Mission, Mount Royal Lower, Mount Royal Upper, Ogden, Parkdale, Parkhill/Stanley Park, Queens Park Village, Ramsay, Renfrew, Richmond, Rideau Park, Rosedale, Rosemont, Roxboro, Rutland Park, Scarboro, Scarboro/Sunalta West, Shaganappi, South Calgary, Spruce Cliff, St. Andrews Heights, Sunalta, Sunnyside, Triwood, Tuxedo Park, University of Calgary, University Heights, Victoria Park, Vista Heights, West Hillshurst, West Mount Pleasant, Windsor Park, Winston Heights Mountview Edmonton Zone 1-12, 20, 21, 28, 42-46, 60, 61, 69-72, 74, 82, 83 Montreal Secteur 1 Montreal: Centre-Ville, 2 Montreal: Centre-Ville peripherique, 3 Montreal: Sud-Ouest, 4 Montreal: Notre-Dame-de-Grace, 5 Montreal: Cote-des-Neiges, 6 Montreal: Plateau-Mont-Royal, 7 Montreal: Villeray, 9 Montreal: Saint-Michel, 10 Montreal: Rosemont, 11 Montreal: Sud-Est, 20 Mont-Royal, 21 Outremont, 22 Westmount, 23 Hampstead, 24 Cote-Saint-Luc, 25 Montreal-Ouest, 26 Saint-Pierre, 27 Verdun Ottawa Ottawa Inner Area, Hull CBD Toronto Minor Planning District 1a-1h, 2a-2l, 3c, 3e, 3g-3i, 4b-4d, 4g, 4h, 6a-6c, 6e-gh, 7b-7d, 14a Vancouver Local Area Arbutus-Ridge, Central Business District, Dunbar-Southlands, Fairview, Grandview-Woodland, Hastings-Sunrise, Kensington-Cedar Cottage, Kerrisdale, Kitsilano, Marpole, Mount Pleasant, Oakridge, Renfrew-Collingwood, Riley Park, Shaughnessy, South Cambie, Strathcona, Sunset, Victoria-Fraserview, West End, West Point Grey Winnipeg Traffic Superzone 1,7, 12, 13,22,26, 30, 33,36 The definitions of the inner city were usually arrived at in concert with local planning staff who had knowledge of the development history of the city. Table 20 above details the inner city definitions for the Canadian cities in the study. Maps outlining the inner city are found in Appendix 1. Each city's inner area usually described as an agglomeration of neighbourhoods, districts or traffic zones as shown in the table. Central Business District (CBD) The Central Business District is defined as the area with the most significant employment concentration in the metropolitan area. The CBDs are usually defined by the central-city planning 78 department and data availability and reliability is high. Table 21 below provides the CBD definitions for the Canadian cities in the study. Maps outlining the CBDs are found in Appendix 1. Table 21 - Canadian central business district definitions (1991) City Administrative unit Names Calgary Community District Chinatown, Downtown East, Downtown West, Eau Claire Edmonton District 1 CBD Montreal Secteur 1 Montreal: Centre-Ville, 2 Montreal: Centre-Ville peripherique Ottawa Ottawa CBD and Hull CBD Toronto Minor Planning District 1e Vancouver Local Area Central Business District Winnipeg Traffic Superzone 36 Outer area The outer area is simply defined as the difference between the inner area and the metropolitan area. This area captures post-automobile era development and transportation patterns. Data items for the outer area are calculated as the difference between the metropolitan area and the inner area. Population The population data are collected for the metro area, inner area and CBD, and are derived for the outer area. The accuracy of this item is important as it is used to standardize many of the other data items for comparative purposes. The population data are considered highly reliable as the study years fall on census years. Municipal governments typically have special reporting done from Statistics Canada (Statscan) to disaggregate data by the administrative units used locally. The matching of population to area is therefore quite precise. In some cases, notably in Vancouver (inner area 1961) and Edmonton (inner area for all years), census tract boundaries may not have corresponded perfectly to area definitions. In these cases, estimations were made based the proportion of the census tract falling within the boundary (with allowances for undeveloped land, etc.). Since these adjustments represented a small proportion of the population counts for the areas in each case, the accuracy of the estimates remains quite good. Furthermore, as Statscan uses the same data collection methodology in each city, comparability is again considered high. Employment The employment figures collected are for "jobs" or "place of work," rather than "labour force." The distinction is crucial as the former measure how many jobs are located in a particular area, while the latter measures how many people living in an area have jobs. For example, the CBDs of most cities will 79 have a very high number of jobs located there (with people commuting in), while its labour force (the number of people living there that have jobs) will be quite small. Moreover, the jobs of the CBD-dwellers may lie outside the CBD. Employment data are obtained through city planning departments. These data have either been obtained through Statistics Canada or through municipal surveys and censuses. Employment data are collected for the CBD, inner area and metro area and are derived for the outer area. Urbanized area Urbanized areas measure the total amount of urbanized land in metropolitan region as defined for this study. The areas used in this study are net areas, as opposed to gross areas. Net areas measure all urban land in the region (residential, commercial, industrial, institutional, transportation, utilities, local parks and open spaces and abandoned urban land), less all the non-urban land (such as undeveloped land, regional scale open spaces, land zoned urban but not yet developed, forests, agricultural land, and waterways). Gross areas will include all land within an administrative boundary such as a census metropolitan area (CMA) or a regional district, regardless of the use of lands within it. Areas are also calculated on the same basis for the inner city and CBD as defined in the preceding sections. The data were mostly obtained from detailed land use inventories provided by planning authorities with non-urban land excluded. Where local land inventories did not exist for certain years (particularly 1961 and 1971 in some cities), I measured the areas manually with a planometer47 from detailed land use maps of the appropriate year. This measuring process was a very labour intensive task and took up to one day for each map. Both the land inventory and planometered land area data are of high quality. Since urban land areas are used to derive urban densities, its appropriate measurement is critical for the formulation of meaningful urban policy. With the ratio of urban to non-urban included in administrative boundaries varying dramatically from region to region, the use of gross densities renders comparisons meaningless. In these cases, urban density has little utility as an indicator of the actual intensity of activity. For example, generous administrative boundaries (such as in Edmonton and Vancouver) will understate densities, while more constrained administrative boundaries (such as in Toronto and Montreal) will overstate them. The implications of over- or understated urban densities are profound for policymaking, particularly in the case transportation and land use planning. Unfortunately, much of the planning literature using urban areas and densities do not to discriminate adequately between gross and net measures, or are not clear about methodology. Table 22 below compares the density figures found in a three studies, each ranked relative to density published in 80 Raad and Kenworthy (1998). In all cases, noteably different rankings are found. In some cases, namely Rothblatt (1994), differences in urban density calculations are dramatic. For example, Rothblatt's density for Edmonton (presumably calculated on regional administrative boundaries) yields a density of 2 p/ha, a figure that is essentially rural. Such statistics are clearly an unsound basis for policy analysis and formulation. Table 22 - Comparison of urban dens ity measures IBI Group (1994)'"' Raad and Kenworthy (1998) TAC (1996)" R/ K '98 Rothblatt (1994) R/K '98 Urban Density50 Rank Urban Density Rank Urban Density Rank Rank Urban Density Rank Rank Toronto (GTA) 26.8 1 25.9 4 24.7 2 4 10.4 2 3 Montreal (RM) 22.6 2 33.8 1 25.2 1 1 11.9 1 1 Vancouver (GVRD) 12.5 7 20.8 6 6.2 5 5 6.1 3 4 Edmonton 15.2 4 29.9 3 9.0 4 3 2.0 4 2 Calgary 13.4 5 20.8 6 - - - -Winnipeg 13.3 6 21.3 5 - - - -Ottawa (RMOC) 20.8 3 31.3 2 19.4 3 2 - - Parking supply Parking supply data are collected for the CBD area only. These data are usually easy to obtain through the central city government as parking inventories are updated regularly. These numbers include both private lots as well as and municipally owned lots and on-street spaces and are counted through surveys or taxation records. The spaces are disaggregated to show the number of on-street and off-street spaces. This classification is useful as it gives an indication of the degree to which a municipality is willing to control travel demand and vehicular flow. For example, ample provision of on-street parking (in proportion to off-street) may indicate a willingness to sacrifice vehicular flow to accommodate parking needs. Meanwhile, a high number of off-street spots indicate loose parking controls and a willingness to accommodate long-term commuter parking. Although inner area parking data would also be useful, these data are not widely collected or regularly updated by municipalities. A planometer is an electronic measuring device that when moved on a flat surface accurately calculates areas. 4 8 IBI Group's densities are based on 1990 figures. The remaining authors quote 1991. Because temporal variation in urban density is quite small, the figures provide satisfactory comparability for illustrative purposes. 4 9 Although TAC's base year was 1991, some cities have provided data for years other than 1991; Edmonton is 1994, Montreal is 1993 and GVRD's urban area is 1991, while its population is 1992. Urban density was calculated for the TAC report using the figures provided for urbanized area and population. 5 0 All urban densities are in persons per hectare (p/ha). 81 Road supply Road supply is a measure of the total number of kilometres of all classes of paved roads in the metropolitan region, from freeways down to local collectors. These are measured in centre-line kilometres. As a result, this measure does not give an indication of the total capacity of a city's road system. For example, two cities may have the same number of centre-line kilometres of roadway, but one may have many more 4 and 6 lane arterials, and therefore more capacity. A more a appropriate measure of capacity is lane kilometres (centre-line kilometres on a roadway multiplied by the number lanes). However, this measure is not widely available and is provided only for selected cities for 1991. The measure of centreline kilometres is still useful as it gives a rough idea of general infrastructure requirements to service its land use pattern (this figure can also be generalized to water and sewer line requirements). The road supply data have been disaggregated by road class, providing data users with a rough idea of the capacity mix of the roadway system. Because of inconsistencies in the classification of roads between municipalities, road classes are generally not comparable. For example, some municipalities may only have two classes of roads whereas others may have as many as five, or more. Vehicles on register Vehicles on register measure the total number of motor vehicles registered in a metropolitan area, less commercial and utility trailers. Generally high quality data broken down by municipality and class of vehicle are available from the various provincial ministries responsible for motor vehicles. Statistics Canada also compiles annual motor vehicle statistics in its annual publication titled Motor Vehicle Registrations, which also breaks data down by municipality. In some cases it is easy to get data that correspond to the metro area definitions. Where data were not available from the provincial sources for a particular year, the Statscan publications were used (this was usually necessary for 1961 and 1971 in most cities). The Statscan publication is drawn from provincial sources and a comparison of the Statscan publication with the provincial sources indicate no great discrepancies. Data accuracy is therefore considered high. Over time, sub-classifications of vehicle registrations have changed from city to city. For example, some cities lump trucks and buses into one category, some disaggregate them, some have one catch-all label for commercial vehicles and some have yet other classifications for vehicles. These categories often change over time. However, passenger cars are always classified separately. For the purposes of this study, two grouping of vehicles are used to overcome this confusion: passenger cars, which simply refer to private automobiles (cars, pickups, vans, sport utility vehicles) and total vehicles, 82 which refers to the total of passenger cars and all other vehicle classes (buses, trucks, commercial vehicles). Private transportation Private transportation indicators were obtained from the regional government agencies in Vancouver, Toronto, Ottawa and Montreal and the city governments in Edmonton, Calgary and Winnipeg. These data are usually obtained through: • computer traffic models (e.g., for VKT, fuel consumption and average network speed); • household surveys (e.g., for average vehicle occupancy); • roadside/screenline counts (e.g., for average vehicle occupancy and vehicle.kms. travelled); Data for the private transportation indicators, where available, is generally quite good, with the exception of fuel consumption in many Canadian cities (see Private Transport Energy Use below). Data items such as VKT and vehicle occupancies were not widely collected prior to 1981 in some cities. In general, data availability and reliability for all private transportation indicators is best for 1981 and 1991. Therefore, comparative analysis between data items and cities is best for these years. Total annual vehicle kilometres travelled (VKT) Total annual VKT refers to the cumulative number of kilometres travelled on all roads by all vehicles (trucks, cars, commercial vehicles) within the metropolitan area during the study year. VKT provides an indication of private motorized vehicle use, in aggregate, in each city. VKT is derived from computer models in most metropolitan areas. In Winnipeg, transportation planning staff provided data based on extrapolation from screenline counts and road volumes. VKT data were difficult to obtain for most cities prior to 1981. Most notably, Toronto, Montreal, and Vancouver were not able to provide any pre-1981 data. Edmonton was not able to provide data pre-1991. In some cities (namely Ottawa and Winnipeg), total VKT and car VKT are not modelled or calculated separately. In these cases, car VKT (see below), which is more widely available, was used to estimate total VKT. Using data available on commercial vehicle traffic and volumes, traffic engineers can make fairly accurate estimations of total VKT. Total VKT is usually 10-15% higher than car VKT in most cities. Total annual VKT in cars Total annual VKT in cars simply indicates the amount of kilometres travelled by private cars (that is, total VKT less commercial vehicles, trucks and buses). Private car VKT is more widely used than total VKT in analyses in this study as car VKT represents the bulk of VKT travelled and is usually more 83 accurate. It is also a more useful and relevant indicator for the purposes of this thesis: formulating urban transportation policies related to dependence on cars for personal travel needs. Average vehicle occupancy Average vehicle occupancy measures the average number of occupants in private cars over a 24 hour period, 7 days a week. Occupancy is usually measured from random surveys conducted by municipal and regional agencies on the driving patterns of residents. Vehicle occupancy is useful as an indicator of the occupancy intensity of cars. Car occupant kilometres Vehicle occupancy is also useful to calculate car occupant kilometres. Car occupant kilometres is the average car occupancy multiplied by car VKT and measures the total amount of passenger travel done by private cars in one year. The car occupant kilometres measure is useful in comparing the amount of passenger travel done by car versus the amount of passenger travel done by transit. Average road network speed The average road network speed measures the average speed of motorized vehicles in a region over a 24 hour, 7 day period. This figure is usually generated from transportation models. However, in Canada, it is not widely available. Generally, only peak-hour speeds are available. Since road network speeds are not widely available on a consistent basis for the Canadian cities, they will not be widely used in this study. However, they are useful as complementary pieces of data for individual metropolitan areas and are therefore provided in the data sheets for reference. Transport energy use Transportation energy use measures the amount of fuel (gasoline and diesel) consumed by private transportation within the metropolitan area. In the Canadian cities, this data is not widely available. While some modelling has been done in 1991 for some urban areas (Toronto, Vancouver and Montreal), accurate modelled data were simply not available from any planning agencies for other cities or other years. I was able to obtain fuel sales data for each of the individual cities for 1973, 1981 and 1991 for all the Canadian cities through Kent Marketing Ltd. 5 1 Kent collects fuel sales data from retail sales records of gasoline vendors in each city. However, these sales do not necessarily measure fuel consumed in the metropolitan region accurately for several reasons. While sales are a relatively accurate reflection of 5 1 The only exception was 1973 data for the City of Calgary. 