Reconceptualising resilience in housing policy: an actor-network approach

ABSTRACT

During the pandemic, our needs and expectations of our housing stock rapidly expanded to cope with demands of lockdowns, working from home and restricted access to other neighbourhoods. Empirical studies over the last few years indicate how concepts such as multifunctionality, responsiveness, social connectivity and localisation became critical. In many cases, the pandemic deepened existing housing-related inequities. Households already struggling with lower incomes found themselves contending with even more challenges: constrained spaces, inadequate heating and the noise of neighbours. Given the strong link between housing and health, there is an urgent need to understand the ability of our current housing stock to meet the demands of such ‘shocks’. Of particular importance is the concept of resilience in housing systems. Resilience has become ubiquitous in policy discourse but is underdeveloped and undertheorised. We thus use actor-network theory (ANT) concepts to develop a multi-level conceptualisation of resilience, arguing that ‘housing resilience’ unfolds on five levels: resilience within an individual dwelling, within housing supply, within the housing supply chain, between dwellings and communities and between dwellings and the natural environment. We then demonstrated the robustness of this conceptualisation by mobilising it as a lens for an analysis, starting with 11 state-level housing policies in Australia. We selected the ‘best practice’ housing policy with the most extensive reference to resilience. We then evidenced how our multi-level conceptualisation allows us to identify critical gaps even in this ‘exemplar’ housing policy, interrogate its structure and prioritise future action plans.

KEYWORDS:

• Actor-network theory
• housing
• social determinants of health
• resilience
• housing policy

JEL CODE:

• R58
• Z21
• Z28

1. INTRODUCTION

The global COVID-19 pandemic triggered fundamental changes in our day-to-day practices, causing a significant change to the ‘state’ of the entire housing system. From the demand perspective of owners and renters, focus shifted from workplace to home, with questions raised about where and how we were supposed to live. Households found themselves rapidly improvising to set up small home offices, accommodating increased cooking and eating at home and using bubble wrap to manage challenges of heating and noise that arose from having to be at home 24 × 7 (Horne et al., 2023). Housing became ‘the front-line defence against coronavirus’, with UN Special Rapporteur Leilani Farha claiming that ‘[h]ome has rarely been more of a life or death situation’ (United Nations Human Rights [UNHR], 2020). Typically, our homes were not built with such demands in mind. Studies show how the lack of access to adequate housing deepened health and well-being inequalities during the pandemic. In the US, underrepresented, low-income minorities including Blacks and Hispanics living in densely populated residential structures became more vulnerable to COVID-19 due to compromised capacity to self-isolate (Okoh et al., 2020). In Australia, low-income households in apartments during COVID-19 experienced challenges such as cramped spaces, noise and hampered access to digital connectivity which in turn compromised access to interactions and digital-based social care (Horne et al., 2023).

These examples call attention to the well-established link between housing and health, with housing now understood as a social determinant of health (Dahlgren & Whitehead, 1991). Even prior to the pandemic, health authorities had already pointed out that access to quality, affordable and appropriate housing are areas of considerable concern (Baker 2013; World Health Organization [WHO], 2018). For example, the rising costs of both owner occupier and rental properties in Australia means that the most vulnerable populations are likely to suffer most from lack of access (Australian Council of Social service [ACSS], n.d.), particularly in regional areas (Rowley et al., 2023). Braubach (2011) suggests that to address this complex policy problem, an interdisciplinary, collaborative approach is best. Crises such as COVID-19 further amplify the need to understand the ability of our current housing stock to address the demand for adequate housing.

From the supply perspective, change was also pervasive. Australia struggled to deliver the construction materials, products and systems necessary to fulfil contracts. Unfortunately, this not only impacted demand with delays and rising costs (Housing Industry Association [HIA] 2023). This disruption also impacted the sector, with numerous large and small companies ceasing to exist (Donaldson, 2023). The fundamental cause for such disruption to supply is caused by Australia’s lack of manufacturing and reliance on imports to cope with the housing pipeline (Haritos, 2020). The housing market routinely displays cycles; however, the COVID-induced shock has led many to question our ability and capacity to ever deliver a sustainable housing construction supply chain. Answers to such questions, for example, whether our housing ecosystem shall revert to pre-COVID, or adapt to a new ‘world’, may lie in an examination of the impact of change at micro, meso and macro levels, as well as how resilient the ‘home’ and the broader housing supply and demand ecosystem are. A high level of disturbance has confronted the housing ecosystem at various levels; this ecosystem must now be reconceptualised.

The ability of housing to respond to change and crisis can be captured under the concept of resilience. Resilience is a concept that has become ubiquitous in housing policy discourse but remains theoretically underdeveloped. This study has two aims. First, we develop a multi-level conceptualisation of housing resilience, using an approach informed by actor-network theory (ANT) concepts. One strength of our ANT-informed approach is that it allows one to bypass conceptual debates on resilience which can hamper theoretical development. Second, we demonstrate the value of this conceptualisation by applying it to an analysis of a ‘best practice’ state-level housing policy. We address the research question: What is the value of this ANT-informed, multi-level conceptualisation of resilience in strengthening Australian state-level strategic housing policies, post COVID-19?

The paper is structured as follows. We explore literature on resilience, highlighting definitions and underlying debates. We then propose ANT as an approach for systematically developing a five-level conceptualisation of resilience. Drawing on definitions and on our conceptualisation, we conduct a comparative content analysis of 11 Australian subnational housing policies to analyse how resilience is defined. We then select a ‘best practice’ policy with the most extensive use of resilience, the New South Wales Housing Strategy 2041. Finally, we demonstrate how even this ‘best practice’ policy can be further strengthened, since our conceptualisation allows us to demonstrate gaps in the policy, interrogate policy structure and formulate coherent action plans.

