UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Housing and location of young adults, then and now : consequences of urban restructuring in Montreal.. Moos, Markus 2012

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
ubc_2012_spring_moos_markus.pdf [ 13.12MB ]
[if-you-see-this-DO-NOT-CLICK]
Metadata
JSON: 1.0072625.json
JSON-LD: 1.0072625+ld.json
RDF/XML (Pretty): 1.0072625.xml
RDF/JSON: 1.0072625+rdf.json
Turtle: 1.0072625+rdf-turtle.txt
N-Triples: 1.0072625+rdf-ntriples.txt
Original Record: 1.0072625 +original-record.json
Full Text
1.0072625.txt
Citation
1.0072625.ris

Full Text

  HOUSING AND LOCATION OF YOUNG ADULTS, THEN AND NOW: CONSEQUENCES OF URBAN RESTRUCTURING IN MONTREAL AND VANCOUVER      by  MARKUS MOOS  BES, The University of Waterloo, 2004 MPL, Queen’s University, 2006     A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS OF THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE STUDIES  (Geography)        THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)    February 2012   © Markus Moos, 2012 ii Abstract Young adults, 25 to 34 years of age, decide on housing, residential location and commuting patterns in an altered context from when the same age cohort entered housing markets in the early 1980s. Neo-liberalization reduced the availability of low- cost, rental housing, and post-Fordist restructuring increased labour market inequality. Societal changes contributed to decreases in household size and delay in child bearing. This thesis asks how the contextual changes factor into young adults’ housing decisions in the Montreal and Vancouver metropolitan areas where restructuring occurred differently, and discusses implications for equity and sustainability. The young adult residential ecology is increasingly concentrated into higher density and amenity-rich neighbourhoods, particularly near transit in Vancouver. The trends are explained by shifts toward the service sector, declining real incomes and growing inter-generational wage inequalities that reduce young adults’ spending power in housing markets, especially in Vancouver with its speculative land market and wealthy immigrants. Holding other characteristics constant, young adults in Vancouver are less likely to reside in single-family dwellings than detached, row or apartment units. In Montreal the trend is toward single-family living. Commuting distances and modes are similar between Vancouver and Montreal but multiple-person households and those with children have longer and more automobile-oriented commutes in Vancouver. The changes reflect higher increases in housing costs and densities in central areas in Vancouver. Montreal has more sustained government support for housing, a larger rental sector and therefore less rampant increases in housing costs. The restructuring of Vancouver’s housing market makes it more difficult than in Montreal to keep accessible the more ‘sustainable’ locations to households of all sizes. Household structure and life- cycle stage, not social status alone, determine location and the commute. A greater sustainability challenge in Montreal will be to stem the shifts toward ownership of single-family dwellings. Generally, young adults’ housing outcomes are more evidently shaped by their position in the labour market, which is increasingly determined by educational attainment. The thesis works conceptually within structuration theory, noting how contexts shape demand but are themselves re-shaped by changing demand. Both contextual and neo-classical arguments have relevance to the overall argument.iii Preface Chapter Three is in part a modified version of an article that has been previously published by the author, with Andrejs Skaburskis, and is reprinted with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved. The student, Markus Moos, identified the research question, prepared the manuscript and conducted the majority of the analysis of research data and writing. Input from the co-author relates mostly to questions of research design and use of data from previous collaborative work.  The thesis also builds on published research conducted by the student as part of his larger research project. These publications are acknowledged below for their conceptual overlap and use of similar literatures and methods in the research. For instance, the statistical models estimating permanent income and housing consumption included in Chapters Two and Six, and the literature describing the approach, are derived and substantially modified versions from research previously published:  Moos, M. & Skaburskis, A. (2008). The probability of single-family dwelling occupancy: Comparing home workers and commuters in Canadian cities. Journal of Planning Education and Research, 27(3), 319-340. ©SAGE Publications Inc. All rights reserved. http://jpe.sagepub.com/content/27/3/319  Moos, M. & Skaburskis, A. (2009). Workplace restructuring and urban form: The changing national settlement patterns of the Canadian workforce. Journal of Urban Affairs, 32(1), 25-53. ©John Wiley and Sons. All rights reserved. http://onlinelibrary.wiley.com/doi/10.1111/j.1467-9906.2009.00476.x/full  Components of the following publications draw broadly on the research conducted for this thesis, and do not pertain to a specific chapter:  Barnes, T., Hutton, T., Ley, D. & Moos, M. (2011). Vancouver: An entrepreneurial economy in a transnational city. In L. Bourne & T. Hutton (Eds.). Canadian Urban Regions: Trajectories of Growth and Change (pp. 291-328). Don Mills, ON: Oxford University Press.  Skaburskis, A. & Moos, M. (2010). Cities as Land Markets. In T. Bunting, P. Filion & R. Walker (Eds.). Canadian Cities in Transition: New Directions in the Twenty-First Century 4th Edition (pp. 225-242). Don Mills, ON: Oxford University Press.  Quastel, N., Moos, M. & Lynch, N. (Under review). Sustainability as density and the return of the social: The case of Vancouver, British Columbia. Urban Geography.    iv Table of Contents Abstract ....................................................................................................................... ii Preface ........................................................................................................................ iii Table of Contents ....................................................................................................... iv List of Tables .............................................................................................................. vi List of Figures .......................................................................................................... viii Acknowledgements ..................................................................................................... x Dedication ................................................................................................................... xi Chapter One: Introduction ........................................................................................ 1 1.1 The Role of Context in Residential Location ..................................................... 7 1.2 The Case Study Cities ....................................................................................... 12 1.3 Methodology: A Research Narrative of Place .................................................. 23 1.3.1 Data sources and geography ..................................................................... 28 1.4 Thesis Overview ............................................................................................... 32 Chapter Two: The Changing Cities ........................................................................ 35 2.1 Defining the Young Adult Cohorts ................................................................... 39 2.2 Young Adult Cohorts in Specific Times and Locations ................................... 43 2.3 Reduced Government Involvement in Housing ............................................... 48 2.3.1 Concurrent changes in the organization of production ............................. 51 2.4 Inner City Revitalization ................................................................................... 56 2.5 Coordinating Land Use and Transport as a Sustainability Strategy ................. 62 2.6 Paying More for Housing ................................................................................. 77 2.6.1 Data summary and preparation ................................................................. 80 2.6.2 A geography of expenditure patterns ......................................................... 87 2.7 Discussion ......................................................................................................... 91 Chapter Three: Global Restructuring and Housing Demand .............................. 94 3.1 Housing and Labour Market Dynamics in a Global Context ........................... 98 3.1.1 The changing profile and settlement patterns of immigrants .................. 102 3.2 Measuring Housing Demand and Neighbourhood Transition ........................ 104 3.3.1 Multivariate analysis of the user cost of housing .................................... 117 3.4 Neighbourhood Transition .............................................................................. 126 3.4.1 Changing housing stock characteristics .................................................. 127 3.4.2 Dwelling values and neighbourhood change .......................................... 135 3.5 Discussion ....................................................................................................... 141 Chapter Four: The Changing Metropolitan Economies and the Young Adult Labour Force ..................................................................................................... 145 4.1 The Young Adult Labour Force ..................................................................... 149 4.2 The Income Distribution ................................................................................. 157 4.3 Generational Income Gap ............................................................................... 167 4.3.1 Income determinants ................................................................................ 173 4.4 Discussion ....................................................................................................... 178 Chapter Five: The Young Adult Residential Ecology ......................................... 184 5.1 Changing Aggregate Location Patterns .......................................................... 188 5.2 Relative Centralization ................................................................................... 206 5.3 A Regression Model of Residential Location ................................................. 211 v 5.3.1 Model Specifications ............................................................................... 212 5.3.2 Model Outcomes ...................................................................................... 219 5.4 Discussion ....................................................................................................... 221 Chapter Six: The Housing and Commuting Decisions of Young Adults ........... 227 6.1 Housing Type and Tenure .............................................................................. 232 6.2 Housing and the Commute ............................................................................. 240 6.3 The Changing Determinants of Housing Type and Tenure ............................ 247 6.3.1 Multinomial logistic regression ............................................................... 249 6.3.2 Household characteristics ....................................................................... 250 6.3.3 Multinomial logistic regression results ................................................... 252 6.4 Commute Distance and Mode ........................................................................ 257 6.4.1 Factors influencing the journey-to-work ................................................. 258 6.4.2 Ordered logistic regression of commute distance ................................... 260 6.4.3 Multinomial regression of commuting mode ........................................... 266 6.5 Discussion ....................................................................................................... 271 Chapter Seven: Conclusions – Growing Just, Sustainably? ............................... 275 7.1 Contours of Post-Fordist Housing and Labour Markets ................................. 279 7.2 Young Adults’ Housing, Location and Commuting Decisions ...................... 283 7.2.1 A “smart growth” generation? ................................................................ 292 7.4 Limitations ...................................................................................................... 297 7.3.1 Challenges for the future ......................................................................... 301 Bibliography ............................................................................................................ 304      vi List of Tables Table 2.1 – Change in the modal split in the journey to work ........................................ 71 Table 2.2 – Proportion of automobile commuters by occupation and educational attainment ............................................................................................................... 75 Table 2.3 – Expenditure on principal accommodation by tenure ................................... 78 Table 2.4 – Summary of young adult households in the provinces of Quebec and British Columbia ................................................................................................................. 83 Table 2.5 – Household expenditure on rent as a function of dwelling and geographic characteristics ......................................................................................................... 86 Table 2.6 – Correlates of the percentage of household income allocated to housing expenditure by year ................................................................................................. 88 Table 2.7 – Correlates of percentage of household income allocated to imputed rent by province .................................................................................................................. 89 Table 3.1 – Means of housing and household characteristics of owners, Montreal CMA 2001 and 1981 ....................................................................................................... 112 Table 3.2 – Means of housing and household characteristics of owners, Vancouver CMA 2001 and 1981 ............................................................................................ 113 Table 3.3 – User cost of housing as a function of housing and household characteristics  .............................................................................................................................. 118 Table 3.4 – Effect of income and immigration status on the user cost of housing, 1981 and 2001 ................................................................................................................ 120 Table 3.5 – Effect of income and immigration status on user cost of housing by place of birth, Montreal ...................................................................................................... 124 Table 3.6 – Effect of income and immigration status on user cost of housing by place of birth, Vancouver ................................................................................................... 125 Table 3.7 – Changing housing stock characteristics ..................................................... 128 Table 3.8 – Principal component analysis of housing stock characteristics ................. 132 Table 3.9 – Beta regression coefficients with housing stock changes as the dependent variable ................................................................................................................. 134 Table 3.10 – Change in census tracts by proportion of recent immigrants .................. 137 Table 3.11 – Change in dwelling value as a function of census tract characteristics, 1981-2001 ............................................................................................................. 138 Table 4.1 – Characteristics of the young adult labour force ......................................... 150 Table 4.2 – Measuring inequality using Gini coefficients ............................................ 166 Table 4.3 – Income as a function of labour force characteristics, Montreal and Vancouver CMA ................................................................................................... 176 Table 5.1 – Correlation coefficients for proportion of population 25 to 34 years of age  .............................................................................................................................. 201 Table 5.2 – Indices of relative centralization ................................................................ 207 Table 5.3 – Principal components (unrotated) .............................................................. 215 Table 5.4 – Linear and spatial regression of young adult residential location in Montreal  .............................................................................................................................. 216 Table 5.5 – Linear and spatial regression of young adult residential location in Vancouver ............................................................................................................. 217 vii Table 6.1 – Proportion of households spending more than 30% of income on shelter costs by household income ................................................................................... 228 Table 6.2 – Households with a maintainer 25 to 34 years of age by housing type ...... 233 Table 6.3 – Proportion of households with a maintainer 25 to 34 years of age who are homeowners by dwelling type .............................................................................. 239 Table 6.4 – Young adults’ changing commuting distance and mode ........................... 241 Table 6.5 – Young adults’ housing costs by the length of the commute, 2006 ............ 242 Table 6.6 – Summary of variables used in multinomial regression of housing type and tenure .................................................................................................................... 250 Table 6.7 – Multinomial logistic regression of housing type and tenure ..................... 253 Table 6.8 – Ordered logistic regression results of young adults’ commuting distance 262 Table 6.9 – Multinomial logistic regression of young adults’ commuting mode, 2006  .............................................................................................................................. 267     viii List of Figures Figure 1.1 – Downtown Montreal and Vancouver ......................................................... 14 Figure 1.2 – Changing urban densities ........................................................................... 15 Figure 1.3 – Montreal and the regional geography ......................................................... 17 Figure 1.4 – Vancouver and the regional geography ...................................................... 19 Figure 1.5 – The changing housing market characteristics ............................................ 22 Figure 1.6 – The Montreal census metropolitan area ..................................................... 30 Figure 1.7 – The Vancouver census metropolitan area .................................................. 31 Figure 2.1 – Age distribution in the Montreal and Vancouver CMAs ........................... 42 Figure 2.2 – Conceptualization of birth cohorts and age strata ...................................... 46 Figure 2.3 – Walkability and rapid transit lines surrounding Vancouver’s downtown .. 65 Figure 2.4 – Walkability and rapid transit in the Vancouver CMA ............................... 66 Figure 2.5 – Walkability index and Metro lines on the Island of Montreal ................... 67 Figure 2.6 – Walkability and commuter rail lines in the Montreal CMA ...................... 68 Figure 2.7 – Cycling infrastructure in Montreal and Vancouver neighbourhoods ......... 73 Figure 3.1 – Average CMA dwelling value and proportion of household maintainers immigrants .............................................................................................................. 96 Figure 3.2 – Montreal CMA zones ............................................................................... 109 Figure 3.3 – Vancouver CMA zones ............................................................................ 110 Figure 4.1 – Young adult household income in Canada 1976-2008 ............................ 147 Figure 4.2 – Occupational distribution of young adult labour force, 2006 .................. 152 Figure 4.3 – Industry distribution of young adult labour force, 2006 .......................... 153 Figure 4.4 – The young adult income distribution in Montreal, 1981 and 2006 .......... 161 Figure 4.5 – The young adult income distribution in Vancouver, 1981 and 2006 ....... 162 Figure 4.6 – Proportion of persons in families below the LICO by the age of the primary maintainer, Montreal CMA .................................................................................. 164 Figure 4.7 – Proportion of persons in families below the LICO by the age of the primary maintainer, Vancouver CMA ................................................................................ 165 Figure 4.7 – Average income by occupation for young adults ..................................... 168 Figure 4.8 – Income gap between young adults and the population 35 years of age and older ...................................................................................................................... 169 Figure 4.9 – Young adult household income in Montreal 1976-2008 .......................... 171 Figure 4.10 – Young adult household income in Vancouver 1976-2008 ..................... 172 Figure 5.1 – Young adult residential locations and housing densities in the five largest CMAs .................................................................................................................... 187 Figure 5.2 – Location quotients of young adults in the Montreal CMA, 1981 and 2006  .............................................................................................................................. 191 Figure 5.3 – Location quotient of young adults in the Vancouver CMA, 1981 and 2006  .............................................................................................................................. 192 Figure 5.4 – Location quotients of young adults in Montreal’s inner city, 1981 and 2006  .............................................................................................................................. 193 Figure 5.5 – Location quotients of young adults in Vancouver’s inner city, 1981 and 2006 ...................................................................................................................... 194 Figure 5.6 – Amenities and housing in Montreal neighbourhoods with high shares of young adults .......................................................................................................... 197 ix Figure 5.7 – New housing developments in Vancouver neighbourhoods with high shares of young adults ..................................................................................................... 198 Figure 5.8 – Relative centralization of the population ................................................. 210 Figure 6.1 – Households with a maintainer 25 to 34 years of age by housing type relative to all households ...................................................................................... 235 Figure 6.2 – Homeownership by age of household maintainer .................................... 236 Figure 6.3 – Distribution of dwelling values by housing type and age of maintainer .. 245 Figure 6.4 – Distribution of gross rent by dwelling type and age of maintainer .......... 246   x Acknowledgments I would like to thank my supervisors, Professor David Ley and Professor Elvin Wyly, for their advice, support and patience as I worked my way through the completion of this thesis. I will always have fond memories of two such generous and thoughtful advisors.   I thank my committee members, university examiners and chair, Professor Daniel Hiebert, Professor Thomas Hutton, Professor Jamie Peck, Professor Carrie Yodanis and Professor Larry Frank. I also owe thanks to numerous other faculty, staff, post-docs and students in the Geography Department at the University of British Columbia for their help, guidance and inspiration at various stages of this work; including Professor Trevor Barnes, Professor Marwan Hassan, Jon Clifton, Ben Crawford, Cory Dobson, Lisa Erven, John Gallagher, Rowan Hicks, Elaine Ho, Katie Kinsley, Sara Koopman, Scott Krayenhoff, Chris Ley, Jason Leach, Nicolas Lynch, Pablo Mendez, Yolande Pottie- Sherman, Sonya Powell, Noah Quastel, John Richards, Elliot Siemiatycki, Bjoern Surborg, Roza Tchoukaleyska, Ren Thomas and Justin Tse. I thank Professor Andrejs Skaburskis for being a continuing source of ideas and guidance, and Professor Mark Roseland for his advice during the comprehensive exam process.  I also would like to thank the various other academics, policy-makers, politicians and real estate agents that provided me with valuable information and sources by taking time to speak to me, respond to my emails, or sharing their ideas in public forums.   I thank my wife, Sarah, for her love, support and great deal of understanding. I thank the Octagon, the Brits, all our other dear friends and both our families for their support and fun times together! I thank my little sister, Beatrice, for her advice on statistics and life in general. Many thanks to Mami and Papi, Beatrice and Michael, Christian and Irene, Urs and Brigitte, Peter and Louise, and Scott, Jessica and Miles.   I thank the Social Sciences and Humanities Research Council of Canada, the University of British Columbia and its Department of Geography and Urban Studies program, IODE Canada and the family of J. Lewis Robinson for their financial support.   The author is responsible for any remaining errors. Any opinions, interpretations and conclusions are the authors’ and do not necessarily reflect those of the persons or institutions acknowledged above.       xi Dedication – To Tante Hedy  for always being an inspiration to speak up for what we believe is right and just  1 Chapter One: Introduction “Urban Canada, at all spatial scales, is being transformed through the intersection of changes taking place in the economy, trade patterns, technology, the demographic structure and immigration, as well as through public policy.” (Bourne, 2007a, p. 3)  This thesis is about changes in residential location and housing decisions due to structural transformations in urban housing markets. There have been in recent decades substantive changes in the spatial structure and social composition of Canadian metropolitan areas. This has raised important questions regarding the emerging character of housing markets and the socio-spatial organization of cities (Bourne, 2007b; Bunting, Filion & Walker, 2010). Most population growth, primarily through immigration, has occurred in the largest cities, contributing to their expanding populations (Bourne, 2007a). Urban growth has materialized in two general ways, the continuing expansion of suburbs and the re-development of central cities, although with increasing internal diversity in structure and socio-economic composition (Bourne & Rose, 2001; Smith, 2006; Ley & Frost, 2006). Growth contributed to escalating housing prices, particularly in the inner cities revitalized by the forces of gentrification and government investment in amenities (Ley, 1996; Skaburskis & Moos, 2008). Higher costs have, along with the increase in the number of smaller households and environmentally- and fiscally-motivated growth management policies, in turn facilitated increases in higher density housing forms in central areas and suburban nodes (Champion, 2001; Filion & Bunting, 2010).  These changes have taken place during a time of global socio-economic restructuring that fundamentally altered the character of the industry and occupational 2 structure of the Canadian economy, now characterized by growth in the service sector and a declining number of manufacturing jobs (Bourne et al., 2011). In the political realm, neo-liberal ideals were brought into effect that resulted in retrenchment of the welfare state (Hackworth & Moriah, 2006). Many, although certainly not all, of the Keynesian-inspired policies that provided a social safety net, and also facilitated suburbanization through infrastructure provision, made way for privatization of government services, including transportation infrastructure that saw increases in user costs (Gillespie, 1983; Fishman, 2005; Filion, 2001; Walks, 2001). The loss of traditionally well-paid, often unionized, employment in the manufacturing sector, coupled with increasing housing prices, a growing number of low-paying service sector jobs and reduced government investment in assisted housing, contributed to growing housing affordability burdens (Bunting, Walks & Filion, 2004; Moore & Skaburskis, 2004; Walks, 2011a).  Growth has also contributed to environmental issues such as traffic congestion, urban sprawl and pollution that pose great challenges for the sustainability of cities (Newman & Kenworthy, 1999; Roseland, 2005; Hanna, 2006). The issues have prompted a set of urban land use and transportation policy responses that aim to reduce automobile use by increasing development densities, providing public transit and containing the spread of cities (Campbell, 1996; Kenworthy, 2006; Banister, 2008). However, various commentators point to the growth agenda inherent in these policies and their potential for redistributive effects; for instance through the displacement of lower-income populations living in denser areas better served by transit and by the continuation of suburbanization by the middle-classes unable to pay inner city housing 3 costs but faced with increasing oil prices (Hall, 1996; Couch & Dennemann, 2000; Krueger & Gibbs, 2007; Quastel, 2009). The ultimate result of the combined changes, as others have already observed (Clapham, 2002; Carr, 2004; Beer, 2006; Calvert, 2010), is that households encounter a very different set of conditions that shape their location and housing decisions than was the case twenty or thirty years ago. Inevitably, households with different incomes, and differing characteristics determining earnings, will face different kinds of challenges that result in new kinds of inequalities (Furlong & Cartmel, 2007; Walks, 2009; Boschmann & Kwan, 2010; Beer et al., 2011; Kershaw, 2011). The debate sees social equity and environmental advocacy collide, two issues at the forefront of public policy debates in several major cities that are trying to “green the city” while struggling to retain affordable housing (Hall, 1996; Marcuse, 1998; Godschalk, 2004; “Briefing: London and Paris”, 2008).  The aim of this thesis is to ask how the location and housing decisions of specifically young adults have changed in the context of these contemporary urban transformations in Montreal and Vancouver, and to document the implications for commuting patterns. In a recent literature review, Calvert (2010) outlines some of the emerging challenges for young adults in Britain as they are moving into the housing market—such as changing age of departure from the parental home, rising debt levels, declining incomes, rising prices and labour market uncertainties. Beer (2006) points to similar changes in Australia, highlighting the changes in housing policy that “emphasize market-based solutions” and the growing “risks” in labour markets for young adults today as compared to when the baby-boomers were entering housing and labour markets (also see Beer et al., 2011). These findings certainly also resonate with 4 Canadian social trends (Beaupré, Turcotte & Milan, 2006). As Bourne & Rose (2001) note, however, the trends of socio-economic transformations are already “reasonably well-documented” but still today less is known “about how they come together, in particular places at particular times, with what impacts and for whom” (p. 107).  While Canadian urban research in recent years has documented severe implications of the transformations for socio-spatial inequalities, housing affordability and poverty (see Walks, 2009), the implications of urban restructuring for young adults warrant further exploration. As young adults are only entering housing and labour markets, their analysis provides useful insight into how households make decisions in a given context and their implications for the commute. Young adults also generally hold a more precarious position in the labour market, thus their analysis provides insight into market conditions (Myles, Picot & Wannell, 1993). Previous empirical studies of young adults have focused on the national or provincial scales warranting more recent study of specific urban contexts (Schrammel, 1998; Skaburskis, 2002; Kershaw, 2011). Moreover, the geography of age cohorts has been understudied until recently when it became evident that age is an increasingly important variable of socio-spatial differentiation that warrants further exploration in its own right (Hagestad & Uhlenberg, 2006; Vanderbeck, 2007; Rosenberg & Wilson, 2010). Also few have studied young adults’ commute patterns explicitly (Thomas, 2007). Although the association between young age and lower automobile use and shorter commuting distances are well-known as they relate to the lower earnings and higher residential mobility of younger workers, less is known about the influence of the urban context (Finnie, 2004; Hanson & Giuliano, 2004; Shearmur, 2006).  5 The comparison of Montreal and Vancouver is particularly useful because it sheds light on the consequences of long-term changes in the housing market in cities where the processes of urban restructuring have played out in different ways. The two cities have distinct housing markets and Montreal with its large manufacturing base was much more directly affected by de-industrialization whereas new economy growth and neo-liberal restructuring are more important factors to consider in Vancouver. Urban planning policies that aim to enhance the sustainability of cities have also been in place for much longer in Vancouver, and are reflected in its urban structure (Tomalty, 1997). The temporal period covered in the thesis is from the early 1980s when restructuring was taking hold in Canadian cities (Rose, 1999) until the mid-2000s. The study period ends just before the recent global recession hit property and labour markets, thus earnings and affordability are likely to have declined since the findings presented here. The characteristics of young adults over this timeframe nonetheless reflect the structural changes influencing Canadian urbanism in two specific metropolitan areas. Thus more generally, this thesis asks about changes in the structure of cities as a result of contemporary societal transformations, contributing to and benefiting from the insight of a long line of chiefly, but not exclusively, Canadian urban research on the relationship between changes in society and space in the urban realm (Massey, 1980, 1995; Ley, 1988, 1996; Bunting & Filion, 1988; Mills, 1989; Bourne, 1993; Wyly, 1999; Walks, 2001, 2009; Germain & Rose, 2000; Filion, 2001; Bourne & Rose, 2001; Hackworth, 2005; Gauthier & Gilliland, 2006).  