Settlement trajectories of nearly 25,000 forced migrants in New Zealand: longitudinal insights from administrative data

ABSTRACT

Aotearoa New Zealand provides differential settlement support to forced migrants, primarily determined by how they receive protection status – as asylum seekers, refugees, or through other visa pathways. Despite these differences, there is limited quantitative evidence of their settlement outcomes related to work, social welfare, education and housing. In response, this study presents administrative data of adults from refugee backgrounds composed of four distinct subgroups (quota refugee, convention refugee, family reunification, and asylum seeker) to explore their access to these main services post-settlement and ascertain longitudinal income trajectories. Using the Integrated Data Infrastructure (IDI), we identified 24,894 working-age adults from refugee backgrounds who first received refugee recognition, an asylum-seeker visa or a family reunification visa between 1997 and 2020. We describe these cohorts’ demographic and socioeconomic characteristics and access to services by drawing from a range of government and census datasets. We then present a mixed model regression to illustrate the association of longitudinal income with years from arrival and other settlement indicators: controlling for age, gender, and refugee subgroups. Findings show outcome disparities between refugee groups and highlight the policy implications for supporting positive settlement outcomes, particularly emphasising the first five years of settlement.

KEYWORDS:

• Refugee
• resettlement
• New Zealand
• settlement outcomes
• Integrated Data Infrastructure (IDI)
• administrative data

Introduction

A range of visa types enables forced migrants to live and work in Aoteraroa New Zealand. These pathways create a differential policy landscape through which different groups can access settlement support and entitlements. In this paper, we focus on four main categories of forced migrants that the New Zealand government agency responsible for granting visas recognises:

• Quota refugees – people whom the United Nations High Commissioner for Refugees (UNHCR) has recognised as refugees before arriving in New Zealand

• Asylum seekers – people who apply for refugee status recognition and protection within New Zealand

• Convention refugees – former asylum seekers whose refugee status has been recognised in New Zealand.

• Refugee Family Support Category (RFSC) – people sponsored by a refugee family member whose refugee status is recognised and are already residing in New Zealand.

As there are different settlement pathways for these refugee subgroups, concerns have been noted across the sector about inequality of services accessible to refugee subgroups (Bloom and Udahemuka 2014; Mahony et al. 2017; Ferns et al. 2022). In addition, far less is known about the settlement outcomes of those from asylum, convention and family pathways, highlighting a significant gap in the literature as it relates to sociodemographic characteristics and outcomes associated with education, language acquisition, housing, and income.

Using administrative data, we identified working-age refugee adults and separated them into the four categories noted above. We then examined their administrative and survey data over 23 years (1997–2020) to describe these groups and address three research questions:

1. What are the access rates to education and state housing for subgroups of refugees?

2. What are the dynamics of employment/social welfare benefits for refugee subgroups over time?

3. What factors contribute to refugees’ income over time?

We first summarise the different settlement provisions across these subgroups and the associated literature on various settlement outcomes. We then use administrative data to compare these associated settlement factors/outcomes and present a mixed model regression that illustrates the association of longitudinal income with years from arrival and other settlement indicators while controlling for age, gender, and refugee subgroups. These findings highlight the policy implications for supporting positive settlement outcomes, particularly emphasising the first five years of settlement.

Forced migration in New Zealand: differential settlement support pathways

New Zealand has maintained a formal refugee resettlement programme since 1987 and has resettled more than 50,000 people since the Second World War. To understand this history, it is imperative to acknowledge that successive governments have defined different groups of forced migrants and the corresponding settlement support levels to which they are entitled.

The largest percentage of people who came to New Zealand as refugees has been through the quota programme. This group is referred to as quota refugees as they are selected from overseas to make up an annual quota comprising approximately 1500 people. Over the last ten years, an average of 394 people per year have applied for protection as asylum seekers when they are already in the country. Asylum seekers whose applications are successfully recognised by the government become known as convention refugees. There is an average of 178 people who receive this convention status over the same ten-year period (see Ferns et al. 2022). Finally, for family reunification pathways, an average of 244 individuals have come via the RFSC programme over the last ten years (data obtained from the Ministry of Business Innovation and Employment [MBIE] 2022). These groups have differential access to settlement support with quota refugees receiving the most government support.

In 2012, the New Zealand government introduced a ‘whole of government’ approach to supporting refugees to integrate with the New Zealand Refugee Resettlement Strategy (NZRSS). The strategy articulates its vision of refugees:

Participating fully and integrated socially and economically as soon as possible so that they are living independently, undertaking the same responsibilities and exercising the same rights as other New Zealanders and have a strong sense of belonging to their own community and to New Zealand. (Immigration New Zealand [INZ] 2012, p. 3)

Within this vision, the Refugee Resettlement Strategy has five overarching aims:

• Self-sufficiency: all working-age refugees are in paid work or are supported by a family member in paid work.

• Participation: refugees actively participate in New Zealand life and have a strong sense of belonging to New Zealand.

• Health and well-being: refugees and their families enjoy healthy, safe and independent lives.

• Education: refugees’ English language skills enable them to participate in education and achieve qualifications and support them to participate in daily life.

• Housing: refugees live independently of government housing assistance in homes that are safe, secure, healthy and affordable. (INZ 2012, p. 3)

While initially intended to be a strategy for all refugees, successive governments have only offered the strategy to quota refugees and thereby excluded people who came through convention and family reunification pathways. Asylum seekers are also excluded.¹ Cabinet papers show that, from the Strategy’s inception, the intent was to include other groups of forced migrants, but this has not happened more than a decade later (see Ferns et al. 2022). Based on the Strategy, quota refugees are placed in resettlement sites where basic needs such as employment, housing, and health services are generally available and receive support services for the first year of settlement. Further, quota refugees participate in a five-week orientation programme whereby they undergo health screening, receive information about settling in New Zealand, and English language training.

In contrast, convention refugees and family reunification members do not receive similar settlement programmes, nor do they receive formal introductions to the available services. Asylum seekers also receive little support and, whilst they have some rights and entitlements related to work and education, there are significant issues around the accessibility of these (Bloom and Udahemuka 2014; Ferns et al. 2022). In response to these differences, we outline the employment, educational, and housing settlement outcomes of four refugee background subgroupings and provide the demographic compositions related to age, gender, and years resettled to outline the policy implications of supporting positive settlement outcomes.