2. (RE)CONCEPTUALISING HOUSING RESILIENCE

Resilience has been defined in various ways, one being ‘the ability of a system to respond to unforeseen events without compromise to core functions’ and the ability ‘to survive, adapt and improve in the face of stress and change, to be able to withstand shocks, but reorganise and rebuild when necessary’ (Crabtree, 2009, pp. 2–3). The concept of resilience has become pervasive across a range of discourses, for example in the realm of policy making. The World Bank has issued a brief on resilient housing, using the term in relation to the responsiveness of dwellings natural hazards (World Bank, 2022). In the state of New South Wales (NSW), Australia, resilience is one of four pillars in a twenty-year housing strategy (New South Wales [NSW] Department of Planning, Industry and Environment 2021). ‘Resilience’ is also a term widely used in the development of housing standards and toolkits, for example, to help ensure that dwellings are ‘bushfire resilient’ and ‘flood resilient’. Unsurprisingly, resilience as a concept has been described as ‘ubiquitous’, but also ‘undertheorised’ (Porter et al., 2018) and ‘widely but uncritically deployed’ (Jacobs & Malpas, 2017, p. 395).

The challenge of theoretical development comes at least in part from debates that underpin resilience as a concept. One debate is linked to the question of ‘resilience of what?’ Some argue that definitions and conceptualisations of resilience will vary depending on whether one is describing the resilience of physical structures (Bigolin et al., 2021) or human entities (Earvolino-Ramirez, 2007), since the former can ‘fail’ while the latter can ‘bounce back’ (Manyena, 2006).

The notion of ‘bouncing back’ points to a second debate in the resilience literature, and rests on the question ‘resilience as what?’ Researchers have spoken of ecological versus engineering resilience, with ecological resilience referring to a system’s ability to find its way to a new state of equilibrium after a shock. Engineering resilience, in contrast, is a system’s ability to find its way back to its original state of equilibrium (Davoudi et al., 2012). Brand and Jax (2007) thus claim there are two definitions of resilience, one referring to a system’s capacity to return to pre-disturbance equilibrium, and another referring to a system’s capacity to arrive at a new post-disturbance equilibrium.

A third debate deals with the scope of analysis in resilience studies, or what we refer to here as ‘resilience across what?’ Sharifi and Yamagata (2018) respond to this question by citing that resilience can unfold on micro, meso and macro levels. Focusing on urban resilience, the authors cite examples such as building typology (micro), city size (meso) and transportation networks (macro) as entities that can be analysed for resilience. Lizarralde et al. (2015) propose that resilience takes place on local, community and urban scales, and reference ‘multi-scale’ resilience.

The dimensions of these debates on resilience have led to seemingly diverging trajectories for research directions. While there is no expectation that research will eventually converge into a single theoretical conceptualisation of resilience, debates left unmanaged can hamper systematic theory building. If ‘resilience’ as a concept underpins policy, strategy and action planning documents in housing, as we have shown, there is a need for a more theoretically robust conceptualisation of housing resilience.

One way to achieve a more systematic, finely grained conceptualisation in ways that bypass, or even reconcile, the sides of these debates, is to approach ‘housing’ and ‘housing systems’ as networks. A housing network can be understood as ‘individuals, institutions, resources, societal norms and expectations, environmental factors, regulatory frameworks and so forth’ (Crabtree, 2009, p. 2). A resilient housing network is one that can respond to shocks and unforeseen events ‘through an ongoing process of feedback, reorganisation and renewal’ (Crabtree, 2009, p. 2). Drawing on this network perspective, we use an approach called actor-network theory (ANT) to understand (a) how ‘housing systems’ are layers of nested networks, which thus leads to (b) a multi-level conceptualisation of housing resilience.

2.1. Actor-network theory (ANT)

There is no single definition of actor-network theory. ANT is an umbrella term overarching a range of theoretical and methodological commitments that are grounded on key concepts.

2.1.1. Human actors, non-human actors, and interdefinition

First, ANT researchers assume that much of reality comprises layers of ‘networks’ made up of human and non-human actors interacting in ways that lead to outcomes. No distinction is made between humans and non-humans, technical and social. All can bring about change. Families, communities, objects such as computers, organisations and occupied dwellings are all networks comprising heterogeneous linked together through a process of translation (Callon, 1999; Latour, 2005; Law, 1992).

2.1.2. Translation and enrolment

Translation involves among other things a ‘prime mover’ enrolling actors into a network. Actors are juxtaposed to one another in roles that are usually simplified so that they can dovetail. For an ANT scholar, these interfaces or relationships are seen as more critical than independent actors’ attributes. Actors’ identities are understood to emerge only in the context of relationships with other actors (Callon, 1999; Law, 1992).

2.1.3. Convergence and stability

The ‘interdefinition’ of actor roles allows convergence, a condition where a network of actors begins to function like a single black box. Successful networks, then, are those that are understood to achieve and maintain convergence. Networks that remain converged are described as stable, but ANT theorists would contend that such stability is always contingent. Non-human actors are seen to play a significant role in making a network stable. For example, computerised systems routinise an organisational network’s activities in ways that make its activities predictable and almost unchanging (Callon, 1991). When network activities are ‘inscribed’ into devices that are difficult to modify, they can persist over time (becoming immutable) and space (becoming mobile), thus ‘immutable mobiles’ can expand the scope of a network (Lower, 2005). That said, networks can be destabilised at any time. Such ‘contingent stability’ makes ANT aligned with the notion of resilience, as it builds on the premise that systems can converge at equilibrium, destabilise, then restabilise at a previous or new state of equilibrium (Callon, 1991, 1999).

2.1.4. Multiplicity

Actor-networks can be nested in other actor-networks, as a family is nested within a community actor-network and a country actor-network. Actors’ roles in different networks might have misaligned goals, a condition that leads to tension (Singleton & Michael, 1993).

