Abstract

Tenure insecurity, including eviction/landlord-ended tenancy and precarious tenure, has multiple social and health consequences. International research has identified a range of risk factors for insecure tenure, many of them consistent with other housing discrimination risks. This study used a 2018 population-weighted survey to compare likelihood of landlord-instigated tenancy end and frequency of moves across social and demographic groups in Aotearoa New Zealand. Non-heterosexuals, young adults, and households with no spare bedrooms had high levels of renting, and were in precarious tenure (high mobility, highly landlord-instigated). Pacific peoples, sole parents, and people with constrained finances also had high renting levels, and were reluctant movers (lower mobility, but highly landlord-instigated). Couples, adults 25 to 44 years, and ‘large urban’ dwellers had mixed renting levels, but were self-mobile (high mobility, by choice). No groups were tenure secure (low mobility, less landlord-instigated).Landlord-ended tenancy risks were generally consistent with previous international research, but ‘non-heterosexuals’ were a newly identified at risk group. Survey data were collected when ‘no cause’ notice was legal. Future research could investigate whether 2021 to 2024 law changes narrowing landlords’ legal reasons to end tenancies lowered the risk of landlord-ended tenancy for vulnerable groups.

Keywords: Tenure security; involuntary mobility; eviction; health determinants; housing discrimination.

Introduction

In this paper, we describe the demography of landlord-ended tenancies and precarious tenure, and identify risk factors for tenure insecurity in Aotearoa New Zealand (NZ) prior to 2021 legislation changes (described in Table 1). We used data from the Statistics NZ General Social Survey (‘the Survey’) to compare the likelihood of landlord-instigated tenancy end and frequency of moves across social and demographic groups in NZ. Non-heterosexuals, young adults, and households with no spare bedrooms had high levels of renting, and were in precarious tenure (high mobility, highly landlord-instigated). Pacific peoples, sole parents, and people with constrained finances also had high renting levels, and were reluctant movers (lower mobility, but highly landlord-instigated). Couples, adults 25-44 years, and ‘large urban’ dwellers had mixed renting levels, but were self-mobile (high mobility, by choice). Importantly, no groups were tenure secure (low mobility, less landlord-instigated).

These risk factors are important because tenure insecurity and precariousness add to the layers of barriers to equitable outcomes for people with limited social and financial resources. In order to develop policies to mitigate tenure insecurity – or when reinstating policies likely to exacerbate tenure insecurity – it is important to understand who these policies are most likely to affect. The findings here establish an important baseline for estimating the impact of 2021 legislation changes removing ‘no cause’ notice, and 2025 changes reinstating it; and are also potentially relevant to other countries moving to improve tenure security (Carr & Kirton-Darling, 2024).

In this paper we use the term landlord-ended tenancy rather than eviction for two reasons. First, eviction has a specific legal meaning in NZ, which we detail later in the paper. Second, because of eviction’s extreme circumstances, the terms eviction or evicted can carry stigma (Grainger, 2021). Tenants may be unwilling to describe themselves as having been evicted when they have left a tenancy in a routine way after the landlord has ended it. Respecting respondents’ right to define their own circumstances, we therefore describe tenancies ended by the landlord, rather than the tenant, as landlord-ended tenancies rather than eviction.

Background

Tenure insecurity, including eviction/landlord-ended tenancy and precarious tenure, is a social and public health issue (Desmond, 2016). Adverse consequences for tenants include not only the effects of actual involuntary move, such as disruption to children’s schooling, or the cost of a move; but also wide-ranging health effects (Bower et al., 2021; Hoke & Boen, 2021; Nathan et al., 2022; Smith et al., 2024), and the psychological (Carrere et al., 2022; Gillespie et al., 2020; Nathan et al., 2019; Pleasence & Balmer, 2008; Vásquez-Vera et al., 2021) and social (Collins et al., 2018) effects of precarious tenure. Tenants may be unwilling to establish community connections, or truly settle into a home, in case they must move again soon (Byrne & McArdle, 2022; Ong ViforJ et al., 2023; Soaita & McKee, 2019; Waldegrave & Urbanová, 2016). Each move requires learning how to manage a new indoor thermal environment (Serjeant et al., 2022), increasing cold and damp-mediated health risks in the interim (Riggs et al., 2021). Furthermore, tenants can also experience follow-on trauma from eviction, such as child maltreatment (Bullinger & Fong, 2021) or loss of identity and sense of home (Power, 2022). Likelihood of worse outcomes has been found to be higher for people whose reasons for housing insecurity are economic, compared to legal (Vásquez-Vera et al., 2022).

