Poverty and Overcrowding among Immigrant Children in an Emerging Destination: Evidence from Finland

This paper aims to analyze the patterns of poverty and housing overcrowding among immigrant children in Finland. We seek to explore whether and to what degree foreign-born children are disadvantaged relative to native children in terms of income poverty and overcrowded housing. Another main objective is to study the patterns of immigrant child poverty and overcrowding in the first years of settlement. We distinguish between four different types of poverty trajectories in the first 5 years after arrival in Finland: (1) no experience of poverty; (2) poor in up to two out of 5 years following arrival (mostly non-poor); (3) poor in three or four out of 5 years following arrival (mostly poor); and (4) poor in all 5 years (chronic poverty). An analogous classification is applied when looking at housing overcrowding. We use data from a compilation of Finnish registers, which contain annual information on all individuals who resided in Finland at any point between 1995 and 2014. The results of a series of logistic regressions show that the disadvantage of immigrant children relative to native children is more pronounced in terms of income poverty than in terms of housing. The most frequent outcome in terms of income poverty in the first years of settlement is no experience of poverty, followed by chronic poverty, i.e., poverty in all 5 years after arrival. The same patterns are found for overcrowding. The multivariate analyses, based on generalized ordered logistic regressions, show substantial heterogeneity across immigrant groups.


Introduction
Finland has been a multi-ethnic society for centuries, mainly due to the strong presence of a Swedish-speaking community on its southern and western shores. However, with the exception of the Sami community in the north of the country, the presence of other ethnic groups has been marginal until recent decades. This changed with an increase in immigrant inflows in the 1990s. A stronger presence of the foreign-born population suggests that, similar to some Southern European countries, Finland is experiencing a migration transition (Katila and Wahlbeck, 2012;Castles et al., 2014). In other words, what until recently was a land of substantial emigration has become a land of immigration. The integration of the foreign-born has become one of the principal concerns of contemporary European societies. Although immigration to Finland is fairly recent and the foreign-born population is still smaller than that of other Nordic countries, immigration and integration policies have played prominent roles in political debates in Finland. The challenges of foreign-born residents' integration in Finland are mainly associated with the country's lack of institutional and societal experience. The aim of this paper is to analyse the patterns of income poverty and housing conditions among immigrant children in Finland. In particular, we seek to explore whether and to what degree immigrant children are disadvantaged relative to native children in terms of these two indicators of well-being. Another principal aim of this study is to explore the associations between the socio-demographic characteristics of immigrant children's families at arrival on the one hand, and, on the other, income poverty and overcrowding in the first years of settlement. Using rich data that combines information on the country of birth and mother tongue, we are able to take into account the substantial heterogeneity of Finland's immigrant population.
There is ample literature showing that poverty can have adverse effects on children's health and academic achievement, as well as on a wide range of socio-economic outcomes in the adulthood (e.g., Lichter, 1997;Duncan and Brooks-Gunn, 2000;Seccombe, 2000). Childhood poverty also has a particular resonance because deprivation in childhood can have adverse, long-term consequences (Gornick and Jäntti, 2012). Additionally, concern for children's well-being may to some degree be driven by "innate feelings of protection towards the young and assumptions of their blamelessness for the situation in which they find themselves" . Our interest in housing conditions is guided by the assumption that a living environment can contribute to the intergenerational transmission of social inequalities. The empirical evidence shows that children's academic achievement, behaviour, and health are negatively affected by growing up in crowded housing, even after controlling for income and other socio-economic characteristics (e.g., Evans et al., 2001;Leventhal and Newman, 2010;Solari and Mare, 2012). Previous studies on the well-being of immigrant children in the Nordic countries suggest that immigrant children are clearly disadvantaged in terms of poverty risks. Lindquist and Lindquist (2012) identify immigrant child poverty as one of major challenges for Swedish family policy, whereas Galloway et al. (2015) conclude their study on Sweden, Denmark and Norway by arguing that immigrant child poverty may be "the Achilles heel of the Scandinavian welfare state". A recent study by Gustafsson and Österberg (2018) illustrates that, while 17% of children who experienced poverty between 1983 and 1985 had an immigrant background, the share was as high as 57% in the period between 2008 and 2010. As these studies do not include Finland in their analysis, our study will complement our knowledge of the well-being of immigrant children in the Nordic countries.