These concepts are understood to describe all networks, whether micro entities (individuals) as well as macro entities (national economies), making ‘micro’ and ‘macro’ analytically indistinct (Law, 1992).

2.2. ANT and housing

We draw on these concepts to frame a network-based understanding of housing and housing systems, with the aim of developing an ANT-informed conceptualisation of housing resilience. Ruming (2008, p. 6) notes ‘that housing is an effect of distributed heterogeneous arrangements of materials rather than the intentional activity of humans alone’. We thus offer the argument that dwellings can be understood from a range of angles using ANT concepts:

• A single dwelling can be part of an actor-network, a household, comprising humans interacting with non-human actors like furniture, fixtures, wall panels, flooring systems, etc. Interactions can change over time, so ‘good design’ can be understood as non-humans adapting to the changing needs of humans.

• A single dwelling, an actor-network, can be seen as an end product of another actor-network. Law and Mol (1995) note that ‘machines, people, social institutions, the natural world are all effects or products’. A dwelling is likewise an ‘end product’ emerging from ‘an imbroglio of building equipment and materials, building design, relations with council staff, consultants, suppliers, unions and industry associations’ (Ruming, 2008, p. 5). This ‘imbroglio’ is the housing supply chain. A completed dwelling is part of a commodity chain, ‘a component of spatially and temporally broad production processes and a site of further practices of production and consumption’ (Crabtree, 2009, p. 7).

• In an Australian context, each completed dwelling is one of the country’s 10.9 million dwellings (ABS, 2022), what is called the ‘housing pipeline’ or ‘housing supply’. Thus dwellings are part of yet another actor-network: the housing stock that is ‘managed’, in different ways, by a government entity cast as the network’s prime mover. The Australian federal government manages by ‘monitoring’ the attributes of the dwellings enrolled in this network in terms of location (metro versus regional), dwelling type (detached, semi- detached), volume (total supply relative to demand) and tenure type. State government also ‘manages’ the state’s housing stock by monitoring, as well as in other ways. For example, it can add to this housing stock directly by developing land, or indirectly by releasing land to be developed by other stakeholders like private partners. These layers of management are important, but fundamentally government does not currently ‘create’ large volumes of this housing stock; it rests with the private sector to actively provide supply. In the past there has been a substantial direct involvement in planning, design, construction and management of what has often been termed social housing by the Australian state and federal governments.

• In addition, each dwelling is an actor embedded in a community actor-network, where it interacts with other actors such as transport systems, parks, sidewalks, schools, other dwellings and community members (Sharifi & Yamagata, 2018).

• Finally, each dwelling is ‘a physically embedded ecological entity’ (Crabtree, 2009, p. 7) nested in an environmental actor-network, where it interfaces with other actors including elements of bushfires, floods and pandemics.

The housing system, then, can be understood as a network of actor-networks, with some networks completely embedded in other networks (dwelling-in-community, dwelling-in-ecology), while others co-exist in the sense that the end product of one (the housing supply chain) is enrolled into another (housing supply). The health and resilience of the housing system requires consideration of the health and resilience of these myriad of networks.

Having framed an understanding that a dwelling can be, (1) an actor-network, (2) an immutable mobile created by a housing supply chain, (3) an actor enrolled in housing stock, (4) a network nested within a community actor-network and (5) a network nested within the natural/ecological environments, we thus offer an emerging conceptualisation of housing resilience, shown as Row 4 in Table 1.

Table 1. Developing a multi-level conceptualisation of housing resilience (refer to Row 4).

	Dwelling as a household actor-network	Dwellings as immutable mobile (end product) from housing supply chains	Dwellings as enrolled actor in housing stock	Dwellings-in-community	Dwellings-in-ecology
Role of dwelling in actor-network	Dwellings emerge and stabilise as actor-networks that arise from the interaction of human actors (family/ occupants) and non-human actors (equipment, furniture, fixtures, etc.)	Dwellings are ‘outcomes’ of the housing supply chains. ‘New housing is the end product of, inter alia, housing market operations, land availability, centralised planning, society perceptions of ownership and dwelling types, builders and developers, survey equipment, expert consultants, site topography’ (Ruming, 2008, p. 6)	Dwellings are enrolled to become part of housing stock. Housing stock changes as new stock is added (ANT: ‘enrolled’), existing stock is adaptively reused (‘redefined’ as an actor), or existing stock is decommissioned (ANT – excluded from the network). There is a distinction between the production of a dwelling (dwelling as end product) and dwelling as actor part of housing stock	Dwellings are enrolled, that is, embedded, into communities and must be interdefined relative to transport actor-networks, employment actor-networks, and other actors like parks, schools, sidewalks (Sharifi & Yamagata, 2018)	Dwellings are enrolled, that is, embedded, into ecological actor-networks (Crabtree, 2009) and must be interdefined relative to actor-network elements that drive floods, bushfires, pandemics
Examples of challenges within an actor-network	During COVID, an increase in the number of human actors within a dwelling might not be met by the enrolment of ‘additional units of space’	During COVID, the ability of housing supply chain prime movers to enrol actors like labour, material was compromised	During COVID, not enough dwellings could be enrolled into housing stock (undersupply); also, the mix of actors, for example regional versus metropolitan housing, was interrogated	During COVID, dwellings and community elements often failed redefine relative to one another. For example, interdefinition fails when there is a gap between dwellings, office locations and public transport, requiring the need for another actor, a bicycle lane, to complete interdefinition	When natural hazards take place, dwellings and ecological elements often fail to interdefine relative to one another. A dwelling might not be fit-for-purpose if it is not built to standard to withstand an ecological shock like flood, bushfire, etc.
Specific examples of resilient responses	Housing components shift as daily requirements of occupants change over their life cycles and through crises, for example through the adoption of livability standards	Housing supply chains have the capacity to enrol resources (raw materials, factory equipment) that allow them to create immutable mobiles (dwellings) that respond to demand	Network ‘prime mover’ can ‘distribute’ or mobilise supply to meet diverse needs (across locations, across population types etc.) and to respond to spikes in demand: emergency housing during pandemics, or influx of immigration	New actors are ‘enrolled’ when the relationship between dwellings and places of work are disrupted, an example being bikes and bike lanes being drawn into the actor-network when public transport is deemed inadequate as an actor	New actor components are enrolled into dwellings, components that can withstand bushfire (for example enrolling specific roof tiles and framing), flood (for example core-filled blocks instead of plasterboard).
Emerging definition of resilient housing actor-network	Household members interact with physical dwelling elements in a routine, everyday fashion such that well-being and flourishing are supported	Housing supply chain shows the capacity to readily enrol the necessary labour force, material supply, funding, while remaining embedded in a larger regulatory and international business actor-network that are aligned with the housing supply chain’s network goals	Prime mover (planner) can enrol a sufficient number of actor-dwellings; dwellings are interdefined such that the housing stock yields the right mix in terms of dwelling type, dwelling location, and to make adjustments to this network as it seeks to respond to another network	There is a responsive relationship between dwellings and other actors in the community network. Dwellings ‘work’ relative to transport, places of employment, social services etc.	There is a responsive relationship between dwellings and actors in the natural environment. Specifically, dwellings respond to environmental challenges in ways that support human flourishing