In NZ, risk factors for eviction have not been quantified. There is qualitative evidence of the effects of landlord-ended tenancy (Chisholm et al., 2022; Chisholm et al., 2017; Zwaan et al., 2017), particularly for older renters (James et al., 2020; James & Saville-Smith, 2016, 2018). Witten et al.’s 2017 study found 30 per cent of surveyed tenants gave the reason they had left their last tenancy as ‘landlord sold the house’, and two per cent as ‘was evicted’ (Witten et al., 2017), but that study did not assess risk factors for having the tenancy ended by the landlord, and some landlord-ended tenancies would have been categorised among the 11 per cent with ‘other’ reasons for move. One quantitative paper has shown that sexuality and gender diverse youth had experienced high rates of landlord-ended tenancy (56 per cent), and that it was significantly associated with having experienced homelessness (Fraser et al., 2022).

Renting in New Zealand

At the time the survey was conducted, roughly a third of NZ households rented their homes, with 83.5 per cent of tenants renting from private landlords, rather than public or community housing providers (Stats NZ, 2020). While 80 per cent of NZ landlords own only one property, the remaining 20 per cent own 73 per cent of private rentals (Bell, 2021). Forty percent of all private rentals are commercially managed (Ministry of Housing and Urban Development, 2022).

In 2018, most tenants (67.5 per cent) were on ‘periodic’ tenancy agreements, which have no fixed end date, rather than on fixed-term tenancies (29.7 per cent) (Stats NZ, 2020), which become periodic at the end of the fixed term unless both parties agree to a new fixed term, or unless one party ends the tenancy. Fixed terms appear to be most commonly one year (Landlords.co.nz, 2016; Moger, 2023; Quinovic, 2020; Stats NZ, 2020), though actual data on fixed term tenancy lengths are scant. Census data from 2018 showed nearly 40 per cent of tenants in private rentals had lived at their current residence less than a year; and less than 20 per cent had lived at their current residence for more than four years (Stats NZ, 2020). In 2019, the NZ Property Investors Federation reported the average tenancy duration amongst its members was two years, three months (Gibson, 2019).

This high mobility is likely to be connected to the legal framework governing tenancies, which has historically made it easy for landlords to end a tenancy with relatively short notice periods. At the time the Survey was conducted, legal grounds for a landlord to end a tenancy included sale of the property; the landlord or landlord’s family member wishing to move into the property; the landlord seeking a legal possession order due to overdue rent, damage to the property, or threatening behaviour by the tenant; a fixed-term tenancy having reached the end of its term; or simply landlord choice, with landlords entitled to give ‘periodic’ agreement tenants 90 days’ notice without providing any reason (Residential Tenancies Act, 1986), leaving all tenancy security subject to the goodwill of landlords. Table 1 summarises the legal grounds and notice periods for ending tenancies, before and after 2021 and 2025 legislation changes.

Tenancies can also end in other ways. A tenancy may be ended immediately when a court or tribunal grants ‘possession’ to the landlord because a tenant has not vacated at the end of a notice period or is otherwise in breach of their obligations, most commonly payment of rent (Toy-Cronin & Bierre, 2022); or when a tenant has breached a court mediation agreement with a termination clause. And, finally, there is eviction, which in NZ law is the act of physically removing the tenant and their belongings from the property and is the final legal step available to a landlord if the tenant remains in the property despite a possession order having been issued. As noted earlier, we use the term ‘landlord-ended tenancy’ throughout this paper to encompass all landlord-instigated tenancy endings, while recognising that in international literature, landlord-instigated tenancy end is categorised as ‘eviction’.

Risk factors for landlord-ended tenancy risk, and housing discrimination

Previous studies in high-income countries have found a range of demographic factors associated with increased risk of eviction. Tsai & Huang’s (2019) review identified four main risk factors for studies from the USA, Canada, the UK, and the Netherlands: financial hardship; sociodemographic characteristics such as larger households or having more children; gender (sometimes); and substance abuse and other health problems. Additional eviction risk factors identified in one or more of their reviewed papers could be loosely classified as demographic, socioeconomic, or health and social risks. Demographic risk factors were gender (though which gender was at risk varied by study); younger age; and being Black or Latino (but being of Asian ethnicity reduced risk). Socioeconomic risk factors identified were financial difficulties, job loss, number of people in the household, no high school education, and social network disadvantage. Health and social risks identified were mental illness, hoarding disorder, experience of violence, and neighbourhood crime (Tsai & Huang, 2019).

