Situating Individuals' Subjective Needs and Aspirations Within The

Situating Individuals’ Subjective Needs and Aspirations within the Wider Context of the Housing Market Ayodeji Temitope Adeniyi Bachelor of Design in Architectural Studies Master of Design in Urban Design

A thesis submitted for the degree of Doctor of Philosophy at The University of Queensland in 2019 The School of Earth and Environmental Sciences

Abstract

This thesis examines how the subjective aspirations (use-values) of renters and owners in the Greater Brisbane Metropolitan region is reflected by the housing market. Conventionally the public recognises how such aspirations are gratified based on house prices (exchange- value). However, the housing market accommodates some factors disproportionately. The ability of a few entities to influence the market may be magnified, while the prospect of people’s aspirations being satisfied in the market is diminished. Consequently, individuals’ needs, and aspirations are only partially explained via supply and demand. Therefore, some of these needs can only be ascertained with proxies such as users specifying their motivations or by studying how people interact with the market. The benefit of these proxies is that they provide a window into what people truly want. This revelation goes beyond what is observable through house prices and homeownership rates, which is how Australia’s designation as a property-owning democracy (POD) is often determined. However, a POD is better characterised by how it reflects economic liberty, opportunity, and balanced competition.

Therefore, I aim to understand housing aspirations by evaluating how individuals’ self-reported preferences progress longitudinally. These preferences are derived from essential social and dwelling needs, as well as individuals’ financial motivations. Additionally, the thesis examines whether the housing market reciprocates those same fundamental needs when consumers interact with the market. The thesis applies a theoretical framework drawn from critical realism because this theory can explain how individuals’ internally derived needs have been facilitated or constrained by external influences within the housing market. This is because critical realism identifies different domains of interpretation based on individuals’ experiences, events and based on external influences, known specifically as causal mechanisms.

The thesis applies a mixed-use methods approach by implementing a survey analysed with methods such as K-means clustering (KMC), and social network analysis (SNA) methods. KMC defines how the housing needs of respondents are clustered to better understand how housing needs progress with age. These housing needs were compiled from the 237 respondents of a large-scale survey that was deployed to Greater Brisbane residents. The SNA was based on property descriptions of houses for sale and rent from 8300 property listings with the intention of revealing the system-level properties (the distinct aspirational, typological, locational clusters) of the housing market, as well as the vertex- level connections (the relationships between isolated property ideals). ii

The thesis finds that the level of agency that people have in terms of actualising their aspirations is limited. For example, residential mobility was found to inefficiently enhance housing aspirations, since individuals’ self-reported preferences progressed in relation to their age and level of market understanding than based on total moves. The inefficiency can be explained by external constraints such as prohibitive housing costs, unpredictable tenure, and complex investment and housing opportunities that are poorly understood by some segments of the market. Many of these constraints were reflected in a cohort divide, both in terms of contradictory attitudes as well as in terms of how the market represented the needs of each cohort. Specifically, through the survey, it was found that the renter cohort was more saddled with life-course priorities than the owner cohort, who prioritised improvements in dwelling and financial adaptations. Moreover, the SNA of prospective land buyers and homebuyers (to a lesser extent) better reflected use-values, than those looking to rent houses.

These findings are significant because they support the premise that the housing market has afforded some residents and investors property rights, tenure and security, which conflict with other participants in the housing market. These conflicts were revealed through a causal coding of what the survey respondents identified as the most disruptive factors in the housing market. Although the Brisbane metropolitan region is said to embody the Great Australian Dream and Australia is said to be a property-owning democracy based on the rate of ownership, these findings question the self-sufficiency of such designations. The POD is meant to embody more than just the rate of homeownership but is supposed to reflect a healthy infusion of use-value ideals within the housing market itself. The findings do not indicate that this has occurred. The housing ladder is supposed to reflect the fulfilment of subjective housing aspirations based on the notion that successful housing moves are measurable by progressive price increases. However, aside from the inefficiency of property transactions which this research has highlighted, the thesis has also shown that many of the ideals of the Great Australian Dream are contradictory.

iii

Declaration by author

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co- authors for any jointly authored works included in the thesis.

Publications included in this thesis

No publications included.

Submitted manuscripts included in this thesis

No manuscripts submitted for publication.

Other publications during candidature

Adeniyi, Ayodeji, Mirko Guaralda, and Raul Dias De Carvalho. 2017. "The Contextualization of Divergent Outlooks in a Greenfield Master-Planned Community: A Pathway towards Reflexivity ". International Planning Studies. doi: 10.1080/13563475.2017.1344541.

Adeniyi, Ayodeji, and Johnson, L., 2018. "I won't move to a One Bedroom Dog Box": The Challenges of Downsizing for older Social Housing Tenants in Queensland, Australia. 2018 Joint Asia-Pacific Network for Housing Research and Australasian Housing Researchers Conference, School of Engineering and Built Environment Griffith University, Gold Coast Campus

Contributions by others to the thesis

No contributions by others.

Statement of parts of the thesis submitted to qualify for the award of another degree

No works submitted towards another degree have been included in this thesis.

Research Involving Human or Animal Subjects

This research has human ethical clearance approval obtained from the School of Earth and Environmental Sciences’ ethics officer Karen McNamara [SEES number 201802-01] on the 3rd of February 2018. A copy of the ethics approval letter is included in Appendix 1.

vii

Acknowledgements

The outcome of this thesis has been the result of personal and interpersonal reflection and dialogue. While this process has provided many lessons and values, the ultimate lesson has been the pursuit of knowledge. However, yearning for information without direction would be a futile endeavour. Fortunately, with the experience and patience of my advisory team, I was equipped with timely and consistent feedback, which simplified the complexities of life as a graduate student. These contributions are too many to list, but my principal supervisor, Dr Thomas Sigler, was largely instrumental in helping to narrow the focus of this thesis, whilst pushing the realm of possibilities and scientific application. Additionally, my associate supervisor Dr Peter Walters bolstered the theoretical basis and social inquiry of this thesis.

External to my advisory team, I owe a debt of gratitude to Dr Dorina Pojani, Dr Sonia Roitman and Dr Kelly Greenop, who were reviewers/chairs during my candidature, and provided feedback directly pertaining to the thesis document. I value the support from Professor Jamie Shulmeister and Dr David Pullar in their capacities as postgraduate coordinators. I am also thankful to Professor Patrick Moss, Professor Lynda Cheshire, and Dr Suzanna Fay for their formal feedback on my methods in the early stages of the thesis. Likewise, I am appreciative of the inspiration and support from Martin North, regarding the economics of housing.

I am privileged to have been part of the Governance and Planning Urban Studies group, which was a productive student initiative. From this group, I received informal feedback from colleagues such as Tigor, Lata, Gillian, Reden, Johanna, Jaime, Temi, Deti, Angela and Jason, who was an encyclopedia for managing various tools and challenges of the PhD. I am also appreciative of the discussions that I had with friends from the ‘penthouse’ in Chamberlain, such as Professor Corcoran, Laurel, Anurodh, Julia, Radek and Rosabella.

Of course, I cannot forget the resourceful Ant, who has been a friend outside of thesis matters, in addition to providing invaluable technical support. Similarly, I am grateful to the ‘R wizard’ Isaac. I am also beholden to Paul, who shared his own PhD experiences during the latter stage of my candidature.

I dedicate this thesis to my friends and family, without whom my journey into academia would not have been possible in the first instance. They are my mum and dad, my siblings Ayodele and Adetayo, my partner Xiaochen, and all other friends and family, whom I cannot name here.

viii

Financial support

This research was supported by an Australian Government Research Training Program Scholarship. This research was also supported by the School of Earth & Environmental Sciences Higher Degree by Research Funding

Keywords subjective aspirations, housing ladder, social network analysis, use-value, exchange- value, life-course, housing needs analysis, territoriality, property-owning democracy

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 060401, Economic Geography, 40%

ANZSRC code: 060806, Social Theory, 30%

ANZSRC code: 120503, Housing Markets, Development, Management, 30%

Fields of Research (FoR) Classification

FoR code: 1604, Human Geography, 35%

FoR code: 1608, Sociology, 35%

FoR code: 1401, Economic Theory, 30%

The University of Queensland

Table of Contents

LIST OF FIGURES ...... xiv LIST OF TABLES ...... xvii LIST OF ABBREVIATIONS ...... xviii LIST OF APPENDICES ...... xix CHAPTER 1 : INTRODUCTION ...... 1 1.1 Subjective Housing Aspirations amid Competing Housing Needs ...... 1 1.2 Problem Statement ...... 5 1.3 Knowledge Gap ...... 6 1.3.1 Purpose Statement ...... 10 1.3.2 Research Questions ...... 10 1.4 Structure and Outline of the Thesis ...... 11 CHAPTER 2 : CONCEPTUAL FRAMEWORKS: THE MULTIPLE DOMAINS OF REALITY AND HOUSING NEED ...... 16 2.1 How Critical Realism Limitations are Addressed ...... 21 2.2 How the Three Housing Facets Satisfy Fundamental Needs ...... 22 2.3 Factors that Constrain the Consumption of Housing Products ...... 26 2.4 Housing Needs Frameworks ...... 28 2.5 Australia as a Property-Owning Democracy ...... 32 CHAPTER 3 : HOW TERRITORIALITY HAS SHAPED THE AUSTRALIAN HOUSING LANDSCAPE ...... 33 3.1 Introducing Territoriality...... 33 3.2 A Brief Introduction of Australia’s Housing Narrative ...... 35 3.3 Introducing the Theory of Means-End Chain ...... 36 3.4 What Form of Housing? The Interchangeability of Housing Terminology ...... 40 3.5 The Territorialisation of Australia’s Settlements: The Emergence of the Great Australian Dream and Housing Commodification ...... 43 3.6 Australia’s Dwelling Profile ...... 46 3.7 Brisbane GCCSA Dwelling Profile ...... 46 3.8 How Models of Neighbourhood Change Explain the Brisbane GCCSA Settlement Pattern ...... 51 CHAPTER 4 : THE THEORETICAL DIVERGENCE BETWEEN HOUSING NEEDS AND MARKET OUTCOMES ...... 56 4.1 The Divergence between Use-value and Exchange-value: Explained through Three Theories of Value...... 56

4.1.1 Introducing the Three Forms of Value and Three Value Theories ...... 57 4.1.2 Cost/Labour Theory of Value ...... 58 4.1.3 Subjective Theory of Value ...... 59 4.1.4 Equilibrium ...... 60 4.1.5 Summary: The Need for an Alternative Proxy for Use-Values ...... 62 4.2 The Contradiction of Use and Exchange-value: Financialisation ...... 63 4.2.1 The Impact of Financialisation on the Housing Terrain: Uneven Development and Extensive Land-Use ...... 65 4.2.2 The Impact of Financialisation on Productivity: Exaggerated Economic Growth and the Stagnation of Wage Growth ...... 66 CHAPTER 5 : RESEARCH METHODOLOGY ...... 71 5.1 Introduction ...... 71 5.2 Research Design ...... 73 5.2.1 Introducing the Components of the Brisbane GCCSA Housing Needs Survey ...... 74 5.2.2 Introducing Individuals’ Housing Needs Composition and Trajectories .... 77 5.2.3 Introducing Individual Attitudes and Explanation of Causal Mechanisms . 77 5.2.4 Introducing how Individuals’ Housing Needs are Reflected by the Market 78 5.2.5 Case Study Selection Criteria ...... 78 5.3 Introducing the Brisbane GCCSA Housing Needs Survey ...... 81 5.3.1 Data Collection and Sampling...... 85 5.3.2 Data Cleaning ...... 95 5.3.3 Interpreting the Brisbane GCCSA Housing Needs Survey: Aspirations and Attitudes...... 97 5.3.4 Potential Limitations of the Brisbane GCCSA Housing Needs Survey ... 100 5.4 Social Network Analysis of Real Estate Advertisements ...... 102 5.4.1 Introducing Social Network Analysis ...... 106 5.4.2 Real Estate Data Collection and Sampling ...... 115 5.4.3 Real Estate Data Filtering ...... 125 5.4.4 Analysing the Social Network Data through Centrality Measures and Community Detection ...... 127 5.4.5 Understanding Centrality and Community Measures through Density .... 131 5.4.6 How Network Centralisation is Defined in this Thesis ...... 134 5.4.7 How the Centrality and Community Measures are Implemented ...... 134 CHAPTER 6 : SOCIAL AND ECONOMIC DETERMINANTS OF HOUSEHOLDS’ HOUSING PROGRESSION OVER THE LIFE-COURSE ...... 136 6.1 Housing Aspirations and the Life-course ...... 136

xii

6.1.1 The Composition of Individuals’ Housing Needs Satisfiers and Life-course Factors ...... 137 6.1.2 The Housing Trajectories of Individuals in the Brisbane GCCSA ...... 141 6.2 Exploring Individuals’ Attitudes in the Brisbane GCCSA ...... 152 6.2.1 Impact of Individual Needs on Housing Attitudes...... 152 6.2.2 Impact of External Factors on Housing Attitudes ...... 156 6.2.3 Summary: How Residents’ Individual Needs Conflict with External Motives ...... 161 CHAPTER 7 : Property Industry’s Facilitation of Household Needs ...... 163 7.1 Social Network Analysis of the Market: Centrality Measures ...... 164 7.2 Centrality Measures of the Market: Weighted Degree Centrality ...... 168 7.2.1 The Density of the Real Estate Networks and Subgraphs ...... 169 7.2.2 The Prominence and Semantics of the Rental Network and Subgraphs 171 7.2.3 The Prominence and Semantics of the Entire Sales Network and its Typological Subgraphs ...... 178 7.2.4 The Prominence and Semantics of the Sales Geographical Subgraphs 187 7.2.5 Subgraph/Network Boundaries: Spatial Adjacency/non-adjacency ...... 193 7.3 Community Measures and Social Network Analysis Conclusion ...... 194 CHAPTER 8 : DISCUSSION ...... 198 8.1 Redefining how Progression is conceived in the Housing Market ...... 200 8.2 The Nexus Between the Facilitation and Prioritisation of Housing Needs Satisfiers ...... 202 8.3 The Conflict of Divergent Territorial Claims: Synthesising the Significance of Individuals’ Needs and External Factors ...... 206 CHAPTER 9 : CONCLUSION ...... 209 9.1 Contributions of this Thesis ...... 210 9.1.1 Situating Individuals’ Subjective Needs within the Wider Context of the Housing Market ...... 210 9.1.2 Establishing an Alternative Means of Measuring Progress in the Housing Ladder: Subjective Aspirations ...... 212 9.1.3 Demonstrating Territorial Conflict and Potential Resolution ...... 213 9.2 Policy Recommendations ...... 215 9.3 Topics for Future Research and Final Remarks ...... 216 LIST OF REFERENCES ...... 218 APPENDICES ...... 232

xiii

LIST OF FIGURES Figure 1.1 The Sequence of Housing Analysis ...... 2 Figure 1.2 The Three Domains of Critical Realism: The Real, the Actual and the Empirical ...... 4 Figure 1.3 Sequence of the Thesis according to Critical Realist Ontology ...... 12 Figure 1.4 Sequence of the Thesis Chapters according to Critical Realist Ontology 12 Figure 2.1 The Capacity to Express Choice within Housing over Time ...... 30 Figure 3.1 Conceptualisation of the Housing Ladder by Typology, Tenure and Family Composition ...... 36 Figure 3.2 Usage of the terms ‘Houses’, ‘Homes’ and ‘Dwelling’, Google Trends (Web Searches: not Seasonally Adjusted)...... 41 Figure 3.3 Housing Suitability (Spare Rooms) as a % of Total Dwellings of the Brisbane GCCSA ...... 48 Figure 3.4 Housing Suitability (Needed Rooms) as a % of total Dwellings of the Brisbane GCCSA ...... 49 Figure 3.5 Dwelling Types as a % of Total Dwellings of the Brisbane GCCSA ...... 50 Figure 3.6 Tenure Types as a Percentage of Total Dwellings in the Brisbane GCCSA, Statistical Area (SA2) ...... 53 Figure 4.1 Market Equilibrium: Showing the Intersection between Market Supply and Market Demand Curves ...... 61 Figure 4.2 Australian Household Debt as a Percentage of Household Disposable Income ...... 69 Figure 5.1 Life-course Queries about the Cultural Aspects of Residential Preference ...... 83 Figure 5.2 Social-Dwelling Attributes: Queries about the Use-value of Residential Preference ...... 84 Figure 5.3 Financial Attributes: Queries about the Financial Aspects of Residential Preference ...... 84 Figure 5.4 Gender Divide from the First 37 Respondents ...... 88 Figure 5.5 Stratified Regions showing the Overlapping Quintiles in Black and their Corresponding Postcodes in Aqua ...... 92 Figure 5.6 List of Postcodes from the Stratified Regions ...... 93 Figure 5.7 Brisbane GCCSA SA4 Geographic Rings ...... 127

xiv

Figure 6.1 Most Relevant Aspirational Themes - Entire Cohort (Mean of Responses) ...... 137 Figure 6.2 Most Relevant Aspirational Themes - Entire Cohort (ZAM)...... 138 Figure 6.3 Most Relevant Aspirational Themes - Owner Cohort (ZAM) ...... 139 Figure 6.4 Most Relevant Aspirational Themes - Renter Cohort (ZAM) ...... 140 Figure 6.5 K-means Clustering of the 27 Variables into 4 Clusters and 3 Groups: Showing how Owner Themes Change According to Age and which Themes are Grouped Together by the Mean Score ...... 143 Figure 6.6 K-means Clustering of the 27 Variables into 4 Clusters and 2 Groups: Showing how Rental Themes Change According to Age and which Themes are Grouped Together by the Mean Score ...... 144 Figure 6.7 The Relationship between Mobility (1-8 Transactions) and the Relevance of the 27 Themes ...... 147 Figure 6.8 Change in the Mean of Relevance of Financial and Social-Dwelling Factors by # of Moves (1-8 Moves) ...... 148 Figure 6.9 Reasons for Moving (Median Moves): Occupation/Level of Understanding ...... 149 Figure 6.10 Sankey of Changes in Relevance of Life-course, Social-Dwelling and Financial attributes (All Cohorts) ...... 151 Figure 7.1 Change in Normalised Rank of Terms by Centrality Methods: Rentals . 166 Figure 7.2 Change in Normalised Rank of Terms by Centrality Methods: Sales .... 167 Figure 7.3 Relationship between Clustering Coefficient and the Visible Edges of All Networks and Subgraphs ...... 170 Figure 7.4 Prominent Terms/Edges of the All Rental Properties Network ...... 173 Figure 7.5 Top 30 Rental Network Terms by WDC Normalised Value ...... 174 Figure 7.6 Prominent Terms/Edges of the Rental (Apartment) Subgraph ...... 175 Figure 7.7 Top 30 Apartment - Rental Subgraph Terms by WDC Normalised Value ...... 176 Figure 7.8 Prominent Terms/Edges of the Rental (House) Subgraph ...... 177 Figure 7.9 Top 30 House - Rental Subgraph Terms by Normalised WDC Value ... 177 Figure 7.10 Prominent Terms/Edges of the All Sales Network ...... 179 Figure 7.11 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the All Sales Network ...... 180

Figure 7.12 Prominent Terms/Edges of the Sales (Apartment) Subgraph ...... 181 Figure 7.13 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the Sales - Apartment Subgraph ...... 182 Figure 7.14 Prominent Terms/Edges of the Sales (House) Subgraph...... 183 Figure 7.15 All Visible Out-degrees and In-degrees (Showing Edge Widths) for the Sales – House Subgraph ...... 184 Figure 7.16 Prominent Terms/Edges of the Sales (Land) Subgraph ...... 185 Figure 7.17 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the Sales - Land Subgraph ...... 186 Figure 7.18 Prominent Terms/Edges of the Sales (Inner) Subgraph ...... 188 Figure 7.19 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the Sales - Inner Subgraph ...... 189 Figure 7.20 Prominent Terms/Edges of the Sales (Middle) Subgraph ...... 190 Figure 7.21 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the Sales - Middle Subgraph ...... 191 Figure 7.22 Prominent Terms/Edges of the Sales (Outer) Subgraph ...... 192 Figure 7.23 All Visible Out-degrees and In-degrees (Showing Edge Widths) for the Sales - Outer Subgraph ...... 193 Figure 7.24 Boundaries of Four Market Types defined by Clustering and Semantics ...... 194 Figure 7.25 Multilevel Communities of the All Rental Network ...... 195 Figure 7.26 Multilevel Communities of the All Sales Network ...... 197

xvi

LIST OF TABLES Table 2.1 Housing Facets and Corresponding Satisfiers ...... 24 Table 3.1 Territorial Boundaries and their Territorial Agents ...... 33 Table 3.2 Means-End Chains Identify Emotional Ends by Retracing Attributes’ Root Causes ...... 38 Table 5.1 How the Research Questions are Addressed Methodologically ...... 72 Table 5.2 The Four Components of the Brisbane GCCSA Housing Needs Survey . 74 Table 5.3 Survey Participation Target and Outcomes ...... 94 Table 5.4 Distribution of Tenure and Typological Types by Response to Principal Dwellings for each Age Group ...... 94 Table 5.5 Distribution of Occupancy by Tenure and Typology ...... 95 Table 5.6 Attributes of the Brisbane GCCSA Housing Needs Survey ...... 97 Table 5.7 Calculation of the Optimal Number of Clusters Using Elbow Method ...... 99 Table 5.8 12 Factors Representing Individuals' Attitudes towards the Housing Market: Explanations are shown in yellow...... 100 Table 5.9 The Different Forms of Meaning in Real Estate Language ...... 109 Table 5.10 The General Features of Social Network Analysis ...... 110 Table 5.11 The General Features of Network Density ...... 111 Table 5.12 Summary of the Centrality Measures Applied in the Thesis ...... 112 Table 5.13 Summary of the Community Measures Applied in the Thesis ...... 112 Table 5.14 Real Estate Property Sales Descriptions Sample ...... 117 Table 5.15 Distribution of Real Estate Sample by Tenure, Typology and Geography ...... 118 Table 5.16 Selected Real Estate Data Themes – Sales Network ...... 120 Table 5.17 Selected Real Estate Data Themes – Rental Network ...... 121 Table 5.18 Model of a Contingency Matrix ...... 122 Table 5.19 Example of a Normalised Adjacency Matrix ...... 124 Table 5.20 Division of Sales and Rental Typologies ...... 126 Table 6.1 Correlation between Changes in the Mean x̅ of rank of Financial, Life Course, and Social-Dwelling Factors ...... 150

xvii

LIST OF ABBREVIATIONS ABS Australian Bureau of Statistics APC Average Priority Cluster CBD Central Business District CC Closeness Centrality CEM Centrality Measure CLC Clustering Coefficient COM Community Measures CSV Comma-separated Values CTV Cost Theory of Value EVC Eigenvector Centrality FHOG First Homeowners Grant FGC Fastgreedy Community (Brisbane) GCCSA Greater Capital City Statistical Area HLA Housing Ladder Analysis HNA Housing Needs Analysis HNS Housing Needs Survey HOC Homeowner Occupiers HPC High Priority Cluster ICC Ipswich City Council KMC K-means Clustering LA Locational-amenity group LCC Logan City Council LEA Locational-economic-amenities LGA Local Government Area LOPC Lowest Priority Cluster LPC Low Priority Cluster MBRC Moreton Bay Regional Council MEC Means-End Chain MLC Multilevel Community OECD Organisation for Economic Co-operation and Development OSNA Ontologically Based SNA POD Property-owning Democracy PTI Price to Income (Ratio) RBA Reserve Bank of Australia RCC Redland City Council ROI Return on Investment RTA Residential Tenancies Authority SA2 Statistical Area Level 2 SA3 Statistical Area Level 3 SA4 Statistical Area Level 4 SEQ South East Queensland SFG Symbolic-lifestyle group SLA Statistical Local Area SNA Social Network Analysis STV Subjective Theory of Value TP Typological-place group TPA Typological-place-amenities WDC Weighted Degree Centrality ZAM Zuccolotto Adjusted Mean

xviii

LIST OF APPENDICES Appendix 1 Human Research Ethics Approval Letter ...... 232 Appendix 2 The Brisbane GCCSA Housing Needs Survey Coverage ...... 233 Appendix 3 Years Lived in the GCCSA (Brisbane) by Age ...... 234 Appendix 4 Tenure by Occupation (Renters) ...... 235 Appendix 5 Tenure by Occupation (Owners) ...... 236 Appendix 6 Occupancy by Dwelling (18-29, 30-39 Cohort) ...... 237 Appendix 7 Occupancy by Dwelling (40-49, 50-59 Cohort) ...... 238 Appendix 8 Occupancy by Dwelling (60-69, 70 and Over Cohort) ...... 239 Appendix 9 Composition of the Mean Number of Housing Moves by Age Bracket ...... 240 Appendix 10 Reasons for Moving: Level of Decision-Making (All Cohorts) ...... 241 Appendix 11 Total Moves: Occupation/Level of Understanding ...... 242 Appendix 12 Property Details Scraper (Outwit) ...... 243 Appendix 13 String Generation for Brisbane GCCSA Page Search ...... 244 Appendix 14 Fast-Scrape of Address List for Property Details (Outwit) ...... 245 Appendix 15 Google AdWords Targeted Location ...... 246 Appendix 16 Facebook Advertisement Audience ...... 247 Appendix 17 Facebook Advertisement Schedule ...... 248 Appendix 18 Advertisement Campaigns: Google AdWords and Instagram Carousel ...... 249 Appendix 19 Google AdWords Audience Insights ...... 250 Appendix 20 Facebook Advertising Audience Insights ...... 251 Appendix 21 Survey of Housing Aspirations Questions ...... 252 Appendix 22 Visible Out/in-degrees (Edge Widths): Rental Network ...... 255 Appendix 23 Visible Out/in-degrees (Edge Widths): Rental Apartment Network ...... 255 Appendix 24 Visible Out/In-degrees (Edge Widths): Rental House Network ...... 256 Appendix 25 Fastgreedy Community Network Plots: Rental and Sales Networks ...... 257 Appendix 26 R Line-by-Line Coding: Description of SNA Plots (Centrality/Community) .. 258 Appendix 27 Line-by-line Coding: Description of the Clustering Coefficient Algorithms... 260 Appendix 28 Summary of the 10 Networks/Subgraphs by Edge Properties ...... 260 Appendix 29 Similarity between the Three Centrality Measures...... 261 Appendix 30 Explanation of Network Subgraphs by Edge Properties ...... 262 Appendix 31 Multi-axis Line Graphs Showing Normalised Values of All Subgraphs ...... 263 Appendix 32 Explanation of Multi-axis Line Graphs ...... 264 Appendix 33 List of Normalised Property Terms (WDC) ...... 265

xix

CHAPTER 1: INTRODUCTION 1.1 Subjective Housing Aspirations amid Competing Housing Needs Housing is one of the fundamental pillars of human life, with its distinctive role in satisfying the need for shelter, financial needs, social interaction, and status. It is for this reason that the United Nations asserts that adequate housing is a universal right (UN Habitat, 2016). This declaration underpinned the importance of housing equity and was first expressed by the United Nations as part of the broader Universal Declaration of Human Rights (UDHR). However, this message is easily diluted, misrepresented, or entirely ignored, due to the many functions that housing must fulfil. Moreover, the number of housing functions and the complexity of those functions are constantly increasing. For example, housing markets have witnessed increased involvement from financial institutions (Fernandez & Aalbers, 2016).

Nevertheless, the declaration that housing is a fundamental right based on its necessity is promising. Adequate housing has been defined by the 1966 International Covenant on Economic, Social and Cultural Rights (ICESCR) as housing inclusive of security of tenure, access to services, schools and affordability (UN Habitat, 2016). Also outlined was the right to a continuous improvement of living standards. These universal declarations and covenants are somewhat useful in coordinating the competing needs of the housing market, but they remain imprecise. Doyal and Gough (1991) stress that the conception of universal needs is dubious and may reflect the fact that self-interest groups desire to blanket diverse needs and aspirations with a single moral legitimacy. The UDHR and OHCHR universal declarations do not comprehensively outline the specifics of housing progression and are only successful in emphasising the universality of housing need. Therefore, the ideas of basic housing needs and universal housing rights insufficiently capture the aspiration that people have for optimal housing. These aspirations are important because they underline whether people’s housing circumstances are aligned with their ideals and life-course. To be clear, the alignment of housing outcomes and housing needs is not guaranteed due to competing motivations in the housing market.

Therefore, to understand how housing outcomes and needs are aligned, housing must be separated into its constituent parts. We can conceive of housing as a basket of goods composed of various housing needs. Individuals do not consume the entire basket of goods, nor is there a single characterisation of how basket of goods are to be delivered (however, housing equity stipulates that there should be a provision of basic needs). Rather individuals have inherent housing needs which lead to a vast composition of housing products. The pursuit of those products represents individuals’ subjective aspirations (what

1 people want). This pursuit is not to be confused with housing outcomes, which is yet another way to describe the basket of goods. Housing outcomes represent what is delivered to individuals. Housing outcomes may or may not be equitable, and they may or may not be entirely representative of individuals’ subjective aspirations. This is because just as housing can be conceptualised as fulfilling different needs, the housing market is also shaped by several processes of a political, social, economic and legal nature (É. D'Arcy & Keogh, 2002). These processes reflect the different motivations that underpin housing needs. The way these processes are coordinated affects how needs transition to aspirations and how aspirations transition to outcomes (Figure 1.1).

Fundamental Housing Needs Housing Housing Needs Satisfiers Aspirations Outcomes (1) (2) (3)

Figure 1.1 The Sequence of Housing Analysis

Of the three ways to study housing (Figure 1.1), housing outcomes are the easiest to understand since these outcomes represent measurable housing quantities such as supply, demand, prices, stock, and transactions. These measures control exchange-value (monetary worth) and do not reveal how needs are satisfied. To reveal needs and their satisfiers, the underlying motivations behind housing outcomes must be understood. These motivations include identity, subsistence, idleness and participation. Even the motivation for financial (freedom) cannot be readily understood based on housing outcomes such as prices. This is because housing needs are subjective, and housing needs satisfiers have to be studied based on immediate or future needs. Upon unearthing how needs are satisfied, the nature of individuals’ subjective aspirations can then be revealed by studying the patterns of satisfaction (the relative differences in prioritisation).

In the housing market literature, it that these subjective aspirations or housing needs are implicit in housing outcomes. This thesis will establish that this is untenable by revealing the barriers that prevent needs from being perfectly nested in prices. One of the primary barriers is due to the competing functions of the needs themselves. In the absence of coordination, needs can counteract one another. Moreover, even when theorists and researchers have recognised the importance of the subjective, such models or research have largely been conducted using tools that are poorly reconcilable with housing outcomes. This thesis will resolve this problem in two ways. Firstly, the housing needs studied in this dissertation will be exhaustive and mirror the housing market’s functionality. Secondly, subjective housing aspirations are not simply what people want, but an observation of

2 consumers’ established behaviours and prioritisation. In other words, their subjective housing aspirations represent whether they appear to be aspiring towards ideals, based on how they have prioritised past housing needs.

Housing needs and financial motivations can be broadly categorised within three facets: dwelling (structural/typological needs), home (social needs) and house (financial needs) (Ruonavaara, 2016). As mentioned, though these needs are complementary, they may be contradictory. This is because individually-derived priorities (use-values) and market-derived outcomes (exchange-values) underscore different processes. While it is unlikely that the entirety of individuals’ needs can be represented in any scientific study or model, in housing studies, it is generally acknowledged that there is a relationship between needs and outcomes. This thesis shows that the social and dwelling needs embody use- value, and together with financial motivations result in price (exchange-value).

Nonetheless, price is only an approximation of subjective needs due to supply and demand forces being unable to reflect aspirations perfectly. Consequently, there have been attempts to quantify housing needs through proxies, such as the amount of labour individuals expend in the production process or by quantifying the relative differences in individuals’ preferences. However, these economic arguments have had varying degrees of success in explaining the subjective. One of the main reasons why these proxies are insufficient is that in housing markets that are increasingly unaffordable or inequitable, the actualisation of needs and aspirations are likely to be more constrained. Therefore, while the housing needs, housing aspirations and housing outcomes continuum outlined in Figure 1.1 is applicable in any housing market, it is even more pertinent for unaffordable housing markets, where the divergence between needs and outcomes are possibly augmented.

One such housing market that is increasingly unaffordable is the case study area of the Greater Brisbane Region. This housing market has exhibited a disparity between the growth of house prices and that of rent and income (Fox & Finlay, 2012 ; Fox & Tulip, 2014). This region is known formally as the Brisbane GCCSA (Greater Capital City Statistical Area). The Brisbane GCCSA is a region situated in the South East of the State of Queensland in Australia. This region includes the City of Brisbane (Local Government Area), which is the capital of Queensland, as well as surrounding urban-rural local government areas, which together form a singular housing catchment. Hence, this region is home to individuals with a diverse range of housing needs and aspirations; these are qualities which are only partly reconciled and understood through concepts such as supply and demand.

The Brisbane metropolitan region was among the least affordable markets out of 309 metropolitan housing markets, based on its median multiple ratio (the ratio of median house prices to median income), which is a widely applied method to gauge housing affordability (Demographia, 2019). Median multiples of 3.0 and under are affordable. According to the Demographia housing study, Brisbane had a median multiple of 6.3 (median house prices are 6.3 times median incomes) and is above the ‘severely unaffordable’ threshold of 5.1. Other unaffordable markets include Sydney with a median multiple of 11.7 and Hong Kong, which is the least affordable market, with a median multiple of 20.9 (Demographia, 2019). These cases of unaffordability represent how the transition from housing needs into housing outcomes may be inhibited, preventing the equitable provision of housing needs and individuals’ subjective aspirations.

Then one of the main undertakings of this thesis is to separate individual needs from market interactions, which is accomplished by introducing a stratified ontology from critical realism. A stratified ontology means that although there is only one reality in the world, that single reality is understood through three different domains: the real, the actual and the empirical. As mentioned earlier, there are competing needs that define individuals’ housing outcomes. These competing needs sustain complex processes that individuals influence but do not fully control or understand individually; this is the real domain. For example, a financially motivated investor is only one agent in the market, individuals’ housing choices are regulated based on planning regimes that they did not develop, though they may have played some small role. These unseen processes are known as causal mechanisms, and they produce actions or events (the actual domain) that may be observed or unobserved. The empirical domain is what an individual subjectively experiences. In other words, what is possible in the housing market, what happens in the housing market and what is experienced in the housing market are all different phenomena (though related). This stratified ontology is important because, through the sequence outlined in Figure 1.1, needs and aspirations are separated from outcomes, albeit without a comprehensive explanation of the underlying processes that shape the housing market. Thus, Figure 1.2 below shows the three domains of the critical realism ontology and explains the why and the how.

What is What Happens What is Possible (Outcomes) Experienced (Causal/Real) (Actual/Event) (Empirical) Figure 1.2 The Three Domains of Critical Realism: The Real, the Actual and the Empirical The critical realist ontology will be used to uncover causal mechanisms that explain the differing financial and spatial motivations within the housing landscape. These causal 4 mechanisms are proposed by firstly observing their effects on the housing market, for example, economic and urban processes that have impacted events in the housing market to varying degrees. Secondly, events within the Brisbane GCCSA and Australia are outlined in dwelling profiles. These events will be experienced dissimilarly by individuals in the housing market, which will be discussed in the core of the thesis: understanding individuals’ subjective housing experiences. Comparing individuals’ experiences with these events and their underlying forces forms an exhaustive knowledge of how individuals’ needs have been facilitated by the housing market.

1.2 Problem Statement Ideally, Australians’ housing circumstances would mirror their housing needs and aspirations, particularly across their entire housing trajectory. In reality, individuals’ housing trajectory in the Australian housing market is increasingly irregular. It is progressively difficult to access the housing market, housing tenure is uncertain and residential mobility, which partially signals disruptions in the housing trajectory has been consistently high (Namazi- Rad, Mokhtarian, Shukla, & Munoz, 2016 ; Sánchez & Andrews, 2011). Consequently, this research identifies the erratic homeownership and rental trajectory of individuals in the Brisbane GCCSA Region as a problem worth investigating. This residential trajectory underpins people’s residential stability, their social interactions, and their financial autonomy. When considering the median multiple of Australian cities, particularly that of the case study Brisbane GCCSA, the number of people who can afford housing is decreasing, although some analysts such as Fox and Finlay (2012) point to the fact that borrowing capacity is independent of the median multiple. Such conflation is the quintessential problem in the Brisbane and Australian housing markets, which is that households are facing financial challenges, in addition to the fact that the appropriate means of evaluating their needs and circumstances are also uncertain.

The uncertainty in the Brisbane housing market has various implications, such as how developers respond to the demand for ‘starter homes’, which are homes at the bottom rung of the housing ladder. The increase of housing and rental stress (households expending more of their monthly incomes on mortgages or rent than affordable) represents a decrease in individuals’ buying power, which makes insufficient housing products such as starter homes more palatable. However, this thesis primarily concentrates on how individual needs are satisfied and the progressiveness of households’ overall housing trajectory amid growing uncertainty, rather than specific housing products. It is not the significance of any one feature of housing that is important to this thesis, but the overall pattern that is suggested

5 by those features. Uncertainty in the housing market, as it pertains to housing tenure, accessibility and residential mobility are subject to historic and ongoing causal influences that mediate an individual’s role within the broader social fabric.

1.3 Knowledge Gap The uncertain housing trajectory of individuals in the Brisbane GCCSA is not a problem worth addressing if it is assumed that adjustments or corrections within the housing market are automatic. Maclennan, Pawson, Kenneth Gibb, Chisholm, and Hulchanski (2019, p. 7) state that a “…. framing assumption has been that housing markets are essentially well- functioning systems, with few inherent market failures”. However, there is no empirical proof that housing markets with severe house price inflation have self-regulating mechanisms capable of reversing problems such as house price inflation in the short term (Maclennan et al., 2019). Since the first International Housing Affordability Survey in 2005 and to date, the Brisbane housing market has been listed as severely unaffordable (Cox & Pavletich, 2005). Therefore, it may be prudent to consider that the use of capital gains as a key yardstick in the Brisbane and Australian markets (Aussie, 2019) reveals only one part of a more complex equation. Countries like Australia, the United Kingdom and Canada invest in what Maclennan and Miao (2017) have termed a ‘rentier’ economy, an economy built on scarce assets rather than productivity. This reality that housing wealth has contributed to the detriment of long-term growth and productivity signifies that sustainable housing futures are not possible if we think of the housing market as a self-contained and self-regulating economic sector. The assumption in neoclassical economics has always been that competitive, up-to-date, rational actors free from external frictions and regulation will efficiently shape the housing market (Kaldor, 1972 ; Maclennan et al., 2019). There is no empirical indication that these assumptions have manifest in housing markets such as the Brisbane GCCSA.

Rather than assume that housing markets are competitive or that individuals are largely up-to-date rational actors, this thesis’ literature investigates the nature of participation in the housing market and the motivations of consumers, government, and private actors in the market. This investigation positions the needs of consumers and how they are satisfied within the context of the underlying mechanisms in society. This context ensures that individuals’ role in the housing market is evaluated, rather than assuming that all actors in the common market are participating on equal terms towards efficient outcomes.

Consequently, this thesis considers that housing outcomes are derived from the coalescence of disparate market knowledge, financial constraints, monopoly behaviour and 6 unpredictable interactions. These differences are also exacerbated by economic advantage/disadvantage, regulatory environments, historical attitudes and territoriality (the contestation/control of geographic space) (Dent, Patrick, & Ye, 2012 ; Storey, 2012). These inconsistencies hamper the efficacy of individuals’ transactions and cloud our understanding of people’s needs and aspirations over time. This is because since housing outcomes are subject to many unpredictable external factors, individuals’ subjective intentions cannot be revealed by aggregate measures (Barker, 2003) such as prices or dwelling stock. The belief that aggregate measures such as price and housing demand represent use-values is because, though individuals lack the capacity and knowledge to equally transact within the housing market, market interactions are sustained by individuals’ perceptions of the market. Thus, the outcomes of these perceptions do not represent objective reality (Houston, 2001). Individuals within housing markets cannot access to the entirety of housing options and information; thus, feedback in the housing market is perpetually imbalanced (ECONorthwest, 2002).

Consumers are unaware of all the possible interactions in the housing market that influence their own decisions; moreover, since buyers in the market are typically financially constrained, individuals struggle to make efficient housing transactions. This is important because poorer individuals are likely to make housing trade-offs which require market awareness. Hence, many individuals cannot fully optimise their housing outcomes. Market prices may further obfuscate subjective needs since in the process of buying, selling, or renting housing, to whatever degree of efficiency, people accrue exchange-value and or some forms of use-values. Hence, since housing is not characterised by a single use, it is unclear to what degree a house immediately satisfies the needs of an individual.

Housing is subjectively consumed, but objectively acquired (Chadbourne, 2015); thus, a comprehensive understanding of how needs are being satisfied should consider both housing needs and the outcomes of market exchange. However, housing research is severed into the subjective based analyses on the one hand, and the market-derived perspectives, on the other hand. The market-derived perspective emphasises access to the housing market. Whether this is the notion that the housing ladder (Khan, 2018) is the mechanism that addresses housing needs, supply and demand analyses or housing needs analyses (Rowley, Leishman, Baker, Bentley, & Lester, 2017), which stipulate the minimum conditions used to describe those in need of housing. These approaches are natural since individuals derive housing products through the market. However, the market-derived analyses offer a limited lens into individuals’ subjective aspirations because house prices

7 are not derived from housing needs (subjective), but rather housing choices (objective). Housing needs are internally derived, while housing choices reflect external constraints and opportunities. Housing aspirations exist in the intersection between housing needs and housing choices in the market (Crawford & McKee, 2018). Therefore, it is important for this thesis not to consider a market-derived approach exclusively.

While housing costs are a broad abstraction of various societal interactions, housing aspirations represent how individuals strive to satisfy their housing needs. Individuals satisfy these needs by using their resources to obtain housing products based on their finances and understanding (housing choices) (Crawford & McKee, 2018). These choices form the relationship between buyers and renters on the one hand, and sellers and lessors on the other hand, and creates a price relationship between what the market can supply and what people can demand. This relationship is quantifiable and measured with market costs, known as house and rental prices. These prices are an aggregation of all interactions and transactions; they are not a representation of housing needs. Prices also reflect two distinct forms of aspiration: for immediate needs, or as a depository of value. Historically people have understood that housing serves a basic requirement for shelter and interaction, but there is an ever-increasing aspiration for housing as a depository of value, meaning it is financially valuable and can be sold or leveraged for other goods in the market. The financial viability of housing does not necessarily mirror its effectiveness for immediate use. For example, land prices per square metre in Australian capital cities have been observed to outpace income growth (Unconventional Economist, 2014). Individuals want to satisfy their immediate needs, but they may also want to maximise their house value. Although these intentions are related, they can diverge for reasons including speculation and scarcity.

However, in a ‘rentier’ housing market that is geared towards maximising people’s housing investments, there is a tendency to assume that individuals’ immediate needs and house price increases are aligned. In such markets, people are expected to not only compete for the gratification of their immediate needs but to also maximise financial gains. While the immediate focus of this dissertation is re-conceptualising individuals’ housing trajectories in terms of their housing needs throughout the life-course, this thesis first establishes why the premise of a non-financially driven housing trajectory may not be embedded in the housing market. This assumption is shaped by the perception of the market as primarily a vehicle of wealth, which influences what constitutes a successful housing trajectory. These perceptions are normative beliefs that sustain the notion that progress in the housing market is based on the financial growth of an individual’s housing portfolio over

8 time. In reality, people will make housing moves that do not correspond to a single pattern because people’s housing aspirations are diverse.

Though needs and market prices may be interrelated, they are not analogous. When these two distinct processes are confused, it further diminishes the ability of individuals and households to attain positive outcomes in the housing market. In the housing market, inefficiency disrupts equilibrium, which subsequently disrupts how the market understands individuals’ housing trajectories if care is not taken.

It is essential to re-conceptualise our understanding of individuals’ needs and housing progress, so that all facets of individuals’ housing needs are recognised, rather than the sum of all their interactions as a single exchange-value. While housing needs and preferences are not a new research field, much of the observations presupposes progression through gradual residential mobility or based on typological yardsticks. Specifically, housing progression is often measured based on a typological continuum; the model suggests that gradual increases in the price of an individual’s house signify upward movement. This process is popularly referred to as the ‘housing ladder’ or ‘property ladder’, which is a metaphor for upward progression in the housing market. Prevailing cultural discourses in Australia mostly emphasise the ‘housing ladder’ to capture the aspirations and subsequent actions of those transitioning from more basic, rental accommodation to more upmarket, owner-occupied dwellings, with the possibility of investment property included in the trajectory (Forrest & Yip, 2012 ; Mudd, Tesfaghiorghis, & Bray, 2001 ; The Committee for Economic Development of Australia, 2017). This approach is a universalist approach because market costs represent the balance of competition in the market and is not a direct representation of the relative needs of individuals. The temptation to simplify the relative needs of individuals into standardised units of measurement, such as frequency of mobility or house prices occurs because housing is sold as a singular product.

There is a need to separate life-course demands from aspirations and evaluate them relative to one another, to understand how individuals’ subjective needs and aspirations are satisfied over the life-course. This is contrary to a cost-driven approach which simply examines the relative value of a house against all other houses in that market. When the life-course is considered as the yardstick for housing progression, the relationship between successive housing and individual needs is more nuanced. If individuals’ housing moves do not increasingly mirror changes based on their life circumstances, then the housing ladder is haphazard rather than progressive. Individuals’ perceptions of the relative importance of

9 their housing needs over the life-course should be identified, in addition to understanding how such needs correspond with what the housing market has to offer.

1.3.1 Purpose Statement In housing policy, the perception of housing reflects the public’s interests; a “knowable set of physical and spatial parameters rather than the behaviour of only one individual in one house” (Tognoli, 1991, p. 656). The diverse housing needs of individuals, which include their financial priorities, social interactions, dwelling needs and particular life-course stages are simplified as a ‘knowable’ set of attributes such as house prices, housing stock and the degree of residential mobility. Therefore, the understanding of individual aspirations are partial, and this study will address this by exploring the nature of people’s choices, how these choices have been limited or facilitated, and what is the significance of these choices.

This thesis will contrast individuals’ motivations and needs with the proposed causal mechanisms of the Brisbane GCCSA, which include territoriality, The property-owning democracy (POD), and the outcomes of market exchanges such as financialisation. The purpose of which will be to understand Brisbane GCCSA residents’ housing aspirations as a relationship between individual outlooks and more complex interactions and social mechanisms based on the multiple domains of reality. The different domains of reality reflect the different vantage points of individuals, as well as the range of potential outcomes. These differences result in several discrepancies, such as the contradiction between various forms of value. Initially, these contradictions have been explored by economists and geographers, who sought to clarify how commodified labour and market exchanges, represent capital imbalances (Harvey, 2012b). There is a further need to distinguish individuals’ housing needs from the causal mechanisms that influence their consumption. These causal mechanisms are embedded in all modes of social interaction, but the financial rationalisation within the housing market is one of the central mechanisms that will be uncovered. Consequently, this thesis will establish if there is a conflict between the financially motivated rationale, which is a means to an end, and the non-financial needs that houses satisfy.

1.3.2 Research Questions The following research question follows from the proposition above. “How well are individuals’ subjective housing needs reflected in their consumption patterns?”. This research question was designed to understand housing consumption through individuals’ subjective priorities, which are more representative of their needs and motivations than market-derived outcomes. This approach differs from previous housing research studies based on the causal links between utility variables such as income and mobility, or needs 10 analysis based on preferences, without identifying how individuals define preferences. The research question above is further segmented into four phases that address these gaps.

Research Question 1 (RQ1): How do individuals prioritise housing needs satisfiers through the life-course?

Research Question 2 (RQ2): How are personal needs responsible for differing housing experiences?

Research Question 3 (RQ3): How are divergent attitudes responsible for differing housing experiences?

Research Question 4 (RQ4): How well are individuals’ needs facilitated by the housing market?

The research questions above are based on revealing how balanced six fundamental needs are but based on how said needs are satisfied in housing. The housing equivalents of fundamental needs are represented in three facets: home, house, and dwelling.

RQ1-RQ3 are derived from a survey of Greater Brisbane residents (Brisbane GCCSA Housing Needs Survey) to understand how their housing needs evolve over the life-course. RQ1 probes how individuals’ housing needs are satisfied by evaluating the relative differences in how 18 social-dwelling and financial satisfiers are prioritised, as well as 9 life- course factors. RQ2 establishes individuals’ perception of personal needs as the basis for differing housing experiences, while RQ3 establishes individuals’ perception of divergent values as the basis for differing housing questions. RQ4 is derived from a social network analysis (SNA) of the Brisbane GCCSA housing market in order to understand the composition of the housing market in relation to individual needs and financial motivations.

1.4 Structure and Outline of the Thesis The structure of this thesis is defined by three branches of philosophy, which inform the chosen methodology and research outputs. The first branch is axiological in nature. This represents what is intrinsically valuable in housing and is embedded in the thesis aim. This is an ethical consideration, which contemplates not only the impacts that housing policy and practices have on people’s livelihood, but that our understanding of value should be inseparable from appropriate policy, action, and ethical outcomes. Shown in Figure 1.3 on the next page is the structure of the thesis, as it relates to the critical realism ontology. Territorial and financial tendencies (which explain the housing market) are firstly introduced in Chapter 3-4, then events and experiences (secondary data) are introduced in the 11 methodology. Subsequently, events and experiences are studied using primary data, which, together with the other data, test the validity of the financial and territorial tendencies (methodological pluralism, explained further in Section 2.1).

Figure 1.3 Sequence of the Thesis according to Critical Realist Ontology The colour scheme of Figure 1.3 relates to the sequence of the structure shown in Figure 1.4 below. Figure 1.4 shows how the various chapters contribute towards methodological pluralism (the triangulation of various forms of knowledge) in Chapter 8.

Chapter 2: Housing Needs

Chapter 2: Housing Needs/Housing Ladder Framework

Chapter 2: Australia as a Property Owning Democracy

Chapters 3: The Interchangeability of Housing Terminology

Chapter 3: Territorial Differentiation of Housing and Meaning

Chapter 3: Australia/Brisbane GCCSA Dwelling Profile

Chapter 4: Financial Theories

Chapter 4: Financial Tendencies/Financialisation

Chapter 5: Introducing: The Brisbane GCSSA Housing Needs Survey

Chapter 5 Introducing: Social Network Analysis

Chapter 6: Housing Aspirations

Chapter 6: Contrasting Housing Attitudes

Chapter 7: Facilitation of The Three Housing Facets

Chapter 8: Methodological Pluralism Figure 1.4 Sequence of the Thesis Chapters according to Critical Realist Ontology Hence, Figure 1.4 explains how the critical realist ontology is embedded in the assumptions and findings throughout the entire thesis. To have a comprehensive understanding of housing needs, firstly, it is necessary to consider what is knowable in the housing market. In positivistic theories, knowledge is predicted and limited to observations, 12 while in idealist theories, knowledge is relative and constructed. Hence, reality or ‘what is’ (ontology) is determined by individuals’ experiential explanations, housing market observations, as well as the causal explanations derived from ontology. Thus, this thesis captures social and economic findings, individuals’ experiences in the Brisbane GCCSA (based on a Housing Needs Survey (HNS)), as well as market observations (derived from the SNA) and finally causal mechanisms, which will all be introduced in the outline below.

Chapter 1 has introduced the context of housing challenges in Australia and the Brisbane GCCSA. The problem statement explained that individuals’ housing trajectories are uncertain, while the knowledge gap explained the negligence of housing needs satisfiers by some researchers and policy-makers. These housing needs satisfiers are the basis for understanding housing aspirations. This process cannot be understood through housing outcomes since those outcomes reveal the results of opportunities and constraints rather than underlying motivations. Hence, an overarching research question based on how well individuals’ housing needs are facilitated by their consumption was devised.

In Chapter 2, the critical realist ontology (the empirical, actual, and real domains of truth) will be explained and how it relates to the housing market. The limitations of this theory will also be detailed. Thereafter Sections 2.2-2.3 will introduce the premise of fundamental needs and corresponding housing facets (home, house, and dwelling). These needs are vital to recognise since aspirations are internalised by individuals, and thus, researching aspirations requires understanding how and why housing needs are transformed. This chapter will accomplish that objective by establishing the role of the three facets of housing consumption and the fact that they manifest as competing processes in housing. Having established that, housing needs frameworks will be introduced which explain how housing needs research can be responsive to individuals’ subjective aspirations. Similarly, the property-owning democracy will be introduced as a policy-model, which can be used to evaluate the housing market as a whole.

Chapter 3 will introduce territorial tendencies through two geographic domains: the public territoriality and domestic territoriality. The significance of these domains is that territoriality is used to establish social and domestic hierarchies. Therefore, it will be explained that the territoriality of individuals in their households is premised on different motivations than that of authorities and developers. Thus, the domestic and public intentions of territoriality are evaluated through consumption models (means-end chains) and the composition of housing stock. Territoriality will also be established by contrasting the needs- based attitudes of individuals with the coordinated outcomes shown in Australia’s housing 13 terminology, housing history, and contemporary dwelling profiles. Lastly, models of neighbourhood change will also be introduced to explain how housing stock and neighbourhoods are shaped by contextual processes such as financialisation.

Chapter 4 will explain the sequence of value, from individual needs (use-value) towards collective needs (social-use-value), and finally, market relations (exchange-value). Use-value is internally derived, and exchange-value is a market outcome. Thus, the need to separate individuals’ use-values from exchange-values is because the latter is largely governed by market constraints and opportunities. It will also be explained that three value theories: cost theory of value, subjective theory of value and equilibrium have been unable to account for the distinction since the means of transacting and the product of transactions can both be reduced to prices. Therefore, resources can be diverted away from the practical uses of housing towards exchange-value by inverting priorities. Lastly, financial tendencies will be introduced, based on the premise that the increasing intervention of financial actors in the real estate sector has led an even greater separation between individuals’ needs and housing outcomes.

Chapter 5 reveals the methodological outline of this research. The chapter firstly introduces the mixed methods research design consisting of a survey instrument and social network analysis (SNA) method. It will be explained that these methods will produce different types of data, including networks, clustering and causal coding. It will be explained how these data reveal the composition of individuals’ housing needs and financial motivations. This will allow audiences to understand whether there is a balance in housing consumption, both from the perspective of the users and from the market-derived data.

In Chapter 6, the composition of individuals’ housing needs, the trajectory of individuals’ aspirations, and finally, individuals’ attitudes towards the housing market, will be established. The composition of individuals’ housing needs and financial motivations will be unearthed through the Brisbane GCCSA Housing Needs Survey’s Likert scale questions. Individuals’ housing trajectories will be represented by comparing the prioritisation of housing needs, using methods such as K-means clustering, which represents the various age cohorts within clusters based on how they had prioritised their housing needs. Additionally, individuals’ attitudes towards the housing market will be derived from an open- ended semi-structured question will capture the socio-cultural attitudes of various individuals.

In Chapter 7, the real estate findings will be analysed using Social Network Analysis (SNA) to uncover how the market facilitates housing needs satisfiers. This is important because, through the survey approach, the experiences of residents in the housing market are uncovered, as well as some causal mechanisms. Through the market-derived housing satisfiers it is possible to understand how individuals’ housing needs are fulfilled within the market context (using real estate advertisements as a proxy). The functionality of the market is understood through the relations between real estate property terms and how these terms are clustered into various communities (social, economic, and dwelling themes within the wider network). Two main measures of the SNA method are employed to accomplish this: centrality measures and community measures. The centrality measures establish the most prominent relationships or ideals within the network, while the community measures establish system-level properties (unsupervised) such as how words that co-occur share meaning. This will provide an understanding of the prominent housing market terms, which forms a window into how well the market facilitates individuals’ housing needs satisfiers.

In Chapter 8, the various domains of knowledge will be combined (methodological pluralism). This process cross-examines the needs, attitudes and aspirations that were uncovered in the survey and SNA with prior literature, which will clarify how subjective aspirations can be contrasted with market-derived outcomes. This triangulation explains that the evolution of the housing market’s territoriality has involved the retention of historical artefacts, which conflict with present-day objectives. Moreover, the tensions between individuals in the private realm and those who chart public affairs will be unveiled. This chapter will discuss the social and economic determinants of households’ progression through the fluid narratives uncovered in Chapter 6, as well as the quantitative boundaries based on the property industry’s perceptions.

In Chapter 9, which is the conclusion, the main contributions of the thesis will be discussed. The first contribution is a framework which details how to situate individuals’ needs within market outcomes, the second contribution details how to evaluate housing progress based on the housing needs satisfiers. The third contribution establishes a paradox, which is that the objectives within the housing market are sustained by all individuals in such a manner that the various cohorts counteract each other’s objectives. The limitations of the research design will also be outlined. The thesis will conclude with policy recommendations based on the property-owning democracy. Lastly, topics for future research and final remarks will be expressed.

CHAPTER 2: CONCEPTUAL FRAMEWORKS: THE MULTIPLE DOMAINS OF REALITY AND HOUSING NEED The conceptual framework of this thesis is based on a critical realist ontology, housing needs and also incorporates the Rawlsian concept of the property-owning democracy (POD), which will be explained in greater detail in Section 2.5. The POD and economic theory. The critical realist ontology is best understood as a resolution of the structure-agency problem in science. The structure-agency problem investigates whether structure is detached from human action and acts back on individuals (reification) or whether structure is treated as a property that is simultaneously created with human action (agency). Agency represents the ability of agents to make a difference through their intentions and choices. Structure represents the rules, resources, relations, and properties of the institutions, in which their choices exist (Giddens, 1984). These functions are relevant for this research because the housing market is a system with structural properties, institutional rules, and individual agents. Without clarifying the functional basis of these attributes, it would be uncertain if the implications of the research outcomes reflected individual intent or the durable forces of the housing market and economy. This structure-agency problem was addressed by Giddens such that structure and agency were bound together. Giddens (1984) explained that social rules were responsible for individual action, which in turn sustained the system. This meant that individual agency was integrated as social rules. This integration is problematic because social rules are defined by the system.

Therefore, the approach that will be embedded in this thesis is that the distinct operation of structure and agency is represented through analytical dualism (Herepath, 2014). This means that although human activity is always produced as social objects, and society cannot exist without human action, structure comes before the actions transformed by it, and structural elaboration comes after it (Archer, Bhaskar, Collier, Lawson, & Norrie, 1998). This separation allows for an analysis of structure and agency and how they are interrelated (Archer et al., 1998). The conceptual structure proposed by Archer et al. (1998) and specifically the multilevel ontology of Bhaskar (2008 [1975]) therefore resolves the dilemma in which structure and agency are viewed as mutually constitutive or equally representing each other in duality (Vandenberghe, 2008). Archer sought to explain that the characteristics of agency and social structure were different, although interrelated. This was in opposition to structuration theory, which was unable to uncover the nature of individuals’ choices since it bound the action of every individual to an external structure. Giddens was able to conclude that there was a need to separate the attained (action) from the unattained

(intent) (Giddens, 1984), but Giddens over-socialises “agents who have thoroughly interiorised their social conditions” (A. King, 2010, p. 254).

Alternatively, critical realists like Archer et al. (1998) propose a vertical hierarchy of agency and structure. Rather than thinking of structure as something that is reproduced at once, critical realism positions the reproduction of structure as an iterative process, whereby people are influenced by structure but also strive to influence structure on their own terms. Rather than referring to the gap between agency and structure as a retrospective internalisation, the disparities that stem from structure and the identity of structure itself are a direct result of individuals’ intended and unintended consequences. This means people’s fragmented social reality predates the reproduction of structure as a whole (Archer et al., 1998). When individuals make decisions, they do not channel all the resources and knowledge of the wider external structure. Instead, individuals have a fragmented perspective and a limited amount of resources and intent. Their decisions will resonate across the entirety of the system, but their aspirations and objectives are based on their limited vantage point.