Immigration to Finland
According to Statistics Finland, there were 387,000 foreign-born residents in Finland at the end of 2018, accounting for more than 7% of the population. Data from the same source suggest that the number of people with foreign backgrounds increased tenfold between 1990 and 2018. Until the 1990s, practically all the immigrant inflow consisted of return migrants and their children (Saarela and Finnäs, 2009). The migration flows at the beginning of the 1990s were clearly dominated by Estonians and Russian-speaking immigrants from the former USSR, although they also included a considerable number of Ingrian Finns, a Finnish-speaking community from the Russian Federation. However, the composition of the immigrant population in Finland diversified considerably in subsequent years, mainly due to two factors. First, the country received substantial numbers of asylum seekers and quota refugees from different parts of the world in the past three decades. Second, Finland's accession to the EU made it easier for labor migrants from developed European countries to settle in Finland. As a result, Finland's immigrant population today is fairly heterogeneous. Apart from Russian and Estonian speakers, who still constitute the largest language groups among migrants, another eight languages are represented with more than 10,000 speakers: Arabic, Somali, English, Kurdish, Farsi, Chinese, Albanian, and Vietnamese (Statistics Finland, 2019).
Finnish studies on immigrant integration largely confirm the patterns of integration that are typical of other European destination countries. In particular, foreign born inhabitants on average have less favorable socio-economic outcomes than natives, with the immigrant disadvantage being especially pronounced among immigrants from less developed countries. Immigrants have lower labor market participation rates and, even when employed, their employment is less stable and they earn less than comparable natives (Heikkilä, 2005;Sarvimäki, 2011;Busk et al., 2016;Sarvimäki, 2017). Although the level of residential segregation in previous decades was modest, it has increased considerably in recent years. In some neighborhoods, the share of the foreignborn population reaches 40%, exceeding 50% for school-age children (Vilkama, 2011). However, a recent study by Kauppinen and van Ham (2019) shows that there is no indication of a general tendency toward self-segregation among non-Western migrants.

Economic Trends and Poverty in Finland
Finland is one of the most prosperous societies in Europe, even though its economy has experienced several shocks over the past three decades. A serious economic crisis took place at the beginning of the 1990s, which resulted in a cumulative drop of 12.6% in the country's real GDP (Gulan et al., 2014). This crisis was followed by economic recovery and growth that lasted until the Great Recession. After the economic recession of the early 1990s, the income poverty rate and income inequality increased in Finland. The increases in poverty and inequality observed during the 1990s were among the highest of the OECD countries (OECD, 2011). While the level of income inequality was relatively stable in the early 2000s, the income poverty rate continued risingalthough at a somewhat slower pace than in the previous decadeuntil the 2008 recession (Blomgren et al., 2012;Moisio et al., 2016). A sharp economic downturn in 2009 has been followed by a relatively slow recovery (OECD, 2018a). During the 1990s recession and its aftermath, the Finnish welfare state experienced changes that can be described mainly as welfare state retrenchments (Kuivalainen and Nelson, 2012;Blomgren et al., 2012).
Poverty prevalence and poverty persistence among children are determined by general economic trends, income distribution, the female labor market participation rate, and the strength of the welfare state in the country of destination (Jenkins and Schluter, 2003). Despite fluctuations in general child poverty trends over the past 20 years, Finland and other Nordic countries are among the countries with the lowest child poverty rates in the world (Bradbury and Jäntti, 2001;Corak, 2006;Gornick and Jäntti, 2012). Finland is also among the few countries where the child poverty rate is lower than the overall poverty rate (Adamson, 2012), which may be at least partly ascribed to universal child benefits and other benefits granted to households with children (TARKI, 2010). As elsewhere in Europe, the characteristics associated with child poverty in Finland are single parenthood (especially single motherhood), a large family, low education, and lack of employment (TARKI, 2010;Gornick and Jäntti, 2012;Karvonen and Salmi, 2016). The relatively positive results in Nordic countries are, however, somewhat undermined by evidence on the persistence of child poverty: namely, it has been shown that the exit rates from child poverty are similar to or even lower than those in some other developed countries (Vaalavuo, 2015).