Having offered our emerging conceptualisation of housing resilience, a key question that thus arises is: What is the value of this ANT-informed, multi-level conceptualisation of resilience in strengthening Australian state-level strategic housing policies, post COVID-19?

3. METHODS

To address this question, we conducted (1) a comparative analysis of 11 strategic state-level housing policies followed by (2) an in-depth discourse analysis based on our five-level conceptualisation of resilience, focusing on a policy that was identified as ‘best practice’, defined shortly.

3.1. Comparative analysis of 11 state-level policies using content analysis

We began with a scoping review of all Australian state-level housing policies, which we analysed from January–July 2023. For context, the function of addressing Australia’s housing priorities is a responsibility shared by multiple stakeholders, including the government, which operates on three tiers: the Commonwealth government, the government of states and territories and local government. The Commonwealth government manages policies such as taxation, financial regulation and welfare; states and local governments are responsible for some homeowner supports, rental regulation and land use planning; local government functions include zoning and setting local council rates (AHURI, 2023). They do not operate independently, and many intergovernmental relationships are involved. Commonwealth funding, for example, is distributed to states and territories upon their fulfilment of specific conditions (Australian Government, 2022), while state governments mobilise local governments to administer state/territory laws. Governments of states and territories are arguably the fulcrum of housing policy as they rest between the Commonwealth and local government. Governments of states and territories are also influential as they are responsible for recognising and prioritising the unique challenges of their respective areas. In states like New South Wales and Victoria, for example, populations are high in capital cities like Sydney and Melbourne, so issues of density, access to transport and homelessness tend to be more pronounced. In the Northern Territory, populations tend to be more sparse, and housing challenges are in areas such as the provision of remote housing and meeting the culture-specific housing needs of indigenous populations.

State-level policies were selected based on the following criteria: (1) strategic level (rather than operational) state level policy, (2) the title of the policy included ‘housing’ or ‘home’ and (3) the start or end date of the policy was from 2018 onwards. A total of 11 policies representing all states and territories in Australia that met these criteria were analysed. All policies were uploaded into NVivo for analysis.

It is acknowledged that by using this selection criteria, other policies that discuss housing issues, possibly from a more oblique angle, would not have been included in this initial pool. For example, the National Construction Code (NCC) as a performance-based code sets the minimum required level of the safety, health, amenity, accessibility and sustainability of buildings. The Australian Building Codes Board on behalf of the Australian Government and each State and Territory government, produces and maintains the NCC. The NCC has clauses specific to housing design and construction but was excluded as it was not explicitly labelled as a ‘housing policy’. Australia’s Disability Discrimination Act of 1992 has clauses that are related to housing, as does NSW’s Housing State Environmental Planning Policy, but they were excluded because they did not have ‘housing’ in the title. Finally, in the health portfolio, housing is seen to be crucial to improved health outcomes, and recent analysis has shown the inclusion of strategies that specifically address the housing needs of the population (Littleton, 2022). Our assumption, however, was that documents explicitly labelled as ‘housing policy’ would have the most comprehensive discussions on housing priorities, as opposed to other policies (construction, disability, environment, health) where housing, while discussed, would be more peripheral. This approach is consistent with our theoretical sampling method. That said, these and other policies may still be subjected to the same analytical approach that we later propose (Table 2).

Table 2. List of state-level policies analysed.

Name of Policy
Housing 2041: NSW Housing Strategy (NSW)
Housing 2041: 2021–2022 Action Plan, NSW Housing Strategy (NSW)
WA Housing Strategy 2020–2030 (WA)
Affordable Housing Strategy 2020–2030 (VIC)
Northern Territory Housing Strategy 2020–2025 (NT)
ACT Housing Strategy 2018 (ACT)
ACT Housing Strategy Implementation Plan 2020 (ACT)
Our Housing Our Future 2019 (SA)
Tasmania’s Affordable Housing Strategy (2015-2025)
QLD Housing Strategy 2017–2027
Queensland Housing and Homelessness Action Plan 2021–2025

We conducted a content analysis across all 11 policies using two sets of terms relevant to our research question, focusing on ‘Resilience/resilient’ and ‘COVID-19/ pandemic’. From here we were able to use these findings to quantify the number of times these words were mentioned and identify a set of discourses related to resilience and housing, post COVID-19, in Australian strategic-level housing policy. From these searches we could also see which policies were doing most in these areas.