Therefore, individuals have different interpretations of reality, and in Bhaskar’s ontology, these interpretations are separated into three tiers: The empirical, the actual and the real (Bhaskar, 2008 [1975]). This fragmented outlook reflects the fact that knowledge is contextual, and individuals will not experience the same contextual processes and outcomes throughout their life. Moreover, individuals will not filter these processes and outcomes in the same way, which means that their experiences are socially constructed (Musto & Rodney, 2016). The empirical refers to people’s observation of experiences. These observations are limited, because individuals’ perception of any event, does not represent the event in its entirety. What individuals perceive or evaluate is an abstraction of an actual object, experience, or situation. This is distinct from the actual event, which refers to what truly occurs in the world. The actual may not be entirely observed since it is composed of the empirical experiences of people, as well as other unobserved experiences. The actual events are shaped by underlying mechanisms that exist in the real but are unobserved until they manifest in some way in actual events. These causal mechanisms can be conceived as a tendency; a power with simply the potential to generate an event

Since actual events cannot be observed in totality, and the (real) causal mechanisms cannot be directly observed, then the real situation which encompasses all the underlying forces, events and experiences can only be partially understood. In housing, an example of the real level might be some enduring quality that governs human agency and enables 17 interactions in the housing market; for example, this could be a tendency that emerges from wealthy individuals’ financial and social capital. This capacity may enable them to rent or buy desirable property, and property records may show that these individuals inhabit high- value houses (actual level). However, the actual also includes non-events; these are potential events that fail to manifest due to counteracting forces. Just because individuals have the requisite capacity to inhabit some houses, does not mean that they would, since there are other causal mechanisms that may counteract their actions. Moreover, the actual involves records that are not wholly experienced or directly observed. When onlookers observe some houses as symbols of financial wellbeing, it is in the empirical level, which is fraught with assumptions. Therefore, the fact that the real unpredictably produces events, and empirical experiences are subjective, means that the knowledge of the structure of society is indeterminate without acknowledging all causal mechanisms and the interactions that sustain them.

Identifying the causal mechanisms that are relevant for the Brisbane GCCSA and Australian housing markets requires the assessment of extant literature, the housing ladder (and its functionality), as well as the implementation and analysis that this thesis carries out. The analysis will represent the fragmented outlook of individual renters and buyers. The concepts that will be sought in these various studies are regularities. Regularities refer to artefacts, such as behaviours and evidence of beliefs that occur consistently (Sayer, 2000). Additionally, there will be a need to examine if the processes uncovered in the extant literature are sustained elsewhere in the conceptual framework.

The significance of uncovering causal mechanisms in studies based on critical realism is due to its multilevel explanatory power; causal mechanisms explain what events are likely to emerge in a system and how people situate themselves in wider society. This is significant because causal mechanisms can be inconsistent and counterproductive. Causal mechanisms are derived not only from individuals’ fundamental needs but largely from shared interactions. Conversely, individuals’ fundamental needs are derived from their inner core beliefs (Morris, Hong, Chiu, & Liu, 2015). Thus, causal mechanisms are not wholly understood by individuals but are viable since they are sustained by individuals’ own normative beliefs and attitudes, as well as social norms that they observe. Normative beliefs are personal beliefs that individuals expect will conform with others’ beliefs: “I believe that housing is unaffordable”. Attitudes can be described as the favourable or unfavourable disposition towards an item or dimension: “housing affordability is problematic”. Lastly, social norms can be described as “‘should do’ and ‘how to do’” (Fang, Ng, Wang, & Hsu, 2017),

18 for example, “I have decided that housing is unaffordable because others are not buying houses”. The three variables are interrelated and all influence behavioural intentions (what individuals intend to do) to varying degrees. Therefore, individuals can develop common intentions based on loosely shared beliefs. This thesis will primarily investigate normative beliefs and attitudes since they relate to specific attributes of the housing market. Conversely, the concept of social norms, encompasses the range of behaviours that people think are acceptable based on shared interactions (what other people are doing or believe) (Fang et al., 2017). Social norms will be briefly discussed in terms of house-price euphoria (investing because others are) in Subsection 4.2.2.

These variables are crucial for the preservation of causal mechanisms because although individuals’ aspirations and external factors often contradict one another, individuals’ behavioural intentions can be shared if people adhere to any one of the variables. For example, even if individuals have an unfavourable disposition towards buying houses due to a specific reason such as debt, observing others in a range of contexts such as investing, and promoting a rite of passage, may influence them. Therefore, in housing, the opportunities and constraints that appear logical in one context may contradict the realities of another context, because housing has broad offerings which can sustain behavioural intentions at the juncture where some of the individuals’ beliefs intersect.

Though normative beliefs such as the Great Australian Dream have been popular for decades, they unequally represent individuals’ beliefs. This imbalance is because the role of individuals in the wider social fabric is unequal. People are socialised into new normative beliefs based on an asymmetrical exchange of preferences (Wendt, 1999). This is because individuals conform their own beliefs to social expectations, and some individuals’ preferences are more visibly expressed than others. This also influences the emergence of social norms since people often prioritise prevailing norms above their own personal values (Becker et al., 2014). Therefore, the housing market is not directly sustained by individuals’ preferences, but also how they interact with or perceive the market.

The indeterminacy of social structure is because structure is regulated at the individual level, but societal outcomes are often evaluated based on competitive feedback in society and the market. Individual action is predicated on an individual’s intent, their position in the market, and knowledge of their position; the last two attributes are the products of causal mechanisms. However, desire and market outcomes are not regulated by the same assumptions and thus require separate methodological approaches (Danermark, 2002). If the housing market is evaluated based on societal outcomes alone, it 19 will conflate the influence of causal mechanisms with individual intent. Berger and Luckmann (1966) refer to causal mechanisms as a form of coercive social control. These social controls carry power relations and symbolism separate from the here and now: “….the objective reality of institutions is not diminished if the individual does not understand their purpose or their mode of operation” (Berger & Luckmann, 1966, p. 79). The fact that the housing market supervenes on the interactions between homeowners, renters, and other agents, does not mean that individual homeowners intend to sustain the housing market in the way that it is constituted.

Housing typologies are reproduced and transformed with some intent from individuals, but largely with counteracting actions from other market participants. Societies are always restructuring the built environment to address their growing needs, but there is also a risk that pre-existing causal mechanisms such as territorial tendencies (tendencies that emerge due to how geographic territory is shaped) and financial tendencies (tendencies that emerge due to prioritising financial value) carry on in a manner that does not reflect their original objectives. The durability of these causal mechanisms is plausible because of emergence; individuals make decisions based on limited empirical experiences (Giddens, 1993 [1976]).

Thus, there are structures close to individuals that embody their immediate understanding and intended actions. At the same time, there are structures farther from individuals that embody the understanding or activity of society as a whole (the actual events of the market). The constant reproduction and transformation of these events are representative of the housing mechanisms, yet individuals’ housing experiences correspond with specific life-course stages such as childhood and adulthood. These stages are the empirical experiences that influence individuals’ actual aspirations and how they mediate signals from the real (causal mechanisms). The first step towards understanding the actual is to unpack fundamental needs, which will be explored in the subsequent sections.

However, this critical realism introduction has firstly shown that the unifying capacity of normative beliefs can present contradictions. One such contradiction is that individuals shape their housing environments within a personal context, whereas the housing market embodies financial, statutory, and territorial ideals at the city, state, and national levels. Thus, planning outcomes can be sustained by loosely shared housing ideals. These contradictions will be explored in Chapters 3-4, but, the limitations of critical realism will be addressed upfront, followed by why fundamental needs are important in the first instance.

2.1 How Critical Realism Limitations are Addressed

This section will underscore some of the potential limitations of critical realism. Critical realism focuses on the mechanisms responsible for social processes, however, since the social world is open (never static and always evolving according to human agency), the artefacts/aggregates of the social world such as house prices and dwelling stock cannot sufficiently explain social mechanisms (Barker, 2003). This is one of the reasons why this research does not consider aggregates but rather systems. The limitation of this conception is that critical realism is then less relevant for policymakers who rely on predictability (Denzin, 2004 ; Næss, 2004). To resolve the critical realist claim that there is no universal claim to truth, critical realism emphasises the need for methodological pluralism, which is the need for reconciling the three modes of reality (Denzin, 2004). This pluralism does not produce a universal truth, but truths that are as substantiated as possible.

Causal mechanisms refer to processes that have a ‘causal’ influence, not the outcome of those influences themselves; thus, the mechanisms are identified in a significantly different manner than social events and individual experiences. There are two main pitfalls that researchers have identified in this process. The first concern is whether causal mechanisms are revealed by the research or produced by the research (Angus & Clark, 2012 ; Kemp & Holmwood, 2003). The second concern is whether causal mechanisms are related to the events and experiences being studied by the researcher. How can the researcher test that the mechanisms have generated a particular event?

However, the causal mechanisms of the housing market are accessible. Individuals can experience housing subjectively and collectively; researchers and consumers engage in the housing market through shared social and economic processes. Thus, the context in which social responses are activated is another way of identifying causal mechanisms. All interpretations of the housing market are interconnected, and this is how the appropriate causal mechanisms are identifiable. Therefore, this thesis will strive to uncover the relevant contexts which influence the individual perspectives and agency of renters and homeowners. Since individuals’ actions are externally mediated, unintended consequences emerge, which are reconciled through normative beliefs, social norms, and attitudes. These emergent properties include financial and territorial tendencies, which are the causal mechanisms that will be investigated in this thesis.

2.2 How the Three Housing Facets Satisfy Fundamental Needs

This thesis aims to understand how individuals’ housing-related use-values and exchange- values are satisfied. These use-values and exchange-values correspond to six fundamental needs: freedom, participation, identity, idleness, subsistence, and protection (Max-Neef, Elizalde, & Hopenhayn, 1991). Max-Neef et al. (1991, p. 26) concede that these needs may not be “….historically and culturally constant”, but they are a comprehensive observation of human activities. These needs are fundamental because individuals are motivated to pursue these needs due to the negative physiological and psychological consequences that would arise if they were deprived of them (King-Hill, 2015). In the context of housing, these fundamental needs are satisfied through themes that can be assigned to one of the three housing facets: home, house, and dwelling. This allocation is done by reviewing housing literature to determine how household activities correspond to the six fundamental needs. The housing facets are classified based on whether the needs are satisfied through structural or typological means: dwelling, satisfied through social means: home, or satisfied via financial means: house.

The research questions have shown that these facets can be understood in terms of their composition, narrative, or how they are collectively balanced. Composition refers to the balancing of individuals’ needs, while narratives refer to individuals’ tendency to attain a coherent housing outlook. These narratives are known throughout the thesis as aspirations, and they lie in the intersection between internally derived fundamental needs and external constraints and opportunities (Appadurai, 2004 ; Hart, 2016). External constraints and opportunities represent diverse interests which may or may not be collectively balanced. Thus, the research questions also examine individuals’ differing attitudes, which will reveal the territorial mechanisms that influence the satisfaction of needs. Firstly, this section will outline what these fundamental needs are and why they should be applied in the thesis.

The assumption that individuals share the same basic capacity to seek out different facets of housing can be initially understood through Maslow’s hierarchy of needs (Maslow, Frager, Fadiman, McReynolds, & Cox, 1970). This hierarchy begins with basic physiological needs, followed by safety needs which allow for stability and structure. Then, individuals develop social needs of belonging, esteem needs (status) and finally, self-actualisation needs (personal growth and potential). This hierarchy is reliant on two concepts: transition and satisfaction. An individual might transition from a lower need like shelter, towards prestige needs, via a process of progressive gratification. Maslow et al. (1970) explain that if basic needs are fulfilled or taken for granted, complex aspirations will become the object 22 of satisfaction. Thus, the pyramid of needs developed by Maslow et al. (1970) is a progressive hierarchy; the lower needs are first satisfied, then the middle needs and finally, the higher needs. Maslow’s hierarchy is based on a gradient of housing needs because when a need is pressing, such as the need for food or water, it is unlikely that other needs will take precedence (Maslow et al., 1970). Yet, according to Max-Neef et al. (1991), pressing needs are rare, and most needs can independently emerge. Thus, all needs are equal and emerge to complement one another, concurrently or as trade-offs, insofar as a need is not pressing to the extent that other needs become insignificant. Subsistence is the only pressing housing need and comprises shelter from the elements, warmth and perhaps sleep. If this need is unfulfilled, the emergence of more elaborate housing needs will be impeded. If an individual has not satisfied the dwelling need of subsistence through shelter, that individual is unlikely to aspire to satisfy the house need of freedom through investments.

However, this thesis considers that there is a multiplicity of simultaneous needs because the research analyses existing renters, buyers, and properties. Due to these simultaneous needs, the classification of home, for example, incorporates much more than identity and incorporates the needs of idleness and participation, just as the concept of dwelling encompasses more than subsistence needs, and includes protection. However, the concept of house is only viewed as the equivalent housing facet of the need for freedom. Max-Neef et al. (1991) state that the universal need for freedom emerged later than other needs. Nonetheless, even though needs may emerge simultaneously, the satisfaction of all needs is not guaranteed. If people do not actively participate in the development of their living environments, then there will be a disconnect between their housing needs and their living conditions (Zavei & Jusan, 2012). If people gratify all their relevant needs, they will reach the pinnacle of self-actualisation and will become more individualised whilst maintaining their social identity (Zavei & Jusan, 2012). If individuals’ higher-level needs are entirely satisfied, then the housing experience is congruent, but if higher-level needs are non-existent or unresolved, then self-actualisation is also unresolved.

Thus, since individuals’ fundamental needs must be satisfied prior to self- actualisation, needs should not be treated as external trade-offs that people indiscriminately choose. This is because needs stem from an innate desire to fulfil individual and social objectives. The manner in which needs are satisfied (satisfiers) differs across cultures and individuals (Max-Neef et al., 1991), but correspond to the same basic structures (housing facets). Metamotivation (the desire to live in harmony with oneself), drives individuals to satisfy and enhance all their needs due to the motivation to find balance and happiness

(Hamel, Leclerc, & Lefrançois, 2003). Thus, the argument is that housing encompasses qualities that must be balanced to prevent negative physiological and psychological effects.

The process of satisfying these inherent needs represents individuals’ aspirations. Aspirations are expressed through satisfiers which are determined by socio-cultural attitudes and practices, socio-political contexts, and environmental contexts. These factors influence structures of activity such as creation, interaction, identification, possession, and functionality. These activities guide how needs can be satisfied. This means that, how a society can create, interact, identify, possess, and generally function influences how their needs are fulfilled. These actions will vary across contexts.

In the housing market, people produce goods by appraising market-revealed preferences (Bruni, 2007 ; Penz, 1986). These goods represent how the satisfiers of fundamental needs are physically expressed through specific products. The market shoulders the consumption and production needs of society, but not exclusively based on the fundamental qualities of any one product but based on how all products and preferences are related and perceived. Max-Neef et al. (1991) emphasise that the needs and satisfiers should not be conflated with their economic equivalents. Though satisfiers are obtained through economic goods, they remain distinct. People create these satisfiers or goods based on an insight of how market-revealed preferences align with their subjective preferences. Hence, the challenge is to reveal housing needs as distinct processes which can then be represented in the three housing facets: house, home, and dwelling (shown in Table 2.1 below). The house facet simultaneously embodies how freedom needs are satisfied, as well as exchange value. This is because, in a market economy, financial freedom (affordability) is vital for physiological and psychological satiation as well as economic gains and losses. Thus, financial (freedom) may be vital or counterproductive, as could home and dwelling if disproportionately consumed. Thus, a balance of housing needs is desirable.

Table 2.1 Housing Facets and Corresponding Satisfiers House Facet Home Facet Dwelling Facet Asset (freedom) Freedom (participation) Typology (subsistence) Value (freedom) Interaction & Belonging (Participation) Structure (subsistence) Investment (freedom) Continuity (identity) Size (subsistence) Financial Security (freedom) Privacy (idleness) Rooms (subsistence) Income (freedom) Personal Identity (identity) Plot (subsistence) Location (freedom) Status (identity) Environment (protection)

Satisfiers (Needs) Satisfiers Longevity (freedom) Place Attachment (identity) Durability (subsistence) Source: Author as derived from Max-Neef et al. (1991); Moore (2000); Ruonavaara (2016); Saegert (1985); Tognoli (1991) 24

The housing themes in Table 2.1 are the different ways that housing is institutionally/theoretically categorised, but these classifications are also how housing is satisfied. For example, the typological difference between a detached dwelling and an apartment, represent the different ways in which shelter is achieved, shelter being a derivative of subsistence. The benefit of anchoring this thesis’ analysis around these institutional satisfiers, is that while they are derived from fundamental needs (universalist perspective), they are still receptive to individuals’ aspirations (pluralistic perspective). The receptiveness of these attributes is important because of the need to reflect the various changes individuals experience in the life-course; these life-course events have a considerable impact on housing decisions (Feijten & Mulder, 2005).

However, since individuals’ aspirations represent both their attainable and unattainable goals (Holme, 1985), it would be detrimental to conflate individuals’ life-course circumstances with their aspirations. It would be more revealing to explore how life-course events influence consumption. Studies from Feijten, Hooimeijer, and Mulder (2008); H. Kendig (1981); H. L. Kendig (1990) have explored the influence that previous housing experiences have on future moves, as well as how life-course events influence transitions between different urban regions and housing types. In such studies, the focus is on causality. For instance, positive variations in income are linked with geographic mobility. This thesis takes a different path, in that the progression of the structures of housing activities forms the object of study.

These structures of housing activity will later be used to gauge variations in individuals’ consumption. The previous paragraphs have explained that physiological and psychological stimuli are responsible for the satisfaction of individuals’ needs. These needs are triggered by life circumstances such as child-rearing, which inform what housing attributes are desirable during the life-course (W. A. Clark & Huang, 2003). However, a household’s progression in tenure, residence or neighbourhood is also influenced by external realities beyond their control. Thus, this thesis examines whether the housing careers of individuals sufficiently captures the environmental, economic, and social motives that they require. Traditionally, housing has often been positioned as a single bundle of services that renters and homeowners trade and consume (W. A. V. Clark & Dieleman, 1996). However, since the various structures of activity (such as those in Table 2.1) are not acquired separately (Van Ham, 2012), a few attributes cannot be used to explain the totality of housing patterns, since individuals’ housing choices are based on more complex decisions.

Therefore, this research aims to understand whether households’ diverse housing needs are satisfied or largely disrupted by trade-offs. The need to have balanced housing needs is especially vital, since a given housing facet may influence how other needs are satisfied. For example, Saegert (1985) found that financial investment influenced the experience of home. Finances enhanced ownership and made people more willing to improve their housing conditions (Saegert, 1985). However, this does not suggest that the various housing facets cannot be understood as distinct processes. The three housing facets are interrelated, but this thesis will clarify the extent to which such interdependencies vary among various individuals and cohorts.

Therefore, the interdependence of the housing facets means that it is vital to establish what circumstances influence households’ prioritisation of these housing themes. This might suggest how much resources households are willing to devote to their housing circumstances. This process is necessary because normative beliefs in the housing market are responsible for sustaining housing needs as social and financial commodities (which will be explained in Chapter 4). Thus, it is not immediately clear if these packaged needs, such as that of “The Great Australian Dream” complement individuals’ pursuit of self-actualisation.

P. King (2004) states that housing needs can be externalised, and economic rationale can overshadow housing needs. This finding suggests that it is possible that some housing themes (Table 2.1) may take precedence. Hence, in this thesis, the three housing facets are not conflated and are separated to allow for an efficient understanding of the historical applications and the continued progression of housing interventions.

2.3 Factors that Constrain the Consumption of Housing Products Fundamental needs are universal and equally relevant since they emerge to relieve discomfort. However, the consequences of these needs and the specific satisfiers (housing products) are likely to be prioritised disproportionately. This is because since the underlying needs are intangible, the ways in which those needs are satisfied are based on the efficiency of economic products (Max-Neef et al., 1991).

In the housing market, the needs of identity, idleness, and participation correspond with home, the need for subsistence and protection corresponds with dwelling, and the need for freedom corresponds with house. These housing needs can emerge simultaneously and equally. However, the consequences of dwelling are dependent on all needs and potentially freedom. Similarly, without the underlying needs embodied in home and dwelling, the consequences of house are inconceivable since these financial consequences are tethered

26 to the structure and use cases of home and dwelling. The concept of home can emerge independently, since this ideal is intangible, though, home cannot be satisfied without the consequences of dwelling and potentially house. If the dwelling aspect of housing is absent, then needs cannot be stimulated through social and private interaction. However, the notion of house is only required within the market economy. This means that conceptions of housing exist in varying forms, though the dwelling and or home needs are indispensable.

Since the satisfaction of needs is based on the efficiency of housing products or goods, then self-actualisation is reliant on individuals consuming housing products that can satisfy all needs with maximum efficiency (Max-Neef et al., 1991 ; Zavei & Jusan, 2012). While the basic satisfaction of goods occurs through standardised housing products, self- actualisation involves specific responses to variations in taste (Gilead, 2013) (the life- course). Therefore, through the housing facets and consequences outlined in Table 2.1 and in Section 2.2, the universality of housing needs and products are expressed. To gauge the specific responses to taste, the progressive accrual of these needs, as well as the prioritisation of specialised financial motivations, are considered.

This process of progressively accumulating the core aspects of housing needs, such as adequate house sizes and house types, is often taken for granted in the housing market. This is because, through social mobility and residential mobility, it is anticipated that people will be able to make gradual changes in their housing circumstances which will lead to higher residential satisfaction. However, it is also possible that in housing markets with financial and socio-economic disparities, mobility inefficiently enhances residential satisfaction. When Sánchez and Andrews (2011) published their findings of residential mobility in OECD countries, it became clear that there is no independent relationship between housing or life- course needs and the incidence of residential mobility across OECD countries.

While in Australia, transaction costs for homebuyers are mid-range (within the context of OECD nations) at approximately 6% of the property value, Sánchez and Andrews (2011) have observed that nearly half of all residential mobility in Australia can be directly attributed to insufficient housing. While just over 50% of the reasons why people move in the Australian housing market can be attributed to reasons other than housing, such as family-related reasons, employment-related reasons and other unstated reasons, but even such reasons may be tangentially related to housing (Sánchez & Andrews, 2011).

In conclusion, this section has established that housing experiences need to contain a balance of all three housing facets so that the highest level of personalisation is attained.

Secondly, it is essential that financial motivations do not dominate since the need for financial freedom is an abstract exchange-value. The overshadowing of use-values will be further discussed in Chapter 3-4. However, simply put, in the case of house, when it is considered as an end, the future production of use-values will be diminished by the prioritisation of stored or differed value (Walker, 1974). Lastly, this section has questioned the use of residential mobility as an effective proxy for residential satisfaction. Housing progression may be better understood by the composition of underlying housing needs, housing aspirations (trajectory/progression/self-actualisation) and finally in terms of attitudes and the perceived improvement of the core elements of housing satisfaction.

2.4 Housing Needs Frameworks Section 2.2 has outlined the universality of needs and how they are satisfied, while Section 2.3 highlighted the factors that impede housing needs consumption. One such assumption is the utilitarian outlook, wherein a product’s worth is measured in an aggregate form such as housing stock or income/price (Rowley & Ong, 2012). Residential mobility was the example given previously, but the more common market assumption is that individuals’ spending behaviour reflects their housing experiences (Rowley & Ong, 2012). However, this utilitarian perspective misrepresents the complexity of subjective needs, interactions, and aspirations since opportunities and constraints are conflated.

Consequently, this section will highlight the need to situate individuals’ housing trajectories within a framework with well-defined boundaries. The proposed framework will be universalist in the sense that individuals’ housing outcomes are understood through predefined housing needs. However, this is not problematic insofar as the framework is sensitive to pluralism (capturing individuals’ diverse needs) (Nussbaum, 2001). The benefit of developing a framework that has pre-defined attributes but is nonetheless comprehensive is that the subjective experiences of disparate individuals can be compared with one another. This comparison reveals how balanced housing provisions are, from a needs- based perspective, rather than as a measure of market competition.

Historically, needs-based research has taken the form of housing needs analysis (HNA). HNA identifies households’ preferences among factors such as local environments, property features, dwelling characteristics, and amenities. HNA is, however, limited since preferences are not depicted as distinct responses that serve different social, biological, and physiological functions. HNA may be longitudinal, prioritising the preferences and the changes of individuals’ housing circumstances, this aspect is an improvement from a utility- based assessment of housing, but there remains a gap regarding the contextual basis for 28 individuals’ consumption. The classification of housing features in typical HNAs does not describe how people’s housing experiences are sustained by distinct motivations. Although from such analyses, a broad conclusion is that consumption is underpinned by financial capacity and alternative housing options; “Variability may be due to constraints, the effect of which are sometimes difficult to distinguish from the effects of wants” (Rapoport, 2000, p. 148). Thus, there is a need to distinguish between utility (quantities of consumption) and aspirations (wants). In Chapter 4, it will be explained that preferences (demand) are not analogous with aspirations but are simply a derivative of utility.

The disparity between individuals’ aspirations and their life circumstances is better observed through housing ladder analyses (HLA) such as housing careers, housing histories, housing pathways, or housing transitions. These frameworks are based on individuals’ housing narratives. The housing careers framework suggest that people’s residential mobility is geared towards increased wealth, comfort and opportunity. Beer, Faulkner, and Gabriel (2006, p. 1) state that “…. Households move to occupy a better dwelling or live in a better neighbourhood”. Thus, the housing careers framework is a combination of how aspirations and life-course triggers influence residential mobility.

Alternatively, the housing histories model does not fixate on the frequency of mobility but on how individuals move between different property types/classes. Housing is thus understood as accumulated property, capable of altering people’s life-course and not simply representative of their life goals (Davis, 1991 ; Faulkner & Beer, 2011). Housing is not simply a container of human ambitions but is a critical aspect of achieving those ambitions through price increases and financial security. The strength of this approach is that it does not decouple the external financial effects from people’s class identities, and the focus is on milestones such as tenure, income generation and family stability.

The housing pathways model builds on the housing careers framework but incorporates social meanings and relationships (Faulkner & Beer, 2011). This model evaluates people’s self-identity and their pathway towards lifestyle attainment, which is vital due to the increasing uncertainty of the life-course (Giddens, 1991). The uncertainty of the life-course is reflected in the increasing importance, competitiveness, and internationalisation of markets (M. Mills & Blossfeld, 2003). Hence, Giddens (1991) asserts that lifestyle opportunities are mainly the domain of the wealthy, and poorer individuals have less likelihood of shaping their lifestyles. Thus, households’ satisfaction is shaped by the limitations of their individual interactions (Clapham, 2002).

Due to the diversity of individual experiences, it is necessary to consider a framework that is both structured and dynamic. The housing transitions model is one such framework, which proposes that housing sequences are fluid and reflect life stages, financial resources, health, tenure and lifestyle values (Faulkner & Beer, 2011). Thus, housing transitions presents a multiplicity of sources influencing housing outcomes; individuals do not achieve one singular aim but gravitate towards different trajectories. This research has considered the housing transitions model and specifically the “housing decision framework”, developed by Faulkner and Beer (2011, p. 32), which includes five categories: demographic, health, housing history, aspirations and employment, as the determining factors for housing transitions. In this thesis, these factors are embodied in the life-course questions (see Questions 18-44, Appendix 21) to account for life-long variations in needs.

The life-course is paramount because the forms of housing that individuals require is flexible and changes at various life intervals. W. A. Clark and Huang (2003) state that throughout the life-course, individuals’ housing experiences are shaped by their biological instincts, relationships, family structures and economic circumstances. Individuals strive to maintain these social interactions, even when they move houses. Thus, the life-course is not separate to how we understand housing but represents when and how housing decisions are made. These phases influence the relative significance of needs (which the Brisbane GCCSA Housing Needs Survey explores). Thus, there is a distinction between life-course stages (Figure 2.1 below), and the general desire for satisfying needs (Table 2.1). Home, house, and dwelling are universal, though their manifestation is unique for everyone.

Figure 2.1 The Capacity to Express Choice within Housing over Time Source: (Faulkner & Beer, 2011) based on P. Williams (2003)

Thus, this thesis’ describes housing aspirations based on consumptive needs as well as external circumstances, since individuals have different consumptive needs based on the

30 life-course stages. This coupling is achieved by ensuring that the life-course needs of all cohorts are equally represented through themes such as employment, education, and family circumstances. Thus, the peculiarity of needs across different life stages is implicit in the broad life-course and housing facets. The equal representation of life-course factors means that despite the differences in the prioritisation of life-course needs, the various cohorts can be compared. However, the differences in agency across the life-course are considered. Life-course stages (young adults, mature adults or elderly) are influenced by individual, social, and environmental needs, as well as capital: social, economic, and political. Thus, as individuals progress through the life-course, their capacity to express choice increases (Figure 2.1 in the previous page). Life-course stages may manifest as either constraints or opportunities, depending on the capacity to express choice. Therefore, even though housing need satisfiers are generally similar (particularly within cultures) individuals’ housing experiences are a product of their awareness, financial capacity and the intervening obstacles that shape their choices.

Therefore, housing researchers have explored various models which explain life- course progression. One such model by Israel (2003) suggests that there are three housing stages: entry, transitionary and the ideal phase.

1. The Baseline Phase is the entry point into the housing market. Housing is chosen based on practicalities, such as convenience, costs, and social networks (Israel, 2003). 2. The Transitionary Phase represents the resolution of housing practicalities, and thus it will be expected that in addition to dwelling attributes, there will now be a considerable emphasis on the house (Israel, 2003). This entails the broader facets of commodities (durability, location, and cost, etc.) (Saegert, 1985). 3. The Ideal Phase signifies ideal housing based on all housing facets: dwelling, house and home (Israel, 2003). Home is vital since it is the expression of self and identity. Symbolic needs and values should be fulfilled at this stage (Ruonavaara, 2016).

However, the problem with this model is its inability to incorporate increasingly dynamic mobility patterns, which have arisen due to unaffordability, globalisation, and changes to cultural norms (Faulkner & Beer, 2011). While Subsection 6.1.2 will explore multiple housing-decisions phases based on age-brackets, this thesis focuses on the factors that enable or impede housing aspirations. This structuralist approach upholds the existing life-course literature whilst delivering findings that are relevant for dynamic housing patterns.

2.5 Australia as a Property-Owning Democracy The property-owning democracy (POD) was largely developed by John Rawls to free people from dependency on the state or free market (Jackson, 2012). Invariably, a housing market that functions as a POD is one in which there is a prevalence of homeownership (D. Rogers, 2016). This is because Rawls viewed property such as housing and businesses as a means of ensuring that a small subset of the population did not monopolise outcomes. A successful POD requires balanced financial territoriality, reduced monopoly power, equitable asset distribution and an investment in real productive wealth, to ensure sustainability.

Hence, this thesis will investigate whether there is a divergence between the territorial needs of individuals and the market. Thus, the functionality of supply-demand, which regulates individuals’ social, economic, and political needs, will be investigated. This exploration will determine if the market is self-regulating, and the role of monopoly power. In terms of asset distribution, this thesis will investigate whether segments of the population are disadvantaged (imbalanced housing needs) or experiencing rental or mortgage stress. The thesis will also assess the significance of growth in households’ financial liabilities (household credit/debt). Finally, this dissertation will explore if there is a disparity between capital gains and wage growth, which will outline the degree of market productivity.

Australia is considered a homeownership nation since its homeownership rate of 67% falls within the 65-85% range ascribed to property-owning democracies (D. Rogers, 2016). A more robust definition is citizenship and direct participation in the moulding of the housing landscape (D. Rogers, 2016). The property-owning democracy “…. Is one in which virtually every citizen has access to productive wealth” (Mandle & Reidy, 2014, p. 519). This does not imply that the property-owning democracy was conceived due to an economic growth rationale. Instead, the POD would allow wealth to be equally distributed among the population, preventing dependency on the state or on trickle-down economics (O'Neill & Williamson, 2012). The POD also involves the ownership of other financial assets (Jackson, 2012). This structure enables individuals to accrue wealth through ownership, rather than there being a capitalist class and a non-wealth holder class. In housing, productive wealth would be reliant on planning processes that facilitate equal distribution and sustainability, such as the coordination of territorial claims and a limitation of features that jeopardise productive wealth such as scarcity, speculation, and unaffordability. Chapters 3-4 will present how scarcity, speculation, and unaffordability are amplified by Australia’s extensive growth. A modern and historical account of Australia’s suburban history will also be given.

CHAPTER 3: HOW TERRITORIALITY HAS SHAPED THE AUSTRALIAN HOUSING LANDSCAPE This chapter theorises that the territorial tendencies (the tendency to control and contest geographic space) (Storey, 2012) within the Australian and Brisbane GCCSA housing markets are a form of causal mechanism. However, there are two forms of territoriality. There are structures that facilitate individuals’ needs (domestic) and the structures that control them (public) (introduced in Section 3.1 and upheld in Section 3.5). These structures diverge since causal mechanisms are not entirely sustained by individuals’ beliefs, but also by people being socialised through normative beliefs. The divergence between these two boundaries of socio-political space has constrained the diversity of housing opportunities (clarified in Sections 3.6-3.8). Assessing how individuals’ aspirations have been moulded through socialisation reflects the position of Archer et al. (1998), which is that a single structure should not altogether be conflated as the medium and the outcome of individuals’ action. Hence, the structures that blur the boundaries of domestic and public interaction will be explained in Section 3.1. Subsequently, Section 3.2 will introduce Australia’s housing narrative, which explains how housing has habitually been framed. Afterwards, the means- end chain theory will explain that housing is framed based on its superficial attributes rather than emotional ends (Section 3.3). Thereafter it will be explained that how housing is framed in Brisbane and Australia is likely to have been influenced by economic processes such as capital gains than changes in housing composition (Section 3.4).

3.1 Introducing Territoriality Territoriality is achieved through social construction; wherein more powerful territorial agents control subservient groups or outsiders through personalisation, defence, socialisation and or institutional power (Porteous, 1976 ; Vollaard, 2009). Territoriality involves the shaping of social interactions through geographical control (Bartel & Graham, 2016). Furthermore, there are two domains of geographical control: public and domestic.

Table 3.1 Territorial Boundaries and their Territorial Agents Territorial Domains Domestic Public Territorial Agents Households Developers, Authorities How to Identify Attitudes & Semiotic Patterns Settlement & Semiotic Patterns

In the public domain, developers and authorities are the territorial agents who dictate the political, social, economic, or environmental structures, while the domestic domain involves households as the territorial agents who express their identity, personalise and defend their home (Porteous, 1976). Table 3.1 above shows that the public is identifiable via settlement and semiotic patterns. Semiotic patterns show developers and authorities’

33 territorial tendencies, since in the process of shaping the public’s geographic territoriality, they influence the symbolism of discourse in the consumer market. This symbolism represents the interpretation of planning outcomes through socially constructed meaning.

Public territorial tendencies can also be revealed by studying settlement patterns directly. There is a continuum from Australia’s historical settlement patterns towards modern-day planning models. Historic patterns allowed for large allotments to facilitate agricultural and productive self-sufficiency (Freestone, 2010). Consequently, Australians’ place attachment facilitated an effective transition to high ownership rates in the anti- communist property-owning democracy paradigm. The extensiveness of this spatial expansion has then influenced the present-day version of the Great Australian Dream, which promises individuals the dream of owning large plots of land. The Great Australian Dream embodies the desire of Australians to own their own home, especially a free-standing dwelling on a large lot in suburbia (Kohl, 2018 ; Yardney, 2018). However, there is a key difference between the spatial expansion of old and that of the present, which is that the paradigm has shifted towards financial value from self-sufficiency (Stapledon, 2007).

However, the central aspect of the Great Australian Dream (place attachment) persists, due to the lasting effect of territoriality on the Australian housing market. This influence endures since defunct structures from the past influence present-day interactions. New or renewed normative beliefs and institutions are being built on historical foundations, and those pre-existing buildings, allotments, or spatial patterns influence future outcomes due to the new territorial claims which sustain them.

Moreover, the persistence of territorial tendencies is not merely ideological. In the housing market, it is uncommon for pre-existing communities or planning to be entirely dismantled. This is partly due to the impracticality of constructing entirely new communities, and partly because viable expansion and growth may be prioritised more so than anticipating future adaptability. In the case of Brisbane, Gallagher, Sigler, and Liu (2019) have shown that the most financially viable redevelopments for developers are detached dwellings; thus, the market favours increased fragmentation to the detriment of urban consolidation. Hence, housing is durable and normative beliefs, and physical planning outcomes are likely to be influenced by what has come before, due to opportunities, constraints, and place attachment. The legacy of historical territoriality influences future territoriality, not due to reification but practicality. This process is a gradual transitioning between different territorial claims, held together by durable institutions, infrastructure or planning models.

Thus, historical, and present-day territoriality overlap since there are shared benefits for authorities and households such as the desire for homeownership and an inclination towards large allotments. However, the domains of domestic and public territoriality are also liable to conflict since normative beliefs are sustained for different reasons. For public authorities, there are aggregate benefits that stem from housing growth, such as increased housing taxes and stimulating construction growth (The Committee for Economic Development of Australia, 2017). Oppositely, in the domestic domain, the focus is on private ambitions, such as privacy and continuity. Growth, privacy, and continuity are not necessarily compatible tendencies, and this will be investigated in Subsections 6.2.1-6.2.3.

3.2 A Brief Introduction of Australia’s Housing Narrative Australia’s geographic territoriality can be examined by evaluating semiotic patterns. These patterns encompass the diverse beliefs, needs, aspirations, and assumptions of individuals in the Australian housing market. These diverse needs and aspirations can coexist in a common marketplace due to some shared normative beliefs, social norms, and attitudes. In this chapter, the focus will be on the former. Normative beliefs in the housing market are largely represented through geographic and physical expressions, but they are also expressed through symbols such as the housing ladder. This is because although meaning has a geographic dimension, that dimension is shaped through interpretation, communication, and negotiation (Horlings, 2015). Thus, symbols such as in the housing ladder reveal some aspects of geographic territoriality. The ladder defines individuals’ housing aspirations within a seemingly progressive hierarchy of consumption, making it appealing to the Australian public. The ladder is assumed to refer to a homeownership trajectory (Ho & Wong, 2009), or at the very least culminate in homeownership as explained by Wulff and Maher (1998). For instance, reverting to private renting from owning is considered a step down the symbolic ladder (Wulff & Maher, 1998). Secondly, an upward move in the ladder signifies increasing asset yields (Lemanski, 2011). These symbolic rungs are shown in Figure 3.1 on the next page.

Thus, the promise of homeownership and profitability are interconnected. Even when the housing ladder is characterised by dwelling types, beginning with basic apartments and culminating in lavish detached dwellings, the underlying basis is price-driven. This sequence originates from Australia’s yearning for detached dwellings; as noted by Baum and Wulff (2003, p. 8), “The expressed preference for homeownership is closely intertwined with the preference for a detached dwelling”. Hence, housing ladder discourse is primarily centred on homeownership, capital gains and the prevailing house types (Baum & Wulff, 2003).

Figure 3.1 Conceptualisation of the Housing Ladder by Typology, Tenure and Family Composition Source: (Faulkner & Beer, 2011) However, the housing ladder is limited since it relies on normative beliefs that poorly represent housing needs and contexts. The primary benefits of the housing ladder are permanence and wealth generation, which are largely beneficial due to regulatory advantages (tax, welfare concessions) and tenure; tenancy laws inhibit permanence and personalisation in the rental market (Daley, Coates, & Wiltshire, 2018). Moreover, individuals who successfully climb the housing ladder and have better outcomes such as better education and health are also likely to be those who can afford housing in the first instance (Daley et al., 2018). Thus, to determine if the housing ladder is simply a self-fulfilling prophecy, in Sections 6.1-6.2 and 7.2, the housing market’s structure will be studied to reveal how regulatory, financial, and territorial tendencies shape the ladder.

3.3 Introducing the Theory of Means-End Chain Individuals’ housing needs and consequences are diverse, as explained in the preceding section. Hence, it may be tempting to consider that people are indiscriminately drawn towards various housing ideals. This section will instead explain how individuals’ behavioural intentions are structured through means-end chains (MEC). MECs explain that individuals’ behavioural intentions when dealing with products in the market are derived from underlying needs (emotional ends). Thus, it is the acknowledgement of the emotional benefits associated with housing that drives consumption. The tangible attributes of housing (how housing is measured and quantified in the market), such as the various forms of housing described in the housing ladder are simply the means to the (emotional) ends. This notion is in contrast with the market-derived perspective of the housing, which suggests that underlying housing needs are secondary and implicit. This approach puts the cart before the horse since housing attributes are evaluated instead of individuals’ underlying needs.

In this section, how underlying emotional benefits transition to housing attributes (housing/dwelling features and products) will be clarified. While it is true that individuals associate the tangible attributes of products with desirable consequences (Veludo-de- Oliveira, Ikeda, & Campomar, 2006), it is more likely that they are drawn to products directly as a result of their housing needs (James, 2019 ; Veludo-de-Oliveira et al., 2006). Such needs include subsistence, which is the need for wellbeing, health, and adaptability (Max- Neef et al., 1991). Individuals with such needs (emotional ends) then consider the consequences that are likely to arise from those emotional ends. These consequences and benefits are the social outcomes that they will experience. Following this evaluation, individuals then consider the attributes of their desired products.

MECs show how consumers’ reconcile the different components of a product (Mahoney, 2003 ; Zaltman, 2003). For example, individuals do their best to form an understanding of how housing is marketed via the attributes of dwelling and house, as well as how those attributes are associated with their underlying needs and wants. This understanding allows people to interface with residential listings in the housing market. Therefore, though MECs suggest that individuals will learn which attributes embody their desired emotional ends, this does not imply that it is the attributes themselves that drive consumption. Individuals attitudes towards specific housing attributes will change based on the suitability of use-values and exchange-values as they progress through the life-course, or based on how they are socialised into new normative beliefs. Moreover, housing is not mono-faceted (it contains a variety of attributes); thus, housing needs can be satisfied through many housing variations. People may purchase a housing product such as inner- city apartments, but this represents the attributes (means) not the emotional end. It is vital not to focus entirely on such attributes, due to the risk of emotional ends being unfulfilled or due to the potential for other consequences or satisfiers that may address people’s needs.

In Table 3.2 on the next page, three MECs are detailed, which show that there are emotional ends based on the six fundamental needs. These needs are derived from the human scale development of Max-Neef et al. (1991), specifically: identity, idleness, participation, protection, subsistence, and freedom. The category of freedom was expanded to incorporate financial freedom. These needs and motivations correspond with the housing framework of Ruonavaara (2016), which consists of the social house, the physical dwelling and the financial home. These elements are conceived as the broad containers of the housing consumption experience (from a universalist standpoint). Individuals desire to belong to a place, interact, receive shelter, or derive financial freedom. Due to varying life

37 circumstances and transitions, the benefits which individuals derive from housing will reflect different compositions of this framework. It is important to unearth these benefits so that the housing needs survey implemented in this thesis is not wholly fixated on housing attributes. Using individuals’ anticipated housing benefits as the basis for user preferences will then allow this study to represent the underlying intentions of consumers.

Traditionally, MECs are formed using one-on-one/in-depth interviews to reveal the benefits of importance to consumers (Botschen, Thelen, & Pieters, 1999). Scholars separate the end (use and exchange values) from the means (attributes) by using laddering to gradually direct individuals’ perceptions of attributes towards the consequences of those attributes, and finally the underlying needs. Coolen, Boelhouwer, and van Driel (2002) used this approach to model housing preference predictors, while Bonatto, Miron, and Formoso (2011) used the model to generate a hierarchy of how value is formed in social housing.

Table 3.2 Means-End Chains Identify Emotional Ends by Retracing Attributes’ Root Causes Emotional End Consequence Attribute Xn Yn MEC 1) Identity Concept of

1 2) Idleness, 3) Participation (Home) MEC 4) Protection 5) Subsistence Particular: Bedroom Count, Specific Type 2 (Physical Wellbeing) Lot Size, House Type of (Dwelling) Fundamental Needs Housing Satisfier (Use-Values) Facet Particular: House Price, MEC Specific Type 6) (Financial) Freedom Rental Price, Bond, Mortgage 3 of (House) Repayment, Capital Gains Fundamental Need Satisfier Housing Facet Y3 (Exchange-Value) Source: Author as derived from Easthope (2004); Max-Neef et al. (1991); Ruonavaara (2016) Conversely, this thesis identifies needs based on predefined fundamental needs (see Section 2.2). These needs are the emotional ends that are required for a positive housing experience. While needs emerge to allay deprivation, they ultimately present potential (Max- Neef et al., 1991). In the housing market, that potential is sought based on overarching identities (attributes) that approximately embody the representation, perception, price, and usefulness of those needs. These identities refer to the notions of dwelling and house. This is because these facets have consequences that are either physically measurable and or can be counted (Veludo-de-Oliveira et al., 2006 ; Zachariah & Jusan, 2011) (also known as commodification) (Marx, 1903 [1887]). Thus, from a market standpoint, these are the least abstract MECs (Ha & Jang, 2013) and include measurable consequences such as bedroom

38 counts and house prices. Conversely, home is the most abstract MEC (from a market standpoint), and its consequences must be indirectly found through dwelling and house. The only way to freely study home is by identifying its emotional ends. Thus, Table 3.2 on the previous page shows that all MEC chains (MEC 1,2,3) have emotional ends (based on the six fundamental needs), though MEC1 does not have direct consequences (satisfiers).

Contrariwise, the underlying basis for the attributes of dwelling and home (starting from Yn) can be unearthed by identifying the specific benefits that they offer the consumer. In the case of dwelling (physical quantities), and for house (monetary quanitites). Then, the emotional ends related with those quantities reveals the underlying needs (ending in Xn).

Moreover, Table 3.2 also shows that retracing the root causes of attributes is also vital because as use-values transition towards exchange-value (Yn to Y3), there are fewer emotional ends being satisfied. This process will be better explained in the next chapter, but simply, since financial freedom allows consumers to access other goods (and associated needs) in the market, it is a form of deferred value (Walker, 1974). This differed or stored value means that emotional ends of house are based on future uses, which is in contrast to the emotional ends of dwelling and home (in particular) which are based on more immediate needs. Therefore, in a housing market that fixates on the concept of house, there is likely to be even less emphasis on retracing underlying needs.

Accordingly, this thesis studies the consequences and emotional ends of the housing facets to identify how attributes can be enhanced. Emotional ends are investigated because they represent the underlying housing motivations, while investigating consequences ensures that individuals’ feedback is comparable, and can also be relevant for policymakers. This holistic study prevents fixating on attributes to the detriment of their underlying needs (commodity fetishism) (Marx, 1903 [1887]). Attributes only partially reflect individuals’ subjective values. People form an understanding of which house types embody their needs, not only on a price relationship but also based on experiences and assumptions about the use-value of different house types. When people aspire to a detached dwelling on a ¼ acre plot, this aspiration reflects various wants, such as price, function, and identity. Yet, attributes (detached) and consequences (¼ acre plot) are not necessarily analogous with individuals’ underlying needs but are based on what they believe is obtainable.

There is no one-to-one correspondence between needs and satisfiers. A satisfier may contribute simultaneously to the satisfaction of different needs or, conversely, a need may require various satisfiers in order to be met. (Max-Neef et al., 1991, p. 199).

In a market, there are thousands of distinct housing configurations that can be classed under a single dwelling or house conception. However, due to the subjectivity and intangibility of emotional ends, use-values are less influential in the market. Hence, the attributes of dwelling or house are often the objects of study. This means that policymakers and developers often presuppose that the embedded home, house, and dwelling needs are relevant. However, the fulfilment of these needs cannot be taken for granted. There are different modes of satisfaction in housing. Specific forms of satisfaction may be loosely associated with particular dwelling types, but satisfaction is an ongoing assessment made by diverse individuals, and a given attribute will satisfy individuals’ needs disparately.

…. however, the form of production and consumption of goods makes goods an end in themselves, then the alleged satisfaction of a need impairs its capacity to create potential. This, in turn, leads to an alienated society engaged in a senseless productivity race. Life, then, is placed at the service of artifacts, rather than artifacts at the service of life. (Max-Neef et al., 1991, p. 25).

Thus, a balanced housing outlook requires the recognition of underlying use and exchange values, in addition to consequences and attributes. If the outlook is imbalanced, then housing attributes are likely to be conflated, which limits the prospect for alternative adaptations (discussed in Section 3.4). Hence, it is important to separate individuals’ fundamental needs from the attributes of dwelling and house, which allows for a better understanding of the basic needs and consequences that embody housing needs.

3.4 What Form of Housing? The Interchangeability of Housing Terminology The previous section has explained how domestic territoriality is constructed in principle. This section will provide evidence to show how the domestic territoriality of individuals and that of the public are inconsistent. This will be shown through the semiotic differences between the housing facets. Housing terminology (symbols of the market) is used as a medium to compartmentalise or facilitate everyday interactions. Since individuals objectively acquire attributes (means) rather than consequences or ends, when the housing facets of home, house and dwelling are interchanged by the public, the underlying consequences or use-values are not immediately evident. However, there are specific differences between housing terminology, in real-world application and this thesis’ ontology.

Individuals substitute different housing terms for varying reasons, so it is vital to study who benefits from this process. Domestic territoriality is shaped through personalisation, which involves the attainment of the broad spectrum of housing aspirations and needs.

Conversely, public territoriality leverages diverse interests for growth. Hence, this section highlights how the interchangeability between the terms ‘homes’ and ‘houses’ does not correspond with the physical state of the housing market. Instead, Australia and Brisbane’s geographic territorialisation has resulted in a less diverse built environment (Queensland Government, 2019) based on spatial expansion and growth. The predominant narrative in the Australian housing market is who has access to homeownership - a financial imperative.

Understanding how individuals’ territoriality are shaped reveals if the medium of personalisation reflects individuals’ housing fulfilment or corresponds with the market-driven territoriality. A simplification of housing as a financial vehicle is more conducive for governance and developmental interests, and conditions away individuals’ needs and housing aspirations. In Figure 3.2 below the gradual decrease in Australia’s search interest (popularity) for the term ‘homes’ and the gradual increase in search interest for the term ‘houses’ shows the divergence between the two terms. The averages for ‘homes’ and houses’ are the relative averages of all the monthly search interests relative to the peak search for ‘homes’ in January 2010. The author calculated the percentage split for national searches as 43% for ‘houses’ and 57% for ‘homes’. The period included web searches from 2004 to October 2019. Searches for the term ‘dwellings’ were negligible relative to the other search terms. It was crucial to input all search terms together since search interest was more reliable when terms were compared to one another. Otherwise, search interest growth would be conflated with the general increase in the Google search corpus.

Figure 3.2 Usage of the terms ‘Houses’, ‘Homes’ and ‘Dwelling’, Google Trends (Web Searches: not Seasonally Adjusted) Source: Google (2019)

The regional dissection for the state of Queensland was computed by Google, showing a stronger search interest for ‘houses’ in the state than nationally. The disparity was even greater for the Brisbane region, with a 13% increase from the national percentage for ‘houses’. This data shows that search interest for the term ‘houses’ is increasing nationally and sub-nationally, particularly in Queensland, Northern Territory and Tasmania. Within Queensland, the strongest search interest for ‘houses’ is found in North Queensland, although Brisbane and other SEQ cities remain well above the national average.

The context of the terms ‘houses’, ‘homes’ and ‘dwellings’ in the google web searches at the national and state levels were parallel to this thesis’ classification. This was tested by examining the related queries for the search terms. ‘Houses’ was related to ‘sales’, ‘lease’ and ‘rent’, ‘homes’ was related to types of living such as ‘container homes’ and ‘affordable’; ‘sales’ and ‘rent’ were relatively less prominent. The results showed that each term had a distinct usage, though there was some overlap. The variation in usage, however, did not correspond with significant changes in the composition of housing types in the Brisbane GCCSA housing market, which will be explained in more detail in Section 3.7. While the changes in dwelling stock were not negligible, they did not reflect a paradigm shift in the physical representation of housing in the Brisbane GCCSA. Although search interest is likely to mirror housing demand, most available houses are existing dwellings. Between 2006 and 2016, 69,063 high density and medium density dwellings were added in the Brisbane GCCSA, representing 43% of newly added stock, but this remains a fraction of total dwelling stock (901,796) (Profile id, 2019). In September 2019 for example, between 81 to 87% of housing for sale on the market were established dwellings (7-10 years old), and upwards of 98% of housing for rent were established dwellings (Domain.com.au, 2018). The gap between separate housing approvals and apartment approvals (adjusted for population) has only been significantly bridged in the past 5 years (Shoory, 2016 ; Warwick, 2016).

The greater change in the housing market has been the increased demand (as early as 2003) from buyers looking to gain housing market exposure (Reserve Bank of Australia, 2003). This demand has been influenced by the desire for capital gains. Thus, housing attitudes have changed more than the composition of housing types and are likely one of the contributing factors towards the increased inclination for people to employ housing vocabulary interchangeably (Room, 1985). This is important because terminology varies, but the built environment is more stable and fundamental needs even more so. Therefore, since the fulfilment of housing needs cannot be understood superficially, this thesis will strive

42 to understand how all housing facets are facilitated by the housing market, as well as how they are prioritised by individuals.

3.5 The Territorialisation of Australia’s Settlements: The Emergence of the Great Australian Dream and Housing Commodification In addition to semiotic patterns, the public domain of geographic territoriality can also be revealed through settlement patterns. Australia’s urban territoriality can be traced to the 19th- century planning objective of transitioning Australia’s penal settlements into organised, controlled, and self-sufficient grids and allotments. This strategy commenced with the Darling regulations in 1829, which was a British-derived regulation consistent with those in many British colonies. Both the original British hierarchy and the Australian derivative involved the distribution of land to productive tenants, though the crown retained control. Society was stratified and controlled based on law rather than economic liberty or design (Anthony, 2007). Thus, the British feudal order soon became unattractive due to consequences of industrialisation such as overcrowding and unhealthy environments. In the early 20th century, ‘Garden City’ ideals emerged in Britain to resolve these problems (Freestone, 2010 ; Troy, 2017). There was an impetus to resolve these problems due to emerging democratic notions of property rights (Troy, 2017). The ‘garden city’ movement was spearheaded by Ebenezer Howard and Raymond Unwin and involved a combination of garden (country) and town aesthetics, which were realised with green belts and satellite towns (Freestone, 2010).

Thus, two forms of organisation were inherited from Britain; the first was centred on control and the latter on freedom and health. The Darling Regulation encompasses both ideals and had three significant features: the size of the allotments, the rectilinear shape of the allotments and the grid pattern of the urban environments. The allotments were large to provide residents with sustenance, rectilinear to allow for efficient resale, and the grid pattern was implemented to tame the landscape and provide order. ½ acre and ¼ acre blocks were commonplace, and the latter would have a significant imprint on Australia’s planning history, as it became the standard allotment implemented throughout the country.

Moreover, in the 19th century, Australian allotments were also influenced by the desire to prevent the spread of fire and diseases (Troy, 2017). Thus, Australian towns featured sizeable allotments with a suburban appearance, though not influenced by garden city ideals, which would emerge much later in the early 20th century (Davison, 1995 ; Freestone, 2010). Australia was not considered a suburban nation at the beginning of the 19th century simply due to the undeveloped urban cores that Australian towns featured. The term suburb 43 was used differently, describing peripheral wastelands. However, by the 1830s, the contemporary definition of the suburb was prevalent. Colonial governments defined clear differences between country, town, and suburban allotments (Davison, 1995). By the 1830s- 1840s, new suburban forms emerged in Australia, known as garden suburbs, and featured specific allocations for open spaces, parks, and recreation.

Development in Australia was low cost with few barriers, and the growth of the Australian suburban template was unprecedented (Davison, 1995). These suburban models also emerged when Australia was still a new nation; thus, Australia’s suburban settlements formed before its urban cores were consolidated (Davison, 1997). By the 1880s, more than 50% of all homes were owned or being purchased, and Australia was dubbed by Adna Weber, as the second most urban nation and most suburban (Davison, 1997).

Despite the need for control via geographic territoriality, suburbs were universally adopted. This is in contrast to British suburbs which were the exclusive domain of the wealthy (Davison, 1995). In Australia, there were fewer land constraints, high incomes, cheap land, and the settlements were well serviced with modern infrastructure. Due to the even distribution of infrastructural services, Australians experienced minimal class differentiation, and suburbs were differentiated by their natural amenity rather than infrastructure. Suburban land was easily attainable for first homeowners (Davison, 1995) and Australians associated homeownership with suburbia and a high standard of living. Australians then valued homeownership above factors such as the quality of neighbourhoods or building standards (Davison, 1993).

Thus, the suburban ideal became deeply entrenched in Australian mindsets. Though the structuring of the communities was shaped by authorities, Australian suburbanisation was a democratic process, as people actively participated in shaping the by building and owning their homes (Davison, 1993). Hence, the Australian mindset was geared towards homeownership above all else. “Australians would prefer their cities to be uglier, and even perhaps a little more dangerous, than to give up the prospect of owning their home, however humble, poorly serviced and unplanned it may be” (Davison, 1993, p. 67).

Therefore, the same mechanisms which enabled suburban growth also facilitated homeownership. Owning one’s home and living in the suburbs were mutual realities from the onset, and this exemplifies the Great Australian Dream. This is important because the factors which influenced this were driven by the territorial considerations of ordering society, enabling self-sufficiency, and promoting homeownership. These processes were not

44 dictated by individual landowners, notwithstanding their participation in constructing homes and holding title. Australians have largely adhered to this territoriality because they have benefitted in several different ways. For example, in the 1950s onwards, when the Menzies’ government promoted homeownership as a means of combating communism (ABC Education, 2010). However, territoriality has a dual purpose of securing property rights for individuals and protecting the state’s interests. This distinction makes it important to examine the motivations and attitudes that led to the proliferation of suburbanisation in Australia.

Moreover, the distinction between individual and market-driven territorialities is less evident in the commodity-centred view of housing in which a limited subset of attributes is used to represent residents’ total housing patterns (Saegert, 1985). While Australian suburbanisation was not initially spearheaded to commodify housing, suburban ideals had the basic elements (price, durability, and location) that advanced the perception of housing as a commodity (Hannonen, 2015 ; Saegert, 1985). These features have taken precedence over cultural preferences and housing needs, and Saegert (1985) states that requirements such as the cost of land relative to location, borrowing costs, zoning regulations, and construction costs are now the biggest determinants of housing consumption. While these factors are influenced by cultural preferences and housing needs to an extent, the strength of that relationship is based on how well the various territorial claims are coordinated. As per Davison (1993), the Australian suburb satisfied five prime wants: privacy, natural amenity, semi-rural surroundings, healthy environment, and private ownership. These prescriptions were more attainable historically but are presently elusive since the efficiency and moderation of land use have been eclipsed by a few growth-driven priorities.

Thus, this thesis will study the coherence of individuals’ attitudes towards the housing market to understand how attainable or idealistic the various motivations are. This analysis probes the normative beliefs which have unified historic suburban idealism, post-war housing, and housing citizenship (D. Rogers, 2016). The Great Australian Dream comprises ideals such as tenure, the property-owning democracy, and suburban living (ABC Education, 2010), and these ideas may be incongruous. The market-driven territoriality governs broad structures such as urban settlement patterns, financial structures, regulations, and typologies that have been standardised for optimisation. However, individuals’ aspirations are not structured to the same degree, outlooks are fragmented, and organised groups such as growth coalitions are uncommon (D. Rogers, 2016).