Other studies not included in Tsai and Huang’s review, carried out in Sweden (Stenberg et al., 1995), Australia (Beer et al., 2006), the Netherlands (Van Laere et al., 2009), and the USA (Friedman & Nakatsuka, 2021; Hobbs, 2021; Lochhead, 2021) had similar findings, or identified the following additional risk factors and at risk groups: receiving welfare benefits, involvement in crime (Stenberg et al., 1995), living alone, or being a solo parent, or older man (Beer et al., 2006). At a regional level, Kang (2021) also identified low car ownership as associated with higher eviction levels; while Merritt & Farnworth (2021) observed that ‘tenant-friendly’ policies can reduce state eviction rates and disparities.

Where eviction risk factors represent membership of a particular population group, they have also all variously been identified as risk factors for housing discrimination. Younger age, gender, ethnicity, health (including mental health) status, household size, having children, and being a survivor of violence (Lapidus, 2002; Ross, 2007), all appear in housing discrimination literature, along with sexual orientation, (Wilson et al., 2021) migrant status (Mensah & Tucker-Simmons, 2021), and having a disability (Aalberts, 1998). Most are also statuses against whom discrimination in the allocation of housing is prohibited under s21(1) of the NZ Human Rights Act 1993 (Soda et al., 1993). As we can see, housing discrimination and eviction are closely linked.

Like tenure insecurity, housing discrimination also has adverse health and psychosocial effects. Race- or ethnicity-based housing discrimination is the most widely explored, with a recent review listing the detrimental health or wellbeing effects of racialised housing discrimination as including pre-term birth, lower life satisfaction, psychological stress, increased mental illness, poor self-reported health, and uterine fibroids (Cross et al., 2023). Mental health and self-rated health are diminished by intersectional discrimination in housing (Lu et al., 2022). This is consistent with adverse health effects from discrimination on other grounds, and in other areas of life (Bastos et al., 2015; Sutter & Perrin, 2016). Housing discrimination, and landlord judgements of what constitutes a ‘good’ tenant can lead to eviction, particularly where regulation and protection of renters’ rights are minimal (Byrne & McArdle, 2022; Power & Gillon, 2022). These findings show the intersection between housing discrimination and eviction risk, and highlight the vulnerability of marginalised groups in the rental market.

Method

This paper draws on data from the 2018 iteration of Statistics NZ’s (Stats NZ) biennial General Social Survey. The core Survey is extensive, with 18 sections, and in 2018 included a ‘housing and physical environment’ supplement. Questions drawn on for this study cover the respondent’s age, sex, ethnicity, sexual orientation, location, household makeup, disability, employment status, economic indicators, crowding, current tenure and landlord sector, previous tenure, and main reasons for their last move (Stats NZ, 2018a).

The Survey interviewed 8,793 people aged 15 years and over, living in private dwellings of all tenures across NZ, over the period 1 April 2018 – 31 March 2019. One person from each selected household was interviewed, following a three-stage sample selection method designed to provide weighted national-level estimates. More details on the Survey methodology and reliability of survey estimates are available from Stats NZ Stats NZ, 2018b; Stats NZ, 2018c).

We accessed anonymised Survey responses in NZ’s Integrated Data Infrastructure (Stats NZ, 2021b). All measures used survey weights provided by Stats NZ. Analysis was carried out in StataSE 16.1. Additional descriptive data were sourced online from Stats NZ (Stats NZ, 2018d). Ethics approval for the study was granted by the University of Otago Human Ethics Committee HD23/043.

Sociodemographic and socioeconomic variables

Variables measured are shown in Table 3. Gender was collected as ‘male’ or ‘female’ and did not collect transgender or other gender identities. We assigned ethnicity as multiple response for Māori, Pacific Peoples, Asian Peoples, Other, and ‘MELAA’ (a Stats NZ composite group of Middle Eastern, Latin American and African); or to a residual ‘NZ European’ group, defined as non-Māori, non-Pacific, non-Asian, non-MELAA, non-Other. We assigned neighbourhood level socioeconomic deprivation quintile using NZDep2018 (Atkinson et al., 2020). Sexual orientation was collected as ‘heterosexual or straight; or ‘gay or lesbian’, ‘bisexual’, or ‘other’, which we aggregated to ‘not heterosexual’ due to sample size. We hope future research has sufficient data to disaggregate. We clustered ‘like regions’ by similarity of early univariate results and included them as much to control for, as to look for, regional variation. Urban classification was assigned using Statistics NZ’s ‘Urban rural indicator’ (Stats NZ, 2021a). Household crowding was measured using the Canadian National Occupancy Standard (Stats NZ, 2012). Participants were defined as having a disability if they answered ‘yes’ to any of a range of questions about visual impairment, hearing, mobility, mental faculties, self-care, and communication.