84 demand for fuel in some metropolitan areas, it is not an accurate measure of consumption within a designated area. For example, some fuel may be bought within the metro area and consumed outside the metro area, or vice versa. Some areas, such as the GVRD, experience a significant amount of cross-border refuelling from residents in southern municipalities going to the U.S. to purchase cheaper gasoline. While the fuel sales data provide acceptable anecdotal evidence of fuel consumed, they should be used with a degree of caution. Journey to work data The modal split for the journey to work measure the proportion of people travelling to work in each mode (car, transit, walking and cycling). Trip lengths measure the average length of trips for work as well as other purposes. In most other countries, this information is collected from census data.52 Ideally, this provides for highly consistent data. However, Statistics Canada has only starting collecting these data as of the 1996 census. Therefore, transportation surveys are the primary source of this information. Public transport indica tors Public transit data were collected from two sources: the individual transit operators in each city and the Canadian Urban Transit Association (CUTA). CUTA is the association of Canadian transit operators and, among its many functions, serves as a research body and clearinghouse of information for them. In Vancouver, Edmonton, Calgary and Winnipeg, there is only one agency responsible for the management and operation of transit services in the region making the collection of transit data relatively straightforward. However, in Toronto, Ottawa-Hull and Montreal multiple transit companies operate within the region. The Ottawa-Hull region has two transit operators: O-C Transpo on the Ontario side and Societe Transport Outaouais (STO) on the Quebec side. In the Greater Toronto area, data were collected from the Toronto Transit Commission (TTC), which covers Metro, as well as from GO Transit (regional commuter transit) and 14 transit agencies operating outside of Metro.53 In Montreal, transit data were collected for STCUM (Societe de Transport de la Communaute urbaine de Montreal), which serves the MUC, as well as for the two off-island operators in Laval and the South Shore. Both Canadian Pacific (CP) and Canadian National (CN) provided data for commuter rail operations in the region. Questions on the census wil l typically ask the mode of transportation used and the work address, which are then used for making mode type and distance calculations. 5 3 For Toronto's 1961 and 1971 data, only TTC statistics were used. This is because data were only available for Metro in these years, as opposed to the entire GTA. See "territorial area definitions" on page 74 for a discussion of the use of Metro data for 1961 and 1971 versus G T A data for 1981 and 1991. 85 In most cities, accurate records of transit system performance have been kept for all study years (back to 1961) as part of the regular reporting of operations. However, some data were compiled using different methodologies, particularly in earlier years (1961, 1971) where data collection standards were not as detailed. Research staff at CUTA assisted by reviewing the data, filling in gaps and providing corrections as necessary. The data for Canadian transit operators were rigorously review and are considered accurate and reliable. Two data items, passenger boardings and trip length, are considered to be "softer" data. The implications are discussed below. Vehicle kilometres travelled Transit vehicle kilometres travelled are the total number of kilometres travelled in revenue operation by a transit operator in one year. The data typically provide a the transit vehicle kilometres by mode (motor bus, trolley bus, LRT, subway, commuter rail, ferry, etc.). This is a standard reporting item for transit operators. Data are considered accurate as operators have detailed knowledge routing and service. Some transit operators include what are known as "deadhead" kilometres (non-revenue driving such as going back to the garage empty), however data screening ensured these kilometres were excluded. Passenger boardings Passenger boardings are the total number of unlinked trips taken by passenger in one year. An "unlinked" trip refers to any boarding of a transit vehicle. A "linked" trip refers to one trip, origin to destination. In many cities, passengers may "transfer" from one vehicle to another to complete a trip, using only a paper transfer. Therefore, calculating unlinked trips precisely is not possible. To calculate "unlinked" trips, transit operators will provide an estimation of the 'transfer rate' (usually approximately 70%) which is then used to "delink" the trip numbers provided. Some transit agencies will automatically report both linked and unlinked trips. The use of linked trips would be simpler in the Canadian context, however, the majority of the transit operators in the global CAAD study report unlinked trips because of the nature of their fare structures (for example, some systems charge a fare upon each boarding or have automated fare systems). To ensure comparability and standardization between all cities the convention of using "unlinked" trips was adopted. Brendon Hemily, Research Manager at CUTA, notes that transit operators in Canada are asked to record the "estimated percentage of revenue passengers transferring on regular service" as part of their annual reporting to CUTA. Hemily indicates that transit systems believe that, on average, their boardings estimates are 93% accurate. However, since the transfer rate probably does not include second or third 86 transfers by revenue passengers, boardings are likely underestimated and "no better than 90%" accurate (Hemily 1998). Average trip length Average trip lengths by public transit are reported by transit agencies and are estimated through regular trip surveys conducted by transit operators of ridership travel characteristics. Sometimes average trip lengths are also reported by regional government agencies collecting information on regional travel. Passenger kilometres travelled Transit passenger kilometres provides an estimation of the total amount of passenger travel on transit. It is calculated by multiplying the number of unlinked trips taken by transit by the average trip lengths of these trips. Since this number involves the multiple of these two estimates, it is considered a "soft" indictor. While passenger kilometres are not highly precise, the estimates of trip lengths and 'unlinked' trips provide an acceptable degree of accuracy for analysis. Transit passenger kilometres is a particularly useful indicator when used in conjunction with car occupant kilometres. Taken together, they indicate the proportions of regional passenger travel taken by car and by transit. Average speed The average speed of transit is simply the total number of revenue kilometres travelled by all vehicles while in revenue service, divided by the total number of hours logged (that is, total distance in kms. divided by total time required hours). These speeds are disaggregated by mode where possible. Average speed of transit is useful as an indicator of transit priority in cities and can be compared to car speeds to determine speed competitiveness. Energy consumption Energy is a major operational cost for transit agencies, therefore its usage is closely tracked for financial reporting. Energy data are supplied by transit agencies in total litres of diesel, gasoline or compressed natural gas (CNG) or in kilowatt-hours (KWh) for trolleys and rail. For ease of use, these are all converted and expressed as joules of energy. 3.3.1 Standardized Data The standardized data appear after the raw data for each city in Appendix 1. These data provide indicators of activity intensity, per capita use, relative infrastructure and service supply, intensity of infrastructure use and transport efficiency. The standardized data combine raw data items to calculate per 87 unit expression of a parameter. This standardization is achieved by normalizing data based on population, jobs, vehicle kilometers, hectares, etc. Al l data items in the 'numerator' are matched with data corresponding to the same year and geographical area in the 'denominator,' ensuring indicators are internally consistent and comparable. Where data items were provided for non-study year because of data availability problems, the corresponding population, jobs or area were used for that year (for example, if 1962 vehicle registrations were provided, the 1962 population would have been used for standardization). 3.3.2 Use of Data and Data Reliability There are several issues one should be aware of in using and interpreting the data presented in this thesis. Firstly, caution should be exercised in the interpretation of raw and standardized data. Standardized data merely allow for comparative analysis of trends and patterns amongst cities. However, in terms assessing quality of life and sustainability, it is the raw data that provide the most useful information about absolute changes in transportation and urban structure. Standardized data can also be deceptive if caution is not employed in its interpretation. For example, a city may experience a levelling, or even decrease, of per capita VKT between two study years. However, VKT may still have increased dramatically in absolute terms. A per capita increase will only result if the growth rate of the data item is greater than that of the population. Growth rates and averages also require caution in interpretation. A growth rate of 10% in VKT in Toronto results in roughly 1 billion extra vehicle kilometers a year, whereas a 10% growth rate in Los Angeles results in 10 billion extra vehicle kilometers driven. The average of Canadian cities, for example, will refer to city averages rather than weighted city averages. This means that Winnipeg (population, 640,000) is given the same weight as Toronto (population, 4.2 million). Again, this provides a picture of the performance of policy in the average city in a country rather than an indication of their overall 'sustainability.' Secondly, relying on quantitative data alone as an assessment of the performance of land use and transportation in a city is insufficient. Qualitative assessments of city characteristics and policies must also be used for a more complete and accurate understanding. For example, two areas with similar and high population densities may not be "equal" in their pedestrian and cycling friendliness. Several other intervening qualitative factors, such as the concentration of densities, the quality of the pedestrian environment, the width of road and availability of traffic calming will also influence pedestrian friendliness. The quantitative data in this thesis only provide a limited range of the information necessary and should be complemented with qualitative information. 88 Finally, while the data used in this thesis are of a high quality for comparative purposes, they do vary in degrees of reliability. Table 23 below provides an assessment of the general quality and reliability of the data items by city and year. Some data in this study are 'hard', while others are 'soft.' Hard data come from census materials, detailed inventorying, registration databases and the like. Meanwhile, soft data are usually those that come from transportation models (in the case of private transportation) and surveys (in the case of trip lengths and transfer rates). While there is a larger margin of error in modelled and surveyed transportation data than in the hard data, they still provide reasonable estimates of the parameters. In the case of surveys, the sample sizes are usually large enough to provide reliable data. Though some modelled data are 'soft,' they may be considered reliable based the relative quality of the modelling and acceptable as they represent a well-informed "best guess" estimate (Litman 1995). Furthermore, most of the models and methodologies used are similar from city to city, maintaining internal consistency.54 Therefore, the indication of error provided for the soft data reflects the quality of the modelling, the degree of internal consistency and the level of trust afforded them as reasonable and confident estimations. Where softer data are used, only notable differences are highlighted. For example, while 5-10% increase in VKT from study year to study year may be within the realm of error, a 10-20% or greater increase may indicate notable change, particularly if such growth persists over time. Other potential errors with data include false accuracy and the multiplication of error. False accuracy involves indicating an unrealistic level of precision from imprecise data sources. For example, vehicle occupancies and transit transfer rates involve surveying and averaging. Indicating a precise result implies precision is possible. Where false accuracy is a concern, data are rounded. Multiplication of error occurs when two parameters with some degree of error are used in conjunction with one another (Hardin 1986). The combined error is larger than those items used as multiples. A level of confidence equal to the multiple of these errors should be place on such derived data. 5 4 For example, the EMMEII transportation model is used in every major region in Canada to model transportation use. 89 _Q _ra "3 OH J I ra Q i CD Q) x> c CO CO D ro "co Q • oo CM 0) _Q CO E o 4-* ra ra Q Montreal 1.661. | Population/Area | in LO LO LO LO LO LO LO LO LO LO | Private Transport | LO LO LO LO in in Transit | LO T J - T J - T J - in LO Montreal 1.961. | Population/Area | LO LO LO LO LO LO LO LO LO I LO | Private Transport | in m in i LO LO Transit | LO •*t T J - T J - in LO Montreal VLQV | Population/Area | in LO LO LO LO LO LO LO LO CO | Private Transport | i 1 • CO • Transit | LO TJ- T J - T J - • LO Montreal 1.961-| Population/Area | LO LO LO LO LO LO i I I I CO | Private Transport | • i • 1 • • • Transit | LO T J - • Tj-Ottawa-Hull L66L | Population/Area | LO LO LO LO LO LO LO LO LO in in LO | Private Transport | LO LO CO CO LO CO Transit | in T J - LO in Ottawa-Hull 1.861. | Population/Area | LO LO LO LO LO LO LO LO i | Private Transport | LO LO CO CO LO CO Transit | m T J - TJ- LO LO Ottawa-Hull VLQV | Population/Area | LO LO LO LO LO LO LO LO in • CO | Private Transport | I 1 • • CM CO 1 Transit | LO T J - T J - T J - I LO Ottawa-Hull 1-961. | Population/Area | CO TJ- m LO LO LO CO CO CO • • CO | Private Transport | CN CM CM • 1 CO 1 Transit | T J - CO CO T J - • T J -Toronto 1-661. | Population/Area | LO LO LO LO LO LO LO LO LO LO LO | Private Transport | LO in LO Tf LO LO Transit | LO TJ- in LO in LO Toronto 1.861. | Population/Area | LO •* LO LO LO LO LO LO LO LO I | Private Transport | in in in 1 TJ- LO LO Transit | o o c^  Toronto LZ6L | Population/Area | LO LO LO LO LO LO LO LO LO LO LO LO | Private Transport | • • • 1 1 I Transit | LO TJ- LO m m LO Toronto 1.961. | Population/Area | LO LO LO LO LO LO LO LO LO in in in | Private Transport | i i 1 1 LO Tj-Transit | LO Tt in LO LO LO Winnipeg 3661. | Population/Area | LO LO LO LO LO LO LO LO LO in in in | Private Transport | <* -<* CO CO CO LO •st Transit | in -* T J - T J - in LO Winnipeg 1.861. | Population/Area | lO LO LO LO LO LO LO LO in LO • LO | Private Transport | CO CO CO LO 1 Transit | LO TJ- CO T J - LO in Winnipeg UGl | Population/Area | LO LO LO LO LO LO LO LO LO in | Private Transport | CO 1 CM LO • Transit | LO T J - CO CO LO in Winnipeg 2961-| Population/Area | LO LO LO LO LO LO LO LO LO LO • rr | Private Transport | CO CO i LO 1 Transit | m 'TJ- CO CO LO CO Calgary 1-661. | Population/Area | LO LO LO LO LO LO LO LO LO LO LO in | Private Transport | LO LO •<t Tj- •<* LO CO Transit | LO co m Tj- LO LO Calgary 1-86I-| Population/Area | LO LO LO LO LO LO LO LO CO LO LO LO | Private Transport | in LO TJ- Tf CM LO 1 Transit | LO co T T TJ- in LO Calgary I.Z6I-| Population/Area | LO LO LO LO LO LO LO LO in in LO | Private Transport | in LO •>*• Tf i LO 1 Transit | LO CO CO CO in T J -Calgary 1-961-| Population/Area | LO M - LO LO •<fr CO in in | Private Transport | Tf TJ. CO i I 1 Transit | in CO CO CO LO TJ-Edmonton 1.661. | Population/Area | LO LO LO LO LO LO I in in | Private Transport | CO CO CO CO CM LO Transit | LO TJ- CO rr LO in Edmonton I-861-| Population/Area | LO CO LO LO LO LO LO LO I • LO | Private Transport | i 1 1 1 i 1 • Transit | in CO CO CO LO Tf Edmonton UZ6U | Population/Area | LO LO LO LO m ' m | Private Transport | • 1 1 1 i 1 Transit | in CO CO CO I CO Edmonton 1-961. | Population/Area | LO I in LO LO LO | Employment I LO I in • m | Private Transport | i 1 1 1 i 1 i Transit | m CO CO CO I CO Vancouver 1.661-| Population/Area | LO LO LO LO LO LO | Employment LO LO LO LO LO in | Private Transport | LO LO Tj- Tj- Tt LO Transit | in CO Tj- Tj- LO LO Vancouver 1.861-| Population/Area | LO LO LO LO LO LO | Employment LO LO LO in LO LO | Private Transport | LO LO T* TJ. T J - i i Transit | LO CO Tj- Tj- in Tj-Vancouver LZ6t | Population/Area | LO LO LO LO LO | Employment LO LO LO in LO in | Private Transport | I • 1 • <N i i Transit | in CO 1 1 LO CO Vancouver I.96I-| Population/Area | LO LO LO | Employment LO I I in in LO | Private Transport | I 1 • 1 i i • Transit | in CO 1 1 i CO | Population/Area | | Total pop. | Urbanized area | CBD pop. | | CBD area | Inner city pop. | | Inner city area | | Employment | Region | | CBD | | Inner city | | Parking | Road Network | Vehicles Registered | | Private Transport | 1 Total VKT | | Car VKT | 1 Car occupant km. | | Avg. road speed | | Private Energy | | Mode Split | Trip Length | Transit | 1 VKT | [ Pass. Boardings | Avg. trip length | Passenger km. | Avg. Speed | | Energy Used | o ON LO y^  *-« c 03 0) O X LU II LO > . CD 3.4 METHODS OF ANALYSIS The data collected for this thesis will be analyzed to determine the extent to which the variables studied influence car use, car ownership and transit use (i.e., manifestations of car dependence). The analysis in the following chapter will consist of a trend and pattern analysis that will identify some of the basic relationships between the variables studied. A correlation analysis will then be used to determine how strongly the variables influence, or are influenced by car use, car ownership and transit use. Correlation analysis serves to measure how well two variables vary together, or how strongly they are related (Gujarati 1988; Kenkel 1984). Of course, no causal relationships can be determined, but knowing the strength of the association between these factors can help to reconcile quantitative findings with theory help and provide some initial directions for policy interventions. While a multiple regression analysis would be preferable in terms of its explanatory and predictive value,55 it would require a much more rigorous statistical treatment that is beyond the scope of this thesis. Issues related to the findings of the correlation analysis will be discussed in greater detail in Chapter 4. 