3.6 Australia’s Dwelling Profile The Australian and Brisbane GCCSA dwelling profiles will be introduced to grasp Australia’s enduring territorialities. Of the 9.9 million dwellings in Australia, 7 million are separate houses, also known as detached dwellings. These are typically found in low-density configurations in the suburbs. There are only 727,095 single storey semi-detached, row, terrace, or townhouses and 542,028 semi-detached, row, terrace, or townhouses with multiple stories. Flats are found in similar numbers, with approximately 500,000 one or two- story flats or apartments, 364,439 three-storey flats or apartments, 533,935 four stories or more flats or apartments, and less than five thousand flats attached to a house. There are also approximately 30,000 flats or houses attached to shops, while the rest of Australia’s dwellings consist of temporary or mobile housing (Australian Bureau of Statistics, 2016b).

Detached houses are the prevalent Australian dwelling type, and the average Australian lives in a three-bedroom, two-vehicle household, owned on a mortgage. These variables are calculated independently by the Australian Bureau of Statistics (2017a). This is similar to the ideal Australian home, which is a four-bedroom house with two bathrooms, a two-car garage, situated on a 665 square metre block, and costing $650,000 (Cahill, 2016). This profile highlights the persistence of the Great Australian Dream and is based on the most highly engaged listings on realestate.com.au. Australians also love a large home, and the REA group report that over 40% of Australians wanted a larger home, while only 5% wanted to downsize. Yet, poor infrastructure and housing unaffordability are changing this trend. Between 2014 and 2017, there was a 16% reduction in average Australian lot sizes (Urban Development Institute of Australia, 2018). “It’s a dramatic change from land sales 30 years ago where “quarter-acre” blocks (1011 square metres) were the norm” (Heagney, 2018).

3.7 Brisbane GCCSA Dwelling Profile The dwelling profile of the Brisbane GCCSA is even less diverse than Australia’s. In 2016, 74.4% of the dwellings were separate houses, compared to 71.1% for Australia, 15.3% of dwellings were medium-density (17.9% for Australia), and 9% of dwellings were high density (9.1% for Australia) (Profile id, 2019). These percentages were derived from the Australian census of 2016 and have not shifted much since the 2006 census. Between 2006 and 2016, the share of separate houses in the Brisbane GCCSA declined by 4.2%, medium-density types increased by 1.1%, and high-density types increased by 2.9% (Profile id, 2019). The limited growth of diverse typologies is more evident among homeowners. 88.4% of homeowners live in separate houses, 5.16% live in semi-detached/row/terrace/townhouses,

46 and 5.53% live in apartments. Conversely, 56.2% renters live in separate houses, 18.03 live in semi-detached/row/terrace/townhouses, and 25.01% live in apartments (Profile id, 2019). Thus, there is a cohort divide in housing densities.

Another way to gauge housing diversity is the number of rooms per typology. While there has been a considerable push for higher density housing (Queensland Government), detached dwellings in the outer suburbs remain more suitable for families than apartments in the inner city, when considering room counts. The Australian Institute of Health and Welfare (2017) view dwellings with one less bedroom than required as overcrowded, while a dwelling with two more bedrooms than required is underutilised. A one-bedroom surplus is not the benchmark since people (such as the elderly) may require an additional bedroom for guests (Adeniyi & Johnson, 2018). Figure 3.3 on the next page shows that excess rooms (top-right and bottom images) correspond with the location of separate houses (detached), shown in Figure 3.5. While Figure 3.4 shows that room shortages correspond with the location of flats (apartments), shown in Figure 3.5. Another way to understand the significance of the asymmetric housing landscape, is that the average household size in Australia is 2.6 persons per household (Australian Bureau of Statistics, 2015), yet this household size is inadequately facilitated. While market forces are relied upon for zoning, delivery, and composition of these housing products, through the coordination of land-use and zoning, developers will have greater flexibility to trial new housing products.

There is a need to trial new diverse housing products since diversification enhances individuals’ capacity to prioritise housing attributes (Gurran et al., 2015). The demand for ‘starter homes’ (less suitable but affordable) (Duke, 2017) reflects the lack of choice afforded to first homeowners at the lower end of the market demand curve (Kelly, Breadon, & Reichl, 2011). This reality has been exacerbated by individuals adhering to existing normative beliefs (Kelly et al., 2011) relating to historic territoriality and present-day place attachment. Moreover, Case (1994) mentions that expectations of future price increases can be a primary motivation in housing choices. Finally, imperfect market knowledge and uncoordinated planning limit individuals’ decision-making. These are all issues which complicate our understanding of demand. Although some degree of spatial polarisation is inevitable, excessive polarisation results in area-type constraints, which are the localisation of typological features by area (Maclennan, 2012). These area-type constraints are the contributors of the polarisation between spare and needed rooms shown in Figures 3.3-3.4. The spare rooms are largely situated far from the Brisbane City inset, while the needed rooms are found in homes near the Brisbane City inset.

Figure 3.3 Housing Suitability (Spare Rooms) as a % of Total Dwellings of the Brisbane GCCSA Source: Author as derived from (Australian Bureau of Statistics, 2016a) 48

Figure 3.4 Housing Suitability (Needed Rooms) as a % of total Dwellings of the Brisbane GCCSA Source: Author as derived from (Australian Bureau of Statistics, 2016a)

Figure 3.5 Dwelling Types as a % of Total Dwellings of the Brisbane GCCSA Source: Author as derived from (Australian Bureau of Statistics, 2016a) 50

3.8 How Models of Neighbourhood Change Explain the Brisbane GCCSA Settlement Pattern

Due to the far-reaching impacts of urban phenomena such as social polarisation, it is important to understand how and why settlements evolve. Various researchers have devised models of neighbourhood change to explain how urban communities develop. These models are: ecological, sub-cultural, and the political economy (Pitkin, 2001 ; Schwirian, 1983). This section introduces ecological models such as the bid-rent theory and the neighbourhood life cycle theory. These models attribute neighbourhood change to structures (such as the economy and migration) and estimate neighbourhood and individual utility based on expected consumption patterns. The bid-rent theory assumes that individuals with similar socioeconomic and demographic profiles share similar preferences and capacities. Broadly, people will trade-off residing near the CBD where employment, services and commuting times are ideal and living where rent is affordable. This relationship exists since the city centre thrives on the profitability of intensive land uses and dense populations, while farther away from the CBD, density is lesser, with more extensive uses (Hardie, Parks, & van Kooten, 2004). Thus, richer individuals tend to live near the CBD, while poorer individuals tend to live farther away, where rent and land are cheaper. However, poorer individuals sometimes opt for more compact living in the city centre, where they can benefit from lower transport costs, employment, and other services (Pitkin, 2001).

The bid-rent relationship can be observed in all major Australian cities: Sydney, Melbourne, Brisbane, Perth, and Adelaide. Although in the 1980s and 1990s, this trend was hardly discernible (Wargent, 2016). Sydney and Melbourne, in particular, have steep gradients, with real house prices exceeding $1,000,000 (2016 dollars) near the CBD (within 10km) and sloping down to approximately $500,000 and $750,000 for Sydney and Melbourne respectively, in the settlements situated 40 kilometres from the CBD (Metcalfe, 2016). The Brisbane curve is more moderate with house prices of approximately $700,000 near the CBD, sloping down towards $500,000 20 kilometres from the CBD (Metcalfe, 2016). The greatest point of difference between the Queensland capital and Australia’s two largest cities is that house prices have a steeper gradient in Melbourne and Sydney within the first 20 kilometres from the CBD.

In Australian capital cities, the CBD, inner and middle rings typically enjoy higher prices because of the feedback loop created by population densities and employment, due to agglomeration, which further incentivise people to invest and live in such areas. This concentration of dwellings also means that land-supply is restricted near the CBD. Importantly such high densities would be unsustainable without a considerable infrastructural backbone. Public transport is most reliable and utilised near the CBD, beyond 20-24 kilometres, public transport patronage by commuters plummet below 10% (Loader, 2018). This is because outer urban growth (auto suburbs) beyond 20-25 kilometres (Loader, 2019a) is typified by low-density urban sprawl (OECD, 2018). The cost of servicing low-density populations with infrastructure is higher than for higher-density areas; thus, population growth has a greater impact in the city centre (active core), inner and middle rings (transit hubs) than elsewhere.

Conversely, density, dwelling size, building heights, house, and land prices, are the defining factors beyond 15 kilometres of the CBD. This is due to stronger land supply and unfettered growth. Thus, population growth puts pressure on prices within a 15-kilometre radius from the city (suburbia) due to desirable infrastructure and building constraints (Wargent, 2016). Most Australian city dwellers live outside of the active core where active transportation such as walking or cycling to work is 50% greater than the metropolitan mean of their respective cities. In Brisbane, few residents live in the active core or in transit suburbs where public transit patronage is 50% greater than the metropolitan mean (Gordon, Maginn, & Biermann, 2015).

The geographic distribution of the active-core, transit suburbs and auto suburbs can be studied using the bid-rent theory. Figure 3.6 on the next page shows Brisbane GCCSA tenure types; the top-right image shows the mortgage belt in a dark blue shade. The mortgage belt houses households with a significant percentage of their household disposable income composed of mortgage debt. This is the typical structure of Australia cities with renters in the middle (the middle-left image), outright homeowners centrally located (the first image), and mortgagors occupying the belt between these geographic areas (Department of Infrastructure and Transport: Major Cities Unit). The mortgage belt is dominated by detached dwellings and lacks a diverse housing mix. Older Australians with less mortgage debt have been relocating from these inner and middle suburbs into the outer fringes (Department of Infrastructure and Transport: Major Cities Unit). Thus, the inner and middle suburbs are the domain

52 of first homeowners and investors. First homeowners are situated there to enter the housing market while being near key infrastructural services; investors, on the other hand, might invest in those extensive margins to benefit from speculative gain.

Figure 3.6 Tenure Types as a Percentage of Total Dwellings in the Brisbane GCCSA, Statistical Area (SA2) Source: Author as derived from (Australian Bureau of Statistics, 2016a)

However, the abovementioned analysis is limited because prices are influenced by dynamic factors such as the quality of housing stock, the vitality of the population and future developments. The neighbourhood life-cycle theory suggests that there are various stages in the life-cycle of a neighbourhood (known as the Hoover Vernon model) to account for this complexity. In the first stage of this life-cycle, new single- family residences emerge, in the second stage people transition to higher density

multi-family homes, which drives up prices, in the third stage people downgrade from the initial structures for higher density living, during this stage, dwellings have deteriorated and are overcrowded to reduce rent. The fourth stage involves the deterioration of dwellings, leading to a thinning population, falling prices, crime, and infrastructural and service breakdown. The final stage is renewal, when the potential value of the neighbourhood exceeds the decline and depreciation of the neighbourhood (Metzger, 2000 ; Schwirian, 1983). Birch (1971) expands this model with the inclusion of a first stage that incorporates low-density rural settlements, and a final stage where the renewed neighbourhood experiences degeneration.

The life-cycle model acknowledges that the functionality of neighbourhoods is shaped by factors beyond suburban limits. Life-cycle progression is often complicated by the failure of neighbourhoods to anticipate demand. Communities struggle to manage socio-economic growth, employment, and infrastructural provision. These failures are consistent with the inaccessibility problems exposed in the bid-rent model. The bid-rent model reflects the nature of monocentric cities and is characterised by the concentration of employment in one primary centre (CBD); hence, prices decrease in relation to one point of origin. This is a trend that is evident in the Australian capital cities; 31% of Brisbane’s jobs are located within 4 kilometres of its city centre. Other job clusters (Chermside, Ipswich, and Upper Mount Gravatt) accounted for 3% of employment (Loader, 2019b). The significance of this economic polarisation has been highlighted by Arribas-Bel and Sanz-Gracia (2014) who explain that polycentric cities featuring more urban centres boast of higher employment densities, higher per capita income and low incidences of poverty. Conversely, uncontrolled developments in the outer suburbia of cities exhibiting the classic bid-rent gradient are susceptible to being disconnected from critical infrastructure, employment, diverse populations, and investment, relative to settlements in polycentric cities.

Nonetheless, the ecological models discussed above are often deemed by researchers to be deterministic, since neighbourhoods are framed as pre-destined to transition between particular life-cycle stages (Pitkin, 2001). However, there are likely to be underlying issues such as socioeconomic disadvantage which result in a long- drawn-out poverty trap. It is not the case that residents of these communities lack agency, but that these communities are stigmatised, and thus a self-fulfilling prophecy reinforces economic disadvantage (Green, Grimsley, & Stafford, 2005 ; Jacobs, 1961).

The bid-rent and life-cycle models have shown that the inherent land-uses, infrastructure, and pre-existing investment that neighbourhoods have, is a contributing factor towards their future growth and development. This is an important point of consideration because the excessive spatial polarisation furthers a lack of diversity in housing outcomes due to decreasing efficiency in the outer suburbs, which constrains the form of growth experienced in those communities.

Hence, it is crucial to question whether the motivations that have sustained spatial polarisation are compatible with the POD or at the very least remain socially or politically relevant for Australians in contemporary times (Davison, 1993). In the 19th century, the Australian suburban landscape was relatively homogeneous, (Davison, 1993), and as shown in Figure 3.6, this homogeneity has not abated in present times. Therefore, housing markets in Australia have been unable to consolidate on the basic POD principles. A century and a half after such diverse housing landscapes were witnessed in Australian settlements there is a stark contrast between the dense unit complexes of Brisbane’s inner city and CBD (active core and transit suburbs) and its predominantly detached suburbia (Figure 3.6). The mechanisms which allow for the provision of private assets over time also encourage the accumulation of property and further the possibility of scarcity driven planning (Brennan & Castles, 2002). Thus, the foundation of the POD (competition) is inhibited, and its realisation partial. This will be further analysed with secondary data in Section 4.2 (an economic perspective of financial impacts), as well as in Subsections 6.2.1-6.2.3 with primary data (by questioning the nexus between the territorial claims). The way competition is stifled, as well as the divergence between individuals’ housing needs and housing outcomes, will be further explicated in the subsequent chapter.

CHAPTER 4: THE THEORETICAL DIVERGENCE BETWEEN HOUSING NEEDS AND MARKET OUTCOMES The importance of this chapter is to clarify why housing needs (the empirical) or use- values, are inherently divergent from market outcomes in the form of exchange-values (the actual). Subsequently, it will then be explained that the functional difference between use-values and exchange-values can be broadened due to individuals exploiting said differences (speculation) for financial profit rather than for their own needs. This financial tendency represents one of the causal mechanisms explored in this thesis and will be introduced in Subsections 4.2.1-4.2.2. Firstly, it will be explained that although use-values are intangible, they stem from material usefulness. This material basis explains why there is a sequence between housing needs towards housing aspirations and finally, housing outcomes.

This sequence is only fragmented because, in the process of individuals cultivating or acquiring use-value, there are constraints and opportunities, which mean that housing outcomes do not perfectly reflect housing needs. Therefore, while it is possible to explain housing outcomes through utilitarian measures such as house prices, dwelling stock and individual income, these measures are inadequate proxies for subjective needs and aspirations. The subsequent sections will explain in greater detail how and why use-values and exchange-values diverge, and why this thesis’ advances the need for more appropriate proxies for housing needs.

4.1 The Divergence between Use-value and Exchange-value: Explained through Three Theories of Value The purpose of this section is twofold. The first objective is to reveal that use-values are intangible and are distinct from other forms of value such as social use-values and exchange-values (price), and consequently use-values must be inferred through various proxies. Some of these proxies are outlined in three of the most notable theories of value: cost theory of value, subjective value theory and equilibrium. The second objective is to explain why use-values are immaterial and distinct, by positing that the inherent functionality of housing markets establishes a perpetual divergence between what individuals need (use-value) and what is exchanged in the market; needs are internally derived, while exchange is externally determined.

These two objectives are important because this thesis aims to understand the relationship between use-values, which embody individuals’ subjective wants and the

56 objective constraints and opportunities in the Brisbane GCCSA housing market. However, while market-derived attributes such as house prices are readily produced, use-values are not directly reproduced in the market. Consequently, the prospect of understanding use-values is predicated on establishing an appropriate proxy and anticipating that use-values will be better reproduced in the future.

4.1.1 Introducing the Three Forms of Value and Three Value Theories This subsection will begin by underlining the three distinct forms of value. The first form of value is use-value. This usefulness emerges when a product or good has satisfied a particular need. This value exists because virtually everything has intrinsic value; that special quality that makes things useful. For example, the intrinsic value of housing includes shelter and the accommodation of social functions. The concept of use-values corresponds with the structures of activity that imply immediate use; thus, dwelling and home are the relevant facets in the framework of Ruonavaara (2016). These structures include habitation, social identity, belonging, as well as the general need to structure human activities (Karjalainen, 1993). This sequence is where value terminates regarding individual needs. However, individual use-vales can have wider social implications, which are social use-values. Social use-values emerge when a product not only satisfies the needs of one individual but is also useful to others (Marx, 1903 [1887]). This transition from use-values to social use-values occurs through economic production, which establishes the product as a commodity. At this point, it will still have varied forms of existence, reflecting the different modes of production. However, following the commodification of the product, it will have an exchange- value, which is something that can be measured, quantified and is representative of social use-values in a general sense (Marx, 1903 [1887]).

The three forms of value above have been identified with varying degrees of success with three value theories: the cost theory of value, the subjective theory of value, and equilibrium in the subsequent subsections. These value theories are not compatible with this thesis’ framework and are introduced to explain why an alternative proxy for use-values is needed (clarified in Subsection 4.1.5). That proxy (housing needs) is used in the Brisbane GCCSA HNS and SNA in Chapters 6-7.

4.1.2 Cost/Labour Theory of Value Therefore, since use-values (subjective) transition to social use-value (cost) and finally, exchange-value (price) subjective values are often inferred through a general price. However, it is hard to infer the usefulness of a product such as housing from its price alone (H. M. Robertson & Taylor, 1957) since prices are also influenced by market constraint and opportunities.

Intuitively this suggests that subjective values should be inferred directly from individuals’ preferences. However, historically, the evaluation of preferences has proven uncertain. Subjective values were problematic for classical economists since, from their perspective; they were unquantifiable. Classical economists considered that subjective values were not a “.…universal measure of value” (Marx, 1903 [1887], p. 81). Though Adam Smith, another classical economist, acknowledged the importance of subjectivity due to its representation of utility or usefulness, he could not compute how the subjective worth of a product was not fixed but driven by circumstance. In some circumstances, goods with immediate uses such as water could be worth more than luxury items such as jewellery (H. M. Robertson & Taylor, 1957). To resolve this paradox, classical economists partitioned value into two: use-value and exchange- value (Holcombe, 1999). The former explained individuals’ subjective wants while the latter was used to explain prices.

To reunite the two forms of value, Marx concluded that there must be something identical in both use-values and exchange-values that allows for quantitative comparison. He recognised socially necessary simple labour time as the common denominator (Marx, 1903 [1887]). This is explained in the following quote:

Both derive from the expenditure of labor power—use value from the qualitative aspect of work as transforming useless matter into useful objects; exchange value from the purely quantitative, commensurable side of work: “abstract labour.” (Varul, 2011, p. 2)

Marx conceived of a value chain beginning with production costs as a proxy for use-values and culminating in exchange-values because he believed that exchange- values alone were incapable of explaining the subjective value of a product. This is because, in the absence of uniform market conditions, the value of exchange is simply a relationship between products (Marx, 1903 [1887]) and use-values and exchange-

values cannot converge. Consequently, the cost theory of value (CTV) uses labour as a proxy for use-value. The theory assumes that in society, individuals will maximise their labour-power and thereby regulate value and determine what is socially useful. In other words, use-value is inferred by how much time and resources a society invests in the reproduction of any good, without delving into the specifics of consumptive attitudes

Thus, by fixating on production costs, Marx was able to explain that use-values were increasingly divergent from exchange-values, due to differences between wage workers and capitalists (Marx, 1903 [1887]). Marx established the market’s inability to perfectly reproduce use-values, explaining how the inefficient transference of use- values in the production process was due to individuals’ interactions being limited by economic and social imbalances. Marx classed these various forms of exploitation and monopolisation as the contributors of surplus-value. Neo-Marxists such as Lefebvre (1968) and Harvey (2010) have built upon Marx’s theses on capital imbalances through their critiques on the production of space and the city. These works will be useful for outlining exploitations in the housing market, though they do not proffer solutions for unearthing use-values.

Marx did not unearth individuals’ subjective intentions but simply differences in the quantity and magnitude of labour and other supply-side inputs because he attributed use-values and consumptive preferences to cost which meant that everything centred on exchange-values or production.

4.1.3 Subjective Theory of Value The CTV provides an inadequate proxy for use-values because it overemphasises production costs. This model is valid if we think of value simply from the standpoint of boundless utility, i.e. needs are abstracted and desired without any constraints. In reality, individuals will appraise the usefulness of something, not simply based on its inherent qualities, but on their ability to acquire it, relative to all their other needs. From this standpoint, the precondition of use-values are actually prices, not production costs. “The subjective value theory of Menger and others starts with the valuation of consumer goods and works its way back through the prices of labour and other inputs accordingly” (Murphy, 2011, p. 1). Use-values are then implicit in the exchange-value, although they are not easily explained by them. The subjective theory of value (STV)

holds that the fairness of price is derived from the process, not the magnitude of price. It is the fluidity of how people prioritise and negotiate market transactions that determine prices, not the amount of labour. This contrasts with the CTV, which subsumes subjective values into definite production costs.

Alternatively, the STV posits that prices not only measure the utility of commodities but also signify how consumers decide how to maximise their satisfaction (Harper, 1967). This concept of subjective value is based on the premise that individuals in the market derive value by selecting from a wide range of alternatives based on which goods are likely to yield them the greatest marginal benefit at any point in time. Thus, it can be concluded that neoclassical economists in the modern era sought to resolve the contradiction that emerges when production costs are used as the proxy for the generation of value. However, these neoclassical economists like Menger (2007 [1871]) simply replaced the idea that value is substantiated through production costs with the premise that value can be substantiated through observed transactions, formally known as ‘revealed preferences’ (Mooya, 2016).

This is a problem because exchange-value (prices) is placed at the beginning of value creation, thereby assuming that prices are created spontaneously, and individuals simultaneously determine exchange-value with those prices in mind. “Although it tried to explain prices, prices were necessary to explain marginal utility” (Mattick, 1977, p. 61). Moreover, the focus on observed transactions is problematic because individuals are unlikely to perfectly rationalise and survey all available options. Thus, many of individuals’ intentions do not manifest materially and influence prices. At best, what prices may infer is how the collective prioritise value based on available choices, whether they reflect their use-values or are simply substitutes.

4.1.4 Equilibrium The final theory of value explained in this thesis is the supply and demand equilibrium model. This model is designed to balance the production centric theories of Ricardo and Marx, with the demand centric theory of Menger (Lagueux, 1997). While the CTV fetishised supply (production costs) and the STV fetishised demand (consumption ‘costs’), equilibrium considers exchange-value to be derived from the intersection between supply and demand. This intersection resolves “…. the growth of the conception of exchange-value (at the expense of use-value), as well as prevents the

advance of the subjective conception of value (at the expense of the objective conception)” (Mooya, 2016, p. 24).

Equilibrium integrates supply-side factors such as production with demand-side considerations such as utility. This process is shown in Figure 4.1 below and assumes that based on individuals and entities competing on the demand-side (market demand curve) of the housing market, and sellers or lessors doing likewise on the supply-side (market supply curve), house prices/rents will have a tendency towards a stable price. This is because as prices decrease, individuals can demand larger quantities, while as prices increase, producers will deliver more houses. Thus, the equilibrium price- point is the price where the number of goods supplied is equal to the number of goods demanded. If there are constraints that limit housing demand, then the demand curve shifts leftwards, and for each price-point, fewer properties will be demanded. Similarly, if there are production constraints, the market demand curve will shift rightward, and fewer houses will be produced at each price-point (Saylor Academy, 2015). Shifts in the demand or supply curve require market information and coordination among suppliers. Without coordination, for example, when there is over-supply, prices decrease, and there may be a tendency to produce even more quantities (Grossmann, 1977). This mismatch between the supply and demand curves is known as disequilibrium and is not only caused by over-supply but also housing shortages.

Figure 4.1 Market Equilibrium: Showing the Intersection between Market Supply and Market Demand Curves Source: (Saylor Academy, 2015)

Thus, equilibrium is a more reliable model for the interaction between supply and demand forces, but it still fails as a proxy for individuals’ subjective needs. What

is explained is how individuals and entities respond to changes in price and or supply within the existing market structure. The likelihood of sustaining not only equilibrium but also a housing market that is perpetually reshaped according to individuals’ needs is reliant on information dispersal and comprehensive coordination between governments, developers, analysts, and community groups. These factors are beyond the scope of equilibrium models, and this limitation will be discussed in Chapter 8.

4.1.5 Summary: The Need for an Alternative Proxy for Use-Values Subsequently, the three forms of value have proven to be inadequate proxies for use- values. There will be some level of discontinuity due to barriers that prevent authorities and developers from coordinating all needs perfectly, with perfect feedback from the individuals. This is partly a consequence of operational constraints and partly a result of territorial contradictions, which have been outlined by theorists such as David Harvey and Karl Marx.

Thus, the characterisation of use-value is likely to prove elusive, regardless of the economic theory in application. The notion that production straightforwardly mirrors use-value is untenable. Ultimately, supply and demand can only reflect the true balance of social aspirations if both the consumer and the producer have leverage. As noted by Lukács (2000 [1923]), it is the fact a buyer can walk away from a transaction that compels the seller to fix an appropriate price. Since households cannot abstain from the housing market, and they must participate in it, whether it suits their needs or not; then, the onus is on balanced competition and coordination to regulate the housing market.

Thus, this subsection has formed a conclusion for Section 4.1 by outlining how use-values must be inferred through means other than prices. This imputation must capture not simply the processes that reflect how people interact with the housing market but how they would like to interact with the housing market. In a matter reminiscent to how Marx attributed labour as the common substance between use- values and exchange-values, this thesis proposes fundamental needs as the bridge between individuals’ subjective housing aspirations and how the housing market facilitates individuals’ housing needs. How these needs are satisfied will be explored in Chapters 6-7.

4.2 The Contradiction of Use and Exchange-value: Financialisation The previous section explained that the market imperfectly reflects use-values due to market and individual constraints. Hence, other means must be devised to develop a better understanding of use-values, specifically subjective needs, and aspirations. However, ‘market constraints’ are not simply the by-product of incidental misinformation or financial disparities between individuals. Housing outcomes can depart from emotional ends since consumption and production underpin different motivations. The degree to which use-values are embedded in the market is defined by the dual function of price. Marx (1903 [1887]) showed that in capitalism, practical uses (utility) and abstract values (price) were established through the same processes (Cook, Davison, & Crabtree, 2016). Thus, housing transactions are not simply stimulated for use-values since housing can also function as a depository of value. An example of this phenomenon could be an investor buyer who already has a principal place of residence, placing one of his/her other homes on the market to derive exchange-value which can then be used for alternative investments or consumption.

Thus, in the process of exchange, every product is reducible to money. Money appears at the beginning of production (reflecting societal demand), as the means of developing houses, but also appears at the end when houses are sold, and a price has been paid (representing the weight of a product for accessing other goods) (Marx & Engels, 2008). Therefore, individuals can be motivated by both the price at the beginning as well as the price at the end. This is not a problem when use-values and exchange-values are balanced, but it is problematic if use-values are not the key drivers of the housing market (Harvey, 2012a). This happens when individuals believe that they can derive more utility from exchange-value than from living in a home. Historically this circumvention would occur in isolated instances, but due to the increased financial significance of housing, the inversion of the use-value and exchange-value process is now occurring at a large scale globally.

Historically, housing consumption was determined by individuals’ immediate needs for two main reasons. Firstly, before capitalism, people would mobilise their resources for their immediate housing needs. Secondly, housing cannot be physically moved to respond to supply and demand (spatial fixity), which creates long turnovers (Gotham, 2009). Even in the advent of capitalism, housing has still largely been consumed as a long-term acquisition, with investments made for retirement. However,

financial actors are now able to transform the illiquidity of housing into liquid and more accessible assets known as securities. These securities include adjustable rate mortgages, real estate investment trusts and mortgage-backed securities, among other options (Gotham, 2009). Securities ensure that people do not have to buy leases or loans; they can invest directly in housing portfolios. Securities are liquid but intangible; investors can buy into the mortgage debts of homeowners rather than into the homes themselves. This process is relevant because financial excess is encouraged since exchange-value becomes the end rather than a means to an end. Hence, financial motivations may undermine the perception of housing as a place to live and magnify the contradictions of capitalism

This global trend is known as financialisation and generally refers to an increase in the magnitude of the financial market (goods, services, and assets) or specific financial securities and institutions. Specifically, this thesis examines the transfer of capital from the real estate sector to the financial sector (Palley, 2013). Financialisation means that people are provided with liquid assets that can be cashed out at any time. Thus, financialisation goes one step beyond commodification. While commodification allows individuals to sell their houses or use it to offset costs, financialisation enables people to refinance and bring forward those benefits (Cook et al., 2016).

According to Gotham (2009), authorities have not viewed financialisation as problematic since it allows governments to transition the welfare state into an economy driven by personal finance. Similarly, individuals may not perceive the process as a predicament since based on their normative beliefs, they position themselves within the pre-existing social narrative and social trajectory. People are also inclined to do what works or is perceived to be working (social norms) (Morris et al., 2015). Paradoxically while financialisation furthers financial speculation, which decreases housing affordability and increases household debt, Australian households are reliant on housing as a vehicle to grow wealth and offset financial liabilities.

This increasing reliance on the market and financialisation means that the traditional relationships that people formed with their environment and homes, wherein their general wellbeing was dependent on having a direct understanding of the use- value of their homes and local amenities have diminished (Karjalainen, 1993). In its place, there is an increased perception of housing as a market product, which

fetishises the production process (Aalbers & Christophers, 2014). Though the production process is important, it should not undermine the need to fulfil underlying emotional ends and housing identities. Thus, it is worrying that due to the shifting focus towards the financial component of housing, consumption patterns have been relatively less studied in housing studies (Gram‐Hanssen & Bech‐Danielsen, 2004).

Consequently, this section has explained that the shift in focus towards production processes and financialisation means that production no longer conforms neatly to consumptive needs. In housing markets such as Australia, assets and capital are in a perpetual state of ‘exchange’, facilitated by mechanisms that are not directly related to real estate attributes and “exist outside the sphere of production” (Fernandez & Aalbers, 2016, p. 84). Since financialisation redistributes wealth through the abstract flow of capital and does not necessarily involve physical exchange, the discernment and management of housing use-values are diminished.

Thus, Subsections 4.2.1-4.2.2 will study the impacts of financialisation that show which financial tendencies shape land-use and economic growth.

4.2.1 The Impact of Financialisation on the Housing Terrain: Uneven Development and Extensive Land-Use This subsection will explain that the inversion of motivations in the housing market has significant consequences for land-use patterns in the Brisbane GCCSA. Pike and Pollard (2010) explain that although financialisation is not disconnected from geography, through borderless capital inflows and outflows focused on value extraction rather than value creation, the geographic terrain is reshaped (public territoriality). The authors also state that houses are treated abstractly, and transactions are based on buying, value enhancement and resale. This dynamic creates new barriers to entry, impeding homeownership and lessening the ability for people to regulate their housing needs. Additionally, investors buy into properties based on speculative interest and sustain the dependence of individuals on financial markets which distorts their existing identities (Pike & Pollard, 2010). Pryke (2006, p. 15) states that financialisation “…. aids the production of futures that are shaped in the name finance and its attendant imaginary”. This implies that future housing outcomes will conform to the requirements of financialisation. This financial tendency will be explored in Subsections 7.2.2-7.2.5.

In the case of the Brisbane GCCSA there is a pre-existing history of extensive land-uses in suburbia. The urban landscape is dominated by detached dwellings, while multi-storey flats are almost entirely situated within the inner core of the metropolitan area. As explained in Chapter 3, historical territorial claims cultivated this urban settlement pattern, but in recent history, this pattern is also likely to have been sustained by the financial imperative for uneven development. In the past 30 years, population growth in the Brisbane GCCSA has mostly occurred in the periphery (shown in the mortgage belt in Figure 3.6); though in the past five years, inner-city/CBD growth has increased (Loader, 2019c). In the periphery, low-density development is legitimised to artificially constrain supply. The extensive expansion of settlements into fringe land enables demand to exceed supply since inefficient expansion increases the cost of unimproved land and infrastructure (Haila, 2016).

According to Stapledon (2007), the rise in fringe land costs is more significant in explaining the increase in Australia’s house price to income ratio population growth, building costs and regulation, although these factors are the likely contributors of the increase in fringe land costs in the first instance. If fringe land has expanded uncontrollably, that is not easily reversible. Stapledon’s findings on the importance of urban fringe land can be understood when considering the housing values to GDP ratio. In the period between 1989 and 2017, the residential structure component of housing values has remained consistent at approximately 1 times the value of GDP, while the residential land component has increased from 2 times GDP to 3.5 times the value of GDP (Unconventional Economist, 2018). This basically implies that construction costs have changed negligibly, but the price of land has soared.

4.2.2 The Impact of Financialisation on Productivity: Exaggerated Economic Growth and the Stagnation of Wage Growth The previous subsection has explained that financial tendencies are responsible for uneven development in the Brisbane GCCSA. However, the proliferation of uneven development does not simply limit housing opportunities; it also comes at a cost that households in the Brisbane GCCSA and Australia are struggling to pay. Hence, the potential influence that financial tendencies have on affordability will be explored in this subsection. This imbalance is observable by identifying the disparity between wage growth and house prices. This relationship means that imbalances in the

66 prioritisation of housing needs do not simply have individual ramifications but affect the economic foundation of Australian real estate.

In traditional media, housing market volatility is mostly attributed to the undersupply or oversupply of housing. This is because housing is inelastic, and the market must anticipate demand. This thesis underscores that the problem is more than a logistical issue but one characterised by artificially driven scarcity. This scarcity fuels volatility and promotes house-price euphoria. House-price euphoria is a phenomenon whereby private developers, financial institutions and homebuyers are hysterically swayed by volatile house price increases and the investments of other individuals (Perkins & Thorns, 1999). These capital gains are of greater significance in economic markets where real estate investments are disproportionately high. Australia is one such market as stated by Credit Suisse (2018, p. 55):

The composition of household wealth in Australia is heavily skewed toward non- financial assets, which average USD 304,500, and form 60% of gross assets. The high level of real assets partly reflects a large endowment of land and natural resources relative to population, but also results from high property prices in the largest cities.

Investment in Australian real estate has been buoyed by the reduction in perceived risk, which has been minimised with the introduction of policies such as negative gearing. Negative gearing allows investors to offset other capital gains or lessen their taxable income (Australian Securities and Investment Commission, 2018). Other forms of artificial government support include capital gains discounts and first homeowner grants (FHOG). These measures distort the market by either allowing investors to privatise their gains and socialise their losses or enabling the market to augment prices. There has been a noticeable increase in the number of geared investors, and investors in general, in the Australian market over the past two decades (Reserve Bank of Australia, 2015), and the market has been artificially stimulated by exogenous government policies, rather than the endogenous buying power of Australian residents (the fundamentals). In economics, when prices exceed underlying fundamentals after off-shooting or undershooting equilibrium, long-term prices are liable to revert to the mean (mean reversion) (Gao, Lin, & Na, 2009). Various studies have produced evidence of mean reversion in the housing market, such as those by

Glaeser and Nathanson (2015) and Gao et al. (2009). Mean prices have been inferred by relationships such as the increase in property transactions as a percentage of Australia’s GDP, as well as the increase in the price to income ratio. Another indicator is the surge in property transactions while Australia’s ownership and affordability have dwindled (Unconventional Economist, 2013).

Australia’s expanding housing market in the absence of sustainable fundamentals (underpinned by use-values) is explainable by the fact that the growth of housing transactions have been supported via the loosening of credit standards rather than the effective buying power of consumers (Digital Finance Analytics, 2019 ; North, 2018a). In Britain M. Robertson (2014) has shown that mortgage finance has often contributed to price increases rather than supply, while in the Australian context, in the past three decades, rolling annual dwelling starts have lagged behind rolling annual population growth (Lawless, 2012). While the fundamental constraint on the supply-side may be attributed to land supply as explained by North (2018a), or fringe land supply as established by Stapledon (2007), these are the constraints that inflate housing costs, not the fuel that sustains high house prices. If supply is constrained and wage growth has stagnated nationally, then where does increased demand stem from? Expensive fringe land can explain why prices would have an upward trajectory but should not account for how prices have been sustained.

The immediate conclusion is that population, migrant growth, and foreign buyers have been abundant. This shies away from distinguishing between demand and effective demand. Demand in the Australian housing market may be growing, but effective demand, which is an indicator of people’s ability and willingness to pay, is negligible. ‘Ability to pay’ is a superficial representation of consumer wealth; lowering interests augments consumers’ appetite for borrowing briefly (C. Rogers, 2006), but is not a sign that Australian housing is affordable (Fox & Finlay, 2012). This is a simplistic view that emphasises access to credit rather than the ability to sustain credit-driven growth. The legitimacy of this theory is especially challenged by the fact that over 1 million Australian households have been experiencing mortgage stress (North, 2019).

Despite the risks associated with mortgage stress and defaults, the allure of housing market investment is sustained by speculation. Due to the dynamic properties of land (having an unfulfilled potential as raw land, which is concretised as improved

68 land), the value of housing can be speculated. Speculation incentivises people to anticipate capital gains or leverage their house through equities. When homeowners buy into a property, their equity is equal to their deposit. However, as repayments are made, and in cases where the total value of the house increases, homeowners are then presented with additional equity, which is often used to leverage on the payments of other debts or to acquire additional properties (Gotham, 2009).

Figure 4.2 Australian Household Debt as a Percentage of Household Disposable Income Source: (Bullock, 2019) The injection of financial capital into real estate has been crucial for the market’s growth. Thus, household debt in the Australian housing market has noticeably increased in the past 25 years (Bullock, 2019), and a considerable percentage of households’ disposable incomes in economies such as the United States, Australia, UK and Germany is consumed by household debt (Organisation for Economic Co- operation and Development, 2017). Figure 4.2 above shows that Australian household debt has since exceeded 100% of disposable incomes, and nearly 60% of that household debt consisted of mortgage debt in 2013-14 (Phillips & Taylor, 2015). In the 1950s, Australia had a relatively low household debt ratio by international standards at 56% (Macfarlane, 2003). Households borrowing to enter the housing market have been a major contributor to this increase, although investor buyer contributions have

also significantly risen, from 18% in the 1990s, 30% in the mid-2000s to nearly 40% in 2010 (Bilston, Johnson, & Read, 2015 ; Bullock, 2019 ; Macfarlane, 2003). Mortgage debt is expectedly higher for people under the age of 30 and those in the 30-50 cohort, than for the elderly. Generally, the latter group have lower mortgage debt and greater value in investment portfolios (Phillips & Taylor, 2015). Even so, Gotham (2009) states that globally, since home equities peak towards the end of the life-course, mortgage debt is now rising as a way of generating wealth, thus, perpetuating imbalance.

Aalbers and Christophers (2014) note that the housing market’s imbalance is due to housing being positioned as the solution for all too many motivations aside from homeownership. Housing sustains profits and equity, finances debt, and forms collateral for defaulted loans. Thus, it is uncertain how the use-values of housing can be prioritised amidst the systematic favouring of housing as an economic asset (Aalbers & Christophers, 2014). Moreover, this imbalance attracts capital investments that would otherwise be invested in the ‘real economy’.

The weak link between credit and productivity in the Australian economy is problematic because credit allows society to borrow against future growth (Das, 2013). Hence, present-day growth is inflated if it is unproductive. The sustenance of the housing market is reliant on constant credit infusion, but it is unclear how many households can carry the debt burden in the future. This is worrisome because the housing market cannot sustain itself. Though investment in the housing market stimulates growth and employment, the level of productivity generated is not enough to offset the increase in house prices relative to wage growth. Wage growth has stagnated relative to house price increases, because of congestion, sprawl, high business costs, labour market underutilisation and a delay in technological diffusion (Gurran et al., 2015 ; North, 2018b). The decline of high paying mining sector jobs is another factor (North, 2018b). Improved housing fundamentals and increased infrastructural costs have also been cited, as well as regulatory constraints. However, analysts such as Stapledon (2007) have explained that real construction costs are not the most significant contributor to price increases. Therefore, the financialisation and debt-fuelled growth model has masked the extent of housing unaffordability and extensive land-use models. The research methodology in Chapter 5 will present a strategy to explore if these structural inefficiencies of the housing market are reflected in how housing needs and aspirations are satisfied.

CHAPTER 5: RESEARCH METHODOLOGY This chapter introduces the research methodology of the Brisbane GCCSA Housing Needs Survey and the SNA method. These methods will be analysed individually but will be triangulated in Chapter 8. The survey employed a retrospective longitudinal design and was created specifically for this thesis. The first section of this chapter introduces the theoretical justification for the mixed-method approach. Then, the research design specifies the objectives of the two methods, which correspond with the four research questions. Next, the survey’s data collection procedures will be described, detailing how the survey was distributed, cleaned, and analysed. A similar process is detailed for the SNA method, showing how the data was targeted and cleaned, and how SNA measures can be used to produce housing market knowledge.

5.1 Introduction It has been clarified that economic value theories conflate use-values with market constraints and opportunities. Thus, understanding individuals’ subjective aspirations require the inference of use-value through alternative means such as proxies and user attitudes. This thesis will primarily infer use-values and financial motivations using two proxies. The first proxy (survey) examines how users prioritise their needs, while the second proxy (SNA) studies how housing market interactions are facilitated. These proxies better reveal people’s genuine needs and aspirations, or lack thereof. Lastly, the methodology captures users’ attitudes towards the housing market. These attitudes reveal the priorities, opportunities and limitations of renters, owner-occupiers, real estate agents, and investors, albeit from the viewpoint of residents alone.

Individuals’ prioritisation of their needs and aspirations were captured through a survey instrument which targeted Brisbane GCCSA residents, while people’s interaction with the housing market was investigated using the structure of housing market terminology. This process involved defining the most important real estate terms in the network and exploring their relationships. This is known as Social Network Analysis (SNA). This mixed-method approach accounts for the subjectivity of participants to varying degrees. The specifics of each approach will be specified in their individual sections: 5.3 for the Brisbane GCCSA Housing Needs Survey, and 5.4 for the SNA method. Table 5.1 on the next page shows an overview of the methodology. The first research question (RQ1) relates to the survey and is shaped by literature which shows that it is the balancing of needs that results in self-

actualisation. The question studies how the prioritisation of housing needs is satisfied. These satisfiers were represented by 18 social-dwelling and financial attributes to show if housing needs were interrupted by life-course priorities or whether the core social-dwelling satisfiers were prioritised (empirical: composition of needs). As individuals progress through their housing careers, they seek to enhance how their needs and motivations are satisfied based on their life-course demands (represented by nine attributes). Thus, it was vital to identify how social-dwelling and financial aspirations improved with time (empirical: the trajectory of needs).

Table 5.1 How the Research Questions are Addressed Methodologically Method/ Measures of Research Question Data Method # Instrument Analysis # Collection Component Themes (Source) Object of Study How do individuals Mean of Preferences prioritise housing needs (a)

Likert scale satisfiers through the K-means Clustering Results and life-course? (b),

RQ 1 Stratified Demographic RQ 1 9 life-course themes, 18 Composition (a) and Sampling, Brisbane Profile social-dwelling & financial Trajectory (b) of Online GCCSA satisfiers Needs (Empirical) Recruitment, Housing

How are personal needs Online Needs Survey Values/Causal responsible for differing Survey, (Hosted on Coding RQ 2 housing experiences? RQ 2 Spreadsheet Checkbox) 6 individual needs factors Attitudes (Empirical) (Excel) Open-Ended How are divergent

Question attitudes responsible for Values/Causal differing housing Coding RQ 3 experiences? RQ 3 6 external factors Attitudes (Empirical) How well are individuals’ Social Centrality and

Web needs facilitated by the Network Entire Method Community Scraping, housing market? Analysis (Rental and Measures

RQ 4 Spreadsheet RQ 4 69-72 economic, dwelling, (Realestate Sales Data) Composition of (Excel) & home satisfiers view.com.au) Needs (Actual) Explanation of Causal

4 Methodological 4

- Mechanisms/ - Critical Pluralism Normative Beliefs All All Realism

RQ 1 Territorial and financial Causal Mechanisms RQ 1 tendencies (Real)

The exploration of individuals’ attitudes (empirical) towards personal needs formed the second component of the research methodology, corresponding to research question 2 (RQ2). The exploration of individuals’ attitudes towards divergent external factors formed the third component of the research methodology,

corresponding to research question 3 (RQ3). The objective of RQ2-RQ3 was to analyse the consistency of individuals’ attitudes, which differed from research on how normative beliefs, attitudes, and social norms predict behavioural intentions (Fang et al., 2017). This thesis instead associated planning and housing outcomes with individuals’ attitudes and normative beliefs to a lesser extent. Thus, housing outcomes were evaluated based on whether they were competing or balanced strategies.

Conversely, the SNA component used language produced by property search to reveal how housing needs were facilitated by the market. Section 7.2 will show that the structure of language mirrors the structure of the housing market (actual). Thus, understanding how the market facilitates the projection of housing needs through property listings, also reveals underlying financial and territorial tendencies. This market-derived proxy corresponded to research question 4 (RQ4).

The final thesis component involved all research questions and corroborated the causal explanations within the entire thesis. This is known as methodological pluralism and was verified using the territorial profiles (Chapter 3), open-ended questions, user-specified priorities, attitudes, and market-derived tendencies. This resolved the critical realist position that there are multiple domains of knowledge.

5.2 Research Design This section and its subsections are an overview of the research methods and how they correspond to the four research questions. Following this outline, the methods will be explained in greater detail in Sections 5.3-5.4. The premise of this methodology is that participants’ subjective preferences can be revealed based on accounts of their housing experiences, in comparison with market offerings. The research assessed the attainment of subjective aspirations by observing the consistency of mobility and semiotic patterns. For the Brisbane GCCSA Housing Needs Survey, this involved understanding whether housing needs were balanced (composition), based on the distribution of the three housing facets (in particular the core social-dwelling needs). The prioritisation of housing attributes by age and mobility was also assessed to determine if individuals had a clear trajectory. Lastly, individuals’ attitudes were contrasted to see how they differed and what implications this had for housing beliefs and outcomes. The SNA method had three components which helped to understand how well housing needs satisfiers were facilitated. These components were based on

the diversity of the central terms in networks, the diversity of communities, as well as the specific relations between property terms (semantics). These approaches formed the backbone of findings in Sections 7.2-7.3.

5.2.1 Introducing the Components of the Brisbane GCCSA Housing Needs Survey This subsection introduces the four components of the Brisbane GCCSA Housing Needs Survey: the demographic profile, dwelling profile, Likert scale queries, and open-ended question (Table 5.2 below).

Table 5.2 The Four Components of the Brisbane GCCSA Housing Needs Survey Component Life-course Social-Dwelling Financial Themes 4. # of dwellings 1. Age Bracket rented/owned, Demographic 2. Gender 5 5. Level of Profile 3. Years lived in Themes understanding of the Brisbane GCCSA housing market Opportunity Screening Questions General Housing Market

for Recall (If eligible go to GHME) Experience (GHME) 1. Occupancy 2. Tenure Dwelling 5 3. Type of dwelling Profile (a) Themes 4. Year moved in 5. Postcode Opportunity If Dwelling Occupied go

for Recall to Life-course queries 10. Rental Income 1. Cultural Beliefs 1. Neighbourhood 11. ROI Renting- 2. Personal/ Quality Capital Social Relationships 2. House Type 12. Flexible/ 3. Environmental 3. Social Capital Permanent Experience 4. Personal/Community Tenure 4. Career/Retirement Identity 13. Deductions Likert Scale Needs 5. Structural Integrity 27 14. Use for other Queries 5. Leaving Childhood 6. Few Restrictions of Themes investments Home Privacy/ Regulations 15. Future Value 6. Forced Move/Buy 7. Diverse Needs and Stability 7. Education/ Adaptability 16. Attractive Cost Training 8. Sufficient Home Size 17. Lower Costs 8. Family Separation 9. House/Community in Future 9. Health or Safety suits Activities 18. Build Savings 9 Life-course Themes 18 Housing Needs Satisfiers Dwelling 6. Reason for moving 2 Profile (b) 7. Suitability of different dwelling or location* Themes Open-ended 1. Impact of Individual Needs 2 Query 2. Impact of External Factors Themes * The question was asked but was not used in the analysis

These components corresponded to the life-course, social-dwelling, and financial attributes of the Brisbane GCCSA Housing Needs Survey. The life-course attributes consisted of three questions from the demographic profile and nine themes from the Likert scale. The social-dwelling aspects were derived from two questions in dwelling profile a, and nine themes from the Likert Scale. The financial facet comprised nine themes from the Likert scale. Additionally, all three attributes were reflected in two themes from dwelling profile b and two themes from the survey’s open-ended question, These profiles and themes corresponded to the demographic, health, housing history, aspirations, and employment components of the housing decision framework (see Section 2.4).

The demographic profile was the first survey component and consisted of life- course and social-dwelling attributes: age bracket, occupational classification (refer to Question 6 of the Survey, Appendix 21), years lived in the Brisbane GCCSA (Question 4 of the Survey, Appendix 21), the number of dwellings rented or owned, level of understanding of the housing market, and gender. This information was used to stratify the survey sample based on age, level of understanding and residential transactions. The age dimension was useful for studying how the 27 Likert scale attributes changed by age. The total number of rental/sales transactions, as well as the level of market understanding (see Question 7, Appendix 21) allowed for an understanding of whether individuals with increased interactions or understanding in the housing market experienced improvements in their housing conditions.

Dwelling Profile: questions related to the attributes of the principal residence (Social-dwelling attributes). This was the first segment of the dwelling profile (a) and consisted of occupancy, tenure, type of dwelling, year moved in, and postcode (see questions 10-16 of the survey, Appendix 21). This profile placed the housing needs satisfiers in Table 5.2 within the context of the rest of the Brisbane GCCSA. For instance, one of the significant differences in the housing market is the renter/owner divide. Thus, tenure data distinguished occupants with mortgages, allowances, assistance, and those who owned outright. The format of the tenure question depended on the form of occupancy nominated, specifically whether it was an ownership-based occupancy or a renter-based occupancy. With this data, it was then possible to understand the differences between the renter and owner cohorts. The

75 occupancy data queried the form of renting or owning, specifically, the type of household composition: living alone, flatmates, a partner, family, or the community.

The second segment of the dwelling profile (b) related to life-course, social- dwelling, and financial attributes and appeared after the Likert scale questions. The profile consisted of the motivation for leaving a residence, and the type of dwelling or address that would be considered a suitable alternative (questions 90-93 of the survey, Appendix 21). The profile was used to construct the reason for moving matrix in Appendix 10, though the alternative residences were not studied due to low response.

The third survey component was the Likert scale queries. The questions were posed in order from the least abstract and salient (life-course) to the most abstract (financial). Thus, people were first asked about the nine life-course themes. The respondent nominated the relevance of these life-course attributes in terms of their expectations at the time of initially renting or buying the principal dwelling (based on the relevance scale of 1-10). Then, individuals’ needs were tied to real-world constraints and opportunities. This association also determined if individuals’ residential mobility was dominated by life-course needs or improvements in housing conditions (dwelling).

Thus, individuals’ social-dwelling and financial priorities were also recorded within 10-year intervals (based on the relevance scale of 1-10). These social-dwelling and financial attributes helped to explain how the satisfaction of individuals’ housing needs was balanced by comparing the increase or decrease in these priorities relative to the life-course, age, and mobility. The open-ended question was the last survey component and revealed the consistency of housing needs satisfiers relative to individuals’ needs and external factors. These patterns were revealed by qualitative causal coding. Participants’ attitudes towards life-course, social-dwelling and financial attributes were examined, to observe if the market’s housing outcomes were sustained due to compatible beliefs.

How these survey components are embedded within the research questions, will be explained in the subsequent pages.

5.2.2 Introducing Individuals’ Housing Needs Composition and Trajectories This subsection explains how the study of housing needs compositions and trajectories resolves the first research question.

Research Question 1 (RQ1): How do individuals prioritise housing needs satisfiers through the life-course?

This research question refers to the trajectory and composition of individuals’ housing needs within the Brisbane GCCSA. The notion that individuals have a housing trajectory is based on the framework from Section 2.4 which highlighted that despite the differences between the life-course stages, the three housing facets central to housing consumption should be balanced since they satisfy fundamental needs. Hence, individuals should exhibit a balance in the consumption of said facets, as well as exhibit a gradual progression in how they prioritise the core facets of housing, such as the appropriate house sizes and types. Understanding individuals’ housing trajectory will be established through K-means clustering, while the composition of housing needs will be understood through the mean/median of preferences and dwelling/demographic attributes. K-means clustering is a machine learning algorithm that allocates data points into clusters based on the distances of variables to the centre of clusters. Thus, it shows which data points are similar and which are different. The objective of the housing needs and trajectories exploration is to reveal the underlying use-values and exchange-values of individuals in the Brisbane GCCSA. This was achieved by capturing individuals’ expectation of relevant needs during their life- course stages, in relation to a specific dwelling (their principal dwelling in that period). These relevant needs were queried in the form of a Likert scale with 27 themes (18 housing needs and 9 life-course themes).

5.2.3 Introducing Individual Attitudes and Explanation of Causal Mechanisms Research Question 2 (RQ2): How are personal needs responsible for differing housing experiences?

Research Question 3 (RQ3): How are divergent attitudes responsible for differing housing experiences?

These research questions related to individuals’ attitudes towards the housing market and were derived from the optional open-ended question at the end of the

survey (Question 519, Appendix 21). The question asked people: “In as many words as you can think of, what do you think are the barriers that make it difficult to remain in one address?”. The open-ended question was answered by 199 participants and allowed for an insight into people’s housing market attitudes. The question identified the barriers that impede individuals’ housing needs and trajectories. These barriers were revealed through the causal coding of the 199 responses. Those themes were separated into individual-needs and external factors and were used to identify how the territorial roles of various housing market participants: renters, homeowner occupiers, investors, real estate agents, authorities, and developers, differed.

5.2.4 Introducing how Individuals’ Housing Needs are Reflected by the Market Research Question 4 (RQ4): How well are individuals’ needs facilitated by the housing market?

This question targeted how well housing outcomes (symbolised by the most prominent real estate property terms) represented the satisfaction of individuals’ housing needs. This question was addressed through the Social Network Analysis (SNA) method, which condensed the multitude of property terms into a few manageable themes. The relationships between the property terms highlighted which property terms were the most prominent. By comparing the functionality of those terms with the housing facets of home, house, and dwelling, it was possible to gauge how different market segments facilitated housing need satisfiers. Hence, the market was segmented into smaller subgraphs based on tenure, geography, and typology, because they shape how individuals’ needs are represented through language. A more comprehensive explanation of SNA is detailed in Section 5.4. Additionally, the property terms were further classified into communities which revealed how the structure of the housing market (without the author’s own ontological imposition) was divided into distinct aspirational groups. These communities were used instead of employing the three housing phases developed by Israel (2003). These categories altogether established the degree of coherence in individuals’ housing trajectories.

5.2.5 Case Study Selection Criteria The Brisbane GCCSA was initially nominated as a suitable case study area due to two factors. Firstly, the underlying nature of housing needs has been enshrouded in metaphors, such as the housing ladder and the Great Australian Dream. Thus, the

region is susceptible to a fixation on attributes rather than underlying housing needs. This would be unproblematic if the housing market were affordable and self-regulating, i.e. the underlying motivations (emotional ends) could be implied. However, housing inequality in the Brisbane metropolitan region is considerable, evidenced by one of the highest median multiples in the developed world. Hence, constraints, such as the periodic increases in house price to income ratios, are likely to complicate individuals’ participation in the housing market (Geck & Mackay, 2018). Moreover, how housing inequality complicates individuals’ participation once they enter the housing market is under-researched. Inequality relates to the disparity between quality or quantities of a good. The quantities of inequality in cities such as Brisbane are relatively well researched in terms of housing affordability and rental/mortgage stress. However, the qualitative aspect of inequality is less understood. Thus, Brisbane city was tentatively considered, though, understanding consumption would also require outlining the boundary of consumption.

The geographical threshold of the Brisbane GCCSA was defined as the boundary of consumption since according to the Australian Bureau of Statistics (2016e), that region alongside other GCCSAs signify the functional extent or labour market of their respective capital cities. Thus, the functional extent encompasses the range of individuals’ travel between home and work. While the Brisbane GCCSA does not conform to a contiguous urban footprint, due to leapfrogging and rural settlements, it does denote a region united by people who share a tendency to shop, work and or socialise in the city’s employment hubs (Australian Bureau of Statistics, 2016e ; Coiacetto, 2007). Thus, the Brisbane GCSCA is represented as a contiguous polygon.

The Brisbane GCCSA functional extent was considered after evaluating three main approaches to defining housing markets. The first approach is hedonic modelling, which is a regression pricing model based on establishing the contributory factor of discrete housing features. This modelling is used to define regions where there are housing products that can serve as substitutes for residents and investors alike. Consumers in such a market choose between houses with similar levels of utility. Another approach is the functional catchment or extent, based on home-work-play linkages (Australian Bureau of Statistics, 2016e ; Coiacetto, 2007). The final approach is based on statistical inference, using census and other datasets to form factor/cluster analyses. Thus, based on broader socio-economic attributes, a housing market is

79 inferred. The final approach is capable of factoring location into calculations such as principal component analysis to determine the level of variation between the tested attributes. This thesis employs the functional catchment approach since the other approaches delineate the areas where individuals have been found to share similar socio-economic attributes and housing outcomes. The objective of those models is to outline small geographic clusters or submarkets with comparable utility, but since the housing market is inefficient, and the life-course is dynamic, individuals are likely to deviate or attempt to deviate from consuming housing uniformly.

Individuals are not seen as consuming housing uniformly in small geographic clusters since the boundaries of household consumption should reflect the limits of their employment, current residence, and recreational environments. This relationship encompasses individuals exercising similar movements and constraints, such as the shared interactions or experiences that individuals experience with commuting distances, access to transportation, labour market competition and complementary land-uses. “…. the outcome on the labour market is not random, but an effect of societal structures dictating the labour market conditions…” (Östh, 2007, p. 19).

Thus, it is vital that individuals can satisfy their social-dwelling, financial and life-course needs within the Brisbane GCCSA functional extent, which is the market consumption threshold. This relationship was explored as part of this research’s second method (SNA). The SNA used social-dwelling, financial and life-course factors to gauge the depth and diversity of the Brisbane GCCSA housing market. This was essential because the fact that individuals are tethered to their work, home and social environments means that if the market does not have a general diversity of housing products, then a lack of viable housing options is likely. This is because housing market inefficiencies such as cost and labour market constraints are ever-present factors that counteract efficient residential mobility. Hence, this thesis primarily defined housing viability based on a bottom-up categorisation of housing facets (derived from fundamental needs) within all housing market segments and less so on the idea that specific submarkets with specific housing attributes, satisfy specific types of renters and owners (see Subsection 7.2.5). The subdivision of the market (see Subsection 5.4.3) was primarily to reveal if the balance of housing facets is consistent across all subgraphs. Though individuals have different needs during the life-course, all needs are classified within the same overarching home, dwelling, and house facets.

5.3 Introducing the Brisbane GCCSA Housing Needs Survey The Brisbane GCCSA Housing Needs Survey (HNS) is a retrospective longitudinal housing needs survey. The retrospective aspect refers to the fact that the data is collected after the housing experiences have already transpired (Street & Ward, 2010), while the longitudinal aspect refers to the fact that multiple intervals and housing circumstances throughout an individual’s life are explored. This data cannot be reliably collected cross-sectionally, which would mean that the data points from the six age groups studied would be derived from a single cohort (the 2018 housing market).