For each included demographic group and response variable, we measured three variables: likelihood households were currently renting; likelihood ‘the last tenancy was ended by the landlord’ for those previously renting; and ‘Mean Mobility Score’ for those currently renting.

Previously and currently renting, and sector of current landlord

Respondents were asked ‘I’d like you to think about the last place you lived before moving to your current address. Did you or your household pay rent for that house or flat?’ Those who responded ‘yes’ were classified as ‘previously renting’. Respondents were classified as ‘currently renting’ if they responded ‘yes’ to the question ‘Do you or anyone else who lives here, pay rent to an owner or agent for this house or flat?’

Mean Mobility Score

Mobility questions are shown in Figure 1. Survey respondents were assigned a mobility score as follows: last moved ten or more years ago=0; last moved five to ten years ago=2; otherwise = (times moved in last 5 years) + 3. These scores were assigned so that having moved more recently, and more often, resulted in a higher score, as illustrated in Table 2. Scores were then averaged across variable groups to create a Mean Mobility Score (Mobility Score, or MMS). A higher Mobility Score indicated people in the group were more mobile than people in groups with a lower Mobility Score.

Landlord-Ended Tenancies

Respondents who were previously renting were asked the main reason they moved out. The range of coded responses is shown in Figure 2. Respondents who had answered ‘the tenancy was ended by landlord e.g. landlord sold house or flat, expiry of lease agreement’ were classified as having had ‘landlord-ended tenancies’ (LET).

Statistical analysis

We first calculated the percentage of respondents who were currently renting in any rental sector, whether previously renting or not (Figure 3 and ‘% Renting’ in Table 3), total and by sub-group. Next, we calculated percentages for the reasons previously renting respondents and continuing tenants (previously renting and also currently renting) had left their last tenancy (Figure 2), and then the percentage of previously renting respondents in each sub-group whose last tenancy had been ended by the landlord (Figure 4 and % LET in Table 3); and the Mean Mobility Scores for those currently renting (Figure 5 and MMS in Table 3).

We modelled single and full multivariate differences between reference and comparison groups for previous renters’ landlord-ended tenancy likelihood and for current renters’ Mobility Scores (Figure 6 and Table 3). For landlord-ended tenancy likelihood we calculated odds ratios using logistic regression. For Mobility Scores, as raw data were moderately over dispersed we tested both negative binomial regression and Poisson regression with the survey-weighted data. Results were identical.

In each full multivariate model, we elected to use the questions ‘have you cut down on spending on shopping or local trips’ and ‘how limited would you feel in buying a non-essential item costing $300’ as better markers of individual financial stress than NZDep2018 quintile. The full models included all variables listed, meaning that results control for other listed variables, and that identified areas of advantage or disadvantage are cumulative.

Tenancy precariousness

After statistical analysis, demographic groups with either high or low likelihood of landlord-ended tenancy or mobility were categorised into ‘tenancy precariousness’ classifications, as follows. ‘High’/’Low’ meant the Mobility Score or likelihood of landlord-ended tenancy was above/below the 95 per cent confidence interval of the mean Score/likelihood and was shaded in pink/green in Table 3Table. At least one of the variables was required to be significantly different from the group baseline comparator (or, if baseline, from at least one other category) for the group to qualify for classification.

• ‘Precarious tenure’: high Mobility Score, high incidence of landlord-ended tenancies.
• ‘Reluctant movers’: low Mobility Score, high incidence of landlord-ended tenancies.
• ‘Self-mobile’: high Mobility Score, low incidence of landlord-ended tenancies, (that is, moving regularly, but by their own choice).
• ‘Tenure secure’: low Mobility Score, low incidence of landlord-ended tenancies.

Strengths and limitations

Statistics NZ conducted the survey and calculated sample weights so as to make results nationally representative. However, the study does have limitations. It is possible some respondents selected ‘housing costs too expensive/moved to a cheaper house or flat’ rather than ‘tenancy was ended by landlord’ if their tenancy had been terminated for non-payment of rent, but only 5.9 per cent of respondents selected the ‘housing costs too expensive’ answer and adding them into the models made no meaningful difference to the results.