Housing
The quality of housing in Finland is high and above the EU average (Eurostat, 2017). Our analysis (results not reported here) shows that, among both natives and immigrants, only a marginal share of housing units have poor amenities. Therefore, as housing amenities do not seem to be an important source of disadvantage for immigrant children in Finland, we do not look at this indicator of well-being. On the other hand, the average size of a housing unit in Finland (88.6 m 2 ) is lower than that in other Nordic countries. In addition, with an average size of around 55 m 2 , the average size of a rented housing unit is smaller than that in most other EU countries (Eurostat, 2012). Finland is characterized by a high level of home ownership, but the social housing sector plays an important role as well. In addition, the housing costs of low-income households are alleviated with a means-tested housing allowance. There are substantial differences in tenure status between natives and immigrants in Finland. In 2010, 72% of natives lived in owner-occupied housing, while this was the case for only 31% of immigrants. On the other hand, 43% of immigrants lived in social/public housing, as compared to 13% of natives. Immigrants were also overrepresented in the private renting market (Andersen et al., 2013).

The Mechanisms of Poverty and Overcrowding among Immigrant Children
Native and immigrant families are arguably quite similar in terms of the principal determinants of poverty. Regardless of a family's origin, disposable income is simultaneously determined by the same three principal factors: (1) the labor market, through its effect on earned income; (2) the state, through its effect on received and paid transfers; and (3) family size, which determines the magnitude of a family's needs (Obućina, 2014). In addition, income distribution is another important determinant of poverty status for both natives and immigrants, as it directly affects the threshold of relative poverty. However, in terms of their relationships with markets and the welfare state, there are sizeable differences between native and immigrant families, as well as between immigrant families originating from different countries (Van Hook et al., 2004). As discussed above, previous studies have demonstrated that immigrants are disadvantaged in the Finnish labor market, in terms of both employment and earnings. These differences emerge, as in most Western countries, due to the lack of countryspecific skills among migrants, but also as a result of discrimination (Larja et al., 2012). In addition, although the Finnish welfare state is residence-based rather than citizenshipbased, it does not seem to be able to offset the effects of the labor market gap between natives and immigrants. Although a large share of immigrant families in Finland receive last-resort welfare benefits, these are often not sufficient for exiting poverty (Nelson, 2013). On the other hand, native recipients on average receive higher amounts of welfare benefits, as they are more often eligible for benefits that are based on previous earnings in Finland (Sarvimäki, 2017). We therefore expect that immigrant children are more likely than native children to be poor, even after controlling for parental labor market participation and family size. At the same time, the factors causing different poverty trajectories between native and immigrant children can also cause substantial differences among children from different immigrant groups. The parents of foreign-born children of Finnish background (mostly return migrants) have better language skills and higher levels of other country-specific skills than do parents of other foreign-born children. Due to considerable similarities northern European institutional settings, it is likely that Nordic immigrants are more familiar than other European and non-European immigrants with the Finnish labor market and welfare state, at least upon arrival. Discrimination is another agent that produces differences across immigrant groups. Liebkind et al. (2016) find that there is an ethnic hierarchy in the Finnish labor market: although less likely to be hired than Finns, Western immigrants are less exposed to discrimination than are other immigrants. Given these considerations, we expect to find substantial differences in poverty among children across immigrant groups, with the children of Finnish background having the most favorable outcomes.