“In retrospective case studies a time line of events and variables that changed over the time period is reconstructed after the events have occurred.” (Street & Ward, 2010, p. 2). In this thesis, the ‘events’ were the principal places of residence of participants, while the changing ‘variables’ were the age bracket, and the life-course changes associated with age, and potentially the dwelling itself. Thus, the research data was collated after the housing moves, and participants’ housing experiences were chronologically and contextually retrieved (A. Mills, Durepos, & Wiebe, 2010). Alternative resources include prospective longitudinal datasets such as the Household, Income and Labour Dynamics in Australia (HILDA) survey, (Australian Government department of Social Services, 2017) since 2001, and Australian Census Longitudinal Datasets, from 2006-2016 (Australian Bureau of Statistics, 2016c). These resources have useful longitudinal housing data, but prospective datasets tend to be limited in terms of the period of study, and partial coverage of housing potentialities (Fehring & Bessant, 2009). The benefit of the Brisbane GCCSA HNS was that the 27 life-course, social-dwelling, and financial themes were tailored to capture the satisfaction of housing needs over a lengthier period.

Retrospective longitudinal studies have been used extensively in public health, consumption studies, and housing histories (Lo, 2013 ; Raimond & Hensher, 1992). Raimond and Hensher (1992) reviewed seven retrospective studies and found that they were a legitimate form of time series data, insofar as events were recallable. Tsemberis, McHugo, Williams, Hanrahan, and Stefancic (2007) notably used a retrospective longitudinal method to capture the residential histories of respondents with psychiatric and substance abuse problems.

Comparatively, the HNS was more straightforward and was implemented to evaluate individuals’ housing trajectories, as well as the attitudes of renters and buyers towards the housing market. The Brisbane GCCSA was selected as it is a functional catchment (housing market) with considerable housing pressures such as unaffordability (M. Thomas, 2017). The Brisbane GCCSA also embodies normative beliefs such as the housing ladder and house price euphoria, which describe housing progression without considering individuals’ subjective aspirations. Thus, the HNS was created to explore aspirations within the eight Local Government Areas (LGAs) of the Brisbane GCCSA: Brisbane City Council, Moreton Bay Regional Council, Logan City Council, Ipswich City Council, Somerset, Redland City Council, Lockyer Valley, and Scenic Rim. The HNS was designed as an online-based questionnaire so people could easily transition between queries tailored for age, occupancy, and tenure.

As explained in the research design, 27 questions (Appendix 21) were posed to individuals based on equivalent survey queries relevant to their specific occupancy and tenure (Figures 5.1-5.3 on the next page). An equivalent survey query is the framing of a question to correspond with different tenures, for example, flexible tenure for renters and permanent tenure for owners. This personalisation was applied since the simplified mutual themes could not be posed to diverse individuals (the terms were too abstract). However, the mutual themes were used to unify the distinct survey queries (the different ways that fundamental needs were satisfied) for analysis. Operationally, respondents were given long-form statements (survey query verbatim) which embodied the nuances specific to their tenure. These nuances were derived from real estate research to construct an exhaustive list of housing potentialities (Baum & Wulff, 2003 ; Kelly et al., 2011 ; Yates, 2009). The queries were broad enough to be universally applicable, but not too general to be analytically insignificant. For example, persons were queried about how their general cultural and religious beliefs shaped their residential mobility, rather than a specific cultural notion.

Figure 5.1 shows how the cultural aspect of the Likert Scale questions was tailored based on whether individuals lived alone (Questions 18-26), lived with partner/family (Questions 27-35) or did not live in the home (Questions 36-44) of Appendix 21.

Figure 5.1 Life-course Queries about the Cultural Aspects of Residential Preference

Figure 5.2 (on the next page) shows how the social-dwelling facet of the Likert scale questions was tailored based on if individuals lived in the home (Questions 45- 53) or did not (Questions 54-62 of Appendix 21).

Figure 5.3 (also on the next page) was tailored based on whether individuals rented (Questions 63-71), owned/lived in the home (Questions 72-80) or if they had invested in the home (Questions 81-89) of Appendix 21.

The themes in Figures 5.1-5.3 satisfied the housing history/aspirations part of the housing decision framework (Section 2.4).

Figure 5.2 Social-Dwelling Attributes: Queries about the Use- value of Residential Preference

Figure 5.3 Financial Attributes: Queries about the Financial Aspects of Residential Preference

5.3.1 Data Collection and Sampling The participants for the Brisbane GCCSA Housing Needs Survey were recruited based on whether they were resident in the Brisbane GCCSA housing market. The unit of analysis was individuals, not households. The degree of involvement in the household buying or renting process was extrapolated by asking participants about their level of decision-making, whether they made the decision alone, together with other people, or whether the decision was made for them.

The online survey targeted approximately 150-300 participants based on a wider population of 1,722,474 people who were 18 and above and resident in the Brisbane GCCSA. This figure represented 76% of the population. The survey sample size was calculated with a confidence rate of 95% and a confidence interval of 5%, which yielded an ideal sample size (based on industry standards) of 385 persons. The attained sample size was 237 people, which fell between confidence rates of 85-90%, with a 5% confidence interval. However, there were 472 housing profiles from these participants, which presented enough data points to map mobility based on age cohort and geography. These data points were also used to study the variance between the three housing facets of the life-course (people’s life circumstances), the social-dwelling measures, and the financial attributes. These facets were studied to understand which housing aspects were prioritised and whether individuals’ interactions within the housing market were balanced. The statistical significance of the sample size was unimportant for the semi-structured open-ended question, which was dependent on whether the range of attitudes about the housing market had been exhausted.

The Brisbane GCCSA HNS was activated on the Checkbox survey platform in two phases. The first phase was between the 13-02-2018 to 01-04-2018 following the ethical approval on the 3rd of February 2018, while the second phase was activated between 12-06-2018 to 25-07-2018 to increase the sample size from 121 responses to 237 responses, to improve the sample’s geographic distribution. The survey covered the core population belts: the Sunshine Coast boundary towards the Gold Coast region on a North-South basis, and the Brisbane region towards Toowoomba, on an East-West basis (Appendix 2). The survey activation gave access to desktop, tablet, and mobile users. Respondents could return to the previous page, and they could reset their responses or make edits, they were also allowed to resume incomplete responses or save their responses for a later time. The default language

for the survey was in English, and the survey title used online was: Project Title: Assessing housing subjectivity: an analysis of value systems through the Australian property ladder. The survey had a landing page which told people about the survey’s conditions, such as age requirements, residency history and incentives. The landing page instructed that participants had to be 18 years and above and have lived in the Brisbane GCCSA for 12 months minimum. This period could be cumulative, although, in Question 9 of the survey, they were asked if they had rented or owned a dwelling for a period of at least six months.

The landing page also had a link to a participation information sheet, which detailed the ethical procedures and privacy. The survey required ethical approval since it was a human research study involving human subjects. The survey was made anonymous to protect individuals’ privacy in compliance with the National Statement of the National Health and Medical Research Council (NHMRC). Most of the questions presented were not highly demanding, and proxies were utilised for all the Likert scale questions and the open-ended question. The only questions that were directly posed were the introductory demographic and dwelling profile questions, and even those were not overly specific. For example, age ranges were used instead of specific ages. Additionally, participants’ income was not asked. The most demanding question was the open-ended question which queried participants about the barriers or challenges in the housing market. However, this question was not likely to trigger any major discomfort. Therefore, the only anticipated risk was anxiety, which was unlikely.

The online survey was hosted on the UQ Server of the external survey tool Checkbox, see Appendix 21 for the survey questions. Checkbox is a survey tool that allows people to conduct personalised surveys, using a variety of graphical tools. In this housing survey, radio buttons, HTML messages, drop-down lists, open-ended answer boxes, ratings and Likert scales were employed (Checkbox Survey, 2020). This online questionnaire tool was used since the survey employed proxies for housing need, which varied based on tenure choice, as well as the varying: ages, decision- making and occupancy. Therefore, the survey questions were written using skip logic or branching. If the survey participant did not fulfil the initial criteria of being 18 years and over, or the residency, rental and ownership requirements, the individual would be directed to the end of the survey, with a message stating their ineligibility. Skip logic was also employed for eligible participants. The first form of branching for eligible

86 participants of the housing survey was based on their age cohort. The age cohorts were as follows: 18-29, 30-39, 40-49, 50-59, 60-69 and 70 and above. The age group that the participant elected in Question 3 of the survey determined how many age cohorts would be visible to them throughout the cohort. Participants were asked about their primary place of residence based on ~10-year intervals (the age brackets). Thus, 18-year-old individuals were only asked about one dwelling, while 70-year-old individuals were asked to construct six dwelling profiles.

The extensiveness of each dwelling profile was contingent on individuals’ level of decision-making and whether they had rented or owned the dwelling for a period of at least six months. If an individual had not fulfilled the latter requirement, then the person would be unable to complete the rest of the profile and would be directed to the next age bracket, if applicable, or the end of the survey. If the individual did not make the renting or buying decision independently or together with others, then the individual would have filled their year of entry into the home, form of occupancy, tenure, type of home, and postcode, but would skip past the 27 themes to the next age cohort, or end of the survey, whichever is applicable.

This thesis employed a stratified convenience sample, which was a combination of random/convenient sampling as well as the systematic approach of online geographic advertising and community targeting, primarily based on a stratified sample. The initial recruitment for the survey was via convenience sampling of personal networks. This was a recruitment strategy that involved sending out the survey link, via personal Facebook contacts, as well as using other personal forms of contact such as text messaging, emails, and word of mouth. Following this form of recruitment, other social media accounts belonging to friends and their contacts were used to disseminate the online link, these accounts were on Twitter, Facebook, and WeChat. Property and urban-related online forums were also used; these forums are freely accessible by anyone to post, join and view. Convenience sampling was effective at maintaining the targeted gender distribution (see Figure 5.4 on the next page); however, this strategy was limited in its geographic coverage. Therefore, geographic sampling was prioritised through community and online advertising, which will be explained below.

Figure 5.4 Gender Divide from the First 37 Respondents Source: Author as derived from Checkbox Survey (2020)

Paid online advertisements were also utilised through Google AdWords (now known as Google Ads), Bing Ads and Facebook Business advertising. Google AdWords allowed a 30-character headline limit and an 80-character description field limit at the time when the Google Ads were deployed in February, March, and July 2018. Sixty clicks were attained in February, 83 clicks in March and ten clicks in July. 139,825 impressions were created altogether; an impression occurs when an ad is shown or viewed on a page. Appendix 18 shows one of the ads that was deployed in those periods. Due to the character limits, it has a concise message and a direct link to the survey; some of the ads also featured links in the bottom of the ad, which linked to sources such as the Facebook page which was created to promote the survey.

Bing Ads (now known as Microsoft Advertising) has a similar appearance and structure to Google Ads; therefore, Google Ads were directly exported to Bing. However, this campaign was only deployed once in July. This campaign was unsuccessful with 0 clicks, although there were 112 impressions.

Facebook Business consists of various advertising platforms; for this survey, only the social media platforms of Facebook and Instagram were targeted. This campaign was initiated in March 2018. The Facebook advertisements had 1038 impressions, while the Instagram ads had 10,855 impressions. In terms of people

reached, Facebook Business is known to apply a rounding threshold for confidentiality purposes. Thus, although the total number of people reached through Facebook Business was 9,726, the subtotal for Facebook was 945, and 8,804 people for Instagram. These figures are comparable with the impressions data, the key difference being that individuals can be exposed to multiple impressions. Regarding the actual campaign results, there were 42 clicks from Instagram, 5 clicks from Facebook, 26 post engagements (interactions such as liking, sharing, or commenting) from Facebook and 3 page likes. The Facebook campaign included a variety of strategies such as linking directly to the survey or linking to the survey’s Facebook page.

In terms of audience targeting, the Google AdWords and Bing Ads used experimental ads that were differentiated by Local Government Area (LGA) or a geographic radius that extended to the boundary with the Sunshine Coast region (similar to the Facebook radius in Appendix 16). This wide net meant that impressions would also be revealed to individuals outside of the Brisbane GCCSA, but the content of the advertisement made it clear who the target audience was, as did the survey itself, should they have made it that far. The survey employed the Google AdWords age bracket of 18-65+, and the range of responses by gender and age can be seen in Appendix 19. The survey also employed keyword targeting (required in all Google AdWords and Bing Ads campaigns); this involved creating keywords that were likely to yield the most impressions but were closely related to the survey context. Hence, the keywords were related to renting, buying and the housing market.

The Facebook Business campaign also used the 18-65+ age bracket for targeting, as well as a geographic radius beginning from the centre of Brisbane and extending towards the boundaries of the Sunshine Coast and Toowoomba and overlapping parts of the Gold Coast (see Appendix 16). The target area could not be specified to the boundaries of the GCCSA (which is unproblematic, since ineligible participants are screened). Targeting in the Facebook Business campaign did not involve keywords since Facebook and Instagram operate using specific connections and interests rather than a search engine model. Thus, individuals were targeted based on the following interests: homeowners, renters, employment, housing, retirement, and family (Appendix 16). This list was tweaked at various intervals for experimentation. This process was applicable to both Instagram and Facebook; one of the images used in the Instagram campaign is shown in Appendix 18.

The online advertising strategy was a form of voluntary sampling since the types of people likely to engage in the survey were specified. A target location was specified, and preliminary estimates on impressions and reach were produced. Due to broad sampling, there was not considered to be any significant bias in reach (see Appendices 16-17 for targeting), although there were inherent differences in interest based on which demographic groups were more likely to participate, as shown in Appendices 19-20. This issue was resolved by the multistage sampling approach, which allowed various segments of the population to be reached by more than one sampling means. For instance, community groups were the most successful method in terms of overall reach, primarily using Facebook groups and pages, and traditional websites and emails to a lesser extent. The Facebook groups/pages were predominantly local community groups aimed at community service announcements and history, or specific activities such as buying, selling, giveaways and social life. Additionally, some pages and groups were focused on renting and buying in the Brisbane GCSCA. The survey recruitment spanned all Brisbane GCCSA postcodes. However, suburbs within the stratified subset were prioritised when searching for community groups, for efficiency. These groups were more frequented by women than men, which was determined by monitoring the link between community posts and survey engagement.

Due to the survey’s mixed sampling, the number of eligible individuals who were shown the survey is unknown. Approximately 3000 individuals interacted with the survey link, but their eligibility is unknown, and some individuals likely visited the link on more than one occasion. If eligible participants are those who confirmed their age, length of stay in the Brisbane GCCSA and the minimum buying/renting of at least one property, then approximately 500 individuals were eligible across all postcodes.

The postcodes were sampled based on the following six attributes, sourced from ABS Brisbane GCCSA SA2/SA3 Data:

1. Age Bracket of Groups (Mode) 2. Percentage of Outright Owners 3. No of Dwellings (Percentage of Bedrooms) 4. Total Household Income Weekly 5. Weekly Rent 6. Dwelling Type

The six attributes were normalised into five classes; therefore, there were a total of 30 regions created based on the attributes above. SA2s and SA3s are the smallest sizes of statistical units utilised by the ABS. For this analysis, the difference between the two was immaterial, since SA3s are created using the aggregate of SA2 data, and this stratification process involved producing entirely new regions by intersecting the SA2/SA3 data with the Brisbane GCCSA mesh blocks. The mesh blocks are a ‘finer’ spatial configuration of land-uses and transportation. The aspect of the mesh blocks that was retained in this analysis is the urban footprint. Each attribute had a spectrum of classes, beginning with the lowest incidence, which was tagged as 1 and ending in the highest incidence, which was tagged as 5. All classes of 1 are intersected with one another, and this process is repeated individually for the 2nd, 3rd, 4th, and 5th quintiles.

The purpose of these quintiles was to produce a diversity of housing features and demographic profiles that were used to complement the data gathering process. The assumption was that by stratifying the spatial differences between values such as ownership and renting, low income and high income, these regions represented a diversity of contexts. The objective was to ensure that the prioritised postcodes (for the community groups only) captured individuals that experienced a range of dwelling circumstances across all age brackets. This does not suggest that the stratified regions represented all the possible combinations of habitation or demography. The quintiles overlapped and excluded specific combinations of individuals in the stratification. What this process ensured is that the postcodes which were targeted in this stream captured all six attributes and their respective quintiles. Moreover, this process was not used to eliminate any part of the region, as all areas of the Brisbane GCCSA region were targeted. The stratified regions were simply considered as priority target areas in this stream, due to their level of diversity.

The normalised classes were produced by firstly forming age brackets of: 0-9 years, 10-19 years, 20-29 years (mode), 30-39 years, 40-49 years, 50-59 years, 60- 69 years, 70-79 years, 80-89 years, 90-99 years, and 100 years and over. The mode (highest total of individuals within a bracket) was used. The percentage of 20-29-year- olds within each Brisbane GCCSA SA2 was then divided by the total of individuals in their respective SA2s, producing a normalised range with five classes. The percentage of outright owners in each SA2 was also normalised by using the total of all tenure

91 types in each SA2 and divided into five classes. The bedroom data used the sum of all dwellings in each SA2 divided by the total number of bedrooms in the respective SA2. This was used to create a normalised range of five classes, highlighting a spectrum of bedrooms per dwelling. The total household income was based on weekly income data, which was normalised by using the highest frequency household income bracket, which was the $2000-$2499 range. This range was then divided by the total of all weekly household income in the respective SA2s, producing a normalised range with five classes. Weekly rent was calculated using the highest frequency weekly rent bracket, which was the $450-$549 range. This range was then divided by the sum of all household rent in the respective SA2s, producing a normalised range with five classes. Dwelling types were calculated using the highest frequency dwelling type in each SA3, as a percentage of the total number of dwellings in the respective SA3s. This normalised value was then divided based on five range classes, which allowed for the diverse range of outcomes in each of the six attributes to be implied.

Figure 5.5 below shows the result of the normalised classes, which were intersected with the mesh blocks (shown in black). This intersection made it possible to identify the corresponding postcodes (shown in aqua), which were selected based on their centroid being situated within 2 km of a boundary of the mesh block intersect.

Figure 5.5 Stratified Regions showing the Overlapping Quintiles in Black and their Corresponding Postcodes in Aqua The overlapping regions were clipped to the mesh blocks, not to the SA2 or SA3 boundaries, and the end output was the larger region shown in cyan as explained above. Figure 5.6 on the next page shows that complete coverage of all LGA regions

in the Brisbane GCCSA was achieved. Though the stratification did not represent all the dwelling and demographic combinations from the six attributes, all six attributes and their corresponding quintiles (30 components) were featured, which were then used as priority target areas for the Facebook community profiles only. In Figure 5.6, the greyed-out postcodes were redundant, since targeting was cross-referenced with suburbs based on a single code; the Gold Coast postcode was crossed out due to this redundancy and since it is outside the Brisbane GCCSA. 4029 refers to the Royal Brisbane Hospital, which could not be targeted. 4072 refers to the University of Queensland, which was conveniently targeted.

Figure 5.6 List of Postcodes from the Stratified Regions

The average duration of the 237 surveys was 19.5 minutes. Survey data handling involved exporting the data from the Checkbox interface into Comma- Separated Values (CSV) Microsoft Excel data documents. CSVs are consistently used in this thesis as they are transferrable between software such as Checkbox, Excel, and Tableau. Thus, the survey scores (mean of preferences) and the causal coding for the open-ended question at the end of the survey were both saved as CSVs.

Table 5.3 Survey Participation Target and Outcomes % Male / Female % of Postcodes Covered Achieved% # Total # Target% Achieved% Target% 18-29 37.5% / 62.5% 18 / 30 48 50% / 50% 30-39 42% / 58% 32 / 44 76 49% / 51% 40-49 27% / 73% 18 / 49 67 49% / 51% 50-59 6% / 94% 2 / 29 31 49% / 51% 60-69 54% / 46% 7 / 6 13 48% / 52% 70+ 100% / 0% 2 / 0 2 45% / 55% All 33% / 67% 79 / 158 237 49% / 51% 102 (70%) 146 (100%) Source: Author as derived from (Australian Bureau of Statistics, 2016d)

Table 5.3 above highlights the survey’s age distribution and geographic coverage at the time of completion. There were 177 postcodes across the 9 SA4s of the Brisbane GCCSA, 146 of which were unique. Of the 146 unique postcodes, at least 102 were covered by the survey. Of the 472 housing profiles, 460 were associated with a postcode, and there were only 12 incidences where a postcode was not recorded. The survey was broadly disseminated based on organic reach, advertising, and community groups. Facebook community campaigns were modulated by examining GIS data to determine if the coverage was representative of the Brisbane GCCSA’s geographic extents. This modulation was achieved using a stratified random sampling of infrastructural, dwelling, and neighbourhood attributes.

Table 5.4 Distribution of Tenure and Typological Types by Response to Principal Dwellings for each Age Group Attribute Cohort/Typology 18-29 30-39 40-49 50-59 60-69 70+ Total % Tenure Owner 66 80 53 23 7 1 230 48.73% Renter* 108 73 42 15 4 0 242 51.27% Total 174 153 95 38 11 1 472 100.00% % 36.86% 32.42% 20.13% 8.05% 2.33% 0.21% 100.00% Typology Detached 98 103 76 28 7 1 313 66.45% Semi-detached 18 24 10 4 3 0 59 12.53% Apartment 54 26 9 6 1 0 96 20.38% Cabin 1 0 0 0 0 0 1 0.21% Dormitory 1 0 0 0 0 0 1 0.21% Boat 1 0 0 0 0 0 1 0.21% Total 173* 153 95 38 11 1 471 100.00% *One renter’s typology appears blank in the export

Table 5.4 above also shows the age distribution of the survey participants but based on their housing profiles (individuals can have between 1-6 profiles, due to the

longitudinal series), rather than their age ranges at the time of answering the survey. The tenure and typology of these respondents are also shown. According to Profile id (2020), 59.3% of Brisbane GCCSA residents owned their homes, while 33% were renters. A further 8.6% were unstated. Comparatively, 48.3% of this thesis’ survey respondents were owners, and 51.27% were renters. Regarding typology, 74.4% of residents lived in separate houses, compared to 66.45% for this thesis’ survey.

Table 5.5 below shows the different occupancy types by tenure and typology. While both renters and owners predominantly lived with a partner or family, the percentage of renters who lived alone was double that of owners. Also, 46% of renters lived with other members of the community. Thus, the owner cohort was a better representation of a family lifestyle. Moreover, owners who lived alone still lived in a variety of house types, but renters who lived alone predominantly lived in apartments.

Table 5.5 Distribution of Occupancy by Tenure and Typology Semi- Occupancy Detached Apartment Cabin Dormitory Boat Total detached own & live alone 11 6 6 0 0 0 23 (10%) own & live with a 176 12 12 1 0 0 201 (87%)

partner/family own & live with 3 1 0 0 0 0 4 (1.7%) renters/community

Owners own without 2 0 0 0 0 0 2 (0.8%) residing Sub-total 192 19 18 1 0 0 230 % 83.48% 8.26% 7.83% 0.43% 0.00% 0.00% 100.00% rent alone 11 27 12 0 1 1 52 (21.5%)

rent with a 85 34 24 0 0 0 143(59.3%) partner/family rent with flatmates 25 16 5 0 0 0 46 (19.0%)

Renters /community Sub-total 121 77 41 0 1 1 241 % 50.21% 31.95% 17.01% 0.00% 0.41% 0.41% 100.00% Total 313 96 59 1 1 1 471

5.3.2 Data Cleaning The Brisbane GCCSA Housing Needs Survey primarily consisted of 27 questions posed using Likert scales (Table 5.2). These themes were expressed differently for each form of tenure or occupancy: rent, own and live, and investment, or live alone, live with partner/family, and do not live in home. Thus, individuals were asked proxy questions that were tailored to their unique circumstances. The personalisation of the

survey questions (Appendix 21) meant that there were 72 questions altogether, and only 27 proxies. Data cleaning ensured that only the broader themes were analysed. Thus, the 27 financial questions were reduced to 9 themes, the 18 social-dwelling questions were reduced to 9 themes, and the 27 life-course questions were reduced to 9 themes. There were 27 remaining themes after the cleaning from a total of 72. These 27 themes are detailed in Table 5.2, using broader thematic names. Figures 5.1-5.3 show how the 72 questions correspond to each theme.

Individuals were queried about the relevance of the housing themes using a Likert Scale of 0-10. 10 was the highest form of relevance, and 1 was the lowest. This scale used 0 to identify null values (nv). Null values either referred to respondents that opted out, or they referred to satisfiers that were deemed irrelevant. This is because the default selection in the Likert scale was 0. Thus, if individuals left the scale at zero, they confirmed on their own terms that the question was irrelevant, or it represented that they did not know how to respond. The survey did not force respondents to fill all scores before progressing. This is because a zero score allowed for greater accuracy, by allowing respondents to rank the factors that were relevant, without having to force responses for the factors that are irrelevant (Manisera & Zuccolotto, 2013).

Thus, the effective range was 1-10 (responses with some degree of relevance) after the scores were cleaned for consistency. This cleaning was done by omitting scores of 0 from all responses; therefore, the aggregate mean and median scores by each individual factor were adjusted. The aggregate mean scores of the older cohorts were also adjusted by deleting null values, except for the cumulative/stacked mean values for each individual factor by age cohort. Yet, simply deleting null values from the scores would result in a low response bias for less frequently answered questions. Deletion would distort the relevance scores of questions which many people deemed inapplicable. The lower the frequency of responses, the more responses were ignored, and the higher the frequency of responses, the fewer responses were ignored. This bias can be adjusted to limit uncertainty by multiplying the null value adjusted scores with the relative frequency of expressed responses (Manisera & Zuccolotto, 2013), which is expressed as:

( ) = ( ) ×∑ 𝑥𝑥𝑖𝑖 𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥 𝑛𝑛 − 𝑛𝑛𝑛𝑛𝑣𝑣 𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑛𝑛 96

In other words, the adjusted mean is equal to the sum of the scores within the sample divided by the number of samples (adjusted for no null values), multiplied by the total sample less the number of null values and divided by the total sample.

5.3.3 Interpreting the Brisbane GCCSA Housing Needs Survey: Aspirations and Attitudes. This thesis strives to understand how well housing needs are satisfied in the housing market. Though individuals have unique circumstances (Section 2.4), the factors listed below in Table 5.6 comprehensively capture the diversity of universal needs. Thus, individuals’ housing progression is analysed by identifying the factors that are progressively accumulated (mean scores), based on age or residential mobility.

Table 5.6 Attributes of the Brisbane GCCSA Housing Needs Survey Life-course Attributes Financial Attributes Social-Dwelling Attributes Life-course House Home Dwelling Freedom, Interaction & Typology, Independence, Maturity, Asset, Value, Belonging, Continuity, Structure, Size, Employment, Marriage, Investment, Financial Privacy, Personal Rooms, Plot, Childrearing, Family, Security, Income, Identity, Status, Place Environment, Social Network, Cultural Location, Longevity Attachment Durability Source: Author as derived from (Moore, 2000 ; Ruonavaara, 2016 ; Saegert, 1985 ; Tognoli, 1991)

These housing needs were interpreted with three components:

1) The composition of individuals’ housing aspirations outlined how balanced the prioritisation of housing needs satisfiers were (Zuccolotto Adjusted Ranking). 2) The housing trajectory of individuals was revealed based on the progression of their social-dwelling, financial and life-course outcomes (K-means Clustering). 3) Based on causal coding, the intersection between individuals’ attitudes towards their individual needs and external factors was identifiable.

As mentioned in the research design, individuals indicated how they prioritised the life-course and social-dwelling attributes (shown in Table 5.6 above) based on a scale of 0-10. The questionnaire was then inserted into CSVs to compute the mean scores of all 27 factors. This formed the composition of housing needs. The life- course attributes formed the first category, followed by the home and dwelling facets, which were amalgamated to form one social-dwelling category. The last category was the financial attributes. There were nine themes for the life-course, nine themes for the

social-dwelling facets and nine themes for the financial attributes. The core social- dwelling factors were determined inductively from the research, based on whether factors such as typology, structure and size were consistently prioritised (based on their mean scores) above the life-course attributes. Establishing how changes in the life-course manifested materially was based on the capacity to express choice (Faulkner & Beer, 2011). The assumption was that if the life-course was the dominant priority, then individuals’ capacity to express choice was restricted.

Specifically, following the adjustment of the mean scores, the expression of life- course factors was studied by analysing the positioning (mean scores) of the life- course factors relative to financial and dwelling factors. Prior to the manifestation of life-course factors as lifestyle outcomes, life-course factors were simply opportunities or constraints. These factors were secondary to cost at the onset. ‘Attractive Cost’ which embodies the attractiveness of rental costs or a house’s price, signalled the capacity of individuals to express their needs. Thus, life-course scores below the cost- line (attractive cost) were desirable, while life-course positions above the cost-line were demanding. The cost-line was determined by the adjusted mean score.

Individuals’ housing trajectories were interpreted using K-means clustering, which is a machine learning algorithm that segments data points into clusters (k) based on the distance between every data point and the centre (mean value) of a cluster, also known as a centroid. This is known as the squared Euclidean distance. Tableau (the data visualization software used) uses the squared Euclidean distance since the square root of the sum is not taken (Beran, 2016). If the data was nearer to another centroid, then it was assigned to that cluster. In the process of reassignment, the mean of each cluster changed, and points were reassigned to the nearest centroid. Clusters were determined in Tableau, beginning with one cluster. The mean of a select variable was used to split the cluster in two, and new centroids were used to reassign the data points (Blum, Hopcroft, & Kannan, 2016). Tableau chose the number of clusters based on the maximum number of k required for convergence. Convergence occurred where the difference in the mean (variation) of all clusters was negligible (Tableau, 2020).

However, the number of clusters was manually set to four clusters based on the elbow method, using the fviz_nbclust function in R (Table 5.7 on the next page). The method shows the optimal difference between the inter-cluster and intra-cluster

distance based on the sum of squares (Kassambara, 2020). This is because, as the n of k increases, the variation in each k decreases. Thus, n trends towards 0 in the table below. Then, if the variation in each k is too little, it is likely that artificial boundaries have divided natural clusters, and if k are too few, then artificial boundaries have clustered dissimilar data points (Żero, 2019). Thus, the optimal point (elbow: dotted line) is where the decrease in variation lessens relative to the increase in k.

Table 5.7 Calculation of the Optimal Number of Clusters Using Elbow Method Renter Cohort K-means Cluster Owner Cohort K-means Cluster

Regarding the interpretation of the K-means results, in Tableau, the variables are known as measures, and this thesis used a single measure for the distances (the mean of the Likert scale scores). Another measure (the participants’ age brackets) was used to compare how ages are categorised within the clusters.

Finally, individuals’ attitudes towards the housing market were interpreted based on manual coding of the open-ended question in Q519 Appendix 21. Attitudes that were related (values coding) were used to identify common causal factors (causal coding). Causal coding is a form of pattern coding which identifies factors based on what events cause what (Saldaña, 2015). Both values coding and causal coding were straightforwardly derived since individuals were explicitly asked to reveal causation. Their attitudes were also easily observed since responses were concise, and the researcher was attuned to participants’ perspectives, having conducted a pilot study using semi-structured interviews (explained in Subsection 5.3.4). The factors from the causal coding were broadly classified within two themes: barriers derived from individual needs and barriers derived from external factors (Table 5.8 on the next page). There were 12 factors in total, which are used to explain the contradictions that pervade the Brisbane GCCSA housing market in Section 6.2 and Chapter 8. These 12 factors were considered comprehensive (satisfying data saturation) since they

addressed the core aspects of housing preferences: tenure, location, design, and price (Gurran et al., 2015) and housing precariousness: instability, and life-course pressures (Faulkner & Beer, 2011). Data saturation is generally understood as the point in which no new data, themes, or coding emerge. Thus, it is also the point where the data can be replicated (Fusch & Ness, 2015).

Table 5.8 12 Factors Representing Individuals' Attitudes towards the Housing Market: Explanations are shown in yellow Impact of Individual Needs (on Attitudes) Impact of External Factors (on Attitudes) How individuals’ needs stimulate change How housing processes stimulate change 1. Individual/family circumstances 1. Housing disputes and divergent attitudes

General life-course needs Rental/regulations and administrative conflict 2. Relationship and household unit 2. Home price changes changes/separation Personal and social relationships Rental price increases and fluctuation 3. Employment/education and family needs 3. Housing tenure Explanation Transport-locational Unsettling/insufficient regulatory environment 4. Financial capacity 4. Housing conditions Financial capital/wealth Gulf between housing costs and conditions 5. Unpredictability/uncertainty 5. Home improvement and upsizing Unpredictability of landlords and life-course Growing households and rental income Factors and and Factors 6. Wider/neighbourhood conditions/changes 6. Home downgrading/downsizing Neighbourhood changes Aging cohort needs

Through observations of housing needs variances, a sequencing of individuals’ housing trajectory and causation coding, this research can identify any housing gaps that may exist between individuals’ aspirations and market outcomes.

5.3.4 Potential Limitations of the Brisbane GCCSA Housing Needs Survey The retrospective longitudinal design was deemed viable following the conclusion of a semi-structured pilot study conducted in June 2017. The five individuals that were interviewed were residents of the Brisbane GCCSA and were able to recall their entire adult housing history with considerable detail. Subsequently, longitudinal housing moves from 237 participants were collected, totalling 472 housing profiles between 1959 and 2018. This range implies a wide range of housing experiences and associated recollections. Such extensive housing histories may be seen as relying on post hoc qualifications of lived experiences, due to the recall effect (Street & Ward, 2010). Hence, individuals’ recollection of housing moves might be perceived as imperfect representations of reality. Residents may also be susceptible to post- rationalisation and may suppress undesirable aspects of their dwelling experiences

100

(Jansen, 2014). This suppression makes it difficult to uncover people’s true aspirations, as opposed to normatively moderated outlooks.

However, it is important to emphasise that it is not simply the age of memory that regulates how easily memories can be retrieved, but also its degree of saliency (Fehring & Bessant, 2009). Housing can be described as salient because according to van der Vaart and Glasner (2007), long-term extended events such as housing tenure are the highest form of autobiographical abstraction and have been associated with enhanced recall (Tourangeau, Rips, & Rasinski, 2000).

Furthermore, although it is plausible that some outlier respondents could struggle to retrieve memories, the literature does not suggest that they would be incapable of recall. Memories of events can be retrieved with some effort and time (M. D. Williams & Hollan, 1981). That effort can involve providing a context, providing a search (within context) and allowing for verification. The context is how the participant describes the memory. While this thesis does not guide the respondents through these processes, the survey pre-empted the recall problem with the sequential nature of the questions posed. This design is likely to stimulate or encourage the process.

Regarding this thesis, a context was automatically suggested as part of the series of dwelling profile questions (see Table 5.2). The first component of the context was a question about living circumstances during a 10-12-year period in the individuals’ life. The second facet of the question stipulated that the house had to be in the Greater Brisbane region. The respondent may conjure up other contexts that support retrieval of events during that period in their life. The periods were roughly: late teenage years/adolescence, early adulthood, adulthood, midlife, retirement age, and the elderly phase. These lengthy contexts should have been easily retrievable since they were likely rich in accessible detail. Within each context, respondents could use cues that evoked their housing circumstances to recall their primary residence during that period and place.

Establishing the primary residence should not be a difficult task due to the saliency of housing memories. Moreover, respondents were asked contextual questions about who they rented/owned with, how they paid for the dwelling, the type/postcode of dwelling and who was involved in the decision-making process. These contexts provided cues that supported the retrieval of information on how

101

rental/ownership decisions were prioritised. Verification was the final step in the retrieval process, upon determining the basic characteristics of the primary dwelling, respondents verified their recollection by forming a link between the dwelling and salient life course events such as leaving their childhood home, their educational circumstances or familial events. Thus, the life-course Likert scale responses necessarily preceded the relatively more abstract and less salient social-dwelling and financial questions. Thus, if there was an anchoring bias (favouring the initial information provided), it was towards objective life-course events that ground the decision-making process (A. Mills et al., 2010).

Individuals’ life experiences had a lasting effect on their attitudes later in life. These experiences encompassed employment, educational changes, career changes, separations from family, aging, financial growth, or stagnation, among other life-defining events. 84% of participants (the question was optional) recalled these same factors with detail in the open-ended question about attitudes towards housing needs and external factors. Moreover, the open-ended question showed that negative housing experiences were not suppressed.

5.4 Social Network Analysis of Real Estate Advertisements This thesis explores how individuals’ needs and aspirations are satisfied, which is distinct from a rationalist understanding of the housing market (Clapham, 2018), which considers needs to be implicit in price. The introductory chapters explained that the housing needs that people have are centred not only on financial outcomes but include the practical uses of housing as a place to live. These housing needs reflect how housing qualities and socio-environmental contexts are associated with specific typologies or attributes. This section will introduce the philosophical underpinning of the SNA method, prior to explaining how words can be representative of the housing market, and how such an application has previously been implemented. SNA is a sociological toolkit that is based on graph theory, intending to understand networked systems through vertex and system-level analysis. The method was chosen because the entire housing market is unknowable. Instead, networks that acted as proxies for the rental and sales markets were created based on the most relevant real estate terms from advertising. In the network, real estate terms (vertices) were the objects of the system. At the system level, the network could be understood as a collection of distinct communities, while at the vertex level, the most prominent ideals were visible.

102

As mentioned, the philosophy guiding the application of SNA was based on the role of property terms as the medium/currency for the feedback between property buyers and renters, and real estate agents. Property terms reflected how individuals’ consumption was facilitated by the market since demand-side needs were mediated through supply-side marketing. Thus, the balance between supply and demand was founded on reciprocity, which influences how property language is shaped. The general assumptions were that demand and supply (production) are interdependent, and the production process was visible in the advertising of housing products. This resolved the inability of the equilibrium model to capture subjective housing needs.

The SNA method was employed in this thesis because social structure as a whole is intangible, the housing market does not exist as an object in the world, we cannot see the housing market, what we observe is how this unseen structure influences observable human behaviour (Lewis, 2000). The market is observable in the form of physical houses, other built environment elements, financial transactions, spoken language, written language of a historical and contemporary nature, building codes, and socio-economic factors etc. Each feature conveys a unique aspect of the market. Some aspects of the housing market are interrelated, but housing features are irreducible to their underlying aspirations. It is impossible to create a network that incorporates all aspects of dwelling, home, house, and every individual need, constraint, capacity, and external factor. These factors form a complex web of variables that often contradict one another and manifest in the abstract, as well as physical attributes. Potential buyers may aspire to ideals that are unrealisable in the housing market, and it has often been the case that only the physical housing artefacts of their aspirations are observable.

This thesis delved beyond said physical limitations by exploring meaning. The study of meaning has two relatively independent roots. Some researchers have studied meaning from a psychological perspective using scales and themes, while information science has focused on understanding meaning by operationalising words and their co-occurrences (Leydesdorff & Hellsten, 2006, p. 234).

In the Brisbane GCCSA Housing Needs Survey, the perspective of renters, homeowners and investors were queried using Likert scales. This method represented the subjective prioritisation of buying and renting over the life-course. Hence, the

103

detailed analysis of individuals interactions established a theory of meaning. This theory of meaning was egocentric and did not signify the network; individuals were queried on their expectations and perceptions. Critical realism emphasises that it is not the intent of individual actors, but the interactions between all actors that results in material or physical outcomes such as financial transactions and words. Thus, the interactions between property terms were used in the SNA to show how the real estate market facilitates the interactions between individuals. This is due to economic feedback. This can be explained with the following hypothesis by Stuart and Botella (2009, p. 2) “….language, and in this case, written text is the vehicle of exchange and transmission of knowledge between the members of a discourse community”.

The above hypothesis is especially relevant to this thesis since 90% of Australian real estate agents advertise online, and 86% of buyers in the Australian housing market discover their new property online (buyMyplace, 2016). Additionally, 65% of renters in the Australian market rent through private agents and 76% of these renters rely on real estate advertisements (Rowley & James, 2018). The percentage of online users and advertisers are co-dependent because real estate platforms exhibit indirect network effects; more users on a real estate platform encourages more real estate agents to advertise on that platform (RBB Economics, 2016). Thus, prominent real estate websites such as realestate.com.au, domain.com.au and realestateview.com.au (which this thesis utilises) cater to the variety of property types and locations available in the housing market.

Key pillars of real estate campaigns are price, sales, property descriptions, and property reports (Boehm, 2018 ; Department for Business Energy & Industrial Strategy, 2017). A survey of homebuyers in England and Wales found that property descriptions were the second most received datapoint in housing sales, and the most helpful (Department for Business Energy & Industrial Strategy, 2017). Although the transparency of agents’ sales history and performance are limited (Department for Business Energy & Industrial Strategy, 2017 ; McCarroll, 2016), this is because there are few mechanisms that validate agents’ performance, though there are some, such as the RateMyAgent platform (McCarroll, 2016). Conversely, property descriptions are easily validated through physical inspections. Real estate agents communicate with both vendors and buyers and hone their advertisements based on the expectations of vendors, as well as buyers’ attributes. If agents do not honour their obligation to

104

vendors, they will lose jobs; if they do not reflect some of buyers’ needs, they will lose sales and rentals. Also, while 5% of agent facilitated rentals and 12% of privately facilitated rentals are discovered via word of mouth (Rowley & James, 2018), this largely represents referrals from agents’ clientele. Thus, agents and their property listings are intermediaries of the interactions between vendors and consumers.

Admittedly, despite this intermediation, agents maintain a greater hold on information than buyers (Gee, 2010). However, the market’s conceptualisation of the housing market remains an effective abstraction of real housing conditions, albeit incomplete. Due to the validation that consumers receive through physical inspections, agents are also inherently selective with language. Selective advertising must be employed since the appeal of housing features are particular to different renter and owner markets (Camilleri, 2018). Thus, real estate listings must be a comprehensive and concise representation of the housing market.

The selectiveness of agents was explained with the Gricean maxims of quantity, quality, relevance, and manner. These maxims were proposed by H.P. Grice and detail principles required for cooperative communication (Grice, 1975). Based on these maxims, property agents will maximise the potential of their text-based property descriptions, since they are rational actors (Zhang, 2004). Quantity means that advertisements provide only the required amount of information, quality relates to the descriptions being as truthful as possible, relevance concerns the appropriateness of the advertisements, and manner relates to the brevity and unambiguity of the listings (Zhang, 2004). These maxims regulate the consistency in the language of real estate agents, ensuring that language complexity has been streamlined, due to brevity, familiarity, and saleability.

These maxims ensure that agents will actively moderate their language based on how individuals have previously responded and how they are likely to respond. This relationship necessarily involves language selectiveness; agents will market properties based on neutral or positive terminology. This is because prospective buyers and renters are searching for properties that they desire. Real estate advertisements can then be thought of as mirroring the more desirable features of properties. Thus, property terms do not represent all the characteristics of a house, but they do reveal the favourable dwelling features that individuals in the housing

105

market observe or consume. This perspective highlights the critical realist notion that causal mechanisms dilute individuals’ needs through their interactions with other people and environments.

5.4.1 Introducing Social Network Analysis Social Network analysis was derived from graph theory (the study of graphs) to formalise social order. Therefore, the underlying structure of SNA is based on mathematical regularities, which can be applied to a mathematical graph, a social network, a concept network, or an ontology network (Stephen P Borgatti, Mehra, Brass, & Labianca, 2009 ; Settas, Sowe, & Stamelos, 2009). Thus, SNA is not limited to social networks but can be used to study any network that has regularities. However, rather than defining an entirely new field of study, researchers have mostly used the term ‘ontology-based social network analysis’ (OSNA). Examples of ONSA include a study by Sam and Chatwin (2012) who developed a model on how to codify emotions and electronic ontologies based on classes and subclasses of data, which were then joined and analysed for similarity. Settas et al. (2009) created rules (having mutual causes, symptoms, or consequences) to study how antipatterns in project management were related, while Kim and Rhee (2019) used Protégé ontology editor to build a network of journal article topics based on how keywords were correlated.

OSNA combines knowledge modelling with the comparative and analytical toolkit of social network analysis. In this thesis, knowledge is modelled using property terms to describe Brisbane GCSSA rental and sale properties. This approach allowed the housing network to be reproduced by identifying the prominence and context of real estate terms based on how they co-occurred in text. Prominence describes which terms had the greatest information flows, while the context was established by grouping terms into distinct housing needs facets and communities.

Traditionally, the market’s complexity is reduced by condensing housing into homologous variables such as house types and using forms of utility such as income to model individuals’ interactions within the market. These approaches assume that individuals are rational actors capable of maximising their potential. However, this thesis shows that individuals’ needs are not necessarily implicit in housing outcomes. Thus, this thesis condensed the elements of the housing market (location, finance, neighbourhood/housing amenities, typology, distance, spatial dimensions, and social

106 constructs) and mapped them against the housing needs layer. This simplified the complexity of the market, as well as highlighted how housing needs were satisfied relative to specific groups: renters, owners, inner region, middle region, outer region, apartment-dwellers, house-dwellers, and landowners.

Moreover, the simplification of the market was not merely a logistical consideration but was also influenced by the need to isolate the most important terms (vertices) and their level of interaction. This is in contrast with replicating the syntax of real estate terminology or general linguistics, wherein some words are typically more used than others. Instead, co-occurrence revealed that a term was important because it was associated with other important terms, not simply that it was widely used in the corpus.

Following the reduction of the market to its most elementary outcomes, the analysis revealed whether the fundamental needs of various Brisbane GCCSA housing divisions (Subsection 5.4.3) were satisfied in a diverse and comprehensive manner. These needs were highlighted in Section 2.2 in the form of three themes: dwelling, house, and home. Hence, the SNA approach emphasised housing market features as well as how they related to housing needs. This is because the property terms were manually classified into the housing facets (home, dwelling and house) based on how they corresponded to the structures of activity (see Table 2.1 and Tables 5.16-5.17 in Subsection 5.4.2). Similar to the approach used by Raamkumar (2016), the categories were predefined so that they could be compared with the automatically generated categories from the community algorithms. This is very different from studies such as that from Nowak and Smith (2017) that extracted housing features from real estate listings but only used them as a predictor of house prices.

This thesis’ methodological approach avoided any such ambiguity by directly studying if segments of the housing market satisfied the entire spectrum of housing needs and whether such satisfaction was balanced. The terms were a form of currency in and of themselves since it was assumed that real estate agents are rational actors who disseminate useful text. These agents act as such because their advertising accuracy will be scrutinised. Thus, language renders the features of the real estate market interoperable, since the common denominator is that all text is written based on market relevance. The Nowak and Smith (2017) approach are to essentially

107 consider that “…. monetary numbers render the objects they measure calculable and relative” (Carruthers, 2010, p. 58), but this thesis showed that the meaning behind housing and built environment features are also measurable. The dwelling, house and home facets can be theorised as the underlying basis for a sustainable housing market. This distinction is important because assessing the properties and relationship of the text, directly questions their substance, whereas reverting to price simply reveals the exchange-value. As mentioned throughout this thesis, exchange-value is predicated on many factors that are not exclusively mediated by individuals’ immediate needs.

While the descriptions within real estate listings may predict house prices to an extent, this thesis focuses on revealing the extent to which listings facilitate housing needs. The central components within the housing networks developed in this thesis should be perceived as the consequence of feedback loops, wherein each vertex (real estate term) influenced the position, outcomes, and attributes of other real estate terms. Thus, how a vertex was connected, shaped its identity. These vertices were the objects of the network, which will be explained further in Subsection 5.4.4.

When the housing market is reproduced as a network, some complexity is lost, but emphasising the edges between terms, preserved key relationships. The various forms of textual relationships are explained in Table 5.9 on the next page. The table includes some literal network terms. Semantics were inherent in the edge relations of the SNA since they represented the contextual relationships between the terms. In real estate advertising, localised property terms such as price, unique, rent, living, dining bedroom, kitchen, and walk, and city-wide attributes (amenities shared by various submarkets such as Brisbane, CBD, City, Gold Coast, drive, and distance) were both reflected in real estate advertising. Thus, irrespective of buyer or renter preferences, fundamental needs should be satisfied in all subgraphs of the Brisbane GCCSA submarket. This is because, if needs are not generally satisfied (irrespective of spatial boundaries), then due to spatial fixity (wherein housing attributes are fixed), underlying housing needs will be dispersed and unequally distributed, which will encourage excessive residential mobility.

Thus, semantics was employed to determine whether needs were satisfied locally or by nonadjacent submarkets. For example, if the weight of the term ‘distance’

108

was considered independently, then its meaning is ‘Fuzzy’, but if it were considered in relation with ‘city’, then the implication would be narrowed to the context of the CBD/city. If this relationship were exhibited in an outer suburban subgraph, for example, it would imply that spatial non-adjacency (the listings are relying on attributes across several submarkets). Thus, preserving the various typological, economic, and social relationships (edges) between property terms ensured that a superficial representation of the terms was not established since a single term can infer numerous meanings (see Table 5.9).

Table 5.9 The Different Forms of Meaning in Real Estate Language Relation Definition Example Implication

Suggested meanings to the user – Spacious Acreage, Large Lot, Connotative derived from the lexical unit Open Plan

Denotative Literal associations and Properties Home House, Dwelling lated to Without boundaries, although derived Inner-Middle-Outer, 7- the word the re Fuzzy Distance from the denotative meaning 10-15km, Short, Far

What the agent does not want the

Inferences buyer/renter to know based on a Derelict Established

mplication is mplication person’s preconceived understanding Neighbourhood I What the agent wants the buyer/renter Implicature to know based on a person’s City Modern, Cosmopolitan contextual preconceived understanding Source: Author as derived from (Crystal, 2008 ; Zhang, 2004)

Meaning can be studied based on similarities, social relations, interactions or flows between property terms (Stephen P Borgatti et al., 2009). SNA emphasises the relations between system properties, not simply each attribute’s individual qualities. Thus, the relative importance of attributes can be shown, not just the aggregate financial value or dwelling stock. This thesis’ SNA emphasised the relationships between prominent housing market features, which signified how responsive the market was to the three housing facets. Prominence was shown through dominant property terms (vertex/nodes) as well as the (edges/ties) between them. Real estate property terms were the objects of the network, in, while edges showed the connections between the terms (Table 5.10 on the next page).

This thesis’ SNA considered listings with greater depth as a positive representation of housing aspirations and those with lesser depth as negative representations. Depth indicated how well the property terms of each network/subgraph corresponded with the dwelling, house (financial), and home

109

(social) descriptors. The categories: dwelling, home, and house, embodied the three means of housing needs satisfaction, and the potential for aspirations to be realised. These categories signified how real estate agents conveyed the potential and breadth of individuals’ consumption. This was a multilevel process that involved finding the most focal terms, network/subgraph density, and their distinct communities.

Tables 5.10-5.13 below show the features of SNA, centrality, and community measures. These network features will be explained later in Subsections 5.4.4-5.4.7. Table 5.10 below shows the components used to construct an SNA plot.

Table 5.10 The General Features of Social Network Analysis Description Function Impact A table consisting of the names of vertices (property terms) as the row and column Example Matrices are used to Matrix headers. If a value exists between two terms Table 5.19 plot the network (adjacency), it means they co-occur intext. Networks have vertices and edges which Tenure shows 2 networks Network show how objects are related. In ontology differences between are analysed SNA, objects are similar since they co-occur. renting and ownership Clusters within a network. The Brisbane How geographies and 8 subgraphs Subgraph GCCSA housing network comprises 8 typologies shape are analysed subgraphs. satisfiers Vertices are the objects of the network. The 69-72 Prominent vertices specific quality or property being observed. vertices Vertices facilitate satisfaction The unit of observation, whether it be people, (property of needs animals, names, terminology, or artefacts. terms) The connections between the objects Widths: Nearly all subgraphs/ Edges (property terms) of the network. Figures 7.4- network edges are 7.23, 7.25- linked. Thus, visible Edge weights are determined by the number 7.26 and edges show only the Edge of connections between a vertex and another Weights Appendices key paths for vertex. Weights multiplied by 60 = Widths 22-25 information flows.

The matrix compiled the weights used to form network plots based on the information flows (edges) between property terms (vertices). Igraph represented these edges using widths, which ordinarily equal the value of weights. However, since the values were too small for adequate projection, a multiplier (60) was used. The rental and sales network were analysed independently and were also divided into smaller networks (known in this thesis as subgraphs). The density, centrality, and community features shown in Tables 5.11-5.13 were necessary to interpret these plots. Regarding

110

the centrality measures, WDC was used to sum all information flows, while CC and EVC were used to reveal what types of terms were connected. See Table 5.11 below, as well as Subsections 5.4.4-5.4.5 for more detail.

Table 5.11 The General Features of Network Density Measures Centrality Measures Function Impact Used in all Words that co-occur Weighted WDC (known as Graph strength in SNA plots share meaning. WDC Degree igraph) is a sum of all the adjacent (edges and sums co-occurrence Centrality edge weights for a given vertex (Csardi, vertices) and (information flows). (WDC) 2015c). in Section Thus > WDC = > 7.3 prominence of data. Code igraph: V(g)$graph.strength <– graph.strength(g) = 𝑠𝑠𝑠𝑠 � 𝑤𝑤𝑤𝑤𝑤𝑤 ∙ The shortest distance (average The mean of𝑗𝑗 all edges Used in Closeness distances) between a vertex and all the = CC. Terms with high Centrality Section 7.1 other vertices in the network (University CC are consistently (CC) (Sankey) of Chieti-Pescara, 2014). related to other terms igraph: Dist <– distances(g, mode="out") Code Dist[Dist == Inf] <– vcount(g), 1/sum( d(v,i), i != v) close <– 1/rowSums(Dist) A value is assigned to each vertex High EVC = connected Eigenvector based on centrality. That value is then Used in to important terms. A Centrality adjusted based on whether the vertex Section 7.1 term with high WDC (EVC) was tied to terms with a high score or a (Sankey) and low EVC has many low score (Csardi, 2015b). weak edges Code igraph: c.e <- evcent(g)$vector $S Λ S^{-1}$ Used to High centralisation The extent to which the network is describe means focal terms are Centralisation distributed among key central vertices SNA plots in more prominent than (Stephen P. Borgatti, 2015) Section 7.2 the rest of the network

Table 5.12 on the next page shows how networks/subgraphs were described using density. The most common definition of density was scarcely used in this thesis since the ratio of edges to the maximum possible edges did not reveal much variation between plots (clarified in Subsection 5.4.5). Clustering coefficients better discerned the density of real estate terminology since it was based on clustering within each individual listing. This density implied that words that were used together shared meaning (Harris, 1954) (explained in Subsection 5.4.2). The implication was that networks with high CLC (density) across the entire network had diminished meaning since terms were used indiscriminately across non-prominent terms, while networks

111

with low overall CLC had higher clustering centrally and their most important (focal) terms had a greater impact.

Table 5.12 Summary of the Centrality Measures Applied in the Thesis Description Function Impact Two/Three (dyad/triad) vertices and all Mutual ties are Dyads show possible edges between them. Only the flows from a-b and edge relations Dyad/Triad mutual or asymmetric edges of dyads are b-a. Asymmetric in Subsections analysed in this thesis; edges that do not flows are only 7.2.2-7.2.4 occur are not analysed. from a-b Triplets are two connected vertices that share Low CLC = more a connection with a third vertex. “friends of isolated terms Triplet friends become friends” (Veenstra, Dijkstra, overall, but focal Steglich, & Van Zalk, 2013) terms are more CLC shows the probability of network edges Subsection interrelated. High being clustered. This is calculated based on 7.2.1 (Density) CLC = less Clustering the number of triplets divided by the total isolated terms Coefficient number of possible edges. See Appendix 27 overall, but focal (CLC) for the code. This is the form of density terms are less studied in this thesis interrelated Only used for Terms in high- Most commonly, density refers to the number cut-off density networks Density of edges in a network/subgraph as a ratio of width/threshold are relatively more the maximum number of possible edges. (Appendix 28) important

Lastly, community measures can also be used to describe networks. These measures were first introduced here as a reference point (Table 5.13 below) but will be described in more detail in Subsections 5.4.4-5.4.5.

Table 5.13 Summary of the Community Measures Applied in the Thesis Community Measures Function Impact Modularity determines if there are more edges within a Used for community Modularity particular group than the edges that exist outside of that models group (Comber, Brunsdon, & Farmer, 2012). Vertices are assigned to communities where they gain Used to MLC the most modularity. Once no vertex can be moved, determine Multilevel groups Community communities form new vertices; these new vertices are how are more (MLC) assigned to new communities if they can gain modularity diverse stable & (Blondel, Guillaume, Lambiotte, & Lefebvre, 2008). housing accurate Vertices are assigned into separate groups, and these features Fastgreedy since it is Community pairs are merged based on which edges gain the most in are, based multilevel (FGC) modularity (University of Chieti-Pescara, 2014). on groups

112

Moreover, Subsection 5.4.7 will detail how the centrality and community measures were applied in this thesis. Firstly, the general functionality of SNA will be described. Earlier, it was clarified that the housing market was reproduced using real estate proxy terms. Connections or edges between these proxy terms (property terms) were able to be established since the vertices (property terms) were like for like. The vertices must either share the same location, membership, attribute, or social relationship and must share some form of interaction or flow.

In this thesis, mutual interaction was represented through the flow of information between real estate property terms. These flows conveyed the market’s representation of how needs were satisfied and was regulated because the information disseminated to renters and buyers was crucial for successful sales and rentals. Agents intermediate between vendors and consumers, which is moderated based on renters and buyers contrasting advertisements against physical inspections or past experiences. The information flows contained within real estate advertisements are formally analysed as edges. The goal of SNA is to understand “…. how these different kinds of edges affect each other” (Stephen P Borgatti et al., 2009, p. 893). Specifically, this thesis revealed how prominent property terms became influential, based on their interactions with other focal vertices, and peripheral terms to a lesser extent. This relationship established which housing needs satisfiers were important in the market. These satisfiers were embodied in the home, dwelling, and house facets.

Moreover, edges (interactions) do not only signify how the network was formed (words co-occurring in the same property extract); they also represent how the network can be analysed. In this thesis, edges were directed, meaning that there was a distinction between the edges’ source and target. A network may also have a weight for its edges. Network edges can be constructed with edgelists or with an adjacency matrix. Edgelists typically contain three columns, the source of the edge (1st column), the target of the edge (2nd column) and the weight of the edge (3rd column). Instead, in this thesis’ adjacency matrix, the source vertices form the column headers, and the target vertices formed the row headers. Normally, for undirected graphs, adjacency matrices are symmetrical since the values between the columns and rows are identical to that between the rows and columns. However, in this thesis, the adjacency matrices

113

were directed. The weights in the matrices were shaped by the number of times two property terms co-occurred; this will be clarified later in this subsection.

Similarly, vertices should not be seen as independent attributes with the same potential network impact, irrespective of how they are interconnected (Marin & Wellman, 2014). Two networks with the same vertex weights will have different dynamics based on edge relations. Consider a network with three terms: ‘home’, ‘privacy’, and ‘dwelling’. In this hypothetical study which uses co-occurring word frequencies, ‘home’ is connected with ‘privacy’ with a weight of 0.5 and with ‘dwelling’ with a weight of 1, while in a second scenario, ‘home’ is connected with ‘dwelling’ with a weight of 0.5 and ‘privacy’ with a weight of 1. In both scenarios, the sum of weights for the vertex ‘home’ is the same, but the networks are different. This is because the identity of ‘home’ is partly derived from its association with ‘privacy’ and ‘dwelling’ in both networks. This relationship can only be observed by isolating the specific interaction or class of interaction. Thus, showing edge weights would communicate each pairing’s strength. Otherwise, the identity of ‘home’ is indeterminate, it may be analogous with ‘dwelling’, or it could more closely reflect the nature of a home environment (‘privacy’).