3.1.1. Search word: resilience/resilient

Across all policies there were 72 explicit mentions of resilience/resilient, with the Housing 2041: NSW Housing Strategy and Action Plan 2041 accounting for over one-third of these instances. Across all policies there was a clear focus on building housing design resilience in relation to climate change and natural disasters, ensuring infrastructure can withstand these types of ‘major shocks and events’, for example through including standalone electricity. There was also some focus on environmental performance standards, and consideration of different types of housing and tenures needed to address future housing resilience. Overall, the main consideration for addressing these factors was portrayed as necessary to ensure social and economic resilience of the community. Thus, the link to housing as a determinant of health was clear, alongside an economic argument, of future proofing communities to be resilient in uncertain times.

The NSW Government continues to deliver a host of measures and restrictions to help rebuild the economy and ensure the state is resilient into the future. Part of this resilience starts at home, in a place of stability and affordability that can weather the storm – NSW Housing Strategy 2041, p. 4.

3.1.2. Search word: COVID-19/pandemic

Four policies from three states (NSW, VIC, WA) mentioned COVID-19, or the pandemic 117 times, with the NSW Housing Strategy and Action Plan 2041 and Victorian Affordable Housing Policy 2020–2030 accounting for almost all of these instances.

Across these policies, the theme of social and economic resilience continued, with the acknowledgement that the COVID-19 pandemic had exposed housing affordability in Australia, and in the recovery phase after the government economic stimulus, there is uncertainty in how the housing industry and communities will recover. All policies placed a focus on the need for additional support for vulnerable population groups. There was also mention of a new kind of housing market including the mobility of population groups between metropolitan and regional areas, a reluctance to purchase a house from younger Australians, and the need for houses to accommodate working from home, rather than relying on offices to provide a suitable work environment, as per pre-COVID-19 times.

‘We know that COVID-19 has increased the need for affordable housing in the short term, and we need to respond now’ – Victoria Affordable Housing Strategy 2030, p. 5.

3.2. In-depth analysis of exemplar policy (NSW housing 2041) using discourse analysis and ANT-informed conceptualisation

Following this first phase of content analysis, we identified a ‘best practice’ policy and made it the sole focus of in-depth discourse analysis, in effect making use of a single case study. The use of single case studies has been criticised by some researchers on the grounds that they are limited in terms of generalisability. These objections have been cogently dealt with by scholars such as Flyvbjerg (2004) and Eisenhardt (1989), who explain how they can be valuable for theory-building, if they are strategically and purposefully selected. Flyvbjerg (2004), for example, explains in detail how generalisability can be achieved by picking a single ‘typical case’, a case which, by definition, shares attributes with many other case contexts. Lessons learned from a typical case study become transferable across these many other similar contexts precisely because it shares their features. Flyvbjerg (2004) also makes arguments for ‘most likely case’ and ‘least likely case’ scenarios. He draws on work exploring oligarchy who was exploring oligarchy in organisations and examined a flat grassroots organisation with a strong democratic culture. The case was initially expected to be among the ‘least likely’ contexts to be oligarchical, yet was found to be so, leading to a conclusion that cascaded across other cases: ‘If A is oligarchical, then B, C, and D, which have a few more oligarchical features, can be concluded to be oligarchical as well.’

Drawing on this approach to strategic sampling, we took a similar approach by selecting a policy that was ‘most likely’ to reflect resilience. If we could choose this ‘most likely’ example, and if we would find it to be deficient in its treatment of resilience, we could then conclude that ‘other policies’ would be deficient as well, following the argument that ‘If it is not true in the best practice case, then it is also not true for all other cases that are less exemplary’ (Flyvbjerg, 2004). In other words, we would mobilise our multi-level conceptualisation of resilience by applying this approach to the NSW Housing 2041 policy, which already made the most extensive use of the term resilience. If gaps or weaknesses related to resilience could be found in this exemplar, then it becomes plausible, though not inevitable, that other policies might also reflect gaps and weaknesses.

The distinctions between content and discourse analysis are important. Researchers have argued that content and discourse analysis have diverging ontological and epistemological assumptions. Content analysis involves counting pre-selected words and phrases. Discourse analysis assumes that texts can have fluid meanings that can shift. When using discourse analysis, synonyms or contextual information from other texts could signify that the fluid meanings associated with ‘resilience’ are present, even if the term itself was not used (Hardy et al., 2004). Our multi-level conceptualisation of ‘resilience’, which unfolds across five levels, was thus used as a tool for discourse analysis. This conceptualisation allowed us to cast a wide analytical net, drawing in texts that might have otherwise been missed using the more restrictive content analysis. Figure 1 shows a sample of our early stages of coding. We refined our analysis iteratively using Table 3, Column 1 as a guide. The number of texts per level of resilience is shown in Table 3, Column 2.

Figure 1. Early sample coding process.

Source: NSW Housing 2041, p. 51, Department of Planning, Industry and Environment 2021. Used with permission.

A screen shot of sample pages of the NSW Housing 2041 policy. The screen shot shows edges of three photographs of people in everyday activities like walking and reading, the fourth shows the edge image of a two-story semi-detached dwelling. Two columns of policy contents are in between these photos. The lines of are highlighted in different colours; a legend on the right mapping the meanings of seven colours is also provided. In the legend, magenta refers to texts speaking on the resilience of housing supply chains. Notably, there are no magenta highlights.

Table 3. Number of texts mapped to levels of resilience.