Positive settlement outcomes: economic, demographic, and broader settlement markers

Settling into a new country as a forced migrant often involves adapting to new systems, language, legal frameworks, and social norms. Within this process of negotiation of one’s past with the present, the literature provides numerous examples of what constitutes successful settlement (Ager and Strang 2008; Strang and Ager 2010; Lundborg 2013; Grzymala-Kazlowska and Phillimore 2018; Paz Aléncar and Tsagkroni 2019). Within this literature, positive settlement outcomes are often related to employment, social security access, education, language acquisition, housing, health and social connections. While there are differing perspectives across this literature, there is a general consensus that positive outcomes are impacted by various forms of settlement support and an appreciation of the diverse cultural and migration histories that people carry with them (Valtonen 2008; Grzymala-Kazlowska and Phillimore 2018).

Numerous integration models and government initiatives recognise economic self-sufficiency as one of the most important factors in successful integration (Ager and Strang 2008; UNHCR 2011; Marlowe et al. 2014; Brell et al. 2020). Without meaningful employment, refugees risk becoming trapped in a cycle of social and economic marginalisation (Brell et al. 2020). This outcome can potentially impact future generations, with earning capacity influencing the ability to afford many of the other resources required to rebuild a life in a new country, among them, housing, education and opportunities to participate equally in the receiving society (Hugo 2014; O’Donovan and Sheikh 2014).

Securing paid work can be a challenging obstacle for refugees because of a low level of host language proficiency, lack of recognition of their previous professional experiences and educational training, limited social connections, and racism (Colic-Peisker and Tilbury 2003, 2007; Elliott and Yusuf 2014; O’Donovan and Sheikh 2014). These obstacles can thereby create an employment gap of disparate outcomes relative to the general population (Brell et al. 2020).

The literature on refugee integration generally recognises that employment is only one marker of several that include a more holistic appreciation of what constitutes successful employment. With respect to these considerations, much of the international literature draws upon qualitative research that provides insightful understandings of the experiences related to host language acquisition (Kosyakova et al. 2022), housing (Fozdar and Hartley 2013) and education (Sidhu and Taylor 2007; Warsame et al. 2014). Other studies demonstrate how demographic considerations can also impact settlement outcomes that include: gender (Cheung and Phillimore 2017), time resettled (McMichael et al. 2017; O'Donnell et al. 2020), language acquisition (Morrice et al. 2019) and age (Bevelander and Pendakur 2014). While our paper does not endeavour to present an exhaustive list of these considerations, we do focus on education, language acquisition and housing as critical indicators that can be mutually constitutive with employment-related outcomes.

Further, people who come to a country through seeking asylum, receiving a family reunification visa, or being selected for a refugee resettlement programmes, can potentially have different demographic characteristics and settlement support needs. However, data on subgroup demographics have been limited. As Bevelander and Pendakur (2014) illustrate, asylum seekers are more likely to have greater human capital and resources to settle (at least initially) and family reunification pathways are likely to have greater social capital by virtue of having a sponsoring family member who is already in the country who can help facilitate community connection and wider opportunities to participate in civic life. Asylum seekers and convention refugees have been shown to be mainly young men (Pew Research 2016).

Within New Zealand, the government provides a snapshot of settlement outcomes, including the rates of employment, access to social welfare, and English language course utilisation (MBIE 2018). According to the latest dashboard on arrivals between 2010 and 2017, the employment rate has increased from a range of 10%–18% one-year post-arrival to 40%–50% five years post-arrival. The reverse trend was also seen for the main unemployment-related benefits, with decreasing rates of 65% at year-one arrival to less than 20% five years later. These statistics provide a snapshot of refugee adults’ settlement outcomes over the first five years from arrival. However, these data use broad definitions of what constitutes ‘employment’, only relate to quota refugees, and do not provide sociodemographic analysis across different refugee subgroups in relation to these outcomes, thereby underscoring the need to look at these groups separately and comparatively.

Methods: administrative data and identification of refugee cohorts

The data for this study were sourced from the Integrated Data Infrastructure (IDI), a large and regularly updated database holding microdata about individuals from a range of government agencies, which is managed by STATS New Zealand. To construct our cohort, we utilised visa data by Immigration New Zealand to identify people who came to New Zealand by the quota, convention, family and asylum pathways previously described. Our cohort comprises individuals who met the following criteria:

• approved for either a quota, convention, family or asylum seeker visa and were resident in New Zealand between 1997 and 2020²

• aged between 16 and 65 years at the time of visa approval to capture working-age refugees; and

• resided in New Zealand for at least one year.

From these parameters, we identified a cohort of 24,894 adults. Where possible, we linked this cohort with the following databases: Inland Revenue (IR), Ministry of Business Innovation and Development (MBIE), Ministry of Social Development (MSD), Housing New Zealand (HNZ), or Ministry of Education (MOE).

Data sources

All data that we present are in relation to people’s outcomes either after arrival in New Zealand (for quota and family pathways) or after a decision on refugee recognition (asylum and convention pathways). Sources of variables used to describe the population for this study are explained below:

• Demographic: age, gender and ethnicity were extracted from the Personal Detail Table, which is of the central tables within the IDI. This table contains everyone who has any information within any database of the IDI. Data on ethnicity in this table is a source-ranked ethnicity variable, which prioritises ethnicity records based on the quality of their source. This gives each individual a binary variable for each of the main ethnic groups: European, Māori, Pacific, Asian, Middle Eastern African Latin American (MELAA), and Other (any ethnicity not included previously). Thus, every individual could have multiple ethnic groups.

• Married/civil union: data were extracted from the Census 2018 first. If this data was missing, we sourced the 2013 Census. If data were missing from both Census, we sought legally married/civil union records from the Department of Internal Affairs. This means if someone did not respond to this question or was missing from the Census, or their marriage was registered outside New Zealand, their legal civil union partnership/marriage is unknown to us. If there was no linkage or relationship record in either, we coded the individual as ‘No/Unknown’.

• English language speaking proficiency: this variable was extracted from Census datasets. In the first instance, we extracted the response to the conversational English question from Census 2018. In cases where there were no links or missing data in Census 2018, we deferred to Census 2013 responses. Noteworthy, this variable was only available for those who ever participated in the Census (n = 14,862).