3.5 CONCLUSIONS Compiling useful comparative urban data is a difficult task. It requires much time, intense scrutiny and a sound methodology. The availability of comprehensive and standardized transportation and urban form data can be extremely useful in policy analysis. However, the compilation and analysis of such comprehensive information has not yet been done in Canada. Where partial data are available, the methodology in collecting them and their comparability are often questionable. A more rigorous process of collecting and screening data can provide reliable data for urban transportation policy analysis. The methodology outlined in this chapter provides a means for collecting data that are reliable and highly comparable. The data judicious use of these data can aid an understanding of some of the For example, one variable, such as car ownership, could be held as dependent and a series of other independent variables (density, transit supply, parking supply) can be simultaneously tested against car ownership for their explanatory value. The resultant regression equation can be used to 'predict' the dependent variable, car ownership, given changing values in density, transit supply or parking supply. 91 major forces contributing to higher or lower levels of car use and ownership. They can also be a useful adjunct to a policy analysis determining which public policies relieve or exacerbate auto dependence. 92 CHAPTER 4 - FINDINGS: TREN D, PATTERN AND CORRELATION ANALYSIS 4.0 INTRODUCTION This chapter provides an analysis of the data collected for this thesis. It draws on the larger sample of cities (46 in total) included in the CAAD update for context, but focusses on developments in the Canadian cities. This chapter has four main components: 1. a general overview of the urban development and transportation trends in Canada from 1961 to 1991; 2. a more detailed analysis of how these urban development trends are manifest in each of the seven Canadian cities; 3. an analysis, focussing on the 1991 standardized variables, of the factors that are correlated with higher or lower levels of automobile dependence; and 4. a summary of the key findings of the trend and correlation analyses. Again, the raw and standardized data referred to in this thesis can be found in Appendix 1, for reference. 4.1 OVERVIEW OF THE TRENDS 1961-1991 Urban development in Canada's seven largest cities in the 1961-91 period has been marked by rapid population increases, urban sprawl, the decentralization of population and employment and inner city decline. These trends have been accompanied by increasing car ownership, use and infrastructure provision, as well as transit decline. While car ownership and use have increased in real terms relative to population increases, transit use has dropped. Furthermore, this decline in transit use came despite the fact that transit service per capita (as measured by transit VKT per capita) has increased in most Canadian cities over this period. One 'bright spot' in the evolution of transportation over this period was a marked revival in transit use between 1971 and 1981 because of a substantial transit investment in most cities. However, transit continued its decline thereafter, and Canadian cities have since continued along the path of increasing automobile dependence. 4.1.0 Urban Growth Trends - Inner City Decline and Sprawl The post-war suburbanization of Canadian cities into exclusively zoned residential sub-divisions was well underway by 1961 (Linteau 1990). Therefore, the 1961 to 1991 data in this study only capture part of this post-war suburbanization picture. However, these data do reveal the continued and accelerated pace of population and employment dispersal throughout Canadian urban regions. 93 Figure 10 below shows the distribution of population growth between the inner and outer areas of the 7 Canadian cities in this study between 1961 and 1991. During this 30-year period, inner city populations declined in most cities to varying extents, while outer area population increased dramatically. Some cities did enjoy moderate inner city growth between 1961 and 1971. However, there was a universally dramatic drop in inner city population in all cities between 1971 and 1981. In this period alone, the decline amounted to 21% of the total inner city population of the study cities. Not only did the outer areas absorb the population losses of the inner areas, they also absorbed most of the residential population increase that resulted from urbanization, immigration and natural population growth. Figure 10 - Inner and outer area population in the Canadian cities. 1961-91 0 ^ ' 1961 1971 1981 1991 Figure 11 below demonstrates how inner cities have declined in regional importance in the 1961-91 period. Inner areas contained a substantial 74.8% of these cities' populations in 1961. This share of regional population declined in each successive study year to represent roughly one quarter of regional populations by 1991. Despite the losses in inner city population during this period, Canadian cities still have inner cities that are relatively well populated and vibrant compared to their U.S. counterparts (Raad and Kenworthy 1998) and some seem to be experiencing a healthy 'reurbanization' post-1981 (though modest compared to outer-area growth). Employment, too, has been increasingly decentralized in Canadian cities, as indicated in Figure 11 below. Unlike inner area population, inner area jobs56 grown in absolute terms in most Canadian cities 'Employment' as a term, differs from 'jobs' in the context of this thesis. Employment refers to the number of people that are employed within a specified area. However, these people may or may not be employed within that same area they live in (e.g., they could live in the inner area and be employed, but that employment may be in the outer area). 'Jobs' refers to the number of jobs located in a specified area (i.e., place of work). For example, the 'number of jobs in the inner area' refers strictly to the number of jobs located within that area, regardless of how many people in that area are "employed". 94 from 1961 to 1991 (some cities saw declines in 1981). However, outer area jobs growth has continued to outpace inner area growth. Figure 11 - Inner area share of regional population and jobs in the Canadian cities. 1961-91 80% r 70% 60% 50% 40% 30% 20% 10% 0% 74 8% 55 9% 60 1% 41 6% 46 9% 44 2% 31 0% 27 0% 1961 1971 1981 1991 H Inner area population as % of metro El Inner area jobs as % of metro Figure 12 - Growth in developed land vs. population in the Canadian cities. 1961-91 120% 57 100% co - 80% CD O » 60% % O O 40% 20% 0% Developed land 88% 1961 1971 1981 1991 The main concern with the declining importance of inner areas in terms of population and jobs is that the outer area shares are increasingly accomodated at very low densities, thereby consuming land and making developments difficult to service by transit. Canadian urban growth in the 1961-91 period has consumed ever increasing amounts of undeveloped land (Figure 12 above). In fact, during this period the growth in land consumption outpaced the growth in population by over a factor of two. Much of this land These figures exclude Edmonton. 95 consumed for urban growth has been prime capability agricultural land (Environment Canada 1981; Manitoba Environment 1995; Warren, Kerr, and Turner 1989). The faster growth in developed urban land versus population resulted in declining population densities. Between 1961 and 1991, the average urban population density of the cities in this study dropped from 36.1 to 25.5 persons per hectare (p/ha). Outer areas (where most growth was concentrated) continued developing at average densities as low as 14.6 p/ha. Furthermore, this development was often exclusively zoned, formless and difficult to serve by transit. Perl and Pucher (1995) indicate that the urban decentralization and suburban development characteristic of much of this outer area development preclude the viability of virtually any mode but the car. The decline in total urban densities and continued spread of low-density outer areas is particularly worrisome given empirical studies showing transit ridership and viability, as well as the suppression of car ownership levels, are greatly diminished at densities below roughly 30 p/ha (Newman and Kenworthy 1989a; Pushkarev and Zupan 1977). As will be pointed out later in this chapter, the decline in urban density has been more accute in some cities than others. Even in cases where density is increasing, sprawl continues unabated. Figure 13 - Average urban densities in 46 World cities. 1990/91 CO j= to c CD n 2 CD £0! In the international context, Canadian cities sits atop a league of low-density cities (Figure 13 above). Canadian urban densities pale in comparison to those in wealthy and developing Asia. However, Canadian densities are still more than twice as high as their U.S. and Australian counterparts, which followed similar car-based urban development models in the post-war years. In turn, the European cities are nearly twice as dense as the Canadian ones. The density profile of Canadian cities midway between its transit-oriented European and auto-oriented U.S. and Australian counterparts provides an interesting 96 positioning for comparative purposes. As I will show shortly, in terms of transportation, the European cities offer an urban model to aspire to while the U.S. and (to a certain extent) Australian cities offer ones to avoid. 4.1.1 Transportation Trends - Increasing Car Dominance and Transit Decline A phenomenon that parralels developments in urban growth in Canada is the increasing dominance of the car's role in urban transportation and transit's declining role. Private car ownership, use and infrastructure provision have all grown faster than population in the 1961 to 1991 period. Meanwhile, transit use in Canadian cities has transit has been declining in real terms, failing to keep pace with population growth. This decline comes despite the fact that transit service provision (in terms of transit vehicle kilometres (VKT) per capita) has increased almost universally in every year since 1961. Similar to the manner in which the growth in developed land outstripped population growth, so too did car ownership. During the 1961-91 period, car ownership growth was 2.5 times the growth in population, rising from 1.3 million vehicles in the seven study cities in 1961 to just over 3.3 million in 1991. Growth was particularly strong in the 1971 to 1981 period, coinciding with the inner city decline and outer area boom mentioned earlier. Figure 14 - Growth in car registrations versus population in the Canadian cities. 1961-91 180% 160% H Car Ownership • 155^ 140% -CO 120% -118% 100% -80% -60% - Population ,^60% 40% -20% H 0% J-base ye; 1961 1971 1981 1991 97 Figure 15 - Average modal split in the Canadian cities. 1961-91 1961 1971 1981 1991 -•-Transit - • - C a r * Walk/Cycle The changing modal shares for cars, transit and walking and cycling (or non-motorized transportation - NMT) reveal an overall decline in transit and NMT shares and an overall increase in the share of car use.58 One particularly worrisome trend is the continued decline of NMT as a share of total travel. One possible explanation for this trend is that roadway severence, automobile priority and traffic domination lead to the deterioration of pedestrian and cycling environments (Appleyard 1981; Calgary 1994a; Moudon et al. 1997; Perkins 1993; Sarkar, Nederveen, and Pols 1997; Zein et al. 1997). Since transit requires high quality pedestrian and cycling environments to improve its catchment potential, this development does not bode well for transit's long-term viability. Again, the bright spot for transit in the 1961-91 period was the ridership gains made between 1971 and 1981, and the concomitant reduction in car use share.59 During this period, provincial governments throughout Canada were able to head off the major financial and ridership hemorraging of transit services experienced in the U.S. by extending healthy subsidies to transit systems to keep fares low and expand services (Perl and Pucher 1995; Pucher 1994; Pucher 1998). However, these gains proved unsustainable. As will be shown in the next section, transit continued its decline in per capita ridership and in per kilometer utilization post-1981, despite the fact that transit service provision is higher than 1961 levels in absolute terms in every Canadian city (as well as in per capita terms in most Canadian cities). Demographic forces have been responsible for part of the shift from transit to cars (CUTA 1991). 5 8 These modal split figures represent the modal split for the journey to work. All-day modal splits usually reveal higher car shares and lower transit and NMT shares. 5 9 This trend is somewhat of a mixed blessing. Although transit shares increased, the total demand for travel increased, and this increase tended entirely towards motorized modes (public and private). So while transit captured a larger share of the increase in travel, car travel still increased in total. 98 However, much of the decline in transit ridership and performance, despite the service increases, can also be can be attributed to the higher servicing needs (Perl and Pucher 1995) and the lower ridership returns characteristic of newer low-density suburban services. Transportation infrastructure geared towards automobile use has also increased over the last 30 years. For example, parking supply growth has consistently outstripped jobs growth in the CBDs of the Canadian cities (Figure 16 below)60. The effect of this has been to make it increasingly easier (and cheaper) for CBD-bound travellers to drive and park their vehicles. Generous parking supply guidelines in many cities have been effectively demand-driven, responding to commercial demands for easy car access to properties. Many Canadian cities have have generous minimum parking standards, while relatively few have parking ceilings (Calgary 1995a). Where ceilings do exist, they are frequently exceeded. Only a select few cities actively enforce tight CBD parking supply. This high supply of CBD parking may actually have the effect of deriving demand for more auto trips (Shoup 1997). Figure 16 - Growth in CBD parking supply versus CBD jobs in four Canadian cities.— 1961-91 80% CD CD CD O C CO 60% -2 40% | 20% 0% Q 52% 2 7 % / / 1 7 %>^^^*2 - 1 % base yeao ^ X ^ - ^ 1 2 % 1961 1971 1981 1991 - • — C B D parking stalls —©—CBD jobs Road infrastructure provision has also increased along with the outward spread of the Canadian cities. Though the use of centre-line kilometers is an admittedly crude measure of car capacity, the data 6 0 The graph understates the degree to which parking supply growth has outstripped jobs growth since it includes Toronto (which has tightened its supply and represents a large proportion of the jobs and parking) and excludes Montreal and Vancouver (which have had parking supply expanding faster than jobs). 