Description of code based on levels of resilience	Number of texts in NSW housing strategy 2041
A – General resilience: Broad references to ‘Resilient; ‘resilience’ and other references to adjustment, flexibility, responsiveness.	15
B – Household resilience: References to resilience in housing design, anything on re-designing within homes because of changes from COVID, e.g. home offices, home school/childcare spaces, need for protection from noise, improvements for self-isolation, etc.	15
C – Housing supply chain resilience: References to how different actor-networks create housing supply, and how (if at all) government supports or collaborate with these other actor-networks sectors.	6
D – Housing supply resilience: References to housing supply as a pipeline, including reference to how this pipeline of dwellings is of a certain volume (level of housing stock) and reflects different attributes (location, housing type, tiers of affordability).	35
E – Dwelling-in-community resilience: References to how dwellings interact with other community elements (transport, schools, park) in ways that support / hamper human flourishing.	16
F – Dwelling-in-ecological environment resilience: Reference to how dwellings interact with / protect households from natural systems (climate, natural hazards, pandemics) in ways that support / hamper human flourishing.	13

A total of 21 text samples related to resilience were identified from content analysis; 32 were identified using discourse analysis. These are summarised in Table 3.

An important point to note is that a unit of text could be coded across multiple categories. The following quote, for example, could be mapped across all five levels, along with the sixth level of ‘general’ resilience:

The four pillars of supply, diversity, affordability and resilience [A-general reference] of housing reflect the vision for Housing 2041 and are the result of a range of interrelated factors. Action in one pillar may influence others. The right type and size of housing (diversity) and housing in the right locations must be planned relative to infrastructure [D-housing supply and its attributes], the market and environmental factors (resilience) [F-environment]. The amount of housing (supply) [D-housing supply-volume] will also impact the cost (affordability) of housing. Some of these factors will be influenced by government and some by delivery partners [C-housing supply chains]. Others will be personal preferences [B-household] based on broader societal changes and trends [E-community].

Once we had completed this analysis, the researchers met to ensure coding was consistent, and any disagreements were resolved.

4. FINDINGS AND DISCUSSION

4.1. Finding 1: an ANT-informed multi-level conceptualisation of resilience reveals important policy gaps that must be addressed

Table 3 shows that most of the sample texts were in relation to the resilience of housing supply, for example in terms of responding to changes in demand for different dwelling locations, housing type and affordability. In contrast, there were only six references to resilience of housing supply chains. Four of these texts foreground the role of government as a leader in housing supply chains (‘NSW government’s land and property development organisations have a unique responsibility and opportunity to deliver’; ‘government-led residential development’). While this foregrounding can be valid given its use within a policy document, the policy glosses over the fact that the volume of housing supply over which the government has direct control is very small. As of 2021, only 3.8% of Australian households lived in social housing, either in the form of government-provided public or community housing provided by not-for-profits (Australian Housing and Urban Research Institute [AHURI], 2022). In 2021–2022, the top 100 residential builders, private firms, accounted for more than one third of the residential market share (HIA 2023).

We do not wish to downplay the importance of public housing for vulnerable communities, but the fact is that most housing in Australia continues to be delivered by housing supply chain actor-networks comprising private actors. The policy overlooks this domain as an opportunity for strengthening resilience. The policy also notes that,

… short-term actions will shape the foundations for future action plans, including (a) facilitating housing supply in appropriate locations, (b) delivering more diverse housing products and tenure types, (c) improving support for vulnerable groups, and (d) driving building sustainability and resilience.

Both (a) and (b) presuppose a healthy housing pipeline, overwhelmingly from the private sector. Yet there is still a persistent focus on government ‘taking a lead role in improving the supply, diversity, affordability and resilience of new housing’.

The policy does note the government will ‘collaborate’ with private players; however, in the entire policy, there are only four superficial references to government partnering or collaborating with these private players (e.g., ‘strengthening partnerships and collaboration … with for profit and not-for-profit developers’).

One possible explanation for this ‘superficial’ treatment is that action plans are found in a separate document, the Housing 2041 Action Plan. We note that an in-depth discourse analysis of the Action Plan has not been included in the scope of this study, due to space limitations. That said, a preliminary review of this action plan was conducted and shows similar results. Figure 2 shows an excerpt of one of five action plans:

Figure 2. Sample of a priority area.

Source: Housing 2041 Action Plan, p. 8, Department of Planning, Industry and Environment 2021. Used with permission.

The figure shows a screen shot showing Priority Area 3 of the Housing 2041 action plan On the left is a text box with comments that highlight how priority areas such as these tend to highlight a narrow government housing agenda for example aims for social housing and management of government-owned assets. The focus on narrow areas is highlighted to support our argument that the action plan focuses on a restricted part of housing supply and housing supply chains, while ignoring domains that directly impact 90+% of housing supply created and managed by private companies.

To take one example: Priority Area 3 begins by foregrounding how the government is ‘committed to working with the community and private sector partners’, which is sound. However, the document then notes the focus of this partnership is ‘to deliver housing projects on government-owned assets or land’, which we argue limits the focus of the partnership to unnecessarily narrow outcomes. If government and the private sector are partnering, we argue that the benefits of this partnership should not be directed to improving 3.8% of housing stock. Partnership benefits should be sought to impact on the remaining 90%+ of the housing stock supplied by the private sector.