• Educational access and attainment: we linked our cohort with Census and MOE datasets to extract data educational attainment. Because only a third of all refugees ever completed education in New Zealand, and for most refugees we did not have an idea of past education, we used the Census as our main source for education. Where possible, we extracted data from the Census 2018, deferring to Census 2013 where it was not. Where there was ‘no response’ or linkage was unsuccessful with Census datasets, we extracted the highest qualification records from the MOE database. If linkages to both Census and the MOE were unsuccessful, we extracted education history data from the MSD datasets. It is important to acknowledge that this data do not necessarily reflect prior education and qualifications achieved before coming to New Zealand (quota and family pathways) or refugee status recognition (convention and asylum pathways). By prioritising Census, we have a subjective measure of education (prior to arrival) for those who never studied in New Zealand.

• Education registration: this variable demonstrated if an individual ever registered for a course or degree in New Zealand, sourced from the MOE.

• Housing support: two variables were created from the HNZ (Housing New Zealand) datasets to describe housing service access; (1) Ever received housing from HNZ; and, if yes, (2) the time from arrival/approval to the first access date.

• Income and social welfare: income data were extracted from Inland Revenue databases, which capture earnings from all forms of taxable income, which include earning, income-tested benefits and Working for Families tax credits (WFF – payments for families with dependent children aged under 18). Taxable income was combined with the WFF tax credit to determine total income which we subsequently adjusted for inflation according to the Consumer Price Index (CPI). The base used was the first quarter of 2020 (https://infoshare.stats.govt.nz). Medians are reported based on yearly incomes. Appendix Table 1 shows the number of refugees for whom we had income information by income year. For this study, we captured income records to the end of 2019 to avoid distortions from the Covid lockdowns from 2020 and the fact that refugee resettlement programmes were halted largely until 2022.

We used the IDI refresh of July 2021 for extracting data for this study.

Statistical analyses

All descriptive and modelling analyses were done in SAS (9.4).

Income data were restricted only to positive incomes (n = 21,672). We specifically aimed to understand factors contributing to wages and salaries income for refugee subgroups over time. For this reason, we did not include those who were solely on social welfare. Since the income variable was significantly skewed with many incomes near zero, we reported median income, and modelled logged of income as the outcome variable. Adhering to Statistics New Zealand’s confidentiality protocols, the minimum and maximum incomes are suppressed. The mean incomes were suppressed if the annual count of a refugee group was fewer than five, and medians were suppressed if there were fewer than ten individuals. All numbers are random rounded, meaning that some of the percentages may not total 100. A p-value of <0.05 was considered significant in all analyses.

For regression models, we used mixed model analyses to test the associates of longitudinal income of refugees. The predicting variables were years from arrival (as a time variant changing every year for each individual’s income data), age, gender, and refugee subgroups. Then, we included education, married/partnered and English-speaking proficiency as further predictors in the final models.

The final model results are presented by arrivals before 2012 and after 2012 to get better model fits for each subgroup of arrivals. We used the year 2012 as a cut point to allow a visual comparison of the income trend if there were any differences when the Refugee Resettlement Strategy was implemented for quota refugees, post-2012, and to provide a more recent picture of income trajectories.

Throughout our analysis and ensuing discussion, we retain a critical and structural awareness to position the reported refugee settlement outcomes holistically to recognise the wider context that settlement occurs (see Malkki 1996). As authors not from refugee backgrounds, we presented our findings and associated analysis in late 2022 to a select group of people from refugee backgrounds and those working in the sector to better ensure our interpretations maintained this structural awareness.

Disclaimer

Access to the data used in this study was provided by Stats NZ under conditions designed to give effect to the security and confidentiality provisions of the Statistics Act 1975. The results presented in this study are the work of the author, not Stats NZ nor individual data suppliers. These results are not official statistics. They have been created for research purposes from the IDI, which is managed by Stats NZ. For more information, please visit https://www.stats.govt.nz/integrated-data/.

Findings: demographic and longitudinal insights

We describe the demographic characteristics of each subgroup and outline the following settlement indicators: employment and social welfare trajectories, access to education and housing, and self-rated English-speaking proficiency. In the tables and figures that follow, it is important to recognise that the information obtained is after the asylum, refugee or family reunification visa is issued – referred to as after decision. This approach means the employment, education, housing and social welfare data do not include prior information for those who were already in New Zealand and then applied for asylum (i.e. the asylum and convention subgroups).

Demographic characteristics

From this table, there are several clear differences in the listed sociodemographic variables across the four subgroups. First, the majority of asylum seekers and convention refugees were men (70% for asylum seekers, 60% for convention refugees). This ratio was almost 50% for family and quota refugees.

In general, the majority of refugees were young upon arrival in the country (or approval of refugee status/ applying for asylum). Family reunification category had the highest proportion of people between 50 and 65 years (16.4%) at arrival. Less than half of quota, family, and convention pathways were known to be a in a legal partnership/marriage (with substantial missing data for asylum seekers).

Taken from the 2018 and 2013 Census, the family subgroup was most likely to report ‘no’ for conversational English (nearly 30%) and followed by quota (24.6%). Convention pathways are more likely to report speaking conversational English at nearly 50%, and this can be partly explained by the fact that they may have had a number of years in New Zealand (including study) to support these skills.

With the high rate of missing data from the Census, caution must be taken around interpretations of these findings and represents an area where further research is required, particularly given its association with positive settlement outcomes.

Table 1 also shows educational attainment in New Zealand by refugee subgroup (after they received their decision as an approved refugee, asylum seeker or family reunification visa) for those with income data. The New Zealand Qualification Framework defines the following levels as:

• L1–L4: secondary school, career/technical certificates, vocational/trade certificates

• L5–L6: Certificates and post-secondary school diplomas

• University: Graduate certificates; bachelor’s degree and higher

  Table 1. Characteristics of the four subgroups of adult refugees resettled in NZ between 1997 and 2020.