Specifically, the thesis adds to the understanding of the impacts of urban restructuring by way of pointed empirical contributions that offer generalizable 6 evidence on aggregate outcomes. There are three broad sets of research questions that guide the analysis:   • How have the conditions of urban development as they emerged since the early 1980s altered the characteristics of the urban housing markets in Montreal and Vancouver? • In what ways are the changes in location and housing decisions of young adults in the two cities reflective of the structuring impacts of the changing urban context? What are the implications for their daily commute (journey to work)?  • What do the patterns imply for current policy debates on environmental sustainability and social equity?   Environmental sustainability is loosely measured in this thesis as household adjustments that are resulting in the kinds of housing, location and commuting patterns commonly understood to be associated with lower resource consumption or pollution; adjustments such as residing in proximity to amenities and transit, higher density housing, shorter commutes and walking, cycling and taking transit to work (Hall, 1996; Walker & Rees, 1997; Newman & Kenworthy, 1999; Gunder, 2006; Moos & Skaburskis, 2008; Ewing, Bartholomew & Winkelman, 2008; Quastel, Moos & Lynch, under review). Following Burton (2000), social equity is interpreted broadly through Rawls’ notions of “distributive justice” in that “primary goods” ought to be “equally distributed” unless an alternate strategy favours those already worst off (p. 1970-1972). The analysis builds on Burton’s measurement of equity outcomes in terms of the income distribution, 7 housing affordability and “better access” to amenities, employment and alternative modes of transportation (p. 1972). The nature of the questions asked also necessitates a conceptual framework that can account for both the contextual and individual level factors acting on location and housing decisions. The contours of such a conceptual framework and the two empirical case studies are discussed next, followed by an overview of the thesis methodology and structure.  1.1 The Role of Context in Residential Location By asking about young adults’ residential decisions within specific urban housing markets, the thesis assigns import to context in understanding household level decisions. Researchers have put different weights on the importance of individual versus structural factors in shaping socio-spatial outcomes (Chouinard, 1997), but the benefits of drawing on multiple theoretical perspectives have been previously established (Bourne, 1981; Kauko, 2001). While this thesis draws on a broad literature from social, cultural and economic geography and planning, it primarily uses insights from urban economics and structural theorists to interpret a quantitative analysis of the determinants of location, housing and the commute. The approach is accommodated under the conceptual umbrella of Giddens’ (1984) structuration theory. It is inspired by Pratt’s (1996) “weaving” of “micro-processes and macro-structures” through structuration (p. 1361), as well as Clapham’s (2002) call for a better understanding of how households make decisions within “the housing context facing them” (p. 59), Guy & Henneberry’s (2000) “integrating” of “social structures” and “economic processes”, and Jarvis’ (2003) “household approach” to “understand the very significant structures 8 of constraint” on residential location and the commute (p. 592). The intent is in part to demonstrate that observable patterns from quantitative analysis can help reveal the ways decisions materialize for different types of households in different housing market contexts. The analysis is not intended, however, to empirically unravel the complex intra-household decision-making processes (Jarvis, 2001) or the multiple “meanings” attached to housing shaped through various societal “relationships and interactions” (Clapham, 2002, p. 64).  Alonso (1964) laid the groundwork for the neoclassical urban economic models of residential location that focus on individual choice. The model includes the usual neoclassical assumptions of the utility maximizing homo economicus, which in this case trades-off housing and journey to work costs within a budget constraint to decide upon a location in a monocentric city. The general ideas of the model have since been expanded considerably to account for changes in urban structure, such as polycentricity, and household composition and the trade-offs with other factors, for instance amenities or transport infrastructure (Shearmur & Charron, 2004; Giuliano, Gordon & Park, 2010; Verhetsel, Thomas & Beelen, 2010). In contrast, the structuralist studies have generally focused on the role of class and race in determining spatial outcomes (McLaffferty & Preston, 1996; Wyly, 1998; Boschmann & Kwan, 2010), and important insight has also been gained from the feminist literature on the internal workings of households and gender relations (Hanson & Pratt, 1988; Jarvis, 2005; Ward et al., 2010). What is being called structuralist studies here actually combines a diverse literature, but as a group it differs from the neo-classical economic studies in that spatial differentiation is 9 conceptualized as the outcome of the workings of social, economic and political processes rather than purely the rational choices of individuals (Clapham, 2002).  Giddens’ (1984) framework of “structuration” is commonly credited with going some distance in reconciling the agency-structure conundrum, forming a basis for research that asks about the “constraints that configure the choices that individuals are able actually to implement” (Pratt, 1996, p. 1359). In simplified terms, structuration theory holds that individuals make decisions within the constraints of the systemic structures, which are simultaneously also the constructs of individual level actions. As Pratt (1996) explains, the “duality” of structure and individual is conceptualized by Giddens through “institutions”, which are “the organizations, both formal and informal, that we establish to regulate our societies” (p. 1362). In this thesis the housing context, which is defined as the price, tenure, stock and policy configurations that the household encounters at a given point in time and space, encapsulates the set of institutions that order the location, housing and commute decisions, and also the history of these factors. The physical urban form, such as the housing stock, the transport infrastructure and the shape of the city, is part of this housing context because it is both a manifestation of institutional settings and the structure for household decisions (Filion, 2001; Vandersmissen, Villeneuve & Theriault, 2003). But also important are the characteristics of households in that they reflect societal changes, for instance declining household size or changing occupational structures and incomes, and these in turn alter the reality within which households make decisions (Rose & Villeneuve, 2006).   The use of the neo-classical economic tools is useful because these take into account the influence of household characteristics, such as household size, that are all 10 too commonly ignored when making comparisons across structural contexts (Clapham, 2002). At the same time, the application of the economic tools in different geographic and temporal contexts is in essence a critique of traditional urban economic theory, which has historically viewed the context of the city as only providing “secondary… generalizations of a historic-spatial relevant nature” (Robbins, 1935; Maclennan, 1977, p. 69). To be clear, it is not the act of abstraction that is being problematized here. Rather questionable, however, is the abstraction of factors known to influence the relationship under consideration, such as the important influence of the history of the housing stock, policies and urban form (Harvey, 1973/2008), some of which have since been integrated into economic analysis (Gibb, 2003). Neoclassical economics is nonetheless still often critiqued on the basis of what Howard & King (2001) have established as its “fallacy of composition” whereby the theory “correctly” assumes “that any particular structure is reducible to actions” but incorrectly infers “that all structures can thereby be eliminated” (p. 788). Alonso removed the context of the city in his model of residential location almost completely whereas cultural, feminist and Marxist geographers have made context, broadly interpreted, a cornerstone of their analysis in that they have examined the ways class, gender and race produce inequalities that relate to social constructs or structural conditions, not individual preferences (Ley, 1985; Hanson & Pratt 1988; Wyly, 1998; Lee, Slater & Wyly, 2008). Harvey (1973/2008) critiques in particular the notion of equilibrium in economic theory, which he argues ignores the “various speeds of adjustments in the urban system” (p. 56). He suggests the existing distribution of resources plays a large role in influencing how different social groups are able to adjust to societal changes. Harvey notes: 11  “Certain groups, particularly those with financial resources and education, are able to adapt far more rapidly to a change in the urban system, and these differential abilities to respond to change are a major source in generating inequalities.” (p. 56)  An analysis of household-level characteristics for different social groups in different contexts is one strategy for reinserting such structural considerations into quantitative analysis of location, housing and commuting decisions in an era of accelerated urban socio-economic restructuring.     It is important to note that adding the specific temporal and geographic context of the city into the analysis of household level variables does not mean including a “limitless number” of explanatory factors but “rather, it means making” context “a question, instead of an answer known in advance” (Mitchell, 2002, p. 52-53)1. One example, derived from Harvey (1973/2008), is that instead of assuming that “competitive bidding” determines residential location patterns as in the economic models, one can ask about the degree to which the governance context distributes land through “bidding”, as opposed to other distributive mechanisms, and the differing outcomes these produce for households (p. 137) (also see Lee et al., 2008). Thus, the temporal and geographic context of the city becomes a variable in the analysis, not as a causal factor but as the place that provides the structure for societal change to occur in particular ways (Filion, 2001).  1 Mitchell (2002) makes this point more specifically in regards to agency and power. “[M]aking this issue of power and agency a question…means acknowledging something of the unresolvable tension” and “requires acknowledging that human agency, like capital, is a technical body, is something made”, he suggests (p. 53). The argument can be extend to the idea that the urban context is “something made” and thus requires ‘unpacking’ and explicit analysis.        12 The framing of location and housing decisions in the context of urban change is particularly useful in terms of the debate regarding the sustainability of commuting patterns (Horner, 2004; Jarvis, 2003). The heavy reliance on urban economic and transportation studies to formulate sustainability policies has meant, as Pratt (1996) argues, that questions regarding “the complex political, economic and social factors” shaping cities have been largely absent from the sustainability literature (p. 1360). A literature that combines, to various degrees, insights from urban studies, feminist, cultural and economic geographies, housing and sustainability is beginning to emerge, often pointing to the insufficiency of density in reducing commute distance or attaining modal shifts due to the effect of rising incomes, and the potential inequities produced as households are differently able to adjust to rising oil prices and policies that push for greater transit use (Garrett & Taylor, 1999; Pratt, 1996; Jarvis, 2003; Danyluk and Ley, 2007; Quastel, 2009). The thesis makes an empirical contribution to this emerging, inter-disciplinary literature by asking about the housing, location and commuting patterns of young adults in the context of two specific urban housing markets.  1.2 The Case Study Cities Although Montreal is often praised for the vibrancy of its urban neighbourhoods and the aesthetic qualities of its heritage buildings, some commentators also describe the city as “ugly” due to the lack of a consistency in architectural form and dilapidated state of some of the housing stock (Hebert, 1989, p. 17; Frost, 1981a). Regardless of whether one agrees with such an assessment, the contrast with Vancouver is evident, which now often appears to receive exceptionally high praise for its urban design in 13 architectural and planning circles (Berelowitz, 2005). The influence of emerging urban planning ideals emphasizing the aesthetic qualities of place are reflected in Vancouver’s urban form (Lynch & Ley, 2010). This exerts influence on the cost and characteristics of housing. Thus more generally, Montreal provides a setting for analyzing residential location and housing trends that differs from Vancouver in important ways: Montreal is distinguished from Vancouver by its lower growth rates (and even decline), fewer physical and regulatory (planning) constraints on urban expansion, a more fragmented regional governance context, and historically higher central city densities but also lower (even negative) density gains in recent years (Filion et al., 2010).  The two cities have also traditionally been characterized by quite different kinds of housing stock (Demchinsky, 1989; Engeland et al., 2005). Montreal’s large rental market in the old, dense inner city surrounded by vast suburbs stands out against Vancouver’s high-density centre containing newly developed condominium towers surrounded by single-family dwellings and secondary centres in the suburbs (Figure 1.1). The density profiles in Figure 1.2 show the dramatic growth in the central city residential component in Vancouver, and the relatively higher densities in the outlying suburbs. Montreal’s historically higher centrally city densities extend further into the inner suburban areas but the outlying suburbs have lower densities than in Vancouver.  At the same time, it is important to remember that Montreal and Vancouver are undergoing similar changes related to demographic shifts, gentrification of the central city and immigration, although again these factors do play out in different ways (Ley, 1996; Heisz, 2006). The two cities differ in size, but both metropolitan regions are 14 experiencing growth related problems such as housing affordability concerns, traffic congestion and sprawl (Bunting et al., 2004; Tomalty, 1997).  Figure 1.1 – Downtown Montreal and Vancouver  Notes: Montreal’s downtown contains a number of high-rise office towers, and the residential components include a mix of row housing, duplexes and high-rise apartment units (Above; May, 2008). Vancouver’s downtown also has an office component but high-rise condominium apartments dominate the skyline (Below; June, 2008).  Montreal and Vancouver have relatively well-maintained central city public transport systems with routes to at least a few of the suburbs but in Vancouver the transit stops have become areas of concentration for higher density development (Filion et al., 2010). Tourism plays an important role in the economies of the two metropolitan areas, and nearby recreational destinations such as Mount Tremblant in Montreal and Whistler in 15 Vancouver benefit from proximity to the large population base and transportation hubs. Neither is the provincial capital of their provinces although Montreal has more public sector functions than Vancouver.  Figure 1.2 – Changing urban densities  Notes: Persons per square kilometer in census tracts, three-kilometer moving average. Source: Calculated using Statistics Canada census tract data (1981a; 2006a).  However, “the peculiar nature of the position of Quebec in the Canadian federal state architecture” also plays a particular role in differentiating urban policy in Montreal from cities in the rest of the country (Boudreau et al., 2007). Montreal and Vancouver differ in their urban policy configuration in the way these are shaped by the workings of local, provincial and federal levels of government. While Vancouver has seen a neo-liberal turn in governance at the provincial and local levels (Mitchell, 2004), the preservation of a quasi-Keynesian welfare state in Quebec despite neo-liberalization 0 2,000 4,000 6,000 8,000 10,000 12,000 0 5 10 15 20 25 30 35 40 45 50 P op ul at io n de ns ity ( pe op le  p er  s qu ar e K M ) Distance from the central business district (KM) Montreal - 2006 Montreal - 1981 Vancouver - 2006 Vancouver - 1981 16 at the federal level means “Montréal policies can scarcely [be] cast in terms of a neo- liberal agenda” (Rose, 2004, p. 288). Commenting on Kaplan’s (1994) study of “Canada’s ambivalent spatial identities”, Wyly (2010) highlights intra-national differences in shaping urban outcomes in Quebec versus the rest of Canada due to the unique Francophone and Anglophone histories. Wyly also describes Mercer & England’s work (2000) that suggests demographic changes, economic restructuring toward services and neo-liberalization are increasing similarities among US and Canadian cities. As is explored in the next chapter, it would seem that these very factors are actually playing out in different ways in Vancouver and Montreal (see Filion et al., 2010), and that particularly the different degree of neo-liberalization and economic growth create distinct housing market outcomes.   The two cities’ different histories, and changing functions in the urban economic hierarchy are reflected in the changing urban structure. Montreal, located in southwestern Quebec (Figure 1.3), is the largest metropolitan region in the province and with a population of approximately 4 million the second largest in Canada, following Toronto. The city was founded as a French colony on the Saint Lawrence River, and grew to become Canada’s largest city and economic centre, competing even with New York for port and manufacturing activities by the late nineteenth century (Marsan, 1981). The city became a national and regional hub of economic activity for surrounding industrializing towns such as Sherbrooke and Trois-Rivières. As Germain & Rose (2000) explain, economic activity began to shift west as early as the Second World War, but the combination of deindustrialization beginning in the late 1970s and 17 the outmigration associated with the election of the indépendantiste Parti Québécois government in 1976 increased population loss and economic decline.  Figure 1.3 – Montreal and the regional geography  Notes: Map created by Chris Ley, University of British Columbia.  Source: Statistics Canada census and cartographic boundary files.  Growth in the high-tech sector and a vibrant cultural economy helped to reinvigorate the economy in recent years, but the decline and decentralization of Montreal’s manufacturing industry especially affected the central city (Shearmur & Rantisi, 2011). Montreal remains an important regional economic and cultural centre as Canada’s largest francophone city but the urban structure displays the archetypal “donut” structure of inner city population losses to the surrounding suburbs associated with de-18 industrialization seen in other large manufacturing centres across the US (Germain & Rose, 2000; Figure 1.2).  Vancouver, located in southwestern British Columbia (Figure 1.4), long served primarily as a gateway to resource exploration in the provincial interior but over time the resource industries played a reduced role in development of the city (Ley & Hutton, 1987; Wynn & Oke, 1992). As Vancouver expanded it gained new economy clusters and became a regional service, education and transportation centre that is now competitive with San Francisco and Seattle in terms of port activities and Pacific trade volumes (McGee, 2001; Hutton, 2008). Vancouver today has a population of 2.25 million making it the third largest metropolitan region in the country. Government functions are more heavily concentrated in Victoria, the provincial capital of British Columbia.  Like Montreal, Vancouver too is the largest metropolitan region in its province. Real estate investment has long characterized Vancouver but the sale of a large inner city property to a Hong Kong developer after the 1986 World’s Fair was a key moment in the globalization of Vancouver property markets (Olds, 2001). Individual ownership and private property have been persisting characteristics of British liberal thought that, as Mitchell (2004) argues, set the stage for present day neo-liberal property relations in Vancouver. Institutionalized planning as a mechanism to separate incompatible land uses serves important functions in facilitating private property markets in Vancouver (Blomley, 2004) but has also shaped urban development patterns by restricting outward expansion through an Agricultural Land Reserve and permitting growth in designated suburban centres that are connected by public transit.  19 Figure 1.4 – Vancouver and the regional geography  Notes: Map created by Chris Ley, University of British Columbia. Source: Statistics Canada census and cartographic boundary files.  The resulting urban form has been likened to “a series of cities in a sea of green” (Cameron, 2007). The city grew dramatically since the 1980s from a “village on the edge of the rainforest” into a gateway city (McGee, 2001) in what has been described as “instant urbanism” (Berelowitz, 2005), so that the urban form reflects contemporary urban planning ideals that arrange the city into nodes and corridors to attain sustainability goals (Hall, 1996; Filion & Bunting, 2010). In part, this is reflected, for instance, in Vancouver’s high central city densities that rebound in the suburbs (Figure 1.2).  The role for land use planning, which has become one of Vancouver’s defining characteristics (Harcourt & Cameron, 2007), was established from the early days, with evident influence on urban development through the successive plans prepared by the 20 then members of the British Town Planning Institute and the earlier British land surveyors who subdivided land for the purpose of selling private property in anticipation of resource booms and the arrival of the transcontinental railway (Hayes, 2007; Berelowitz, 2005; Hodge & Gordon, 2008).  In contrast, land development in Quebec by the French colonists was based on “egalitarian principles” through the use of the “côte”—rectangular plots of land that all adjoin a common resource such as a river (Germain & Rose, 2000, citing Marsan, 1981). The “côte” had a “stabilizing influence on the rural landscape” that later came to be a “powerful factor of uniformity in urban development” and lent “the old populous districts of Montreal…their strong gregarious and egalitarian character” (Marsan, 1981, p. 42-43). Urban development in Montreal is arguably still shaped in recent periods by social policy with relatively greater emphasis on notions of egalitarianism (Seguin & Germain, 2000). Montreal’s French history as compared to the British background, which is more predominant in the rest of Canada, including Vancouver, is perhaps one of the most obvious distinguishing factors between the two metropolitan areas. However, as Marsan (1981) notes, to view Montreal purely through its position within francophone Canada would overlook the important role the city “has always played in the history and development of Quebec and of Canada” (p. xxvi). The comparison of French and British histories to explain current trends also ignores the unique yet shared development experiences of Montreal and Vancouver since their founding as European colonies2 as well as the diversity of past and current immigrant 2 The author owes thanks for this point to fellow students in David Ley’s graduate seminar (Department of Geography, University of British Columbia, Fall, 2006) who discussed how an exaggerated focus on Canada’s British history overlooks the nation’s development since colonization.  21 populations (Germain & Rose, 2000). Yet the “hegemonic tendencies” of dominant ideologies, for instance as they relate to the liberal versus social egalitarian notions of property arising from the Anglophone versus Francophone history of British Columbia and Quebec, can leave long-lasting impressions on the urban landscape (Mitchell, 2004; Kaplan, 1994). These differences are valuable to keep in mind in this comparative study as long as they are understood in relative terms and in reference to the historic context (see Boudreau et al., 2007).  The different development trajectories are, for instance, apparent in the contemporary housing stock (Figure 1.5). Less favourable economic conditions in Montreal have resulted in lower prices and rents than in Vancouver. High rental vacancy rates peaking in the 1990s are partly attributable to low household formation rates among young adults, who represent a large share of renters and saw their incomes decline (CMHC, 2004; Heisz, 2006). Due to low incomes the cost of owning, as measured by the ratio of an owner’s major payments (OMP) to household income, was actually higher in Montreal than in Vancouver until the 1990s3. Urban densification strategies and rising prices are contributing factors in the decline of the share of single- family dwellings in Vancouver (Tomalty, 1997). The shift away from single-family dwellings in Vancouver has not resulted in a decline in ownership levels, which still remain much higher than in Montreal.    3 The ratio of housing costs to income is often used to compare general affordability levels over time or between places. The figures should not be used to make inference of the relative affordability of renting versus owning because the estimate does not take into account the income differences between tenure. 22 Figure 1.5 – The changing housing market characteristics                     Notes: The vacancy rates in 1971-1991 are for apartment structures of six units and over and in 1992- 2006 for row and apartment structures of three units and over. Housing price, rent (one-bedroom only) and vacancy rate are annual data for privately initiated structures only. The housing price and rent data prior to 1990 and 1992 respectively are linear approximations using decennial census data. The housing cost to income ratios are only calculated for census years. OMP: Owner’s major payments. All data are for the census metropolitan areas. Nominal value ($) shown for housing prices and rents.  Source: Calculated using the Statistics Canada census and data compiled by CMHC (2007a) from their Rental Market Survey and the Canadian Real Estate Association’s MLS®.  Choko & Harris (1990) suggest several factors behind Montreal’s high rental rates. The Montreal economy was historically dominated by the Anglophone business elite, they explain, leaving the property market as one of the few investment opportunities for Francophones. This resulted in an over-supply of rental housing at low 0 100 200 300 400 500 600 1979  1982  1985  1988  1991  1994  1997  2000  2003  2006  M L S ( N om in al  $1,000)  Average Housing Price Vancouver Montreal 0 200 400 600 800 1000 1979  1982  1985  1988  1991  1994  1997  2000  2003  2006  R en t ( N om in al  $)  Average Rent Vancouver Montreal 0 2 4 6 8 10 1979  1982  1985  1988  1991  1994  1997  2000  2003  2006  R en ta l V ac an cy R at e (%)  Vacancy Rate Vancouver Montreal 0.120 0.140 0.160 0.180 0.200 0.220 0.240 0.260 2006 2001 1991 1981 H ou si ng C os t /  I nc om e  Housing Cost to Income Ratios  Vancouver OMP / Income Montreal OMP / Income Vancouver Rent / Income Montreal Rent / Income 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 2006 2001 1991 1981 P ro po rt io n  Si ng le -fam ily D w el lin gs  Single-Family Dwellings Vancouver Montreal 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 2006 2001 1991 1981 Pro por ti on  Owne r Occupie d Home Ownership Vancouver Montreal 23 cost. The inertia of this stock, in combination with low income levels, little wealth generating potential for renters and higher cost of self-building (historically an important avenue to ownership), helped to create and reinforce a “local culture of property” that kept ownership levels low (Choko & Harris, 1990). In recent years, ownership levels have increased but unlike in Vancouver where condominium apartments have played a large role, suburbanization has been an important factor behind the growing single-family dwelling ownership in Montreal (Germain & Rose, 2000; Harris, 2011). The analysis later returns to a more detailed discussion of these trends.   1.3 Methodology: A Research Narrative of Place  The following section outlines some of the broad parameters of the thesis methodology whereas specific technical details are left to individual chapters. The research takes guidance from urban and housing economics and the branches of urban geography and planning analyzing the changing social space, economic activity and housing markets in cities. The primary empirical evidence is quantitative data from the Statistics Canada censuses and other surveys (e.g., Survey of Household Spending) conducted between the early 1980s and mid-2000s that are interpreted in light of the academic literature, government reports and policy documents detailing urban societal change. Also informing the discussion are conversations with key informants, newspaper coverage and visual observations of the urban housing stock. The highly empirical focus of the research is motivated by Gottmann’s (1961) use of maps and 24 statistics to point to broad processes of urban change (Pawson, 2008). Also inspiring in this regard is Harris’ (2008) methodological suggestion to:  “steep oneself in a complex body of data…and out of that steeping to explore the relevant theoretical literature, and then, in whatever permutations and combinations put the two together.” (p. 413)  The various information sources are combined into what can be referred to as a ‘research narrative of place’4. The idea of referring to the analysis as a ‘narrative’ is inspired by those who have used the concept of “story-telling” to conceptualize the research and public policy-making process (e.g., Usher, 1997; Throgmorton, 2003). The use of the term ‘narrative’ to conceptualize the research method is also motivated by contemporary epistemological critiques of quantitative research. The work of critical geographers brought to mind potential contradictions in this thesis in regards to the assertions that context matters, yet an explicit aim to produce empirical generalization from data that necessarily abstract reality (Schwanen & Kwan, 2009)—abstraction is commonly critiqued for its reductionism and claim to superiority and neutrality which works, it is argued by critics, to preserve the status quo and produce “non-local” “expert knowledge” (Philo, Mitchell & More, 1998; Gieryn, 2008, p. 799). The thesis is not able to resolve any such contradictions at an epistemological level, nor does it claim to contribute to the epistemological literature. The intent here is to merely point to the research that argues that the critiques of positivism are often incorrectly extended to all 4 In some sense, this research uses census data and other sources to produce what Macfarlane (2009) suggests editors refer to as a “write-around”, where “…a portrait of the subject”, in this case the two cities and the young adult population, “is constructed from sources—interviews with colleagues and friends, newspaper stories, drafts of speeches, and personal observations—that excludes the subject himself” (p. 13). 25 quantitative research in that claims to neutrality are a component of a particular politics not a methodology (Sheppard, 2001; Barnes, 2009; Schwanen & Kwan, 2009). Also, viewing the research as a narrative works, in some sense, to destabilize the notion that the approach taken here inherently elevates one source of information, quantitative or qualitative, as superior (Philip, 1998). Each source serves a specific purpose to advance the narrative, this narrative being, by dictionary definition, one, not the, interpretation of the world. The narrative component highlights the “temporal element” and normative decisions made in the presentation of research findings through language that as Solnit (2001) declares “cannot be perceived all at once” but “unfolds in time” (p. 268). The research narrative “differs from a fictional account because it embraces [the] data but it remains a story because it must have a beginning, end, and middle” (Yin, 2009, p. 130), and requires normative interpretation as “data do not speak” (Carter, 2009, p. 475).  Any contradiction of using quantitative data to make an argument about the importance of context is thus reduced in the way numbers are used in this thesis, “not to abstract but to measure” the general condition of a particular population and context (Sorkin, 2009, p. 28). Inspiration is drawn from what has been called a “methodologically inspired and infused quantification”, which recognizes that “counting matters” (Carter, 2009, p. 466 & 475) and provides a “technique for organizing one particular type of information” (Guy & Henneberry, 2000, p. 2011) necessary to draw conclusions about societal trends and to reveal the social injustices and environmental implications they may produce (Ellis, 2009; Wyly, 2009).  It should also be noted that the case of young adults in Montreal and Vancouver are used to generalize to “theoretical propositions and not to populations or the 26 universes” (Yin, 2009, p. 15)5. Thus, following Yin, the characteristics of young adults in the two cities are seen to be revealing of the way specific contextual features shape location and housing decisions; they are not merely numerical examples of young adults ‘everywhere’. Methodologically, the research question necessitates an approach that compares housing decisions across different contexts while taking into account the differences in the household level characteristics. All statistical models in this thesis are therefore specifically developed to operationalize the combined insights from the neo- classical and structural theories of the factors shaping housing, location and commute patterns. In the neo-classical economic theories, the housing decisions are made at the level of the household. The decision depends on household size and composition that shape the space requirements, constrained by the households’ budget, and the factors that determine the ability to earn an income (Bourne, 1981; Goodman, 1986; Quigley & Raphael, 2004; Skaburskis & Moos, 2010). Following this theory, the housing decisions are analyzed using multivariate models to take into account the household level factors.  