6 1 These cities are Calgary, Edmonton, Winnipeg and Toronto. Vancouver, Ottawa and Montreal were left out because of a lack of comprehensive time series data. 99 in this thesis show roughly a doubling of road length per capita in most cities. For example, lane kilometre data, which more accurately measure road supply, show substantially higher road provision in Ottawa and Winnipeg (RMOC 1994; Winnipeg 1995). The increasing ease of access by automobile to the CBDs, and throughout the regions at-large, has likely produced in many of the synergic side effects on land use and environmental quality discussed in Chapter 2. Figure 17 below captures the basic problem with transsportation developments in the Canadian cities in this study over the last 30 years. While automobile ownership (cars per 1000 people) and automobile use have grown faster than population since 1961, transit use has consistently lagged behind. Furthermore, the increase in car use is accompanied by dropping average vehicle occupancies (Figure 18 below). While this relative loss in the importance of transit is of concern, it is even more significant given the large increases in private motor-vehicle ownership and use in absolute terms. Figure 17 - Growth in car ownership, car use and transit use versus population in the average Canadian citv. 1961-91 100% 1961 1971 1981 1991 -•—Cars/1000 - * — VKT/capita • Transit boardings/capita •Population 100 Figure 18 - Average vehicle occupancies in the Canadian cities. 1961-91 —=T53 1 1.65 £ 1.60 1.55 §• 1.50 o o 2 1.45 co o CD 2> 1.40 1.30 1961 1971 1981 1991 Figure 19 - Average transit share of m otorized travel in 46 World cities, by region 1990/91 70% o 60% cn 50% 40% CD > 2 xs CD N 3 30% o E S 20% o o CO CO c CO 10% 0% 64.1% 22.6% 9.6% 3.1% 41.0% f£>f 0<v In the international context, Canada is relatively automobile dependent. Canada's position in terms of transportation orientation mirrors its position in terms of land use. For example, its transit use is among the lowest in the world as a proportion of total motorized urban travel (Figure 19 above). Canada outperforms the U.S. and Australia, falls well behind the European cities and is nowhere near the very high levels of transit use found in Asia. In terms of automobile use, similar patterns arise. U.S. and Australian cities drive an average of 67% and 8% more VKT per capita than Canada, respectively. Meanwhile, the Canadian cities drive 50% more on average than the European cities. Some indicators further describing Canada's relative position in terms of urban transportation performance are presented 101 in Table 24 below. As will be discussed in subsequent sections, these same distinct transportation patterns emerge for cities depending on their urban density, transit supply and provision of automobile infrastructure. Table 24 - Land use and private and public transportation indicators in World regions. 1990/91 Region Land Use Intensity Car Use Cars Owned CBD Parking Transit Use % of workers using3 pop. + jobs/ha VKT/ capita cars/ 1000 pop. stalls/ 1000 jobs boardings/ capita car transit NMT United States 22.3 12,336 608 468 63 86.3 9.0 4.9 Australia 17.5 8,034 491 483 92 80.4 14.5 5.0 Canada 42.9 7,406 499 408 161 74.0 20.0 6.0 Europe 81.4 5,026 392 230 318 42.8 38.8 18.4 Wealthy Asia 275.1 2,688 109 72 487 20.2 59.6 20.2 Dev'ping. Asia 215.9 2,093 110 240 356 41.5 31.8 26.8 Note: Modal splits may not total 100% due to rounding. In terms of trends, Canada, like Australia and the United States, experienced relatively low growth (11-12%) in vehicle ownership per capita between 1981 and 1991 (see Figure 20 and Table 25 below). However, these growth rates understate the true extent of the expansion in vehicle ownership as they added to already high levels of car ownership (in both per capita and absolute terms). For example, developing Asia's per capita increase of 104% yielded 56 additional cars per 1000 people whereas Canada's much lower 12% growth yielded an additional 52 cars per 1000. During this same period, Figure 20 - Change in car ownership a nd use in the World cities, by region. 1981-91 14,000 700 600 CD CL O 500 0 CL O Q 400 O 1— CD 300 CL cn i _ CO 200 O 100 0 608 547 499 447 • 4 6,022 ,404 0,013 443 2,336 491 *7 083 034 392 332 3,026 2,688 1,814^>_ 109 70 110 S B 12,000 10,000 m CD •4—' 8,000 g L _ CD CL 093 6,000 4,000 2,000 0 > 80/81 & 90/91 Cars •80/81 & 90/91 VKT 102 Canadian cities experienced strong growth in car use relative to other regions. Average car VKT in Canadian cities grew by 23% from an average of 6,022 VKT to 7,406 VKT. While Europe and the U.S. also experience the same level growth, Canada's performance was worse than Europe's, since theirs grew from relatively low level, and better than the U.S., since the U.S. grew from already very high levels of VKT. Table 25 - Change in average car own ership and car use in World cities, by region. 1981-91 Car Ownership Car Use 1981 1991 % change 1981 1991 % change United States 547 608 11% 10,013 12,336 23% Australia 443 491 11% 7,083 8,034 13% CANADA 447 499 12% 6,022 7.40G Europe 332 392 18% 4,120 5,026 22% Wealthy Asia 70 109 56% 1,814 2,688 48% Developing Asia 54 110 104% 1,257 2,093 67% 4.2 A CLOSER LOOK AT THE CITIES While the preceding overview of urban developments paints of general picture of how Canada has developed as a nation, a disaggregated look at Canadian urban realities reveals distinctiveness of trends and patterns in individual cities that can lay the foundation for policy analysis. This section provides a comparative analysis of the development of individual Canadian cities. The comparisons clearly show that regions with lower urban densities, higher infrastructure provision for automobiles and lower levels of transit service have higher car ownership and use, as well as lower transit patronage.62 4.2.0 Urban Growth Trends The decline in the relative importance of inner areas combined with strong outer area growth has been characteristic of most Canadian cities. This is particularly the case in Calgary, Edmonton, Montreal and Ottawa. However, Canadian cities on the whole still have quite well populated inner areas. Some cities, such as Toronto and Vancouver, actually seem to be experiencing a notable 'reurbanization' of their inner areas. Between 1981 and 1991, they each experienced inner area population increases of over 47,000 and 43,000 people, respectively. However, despite such gains (and the general stabilization of 6 2 No discernible relationships between these factors and NMT use could be determined because of the lack of availability of NMT data beyond course modal split data. 103 inner area populations throughout Canada), the relative importance of inner areas continue to diminish amid low-density sprawl on the periphery. Figure 21 below traces the growth in population of both the inner areas and total metropolitan areas of the Canadian cities in this study. While all the cities have faced various degrees of inner area decline, all have experienced a burgeoning of their outer areas. Calgary and Edmonton continued to experience losses in inner area population up until 1991, while Winnipeg and Ottawa seem to have witnessed a stabilization of their inner area populations. Montreal continues to see a substantial bleeding of its inner area population in the face of an explosive growth in its suburbs. Figure 21 - Inner and metro area populations in 7 Canadian cities. 1961-91 1,600,000 flOnOna 3,500,000 3,000,000 c o 2,500,00012 Q_ 2 , 0 0 0 , 0 0 0 § . co CD 1,500,000™ c CO i,ooo,oooi Q. o 500,000 is 0 The increasing concentration of population growth into outer areas, means that the importance of centralized activity in inner areas is declining (Figure 22 below). In Edmonton and Calgary, this declining importance of the inner areas has partly been due to inner city population losses. But most of this loss has been a function of rapid metropolitan growth, with the majority of the population increase settling in outer areas. Their metropolitan populations increased by 2.2 and 2.8 times, respectively, and almost all of this increase settled in the outer areas. While Edmonton's inner city population declined from 213,000 to 196,000 people between 1961 and 1991 (a drop of 8.7%), its outer area increased from 63,000 to 418,500 (an increase of 565%). Calgary's pattern of suburban settlement is even more unbalanced with an outer area growth of 731%. While Calgary and Edmonton's declining inner area importance is almost singularly accounted for by metropolitan growth, Montreal's large decline is a dual function of massive hemorrhaging of their inner city population base and the suburbanization of new settlement. Between 1961 and 1991, 104 Montreal's inner area population declined from 1.361 million to 883,000 (a drop of 54%). It is difficult to pinpoint precisely what ails Montreal in that there are many complex forces at play. For example, the loss of inner city population and economic importance has been attributed to political uncertainty due to separatism (Jacobs 1980) as well as economic restructuring and industrial obsolescence (Barber 1995; Coffey 1994). However, metropolitan population in Montreal still increased significantly between 1961 and 1991 and most of this increase settled in relatively low-density suburbs in Laval and on the South Shore. Significant enabling factors that facilitated this dramatic outer area growth in Montreal were political fragmentation, significant regional highway investments, the extension of trunk sewers lines and the relaxation of regulations governing the subdivision of rural land (Frisken 1994b; Sancton 1994; Trepanier 1994). Figure 22 - Proportion of regional population living in Canadian inner areas. 1961-91 - 90% r : CD i Vancouver and Toronto provide interesting cases in that have both experienced a notable process of 'reurbanization' in their inner areas. Vancouver's inner city experienced a net growth in population of 18% between 1961 and 1991, the largest gain of any city. While much of the post-1981 growth in Vancouver has been driven by immigration, conscious policies by successive civic administrations since the mid-1960s have encouraged significant densification and infill in areas such as the West End and False Creek. Additionally, densification was encouraged throughout city neighbourhoods by encouraging the conversion of single family dwellings to multi-unit dwellings. Nonetheless, this inner area population increase pales in comparison to the 165% increase in Vancouver's outer area that brought an additional 685,000 people into the region at very low densities. 105 Toronto, too, has been able to reverse its inner city decline since the mid-1970s through conscious policy and the benefits of immigration (Bourne 1992). However, like Vancouver, its outer areas have also experienced rapid growth far exceeding that of the inner city. In fact, this growth (and the decline in inner city share of regional population) is even more pronounced when the entire GTA is considered.63 Figure 23 - Metro. CBD. inner and outer area densities in the Canadian cities. 1961-91 I I Metro Area —©— CBD —•— Inner Area —A—Outer Area Figure 23 above shows the relative densities of the Canadian cities between 1961 and 1991. Metro Toronto and Montreal are by far the most dense Canadian cities (41.5 and 33.8 p/ha, respectively),64 followed by Ottawa and Edmonton in the 30 p/ha range and Winnipeg, Calgary and Vancouver in the 20 p/ha range. Toronto and Montreal have the highest inner area densities in the 60 p/ha range. Ottawa, Vancouver and Winnipeg follow them in the 40 p/ha range, with Edmonton and Calgary having relatively low density inner areas in the mid-20 p/ha range. In terms of outer area density, Metro Toronto is the most dense at 35.4 p/ha. Edmonton, Ottawa and Montreal fall in at the 30 p/ha range and Calgary, Winnipeg and Vancouver have lowest density outer areas. Since Metro Toronto has been largely built-out since 1981, outer area growth and some transportation data for Metro will understate GTA-wide developments to some extent. Again, the lack of availability of comprehensive data for the GTA precludes their use in lieu of Metro data for the comparative and statistical analysis in this chapter. Metro data alone should therefore be interpreted with caution as they only provide a partial picture of regional developments. Instead, I will discuss GTA-wide developments throughout this chapter, where GTA data are available. Nonetheless, since the boundaries of Metro are fixed, they do provide useful information about how a transportation system may perform given a degree of urban containment. 6 4 The GTA has a population density of 26.9 p/ha. 106 As mentioned in the previous section, the outward spread of cities with developed land growth exceeding population growth has resulted in dropping urban densities (Figure 23 above). A common thread in the density profile of all the Canadian cities is the decline of inner area densities and the proliferation of lower density outer areas. The most precipitous drop in urban densities was experienced in Montreal between 1961 and 1971 (dropping from a very high 57.6 p/ha down to 39.0 p/ha, with the decline stabilizing by 1981). Again, this decline is largely due to a depopulation of the inner area and the accompanying spread in the growth of the outer areas. Only Vancouver and Toronto have experienced rising metro-wide urban densities. In Toronto, the increases in density are the result of infill development and redevelopment. In Vancouver, inner area density has similarly increased due to redevelopment and infill, as mentioned earlier. However the outer area density increases seem to be coming on the heels of restrictions of the conversion of agricultural land (due to the creation of the ALR in 1974) as well as policies aimed at concentrating significant amounts of new higher density development around 'regional town centres.' Figure 24 - Metro densities in the Canadian cities. 1981-1991 45 CO SI c/i c CD Q 40 35 30 25 20 15 41.5 39.6 33.9 28.8 20.8 21.2 18.4 20.8 27.4 33.8 31.7 22.5 21.3 31 3 ft* S6 ' •tip • 1981 Metro Density S1991 Metro Density Figure 24 above shows that although the densities of Canadian cities are no longer dropping as dramatically (at least between 1981 and 1991), they are still dropping nonetheless. This trend is more worrisome than may first appear. Firstly, many urban densities are decreasing from already low levels (e.g., Calgary and Winnipeg) and the low-density spread of outer areas is accounting for the majority of the decline. Secondly, moderately declining densities still indicate sprawl - substantially so in rapidly growing cities. For example, Figure 25 below shows that in Winnipeg, Ottawa and Calgary, the spread of the urban envelope continued to be quite dramatic despite only moderate drops in urban density. For 107 example, Calgary's density decreased by less than 2% while its urbanized area grew by more than 22% (see Table 26 below). Figure 25 - Urbanized areas in Calgary. Ottawa and Winnipeg. 1961-91 35,000 30,000 25,000 20,000 15,000 10,000 5,000 ^ -A 34,173 27,392 A 30,146 29,023 -.18:140 . -H 23,188 a-t-T,753- " >* j r ' • -9,237 1961 1971 1981 1991 •Winnipeg — -A— Calgary - - « - - Ottawa Table 26 - Effect of density changes on developed areas in three cities between 1981 and 1991 City Winnipeg Calgary Ottawa Decrease in density (p/ha) -1.2 -0.4 -0.4 % decrease in Increase in % increase in density urban area (ha) urban area -5.6%' +4,083 ~~ +15.7% -1.9% +6,181 +22 V., -1.3% +5,835 +25 2' , Cities such as Vancouver, where urban densities have actually increased, provide an even more dramatic example. Even with moderately increasing average densities, the prevalence of low-density development can still result in substantial sprawl. Vancouver's average density increased by 13% between 1981 and 1991. However, it's developed area continued along its three decade long growth path and increased by 17% (Figure 26 below). Although population growth in Vancouver finally outstripped developed land growth on average, its outer areas continue to sprawl at low densities.65 It is difficult, given the data to determine the average density of new development in Vancouver from 1981-1991. A rough calculation of the average density that this new population settled at (outer population increase divided by outer area increase) yields an average density of 31.3 p/ha. However, much of the population increase (and density increase) settled in land previously considered outer area 'developed' land, but which was subsequently redeveloped (e.g., Metrotown and New Westminster). Thus, one can safely assume that the majority of 10,532 ha of new urban land in Vancouver between 1981 and 1991 developed at densities well below 20 p/ha. 108 Figure 26 - Metro density versus developed area in the GVRD. 1961-91 30 1961 1971 1981 1991 Density • Developed Land Figure 27 - Inner and outer area employment trends in the Canadian cities. 