  Variables		Asylum(N = 6096)	Convention (N = 5430)	Family(N = 2529)	Quota (N = 10,839)	Total
  		N (Col%)	N (Col%)	N (Col%)	N (Col%)
  Age (at first decision or first arrival)	16–29	2433 (39.9)	2118 (39.1)	1185 (46.8)	5229 (48.2)	10,965
  	30–39	2352 (38.6)	2103 (38.7)	510 (20.2)	3174 (29.3)	8139
  	40–49	969 (15.9)	840 (15.4)	420 (16.6)	1542 (14.2)	3774
  	50–65	342 (5.6)	369 (6.8)	414 (16.4)	894 (8.2)	2016
  Gender	Male	4242 (69.6)	3291 (60.6)	1263 (49.9)	5499 (50.8)	14,298
  	Female	1854 (30.4)	2142 (39.4)	1266 (50.1)	5337 (49.2)	10,596
  Legally married/in a civil union relationship	Yes	1254 (20.6)	2298 (42.3)	918 (36.3)	4944 (45.6)	9414
  	No/Unknown*	4842 (79.4)	3135 (57.7)	1611 (63.7)	5892 (54.4)	15,480
  Conversational English(Census)	Missing	2439 (57.6)	1350 (27.6)	510 (23.1)	2520 (24.3)	6819
  	Yes	1443 (34.1)	2868 (58.7)	1041 (47.2)	5283 (51.1)	10,632
  	No	354 (8.3)	669 (13.7)	657 (29.8)	2547 (24.6)	4230
  **Education attainment	Missing	1095 (25.8)	177 (3.6)	303 (13.7)	867 (8.4)	2442
  	L1–L4	1053 (24.8)	1755 (35.9)	834 (37.8)	5265 (50.9)	8907
  	L5–L6	267 (6.3)	639 (13.1)	153 (6.9)	849 (8.2)	1911
  	University	591 (13.9)	1155 (23.6)	153 (6.9)	846 (8.2)	2742
  	Overseas qualification	411 (9.0)	534 (10.9)	204 (9.3)	687 (6.6)	1836
  	No qualification	714 (16.8)	543 (11.1)	408 (18.5)	1590 (15.4)	3255
  	Non-response	108 (2.5)	87 (1.8)	153 (6.9)	246 (2.3)	591

  *Unknown is for those people who did not respond to this question in the Census or had marital records outside of NZ that were not registered in NZ. **Education data is provided for all with an income data (Asylum = 4239, Convention = 4887, Family = 2205, Quota = 10,350).

The overseas qualification data is obtained from the Census and does not provide information about the level of qualification.

Education attainment was one of the most difficult variables to find information on for in this population, with no information found for 7.7% of refugees and 25.8% of asylum seekers (among people with income data). However, of those for whom we have data, we found convention and asylum groups as most likely to have a university qualification. Quota refugees were the most likely to have lower level (L1–L4) qualifications – we return to this difference in the discussion to emphasise the importance of the quota system being a humanitarian programme.

Table 2 outlines the proportions who ever accessed education in New Zealand (inclusive of English language courses) with 66% of quota refugees registered for a course, compared with 54.8% family and 46.2% of convention refugees. Across the subgroups, quota and family members have the highest rates of registering for a course in New Zealand and are far more likely to register for this course sooner than the convention and asylum counterparts. Convention and quota refugees also have the highest rates of tertiary registration compared with family subgroup and asylum seekers.

Table 2. Access to education and Housing New Zealand (HNZ) for adults arriving between 1997 and 2020, by subgroup.

		Asylum (N = 6096)	Convention (N = 5430)	Family (N = 2529)	Quota (10,839)	Total
Ever registered in any course in NZ	Yes	1368 (22.4)	2511 (46.2)	1386 (54.8)	7215 (66.6)	12,480
Months from arrival/refugee claim to the first registration	No	1368	2511	1386	7215	12,480
	Mean	62.2	34.7	10.46	15.2
	Median	53	23	4	5
	25th percentile	27	11	2	2
	75th percentile	85.5	46	11	17
Ever registered for tertiary education	N	1173 (19.3)	2784 (51.2)	903 (35.7)	4971 (45.9)	9834
Accessed Housing NZ after arrival/decision	Yes	258 (4.2)	1047 (19.3)	600 (23.7)	7260 (67.0)	9162
	No	5838 (95.8)	4386 (80.7)	1929 (76.3)	3576 (23.0)	15,735
Months from refugee status or arrival to the first HNZ	Mean	69.13	52.05	27.73	15.40
	Median	41.5	35	19	2
	25th percentile	23	17	7	1
	75th percentile	110	69	39	13

Note: Cells display N and column % in parentheses.

Refugees from different subgroups had disproportionately accessed HNZ housing as shown in Table 2 with 67% of quota refugees ever having accessed the service, followed by 24% family, 19% convention and only 4% of asylum seekers. In addition, the median months to the first access day is much longer in convention (35 months) and family (19 months) than the quota subgroup (2 months). These data need to be considered alongside wider housing policies and support. It is likely that people receive an accommodation supplement through a Work and Income benefit meaning that a significant portion of a person’s rent may be covered but they are not residing in a HNZ home. Further, family reunification arrivals are not eligible for public housing until they have been here for 2 years, which skews the data for this group.

Employment and social welfare

For income outcomes, we only used those with at least one year of Inland Revenue income data for the years 2000–2020, resulting in 21,672 individuals with at least one-year income data from Inland Revenue Department (IRD).³ Figures 1–3 display the percentages of refugees whose main source of income was from (1) wages and salaries, (2) self-employment, and (3) a benefit (some form of social welfare support).

Figure 1. Proportions of refugees with main source of income from WAGES AND SALARIES, by years from arrival.

Figure 2. Proportions of refugees with main source of income from SELF-EMPLOYED, by years from arrival.

Figure 3. Proportions of refugees with main source of income from a BENEFIT, by years from arrival.

Caution needs to be taken when looking at these figures. The figures only report proportions of an individual’s main source of income. This means that the percentages of people on a benefit as a main source of income may still be participating in some form of paid work.

Wages and salaries as main income source

While there are disparities across the four subgroups, it is clear that the largest increases of wages and salaries as the main source of income are greatest for quota and convention refugees in the first five years. After this time, there is a considerable flattening of the curves. Family pathways and quota follow the same upward trend, with the difference that family starts at a higher proportion (25%), compared with quota refugees. This is not too surprising seeing that the reunification criteria involve an applicant sponsoring a family member. After 5–7 years, family pathway continues the upward trend to higher proportions on wages and salaries whereas quota remains largely fixed at around the 40% range. Asylum seekers show a fairly steady trend of 60% to 70% of people whose main income is wages and salaries.