The multivariate models also contain two specific methodological features to permit comparison of how the influence of contextual factors, based on the structural theories, differs between time periods and metropolitan areas. First, the models are generally constructed separately for each metropolitan area, permitting comparison of the magnitude and direction of regression coefficients in different housing contexts, as is done for instance in the literature on housing sub-markets (e.g., Cho, 1997). When the models include both metropolitan areas, dummy variables are used to detect 5 A second aim of using case studies also arises from the broader purpose of the discipline of geography to “show the world to be persistently diverse” but realizing that the local arises from “multiscaled relations” (Castree, 2005) so that geography necessarily amounts to more than simple “uniqueness of place” (Cox & Mair, 1989; Sharpe, 2009).  27 metropolitan-specific effects. Second, some versions of the models combine datasets from different years, allowing the use of a dummy variable to test for temporal changes. An explicit example of this approach is Vandersmissen et al.’s (2003) analysis of commuting patterns over time in Quebec but the inclusion of temporal lags is widely used in the housing literature.  Following structuration theory, the quantitative models thus permit individual- level variables to influence housing, location and commute patterns but these relationships necessarily are shaped by the specific structural context of the city, and in turn re-shape the context for future cohorts. The analysis measures empirically the nature of these effects in two metropolitan areas where restructuring occurred very differently. This contributes to better understanding of the nature of individual decisions by comparing the urban form as it emerged in Vancouver and Montreal since the early 1980s. Neo-classical economic theories provide a guide for understanding and operationalizing individual and household level variables in the statistical models throughout the thesis, while structural theories provide guidance for interpreting the variables as the outcomes of larger societal processes and contextual factors. From a geographic disciplinary perspective, it is certainly expected that ‘context matters’—the intent here is to provide an empirical analysis of how the context has changed and how individual-level relationships are shaped by different contextual conditions using statistical models.    28 1.3.1 Data sources and geography The Statistics Canada census data (1981, 2001, 2006) used for analysis come from three types of files—census tracts, individual public-use micro data files (PUMFS) and household PUMFS—that are available for academic use through Canadian university libraries6. The analysis is conducted for the Montreal and Vancouver census metropolitan areas, simply referred to as the CMAs, or the metropolitan areas for shorthand. CMA boundaries are determined by Statistics Canada according to commuter flows (Statistics Canada, 2001c). At the intra-urban scale, census tract files provide data as averages or counts for areas containing 2,500 to 8,000 people. The size of tracts thus varies depending on population density. Because census tract boundaries change over time to keep population counts reasonably consistent, and new tracts are added as the CMAs grow, the more recent tract data are matched to the 1981 boundaries to facilitate temporal comparisons. The 1981 and 2001 data were available from previous collaborative research (Skaburskis & Moos, 2008), and are expanded for this thesis to include the 2006 data, the most recent census at the time of writing.  The PUMFS are 3 percent population samples. They allow analysis of individuals and households in CMAs but do not reveal their intra-urban location. Variables are sufficiently similar across census years so that with some modifications the files can be merged to facilitate temporal analysis. The use of census data limits the analysis to a particular scale and to a pre-determined set of variables. The census is particularly useful, however, for this kind of analysis in that it is the only data source 6 The combination of observations from different census years into one dataset with “comparably coded variables” provides a useful means for examining societal change (Dillon, 1997, p. 381). The 1981, 2001 and 2006 PUMFS were combined into one database for the purposes of this thesis. This database then also formed the basis for analysis published in Moos & Skaburskis (2010).     29 that provides a consistent database with at least some information on the location, labour, housing and commuting characteristics of the population in all areas of the two CMAs. The 2006 data were available for individuals but the household data had not been released when the analysis was conducted, thus the 2001 data are used for household level analysis.   Interpretation of the data was facilitated by a vast literature ranging across diverse fields of study gathered using searches of scholarly databases, web searches and article bibliographies. Also informing the data analysis is a number of newspaper articles, observations from visiting the cities and conversations with key contacts. Current newspaper articles were obtained from casual reading while a historic search of the Vancouver Sun, Montreal Gazette and The Globe and Mail provided articles pertaining to young adults, housing and commute issues from the late 1970s to 20067. Efforts were made to visit several census tracts where the data showed particularly high concentrations of young adults or dramatic changes in the housing stock, and to make use of the public transit system in both metropolitan areas8. Sixteen key contacts with long-term knowledge of the two cities were asked to comment on the changes in the housing markets, policies and young adult location decisions9. The intent behind 7 Newspaper archives were accessed through the University of British Columbia Library databases. Several variations of the terms ‘commute’, ‘housing’ and ‘young adults’ were combined to search for articles relevant to Montreal and Vancouver. The search yielded several hundred results but only select articles were read in full. Articles were selected for reading when titles and abstracts suggested relevance to housing and commute trends in the two metropolitan areas. I owe thanks to Yolande Pottie-Sherman (2008) for inspiring me to think about media analysis in a more systematic fashion.    8 The researcher lived in Vancouver while conducting the research for this thesis (2006-2010) and visited Montreal on three occasions (April 2008; July 2009; March 2010).   9 The key contacts are a mix of current and former senior public servants (5), former politicians (2), senior real estate agents (5) and public-policy researchers/academics (4). The key contacts were explicitly 30 consulting newspapers, visiting the cities and talking to key contacts was to gain more insight than what could be attained from the numbers alone (Dandekar, 2003; Phillips, 2010). The information enhanced and guided the data analysis. However, findings from these other sources are not presented on their own but rather are cited where they assist in adding depth in interpretation.   Figure 1.6 – The Montreal census metropolitan area  Notes: Map created by Chris Ley, University of British Columbia.  Source: Statistics Canada census and geographic boundary files.  identified for their long-term knowledge of Vancouver and Montreal. Conversations ranged in length from 20 minutes to over 1 hour, and were semi-structured. The conversations took place during the fall and winter of 2008 and 2009. Approval was obtained for the research from the University of British Columbia Office of Research Ethics.  31 The metropolitan areas include many separate municipalities, and the analysis at times makes reference to the names of these municipalities when discussing intra-urban trends. The central municipalities are the City of Montreal and the City of Vancouver (Figures 1.6 & 1.7). The City of Montreal is located on the Island of Montreal along with several other municipalities that form the Communauté urbaine de Montréal. Bridges connect the Island to Ile Jesus to the north where the suburban municipality of Laval is located. Across the bridge on the southwest end of the Island of Montreal is Longueil that is also connected to the Island by the underground Metro. The Metro has also been extended to Laval in recent years. The suburban municipalities are commonly referred to as the north and south shores.  The City of Vancouver is juxtaposed with the City of Richmond to the south and the City of Burnaby to the east. A bridge connects downtown Vancouver, through Stanley Park, to the north shore municipalities of North and West Vancouver. The lands immediately to the west of the City of Vancouver do not have municipal status and are referred to as electoral area A. The lands are held in the form of an endowment by the University of British Columbia and are under directive of the Province. The SkyTrain, an elevated rapid transit system, extends from the City of Vancouver through Burnaby and New Westminster into Surrey. The suburban municipalities extend to the US border on the south and to the Abbotsford CMA to the east. The Vancouver CMA follows the boundary of Metro Vancouver, an upper-tier regional governing body.       32 Figure 1.7 – The Vancouver census metropolitan area  Notes: Map created by Chris Ley, University of British Columbia.  Source: Statistics Canada census and geographic boundary files.  1.4 Thesis Overview Following this introduction, the thesis has six remaining chapters. The second chapter, titled ‘The Changing Cities’, deals further with conceptual issues and the changes in the housing context. It elaborates on the use of a cohort approach in concert with structuration theory as a conceptual framework for the empirical analysis. The chapter also describes more specifically the societal restructuring associated with changes in governance, the economy, demographic transitions and planning policies aimed at increasing the sustainability of commuting and land use patterns. Because the changes are discussed in relation to the specific cases of Montreal and Vancouver, the chapter also helps describe empirically the changes in housing market context. 33 Highlighted are the changes in the labour and housing market associated with neo- liberalization and post-Fordist restructuring, inner city revitalization and the coordination of land use and transportation as a sustainability strategy. The emerging inequalities arising from restructuring are discussed. Chapter Three, ‘Global Restructuring and Changing Housing Demand’, delves in more detail into one specific aspect of housing market restructuring, analyzing the changes in the relationship between income and housing consumption due to immigration. The analysis helps build the economic relationships that link housing and household characteristics. Chapter Three also provides an analysis of the spatial dimensions of changes in the housing stock in the two cities. It helps reveal important differences in context within which young adults make decisions. Chapter Four, ‘The Changing Metropolitan Economies and the Young Adult Labour Force’, describes the changing labour and household characteristics of the young adult labour force. The main purpose is to detail the changes in the earnings of young adults, and the inequality implications. The analysis identifies the determinants of income and how these have changed over time in relative terms; and it links the findings to the theme of inter-generational equity as a component of the sustainability debate. The fifth and sixth chapters include the empirical analysis of the young adult residential location, housing and commute decisions. Chapter Five, ‘The Changing Residential Ecology’, analyses young adults’ changing residential geographies in relation to other characteristics of the urban form. It considers how young adults locate in relation to the social space, housing characteristics, commute mode, walkability and distance to transit and the central business district. These variables are used as measures 34 of whether over time, and compared between CMAs, young adults are locating and commuting more or less sustainably, and whether these patterns are distributed equitably.  Chapter Six, ‘The Housing and Commute Decisions’, analyses young adults at the household scale, considering the changes in housing type, tenure and expenditure over time. In Chapters Five and Six, multivariate models are used to investigate the changes in location and housing, holding household level factors constant, and their implications for commute patterns. The concluding chapter brings together the key empirical findings but its primary purpose is to serve as a platform to relate the empirical findings to the broader conceptual themes introduced in this and subsequent chapters. Empirical conclusions deal with the sustainability and social equity implications of the changes in the societal context and the nature of young adult decisions in the two metropolitan areas. Conceptually, the analysis permits conclusions regarding the ways societal transformations are reflected in urban space (Mills, 1989).  35 Chapter Two: The Changing Cities Part of the question addressed by this thesis is how the two metropolitan areas have changed over the study period. There have been numerous accounts of the formative changes in Canadian society on the city structure, governance, economy and social space (e.g., Ley, 1996; Bourne & Rose, 2001; Hutton, 2004; Bunting et al., 2010; Barnes et al., 2011). Common themes highlighted by these accounts are the transition toward a post-Fordist socio-economic structure and post-industrial central city space, increasing globalization and neo-liberalization, demographic transitions associated with the aging of the population and increasing diversity of households, and growing concern over environmental issues materializing in sustainability policies. Filion & Bunting (2006) suggest that the outcomes of these contemporary societal changes in cities are greater “unevenness”, “uncertainty” and concern over “sustainability”. They suggest that the themes are not new but that they           “…raise more concern at present than in the past—especially in regard to environmental sustainability. They have also become more manifest as differences between people and places have been accentuated over the last decades, and as globalization has eroded local control over the economy and thus raised uncertainty levels.” (p. 1).  Added to their list may be the greater ‘connectivity’ of people and places arising from globalization and technological change (Castells, 2010), and the heightened emphasis on ‘flexibility’ and ‘efficiency’ as desirable attributes of social and economic systems (Peck, Theodore & Ward, 2005; Siemiatycki, 2005). The implications of these changes for the housing market context within which young adults make decisions are wide- ranging (Beer et al., 2011). Unevenness, for instance, has materialized in more 36 segmented housing markets and increasingly segregated social space (Walks, 2010; Rose & Villeneuve, 2006). Uncertainty, in part a product of increasing flexibility in labour markets, means that households change their expectations about the permanency of work locations and their earnings. Some have argued these changes would increase the attractiveness of rental markets and central locations in that there are fewer transaction costs to moving when renting, and that more employment opportunities can be reached from central residential locations, particularly for dual-earner households (Costa & Kahn, 2000; Skaburskis & Moos, 2010).  At the same time, housing, which serves the dual function of investment and shelter (Bourne, 1981; Pozdena, 1988), has arguably become a more important component of household investment in a context of neo-liberalization that emphasizes the role of individual rights and privileges, which in the western context remain tied to property ownership (Blomley, 2004; Ronald, 2008). Flexibility and uncertainty have also translated into more part-time work and outsourcing, to which some households may respond by becoming self-employed or by including office space in their residential dwellings (Moos & Skaburskis, 2008). Growing entrepreneurialism is ostensibly required both on the part of households and governments in a context of growing global competition and less certainty about the future (Leitner, 1990; Larner, 2000). The entrepreneurialism and efficiency associated with neo-liberalization spur renewed interest in self-sufficiency and localism, which are also components of some emerging environmental movements (Kohler & Wissen, 2003; Geddes, 2005). But in terms of local government, the environmental concerns have largely materialized in urban planning as a set of policies aimed at coordinating land use and transportation 37 patterns to reduce vehicle use and emissions, and by limiting the spread of cities through growth management policies (Hall, 1996; Gunder, 2006; Moos & Skaburskis, 2008). The changes have opposite effects in that increasing property ownership, both in terms of physical size and investment amount, seemingly help households attain a degree of economic, and environmental, self-sufficiency in a context of uncertainty and flexibility. Yet at the same time this same uncertainty and flexibility, and also growing emphasis on sustainability, would make smaller, denser, and therefore less resource intensive housing options potentially more desirable. Households entering the market would balance these concerns in combination with their demands for housing space depending on life-cycle stage and household composition, which are also becoming more uneven (Rose & Villeneuve, 2006). Beer (2006) categorizes the changes in housing markets facing the different young cohorts over the past 30 years by changes in “sequence” (e.g., delay in child bearing, divorce), “meaning” (e.g., investment versus shelter, retirement plan), differentiated “choice/constraints/risks” (e.g., housing supply, labour market segmentation, inheritance) and “housing policy” “re-orientation” toward the market (also see Beer et al., 2011).  The label of heightened complexity might be an apt description of these changes that Dear & Flusty (2002) have argued produce a “contingent mosaic of variegated monocultures” in the “post-modern” urban field that is no longer focused on one specific central business district (p. 227). In contrast others have shown the persistence of the generalizable forms in social space and housing characteristics that continue to display the spatial patterns of the city organized into nodes, sectors and concentric circles by demographic variables as theorized by the Chicago School’s social ecologists 38 (Shearmur & Charron, 2004; Hackworth, 2005). Underlying this debate are disagreements not just on the changing urban socio-spatial structure but also as to whether the drivers of the changes arise from individual, such as life-cycle stage or household characteristics, versus structural factors, such the growing entrepreneurialism of states that results in their investing in central city amenities (Larner, 2000; Shearmur & Charron, 2004; Skaburskis & Moos, 2008). As is often done, these views can be considered as complementary explanations of changing residential geographies and housing markets (Bourne, 1981; Ley, 1988; Kauko, 2001) and structuration theory would see the two, individual and structural, as reciprocally reinforcing one another.  The issues raised above are further explored throughout this thesis. This chapter deals specifically with how the societal changes have altered government involvement in housing and the organization of production, and their implications in the structure of cities, particularly through inner city revitalization and the coordination of land use and transport patterns, in the specific case of Montreal and Vancouver. The overarching aim is to paint a broad picture of the two metropolitan areas’ housing market contexts, and their changes since the early 1980s to set the stage for the analysis of young adult location, housing and commuting decisions in the chapters that follow. The chapter begins by defining the young adult cohort in more detail and in connecting cohort changes to structuration theory demonstrating how young adults at a given point in time are operating under different formative contexts, which they themselves alter as they make new kinds of decisions and eventually replace older cohorts. After discussing the changes occurring in Montreal and Vancouver, the chapter analyses quantitatively differences in the housing expenditure patterns using data from the 1982 Family 39 Expenditure Survey (FAMEX) and the 2005 Survey of Household Spending (SHS). Linear regression models are constructed using geography, housing and household characteristics as independent variables. The variables identifying the geography and survey year help to compare young adults’ spending patterns in the different temporal and metropolitan contexts. In line with the objectives of the thesis, this describes how households make decisions about housing under different conditions.  2.1 Defining the Young Adult Cohorts Demographer Rindfuss (1991) argues that there is an “inherent ambiguity” in determining an appropriate age range to define “anything so nebulous” as young adults (p. 494). As Calvert (2010) discusses in her literature review, sociologists commonly refer to “transition markers” such as completing education, moving out of the parental home, obtaining employment, marriage and childbirth to signify transition from youth to adulthood, with young adults being in the early stages of having attained these markers. However, the decline of traditional norms surrounding family formation and marriage, growing educational attainment and young adults’ own changing and varying perception of adulthood, question any universality the markers may have held; although the “destandardisation argument” does remain “subject to debate” as some find a delay in attainment of markers as opposed to greater diversity in life courses (see Calvert, 2010, p. 9; Shanahan, 2000; Elchardus & Smits, 2006). The young adult 40 category is defined here as those from 25 to 34 years of age10. The important distinguishing factor is that young adults are more likely to be influenced by the current housing context than the older cohorts. The use of the 25 to 34 age group in the 1981, 2001 and 2006 census data captures individuals that would be considered part of the baby boomers, generation x and generation y “when they were at the start of their housing careers” (Skaburskis, 2002, p. 378). Those younger than 25 are more likely to be full-time students whose location and housing decisions are tied to the parental home or an educational institution, whereas those over the age of 34 would rarely be considered young adults.11 In fact, some accounts use 30 as the age of attaining adulthood (Rindfuss, 1991) but the later cut-off is useful since several transition markers today evidently extend into the mid-30s (Clark, 2007). The exact age cut-offs are also constrained by the use of census data that groups the population into five- or ten-year cohorts in some of the data files.  The analysis in this thesis uses three different categories of young adults. The first is the total young adult population (or cohort) determined solely by age using the 1981 and 2006 individual PUMFS and the census tract data. The young adult cohorts are similar in size in Montreal and Vancouver, constituting about 18 and 14 percent of the total population in 1981 and 2006 respectively. The second category of young adults restricts the population to those in the labour force, which includes those employed or 10 The comparison of fixed age cohorts over time in a context of an aging population means that young adults are on average ‘older’ in the 2001 and 2006 data than in 1981, which may result those in the more recent census data being more established in terms of the transition markers (see Boyd & Norris, 1999).  11 According to the 2006 census, almost 60 percent of those 20 to 24 years old are attending school in Montreal and Vancouver, compared to only about 24 percent and 23 percent of those 25 to 34 years old attending school in Montreal and Vancouver respectively.  41 unemployed but excludes those not actively looking for work. When using the 1981 and 2001 household PUMFS a third categorization of young adults is used based on the age of the primary household maintainer. The category is also restricted to those in the labour force. The category used in the household file is not directly comparable to the individual file in that it excludes young adults who may be residing with a primary maintainer outside the young adult age range. The size of the young adult labour force is similar in Montreal and Vancouver at about 28 and 21 percent of the total labour force in 1981 and 2006 respectively. Well-known is the declining size of the young adult cohort due to the aging of the population (Figure 2.1; Foot & Stoffman, 1996). The historically larger size of the young adult cohort is attributable to the baby boom— occurring between the late 1940s and mid 1960s—which some observers explain as an outcome of increasing fertility rates in a context of rising living standards and enduring patriarchal family values in the post-war period (Bean, 1983; Roberts et al., 2005).  The definition in the household PUMFS excludes the growing share of adult children living in their parents’ home, either never having moved away or having returned after leaving initially (Boyd & Norris, 1999). A Statistics Canada study found that the percentage of young adults still living at home doubled from 1981 to 2001 from 12 to 24 percent for those 25 to 29 years of age and from 5 to 11 percent for those 30 to 34 (Beaupré et al., 2006). The study notes that the trends are at least in part attributable to the recession of the 1990s and declining overall economic prospects. The trend is of course highly relevant in terms of the questions posed in this thesis regarding the implications of context on housing decisions but the datasets used do not permit analysis of young adults living at home separately.  42 Figure 2.1 – Age distribution in the Montreal and Vancouver CMAs   Notes: Percentage of population in each age cohort. Montreal (top) and Vancouver (bottom) census metropolitan areas. Source: Calculated using the Statistics Canada PUMFS (1981c; 2006b).  6 4 2 0 2 4 6 0-4 10-14 20-24 30-34 40-44 50-54 60-64 70-74 80-84 90-94 100+ Male1981 Male2006 Female1981 Female2006 MALE FEMALE 6 4 2 0 2 4 6 0-4 10-14 20-24 30-34 40-44 50-54 60-64 70-74 80-84 90-94 100+ Male1981 Male2006 Female1981 Female2006 MALE FEMALE 43 However, knowledge of the parents’ characteristics with adult children living at home helps reveal how inclusion of this group when using the individual PUMFS and census tract data impacts the results. Turcotte (2006) finds that single-family homeownership, immigrant status, Asian and South American place of birth and residing in large urban centres increase the probability of having an adult child living at home. Thus, inclusion of young adults living at home in this thesis likely overestimates young adults’ own housing consumption and their tendency to reside in single-family dwellings, particularly among immigrants.    2.2 Young Adult Cohorts in Specific Times and Locations  The assumption underlying the cohort approach is that each generation, or birth cohort12, is distinguished from those preceding, having confronted “similar opportunities and constraints”, which leads to distinguishable patterns of behaviours, norms, values and beliefs (Myers, 1999; Carr, 2004, p. 453; Twenge, 2006). Norman Ryder’s (1965) work is often credited for the development of the cohort approach13. Ryder argues:  “in an epoch of change, each person is dominated by his birth date. He derives his philosophy from his historical world, the subculture of his cohort” (p. 855).   12 The terms ‘generation’ and ‘cohort’ are sometimes used interchangeably. Following Riley (1987), the term ‘cohort’ is used here to refer to people born in a similar time period, and ‘generation’ to refer to kinship instead.   13 French demographer Jaque Vallin referred to Norman Ryder as the “father of a method that no serious demographic textbook can afford to overlook” (cited in Quinones, 2010).    44 Constable (1996) notes that cohort differences need to be understood as context specific because they are contingent upon the pace of social change itself—and it is therefore perhaps not too surprising that in the recent times of “high modernity” (Giddens, 1984), characterized by rapid changes in technology and social and economic organization, the concept of generational differences is seemingly amplified in popular consciousness (Furlong & Cartmel, 2007).14 In the media, generational differences are therefore often essentialized and believed to be omnipresent. This apparent belief that “we resemble our times more than we resemble our parents”15, as an Arab proverb proclaims, arguably has the effect of excluding historically contingent variables, such as class or ethnicity, from our understanding of societal outcomes. Furlong & Cartmel (2007) draw on the works of Giddens to argue that the importance of class, ethnicity or gender has not disappeared but certainly shifted due to greater “individualization” and “uncertainty” of flexible labour markets. Young adults, according to Furlong & Cartmel, thus not only encounter increasingly less favourable conditions but also a changed context where responsibility for ‘success’ is increasingly defined in terms of individual effort. In other words:   14 A title search for books using the phrases “generational change” or “generation x” or “generation y” from 1990 to 2010 yields 55 relevant book titles using the University of British Columbia Library Catalogue, including several translated versions of Douglas Coupland’s (1991) popular book “Generation X: Tales for an Accelerated Culture” that arguably foreshadowed growing interest in generational differences.   15 There appear to be several different versions of this proverb in use. The quote is adapted here for gender-neutrality. Twenge (2006) uses the proverb in her study of the changing attitudes, norms and behaviours of young adults in different cohorts to allude to the importance of generational differences. I came across Twenge’s work while writing this thesis thanks to a CBC One radio program, “Ideas in the Afternoon” hosted by Mary O’Connell on January 3, 2011.  45 “The intersection of life course stage with key social structures at particular historical moments matters, and matters greatly” (McDaniel, 2004, p. 32-33).  The use of a cohort approach is useful in this thesis because it provides insight on how individuals’ decisions reflect the particulars of different time periods; and because it serves as a conceptual framework that connects individuals to structures through the process of aging of successive cohorts, each growing up under different kinds of conditions (Beer, 2006). Riley’s (1987) description perhaps best elucidates the basis of the cohort approach. She explains two “dynamisms” that make aging a process of social change: First, since everyone ages there are necessarily “successive cohorts” aging together. Second, at any given time societies, and their institutions, are composed of numerous “age strata”. Riley describes how these two processes can be visualized as a series of diagonal (cohorts) and horizontal (strata) lines in a two dimensional space of time (x-axis) and age (y-axis) (Figure 2.2). The implication is that as time passes, society and its institutions are altered by the “dynamisms of aging” (p. 4):  “[T]he people in a particular age strata are no longer the same people: they have been replaced by younger entrants from more recent cohorts, with more recent life experiences.” (Riley, 1987, p. 4)   Therefore, the process of aging, while not explicitly treated by Giddens, actually provides a mechanism of how individuals serve to “constitute and reconstitute” social institutions through the “duality” of structuration (Giddens, 1984, p. 25). Giddens does not directly theorize how institutions change over time despite acknowledging the existence of time and life-cycles (Johnston, 2006). Some interpret the occurrence of change in structuration theory through “thresholds”, which means behaviours that 46 begin as “deviant” within existing structures eventually tilt the scale of what is socially acceptable, thus forming new structures for subsequent behaviours (Mills, 2004). Adding the element of aging highlights the way this “threshold” may actually be reached as younger cohorts replace older ones in the age strata and selectively transfer emerging behaviours into existing settings, such as the workplace, housing markets or political institution, whereby structures are then remade.  Figure 2.2 – Conceptualization of birth cohorts and age strata  Notes: The shaded areas depict the young adult cohorts, 25 to 34 years of age in the 1981 and 2006 censuses. Select events in the urban histories of Montreal and Vancouver are shown to illustrate the different contexts within which young adults make location and housing decisions in different time periods and cities.  Source: Based on Riley (1987).   The portraying of society as an age stratum also links the past and the future to the present in that at any given time, society is composed of a series of cohorts each 0 10 20 30 40 50 60 70 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 A ge  Time Cohor ts A ge  S tr at a Montreal  Metro & Expo Montreal  Olympics Montreal  MCM  Elected Vancouver  SkyTrain & Expo Vancouver Olympics TEAM Elected in Vancouver Business Migrants Quebec Referndum Quebec Referndum Vancouver Housing Bubble 47 with different histories, which then as a whole serve as a structure for future cohorts. Pred (1984) argues that these connections are absent from structuration theory but can be incorporated by drawing on time-space geography through an understanding of the “paths” of “individual biographies” (p. 281). The cohort approach can also introduce these notions of how individual histories work to build structures but it is more limited than Pred’s approach of using “individual biographies” in that cohorts are an aggregation of individuals’ histories. It is, for instance, difficult to account for the varied histories of immigrants and locally born young adults under the same conceptual umbrella of a shared cohort experience. The same applies for young adults with varied genders or class and ethnic backgrounds. The cohort approach is therefore not unproblematic as it can have the effect of ‘flattening’ place and temporally specific conditions. The issues associated with the aspatial treatment of cohorts are arguably amplified by the growth in global migration patterns.  