1961-91 90% 80% 70% •] 60% 50% 40% 30% 20% 10% 0% no place of work data available for Edmonton 1961-01 120 100 "TO" 80 •§ £• W 60 S 40 | 20 ff>* OP I Proportion of jobs in inner area —•— Inner area job density —*- Outer area job density Finally, employment distribution in the sample of Canadian cities has followed a pattern similar to population in terms of declining concentration in inner cities and dispersal into lower density outer areas (Figure 27 above). Although the inner city share of employment also decreased over the 1961-91 period, most cities appear to have stabilized their inner city proportions of population since 1981, or have declined only slightly. One positive note is that inner area employment densities (including the CBD) continued to increase due to continued jobs growth, making these areas more serviceable by transit. In absolute terms, though, outer area employment growth continued to be strong in all cities and these jobs 109 were invariably accommodated at low, or extremely low, densities. With the exception of Toronto and Ottawa, all of the cities had outer area densities of less than 10 jobs/ha.66 4.2.1 Transportation Trends All of the Canadian cities in this study show increasing levels of car ownership coupled with increasing levels of car use over the 1961-91 period (Figure 28 below). These increases were strongest in Vancouver, Calgary and Montreal, particularly between 1981 and 1991.67 For example, these cities each showed per capita increases in car VKT driven of 23%, 30% and 45%, respectively. Ottawa, Winnipeg and Toronto, on the other hand, seem to have experienced relatively moderate growth in car ownership and use.68 Car ownership and use As at 1991, Vancouver, Calgary, Edmonton and Winnipeg stand out as cities that display the highest overall levels of car ownership and use. This generally corresponds with a combination of lower densities, lower transit service provision and greater infrastructure dedicated to the automobile. Meanwhile, Toronto, Montreal and Ottawa have lower levels of car ownership and use as well as correspondingly higher levels of transit provision and use. These cities also have generally higher population and employment densities and a relatively lower level of auto infrastructure provision. The per capita data presented in Figure 28 below present only a partial picture of the real growth in motorization in Canadian cities. Although it communicates the increase in driving done by the average person, it does not communicate the total (absolute) increase. Table 27 below shows the increase in per capita VKT, as well as absolute levels of population and VKT growth in the Canadian cities from 1981-91. While there were substantial per capita increases in Vancouver, Calgary and Montreal, the absolute increase in car VKT are even stronger. Each of these cities showed increases in the neighbourhood of Outer area densities do not always provide a complete picture of employment dispersal, particularly in regions where significant poly-nucleation (i.e., sub-centres) exists. For example, Toronto and Vancouver stand out as two cities that have managed to create concentrated sub-centres of activity. However, even the largest of these sub-centres pale in comparison to the CBDs. Therefore, the low average outer area densities serve as sufficient indicators of dispersal for most cities, especially those without notable poly-nucleation. 6 7 Edmonton appears to have experienced a strong increase in car orientation, however poor data availability pre-1991 precludes accurate trend analysis. 6 8 The apparent drop in car ownership levels (per 1000 population) in Toronto between 1981 and 1991 is due in large part to changes in the collection of vehicle registration data during this period. During this period, the Ontario Ministry of Transportation (MTO) went from bi-annual to annual reporting of "inactive" vehicles. This substantially increased the number of vehicles deemed "inactive". It is therefore not possible to make accurate comparisons of vehicle ownership levels between these years, though it may be possible to assume 1981 vehicle registrations are overestimated. 110 60% ~ even greater than the very strong 52% growth found in Los Angeles, one of the most auto-oriented cities in the United States. a) a. o co CL O O O CD a. co O 0 -P ft Figure 28 - Change in car ownership and use in the Canadian cities. 1961-91 700 600 500 400 300 200 100 kud_au r—i W n W >4» iSS ^vo^ «e» x «>eQ Vehicle Registrations • Car VKT per capita 10,000 9,000 8,000 7,000 CO 6,000 Q. CO 5,000 £ 4,000 H 3,000 > 2,000 1,000 0 Table 27 - Increases in VKT/capita. population and VKT between 1981 and 1991 % Increase Between 1981-1991 Vancouver Calgary Edmonton Montreal Winnipeg Ottawa Toronto car VKT/capita 23% 30% ? 45% 12% 2% 18% population 32% 20% 18% 10% 9% 23% 6% car VKT 63% 56% p!iw|li| 64% 23r c 26% 26% While all the cities have experienced car VKT growth far outstripping population growth, nowhere was this more acute than in Montreal. Their population only grew by 10%, while driving increased 64%. In fact, the increase in car VKT in Montreal of 6.2 billion car kilometres was the greatest of any city in Canada and exceeded the total 1991 kilometrage in either of Edmonton, Calgary, Winnipeg or Ottawa. Vancouver's total growth in VKT amounted to 5.0 billion VKT. This too was greater than the total VKT clocked in several of the medium sized Canadian cities, and is in some respects of greater concern than Montreal's increase given Vancouver's population is less than half of Montreal's. I l l Of course, such strong growth in motor vehicle use increases the many externalities associated with car use mentioned in chapter 2. This is particularly the case since technological improvements to mitigate these impacts can rarely keep up with such strong growth. One final interesting aspect of driving patterns in Canadian cities is the difference between VKT clocked in the outer areas, versus VKT driven region-wide. VKT driven in the outer areas is well above the annual regional per capita VKT in the GTA, Montreal and Vancouver, as indicated in Table 28 below. Likewise, the inner areas of these cities are responsible for far less than the regional average of VKT. Table 28 - Outer area VKT and other transport variables in 3 Canadian and 3 U.S. cities. 1991 Metropolitan Region Toronto (GTA) Montreal Vancouver USA three city Avg.69 Car VKT/capita Metro Inner area7" Outer area Urban density (\ Metro Inner area Outer nrea Transit VKT/capita Metro Inner area Outer area Transit boardings/capita Metro Inner area Outer area °'o of regional pass kms. on transit 5,680 5,019 6,448 25 9 41 5 18.1 65 98 49 210 350 48 *5-\-4,746 3,130 /443 33.8 43.2 28 5 60 78 37 222 351 50 1~2.8% 8,361 5.G73 9.5i>y 20.8 40^ 17.4 50 o71 117 iSiil 6.5% 11,155 jjjjjMj 13,033 19.7 60 15.9 28 IlliilB lliil§lli 63 lilllll 18111111 ~ 6%" These cities only include San Francisco, Los Angeles and New York as inner/outer area VKT splits are not available for other cities. New York is by far the most transit-oriented and dense city in the U.S., therefore, the averages tend to overstate San Francisco and Los Angeles' performance as well as that of the U.S. in general. For example, New York's percent of passenger travel on transit is 11%, while San Francisco and Los Angeles manage just 5% and 2%, respectively. 7 0 Because of limitations with transit data and of the models generating the VKT estimates, the definition of inner areas differs somewhat for this analysis than that used in the rest of the thesis. Vancouver's inner area includes a slightly larger area incorporating the Killarney neighbourhood, which is usually considered outer area in this study (in total, the area modelled is equivalent to the entire City of Vancouver proper). The impact of the inclusion of this small neighbourhood is likely very small. Its exclusion would likely drive Vancouver's inner area estimated VKT down by only a very marginal amount. The inner areas of Toronto and Montreal are simply Metro and MUC, respectively. Their outer areas are simply the difference between these and the regional totals. 7 1 Vancouver's estimated inner area boardings were provided by BC Transit (Rees 1998). 112 For example, Vancouver's inner area shows only 5,673 VKT/capita, well below the regional average of 8,361 VKT/capita. This contrasts sharply with the 9,722 VKT/capita driven in Vancouver's outer area which is well above the regional average and 80% higher than inner area VKT. This inner-outer area dichotomy is also reflected in the relative standing of each city in terms of density gradient and public transport use. The higher densities of inner areas appear to be a significant enabling factor for achieving higher levels of transit use, while the lower density, more auto-dependent nature of outer areas reflects itself in terms of lower transit ridership there. Toronto's ex-Metro suburbs (i.e., the surrounding "905" region of the GTA) has only 48 boardings per capita in contrast to Metro's 350, thus pulling down the regional average down to 210 boardings. In Montreal and Vancouver as well, the same inner/outer area patterns are evident. The Montreal-Toronto inner-outer area comparison is particularly interesting since Montreal is able to attain higher levels of ridership in both it's inner and outer areas, despite markedly lower levels of transit service. In each of these three Canadian cities, it appears as though transit service has much higher returns per transit VKT in inner areas than outer and that density plays a crucial role in this. The higher levels of car use and lower levels of boardings per VKT in inner versus outer areas have tremendous implications for transit productivity and viability. These issues will be discussed in greater detail in section below (Transit demand, service and utilization). Auto infrastructure: roads and parking Increasing use of private motor vehicles has accompanied increasing infrastructure provision for motor vehicles in most Canadian cities. All cities have experienced continued growth in their road network lengths, most notably Vancouver, Calgary and Edmonton72. However, the inability of road construction to keep pace with strong growth in motor vehicle ownership and use has led to decreasing road length per capita and increasing vehicles per kilometre of road in all cities. Congestion has also risen with VKT use. For example, in Vancouver, the number of vehicle kilometres per kilometre of road73 increased 40% between 1981 and 1991. Time series data are unavailable for Winnipeg and Montreal. 7 3 Since this study measures road network length rather than lane kilometres (which is a better gauge of capacity) vehicle kilometres per kilometre of road is only a crude measure of congestion. Vehicle kilometres remaining constant, cities with wider roads (more lanes) and higher traffic priority will have less congestion and intensity of road infrastructure use. In the case of Vancouver, where there are fewer freeways and wide roadways than Calgary, traffic intensity will be higher. 113 Figure 29 - CBD parking supply and employment in the Canadian cities. 1961-91 CL CL CO D) C L_ CO CL TJ C CO CO Q CO O 300,000 250,000 200,000 150,000 100,000 50,000 0 700 h 600 500 400 300 200 100 CO .£> O O O O CD CL JO t o D> c 1* i_ CO CL Q CO o B B Parking/1000jobs 1961-91 —•— CBD jobs. 196J -91. . A CBD Pc 3arking Supply1961-91 Most Canadian cities have also experienced net increases in CBD parking supply (as expressed in CBD parking stalls per 1000 CBD jobs) during the 1961-91 period (Figure 29 above). Out of all the Canadian cities in the study, Toronto, Ottawa and Montreal are the cities show relatively low (sub-350 stalls per 1000 jobs) and only Toronto and Ottawa have experienced net declines in parking supply. Toronto has managed to keep an exceptionally low parking supply, tightening it over the decades to its current low level of 176 stalls per 1000 jobs. Toronto's low supply has been the result of a decades long policy of accommodating 'essential' drivers and discouraging discretionary, non-essential commuting by car (Calgary 1995a). In Ottawa, conscious policies aimed at controlling the supply and price of parking have also been actively pursued. In 1975, the federal government, in cooperation with provincial and municipal governments, eliminated free parking for employees and began efforts to rein in supply explicitly as a means of increasing transit use and enhancing equity (De Leuw 1976).74 As a result parking supply decreased from 284 stalls/1000 workers in 1971 to 230 stalls in 1991 and transit use increased dramatically. Vancouver, Calgary, Edmonton and Winnipeg have the most generous CBD parking supplies in Canada. In contrast with Toronto and Ottawa, Vancouver has moved from a relatively tight supply of parking at 343 stalls per 1000 jobs to a much more generous supply of 443 stalls. Part of this is accounted for by a slight decline in CBD employment through a regional redistribution in jobs over the As the largest employer in the Ottawa-Hull region federal government employees are responsible for the majority of travel demand in the region. 114 period, however much is due to a large expansion in parking supply - from 33,617 stalls in 1981 to 41,915 stalls in 1991. The regional significance of CBD parking supply in regulating travel choices can be inferred by determining the level of regional employment concentrated in the CBD. Figure 30 below shows the proportion of regional workers that would be subjected to given level of parking supply in each city. For example, Winnipeg and Calgary have relatively large portions of their regional workforce subject to generally ample parking supply conditions. Ottawa and Montreal, on the other hand, have significant shares of their regional workforce subject to tight parking supply conditions. Vancouver and Toronto indicate relatively low shares of regional employment in their CBDs. However, this somewhat understates the importance of their CBD parking supplies as both cities have significant concentrations of employment in areas just outside their officially designated CBDs (or what each refer to as their "central area") and both have polynucleated regions with significant concentrations of employment located in regional sub-centres.75 For example, Toronto's parking supply is even tighter looking at its 'central area' rather than just the CBD alone. A paltry 120 spots are provided per 1000 workers bound for this area, yet it contains over 405,000 jobs - 32% of all the jobs in Metro. While Vancouver's central area parking supply is unknown, its proportion of jobs in its central area is lower than Toronto's at 15%. Figure 30 - Proportion of regional employees working in CBDs in Canadian cities. 1991 30% IMMW^^ 25% Q m (J I 15% i 13.4% o % 10% 5% 0% I M L 21.0% 19.4% 2 0 - 0 % 25.9% 24.2% 23.5% 13.7% 12.1% oOS6* \c#S rttv^ rt«6^ X\Oe^ sMS*** <<f& O* 7 5 Parking supply in these sub-centres was not counted in this study. The supply is likely tight relative to other cities' outer areas in general since land is more scarce and stricter parking policies cap their supplies. 115 Parking and transit There appears to be a strong link between parking supply and the degree to which automobiles are used for commuting purposes. For example, Canadian regions with parking supplies in the 400-600 stalls per 1000 workers range are able to achieve journey to work modal splits in the range of 11-20%. Meanwhile, the European cities, with an average of 230 stalls/1000 workers obtain an average transit share of 39%. Some of the most car-oriented cities in the U.S., such as Phoenix and Detroit have very generous parking supplies (906 and 706 stalls per 1000 workers, respectively) and virtually negligible transit shares (2.1% and 2.6%, respectively). Calgary's decline in parking supply per employee in 1981 provides an excellent case study explaining the strong possible link between parking supply and transit use. Figure 31 below shows the available supply of parking per 1000 jobs in Calgary in 1971, 1981 and 1991, as well as the corresponding share of commuters driving or taking transit/nonmotorized modes to work in the CBD76. The sharp decline in parking supply between 1971 and 1981 (from 565 to 425 stalls per 1000 jobs) came on the heels of a booming economy and sharp growth employment. While the parking supply did grow during this period in absolute terms, it did so at a rate far below jobs growth. This tightening of parking supply in 1981 corresponds with a sharp decline in the proportion of people driving to the CBD (from 62.5% to 46.4%) and a sharp increase in people taking transit and nonmotorized modes (from 37.5% to 53.6%). Figure 31 - Parking supply and modal split in Calgary. 1971-1991 700 CO _Q O O O o J2 "co to O CO O CO co O 600 % 500 400 300 -t 1971 1981 1991 70% CBD stalls/1000jobs' •Transit and NMT % •Auto % Also see Figure 29 for additional CBD data for Calgary. 116 This positive trend reversed itself in the 1981 to 1991 period as stall growth outstripped employment growth by over 2.3 times,77 bringing the parking supply back up to 522 spots per 1000 jobs from 425 in 1981.78 In other words, for every 1000 new jobs in the CBD, 2341 additional parking stalls were added. Transit use declined amid this increase in parking supply, despite the expansion of L R T and higher levels of transit service (in absolute and per capita terms) available by 1991.79 During this period, Calgary also had one of the strongest region-wide VKT growth figures in Canada for this same period. Furthermore, during the 1981-1991 period, the decentralization of employment continued in Calgary (as it did elsewhere), subjecting those periphery-bound commuters to an even less stringent parking supply than that found in the CBD. Energy use Not surprisingly, those regions with the most regional travel by car, the greatest infrastructure provision afforded to the automobile and the lowest public transit patronage, have the highest levels of gasoline use in the country (Figure 32 below). With nearly half the driving and three times the transit use the average U.S. city, Canadian cities consume much less gasoline. Figure 32 - Gasoline use in seven Canadian cities and the U.S.. 1991 60,000 55,807 CL ra o "5 o 3 o O) a> E. <D in 3 0) c o in a CD 32.018 27,706 30,746 26.705 The Alberta "oil boom" slowed in the early 1980's and as a result many developers were unable to develop downtown land profitably. Some of the initial growth in parking supply came in the form of ground level lots that were converted to parking in order for land owners to cover holding costs on their land. However, a substantial number of new stalls remained by 1991, despite a rebound in the economy. 