Details of the individual or household were collected at the time of the Survey, which was administered after any move. This is an important limitation to the household demographic data used in our analysis. While few people will have changed their sex, sexual orientation, or ethnicity, other characteristics have the potential to have changed when the respondent moved. In most instances, the potential bias arising from this data gap is towards the null. Risk by age will have been skewed slightly upwards, however the age groups are wide and average length of tenancy short, so the skew will be small.

In reporting demographic characteristics as risk factors for landlord-ended tenancy, we recognise that except for single-person households, the risk attaches not (only) to the individual, but also to the person’s household. However, it also means that characteristics of other household members associated with higher forced moves were unable to be measured, potentially underestimating any effect.

The inclusion criteria for the study will also have excluded people living in ‘emergency’ or ‘transitional housing’ (Human Rights Commission, 2022) from the analysis, estimated at 2,879 and 2,341 people respectively in June 2018 (Ministry of Social Development, 2018). Although some of these people may have been in employment, and total numbers represent only a small proportion of total unemployed, it is possible that findings for people who were unemployed could have been skewed by those in worst housing deprivation being excluded from the analysis. The 2018 data show a lower likelihood of landlord-ended tenancy than Witten et al.’s 2015 tenant survey (Witten et al., 2017). It is not possible to tell whether the difference is due to changes in property investor sales activity, to different sampling methodologies, or to some other difference between the data sources.

It is also important to note that while the Survey sampling resulted in a roughly representative proportion of public and community housing sector tenants (16 per cent of currently renting respondents), at the time of the survey NZ’s central government public housing provider was phasing in a ‘Sustaining Tenancies’ programme, aiming to work with tenants to avoid eviction except in rare circumstances (Mailing, 2022). The subsequent government ended the ‘Sustaining Tenancies’ policy in 2024, adding further tenure insecurity to present-day renters.

Other factors identified as risks in other studies, such as job loss, mental illness, hoarding disorder, experience of violence, neighbourhood crime, or substance abuse problems, were not able to be measured in this study.

Results

Full results are shown in Table 3.

Descriptive analysis

Rental tenure

Just under a third (32.1 per cent) of Survey respondents were ‘currently renting’. Likelihood of living in rental tenure varied extensively across different demographic categories. Rental tenure decreased with age, with 49.2 per cent of 15 to 24-year-olds living in rental accommodation, compared to 12.0 per cent of people aged 65+ years. Rental tenure also increased with increasing socio-economic deprivation (NZDep quintile). Others with high levels of rental tenure (significantly higher than average) were Māori, Pacific, Asian, and MELAA peoples; people who were not heterosexual; people who were unemployed; ‘other’ households (which include house shares) and single-parent households; households with no or 1+ bedrooms needed; people who had cut back a little or a lot on shopping or local trips; and people in major urban areas. People with lower levels of rental tenure (significantly less likely than average) were NZ Europeans; not in the labour force; in a couple only household or a household with a parent couple; and people in a household with a spare bedroom. Levels of rental tenure also increased with increasing settlement/urban size. Both males and females were equally likely to live in rental tenure (other gender identities were not collected in the data, and the data did not capture whether people were transgender).

Landlord-ended tenancy

Of respondents who were previously renting, 17.6 per cent said that the main reason for their most recent move was that their tenancy had been ended by the landlord. This reason was the second most common reason for moving, after ‘moved from rental to owned dwelling’ (20.6 per cent), and ahead of ‘social reasons’ (12.8 per cent), such as relationship change, or wanting to live closer to friends or family. For continuing tenant respondents, 25.4 per cent reported their last tenancy was ended by the landlord (Figure 2).

There were fewer differences in likelihood of landlord-ended tenancy than there were for likelihood of renting. Households with single parents had a higher than average percentage of landlord-ended tenancies. Asian peoples, people in couple-only households, and people in rural areas were less likely than average to have had their last tenancy ended by the landlord.

Mobility

Nearly three-quarters (72.0 per cent) of currently renting respondents had lived at their current address less than five years, and a further 15.2 per cent between five and ten years; meaning 87.2 per cent had lived at their current address less than ten years (Figure 7). The Mobility Score for all respondents (i.e. all tenures) was 2.84, but for those currently renting it was 4.41. Mobility decreased as age increased. People with (significantly) higher than average mobility were NZ Europeans, people in ‘Other’ households, people in NZDep 1-2 and NZDep 5-6 areas. Groups whose mobility was significantly lower than average were Pacific Peoples; people not in the labour force; single-person households; people in households with children (parent couples or single parents); people in households needing a spare bedroom; and people in NZDep 9-10.