However, the fact that migration changes the cohort composition does not imply that aging is not a driver of social change; it merely acknowledges that it is not the only one. Furthermore, the idea of the cohort approach as applied here is not to suggest that all young adults necessarily have similar histories but rather that at any given time and place there are broad social structures and conditions within which young adults make decisions; and at the same time recognize that there are intra-cohort differences in the ability to make these decisions due to factors such as class, gender, ethnicity that require exploration (Furlong & Cartmel, 2007). Place and temporal specific conditions can be introduced more explicitly into cohort analysis by considering how cohort characteristics originate from local conditions, which also are a product of global-local 48 interactions (Swyngedouw, 2004; Chapter Three). Figure 2.2 overlays select events in the history of urban development in Vancouver and Montreal to illustrate the way aging occurs in a geographic-specific context. Young adults in Montreal and Vancouver would of course also experience similar kinds of broad changes occurring society-wide. The question explored primarily in the remainder of this, and also the following chapters deals with both these local and global changes that reshaped the conditions within which young adults make housing decisions, and how these differ between the two metropolitan areas. Applied is what Myers (1999) refers to as a “quasi-panel approach” in using “cohort data from repeated cross-sections” (p. 478).  2.3 Reduced Government Involvement in Housing An important change along the timeline of the two young adult cohorts is the embrace of a neo-liberal political ideology in the late 1970s and early 1980s. Neo- liberalism, characterized by the three “major tenets” of laissez-faire government, free markets and the liberties of the individual, is often understood as a set of policies that differ from those of the preceding Keynesian welfare state based on interventionist, redistributive and egalitarian principles aimed to manage the shortfalls of markets (Larner, 2000; Hackworth, 2007, p. 4). The “roll-out” of neo-liberalism became widespread in North American in the 1990s (Peck, 2001), and actually required a substantial degree of state involvement to “impose market rule on all aspects of social life” (Brenner & Theodore, 2002, p. 352). Neo-liberal policies also became largely “depoliticized” (Peck & Tickell, 2002) and the rule of the market “naturalized” (Peck, 2001; Jessop, 2002). In Canada, neo-liberalization had a particularly profound impact 49 on housing through the changes in state involvement in housing provision that lowered the supply of affordable housing and led to a decline in welfare state functions, reducing income support (Bunting et al., 2004).  Canadian housing policy has been described as having moved from goals of “social development” in the 1960s and 70s to “financial restraint” into the early 1980s, “disentanglement” associated with a reduced government presence in the early 1990s and finally almost complete “disengagement and privatization” starting in 1994 (Carroll & Jones, 2000, p. 279). The federal government began its involvement in “building a non-market social housing sector as part of a broader social safety net” in 1964 (Hulchanski, 2004, p. 179). Some observers characterized housing in Canada during the 1960s and 70s as relatively equitable and affordable (Sewell, 1994). Federal involvement began to decline in the 1980s but still funded new supply, such as cooperative housing (Cole, 2008). In the mid-1990s, the federal government made an “unprecedented… decision to abandon a federal role in social housing” altogether (Shapcott, 2004, p. 198). Also restructured was the role of the Canada Mortgage and Housing Corporation (CMHC) by emphasizing its function as a mortgage insurance provider, eliminating its former role in rental housing provision (Harris, 2006; Dalton, 2009). The cuts in funding at the federal level were mirrored by almost all provinces and led to the downloading of responsibilities to municipalities, which often lacked resources to carry out programs, and therefore involvement of not-for-profit providers increased (Hulchanski & Shapcott, 2004; Germain & Rose, 2000).  In terms of provincial policy frameworks, British Columbia has undergone neo- liberalization (Mitchell, 2004), whereas Quebec has generally resisted the trend (Rose, 50 1999). However, in terms of housing both Provincial governments have been said to retain more “extensive housing programs”, particularly when it comes to support for co-operative housing (Evans, 2007, p. 6; Cole, 2008)16. British Columbia is certainly more neo-liberal than Quebec, but less so than most US states and definitely less than other Canadian provinces, notably Alberta and Ontario (McBride & McNutt, 2007; Stewart, 2009). The situation in British Columbia changed in 2001-2002 when a right- of-centre Liberal party replaced the former left-of-centre New Democrats, introducing “sweeping changes to the social programs”, and making “severe cuts” (Tang & Peters, 2006, p. 573). Locally, the City of Vancouver and the City of Montreal retained housing programs as cuts occurred but the delivery, particularly in Vancouver, shifted more toward provision of emergency shelter as opposed to building stock, whereas the latter goal remains more prominent in Montreal (Germain & Rose, 2000). One key contact noted that staff working on housing in Vancouver increased since the 1980s but that much more time is now devoted to acquiring funding as provincial and federal contributions declined since the 1970s (cf. Hackworth & Moriah, 2006). In contrast, Quebec signaled “definite intentions” to retain a government presence in housing markets and “acknowledged the importance of social housing policy” by compensating for the loss of federal funding (Vaillancourt et al., 2001, p. 13). The City of Montreal thus had greater abilities to pursue “housing and neighbourhood initiatives since the 1970s”, including building new stock, thus “earn[ing] it the justifiable reputation of the most interventionist municipality in Canada as regards the housing sphere” (Germain & Rose, 2000, p. 163, citing Gaudreau, 1992, and Hulchanski et al., 1990).  16 Conversation with key contact, January 30, 2009: Former municipal planner. 51 Young adults today are facing a housing market with less government intervention in Vancouver whereas Montreal appears to have retained many of the social welfare functions of the Fordist era. The effects of reduced government intervention in housing is believed to be especially influential for low-income earners who would see the supply of lower priced units decline as private developers have few incentives to build low-cost rental housing (Hulchanski & Shapcott, 2000). The effects of neo-liberalization are likely to be felt unevenly (Walks, 2010).  The long-term implication is that the federal government is today more “poorly placed to represent the interests of low income households who are experiencing new and enduring forms of disadvantage” (Dalton, 2009, p. 19).   2.3.1 Concurrent changes in the organization of production Another material difference for young adults today is the change in the organization of production arising from shifts toward post-Fordism that occurred in concert with neo-liberalization. The post-war years, commonly referred to as the Fordist period, were characterized by the Keynesian welfare policies and an economy based on mass production that offered high wages for the middle class and stability in employment arising from unionization (Filion & Bunting, 2006). The shift toward post- Fordism beginning in the 1970s brought with it increasing internationalization and greater flexibility in the production process, and reduced protection for workers (Harvey, 1989). The “social context” began to change in the 1970s but Fordist policies remained for some years so that young adults in the 1980s came of age amidst this transition (Filion, 2001). On the one hand, the transition resulted in a process of de-52 industrialization of central city neighbourhoods and the decline of urban economies previously relying on manufacturing, bringing with it large job cuts due to rationalization and outsourcing in the production process (Lloyd, 2006). Growth, on the other hand, was beginning to occur in the service and technology sectors of what has come to be termed more broadly as the “new economy” that also includes the growing employment in the ‘creative’ industries and quaternary sector occupations (Hutton, 2008).   For workers, the implications of the changes in production have been increasing income inequality, as employment in the post-Fordist economy is more concentrated in managerial and professional occupations and lower level services (Sassen, 1991; Walks, 2001). There is also less certainty about long-term job security and earnings as labour has become more “flexible” in production and neo-liberalization reduced the protection offered by the welfare state, and unions, under Fordism (Peck et al., 2005). At the urban scale, one implication has been that economic development is more directly tied to the ability to compete globally to attract increasingly mobile capital (Leitner, 1990; Hall, 2006). Localities are using “entrepreneurial” strategies focused on economic competitiveness as opposed to the “managerial” policies during Fordism that “primarily focused on the local provision of services, facilities and benefits to urban populations” (Harvey, 1989, p. 3). As “cities have become strategically crucial arenas in which neo-liberal forms of creative destruction have been unfolding during the last two decades” (Brenner & Theodore, 2002, p. 369), the uneven effects of an “unforgiving” competition have become visible as resources are channeled to 53 individuals, neighbourhoods and cities “with economic potential, rather than those in need” (Peck & Tickell, 2002, p. 394). Montreal and Vancouver followed very different paths in the context of neo- liberal and post-Fordist restructuring (Bourne et al., 2011). Montreal was known as a “gateway to the continent” at the height of Fordism (Marsden, 1981), “labour-intensive” manufacturing industries “supported” by a strong welfare state, “poor rural population” and “Anglo money” (Naylor, 1990, p. 71). Over the course of the 1960s and 70s, the effects of de-industrialization materialized in the loss of manufacturing jobs and “decline in the Canadian urban system…in terms of the loss of head offices and financial institutions” (Slack, 1981, p. 27). As Germain & Rose (2000) note, Montreal’s decline thus already occurred before the referendums (in the early 1980s and mid 1990s) that raised fears about Quebec separating from the rest of Canada. But the political instability certainly contributed further to the unfavourable economic conditions. In contrast, Vancouver, although impacted by de-industrialization, had at the time of restructuring a much smaller population base and manufacturing sector, being focused more heavily on resource industries (McGee, 2001). Vancouver’s transition to post-Fordism thus could occur more readily with new growth whereas Montreal was faced with a large, unemployed, low-income population but an economic context that increasingly demanded education and fewer workers (Naylor, 1990; Hutton, 2008).   The embrace of “entrepreneurial” urban strategies to attract new capital is being pursued by municipalities by way of “urban spectacles” such as large scale sports events, festivals and cultural events (Hall & Hubbard, 1996; Gotham, 2002), and is exemplified in the Vancouver World’s Fair (EXPO) of 1986 that saw the cooperation 54 of governments, at all scales, and private sector interests in attracting investment to Vancouver (Mitchell, 2004; Olds, 2001). Interestingly, however, Montreal has an even more sensational history of hosting large events, particularly under the leadership of Mayor Drapeau known for his legacy of mega-projects, including Expo 67 and later the Olympics in the 1970s.   “Jean Drapeau side-stepped the other levels of government and handed down a series of “faits accomplis” in international relations. The government of Montreal conducted affairs with cities in other countries, and with other national governments, signing agreements to realize projects of benefit to Montreal and the commercial enterprise doing business here.” (Leonard & Leveillee, 1986, p. 56)  In one sense, Major Drapeau’s efforts go to show that “the role of the local state in actively promoting conditions favourable to capital accumulation” is likely not an exclusive condition of neo-liberalism (Hall & Hubbard, 1996, p. 155). However, the Vancouver EXPO is an example of how the state reworked the rules that governed international events so as to facilitate foreign investment in the provincial economy, particularly from Asia, which in practice became investment in local real estate (Olds, 1995; Mitchell, 2004). While Montreal’s “urban spectacles” in the 60s and 70s certainly aimed to increase the city’s international profile, the Vancouver EXPO was used to “facilitate the process of de-industrialization” (Mitchell, 2004; Hutton, 2004) and transition the urban economy so that real estate markets could become a “vehicle for capital accumulation” (Blomley, 2004; Newman & Wyly, 2006, p. 31).    The transition to post-Fordism alters the housing context for young adults in that urban growth, and housing costs, are increasingly tied to the metropolitan position in the urban hierarchy. The neo-liberal strategies of “making the city attractive” are also 55 used to compete for global investment in Montreal (Belanger, 2000; Darel, 2004; Ley, 1996). But the city’s difficulties in overcoming economic decline have meant that its housing market has not been exposed to the same inflationary impacts to the same extent as is the case in Vancouver (Walks & Maaranen, 2008). While in Montreal the state is playing in recent years a more active role in promoting central city housing developments, this has been linked more directly to the “interregional scale of municipal ‘competitiveness’” than the case of neo-liberalization responding to global pressures (Rose, 2009, p. 425). In contrast, foreign investment in real estate disconnected Vancouver’s housing markets from local conditions, in part through immigration that took on new dimensions due to the increase in wealthy business migrants explored in more detail in Chapter Three, but importantly also through neo- liberal interventions by the state (Mitchell, 2004).  The changes have created a condition where gains from real estate markets have become economic development tools and cities are being marketed for their amenities so that the urban policies in effect actively facilitate higher prices in concert with the creation of ‘desirable locations’ (Gurran, 2008; Kipfer & Keil, 2002; Kern, 2010). In Montreal, economic decline has kept affordability in check by preventing large housing market gains. Yet at the same time the city has also seen declines in employment opportunities so that there is a larger segment of low-income earners and those unable to find work than in Vancouver for whom affordability remains a concern. Housing cost inflation in Vancouver is, however, nothing new. It should be remembered that Vancouver experienced a short, sharp real estate bubble in the early 1980s that dramatically raised prices before it burst (Skaburskis, 1988). Interest rates are also an 56 important component of housing affordability in terms of ownership, and these rates were (briefly) as high as 20 percent in the 1980s. One key contact noted that housing policies in the early 1980s actively tried to address the high cost of borrowing, whereas today the policies have shifted to addressing the higher purchasing cost of housing (also see Beer, 2006).17 One implication for young adults is that while the high price of housing is more prohibitive today, the lower borrowing costs would serve as an incentive to enter the housing market more quickly to ‘lock in’ low interest rates. There were also two recessions between 1980 and the mid 2000s, which had the effect of lowering prices but at the same time also reducing employment prospects (Yelowitz, 2006). The ‘beneficiaries’ of recessions are those with stable employment as prices decline but young adults entering the market would likely face greater challenges (e.g., Brethour, 2008). Today’s young adults may face “twin pressures”, as Beer (2006) suggests, from the growth in the service sector that increases desirability of central locations and the heightened competition for housing from the larger cohort of baby boomers. In addition there are the changes in the labour force arising from post-Fordist restructuring, which have generally been associated with a polarizing income structure impacting households’ ability to afford housing unevenly (Sassen, 1991; Walks, 2001). The discussion returns to this issue in more detail in later chapters.   2.4 Inner City Revitalization  Post-Fordist restructuring contributed to a return of investment to the inner city that had been in decline since the 1960s due to de-industrialization. The economic 17 Conversation with key contact, December 9, 2008: Former politician. 57 restructuring occurred along with demographic trends and cultural changes that began to favour urban living (Ley, 1996). “The stability of post-war Fordism” was predicated on several cultural factors such as “an ethic of conformity, the primacy of the nuclear family, and the embrace of rationalized consumption” (Lloyd, 2006, p. 35). The changes in household structure and demography beginning with the decline of traditional norms regarding family formation and the equalization of women’s rights in the 1960s worked along with growing post-modern ideals and a rejection of the homogenous suburbs to facilitate renewed interest in inner city living (Mills, 1989; Ley, 1988). The changes contributed to the gentrification of the inner city by a “new middle class” of quaternary sector workers with implications for displacement of lower income populations (Ley, 1996; Atkinson, 2000). The changes reshaped the housing landscape both in terms of price gradients and the amenity and housing stock available in the central city so that young adults’ housing decisions today are made in a context where the central city real estate market has increased in importance and size due to gentrification while suburbanization is continuing, in part, driven by high central city housing costs (Skaburskis & Moos, 2008).   According to the stage model of gentrification, the process begins in specific neighbourhoods through an influx of artists and students attracted by the pragmatic and counter-cultural appeal of residing in industrial or declining low-income areas (Ley, 2003a). The presence of the artists works to attract more residents that often use their sweat-equity to upgrade the existing housing stock, thus raising prices and eventually displacing the original artists and low-income residents. In Vancouver and Montreal, gentrification in the 1970s and 1980s also commonly took the form of rental 58 conversions to condominium apartments, creating severe opposition from residents concerned about the loss of rental housing in the city (Ley, 1996; Harris, 2011). Both cities eventually placed a moratorium on conversions in select neighbourhoods but in Montreal a by-law remains in effect today that would allow the city to ban conversions to condominium apartments if the percentage of rental apartments fell below a certain level. However, a key informant noted that this is rarely enacted in practice. Nonetheless, gentrification is more widespread and rent increases associated with new developments are higher in Vancouver than in Montreal, which relates to the differences in the restructuring of housing policy:  “In Montreal, almost three-quarters of the [census] tracts experiencing some gentrification remained in an incomplete state by the end of the study period (2001), potentially due to the greater state of original disinvestment in inner-city Montreal…[p]rovincial and municipal policies of siting new social housing and rental apartments in such neighbourhoods…[t]he marginality of Montreal’s position within the global economy and its more limited occupational transformation.”  (Walks & Maaranen, 2008, p. 65).  As inner cities again became more desirable places, gentrification increasingly took the form of large-scale condominium apartment developments built on former industrial lands, something much more common in Vancouver than in Montreal. In fact, because condominiums have “become the principal form of residential property ownership in the inner city”, any “account of Vancouver’s transformation would be incomplete without considering the role of condominiums” (Harris, 2011, p. 24). As Kern (2010) argues, the condominium apartment has been described as “a spatial fix for the deindustrialized city” that permitted the expansion of the social and cultural ideals of suburban homeownership to the central city (p. 664). Kern also assigns the growth of 59 condominium apartments to their ease as an investment tool; and property markets are therefore increasingly for their investment as opposed to their use value.  “Urban housing markets have become important sites for neoliberalization, as witnessed by the elimination of rent controls, state withdrawal from housing provision, and the facilitation of speculative investment in inner-city sites.” (Blomley, 2004, p. 31).  Thus, as Blomley goes on to explain citing Smith (2002), inner city gentrification became entangled not only with the post-Fordist restructuring that provided the de- industrialized spaces and a quaternary sector workers willing to live in central areas. Gentrification is also now, notes Blomley, a facet of neo-liberalization since the state, largely by way of facilitating private investments, actively promoted investment in inner city real estate markets as an economic development strategy (also see Mitchell, 2004).    One key aspect of successfully marketing inner city real estate has become the amenity component (Kern, 2010). Perhaps now seeming ironic given today’s corporate dimension of inner city revitalization projects, it was indeed opposition to “pro- business” governments that resulted in the election of the Electors Action Movement (TEAM) in Vancouver’s municipal council in the early 1970s, and replaced “boosterism” with “the pursuit of the quality of life; the humane city would supersede the city efficient” (Ley, 1987, p. 45). The change in political ideology, as Ley explains, was connected to broader rejections of modernist principles that guided city planning in the 1950s. It resulted in the elimination of a proposal to build an expressway through Vancouver’s inner city and instead focused on increasing urban amenities; “park space was expanded, pedestrian areas opened up, new cultural and recreational amenities developed” (Ley, 1987 p. 45). The changes were exemplified in the transitions in False 60 Creek from, as one key contact put it, “an industrial slum to a recreational and residential space”.18 The support for an increasing inner city amenity component by TEAM resonated with the politically left-leaning, generally well-educated, quaternary sector workers, reinforcing gentrification pressures (Ley, 1996). De-industrialization in favour of a post-industrial urban landscape in Vancouver was also facilitated by the re- zoning of industrial lands by the City (Hutton, 2004).  In Montreal, similar kinds of political pressures began to build in the 1960s through opposition particularly to long-term mayor Drapeau who had pursued a series of modernist mega-projects, including the construction of central city highways that unlike in Vancouver were actually completed. Issues of social justice entered public debate in the 1960s in Montreal as elsewhere due to the lobbying of the social movements of the times (Leonard & Leveille, 1986). However, it took almost two decades until the political administration changed its course:  “Around the 1980s, the government of Drapeau-Lamarre modified its public discourse and its priorities instead of simply waiting for another extravaganza. It became concerned with the revitalization of commercial streets, renovations of housing in old neighbourhoods, development of new industries and, in general, a regeneration of the traditional city.” (Leonard & Leveillee, 1986, p. 55)  First as the opposition to Drapeau during the early 1980s, and then winning the municipal election in 1986 the Montreal Citizens’ Movement (MCM) also aimed to focus on the social aspects of the city, representing a similar constituency of left-leaning urbanites as TEAM did in Vancouver (Ley, 1987; Demchinsky, 1989). Drapeau began 18 Conversation with key contact, January 20, 2009: Former politician. 61 focusing on neighbourhood issues through housing programs intended to attract the middle-class back to the central city; and the MCM aimed to work for more citizen engagement and affordable housing that was beginning to decline because of Drapeau’s neighbourhood improvement programs that attracted gentrification (Leonard & Leveillee, 1986; Germain & Rose, 2000). Similarly, the embrace of “social diversity” by the left-leaning quaternary sector workers certainly also helped build support for social and affordable housing strategies being pursued by TEAM in Vancouver (Ley, 1987).   The combination of the changes in the political ideology and economic restructuring that increased the quaternary sector occupations in the inner city had two key implications for urban housing markets. First, the changes made it much easier to co-locate jobs and housing as new service industries were generally non-polluting and the density and proximity dimensions of the inner city became a desirable aspect both in the consumption and production of services in the new economy (Ley, 1996; Hutton, 2004). Second, the success of increasing urban amenities and the emphasis on homeownership contributes to rising housing costs and the loss of rental stock, which makes it more difficult to address ideals of social inclusivity, which the “new middle class” originally desired (Ley, 1987; Blomley, 2004; Kern, 2010; Harris, 2011). These pressures are less severe in Montreal having experienced de-industrialization more severely and retained a stronger government presence in protecting affordable housing (Walks & Maaranen 2008; Germain & Rose, 2000). While policies in Montreal and Vancouver both emphasized inner city neighbourhood revitalization, they took place in 62 different contexts so that gentrification of neighbourhoods was much more pronounced in Vancouver.  2.5 Coordinating Land Use and Transport as a Sustainability Strategy  A key difference between the two metropolitan areas is that investment in the inner city in Vancouver became part of a larger regional planning strategy to coordinate land use and transportation patterns (Filion et al., 2010). As early as the 1970s a report on a public opinion survey of residents, local politicians and urban visionaries referred to the need to connect different parts of the urban area with transit, constraining the automobile and limiting urban growth to protect Vancouver’s quality of life (Clint, 1974; Brunett-Jailly, 2008; Hutton, 2009). One former planner noted that “accessibility”, by co-locating jobs, housing and amenities at higher densities, had long been a planning objective in Vancouver as opposed to the “mobility” objective of efficiently moving people between two places more common in planning in other cities.19 The “accessibility” policies incorporated Smart Growth and other sustainability planning goals that aim to limit urban expansion and raise local densities to reduce automobile use by creating conditions that make it easier to walk, cycle or take transit (Tomalty, 1997; Newman & Kenworthy, 1999; Roseland, 2005). As Hall (1996) explains, sustainability became popular on a global scale in the late 1980s through the Brundtland Report. The Brundtland report famously defined sustainable development as “development which meets present needs without compromising the ability of future generations to achieve their own needs and aspirations”.   19 Conversation with key contact, January 30, 2009: Former municipal planner. 63 It was given specific meaning in urban planning through land use and transport policies (also see Moos & Skaburskis, 2008; Quastel et al., under review). Hall suggests that the pragmatic ‘definition’ of sustainability in planning arose in part from the work of Newman & Kenworthy (1999) who demonstrated an evident relationship between the density of urban form and automobile use (also see van Diepen & Voogd, 2001; Ewing et al., 2008). Many argue that sustainability has nonetheless remained “an empty term” taking on many meanings depending on context (Gunder, 2006). The implementation of densification policies has raised difficult questions regarding the “right to retain neighbourhoods” (Godschalk, 2004) and the connections between the capitalist growth imperative and densification policies (Kipfer & Keil, 2002; Krueger & Gibbs, 2007) have led some places to implement alternative “slow” or “no growth” strategies (Roseland, 2005; Mayer & Knox, 2006). These have in turn also generated critiques because of the implications of social exclusion of “no growth” in a context of increasing global immigration (Neumayer, 2006). It is interesting that when the land use and transport policies now being called sustainable were initially implemented in Vancouver, the quality of life factor appeared to be a dominant driver:    “Environment, sustainability, nahh, believe you me, no one talked about it...we did talk about compact communities [in the 1970s]...how to build communities where people could walk to neighbourhood shops...this translates into what today is being called sustainability, but this was not our explicit goal.”20   So Vancouver differs from Montreal in that it had already begun to lay the groundwork for “creating cities that are denser and more compact” in the 1970s for what came to 20 Conversation with key contact, December 9, 2008: Former politician.  64 be seen as examples of more sustainable urban development patterns in the 1990s when sustainability “emerged as almost a Holy Grail” in the planning profession (Hall, 1996, p. 413; Brown, 2006; Gunder, 2006; Quastel et al., under review).  The establishment of an Agricultural Land Reserve that contained urban growth and the Liveable Region Strategic Plan, identifying growth centres and visions for protecting environmental assets, helped in the mid-1970s to place the ideals of ‘compact communities’ into policy documents with an overarching aim to maintain “liveability” (Harcourt & Cameron, 2007). The Agricultural Land Reserve and the designation of specific growth centres were certainly not fully successful as one former city planner noted but a provincial policy-maker argued that these policies became two of the “strongest determinants of development patterns” since the 1990s.21 The policies had a net effect of centralizing amenities in and around transit corridors and the growth centres, shown in Figures 2.3 and 2.4 using a walkability-index that measures proximity to surrounding amenities.22      21 Conversation with key contact September 26, 2008: Former municipal planner. Conversation with key contact, December 18, 2008: Provincial policy-maker.  22 The walkability index was calculated using an Internet-based program called WalkScore (2010). The index ranges from 0 to 100, with walkability increasing with higher values. The application uses straight- line distances from a street address (or intersection) to amenities (retail, schools, transit, restaurants, parks) within a .25 mile radios to calculate the index. The street intersections nearest the centroids of census tracts were used to approximate the walkability. In sparsely populated tracts, walkability was calculated at a central intersection in the most populated area. The index as used here is only a very crude approximation of walkability but it serves its purpose to capture relative differences in walkability across tracts. The walkability profiles of the census metropolitan areas fit expectations based on knowledge of the streetscape and other more sophisticated measures of walkability (e.g., Frank et al., 2010). 65 Figure 2.3 – Walkability and rapid transit lines surrounding Vancouver’s downtown   Notes: Rapid transit did not exist in 1981 but there was frequent bus service in the corridors eventually served by rapid transit. By 2006, mass rapid transit connecting the core included express buses from Commercial-Broadway station to UBC (99B) and downtown to Richmond (98B), two SkyTrain lines (Expo and Millennium Line) splitting at Commercial-Broadway station, the SeaBus ferry to North Vancouver and the West Coast Express commuter rail from downtown heading east along Burrard Inlet. Source: Created using Statistics Canada and DMTI (2009) shape files, WalkScore (2010) data and reference maps from TransLink (2011).                          !! !! ! ! !! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! !! ! !!! !! ! ! ! ! ! ! ! ! ! ! 4  Vancouver census tracts  Walkability index 0 - 49 50 - 69 70 - 89 90 - 100 Mass Rapid Transit 0 5 102.5 Kilometers UBC North Vancouver Richmond Burnaby Downtown False Creek Burrard Inlet Commercial- Broadway Burrard- Broadway66 Figure 2.4 – Walkability and rapid transit in the Vancouver CMA  Notes: A rapid transit system did not exist in 1981 but there was frequent bus service in the corridors that eventually became served by rapid transit. By 2006, mass rapid transit consisted of an express bus from Commercial-Broadway station to UBC (99B), an express bus from downtown to Richmond (98B), an express bus connecting Port Moody and Coquitlam to Burnaby (97B), a SkyTrain (ExpoLine) from downtown to Burnaby, New Westminster and Surrey to the south-east, a SkyTrain (the Millennium Line) from downtown heading north into Burnaby at Commercial-Broadway, the SeaBus ferry from downtown to North Vancouver and the West Coast Express commuter rail from downtown along Burrard Inlet to Port Moody and Maple Ridge. Source: Created with Statistics Canada and DMTI (2009) shape files, WalkScore (2010) data and reference maps from TransLink (2011).  The situation regarding planning was different in Montreal in that local politicians in the early 1980s argued for the strengthening of the downtown and opposed the kind of nodes and corridor system pursued in Vancouver. As Leonard & Leveille (1986, p. 94) note, “both the MCM and the then governing Civic Party objected to the creation of what they called “artificial growth centres” in the suburbs”, so that the high-amenity and density areas are centralized (Figures 2.5 & 2.6). Montreal !!!! ! ! !! ! !! ! !!! ! !! ! !! ! ! ! !! ! !! ! ! ! ! ! ! ! ! !! !! ! ! !! !!! !! !! !!!! !!!!! !! !!! ! ! !! ! ! !! !!!!! 