7 8 To maintain a parking ratio of 425 spots per 1000 jobs, the parking supply should have increased by only 1864 stalls. Instead it increased by 10,262 stalls, 5.5 times what it should have. 7 9 Additionally, much of the service increase was heavily CBD focussed. 117 Transit service, demand and utilization Most Canadian cities show a pattern of transit decline between 1961 and 1971, revival between 1971 and 1981 and decline once again between 1981 and 1991. This is reflected in terms of boardings80 per capita (demand) and transit VKT per capita (service) in Figure 33 below. In most cities throughout this period, transit demand and service move in the same direction. (Tables 29-30 below highlight some of the changes in the transit service, demand and utilization discussion that follows.) Only Vancouver and Toronto actually showed increases in transit travel demand between 1981 and 1991, while all other cities faced declining demand. Vancouver's transit demand increased nominally from 114 boardings per capita to 117 boardings (or 3%) between 1981 and 1991, while its transit service increased 10%, mostly because of the introduction of SkyTrain and a realignment of bus routes. Toronto's increase was more substantial. Between 1981 and 1991 transit demand in Metro Toronto increased by 20%, from 292 to 350 boardings per capita. This demand was accompanied by a 22% increase in transit VKT per capita,81 while population increased only nominally. Figure 33 - Transit demand and supply in 7 Canadian cities. 1961-91 400 350 Transit boardings —•— Transit service, VKT/capita Overall, Toronto and Montreal have the highest per capita transit ridership in North America.82 Both cities have relatively high urban densities and transit friendly urban forms, particularly in their inner The terms "boardings" and "trips" will be used interchangeably in reference to transit travel demand. 8 1 Transit demand GTA-wide is obviously lower, but still very high by Canadian, and North American, standards. Available data indicated GTA-wide transit demand also increased during the 1981 to 1991 period from 198 boardings to 210 boardings per capita. Although Metro's surrounding suburbs (GTA less Metro) also increased their transit usage, they did so from relatively low levels 41 boardings per capita in 1981 to 49 boardings in 1991. 8 2 Montreal's transit use is actually higher than Toronto's if one considers the entire GTA as the region. The Montreal Region managed 222 boardings per capita region-wide in 1991 while the GTA garnered 210 boardings. 118 cities. Both cities also invested heavily in subway systems in the 1950's and 60's to serve as "spines" for their transit systems. In addition, Toronto has maintained an extensive network of streetcars (trams) that have served the dual function of maintaining high levels of transit service and visibility as well as limiting automobile capacity on arterial roads. Ottawa has also been a transit leader in Canada. As at 1991, it had the fourth highest ridership levels per capita in North America, after Toronto, Montreal and New York. Between 1971 and 1981 Ottawa managed to double its transit boardings (from 75 to 155 annual boardings per capita). This increase coincided with a 116% in transit service VKT. Ottawa managed this impressive rebound in transit ridership using an entirely entirely bus-based system. In fact, Ottawa's high of 155 boardings per capita in 1981 came even before the completion of the Ottawa-Carleton Transitway.83 Ottawa's transit ridership actually declined between 1981 and 1991 (by 13%), well after major portions of the busway were complete. Whether these declines would have been even greater without the busway, or whether they were aggravated by a 14% decline in system-wide service VKT is not easily discemable from the data at hand. Overall, transit service kilometers have increased in both absolute and per capita terms in most Canadian cities between 1961 and 1991.84 Table 29 below indicates that the biggest increases in service kilometres over this period were observed in Toronto, Ottawa and Calgary,85 with most of the growth coming between 1971-81. Between 1981 and 1991, service increased substantially yet again in Toronto, increased marginally in Vancouver and Calgary and dropped in Montreal, Winnipeg, Ottawa and Edmonton. However, few cities were able to parlay these increases in service kilometres into lasting gains in transit demand (Table 30 below). Despite substantial expansion in service since 1961, demand has actually decreased, or increased only nominally, in most cities. Only Toronto experienced a strong growth in transit demand since 1961, with an increase of 127%.86 A look at the more recent changes in boardings per capita between 1981-91 reveals a drop in demand in all cities except Vancouver and Toronto. The declines in Montreal, Winnipeg, Ottawa and Edmonton all correspond to declines in transit service VKT per capita during that period. Meanwhile, the sizeable drop in ridership in Calgary (-22%) In the early-1980's, Ottawa began construction of the Transitway, a busway network which uses mostly articulated buses in exclusive rights of way, grade separated from general traffic as most rail rapid transit systems operate. By 1991, the busway carried 210,000 passengers per day, more than almost any light rail system in North America (McCallum and Beere 1997). However, Ottawa's busway has not been able to attract higher density development similar to that found in along Vancouver and Toronto's rail lines, despite effort to encourage it. 8 4 Only Winnipeg had a net decline in transit service VKT over this period. 8 5 Although figures are not available for the Montreal region back to 1961, transit service VKT in the MUC increased 79%. Meanwhile, boardings per capita increased by 42% while utilization declined by 21%. 8 6 Again, while metro-wide figures are not available for Montreal, the MUC showed a gain in transit patronage of 42%. 119 came despite and a notable increase in transit service (8%) and the expansion of LRT, indicating that other forces were strongly influencing transit demand there. Table 29 - Change in transit service levels (VKT/capita) in seven Canadian cities. 1961-91 Transit VKT per capita % change 1961-71 1961 1971 1981 1991 1961-71 1971-81 1981-91 1961-91 Toronto 48 58 81 98 20% 39% 22% 104% Montreal 63 60 -5% Winnipeg 43 43 44 40 - 1 % 3% -9% -7% Vancouver 37 25 46 50 -31% 82% 10% 37% Ottawa 32 24 65 56 -25% 174% -14% 76% Edmonton 33 31 63 51 -6% 101% -19% 53% Calgary 30 31 46 50 3% 49% 8% 66% Table 30 - Change in transit demand (boardings/capita) in seven Canadian cities. 1961-91 Transit boardings per capita % change 1961 1971 1981 1991 1961-71 1971-81 1981-91 1961-91 Toronto 154 132 292 350 -14% 120% 20% 127% Montreal 228 221 -3% Winnipeg 148 136 134 98 -8% - 1 % -27% -34% Vancouver 138 88 114 117 -36% 29% 3% -15% Ottawa 115 72 155 135 -38% 116% -13% 17% Edmonton 98 108 140 109 9% 30% -22% 10% Calgary 94 73 120 94 -23% 65% -22% 0% Table 31 - Chanae in transit utilization (boardinqs/VKT) in seven Canadian cities. 1961-91 Transit boardings per VKT % change 1961 1971 1981 1991 1961-71 1971-81 1981-91 1961-91 Toronto 3.3 2.3 3.6 3.5 -28% 56% -5% 6% Montreal87 3.6 3.7 2% Winnipeg 3.4 3.2 3.0 2.4 -7% -4% -20% -29% Vancouver 3.8 3.5 2.5 2.3 -7% -29% -6% -38% Ottawa 3.6 3.0 2.4 2.4 -17% -21% 1% -34% Edmonton 2.9 3.4 2.2 2.1 16% -35% -4% -28% . Calgary 3.1 2.4 2.6 1.9 -25% 11% -27% -39% Not surprisingly, the cities that lost transit patronage in this period also experienced a declining importance of transit relative to the car (Figure 34 below). One development that is of particular interest is that even where transit demand has risen (i.e., Toronto and Vancouver), the transit share of total 120 motorized transportation has actually declined. In other words, in real 'inflation adjusted' terms, transit use is declining. The increase in transit demand in these cities was more a function of an increased demand for travel of which transit took a smaller share relative to the automobile.88 Figure 34 - Transit share of motorized travel 30% - r @ 1981 a 1991 Another pattern that is evident in examining these changes in transit supply and demand is that where transit supply has dropped, the percentage decline in transit demand dropped at a rate greater than supply. Furthermore, where transit supply increased, the percentage increase in transit demand grew at a rate less than supply. For example, Winnipeg's service VKT dropped by 9% between 1981-91, while its ridership dropped by 27%. Meanwhile, Vancouver's expansion in VKT per capita of 10% only yielded an increase of 3% in boardings per capita. In other words, transit is getting 'less bang for the buck.' There have been, for the most part, declining marginal gains in ridership for service increases. This contrast in increasing service provision (Table 29) with demand failing to keep pace (Table 30) is manifest itself in poor or declining "utilization" (boardings per VKT) in most Canadian transit systems (Table 31). Although Table 29 indicates expanding service per capita between 1961 and 1991, Table 31 indicates this increase in service was poorly used. The reason for the declining utilization in transit appears to be that most of the increase in service VKT was provided to lower density suburban areas. This increase in service VKT often served a more sparsely populated catchment, therefore Trend data are not available for the entire Montreal area for 1961 and 1971. However, the MUC's change in boardings between 1961 and 1991 was 42%. 8 8 Total "travel" is measured measured by passenger kilometres (i.e., number of passengers in a vehicle x kilometres travelled). The strong increase in car passenger kilometres is likely related to an increase in sprawl-induced long haul urban travel, the preponderance of which is done by automobile. 121 ridership potential is much lower. This supports the general axiom of transit planning that relatively high densities are required to deliver accessibility to transit services, which can then be translated into ridership. Table 32 - Land use and transit service in 7 Canadian cities. 1991 Density Service Demand Utilization | intensity p/ha VKT/ha a. Boardings/cap Board ings/VKTI Montreal (MUC) 43.2 3361 351 4.5 Montreal (Region) 33.8 2034 222 3.7 Toronto (Metro) 41.5 4378 350 3.5 Toronto (GTA) 25.9 1685 210 3.4 Ottawa 31.3 1750 135 2.4 Vancouver 20.8 1046 117 2.3 Edmonton 27.4 1409 109 2.1 Winnipeg 21.3 862 98 2.4 Calgary 20.8 1033 94 1.9 The strong role density plays in determining ridership levels and service viability was demonstrated in Table 28 above. The higher density areas of both MUC (Montreal) and Metro (Toronto) have substantially higher levels of service and ridership than their outlying areas. Table 32 above provides a comparative overview of land use and transit service and performance in various Canadian regions. The more dense cities are able to offer higher levels of service intensity (VKT per hectare of developed land) and in so doing can provide more frequent service to a greater number of people.89 In general, the more compact cities also have higher transit utilization: for every kilometer of revenue service provided, more people get on the bus. Declining transit utilization has tremendous implications in terms of productivity, subsidies and overall system sustainability. The servicing of vast tracts of suburban land has meant declining service intensity and a diminished ability to attract patronage. In the short run, having to travel more kilometres to attract fewer passengers (low system utilization) means ever increasing operating subsidies are required (for labour, fuel and maintenance) in order to keep the system running. In the long run, these services The use of transit VKT per capita as a measure of service only paints a partial picture of true service levels. In a comparative sense, it tends to overstate the level service in low density cities, since transit services travel greater distances to service the existing population base. Furthermore, higher density cities are more like to have rail services. Since each rail car (which is the basis for rail VKT figures) carries more passengers, and since rail services are typically operated in dense corridors, cities with rail VKT actually deliver higher levels of service, and are more accessible, per VKT, than buses. Therefore, given two cities with equal transit service VKT, the higher density city likely offers more frequent and accessible service per VKT. Using VKT/ha normalizes for density and provides a crude estimation of how 'intense' service is. 122 prove unsustainable and become vulnerable to cuts when purse strings tighten. In many cities, service expansion into sprawling areas has heavily taxed the resources of transit systems (Perl and Pucher 1995). BC Transit's expansion in the 1980's, for example, focussed heavily on extending new and more frequent services low-density areas. The resulting decline in system utilization has drained resources away from improvements to already viable transit services within higher density areas. Toronto's gains in ridership and utilization since 1971 were due in large part to the fact that service increases were fortified within a fully developed boundary (i.e., Metro). Since the TTC's service area was constrained by Metro's largely built out urban boundary, service expansion was not diluted by the low density spread of serviceable area that has compromised transit performance in Metro itself pre-1971, and in the GTA and other Canadian cities today. The intensification of land uses within Metro has helped Toronto to reap dividends from service expansion.90 Although the level of transit service is an important, even critical, enabling factor in increasing transit patronage, the data presented above indicate increases to transit service alone will not guarantee ridership. Decreasing transit provision is just one of the factors influencing declining transit patronage and utilization. While transit decline has surely been influenced by other factors such as the declining operating and capital costs of the car (Perl and Pucher 1995; RMOC 1995), sharp increases in transit fares (Pucher 1998) and demographic change (CUTA 1991),91 the data presented here indicate that auto availability, auto infrastructure provision, land use and transit supply also play important roles. In the case of Ottawa, it seems cuts in service since 1981 and lower-density urban growth may be key contributors to lower demand. In the case of Calgary where transit supply increased dramatically, it seems road and parking infrastructure supply, in addition to outward urban growth has contributed to the loss of transit share. In Metro Toronto, land use intensification and controlled parking supply may have been the key catalysts that allowed the expansion of transit service to truly reap rewards by building a strong transit catchment base and by controlling ease of access by car. The cities summarized The transport and land use characteristics of the cities discussed thus far reveal a web of factors that influence levels of auto dependence. The factors appear to act as levers, serving to increase or decrease auto use, and likewise for transit. Some of the key characteristics defining the transport and land Of course, the trends outside of Metro are still of concern, however, the gains in Metro provide an interesting cas study of the potential urban containment policies may have for encouraging transit use. 9 1 Certain 'demographic' changes and sprawl are in many ways proxies for one another. For example, the increasing affluence and aging of baby-boomers leads to higher car ownership and less transit use. However, high car ownership and use may simply reflect preferences for single family suburban dwellings, which in many cases are automobile dependent. 123 use patterns of Canadian cities are placed in the context of some of the global cities and are summarized in Table 33 below. Table 33 - Land use and transportation characteristics of selected World cities. 1990/91 City Land Use Cars Car Parking Transit Transit Intensity Owned Use Supply Use Service pop+jobs/ cars/ VKT/ stalls/ Boardings/ VKT/ ha 1000 pop. capita 1000 jobs Capita Capita Phoenix 16 644 11,608 906 15 10 Detroit 19 693 11,239 706 24 14 Portland 20 764 10,114 403 46 27 San Francisco 25 604 11,933 137 112 49 Perth 15 522 7,203 631 54 47 Vancouver • • • 1 564 8.361 443 117 50 Calgary 630 7,913 522 Edmonton 527 7,062 594 109 Winnipeg 412 6,871 546 98 Ottawa 510 5,883 230 135 Toronto (GTA) 515a 5,680 176 210 65 Toronto (Metro) 65 431 5,019 176 350 98 Montreal 49 420 4,746 347 222 Copenhagen 45 283 4,558 223 164 121 Amsterdam 71 319 3,977 354 325 60 Stockholm 92 409 4,638 193 348 133 Munich 91 468 4,202 266 404 91 Vienna 106 363 3,964 187 422 73 Tokyo 178 225 2,103 43 461 89 Hong Kong 440 43 493 33 570 140 Note: a. Estimated from "Transportation Today and Tomorrow Survey." Joint Programme in Transportation, University of Toronto, 1991. Various degrees of car orientation are evident upon examining the data and clear patterns emerge. The cities with the highest car use also have relatively low land use intensity, high car availability, abundant CBD parking, low transit use and sparse transit service. Since each of these factors represent 'policy intervention points' and may influence auto dependence to various degrees, one of the key questions for transportation planning is: what is the relative importance of these each of these factors in determining auto dependence? Knowing which factors most strongly influence, and are influenced by, a city's car orientation can help to steer planning towards policies that offer the greatest marginal benefits in reducing car dependence and improving access. 