Importantly, as these data reflect only working-age adults, they do not capture children or those under 18 who came via the family reunification pathway. We anticipate that these young cohorts would evidence higher rates of income over time and we return to this consideration in the discussion.

Self-employment as the main income source

Self-employment as the main source of income represents the lowest proportion across all subgroups. Despite these smaller percentages, all subgroups generally show a positive trajectory over time. Nearly 10% of convention, asylum seekers, and family groups were working as self-employed 15–18 years from arrival.

Benefit as the main income source

The first five years for all refugee subgroups indicate a shift from the benefit to wages and salaries or self-employment as the main source of income. After this time period, people on the benefit still trends slightly downwards with the exception of the family subgroup, which drops further below 20% after 15 years. Seeing that family reunification places are supported by a person who is already in New Zealand, this is somewhat expected as these family members may have more social and human capital relative to other groups.

The three figures generally illustrate an upward trend for wages and salaries and downward trend for benefits over time. However, after five years, the upward trend for wages and salaries has slowed or flattens out. Overall, the greatest reduction in the benefit being the main source of income for all groups occurs in the first five years.

Contributing factors to income: insights from mixed models

From regression modelling, we found that tertiary-level qualification, self-reported English-speaking proficiency, years from arrival, and being male were all contributing factors to higher income levels. These findings were consistent for both groups arriving between 1997–2012 and more recent arrivals 2012–2019. However, significant differences between groups of refugees remained in place across both time periods. Those who arrived after 2012 earned more than those who arrived before this time (1997–2011). For both cohorts, those who self-identified as speaking English had higher incomes.

We identified 6,096 asylum seekers who did not become convention refugees during the study period. However, some of those people acquired residency through other pathways, whereas most of them did not. Because we found a difference in mean income between the two groups of asylum seekers, we divided them into two subgroups for the mixed model regression as follows:

1. Non-resident asylum (n = 3858): those who did not achieve residency and were either still awaiting their asylum claim decision or had left the country; and

2. Resident asylum (n = 2231): people who applied for asylum but during the study period successfully became a resident via a different visa pathway (such as skilled worker or through marriage/partnership/family relations).

Caution is required when looking at the policy implications for asylum seekers as these two groups can have very different trajectories.

As Table 3 illustrates, there are significant differences between subgroups of refugees in terms of income. These differences remained significant even after accounting for other variables. This table shows that only the married/partnered variable did not contribute to the long-term income of the refugee subgroups.

Table 3. Significant variables contributing to income (results from the mixed model) for arrivals before 2012 and after 2012.

Before 2012 arrivals (1997–2011) (N = 11,313)			After 2012 arrivals (2012–2019) (N = 4188)
Effect	^F Value	p-value	&F Value	p-value
Refugee Subgroup	F(4) = 26.76	<.0001*	F(4) = 43.9	<.0001*
Married	F(1) = 0.4	0.5259	F(1) = 1.98	0.159
Sex	F(1) = 232.4	<.0001*	F(1) = 23.33	<.0001*
Years arrived from income year	^^F(22) = 327.31	<.0001*	$F(7) = 312.56	<.0001*
Qualifications	F(6) = 7.2	<.0001*	F(6) = 2.41	0.0251*
Refugee subgroup*Qualification	F(24) = 2.24	0.0005*	F(24) = 2.02	0.0022*
English	F(2) = 15.15	<.0001*	F(2) = 5.91	0.0027*
Refugee subgroup*English	F(8) = 1.8	0.0718	F(8) = 2.39	0.0144*
Refugee subgroup*Married	F(4) = 1.32	0.2617	F(4) = 0.44	0.7776
Refugee subgroup*Sex	F(4) = 7.48	<.0001*	F(4) = 4.55	0.0011*
Age group	F(3) = 38.02	<.0001*	F(3) = 4.16	0.0059*

Notes: ^Denominator = 12,000; ^^Denominator = 74,000 (for years from arrival); & denominator = 4131; $ denominator = 8950. * indicates significant p-values.

Figures 4 and 5 also show that quota refugees were earning less compared with family, convention, and asylum seeker groups.

Figure 4. Mean income averaged over the years by qualifications and refugee subgroups, adjusted for age and gender (BEFORE 2012).

Figure 5. Mean income averaged over the years by qualifications and refugee subgroups, adjusted for age and gender (AFTER 2012).

Further, quota refugees were earning less income regardless of their qualification, after controlling for all other factors. The mean income of convention refugees who arrived after 2012 was higher than that of before-2012 arrivals. The same was true for asylum seekers (both resident and non-resident) who earned more income after 2012 when compared with before 2012 arrivals (see Appendix Tables 2 and 3).

The effect sizes shown in Table 4 (provided at the end of the paper) show that asylum seekers and convention refugees earn significantly more than quota refugees after controlling for all other variables. Despite the fact that arrivals post-2012 had a maximum of eight years of income data in the study (and pre-2012 arrivals having up to 20 years of income data), the results of both models show higher incomes for convention and family refugees in the long term.

Table 4. Mixed model regression coefficients with longitudinal income as the outcome.