4  Vancouver census tracts  Walkability index 0 - 49 50 - 69 70 - 89 90 - 100 Mass Rapid Transit 0 10 205 Kilometers Downtown UBC North Vancouver Downtown Richmond Burnaby New Westminster Surrey Port Moody Coquitlam Maple Ridge Langley67 also did not make a direct effort at regional or sustainability-motivated urban planning until the mid-2000s, although elements of sustainability are reflected in planning policies (Brown, 2006; Filion & Bunting, 2010; Fischler & Wolfe, forthcoming). A regional growth boundary actually existed for some time, as one Montreal academic noted in conversation, but it ultimately has had little effect because of the large amount of undeveloped land still available inside the boundary.   Figure 2.5 – Walkability index and Metro lines on the Island of Montreal  Notes: In Montreal, the Metro lines are referred to by colour. In 1981, the Metro included the ‘Green line’ from Angrignon to Honore-Beaugrand, the ‘Yellow line’ from downtown to Longueuil and the ‘Orange line’ running from Henri-Bourassa to Place-Saint-Henri. By 2006, the Metro also included the ‘Blue line’ from Snowdown to Saint-Michel and an extension of the ‘Orange line’ to Cote-Vertu.  Source: Created using Statistics Canada and DMTI (2009) shape files and WalkScore (2010) data. Metro lines based on spatial files from Section Information et production graphique, Société de Transport de Montréal (STM) received with permission from McGill University Library.  ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! 4  Montreal census tracts  Walkability index 0 - 49 50 - 69 70 - 89 90 - 100 Metro 0 5 102.5 Kilometers Mont Royal F le u v e  S a in t L a u re n t Cote-Vertu Henri-Bourassa Vieux Montreal Saint-Michel Snowdown Angrignon Longueuil Honore-Beaugrand Place- Saint-Henri68 Figure 2.6 – Walkability and commuter rail lines in the Montreal CMA  Notes: At the time of the 1981 census only the lines to Vaudreuil-Hudson and Deux-Montagnes had regular service. By 2006, all the lines shown above had some regular commuter rail service. Source: Created using Statistics Canada and DMTI (2009) shape files and WalkScore (2010) data. Commuter rail lines based on spatial files from Section Information et production graphique, Société de Transport de Montréal (STM) received with permission from McGill University Library.  Germain & Rose (2000) argue, citing Frisken (1994), that “the Montreal metropolitan region has never been able to successfully integrate the development of transportation and land use planning” (p. 108). Several unsuccessful attempts were made over the years to consolidate what remains a very “fragmented regional governance” context (Boudreau et al., 2007). Filion et al. (2010) argue that the highly fragmented nature of urban governance and high degree of municipal competition contributed to suburbanization in Montreal as outlying municipalities could provide ready infrastructure and housing types not available in the historically dense inner city (also see Rose, 2009).  The policies during the 1980s that aimed to attract people to the inner city in the face of decline, such as Operation 2000 logements, actually increased single-family and 4  Montreal census tracts  Walkability index 0 - 49 50 - 69 70 - 89 90 - 100 Commuter Rail 0 10 205 Kilometers Vaudreuil-Hudson Deux-Montagnes Blainville-Saint-Jerome Candiac Mont-Saint-Hilaire69 townhouse developments in the inner city (Germain & Rose, 2000), contrasting with Vancouver’s active attempts to raise development densities in the downtown (Blomley, 2004; Hutton, 2004). While at the provincial level Quebec has been proactive in implementing sustainability strategies, land use and transport strategies have not received as much attention—there has been “less of a political urgency” for sustainability in Montreal as a result of “spotty ideological commitment, fragmented governance structures and a favorable urban and regional ecology” that permitted expansion (Fischler & Wolfe, forthcoming, p. 33). Nevertheless, Montreal’s inner city has long had the high density and amenity components that today are promoted as sustainable since its early industrial origins (Newman & Kenworthy, 1999). Even without the proactive efforts present in Vancouver, Montreal manages to have superior transit usage rates (Danyluk & Ley, 2007; Filion et al., 2010).         Vancouver’s regional planning efforts were aided by the creation of a regional government (first the GVRD, later Metro Vancouver) and a regional transit agency (TransLink) (Harcourt & Cameron, 2007). One policy-maker noted that the province actually played a key role in facilitating planning goals in that it created a “governance apparatus” that permitted these regional planning functions.23 In the mid 1980s, the then NDP provincial government took an interest in funding public transit and “key people”, as another provincial policy-maker noted, viewed transit “as a basis for shaping urban form”.24 As part of the EXPO in 1986 – originally named Transpo 86 – an elevated rapid transit system (SkyTrain) was built to connect the downtown with several of the suburban growth centres (Figure 2.4). It initially had one line built in 23 Conversation with key contact, December 18, 2008: Provincial policy-maker. 24 Conversation with key contact, December 5, 2008: Provincial policy-maker. 70 conjunction with the World Exhibition (the EXPO line). Developed over the 1980s and 1990s, Vancouver now has a multi-modal rapid transit system consisting of express buses, sea ferries, commuter rail and the SkyTrain network. Montreal also created a regional government in the 1970s (the Montreal Urban Community) but it was arguably much less influential in shaping development patterns, and until the early 2000s only included select municipalities on the Island of Montreal. Also founded as part of the MUC in the early 1980s was a regional transport commission, the Societe du transport de la communaute urbaine (STUCM) but it took until 1989 when the province introduced funding that a fully regional transit agency was established that included Laval and the municipalities on the north and south shores (Leonard & Leveillee, 1986). However, as Germain & Rose (2000) explain, the process of establishing the regional agency was fraught with disagreements and years of uncoordinated transit development that led to declines in ridership. The rapid transit system began with the construction of the Metro in the 1960s and expanded over the years (Figure 2.5), but also not, of course, without several gaps as funding came and went with different governments. In the late 1970s and early 1980s, the provincial government re-opened discussions regarding suburban commuter rail lines and integration with the Metro (Frost, 1981b), but rapid transit use still remains mostly concentrated in the inner city.   The combination of policies aimed at increasing “livability” and facilitating de- industrialization through re-zoning, it is commonly suggested, have been successful in attracting growth in Vancouver, particularly in the “new economy” sectors, while sustaining a high quality of life and access to the surrounding natural environment (Berelowitz, 2005; Boddy, 2004; Hutton, 2008; Harris, 2011). Also a relative success in 71 the context of North American cities, dominated by sprawl and the automobile, is that Vancouver has attracted more residential development to central areas than any other city and made walking, biking and public transit realistic alternatives to driving a car (Newman & Kenworthy, 1999; Filion et al., 2010). Montreal because of its longer history of urban density and rapid transit provision, and lower incomes, has long seen higher transit usage and walking and cycling in the central city. Over 60 percent of commuters living in the City of Vancouver travelled by automobile in 2001 as compared to just under 49 percent in the City of Montreal (Danyluk & Ley, 2007, p. 2203). But suburbanization, facilitated by provincial investments in highway infrastructure, also contributed to continuing automobile use in Montreal, which is also the case in Vancouver (Filion et al., 2010; Fischler & Wolfe, forthcoming; Table 2.1).   Table 2.1 – Change in the modal split in the journey to work  Montreal CMA Vancouver CMA   2006 ∆ 06-81 2006 ∆ 06-81 Automobile .705 .072 .742 -.003 Public transit .215 -.053 .169 -.011 Bicycle .015 .013 .016 .006 Walking .057 -.034 .063 -.001 Other mode .008 .002 .011 .008 Notes: Employed labour force only. Automobile includes car commutes as a driver or passenger. Bicycle category combines motorbikes and bicycles in 1981 only. Proportions shown are based on weighted number of cases. Statistical significance of difference in distribution between years is p<0.0001 in Montreal and p<0.05 in Vancouver based on unweighted samples. Source: Calculated using Statistics Canada PUMFS (2006b) and Statistics Canada Travel to Work Survey (1981d).  Montreal’s inner city is arguably equally amenity-rich, perhaps in terms of consumption even more so, than Vancouver’s but the combination of economic decline, reluctance toward neo-liberalization and the stronger commitment to affordable (rental) 72 housing provision have kept the inner city relatively more affordable. In contrast, it is the creation of the amenity-rich urban environments that have been associated with gentrification and rising housing costs in Vancouver (Ley, 1996; Blomley, 2004; Quastel et al., under review). One explanation that has been proposed is that the centralization of residential development is actually increasingly fed by the environmental concerns of “eco-gentrifiers” (Quastel, 2009), who, as it turns out, do not necessarily reduce automobile use due to their higher incomes (Danyluk & Ley, 2007) yet reside downtown to forego long travel distances and the growingly congested suburbs.  The “eco-gentrification” hypothesis foresees an urban accessibility surface increasingly defined by public transit, bike lanes and the walkability of pedestrian environments (Figure 2.7), introducing a class conflict into the urban sustainability debate as lower income households, who may not necessarily afford a car, are priced out of the inner city or at least out of its ownership market (Marcuse, 1998; Burton, 2000; Quastel, 2009; Filion & Bunting, 2010). Affordable housing developer Howard Rotberg (2008) goes as far to typify “Vancouverism” by a “narcissistic” culture that brags about real estate development at high densities to protect existing neighbourhoods and natural areas at the expense of the younger generations and low-income populations who are priced out of the market. Whether or not this is a distinctively local occurrence, or purely connected to gentrification, is debatable as other observers have pointed to a coalescing of conditions, including the aging of the population, rising gas prices, environmental concerns and declining household size, that would see cities generally, and households of different socio-economic status, move toward higher density patterns 73 and support transit, cycling and pedestrian infrastructure (Hall, 1996; Champion, 2001; Filion & Bunting, 2010).   Figure 2.7 – Cycling infrastructure in Montreal and Vancouver neighbourhoods     Notes: Traffic-separated bike lanes in downtown Montreal (top left), and the Bixi bikes available for rent for a small fee throughout the inner city (top right)25. Traffic calming in Vancouver’s Kitsilano neighbourhood (bottom left). The City of Vancouver has designated several bicycle routes on neighbourhood streets off main arterial roads (bottom right).  Source: Montreal photographs taken July 2009. Vancouver photographs taken June 2010.  The arguments hold in a general sense in that pressures on price and their equity implications are more acute in Vancouver than for instance in Montreal, although even here observers are beginning to question the exclusivity of housing developments built according to sustainability principles (Poitras, 2009). But changes in the share of 25 One key informant indicated that the geography of the Bixi bikes initially coincided closely with the pattern of gentrification but that less affluent neighbourhoods began to ask for the Bixi’s as well. April, 2008. 74 workers commuting by car certainly point to a tendency toward “eco-gentrification” more prevalent in Vancouver where the proportion of workers in the social sciences, arts and culture occupations and those with university education traveling to work by car have decreased by 14 and 13 percent respectively from 1981 to 2006, whereas automobile commutes increased for primary sector and clerical workers (Table 2.2).26 In Montreal, automobile commutes have increased for those in clerical and manual occupations but the changes are not statistically significant for occupations characteristic of the quaternary sector implicated in gentrification or those with university education.   The trends point to the broader changes in the factors influencing urban development as young adults today would make different kinds of decisions compared to twenty or thirty years ago when environmental issues were not as high on the political agenda (Beer, 2006; Filion & Bunting, 2010). It should not be forgotten that densification substantially increased the housing supply in and around transit and the downtown in Vancouver that would offer young adults today a larger diversity of housing. But this housing is becoming more expensive and smaller, and the remaining lower density stock, or units with a sufficient number of bedrooms to accommodate larger households, has increased in cost so that it is becoming out of reach for an 26 Occupational categories are groupings of related categories from the census: Managerial (1981: Managerial, natural sciences, engineering, math; 2006: Management, business and finance, natural and applied sciences). Health (1981: Medicine and health; 2006: Health). Social sciences (1981: Social sciences, teaching, art, literature, recreation; 2006: Social sciences, education, government services, religion, art, culture, recreation and sport). Sales and services (1981: Sales, services; 2006: Sales and services). Clerical (1981: Clerical and related; 2006: Clerical and administrative occupations). Manual (1981: Processing, machine, fabrication, assemble, repair, construction trades, transport equipment operating; 2006: Trades, transport and equipment operators, processing, manufacturing and utilities). Primary (1981: Farming, horticulture, animal husbandry, other primary; 2006: Occupations unique to primary industry, e.g., farming, mining, forestry). Statistics Canada (1981a; 2006a).     75 increasing proportion of the population (Gray, 2006; Lee et al., 2008; Rotberg, 2008). This creates tensions between the environmental objectives of higher density development and social objectives of equality (Burton, 2000; Quastel, 2009).   Table 2.2 – Proportion of automobile commuters by occupation and educational attainment  Montreal CMA Vancouver CMA   2006 ∆ 06-81   2006 ∆ 06-81   Occupation       Managerial .761 .009  .794 -.043  Health .717 .114 ** .759 -.043  Social sciences, arts and culture .651 -.034  .708 -.140 *** Sales and services .628 .026  .661 .002  Clerical .644 .120 *** .710 .096 ** Manual .847 .150 *** .869 .014  Primary .816 .035  .856 .243 * Educational attainment       Less than high school .704 .074 *** .728 .011  High school .708 .126 *** .758 -.023  College or trades .741 .104 *** .777 .068  University degree .669 -.038  .715 -.129 *** Notes: Auto commutes (driver or passenger) versus walking, cycling or taking public transit. Employed labour force only. Proportions shown are based on weighted number of cases.  Statistical significance of difference between years ***p<0.0001; **p<0.01; *p<0.05 based on unweighted samples. Source: Calculated using Statistics Canada PUMFS (2006b) and Statistics Canada Travel to Work Survey (1981d).   As Hall (1996) suggests:      “So the phenomenon of environmentally-conscious NIMBYism looms ever larger; and it is very difficult to combine it with any concept of social equity, whether for the less fortunate in the local community, or still more for the less fortunate in other places, or for younger generations, or for generations still unborn.” (p. 422)  Smaller households can be more readily accommodated in central, high-density areas advocated by sustainability policies; and housing affordability may therefore also 76 become an increasingly important variable in the decision as to whether to form families or have children (Skaburskis, 1994; Lauster, 2010).  The coordination of home and work locations have also become complicated over the past twenty to thirty years by changes related to the shifts in the organization of production. Post-Fordist restructuring increased the employment component for centrally located quaternary sector workers. But de-industrialization resulted in the suburbanization of employment particularly in manufacturing industries, and warehousing; and the emergence of business parks and ‘edge cities’ also increased the office and retail components of suburban areas, particularly of back-office functions (Hutton, 2010). This partly explains the increases in auto commutes for clerical workers (Table 2.2). The employment components have grown substantially in suburban areas, yet in the case of Montreal and Vancouver the central cities still remain the largest concentration of employment (Heisz et al., 2005). Shearmur & Coffey (2002) in their analysis of the “space economy” conclude that Montreal remains “a city with a strong CBD, around which economic activity tends to converge” whereas Vancouver, which they call a “paradoxical case”, also “has a strong and fast-growing CBD” but “stands out as having a large number of…isolated centres” capturing a “significant proportion of employment growth” (p. 594).  The impact on commuting patterns has been an increase in suburb-to-suburb commutes and so called reverse commuting but the persistence of central employment growth, and “scattered, low-density developments” of suburban employment, has also meant increased congestion, particularly in Montreal (Heisz et al., 2005; Hutton, 2010, p. 121). Average commute distances, and time, continue to be shorter in inner areas, and 77 the continuing expansion of the suburbs and inner city growth provide a setting today where central locations may increasingly be sought after due to the loss of time associated with being ‘stuck in traffic’. This is particularly the case among the increasingly educated who place a higher perceived value on their time and the young adults who according to social surveys dislike the commute more than older workers in Canada (Wheaton, 1977; Turcotte, 2008a; Skaburskis & Moos, 2010, p. 238).   2.6 Paying More for Housing The difficulty in finding and affording housing for Vancouver’s young adult labour force as compared to the past and to other metropolitan areas is a common theme in public discourse, perhaps most dramatically captured by a headline that claims the city “eats its young” (Beers, 2007; see Rotberg, 2008). Another newspaper article that also highlights the rise in housing costs in Vancouver, and the more favourable rental market in Montreal, is particularly germane in that it compares the housing experience of one young adult woman in the two metropolitan areas. The woman asserts: “In Vancouver, a renter’s rat race: Just here from Montreal, I figured finding a decent, non-frills place would be easy. Crazy me” (Addleman, 2009). Caution is undeniably warranted in taking the anecdotal, and perhaps sensational, newspaper reports as representative of larger trends but the high cost of Vancouver’s housing market is certainly also exposed in systematic accounts (Ley & Tutchener, 2001; Luffman, 2006; CMHC, 2007d).  Housing expenditure is one of the largest components of household spending— Canadians allocate just below twenty percent of their household expenditures to pay for 78 shelter costs, followed by transportation and food (Statistics Canada, 2006e; Chawla, 2007). The percentage of household income allocated to housing expenditure is also an important indicator of affordability—households paying more than 30 percent of their income towards their principal accommodation are commonly considered to be at risk of affordability issues (Lefebvre, 2003). Low-income households, lone-parent families, women, ethnic minorities, the elderly and those residing in the largest cities are well- known to be more likely to face affordability issues in Canada (Moore & Skaburskis, 2004). The increasing share of income that young adults allocate to housing over time could be in part related to stagnant incomes (Chapter Four) but nonetheless speaks to the rising cost of housing as a share of total expenditures in Montreal and Vancouver (Table 2.3). The expenditures are lower in Montreal than in Vancouver due to the differences in incomes and cost of housing. The renters in Vancouver saw their real housing expenditures decline somewhat, which likely relates to the declining incomes of renters observed more generally (Hulchanski & Shapcott, 2004).   Table 2.3 – Expenditure on principal accommodation by tenure  Montreal  Average expenditure Share of Income  1982 2005 1982 2005 Renters 5,818 6,090 .132 .167 Owners 10,752 13,247 .139 .146  Vancouver  Average expenditure Share of Income  1982 2005 1982 2005 Renters 9,612 9,252 .189 .205 Owners 14,499 18,430 .169 .225 Notes: Dollar values adjusted to $2006 using Bank of Canada rates (2010). Includes households with a maintainer 18 to 39 years of age. Expenditure shown for rented or owned principal accommodation by tenure of the household. Source: Survey of Household Spending and Survey of Family Expenditure Statistics Canada custom tabulations.   79  Both young adults who are renters and those who are owners pay a larger percentage of their income toward housing in Vancouver where dwelling values and rents are also higher. The trends are revealing of the increasing cost of housing that has become a seemingly accepted characteristic of housing markets in major cities (Hackworth, 2007; “Briefing: London and Paris”, 2008; Kern, 2010). Rising housing costs are in one sense an indicator of the desirability of a particular location as some households are willing to pay more (Glaeser & Gottlieb, 2006), but increases in the share of income allocated toward housing also means that households would have to downscale their housing consumption to maintain similar levels of expenditure on other goods and services. Thus, increases in the cost of housing raise questions regarding affordability and the potential displacement of low-income populations (Ley, 1996; Gurran, 2008; Harris, 2011). Kern (2010), drawing on work by Blomley (2004), sees these trends connected to the neo-liberal restructuring of urban housing markets whereby urban space is increasingly privatized, social and assisted housing policies disappear and space is allocated on the principles of “highest and best use”.  While location has long been a question of being able to pay more than someone else, neo-liberalization, which helped produce an inner city condominium market in Vancouver, is a dramatic departure from the more inclusive Keynesian-inspired urban policies of the 1970s (Mitchell, 2004). Different in the Montreal (and Quebec) case are governments that “have continued to support the construction of social housing, and that require the inclusion of affordable housing as part of new infill development on brownfield sites” (Walks & Maaranen, 2008, p. 60). Also influencing the housing expenditures are the global exposure of Vancouver’s housing market, and the tighter 80 land constraints that contrast with Montreal’s less favourable economic conditions that alleviate pressures on housing markets.     The intent of the analysis that follows in this section is to reveal empirically how the differences in urban restructuring between two metropolitan areas have shaped young adults’ housing expenditures. Since the average expenditures shown in the table above do not account for other changes in the decisions relating to dwelling type and tenure or household characteristics, a multivariate analysis is used to test changes over time in the share of income allocated to housing than similar households in the two metropolitan contexts. If the housing context in Vancouver is more restrictive due to neo-liberal restructuring, the similar households should be paying a higher percentage of their income for similar kinds of housing. The multivariate models also permit insight into the ways the constructed permanent and monetary incomes, and by inference their determinants such as education, occupation and gender, are associated with housing expenditures, revealing potential inequalities in the way housing markets have changed.   2.6.1 Data summary and preparation  The analysis of housing expenditure over time presents several challenges related to sample size and data comparability. The Survey of Household Spending (SHS) (known as the Survey of Family Expenditure (FAMEX) prior to 1997) that provides data on spending patterns of Canadian households includes a much smaller number of cases than are available in the census files. The total number of cases in the 2005 SHS and the 1982 FAMEX are 15,222 and 10,938 respectively as compared to 81 312,513 cases available in the 2001 PUMFS (“Survey of household spending”, 2009).27 The SHS and FAMEX data are not available publicly at the same level of geography as the PUMFS due to privacy concerns. The SHS and FAMEX identify households’ location by province and size of urban area, thus permitting analysis for urban areas with populations larger than 100,000 in Quebec versus British Columbia.28 However, since it is necessary for the purposes of this analysis to further restrict the sample of large urban areas to include only young adults, sample size becomes too small to conduct analysis that yields statistically significant results, and in some cases cross- tabulation yields cell counts below Statistics Canada’s minimum data reliability threshold (Statistics Canada, 1982a; 2005b).  Due to the issues with sample size, the analysis using the SHS and FAMEX micro-data files include all sample households in each of the two provinces with a maintainer 25 to 34 years of age. This obviously limits direct comparability to the analysis conducted for the two metropolitan areas using the census data but several tests using interaction terms reveal that the relationships in question—spending patterns over time and by household characteristics—do not actually vary by urban size or urban/rural distinctions.29 The contextual differences between Montreal and Vancouver also extend, 27 The 1982 FAMEX is closest to the 1981 census (previous and subsequent FAMEX conducted in 1978 and 1984 respectively). The SHS is conducted annually but the 2005 income figures are comparable to the 2006 census that asks about income from the previous year (Statistics Canada, 2005a).     28 Custom data were purchased from Statistics Canada to obtain summaries at the metropolitan scale but the definition of young adults is expanded to those 18 to 39 year old due to small sample size and does not permit further cross-tabulation (Table 2.3).  29 When the data were restricted to include only the largest urban areas in each of the two provinces, the overall conclusions remain the same, although one cannot be certain as to the statistical significance due to the high variance. It is expected, however, that the analysis of provincial data would result in lower housing expenditure estimates since housing costs are generally higher in the large metropolitan areas. 82 albeit in more complex ways than can be accounted for here, to the provinces as a whole, thus still making relevant an analysis that compares Quebec to British Columbia. The de-industrialization and economic decline certainly afflicted housing markets across Quebec, whereas rapid housing market appreciation occurred across British Columbia. There are also the differences in provincial policies that made neo-liberalization of the state more common in British Columbia than in Quebec.  Table 2.4 shows a summary of the variables separately for each province and census year.30 The data reflect the increase in the percentage of young adult households with a female maintainer and the decline in the presence of children and number of rooms, trends explored in more detail in Chapter Four. The summary shows a higher share of young adult households in rented multiple-dwellings and lower shares in the more recently constructed housing in Quebec than in British Columbia, findings which will be put to a multivariate test in Chapter Six. The variable measuring urban size shows a decline in the proportion of young adult households in the rural areas and cities with populations less than 100,000. It reflects the movement of young adults into larger urban areas and the overall urbanization of the population that in Canada is primarily The regression models do include a variable to account for the differences in the magnitude of spending patterns by urban size.  30 The data include young adult households 25 to 34 years of age and exclude seasonal households. Categorical variables were collapsed into dummy variables due to cell counts below 30. The variable identifying dwelling type and tenure even when reduced into four categories still had one cell with counts below 30 when tabulated for the two years and provinces separately. Thus, this variable combines the 1982 and 2005 data for the purpose of the summary table.  83 taking place in and around the largest cities (Bourne, 2007a; Moos & Skaburskis, 2009). Also shown are the permanent and monetary incomes.31   Table 2.4 – Summary of young adult households in the provinces of Quebec and British Columbia  Quebec British Columbia  1982 2005 1982 2005 Female = 1 .192 .446 .217 .418 Children present = 1 .548 .383 .557 .359 Own single-detached house .311 .322 Rent single-detached house .077 .202 Own multiple-dwelling .069 .099 Rent multiple-dwelling .544 .376 Dwelling built > 1960 = 1 .583 .694 .607 .854 Population < 100,000 = 1 .378 .222 .445 .198 Number of Rooms 5.057 4.974 5.779 5.252 Permanent income (000s) 56.097 51.904 59.413 56.539 Monetary income (000s) 2.714 5.884 5.982 6.652 Housing expenditure (000s) 7.085 8.278 10.791 12.680 Housing expenditure (%) 12.081 14.341 16.481 20.066 Imputed rent (000s) 5.209 5.638 7.477 7.733 Imputed rent (%) 8.882 9.767 11.420 12.237 Notes: The 1982 and 2005 data are combined when cell counts are below Statistics Canada’s recommended minimum for release (<30). Housing expenditure and imputed rent shown as a percentage of household income. Dollar values adjusted to $2006 using Bank of Canada rates. Source: Calculated using data from the Statistics Canada Survey of Household Spending (2005b) and the Survey of Family Expenditure (1982b).    In an analysis of labour and housing market dynamics, it is especially important to distinguish between the permanent and monetary incomes.32 The housing literature has long emphasized the need to estimate separately the differing effects of regular 31 The linear regression included 10,970 cases and showed an r-squared of 0.388 in 2005 and 8,858 cases with an r-squared of 0.375 in 1982. The incomes are predicted using all households earning a positive income, excluding seasonal households and those with no maintainers in the labour force. The coefficient estimates are similar to those found in the census data (Chapter Three) with almost all variables producing statistically significant coefficients at the p<0.0001 level, except some of the categories in the variable identifying the metropolitan areas.   32 The discussion of permanent income in this and the subsequent paragraph draws on research previously published by the author (see Moos & Skaburskis, 2008).  84 (monetary) earnings and the long-term (permanent) income potential on housing demand (Goodman, 1986). The permanent income, the expected earnings over a longer time frame, is believed to be the basis for housing decisions as households pay for their housing over several years and want to minimize search, moving and other transaction costs. Financial institutions also base their decisions to lend on households’ stability of earnings, which is a function of permanent income.  The method of estimating permanent and monetary income is based on work by Goodman (1986), Goodman & Kawai (1984), Skaburskis (1996) and subsequent application of the technique (e.g., Moos & Skaburskis, 2008; 2010). To estimate the permanent income, the natural log of household income is regressed against factors influencing the household’s ability to earn an income over time. The exponent of the predicted value is the estimate of the permanent income, and the monetary income is the actual income less the permanent income. The independent variables generally included in the estimates of permanent income are gender, number of earners, age, occupation and level of schooling. A variable identifying census metropolitan areas is also added here to identify regional differences in earnings. One limitation in the use of the SHS and FAMEX data is a change made by Statistics Canada in what is included under housing expenditure (Lafrance & LaRochelle-Cote, 2011). The main difference is that the 2005 table includes the mortgage principal and interest payments whereas the 1982 data only include mortgage interest payments.33 Therefore, it is not known whether any increases in housing 33 The two variables used to measure housing expenditure are differentiated by tenure as ‘rented principal residence’ and ‘owned principal residence’ in the surveys. The variable for renters includes spending on rent, parking, tenant insurance and repairs not covered by the landlord in 1982 and 2005. The variable 85 expenditure of owners shown in Tables 2.3 and 2.4 are actually due to changes in the cost of housing or simply arise from discrepancies in variable definition. Adapting approaches used by Crossley & Curtis (2006) and Lafrance & LaRochelle-Cote (2011), the issue is addressed by estimating an imputed rent as a measure of housing expenditure. As these authors explain, imputed rent is the amount a household could be expected to pay given the rents charged for the type of housing they are consuming. The changes over time in imputed rent do suggest an increase in total housing expenditure and as a percentage of household income for a standardized dwelling (Table 2.4). There are also differences in the number and types of categories used in the 2005 SHS and 1982 FAMEX to define dwelling characteristics, such as type and year of construction, and urban size. These issues are resolved by combining categories for these variables when the 1982 and 2005 tables are used jointly, which results in some loss in detail.   