There is little doubt that income is one of the most important determinants of a family's housing conditions in general and of housing size in particular. However, economic resources alone cannot fully explain the differences in overcrowding between native and immigrant children. At least two other factors come into play when explaining the immigrant disadvantage in housing. First, immigrants are less familiar with the local housing market and usually do not have strong language skills upon arrival, with the exception being immigrants of Finnish background. Such impediments may complicate and prolong the search for adequate housing (Kauppinen and Vilkama, 2016). The second factor is the nearly universal finding that immigrants are discriminated against in the housing marketand the empirical evidence suggests that Finland is no exception to this pattern (Öblom and Antfolk, 2017;Auspurg et al., 2019). The role of discrimination is especially important in the first years of settlement, when purchasing an apartment is still a distant option, and a large share of migrants are confined to the rental market (Dhalmann and Vilkama, 2009). Some landlords may be simply unwilling to rent apartments to immigrants, which may also make the housing search a complicated and lengthy process. Other landlords demand higher rent from immigrant tenants, which implies that immigrant families pay the same price for less living space (Andersen et al., 2013). Bearing in mind these considerations, our expectations for housing are identical to those concerning poverty, specifically: we expect native children to be less likely than immigrant children to live in overcrowded housing. We also expect to find considerable differences among children from different immigrant groups, with foreignborn children of Finnish background being the least likely to experience overcrowding.

Data and Methodology
We use data from a compilation of Finnish registers: the Population Register, the Family Register, the Migration Register, and the Finnish Longitudinal Employer-Employee Data (FLEED). The Population Register, the Family Register, and FLEED contain annual information on all individuals who resided in Finland at any point between 1995 and 2014, which is also the time span of this study. The Migration Register contains information on the month and year of immigration to and emigration from Finland. The information in different registers can be merged via an anonymized personal ID. Apart from a wide array of socio-demographic and contextual characteristics for the general population, we also have access to detailed immigrant-specific information, such as the country of birth, native language, and date of immigration. We focus on the well-being of children, i.e., individuals aged 17 or younger. Two dimensions of well-being are addressed in this paper. First, we look at relative income poverty. Consistent with the dominant approach in European literature, the poverty line is equal to 60% of the median household income. The reference population for determining the poverty threshold is the entire population residing in Finland. The modified OECD-scale is used to adjust for family size (1 for the first adult, 0.5. for each additional adult and 0.3 for each child). The second dimension of well-being that we look at is overcrowding. Here, we use the definition of overcrowding adopted by the European Union. According to this definition, a person lives in an overcrowded household if the household does not have at its disposal a minimum number of rooms equal to: one room for the household; one room per couple in the household; one room for each single person aged 18 or more; one room per pair of single people of the same gender between 12 and 17 years of age; one room for each single person between 12 and 17 years of age who is not included in the previous category; and one room per pair of children under 12 years of age.
The empirical analysis of both indicators is divided into two parts. In the first part, we perform separate regressions for each year between 1995 and 2014, and then compare poverty and overcrowding rates among immigrant children with those of Finnish-born children. Dependent variables are being poor (in the analysis of poverty), and residing in an overcrowded dwelling (in the analysis of overcrowding). The multivariate analysis is based on logistic regression. All independent variables refer to the year of the analysis. In the second part of the empirical analysis, we focus only on immigrant children and explore the patterns of poverty and overcrowding in the first five full years after arrival in Finland. 1 This analysis includes foreign-born individuals who arrived in Finland as children between 1994 and 2009 and who were 12 years old or younger in the year of arrival (hence younger than 18 at the end of their fifth full year in Finland). The characteristics of this population are shown in Table A.3 in the Appendix. The descriptive analysis of poverty and overcrowding after arrival is based on sequence analysis. When analysing poverty, the dependent variable has the following outcomes: (1) not poor in any of the first 5 years after arrival (no poverty); (2) poor in up to two out of 5 years following arrival (mostly non-poor); (3) poor in three or four out of 5 years following arrival (mostly poor); and (4) poor in all 5 years after arrival (chronic poverty). Similarly, when looking at overcrowding, the dependent variable has the following outcomes: (1) adequate housing in each of the first 5 years after arrival in Finland (no overcrowding); (2) overcrowding in up to two out of 5 years after arrival (mostly adequate housing); (3) overcrowding in three or four out of 5 years following arrival (mostly inadequate housing); and (4) overcrowding in all 5 years after arrival (chronic overcrowding). Since there is a hierarchical relationship between the outcomes, we employ generalized ordered logistic regression for the purpose of multivariate analysis. This approach relaxes the proportional odds assumption of ordered logistic models. At the same time, unlike multinomial logistic regression, it respects the ordinal nature of the dependent variable (Williams, 2006). 