		After 2012 (N = 4,188) (2012–2020 arrivals)				Before 2012 (N = 11,313) (1997–2011 arrivals)
Effect	Category	Beta Estimate	SE	t-value	p-value	Beta Estimate	SE	t-value	p-value
Intercept		8.11	0.12	70.6	<.0001	7.45	0.1	87.01	<.0001
Refugee group	Non-resident Asylum	1.22	0.43	2.85	0.004	0.51	0.24	2.09	0.04
	Convention	0.48	0.22	2.17	0.03	0.29	0.11	2.57	0.01
	Family	0.26	0.15	1.76	0.078	0.63	0.14	4.49	<.0001
	Resident_Asylum	−0.51	0.75	−0.68	0.496	0.37	0.12	3.24	<0.0001
	Quota	0				0
Married	Yes	0.04	0.05	0.7	0.478	0.09	0	2.72	0.01
	No/ Unknown	0				0
Gender	Female	−0.66	0.05	−12.6	<.0001	−0.54	0	−17.42	<.0001
	Male	0				0
Age group (ref = >50 years)	16–30 years	0.22	0.08	2.7	0.007	0.51	0.1	9.16	<.0001
	30–40 years	0.22	0.08	2.6	0.009	0.49	0.1	8.59	<.0001
	40–50 years	0.23	0.09	2.6	0.008	0.32	0.1	5.34	<.0001
	50–65 years	0.00				0
Years from Income year	1	0				0.41	0	16	<.0001
	2	0.9	0.04	24.4	<.0001	0.7	0	28.3	<.0001
	3	1.16	0.04	30.1	<.0001	0.85	0	34.13	<.0001
	4	1.37	0.04	33.9	<.0001	0.96	0	38.22	<.0001
	5	1.53	0.04	35.5	<.0001	1.07	0	42.38	<.0001
	6	1.67	0.05	33.8	<.0001	1.16	0	45.15	<.0001
	7	1.8	0.07	27.6	<.0001	1.2	0	46.33	<.0001
	0	0
	8					1.29	0	49.26	<.0001
	9					1.34	0	49.84	<.0001
	10					1.33	0	48.51	<.0001
	11					1.38	0	48.81	<.0001
	12					1.42	0	48.61	<.0001
	13					1.4	0	46.52	<.0001
	14					1.44	0	46.14	<.0001
	15					1.47	0	45.65	<.0001
	16					1.5	0	44.88	<.0001
	17					1.54	0	44.31	<.0001
	18					1.57	0	43.17	<.0001
	19					1.63	0	41.81	<.0001
	20					1.62	0	37.72	<.0001
	21					1.69	0.1	31.74	<.0001
	22					1.71	0.1	14.91	<.0001
	0					0
Qualification	L1–L4	−0.1	0.07	−1.42	0.157	0.21	0.1	3.85	<0.0001
	L5–L6	−0.28	0.12	−2.28	0.0226	0.24	0.1	3.57	<0.0001
	Missing	−0.03	0.16	−0.2	0.8386	0.22	0.2	1.14	0.25
	Non-response	−0.01	0.14	−0.08	0.9375	0.16	0.3	0.63	0.53
	Overseas	−0.11	0.1	−1.16	0.2476	0.08	0.1	0.86	0.39
	Uni	−0.19	0.12	−1.53	0.126	0.35	0.1	5.36	<.0001
	No Qual	0				0
Qualification*group (Ref = 'Quota’, ‘No qualification’)	Non-resident asylum:L1–L4	0.03	0.43	0.08	0.9373	−0.06	0.1	−0.51	0.61
	Non-resident asylum:L5–L6	0	0.46	0	0.9963	−0.21	0.2	−1.07	0.29
	Non-resident asylum:missing	0.16	0.46	0.34	0.737	0.22	0.2	1.06	0.29
	Non-resident asylum:Non-response	−0.02	0.44	−0.04	0.9697	0.34	0.5	0.67	0.5
	Non-resident asylum:Overseas	−0.5	0.49	−1.03	0.3019	0.47	0.4	1.33	0.18
	Non-resident asylum:Uni	0.19	0.43	0.45	0.6551	−0.36	0.2	−2.35	0.02
	Non-resident asylum:No qual	0				0
	Resident_Asylum: L1–L4	0.1	0.69	0.14	0.8895	−0.15	0.1	−1.51	0.13
	Resident_Asylum: L5–L6	0.32	0.72	0.45	0.6539	−0.12	0.1	−0.98	0.33
	Resident_Asylum: Missing	−0.3	1.75	−0.17	0.8628	0.17	0.2	0.71	0.48
	Resident_Asylum: Non-response	0.98	0.7	1.39	0.1653	0	0.3	0	1
	Resident_Asylum: Overseas	0.39	0.71	0.55	0.5833	0.04	0.1	0.29	0.77
	Resident_Asylum: Uni	0.77	0.72	1.07	0.283	−0.09	0.1	−0.8	0.42
	Resident_Asylum: No qual	0				0
	Convention: L1–L4	−0.24	0.22	−1.06	0.2878	−0.2	0.1	−2.26	0.02
	Convention:L5–L6	0.4	0.26	1.55	0.1219	−0.18	0.1	−1.73	0.08
	Convention: Missing	0.72	0.33	2.17	0.0301	0.48	0.3	1.91	0.06
	Convention: Non-response	0.16	0.3	0.54	0.5925	0.47	0.4	1.2	0.23
	Convention: Overseas	0.04	0.25	0.18	0.8594	0.24	0.1	1.95	0.05
	Convention: Uni	0.4	0.24	1.65	0.0992	−0.01	0.1	−0.08	0.94
	Convention: No qual	0				0
	Family: L1–L4	−0.16	0.15	−1.08	0.2821	−0.3	0.1	−2.39	0.02
	Family: L5–L6	−0.08	0.23	−0.32	0.7457	−0.41	0.2	−2.44	0.01
	Family: Missing	0.19	0.26	0.71	0.4751	0.17	0.4	0.41	0.68
	Family: Non-response	−0.05	0.21	−0.21	0.8302	−0.23	0.4	−0.56	0.58
	Family: Overseas	0.27	0.2	1.35	0.1777	0.14	0.2	0.82	0.41
	Family: Uni	0.24	0.22	1.09	0.2775	−0.22	0.2	−1.3	0.19
	Family: No qual	0				0
Conversational English (ref = 'No’)	Yes	0.05	0.06	0.9	0.3678	0.18	0	4.22	<.0001
	Missing	0.12	0.12	0.95	0.3411	−0.07	0.1	−1.28	0.2
	No	0				0
English*Group (ref = 'Quota’, No’)	Non-resident Asylum: Yes	−0.1	0.27	−0.38	0.7068	0.16	0.3	0.6	0.55
	Resident Asylum: Yes	0.88	0.49	1.78	0.0748	0.2	0.1	2.32	0.02
	Convention: Yes	0.23	0.13	1.74	0.0812	0.1	0.1	1.31	0.19
	Family: Yes	−0.04	0.12	−0.32	0.7516	−0.03	0.1	−0.32	0.75
	Non-resident Asylum: Missing	−0.55	0.33	−1.66	0.098	−0.03	0.2	−0.13	0.9
	Resident Asylum: Missing	−1.25	1.04	−1.19	0.2327	0.34	0.1	3.32	<0.0001
	Convention: Missing	−0.14	0.22	−0.64	0.5249	0.1	0.1	1.05	0.29
	Family: Missing	0.02	0.2	0.11	0.9126	0.06	0.2	0.34	0.73
	Quota: Missing	0				0
Group*married (Ref = 'Quota’, No/unknown’)	Non-resident Asylum: Yes	−0.06	0.21	−0.3	0.7624	−0.23	0.2	−1.52	0.13
	Non-resident Asylum: No/Unknown	0				0
	Residen asylum: Yes	0.36	0.3	1.18	0.2386	−0.09	0.1	−1.64	0.1
	Residen asylum: No/Unknown	0				0
	Convention: Yes	0.06	0.1	0.57	0.5699	−0.03	0.1	−0.56	0.58
	Convention: No/Unknown	0				0
	Family: Yes	−0.05	0.1	−0.45	0.653	0.04	0.1	0.46	0.64
	Family: No/Unknown	0				0
Group*sex (Ref = Quota, male)	Non-resident Asylum: Female	0.31	0.15	2.09	0.0371	0.36	0.1	4.22	<.0001
	Non-resident Asylum: Male	0				0
	Resident Asylum: Female	0.9	0.29	3.11	0.002	0.17	0.1	2.9	<0.0001
	Resident Asylum: Male	0				0
	Convention: Female	0.15	0.1	1.56	0.1181	0.17	0.1	3.36	<0.0001
	Convention: Male	0				0
	Family: Female	0.25	0.1	2.6	0.0094	−0.06	0.1	−0.68	0.5
	Family: Male	0				0