Imputed rent is predicted for young adults using a regression model that estimates the natural log of rent as a function of housing and geographic characteristics for all tenant households (Table 2.5). The results are similar to those of a previous estimate of imputed rent by Lafrance & LaRochelle-Cote (2011), although they did not include year of construction and urban size (and charges for water, electricity and heat that they include are excluded here). Duplex and apartments are associated with higher rents, which as Lafrance & LaRochelle-Cote suggest is likely due to the location of multiple-dwellings in central areas of larger cities. In 1982, the newer buildings are associated with higher rents but this relationship becomes less clear over time, perhaps ‘owned living quarters’ includes spending on maintenance and repairs, condominium charges, property taxes, homeowner insurance and mortgage payments.  86 due to the value appreciation of the older housing due to gentrification (Skaburskis, 2006a).   Table 2.5 – Household expenditure on rent as a function of dwelling and geographic characteristics 1982 FAMEX 2005 SHS Variables Coeff.  Variables Coeff.  Semi-detached dwelling .109  Semi-detached dwelling .071  Row house .077  Row house -.017  Duplex .201 *** Duplex .133 * Apartment .241 *** Apartment .225 *** Other dwelling -.111  Other dwelling -.093  Built 1946 – 1960 .086 * Built 1946 – 1960 .101 * Built 1960 – 1970 .169 *** Built 1961 – 1970 .093 * Built 1971 – 1975 .135 * Built 1971 – 1980 .063  Built 1976 – 1980 .313 *** Built 1981 – 1990 .033  Built 1981 .122  Built 1991 to 2005 -.008  Built 1982 .036  Number of rooms .222 *** Number of rooms .206 *** Number of rooms^2 -.017 *** Number of rooms^2 -.013 ** Number of bathrooms .106  Number of bathrooms .160  Number of bathrooms^2 -.016  Number of bathrooms^2 -.013  Nova Scotia .185 * Quebec -.070  New Brunswick .120  Ontario .061  Quebec .130  Prairies .141 ** Ontario .461 *** British Columbia .278 *** Manitoba .091  Population 30,000 – 99,999 -.102 *** Saskatchewan .088  Population less than 30,000 -.284 *** Alberta .361 *** Rural areas -.558 *** British Columbia .478 *** Constant 6.948 *** Territories .311 ** N-cases  3,359   Population < 100,000 -.215 *** R-Squared .212  Rural areas -.249 **    Constant 7.649 ***    N-cases  2,899      R-Squared .165  Notes: Sample weights are used in estimation of linear regression models. Table includes households who are renting. The base in 1982 is the households in single-detached houses, built before 1946 in Atlantic Canada and urban areas with populations greater than 100,000. The base in 2005 is the households in single-detached houses, built before 1946 in Newfoundland and Labrador and urban areas with populations greater than 100,000. ***p<0.0001; **p<0.01; *p<0.05    Source: Calculated using data from Statistics Canada Survey of Household Spending (2005b) and the Survey of Family Expenditure (1982b).    87 The value of rent increases with the number of rooms but the effect declines as the number of rooms increases. The variable identifying urban size points to the higher rents in large metropolitan areas. The variable identifying provinces shows that rents are higher in British Columbia compared to the base for otherwise similar kinds of dwellings, whereas there is no statistically significant distinction between Quebec and the base. The finding clearly points to the higher cost of the rental market in British Columbia compared to the rest of Canada that is independent of the effect of urban size and housing characteristics on rents.  2.6.2 A geography of expenditure patterns  The first two multivariate models find that young adult households spend a higher percentage of their income on housing in British Columbia than in Quebec, despite lower incomes in the latter (Table 2.6).34 The models are constructed separately for the 1982 and 2005 surveys. The dependent variable is the percentage of household allocated to housing expenditures and the independent variables denote the province, urban size and housing and household characteristics. The regressions account for several household level variables that would influence the share of income allocated to housing. Therefore, the coefficients for the variables denoting province point to how the 34 The models using FAMEX and SHS include Stata’s pweight and jackknife estimation options, using Statistics Canada’s sampling weights, to estimate the variance in the case of “complex survey design” (Statistics Canada, 2005a). The pweight command with jackknife specified is a “robust variance estimation technique” that “adjusts for the design characteristics” using weights and repeated estimation using sub-samples “so that variances, standard errors and confidence intervals” are more accurate reflections of their true values (“Choosing the correct weight syntax”, 2010). Since the public files do not include all information regarding sampling stratification and clustering, the variance calculated is still likely an underestimate (Statistics Canada, 2005a). There are different views on whether the use of weights actually improves the reliability of variance estimates—Nichols (2007) and von Hippel (n.d.) point to sources that discuss the issue in further detail. 88 differences in housing context, including the degree of neo-liberal restructuring, and housing costs, combine to create a specific geography of housing expenditure patterns. The magnitude of the coefficient for province shows that in 1982 a young adult household in Quebec spent about 5.8 percent less of their income toward housing than those in British Columbia. In 2005, the difference increased to about 8.5 percent. The second set of models combines the 1982 and 2005 data for each province. It finds that young adults allocate a higher percentage of their income toward the imputed rent in 2005 than in 1982 but the increase over time is larger in British Columbia (Table 2.7).    Table 2.6 – Correlates of the percentage of household income allocated to housing expenditure by year  1982 FAMEX 2005 SHS  Coeff. Beta  Coeff. Beta  Quebec = 1 -5.767 -.283 *** -8.525 -.272 *** Female = 1 1.898 .080 * -.005 .000  Child present = 1 -.512 -.027  -2.819 -.091 * Rent single-detached house -5.110 -.177 *** -9.780 -.207 *** Own multiple-dwelling -.060 -.001  -5.448 -.115 * Rent multiple-dwelling -3.352 -.177 *** -9.854 -.329 *** Dwelling built > 1960 = 1 2.201 .115 ** -.740 -.021  Population < 100,000 = 1 -3.935 -.204 *** -7.356 -.202 *** Number of rooms .171 .034  .476 .067  Permanent income (000s) -.160 -.317 *** -.257 -.340 *** Monetary income (000s) -.162 -.400 *** -.237 -.451 *** Constant 30.311  *** 48.815  *** N-cases 857   453   Adjusted R-squared .290   .321   Notes: Standard errors used to determine significance levels calculated using jackknife estimation procedures. Sample weights are used in estimation of the models. Includes households with a maintainer 25 to 34 years of age. The base is the British Columbia households with male maintainers, no children present, owning a single-detached house, dwellings built before 1960 and residing in a city with a population of at least 100,000. ***p<0.0001; **p<0.01; *p<0.05    Source: Calculated using data from Statistics Canada Survey of Household Spending (2005b) and the Survey of Family Expenditure (1982b).     89 Table 2.7 – Correlates of percentage of household income allocated to imputed rent by province  Quebec British Columbia  Coeff. Beta  Coeff. Beta  Female = 1 .611 .029  1.946 .060  Children present = 1 -1.214 -.063 ** -2.040 -.067 * Population < 100,000 = 1 -5.232 -.252 *** -5.327 -.164 *** Year = 2005 6.415 .333 ** 10.502 .346 * Permanent income (000s) -.184 -.361 *** -.212 -.280 *** Permanent income * Year -.088 -.267 * -.140 -.291 * Monetary income (000s) -.188 -.476 *** -.232 -.442 *** Constant 24.533  *** 31.251  *** N-cases 779   531   Adjusted R-squared .539   .416   Notes: Standard errors used to determine significance levels calculated using jackknife estimation procedures. Sample weights are used in estimation of the models. Includes households with a maintainer 25 to 34 years of age. The base is the 1982 households with male maintainers and no children present. ***p<0.0001; **p<0.01; *p<0.05    Source: Calculated using data from Statistics Canada Survey of Household Spending (2005b) and the Survey of Family Expenditure (1982b).    The geography of expenditure patterns across provinces reflects in part what is often described as the ‘high desirability of British Columbia real estate’ and the associated quality of life factors. Housing types in the regressions are therefore not completely similar in that not accounted for are the external factors associated with particular housing locations. One such factor of Vancouver’s real estate market is the aesthetic aspect of the surrounding natural environment. Berelowitz (2005, p. 161 & 175) suggests, for instance, that “the city acts as a sort of mirror or a vast display case for the aesthetic consumption of nature” and that people “are prepared to pay dearly for their piece of visual paradise”. The provincial variable also encompasses information on economic success as British Columbia was experiencing growth from Asia-Pacific trade, strong natural resources demand and the expansion of the new economy in Vancouver, whereas Quebec was continuing to struggle to adjust from decline associated with de-industrialization and the separatist movement (Ley & Hutton, 90 1987; Germain & Rose, 2000; Barnes et al., 2011). However, the regressions do not account for the more subtle differences in housing quality such as overcrowding or even simply residing with roommates that households might use to reduce housing expenditures in high-priced markets.  The data also importantly point to the potential repercussions for affordability and displacement of institutional restructuring that facilitates real estate price gains and reduces public intervention in the housing market on the premises of neo-liberal (entrepreneurial) governance to enhance the attractiveness of place for global investment (Kipfer & Keil, 2002; Kern, 2010). Households have to allocate more of their income in higher priced locations toward housing and those with lower incomes pay an even larger share of their income. The coefficients for the permanent and monetary income variables show that the percentage of income allocated to housing decreases as income increases. The permanent income variable includes information regarding education, occupation and gender, thus pointing to the ways labour market restructuring that is often shown to increase class and gender income inequality works its way into the housing market to produce affordability concerns (see Chapter Four). The findings also resonate with evidence from the housing affordability literature that suggests low incomes, not just price increases, are behind growing affordability concerns (Lefebvre, 2003; Moore & Skaburskis, 2004). In the model in Table 2.7, the variable indicating the year of the survey data was also included as interaction effects with the income variables but only the interaction with permanent income that produced statistically significant outcomes is shown. The interaction points to the strengthening of this negative relationship between income and the percentage of income allocated to 91 housing over time. It points to the greater potential for affordability concerns for low- income households today than in the past.  The findings for the variable measuring urban size are also interesting in that they show an increase over time in the difference in housing expenditures between large cities with populations over 100,000 and the smaller cities and rural areas. This corresponds with Ley’s (2007) finding that growing domestic outmigration from large metropolitan areas is related to housing affordability issues. The reverse of this trend is that larger cities with high priced markets potentially become out of reach of those in areas where they are spending relatively less of their income on housing, which would constrain migration between places with various successes in integrating into the new economy. The finding of increasing differentiation of housing expenditure by urban size also relates in more general terms to the way the Canadian urban system has developed over the past twenty to thirty years. Most economic and population growth occurred in large urban centres (Bourne, 2007a)—the relative increases in the share of income allocated to housing over time in the larger cities points to the way the restructuring of the urban system raises housing expenditures by increasing competition for land and housing in large cities. Housing expenditures define the geographies of affordability across provinces and the urban system. They shape young adults’ housing and commute decisions as will be explored in subsequent chapters.  2.7 Discussion  Evidently young adults are entering housing and labour markets under quite different conditions today than in the past. Urban growth and investment in inner city 92 amenities and real estate markets have resulted in increases in the cost of housing, particularly in Vancouver. The changes have also, however, dramatically altered the housing stock characteristics, for instance increasing the supply of smaller condominium units and the availability of rapid transit. At the same time, flexibility in the labour market associated with neo-liberalization has meant decline in the security of employment and earnings that shape housing and location decisions. In some sense, the perceived imperative of homeownership is often argued to have increased with neo- liberalization (Ronald, 2008) but rising prices and declining government support actually make it more difficult to enter the market (also see Calvert, 2010; Beer et al., 2011). One expected outcome, which pans out in the spending data, is that young adults would spend an increasing share of their income on housing, reflecting both the increasing amenity component and urban growth that have impacted price as well as perhaps an increase in the investment component households allocate toward housing. Housing expenditures, despite also having increased, remain lower in Quebec than in British Columbia, and reflect in part the differences in the degree of neo-liberalization of housing policy and the negative impact post-Fordist restructuring has had on the Quebec, and Montreal, economy.   An important finding of the analysis of housing expenditures is the relatively greater effect of income on housing consumption than in the past. This points to the fact that higher income households are increasingly allocating more of their income toward housing, which suggests some internalization of neo-liberal ideals that emphasize the investment, as opposed to the shelter, component of home ownership (Bourne, 1981; Larner, 2000). The findings squarely point at the growing affordability burdens faced by 93 lower income households who are competing with higher income earners who not only have more to spend on but also allocate a larger share of their earnings toward housing and are being compensated more due to the inequities emerging in labour markets (see Walks, 2010). Subsequent chapters continue to add to this discussion of the differences in the housing context and their implications for earnings and housing market outcomes at the neighbourhood and household scales. Chapter Three considers in detail the role of immigrants; immigration is now the largest component of population growth in major metropolitan areas and analysis of its transformative effects on especially the Vancouver housing ownership market contributes to the thesis’ aims of adding insight into how housing markets have changed over time. The analysis in Chapter Three also provides an opportunity to add a quantitative dimension to the changes in the housing context broadly established in the discussion above. Chapters Three and Four further analyze two themes only touched upon briefly in this chapter, and these are the important relationship between housing consumption and the demographic characteristics of the households and the way labour market restructuring is altering the income distribution, which influences young adults’ abilities to afford housing.     94 Chapter Three: Global Restructuring and Housing Demand35  An important driving force of urban restructuring in Canada since the 1980s has been the increasing influence of globalization that manifests itself as processes of institutional and socio-economic restructuring at different scales. Cities play a unique role, becoming nodal centres of economic and population growth (Castells, 2010). The internationalization of the production process and competition among locales for capital and labour are characteristics of a globalizing world system in which people and places are ever more tightly connected. While there is already a rich urban literature that documents the social and economic transformations of the postwar, intra-urban landscape, much remains to be explored about how globalization processes have materialized as shifts in the internal structuring of Western cities (Marcuse & van Kempen, 2000; Olds, 2001; Walks, 2001), especially through changes in urban housing markets (Bartelt, 1997; Ley & Tuchener, 2001).  In a globalizing world it is of course difficult, if not futile, to isolate truly local versus global forces that shape the internal organization of cities, especially if one accepts that the processes operating at different scales are interrelated (Swyngedouw, 2004). Nonetheless, if certain aspects of cities are qualitatively and quantitatively different than when economies were more nationally contained, it is possible to speak about the local outcomes of the globalization process (see, for instance, Bartelt, 1997). The growing involvement of foreign capital in land speculation and real estate development projects has received prior treatment in the literature (Olds, 1995), but less 35 This chapter is in part a modified version of an article that has been previously published by the author, with Andrejs Skaburskis, and is reprinted with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved. 95 attention has been paid to immigrants as agents of the globalization process in residential housing-market functions (Carter, 2005; Saiz, 2007). This chapter focuses on the changing relationship between housing demand and income in the Montreal and Vancouver metropolitan areas, and how the changing nature of this relationship due to immigration influences the ways neighbourhoods change. The analysis thus further contributes to the objective of the thesis to illuminate the changes in the urban housing market context within which young adults make decisions; and it also develops the methodologies applied in subsequent chapters to understand the relationship between housing and household level variables. A specific focus on immigration as a facet of the changing housing context is warranted since for many observers there is little doubt that immigration has become a more important factor in urban housing market growth in Canada, particularly in Vancouver (Figure 3.1; Ley & Tutchener, 2001). Hou & Bourne (2006) explain how net domestic migration and high levels of international immigration have made the latter a more significant influence on residential markets. This is especially the case in Toronto and Vancouver that are the main destination for immigrants arriving in Canada, and where immigration is the largest component of population growth (Heisz, 2006). Montreal is also one of the three largest recipients of immigrants to Canada but as a proportion of household maintainers the share remains smaller than in several other medium-sized CMAs, perhaps in part due to the large size of immigrant households in Montreal (Rose, Germain & Ferreira, 2006). Due to the larger size of the rental market and relatively lower pressure on its ownership market, aggregate property value effects are likely less pronounced in Montreal. In contrast, Ley et al. (2001) show that 96 immigration was the most important correlate of changes in residential dwelling value from 1986 to 1996 in Vancouver when differentiating between the effects of gentrification, social polarization, and immigration36. The impact on housing demand from growing immigrant flows are compounded by the shifting profile of migrants due to changes in Canadian immigration policy that favour skilled and economic migrants (Hiebert, Mendez & Wyly, 2008).   Figure 3.1 – Average CMA dwelling value and proportion of household maintainers immigrants   Source: Calculated using Statistics Canada PUMFS (1981b; 2001a). Modified and reproduced with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved.  36 Muller & Espenshade (1985) in Los Angeles and Burnley et al. (1997) in Sydney also find positive correlations between housing prices and immigration. Saiz (2007) confirms the positive association in the U.S. while controlling for other factors affecting housing prices. Carter (2005, p. 266) summarizes further changes in Canadian cities, such as “the growth of exclusive, prosperous immigrant neighborhoods, the development of monster homes, new architectural designs, ethnic businesses and changing household growth patterns.”   0 0.1 0.2 0.3 0.4 0.5 0.6 50,000 100,000 150,000 200,000 250,000 300,000 350,000 400,000 P ro po rt io n of C M A  M ai nta in er s Imm ig ra nt s Average CMA Dwelling Value ($2001) 2001 1981 Vancouver Vancouver Montreal Montreal R^2 = 0.721 R^2 = 0.250 97 The implications of the large influx of wealthy and skilled, particularly Asian, migrants to Vancouver have received substantial prior attention (Li, 1994; Ley, 1995, 2010; also see Winders, 2000, for a review), but understanding of its impact on the housing market can be further elucidated by an examination of the determinants of the demand for owned housing, particularly in comparison to Montreal that has a different immigrant profile and settlement patterns (Rose et al., 2006). Income is emphasized because global restructuring alters the way earnings are allocated to housing. While immigrants continue to be socio-economically diverse, the arrival of increasingly skilled and wealthy migrants would change the relationship between local labour market earnings and housing as immigrants arrive with established savings, or in some cases even continue to earn income outside the country. The picture is further complicated by growth of well-educated native-born workers moving into central areas, an emphasis on real estate as an investment, and urban growth induced price escalations that change the ways households consume housing in globalizing cities, and thus reshape the conditions within which young adults entering the market operate. The chapter begins with a conceptual discussion of how others in the literature have placed housing in a global context. It considers the changing dynamics between housing and labour markets in a context of global restructuring, and then the particularities of immigration in Montreal and Vancouver.  The analysis explores the changes in the housing stock, such as prices, rents and dwelling types and densities, over time and their intra-urban spatial dimensions so as to add a quantitative dimension to the changes in the housing context discussed in the previous chapter using the 1981 and 2006 census tract data. The empirical analysis of the determinants of housing 98 demand uses the 1981 and 2001 PUMFS. The user cost of housing, a measure of housing demand (or consumption), is estimated as a function of a household’s demography and income, holding housing characteristics constant. Census tract data are used to explore intra-urban dimensions by way of an analysis of the determinants of changes in dwelling values, which provides further context on the changing housing markets in Montreal and Vancouver.  3.1 Housing and Labour Market Dynamics in a Global Context  There is a clear recognition that while housing “tended to have more the flavor of Main Street than of Wall Street” (Logan, 1993, p. 35)—and the local orientation of housing markets justified the analytical focus at the intra-urban scale—housing must now be studied in light of regional as well as global scale effects (Coakley, 1994; Bartelt, 1997; Ley & Tutchener, 2001; Terrones, 2005; Ley, 2007). Following Dymski & Isenberg (1998), globalization is understood here as a process whereby institutions at various scales are setting very specific conditions that facilitate the movement of capital and people across national boundaries, plus the neo-liberal restructuring that is working to configure institutions in a manner that enhances these flows (Smith, 2001; Mitchell, 2004; Aalbers, 2009). Bartelt (1997) explains how these global “capital and labor flows in the modern world economy are mediated through existing urban locales” (p. 124), and his analysis of the ways in which the flows reshape and modify the local housing markets reveals impacts of globalization. Also, Badcock (2004) examines how global capital flows exploit national variations in interest rates in real estate investment decisions and the implications for local housing and lending markets, thus emphasizing 99 the changing local conditions in facilitating globalization processes. The analysis in this chapter follows along a similar vein as these previous studies by examining how immigration reconstitutes the local housing market by altering the determinants of housing demand.  Although the factors affecting urban housing markets are diverse and multi- scalar, the changes in labour markets arising from globalization offer one explanation for the shifts in housing demand and the internal structure of cities. Labour markets are closely related to housing demand, the latter being a function of household formation, income, wealth, and preferences for housing space (Bourne, 1981). Households may also alter their labour market participation to match their preferences for housing consumption. The connection between labour market income and housing consumption became especially apparent as many North American cities transitioned from industrial to service-oriented economies due to the internationalization of the production process. The growing numbers of high-income earners in the tertiary and quaternary sectors of the economy raised the profitability for high-priced housing, especially in the inner city through gentrification, often resulting in greater affordability burdens for renters and those in lower-level occupations (Badcock, 2004; Hulchanski & Shapcott, 2004; Hutton, 2008). The higher earner households may spend more of their income toward housing to pay for proximity to amenities or to invest in the real estate market (Hackworth, 2005; Meligrana & Skaburskis, 2005). In contrast, the job losses and diminishing employment benefits and security associated with de-industrialization of North American cities contribute to neighbourhood decline and housing stock deterioration (Bartelt, 1997).   Immigration factors into the processes that alter local housing market functions 100 by shifting the composition of the labour force (Saiz, 2007). By expanding population growth, immigrants increase local demand for housing, pushing up prices unless accommodated by new supply or offsetting native out-migration (Ley, 2007). Immigrants differ in their household formation rate, changing their demand for housing; and housing consumption also differs by race and ethnicity (Skaburskis, 1996). Immigration can further increase the size of the labour pool, especially in lower-level occupations, decreasing wages which then would reduce spending power and raise demand for lower-level housing submarkets. When immigrants arrive with savings or earn income outside the country, they also de-couple the relationship between local housing and labour markets. There were likely always people whose housing consumption was supported by means other than their labour market participation. However, the scale and changing profile of immigrants, and also the increasing cost of housing in cities where real estate markets are seen as investment tools (Blomley, 2004; Hackworth, 2005), reshape the relationship between income and housing consumption when measured at the local scale. Wealthy migrants who buy directly into ownership markets keep housing markets afloat without necessarily resulting in corresponding increases in local labour market incomes, whereas the domestic labour force is spending higher shares of their income on housing either for investment reasons or to pay for particular housing locations, which include the rising cost of living in an expanding city. The result at the neighbourhood level is that the socioeconomic characteristics no longer correspond as closely with investment and construction of housing at different stages of the neighbourhood cycle as assumed in traditional models of neighbourhood transition (Bourne, 1981). The traditional way of thinking about neighbourhood 101 transition was based on an economic theory of blight-decay-abandonment cycles, building on theories of succession (Park, Burgess & McKenzie, 1925; Hoyt, 1938). The models equate neighbourhood desirability with higher housing values. Expanding areas benefit from increasing values while older areas stabilize and later exhibit declining values as the housing stock ages and “filters down”. Due to gentrification and rehabilitation, models began to allow for interruptions in the depreciation cycle (Lowry, 1960; Metzger, 2000). The continuing social upgrading of central city neighbourhoods and the “entrenchment of wealth” in elite neighbourhoods counters the assumption of an assumed ‘natural’ cycle of neighbourhood change that sees decline as somehow inevitable (Ley, Tutchener & Cunningham, 2001). “Filtering” is not occurring, at least not in Canadian cities, and is arguably working to increase housing affordability burdens (Skaburskis, 2006a).  In the case of immigrants, neighbourhood change was traditionally studied through the invasion of inner-city reception points and subsequent succession into established neighbourhoods; but it is now commonly understood that neighbourhood change needs to be viewed in a broader spectrum of immigrant experiences (Mark & Goldberg, 1985; Ley & Murphy, 2001). Today, wealthier immigrants often move directly into ownership markets and can raise demand for housing locally, especially if they display strong locational preferences to form ethnic communities, which is particularly the case in Canada in Vancouver and Toronto (Logan, Alba & Zhang, 2002; Hou & Milan, 2003; Ley, 2003b, 2010; Hou, 2004). The changes alter the ways neighbourhoods transition as the linkages between income and housing stock changes become strengthened on the one hand through gentrification but de-coupled on the other 102 by influx of savings and income through immigrants.  3.1.1 The changing profile and settlement patterns of immigrants  The demographic profile of immigrants has changed across Canada over the past 30 years due to shifts in immigration policy that influence the location and housing patterns (Ley & Murphy, 2001). The Canadian Immigration Act of 1967 introduced a point system and business migration program, which was expanded in 1978 and 1986. Whereas people were still admitted under humanitarian (refugee) and family reunification criteria, the majority of immigrants in recent years have been admitted under the points system (Hoering & Walton-Roberts, 2006). In particular, Chinese and Hong Kong immigrants to Vancouver tended to have higher educational levels, occupational skills, and financial assets than immigrants in the past and those moving into other Canadian cities such as Montreal (Hiebert, 1999; Hou, 2004). Migration to Vancouver was partly triggered by the 1997 return of Hong Kong from British to Chinese rule (Badcock, 2002). Vancouver became one of the main destinations for business migrants to Canada. “The combined figures from Hong Kong and Taiwan amounted to some 63,000 business immigrants between 1980 and 2000, or 63.5% of the BC total” (Ley, 2003b, p. 430). Only about 5% of those arriving in Vancouver after 1980 were refugees. Vancouver, valued for its quality of life, became a commuter suburb for wealthy transnational workers from the Asian Pacific Rim (Crossette, 2001, p. A6; Ley, 2010).37 Arriving with established wealth, and commonly continuing to earn 37 The increase in wealthy immigrants and their transnational lives have been documented in the academic literature (Mitchell, 2004), highlighting the conditional affiliation with a single national 103 income outside the country, wealthy migrants to Vancouver can support much higher levels of housing consumption than their locally earned income would suggest.  In contrast, among the three largest Canadian metropolitan areas Montreal receives the lowest share of immigrants, “the largest share of refugees”, “the least affluent immigrants” and the largest continuing influx of European immigrants such as from “France and its former colonies…Haiti and the [North African] countries of the Maghreb” (Hiebert et al., 2006, p. 2). More than 80 percent of recent immigrants to Montreal are renters, yet all immigrants “are more likely to be homeowners than Canadian-born Montrealers, though the gap is closing”, particularly among the young (Rose et al., 2006, p. 11). Since there are fewer recent immigrants and even fewer enter the ownership market, the effects on dwelling values have not traditionally been a point of discussion in Montreal as they have been in Vancouver. The settlement patterns of recent immigrants also vary between Vancouver and Montreal. In Vancouver, the suburbanization of jobs, more widespread gentrification of the inner city and increasing diversity of immigrants contributed to the suburbanization of the immigrant social landscape (Balakrishnan & Kralt 1987; Hiebert, 1999; Anisef & Lanphier, 2003). In Montreal, all immigrants, even as homeowners, are “largely concentrated in the center of the region, that is to say on the Island of Montreal and in adjacent parts of the South Shore and Laval” (Rose et al., 2006, p. 8). Whereas low-income immigrant groups still reside in Vancouver’s inner city (Hou & Milan, 2003), most increases in the immigrant jurisdiction among “flexible citizens” (Ong, 1999). The experiences of “satellite kids” and “astronaut families” in Vancouver have received particular attention (Waters, 2003; Kobayashi & Preston, 2007). Some immigrant families who bought houses in Vancouver return to their home country but their “satellite kids” remain in Canada to attend school. In the case of the “astronaut family”, only the husband returns to work in Asia shortly after the family immigrates to Canada. 104 population in Vancouver’s inner city since the 1980s involved wealthier, often Chinese, immigrants moving directly into ownership markets in existing middle- to upper- income neighbourhoods (Laryea, 1999; Li, 1994).  However, a word of caution is warranted: the examination of ownership markets, particularly in Vancouver, portrays a somewhat romanticized view of immigrants’ ability to attain housing. Integration remains a slow process because many recent immigrants often move into rental housing (Fiedler, Schuurman & Hyndman, 2006; Mendez, Hiebert & Wyly, 2006), while the arrival of wealthy business migrants that displace former local residents in Vancouver (Mitchell, 2004) contrasts with those needing to “double up” and live in overcrowded conditions due to severe affordability burdens that are increasing for recent immigrants in Montreal and Vancouver (Jakubec, 2004). It thus must be remembered that the analysis in this chapter focuses on the determinants of demand for owner-occupied housing during a period when the migration of skilled and wealthy business migrants dominated immigrant flows to Vancouver (Ley et al., 2001; Hou, 2004). The trends are being compared to Montreal where ownership markets constitute a smaller share of the total housing stock, fewer immigrants arrive and an even smaller proportion of recent immigrants are homeowners (Rose et al., 2006).  