In this thesis, SNA was analysed with the R software, which was designed for statistical computing and graphics. R utilises its own computing language known as the R language, which allows users to manipulate data using functions in lines of code. There are many mathematical operations that can be run with the language. R also allows the installation of plugins (packages) and does not require external programs. This feature makes it advantageous for data reproducibility since research processes (code) can be specified entirely in the software (Appendices 26-27). R has functions that are applied in the base package (the native functionality of the software) such as multiplication of datasets and importing Excel CSVs, but the specific functions that are required for SNA such as graphing the edges and vertices based on adjacency, rendering the vertices and determining the weights of edges graphically and mathematically, are computed using an R package. R packages such as the igraph package used in this thesis, extend the functionality of base R. Igraph allows users to add vertices, edges, specify network layouts, and conduct the analysis of centrality and community algorithms. This is also done using line-by-line coding by attributing the specific qualities that the user desires to functions recognised by R and igraph.

114

5.4.2 Real Estate Data Collection and Sampling This subsection details the data collection and sampling processes of the SNA method. The data collected was in the form of Brisbane GCCSA rental and sales property listings, sourced from the realestateview.com.au website in April 2018. Property listings were collected since they described the state and context of dwellings on the market (at the time of collection). This is key because RQ4 aims to uncover how the satisfiers of household needs are facilitated by the property industry. Property descriptions reflect the housing market’s perspective on housing since these descriptions are necessary for marketing, sales, and leasing properties. The interactions between the real estate terms represent how economic feedback is facilitated between renters, buyers, real estate agents, investors, and developers in the housing market. Since marketing, leasing, and property sales are reliant on efficient communication, efficiency is refined over time, based on feedback.

The data extraction process yielded 850085 words for the property sales data, which had to be further refined into 72 concepts. Similarly, 18513 words were harvested for the rental data, which had to be refined into 69 concepts. This refinement was necessary to allow for a more relevant and manageable network of relations.

The real estate listings covered 35-45% of all Brisbane GCCSA rental market listings (5000 Records), as well as 20-45% of all Brisbane GCCSA sales market listings. (3300 Records). The auto-coded real estate data were manually assigned to themes and colour-coded based on the home, house, and dwelling classifications.

The data was harvested by conducting a search for “Sales” “Brisbane – Greater Region” in realestateview.com.au. Harvesting or scraping refers to the extraction of online content, such as textual data. Web scraping was used to collate datasets by amalgamating the targeted content and disregarding noise (Mihalcea, 2017). Property descriptions were the textual data that were targeted in this study. The URL for the “Sales” “Brisbane – Greater Region” search was copied into the Outwit Hub software for scraping. This software extracted, organised, and exported textual data as CSVs, which would be too tedious to manually compile. Within Outwit, the web link for a single page (page=1) was expanded to include all pages of the Brisbane – Greater Region by using the string edition & generation panel (see Appendix 13). Additional rows were inserted into the query string, which is a web address scraper designed to extract the

115

URLs of all the pages by using the auto-explore pages function. The “Fast scrape (include selected data)” mode was used, and the web address scraper was selected from the drop-down menu (see Appendix 14). This generated an address list for each sales property listed in the Brisbane GCCSA, which was stored in the software’s scraped section and then saved in a sales-list in the queries. Then a scraper was built that fast scraped the URLs of all sales-list property addresses, which populated the property description, postcodes, and typology fields (see Appendix 12). The same process was repeated for the rental data, with “Rent” in the real estate search.

The raw real estate data was then organised with the following headers: ‘postcode’, ‘typology’, and ‘property descriptions’. These headers were used to populate the housing themes, as well as for analysis. A sample of the property description header (derived from Brisbane GCCSA property advertisements) is shown in Table 5.14 on the next page. Highlighted in turquoise are the relevant property terms. How relevance is determined will be explained in pages 119-120.

However, the significance of Table 5.14 is how the advertising mechanism of the market can be condensed using select terms (highlighted). Reducing that mechanism to only its most important components eliminates noise - a combination of general articles and contextual nuances (Bourgeois, Cottrell, Lamassé, & Olteanu, 2015). In the table below, ‘home’ co-occurs with ‘location’, the two important terms in the first listing. The co-occurrence between these two terms signifies a flow of information that is more significant than the individual frequency of each term. This is because words have regularity.

Harris (1954) was the first to determine that words that are similarly distributed have a shared meaning. Co-occurring words share meaning because they likely relate to the same semantic unit (Lancia, 2007). For example, in the first sentence of the second listing (Table 5.14), ‘shops and ‘beach’ both relate to ‘walk’. Similarly, the flow of information between important terms that co-occur is more universally significant than the flow of information between non-important vertices. For example, the occurrence between ‘quietest’ and ‘Aspen’ (Table 5.14) is contextually important but is not universally meaningful in comparison with other property listings. Also, in addition to the regularity of similar types of words proposed by Harris (1954), in SNA, prominent terms are generally related to other prominent terms (Griffin, 2017).

116

Finally, this thesis evaluates the premise of target marketing (selective marketing will be used henceforth to avoid confusion with market targeting). Selective marketing refers to the advertisement of a product based on the quality of the product (market positioning), a segment of the population (market targeting), or based on the shared characteristics of individuals in a market (market segmentation) (Camilleri, 2018). These strategies influenced which property terms were emphasised, and this was reflected in the weight (co-occurrence), attribute (what facet) and depth (how balanced the housing needs satisfiers were) of the property terms. Hence, advertising strategies influenced the focus of property listings based on their typology, geography, and tenure. This will be discussed in Subsections 7.2.2-7.2.5.

Table 5.14 Real Estate Property Sales Descriptions Sample Typology Property Description Positioned in the quietest street within Aspen In The Sun, the location of this House standalone home is extremely central. Superbly positioned only minutes walk to shops, beach; transport. Hard to Unit find a ground floor 2 bedroom unit with an ensuite, open plan living; dining areas plus a spacious covered outdoor eating entertaining area. Source: (Realestate.com.au, 2018)

Following the organisation of the raw data into headers, the sales and rental data were cleaned (eliminating HTML tags and references). For typological headers, real estate websites often use inconsistent names, i.e. separate, detached and detached dwelling, which all represent the same information. Thus, corresponding typological classifications were merged. Shown above is a sample of the data, without the postcode header. The sample shows that each property row is associated with a single typology.

The typologies and geographic classifications are shown in Table 5.15 on the next page; there were 4696 property listings in the sales data and 3382 listings in the rental data. Consistent with the typological patterns explained in Section 3.7 and Subsection 5.3.1, the rental cohort has a more diverse typology. The geographic distribution of the sales cohort is also shown. Most Brisbane GCCSA residents live in the middle region, and only 11% live in the well-serviced inner region (see Subsection 5.4.3 for spatial boundaries). The sub-total of this division was greater than the subtotal of the typological sub-division since postcode regions (shapefiles) from the Australian Bureau of Statistics (2016f) were used, not suburb level postcodes.

117

Table 5.15 Distribution of Real Estate Sample by Tenure, Typology and Geography Tenure Typology Total % Geography Total Postcodes % Apartment 998 21.25% Inner 528 24 10.80% House 2364 50.34% Middle 2497 87 51.09% Sales Land 1334 28.41% Outer 1862 66 38.10%

Sub-total 4696 100.00% Sub-total 4887 177 100.00% % 58.13% Apartment 1427 42.19% Rentals House 1955 57.81% Sub-total 3382 100.00% % 41.87% Total 8078 100.00%

There are no official boundaries for suburb level postcodes. Postcode areas are created by assigning suburb level postcodes based on how many dwellings they contribute. However, the real estate data was filtered based on the suburb level postcodes used by real estate websites. Thus, 31 postcodes and 191 properties overlapped. Alternatively, dwellings’ geographic coordinates could have been used; however, not all addresses were disclosed. The overlap represents edge properties and thus does not distort the broader geographic division (see Subsection 5.4.3).

The sales and rental listings were amalgamated into two separate Excel documents and PDFs and were uploaded into the Leximancer text-mining software. The software reveals the main concepts of complex datasets. In this thesis, isolating only the key property terms is crucial; else the SNA would have too many terms for analysis. In the SNA sales document, there were 850,085 words in the property description field from 4696 property listings, which amounted to 181.02 words per listing. A sentence range of 15-25 words translated to either 6 or 10 sentences per block as the optimal setting in Leximancer. Only 1,2,3,4,5,6 and 10 sentences were viable; 6 sentences per block was used. Sentence blocks are the blocks of text that determine if a word is relevant based on the frequency of occurrence and co- occurrence within each block (D. A. Thomas, 2014). For the rental data, there were 518,513 words in the property description field from 3382 property listings, which amounted to 153.27 words per listing and 6 sentences per block as the optimal Leximancer setting. Leximancer identifies ‘name like’ (proper nouns) and ‘word like’ concepts by association, within contexts and text excerpts, as well as based on the frequency of occurrence. The significance of context blocks is that the relevant

118

property terms were identified locally based on the average number of words per listing, not the size of the entire corpus or housing market.

Leximancer generates the number of targeted concepts (property terms) automatically or by specifying a target. 139 and 149 property terms were generated for the rental and sales categories, respectively. Textual relevance was confined to such a limited range due to an accelerated decrease in relevance as rank reduced. This decrease could plausibly represent a power-law distribution in both textual corpora. This law means that in textual corpora with enough depth, the frequency of any word will decrease proportionally to its rank (Moreno-Sánchez, Font-Clos, & Corral, 2016). Moreover, in networks displaying the power-law distribution, a few vertices are disproportionately connected, relative to other vertices (Himelboim, 2017). The targeted sales terms had a p-value of 0.57, and the rental range had a p-value of 0.82. A p-value greater than 0.1 denotes a plausible power-law distribution (Clauset, Shalizi, & Newman, 2009). Other distributions may also explain why the frequencies of property terms suddenly plateau by the 130-150-word relevance threshold.

The lowest-ranked frequency (count) generated for the sales and rental markets were 62 and 92, respectively. In Leximancer, the relevance threshold measures how frequently a term is contextually employed. The relevance of the terms was decided by its context (sentence blocks). Upon automatically sorting the terms by relevance, the applicability of the most significant terms was manually determined based on the frequency of occurrence. Moreover, data saturation was reached halfway before the maximum threshold; hence, the targeted spectrum provided analytical latitude. Data saturation is the point in which concepts can longer be assigned as housing needs satisfiers (Fusch & Ness, 2015). Nearly half of the generated concepts did not satisfy this criterion. Concepts that did not meet both criteria were purged, leaving 72 real estate terms for sales and 69 real estate terms for rentals.

The real estate terms corresponded to locational/typological features, amenities, cost, future development, financial opportunities, social needs, and status. These themes (as with the survey data) were classified under the three housing needs facets: home, house, and dwelling. Thus, by analysing how holistically the market represented these facets, the degree to which the market facilitated needs was measurable. These facets reflected geographic, typological, and socio-economic

119

features (conceptually defined in Table 2.1 and precisely in Subsection 5.4.1). The attribution of the property terms to one of the three facets was guided by the literature on the functionality of housing features. The practical purpose of a term determined its identity, which is highlighted in Tables 5.16-5.17 as structures of activity.

Table 5.16 below shows a selection of some of the sales terms extracted from Leximancer and how they were filtered and are related to the housing facets. The header “count” refers to a word’s frequency in Leximancer. “Relevance %” is a category based on the percentage that a real estate term constitutes from the total.

Table 5.16 Selected Real Estate Data Themes – Sales Network Real Estate Terms Structures of Activity Count Relevance % Brisbane Amenity 403 11 house Work/Play/Type/Size 501 14 development Economic 345 10 price Economic 283 8 unique Economic/Identity 263 7 potential Economic 257 7 sought Economic/Identity 256 7 House popular Amenity/Type 208 6 investors Economic 198 6 future Resale value/ROI 176 5 prime Economic 155 4 sale Cost 148 4 Modern Work/Play/Type/Size 274 8 Walk Amenity 147 4

living Work/Play/Type/Size 132 4 quality Age & Structural Integrity 127 4 water Amenity 796 23 schools Amenity 749 21

Dwelling transport Amenity 671 19 shopping Amenity 541 15 drive Amenity 521 15 secure Privacy/Continuity 484 14 located Amenity 1009 29 distance Amenity 690 20 local Amenity 536 15

master Amenity 321 9 kitchen Work/Play/Type/Size 210 6

Home Land Work/Play/Type/Size 128 4 bedroom Work/Play/Type/Size 125 4 block Type 1677 48 garage Age & Structural Integrity 980 28 Source: Author as derived from Domain.com.au (2018); Realestate.com.au (2018); realestateVIEW.com.au Ltd (2019)

120

Table 5.17 below shows a selection of some of the rental terms extracted from Leximancer and how they were filtered and are related to the housing facets. The real estate terms vary since the rental listings relate to a different housing market segment, but the structures of activity are comparable since the overarching housing facets are common to all individuals.

Table 5.17 Selected Real Estate Data Themes – Rental Network Real Estate Terms Structures of Activity Count Relevance % CBD Amenity 326 6 Brisbane Amenity 215 4 Brisbane CBD Amenity 180 3 tenants Economic 496 10 House city Amenity 429 8 rent Economic 291 6 Pets Identity 583 11 Bbq Amenity 234 5 area Amenity 3782 73

living Work/Play/Type/Size 2991 58 home Work/Play/Type/Size 2712 53 dining Amenity 1739 34 shops Amenity 1625 32 Dwelling access Amenity 1553 30 walk Amenity 1534 30 transport Amenity 1256 24 family Identity 1227 24 Located Amenity 521 10 Bedrooms Work/Play/Type/Size 378 7 kitchen Work/Play/Type/Size 3309 64 bathroom Work/Play/Type/Size 2139 41

Home garage Work/Play/Type/Size 1840 36 room Work/Play/Type/Size 1742 34 space Work/Play/Type/Size 1617 31 Source: Author as derived from Domain.com.au (2018); Realestate.com.au (2018); realestateVIEW.com.au Ltd (2019)

Following the division 'of the data into the dwelling, home and house facets, the next methodological step involved the construction of a contingency table. This table was created by adding the 72 sales terms and the 69 rental terms as headers to the right of their respective documents, containing the property descriptions and other headers. The following Excel formula was input into every row under each real estate term: =IFERROR(IF(SEARCH("*Real Estate Term*",$N2,1),"1"),"0")

N2 represents the position of the property description; therefore, for the Sales list containing 4696 properties, the series would continue as $N3, $N4...$N4697, similarly, for the rental property series with 3382 properties, the series could continue

121

as $N3, $N4…$N3883. The IFERROR in this formula is a conditional statement meaning that if the *Real Estate Term* is found in the property description field then “1” is assigned, otherwise “0” is assigned. The search terms were carefully written to account for minor differences in their form, i.e. both ‘located’ and ‘location’ would register as the same real estate term. In this binary system, if a word existed in the property description, it was denoted as a 1; otherwise, it was denoted as a 0. This resulted in an association profile wherein search terms that coexisted with other terms had a value of 1. After the contingency matrix was populated, the file was saved as a CSV with only the real estate headers and associated binary values (1s and 0s).

Table 5.18 Model of a Contingency Matrix Property Description Brisbane Gold Coast Brisbane CBD CBD House Development Price Brisbane CBD Unit 1 0 1 1 0 0 0 Market Prices 0 0 0 0 0 0 1 CBD Parking 0 0 0 1 0 0 0 Sale Price 0 0 0 0 0 0 1 Granny Flat 0 0 0 0 0 0 0 Final Sale Price 0 0 0 0 0 0 1 Gold Coast House 0 1 0 0 1 0 0

Table 5.18 above is a simplified contingency matrix showing how the data are represented (only in 1s and 0s). The complete contingency matrices for the rental and sales networks and their respective subgraphs (see Table 5.20) were later multiplied as adjacency matrices, like the matrix shown in Table 5.19, later in this subsection.

The contingency table was transformed into an adjacency matrix in the programming software R using the following matrix formula x %*% y to multiply each row with the value it co-occurs with. This is described in the following quote

Multiplies two matrices if they are conformable. If one argument is a vector, it will be promoted to either a row or column matrix to make the two arguments conformable. If both are vectors of the same length, it will return the inner product (as a matrix)” (Eidgenössische Technische Hochschule Zürich, 2018).

The relevant code in the R software is: matrix <- t(CSV)%*%as.matrix(CSV)

CSV is an assignment operator (reference) to the Excel .csv file of the contingency matrix. Since the values in the contingency matrix were either 1s or 0s, the co-occurrence of two terms is signified if they both have a value of 1 in the same

122 row; when the matrix was multiplied, the value shared between two cells remained either a 1 or 0. The value of the multiplication between two cells of real estate terms (columns) is the shared value of their interaction. All real estate interactions were multiplied resulting in a symmetrical adjacency matrix, with the real estate headers positioned in both the x (rows) and y (columns) of the spreadsheet. The number of interactions between a term such as ‘Brisbane and another term such as ‘CBD’ is a shared value and was written twice, whether the relationship was between row (Brisbane) and column (CBD) or between column (CBD) and row (Brisbane). The exception being for the interaction between two same terms (found in the diagonal of the matrix), which were only represented as one value. Since the diagonal values (known as self-loops) were self-referencing, they were omitted in the analysis.

While this approach would be sufficient for the basic construction of network centrality, it was also important to emphasise the relative significance of the interactions between concepts (co-occurrence). This interaction was accentuated by normalising the occurrence of concepts at the individual property level based on the total number of occurrences for that concept. In other words, if there were five terms: a, b, c, d, and e that all co-occurred with e, the normalised weight for a to e would be based on its weight divided by the sum of a to e, b to e, c to e, d to e, and e to e. Thus, the relative significance of a to e is shown, not its absolute centrality. This method is used instead of a similarity or distance measure such as cosine similarity. The implementation of cosine similarity would factor the relative difference between the cosine similarity of the total occurrences of two concepts, and the co-occurrence between those two concepts and all other concepts.

Alternatively, the formula below is then used to normalise the matrix by column. Each cell within a given real estate term (column) is divided by the sum of that column, resulting in a normalised value between 0 and 1.

matrix2 <- matrix / t(replicate(nrow(matrix), colSums(matrix)))

This code ensured that the edges revealed the relative weight of the relationships within each term, since edges were weighted by their relative links. Thus, the matrix was no longer symmetrical. The total sum of all rows was equal to the total sum of all columns, but the individual totals across each vertex varied since some terms co-occurred more frequently; some rows had a value greater than 1, and some

123

terms had a value less than 1 (Table 5.19 below). This normalisation resolved the bias of vertices (terms) that had a frequent repetition of the same terms in the property description fields but low levels of co-occurrence with other terms. For instance, the term ‘development’ has multiple instances in the same property excerpt, and thus it has one of the highest normalised values in its own interaction (development, development) with a value of 0.40, but its level of co-occurrence (interaction) with other terms is less than average, resulting in the lowest SUM ROWS in the table.

Table 5.19 Example of a Normalised Adjacency Matrix Gold Brisbane SUM Brisbane CBD House Development Price Coast CBD Rows Brisbane 0.33 0.27 0.24 0.23 0.16 0.19 0.13 1.55 Gold 0.16 0.30 0.11 0.09 0.09 0.11 0.08 0.95 Coast Brisbane 0.10 0.08 0.24 0.15 0.05 0.03 0.04 0.68 CBD CBD 0.15 0.10 0.24 0.31 0.09 0.07 0.10 1.06 House 0.13 0.12 0.11 0.12 0.46 0.15 0.14 1.23 Development 0.06 0.06 0.02 0.04 0.06 0.40 0.03 0.67 Price 0.07 0.07 0.05 0.07 0.09 0.05 0.47 0.87 SUM Cols 1.00 1.00 1.00 1.00 1.00 1.00 1.00 7.00

As explained above, the social networks were initially undirected since the co- occurrence of words is a mutual relationship. This meant that the ‘a to b’ and ‘b to a’ relationship between edges were one and the same. This was represented in the initial adjacency matrix, where the values from row to column were the same as the values from column to row. However, since the networks were normalised, the networks became asymmetrical, and the values from column to row differed from the values from row to column. The sum of the columns equated to 1 since it is where the normalisation occurred; the sum of the columns represented the division of a term by every property term relating to that property term. The property terms that constituted a larger share of the relations were consistently weighted higher as a fraction, while the terms with smaller shares of the relations in various columns were consistently weighted lower. Thus, highly weighted terms exceeded the value of 1 across the rows, while lower weighted terms had values that were less than 1 across the rows.

Due to this asymmetricity, it is important to consider how the network edges were represented in the igraph software, which will also be documented in the form of

124 coding (Appendices 26-27). The representation was shaped by two realities. The first consideration is that the igraph software did not sum the reciprocal edges if the network was graphed as a directed network; only the highest weighted edge in the dyad was graphed. Therefore, to represent the complete value of the dyad, the network was initially represented as an undirected network, which summed the asymmetric edges. Afterwards, the matrix was reconstituted as a directed network. This reconstitution was important because the normalisation process established a directedness between the edges. The value of a term ‘a’ as a percentage of all edges relating to the term ‘b’ is not the same as the value of a term ‘b’ as a percentage of all edges relating to the term a. This is because the composition of the vertex ‘a’ is likely to be different from the composition of the term ‘b’. Hence, ‘a’ can be more important within the context of b than b is important within the context of a. This distinction explains why the direction of edges mattered in the networks and subgraphs.

Regarding the how: in the network matrix, every property term had a column header; this column header refers to the out-degree (outgoing edges) of each term. Each property term also had a row header; this row header refers to the in-degree of every edge (the incoming edges). The network edges were normalised by finding the weight (co-occurrence) between a property term (column) and its associated pair in the row and dividing that value by the total of every occurrence of the primary term in the column. Thus, normalisation was by out-degree. The normalised edge value was the fraction of every edge originating from a property term, as a percentage of all the edges originating from that same term. The in-degree, on the other hand, was a by- product of this same normalisation process. These individual weights (in-degrees and out-degrees) were summed (excluding the diagonals), and the total was the strength of a single vertex (property term). This process clarified the direction of the edges since out-degrees reveal the key terms and should not be conflated with in-degrees.

5.4.3 Real Estate Data Filtering Following the collection and sampling of property data, the two networks (sales and rentals) were studied separately to highlight the differences in tenure between renters and homeowners. These two networks were further split into eight subgraphs based on geography and typology, reflecting the different locations and housing types of the Brisbane GCCSA. Typology was separated into the house, apartment, and land subgraphs. The network hierarchy is shown in Table 5.20 below. The sales and rental

125

classifications formed the overarching networks. The sales network was firstly subdivided into three typological subgraphs (house, apartment, and land) based on the specific property types available in the market. These subgraphs were created by reviewing real estate websites to understand how Brisbane GCCSA house types were classified. Table 5.20 shows the sub-components that constitute the definitions. The rental network was similarly divided into typological subgraphs. However, while the house and apartment subgraphs corresponded to both the sales and rental network, the land subgraph only corresponded to the sales network.

Table 5.20 Division of Sales and Rental Typologies Summary Network Hierarchy Networks (2) Sales (1) & Rentals (1) Sales Only Tenure For Sale, For Rent Subgraphs House Apartment Land (5) (2) (2) (1) House Apartment Land Development Unit Acreage Duplex Townhouse Typology Semi-Detached Block of Flats Blank Villa Terrace Rural Studio Subgraphs No Geographical Subgraphs Inner Middle Outer (3) for Rent (1) (1) (1) Brisbane - North, Moreton Bay Geography - Brisbane Brisbane - East, - North, Statistical Not Applicable Inner Brisbane - West, Ipswich, Area Level 4 City Brisbane - South, Logan - Moreton Bay - South Beaudesert Source: (Domain.com.au, 2018 ; Realestate.com.au, 2018 ; realestateVIEW.com.au Ltd, 2019)

Geographically, the housing market was split into three regions: Inner, Middle and Outer (Figure 5.7 on the next page). Additionally, the overarching sales network is also divided into three geographical regions: Inner, Middle and Outer. These regions are not computed using the same definition from Sections 3.7-3.8. That definition was specific to the Brisbane metropolitan area, whereas this definition was created to relate to the Brisbane GCCSA. This geographic separation creates three subgraphs which only correspond with the sales network. Consequently, two networks and eight subgraphs are analysed in Chapter 7: the sales and rental networks, five typological subgraphs and three geographical subgraphs. These networks and subgraphs are all

126

represented in 10 separate plots using the traditional graph theory/SNA network configuration explained in Subsection 5.4.1.

Figure 5.7 Brisbane GCCSA SA4 Geographic Rings 5.4.4 Analysing the Social Network Data through Centrality Measures and Community Detection There are two ways in which the SNA data is analysed. The first is based on centrality measures. Centrality measures are used to represent the connections between vertices (property terms) based on their importance in the network. There many forms of centrality such as degree centrality (how connected a vertex is), weighted degree centrality (WDC): how many edges a vertex has (known informally as prominence in this thesis), betweenness (frequently occurring in-between the shortest paths between two vertices), closeness (shortest average path length of a vertices’ edges) and eigenvector centrality (the importance of a vertices’ edges). The second way in which the SNA data is analysed is using community detection. While centrality measures

127

emphasise the relations between a single vertex and all other vertices, communities in SNA emphasise the shared properties among vertices. This is useful in determining which vertices have similar functionalities in a network (Fortunato, 2010). Since networks are based on flows of information, interactions, social relations or similarities (Stephen P Borgatti et al., 2009), it is then expected that there will be distinct clusters of information, relationships and attributes. In this thesis, such relationships are important because shared values (any form of value) underpin the market economy and social unity.

The various centrality measures of the SNA are explained in this subsection. The first measure is WDC, known as graph strength in igraph. WDC weighs the strength of connections, rather than the number of connections, which is the case for degree centrality. Consequently, this measure is used in the analysis rather than degree centrality. Although the more widely known measure is degree centrality, that metric simply represents how connected a vertex is to other vertices by measuring the number of connections (edges). In this thesis, the relationship is uniform across all vertices since essentially all property terms are connected. Property terms with high degree centrality are those that occur in numerous contexts. However, there is virtually no distinction in this thesis, since all property terms occur in all contexts. WDC, on the other hand, highlights the incidence of a theme (how frequently the property term occurs in each context), which communicates that a relationship does not simply exist but is significant. Since the matrix which is responsible for the creation of the network is normalised within each vertex, there is no bias for property terms that exhibit a high frequency of usage but only in relation to a few terms. The networks in this thesis are densely connected with edge densities ranging from 89.7% to 100%. Hence, the degree of the graphs is uniform, and do not convey any significant information. A weighted degree must be considered. WDC resolves this problem using the sum of all the adjacent edge weights for a given vertex (Csardi, 2015b).

One of the limitations of measures such as degree centrality and even WDC is that they emphasise the sum of connections or weights. Conversely, closeness centrality is derived from the average distance (weight) and is a measure of proximity. In igraph, closeness centrality refers to the “inverse of the average length of the shortest paths” (Csardi, 2015a). The length of the paths (network distance) is computed using the weights of the paths. Igraph calculates the closeness based on

128 weights, which are then used as costs, in determining how ‘costly’ it is for vertices to be bridged. Ordinarily, this would mean that edges with high weights would be represented as having a lesser closeness than edges with lower weights. This explains why the inverse of closeness is used in igraph. Closeness represents the average distance between a vertex and all the other vertices in the network (University of Chieti- Pescara, 2014). This is useful for determining how reachable the vertex is to other parts of the network. Vertices with low degree centralities and low connectedness can still have a high closeness centrality if it is near a vertex with better connections. This measure is useful for outlining which property themes are likely to have a high impact when occurring in the text, due to a high capacity to efficiently transmit information (Griffin, 2017). Therefore, this measure favours property terms that exhibit a consistently high weight across all its edges.

Akin to closeness centrality, eigenvector centrality focuses on the relative strength of a vertex. Eigenvector centrality was determined by assigning the value of centrality based on the weight of adjacent vertices rather than simply the number of connections (Csardi, 2015b ; Griffin, 2017). Thus, the eigenvector relationship was useful for highlighting which vertices were connected to other highly weighted terms, and which had edges with terms of low centrality.

Complementing the centrality measures were community measures which represented the relationships of several vertices (how vertices were grouped), versus the relationship between individual vertices. The most fundamental community measure is modularity. Modularity determines whether there are more edges within a group than the edges that exist outside of that group (Comber et al., 2012 ; University of Chieti-Pescara, 2014). The density of a potential cluster (known as a subgraph) is compared with the density of the existing network, based on comparable structural relationships to achieve modularity. This means that a random graph with the same communities but with random edges between its vertices is subtracted from the density-derived subgraph. The idea is to emphasise the strength within each community relative to the edges outside of that subgraph (Bearman & Hoffman, 2017).

The most effective means of assessing the quality of community measures is by comparing the modularity between their clusters (Blondel et al., 2008). Since modularity establishes if the edges and property terms within a group/cluster are more

129 comparable compared to the rest of the network, communities based on modularity reveal property terms that are closely related. This is an automatic and inductive means of identifying property classifications. The benefit of this process is that a comparison can be made between the manually coded housing facets (explained in the previous subsections), which represent what individuals universally want and the automatically coded communities which represent housing outcomes.

The two community measures that were used in this thesis were based on network optimisation. Other community detection methods rely on divisive or agglomerative algorithms. The divisive approach is a top-down sequence wherein inter-community links within the existing community are found and then severed, while for network optimised algorithms, similar edges are used for community formation. While the effectiveness of divisive and agglomerative community detection can be measured through modularity, for optimisation algorithms, modularity (when used as the objective function) is crucial in the initial formation of communities since the focus is on the positive gain that can be acquired through partitioning (Blondel et al., 2008).

One such community measure that utilises modularity is the multilevel community (MLC) proposed by Blondel et al. (2008). MLC is a form of community detection wherein vertices belong to one community each and are then progressively assigned to communities based on which groups yield the greatest gain in modularity. Once vertices cannot yield a positive gain in modularity, the communities are treated as new vertices and are then reassigned once again. Thus, the process is iterative (multi-level) and concludes once the modularity in the network can no longer be increased. This community detection process was useful for outlining the boundaries of similar property terms. When running MLC, communities are liable to change during each iteration; however, a seed ensured that the same communities were always generated. The objective of the SNA (how balanced housing needs satisfiers are) was met since the algorithm established groups based on vertices of similar density. Vertices (property terms) of similar density were related since they had similar structures. These structures reflected distinct types of housing needs. Thus, minimal variation in the distinct communities signified that housing needs were imbalanced.

Like MLC, the Fastgreedy community (FGC) is traditionally employed as a mutually exclusive tool; thus, vertices do not belong to more than one community

130

(Valejo, Valverde-Rebaza, & de Andrade Lopes, 2014). This community detection model was pioneered by Clauset, Newman, and Moore (2004) and only handles directed edges, though weighted edges are allowed (Harenberg et al., 2014). FGC is formed by placing every vertex in the network in separate groups and then forming pairs based on which edges yield the most increase in modularity. Thus, edges are formed based on which vertices have the most in common. This process is continued until there is no longer an increase in modularity, or until the smallest decrease in modularity is no longer feasible (University of Chieti-Pescara, 2014). FGC is one of the traditional SNA algorithms, developed in the mid-2000s, and lacks the state-of- the-art input parameters of the more advanced algorithms developed in the past 10 years. Most of the state-of-the-art community detection models also facilitate directed edges, and many allow for overlapping vertices, which FGC is unable to process.

The multilevel algorithm will reveal the distinct segments of the real estate networks/subgraphs (Section 7.3), while the supplementary Fastgreedy algorithm (Appendix 25) corroborates the consistency in the grouping of property terms, notwithstanding the algorithmic differences.

5.4.5 Understanding Centrality and Community Measures through Density The objective of the centrality measures and community detection algorithms were to outline the connections (edges) between vertices as well as their distinct communities. These measures revealed how well the networks and subgraphs facilitated the information flows between terms. Networks and subgraphs with a balance of home, house, and dwelling terms were considered as adequately satisfying housing needs.

However, comparisons across different networks are difficult without addressing network density. Network density is mostly analysed based on the number of possible edges divided by the total number of edges. This calculation shows that density is inversely related to the scale of networks (De Nooy, Mrvar, & Batagelj, 2018). For example, in social networks, even though the number of individuals within each network can vary substantially, the number of people that everyone can connect to has a natural limit. Thus, larger networks tend to be less dense. Therefore, this SNA analysis addresses density by normalising weighted edges. However, network density is used to adjust the cut-off thresholds of visible edges since in networks with a greater density, the relative importance of edges is greater (see Appendix 28).

131

There are two means of addressing the differences in network density. The first is that the relevance of the property terms was determined locally in Leximancer. The terms that appeared the most within six sentence blocks of text were the property terms (vertices) that were analysed. Vertices were not assigned based on the overall size of the network but based on the contextual relevance of every individual block of text. The cumulative relevance of these terms then led to a range of property terms that were filtered based on how relevant, unique, or unambiguous, the terms were.

Additionally, density was also addressed through normalisation (except for the clustering coefficient analysis, clarified later in this subsection). This normalisation was performed by dividing the sum of each property term within a column by the total sum of all edges in that column (explained with Table 5.19). Thus, every edge weight (WDC) was produced by the out-degrees of a vertex (the edges from a property term). The in-degrees (edges coming into a property term) were not discounted, but the in- degree of one term is the out-degree of another term. Thus, the row-column relationship in the adjacency matrix signifies the out-degrees of property terms, and it is at this level that terms were normalised. The column-row relationship is incidental.

Therefore, this thesis does not analyse density through the number of edges as a proportion of the maximum number of possible edges in the subgraphs as defined by De Nooy et al. (2018). This is due to the bias between network sizes, but also because complete networks are networks with maximum density, i.e. all the possible edges occur, and the networks and subgraphs that are presented in this thesis have a maximum or near-maximum density. Such networks reveal very little of significance.

One of the more effective means of identifying network density is via the clustering coefficient (CLC) algorithm. CLC is the probability of the vertices within a network being clustered (Csardi, 2015d). This coefficient is either represented as a measure of network density (local clustering) or by considering the density of triplets (global clustering). A triplet is formed between three vertices (property terms). Two closed edges are an open triplet, i.e. if ‘home’ were connected to ‘house’, and ‘house’ was also connected to ‘bedroom’, but ‘home’ was not connected to ‘bedroom’. A closed triplet would be if ‘home’ and ‘house’ were connected and they were both connected to another term, such as ‘bedroom’. This is the definition of clustering used in this thesis, tested using two forms of global clustering: clustering coefficient by graph

132

or by edgelist (Appendix 27). The Csardi approach processes the network graph (multiplied matrix), while the Opahl and Panzarasa approach processes edgelists. In both methods, self-loops are disregarded or deleted. Self-loops are the edges between a property term and itself (the diagonal of an adjacency matrix). This is because a self- loop is self-referencing and does not reveal how multiple terms are interconnected.

Clustering predicted two causal tendencies, introduced in Subsection 7.2.1, and tested in Subsections 7.2.2-7.2.5. The first is the territorial tendency, which suggests that networks with a low overall clustering coefficient will have a more balanced composition of the housing needs facets. The financial tendency suggests that networks have a low overall clustering coefficient because density is concentrated among focal terms. This selectiveness reflects the tendency towards focal terms that communicate a precise message, aligned with financial terms.

Although the CLC used in this thesis did not use normalised edge weights, this was unproblematic since the relevance of the property terms was based on isolated excerpts of text in Leximancer, and the co-occurrence of the triplets was established among individual property listings. Hence, the CLC relationship was based on the fraction of closed triplets within isolated excerpts and not the entire network. The application of the clustering coefficient is shown in Appendix 27. Edge weights were not normalised because clustering does not vary significantly between networks if properties such as degrees are similar. Thus, the graph method of Csardi (2015d) was unfitting for this thesis. Rather than integrating the weighted degree of edges, Csardi’s method used the degrees of property terms as the weight. This is problematic because this thesis’ edge densities were either 1 (maximum value) or almost 1. Transitivity computed using this method would be negligible unless arbitrary cut-off weights were applied to the lower weighted vertices (Opsahl & Panzarasa, 2009). The approach by Opsahl and Panzarasa (2009) used igraph graph objects as the input (graph objects are the result of the matrix that has been multiplied across rows and columns). Thus, this form of CLC studied the relationships between the sum of all co-occurrences.

The formula proposed by Opsahl and Panzarasa (2009) was strategically employed to inject weighted degrees into the analysis. As with the Csardi formula, it was found that the weights being computed were of degrees; however, the key difference afforded by the Opsahl and Panzarasa approach is that an edgelist was

133

used as the input. When the contingency tables (rather than the adjacency matrices) of the individual subgraphs were retrieved within the formula, they formed an edgelist between each individual property term. Therefore, the clustering coefficient analysed the co-occurrences at the property level, not at the network level. This weighting resolved the weighting/edge density problem and revealed significant variation between the subgraphs. That variation refers to the probability of closed triplets within each property sampled in the Brisbane GCCSA occurring as a fraction of all triplets. Networks with a higher clustering coefficient have property terms that more frequently relate to one another and a mutual property term.

5.4.6 How Network Centralisation is Defined in this Thesis The centrality and density measures discussed thus far reveal the significance of individual vertices within networks (point centrality). However, entire networks can display centrality; this is known as centralisation. Centralisation is a measure of the extent to which the network is distributed among key central vertices (Stephen P. Borgatti, 2015). Centralisation can be computed with a number of measures such as degree, eigenvector, betweenness and closeness (Csardi, 2015a). These measures seek to identify the extent to which the central vertices vary from the rest of the network. Centralisation within the context of this thesis was conceived as the distribution of edge weights among focal property terms (see Subsection 7.2.1), which is reflected in the weighted degree positions of the property terms. A centralised network has few edges that are considerably more highly weighted than the rest of the network. While a decentralised network is more evenly distributed. Therefore, in this thesis, centrality measures identified the core or central property attributes of networks, while centralisation and the clustering coefficient described broader relationships such as density and distribution.

5.4.7 How the Centrality and Community Measures are Implemented The primary centrality measure of this thesis (weighted degree centrality) and the primary community measure (multilevel algorithm) were created within R’s igraph package. Appendix 26 shows a description of the line-by-line coding used in the R program to create the SNA network and community plots. The sub-headings denote whether the sub-component was used for the centrality measure, the community measure, or both; the centrality and community measures share much of the same coding. The main difference is whether vertices were identified according to their

134

community membership or based only on the layout configuration. While the networks were structured entirely by the Fruchterman Reingold layout and the WDC measure, which was used for the vertices, community membership was determined by the dense connections among similar vertices or how those connections related to the entire network. Thus, the objective of community membership was to partition the vertices into clusters of terms with similar meaning. Contrariwise, the Fruchterman Reingold layout assigned forces to the edges of property terms. These edges can be considered springs which propel property terms towards the centre of the network or outwardly (Yuliana, Sukirman, & Sujalwo, 2017). This layout primarily emphasised the strength of the property terms with the highest centrality; the secondary objective ensured that vertices did not overlap. Igraph has other layouts such as the Davidson Harel and Kamada Kawai layouts, but those were found to be more asymmetric and exhibited significant vertex overlap.

Consequently, the network graphs and community detection models were designed to share much of the same qualities, such as the centrality of the highest weighted property terms. These plots used the Fruchterman Reingold layout and had a radial appearance, whereby the top centrality scores were central, and the lowest scores were peripheral. Thus, the placement of the vertices was not arbitrary, though the spacing between vertices was adjusted for uniformity. The arrows in the network plots also illustrated the flow of information since the subgraphs were directed. These inherent qualities facilitated an intuitive analysis, although most of the insights into the functionality of the housing market lay in subtle differences across the centrality and community measures, as well as the nuances within each individual network.

Section 5.4 introduced SNA and its measures of analysis, data collection and processing, as well as the functionality of the centrality and community measures. WDC was introduced as the primary centrality measure. Additionally, eigenvector and closeness centrality were also employed in the thesis for comparison. These centrality measures were useful for identifying the property terms with the greatest housing market impact, though the relationship between multiple property terms was better explained through clustering, centralisation, and community algorithms.

135

CHAPTER 6: SOCIAL AND ECONOMIC DETERMINANTS OF HOUSEHOLDS’ HOUSING PROGRESSION OVER THE LIFE-COURSE This chapter provides evidence for research questions 1 to 3 of the thesis, based on data from the Brisbane GCCSA Housing Needs Survey (HNS). The HNS sought to reveal how individuals’ housing needs satisfiers are composed, individuals’ housing trajectory, as well as individuals’ attitudes towards the housing market. The satisfiers of individuals’ housing needs are ranked by their mean scores to reveal the housing needs composition. There are 18 needs in total; the scores of these themes are then analysed with K-means clustering, bar graphs, line graphs and matrices to determine individuals’ housing trajectories. Lastly, individuals’ housing attitudes are determined through values/causal coding of the individual needs and external factors that explain housing market instability.

The comprehensiveness of these analyses is possible since the survey spanned 102 unique addresses (see Appendix 2 and Subsection 5.3.1). In total, there were 237 participants with 472 housing histories, which reveal individuals’ housing prioritisation based on the longitudinal (retrospective) housing perspectives of individuals in relation to their life-course circumstances. Understanding these housing circumstances is vital since an upward trajectory is presupposed due to normative beliefs. These beliefs include the housing ladder, which likens residential mobility to career progression, and suggests that individuals in 20th-century Australia and beyond share the same housing progression (Faulkner & Beer, 2011). This progression refers to the transition from renting to buying and then to a larger family home. However, this is a narrow definition of progression, centred on the face-value of residential mobility. Alternatively, preferences based on the comprehensive range of housing needs satisfiers are a better characterisation of progression. Thus, rather than relying on prices which confound needs with external interactions, the Brisbane GCCSA HNS queries internally-derived priorities that represent housing narratives based on, which are the socio-economic determinants of household progression over the life-course.

6.1 Housing Aspirations and the Life-course

This section provides evidence for the social and economic determinants of individuals’ housing progression. This is based on how respondents of the online survey represented their housing choices, within the three classifications of their life- course events, their social-dwelling aspirations, and their financial motives. This will

136

be done to address research question 1: How do individuals prioritise housing needs satisfiers through the life-course?

RQ1 is based on six dwelling profile themes, five demographic profile themes and the 27 Likert scale themes, which were detailed in Table 5.2 of the research design (Section 5.2). Eighteen of the 27 attributes represent individuals’ housing needs, while 9 of the themes represent life-course triggers. These themes allow individuals’ housing histories to be analysed based on how balanced the three housing facets are and based on age/occupational cohorts. The demographic and dwelling profile themes also ensure that individuals’ histories relate to real-world contexts which can be used as data points for the housing trajectories. The 27 themes are designed to encompass the breadth of diverse housing experiences since they are proxy questions. Otherwise, the complexity of housing experiences would be untenable. Aspirations can be deduced from these themes by ascertaining whether the core dwelling attributes (such as house type and size) are prioritised or whether life-course factors are overbearing.

6.1.1 The Composition of Individuals’ Housing Needs Satisfiers and Life-course Factors The importance of this subsection is to outline the composition of individuals’ housing needs satisfiers and life-course triggers, as well as their housing trajectories. This composition is represented with a series of bar graphs based on the relevance of the 27 social-dwelling, financial and life-course factors which were derived from the HNS.

Figure 6.1 Most Relevant Aspirational Themes - Entire Cohort (Mean of Responses)

137

Figure 6.1 on the previous page shows how the mean number of responses varies across all 27 themes. Social-dwelling attributes (in red) were the most relevant for the most number of participants, financial attributes (in green) were the second most applicable, while life-course attributes (in blue) were the least applicable. Due to this disparity, a Zuccolotto adjusted mean (ZAM) was used for Figures 6.2-6.4

Figure 6.2 Most Relevant Aspirational Themes - Entire Cohort (ZAM)

Figure 6.2 above shows the mean of the entire cohort’s priorities after applying the Zuccolotto formula, which addresses the low response bias (explained in Chapter 5). Following the adjustment, the range changes from 1-10 to 0.22-2.41. The formula is useful for showing the relative changes in rank. The most relevant attributes were the social-dwelling attributes, followed by the financial facets. The life-course factors are the least relevant. This is because social-dwelling attributes such as ‘house/community suits activities’, ‘sufficient home size’ and ‘house types’ are factors that are more reflective of the desires of the broad cohort, while the financial attributes are less relevant since not all financial instruments are relevant for individuals. For example, people found conventional financial attributes such as buying into tenure, purchase/rental costs and appreciating value over time as more relevant, relative to more precise strategies such as tax deductions and rental income. Similarly, the variability in the relevance of various life-course attributes is because individuals or households do not typically experience factors such as leaving home, change in employment, and change in health consistently through the life-course.

138

Figure 6.3 Most Relevant Aspirational Themes - Owner Cohort (ZAM)

The Zuccolotto formula is also applied to the owner cohort to reveal the priorities specific to homeowner occupiers (HOCs) and investors (Figure 6.3 above). Following the adjustment, the range changes from 1-10 to 0.1-1.9. A key difference from the overall cohort is that the importance of ‘flexible tenure’ and return on investment (ROI) is even greater for the owner cohort. However, the core social- dwelling remain just as significant. The relative low prioritisation of life-course factors is generally attributable to individuals being exposed to a few life-course issues at once. However, the relative insignificance of life-course factors relative to the renter cohort can be attributed to the financial and regulatory advantages of the owner cohort.

Ownership affords greater property rights since homeowners are a wealthier cohort (on average), and the housing market is orientated towards homeownership. While there are more low-income owners than low-income renters in absolute terms, the proportion of low-income renters as a percentage of all renters is greater than the number of low-income owners as a percentage of all owners (Australian Bureau of Statistics, 2016d). This disparity effectively means that members of the owner cohort are wealthier stakeholders in Australia’s non-financial sector and are better positioned to accumulate social-dwelling attributes such as desirable neighbourhood quality and house types. The low-income calculations are based on the estimated number of low- income households and the estimated number of rental and owner households

139

provided by the Australian Bureau of Statistics (2016d); 36% of owner households were low-income, while 48% of renter households were low-income.

Figure 6.4 Most Relevant Aspirational Themes - Renter Cohort (ZAM)

The Zuccolotto adjustment was also applied to the renter cohort; the range changes from 1-10 to 0.11-1.65 (Figure 6.4 above). The renter cohort has lower relevance scores for financial attributes, relative to the owner cohort. This is unsurprising as there is a greater number of households experiencing housing stress than among the owner cohort. While ‘attractive cost’ is a significant priority for this cohort, specialised instruments such as ‘deductions’ and ‘rental income’ are not. Moreover, renters have a relatively lesser prioritisation of tenure, which is one of the significant facets of the Great Australian Dream, as it involves ownership and democratic liberties. The lower prioritisation signifies that renters find tenure flexibility to be less advantageous compared to owners’ prioritisation of the permanency of tenure. Furthermore, the tenurial disadvantage of renters is also evidenced by their low prioritisation of ‘diverse needs and adaptability’. This low prioritisation reflects the perception among renters that they cannot adapt their housing to suit their needs. Overall, these differences reflect the regulatory disadvantage of the rental cohort who prioritise life-course factors more than the owner cohort.

Although prioritising life-course factors does not inherently suggest a negative outlook, there are a few themes within this category that signify vulnerability. This includes the ‘forced move/buy’ category, which consists of evictions, and the

140

destruction of property. ‘Career-retirement needs’ is another factor that is potentially suggestive of vulnerability. As will be explained through the open-ended passages from the survey, themes such as ‘education/training’ and ‘career/retirement needs’ reflect changes in lifestyle, status, and needs, as well as having a locational dimension. The latter, which involves moving as education or employment circumstances change, is the element of volatility. Thus, in the rental cohort, the greater relevance of locational triggers, as well as forced moves, is suggestive of greater vulnerability, relative to the owner cohort.

There are, however, some consistencies across both tenure types. ‘Sufficient home size’ and ‘house type’ are a constant among all cohorts. This is because the suitability and quality of the house, community and neighbourhood are indispensable.

6.1.2 The Housing Trajectories of Individuals in the Brisbane GCCSA This subsection will uncover individuals’ housing trajectories by comparing how the 27 social-dwelling, financial and life-course factors (derived from the Likert Scale) were prioritised. These factors will be studied relative to age and the number of housing transactions (total houses owned/rented). These patterns will be shown (without the Zuccolotto formula) with scatterplots, stacked bar graphs, line graphs and matrices. The Zuccolotto formula forms a bias wherein each factor’s range is overly dependent on its response rate. This is not an issue when the total responses (Subsection 6.1.1) are used but is unviable when age and mobility are used to compare progression.

The first relationship uses K-means clustering (KMC) to show how the 27 themes are clustered based on priorities and age, as well as the degree to which the themes diverge or converge. KMC reveals intuitive relationships that are not otherwise visible. The scatterplots in Figures 6.5-6.6 (Pages 143-145) show how 27 themes within the owner/renter cohorts are ranked in descending order from the attribute with the highest relevance mean score across all five age brackets to the attribute with the lowest relevance scores. Outliers with 0 relevance scores were removed, as well as the entire 70 and over cohort (due to low representation). The plot sizes correspond with the cohorts’ age brackets, while the x-axis measures the relevance for each cohort from 0, the lowest to 10, the highest. The central line in the chart is the ‘average’ line (mean score of all plots). The figures can be read horizontally (how ages cluster) or vertically (how the themes within the dotted groups converge or diverge).

141

As explained in Subsection 5.3.3, data points (relevance scores) are assigned to the nearest centroid based on similarity (the shortest distance between scores and the centroid). After the creation of these clusters, groups can then be identified based on convergence, having parallel themes, or having divergent themes. Convergent themes are prioritised by the five age cohorts similarly, i.e. the circle markers are confined to no more than two clusters. Parallel themes are prioritised similarly by the five age cohorts but are confined to three clusters. Lastly, divergent themes are prioritised dissimilarly by the five age cohorts and are confined to four clusters.

The convergence of prioritisation would signal that financial and social-dwelling factors (which represent needs) are consistently prioritised, which communicates agency and capital across different age groups. However, some degree of divergence is expected among the life-course factors, which represent constraints and opportunities that emerge unpredictably or during specific intervals.

Therefore, the relationship between opportunities/constraints and social- dwelling/financial strategies will be explored. This association is important because as mentioned in Section 2.2, aspirations represent the intersection between housing needs and external constraints and opportunities. If individuals are unable to prioritise solutions that can transform their life-course constraints into opportunities, then their housing experiences will suffer. Consequently, a cohort comparison between renters and homeowners will be explored below, to assess if there are differences in the prioritisation of needs and life-course events.

Firstly, the owner cohort is analysed using Figure 6.5 on the next page. The figure shows that there are three different groups across the four priority clusters. The convergent group is anchored by the core social-dwelling attributes. These attributes have a positive effect on housing experiences since they relate to appropriate house sizes, types, and structure. Moreover, tenure is also highly prioritised, as well as housing costs. Thus, the basic elements of housing stability are shown in this primarily high priority cluster. This stability is even more noteworthy since all age cohorts converge within this group. Only three themes within this group have an average priority: the 18-29-year-old mean for ‘attractive cost’ and ‘structural integrity’, the 18- 29-year-old, and 30-39-year-old mean for ‘few restrictions on privacy/regulations’.

142

Figure 6.5 K-means Clustering of the 27 Variables into 4 Clusters and 3 Groups: Showing how Owner Themes Change According to Age and which Themes are Grouped Together by the Mean Score

Similarly, the owner cohort’s parallel group is uniform across the five age groups. Except for four outliers, this group is also confined to only two clusters. The outliers are the 60-69-year-old mean for ‘ROI renting/capital’, the 18-29-year-old for ‘social capital’, and the 18-29-year-old mean, and 60-69-year-old mean for the ‘rental income’ factor. This group contains a balance of life-course, financial, and social- dwelling factors. The most divergent factors within this group are the financial factors of ‘ROI renting/capital’ and ‘rental income’, while the most convergent is the social- dwelling factor of ‘social capital’. Owing to its balance of housing needs, constraints, and opportunities, this group contains factors with an average prioritisation.

Lastly, the divergent group contains the lowest prioritisation within the owner cohort, which is attributable to the prevalence of life-course factors. This is because life-course factors are inherently age-specific events. Thus, factors such as ‘leaving (the) childhood home’ and ‘separation from family’ are very divergent. Conversely, the financial factor ‘use for other investments’, and the life-course factors of

143

‘career/retirement needs’, and ‘personal/social relationships’ are the most convergent within this group. The latter two factors are generally applicable. The implication of the general convergence of housing needs across the owner age cohorts is that housing outcomes are normally/highly prioritised, and there is no inherent disadvantage by age. Moreover, since housing needs are prioritised above life-course factors, life- course factors are likely to manifest as opportunities rather than constraints.

Contrariwise, the renter cohort is markedly more divergent. Consequently, there is more variation between potential clusters compared to the sales cohort (as shown in Subsection 5.3.3, up to 7 clusters could have been used; however, four clusters were sufficient). Were 7 clusters chosen, the high priority cluster would begin at a score of 7.1 instead of 6.75. However, this would artificially separate the most compact segment of the network. The minor differences in the ranges of the renter and owner K-means reflect the fact that clusters are calculated based on the relative distances of mean scores within each cohort.

Figure 6.6 K-means Clustering of the 27 Variables into 4 Clusters and 2 Groups: Showing how Rental Themes Change According to Age and which Themes are Grouped Together by the Mean Score

144

For the renter cohort (Figure 6.6 on the previous page) there was almost universally a low affinity among the 27 themes and five age brackets (thus the within- cluster sum of squares would have more variation, compared to the owner cohort). Unlike the owner cohort, the core social-dwelling attributes of the renters are not universally highly prioritised. Instead, only the ‘sufficient home size’ and ‘house/community suits activities’ factors are wholly highly prioritised. However, the renter cohort highly prioritises the ‘career/retirement needs’ and ‘health or safety’ factors, which are prioritised above core social-dwelling factors such as ‘diverse needs and adaptability’, ‘structural integrity’ and ‘few restrictions of privacy/regulations’ respectively. This signifies that some constraint/opportunity events have a greater bearing than the outcomes that could address them.

The unusual burden of life-course prioritisation is also demonstrated by the interspersing of the factors across all four clusters of the cohort. This dispersal is best exemplified in the divergent group, wherein the extent of convergence across the five age cohorts is even more pronounced. With the exception of the ‘separation from family’, education/training’, ‘leaving (the) childhood home’, ‘flexible/permanent tenure’ ‘personal/social relationships’, ‘deductions’, and ‘cultural beliefs’ factors, the other themes are significantly divergent. While the highly divergent themes include the unpredictable life-course event of ‘forced move/buy’ the social-dwelling, factors of ‘personal/community identity’, ‘social capital’, and ‘diverse needs and adaptability’ are inconsistently prioritised across the age cohorts. Therefore, the obscurity of these social-dwelling factors further reveals the burden of the relatively high prioritisation of life-course themes. Though it is somewhat more natural for specialised financial strategies to be polarising, Figure 6.6 also shows that financial factors also have a lower mean score than life-course factors. Thus, precariousness is higher than in the owner cohort, yet innovative means of alleviating uncertainty are sparsely employed.

Figures 6.5-6.6 have shown that all ages of the owner cohort universally prioritise social-dwelling and financial needs, although life-course factors are relatively more divergent. Conversely, the renter cohort is almost entirely fragmented and exhibit vastly different housing priorities. This underscores the instability of renting in the Brisbane GCCSA and a significant hurdle for housing progression. Though the life- course is characterised by different events and normative standards, the basic elements of housing stability are also divergent.

145

As mentioned, the scatterplot can also be interpreted in terms of how prioritisation progresses with age. In the owner cohort, there is generally an increase in the relevance of the 27 themes the older individuals become. In the renter cohort the ‘sufficient home size’, ‘structural integrity’, ‘health & safety’, ‘environmental experience’, and ‘education and training’ factors are the only themes that exhibit a linear progression in housing needs and life-course prioritisation. While these factors are crucial elements for housing stability, no financial factors exhibit a linear progression, and several core social-dwelling and life-course factors, also do not exhibit progression by age. This irregular progression is even more problematic, considering the divergence of priorities by age. This irregularity is significant since tenure is more uncertain among renters. This signifies that precarious life-course events, which trigger housing moves, are not harnessed as opportunities, but rather represent constraints, with few discernible patterns, relatively to the owner cohort.

Conversely, among the owner cohort, linear age progression is present in these factors: ‘environmental experience’, ‘attractive cost’, ‘structural integrity’, ‘house type’, ‘neighbourhood quality’, ‘personal/community identity’, and ‘sufficient home size’. These variables were predominantly associated with neighbourhood/community interactions and social-dwelling factors. For example, the gradual progression in the relevance of environmental, suburban, and urban experiences is because housing experience increases as people age. Conversely, the life-course factors further down in the scatterplot exhibit a weaker correspondence to age, since the life-course questions are not skewed towards any age group. Similarly, financial factors do not exhibit linear age progression due to the specialisation of the strategies. An exception is the ‘deductions’ factor, which is polarised overall but exhibits a clear progression as individuals age. This is since it is a more prevalent financial strategy. The ‘leaving childhood home’ factor is the only theme to exhibit an inverse age progression; the older people get, the more likely they have already left their parental home.

Some of the factors that do not exhibit an increase in prioritisation as individuals age are the ‘personal/social relationships’, and ‘social capital’ themes since these are likely to be more relevant to individuals’ personal outlook and needs than their age. Overall the owner cohort exemplifies housing progression, since, in addition to the compactness between the various age groups’ prioritisation and linear progression, the age groups with the lowest relevance scores are primarily the 18-29

146 cohort, while the cohorts with the highest mean scores are primarily the 50-59, and 60-69 cohorts. This progression is crucial since the Australian Bureau of Statistics (2017b) find that as individuals age, they are less likely to move in general. Figure 6.7 below highlights the relationship between mobility and the relevance of all 27 themes, divided into four residential transaction classes within 1-8 moves. There were four breaks: 1-2, 3-4, 5-6 and 7-8. The range of residential mobility was from 1-30 residential transactions. Individuals with less than one move were ineligible to participate in the survey. Australia’s residential mobility is 5.1 moves per individual, calculated based on an age spread of 17-50 years (Bernard, Forder, Kendig, & Byles, 2017). In this thesis, individuals’ entire rental and ownership career were integrated, resulting in a mean of 7.3 moves per individual and a median of 6 moves per individual.

Figure 6.7 The Relationship between Mobility (1-8 Transactions) and the Relevance of the 27 Themes

There was a weak relationship between residential transactions and most of the themes. Thus, mobility is largely a poor index for assessing how housing needs are accumulated since mobility reflects diverse housing trajectories. Wealthy individuals may leverage the flexibility of housing tenure, while financially stressed

147 individuals may be prone to lateral or even regressive housing moves. Additionally, some individuals may move more by virtue of having been in the housing market for longer. Only factors that directly benefit from increased or decreased housing moves should be evaluated with residential mobility. For example, ‘career/retirement needs’, which has an inherent locational trigger. Similarly, an inconclusive pattern is exhibited when examining the trend of financial and social-dwelling relevance by the number of housing moves (Figure 6.7).

Regarding the more universally relevant social-dwelling factors, there is a more consistent increase in relevance in relation to residential mobility. Though, when considering financial relevance, the visible pattern is the increase in relevance of people making 1 to 4 housing transactions and a noticeable drop from individuals making 4-8 housing moves (Figure 6.8 below).

Figure 6.8 Change in the Mean of Relevance of Financial and Social-Dwelling Factors by # of Moves (1-8 Moves) This implies that people with more frequent housing moves are making inefficient moves. These individuals are often the most vulnerable/disadvantaged, while specialised financial attributes are more readily accrued by wealthier individuals with more stable tenure. However, the more individuals moved, the greater their perceived understanding of the market. Figure 6.9 on the next page shows participants’ median moves according to their occupation and understanding. The survey asked participants to select from three levels of understanding: great understanding, some understanding and little understanding. Most participants chose ‘some understanding’,

148

trailed by those with ‘little understanding’ and lastly those with ‘great’ understanding’. The grand total row refers to median housing moves within each category. An increase in median moves, corresponded with an increase in perceived market understanding.