Our ANT approach thus points to one significant improvement to policy. The full housing supply chain actor-network must be treated as foundational to the housing system and prioritised in terms of policy reform. There is very limited ‘housing stock’ for the government to manage if the housing supply chain that produces most of this stock is left to flounder. Government must look beyond its responsibility for public housing and targeted development of its land. Instead, government must acknowledge that it also has a significant responsibility to strengthen overall housing supply chain resilience, which entails creating a robust playing field for private players. This would mean acknowledging and seeking to address, possibly as a regulator, the ‘shocks’ that buffet the housing supply chain actor-network, including uncertainty in the supply of building materials, fluctuating shipping costs, the impacts of migration changes and the exodus of labour from the industry (HIA 2023). Too much focus on its limited role as a developer has obscured other ways that government can strengthen housing supply chains. Government can, for example, seek to design and implement ‘fit for purpose’ regulation that is responsive to needs, easy to monitor and implement and supports a level playing field in residential construction. Governments can also seek to change the industry’s culture, starting with counterproductive tendering practices, including the excessive focus on cost criteria that perpetuates fragmentation and unhealthy cash-driven business models. However, changing the tendering practices does not really address the 90%+ and certainly does not address the lack of manufacturing and ability to service the housing needs. Our manufacturing capacity will need significant economic drivers to re-establish self-sufficiency in the sector to reduce international reliance. This provides an important yet untapped direction for government; i.e. public private collaboration to reduce risk in the overall housing supply chain ecosystem.

4.2. Finding 2: an ANT-informed multi-level conceptualisation of resilience shows how values like resilience can exist as part of a hierarchy, and how resilience has the potential to be a key organising principle of housing strategy

Following Finding 1, we argue that if housing supply chains are resilient, a foundation is established so that housing supply can become more resilient; producers can now create a housing pipeline that responds to changing housing needs. The ability to discern this argument is hampered because of the way resilience is currently positioned in the policy document. In some, not all, areas, ‘resilience’ appears to be narrowly defined as responsiveness to ‘environmental factors’, as shown in Figure 3:

Figure 3. One way that resilience is operationalised in Housing 2041.

Source: Housing 2041, p. 21. Department of Planning, Industry and Environment 2021. Used with permission.

A screen shot from Housing 2041 showing three bullet points that define resilience, but in very narrow ways. ‘Resilience’ is narrowly defined as the improvement of a standard called BASIX, as well as environmental and ecological responsiveness.

Resilience, then, is currently presented as one of four values simply sitting alongside diversity, supply and affordability:

The right type and size of housing (diversity) and housing in the right locations must be planned relative to infrastructure, the market and environmental factors (resilience). The amount of housing (supply) will also impact the cost (affordability) of housing.

In contrast, our ANT conceptualisation of resilience drives us to offer tighter, hierarchical interrelationships among these four pillars, as depicted in Figure 4:

Figure 4. Proposed hierarchy of values.

A four-level figure, broken down from left to right, showing how Resilient Housing Supply can be broken down into two dimensions related to supply and demand. Demand resilience is then linked to housing affordability, while supply resilience is broken down into four types of diversity. The figure emphasises that resilience is the ‘root’ value while affordability and diversity are subsumed within it.

Figure 4 shows a hierarchy, and we argue that the policy should be reorganised to reflect this hierarchy. Resilience is the foundational value. When housing supply is resilient, both ‘adequate supply’ and ‘appropriate diversity’ are achieved. ‘Affordable housing’ then emerges as one dimension of diversity: it foregrounds that there is a demand for housing at different price levels, one of which is ‘affordable housing’. This is different from ‘housing affordability’ which, in turn, unfolds as the general affordability levels of housing supply.

Instead of building a policy based on ‘[t]he four [separate] pillars of (a) supply, (b) diversity, (c) affordability and (d) resilience’, we would argue for a policy with resilience (d) as a foundation, leading to the pillars of (a) adequate supply and (b) appropriate diversity. Housing affordability and affordable housing are embedded into (a) and (b).

4.3. Finding 3: an ANT-informed, multi-level conceptualisation of resilience paves the way for identifying priority strategies, starting with those that address multiple levels of resilience. One example we provide here is offsite manufacturing

We have so far argued three points. First, housing resilience unfolds on five levels (our conceptualisation). Second, housing supply chain resilience is foundational and policy documents such as this should focus more on this area (Finding 1). Third, resilience can be a central principle for organising and shaping policy, instead of being one ideal sitting alongside others (Finding 2).

Drawing these threads together, action planning within a policy context becomes much more streamlined. Currently there are five Priority Areas listed under the Housing 2041 Action Plan (Figure 5):

Figure 5. Priority areas, Housing 2041 Action Plan.

Source: Housing 2041 Action Plan, p. 6, Department of Planning, Industry and Environment 2021. Used with permission.

An image showing five green boxes arranged from left to right and numbered from 1 to 5 as priority areas in the Housing 2041 Action plan. Box 1 mentions access to data for decision-making; 2 is on providing regulatory and guidance frameworks for decision-making; 3 is on managing government assets; 4 is on research for innovative building and 5 is on partnering with local government and communities. Again, private sector players are glossed over in these box contents.

Our five-level conceptualisation provides a basis for assessing the robustness of these priority areas. While we note that a full-scale analysis of this Action Plan is outside the scope of this study, a preliminary analysis already shows some limitations. As discussed earlier, Priority Area 3 draws on the fuel of government and private sector partnerships but funnels the strength of this partnership into a limited domain: government-led development, a tiny slice of overall housing supply. Priority Area 5 also appears quite limited, as it foregrounds ‘local governments and communities’ but fails to mention the private entities that make up most of the housing supply chain. With respect to Priority 1 and Priority 4, there have been long standing quality research efforts in urban and regional housing pursuits with the work of the Australian Housing and Urban Research Institute, and allied researchers. However, the focus has tended towards the upstream decision making with respect to urban planning and social geography concerns, and far less attention on the design, construction and organisation of the private sector, which is where significant disruption has occurred (HIA 2023).

One example of a priority area that we would propose, because it responds to all five levels of resilience, would be the increasingly widespread adoption of offsite manufacturing (OSM) in residential construction. OSM, which has been widely adopted in Scandinavia and Japan, has been linked to benefits such as reduced building costs, shorter build schedules, greater predictability in cost and scheduling, better quality and improved environmental performance (Marin-Guzman & Cranston, 2019). Despite strong advocacy for adoption, widespread OSM adoption remains embryonic in Australia, with only a quarter of the number of prefabricated manufacturers that Sweden does (Steinhardt et al., 2020).