3.2 Measuring Housing Demand and Neighbourhood Transition  The aim of this analysis is to ask in general how housing consumption is related to the characteristics of the household in the Montreal and Vancouver metropolitan areas in two time periods. Of specific interest is how the relationship between income 105 and housing differs between recent immigrants and the rest of the population. The hypothesis is that due to the changing profile of immigrants to Vancouver, recent immigrants identified in the 2001 census would allocate a smaller share of their locally earned income to their housing consumption than is the case for local residents, suggesting a de-coupling of local housing and labour markets. The analysis tests the reverse argument that the difference between recent immigrants and the local population should not have been visible in the early 1980s. The attribution of changing relationships to an influx of business migrants in Vancouver is tempered by a comparison to Montreal that has received fewer wealthy migrants. But nonetheless the impacts of the increasing skill requirement of immigrants could bring similar changes to the relationship between income and housing in Montreal but have lower aggregate impact in a city where ownership markets provide a much smaller component of the overall housing stock. The investigation turns to the neighbourhood scale to shed light on how changes in housing values are linked to changes in the socioeconomic composition, again focusing on the presence of recent immigrants.   The first task of the analysis is to develop multivariate regression models of housing consumption to discern the relationship between a household’s income (labour market participation) and housing demand. The model controls for numerous factors affecting housing consumption and distinguishes recent immigrants (estimated to be those who arrived within 20 years before census administration) from the rest of the population. In the 2001 census, the recent immigrants account for 24 percent of Vancouver’s metropolitan population compared to 14 percent of Montreal’s. Migrants are also distinguished by place of birth, but only in broad categories (Canada, Europe, 106 Asia, other) available in a temporal analysis comparing censuses 20 years apart. Variables affecting housing consumption are linked to the variable denoting recent immigrants to examine differences in housing consumption. Of most interest will be the interaction between the immigration variable and the household’s income. The regressions show differences in housing consumption between recent immigrants and the rest of the population with otherwise similar household characteristics, living in similar types of dwellings. Separate regression equations are also constructed for recent immigrants and the local population to compare the magnitude and direction of resulting coefficients.   The user cost of housing is derived from the Statistics Canada PUMFS and used as a measure of housing consumption that recognizes the dual functions of housing as shelter and investment. It more directly reflects the costs of homeownership than dwelling value or resale price and accounts for the impact of changing interest rates on the annual cost of owning (Pozdena, 1988; Himmelberg, Mayer & Sinai, 2005). The user cost of housing is usually calculated as the sum of the opportunity cost of capital, property taxes, maintenance costs, and risk premium of owning versus renting minus expected capital gains and applicable tax deductions. Given the data availability of the PUMFS, the user cost of housing is only roughly approximated for each census year as follows:   User cost of housing = Property Taxes + Utilities + (Dwelling Value) × (Interest rate + Risk Premium + Depreciation Rate – Expected Capital Gain).  107 Property taxes and utilities are reported in the census as “owner’s major payments”.38 Dwelling value is the dollar amount expected by the owner if the dwelling were to be sold. A long-term, low-risk interest rate is used to account for the opportunity costs of owning a home (Himmelberg et al., 2005). The interest rate is set to the 1980 and 2000 yearly average treasury rates of 12.8 and 5.7 percent, respectively, available from the Bank of Canada. All dollar amounts are adjusted for inflation using Bank of Canada rates to 2001 dollars. The risk premium of owning a home and the depreciation rate (or maintenance costs) are set at 4.0 and 3.0 percent, respectively (Green & Hendershott, 1993). Expected capital gains are estimated by the long-term, real appreciation rates of Canada’s 13 largest urban housing markets from the 1981 to the 2001 census (4.4 percent).  The user cost of housing is then estimated in additive form as a function of the following independent variables: the household’s long-run income potential, temporary (transitory) income, proportion of income derived from investments, household size, age of primary maintainer, gender of primary maintainer, and presence of children. The variables take into account several household characteristics that influence the consumption of housing. Independent variables describing the physical attributes of the home (dwelling type, period of construction, number of rooms, and repair status) are also included so the coefficients on demographic or income variables measure the impact on housing demand for a similar amount and quality of housing (Green & Hendershott, 1993). 38 The “owner’s major payments” variable includes mortgage payments unless the household owns the home out- right. Including mortgage payments incorrectly inflates the user cost of housing. Hence, for households with mortgage payments, taxes/utility payments were predicted using a regression model that relates tax/utility costs to the characteristics of the home. 108 The second task of the chapter is to examine whether immigration alters the traditional depiction of neighbourhood transition where changes in dwelling value parallel the socioeconomic status of a neighbourhood. This task is tackled by way of a multivariate regression model that estimates change in average census tract dwelling value from 1981 to 2001 as a function of independent variables describing changes in the tract’s housing stock and socioeconomic composition. A variable on household income is included, expecting a positive relationship with changes in dwelling value as expected by our traditional understanding of neighbourhood transition. Variables describing the proportion of recent immigrants residing in a tract test for the association between immigration and dwelling value, and interaction effects with income reveal whether the relationship between change in dwelling value and income is mediated by immigration.  Variables measuring dwelling density, proportion of the stock single-family dwellings, and proportion of the stock built before 1946 control for the changing physical characteristics of census tracts that impact dwelling value. A variable measuring the distance from the census tract to the CBD, and also CMA zones, analyzes value changes in relation to the tract’s location in the larger urban context. The categorization by CMA zones—inner city, old suburbs, new suburbs, exurbs—has been previously utilized to divided Canadian urban regions into areas by their “dominant period of urban development” (Filion & Bunting, 2010; Walks, 2001; Skaburskis & Moos, 2008; Figures 3.2 & 3.3).   109 Figure 3.2 – Montreal CMA zones  Source: Calculated using Statistics Canada census data and shape files (1981a).     4 Zones Inner city Old suburbs New suburbs Exurbs 0 10 205 Kilometers Laval Montreal LaSalle Longueuil Vaudreuil- Dorion Brossard Mercier Varennes Richelieu Mont- Saint-Hilaire Repentigny Terrebonne Blainville Saint- Eustache Saint- Philippe Montreal- Est Dorval Fl eu ve  Sa in t-L au re ntl110 Figure 3.3 – Vancouver CMA zones  Source: Calculated using Statistics Canada census data and shape files (1981a).    3.3 Changing Determinants of Housing Demand The analysis is particularly concerned with the changing importance of income in determining housing demand due to changes in the labour force arising from immigration. Several other household-level variables that influence housing demand require inclusion in the multivariate models to account for their effect on the user cost of housing (Green & Hendershott, 1993; Himmelberg et al., 2005). Tables 3.1 and 3.2 show the variables included in the model describing the characteristics of households in the 1981 and 2001 census data. Included are only the households residing in owner- 0 5 10 15 202.5 Kilometers 4 Zones Inner City Old Suburbs New Suburbs Exurbs VANCOUVER CBD l BURNABY RICHMOND WEST VANCOUVER NORTH VANCOUVER DELTA SURREY COQUITLAM LANGLEY Burrard Inlet S tr ai gh t o f G eo rg ia Canada-USA Border MAPLE RIDGE111 occupied dwellings. Permanent and temporary incomes are estimated as described in Chapter Two.39 The household income can be influenced by the decisions regarding housing expenditure as households may work more to afford more expensive housing (Kohlhase, 1986), and as discussed above, immigration status also affects labour market outcomes. Thus, the permanent income is estimated without inclusion of the immigration variable or information regarding housing. This permits the use of the permanent income variable in the estimation of housing demand to determine if differences in the income prospects of households affect housing consumption differently for recent immigrants than for the rest of the population. The effect of monetary income shows how housing demand varies with participation (and earnings) in the local labour market. Real permanent income has increased for recent immigrants over the 20-year time frame in Montreal and Vancouver (Tables 3.1 & 3.2), reflecting the growing emphasis of immigration policy on educated and skilled migrants.    39 Separate models are estimated using the 1981 and 2001 PUMFS data, including all household records, except those reporting negative incomes (output not shown). The number of cases and adjusted r-squared values are 31,166 and 0.305 in the 1981 model and 106,555 and 0.243 in 2001. With the exception of select dummy variables identifying census metropolitan areas, the p-values are less than 0.0001 for the regression coefficients in both years. The direction of coefficients remains the same in both years with male maintainers, more earners, older age, higher order occupations and higher levels of schooling positively associated with household income. The magnitude of the coefficients show the impact of gender on income declining and the impacts of education and higher occupational status increasing from 1981 to 2001. The implications of these findings are discussed further in Chapter Four. 112 Table 3.1 – Means of housing and household characteristics of owners, Montreal CMA 2001 and 1981  Recent Immigrants Rest of the Population  1981 2001 1981 2001      User Cost of Housing ($1,000) 33.939 18.395 29.267 16.741 Dwelling Value ($1,000) 168.136 158.764 140.104 141.514 Average Number of Rooms 6.705 6.734 6.458 6.793 Single-Family Dwelling .452 .440 .612 .608      Dwelling Built Pre-1946 .114 .104 .159 .108 Dwelling Built 1946 – 1980 .858 .555 .818 .526 Dwelling Built 1981 – 2001 .028 .342 .023 .366      Only Regular Maintenance .753 .548 .769 .643 Minor Maintenance Required .174 .335 .175 .297 Major Maintenance Required .073 .116 .057 .061      Permanent Income ($1,000) 60.617 70.951 55.753 67.209 Temporary Income ($1,000) 10.051 -4.109 14.171 15.255 Prop. Investment Income .043 .044 .074 .049 Household Size 4.178 3.743 3.463 2.739 Maintainer Age 43 50 48 56 Dependent Children Present .740 .625 .528 .353 Maintainer Female .052 .274 .166 .324      Maintainer Place of Birth     Canada - - .858 .838 Asia .155 .381 .002 .020 Europe .632 .308 .133 .109 Other .213 .311 .007 .033 Notes: All dollar values adjusted for inflation to $2001 using the Bank of Canada consumer price index. Recent immigrants: Immigrated to Canada within 20 years before census enumeration. Source: Calculated using Statistics Canada household PUMFS (1981b; 2001a).        113 Table 3.2 – Means of housing and household characteristics of owners, Vancouver CMA 2001 and 1981  Recent Immigrants Rest of the Population  1981 2001 1981 2001      User Cost of Housing ($1,000) 63.044 30.689 60.659 28.993 Dwelling Value ($1,000) 366.831 309.571 352.900 291.783 Average Number of Rooms 6.926 6.608 6.716 6.955 Single-Family Dwelling .861 .572 .848 .619      Dwelling Built Pre-1946 .181 .046 .212 .099 Dwelling Built 1946 – 1980 .769 .292 .741 .429 Dwelling Built 1981 – 2001 .050 .662 .047 .472      Only Regular Maintenance .789 .703 .807 .699 Minor Maintenance Required .183 .214 .154 .227 Major Maintenance Required .028 .084 .040 .074      Permanent Income ($1,000) 64.712 73.991 53.689 71.426 Temporary Income ($1,000) 14.232 -7.022 18.557 21.268 Prop. Investment Income .054 .108 .100 .066 Household Size 4.279 3.808 3.041 2.657 Maintainer Age 42 51 51 58 Dependent Children Present .704 .581 .401 .301 Maintainer Female .074 .253 .242 .335      Maintainer Place of Birth     Canada - - .733 .680 Asia .424 .799 .032 .106 Europe .435 .113 .223 .174 Other .139 .088 .012 .041 Notes: All dollar values adjusted for inflation to $2001 using the Bank of Canada consumer price index. Recent immigrants: Immigrated to Canada within 20 years before census enumeration. Source: Calculated using Statistics Canada household PUMFS (1981b; 2001a). Modified and reproduced with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved.     114 It is interesting to note that recent immigrants already had higher income prospects than the local population in the 1981 census, and the difference vis-à-vis the rest of the population actually decreased by the 2001 estimates. In other words, while it is certainly true that the share of skilled and well-educated migrants to Canada increased since the shift to the points system in immigration policy, recent immigrants already exceeded the rest of the population with respect to their earning potential in 1981. The negative temporary income of recent immigrants shows that their monetary earnings in the labour market in 2001 are below their earning potential. This income penalty can result from non- acceptance of foreign professional accreditation as labour market integration remains a slow process in Canada so that immigrants earn less than their labour market profiles would predict (Jakubec, 2004).  The negative temporary income is also in line with the arrival of wealthier migrants who are earning little or no income in the local labour market (Ley, 2003b; 2010), and the absolute magnitude of the negative monetary income is higher in Vancouver, the city that received the largest relative influx of Asian migrants. Table 3.2 also shows the proportion of income received from investments, which are another indication of the increasingly wealthy migrants arriving in Vancouver. It is the only crude measure of household wealth available in census data. Household wealth plays a potentially crucial role in housing consumption. Recent immigrants in Vancouver derive more than 10 percent of their income from investment in 2001, almost twice as much as for the recent immigrants in 1981 and higher than for the rest of the population whose share of income derived from investment decreased from 10 to 7 percent. In Montreal, the percent of income recent immigrants earn from investment is lower than in 115 Vancouver and remained relatively constant from 1981 to 2001 (Table 3.1).   The variables measuring user cost of housing (and dwelling value) show that recent immigrants consume more housing than the rest of the population in Montreal and Vancouver, a finding also of the 1981 census. Due to high interest rates in the early 1980s, the user cost of housing actually decreased from the 1981 to 2001 census. The value of dwellings also declined in real terms in Vancouver due to a housing bubble that inflated prices leading up to the 1981 census (Skaburskis, 1988). In Vancouver, recent immigrants identified in the 1981 census were more likely to live in single-family dwellings with more rooms compared to the local population. The share of households in single-family dwellings and their average number of rooms decreased from the 1981 to the 2001 census, which relates to the higher-density development patterns facilitated by higher prices and the urban growth management strategies discussed in Chapter Two. In Montreal, the recent immigrants are less likely to reside in single-family dwellings than the rest of the population in both census years, and the change over time is not different from zero at a statistically significant level. It is known that the difference in single-family dwelling occupancy between immigrants and the local population has traditionally disappeared with length of stay in Canada (Mendez et al., 2006). The lower share of recent immigrants in single-family dwellings and the relatively higher proportion in older dwellings in Montreal reflects the more centralized location patterns of immigrants in this metropolitan area. In Vancouver, the recent immigrants have increasingly come to reside in newer buildings in the expanding suburban fringe.   The comparison of household characteristics between Montreal and Vancouver 116 reflects these differences in the transformations in the owner-occupied housing stock and the immigrant social landscape. The percentage of the rest of the population residing in single-family dwellings declined in Vancouver so that the magnitude in 2001 is almost the same as in Montreal, whereas the percent of recent immigrants living in single-family dwellings remains higher in Vancouver where the immigrant residential location patterns have become more suburbanized. Reflecting the differences in dwelling type, household size of the rest of the population is smaller in Vancouver than in Montreal whereas the recent immigrant households are on average larger in size in Vancouver than in Montreal. Household size is included in the regression as larger households will have greater expenditures on food, clothing, and other goods, thus leaving less to be spent on housing (Ho & Chiu, 2002). Household size is lower for recent immigrants and the local population in 2001 as compared to their counterparts in 1981.  Age is included in the analysis as it is expected to be positively related with housing consumption. Older maintainers had more time to accumulate equity. Maintainers are generally older in 2001 than in 1981 due to the aging of the population. The recent immigrants are younger than the rest of the population in both years. The presence of children could increase housing consumption as households might pay a premium for housing with yards or to reside closer to desirable schools, although the presence of children also reduces income available to spend on housing (Murdie et al., 1999). Recent immigrants are more likely to have children in their home than the rest of the population, and the Montreal households are more likely to have children than those in Vancouver. Gender is also associated with housing consumption decisions, partly due 117 to differences in income and household size (Skaburskis, 1997). The proportion of females who are household maintainers is lower among recent immigrants than the rest of the population but the difference becomes smaller by 2001.  3.3.1 Multivariate analysis of the user cost of housing Table 3.3 shows the regressions with the user cost of housing as a function of housing and household characteristics in 2001. The regressions pool all owner households, and interaction terms are included to identify the recent immigrants. Interaction terms were removed from the model if they were not statistically significant40. The standardized beta coefficients permit comparison of the relative contribution of each independent variable in the prediction of the user cost of housing. Table 3.4 shows the regressions constructed using the 1981 data. For brevity, only selected variables are shown. As expected, the regressions show that households with older maintainers and the presence of children are associated with a higher user cost of housing. Dwellings with more rooms and in better state of repair command a higher user cost of housing in both metropolitan areas. Also, housing built before 1946 has positive coefficients, showing that older housing primarily located in the inner city is more highly valued due to gentrification (Filion & Bunting, 1990; Skaburskis, 2006a), whereas the areas built-up between 1946 and 1980 that comprise the old post-war suburbs are devaluing (Skaburskis & Moos, 2008).   40 Comparison of coefficients in the models constructed for recent immigrants and the rest of the population separately reveals the same overall conclusions as those presented here. 118 Table 3.3 – User cost of housing as a function of housing and household characteristics  Montreal CMA Vancouver CMA  Coeff. Beta  Coeff. Beta  Recent Immigrants 3.614 .096 ** 2.261 .080 * Average Number of Rooms 1.514 .283 *** 1.670 .291 *** Single-Family Dwelling -1.918 -.101 *** 4.253 .167 ***   ×Recent Immigrants 1.165 .022 * 1.697 .049 **        Dwelling Built Pre-1946 3.189 .101 *** 3.840 .092 *** Dwelling Built 1946 – 1980 -1.209 -.067 *** -1.966 -.081 *** Dwelling Built 1981 – 2001 Base             Only Regular Maintenance Base      Minor Maintenance Required -.790 -.041 *** -.853 -.031 ** Major Maintenance Required -.956 -.025 ** -2.968 -.061 ***        Permanent Income ($1,000) .104 .272 *** .091 .205 ***   ×Recent Immigrants -.047 -.095 *** -.063 -.180 *** Temporary Income ($1,000) .059 .237 *** .042 .150 ***   ×Recent Immigrants -.026 -.026 ** -.041 -.065 *** Prop. Investment Income 12.380 .131 *** 9.159 .091 *** LN(Household Size) -3.029 -.116 *** -2.022 -.065 ***   ×Recent Immigrants 1.846 .068  3.964 .199 *** Maintainer Age .073 .085 *** .134 .120 *** Dependent Children Present 1.657 .092 *** 1.710 .072 ***   ×Recent Immigrants -1.481 -.034  -2.251 -.069 ** Maintainer Female .037 .002  .131 .005  Constant -1.114   2.926  **        N-cases 12410   7492   R^2 .292   .328   Notes: Table only includes households in owner-occupied dwellings. Natural log of variables used when it improved model fit. Recent immigrants: Immigrated to Canada within 20 years before census enumeration. *p<0.05, **p<0.01, ***p<0.001. Source: Calculated using Statistics Canada household PUMFS (1981b; 2001a). Modified and reproduced with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved.  One key difference between metropolitan areas is that the user cost of housing is higher for single-family dwellings as compared to other dwelling types in Vancouver whereas this effect is reversed in Montreal. In Vancouver, the single-family dwellings cost more than other dwelling types, holding constant the number of rooms, repair 119 status and period of development. Since the variable identifying number of rooms roughly controls for the effect of size, the coefficient for single-family dwellings in Montreal is likely detecting the lower per unit cost of single-family housing in the suburbs as compared to the higher density units in central areas. It detects the impact of location on cost not accounted for in the model, and reflects the decentralized nature of the single-family dwelling stock in Montreal. It should be noted that Montreal is somewhat unique in the Canadian context in terms of the lower user cost of housing for single-family dwellings as the coefficient has a positive sign when the regressions are constructed for other metropolitan areas in Canada. Single-family dwellings are often thought to reap a premium due to their location in established neighbourhoods, however in the case of Montreal that has a much smaller share of single-family housing the premium appears to be simply due to size.   An important finding to note in terms of the aims of this analysis is that recent immigrants in 2001 exhibit higher housing consumption than similar households living in similar types of dwellings in Vancouver and Montreal (Table 3.3), but that this was already true in the early 1980s in Vancouver (Table 3.4). In fact, comparing the beta coefficients for the variable denoting recent immigrants in Vancouver reveals that immigrant status contributed more to the prediction of housing consumption in the 1981 data, before globalization was generally described as a noteworthy force of urban change. To some extent this is not too surprising given that immigrants who arrived between 1981 and 2001 would have encountered a more expensive market and a higher- density urban setting, reducing the difference in housing consumption between immigrants and the rest of the population with similar households. In Montreal, the 120 recent immigrants who are owners did not have a higher user cost of housing in 1981 but by 2001 the magnitude of the coefficient is actually larger than in Vancouver.   Table 3.4 – Effect of income and immigration status on the user cost of housing, 1981 and 2001 Montreal CMA 1981 Census 2001 Census  Coeff. Beta  Coeff. Beta  Recent Immigrants 3.619 .079  3.614 .096 ** Permanent Income ($1,000) .103 .270 *** .104 .272 ***   ×Recent Immigrants -.049 -.074 * -.047 -.095 *** Temporary Income ($1,000) .089 .237 *** .059 .237 ***   ×Recent Immigrants .043 .033 * -.026 -.026 ** Prop. Investment Income 2.553 .036 ** 12.380 .131 *** Constant 11.212  *** -1.114          N-cases 4670   12410   R^2 .300   .292   Vancouver CMA 1981 Census   2001 Census    Coeff. Beta  Coeff. Beta  Recent Immigrants 11.114 .249 *** 2.261 .080 * Permanent Income ($1,000) .073 .157 *** .091 .205 ***   ×Recent Immigrants -.036 -.061  -.063 -.180 *** Temporary Income ($1,000) .040 .101 *** .042 .150 ***   ×Recent Immigrants -.016 .017  -.041 -.065 *** Prop. Investment Income 5.764 .069 *** 9.159 .091 *** Constant 31.417  *** 2.926  **        N-cases 3079   7492   R^2 .248   .321   Notes: Table only includes households in owner-occupied dwellings. Regressions include control variables describing housing stock and household characteristics. Recent immigrants: Immigrated to Canada within 20 years before census enumeration. *p<0.05, **p<0.01, ***p<0.001.  Source: Calculated using the Statistics Canada household PUMF (1981b; 2001a). Modified and reproduced with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved.  The regression results in Table 3.3 suggest that long-term income prospects rather than their current earnings (temporary income) explain recent immigrants’ housing consumption in Vancouver. The coefficient for the interaction effect between recent immigrants and temporary earnings is negative and similar in magnitude to the 121 effect of the temporary income variable so that when the regressions are constructed solely for recent immigrants, the effect of the temporary income is not different from zero at a statistically significant level. These findings point to a de-coupling of housing from local labour markets. Table 3.4 repeats the analysis using 1981 data and shows that this de-coupling effect was not visible in the aggregate data before Vancouver experienced globalization of its real estate market and an influx of more skilled and wealthy migrants. However, in both metropolitan areas the permanent and temporary incomes contribute less to the prediction of recent immigrants’ housing consumption than they do for the rest of the population. But in Montreal the effect of temporary income on housing consumption was actually larger in magnitude for recent immigrants in 1981 than for the rest of the population. In the 2001 data the temporary income variable still does have an effect on housing in Montreal albeit less so than for the rest of the population.   Thus, recent immigrants have a lower income elasticity of housing demand in 2001, meaning that they are less responsive to income changes in their housing consumption than the rest of the population. In Vancouver, the difference is almost completely absent in 1981. At least one explanation is that recent immigrants today, other things being equal, tend to arrive with established wealth or continue to earn income outside the country. This renders them less likely to make out-of-pocket expenses on housing from locally derived income. Because recent immigrants actually reside in more expensive dwellings than the rest of the population and have a higher user cost of housing (Table 3.1), the results indicate that they likely have “invested 122 more heavily in equity” (Hiebert et al., 2008, p. 40). The occurrence of this “immigration effect” has been previously documented by Hiebert et al.:   “immigrants draw on wealth and/or income transferred to Canada or save high proportions of their income in Canada as renters (or both) in order to purchase housing quickly and with lower mortgage payments.”  The regressions also include the share of income earned from investments (Table 3.3). The beta coefficient is larger for this variable when the regressions are only constructed for recent immigrants as compared to regressions for the rest of the population but in the combined data an interaction effect is not statistically significant. The effect of investment on the user cost of housing is higher in the 2001 data than in 1981, hinting at the increasing importance of household wealth in supporting housing consumption. The magnitude of the coefficient is larger in Vancouver in 1981 but larger in Montreal by 2001. Housing prices were escalating quickly in the early 1980s in Vancouver (Skaburskis, 1988), which may have inclined those with investment income to increase their housing consumption. Since the ownership market is much smaller in Montreal than in Vancouver but has historically been an important source of investment income for owners who reside in and also rent multi-unit dwellings (Choko & Harris, 1990) investment income could be expected to be a more important predictor of housing consumption in Montreal than in other metropolitan areas as is the case in the 2001 data.  The negative sign for household size indicates that as household size increases, more must be spend on other expenditures, leaving less to be spent on housing (Ho & Chiu, 2002). Yet for recent immigrants in Vancouver, household size has a positive sign. This can be explained, at least in part, by the larger number of employment income 123 recipients in immigrant households and the larger share of South Asian immigrants, which have a strong cultural tradition of living with extended family, than in Montreal (Rose et al., 2006). As expected, the presence of children increases housing consumption. However, the effect is negative in the case of recent immigrants in Vancouver. One explanation is that immigrant households with children reduce housing consumption due to economic necessity—although the variable does not show a statistically significant coefficient when the regression is constructed only for the recent immigrants. The positive coefficient for the variable identifying single-family dwellings confirms the commonly held notion that immigrants, as compared to similar households, place greater value on single-family living for cultural and status reasons (Murdie et al., 1999). One important variable that has been missing from the analysis thus far is place of birth. Tables 3.5 and 3.6 repeat the regression analysis by place of birth. Asian, European, and Other migrants are compared to the rest of the population, in each case excluding the remainder of the recent migrants. The results show that the findings in the previous regressions for Vancouver are due to those of Asian migrants. As shown in Table 3.1, the profile of recent immigrants has shifted from European migrants to those from Asia, the latter dominating immigrant flows to Vancouver since the 1980s. It is also primarily Asian migrants who have been documented to arrive with significant amounts of equity.        124 Table 3.5 – Effect of income and immigration status on user cost of housing by place of birth, Montreal Montreal CMA       Regression Coeff. Asia  Europe  Other  1981 Census       Recent Immigrants 6.073  7.075 * -9.773 * Permanent Income ($1,000) .104 *** .103 *** .104 ***   ×Recent Immigrants -.048  -.080 * -.022  Temporary Income ($1,000) .089 *** .089 *** .089    ×Recent Immigrants .007  .052 * .082 * N-Cases 4297  4504  4325  R^2 .299  .301  .302  2001 Census       Recent Immigrants 5.032 * 2.627  3.508  Permanent Income ($1,000) .104 *** .104 *** .104 ***   ×Recent Immigrants -.060 ** -.041  -.051 * Temporary Income ($1,000) .059 *** .059 *** .059 ***   ×Recent Immigrants -.027 * .007  -.058 *** N-Cases 11950  11888  11886  R^2 .296  .299  .296  Notes: Table only includes households in owner-occupied dwellings by place of birth of recent immigrants compared to the rest of the local population. Regressions include control variables describing housing stock and household characteristics. Recent immigrants: Immigrated to Canada within 20 years before census enumeration. *p<0.05, **p<0.01, ***p<0.001 Source: Calculated using the Statistics Canada household PUMF (1981b; 2001a).   In Montreal this effect is also visible even though fewer wealthy Asian migrants have been documented to arrive: Since there were fewer wealthy migrants, the decrease in the effect of monetary earnings on housing consumption among certain immigrant groups in Montreal necessarily relates at least partly to their higher propensity to save more equity before entering into homeownership rather than purely the transfer of wealth from outside the country. However, due to the smaller size of the ownership market in Montreal the recent immigrants who are owners would inherently be wealthier than the rest of the population so that the influx of capital in de-coupling housing from labour market still serves as an adequate explanation; yet, any aggregate 125 effects of immigration are less visible (Figure 3.1) due to the smaller scale of immigrant flows to Montreal and fewer other inflationary pressures on the housing market.  