124 4.3 DATA CORRELATION ANALYSIS The analysis above reveals certain factors associated with higher and lower levels of automobile dependence in Canadian cities and this section follows with a basic correlation analysis of some of the key variables examined in this thesis. It is particularly focussed on identifying the factors most strongly correlated with car ownership (cars/1000 people), car use (car VKT/capita) and transit use (boardings/capita).92 I will refer to these three indicators as the "auto dependence indicators." The basic correlation analysis presented in this section takes car ownership, car use and transit demand and assesses the degree to which they are positively or negatively associated with other variables. For example, I will determine how strongly transit use is associated with such factors as density, transit supply and parking supply. The strength of correlation for the data items presented are given a qualitative designation and are then discussed. These correlations are also used to reconstruct the auto dependence feedback diagram (presented earlier in Chapter 2), showing the strength of correlation between various factors identified as being involved in auto dependence feedback. The correlation analysis, as with the previous discussion in this chapter, focusses on Canada, but uses the world cities for context.93 4.3.0 Correlations of Key Factors Table 34 and Table 35 below provide a summary of the correlations between some of the key factors studied in this thesis.94 The full correlation tables, including levels of significance, the number of cases for each factor and a key defining variable short forms are found in Appendix 4. Using the correlation charts Table 34 presents the correlations for the 7 Canadian cities in the study and Table 35 presents the correlations for 41 world cities.95 Thirty-two variables in the tables are compared to car ownership, car Since boardings per capita focussed only on region-wide transit, and some explanatory factors have a more narrow geographical scope (e.g., CBD, inner area and parking variables), journey to work modal split on transit is also used as a supplementary transit indicator. Since much of the transit demand in these areas is journey to work oriented, correlations between parking supply and employment, for example, can help to explain why transit mode share may be higher or lower. It must be cautioned, however, that since the transit mode split data is only focussed on the journey to work, it is not useful in its explanatory function for any of the region-wide variables. Therefore, low levels of correlation with region-wide variables, particularly for cities where transit is heavily work oriented (e.g., North America and Australia), are not meaningful. 9 3 The sample of World cities used in the correlation analysis excludes those in developing Asia. These cities have a dramatically different economic environment (for example, infrastructure decisions are much more constrained by lack of resources), so removing them and comparing only developed-nations cities will provide more meaningful correlation data. 9 4 The large volume of data collected precludes the detailed analysis of each factor. Furthermore, many of the factors are co-related or are proxies for one another. Therefore, the most relevant variables have been chosen for analysis. 125 use, transit use and transit mode split to work. The resultant correlations have been rounded to the nearest hundredth and are expressed as positive or negative values.96 The tables, and their more detailed counterparts in Appendix 4, should be used carefully. Many of the variables compared in the table co-vary, are surrogates for one another or result in correlations that are meaningless. For example, metropolitan areas with high population densities are also likely to have high employment densities. Likewise, activity density, which describes the total employment and population density, will also be high. Therefore, each of these three variables will likely have similar correlations to car use. On the other hand, certain related variables can provide different dimensions on factor relationships. Inner area employment densities do not seem to bear as strong a relationship to car ownership, car use, transit use as for inner area population densities. However, the relationship strengthens for the journey to work transit modal split. Various levels of statistical significance have been attached to the correlation data and these are indicated in Appendix 4. While many of the strong correlations that will be discussed are also "statistically significant" (particularly in the case of the larger 41 cities sample), I have not screened for statistical significance for two reasons. Firstly, because the Canadian cities are fewer (and therefore have fewer data "cases"), the data are less likely to establish correlations that are "significant" in the statistical sense, unless the correlation is extremely strong. Many of these relationships are just shy of being "statistically significant" by the slightest margin by virtue of the small sample size. The larger sample of 41 cities (or cases) show most of these same relationships to be similarly strong and statistically significant to the 0.01 level (meaning there is a 99% chance the relationship is not a random occurrence). Secondly, while the 41 world cities are a 'sample' of the world's many large cities, the 7 Canadian cities sampled, for the most part, comprise the large Canadian cities (population over 600,000). Therefore, establishing statistical significance is irrelevant since the sample is the entire 'population' of large Canadian cities. Any correlations that result are therefore "significant" by definition since they include all possible large Canadian cities. While sampling further World cities may result in changing correlations, sampling other large Canadian cities is not possible. The World cities provide a large enough sample size to make inferences that bear greater statistical significance. This includes all the Canadian, Australian, U.S., European and Wealthy Asian cities. It excludes all the Developing Asia cities (see note 93). 9 6 A positive correlation means the two variables move in the same direction - an increase in one variable leads to an increase in the other, and vice versa. A negative correlation means the two variables move in opposite directions -an increase in one variable leads to a decrease in the other, and vice versa. 126 Giving meaning to the correlations Managing and interpreting the large volume of correlation data generated can be unwieldy and confusing. Furthermore, given that there is a range of error with the various data items (see Data Quality and Reliability, Chapter 3), discussing minor differences in correlation does not contribute to a greater understanding of the relative of importance of factors influencing auto dependence. In order to make the large volume of correlation data useful for guiding discussion and policy analysis, qualitative descriptions will be given to correlations that fall within designated ranges. By assembling and sorting all the correlation data, it is possible to identify clear groupings of data, making the assignment of qualitative designations to them a relatively straightforward task.97 The correlations grouped and described as follows: Correlation range Descriptor Symbol 0.83- 1.00 very high correlation ^ ^ • • • M M 0.74-0.82 high correlation — — 0.67-0.73 significant correlation — — — — • 0.60 - 0.66 moderate correlation • • < 0.60 weak correlation • These descriptors, and the accompanying symbols, will be often used in lieu of the actual correlation numbers in discussion and analysis that follows. For simplicity, this classification will be used for both the Canadian and World cities. However, it is recognized that since the taxonomy is based on the correlation results from the smaller Canadian cities sample, a lower threshold for each data range would be more appropriate for the World cities. For example, data deemed to have a "very high correlation" may more appropriately fall between 0.75 and 1.00 correlation coefficients, rather than the existing 0.83 - 1.00, and the threshold for the correlations deemed to be "weak" may fall below 0.60. Using one conservative classification system for both the Canadian and World cities allows the analysis to be both statistically sound, yet straightforward for the purpose at hand. While the correlation ranges identified for designation are arbitrary to some extent, they do reflect apparent "clusters" of correlation data. The use of all correlation coefficients greater than 0.65 for the Canadian cities sample is supported by statistical tables which hold that values over 0.65. 127 oo N3ai/\iaino iSN3aino N 3 a o v i n o N3ai/\I3±3I/\I ISN3ai3l/\l V3<dV±3l/\l N3aOV131Al N30IAI3NNI ISN30NNI N300VNNI d " S V 9 isN3aaao N 3 a o v a a o N3ai / \ i3aao d l w y v o o o o T a v o CO CM O co oo co CD CM co CN CO LO LO co CM CO r--o o CM CO co cn LO 0 0 r- co TJ- oo b d co «-o d d CM co r- oo d d co o CM d d o o o i -co d CM CD TT co °° co P P d CD d LO d cS o d 5 °-i * Qi a: < < o o co 9 d d CO co ° oo oo ^ CO CD 9 d d LO i : LO r-n . •<* o 9 d d r - CO CD r-d d CM co if CO CD ^ IS: » N s O O o o co 1 -co LO r- oo d d h- o 0 0 N -d d LO oo co co d d CO h-LO LO LO d d • i a. z H O O a H 0. NNIdOdUd aaododyd N v y i d o y d l lAINdOdd syvodoad NNiaoryd a a o a o r y d OOOL-O>W co CO CD d d o LO LO CO LO LO LO 0 0 LO oo o d I LO oo d LO d 9 o LO LO o co co d d co o o oo co co CM d d CM o CD o d CD CO d d CD CO LO h-oo o LO o d CD O co r-d d S °-o I ° 5 I * Qi 01 < < O O co CD 0 0 CO CD CD LO O O CD CO O CO CO CD CO CD CO d d CO co d co LO r--0. 0. Q Qi m H o I-oo CO r--oo d co CM co CD d d CM CD CD LO CM d o o o CO CD CD O oo o r--oo H H 0. O Qi 0. U NaaiABino iSNaaino Naaovino N301AI3131AI isN3ai3n V3cdV13IAI N300V13IAI N30IAI3NNI ISN30NNI N3aOVNNI d~SV9 iSN3aaao N 3 a o v a a o N3ai^i3aao d ~ i / w y v o o o o T y v o CD CM r- co LO co oo co oo oo r--CM CO co co oo CO LO CM d oo o o CO d CM co o co co co co co CO CM LO CO LO CO LO CO CM o o co d CD CO CO CM d d o o •«- o o Q., 1 < < o o LO CO d d CO LO co d d CD d d o LO co oo d d CM r--d d CO CM CM d • d • co oo d d CM o CD d d CD CO CO d o co o CD d d CO CO oo oo d I d • CM CO LO LO o d IllBI LO CM LO d d o co d d co oo oo d ||:;:|J!||: d I co oo d d a: H o 2 H a. O a. ddJJHAyil d N3NVcJl adsyiioi dd>ld±0± d~N3101 d d a a a i o i i / \ i»daaio i d~avoy NNIdOd^d aaododdd N V y i d O d d i^ Ndoyd syvodoyd NNiaoryd aaoaoryd 0001. 0>ld CO ->- ,j s to oo o d d CM co oo oo d oo LO co co oo o co o o o < CO co CO o o d •«- o co CM CO co d d d CO CO d d oo CM co co CO co LO at < o CO oo o o co CO 0 0 d d co j^-. <N co p o d oo co t co co oo oo oj o d d CD co o d d oo oo d d 9 co LO d co oo o d CO co oo oo co co CO co o o CO o co CM d oo CO o o co d co o LO 0. 0. Q ca O O H Z I— 0. o DC a. o V. ft © o o o x> a .-s o « W s h «* • & * ft 3 H O u H II W pa d & !^ o..a " -a c « S S3 * a o a § ° 3 <L> ii .3 ^ CO 2 « 4.3.1 Relative Strength of Factors Influencing Auto Dependence Analysis of the correlation data in Table 34 and Table 35 reveals several key factors associated with higher and lower levels of automobile dependence. By excluding various data items because they covary with others, are surrogate measures of one another, or are not relevant to the analysis, a core set of influencing factors are identified. These factors, their relative strength of correlation and their relationships between one another are mapped out for the Canadian and World cities in Figure 35 and Figure 36 below.98 Figure 35 - Correlations of factors involved in auto dependence feedback in the Canadian cities -0.63 ^0.77 Transit Use "—^ ^ ^ 0 . 8 7 « (boardings/capita) ^^^^^ y ^ ^ ^ ^ . ' Transit Share (% of trips on JTW) Car Use ^ \ (car VKT/person) 0.55; -0.77. 0.70 c Gasoline Use 9 8 Some additional factors not summarized in Table 34or Table 35, but included in Appendix 4, will be drawn in the analysis on occasion. 130 Figure 36 - Correlations of factors invo Ived in auto dependence feedback in the World cities fMetrol < ' , / / All metro area size / ^ correlations very 0.78 * weak (generally, <±0.3) Transit Share (% of trips on JTW) Car Ownership (cars/1000 people) > > 0.84 f Car Use (car VKT/person) >r > 86 < 0.97 0.83 Gasoline Use Of the factors studied, urban density, transit supply and CBD parking supply are most strongly correlated with the variables of interest: transit use, car use and car ownership. Urban density Urban density appears to exert the strongest and broadest influence on the three auto dependence indicators. Much of the correlation data for the density variables" in the Canadian cities fall in the -0.8 to -0.9 range for car use and the 0.8 to 0.9 correlation coefficient range for transit use. Figure 37 below shows the relationship between density and car use in the Canadian and U.S. cities for 1991 (-0.86 correlation coefficient; -0.89 for Canadian cities only). Montreal, Ottawa and Toronto lead the North American cities in terms of relatively low levels of car use per capita. Not coincidentally, these cities also have the highest level of transit use. The "outlying" cities on this graph (sitting further from the curve) are of particular interest and some will be discussed in greater detail below. These include all the employment, population and activity density variables for the CBD as well as the inner, outer and metro areas (see Appendix 4). 131 Figure 37 - Density and car use in the Canadian and U.S. cities. 1990/91 2,500-| , , , , 1 1 5 10 15 20 25 30 35 Urban density (persons/ha) The axiom that higher urban densities facilitate higher transit ridership appears to hold true for Canadian cities. The Canadian cities demonstrate both high (0.76) and very high (0.91) correlations between inner area and metro-wide population densities, respectively. This confirms the observations presented earlier in this chapter, and noted by other authors (Newman and Kenworthy 1989a; Patterson 1993; Perl and Pucher 1995; Pushkarev and Zupan 1977), that higher urban densities are strongly associated with higher transit use. Indeed, density appears to be a key enabling factor for higher transit use. 0 below plots the relationship between density and transit use for the Canadian and U.S. cities. Montreal and Toronto far outperform all other North American cities in transit use. Transit use in these two cities appears to "take o f f after the next closest city, New York, managing transit ridership levels approximately 50% and 230% greater, respectively. Much of this higher transit demand can be attributed to the relatively higher urban densities in those two cities. The much higher transit ridership levels in these two cities seems to support the assertion by other authors (Newman and Kenworthy 1989a, for example; Pushkarev and Zupan 1977) that sharp increases in transit ridership are observed in cities where densities are over 30 p/ha. This theory would suggest that several Canadian cities, particularly Ottawa, Edmonton and Vancouver, are under-performing in transit use given their metro and inner area densities. Toronto's density and car use are for the GTA. 132 Figure 38 - Density and transit use in the Canadian and U.S. cities. 