The main findings in terms of mean differences between refugee subgroups and demographic covariates from Table 4 are:

• Time settled: The main indicator of income was time from arrival. For all subgroups in both models, the longer people stayed in the country, the more they earned (p < 0.0001).

• Gender: as shown in Appendix Tables 2 and 3, the mean income of women was significantly lower than that of men of the same subgroups, except for asylum-residents. For both genders, convention and family subgroups were earning significantly higher income than quota subgroup of the same gender (p < 0.001).

• Age: Age at arrival/decision date was inversely associated with income. In the before-2012 arrival group, younger age groups (16–30) had significantly earned more over time than those aged 40–65 years old. For arrivals after 2012, those aged 16–30 earned more than those aged 50–65.

• Subgroup: All subgroups were earning significantly more than quota refugees for both before 2012 and after 2012 cohorts (Appendix Tables 2 and 3).

• English: Self-reported English proficiency significantly contributed to higher incomes in the quota, convention and family subgroups. The exceptions to this where English proficiency did not demonstrate significant differences were with the family subgroup (after 2012) and the asylum subgroups of residents and non-residents (before and after 2012).

Overall, this modelling shows that quota refugees are not doing as well when comparing incomes to the other groups. Regardless, what continued to be the most consistent indicator of increasing income for all refugee subgroups was time from arrival in New Zealand.

Discussion: implications for refugee settlement support provision

Before articulating the policy implications of these economic, demographic and wider settlement findings, we find it paramount to emphasise that the refugee pathways to New Zealand are ultimately about protection. While understanding economic trajectories is important, it is equally important to acknowledge New Zealand’s humanitarian commitments to responding to what is currently the largest number of forcibly displaced people worldwide. The country’s pledge to playing a part in providing safety and security for through its quota, asylum, family reunification and other humanitarian pathways is a critical response to these crises that recognises the vulnerability of those who continue to live in situations of endemic precarity and danger. Within our study, the observation across several economic indicators that quota refugees are not faring as well and that they have lower reported levels of education suggests that New Zealand’s quota programme is a humanitarian one (see Beaglehole 2013) and should be celebrated as such.

Secondly, we acknowledge the challenges that forced migrants can experience in trying to obtain work also reflects systemic, structural, and wider societal receptiveness to forced migrants alongside the individual trajectories we present (see also Ager and Strang 2008; Cassim et al. 2022; Ferns et al. 2022). This appreciation is necessary to ensure that the conclusions of this study do not fall into the problematic tropes that primarily position settlement success as an individual responsibility.

Overall, this study provides novel understandings about different refugee subgroups and their settlement trajectories. We outline the policy implications across these four subgroups with three main recommendations.

1. Across the income, social welfare, and educational data, our study suggests that quota refugees are in greatest need of protection and supports the longstanding narrative that the government does not ‘cherry pick’ those who are most likely to contribute to the economy – thus it is a humanitarian programme. Seeing that targeted settlement support is provided only in the first year, there is a case for extending this to five years where marked improvements in terms of income and the percentages of people moving into paid employment occur.

2. Though these findings indicate a need for more assistance for quota groups, we do not suggest that the Refugee Resettlement Strategy should be limited to quota refugees alone and should be inclusive of other groups. It is clear that all groups follow a similar trend of making positive strides in the first five years and this is where targeted support could generate best long-term outcomes.

3. Taking a stronger intersectional focus is important for policy. Across the four groups, there are evident demographic differences, and the mixed model demonstrates that age, time settled and gender are important considerations for economic outcomes. Policy solutions need to be responsive to these realities.

The five-year window

Overall, this study demonstrates that, for quota, convention and family pathways, the first five years is where the greatest shifts to income and wages are seen and this aligns with other studies that demonstrate the importance of settlement support and establishing a welcome for forced migrants (Legrain 2016). These findings generally align with other large quantitative studies that show promising increases in the proportions of refugees employed over a ten-year period in Norway (Bratsberg et al. 2017) and Denmark (Schultz-Nielsen 2017), although our study suggests a smaller time window to realise these outcomes. After this five-year timeframe, positive economic trajectories in terms of proportions of people earning income through paid work versus social welfare tend to flatten out and appear entrenched.

This study demonstrates a significant income gap between different refugee subgroups. We can see from the trends that, for about 40% of quota refugees, the main source of income remains a benefit even after many years of settlement – but we cannot know the underlying reasons for this from these data alone. Other studies have suggested that refugees may end up working in ‘survival’ jobs for a long time, regardless of their job history or education (Lumley-Sapanski 2021). Further, additional gaps remain about the exact hours the refugees worked for each industry, type of contracts (short term or long term) or job satisfaction measures. Future studies could focus on these time periods to understand what occurs over the first five years and after, alongside an analysis of job opportunity and quality which has been explored in wider societal contexts (see Starr 2019).

Refugee Resettlement Strategy

While there were not marked differences in income between the pre-2012 and post-2012 outcomes for quota refugees, it is important not to view the New Zealand Refugee Resettlement Strategy solely within these parameters. As noted earlier, the strategy takes a more holistic approach to support that acknowledges the range of supports that people need. As the indicators that are used to capture the ‘success’ of the strategy through the refugee data dashboard are very broad, incorporating more specific measures about the type of work, hours employed and considering these across a range of demographic considerations (noted further below) could help development of a more tailored and responsive strategy.