Table 3.6 – Effect of income and immigration status on user cost of housing by place of birth, Vancouver Vancouver CMA       Regression Coeff. Asia  Europe  Other  1981 Census       Recent Immigrants 12.862 ** 4.392  19.866 ** Permanent Income ($1,000) 0.073 *** 0.072 *** 0.073 ***   ×Recent Immigrants -0.043  -0.019  -0.037  Temporary Income ($1,000) 0.04 *** 0.038 *** 0.039 ***   ×Recent Immigrants 0.002  -0.037  -0.007  N-Cases 2,802  2,814  2,675  R^2 0.252  0.254  0.255  2001 Census       Recent Immigrants 3.018 * -1.172  4.498  Permanent Income ($1,000) 0.091 *** 0.089 *** 0.088 ***   ×Recent Immigrants -0.076 *** 0.006  0.005  Temporary Income ($1,000) 0.042 *** 0.042 *** 0.042 ***   ×Recent Immigrants -0.044 *** 0.005  -0.028  N-Cases 7,114  6,012  5,976  R^2 0.32  0.333  0.328  Notes: Table only includes households in owner-occupied dwellings by place of birth of recent immigrants compared to the rest of the local population. Regressions include control variables describing housing stock and household characteristics. Recent immigrants: Immigrated to Canada within 20 years before census enumeration. *p<0.05, **p<0.01, ***p<0.001 Source: Calculated using the Statistics Canada household PUMF (1981b; 2001a). Modified and reproduced with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved.  The commonly held assumption of successive housing careers is largely based on European immigrant experiences, and also on immigrants with lower socioeconomic status regardless of ethnicity (Hiebert et al., 2008). Between 1960 and 1980, European and other migrants were commonly working in manual or trades occupations. The migrants used their ability to excel in the local labour market to eventually move up the housing ladder. It was their locally earned income that supported their housing 126 consumption. This effect is evident in Montreal where recent immigrants of European or Other origins even allocated a higher share of their monetary income to housing than the rest of the population in 1981, but this affect disappeared in 2001 for European immigrants and reversed for those of Other origins. The new story, particularly in Vancouver, is one based on Asian migration where income and housing demand correspond to a lesser extent for recent migrants who are increasingly arriving with wealth and continue to earn income outside the country. Again, obviously immigrant experiences have always been and remain varied, but the aggregate picture shows a de- coupling of housing from local labour markets for recent Asian immigrants. Interestingly, for the rest of the population (and also immigrants from European and Other origins) the magnitude of the beta coefficients for the income variables increased from 1981 to 2001. In other words, with the exception of Asian migrants, income has become more important in explaining housing consumption.   3.4 Neighbourhood Transition  Computing a simple Pearson correlation between changes in average household income and dwelling value at the census tract level from 1981 to 2001 reveals a positive correlation of 0.334 in Vancouver and 0.443 in Montreal. Given the multitude of factors acting on dwelling value, this admittedly modest correlation nonetheless indicates that changes in neighbourhood housing values are linked to changes in socioeconomic status as predicted by the models of neighbourhood transition discussed above. The correlation increases to 0.442 in Vancouver and 0.503 in Montreal when only the census tracts with lower than CMA average concentrations of recent immigrants are 127 considered; it decreases to 0.260 in Vancouver and 0.406 in Montreal when computed for the census tracts with above-average concentrations of recent immigrants. The correlations suggest a de-coupling of housing from local labour markets at the neighbourhood scale too but evidently this effect is more pronounced in Vancouver. The analysis below examines this observation in greater detail using the census tract data. First, however, it is necessary to establish in more general terms the changes in the characteristics of the housing stock and households across the CMA zones.   3.4.1 Changing housing stock characteristics  All households, including immigrants, are making decisions about housing within changing cities and housing markets (Hiebert, 1999). Thus, it is important to analyze the housing consumption in the context of the characteristics of the housing stock. There are both commonalities and differences between how the societal changes discussed in Chapter Two materialized in the housing stock in Montreal and Vancouver. The historical differences as well as the contemporary societal transformations occurring since the late 1970s and early 80s are reflected in the changing characteristics of the housing stock. Table 3.7 shows the changes by the CMA zones—inner city, old suburbs, new suburbs and exurbs. The increases in the number of units are highest in the new suburbs and indicate that suburbanization continues to be an important dimension of urban growth in the post-Fordist period in Vancouver and Montreal. However, clearly visible is the large increase in the number of units in Vancouver’s 128 inner city, at over 68,000 units from 1981 to 2006 it is more than double the increase in Montreal41.   Table 3.7 – Changing housing stock characteristics  Montreal Vancouver  2006 2006-1981 2006 2006-1981 Average dwelling value ($1,000) Inner city 342.733 163.169 614.330 181.096 Old suburbs 261.368 90.443 584.822 148.776 New suburbs 222.827 90.694 451.057 83.760 Exurbs 287.300 125.306 610.051 183.521 Average gross rent Inner city 735.882 76.172 1018.754 93.462 Old suburbs 692.615 -36.484 986.673 .786 New suburbs 725.064 -38.186 950.777 -67.942 Exurbs 701.933 40.779 1151.129 279.271 Value of owned dwellings per square km Inner city 364.653 254.491 750.456 432.816 Old suburbs 131.362 67.375 221.720 106.779 New suburbs 39.469 30.586 133.866 92.475 Exurbs 13.480 8.307 9.781 6.741 Value of rental buildings per square km Inner city 532.885 54.618 366.585 126.737 Old suburbs 101.721 6.012 52.586 8.501 New suburbs 9.323 4.280 22.410 10.779 Exurbs 1.279 -.325 .859 .544 Proportion of dwellings owner-occupied Inner city .271 .095 .463 .045 Old suburbs .464 .045 .642 .043 New suburbs .777 .057 .761 .047 Exurbs .869 .100 .844 .012 Proportion of dwellings single-detached Inner city .014 -.010 .165 -.263 Old suburbs .240 -.013 .376 -.208 New suburbs .583 -.038 .422 -.259 Exurbs .858 .076 .713 -.167 (Continued on next page)  41 The population density within the “built-up city” increased by 25.1 percent in Vancouver from 1971 to 2006, whereas Montreal saw a decline of 32.8 percent (Filion et al., 2010, p. 545). Between 1991 and 2001 the population within two-kilometers of the downtown increased by 28.2 percent in Vancouver and also increased in Montreal by 6.1 percent (Filion & Gad, 2006). 129 Figure 3.7 (continued from previous page) Changing housing stock characteristics  Montreal Vancouver  2006 2006-1981 2006 2006-1981 Proportion of high-rise apartments (5 stories+) Inner city .153 .005 .276 .097 Old suburbs .092 .007 .125 .028 New suburbs .036 .007 .045 .017 Exurbs .000 -.001 .004 .003 Average number of rooms per dwelling Inner city 4.524 .035 4.903 -.125 Old suburbs 5.317 .163 6.191 .066 New suburbs 6.566 .459 6.689 -.062 Exurbs 7.437 .898 7.762 .512 Dwelling density (units per square km) Inner city 3927.834 443.759 2640.722 885.374 Old suburbs 1084.266 189.998 590.236 150.240 New suburbs 227.860 134.536 389.825 232.100 Exurbs 54.023 12.471 18.995 10.430 Number of dwelling units Inner city 273770 30930 204920 68705 Old suburbs 716700 125589 265665 67623 New suburbs 439725 259627 319890 190461 Exurbs 6580 1519 28985 15915 Source: Calculated using Statistics Canada (1981a; 2006a) census tract data.  Revitalization and revalorization of the inner city is thus evident in both metropolitan areas, the value of dwellings and rents increasing most in the inner cities both in absolute terms and on a per area basis (Skaburskis & Moos, 2008). Rents actually decreased in real terms in Montreal’s old and new suburbs and in Vancouver’s new suburbs. The decline in real rents in these areas is likely related to the declining incomes of renters who are dispersing due to rising housing costs in the inner city (Bunting et al., 2004)—the rental stock increasingly consisting of relatively lower quality units such as basement apartments. There have also been changes in the tenure composition, both metropolitan areas seeing increases in homeownership. The increases are highest in Montreal’s inner city but as mentioned in previous chapters ownership 130 rates remain substantially lower in Montreal than in other Canadian cities. However, it is particularly in Montreal’s inner city where ownership rates are low, whereas in the new suburbs and exurbs ownership rates actually exceed those in Vancouver. Increases in homeownership in Montreal occurred often through the conversion of duplexes, two or three story attached buildings historically often co-occupied by tenants and the owner of the building (Germain & Rose, 2000). In Vancouver, it is the growing condominium market that contributes to homeownership increases in the inner city (Harris, 2011). The loss of the older stock built before 1946 points to the revitalization of the inner city in Montreal and Vancouver.   The most notable difference between the two metropolitan areas in terms of these select housing characteristics are the decline in the proportion of single-family dwellings and number of rooms and related increases in high-rise apartments and dwelling densities in Vancouver. These changes relate to the increasing intensification of the inner city but point to the general increase in higher density development across the metropolitan area in large part associated with growth management policies (Tomalty, 1997) and rising cost of housing (Skaburskis & Moos, 2008). Montreal’s inner city continues to be characterized, however, by lower proportions of single-family dwellings and higher dwelling densities than Vancouver’s. Just over one percent of the inner city stock in Montreal is in the form of single-family dwellings so that suburbanization is particularly differentiated from central areas by the availability of a different housing type. In Vancouver, single-family dwellings still constitute almost 17 percent of the inner city housing stock but the increases in land values in central areas also means that most of this stock is in higher income and wealthy neighbourhoods 131 where housing costs are high. Notable is that the new suburbs in Vancouver have actually higher dwelling densities than those in Montreal. In part, these higher densities are due to the concentric circles combining the low-density suburban areas with areas near the suburban growth centers where there is a presence of high-rise apartments and other higher density housing units. The higher suburban densities, however, also reflect an increasing proportion of row and semi-detached housing in Vancouver.      Table 3.8 shows the outcome of the pca including select variables measuring changes in the housing market from 1981 to 2006. Akin to factorial analysis used heavily in study of the urban social ecology, pca is a useful technique to reveal how several variables are mutually correlated in that it reduces the original set of variables into uncorrelated components (Ley, 1988; Wyly, 1999; Meligrana & Skaburskis, 2005). However, where factorial analysis aims to reveal “underlying, but unobservable structures in the data”, pca simply explains “the variance-covariance structure through a few linear combinations of the original variables” where a large proportion of the variance can be explained by the “first one, two or three components” (Johnson & Wichern, 1992, pp. 356, 359, 396). Components with Eigenvalues above one are retained, as is common practice (Andrey & Jones, 2008). In this case, pca is a means of revealing in what ways housing market changes occurred as a combination of cost and stock characteristics, and than to relate the components to the spatial structure of the metropolitan areas using linear regression analysis to study the geography of the changes. In the Montreal case, the first component scores highly on the variables measuring changes in the value of the owner-occupied housing stock and the increase in the proportion of owned and single-family dwellings. In Vancouver, the first component 132 points to the increases in the per area values of owner-occupied and rental housing and the increases in homeownership.   Table 3.8 – Principal component analysis of housing stock characteristics Montreal Component scores Change 2006-1981 First Second Third Dwelling value .479 -.109 .028 Gross rent .256 .602 -.094 Dwelling value per area .531 .030 .047 Gross rent per area .226 .632 -.114 Proportion owned dwellings .458 -.187 .449 Proportion single-family dwellings .355 -.381 -.080 Number of units -.189 .211 .876 Eigenvalue 2.229 1.334 1.028 Proportion .318 .191 .147 N-cases 643   Vancouver Component scores Change 2006-1981 First Second Third Dwelling value .095 .501 .490 Gross rent .188 .565 -.016 Dwelling value per area .578 -.015 .266 Gross rent per area .533 -.152 .329 Proportion owned dwellings .456 -.010 -.495 Proportion single-family dwellings .275 .273 -.577 Number of units .234 -.576 .052 Eigenvalue 2.231 1.542 1.415 Proportion .319 .220 .202 N-cases 245   Source: Calculated using Statistics Canada (1981a; 2006a) census tract data.  Table 3.9 shows the standardized beta coefficients of a series of regressions that relate the components to independent variables measuring distance to the central business district, distance to rapid transit stations, walkability, categorical variables the distinguish the inner city from the suburbs and exurbs, and the changes in dwelling density. The negative relationship between the first component and the distance variables and the dummy variables differentiating the outer areas from the inner city as 133 well as the positive relationships with walkability means that higher scores in the first component occur in the central areas in both Montreal and Vancouver. In Montreal, these findings point to the increases in dwelling values in central areas as well as the increasing ownership components. In Vancouver, the first component points squarely at the increasing condominium stock in the inner city that is associated with increasing home ownership and higher per area housing costs and rents. Most interestingly, however, is that in Vancouver the change in density is the best predictor of the first component among the variables measuring the spatial structure, with an r-squared of 0.706 and a standardized beta value of 0.841 for the density variable (as compared to an r-squared of 0.065 and a beta 0.258 for the density variable in Montreal). It highlights the homeownership and rising housing cost components associated with densification of the Vancouver metropolitan area.  The second component in Montreal shows a positive relationship with changes in rents and a negative relationship with changes in the proportion of single-family dwellings. In Vancouver, the second component is positively associated with changes in the total value of dwellings and rents and negatively associated with changes in the number of units. The regressions show that in Montreal the second component occurs in the inner city and new suburbs and areas experiencing increases in dwelling densities. In Vancouver, the second component is associated with proximity to the central business district and negatively associated with areas in the new suburbs and those experiencing increases in dwelling densities. The component evidently points to the increases in the cost of housing in remaining lower density areas in transit accessible and central locations. Finally, the third component in Montreal shows high scores 134 associated with changes in owned units and the number of dwelling units added. The third component in Vancouver again points to areas experiencing increases in dwelling values and the per area value of the rental stock, and is negatively associated with changes in homeownership and single-family dwellings.   Table 3.9 – Beta regression coefficients with housing stock changes as the dependent variable Montreal First  Second  Third  Distance to centre -1.484 *** -.101  .160  Distance to centre^2 1.033 *** .158  .165  R-squared .378  .002  .100  Distance to transit -.797 *** -.046  .355 *** Distance to transit^2 .436 *** .120  -.088  R-squared .205  .004  .075  Walkability .515 *** .041  -.284 *** R-squared .264  .000  .079  Old suburbs -.628 *** -.219 *** -.054  New suburbs -.513 *** -.005  .602 *** Exurbs -.098 ** -.109 ** .002  R-squared .373  .050    Change in dwelling density (2006-1981) .258 *** .343 *** .206 *** R-squared .065  .116  .041  Vancouver First  Second  Third  Distance to centre -1.183 *** -.573 ** -.503 * Distance to centre^2 1.006 *** .353  .259  R-squared .140  .060  .065  Distance to transit -.524 ** .079  -.102  Distance to transit^2 .425 * -.062  .025  R-squared .027  .007  .002  Walkability .406 *** -.052  -.013  R-squared .162  .001  .004  Old suburbs -.390 *** .032  -.315 *** New suburbs -.283 *** -.329 *** -.406 *** Exurbs -.129 * .054  -.149 * R-squared .091  .116  .105  Change in dwelling density (2006-1981) .841 *** -.343 *** .205 ** R-squared .706  .114  .038  Notes: Linear regression outputs of principal component scores (Table 3.7) as dependent variables and measures of the metropolitan spatial structure as independents. Standardized beta coefficients shown. Polynomials of variables were included when it improved model fit. ‘Inner city’ is the base for the regression using the CMA zones as independent variables. Source: Calculated using Statistics Canada (1981a; 2006a) census tract data. 135 The regressions place the third component measuring changes in owned units and unit growth in Montreal into the new suburbs and areas away from transit and high walkability. In Vancouver, the third component again points to increases in housing costs in the inner city and areas experiencing increases in density but it is not associated with proximity to transit. It points to tracts in the southwest of the inner city that have maintained higher single-family dwelling stock and ownership levels. Overall, the results show that households entering housing markets, whether they are immigrants or young adults, are facing a housing market context where the inner cities have seen significant revitalization but evidently suburban growth continues to be an important component of urban change. Clearly, there are evident differences in the spatial dimensions of housing costs between Montreal and Vancouver that would influence where households locate based on their incomes. Importantly, since the 1980s densification, associated with higher prices, has emerged as a particularly dominant factor of housing market change in Vancouver.  3.4.2 Dwelling values and neighbourhood change Table 3.10 first considers the relationships between changes in dwelling value and the presence of recent immigrants descriptively. The data show that in Vancouver’s old and new suburbs, higher absolute increases in dwelling value occurred in tracts with lower relative proportions of recent immigrants. But these tracts also had higher increases in income, matching our theoretical expectations of changes in a neighbourhood’s socioeconomic status mirroring its housing outcomes. However, in the inner city the tracts with higher relative proportions of recent immigrants display higher 136 increases in dwelling value than the tracts with lower shares of recent immigrants, yet the former exhibits lower increases in household income. The findings suggest that the de-coupling of housing from income parameters materializes primarily in Vancouver’s inner city. In Montreal, an effect of immigration on changes in dwelling values is not evidently discernable. Increases in dwelling values in the inner city are matched by increases in incomes that are similar in scale in the tracts with above and below average proportions of recent immigrants. In the new suburbs, the increases in dwelling values and incomes are lower in the recent immigrant neighbourhoods. Only in Montreal’s old suburbs is there a higher increase in dwelling values in the recent immigrant neighbourhoods and the change in household income, although positive, is small in magnitude. The effects of immigration on ownership markets are more evident in Vancouver, with more wealthy migrants and a larger ownership market. It should be taken into account, however, that immigrants in Vancouver could have moved into neighbourhoods that are already experiencing value increases simply due to other factors. In other words, a multivariate test is needed. The results of the regression model examining changes in tract dwelling value as a function of stock and income characteristics are shown in Table 3.11. Included in the model are 1981 dwelling value, household income, and the proportion of the stock built before 1946. The distance variable places the tract in the larger CMA context based on the neoclassical economic assumption that value, and its increase over time, still depends on accessibility to the CBD (Alonso, 1964; Skaburskis & Moos, 2008). Distance is only a crude measure of accessibility and does not change over time with transport improvements or changes in activity patterns (Maclennan, 1977). The model holds 137 constant the initial tract conditions and introduces variables of neighbourhood change: age of stock, number of owned units, share of single-family dwellings, and household income.  Table 3.10 – Change in census tracts by proportion of recent immigrants Montreal CMA Tracts with Prop. Of Recent Immigrants in 2001  ≤ 1.2×CMA Mean > 1.2×CMA Mean Inner City n=102 n=118 Proportion of Recent Immigrants .091 .383 Change in Household Income ($1,000) 4.729 5.090 Change in Dwelling Value ($1,000) 36.136 32.250 Old Suburbs n=208 n=135 Proportion of Recent Immigrants .073 .362 Change in Household Income ($1,000) -2.481 .397 Change in Dwelling Value ($1,000) -.539 2.774 New Suburbs n=81 n=10 Proportion of Recent Immigrants .031 .324 Change in Household Income ($1,000) 8.381 2.365 Change in Dwelling Value ($1,000) 10.319 2.793 Vancouver CMA Tracts with Prop. Of Recent Immigrants in 2001  ≤ 1.2×CMA Mean > 1.2×CMA Mean Inner City n=26 n=35 Proportion of Recent Immigrants .188 .446 Change in Household Income ($1,000) 14.398 3.186 Change in Dwelling Value ($1,000) -53.269 -18.221 Old Suburbs n=73 n=33 Proportion of Recent Immigrants .179 .423 Change in Household Income ($1,000) 4.618 -.675 Change in Dwelling Value ($1,000) -50.266 -51.883 New Suburbs n=54 n=17 Proportion of Recent Immigrants .141 .375 Change in Household Income ($1,000) 7.063 -5.472 Change in Dwelling Value ($1,000) -68.336 -101.769 Notes: All dollar values adjusted for inflation to $2001 using the Bank of Canada consumer price index. Recent immigrants: Immigrated to Canada between 1981 and 2001.  Source: Calculated using Statistics Canada census tract data (1981a; 2001b). Modified and reproduced with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved.    138 Table 3.11 – Change in dwelling value as a function of census tract characteristics, 1981-2001 Montreal CMA Coeff. Beta  Dwelling value 1981 ($1,000) -.444 -.719 *** Distance to the center -1.181 -.209 *** Household income 1981 ($1,000) .768 .324 *** Change in income 1981-2001 ($1,000) 2.065 .690 *** Inner city immigrant tracts 6.604 .054     x Change in income 1981-2001 ($1,000) -.112 -.017  Old suburbs immigrant tracts 8.744 .077 *    x Change in income 1981-2001 ($1,000) -1.079 -.232 *** New suburbs immigrant tracts 3.851 .010     x Change in income 1981-2001 ($1,000) -.541 -.019  Change in owned units 1981-2001 -.006 -.095 ** Change in proportion single-family dwellings 1981-2001 44.658 .052  Proportion of dwellings built pre-1946 in 1981 37.336 .238 *** Change in proportion dwellings built pre-1946 1981-2001 88.783 .213 *** Constant 41.433  *** R-Squared .508   N-Cases 648   Vancouver CMA Coeff. Beta  Dwelling value 1981 ($1,000) -.532 -1.040 *** Distance to the center -.740 -.130 * Household income 1981 ($1,000) 2.589 .805 *** Change in income 1981-2001 ($1,000) 2.347 .482 *** Inner city immigrant tracts 41.353 .227 ***    x Change in income 1981-2001 ($1,000) -1.429 -.123 * Old suburbs immigrant tracts 37.107 .193 ***    x Change in income 1981-2001 ($1,000) -.602 -.033  New suburbs immigrant tracts -6.006 -.024     x Change in income 1981-2001 ($1,000) -2.275 -.078  Change in owned units 1981-2001 -.004 -.083  Change in proportion single-family dwellings 1981-2001 21.794 .041  Proportion of dwellings built pre-1946 in 1981 143.587 .431 *** Change in proportion dwellings built pre-1946 1981-2001 51.463 .079  Constant -47.128  *** R-Squared .511   N-Cases 244   Notes: Immigrant census tracts is a 0/1 dummy variable where 1 = [(proportion of recent immigrants in tract) / (mean CMA proportion of recent immigrants)] > 1.2. Recent immigrants: Immigrated to Canada within 20 years before census enumeration. *p<0.05, **p<0.01, ***p<0.001. The Vancouver results differ slightly from previously published results due to refinements in variable definitions. Source: Calculated using the Statistics Canada census tract data (1981a; 2001b). Modified and reproduced with permission from Urban Geography, Vol. 31, No. 6, pp. 724-749. ©Bellwether Publishing, Ltd., 8640 Guilford Road, Columbia, MD 21046. All rights reserved.  139 The regression holds constant impacts on value from redevelopment (i.e., loss of the older stock), but not changes arising from renovations, changes in housing size or quality, or local bylaws and regulations that can affect value differently across the city (Mark & Goldberg, 1985; Bourne & Bunting, 1993). Finally, three dummy variables identify the location of census tracts with proportions of recent immigrants in the 2001 census, 1.2 times that of the CMA average42. The dummy variables might also be picking up other effects on price not captured by the controls. The effects of the housing stock variables are similar between the two metropolitan areas. The coefficient for 1981 dwelling value is negative, suggesting some cyclicality in local housing investment over time (Beauregard, 2005). Neighbourhoods with higher proportions of pre-1946 stock in 1981, and those that retained higher proportions of this stock, saw larger increases in dwelling value, consistent with prior findings (Skaburskis & Moos, 2008). The coefficient for the distance variable confirms the negative relationship between distance from the CBD and value gains. The distance decay effect is larger in Montreal as evident by the larger size of the coefficient than in Vancouver, and the larger beta value also suggests that centrality explains more of the changes in dwelling values between 1981 and 2001 in Montreal. The 1981 household income variable shows a positive coefficient. This is consistent with findings regarding wealth accumulation in high-income areas (Ley et al., 2001). The change in income is also positively related with value increases, in line with models of neighbourhood transition that suggest value changes occur with income shifts. 42 Sensitivity analysis revealed that altering the categories does not alter the overall conclusions; neither does including the proportion of recent immigrants as a continuous variable. The use of dummy variables facilitates the interpretation of its interaction with the variables describing CMA zones and change in household income simultaneously. 140 The larger size of the beta coefficient for this variable in Montreal suggests that the change in income is a better predictor of changing dwelling values than in Vancouver that has been thought to be more influenced by global real estate investments.  In Vancouver, the case study neighbourhood dummy variables suggest positive price gains for locations with high concentrations of recent immigrants, but only in the inner city and the old suburbs. The interaction term between the variable identifying tracts with high proportions of recent immigrants and household income in the inner city shows a negative coefficient that is statistically significant. The results are consistent with the flow of wealthy migrants into Vancouver’s inner-city neighbourhoods, entering directly into ownership markets with equity or income derived from foreign sources. There are also additional value gains in recent immigrant neighbourhoods in the old suburbs, but there the interaction term with income does not point to a de-coupling of housing markets from socioeconomic characteristics of the neighbourhood. The potential for foreign equity to have influenced Vancouver’s housing market is most evident in the inner city where price gains have been most marked.  In Montreal, the case study neighbourhood dummy variables or their interaction with the income changes do not display statistically significant coefficients in the inner city or the new suburbs. In the old suburbs, however, there is a positive effect associated with the presence of recent immigrants, and the income changes in these neighbourhoods explain less of the changes in dwelling values than is the case for the entire metropolitan area. It is in Montreal’s old suburbs where dwelling values and incomes have declined since 1981 and thus the immigrant population, which is 141 generally less wealthy than in Vancouver, has been able to buy into ownership markets. Despite effects due to immigration, comparison of beta coefficients would suggest that income remains the most informative predictor of changes in neighbourhood dwelling value in both metropolitan areas.   3.5 Discussion  This chapter positioned immigrants as agents of the globalization process that can shape local housing markets. It also asked about the spatiality of the changes in the housing context within which immigrants, and the young adults analyzed in subsequent chapters, make decisions about housing. The analysis points to densification and increases in dwelling values and rents in higher density and central areas a key factor shaping the housing context in Vancouver that is different from the past and from current conditions in Montreal. The analysis also used data from the 1981 and 2001 Statistics Canada census to investigate the changing relationship between income and housing demand (measured by the user cost of housing) in a multivariate framework as a growing share of recent immigrants to Vancouver arrived with established wealth. The analysis revealed that recent immigrants consumed more housing than similar non- immigrant households, but that was already occurring at the time of the 1981 census, and that this is occurring in Montreal as well where there were fewer wealthy migrants.  One pertinent finding is that the positive relationship between income and housing demand is mediated by recent immigrant status, which was not the case in Vancouver in 1981. The results can be attributed to the shifting profile of immigrants arriving in Vancouver. In the post-1990 period, immigrants, particularly from Asia, 142 increasingly arrived with established wealth and many were known to continue earning income outside the country. However, since immigrants’ housing consumption is also less tied to their local labour market earnings in Montreal, there appears to be a general “immigration effect” at play whereby immigrants purchase housing with greater equity (Hiebert et al., 2008). At the census tract level, the presence of recent immigrants was positively associated with higher dwelling value appreciation even after other factors that affect dwelling value change were taken into account. In Vancouver, this effect is visible in the inner city and old suburbs, but not in the new suburbs, whereas in Montreal the effect is present only in the old suburbs. While there may also be effects from non-immigrant investors, the distinctive character of recent immigrant flows plays an important part in disconnecting Vancouver’s housing from local labour markets.  However, it must be remembered that for many immigrants “paying for housing is a major struggle” (Hiebert et al., 2008, p. 31). This chapter did not examine effects on rental markets or impacts on affordability. The focus on owners also overlooks the difficulties of those unable to attain ownership. Even though the potential for upward pressures on price can be inferred in Vancouver, care needs to be taken that immigrants do not become scapegoats for the city’s housing affordability crunch. Immigration occurred concurrently with other factors of housing market restructuring, notably the retraction of the welfare state. Furthermore, Vancouver could also be somewhat exceptional in terms of housing market effects associated with wealthy and skilled migrants. The findings here point to one particular way that globalization processes influence housing markets. Globalization, and the (neoliberal) institutional restructuring that facilitate it, results in new types of labour flows that alter the relationships that help 143 explain housing consumption and neighbourhood transition. Local labour market indicators become less useful in analyzing housing markets in gateway cities where large numbers of migrants arrive with foreign equity as is the case in Vancouver.  Even in Montreal immigrants allocate a lower share of their income toward housing despite having fewer wealthy migrants. But ownership markets constitute a smaller proportion of the overall housing market in Montreal and there are fewer pressures on price. Thus, immigration is not at the aggregate scale related to dwelling price increases in Montreal yet immigrants who are owners would necessarily be those with relatively higher equity, which would explain their ability here too to allocate lower shares of their income toward housing. The presence of this effect in the old suburbs where prices have declined suggests, however, that immigrants are not buying into expensive ownership markets to the same extent as they are in Vancouver where there has been a high influx of wealthy Asian migrants.  The analysis in this chapter also permits insight into the relationship between household and housing variables on housing consumption and neighbourhood change in more general terms that provide important insight for the analysis of young adults’ housing decisions. At the household scale, the findings point to the continuing importance of household characteristics that change over the life-course, such as size and the presence of children, on housing consumption but evidently these relationships are being transformed by immigration in that some immigrants reside in larger households and have more income earners present. The findings also show that the positive effect of income on housing consumption, and changes in neighbourhood dwelling values, have increased over time. The finding suggests greater inequality 144 among owners but also increasing difficulty to attain homeownership for low-income households. It takes more income than in the past to attain similar levels of housing consumption.  The finding for the income variables also indicates that households, with the exception of recent immigrants, are allocating a larger share of their income towards housing than in the past, and the beta values show that the effect of the income variable continues to be one of the strongest determinants of changing dwelling values at the neighbourhood scale. Interpreted in another way, the findings also indicate that households experiencing stagnant or declining incomes would face increasing affordability burdens, which is an important change in the context facing the young adult households. Thus, the changes in the labour market and household composition are of great importance in influenc