Author: Gwilym Pryce (on behalf of the Life at the Frontier research team)
Project: Life at the Frontier: The Impact of Social Frontiers on the Social Mobility and Integration of Migrants
Key findings:
- We have evaluated the effectiveness of the Norwegian Refugee Dispersal Policy, estimating whether the employment rate in refugee settlement neighbourhoods had an impact on refugee labour market outcomes. We found only a modest positive effect.
- Using Scottish Census data from 1971 to 2011, we find strong evidence of path dependency in migration location patterns over a 40-year period. This means that a significant reduction in the levels of segregation through policy intervention is likely to prove difficult.
- We have developed a research design for estimating the effectiveness of the Norwegian Introduction Programme which provides training for new migrants. We hope to report initial findings in the next few months.
- We have exploited the difference in housing policy between Scotland and England to investigate whether increasing new refugees’ access to social housing in Scotland helped reduce homelessness. We find a significant effect.
- We are in the process of exploring the use of register data from Sweden and Norway to develop Chetty and Hendron (2018) style estimates of the impact of neighbourhoods on educational attainment and labour market outcomes.
Introduction
The overarching focus of the Life at the Frontier project, and the wider NordForsk research programme of which it is a part, is on the integration of migrants. Integration, however, remains a contested term with some scholars (e.g. Adrian Favell) dismissing the idea of integration as nothing more than the repackaging of assimilation. However, we believe the term can have a meaningful interpretation if it is defined in a sufficiently inclusive way. A particularly useful definition is that set out in a 2004 UK Home Office report:
“Successful integration entails migrants being able to achieve outcomes within employment, housing, education, health etc. which are equivalent to those achieved within the wider host communities”
(UK Home Office report, 2004)
One advantage of this approach is that it emphasises equality of opportunity and outcome. As such, it frames the discussion around migration integration in terms of social justice, social inclusion and social mobility in the host nation, rather than placing the onus on migrants. Moreover, it raises the question of whether society as a whole is ‘integrated’. If a country is fragmented and stratified by income, social class, religion and race, it makes it less likely that migrants can become integrated into employment, housing and education.
This definition also exposes the weaknesses of static notions of integration as it carries within it an implicit emphasis on the dynamic nature of integration over the life course. Static measures – such as the extent to which migrants from a particular country have employment rates similar to those of natives – can be misleading as they fail to account for the migrant life course position, especially their trajectory in the labour market. A more insightful approach is one that explicitly considers life-course trajectories and that compares them with natives and other relevant groups. The implication is that measures of integration should focus on social mobility and its various drivers.
So what might be the main factors that determine the social mobility, and hence integration, of migrants? In this article I highlight our ongoing work with respect to the following drivers:
- Neighbourhood context
- Settlement policy
- Segregation dynamics
- Social housing policy
- Training for new migrants
- School quality
- Housing market discrimination
Neighbourhood effects
One of the challenges facing countries receiving refugees is where to house them. The cheapest option, at least in the short run, is to locate them in areas with low-cost housing. But there is a problem with this approach. Such areas often have the highest rates of poverty and deprivation which may not be conducive to migrants achieving high rates of social mobility and integration. The characteristics of neighbourhoods – employment rates, levels of crime, concentration of offenders, quality of housing, levels of pollution etc. – can all have significant impacts on the life outcomes of those who live there.
There is a long tradition of estimating these “neighbourhood effects”, especially the impacts of spatial concentration of poverty on life outcomes. Galster and Sharkey’s (2017) widely cited review of the evidence suggests that neighbourhood context not only has a significant direct effect on cognitive development, educational attainment, teen fertility, physical and mental health, labour force participation, earnings, and crime, but that there will likely be important indirect effects as well. These indirect effects arise from deprivation rates in surrounding areas and/or at multiple spatial scales, and the intergenerational effects of neighbourhoods on relationship matching, household formation and the attributes and behaviours of parents.