Figure 6.9 Reasons for Moving (Median Moves): Occupation/Level of Understanding Therefore, housing transactions become inefficient after 1-6 moves for social- dwelling factors, and after 1-4 moves for financial factors, yet market understanding continues to improve. This relationship can be explained by the Sankey chart (Figure 6.10 on page 151). The Sankey shows the relative change in the rank (mean of preferences) of each of the 27 attributes by age. The Sankey shows that as individuals aged, their prioritisation of social-dwelling attributes increased. However, as individuals matured further, exogenous life-course circumstances eventually returned to the fore. This means that in the housing market, economic capital, social capital, and market understanding help to ease early life-course demands, however, towards the tail end of individuals’ housing trajectories, market understanding and other resources are less effective at alleviating life-course factors.

149

Figure 6.10 also shows that the social-dwelling attributes are the most stable across the age cohorts. Conversely, the financial and life-course factors are less consistent across the age groups. Table 6.1 below shows that social-dwelling and financial attributes are negatively correlated with the life-course (-0.96 and -0.97 R2 respectively, 0.05 significance level; calculated based on the correlation of the mean of the changes in rank). This means that life-course pressures suppress the prioritisation of financial and social-dwelling satisfiers. This is understandable because financial strategies require long-term visioning and social-dwelling stability. Thus, social-dwelling and financial attributes are positively correlated (0.85 R2, 0.10 significance level). A general increase in the prioritisation of social-dwelling factors coincides with a general increase in the prioritisation of financial factors.

Table 6.1 Correlation between Changes in the Mean x̅ of rank of Financial, Life Course, and Social-Dwelling Factors Legend 18-29 to 30-39 to 40-49 to 50-59 to FI: R2 SD: R2 LC: R2 x̅ increase x̅ decrease 30-39 40-49 50-59 60-69 Financial (FI) -1.44 1.44 -3.22 0.78 0.85 -0.97 Social-Dwelling (SD) 0.67 2.22 -2.33 0.11 0.85 -0.96 Life-course (LC) 0.78 -3.67 5.56 -0.89 -0.97 -0.96

The impact of the life-course factors on social-dwelling and financial satisfaction means that climbing the housing ladder is a poor indication of how needs are being satisfied. Moreover, a further implication of the life-course and financial fluctuations, is that while there are increased changes in prioritisation as individuals progress through the life-course, individuals make more housing moves in the early stages of the life- course. This further corroborates the notion that the timeliness of housing moves has a greater impact on changing housing circumstances than the number of moves.

150

Figure 6.10 Sankey of Changes in Relevance of Life-course, Social-Dwelling and Financial attributes (All Cohorts)

151

6.2 Exploring Individuals’ Attitudes in the Brisbane GCCSA

This section will explore individuals’ attitudes towards the housing market using causal coding. These attitudes were derived from 199 of the 237 survey respondents and are represented within two themes: understanding the impact of individual needs on housing attitudes and understanding the impact of external factors on housing attitudes. The first theme (individual needs) comprises six attributes and is designed to address Research Question 2 (RQ2): How are personal needs responsible for differing housing experiences? The second theme (external factors) also comprises six attributes and is designed to address Research Question 3: How are divergent attitudes responsible for differing housing experiences? Altogether the twelve factors (Table 5.8, Subsection 5.3.3) represent respondents’ perceptions of the overarching problems facing household tenure in the Brisbane GCCSA.

Throughout the thesis, the intersection between internally derived and externally derived needs have been explored. In a similar vein, the separation of individuals’ attitudes into the above-mentioned themes makes it possible to gauge the extent of the contradiction between the various housing market participants, and how they relate to the broader normative beliefs and attitudes that were explained earlier in the thesis. Contradictions are framed as the gulf between expectations and attainable outcomes. Contradictions may stem from internal needs such as family circumstances or external factors such as disputes with external parties.

These factors are important because the housing market is shaped by the perceptions of individuals, authorities, developers, and corporate entities which informs their decision-making. Since individual needs and their understanding of external factors are based on empirical housing experiences, as well as their understanding of actual housing events that they have observed; some causal mechanisms can be corroborated by examining if the regularities that were proposed in Chapters 3-4 are present in the contexts communicated by the participants.

6.2.1 Impact of Individual Needs on Housing Attitudes ‘Individual/family circumstances’ is the first of the individual needs attributes, and it is centred on individual and family circumstances. These circumstances include the general housing needs of the household unit. This category does not incorporate needs associated with changes to the household unit or demands based on

152

employment opportunities and access. Changes to the household unit form a separate but interrelated category. Therefore, the life-course category embodies general individual and family needs, while the household-unit category incorporates structural changes and transitions. The distinction between these two categories is important, because general housing needs indicate a more predictable and controllable experience, while household unit changes involve relationship transitions. This classification also distinguishes employment and employment needs from employment-related accessibility. Consequently, general employment needs such as having a job, or having flexible tenure, are then separated from locational qualities.

This category explores issues such as growing and changing family needs, unspecified lifestyle changes and children's needs. One of the reasons why housing needs were coupled with life-course circumstances, within the survey design, was due to how needs manifest. Although we can think of stages within the life-course as ‘expectations’ and ‘norms’ (Giddens, 2009), there are specific events that need to occur before changes can be realised. Housing consumption is financially tasking, and thus, people’s housing trajectories in the face of growing economic challenges may not always adhere to expectations. Therefore, it is more accurate to determine which events are associated with yearnings or outcomes. An example of one individual’s changing circumstances is shown with the following quote from participant # 115, 30- 39-year-old male. “Changing status/career. Example: From student to new employee or from employee to student; not being F/T employed/ needing to change jobs”.

As per the above quote, it is folly to presuppose that life-course events adhere to predetermined normative patterns. Due to the significant pressures that people face, various facets of a life-course stage may be contradictory. In this thesis, lifestyle is largely described as the successful translation of multi-faceted life-course needs into a cohesive narrative, yet lifestyle changes can also be based on a more lopsided prioritisation of norms such as status. This effectively means that it is not simply the presence of life-course pressures and circumstances that defines lifestyle adjustments; the relative absence of life-course burdens may also foster a lifestyle adaptation.

The actualisation of life-course needs was found to be complicated in family- settings. Families compound the number of variables that inspire life changes. The

153

process of moulding lifestyles from life-course events is facilitated by the fact that life events unearth new needs. It should be stated that life-course changes do not necessarily suggest that some life circumstances disappear and are supplanted by new ones. Life-course changes may simply imply changes in the priority of different needs at different stages.

The second individual-needs category relates to ‘relationships and household unit changes/separation’. Issues within this category are exogenous, in that they involve social networks, or they involve changes between the individual and other partners, or housemates, for example. These interactions are viewed as less predictable, as they are reliant on external factors. Hence, factors such as relationship changes, or changes within the household unit, as well as community/social interaction and support are explored in this group. The unpredictability of this category is exemplified in the following quote from participant #90, 18-29-year-old male. “At that stage in my life it was purely my job moving me around. But I feel as people are looking for a change more regularly and life is always changing so you must adjust to cope. That could mean moving house or getting a housemate”.

There is also an ‘employment/education, and family needs’ category centred on the transport and locational issues associated with those needs. This class was separated from individual/family circumstances due to the distinct function of locational demands. Locational preferences are one step removed from the individual’s or household’s needs. The location is not the inherent need, but rather the facilities that are sought. This is especially the case because, in this category, the dynamics of status and the financial or future value of location are not being assessed. What is being studied is how easily people can access qualities such as employment, education, amenities, services, and family. Due to the significance of these factors on people’s livelihood, this category influences where individuals will live and may even contradict other preferences that they may have. “My job can take me anywhere in Queensland. I've also moved in the past because of dodgy r so estate agents that were managing the property I rented”. This quote by participant #164, 30-39-year-old female, shows how employment can be restrictive, while in other circumstances, it may be thought of as providing opportunity. This relationship held true both on a local scale as well as on a wider scale. People might move interstate to be nearer to housing, or they might move between suburbs for ease of access. Many individuals

154

were prone to shuffling around due to exogenous circumstances. Untenable circumstances may result in individuals changing housing market altogether.

Furthermore, this category increases in complexity, proportionally to the size of the household unit. Balancing multiple employment locations was a reality among those with family or friends living together. Such potential might be determined by factors such as the inclusion of a new housemate. Since progression can be thought of as inherent or dependent on a trigger, it is often tied to changes in financial capacity. While progression can be perceived as the ability to make unstable situations stable, there were individuals who proactively sought change.

‘Financial capacity’ forms the fourth individual-needs category. In the same vein as the previous categories, this group relates to individual capacities. Thus, financial impacts (which will be discussed in Subsection 6.2.2) form a different category altogether, within the external factors category. Conversely, this class relates to individuals’ and households’ financial capacities as well as how their housing helped facilitate their financial needs, such as through rental income. This quote from participant #154, 50-59-year-old female shows the importance of capital. “Finances . Need to move somewhere cheaper”. This category then involves how individuals can approach housing based on their economic standing, rather than the financial impacts that may emerge from participating in the housing market, such as housing indebtedness and housing stress. Thus, Individuals also found that their potential for adapting to life changes was contingent on the size and composition of the household unit.

The fifth individual-needs category is based on the ‘unpredictability/uncertainty’, particularly relating to landlords and the life-course. Some of the issues that individuals experienced included the perceived exploitation of renters, which was often compounded with the understanding that rental regulations and management were lacking. This problem is explored as an individual needs factor, as it shapes how individuals perceive the housing market. This is underscored by another key dimension with this category, which was the discernment that rental owners/investors were unpredictable. This uncertainty is shown in the following quote by participant #70, 30-39-year-old male “Short term leases with no power to extend lease if land lord

155

wants to evict you for any reason. Annual rent increases. Cost of living not parity with wage growth”.

The last individual-needs dimension involves ‘wider/neighbourhood conditions and changes’. While many demand-side categories pertain to how neighbourhoods can accommodate various needs, such as through ease of access to amenities, this class is associated with broader changes to the neighbourhood. The dynamics within a community is, therefore, something that was at the forefront of many residents’ minds. Then, this category comprises interactions between individuals and their community. For that reason, any change to a community’s internal structure underscores a significant transformation of the character and or functionality of a locale. These changes were triggered by factors such as an increase in activity, demolition, noise, traffic, crime, construction, decrease in parks/space and a loss of preservation. A select few also attributed a significant change in the character or functionality of a community to immigration issues. Growth can be particularly problematic for residents as it encompasses much of the aforementioned factors. This is communicated by participant #48, 28-29-year-old female, in the following quote. “Changing aesthetic, demographic of the community. Overpopulation. Too built up”.

It was also established that apart from continuity, personal needs were paramount. It was found that notwithstanding the presence of some neighbourhood amenities, they may be incompatible with the demography/lifestyle of certain individuals or households, as those amenities did not reflect their present identity.

6.2.2 Impact of External Factors on Housing Attitudes External-factor dimensions are the repercussions of the challenges, trade-offs, and frictions, which constrain individual-needs. The first of these are incorporated in the ‘housing disputes and divergent attitudes’ category. Although the reasons that lead to individuals relocating or seeking alternative residences are multi-faceted, conflict is one of the key drivers. Due to the regulatory and housing management concerns, many respondents believed that there were divergent expectations between renters and owners. This divergence occasionally resulted in conflict and disputes, as well as evictions. However, this category does not delve into the specific circumstances of terminations or notices to leave, which will be explored as part of the housing tenure classification. Rather this category explores respondents’ value constructs, which then

156

suggests why conflict emerges. One such experience was documented by a participant in the following quote by participant #195, 18-29-year-old female. “Difficult neighbours. Cost of owning a large home. Takes up too much time maintaining your own home”. The quote underscores the yearning for stability and healthy relations. Such disputes between neighbours and flatmates occur frequently. Thus, this uncertainty has been incorporated as part of the rental decision-making process.

There was also a divergence between the objectives of investors and those of renters. A significant segment of the respondents perceived the market as favouring the investors, managers, and landlords. Some individuals had an alternative perception of housing and viewed it as an investment vehicle and asset, believing that emotional attachments conflicted with buying and selling as the need arises. These perspectives represent differences between individuals who must rent at the behest of investors and how investors often perceive the market.

The external factor categories shed light into a different facet of finances and socio-economic wellbeing. While the individual-needs category of ‘Financial Capacity’ explored what and how individuals were able to participate in the housing market, the ‘home price changes’ category focuses on financial challenges that emerge as a direct consequence of housing. This includes rental price increases and fluctuations, as well as how the cost of living increases relative to income growth. Rental increases were perceived as both a cause of housing problems, as well as a by-product of other issues. There is a compounding effect, whereby living in proximity to other people with volatile rental tenure/arrangements can add to volatility.

People also blamed investors and the market for unpredictable and unjustifiable increases in rent. They perceived an inflationary cycle of housing prices with capital gains outstripping wage growth. These increases often led to unwanted mobility. This perspective is seen in this quote by participant #123, 40-49-year-old female. “* Landlords not keeping up maintenance to property therefore turning it into a slum - rent increases while house quality/safety decreases* Rent increases being higher than average cost of living* Harassment from Real Estate Agent”. However, some respondents determined that understanding the market might ultimately mean making numerous moves, to develop an understanding of suburbs and the market.

157

‘Housing tenure’ was one of the most unsettling external factor dimensions. This was due to housing unpredictability and the regulatory environment, which was outlined in the impact of individual needs subsection. While some individuals were able to learn from their rental experiences, the general problem with housing tenure, as noted by the respondents was the sheer unpredictability. This meant that the renter cohort often experienced tenure insecurity, as they could not pre-empt when their place of residence would be sold or occupied by the owner. As a result, a general gripe among the renters was their inability to secure long-term tenure. Many individuals found that rentals were restrictive, in that they were unable to modify their home to suit their needs. This contributed to a general understanding that homeownership tenure was desirable. This yearning for homeownership was complicated by the fact that there was a weak transition between renting to ownership, according to the respondents.

Residents’ capacity to actualise their housing needs was equally as indicative of their housing trajectory, as their subjective aspirations. If residents were unable to safeguard a consistent standard of living, then their aspirations were effectively hindered. One of the fundamental ways in which individuals could preserve and influence their standard of living was through tenure. This is evidenced in the excerpt below from participant #175, 40-49-year-old Male.

Rental properties are often not secure, as leases are short term (6-12months) which makes it difficult to make long-term plans. Rental prices tend to rise regularly and sometimes sharply, and tenants have very little control over the conditions of their rental agreements. Tenants are also heavily dictated to, about how they live in rental properties, with pets, changes to gardens and internal changes, restricted. All these factors make rental properties feel like an insecure environment and make it difficult to settle comfortably.

Additionally, one of the differences of opinion was on how rental properties could serve the individual needs of renters, such as how their home could reflect their personal identity. The following quote by participant #146, 40-49-year-old male, highlights how the problem is perceived.

158

It is very difficult to make changes to a rental property to reflect changing situations (e.g. pets) or other lifestyle requirements where the property is less than ideal (e.g. air conditioning). Usually, the only choice is to move.

It was uncommon for individuals to draw parallels between tenure types. Some respondents perceived uncertainty in the rental market while failing to identify any uncertainty in the housing market. The rental cycle was intuitively perceived as being more volatile by many individuals.

‘Housing conditions’ were complicated by factors such as a disparity between dwelling value, quality, and housing costs. People expected to experience housing standards that were proportional to the asking price. This is seen in this quote by participant #209, 30-39-year-old female “Getting to a point financy(sic) where you can afford to stay in one property and also still enjoy and be comfy in that property”. This is in addition to a demand for housing that addresses their particular needs, which means that as a household’s life circumstances evolve, the utility and functionality of the house in which they inhabit might lessen. The value of housing is, therefore, conditional on an individual’s stage in life.

While the future functionality of a house was a significant issue among the respondents, the most significant aspect for the rental cohort was the poor maintenance of their home or the condition of the property in general. On the one hand, many respondents believed that changing life circumstances were expected, and therefore changing their housing arrangements was part and parcel of the housing experience. Whereas on the other hand, maintenance issues were more problematic, as they believed that such problems should be more manageable. The lack of maintenance of properties was a problem that many respondents felt was out of their control as renters. Life-course changes such as a growing family were of their own doing and could be internalised, which is in stark contrast to property disputes which were at the behest of the owners and managers of the property. This was communicated with numerous accounts of residents experiencing difficulties getting rental managers to maintain their property.

While control over their home was less attainable for renters, the opposite holds for the owner cohort. A considerable number of the participants had a need for ‘home improvement and upsizing’. Upsizing was common among growing households as well

159

as allowing households to rent out to the public for additional income. Upsizing was perceived as a likely event because the spatial needs of individuals or households are relative. It is often tasking to accommodate dynamic needs in a fixed dwelling over time. The efficacy of housing typology is closely related to the life-course, since the course of one’s life is a dynamic and ongoing process, as described by one participant. “Changing needs of the house as family grows”. Consequently, some participants found that they required more space for their needs, as stated in the following quote.

Size of family, needing more room, needing another living/dining area for the kids, wanting to adjust the garden to suit our needs, wanting to add outdoor decking or more shade.

Increased spatial needs can also be resolved through renovations, but such improvements are also costly, as communicated by participant #17, 30-39-year-old male, “Changing life circumstances including: - relationship changes- employment type and location- suitability of dwelling to changing circumstances- cost, time and effort of renovations”.

On the other hand, ‘home downgrading/downsizing’ was a consideration among the aging cohort. This is communicated by participant # 107, 40-49-year-old male, in the following quote. “House size compared to occupants. Inevitable need to downsize” Older people often live in residences that once housed their children, and once served various needs in their younger life. Therefore, they may elect to downsize their home if they feel that their housing over services their needs. Alternatively, they may also elect to downgrade their home. Downgrading in this context, is a financial consideration, as it means moving to a house with lesser upkeep, and possibly accessing liquidity if they own and sell the property; downgrading may also refer to lessening the upkeep and functionality of their existing home. Although downsizing may be closely related to downgrading, downsizing encompasses much more than financial objectives. Downsizing was used to refer to the size of the home relative to the needs of the inhabitants. The connection between downsizing and downgrading is because minimizing spatial consumption might also result in cheaper housing. However, downsizing is a complex process that may be initiated without necessarily factoring finance or having the perception of downgrading.

160

Of all the categories, the most significant issue was found in the consequences of financial impacts on residents. This is crucial considering that these impacts are tied to increasing rental and mortgage costs, which is a different problem altogether to the financial capacity dimension outlined in the individual needs subsection. While a case may be made that financial intervention (credit supply) is capable of increasing accessibility to the market (Fox & Finlay, 2012), there is another subcomponent of financial difficulty tied to household costs and financial management. This is corroborated by the extensive household surveys conducted by, in which mortgage stress was observed to be more widespread than the official figures, which fail to consider inflation and rising costs aside from housing itself (North, 2019). Therefore, the present situation is that the strong growth observed in the Australian and Brisbane GCCSA housing markets has not translated to less demanding financial conditions for households. Individuals experience this financial stress within their individual households, but mortgage and rental costs are tied to the functionality of the Australian property-owning democracy. The promise of financial independence is presupposed alongside the ownership prerogative.

One of the reasons why the unrealised potential of the property-democracy may be overlooked is that the plight of residents in the housing market has often been oversimplified as a problem limited to getting on the proverbial housing ladder. However, mortgage stress has been shown to be as acute as rental stress, and worryingly the most significant external factors from the open-ended question aside from financial impacts were related to residential disputes and tenure. Notably, there was considerable friction observed between renters and landlords. This friction was observed in terms of day to day interactions and management, but also embodies the long-term uncertainty associated with the absence of tenure security experienced by renters. Consequently, the unrealised potential of Australia’s property democracy has not alleviated the financial security of homeowners and at the same time, there is a feeling among many residents that the owner class have divergent interests to those who are not members of the property class.

6.2.3 Summary: How Residents’ Individual Needs Conflict with External Motives This subsection is a summary of the contradiction that exists between residents’ attitudes towards their housing needs and their attitudes towards external factors. This component of the research findings has substantiated the legitimacy of a progressive

161

yearning for individuals to aspire towards homeownership from their renter background. The nature of renting as a lower level of commitment, financially (at least in the short term) and contractually, means that a larger percentage of the lower socio- economic cohort are represented, as well as vulnerable and youthful individuals. Thus, the findings from Subsection 6.1.1, which showed that very few individuals were capitalising on their rental flexibility to invest money that they would otherwise commit to homeownership elsewhere is expected.

However, the findings have also shown that the reason why the rental market is characterised by precarity in the first instance, is because individuals’ capacity to harness their individual needs is directly limited by the absence of external regulatory support, comparable to that of the homeownership market. This was reflected in findings from this subsection which highlighted individuals’ perception of financial unpredictability, regarding fluctuating rental costs and regulatory constraints. These unequal regulatory conditions emerge since investment and debt are the key drivers of the market, and this causal financial tendency makes the housing ladder structurally discontinuous.

The discontinuity in the housing ladder proves that the core normative beliefs and social norms within the housing market are loosely sustained due to the premise of mobility, not due to the benefits afforded to all participants. This is problematic because Section 6.1 outlined that residential mobility is inefficient and is only a useful predictor of improved housing conditions during specific life-course junctures.

162

CHAPTER 7: Property Industry’s Facilitation of Household Needs The aim of this chapter is to uncover how well the housing market reflects household needs. The primary objective in fulfilling this aim is to reveal the composition of housing needs in terms of actual housing market features. The value of deriving the more abstract composition of housing needs from participants is that it forms a tangible representation of how aspirations have been mediated in the market. Therefore, this chapter is a representation of how the market is presently functioning, though it does not explain the intricacies of individuals’ needs, which was observed in Chapter 6. As explained in Section 5.4, it is generally assumed by the public that real estate agents will maximise their advertising space to communicate the most neutral or positive features in the least amount of words. Consequently, these words can be considered as a concise reproduction of the market’s functionality.

The implementation of SNA within the research methodology is designed to complement the housing needs survey (HNS) by revealing the market conditions that are not evident within the survey. The HNS is designed to uncover patterns of housing aspiration by probing the relevance of 27 housing themes. Though these themes are influenced by market realities, they are represented through individual housing experiences. The distinction between experiences and actual events occurs because individuals in the housing market are unable to experience the market in its entirety. This is aligned with the critical realist ontology which stipulates that events in the housing market may be observed or unobserved. The Brisbane GCCSA HNS thus has a limited view of the actual composition of the housing market. For this reason, understanding how the housing market facilitates housing needs is required. Since housing aspirations lie in the juncture between needs and external possibilities, this method is useful, given that it outlines housing outcomes (in the form of housing features/ideals), which are mapped against the housing needs satisfiers.

This sequence represents the relationship between individuals’ needs and housing market features. The greater an individual’s agency is, or the more enmeshed an individual is within the market’s causal mechanisms, the higher the degree to which the market will reflect their needs. For example, if permanence were a factor that the renter cohort was unable to express, the market would not accurately reproduce this quality in the network plots. This relationship is emergent because the features of the housing market cannot be attributed to the sum of individual actions based on

163

retroductive reasoning. What can be inferred is the degree to which the market mirrors housing demand (Demographia, 2019); which partially embodies housing needs (see Chapter 4). This mirroring can be inferred based on the co-occurrence of textual language within SNA, which highlights the relative prominence of real estate terminology (which is analysed according to the three housing facets). Prominence in the context of this thesis is defined by weighted degree centrality and relates to the central vertices and edges, the visible edges, and the colouring of the edges

In the subsequent sections of this chapter, the two overarching measures of SNA analysis will be analysed. The first measure is based on centrality values which explain the relationship between real estate property terms and how they relate to individuals’ needs (vertex level analysis), whereas the second measure is based on community detection (system-level analysis). Community detection reveals communities within housing networks that share similar edge interactions and meaning. A secondary benefit is that terms at the intersection where communities meet are responsible for bridging distinct housing components. In community models, ‘communities’ are not derived from a predefined ontology, and emerge directly from the analysis without any adjustment from the researcher. However, the communities in Section 7.3 are examined, and prescribed names based on the common relationships observed within each cluster.

With regards to the property terms within the networks, the most significant relationships are not the connotative, denotative, and fuzzy relations, which are based on the direct associations and variations of property terms (see Table 5.9). What will instead be explored through semantics will be the inferences and implicature of the terms. Inferences being what the real estate agents want to communicate passively, and implicature being what they want to communicate actively. These relationships will be discussed at the conclusion of each subgraphs’ summary in subsections 7.2.2- 7.2.4. However, what can be immediately discerned from the literal/associative meanings of the property terms is the absolute weighting of the various terms.

7.1 Social Network Analysis of the Market: Centrality Measures

This section aims to reveal the relationships between the 69 real estate terms of the rental network and that of the 72 real estate terms of the sales using the distribution of centrality scores. These scores are derived from the number of times that each

164

property term co-occurs with another term in a single property listing. Therefore, an account of how relevant or significant the terminology is within the housing market is constructed. The centrality measures are firstly analysed within two separate networks based on the sales and rental markets, since the information that the real estate agents convey to people leasing homes is divergent from the information conveyed to buyers.

Three centrality measures: weighted degree centrality (WDC), eigenvector centrality (EVC) and closeness centrality (CC) were explained in Section 5.4; WDC was the most accurate representation of overall connectedness. Closeness centrality is biased towards property terms that are context-specific since it is computed using the average distances (weights), not the aggregate. This is because vertices that occur more generally will more likely have lower weighted edges, lowering the mean. Eigenvector centrality is based on an identification of the real estate terms that are associated with highly influential terms. There is considerable similarity in these centrality scores with those derived from the WDC measure because property terms with high weighted degree scores are densely connected. Since a term with a high eigenvector is reliant on being connected to many high eigenvector property terms, it is also a reflection of sub-network density (Csardi, 2015b).

Figures 7.1-7.2 on the next pages show how the ranking by normalised centrality changes between the three centrality measures. Appendix 29 provides a comprehensive outline of how to interpret those distributions. The Sankey charts show that variation primarily occurs outside the densely distributed network segments. This variation in the lesser half of the Sankey charts represents that fact that fringe terms (low centrality outliers) may have a low WDC, but still have an average closeness centrality that is relatively high. Vice versa, some highly weighted property terms are related to only a segment of the network, thus have a lower average centrality (closeness centrality).

The rental network appears to have greater consistency among its focal property terms (rooms, bedrooms, kitchen, area, living, built, located, walk, and open). However, this highlights that the basic housing typology features are employed more consistently than home ideals. Conversely, the sales network has a similar dwelling composition for the WDC and EVC measures, but its home terms have a higher normalised closeness centrality comparatively. This means that in the sales market,

165

the home terms are more selectively defined, and more indiscriminate in the rental market. These Sankey charts supplement the analysis in Subsections 7.2.2-7.2.4, by highlighting the contextual difference between how property terms function in the sales and rental networks.

Figure 7.1 Change in Normalised Rank of Terms by Centrality Methods: Rentals

166

Figure 7.2 Change in Normalised Rank of Terms by Centrality Methods: Sales

167

7.2 Centrality Measures of the Market: Weighted Degree Centrality

Of the three centrality measures: weighted degree centrality (WDC), eigenvector centrality and closeness centrality that have been introduced, the weighted degree WDC approach best represents the magnitude of information flows, and the extensiveness of the connections between property terms. The other centrality measures focus on the strategic or contextual significance of terms. Thus, this section primarily explores the WDC of the real estate network in relation to the three divisions: geography, tenure and typology, which were introduced in Subsection 5.4.3. These divisions and their subclassifications altogether constitute ten networks/subgraphs. Analyses at the subgraph level reveal how the aspirations of different segments of the housing market have been facilitated. In this section, the ten network plots will anchor the RQ4 findings in Subsections 7.2.2-7.2.4, and these subsections will be preceded with a description of each subgraph’s density in Subsection 7.2.1. The SNA plots show the prominent vertices (terms) and edges, while the density discussion describes how well the prominent terms and edges can spread across the subgraphs.

The sum of all co-occurrences or edges to a given vertex determines its WDC. The most prominent real estate terms have a greater WDC score and have the largest size and labels (shown in the centrality and community SNA plots in Subsections 7.2.2-7.2.4 and Section 7.3). As mentioned in Subsection 5.4.2, the property terms (vertices) are colour-coded as dwelling, home and house vertices based on how they correspond to the structures of activities (shown below).

1) House: Amenity, Work, Play, Type, Size, Economic, Identity, Resale value, ROI 2) Dwelling: Work, Play, Type, Size, Amenity, Age & Structural Integrity, Amenity, Privacy/Continuity 3) Home: Amenity, Work, Play, Type, Size, Age & Structural Integrity

The network plots can also be interpreted by the number of edges (connections) visible, the size of those edges, and by edge colours. The size of edges is regulated by the normalised co-occurrence between a given property term and other property terms. The size of the vertices is then equal to the sum of all normalised edges. Thus, the individual edge relationships represent the relative prominence between two terms as a normalised edge weight between 0 and 1. The visibility of the edges is determined

168

across all figures by deleting edge weights that fall below the threshold of the sum of all weights. The threshold is the edge density multiplied by a factor of 2.5. Therefore, the threshold takes into account the differences in the number of edges within each subgraph.

7.2.1 The Density of the Real Estate Networks and Subgraphs The significance of density within SNA is how easily information (real estate text) can be disseminated. The flow of information is aided by density, but density refers to the concentration of edges in the core of networks/subgraphs, not the overall distribution of subgraphs/networks. Density is exemplified in the relationship between visible edges (asymmetric or mutual) and triplets (density). This approach is similar to the modelling applied by Faust (2006), except the author used triads instead of triplets (see Table 5.12).

This relationship (clustering coefficient) highlights that network density influences the distribution of property terms (shown in the Figure 7.3 scatterplot on the next page). Simply, the coefficient shows that the probability of vertices being clustered reduces the number of visible edges, and the prospect of information being easily disseminated. The scatterplot is broadly divided into three categories: The rental network is shown in light blue, while its subgraphs are shown in teal. The sales network is shown in red, while its subgraphs are shown in orange. The sales land network is, however, an exception and is shown in light orange. The mean of the entire property market is shown in green.

The probability of a high clustering coefficient reduces the efficacy of distributed information because the weights or edges between property terms produce a zero- sum relationship. Networks with more uniformly distributed edges (higher clustering coefficient) will have fewer highly weighted edges. Due to the normalisation of the edges, uniformity means that the strength of the relationships between terms is diluted. Thus, in uniform networks, property terms are generally associated with one another, whereas asymmetric networks have prominent terms that are relevant in specific contexts. This phenomenon exemplifies the concept of selective marketing, explained in Subsection 5.4.2 (Camilleri, 2018). A formal SNA explanation of uniform and asymmetric subgraphs is centralisation and decentralisation, respectively. Subgraphs with a high clustering coefficient are decentralised, and subgraphs with a low

169

clustering coefficient are centralised. Centralisation relates to how property terms are organised around focal points (Stephen P. Borgatti, 2015). In a centralised network, most information flows are among the prominent terms, while the peripheral vertices are relatively isolated. This results in a greater number of visible edges (see Figure 7.3 below). In decentralised networks, information flows are dispersed throughout the whole network, and language use is indiscriminate and inefficient. This results in a lower number of visible edges.

Figure 7.3 Relationship between Clustering Coefficient and the Visible Edges of All Networks and Subgraphs

Consequently, the clustering coefficient reveals that the sales land subgraph will have key property terms with the greatest likelihood of being disseminated to the public. The other sales subgraphs/network have an average likelihood of information being effectively disseminated. Lastly, the rental network and subgraphs have a less than average probability of information being effectively propagated. This relationship can be explained by two causal mechanism tendencies. The first is the territorial tendency (territoriality reflects how needs are imagined) (Storey, 2012) which suggests that subgraphs that integrate a greater variety of housing need satisfiers also effectively and efficiently employ selective marketing. Although the land subgraph 170

embodies raw unfulfilled potential, it is also an opportunity for the prospective representation of housing needs. The sales subgraphs/network are less typologically diverse than the rental network/subgraphs (see Table 5.15, Subsection 5.4.2), but normative beliefs, ideals, and regulatory freedoms are greater in the sales market. Lastly, the rental market contains a greater balance of apartment and detached housing, but the significance of these housing settings is ideologically lesser.

The second causal mechanism is the financial tendency. This tendency suggests that real estate advertisements with a direct financial focus use more efficient terminology. Thus, the sales land subgraph, which has the most direct financial focus, can be marketed more efficiently. Efficiency is gauged in terms of clustering, wherein networks that have lower clustering (density) among their non-focal terms and higher clustering among their focal terms, exhibit greater information flows. On the other hand, the sales subgraphs/network is more complex, and thus, sales terminology is relatively less efficient. Lastly, the rental network and subgraphs have a higher overall cluster coefficiency. Thus, the lower density among focal terms likely reflects the fact that the rental market has a lesser financial focus than the sales network.

The territorial tendency can be tested by comparing whether the composition of housing need facets within the rental and sales subgraphs/networks are balanced (Subsections 7.2.2-7.2.4). The financial tendency can be tested by measuring the regularity of prominent house themes across all subgraphs. If these tendencies are together validated, it will reinforce the notion that the housing ladder is a self-fulfilling prophecy, since the market’s focus influences how housing is imagined.

7.2.2 The Prominence and Semantics of the Rental Network and Subgraphs The previous subsections underlined the decentralised and centralised nature of the rental and sales subgraphs/networks, respectively. For the rental network, selective marketing was diminished, and the learning capacity was weakened. Thus, the initial centrality and centralisation analysis suggests that the market’s feedback of renters’ housing needs was inefficient, while the most efficient was the sales land subgraph. This conclusion is based on use-value and financial tendencies. These tendencies can be validated in this subsection by testing the regularity in which information is spread throughout the networks/subgraphs.

171

Regarding the territorial tendency, it is vital to recognise how well the rental property terms correspond to the facets of the housing ternary (home, house, and dwelling) since this represents the extensiveness and selectiveness of the market’s representation of housing needs. If property terms within the sales subgraphs/network are more thematically extensive than the rental network, the relationship between clustering and housing needs satisfiers will be substantiated. In contrast, financial tendency suggests that the integration of a network/subgraph with the market’s territoriality (financial focus) improves its efficiency. This premise can be tested by observing the prominence of house themes within the various networks/subgraphs.

Lastly, this subsection will also uncover the extent to which housing needs satisfiers are related to one another. This relationship will clarify if holistic housing outcomes are being delivered or simply niches within each network/subgraph. These measurements are initiated by observing the prominent themes (vertices) and the most important edges within the networks/subgraphs. These visible edges are colour coded for each network/subgraph; the darker the edges (peach to maroon), the higher the relative impact within the network. If a network exhibits visible edges across the ‘Upper’, the ‘Mid’, and the ‘Low’, of the colour spectrum, this means that the network/subgraph has edges that are relatively more influential than networks/subgraphs without colour differentiation. Aside from the colour differentiation, the most prominent edges in the graphs are centrally located, due to the Fruchterman-Reingold layout. This layout ensured that the network developed a radial spoke appearance due to the outwardly springing of the edges (evidenced by the arrows) with the greatest weights from the centrally located property terms.

The first network/subgraph that will be analysed is the all rental network (Figure 7.4 on the next page). The extent to which the market satisfies housing needs (the territorial tendency), can be tested by identifying the composition of satisfiers within the network. The network reveals that no prominent edges originate from the house related terms. Instead, prominent edges originate from dwelling related attributes such as: ‘kitchen’, ‘bedrooms’ and ‘bathroom’. The next tier of edges stems from the home attributes such as: ‘living’, ‘area’ and ‘shops’. Thus, the property terms reveal an incomplete representation of the housing ternary, since financial freedoms are peripheral in the network. This conclusion is also corroborated by the peripheral placement and size of the house vertices.

172

Figure 7.4 Prominent Terms/Edges of the All Rental Properties Network

In a similar vein as the housing needs survey, this SNA research captures broader house needs satisfiers. For the rental cohort, these include locational amenities (which are located in the network periphery), as well as a few dwelling related terms that are more dominant. Hence, the relationship between these dwelling terms exhibits a low level of complexity. This lack of complexity is also evidenced by few visible property terms forming closed triplets (“friends of friends become friends”) (Veenstra et al., 2013). Thus, this network has negligible peripheral edges rather than influential property terms (see Appendix 22). This relationship validates the territorial tendency, which suggests the rental market has an imbalanced representation of housing needs satisfiers. Also, the financial tendency that subgraphs/networks with a stronger financial focus will have the greatest efficiency is validated since house terms are relatively insignificant in the network.

Finally, the thematic groupings of this network are considered. There are fewer edges to analyse in this subsection due to the broad distribution of weights in the rental networks. Consequently, this subsection focuses on meaning as it relates to the

173

vertices (Figure 7.5). Highlighted in yellow are the standout property terms other than the generic dwelling attributes. There is a focus on spaciousness, location, accessibility and amenity. These terms do not have as much of an impact as the dwelling terms as mentioned above; however, based on closeness centrality, the locational terms are more consistently related to other terms. This means that the locational terms are important in context, though relatively isolated from the rest of the network. Conversely, the ‘home’ term is less robust than suggested. Its key relations are with terms such as bedroom, while its closeness centrality is low relative to its prominence. These relationships suggest that real estate agents rely on the typological features of housing first and foremost (market positioning) (Camilleri, 2018), which is market differentiation by product, e.g. ‘bathroom’, ‘bedrooms’, and ‘kitchen’. Then, they rely on locational attributes. Thereafter they rely on cultural aspects such as ‘home’, ‘pets’, ‘yard’ and ‘space’. Yet, the term ‘home’ has a lower closeness centrality relative to its prominence (weighted degree/eigenvector), which suggests that in the rental network, the term is used relatively indiscriminately or generally.

Figure 7.5 Top 30 Rental Network Terms by WDC Normalised Value

The rental network can be further analysed by delving into the typological distinction between apartments and houses. In the rental apartment subgraph (Figure 7.6 on the next page) the most prominent edges stem from even fewer central vertices than the entire rental network. These prominent vertices are primarily the ‘room’ and ‘bedrooms’ terms. Hence, it is possible to define the absolute centre of the subgraph as relating to bedrooms or rooms. The absolute centre is the single-most central property ideal. The sparseness of the visible subgraph edges can be explained

174

by the higher standard deviation. In such decentralised subgraphs/networks, the normalised mean (0.32 in this case) is nearer to the mid-range (0.5) than in centralised subgraphs/networks (see Appendix 31). This distribution implies that the edges are more evenly distributed across the subgraph. The more evenly distributed the edges are, the fewer prominent edges there are with higher weights. Thus, the low clustering of prominent property terms reduces the extent to which housing needs are satisfied in this subgraph. This subgraph is dominated by a few dwelling related terms and even fewer home property terms than the entire rental network. Moreover, the house terms are peripheral and disconnected from visible terms. These relationships reveal that only the basic structural aspects of housing are easily disseminated to the public. Thus, the territorial tendency is validated due to the ineffectiveness of selective marketing in this subgraph.

Figure 7.6 Prominent Terms/Edges of the Rental (Apartment) Subgraph

Like the all rental network, the most prominent out-degrees by weight (as shown in Appendix 23) stem from the dwelling terms: ‘bed’/’bedrooms’. Thus, market positioning is employed. The most prominent non-dwelling related terms can, however, be seen in Figure 7.7 below. However, there are even fewer non-dwelling vertices than in the all rental network. The only notable non-dwelling connected

175

vertices among the visible edges are: ‘shops’, ‘pets’, ‘transport’, and ‘city’. These terms communicate the importance of locational amenity within this subgraph, as well as issues surrounding rental restrictions or lack thereof regarding pets. Thus, the non- dwelling focus stresses external opportunities and constraints but lack internal ideals or external aspirations. The term ‘home’ is more peripheral than the term ‘bus’.

Figure 7.7 Top 30 Apartment - Rental Subgraph Terms by WDC Normalised Value

In contrast to the rental apartment subgraph, the rental house subgraph (Figure 7.8 on the next page) is the most centralised rental subgraph/network with a greater visible edge density in the central core of the graph. This subgraph’s normalised mean (0.296) is the lowest and farthest from the mid-range among the rental subgraphs/network (see Appendix 31). This means that this subgraph has more extremes, above and below the mean; thus, there are more prominent edges than the other rental subgraphs/network. Hence, the facilitation of housing needs satisfiers is relatively more balanced than the other rental subgraphs/network, although still insufficient. Though the ‘tenants’ property term is peripheral, it is well connected with dwelling property terms, which signifies a partial selective marketing of economic (house) related tenure. The balance between home and dwelling property terms is also more balanced relative to the other rental subgraphs/network, but the overall exposure to house terms remains imbalanced. Thus, the territorial tendency mechanism is validated because the peripheral terms housing needs satisfiers are isolated. Moreover, the non-integration of rental house terms in the advertising mechanism also substantiates the financial tendency mechanism.

176

Figure 7.8 Prominent Terms/Edges of the Rental (House) Subgraph

Regarding the specific edge relations of the rental apartment subgraph, the highest weighted edges are mostly between the ‘bedroom’ related terms; thus, market positioning is employed, and there are few relationships of significance. Instead, Figure 7.9 below, shows the most prominent non-dwelling attributes such as: ‘home’, ‘shops’, ‘pets’, ‘dining’, ‘schools’ and ‘family’, which are more informative. These locational/accessibility attributes mostly correspond with the apartment subgraph. Though, a clear distinction is the more central presence of home-related attributes.

Figure 7.9 Top 30 House - Rental Subgraph Terms by Normalised WDC Value

177

7.2.3 The Prominence and Semantics of the Entire Sales Network and its Typological Subgraphs Contrary to the rental market, the sales network has more efficient selective marketing, based on preliminary centrality and centralisation analyses in Section 7.1 and Subsection 7.2.1. Since causal mechanisms are corroborated based on the regularity of patterns, the importance of this subsection will then be to further understand if the territorial tendency mechanism is valid. This involves understanding if the sales network/subgraphs have a balanced representation of the home, house, and dwelling satisfiers. If these housing needs facets are balanced, it will corroborate the territorial tendency, which suggests that the peripheral isolation of housing needs reduces advertising efficiency. Moreover, the financial tendency mechanism, which suggests that an increase in financial focus explains advertising efficiency will also be tested. This premise can be tested by observing the prominence of house themes within the sales subgraphs/network. Firstly, the entire sales network (all sales) will be studied, then the typological subgraphs (house, apartment, and land) will be explored.

The all sales network is less centralised than the other sales subgraphs/network since it has fewer visible edges which account for a considerable portion of the normalised distribution (Figure 7.10 on the next page). This network had the lowest standard deviation across all subgraphs/networks. Generally, speaking, the lower the standard deviation, the more asymmetric the network is, and the higher the number of highly weighted edges there are likely to be. Hence, in contrast to the rental network, the sales network features central terms which are more complex. Regarding the composition of housing needs satisfiers, all facets of the housing ternary (home, house, and dwelling) are represented in this network. There are also identifiable niches within the network, such as a focus on land (‘land’, ‘block’, ‘price’, ‘dream’), location (‘unit’, ‘distance’, ‘affordable’), investors (‘bedrooms’). Thus, this subgraph targets a broad demographic, and diverse marketing is employed (market segmentation). Market segmentation considers heterogeneous customers, while the rental market appeared to focus on market positioning.

Dwelling attributes are the most focal vertices, but home-related terms such as ‘home’, ‘walk’, ‘park’, and ‘school’ are very prominent in the network. However, the terms ‘home’ and ‘school’ are not selectively employed (relatively speaking), since

178

relative to other property terms, their closeness centrality is lower compared to their weighted degree and eigenvector centrality.

Figure 7.10 Prominent Terms/Edges of the All Sales Network

This is likely due to market segmentation resulting in the conflation of terms across different housing typologies and geographies. This network also exhibits complexity in the form of house terms such as ‘potential’, ‘house’, ‘CBD’, ‘price’ and ‘investors’, which were largely absent in the rental subgraphs/network. Thus, the use- value and financial tendency mechanisms are substantiated, due to the market targeting of agents, investors and developers, which is reflected in the depth of housing needs in the network.

The depth of the sales network is also exhibited in the edge relationships (see Figure 7.11 on the next page). There is greater diversity in out-degree-in-degree relationships in this network. The most informative edges (show in yellow) are the: ‘home-dream’ (shown in Figure 7.10 in maroon), ‘home-block’, ‘unit-community’, ‘land- dream’, ‘home-flat’, ‘bedroom-investor’, ‘home-trees’ and ‘home-living’ edges. These edges and vertices communicate the importance and diversity of home in the sales

179

network. Home is associated with dwelling related terms such as ‘block’ and ‘flat’, but also to aspirational terms such as ‘dream’. Similarly, the relationship between ‘land- dream’ and ‘unity-community’ edges establishes an overlap between the dwelling attributes and the aspirational/cultural attributes.

Figure 7.11 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the All Sales Network Another way of dissecting the sales network is by typology. There are three typologies in the sales network: sales apartment, sales house, and sales land. The sales apartment subgraph (Figure 7.12 on the next page) is a centralised subgraph with extensive focal property terms and few isolated terms. The most prominent terms are ‘bed’ and ‘bedroom’, but the terms: ‘living’, ‘kitchen’, ‘location’, ‘park’, and ‘walk’ are also prominent, and altogether form the upper 50th percentile of WDC centrality. There are, however, some significant differences compared to the sales all network. The prominence of ‘home’ and ‘school’ is considerably lesser. Therefore, home-related terms are relatively more peripheral in this subgraph. Similarly, compared to the entire sales network, the term ‘land’ is considerably less prominent since the terms ‘apartment’ and ‘units’ place less emphasis on the land component. House terms such as ‘price’, ‘sought’, ‘unique’, ‘affordable’, ‘potential’, ‘investors’, ‘CBD’, and ‘house’ are some of the connected vertices in this subgraph. These terms reveal an aspirational facet that was absent in the rental apartment subgraph. Hence, the integration of

180

house themes substantiates the financial tendency observation. Also, the balance of housing facets validates the territorial tendency of increased housing needs satisfaction relative to the amount of marketing efficiency (control).

Figure 7.12 Prominent Terms/Edges of the Sales (Apartment) Subgraph

Figure 7.12 shows that all edges are either in the upper or the mid spectrum. The upper-range edges primarily stem from the dwelling vertices, but all housing facets are represented in this range. Moreover, while the most common edges are those which originate from the central terms towards the periphery, there are a few edges that cut across the subgraph, such as the ‘drive-driveway’ edge, though these are mostly derivatives of the same ideals (denotative, see Table 5.9).

Thus, Figure 7.13 on the next page is considered more insightful in terms of the implicature of the terms (see Table 5.9). Some terms signify market positioning (the quality of the product) such as ‘bed/bedroom-dream’, ‘bed/bedroom-ideal’, ‘bedroom- home’, ‘bedroom-sought’, ‘bedroom-modern’, ‘bedroom-quality’, and ‘kitchen-modern’ edges. However, some terms also exemplify market segmentation (the distinct characteristics of homeowners and investors) such as the ‘unit-community’, ‘bed- investor’, ‘bedroom-location’, ‘living-apartment’, ‘bed/bedroom-affordable’, and ‘living-

181

water’ edges. The ‘unit-community’ edge in particularly emphasises the appeal of apartments in a manner that envelopes broader typological and social characteristics.

Figure 7.13 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the Sales - Apartment Subgraph

The sales house subgraph (Figure 7.14 on the next page) is the second subgraph of the sales typological division. This subgraph is highly centralised, with many focal property terms, and few isolated terms. Dwelling terms such as ‘bed’ and bedroom are the key terms in this subgraph. However, the term ‘home’ is the third most significant vertex, which signifies the association of the home ideal with the house typology in particular. Other home-related terms such as ‘walk’ and ‘school’ are also more important in the sales house subgraph compared to the sales apartment subgraph. However, ‘location’ and ‘living’ are relatively less important.

Thus, overall there is a greater emphasis on home ideals in this subgraph, but it is tailored to the quintessential ideal of the Australian home. House terms were also more prevalent in this subgraph and included terms such as: ‘prime’, ‘potential’, ‘unique’, ‘sought’, ‘affordable’, ‘popular’, ‘unique’, ‘price’, ‘house’, ‘sale’, and ‘investor’. These terms generally had a higher normalised weighted degree centrality compared to the sales apartment subgraph. Thus, the financial tendency observation is upheld. This tendency posits that financial centric sales subgraphs/network such as the land

182

and house subgraphs have a higher clustering among their most important terms, and lower clustering overall because real estate marketing has an inherent financial bias.

Figure 7.14 Prominent Terms/Edges of the Sales (House) Subgraph

Thus, the representation of individuals’ housing needs satisfiers appears to be more balanced in this subgraph. Regarding the territorial tendency, this subgraph is one of the outliers (Figure 7.3). The number of visible edges is greater relative to the amount of overall clustering. However, as shown in Appendix 27, this is because the lower weighted vertices have a higher clustering coefficient than normal. Conversely, the highest weighted vertices have a clustering coefficient that is higher than the sales apartment subgraph. This suggests that the sales house subgraph is a broader subgraph/network than most, and market segmentation is employed in this subgraph to a greater extent than other subgraphs/network. This assessment will be further evaluated based on the edges shown in Figure 7.15 on the next page.

183

Figure 7.15 All Visible Out-degrees and In-degrees (Showing Edge Widths) for the Sales – House Subgraph

‘Home’ was the most important term in relation to the terms: ‘island’, ‘living’, ‘modern’, ‘location’, ‘walk’, ‘established’, ‘prime’, ‘ideal’, ‘tree’, and ‘dreams’ etc. Other home-related out-degrees include ‘living-island’, and ‘unit-community’. The significance of these home out-degrees is that they reveal different niches based on locational and financial aspirations (‘growth’, ‘ideal’, ‘affordable’, ‘investors’, ‘potential’ and ‘price’), as well as product quality. The latter is exemplified with terms such as ‘quality’ and ‘prime’. Thus, the premise that market segmentation (catering to heterogeneous groups) is used in this subgraph is further substantiated.

The last subgraph of the sales typological division is the sales land subgraph (Figure 7.16 on the next page). This is the most centralised subgraph (no vertex is completely isolated), meaning that this subgraph’s vertices are more related to one another than in other subgraphs/networks. This subgraph’s standard deviation reveals that most edges are close to the mean, although it has outliers with an edge width > 5. However, even non-outliers are more prominent relative to other plots. This relationship signifies that this subgraph is significantly contextualised, with specific relationships between ideals, rather than generalised and uniform co-occurrences.

184

Figure 7.16 Prominent Terms/Edges of the Sales (Land) Subgraph

The dwelling terms: ‘block’, ‘land’, and ‘location’ are the most focal vertices and have a weighted degree centrality that is considerably greater than all other terms. However, home terms such as ‘home’, ‘school’, ‘park’, ‘water’, ‘tree’, and ‘drive’ are the next tier of terms in importance. This is because, although this subgraph contains vacant addresses, the potential home or prospective house-land packages are advertised. This relationship is also revealed by relatively prominent house terms such as ‘dream’, ‘ideal’, and ‘unique’. These vertices are more prominent in the land subgraph than in the sales house subgraph. This represents the unfulfilled potential of raw land (Subsection 4.2.2).

Thus, this subgraph broadly satisfies housing needs. Hence, the territorial tendency observation is substantiated, with the caveat that, while vertices such as ‘home’ and ‘dream’ are more prominent in this subgraph than in most subgraphs/networks, inferences are being made by real estate agents, due to the prospective nature of this subgraph. Regarding financial tendency, although house terms are not more prominent in this subgraph than in the other sales subgraphs/network, the most prominent dwelling terms of this subgraph: ‘land’, ‘block’, and ‘location’ are the basic building blocks of geographic territoriality.

185

Figure 7.17 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the Sales - Land Subgraph

186

The dwelling terms can be further explored through their specific edge relations. Figures 7.17 on the previous page shows that an upper spectrum cluster comprised ‘water-luxury’, ‘land-luxury’, ‘block-luxury’, ‘Island-luxury’ and ‘land-island’ edges. This cluster can be simplified as a triangular relationship between ‘land-island-luxury’. This signifies that the highest weighted outliers in this subgraph vary from the rest of the subgraph. This is because the prominence of the term luxury is misleading. The term does not feature as a visible out-degree in Figure 7.17, rather it consistently features as an in-degree, because seven terms are considerably more associated with the term, relative to all other terms. Hence, this cluster does not signify prominence, though it does exemplify market segmentation. Market segmentation is the key selective feature of this subgraph since land is differentiated based on price, and aspirations (‘future’, ‘affordable’, ‘unique’), as well as quality.

7.2.4 The Prominence and Semantics of the Sales Geographical Subgraphs The final means of understanding the sales network is through geographical subgraphs, of which there are three: sales inner, sales middle, and sales outer. Subsection 7.2.1 has shown that the sales geography subgraphs have a lower propensity for information transfer, relative to the sales typological subgraphs. This subsection will then be useful to expand on the characteristics of that relationship, in terms of both the territorial and financial tendencies. Additionally, this subsection will uncover if the dissemination of property ideals has a spatial/geographic element. This dissemination, which is also known as market feedback, will be assessed based on the subgraphs’ housing needs compositions. This composition will be examined in the form of social network graph plots which highlight the weighted degrees of vertices and edges, as well as through out-degree and in-degree prominence of said edges.

The first geographical subgraph is the sales inner subgraph (Figure 7.18 on the next page), which is a largely centralised subgraph. Nevertheless, there are still some isolated property terms, particularly house related terms. Most edges emanate from the central dwelling property terms of ‘bed’, ‘bedroom’, ‘kitchen’, and ‘location’. Hence, this subgraph bears many similarities with the sales apartment subgraph (see Figure 7.13). However, key differences include the relative increase in prominence of the terms: ‘home’, ‘city’, ‘CBD’, ‘trees’, and ‘land’. This relative environmental diversity reflects the fact that the Brisbane Inner City SA4 still extends over 10 kilometres in some directions, notwithstanding its compact size (see Figure 5.7). The most

187

prominent home terms in this subgraph are ‘living’, ‘park’, ‘walk’, and ‘home’, while the most prominent house terms are locational in nature: ‘Brisbane’, ‘city’, and ‘CBD’. Thus, implicature (what agents want individuals to conclude) is employed in this subgraph. Hence, this subgraph exemplifies market targeting (a segment of the population is targeted). The degree to which this subgraph satisfies housing needs is proportional to its level of market control (efficiency), which substantiates the territorial tendency mechanism. Moreover, the relative obscurity of the house terms relative to the overall selective marketing also validates the financial tendency mechanism.

Figure 7.18 Prominent Terms/Edges of the Sales (Inner) Subgraph

Regarding edge relations, there is little differentiation (no outliers) since they are all within the upper spectrum. This is also shown in Figure 7.19 on the next page. The most discernible feature of the figure is that most visible out-degrees are the terms ‘bed’ and ‘bedroom’, though the term ‘kitchen’ also features to an extent. In terms of in-degrees, there is more variation. The prominent non-dwelling edges are: ‘home- dream’, ‘unity-community’, and ‘CBD-Brisbane-CBD’ (though this is denotative). These edges reveal that there is some degree of market segmentation. However, the dominant selective strategy is market targeting based on locational attributes.

188

Figure 7.19 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the Sales - Inner Subgraph

The sales middle subgraph (Figure 7.20 on the next page) is a partly centralised subgraph since a third of its property terms are disconnected. The subgraph still features a diverse range of out-degrees and is not dominated by the bed and bedroom vertices as is the case in most subgraphs/networks. Instead, out- degrees such as ‘home’, ‘land’ and ‘kitchen’ feature prominently. The ‘home’ vertex, in particular, is the most prominent property term in this subgraph, making it one of only two subgraphs/networks where ‘home’ is the leading term. The term ‘land’ is similarly more prominent in this subgraph than in all subgraphs/networks with the exception of the sales land subgraph. The term ‘location’ also has a higher weighted degree centrality than the sales inner subgraph, although relative to its position in the subgraph it has a lower rank. Similarly, the ‘school’ and ‘shopping’ terms have higher weighted degree centralities compared to the sales inner subgraph. Thus, in this subgraph, location relates to localised amenity more so than city-wide amenities. This is because the sales inner subgraph relies more on implicature – the Brisbane CBD term evokes amenities that are implicit. There is, however, no equivalent to the prominence of ‘home’ and ‘land’ in the sales inner subgraph. This divergence communicates one of the core facets of the Great Australian Dream, which is that an increase in the prominence of ‘land’ is often associated with the legitimacy of ‘home’.

189

Figure 7.20 Prominent Terms/Edges of the Sales (Middle) Subgraph

Based on its low overall cluster coefficiency, this subgraph would be expected to have one of the highest capacities for market feedback. However, this subgraph is one of the outliers shown in Figure 7.3. This subgraph’s visible edges are lesser relative to its overall clustering. As shown in Appendix 27, this is because the lower weighted vertices have a lower clustering coefficient than normal. The highest weighted vertices in this subgraph are however higher than the rental network. This dynamic means that the sales middle subgraph is asymmetric, with relatively obscure terms (such as the house vertices), but with relatively more selective focal relations.

Thus, based on the edge relations shown on the next page in Figure 7.21, it can be concluded that market targeting (focused on a home appeal) is employed in this subgraph. The most relevant edges in Figure 7.21 are the ‘home-dream’, ‘block- flat’, ‘land-flat’, ‘block-dream’, ‘land-dream’, ‘unit-community’, ‘land-future’, ‘home- established’, ‘home-trees’, ‘land-price’ and ‘home-future’ edges. These relationships highlight the significance of ‘home’, ‘land’, and ‘unit’ to a lesser extent. However, the in-degrees of ‘community’, ‘dream’, ‘future’, ‘established’, and ‘flat’ are insignificant.

190

The prominent out-degree vertices (the most important terms among all terms occurring with an in-degree) however clarify the idealism of this subgraph.

Figure 7.21 All Visible Out-degrees and In-degrees (Showing Edge/Tie Widths) for the Sales - Middle Subgraph

The last sales geographical subgraph is the sales outer subgraph, shown in Figure 7.22 on the next page. The subgraph is partly centralised since it is largely disconnected with relatively few focal vertices. Peripheral terms include those in an edge-induced subgraph between the vertices: ‘CBD-Brisbane CBD-Brisbane-Gold Coast’. These locational-related terms are isolated from the rest of the graph (based on the weights). The most prominent term is ‘home’, making this the only subgraph/network alongside the sales middle subgraph to have a home term as the leading vertex. The other prominent home terms are ‘living’, ‘walk’, ‘school’, ‘park’, and ‘shopping’, which are terms also prominent in the sales middle subgraph, however, ‘school’ and ‘shopping’ are even more prominent in this subgraph. Nevertheless, the term ‘drive’ has a higher weighted degree centrality in this subgraph than in all subgraphs/networks, which suggests that the amenities are not localised, but rather city-wide attributes. This means that housing needs satisfiers that are represented beyond the neighbourhood level may be embellished using fuzzy terminology (see Table 5.9). For instance, the dwelling terms ‘location’, ‘distance’, and walk have a higher normalised weighted degree centrality than in all other subgraphs/networks.