We argue that OSM should be included as a priority action item as it strengthens the resilience of all five actor-networks, collectively leading to ‘resilient housing’. Prefabricated houses can be built ‘in days instead of years’, making for more responsive housing supply and helping address the housing shortage. Such houses have also been found to be environmentally more responsive, for example in terms of energy-efficiency (Sweeney, 2023). OSM has been used by the NSW government to provide basic temporary housing needs of vulnerable communities, specifically people displaced by floods (Chenery, 2022). Prefabricated modular components can support the implementation of rapid design solutions within structures like dormitories during pandemic outbreaks (Dişli & Arslan, 2023), a practice that can be extended to individual dwellings.

Most importantly, OSM can strengthen housing supply chains. In Australia, OSM advocates have pushed for an escalation in OSM adoption, with one leading university aiming for growing the prefabricated market share from five percent to 15 percent by 2025. This rate of escalation will lead to 20,000 new jobs and $30 billion in economic growth (Bleby, 2018). The government’s role can be to support private housing supply chains through novel methods of demand aggregation, so that multiple projects across regions can be bundled and OSM suppliers, previously constrained to the limits of single projects, have increasing opportunities to achieve economies of scale. OSM appears and reappears on the housing supply debate with frequent and monotonous regularity with slight adjustments and movement forward. If Australia is to address the resilience, we need to develop a coherent evidence-based approach to an offsite manufacturing agenda, which links Priorities 1 and 4. Our housing research agenda has always tended to focus on narrow planning and land supply concerns with little attention to the entire supply chain. The costs, delays and disruption to supply we continue to experience provide clear signs that this distressed sector needs refocus and the collaborative practice of public and private sectors to achieve resilience.

5. CONCLUSION

We began this study by discussing resilience in relation to the COVID-19 pandemic. As we developed our argument, we built an ontological understanding of the housing system as heterogeneous networks nested into each other, leading to a theoretical multi-level conceptualisation of resilience with explanatory power for the resilience of individual dwellings, dwellings in community, dwellings in ecology, housing supply and housing supply chains. Our conceptualisation cuts through complex dichotomies and debates that have unnecessarily hampered research and policy-making related to resilience. Because it is ANT-informed, our conceptualisation does not need to distinguish between technical versus social resilience, or resilience at micro, meso and macro levels, since ANT treats these concepts as analytically indistinct (Law, 1992). Furthermore, the ANT concept of ‘contingent stability’ means there is no need to distinguish between ‘old’ and ‘new equilibrium’ given that equilibrium is something that is constantly being recreated (Callon, 1999). Finally, our multi-level conceptualisation provides a way to position resilience within a hierarchy of values, allowing us to clarify how it can be a first order principle relative to other values like diversity, supply and affordability.

The actor-network approach we used for developing an understanding of resilience can arguably be expanded to understandings of other housing-related concepts. For example, in the international community, the right to housing is recognised in the International Convention on Economic, Social and Cultural Rights (ICESCR), the Universal Declaration on Human Rights (HDR) and UN Habitat III. References are made to the concept of ‘dignity’ as underpinning housing as a human right, yet limited consensus has been achieved it terms of its definition. The same can be said about the concept of ‘human flourishing’. ANT concepts can provide the scaffolding for developing more robust definitions for these concepts, which have long since been debated and have yielded sometimes widely inconsistent interpretations in national and international courts (Moon, 2018). Emerging definitions guided by ANT, as we have shown, have the advantage of considering humans and non-humans and cutting across the micro, meso and macro units of analysis, bypassing often thorny debates. In fact, the beginnings of an actor-network approach can already be seen in the work of researchers who have begun to look at how ANT can inform ‘care theory’, and how ‘the insights and methodology [of ANT proponent Bruno Latour] cultivate the possibility of a care network assemblage approach’ (Flower & Hamington, 2022, p. 45).

We used our conceptualisation to conduct a discourse analysis, thus providing methodological guidance for closely examining a ‘best practice policy’ that built on resilience as one of four pillars. While we do acknowledge that some researchers are concerned about the so-called limits of a single case study, we align with researchers (Eisenhardt, 1989; Flyvbjerg, 2004; Gioia et al., 2013) who note that this approach is valid for theoretical sampling and can yield generalisable results (Flyvbjerg, 2004). Our methodology has allowed us to identify gaps in the policy and, more fundamentally, has shown how policy can be restructured using resilience as a central principle, while paving the way for identifying priority action plans. The approach we offer here is useful for developing as well as critiquing agendas, plans and instruments, in the field of housing and beyond.

AUTHOR CONTRIBUTION STATEMENT

Conceptualisation, Z.P.; methodology, C.L. and Z.P.; validation, Z.P., C.L. and K.L.; writing – original draft preparation, Z.P.; writing – review and editing, Z.P., C.L. and K.L. All authors have read and agreed to the published version of the manuscript.

ACKNOWLEDGEMENTS

Z. Pablo and K. London would like to acknowledge the members of the multi-institutional research team of the ARC Linkage Project ‘Constructing Building Integrity: Raising standards through professionalism’ (LP190101218) for insights on the theoretical conceptualisation of resilience in this paper. Z. Pablo would like to acknowledge the teaching team of the course Sustainable Development Goals: Law and Policy (Cambridge University) for insights on possible implications of the work on international law and policy.

DATA AVAILABILITY STATEMENT

Data from NSW Housing 2041 is publicly available. The State of New South Wales through Department of Planning, Industry and Environment 2021 notes in the policy document that ‘ [y]ou may copy, distribute, display, download and otherwise freely deal with this publication for any purpose, provided that you attribute the Department of Planning, Industry and Environment as the owner’.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.