Over the study period, quota refugees having had the fastest and highest access to housing and education services than the other three groups and demonstrates differential and arguably discriminatory treatment relative to the other subgroupings we present. While family and convention pathways may have greater access to social support and existing networks and resources in contrast to quota refugees, we also know that this is not the case for many (see Bloom and Udahemuka 2014). As it is clear that all groups follow a similar trend of making positive strides in the first five years, targeted support translated through the Strategy could generate best long-term outcomes and respond to the critiques that the current Strategy is discriminatory to other forced migration groupings (see Ferns et al. 2022). Whether asylum seekers should be part of this strategy, or if they should be supported through other means, is a discussion beyond the scope of this paper, but there is little question that they should be supported while their legal right for refugee recognition is under review. While economic outcomes for convention, family and asylum subgroups are better than the quota counterpart, extending support coverage to these groups could help realise quicker positive outcomes with somewhat higher short-term funding costs and arguably much greater long-term economic gains.

It is also worth acknowledging that the New Zealand Refugee Resettlement Strategy is only a strategy – it is not a policy (see Marlowe et al. 2014). However, this strategy does provide a language and a basis for measuring success and where resources to support settlement initiatives can flow. In this sense, the importance of the strategy is clear to help respond to some of the economic trends and other settlement indicators (access to education, language support, and housing) that we cover.

Intersectional approaches to policy

This study demonstrated that there are significant differences between different refugee subgroups in relation to access to services and aligns with other studies arguing for more inclusive approaches to settlement support within New Zealand (see also Bloom and Udahemuka 2014; Ferns et al. 2022). We summarise some of the main considerations below:

• Subgroup implications – Our study illustrates that quota refugees have the least-positive economic trajectories relative to the convention, family and asylum groupings. This observation reinforces that within New Zealand’s refugee quota that there are dedicated categories of ‘women at risk’ and ‘medical/disabled’ – it is not surprising by having targeted programmes towards those with potentially higher vulnerabilities that their settlement outcomes may not be as strong (see DeSouza 2011). Again, this further suggests the importance of longer-term settlement support strategies within the noted five-year window and a responsiveness to the demographic differences across the four subgroups.

• Education – It is also worth acknowledging that convention refugees and asylum seekers (particularly those who were able to stay in New Zealand under a different visa as already discussed) could have been studying in the country for a considerable time before applying for asylum or acceptance of the refugee status. This could explain some of the differences in access rates for education and higher mean income. It is important to acknowledge that these data do not reflect prior education and qualifications achieved before coming to New Zealand (quota and family pathways) or refugee status recognition (convention and asylum pathways). Further research into these educational backgrounds along with associated skill recognition/accreditation in New Zealand would be an important step in understanding economic settlement trajectories.

• Gender – The quota and family reunification demographics are balanced in terms of gender. The quota counterparts are generally younger, while the family pathway has the highest proportion of older members. The asylum and convention pathways had a much higher representation of males and this is reflected in the international literature (Pew Research 2016). These demographics have important implications, as they relate to what constitutes settlement success and where to target particular settlement support – which may relate to the flexible delivery of services that are responsive to gendered considerations that include caring responsibilities, English language competencies, access to transport, and others (see DeSouza 2011). It is also noteworthy that the large gender gap in income for all refugee subgroups is largest in the quota group. This observation could likely be explained considering some of the specific selection criteria for the quota that include ‘women at risk’, which demonstrates a commitment to settling highly vulnerable groups.

• Age – This study also showed that adults who arrive at a younger age have a better income trajectory over the years. This could be due to a higher chance of accessing education, such as tertiary learning opportunities. Nevertheless, this difference implies that targeted support may be needed for older refugees in the first few years.

• English language – The number of refugees who said ‘no’ to having conversational English in the New Zealand Census was striking. While there are large amounts of missing data, the high proportions that still report this suggest that greater support and resourcing are needed. This would be more helpful when services are offered in the first five years of a refugees’ arrival. This measure was merely a subjective measure for English speaking proficiency captured at one point in the Census. This indicates the limitations of this one measure in adequately reflecting the English sufficiency of refugees in the long term and the need for surveys and qualitative research to more comprehensively understand English language proficiency over time.

Conclusion

A major strength of this study is the large cohort size available to look at settlement outcomes over time across the four subgroupings and the nationally representative sample population. In future studies, this will allow comparisons with general population and immigrants (the focus of a future paper). Further, this study ascertains, for the first time, the differential demographic composition and income and employment settlement outcomes across the four subgroups of quota, convention, family, and asylum. Despite this strength, we also acknowledge the limitations of administrative data and the need, particularly, for qualitative research to illuminate why these disparities exist across the subgroupings. This is important for recognising the demographic differences across these four groups and the need to continue to monitor these trends over time. We acknowledge that there is missing information within these data sets and reinforces the need for further targeted surveys and qualitative studies to understand the outcomes and barriers to access as it relates to services and employment.

As already emphasised, this study reflects only working-age adults and does not capture children or those who were under 18 at first decision, or time of arrival. We anticipate this population would show higher rates of employment and income over time. Further research could focus on this group as it is well established that the first generation of refugees resettled in a new country often invest heavily in the future of the second (for instance, see Waters et al. 2010; DeSouza 2011).

We have shown how the pathways by which forced migrants find protection in New Zealand create differential access to services and settlement support. Our study demonstrated the importance of the first five years of the settlement experience where the greatest shifts of people receiving their main income from wages and salaries occurs and the reduction of those receiving a benefit. Extending settlement support over this period of time and tailoring this to the demographic differences across various subgroups may help to further accelerate positive trajectories over this important initial five-year window before the general levelling off of outcomes occurs.

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.

Additional information

Funding

This work was supported by Royal Society Te Apārangi [Rutherford Discovery Fellowship, ID# UOA1801].

Notes

1 This strategy is currently undergoing a ‘refresh’, which includes potential inclusion of these different groups.

2 1997 was the first year in the decision table of the Immigration New Zealand database where refugees could be identified from the criteria of every specific decision.

3 Data are sourced only from 2000 to 2020 as there were limited data for the study cohort from 1997 to 1999.