191

Figure 7.22 Prominent Terms/Edges of the Sales (Outer) Subgraph

Owing to this subgraph’s average centralisation, the house vertices are also considerably prominent. The terms: ‘CBD’ and ‘Brisbane’ are prominent out-degrees, and terms such as ‘unique’ and ‘sought’ feature as in-degrees. More importantly, the terms ‘Brisbane’ and ‘house’ have a higher normalised WDC than in other subgraphs/networks. Moreover, peripheral vertices such as ‘sought’, ‘affordable’, ‘investors’, ‘unique’, and ‘potential’ relate to the central vertices, whereas home terms such as ‘shopping’, and ‘transport’ are disconnected. This is unusual because, in other subgraphs/networks, an increase in the prominence of peripheral house edges also coincides with an increase in the prominence of peripheral home edges. Furthermore, home terms are some of the leading vertices in this subgraph, and the term ‘home’ is almost as related with house terms as it is to other home vertices (see Figure 7.23 on the next page). Also, the fact that ‘shopping’, a term with a high normalised weighted degree centrality is detached, means the term is used indiscriminately. Hence, this subgraphs’ locational amenity is fuzzy and generally employed, while home-home and home-house edges embody market targeting. ‘Home-home’ edges are ‘home-dream’, ‘home-living’, ‘home-trees’, ‘home-quality’, ‘home-water’, ‘home-modern’ while home- house edges are ‘home-potential’, ‘home-unique’, and ‘home-sought’.

192

Figure 7.23 All Visible Out-degrees and In-degrees (Showing Edge Widths) for the Sales - Outer Subgraph

Due to this subgraph’s average centralisation and clustering, the depth of its housing needs satisfiers are average and focused on market targeting. This pattern validates the territorial tendency mechanism. Regarding the financial tendency mechanism, the highest clustered terms are the home and house terms, while the rest of the subgraph has a low clustering coefficient. This selectiveness suggests that this subgraph is aligned with the market’s financial imperative. Geographic territoriality is also shown in how fuzziness engenders placelessness (indiscriminate language use).

7.2.5 Subgraph/Network Boundaries: Spatial Adjacency/non-adjacency The previous subsections have outlined that some of the subgraphs/networks have satisfied housing needs based on localised attributes (spatial adjacency), while some have been more reliant on city-wide attributes (spatial non-adjacency), see Subsection 5.4.1. Figure 7.24 on the next page shows which subgraphs may be adjacent (relating to the same submarket). Spatial adjacency is a combination of spatial continuity, substitutability, and similarity. Spatial continuity means that a submarket occupies a continuous spatial space (C. Wu & Sharma, 2012). Substitutability means that price, location, structure, and neighbourhood quality similarly contribute to prices. Lastly, housing attributes must be composed of similar quantities and type (Y. Wu, Wei, & Li, 2020). Submarket adjacency can be observed by evaluating the similarity of adjacent neighbours (Mennis & Guo, 2009). Subgraphs/networks with similar clustering share similar meaning/attributes. Spatial non-adjacencies are expected in the typological subgraphs and sales/rental networks since they are geographically dispersed in the Brisbane GCCSA. This dispersal partly explains the market segmentation of the sales

193

network and typological subgraphs. Though marketing strategies are more influenced by financial and territorial tendencies. For example, the rental market which is less aligned with the market, emphasises product differentiation.

Figure 7.24 Boundaries of Four Market Types defined by Clustering and Semantics Conversely, the sales geography subgraphs reflect spatial boundaries. House (economic) terms were gradually more focal from the inner to the middle, and outer subgraphs, even though in absolute terms, exchange-value increases towards the CBD (Section 3.8). Thus, there is an inversion of the importance of use-values (Section 4.2). Market targeting was used in all sales geographic subgraphs, but the inner subgraph emphasised location, while the middle and outer subgraphs emphasised home appeal. Thus, these subgraphs were generally more defined and indicative of discrete submarkets. This is especially true for the sales inner and middle subgraphs (circled green). However, the sales outer subgraph (circled red) reveals some area-type constraints: the fuzzy terminology used means that needs are satisfied based on attributes outside of its spatial boundaries. These patterns have shown that the housing market’s focus has influenced how needs are satisfied in the housing market. Thus, the housing ladder captures financial and territorial tendencies, which may not reflect how best needs can be satisfied.

7.3 Community Measures and Social Network Analysis Conclusion

The previous network graphs and out-degree-in-degree figures have uncovered the individual relationships between vertices, as well as their prominence. This section will

194

reveal how those individual vertices function together as distinct communities within the wider network, as well as provide an overall conclusion for the implications of market feedback. Communities embody the locational, amenity, typological, cultural and aspirational themes within housing networks. Thus, identifying how networks are segmented reveals how well they facilitate individuals’ housing needs satisfiers. The community measure used in this section is the multilevel algorithm (specified in Subsection 5.4.4). This measure reveals differences in the density of network clusters (number of edges to weights relation). The density of edges reflects the spread of information in the real world due to the Gricean Maxims (Section 5.4), which explain that real estate agents will maximise their property descriptions to express market information; the construction of that information varies in relation to the context of the application. That variation is reflected in the densities of edge relations.

Figure 7.25 Multilevel Communities of the All Rental Network

Subsections 7.2.1 and 7.2.2 highlighted the narrowness of the rental market since the most prominent rental terms inefficiently conveyed housing needs. Also, there were limited edge relations involving financial and aspirational themes. This narrowness is further substantiated in Figure 7.25 above, which shows a multilevel community of the rental network. There are two communities: the locational-amenity

195

group (community 1) and a typological-place group (community 2). The overlap between the communities reflects the force-directed placement of the most prominent property terms. In other words, from a community standpoint, the overlap is only visual. However, analytically the overlap is non-trivial since the intersection between a vertex such as ‘modern’ is a product of its comparable centrality with property terms such as ‘home’ and ‘kitchen’. More importantly, Figure 7.25 represents that the rental network is not nuanced. For instance, there is no distinct group for housing aspirations. The extensive emphasis on typological and locational terms reflects the market positioning strategy. As mentioned in the previous section, this strategy is driven by where a property sits in relation to other properties, more so than how well the property corresponds with various market segments or a particular demographic.

Conversely, the sales network (Figure 7.26 on the next page) shows a greater complexity of definitions and concepts. There are three communities: a locational- economic-amenities group (community 1), a typological-place-amenities group (community 2) and a symbolic-lifestyle group (community 3). Thus, there are two forms of amenity clusters. Community 1 refers to locational amenity, while community 2 refers to dwelling amenity. The central term ‘home’ functions as an intermediary term bridging other key dwelling attributes such as ‘kitchen’, ‘bedrooms’, and ‘living’ with the symbolic-lifestyle elements and the locational elements such as ‘location-unit’ and ‘school’.

Moreover, the community groups of both networks also represent market feedback, which has also been explained in the form of clustering (density), closeness centrality, eigenvector centrality, and weighted degree centrality (measures degree of prominence). The centrality measures underscored the use-value of the two networks and eight subgraphs. The territorial tendency (how geography is imagined through emotional ends) communicated how the satisfiers of housing needs were facilitated by the market, while the financial tendency mechanism compared the notion that sales and extensive growth are more leveraged in real estate advertising than other interests. Thus, it was also apparent that agents in the market employed various strategies such as fuzziness, implicature, and inference to strategically present housing products. Thus, the property terms unearthed in Section 7.2 cannot be framed as a 1:1 representation of how individuals’ housing needs and aspirations are satisfied. However, they do reveal causal mechanisms of the market.

196

Figure 7.26 Multilevel Communities of the All Sales Network

Since causal mechanisms reveal events (actual) that are likely to emerge in a network and how people position themselves in society. Accordingly, Chapter 7’s findings represent the tendency for rental and sales housing products to reflect housing needs, financial motivations, and aspirations. This causal mechanism does not suggest that the market targeting employed in the sales geographic subgraphs, the market segmentation that is chiefly employed in the sales all, and sales typological subgraphs, and the market positioning employed in the rental network/subgraphs are deterministic. A renter may experience a holistic housing experience, while a homeowner may not. Individuals have agency, and their empirical (lived) experiences suggest that they negotiate numerous causal mechanisms in a manner that is not entirely knowable. However, as mentioned in Section 2.1, it is also true that individuals themselves sustain causal mechanisms because insofar as a mechanism partially adheres to an individual’s beliefs, the person can map his or her internal aspirations onto the external world. Nonetheless, there is much variation in how well individuals’ needs are homologous with normative beliefs. The stronger feedback between owners and the market suggests that the tendency for corresponding needs and outcomes should be higher. This nexus will be further investigated in the subsequent chapter.

197

CHAPTER 8: DISCUSSION The thesis has sought to answer the main research question, which is: how well are individuals’ subjective housing needs reflected in their consumption patterns?”. Multiple domains of truth were researched to answer this question. Empirical truths based on individuals’ prioritisation of housing needs, actual events, and finally, causal tendencies (financial and territorial), were all uncovered. These truths were reconciled using methodological pluralism. Geographic patterns revealed consistency in extensive development (low-density/outer suburban growth). Likewise, semiotic patterns revealed a gradual increase in economic marketing the further away from the CBD. Pluralism was also realised based on the period of activation. Section 3.4 showed that circa 2011 the usage of house-related terminology increased, a period which coincided with the increase in the value of residential land relative to GDP (Unconventional Economist, 2018). Lastly, pluralism is realised based on how causal tendencies were activated. The study of attitudes showed that public territoriality did not wholly resonate with domestic needs. Individuals primarily yearned for enhanced dwelling conditions to address life-course needs. However, the implications of the current growth model (neighbourhood changes, housing costs and regulatory uncertainty), appeared to disrupt their aspirations. Therefore, the way in which individuals sustained the housing market was based on conforming loosely to normative beliefs which counteracted other aspects of their aspirations.

These relationships are important because this thesis has also argued that housing markets are not self-regulating. The market is not self-regulating due to information asymmetries (exemplified in the SNA marketing), monopolies, transaction costs and a general competitive imbalance (Helbing, 2013). The implication is that needs are not implicit. Different domains have different intentions. Housing products are developed based on rough feedback between consumers and developers. Developers look to what has been successful in the immediate past to develop housing products (E. D'Arcy & Keogh, 1998), while what works from the consumer’s perspective is largely defined by the reality that there is no substitute for housing. Individuals have their own needs, but their ability to seek out those preferences in the housing market is reliant on their level of awareness and financial capacities. Moreover, the crux of the matter is that since the delivery of housing needs is contradictory, the success of the housing market is not simply reliant on how well

198

individuals and investors stake their claims, but on how well their disparate objectives are coordinated. The levers of government cannot be understated as the necessary mechanisms for coordinating the capitalist imperative with social equity. Neighbourhood planning schemes, city masterplans and other assessable planning schemes play a key role in this process. This is because even when developers look towards housing outcomes that are innovative and robust, the feasibility is limited from the onset by how well infrastructure, employment and complementary land-uses have been organised.

However, even municipalities, state/federal governments and their various apparatus may have motives that contradict individuals’ aspirations. Thus, the complexity of the housing market means that the barriers which constrain housing needs can emerge just as easily as the opportunities which facilitate them. Moreover, developers are more likely to sustain existing growth models than innovate, since the level of risk they are willing to absorb is predicated on how well community groups, authorities and other stakeholders can balance environmental, political and economic costs in the first instance. Thus, the coordination of disparate housing motivations is largely an iterative process, bereft of a ‘common substance’ that can situate market outcomes within the context of individuals’ subjective housing aspirations.

This thesis has made a case for such a conduit. Though researchers such as Badcock (1994) have found the conceptualisation of housing careers and the housing ladder to be problematic due to the fact that the life-course is dynamic, it is useful to consider the universalist route as well, and consider what binds the seemingly irreconcilable housing trajectories together. What is the ‘common substance’ that unites the production side of housing with the demand side, without conflating outcomes with preferences, or conflating actions motivated primarily by exchange- value, with those motivated primarily by use-value? That conduit is based on defining housing needs satisfiers. These satisfiers revealed the dynamism of individuals’ subjective housing priorities (composition) in relation to structured housing market themes and outcomes.

With the composition of individuals’ needs established, it was then possible to determine the more relevant processes that underpinned individuals’ housing trajectories. Similarly, in terms of housing market outcomes, it was important to

199

determine how well the real estate property terms conveyed housing needs. This was key because due to feedback, this would imply how well housing needs were balanced. These findings help to redefine the concept of housing progression, help to place individuals’ subjective aspirations within market outcomes, as well as reveal if the territoriality of individuals within the Brisbane GCCSA is compatible with that of the public. These topics will be discussed in the subsequent sections.

8.1 Redefining how Progression is conceived in the Housing Market

The purpose of this section is to situate this thesis’ findings on housing progression within the broader housing market literature. This dissertation represents a departure from the existing housing market outlook since housing needs are not implicit in housing outcomes. Therefore, the basis for understanding a successful housing trajectory is built on the basic principles that drive metamotivation rather than the housing ladder. Metamotivation is the longing to be in harmony with oneself and is shaped by the need to protect oneself from external forces, cultivate social interactions and eventually attain self-actualization, by gaining command over one’s domestic domain. This is the underlying framework that renders needs fundamental. Though needs can be satisfied in many ways, the role of housing as a satisfier of needs is crucial, because housing has become a cornerstone of overall life-course expenditure. Housing has become the primary means of consumption, and the primary means of satisfying needs in physical form.

Therefore, the housing ladder construct of typological ascension, or the notion that the most significant hurdle faced by individuals is participation in the housing ladder, underestimate the importance of fundamental needs. Typological ascension is the assumption that housing transactions will reflect progressive moves between rungs, from rental housing to purchased property, and larger, more upmarket housing later in life. Though this thesis’ data does support the gradual increase in the prioritisation of house sizes by age cohort, the data has revealed that improvements in housing conditions cannot be effectively gauged by the number of housing transactions. Conversely, it is the capacity to make appropriate housing decisions (the absence of life-course disruptions, for example) that is relevant. Moreover, the thesis found that some of the highly valued attributes by the owner cohort (such as tenure, cost, and adaptability), are factors that would be equally prioritised by renters under

200

the appropriate regulatory conditions. For example, the survey profile shows that owners represented a family lifestyle, but the uncertainty that disrupts family-living under rental conditions are largely regulatory in nature.

Moreover, this thesis does not use a single characterisation of the housing market to define aspiration since this approach produces a partial perspective. This is because fixating on housing attributes presupposes the dwelling types or tenure that is required. For example, Badcock (1994) explains that ownership can simultaneously enable entrapment and enrichment for individuals and households. Additionally, with regards to downsizing among the elderly, is not necessarily a down move, if it is planned, furthermore, changes in typology might be driven by location rather than an attachment to a particular house type. Findings from the Brisbane GCCSA HNS also showed that both renters and buyers prioritised housing suitability above house types. This shows that the underlying emotional ends (the needs that the house meets) are more crucial than housing attributes (how the housing product is represented).

However, if specific house types are associated with desirable outcomes based on a single housing dimension (such as a capital gains imperative), then underlying emotional ends may be marginalised. For example, the HNS showed that the anticipated future value of housing was more prioritised than the structural integrity of homes. This is consistent with findings from Unconventional Economist (2018) that show that the value of residential structures relative to GDP has been constant for at least 30 years, while the value of land relative to GDP has soared. Thus, it is important to understand that housing progression (as it is currently framed) does not necessarily imply that all the critical aspects of housing are enhanced. The HNS has shown that aspirations are a product of independent factors that are often related yet remain distinct. These include capital gains and typology but are not limited to these qualities.

Hence, the comprehensive list of 27 attributes was subject to cumulative and clustering analyses to determine whether graduation was based on residential mobility or age; the findings concluded that age was more representative of progression in the Brisbane GCCSA housing market than mobility. However, it is important to note that the K-means clustering showed that the enhancement of housing priorities based on age was less evident for the renter cohort. The rental age cohorts’ priorities were more random and varied. Still, the general inefficiency of residential mobility is even more

201

problematic because when individuals with considerable housing moves (in excess of the mean), make more than the projected 1-2 moves (Bernard et al., 2017) between the ages of 18 and 29, they are not necessarily making much progress in the market.

Progression in the housing market is better characterised by housing transactions that are made during significant junctures in an individual’s life; these junctures come with increased financial capacity and valuable experience. Individuals with frequent housing transactions in a short period of time are likely to represent one of the following groups: individuals with disruptive life-course experiences, or individuals with speculative investment in the housing market. The former is doubly problematic because it suggests weaker underlying capacities to financially progress in the housing market, in addition to the volatility that is likely to diminish or disorientate the immediate satisfaction of housing needs. Individuals with low-income or disabilities are more prone to uncertainties which limit their expression, and these uncertainties, in turn, complicate the housing narratives that they have established. On the other hand, investor buyers who dominate the private rental market are likely to be financially secure.

8.2 The Nexus Between the Facilitation and Prioritisation of Housing Needs Satisfiers

This section investigates the intersection between individuals’ housing needs satisfiers and how those same needs have been facilitated in the housing market. The former was determined through the implementation of the Brisbane GCCSA Housing Needs Survey (HNS), while the latter was established following the implementation of an SNA of the Brisbane GCCSA housing market. This mixed-methods approach revealed the processes that best characterised the accumulation of housing needs, the attributes that were consistent across cohorts and those that were polarising.

Regarding the satisfiers of housing needs, one of the critical considerations was how best to define aspirations. In the background of this dissertation, aspirations were described as the intersection between housing needs satisfiers and external barriers and opportunity. Therefore, there were two considerations during analysis. The first was that aspirations are more than just the composition of needs; aspirations must relate to a trajectory (a pattern of change). The second consideration is that the factors that trigger change must not be confounded with aspirations, because they represent both aspirations and compulsion. Therefore, the fact that renters prioritised life-course

202

factors to a greater extent than the owner cohort reflects both the desire for life-course flexibility (opportunities), as well as housing compulsion/disruption. Faulkner and Beer (2011) find that compulsion or precariousness does not only relate to the vulnerability of tenure but can encompass all aspects of the life-course, including finances, health, and employment.

The life-course attributes are, however, useful to understand how mobility is being triggered. From the open-ended question, it was evident that life-course events influence the prioritisation of the dwelling-based attributes. When people feel that they are working towards an ideal, they associate aspects of dwelling, such as house-size, privacy, and the locational features with that ideal. This finding is not new to the body of literature (Feijten & Mulder, 2005). What remains to be seen is how people can navigate and realise the ‘ideal’.

However, though it is assumed by some housing needs researchers that life- course changes are necessary to spur continued housing progression (Faulkner & Beer, 2011), the relationship between life events and the relevance of various neighbourhood and dwelling needs is much more complex. The data has shown that the core social-dwelling needs are prioritised consistently regardless of changes in the life-course triggers. This reflects the fact that those social-dwelling needs are universally applicable, as well as the notion that higher-level needs are required for self-actualization.

However, social-dwelling and financial prioritisation were generally dependent on life-course prioritisation. While life-course factors present opportunities for residential mobility or improved housing conditions, an increase in life-course events only corresponds with the progress of social-dwelling and financial aspirations if life- course events are not inhibiting. For this to happen, there must be an underlying financial capacity, and there must not be a significant increase in life-course events, as it implies volatility. Life-course events are useful in two ways: they inspire opportunities for change, and they produce a template for change. Life-course circumstances that are appropriately addressed transform into lifestyle changes, while those that are unexpected, and overbearing represent compulsion.

Therefore, aspirations and compulsion are distinct, and the prioritisation of life- course factors by the renter cohort is not inherently positive. It was found that the

203

individuals with the least amount of housing market knowledge nominated the greatest life-course relevance, while those with the greatest social-dwelling relevance considered themselves to have the greatest housing market knowledge. The prioritisation of dwelling attributes in this instance represents the formation of an aspirational trajectory, while the prioritisation of life-course attributes is simply indicative of constraints and opportunities. The data showed that while life-course events may be planned for, they may also represent compulsion.

In terms of how individuals have been able to mould their housing experiences, the most consistent predictors were the core social-dwelling attributes, since these factors were prioritised above housing costs and because they were not particular to any group. This is striking because the renter cohort had lower prioritisation of these attributes compared to the owner cohort. Moreover, the renter cohort also had lower prioritisation of specialised financial instruments. This suggests that the housing priorities of renters are less developed. This is revealing because this data is supported by the market-derived data from Chapter 7. While the sales subgraphs/network had a relatively uniform distribution of the dwelling, home, and house concepts, for the rental market fewer than 10% of the concepts could be associated with the house facets; the dwelling component accounted for 50% of the concepts employed.

The renter cohort was also presented with a narrower band of information on amenities and lifestyle facets. This restriction suggests that the market insufficiently captures renters’ feedback. This suggests that the public’s territoriality has discounted some of renters’ aspirations. The capacity to adapt a rental property or the capacity for more stable tenure are absent in the market; thus, renters lack the capacity to express these housing aspirations. The absence of these features in the market, then further reinforces the normative beliefs about the housing ladder, and influences renters’ decision-making. Though based on the findings from the qualitative open- ended passages, this does not imply that renters forego these aspirations while renting; instead, the uncertainty becomes an impediment that stimulates inefficient housing decisions.

Moreover, by gauging the distribution of the rental and sales subgraphs/networks outlined in Subsections 7.2.2 to 7.2.4, it was found that the rental

204

network from a centrality point of view had too few terms dominating the relationships between property terms, while from a community standpoint, the multilevel algorithm highlighted a rental network lacking in bridging terms. Bridging terms being property terms that connect the disparate parts of the networks. Specifically, in the rental community graphs, there were two distinct communities: a locational-amenity group (LA) and a typological-place group (TP). The LA group was more abstract and related to symbolic or cultural boundaries such as the city, CBD, as well as amenities such as ‘cafés’ and ‘restaurants’. The TP group was more defined and related to terms such as the ‘bedrooms’ within the house, the ‘yard’ and locality. There was very minimal overlap between these community groups.

Conversely, for the sales network, there were three distinct communities: a locational-economic-amenities (LEA), typological-place-amenities (TPA) and a symbolic-lifestyle group (SFG). There were significant differences from the renter community. The LEA group had a house facet incorporated, with terms such as ‘growth’ and’ sales’, as well as had service-related amenities such as shopping and schools. The TPA group had a distinct form of amenities (cultural) with terms such as ‘cafés’ and ‘boutique’s, while there was an additional group (SFG) which featured aspirational concepts such as ‘future’, ‘dream’, and ‘home’. However, while the sales community was more nuanced, there was a tight overlap between bridging terms such as ‘home’ (SFG), ‘bedroom’ (TPA) and ‘location’ (LEA). The fact that the sales network consisted of language with greater depth relative to the rental network reveals that aspirational ideals have been coupled with homeownership. This is related to the growth trajectory and permanence of homeownership. This means that the rental cohort are afforded a less discernible lifestyle, by virtue of their tenure, and it also means that even where such opportunities exist; since these attributes are not prioritised through market feedback, renters may be less privy to their opportunities.

Naturally, the way in which the rental and sales markets function is distinct, the latter consists of ‘active’ investors, homebuyers, while the former consists of ‘passive’ renters. The active cohort is tasked with going beyond their level of familiarity and experience, forming new experiences as they scour the housing market. The passive cohort is relatively more constrained by social mobility and their personal networks. Information is one of the critical elements that help shape the housing market. It is required for renters, owner-occupiers and investors to make informed decisions about

205

their present and future needs, but this information is not always forthcoming (E. D'Arcy & Keogh, 1998). Renters especially, do not obtain the required amount of information from pre-rental inspections, partly, due to time constraints (Ballard, 2017). Furthermore, even when information is sufficient at the time of the transaction, findings have suggested that many housing complications occur in the middle of tenure. While it is presently improbable for the market to adjust for such uncertainties, it is expected that factors with a greater degree of permanence should be reflected through feedback in the housing market. These factors include the depth of neighbourhood, dwelling and locational facets, as well as insights into amenities.

What the findings have suggested is that there is a connection between how housing is consumed and the information that is relayed. Therefore, the unsupervised SNA communities (Section 7.3) discussed above corroborate the premise of Harris (1954) that words that occur together share meaning. This tendency revealed that housing presented for sale is marketed with information about financial and growth opportunities, while undeveloped land is devoid of information on amenities. This appears to be straightforward and unproblematic. However, in the market, goods are not valuable because they have intrinsic saleability, rather they can be sold because they are of use to people. Therefore, many attributes that are representative of financial purposes also simultaneously represent social uses. The uncertainty of rental tenure and the relative absence of financial applications has therefore eliminated or rendered less visible, information that is of practical use to renters. For example, schools were significantly less marketed in the rental market. Thus, it appears that how people partake in the housing market has a significant bearing on their outcomes and perceptions, independent of other factors. There are fundamental differences between renters and owners’ territorial claims, which will be investigated in the next section.

8.3 The Conflict of Divergent Territorial Claims: Synthesising the Significance of Individuals’ Needs and External Factors

While the differences between the domestic and public facets of geographic territoriality were outlined in Table 3.1 in Chapter 3, the findings from Subsections 6.2.1-6.2.3 (Research Questions 2-3) have revealed that the perception of conflict in the housing market is framed not simply through the tension between domestic and

206

public territoriality but through the lens of different forms of public territoriality. The historical expressions of public territoriality (growth with a homeownership imperative) were pitted against the modern adaptations of public territoriality (growth with a financial imperative), just as significantly as the domestic territoriality involving the control of rental tenure was pitted against the public territoriality of homeownership and investment.

Regarding the divergence between public and domestic territoriality, Subsections 7.2.2-7.2.5 showed that how geographic space was imagined was constructed based on homeowner ideals. Renters’ needs were not simply inadequately captured, as explained in the previous section, but ideals within the rental network were indiscriminately employed. Thus, the network is imagined as a transitory geographic space for renters, while for homeowners, the basis for normative beliefs such as the housing ladder and the Great Australian Dream, were well defined.

Another factor in the private versus public contention of social-political space is the divergence in power between landlords and renters. The limitations of rental regulations, particularly regarding long-term tenure and how residents were able to modify their housing, was frequently raised in the responses. Tenure was problematic not simply because of the short length of tenure, but due to the general unpredictability of tenure, which involved the possibility of evictions as well as renters being unable to pre-empt the potential sale of their homes. Therefore, the unpredictability of rental tenure, as well as the regulatory restrictions, mean that the premise of rental tenure being the lowest rung of the housing ladder is largely a self-fulfilling prophecy, since the unpredictability and restrictions are directly tied to the market’s priorities. This represents the significance of external factors on individuals’ rental experiences.

Contrariwise, the role that individual needs have on the private-public contradiction is largely influenced by family dynamics. This is because changes in the composition or circumstances of families trigger new needs, needs that are often difficult to realise. Thus, even though the rental market offers the promise of flexibility, this incentive is mostly beneficial for a subset of the population with stable family dynamics (fewer life-course demands), financial capacity (identified based on the accrual of social-dwelling factors above the cost-line, as well as based on age cohorts) and market awareness. Even when considering that some individuals seek out change

207

in housing circumstances, regardless of the stability of their other life-course states, such individuals would remain inhibited if their financial capacity were insufficient.

The more significant life-course barrier is brought forth by inconsistencies in how individuals adhere to the public facet of geographic territoriality. As mentioned in Section 2.1, individuals’ divergent attitudes are united through shared modes of interaction, such as the housing market itself, as well as through specific principles such as the Great Australian Dream. The caveat is that the way in which people sustain these mechanisms remains limited by their individual understanding and responses. This is because Australia’s territorial history is a composite of many ideals such as garden city ideals, escalation of post-war homeownership from the 1950s, and the more recently financialised housing landscape. In some contexts, these ideals complement one another, while in other circumstances they are contradictory. For instance, the extensive urban mode of development derived from the garden city era presented homeowners with tenure and self-sufficiency, all the while enabling speculative fringe growth. Thus, there is an underlying emotional basis to the capitalist imperative observed today. Yet in terms of the day-to-day interaction between individuals, this connection is not necessarily acknowledged. The open-ended survey question revealed that some individuals were mystified by others’ emotional attachments towards the market. Rather they believed housing decisions should be underpinned by clear-cut supply and demand forces.

However, presenting supply and demand forces as final and automatic reintroduces the notion of a self-regulating market. E. D'Arcy and Keogh (1998) throw caution to the wind by explaining that housing markets have prevailing trends which influence how individuals tailor their housing strategies.

208

CHAPTER 9: CONCLUSION This thesis has reconciled the subjective considerations of housing needs/housing ladder analyses (individuals’ housing aspirations) with the objective/quantifiable measures of housing market research. This reconciliation was prompted by the realisation that housing needs and housing outcomes are inherently divergent. It is not possible to understand what individuals want in the housing market by simply analysing variations in consumers’ demand for housing or producers’ housing supply. Throughout this thesis, it has been established that since there is no substitute for housing, it is the amount of leverage that individuals have that determines their housing outcomes. This realisation distinguishes individual agency from realised housing outcomes. This separation was viable because while individuals’ priorities are bound to their life-course or environment, their aspirations reveal their broad intentions. Thus, housing aspirations reveal whether housing needs are satisfied in a balanced and progressive manner.

Individuals’ housing needs satisfiers were classified within the categories of house, home, and dwelling. These housing facets were based on the premise that there are individuals who have inherent needs which must be satisfied to alleviate physiological and psychological stresses. This thesis has reverted to these basic functions, which provided the housing ladder analysis with a solid foundation, not susceptible to changes in the prevailing attitudes towards the housing market. Prevailing market trends have instead been used to describe how such mechanisms influence financial or social strategies. In the Brisbane GCCSA, much of the prevailing discourse is centred on house-price euphoria and the notion of a progressive housing ladder. Subsequently, this financial imperative has also shaped the psyche of individuals and how progress is conceived. There has been a disproportionate focus on house prices and residential mobility, even though these measures are not the most appropriate for a comprehensive understanding of housing aspirations.

This imbalance is problematic because the more consumers, investors and authorities shift their focus towards extracting financial value, rather than prioritising housing needs, the greater the divergence between housing outcomes and housing needs. Housing needs cannot be taken for granted because the exchange-value of housing represents a store of financial wealth. That store of financial wealth does not have to correspond perfectly with housing needs since competition in the market is

209

what regulates supply and demand. Due to the commodification and financialisation of the housing market, the notion that the underlying emotional ends in housing are inherently balanced is untenable.

9.1 Contributions of this Thesis

This section will present the contributions of this thesis to the body of knowledge. These contributions were guided by the critical realism ontology as well as the mixed- methods findings. There are three main contributions that will be outlined in this section. The first is the introduction of a conduit that bridges the internally derived subjective needs of individuals with the externally derived outcomes in the housing market. The second contribution is that by studying how the same subjective needs are accumulated longitudinally, an alternative means of measuring housing progress has been established. The third contribution shows that the financial and territorial tendencies of the housing market equally satisfies the yearning that individuals have for the Great Australia Dream, and housing ladder progression, as well as impedes their housing aspirations.

9.1.1 Situating Individuals’ Subjective Needs within the Wider Context of the Housing Market Situating individuals’ needs within the wider housing market context is important because it allows researchers, policy-makers, analysts and others who study the market to be able to compare market outcomes with a holistic list of measures, measures which stem from the consumers directly, rather than measures such as house prices, which are derived from market competition. Price is a derivative because it is a product of exchange that is far removed from the underlying needs and motivations that guide the decision-making of renters, homeowner occupiers and investors. Historically, classical economists were not sure how to incorporate these subjective needs in a market context (Marx, 1903 [1887]).

Contemporaneously, this divide has been addressed by presenting consumers with housing outcomes and then encouraging them to reveal their trade-offs (Kelly et al., 2011). This is a pragmatic means of presenting renters and owners alike with tangible forms of housing, which they can choose from, based on attributes such as dwelling and location. This is because consumers in the market understand the consequences of different forms of dwelling and house, as explained in the means-

210

end chains: MEC 2 and MEC 3 (Table 3.2). These consumers know the forms of housing that they prefer, relative to other houses, and they understand some of the consequences of those house types and constructs. However, there are some limitations to this approach. Firstly, individuals are encouraged to trade-off a few dwelling, locational or typological options; those options are not exhaustive of the various motivations that exist in the housing market. Apart from the premise that personalisation (the highest state of progression in housing) is hinged on a comprehensive fulfilment of various housing needs (Zavei & Jusan, 2012), this thesis has also established that needs contradict one another.

Moreover, “…. The broad institutional climate will frame the perceptions and expectations of actors in the property market” (É. D'Arcy & Keogh, 2002, p. 23). The second consideration, as mentioned in the discussion chapter, is that developers constantly look to the immediate past, future innovation is reliant on their risk appetite and how governments incentivise those modes of development. From the consumer standpoint, it is feedback or mobility that resolves their needs. Feedback is reliant on market understanding, tenure, geographical positioning, and income, while mobility is inefficient. Thus, there are segments of the market who are unable to communicate their needs (mostly the renter cohort), and even if developers do receive that feedback, there is a degree of hesitancy in developing what comes next.

Therefore, information is paramount, existing housing needs analyses cannot be discounted, but there is a need for developers and governments to be armed with complementary means of understanding market needs. Clapham (2018) has highlighted that there is uncertainty in terms of what form of research should be conducted in the housing space. There is the rationalist housing needs research (based on utilities such as house prices, dwelling stock, the political housing needs research (based on the discourse analysis of different groups) and finally the structuralist approach based on the explanation of constraints, opportunities and impacts. Invariably by situating individuals’ housing priorities (structuralist) within the context of housing outcomes (rationalist), and by evaluating the attitudes of individuals towards the market (political), this thesis has developed a model that appeals to all approaches.

211

Such a comprehensive model is important because risk has been identified as a major barrier in the realisation of adequate housing, and the contribution of a comprehensive conduit between what individuals want and housing outcomes can eliminate some of that risk. This can be accomplished by understanding not simply why individuals trade-off housing attributes in isolated circumstances, but how their overall predispositions can be accommodated in the market.

9.1.2 Establishing an Alternative Means of Measuring Progress in the Housing Ladder: Subjective Aspirations Subsequently, the second main contribution of this thesis is that housing needs satisfiers were harnessed to determine the overall predisposition of individuals in various cohorts/segments, such as renter/owner, geographic and typological. In the general public, the notion of a general predisposition (or what has been termed as housing trajectories in this thesis) is poorly developed. The notion is more established in the various forms of housing ladder analyses, which have been detailed by Faulkner and Beer (2011). This research has expanded on those analyses which emphasis the periods in the life-course where particular housing decisions emerge, though the focus in this thesis has been on what enables progress. This thesis established that when individuals moved (the period in their life where they had sufficient market knowledge and financial capacity) was more likely to reveal a discernible housing trajectory than generally observing residential mobility. This was due to a decreased likelihood of involuntary moves.

In Chapter 6 of this thesis, evaluating progress involved understanding how individuals’ needs were accumulated, as well as understanding the relationship between variables such as decision-making and understanding, accumulation and residential mobility/age. Similarly, in Chapter 7, the capacity for aspirations could be understood, albeit to a lesser extent, in terms of how balanced the home, house, and dwelling facets were. Additionally, the capacity for property terms (developed by property agents) to reach consumers and investors alike was examined.

This is an important contribution to the body of knowledge for two reasons. Firstly, while risk was outlined as a significant barrier for housing innovation, one of the most significant enablers to produce housing is feedback. The ability for property agents on the one hand to reflect attributes that are potentially representative of

212

individuals’ needs, would be a partial indicator of successful housing aspirations. The more significant conclusion was that progression is not linear (in terms of successive housing moves) and is not best characterised by the pricing of dwellings or changes in typology (the housing ladder narrative). While it was clear that the constant themes across all cohorts largely reflected the prioritisation of the core dwelling attributes such as house types, Subsection 6.1.2 highlighted that the prioritisation of most other factors varied throughout individuals’ housing careers. Moreover, through the exploration of why people moved, it was clear that the predominant issue was life- course circumstances, not dwelling related reasons (directly). This would suggest that the focus on house type or house size is largely a measure of individual and family needs.

Consequently, shifting the market’s focus towards the adaptability and suitability of housing, would satisfy individuals’ most pressing housing needs, and alleviate inefficiencies in residential mobility to some extent. Moreover, the fact that an entire segment of the housing market (the rental market), has been imagined as a placeholder, further encourages inefficiencies in housing satisfaction.

9.1.3 Demonstrating Territorial Conflict and Potential Resolution This thesis has advanced the notion of how the property-owning democracy (POD) should function. The POD is a model aimed at securing individuals’ participation in the market, such that individuals’ liberties are non-conflicting, and there is equality of opportunity. Hence, defining how the Brisbane GCCSA addresses the POD is an important contribution since it is the overarching ideal that can unite the disparate aspirations and objectives of individuals, developers, and authorities.

This thesis has introduced different forms of territorialities to investigate the functionality of the POD. Territorialities being the ways in which the dominant agents in a domain impart or defend their social and or political boundaries. It was established that the private territoriality of domestic residents, who are tasked with defending their private abodes, differs from the objectives of the public territoriality of governments and developers, who throughout history have encouraged order, homeownership, and growth. While the territoriality of Australian residents does overlap with that of their government and developers, it has also been established that there are other positions that are contradictory.

213

The best exemplification of this duality between overlapping and divergent ideals is found in the notion of the POD. This is because the concept is versatile, flexible and can be interpreted in a variety of ways. It is the adaptability of housing for various needs that allows individuals’ fragmented outlooks to be reconciled with that of other agents and entities in a common market. This is how normative beliefs are sustained. The POD can be broken down into three segments: property, ownership, and democracy. In the 19th and 20th centuries, property was important from the perspective of self-sufficiency (Freestone, 2010). Ownership was naturally associated with property because individuals were tasked with maintaining their own livelihood. Land was readily available, and residents were able to build their own homes. This was a specific form of liberation, one that the Australian government would look to further amplify in the mid-20th century as a means of buttressing democratic principles. In other words, the property-owning facet was then positioned to shape the democratic ideals of the anti-communist government (ABC Education, 2010). Homeownership peaked in that period, all the while, the notion that democratic liberties should shape the direction of housing first and foremost was largely undeveloped.

Gradually the property aspect of ownership would be prioritised above democratic liberties, which is a concept that has been poorly defined. As explained in the body of the thesis, Australia, at present has grown to be a nation with a disproportionate stake of household wealth in non-financial assets, particularly real estate (Credit Suisse, 2018). Hence, the property aspect has evolved into a sophisticated and dominant part of the Australian economy. Property at this stage has transcended the earlier definition of the physical dwelling and has largely integrated the financial considerations of house. However, if we are to consider how the property- owning democracy is intended to function, from a Rawlsian perspective, the democratic facet is just as critical. Democratic liberties determine how people can guarantee not simply their housing freedoms, but their economic freedoms in general (Mandle & Reidy, 2014 ; O'Neill & Williamson, 2012). This model necessitates that most participants in the market economy (which includes the housing market) are asset owners and have the means to prevent monopoly power. Therefore, in the case of Australia, the property-democracy is not fully realised. While 67% of Australians are homeowners, this rate is falling, and there is a considerable disparity between the participants in the housing market.

214

The investor class typically enjoy better financial leverage, with fewer debt obligations, while many homeowners are saddled with significant debt. As a result, there are numerous antithetical positions in the housing market. This includes the fact that the very mechanism (homeownership) that has been entrusted to enrich Australians as stakeholders are the same structure that is partly responsible for the fragmentation of Australian society. This is related to another antithetical position, which is that the individuals and entities with monopoly power are inclined to stimulate prices and development to the point in which development is unsustainable. Moreover, property zoning rights in Australia are not liberally regulated, which signifies that even when landowners have access to land, they may not have the rights to develop, this means that competition among developers is doubly restricted (Leith van Onselen, 2013). This unsustainability has been upheld with the infusion of lax lending regimes, which has provided households with credit to partake in the purchase of housing with median prices that are ten times income.

These problems would have been prevented if Australia had truly adhered to the tenets of the property-owning democracy concept. Particularly, Australia has failed to embrace the stipulation that the POD should be centred on productive wealth (O'Neill & Williamson, 2012). The POD is meant to encompass a broad range of productive assets, they may be financial assets, but investments that are nonetheless tied to the real economy (Jackson, 2012). The two main components of the real economy are measurable through output and wages.

9.2 Policy Recommendations This thesis’ policy recommendations will be made through the lens of the property- owning democracy. The conflicts that have emerged between the various participants in the market have underscored the need for better coordination between objectives such as housing market growth, tenure, and housing needs. This thesis has shown that no single housing market objective can remedy the needs that emerge elsewhere. For example, regulations disproportionately favour investors and homeowner occupiers. Renters are faced with tenure uncertainty and an insufficient means for adapting their home to meet their needs.

Moreover, there is a need to prevent artificial scarcity, which is the encouragement of extensive development in fringe land and regulates speculation,

215

since these models contradict other liberties. While it is recognised that a significant segment of the population has developed an affinity for such development, it has also been clear that Brisbane and Australian housing solutions have been bereft of versatility and diversity. For example, apartments can be developed to cater for families. This will reorient the perception of Australians in the market.

Regarding socio-economic opportunity (asset distribution and affordability), it is suggested that positioning the housing market as a mainstay of the economy is impractical if the two components of the real economy (output and wages) are neglected. This means that the focus needs to shift away from demand towards effective demand. It is folly to consider the debt burden of Australian households as a measure of their buying power, as opposed to wages. While there is nothing inherently wrong with debt (provided it is manageable), its purpose is to allow the market to borrow from future growth. Therefore, credit must be used to stimulate growth, which in turn can be used to service future debt. In other words, household debt cannot be disproportionately concentrated in the real estate sector, which is largely functioning as a conduit for financial securities. Hence, it is recommended that household liabilities in this sector decreases and prices should be allowed to grow in tandem with wages. This will allow for an increase in financial, social, and notional liberty.

In terms of real productive wealth, in the property market that may involve optimising land uses. These recommendations will restructure the housing market, such that use-values are better reflected in exchange-values. I.e. housing needs become less divergent from housing outcomes. In that spirit, the most significant recommendation that this thesis can offer is that in much the same way that governments have supported housing market growth, there is a need for policies that can stimulate all housing needs facets. For example, it has been established that the renter cohort has a limited aspirational and dwelling outlook. Much of their focus has been consumed by considerations of stability.

9.3 Topics for Future Research and Final Remarks The prospects for this research are many. Expanding the scope of the satisfiers of housing needs is one area of consideration. This would also involve expanding the magnitude of property terms that were studied in the social network analysis. While it is acknowledged in SNA research that only a few terms dominate networks, there is

216

room to explore the importance that fringe property terms have on the composition of the networks.

Ultimately, the objective of this research direction is that the means of evaluating housing outcomes are reassessed. Presently, the Australian housing market is beset with a commodity-driven focus. The problem with such a perspective is that commodification promotes standardisation, and standardised products are differentiated based on price. To augment price, producers must either be more efficient or at the very least, ensure that other producers cannot be more efficient (artificial scarcity/monopolies). We measure outcomes based on these efficiencies, rather than based on the holistic consumption of housing needs. Rather than delivering the form of housing that we would like, productively; we deliver our houses based on productivity. Moreover, what we deem to be productive is doubtful. This is related to the broken windows fallacy, where we measure the activity that has occurred and is visible, whilst ignoring the benefits that could have been derived from activities that were neglected and are invisible.

Although capital gains are visibly highlighted; high median multiples are relatively invisible. Moreover, this divergence between fundamental needs and housing outcomes is irreconcilable without intervention. This is because growth forms its own directive. Productivity represents economic value, and value extraction can be singularly conceived or multifaceted. Therefore, the form of productivity that we sustain in the housing market must not counteract individuals’ subjective needs and aspirations. Ultimately, the process is irreconcilable without coordination and intervention because through the fear of missing out; individuals will sustain the prevailing housing paradigm. If house prices and unfettered ownership are the principal objectives in Australian housing markets, then it is of no surprise that Australian cities have been blighted with unsustainable growth. If, however, we consider housing needs and aspirations as the preeminent yardstick, then so too will housing outcomes reflect that objective.

217

LIST OF REFERENCES Aalbers, M. B., & Christophers, B. (2014). Centring Housing in Political Economy. Housing, Theory and Society, 31(4), 373-394. doi:10.1080/14036096.2014.947082 ABC Education. (2010). Menzies' Great Australian dream. History. Adeniyi, A., & Johnson, L. (2018). "I won't move to a One Bedroom Dog Box": The Challenges of Downsizing for older Social Housing Tenants in Queensland, Australia. Paper presented at the 2018 Joint Asia-Pacific Network for Housing Research and Australasian Housing Researchers Conference, School of Engineering and Built Environment Angus, J. E., & Clark, A. M. (2012). Using Critical Realism in Nursing and Health Research: Promise and Challenges. Nursing Inquiry, 19(1), 1-3. doi:10.1111/j.1440-1800.2011.00580.x Anthony, T. (2007). Criminal Justice and Transgression on Northern Australian Cattle Stations. In I. Macfarlane & M. Hannah (Eds.), Transgressions: Critical Australian Indigenous Histories (pp. 35-61). doi:10.22459/T.12.2007.03 Appadurai, A. (2004). The Capacity to Aspire: Culture and the Terms of Recognition. In V. Rao & M. Walton (Eds.), Culture and Public Action (pp. 59-84). Stanford University Press: Palo Alto. Archer, M., Bhaskar, R., Collier, A., Lawson, T., & Norrie, A. (1998). Critical Realism: Essential Readings. Abingdon: Routledge. Arribas-Bel, D., & Sanz-Gracia, F. (2014). The Validity of the Monocentric Cty Model in a Polycentric Age: US Metropolitan Areas in 1990, 2000 and 2010. Urban Geography, 35(7), 980-997. doi:10.1080/02723638.2014.940693 Aussie. (2019). The Aussie/CoreLogic 25 years of Housing Trends. Property Market Reports. Retrieved from https://www.aussie.com.au/home-loans/property-reports/25years.html Australian Bureau of Statistics. (2015). Housing Occupancy and Costs, 2013-14. Retrieved from http://www.abs.gov.au/ausstats/[email protected]/Lookup/by%20Subject/4130.0~2013- 14~Main%20Features~Housing%20Mobility~10003 Australian Bureau of Statistics. (2016a). 2016 Census - Counting Dwellings, Place of Enumeration (MB). Retrieved from https://auth.censusdata.abs.gov.au/webapi/jsf/tableView/tableView.xhtml# Australian Bureau of Statistics. (2016b). Census TableBuilder. Census of Population and Housing, 2016, TableBuilder. Retrieved from https://www.censusdata.abs.gov.au/webapi/jsf/dataCatalogueExplorer.xhtml Australian Bureau of Statistics. (2016c). Dataset: Australian Census Longitudinal Dataset, 2011- 2016. Retrieved from https://tablebuilder.abs.gov.au/webapi/jsf/tableView/tableView.xhtml Australian Bureau of Statistics. (2016d). Greater Brisbane. 2016 Census Community Profiles. Retrieved from http://quickstats.censusdata.abs.gov.au/census_services/getproduct/census/2016/communi typrofile/3GBRI?opendocument Australian Bureau of Statistics. (2016e). Greater Capital City Statistical Area (GCCSA). Australian Statistical Geography Standard (ASGS): Volume 1. Retrieved from https://www.abs.gov.au/ausstats/[email protected]/Lookup/by%20Subject/1270.0.55.001~July%20 2016~Main%20Features~Greater%20Capital%20City%20Statistical%20Areas%20(GCCSA )~10003 Australian Bureau of Statistics. (2016f). Postcodes & Postal Areas. Geography. Retrieved from https://www.abs.gov.au/websitedbs/censushome.nsf/home/factsheetspoa?opendocument& navpos=450 Australian Bureau of Statistics. (2017a). Census reveals the ‘typical’ Australian. Retrieved from http://www.abs.gov.au/AUSSTATS/[email protected]/AUSSTATS/[email protected]/mediareleasesbyRelea seDate/5E54C95D3D5020C6CA2580FE0013A809?OpenDocument Australian Bureau of Statistics. (2017b). Census: Younger Australians more likely to Make a Move. Retrieved from https://www.abs.gov.au/AUSSTATS/[email protected]/mediareleasesbyReleaseDate/64EEC2403E 851326CA2581BF0036648E?OpenDocument

218

Australian Government department of Social Services. (2017). Household, Income and Labour Dynamics in Australia (HILDA) Survey. Retrieved from https://www.dss.gov.au/sites/default/files/documents/01_2017/ncld_hilda_fact_sheet_12_ja nuary_2017.pdf Australian Institute of Health and Welfare. (2017). Housing Assistance in Australia 2017. Retrieved from https://www.aihw.gov.au/reports/housing-assistance/housing-assistance-in-australia- 2017/contents/dwelling-and-household-size Australian Securities and Investment Commission. (2018). Negative and Positive Gearing. Money Smart Financial Guidance. Retrieved from https://www.moneysmart.gov.au/investing/invest- smarter/negative-and-positive-gearing Badcock, B. (1994). ‘Snakes or Ladders?’: The Housing Market and Wealth Distribution in Australia*. International Journal of Urban and Regional Research, 18(4), 609-627. doi:10.1111/j.1468- 2427.1994.tb00289.x Ballard, A. M. (2017). Loose Praxis: Methodological assemblage in the social complexity of the Australian private rental system. The University of Queensland, Brisbane. Barker, D. (2003). Emancipatory for Whom? A Comment on Critical Realism. Feminist Economics, 9(1), 103-108. doi:10.1080/13545700110059270 Bartel, R., & Graham, N. (2016). Property and Place Attachment: A Legal Geographical Analysis of Biodiversity Law Reform in New South Wales. Geographical Research, 54(3), 267-284. doi:10.1111/1745-5871.12151 Baum, S., & Wulff, M. (2003). Housing Aspirations of Australian Households. Retrieved from https://www.ahuri.edu.au/__data/assets/pdf_file/0017/2177/AHURI_Final_Report_No30_Ho using_aspirations_of_Australian_households.pdf Bearman, P., & Hoffman, M. (2017). Social Network Analysis. Social Network Analysis. Becker, M., Vignoles, V. L., Owe, E., Easterbrook, M. J., Brown, R., Smith, P. B., . . . Brambilla, M. (2014). Cultural Bases for Self-evaluation: Seeing Oneself Positively in Different Cultural Contexts. Personality and social psychology bulletin, 40(5), 657-675. doi:10.1177/0146167214522836 Beer, A., Faulkner, D., & Gabriel, M. (2006). 21st Century Housing Careers and Australia's Housing Future: Literature Review. Retrieved from https://www.ahuri.edu.au/research/final- reports/128 Beran, B. (2016). Understanding Clustering in Tableau 10. Retrieved from https://boraberan.wordpress.com/2016/07/19/understanding-clustering-in-tableau-10/ Berger, P. L., & Luckmann, T. (1966). The Social Construction of Reality London: Penguin. Bernard, A., Forder, P., Kendig, H., & Byles, J. (2017). Residential mobility in Australia and the United States: a retrospective study. Australian Population Studies, 1(1), 41-54. Retrieved from http://www.australianpopulationstudies.org/index.php/aps/article/view/11/6 Bhaskar, R. (2008). [1975]. A Realist Theory of Science. Abingdon: Routledge. Bilston, T., Johnson, R., & Read, M. (2015). Stress Testing the Australian Household Sector Using the HILDA Survey. Retrieved from http://www.rba.gov.au/publications/rdp/2015/2015- 01/descriptive-statistics.html Birch, D. L. (1971). Toward a Stage Theory of Urban Growth. Journal of the American Institute of Planners, 37(2), 78-87. doi:10.1080/01944367108977361 Blondel, V. D., Guillaume, J.-L., Lambiotte, R., & Lefebvre, E. (2008). Fast Unfolding of Communities in Large Networks. Journal of Statistical Mechanics: Theory and Experiment, 2008(10), 1-12. doi:10.1088/1742-5468/2008/10/P10008 Blum, A., Hopcroft, J., & Kannan, R. (2016). Clustering. In Foundations of Data Science (Vol. 5, pp. 208-244). Retrieved from https://www.cs.cornell.edu/jeh/book.pdf Boehm, P. (2018). 10 Options if Your Property Doesn’t Sell. Retrieved from https://propertyupdate.com.au/10-options-property-doesnt-sell/ Bonatto, F. S., Miron, L. I. G., & Formoso, C. T. (2011). Evaluation of Social Housing Projects based on the User Perceived Value Hierarchy. Ambiente Construído, 11(1), 67-83. doi:10.1590/S1678-86212011000100006 219

Borgatti, S. P. (2015). Centrality and Centralization. Chapter 5. Retrieved from http://www.analytictech.com/mb119/chapter5.htm Borgatti, S. P., Mehra, A., Brass, D. J., & Labianca, G. (2009). Network Analysis in the Social Sciences. Science, 323(5916), 892-895. Retrieved from https://science.sciencemag.org/content/sci/323/5916/892.full.pdf Botschen, G., Thelen, E. M., & Pieters, R. (1999). Using Means-End Structures for Benefit Segmentation an Application to Services. European Journal of Marketing, 33(1/2), 38-58. doi:10.1108/EUM0000000004491 Bourgeois, N., Cottrell, M., Lamassé, S., & Olteanu, M. (2015). Search for Meaning Through the Study of Co-occurrences in Texts. Paper presented at the International Work-Conference on Artificial Neural Networks, Palma de Mallorca, Spain. Brennan, G., & Castles, F. G. (2002). Australia Reshaped: 200 years of Institutional Transformation. Cambridge: Cambridge University Press. Bruni, L. (2007). Handbook on the Economics of Happiness. Cheltenham: Edward Elgar Publishing. Bullock, M. (2019). Property, Debt and Financial Stability. Urban Development Institute of Australia (WA) Perth – 20 March: Reserve Bank of Australia Retrieved from https://www.rba.gov.au/speeches/2019/pdf/sp-ag-2019-03-20.pdf buyMyplace. (2016). Investor Presentation. Retrieved from https://www.asx.com.au/asxpdf/20160315/pdf/435tz96hf0mpjb.pdf Cahill, D. (2016). What does Australia’s ideal house look like? Retrieved from https://www.realestate.com.au/news/what-does-australias-ideal-house-look-like/ Camilleri, M. A. (2018). Market Segmentation, Targeting and Positioning. In Travel Marketing, Tourism Economics and the Airline Product (pp. 69-83). Cham, Switzerland: Springer. Carruthers, B. G. (2010). The Meanings of Money: A Sociological Perspective. Theoretical Inquiries in Law, 11(1), 51-74. Retrieved from http://www7.tau.ac.il/ojs/index.php/til/article/viewFile/733/692 Case, K. E. (1994). Land Prices and House Prices in the United States. In Y. Noguchi & J. M. Poterba (Eds.), Housing Markets in the United States and Japan (pp. 29-46). Chicago: The University of Chicago Press. Chadbourne, M. J. (2015). Residential Satisfaction in the Changing Urban Form in Adelaide: a Comparative Analysis of Mawson Lakes and Craigburn Farm, South Australia. (Master of Philosophy ). University of Adelaide, Retrieved from https://digital.library.adelaide.edu.au/dspace/bitstream/2440/92336/3/02whole.pdf Checkbox Survey. (2020). Secure Data Collection and Real-time Reporting are more Important than Ever. Retrieved from www.checkbox.com Clapham, D. (2002). Housing pathways: A post modern analytical framework. Housing, Theory and Society, 19(2), 57-68. Retrieved from https://www.researchgate.net/profile/David_Clapham5/publication/232980113_Housing_Pat hways_A_Post_Modern_Analytical_Framework/links/5587cd3808ae7bc2f44d5ab2.pdf Clapham, D. (2018). Housing Theory, Housing Research and Housing Policy Housing, Theory and Society, 35(2), 163-177. doi:10.1080/14036096.2017.1366937 Clark, W. A., & Huang, Y. (2003). The Life Course and Residential Mobility in British Housing Markets. Environment and Planning A, 35(2), 323-339. doi:10.1.1.453.2270 Clark, W. A. V., & Dieleman, F. M. (1996). Households and housing: choice and outcomes in the housing market. New Brunswick, N.J: Center for Urban Policy Research. Clauset, A., Newman, M. E., & Moore, C. (2004). Finding Community Structure in very Large Networks. Physical Review E, 70(6), 066111. doi:10.1103/PhysRevE.70.066111 Clauset, A., Shalizi, C. R., & Newman, M. E. (2009). Power-law Distributions in Empirical Data. SIAM Review, 51(4), 661-703. Retrieved from https://arxiv.org/abs/0706.1062 Coiacetto, E. (2007). Residential Sub-market Targeting by Developers in Brisbane. Urban Policy and Research, 25(2), 257-274. doi:10.1080/08111140701344833

220

Comber, A. J., Brunsdon, C. F., & Farmer, C. J. (2012). Community Detection in Spatial Networks: Inferring Land use from a Planar Graph of Land Cover Objects. International Journal of Applied Earth Observation and Geoinformation, 18, 274-282. doi:10.1016/j.jag.2012.01.020 Cook, N., Davison, A., & Crabtree, L. (2016). Housing and Home Unbound: Intersections in Economics, Environment and Politics in Australia. Abingdon: Routledge. Coolen, H., Boelhouwer, P., & van Driel, K. (2002). Values and Goals as Determinants of Intended Tenure Choice. Journal of Housing and the Built Environment, 17(3), 215-236. Retrieved from https://link.springer.com/content/pdf/10.1023/A:1020212400551.pdf Cox, W., & Pavletich, H. (2005). Demographia International Housing Affordability Retrieved from http://www.demographia.com/dhi-200502.htm Crawford, J., & McKee, K. (2018). Hysteresis: Understanding the Housing Aspirations Gap. Sociology, 52(1), 182-197. Retrieved from https://pdfs.semanticscholar.org/aaa9/acd3a15977a03416457e30e4e04381a9a3ab.pdf Credit Suisse. (2018). Research Institute: Global Wealth Report 2018. Retrieved from Credit Suisse AG: http://publications.credit-suisse.com/tasks/render/file/index.cfm?fileid=B4A3FC6E- 942D-C103-3D14B98BA7FD0BCC Crystal, D. (2008). A Dictionary of Linguistics and Phonetics. Malden: Blackwell Publishing. Csardi, G. (2015a). Centralization of a Graph. R igraph Manual Pages. Retrieved from https://igraph.org/r/doc/centralize.html Csardi, G. (2015b). Find Eigenvector Centrality Scores of Network Position. R igraph Manual Pages. Retrieved from https://igraph.org/r/doc/eigen_centrality.html Csardi, G. (2015c). Strength or Weighted Vertex Degree. R igraph Manual Pages. Retrieved from http://igraph.org/r/doc/strength.html Csardi, G. (2015d). Transitivity of a graph. R igraph Manual Pages. Retrieved from https://igraph.org/r/doc/transitivity.html D'Arcy, E., & Keogh, G. (1998). Territorial Competition and Property Market Process: An Exploratory Analysis. Urban Studies, 35(8), 1215-1230. doi:10.1080/0042098984330 D'Arcy, É., & Keogh, G. (2002). The Market Context of Property Development Activity. In Development and Developers: Perspectives on Property (pp. 19-34). doi:10.1002/9780470757192.ch2 Daley, J., Coates, B., & Wiltshire, T. (2018). Housing Affordability: Re-imagining the Australian Dream (0648230740). Retrieved from https://grattan.edu.au/wp- content/uploads/2018/03/901-Housing-affordability.pdf Danermark, B. (2002). Interdisciplinary Research and Critical Realism The Example of Disability Research. Alethia, 5(1), 56-64. doi:10.1558/aleth.v5i1.56 Das, S. (2013). Debt Dilemma: the End of Growth? ABC News. Retrieved from https://www.abc.net.au/news/2013-04-22/das-the-end-of-growth3f/4643690 Davis, J. E. (1991). Contested Ground: Collective Action and the Urban Neighborhood. London: Cornell University Press. Davison, G. (1993). The Past & Future of the Australian Suburb. Australian Planner, 31(2), 63-69. doi:10.1080/07293682.1993.9657608 Davison, G. (1995). Australia: the First Suburban Nation? Journal of Urban History, 22(1), 40-74. doi:10.1177/009614429502200103 Davison, G. (1997). The Great Australian Sprawl. Historic Environment, 13(1), 10. Retrieved from http://search.informit.com.au.ezproxy.library.uq.edu.au/documentSummary;dn=7694024653 33798;res=IELHSS De Nooy, W., Mrvar, A., & Batagelj, V. (2018). Part II - Cohesion. In Exploratory Social Network Analysis with Pajek (Vol. 46, pp. 73-96). Retrieved from http://vlado.fmf.uni- lj.si/pub/networks/course/ch03/Chapter3.pdf Demographia. (2019). 15th Annual Demographia International Housing Affordability Survey. Retrieved from Belleville, Illinois: http://www.demographia.com/dhi.pdf Dent, P., Patrick, M., & Ye, X. (2012). Culture, Markets and Institution. In Real Estate: Property Markets and Sustainable Behaviour (pp. 23-36). London, UNITED KINGDOM: Routledge. 221