The findings of neighbourhood effects research have, however, been controversial. This is due to the difficulties of distinguishing the causal effects of a neighbourhood’s social and physical environment from other factors. It is possible, for example, that people living in deprived neighbourhoods tend to have lower health outcomes not because such neighbourhoods cause poor health, but because households with poor health are more likely to move to such neighbourhoods. Imagine, for example, a family living in an affluent neighbourhood where the main earner loses their job due to chronic illness. It is not long before the family struggle to meet their housing costs and are forced to move to a more deprived neighbourhood where rents are cheaper. Screening out these “selection effects”, and other confounding factors such as family characteristics and unobserved natural ability, has proved difficult.
Recently, however, neighbourhood effects research has received a significant boost through the pioneering work of Princeton professor, Raj Chetty. By drawing on very large US datasets that allow Chetty and his team to follow household moves into different US counties, they have developed robust methodology for isolating the causal effect of area attributes on educational outcomes. Their research has, for example, confirmed the beneficial effects of moving to a more affluent neighbourhood on educational attainment. Crucially, younger siblings who benefitted from longer exposure to the new neighbourhood than their older siblings, experienced a larger educational boost. Chetty and Hendren (2018) find that boys’ outcomes are more place-sensitive than girls’ outcomes. Good places to grow up tend to have lower income inequality, higher school quality, less crime, larger proportions of families with two parents and lower segregation.
Chetty’s work highlights how the large geographical differences across the US in poverty and disadvantage affect social mobility. Similar disparities have recently been found in the UK. Estimates by Buscha et al. (2021) reveal large geographical differences in intergenerational social mobility rates, not only between regions but also within them. In fact, “all regions in England and Wales contain districts in the top and bottom 20th percentile of social mobility nationally.”
These findings are important for our own research because they may have implications for the impact of geographical inequality on the life outcomes of migrants. Moreover, the fine-grained spatial differences in mobility rates uncovered by Buscha et al. in the UK may suggest that it matters not only which region a migrant moves to – whether it is a comparatively affluent region such as London or a comparatively poorer region such as Wales – but also which local authority district they select within those regions. With this in mind, we are currently developing a research grant application to the ESRC to estimate the extent to which the social mobility of migrants is determined by geographical context.
We are also exploring the potential to replicate the Chetty and Hendren (2018) design using Swedish and Norwegian register data. In particular, we are interested in investigating whether similar levels of geographical inequality in social mobility are to be found in countries with much more generous welfare regimes than in the Chetty and Hendren study. We shall also explore whether residential location in childhood has a causal effect on income in adulthood. As far as possible we will use the same modelling approach, the same variables, covering the same time-period as the Chetty and Hendren study. One difference will be the size of geographical units used to track movers. Chetty and Hendren (2018) use counties and commuting zones. But these are too large to be applicable in a Nordic setting. So instead, we shall use municipalities as our geographical unit. There are at least two important reasons for this. Firstly, it is the municipalities in Norway and Sweden that are responsible for nurseries, compulsory schooling, local communities and much of the local planning in general. Municipalities are therefore the most policy-relevant administrative unit. Secondly, moves at a lower level of aggregation might not lead to children changing schools, and they may even retain the same friends and social networks. We hope to report on our findings before the end of the project (March 2023).
Impact of Norwegian Refugee Dispersal Policy on the labour market integration of refugees
An alternative to the Chetty and Hendron (2018) approach is to make use of a quasi-experiment such as the one arising from a policy intervention. Norway’s approach to the spatial dispersal of refugees is such a policy. It “directly places refugees aided by the United Nations High Commissioner for Refugees (UNHCR) within Norwegian neighbourhoods” (Bratsberg et al. 2020). Bratsberg et al find that “within each municipality, the specific neighbourhood a refugee is placed in is as-if random conditional on the year of arrival” (Bratsberg et al. 2020). This provides researchers with large random variations in the characteristics of settlement neighbourhoods, which means that the Norwegian refugee dispersal policy offers an ideal quasi-experiment opportunity for exploring the impact of neighbourhood characteristics on the labour market integration of refugees.
For example, we might expect refugees that are assigned to neighbourhoods with high employment to do better in the labour market because they will have better access to positive role models and peer effects. We were able to use Norwegian register to estimate whether this effect was significant. We found that there were indeed positive effects on the refugee’s later employment probability of having ‘good neighbours’ in the first neighbourhood (i.e. being placed in an initial neighbourhood with a higher employment rate), but that these effects were small and often statistically insignificant. A one standard deviation higher employment rate in the initial neighbourhood implies a less than one percentage point increase in the refugees’ later employment probability. These results suggest that the employment rate in the initial neighbourhood matters little for the refugees downstream in labour market integration.