1990/91 60 0 -I , , , , ? , , 1 0 50 100 150 200 250 300 350 400 Transit Use (boardings/capita) At the lower end of the urban density gradient, the seven U.S. cities hovering in the 8-13 p/ha density range have transit ridership levels of between 15 and 30 boardings/capita annually. Chicago, Washington, San Francisco and Boston all manage to achieve high transit ridership relative to the other U.S. cities, and even relative to some of the Canadian cities. Several factors appear to account for this difference. Unlike the U.S. cities with low transit use, Chicago, Washington, San Francisco and Boston all have extensive rail networks upon which their systems are based (very few of the other cities have any rail whatsoever). Public transport is therefore faster and more accessible. These cities also have somewhat higher metropolitan population densities compared to the other seven U.S. cities as well as very dense inner areas which account for a large proportion of the transit ridership. Finally, and perhaps most importantly, these cities all have extremely tight CBD parking supplies, even by Canadian standards. Since significant regional employment in these cities is still located in the CBDs, many commuters have relatively constrained modal choice options (parking will be discussed further below). Outer area densities also demonstrate a high degree of correlation with car use and transit use. Outer area density is highly negatively correlated with car use (-0.82). Conversely, density has a high positive correlation with transit use. Outer area employment density also exhibits a very high positive correlation (0.92) with transit use. This does not mean that people in outer areas use transit often for work. Rather, it indicates that where employment densities in outer areas are high enough, transit can provide viable service such that it 133 can attract ridership. Areas with higher density outer areas, such as Toronto, Montreal and Ottawa, are able to attract notable ridership, thereby boosting region-wide ridership relative to other cities. Higher urban densities not only make car use less necessary and transit more accessible, they also make transit more viable, as discussed in section above. There is a high positive correlation between transit utilization (boardings/km) and metro area population density (0.81 correlation coefficient). This relationship is even stronger between transit utilization and inner area density (0.93 correlation coefficient). Transit systems operating within cities of higher densities will likely have lower servicing requirements, or will be able to deliver higher levels of service to a greater number of people. Density therefore appears to have important implications for transit cost-effectiveness. One area where correlations with density are not particularly strong is with car ownership. Although there is a high negative correlation with car ownership and inner area density (-0.76), there is a weak negative correlation with metro density (-0.56). This implies that, in the Canadian context, metro-wide changes in density do not affect car ownership to a great extent. One possible explanation is that metro-wide densities are fall with in a comparatively narrow (and low) range. Therefore, urban densities in Canada may not have passed a sufficiently high threshold such that car ownership is not as necessary and accessibility needs can be satisfied to a significant extent by other modes. This is particularly the case for lower density cities like Vancouver and Calgary, where car ownership approaches U.S. levels. A comparison with the World cities sample, where there is greater upward density variation, supports this observation. These cities (which include the Canadian cities) show a high negative correlation (-0.78) between metro activity density and car ownership. At the higher density end of the spectrum, relatively high degrees of accessibility means that car ownership, or multiple car ownership, is not a necessity. Since most Canadian cities seem to fall short of the "threshold" density for realizing low levels of car ownership like those found in Europe and Asia, owning a car is still necessary to meet some access needs. This is particularly the case at off-peak times where transit service in many Canadian cities can be poor to non-existent. The fact that there is a high correlation between inner area density, where densities reach or surpass European levels, and region wide car ownership does give some credence to the assertion that increases in density in the Canadian context could result in a declining need for car ownership. There are likely many other qualitative aspects of urban form not studied here that greatly affect the levels of car and transit use in a city. These may include land use mix, the quality of the urban fabric and the basic walkability and cyclability of the streets. However, at first glance, the net density of a city does appear to act as a suitable barometer in the Canadian context to measure the degree to which a city is auto-oriented. 134 Transit service While density is a critical enabling factor for cities to achieve lower levels of auto use and higher levels of transit use, it alone is not sufficient. Transit service levels are also critically important (see Figure 33 above). After all, i f transit service is widely available, transit use is unlikely to be high, regardless of the density. Amongst the Canadian and World cities, transit supply is very highly, positively correlated with transit use at 0.96 and 0.84, respectively. Figure 39 below demonstrates that there is a wide range of transit potential ridership levels at various levels of density. It is clear that as density increases, higher levels of transit ridership become possible. Figure 39 - Density and transit use in Europe. Australia. Canada and the U.S.. 1990/91 80 re .C c 0 Q c o re 3 Q. O CL 70 60 50 40 -| 30 I 20 10 0 / * Brussels • Vienna Munich Frankfurt Pans ^urich • % Toronto Montreal Lqs Angeles * / * Vancouver / Winnipeg; * • New Yock-/ V ^Sydney 100 200 300 400 Transit Use (boardings/capita) 500 600 It is instructive to compare the transit use outcomes of various cities at similar levels of service or similar levels of density. Table 36 below shows the influences of both density and transit service on transit ridership levels in fifteen cities. The shaded areas provide reference points for groupings of cities. In examining the table, it becomes clear that both density and transit use exert strong influences on ridership. In the first grouping of cities, Los Angeles has the highest urban density of the four cities, but less than 50% of the transit boardings of Vancouver and roughly 1/3 the transit boardings of New York. Winnipeg, Vancouver and New York all have similar densities and ridership increases with higher transit supply. In the second grouping of cities, Montreal is slightly denser than Ottawa. Yet the slightly higher VKT yields many more trips per capita. Frankfurt and Montreal have similar ridership levels although Montreal provides 25% more service VKT. Frankfurt's higher density appears to lower 135 servicing requirements to deliver the same ridership. On the other hand, San Francisco and Frankfurt have similar service levels, yet San Francisco's much lower urban density yields few rides per VKT. Table 36 - Influence of transit service and density on transit use in 15 World cities. 1990/91 City Transit Use (boardings/capita) Transit Service (VKT/capita) Metro Density (p/ha) Transit Utilization (boardings/VKT) Los Angeles 55.0 19.8 23.9 2.8 Winnipeg 97.7 40.5 21.3 2.4 Vancouver 117.2 50.3 20.8 2.3 New York 155.2 62.8 19.2 2.5 Ottawa 134.6 55.9 31.3 2.4 Montreal 221.5 60.2 33.8 3.7 Frankfurt 216.6 47 9 46.6 4.5 San Francisco 112.0 49.3 16.0 2.3 Sydney 160.3 <J4 0 16.8 1.7 Toronto 350.0 98.4 4 1.5 3.6 Zurich 514.9 148.1 47.1 3.5 Amsterdam 324.5 60.3 48.8 5.4 Paris 295.0 71.0 46.1 4.2 Vienna 421.8 72.6 68.3 5.8 Munich 403.5 91.4 " 53.6 4.4 In the third grouping of cities, Toronto and Sydney are similar in population and in service VKT. Yet, Toronto's transit ridership is 118% greater than Sydney's. Again, Sydney's much lower urban density figures appear to explain much of the difference.101 In the final group of European cities, Zurich, Amsterdam and Paris all have similar population densities, yet the much higher transit service levels in Zurich deliver much higher ridership levels.102 Paris and Vienna have similar service VKT, yet Vienna's ridership is appreciably higher. Vienna and Munich both have similar ridership levels, balancing inversely related service and density levels. While adding service can do much to boost transit ridership in cities, it is not sufficient unto itself. Cities at the lower end of the density spectrum have ridership levels that fluctuate only within a narrow and low range. This is especially true of the sub-15 p/ha cities. Only those with relatively high transit service levels (and tight parking supplies to add inducements) are able to attain ridership nearing 100 boardings/capita. However, these cities reach an upper limit of ridership potential, which is only surpassed by cities with higher densities. Density appears to be the condition without which high levels of ridership are not attainable. Toronto also has better bus-rail service balance and integration which helps to boost its ridership relative to Sydney's. 136 The very high correlation (0.96) between transit use and transit service in the Canadian cities supports the observation made earlier in the chapter that simply providing adequate levels of service translates into higher ridership. It also suggests that many Canadian cities are underserviced given their densities relative to similarly performing cities elsewhere in the world. Parking supply, transit and cars CBD parking supply exerts a very strong influence on the use of transit. There is a very high correlation (-0.87) between CBD parking supply and transit mode share for the journey to work. This relationship is even stronger when the CBD mode split is examined.103 CBD-bound modals splits for the Canadian cities reveal a negative correlation coefficient of -0.92 for the relationships between CBD parking supply and both the transit and the combined transit-NMT mode splits. Not only is the relationship between parking supply and CBD modal split very strong, but transit use and NMT use are also very responsive changes in parking supply, as demonstrated with the case of Calgary parking (see Figure 31 above). Figure 40 - CBD parking supply and CBD mode split in Canadian cities. 1991— 80% c ft '35 c CO o lit k_ 0) **< 3 E E o o 70% 60% 50% 40% 30% - X N ^ A Toronto + Ottawa"'*' + Montreal + Vancouver ^^*"*««^. Winnipeq C a l g a r y ^ - * ^ ^ + Edmonton 100 200 300 400 500 600 CBD Stalls per 1000 Employees 700 Since, Frankfurt has a similar density, it is also useful to include it in a comparison with this group. 1 0 3 The journey to work modal split data collected in this studied are to region-wide totals. However, since most Canadian cities have transit systems that heavily cater to CBD-bound trips for the journey to work, this measure is a suitable barometer for transit use to the CBD. However, additional data regarding journey to work modal split to the CBD were obtained to confirm this relationship. 1 0 4 CBD-bound transit and NMT mode share data were obtained for Edmonton, Montreal, and Ottawa from TAC (1996). 137 The CBD parking supply also shows a high (-0.77) and significant (0.70) correlation between transit use (boardings) and car use. Since the parking supply data only cover one small (albeit important) portion of the urban area, it cannot reflect the true impact of regional transport demand. The strong correlations with these other factors, however, may also indicate that: the impact is significantly strong enough in the CBD to affect total regional travel over the year; cities with tight CBD parking supply may also have relatively tight parking supply elsewhere in the city; or tight parking supply in the CBD may have synergetic impacts on transit supply and transportation demand region-wide.105 Better collection of disaggregated parking and transportation demand data for the region would provide insights about the dynamic of these relationships region-wide. Such disaggregated data would also help describe the relationship between parking supply and car ownership. Currently, the CBD focussed parking data and region-wide ownership data yield no useful information on these dynamics. Nonetheless, the relationships presented here do confirm the strong association between parking supply and transportation behaviour observed by others (Calgary 1995a; Moore and Thornes 1994; Morall 1996; Shoup 1997). Adjusting parking supply may be an extremely effective, quick, low cost policy lever for boosting transit ridership and NMT use, while reducing car use and ownership. Other factors of importance Several other factors can be identified as having a moderate to high correlation with the auto dependence variables in the Canadian cities. These are as follows: 1. Car ownership and car use have a high positive correlation (0.76). This relationship is even stronger when the larger world cities sample is used (0.84). This may suggest that car availability is a key determinant of use. Since car ownership is expensive and car use is cheap (Litman 1998a), cars will be used once available as a means of extracting value from the sunk, or fixed, costs. This is particularly the case where low densities or lack of transit service require car ownership to meet basic access needs. In the case of high auto dependence cities (e.g., those in Canada, Australia and the U.S.) transit and NMT can rarely meet all of the access needs of residents, thereby requiring the high levels of car ownership (availability) that feed the cycle of auto dependence. 2. Transit supply and car use have a moderate negative correlation (-0.63). This relationship is more pronounced when the larger World cities sample is used (-0.81). Again, the stronger correlation found when a larger group of higher density cities is included may indicate that increasing transit supply plays a greater role in reducing traffic at higher densities. Interestingly, parking supply also has strong correlations with density. This, of course, may simply describe other phenomena for which parking supply is a proxy. For example, cities with low parking supply could have high land costs, which makes low-density development and the building parking expensive. Or, perhaps the cumulative effect of tight parking supply over time has spurred investment in transit, which in turn has shaped land uses. 138 3. Road supply and transit use have a moderate negative correlation (-0.63). However, no clear relationships can be identified between road supply and car ownership and car use (0.43 and 0.26, respectively), although the correlations move in the expected direction. This unexpected result could mean that the road-transit relationship is a chance one. It could also likely stem from the low number of cases, the relatively poor quality of Canadian road supply data and the use of centre-line kilometre instead of lane kilometre data. The use of the World cities data to identify road supply relationships helps to clear up some of the confusion (see below). Since the Canadian cities results are limited by data quality, have a small sample size and reflect only a narrow range of urban densities (i.e., they are relatively homogenous), the World cities correlation data help to identify additional relationships not clear from the Canadian sample alone. These additional correlations of interest from the World cities sample106 are: 4. Road supply is significantly correlated with car ownership and car use (0.69 and 0.67, respectively) and is highly correlated with transit use (-0.80). These data highlight the relationship between road infrastructure supply and auto dependence. The axiom 'build it and they will come' seems to hold true when the larger sample is considered. Since road length figures are only centreline kilometres, they do not completely reflect roadway capacity. It is expected that these relationships would be stronger using lane kilometres. As was pointed out with several other correlations, road supply appears to have synergetic effects with other factors. The causality is uncertain, but it is clear that many of these effects are mutually reinforcing. 5. NMT modal share is significantly correlated with transit use and car ownership (0.69 and -0.70) and is moderately correlated with car use (-0.66). One likely explanation for these correlations is that lower levels of car ownership and use also have other complementary conditions (such as higher densities and better pedestrian and cycling environments) that facilitate more walking, cycling and transit. Better transit services also appear to facilitate (and require) more walking and cycling. The cities with the highest NMT modal splits are those in Europe and Wealthy Asia that have higher densities, better public transport systems and less need for cars to meet basic access requirements. One final finding of interest is that city size (metropolitan area in hectares) demonstrates very weak correlations with all of the variables studied (generally, less than ±0.3) in both the Canadian and World sample. This dispels the myth that increasing auto dependence is a necessary by-product of urban growth. 4.4 CONCLUSIONS The trend data presented in this chapter indicate that Canadian cities are becoming increasingly automobile dependent. Canadian cities are sprawling. While most inner areas remain healthy, or are enjoying a revival, their relative importance in terms of metropolitan activity is declining in all cases and growth is increasingly being accommodated in low-density suburbs. Again, the world sample includes the seven Canadian cities. 139 Most often, these outlying suburban areas are difficult to service by transit and car ownership is virtually a necessity to meet basic transportation needs. Even where transit use has grown in absolute terms, such as in Toronto and Vancouver, it is declining in real terms relative to population, car use and car ownership growth. At the same time, this cycle of auto dependence is being exacerbated by increasing infrastructure provision for the automobile (in terms of road space and CBD parking provision) and declining transit service. Low-density sprawl sends transit into an economic tailspin. Transit service areas increase while the passenger catchment areas decrease. This reduces ridership, requires increased subsidies and puts further pressure on transit operators to cut services. The data presented and analyzed in this chapter add a quantitative dimension to the "cycle of dependence" illustrated in Chapter 2. Increasing sprawl requires increasing car ownership that leads to more driving. Transit, walking and cycling drop as the urban fabric cannot support them. More driving results, which leads to increased demands for automobile infrastructure, which increases sprawl. Transit viability is again reduced. The cycle repeats itself, in many cases resulting in almost complete capitulation to the automobile. Comparative analysis within Canada, and between Canadian and other World cities, reveal commonalities that define levels of automobile dependence. Those cities with higher urban densities, higher transit service provision and lower automobile infrastructure provision all have lower levels of c