Denzin, N. K. (2004). Explaining Society: Critical Realism in the Social Sciences. Contemporary Sociology, 33(2), 249. Retrieved from https://search.proquest.com/openview/ab2afd500e44d9c0328a6442075e1cc0/1?pq- origsite=gscholar&cbl=47867 Department for Business Energy & Industrial Strategy. (2017). Research on Buying and Selling Homes. Retrieved from https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_d ata/file/653581/buying-selling_homes-research.pdf Digital Finance Analytics (Producer). (2019, March 13, 2019). What Are Households Thinking Now? Retrieved from https://www.youtube.com/watch?v=KuGg24Wv12o Domain.com.au. (2018). Retrieved from www.domain.com.au Doyal, L., & Gough, I. (1991). A Theory of Human Need. New York: Guilford Press. Duke, J. (2017). What First-Home Buyers have to Compromise on to Buy a Sydney Home for Less than $650,000. Retrieved from https://www.domain.com.au/news/what-firsthome-buyers- have-to-compromise-on-to-buy-a-sydney-home-for-less-than-650000-20170704-gwwe4x/ Easthope, H. (2004). A Place Called Home. Housing, Theory and Society, 21(3), 128-138. doi:10.1080/14036090410021360 ECONorthwest. (2002). City of Ashland Housing Needs Analysis Retrieved from Eugene, Oregon: http://pages.uoregon.edu/rgp/PPPM613/downloads/Ashland%20Housing%20Needs%20An alysis.pdf Eidgenössische Technische Hochschule Zürich. (2018). Matrix Multiplication. R Documentation. Retrieved from https://stat.ethz.ch/R-manual/R-devel/library/base/html/matmult.html Fang, W.-T., Ng, E., Wang, C.-M., & Hsu, M.-L. (2017). Normative Beliefs, Attitudes, and Social Norms: People Reduce Waste as an Index of Social Relationships When Spending Leisure Time. Sustainability, 9(10), 1696. Retrieved from http://www.mdpi.com/2071-1050/9/10/1696 Faulkner, D., & Beer, A. (2011). Housing Transitions Through the Life Course. Bristol: Policy Press. Faust, K. (2006). Comparing Social Networks: Size, Density, and Local Structure. Metodoloski zvezki, 3(2), 185. Retrieved from https://www.stat-d.si/mz/mz3.1/faust.pdf Fehring, H., & Bessant, J. (2009). The Life Course Research Design for Transitional Labour Market Research. Journal of Education and Work, 22(2), 81-90. Retrieved from https://researchbank.rmit.edu.au/view/rmit:8376/b2006016560.pdf Feijten, P., Hooimeijer, P., & Mulder, C. H. (2008). Residential Experience and Residential Environment Choice over the Life-course. Urban Studies, 45(1), 141-162. doi:10.1177/0042098007085105 Feijten, P., & Mulder, C. H. (2005). Life-course Experience and Housing Quality. Housing Studies, 20(4), 571-587. doi:10.1080/02673030500114474 Fernandez, R., & Aalbers, M. B. (2016). Financialization and Housing: Between Globalization and Varieties of Capitalism. Competition & Change, 20(2), 71-88. Forrest, R., & Yip, N.-m. (2012). Young People and Housing: Transitions, Trajectories and Generational Fractures. Oxon: Routledge. Fortunato, S. (2010). Community Detection in Graphs. Physics Reports, 486(3), 75-174. doi:10.1016/j.physrep.2009.11.002 Fox, R., & Finlay, R. (2012). Dwelling Prices and Household Income. Reserve Bank of Australia Retrieved from https://www.rba.gov.au/publications/bulletin/2012/dec/pdf/bu-1212-2.pdf Fox, R., & Tulip, P. (2014). Is Housing Overvalued? Retrieved from https://www.rba.gov.au/publications/rdp/2014/pdf/rdp2014-06.pdf Freestone, R. (2010). Urban Nation: Australia's Planning Heritage. Collingwood: CSIRO Publishing. Fusch, P. I., & Ness, L. R. (2015). Are We There Yet? Data Saturation in Qualitative Research. The qualitative report, 20(9), 1408. Retrieved from https://nsuworks.nova.edu/tqr/vol20/iss9/3 Gallagher, R., Sigler, T., & Liu, Y. (2019). Targeted Urban Consolidation or Ad Hoc Redevelopment? The Influence of Cadastral Structure and Change on the Urban Form of Brisbane, Australia. Urban Geography, 40. doi:10.1080/02723638.2019.1663058

222

Gao, A., Lin, Z., & Na, C. F. (2009). Housing Market Dynamics: Evidence of Mean Reversion and Downward Rigidity. Journal of housing economics, 18(3), 256-266. doi:10.1016/j.jhe.2009.07.003 Geck, M., & Mackay, S. (2018). Housing in Queensland: Affordability and Preferences. Retrieved from https://apo.org.au/sites/default/files/resource-files/2018/04/apo-nid140056- 1186571.pdf Gee, H. (2010). Residential Real Estate Data on the Internet: Benefits and Limitations. Journal of Business & Finance Librarianship, 15(2), 104-122. doi:10.1080/08963560903554658 Giddens, A. (1984). The Constitution of Society: Outline of the Theory of Structuration: University of California. Giddens, A. (1991). Modernity and Self-Identity: Self and Society in the Late Modern Age. Stanford Stanford University Press. Giddens, A. (1993). [1976]. New Rules of Sociological Method: a Positive Critique of Interpretative Sociologies. Cambridge: Polity Press. Giddens, A. (2009). Sociology In P. W. Sutton (Ed.), (6th ed.). Cambridge: Polity Press. Gilead, T. (2013). Educational Insights of the Economist: Tibor Scitovsky on Education, Production and Creative Consumption. Studies in Philosophy and Education, 32(6), 623-639. doi:10.1007/s11217-013-9355-6 Glaeser, E. L., & Nathanson, C. G. (2015). Housing Bubbles. Retrieved from Cambridge: https://www.nber.org/papers/w20426.pdf Google. (2019, September 5, 2016). Google Trends: Compare. Retrieved from https://trends.google.com/trends/explore?date=all&geo=AU&q=houses,homes,dwellings Gordon, D. L. A., Maginn, P. J., & Biermann, S. (2015). Estimating the size of Australia’s Suburban Population. Retrieved from Perth, University of Western Australia: https://resources.patrec.org/publications/perspectives/SuburbanPopulation2015.pdf Gotham, K. F. (2009). Creating Liquidity out of Spatial Fixity: The Secondary Circuit of Capital and the Subprime Mortgage Crisis. International Journal of Urban and Regional Research, 33(2), 355-371. doi:10.1111/j.1468-2427.2009.00874.x Gram‐Hanssen, K., & Bech‐Danielsen, C. (2004). House, Home and Identity From a Consumption Perspective. Housing, Theory and Society, 21(1), 17-26. doi:10.1080/14036090410025816 Green, G., Grimsley, M., & Stafford, B. (2005). The Dynamics of Neighbourhood Sustainability. York: Joseph Rowntree Foundation York. Grice, H. P. (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Studies in Syntax and Semantics III: Speech Acts (pp. 183-198). New York: Academic Press. Griffin, R. H. (2017). Primate Social Networks: Basics of Visualization and Analysis with Igraph. Retrieved from http://www.randigriffin.com/2017/04/26/primate-social-networks-in- igraph.html Grossmann, H. (1977). Marx, Classical Political Economy and the Problem of Dynamics Part II. Capital & Class, 1(3), 67-99. doi:10.1177/030981687700300104 Gurran, N., Phibbs, P., Yates, J., Gilbert, C., Whitehead, C., Norris, M., . . . Goodman, R. (2015). Housing Markets, Economic Productivity, and Risk: International Evidence and Policy Implications for Australia—Volume 1: Outcomes of an Investigative Panel. Australian Housing and Urban Research Institute, Melbourne. Retrieved from https://www.ahuri.edu.au/__data/assets/pdf_file/0023/5675/AHURI_Final_Report_No254_H ousing_markets_economic_productivity_and_risk_international_evidence_and_policy_impli cations_for_Australia-Volume_1.pdf Ha, J., & Jang, S. (2013). Attributes, Consequences, and Consumer Values: A Means-End Chain Approach across Restaurant Segments. International Journal of Contemporary Hospitality Management, 25(3), 383-409. doi:10.1108/09596111311311035 Haila, A. (2016). Urban Land Rent: Singapore as a Property State. West Sussex: John Wiley & Sons. Hamel, S., Leclerc, G., & Lefrançois, R. (2003). PERSPECTIVE: A Psychological Outlook on the Concept of Transcendent Actualization. The International Journal for the Psychology of Religion, 13(1), 3-15. doi:10.1207/S15327582IJPR1301_02 223

Hannonen, M. (2015). A Field Theory of House Prices: An Empirical Study of the Helsinki Submarket: Suomen E-painos Oy. Hardie, I. W., Parks, P. J., & van Kooten, G. C. (2004). The Economics of Land Use at the Intensive and Extensive Margins. In H. Folmer & T. Tietenberg (Eds.), International Yearbook of Environmental & Resource Economics (pp. 101-138). Cheltenham, United Kingdom. Harenberg, S., Bello, G., Gjeltema, L., Ranshous, S., Harlalka, J., Seay, R., . . . Samatova, N. (2014). Community Detection in Large‐scale Networks: a Survey and Empirical Evaluation. Wiley Interdisciplinary Reviews: Computational Statistics, 6(6), 426-439. doi:10.1002/wics.1319 Harper, F. A. (1967). The Subjective Theory of Value Innovator. Retrieved from https://history.fee.org/media/4074/1561_the-subjective-theory-of-value.pdf Harris, Z. (1954). Distributional structure. Word, 10(2-3), 146-162. doi:10.1080/00437956.1954.11659520 Hart, C. S. (2016). How Do Aspirations Matter? Journal of Human Development and Capabilities, 17(3), 324-341. doi:10.1080/19452829.2016.1199540 Harvey, D. (2010). Social Justice and the City. London: University of Georgia Press. Harvey, D. (2012a). Professor David Harvey on “The Contradictions of Capital”. The 2012 Distinguished Lecturer Series. Retrieved from https://integratinghorizons.com/professor- david-harvey-on-the-contradictions-of-capital/ Harvey, D. (2012b). Professor David Harvey on “The Contradictions of Capital”. Retrieved from https://integratinghorizons.com/professor-david-harvey-on-the-contradictions-of-capital/ Heagney, M. (2018). Australians are Paying More for Less Land: Here’s Why. Retrieved from https://www.domain.com.au/news/australians-are-paying-more-for-less-land-heres-why- 20180916-h15dk1-763835/ Helbing, D. (2013). Economics 2.0: The Natural Step Towards a Self-Regulating, Participatory Market Society. Evolutionary and Institutional Economics Review, 10(1), 3-41. doi:10.14441/eier.D2013002 Herepath, A. (2014). In the Loop: A Realist Approach to Structure and Agency in the Practice of Strategy. Organization Studies, 35(6), 857-879. doi:10.1177/0170840613509918 Himelboim, I. (2017). Social Network Analysis (Social Media). In J. r. Matthes (Ed.), The International Encyclopedia of Communication Research Methods (pp. 1-15). doi:10.1002/9781118901731.iecrm0236 Ho, L. S., & Wong, G. W.-c. (2009). The First Step on the Housing Ladder: A Natural Experiment in Hong Kong. Journal of housing economics, 18(1), 59-67. doi:10.1016/j.jhe.2009.02.003 Holcombe, R. G. (1999). Great Austrian Economists, The. Auburn: Ludwig von Mises Institute. Holme, A. (1985). Housing and Young Families in East London. London: Routledge & Kegan Paul Books. Horlings, L. (2015). The Worldview and Symbolic Dimension in Territorialisation. In J. Dessein, E. Battaglini, & L. Horlings (Eds.), Cultural Sustainability and Regional Development: Theories and Practices of Territorialisation (pp. 43-58). Retrieved from http://ebookcentral.proquest.com/lib/uql/detail.action?docID=3570320 Houston, S. (2001). Beyond Social Constructionism: Critical Realism and Social Work. British Journal of Social Work, 31(6), 845-861. doi:10.1093/bjsw/31.6.845 Israel, T. (2003). Some Place Like Home: Using Design Psychology to Create Ideal Places. Chichester, West Sussex: Wiley. Jackson, B. (2012). Property‐Owning Democracy: A Short History. In M. O'Neill & T. Williamson (Eds.), Property‐Owning Democracy: Rawls and Beyond (pp. 33-52). Chichester: Blackwell Publishing Ltd. Jacobs, J. (1961). The Death and Life of Great American Cities Vintage Books: New York. James, A. (2019). Means End Chains and Laddering. Retrieved from https://www.e- education.psu.edu/ba850/node/647 Jansen, S. J. T. (2014). Why is Housing Always Satisfactory? A Study into the Impact of Cognitive Restructuring and Future Perspectives on Housing Appreciation. Social Indicators Research, 116(2), 353-371. doi:10.1007/s11205-013-0303-1 224

Kaldor, N. (1972). The Irrelevance of Equilibrium Economics. The Economic Journal, 82(328), 1237- 1255. Retrieved from https://www.jstor.org/stable/2231304 Karjalainen, P. T. (1993). House, Home and the Place of Dwelling. Scandinavian Housing & Planning Research, 2(10), 65-74. Retrieved from 10.1080/02815739308730324 Kassambara, A. (2020). K-Means Clustering in R: Algorithm and Practical Examples. Retrieved from https://www.datanovia.com/en/lessons/k-means-clustering-in-r-algorith-and-practical- examples/ Kelly, J.-F., Breadon, P., & Reichl, J. (2011). Getting the Housing we Want. Retrieved from Melbourne: http://grattan.edu.au/wp- content/uploads/2014/04/117_report_getting_the_housing_we_want.pdf Kemp, S., & Holmwood, J. (2003). Realism, Regularity and Social Explanation. Journal for the Theory of Social Behaviour, 33(2), 165-187. doi:10.1111/1468-5914.00212 Kendig, H. (1981). Buying and Renting: Household Moves in Adelaide Canberra Australian Institute of Urban Studies. Kendig, H. L. (1990). A Life Course Perspective on Housing Attainment. In D. Myers (Ed.), Housing demography: Linking demographic structure and housing markets (pp. 133-156). Madison: University of Wisconsin Press. Khan, C. (2018, June 28, 2019). Is the Property Ladder just a Myth? Property. Retrieved from https://www.theguardian.com/money/shortcuts/2018/jan/07/is-the-property-ladder-just-a- myth Kim, H. H., & Rhee, H. Y. (2019). An Ontology-Based Labeling of Influential Topics Using Topic Network Analysis. Journal of Information Processing Systems, 15(5). King-Hill, S. (2015). Critical Analysis of Maslow’s Hierarchy of Need. The STeP Journal (Student Teacher Perspectives), 2(4), 54-57. Retrieved from http://insight.cumbria.ac.uk/id/eprint/2942/1/KingHill_CriticalAnalysisOfMaslows.pdf King, A. (2010). The Odd Couple: Margaret Archer, Anthony Giddens and British Social Theory. The British Journal of Sociology, 61, 253-260. doi:10.1111/j.1468-4446.2009.01288.x King, P. (2004). Private Dwelling: Contemplating the Use of Housing. New York: Routledge Kohl, S. (2018). A Small History of the Homeownership Ideal. Retrieved from Cologne: https://www.mpifg.de/pu/mpifg_dp/2018/dp18-6.pdf Lagueux, M. (1997). Menger and Jevons on Value: a Crucial Difference. Montreal: Université de Montréal, Faculté des arts et des sciences. Lancia, F. (2007). Word Co-occurrence and Similarity in Meaning: Some Methodological Issues. Retrieved from https://mytlab.com/wcsmeaning.pdf Lawless, T. (2012). Four reasons why Australia’s housing sector hasn’t responded to demand. Research Blog Industry Analysis Retrieved from http://blog.corelogic.com.au/2012/09/four- reasons-why-australias-housing-sector-hasnt-responded-to-demand/ Lefebvre, H. (1968). The Right to the City. In Le Droit à la ville (Vol. 3, pp. 108-123). Paris Anthropos. Leith van Onselen. (2013). The Aussie Urban containment Racket. Retrieved from https://www.macrobusiness.com.au/2013/09/the-aussie-urban-containment-racket/ Lemanski, C. (2011). Moving up the Ladder or Stuck on the Bottom Rung? Homeownership as a Solution to Poverty in Urban South Africa. International Journal of Urban and Regional Research, 35(1), 57-77. doi:10.1111/j.1468-2427.2010.00945.x Lewis, P. (2000). Realism, Causality and the Problem of Social Structure. Journal for the Theory of Social Behaviour, 30(3), 249-268. doi:10.1111/1468-5914.00129 Leydesdorff, L., & Hellsten, I. (2006). Measuring the Meaning of Words in Contexts: An Automated Analysis of Controversies about 'Monarch butterflies,' 'Frankenfoods,' and 'stem cells'. Scientometrics, 67(2), 231-258. doi:10.1556/Scient.67.2006.2.6 Lo, K. (2013). Approaching Neighborhood Democracy from a Longitudinal Perspective: An Eighteen-Year Case Study of a Homeowner Association in Beijing. Urban Studies Research, 2013, 639312. doi:10.1155/2013/639312

225

Loader, C. (2018). What Explains Variations in Journey to Work Mode Shares Between and Within Australian Cities? Retrieved from https://chartingtransport.com/2018/12/06/what-explains- variations-in-journey-to-work-mode-shares-between-and-within-australian-cities/ Loader, C. (2019a). Are Australian Cities Sprawling with Low-Density Car-Dependent Suburbs? Retrieved from https://chartingtransport.com/2019/01/30/are-australian-cities-sprawling- with-low-density-car-dependent-suburbs/ Loader, C. (2019b). Suburban Employment Clusters and the Journey to Work in Australian Cities. Retrieved from https://chartingtransport.com/2018/07/08/suburban-employment-clusters- and-the-journey-to-work-in-australian-cities/ Loader, C. (2019c). Where is population growth happening in Australia? Retrieved from https://chartingtransport.com/2019/01/08/where-is-population-growth-happening-in- australia/ Lukács, G. (2000). [1923]. A Defence of History and Class Consciousness: Tailism and the Dialectic. London: Verso. Macfarlane, I. J. (2003). Do Australian Households Borrow Too Much? [Press release]. Retrieved from http://www.rba.gov.au/speeches/2003/sp-gov-030403.html Maclennan, D. (2012). Understanding Housing Markets: Real Progress or Stalled Agendas? In D. F. Clapham, W. A. V. Clark, & K. Gibb (Eds.), The SAGE Handbook of Housing Studies. doi:10.4135/9781446247570 Maclennan, D., & Miao, J. (2017). Housing and Capital in the 21st Century. Housing, Theory and Society, 34(2), 127-145. doi:10.1080/14036096.2017.1293378 Maclennan, D., Pawson, H., Kenneth Gibb, Chisholm, S., & Hulchanski, D. (2019). Shaping Futures: Changing the Housing Story Final report. Retrieved from https://cityfutures.be.unsw.edu.au/documents/539/shaping_futures_final_report_WEB.pdf Mahoney, M. (2003). The Subconscious Mind of the Consumer (And How To Reach It). Retrieved from https://hbswk.hbs.edu/item/the-subconscious-mind-of-the-consumer-and-how-to- reach-it Mandle, J., & Reidy, D. A. (2014). A Companion to Rawls. Chichester: Wiley Online Library. Manisera, M., & Zuccolotto, P. (2013). A Proposal for the Treatment of “don’t know” Responses. Retrieved from http://syrtoproject.eu/wp-content/uploads/2014/05/13_UNIBS6.pdf Marin, A., & Wellman, B. (2014). Chapter 2 | Social Network Analysis: An Introduction. In J. Scott & P. J. Carrington (Eds.), The SAGE Handbook of Social Network Analysis (pp. 11-22). doi:10.4135/9781446294413 Marx, K. (1903). [1887]. Capital: a Critical Analysis of Capitalist Production (9th ed.. ed.). London: London: Sonnenschein. Marx, K., & Engels, F. (2008). Wage Labour and Capital United States: Wildside Press. Maslow, A. H., Frager, R., Fadiman, J., McReynolds, C., & Cox, R. (1970). Motivation and Personality (Vol. 2). New York: Harper & Row Mattick, P. (1977). Economics, Politics, and The Age of Inflation. New York: Routledge. Max-Neef, M. A., Elizalde, A., & Hopenhayn, M. (1991). Human Scale Development: Conception, Application and Further Reflections. New York: Apex Press. McCarroll, P. (2016). Vendor, Buyer Feedback An Important Tool In Increasing Real Estate Transparency. Retrieved from https://www.yourinvestmentpropertymag.com.au/news/vendor-buyer-feedback-an- important-tool-in-increasing-real-estate-transparency-213491.aspx Menger, C. (2007). [1871]. Principles of Economics. Auburn, Alabama: Ludwig von Mises Institute. Mennis, J., & Guo, D. (2009). Spatial Data Mining and Geographic Knowledge Discovery—An Introduction. Computers, Environment and Urban Systems, 33(6), 403-408. doi:10.1016/j.compenvurbsys.2009.11.001 Metcalfe, L. (2016). How Far Will You Have To Go? The Price of Proximity. Retrieved from https://www.microburbs.com.au/blog/property-proximity-price-jane-slack-smith/

226

Metzger, J. T. (2000). Planned Abandonment: The Neighborhood LifeCycle Theory and National Urban Policy. Housing Policy Debate, 11(1). Retrieved from https://www.innovations.harvard.edu/sites/default/files/hpd_1101_metzger.pdf Mihalcea, G. I. R. (2017). Web Crawling and Scraping. In G. I. R. Mihalcea (Ed.), Text Mining: A Guidebook for the Social Sciences (pp. 34-41). doi:10.4135/9781483399782 Mills, A., Durepos, G., & Wiebe, E. (2010). Encyclopedia of Case Study Research. doi:10.4135/9781412957397 Mills, M., & Blossfeld, H.-P. (2003). Globalization, Uncertainty and Changes in Early Life Courses. Zeitschrift für Erziehungswissenschaft, 6(2), 188-218. doi:10.1007/s11618-003-0023-4 Moore, J. (2000). Placing Home in Context. Journal of Environmental Psychology, 20(3), 207-217. doi:10.1006/jevp.2000.0178 Mooya, M. M. (2016). Aristotle to Marshall: The Evolution of Economic Value Theory. In Real Estate Valuation Theory: A Critical Appraisal (pp. 23-41). Berlin: Springer. Moreno-Sánchez, I., Font-Clos, F., & Corral, Á. (2016). Large-Scale Analysis of Zipf’s Law in English Texts. PLOS ONE, 11(1), e0147073. doi:10.1371/journal.pone.0147073 Morris, M. W., Hong, Y.-y., Chiu, C.-y., & Liu, Z. (2015). Normology: Integrating Insights about Social Norms to Understand Cultural Dynamics. Organizational Behavior and Human Decision Processes, 129, 1-13. doi:10.1016/j.obhdp.2015.03.001 Mudd, W., Tesfaghiorghis, H., & Bray, J. R. (2001). Some Issues in Home Ownership. Policy Research Paper No. 17. Department of Family and Community Services Canberra. Retrieved from https://www.dss.gov.au/sites/default/files/documents/05_2012/prp_no_17.pdf Murphy, R. P. (2011). Problems with the Cost Theory of Value. Retrieved from https://mises.org/library/problems-cost-theory-value Musto, L. C., & Rodney, P. A. (2016). Moving from Conceptual Ambiguity to Knowledgeable Action: Using a Critical Realist Approach to Studying Moral Distress. Nursing Philosophy, 17(2), 75- 87. doi:10.1111/nup.12104 Næss, P. (2004). Prediction, Regressions and Critical Realism. Journal of Critical Realism, 3(1), 133-164. doi:10.1558/jcr.v3i1.133 Namazi-Rad, M.-R., Mokhtarian, P., Shukla, N., & Munoz, A. (2016). A Data-Driven Predictive Model for Residential Mobility in Australia–A Generalised Linear Mixed Model for Repeated Measured Binary Data. Journal of choice modelling, 20, 49-60. doi:10.1016/j.jocm.2016.04.006 North, M. (2018a). Who Really Benefits from High Home Prices? Retrieved from https://digitalfinanceanalytics.com/blog/tag/house-prices/ North, M. (2018b). Why Is Wages Growth So Low? Retrieved from http://digitalfinanceanalytics.com/blog/why-is-wages-growth-so-low/ North, M. (2019). Mortgage Stress Higher Post the Election. Retrieved from http://digitalfinanceanalytics.com/blog/mortgage-stress-higher-post-the-election/ Nowak, A., & Smith, P. (2017). Textual Analysis in Real Estate. Journal of Applied Econometrics, 32(4), 896-918. Retrieved from http://busecon.wvu.edu/phd_economics/pdf/15-34.pdf Nussbaum, M. C. (2001). Women and Human Development: The Capabilities Approach (Vol. 3). Cambridge: Cambridge University Press. O'Neill, M., & Williamson, T. (2012). Introduction. In M. O'Neill & T. Williamson (Eds.), Property‐ Owning Democracy: Rawls and Beyond (pp. 1-14). Chichester: Blackwell Publishing Ltd. OECD. (2018). Rethinking Urban Sprawl: Moving Towards Sustainable Cities. Retrieved from Paris: https://www.oecd.org/environment/tools-evaluation/Policy-Highlights-Rethinking-Urban- Sprawl.pdf Opsahl, T., & Panzarasa, P. (2009). Clustering in Weighted Networks. Social Networks, 31(2), 155- 163. doi:10.1016/j.socnet.2009.02.002 Organisation for Economic Co-operation and Development. (2017). Statistical Insights: What does household debt say about financial resilience? National Accounts.

227

Östh, J. (2007). Home, Job and Space: Mapping and Modeling the Labor Market. Uppsala University, Uppsala. Retrieved from https://www.diva- portal.org/smash/get/diva2:169584/FULLTEXT01.pdf (72) Palley, T. (2013). Financialization: The Economics of Finance Capital Domination. New York: Palgrave Macmillan. Penz, G. P. (1986). The Interest Conception of Consumer Sovereignty. In G. P. Penz (Ed.), Consumer Sovereignty and Human Interests (pp. 5-24). Cambridge: Cambridge University Press. Perkins, H. C., & Thorns, D. C. (1999). House and Home and their Interaction with Changes in New Zealand's Urban System, Households and Family Structures. Housing, Theory and Society, 16(3), 124-135. doi:10.1080/14036099950149983 Phillips, B., & Taylor, M. (2015). Buy Now, Pay Later: Household Debt in Australia. Retrieved from AMP: http://www.keepmeposted.org.au/wp-content/uploads/2017/04/ Pike, A., & Pollard, J. (2010). Economic Geographies of Financialization. Economic Geography, 86(1), 29-51. doi:10.1111/j.1944-8287.2009.01057.x Pitkin, B. (2001). Theories of Neighborhood Change: Implications for Community Development Policy and Practice. Retrieved from http://www.urbancenter.utoronto.ca/pdfs/elibrary/UCLA_theories_of_neighborho.pdf Porteous, J. D. (1976). Home: The Territorial Core. Geographical Review, 66(4), 383-390. doi:10.2307/213649 Profile id. (2019). Dwelling Type. Greater Brisbane. Retrieved from https://profile.id.com.au/australia/dwellings?WebID=270&EndYear=2006&DataType=EN Profile id. (2020). Housing tenure. Greater Brisbane. Retrieved from https://profile.id.com.au/australia/tenure?WebID=270&BMID=10 Pryke, M. (2006). Speculating on Geographies of Finance. University of Manchester/OU CRESC WP Series. CRESC, Open University. Milton Keynes. Retrieved from http://hummedia.manchester.ac.uk/institutes/cresc/workingpapers/wp24.pdf Queensland Government. (2019). Density and Diversity Done Well Open Idea Competition. Retrieved from https://www.hpw.qld.gov.au/aboutus/OQGA/Pages/DensityDiversityDoneWellOpenIdeaCo mpetition.aspx Raamkumar, A. S. (2016). What's in a Country Name - Twitter Hashtag Analysis of #singapore. Computer Science: Social and Information Networks. arXiv.org. Retrieved from https://arxiv.org/abs/1601.04135 Raimond, T., & Hensher, D. A. (1992). Review of Panel Surveys and Other Longitudinal Techniques. An Annotated Bibliographic Review [Press release]. Retrieved from https://ses.library.usyd.edu.au/bitstream/handle/2123/19035/ITS-WP-92-19.pdf Rapoport, A. (2000). Theory, Culture and Housing. Housing, Theory and Society, 17(4), 145-165. doi:10.1080/140360900300108573 RBB Economics. (2016). Economic Assessment of PMG’s Application for Authorisation Retrieved from https://www.accc.gov.au/system/files/public-registers/documents/D16%2B81044.pdf Realestate.com.au. (2018). Retrieved from https://www.realestate.com.au/buy realestateVIEW.com.au Ltd. (2019). realestateVIEW. Retrieved from https://www.realestateview.com.au/ Reserve Bank of Australia. (2003). Productivity Commission Inquiry on First Home Ownership. Retrieved from https://www.rba.gov.au/publications/submissions/housing-and-housing- finance/inquiry-productivity-commission-on-first-home/pdf/inquiry-productivity-commission- on-first-home.pdf Reserve Bank of Australia. (2015). Proportion of Investment Housing Relative to Owner-Occupied Housing. House of Representatives Standing Committee on Economics Retrieved from https://www.rba.gov.au/publications/submissions/housing-and-housing-finance/inquiry-into- home-ownership/proportion-investment-housing-relative-owner-occ-housing.html

228

Robertson, H. M., & Taylor, W. L. (1957). Adam Smith's Approach to the Theory of Value. The Economic Journal, 67(266), 181-198. doi:10.2307/2227781 Robertson, M. (2014). Case Study: Finance and Housing Provision in Britain. Retrieved from http://fessud.eu/wp-content/uploads/2013/04/Case-Study_-Finance-and-Housing-Provision- in-Britain-working-paper-51.pdf Rogers, C. (2006). The Principle of Effective Demand: The Key to Understanding the General Theory. University of Adelaide. School of Economics. Retrieved from https://www.adelaide.edu.au/directory/colin.rogers?dsn=directory.file;field=data;id=10692;m =view Rogers, D. (2016, April 13, 2017). Australia’s Housing Affordability Problem Explained in Five Historical Steps. Retrieved from http://theconversation.com/australias-housing-affordability- problem-explained-in-five-historical-steps-64794 Room, A. (1985). Dictionary of Confusing Words and Meanings. London: Routledge. Rowley, S., & James, A. (2018). The Private Rental Sector in Australia: Public Perceptions of Quality and Affordability. Retrieved from https://bcec.edu.au/assets/BCEC-Private-Rental-Sector-in- Australia-Report_final_web.pdf Rowley, S., Leishman, C., Baker, E., Bentley, R., & Lester, L. (2017). Modelling Housing Need in Australia to 2025. Retrieved from Melbourne: https://www.ahuri.edu.au/__data/assets/pdf_file/0023/14297/AHURI_Final_Report_287- Modelling-housing-need-in-Australia-to-2025v2.pdf Rowley, S., & Ong, R. (2012). Housing Affordability, Housing Stress and Household Wellbeing in Australia. Retrieved from Western Australia Research Centre: https://www.ahuri.edu.au/__data/assets/pdf_file/0010/2215/AHURI_Final_Report_No192_H ousing_affordability,_housing_stress_and_household_wellbeing_in_Australia.pdf Ruonavaara, H. (2016). Theorising the Three Facets of Housing: a Sociological Perspective. A Sociology of Housing: Department of Social Research University of Turku. Saegert, S. (1985). The Role of Housing in the Experience of Dwelling. In I. Altman & C. M. Werner (Eds.), Home Environments (pp. 287-309). Boston, MA: Springer US. Saldaña, J. (2015). An Introduction to Codes and Coding. In The coding manual for qualitative researchers (Vol. 3, pp. 1-31). Sam, K. M., & Chatwin, C. R. (2012, 11-13 June 2012). Ontology-based Text-mining Model for Social Network Analysis. Paper presented at the 2012 IEEE International Conference on Management of Innovation & Technology (ICMIT). Sánchez, A. C., & Andrews, D. (2011). Residential Mobility and Public Policy in OECD Countries. OECD Journal: Economic Studies, 2011/1. doi:10.1787/eco_studies-2011-5kg0vswqt240 Sayer, A. (2000). Realism and Social Science. In A. Sayer (Ed.), Realism and Social Science. doi:10.4135/9781446218730 Saylor Academy. (2015). Housing Supply and Demand. Macroeconomics Theory through Applications. Retrieved from https://saylordotorg.github.io/text_macroeconomics-theory- through-applications/s08-01-housing-supply-and-demand.html Schwirian, K. P. (1983). Models of Neighborhood Change. Annual review of sociology, 9, 83-102. doi:10.1146/annurev.so.09.080183.000503 Settas, D. L., Sowe, S. K., & Stamelos, I. G. (2009). Detecting Similarities in Antipattern Ontologies using Semantic Social Networks: Implications for Software Project Management. The Knowledge Engineering Review, 24(3), 287-307. doi:10.1017/S0269888909990075 Shoory, M. (2016). The Growth of Apartment Construction in Australia. Retrieved from https://www.rba.gov.au/publications/bulletin/2016/jun/pdf/bu-0616-3.pdf Stapledon, N. D. (2007). Long Term Housing Prices in Australia and Some Economic Perspectives. (Doctorate of Philosophy). The University of New South Wales Sydney. Retrieved from http://unsworks.unsw.edu.au/fapi/datastream/unsworks:1435/SOURCE02 Storey, D. (2012). Territories: The Claiming of Space. Retrieved from https://ebookcentral-proquest- com.ezproxy.library.uq.edu.au/lib/uql/detail.action?docID=958674

229

Street, C., & Ward, K. (2010). Retrospective Case Study. In A. J. Mills, G. Durepos, & E. Wiebe (Eds.), Encyclopedia of Case Study Research. doi:10.4135/9781412957397 Stuart, K., & Botella, A. (2009). Corpus Linguistics, Network Analysis and Co-occurrence Matrices. International Journal of English studies, 9(3), 1-20. Retrieved from https://dialnet.unirioja.es/descarga/articulo/3156132.pdf Tableau. (2020). Find Clusters in Data. Tableau Desktop and Web Authoring Help. Retrieved from https://help.tableau.com/current/pro/desktop/en-gb/clustering.htm The Committee for Economic Development of Australia. (2017). Housing Australia. Retrieved from https://www.ceda.com.au/CEDA/media/General/Publication/PDFs/HousingAustraliaFinal_Fl ipsnack.pdf Thomas, D. A. (2014). Searching for Significance in Unstructured Data: Text Mining with Leximancer. European Educational Research Journal, 13(2), 235-256. doi:10.2304/eerj.2014.13.2.235 Thomas, M. (2017). Housing Affordability in Australia. Retrieved from https://www.aph.gov.au/About_Parliament/Parliamentary_Departments/Parliamentary_Libra ry/pubs/BriefingBook45p/HousingAffordability Tognoli, J. (1991). Handbook of Environmental Psychology In (pp. 655-690). Malabar, Florida: Krieger Pub. Co. Tourangeau, R., Rips, L. J., & Rasinski, K. (2000). The Psychology of Survey Response. Cambridge: Cambridge University Press. Troy, P. N. (2017). Physical Determinism and Australian Cities. In N. Sipe & K. Vella (Eds.), The Routledge Handbook of Australian Urban and Regional Planning. New York: Routledge. Tsemberis, S., McHugo, G., Williams, V., Hanrahan, P., & Stefancic, A. (2007). Measuring Homelessness and Residential Stability: The Residential Time-line Follow-back Inventory. Journal of Community Psychology, 35(1), 29-42. doi:10.1002/jcop.20132 UN Habitat. (2016). The Right to Adequate Housing. Retrieved from Geneva: https://www.ohchr.org/Documents/Publications/FS21_rev_1_Housing_en.pdf Unconventional Economist. (2013). The History of Australian Property Values. Australian Property. Retrieved from https://www.macrobusiness.com.au/2013/02/the-history-of-australian- property-values/ Unconventional Economist. (2014). The Consequences of our Monopoly Housing Market. Australian Property. Retrieved from https://www.macrobusiness.com.au/2014/05/the-consequences-of- our-monopoly-housing-market/ Unconventional Economist. (2018). 2018 Demographia International Housing Affordability Survey. Retrieved from https://www.macrobusiness.com.au/2018/01/2018-demographia- international-housing-affordability-survey/ University of Chieti-Pescara. (2014). Network Science. Network. Retrieved from https://www.sci.unich.it/~francesc/teaching/network/ Urban Development Institute of Australia. (2018). 10th Annual State of the Land Report. Retrieved from http://www.udia.com.au/_literature_239258/2018_State_of_the_Land Valejo, A., Valverde-Rebaza, J., & de Andrade Lopes, A. (2014). A Multilevel Approach for Overlapping Community Detection. Paper presented at the 2014 Brazilian Conference on Intelligent Systems, Sao Paulo, Brazil. van der Vaart, W., & Glasner, T. (2007). Applying a Timeline as a Recall Aid in a Telephone Survey: a Record Check Study. Applied Cognitive Psychology, 21(2), 227-238. doi:10.1002/acp.1338 Van Ham, M. (2012). Housing Behaviour. In D. F. Clapham, W. A. V. Clark, & K. Gibb (Eds.), The SAGE Handbook of Housing Studies. doi:10.4135/9781446247570 Vandenberghe, F. (2008). Structure, Agency and the Internal Conversation. Retrieved from http://www.journaldumauss.net/IMG/article_PDF/article_362.pdf Varul, M. Z. (2011). Encyclopedia of Consumer Culture. In D. Southerton (Ed.). doi:10.4135/9781412994248 Veenstra, R., Dijkstra, J. K., Steglich, C., & Van Zalk, M. H. W. (2013). Network–Behavior Dynamics. Journal of Research on Adolescence, 23(3), 399-412. doi:10.1111/jora.12070

230

Veludo-de-Oliveira, T. M., Ikeda, A. A., & Campomar, M. C. (2006). Discussing Laddering Application by the Means-End Chain Theory (1052-0147). Retrieved from http://www.nova.edu/ssss/QR/QR11-4/veludo-de-oliveira.pdf Vollaard, J. P. (2009). Political Territoriality in the European Union: the Changing Boundaries of Security and Healthcare. (Doctoral Thesis). Leiden University, Retrieved from http://hdl.handle.net/1887/13883 Walker, R. A. (1974). Urban Ground Rent: Building a New Conceptual Framework. Antipode, 6(1), 51-58. doi:10.1111/j.1467-8330.1974.tb00584.x Wargent, P. (2016). Why do Property Prices Rise Faster Closer to the City? Retrieved from https://propertyupdate.com.au/property-prices-rise-faster-closer-city/ Warwick. (2016). Building Approval Data Shows Developers Slowing Activity. Retrieved from https://www.cmaust.com/single-post/2016/11/30/Do-We-Keep-Building- Wendt, A. (1999). Social Theory of International Politics. Cambridge University: Cambridge Press. Williams, M. D., & Hollan, J. D. (1981). The Process of Retrieval from Very Long-Term Memory. Cognitive Science, 5(2), 87-119. doi:10.1207/s15516709cog0502_1 Williams, P. (2003). Housing and Social Change: East-West Perspectives. In R. Forrest & J. Lee (Eds.), Home Ownership and Changing Housing and Mortgage Markets: The New Economic Realities (pp. 162-182). London: London: Routledge. Wu, C., & Sharma, R. (2012). Housing Submarket Classification: The Role of Spatial Contiguity. Applied Geography, 32(2), 746-756. doi:10.1016/j.apgeog.2011.08.011 Wu, Y., Wei, Y. D., & Li, H. (2020). Analyzing Spatial Heterogeneity of Housing Prices Using Large Datasets. Applied Spatial Analysis and Policy, 13(1), 223-256. doi:doi.org/10.1007/s12061- 019-09301-x Wulff, M. N., & Maher, C. (1998). Long- term Renters in the Australian Housing Market. Housing Studies, 13(1), 83-98. doi:10.1080/02673039883506 Yardney, M. (2018). The Evolving Great Australian Dream [Infographic]. Retrieved from https://propertyupdate.com.au/the-evolving-great-australian-dream-infographic/ Yates, J. (2009). Tax Expenditure and Housing. Retrieved from http://library.bsl.org.au/jspui/bitstream/1/6023/1/Yates_tax_expenditure_and_housing_2009 .pdf Yuliana, I., Sukirman, S., & Sujalwo, S. (2017). A Comparison of Community Clustering Techniques: Fruchterman-Reingold and Wakita-Tsurumi. Paper presented at the ISETH |56| The 3rd International Conference on Science, Technology, and Humanity, Universitas Muhammadiyah Surakarta. Zachariah, Z. B., & Jusan, M. B. M. (2011). Means-End Chain Model Framework for Measuring Housing Environment Choice Behavior. Journal of Civil Engineering Architecture, 5(6). Retrieved from http://www.davidpublisher.org/Public/uploads/Contribute/56c186341a376.pdf Zaltman, G. (2003). How Customers Think: Essential Insights Into the Mind of the Market. Cambridge: Harvard Business Press. Zavei, S. J. A. P., & Jusan, M. M. (2012). Exploring Housing Attributes Selection based on Maslow's Hierarchy of Needs. Procedia - Social and Behavioral Sciences, 42, 311-319. doi:10.1016/j.sbspro.2012.04.195 Żero, O. (2019). Top Three Mistakes with K-Means Clustering during Data Analysis. Retrieved from https://zerowithdot.com/mistakes-with-k-means-clustering/ Zhang, G. (2004). Grice's Maxims and the Principle of Selectiveness: an Advertising Language Perspective. Foreign Language and Literature Studies, 38, 127-144. Retrieved from https://espace.curtin.edu.au/bitstream/handle/20.500.11937/24344/19757_downloaded_str eam_275.pdf?sequence=2

231

APPENDICES Appendix 1 Human Research Ethics Approval Letter

232

Appendix 2 The Brisbane GCCSA Housing Needs Survey Coverage

Postcode (coverage) is shown in blue/purple, boundary of the Brisbane GCSSA is shown in red.

Source: Author as derived from (ESRI/ArcGIS)

233

Appendix 3 Years Lived in the GCCSA (Brisbane) by Age

234

Appendix 4 Tenure by Occupation (Renters)

235

Appendix 5 Tenure by Occupation (Owners)

236

Appendix 6 Occupancy by Dwelling (18-29, 30-39 Cohort)

237

Appendix 7 Occupancy by Dwelling (40-49, 50-59 Cohort)

238

Appendix 8 Occupancy by Dwelling (60-69, 70 and Over Cohort)

239

Appendix 9 Composition of the Mean Number of Housing Moves by Age Bracket

240

Appendix 10 Reasons for Moving: Level of Decision-Making (All Cohorts)

241

Appendix 11 Total Moves: Occupation/Level of Understanding

242

Appendix 12 Property Details Scraper (Outwit)

243

Appendix 13 String Generation for Brisbane GCCSA Page Search

244

Appendix 14 Fast-Scrape of Address List for Property Details (Outwit)

245

Appendix 15 Google AdWords Targeted Location

246

Appendix 16 Facebook Advertisement Audience

247

Appendix 17 Facebook Advertisement Schedule

248

Appendix 18 Advertisement Campaigns: Google AdWords and Instagram Carousel

Approved Google AdWords Advertisement

Instagram Advertisement (Image from the Ad Carousel)

249

Appendix 19 Google AdWords Audience Insights

250

Appendix 20 Facebook Advertising Audience Insights

251

Appendix 21 Survey of Housing Aspirations Questions

252

253

254

Appendix 22 Visible Out/in-degrees (Edge Widths): Rental Network

Appendix 23 Visible Out/in-degrees (Edge Widths): Rental Apartment Network

255

Appendix 24 Visible Out/In-degrees (Edge Widths): Rental House Network

256

Appendix 25 Fastgreedy Community Network Plots: Rental and Sales Networks

Fastgreedy Communities – All Rental Network

Fastgreedy Communities – All Sales Network

257

Appendix 26 R Line-by-Line Coding: Description of SNA Plots (Centrality/Community)

Line-by-Line Coding Description Centrality Measure (CEM) & Community Measures (COM) Sub-Component Set Seed library(igraph) Sets the seed for reproducibility: 800 set.seed(800) for Centrality and 450 for Community Measures. Sub-Component Matrix Multiplication CSV <- read.csv ("Research/Thesis/Network Read the contingency table and Analysis/Matrix File.csv") create a matrix by multiplying the matrixa <- t(CSV)%*%as.matrix(CSV) rows and columns. Reconstitute the matrixa2 <- matrixa / t(replicate(nrow(matrixa), adjacency matrix by normalising the colSums(matrixa))) cells within each column. Graph the g <- graph.adjacency(matrixa2, mode = c("directed"), adjacency matrix. Simplifying the weighted=TRUE, diag = FALSE) network omits redundant edges so a g <- simplify(g) single edge is used for a dyad. Community Measures (COM) Traditional COM are read as g1 <- graph.adjacency(matrixa2, mode = c("directed"), undirected graphs, the network is first weighted=TRUE, diag = FALSE) graphed in directed mode, then re- g <- simplify(g) graphed as undirected. If the graph is g <- as.undirected(g1, mode = c("mutual")) initially undirected, igraph uses only the highest value edge of a dyad. Centrality Measure (CEM) & Community Measures (COM) Sub-Component Attribution V(g)$graph.strength <- graph.strength(g) V(g)$frame.color <- NA; V(g)$label <- V(g)$name layout1 <- layout.fruchterman.reingold(g) V(g)$label.color <- c("white"); V(g)$label.cex <- 1.1* These codes set the attributes for the V(g)$graph.strength /max(V(g)$graph.strength)+ 0.6 vertices (V) and the edges (E) V(g)$size <- 10*V(g)$graph.strength/ max(V(g)$graph.strength)+ 1; E(g)$width <- (E(g)$weight)* 60 Centrality Measure (CEM) Only typoname <- sum(matrixa2) Sets the colour coding for the home, ifelse(typoname== "72", dwelling, and house using if-else logic. house <- V(g)$label[1:17], #sales The if argument relates to sales, while house <- V(g)$label[1:5]) #rental else relates to the rental network. The ifelse(typoname== "72", $label attribute references the 72 sales home <- V(g)$label[18:44], #sales terms/and 69 rental terms in the Excel home <- V(g)$label[6:34]) #rental CSV, the headers were pre-organised ifelse(typoname== "72", to match with the cuts shown here. dwelling <- V(g)$label[45:72], #sales 1:17/1:5 (house), 18:44/6:34 home), dwelling <- V(g)$label[35:69]) #rental 45:72/35:69 (dwelling) etc. The ifelse V(g)$color <- ifelse(V(g)$name %in% house, "darkgreen", argument tells the program what colour ifelse(V(g)$name %in% home, "orange","blue")) to assign based on the cut. A function is created to assign colours pal1 <- c("blue", "orange", "darkgreen") based on a numeric cut (defined by pal2 <- c("#800026", "#E31A1C", "#FC4E2A") edge width breaks). The cut is set to pal3 <- function(v) {; third <- v[3]; second <- v[2] “3” since, 3 breaks correspond with the first <- v[1]; answer <- c(third, second, first) 3 colours. The cut divides the edge weights into thirds. 258

return(answer)}; pal4 <- pal3(pal2); fine = 3; pal = colorRampPalette(pal4); E(g)$color = pal(fine)[as.numeric(cut(E(g)$width, breaks = fine))] Sub-Component Filter Graph Centrality measure (CEM Only) maxedges <- sum(matrixa2)^2-sum(matrixa2) Identify the # of possible edges in the EdgeCount <- ecount(g); EdgeDensity <- edge_density(g) network by squaring the # of property threshold <- edge_density(g)*2.5 terms, subtracting the # of property EdgeMax <- max(E(g)$width) terms (since the square matrix will self- SumWeights <- sum(E(g)$weight) loop). Identify edge density, highest Meanwidth <- mean(E(g)$width) weighted/ mean widths. Sub-Component Graph Plotting Deletes all edge widths This section explains how the graph is threshold, "dimgrey", rgb(0,0,0,0)), edge.curved = 0.1, plotted to create the network figures edge.arrow.size=0.0, layout=layout1, which are shown in Section 7.3. The vertex.shape="sphere", main="FG Com") layout is called, the edges are coloured G22 <- gl(1, 2, 3, 4, labels = c("Very Low", "Low", "Mid", based on community membership. The "High", "Very High")); value<-runif(1.4); n<-6 legends are also created based on the size_vec<-seq_len(n); sizeCut<-cut(value,n) colour sets (mco, nco). scaled <- 1 + ((2-1) * (size_vec - min(size_vec) ) / ( max(size_vec) - min(size_vec) ) ) legend('topleft',legend=levels(G22),pt.cex=scaled,col='bla ck',pch=21, pt.bg='grey', title="Cen") Group2 <- gl(1, 2, 3, labels = c("Com1", "Com2", "Com3")); legend("topright",bty = "n", pch=c(16), pt.cex=3.5, legend=levels(Group2), col=mco, border=NA) Centrality Measure (CEM) & Community Measures (COM) dev.copy(device = png, filename = "Thesis/Network Save the plot in folder Analysis/Plot.png", width = 820, height = 500); dev.off() 259

Appendix 27 Line-by-line Coding: Description of the Clustering Coefficient Algorithms

Clustering Coefficient Method (Graph) Description Method formulated by (Csardi, 2015d). There are two primary distinctions. The global level which computes the clustering for the entire network, and the local level which computers clustering per property term. It was found that both levels were based on the degree distribution of the network (g). Code transit <- transitivity(g, type = c("global"), vids = NULL) Clustering Coefficient Method (Edgelist): This method is used in the thesis. Explanation of Outliers Description The code begins with a tnet (R Package) test of the Sales Apartment Clustering

network. The package was designed to analyse Bottom 36 terms 0.1654511 weighted one-mode, two-mode and longitudinal Top 36 terms 0.5171209 networks. The test confirms that the adjacency Mean Above Top 10 terms 0.7117295 matrix (CSV) will be processed as a weighted one- Sales House Clustering mode edgelist. The clustering coefficient is then Bottom 36 terms 0.2056923

implemented as five measures: binary(bi), arithmetic Top 36 terms 0.5159588 mean(am), geometric mean(ma), maximum(ma) and Top 10 terms 0.7247873 minimum(mi). These measures reflect the Sales Middle Clustering differences in how the weights (degrees) are Outliers Bottom 36 terms 0.1159468 summed (Opsahl & Panzarasa, 2009). Top 36 terms 0.4489248 The overall clustering coefficient of all Top 10 terms 0.6061877 networks/subgraphs is shown in Figure 7.3, but the

All Rental Clustering coefficient of outliers is shown in yellow. The overall Bottom 35 terms 0.2073547 coefficient of sales house is higher due to a

Mean Top 34 terms 0.5591823 relatively high bottom clustering, overall coefficient Below Top 10 terms 0.5689681 of sales middle is lower due to its relatively low bottom clustering. Code library(tnet) directed.net <- as.tnet(CSV, type="weighted one-mode tnet") clustering_w(directed.net, measure=c("bi", "am", "gm", "ma", "mi"))

Appendix 28 Summary of the 10 Networks/Subgraphs by Edge Properties Max # of Edge Visible Mean Cut-off Width Max Edge Sum of Network/Subgraph Edges Density Edges Edge Width (Baseline) Width Width All Rental Network 4692 1 44 0.846 2.5 2.756 3969 Apartment (Geog) 4692 1 15 0.847 2.5 2.639 3974 House (Geog) 4692 0.997 73 0.847 2.494 2.847 3965.3 All Sales Network 5112 0.999 72 0.801 2.499 5.164 4090.7 Apartment (Geog) 5112 0.983 131 0.819 2.457 4 4115 House (Geog) 5112 0.998 192 0.804 2.496 3.899 4103.6 Land (Geog) 5112 0.897 326 0.873 2.242 8 4003 Inner (Type) 5112 0.973 100 0.828 2.433 3.358 4121.5 Middle (Type) 5112 0.999 90 0.799 2.499 5.873 4080.7 Outer (Type) 5112 0.998 77 0.802 2.497 3.401 4096.4 Mean 4986 0.9844 112 0.8266 2.4617 4.1937 4051.9

260

Appendix 29 Similarity between the Three Centrality Measures.

Component Similarity between the Three Centrality Measures Normalised The Sankey charts are normalised within a scale from 0-1 using the scores of following formula: centrality min ( ) = measures max( ) min( ) 𝑥𝑥𝑖𝑖 − 𝑥𝑥 Normalisation Raw weight is subtracted𝑧𝑧𝑖𝑖 by the minimum weight in the range and Definition divided by the maximum weight 𝑥𝑥in− the range𝑥𝑥 subtracted from the minimum weight in the range. Normalisation normalised <– d (x-min(x))/(max(x)-min(x)). in igraph Spreadsheet Property terms are then ranked from 1-69 (rental) or 1-72 (sales) based Rank on the 0-1 scale. Process applies to all three centrality measures. Rank Colour The charts are colour coded by the normalised weighted degree score. Coding (WDC) The highest weighted normalised values of the weighted degree centrality (WDC) measure are shown in a gradient from light blue to dark. Changes in the WDC, Eigenvector Centrality (EVC) and Closeness Centrality (CC) ranks are shown. Sankey Sankey is based on a sigmoid function which shows the changes Interpretation between centrality measures in the form of an “S” curve. The lesser the (Slope) similarity, the greater the slope of the curves. Sankey The numeric values (0-1) are unique to each centrality measure. Thus, Interpretation the positional changes can also be understood mathematically; the (Value) lower the score, the lower the centrality score. Sankey The further down the pecking order of the 69 terms, the greater the slope of the curves. Rental 50% of the normalised values of the WDC and EVC networks are within (Distribution) the first 15 terms (20% of the rental network terms). The CC network is more uniformly distributed and standardised. Terms such as City and CBD with low centrality scores in the WDC and (EVC) measures can have significant CC since CC is computed using average distances (weights). The average change in rank from WDC to EVC is ~1.5, and the average change in rank from EVC to CC is ~4. Rental Sankey In the rental Sankey the four core property terms: room, area, (Key Terms) bedrooms, kitchen have significant weight in the network (WDC) and have a high affinity with other high centrality terms (EVC), which have high average weights (CC). These terms form a dense network subgraph. In the periphery, the relationship between these measures is less strong, and terms may have some qualities of centrality, but not all; hence the fluctuation in the slopes for lowly ranked terms. Sales 50% of the normalised values of the WDC and EVC networks are within (Distribution) the first six terms (8% of the sales network terms). 50% of the distribution of the CC measure corresponds with 40% of sales terms. The uneven distribution in all three measures means that the sales network has a more focal subgraph at its core. The average change in rank from WDC to EVC is 1, and the average change in rank from EVC to CC is 3.5. Sales Sankey The key sales terms such as Bed, Bedroom, Home and Location(ed) (Key Terms) are ubiquitous in the network and are relatively more dominant than the equivalent rental network terms.

261

Appendix 30 Explanation of Network Subgraphs by Edge Properties

Explanation of Network Subgraphs by Edge Properties Edge The maximum number of possible edges (max # of edges), edge Properties density, visible edges, mean of edge widths, cut-off width (threshold) and sum of widths are shown in Appendix 31 Visible Edges A cut-off threshold (Appendix 31) is implemented to emphasise only the most prominent edges; this is done for analytical clarity. Sum of Width The sum of width is the number of co-occurences (weight) within a property term multiplied by 60 (for emphasis). The weight is the sum of every instance where two terms coincide across all the properties of a given subgraph. The weight is likely to be greater than one prior to normalisation. Normalisation results in edge weights between 0-1. Column/Row Since weights are normalised within each column of the matrix, each Sums (see edge has a value from 0-1. The sum of all columns equate to 1, but the Table 5.19) sum of a given row could be anywhere from 0-69 (rental) or 0-72 (sales). Average Edge The average edge weight sum of all property terms is 67.532 (diagonals Weight Sum are discounted). Significance If every edge in the network were to have the same weight, the average of Visible normalised edge weight would be ~0.015 (1/67.532). Multiplied by 60 = Edges Edge width of 0.89. Thus, the average cut of width is ~3 times greater than the average edge weight, if the matrix was uniform. Thus, the visible edges are disproportionately more influential than other edges. Without a cut-off threshold, these negligible edges would stand out. Colour Each subgraph is cut into three equal sections. Edges that are in the Spectrum upper 33.33% of the width/weight distribution are labelled “Upper”, edges in the middle 33.33% are labelled “Mid”, and edges in the lowest 33.33% are labelled “Low”. Colour Due to some networks being more assymetrical, some subgraphs will Spectrum exhibit the entire spectrum. The ranges are as follows: Distribution Subgraph Threshold Ranges Low Legend Visible Edges Mid Upper R All 2.5 0 0.90948 1.81896 2.756 E Apartment 2.5 0 0.87087 1.74174 2.639 N T House 2.494 0 0.93951 1.87902 2.847 All 2.499 0 1.70412 3.40824 5.164 S Apartment 2.457 0 1.32 2.64 4 A House 2.496 0 1.28667 2.57334 3.899 L Land 2.242 0 2.64 5.28 8 E Inner 2.433 0 1.10814 2.21628 3.358 S Middle 2.499 0 1.93809 3.87618 5.873

Outer 2.497 0 1.12233 2.24466 3.401

262

Appendix 31 Multi-axis Line Graphs Showing Normalised Values of All Subgraphs

263

Appendix 32 Explanation of Multi-axis Line Graphs

Concepts Explanation of Multi-axis Line Graphs The x-axis signifies the number of edges in each network/subgraph, and the y- Multi-axis axis refers to the normalised values (weights) of all edges in the subgraphs. The Graphs highest weighted edge in each network is assigned a value of 1, and the lowest weighted edge is assigned a value of 0. Thus, all subgraphs are comparable. The gentler the slope, the higher the standard deviation, since there is more Gentle variation across a greater number of data points. This broad distribution also Slope means that on average edges are farther from the mean. The steeper the slope, the greater the number of focal terms. This means that the Steep weights in the graph are better distributed among the key vertices, rather than Slope uniformly distributed with a diluted effect across the entire network. The rental network/subgraphs have a gentle slope, and thus their highly weighted property terms are fewer, and the subgraphs are decentralised. All the visible Rental edges lie between the normalised values of 1 and 0.877. Thus, there are a greater Network number of peripheral terms without leading edges. Networks with high average density have high decentralisation and vertices with fewer information flows. This network/subgraphs have considerably more focal terms, in contrast to the dense rental networks. The visible edges in the sales subgraphs/network have a Sales All normalised value ranging from 0.48-1, for the all sales properties network, 0.62-1, & for the sales apartment network, 0.64-1, for the sales house network and 0.28-1, Typology for the sales land network. Thus, these highly centralised subgraphs have more information flows. These subgraphs are more centralised than the rental network, but less centralised than the sales typological subgraphs, except for the sales middle Sales subgraph (which is equally centralised). There are more highly weighted terms Geography (proportionally and overall) than the rental networks, though fewer than the typology network.

264

Appendix 33 List of Normalised Property Terms (WDC)

265