Segregation dynamics
One of the reasons why the initial settlement of refugees may have little impact is that migrants may, in time, move away. A potentially important factor in driving thee relocation decisions is “homophily” – the desire to live near other households of the same ethnic group or country of origin. Over time, homophily can lead to cumulative increases in the concentration of ethnic groups in particular locations. To explore the implications of this, we develop a theoretical simulation model which suggests two key processes that drive the long-term patterns of segregation. The first is “spatial persistence”, which refers to the potential of initial cohorts of migrants to create path dependency in the settlement patterns of future migrants. In theory, this will lead to spatial persistence because the share of migrants currently flowing into a particular neighbourhood may be determined by the previous patterns of migrants in that same neighbourhood from a decade or more ago. If there is a high degree of spatial persistence, this is likely to indicate a high degree of path dependency in future patterns of migration flows, and may imply that the geography of migrant residence becomes resilient to change very early on. The second key factor is the “homophily horizon” – the degree to which migrants take into account the wider geography of the ethnic or migrant mix when making their location decision. Simulation work by Bakens and Pryce (2019) implies that an increase in the homophily horizon rapidly increases the rate at which a society becomes segregated. It is therefore a potentially important concept for understanding the long-term drivers of segregation.
Understanding the evolutionary dynamics of segregation has important implications for policy. If, for example, migrant settlement patterns tend to remain relatively fluid even after successive rounds of migration, then policy makers have flexibility with respect to the timing of an intervention that seeks, for example, to change the degree of segregation. On the other hand, if migration settlement patterns become inert relatively quickly, effective intervention may only be possible during a relatively narrow window of opportunity that rapidly closes after the early stages of migration.
We have developed an empirical model to demonstrate how these effects could be estimated using panel data regression methods. Using Scottish Census data from 1971 to 2011, we find strong evidence of path dependency in migration location patterns over a 40-year period. We find that some migrant groups have significantly wider homophily horizons, especially those from poorer nations. These findings may have important implications for the decision of migrants to relocate to other neighbourhoods rather than remaining in their initial settlement housing, which in turn may significantly affect the impact of resettlement programmes. In relation to this, the particular form that segregation takes, particularly whether it generates abrupt rather than gradual social boundaries between neighbouring communities, may also be important for determining the life outcomes and lived experiences of migrants (see our accompanying report for the Nordic Council of Ministers entitled, “Segregation & Social Frontiers”).
The effectiveness of training programmes for new migrants
So far I have mainly discussed our work on the geographical determinants of migrant integration. However, many of the interventions open to policy-makers are not specifically geographical. For example, refugee induction programmes that offer high-quality training might help boost the life-chances of migrants. For example, the Norwegian Introduction Programme provides language, employment and cultural training for new migrants. Attendance on the programme is mandatory for most refugees and family reunification migrants (with non-Nordic spouses). Migrants are given a special financial reward for participating in the programme based on the number of hours of attendance. Crucially, the amount paid changes based on age: under 25s earn benefits at 2/3 the rate of those over 25.
This age cut-off is useful for assessing the effectiveness of the programme because it can be used to measure the effects of cash incentives on programme attendance and its knock-on effect on integration outcomes such as education and employment. We have developed a robust research design for evaluating the programme based on this feature which we are in the process of implementing in order to estimate the following impacts:
- The effect of the age-cut off (via benefit rate) on programme participation
- The effect of the age-cut off (via benefit rate) on employment status, wages, and further education/ training.
- The direct effect of programme participation on integration outcomes, estimated using the results of 1 and 2.
- The variance in programme attendance effects across municipalities: i.e. how the effects identified in 1 and 2 vary across different intervention sites.
- Whether the programme effects differ for different migrant groups by gender and other characteristics.
We anticipate that our initial results will be available in the next few months.
Homelessness impacts of Scottish refugee housing policy
One of the criteria for the successful integration of migrants listed in the 2004 UK Home Office definition was housing. This can refer to the quality and location of housing, but in many countries, migrants – especially refugees – often struggle to find stable housing tenure of even the most basic kind. One of the interventions available to host governments is to allow refugees to apply for social housing. This is not the case in England. However, a policy difference emerged in 2001 between England and Scotland when the Scottish government allowed refugees to apply for social housing.
This created a quasi-experiment for researching whether increasing new refugees’ access to social housing in Scotland helped reduce homelessness relative to England. We investigated this by testing whether the Scottish policy reduced refugee homelessness eight months after permission to remain, using nonresponse rates of the 2005-2009 Survey of New Refugees as a lower-bound proxy for homelessness. We found that refugees assigned to Glasgow, Scotland had a significantly lower homelessness rate than comparable refugees assigned to live elsewhere in the UK. We attribute this effect to allowing refugees priority access to social housing, discounting potential confounders and other mediators.
Policy implications
The integration of migrants should be thought of in dynamic terms. This means avoiding simple static comparisons of migrants and non-migrants. Instead, policymakers should focus on measures of social mobility among migrants, investigate whether these rates are lower than among the native-born population, and explore what drives these discrepancies. A particularly important set of drivers are those associated with “neighbourhood effects” – those arising from differences in the geographical concentration of poverty and disadvantage. We know from US and UK research, for example, that social mobility rates can vary hugely between areas, even within the same region or metropolitan area. As yet, we are not aware of any robust estimates in the UK, Norway or Sweden that explore how social mobility rates differ between migrants and natives, or between migrant groups. In principle, the integration of migrants could depend significantly on the local rates of social mobility. Nordic countries may be less affected by such variations as they have lower levels of geographical inequality more generally. Indeed, in our study of the Norwegian Refugee Dispersal Policy, we found that the employment rate in refugee settlement neighbourhoods had an only a very modest impact on refugee labour market outcomes. However, this may have been due to migrants relocating to neighbourhoods with high concentrations of their own group. Designing effective policy responses to refugee integration will therefore require careful analysis of how internal migration moderates the effectiveness of initial settlement initiatives, which will in turn require an understanding of the drivers and impacts of segregation. Crucially, policy-makers need to take a life-course approach to integration, one that follows and assists migrants and refugees over the various stages of their education, employment and housing trajectories, not just when they first arrive in the country. There is also an imperative to understand and address the drivers of geographic variations in social mobility within all three countries and the extent to which particular migrant groups cluster in low-social-mobility areas.
References
- Andersen, H.L., Meng Le Zhang, Bertha Rohenkohl, Urban Lindgren and Liv Osland (2022) Estimating the effect of place on intergenerational social mobility: How different are results in a Scandinavian welfare setting? Nordforsk Life at the Frontier Working Paper (in preparation).
- Andersen, H.L., Osland, L. and Le Zhang M.L. (2022) “Labour market integration of refugees and the effects of the neighbourhood”, NordForsk Life at the Frontier Working Paper (in preparation).
- Bakens, J., & Pryce, G. (2019). Homophily horizons and ethnic mover flows among homeowners in Scotland. Housing Studies, 34(6), 925-945.
- Chetty, Raj, and Nathaniel Hendren. 2018. ‘The Impacts of Neighborhoods on Intergenerational Mobility I: Childhood Exposure Effects*’. The Quarterly Journal of Economics 133 (3): 1107–62.
- Heidrich, S. (2017) Intergenerational mobility in Sweden: a regional perspective. Journal of Population Economics 30, 1241–1280.
- Olner, D. Meen, G. and Pryce, G. (2022) Dynamics of Migrant Segregation: Spatial Persistence & Homophily Horizons in Migrant Settlement Patterns in Scotland (1971-2011), NordForsk Life at the Frontier Working Paper (in preparation).
- Zhang, M. L., Henrik Andersen, Liv Osland (2022) The impact of the Norwegian Introduction Programme on migrant integration: A natural experiment using changes in age-related benefits, NordForsk Life at the Frontier Working Paper (in preparation).
- Zhang, M. L., Sin Yi Cheung, Jenny Phillimore (2022) Did increasing new refugees’ access to social housing reduce homelessness? Evidence from a quasi-experiment, NordForsk Life at the Frontier Working Paper (in preparation).