2 It is also important to emphasize that, unlike in the first part of the analysis, all independent variables in the second part refer to the year of the child's arrival in Finland. Around 2000 immigrant children are omitted from the overcrowding analysis. These are children who, during any of the first five full years in Finland, lived in a housing unit inhabited by two or more families, or for whom the structure of the housing was unknown. 3 The main variable of interest in logistic regressions on native-immigrant differentials is an indicator variable for the foreign-born children. These models include a number of other covariates. In order to examine whether the socio-economic vulnerability of children varies by age while also considering that housing needs increase as children get older, the models control for the age of the child. We distinguish between the following categories: age 6 or younger; 7-12; and (only in the first part of the analysis) 13 or older. In order to take into account the families' resources and needs, we include in the models a number of family socio-economic and demographic characteristics. We control for the age of either the older parent or the only parent of a child living in a single-parent family. This variable is categorized as follows: age 30 or younger; 31-40; 41-50, and older than 50. The analysis also takes into account the family type, where we distinguish between: married couple with child(ren) (reference category); mother with child(ren); father with child(ren); cohabiting couple with common child(ren); and cohabiting couple with non-common child(ren). Also included is a continuous variable indicating the number of other children in the family. We further control for the highest education level of parents, using the classification of Statistics Finland: less than upper secondary education or unknown 4 (reference group); upper secondary; lowest level tertiary; lower-degree level tertiary; higher-degree level tertiary, and doctorate or equivalent level. Labor market attachment among the adults in the family is measured as the percentage of adults in the family who are employed. The models also include two contextual variables. Because the structures of economic opportunities as well as of available housing may differ in urban and rural regions, we introduce a categorical variable indicating the degree of urbanization: inner urban area (reference category); outer urban area; peri-urban area; and rural area. Another categorical variable takes into account the possible impact that regional trends and policies have on poverty trajectories and housing conditions. This variable has 19 categories, one for each Finnish region. 5 With the exception of the dummy variable for foreign-born children, all other independent variables from the analysis on immigrant-native differentials are included in our models examining the well-being of immigrant children in the first years of settlement. Furthermore, as this part of the analysis covers only foreign-born children, it also includes several immigrant-specific variables. The heterogeneity of the Finnish immigrant population makes it necessary to control for immigrant group. A nonnegligible share of foreign-born children has at least one parent of Finnish background, and they are also registered in our data as children of Finnish background. These are mostly foreign-born children of return migrants and they constitute a separate group, regardless of their country of birth, with the exception of those born in the former USSR. 6 A non-negligible share of children (especially from earlier immigrant cohorts) is registered as having been born in countries that no longer exist, such as the USSR or Yugoslavia. Therefore, instead of simultaneously including former and newly emerged countries in our models, for these two cases we assign the former country as the country of birth, but combine this information with that of native language when identifying immigrant groups. 7 For Iran and Iraq, we also combine information about country of birth and native language in order to account for pronounced ethnic heterogeneity in immigrant inflows from these countries. We identify as many as 17 immigrant groups, classified as follows: foreign-born children of Finnish background (reference group); former USSR, Estonian language; former USSR, Russian language; former USSR, Finnish language; former USSR, other languages; Nordic countries; former 4 One shortcoming of information on education in Finnish registers is that it is not possible to distinguish between individuals with less than upper secondary school and individuals for whom the education level is unknown. This is why they are considered a single group in this study. 5 For the detailed regional classification, see: http://tilastokeskus.fi/meta/luokitukset/maakunta/001-2017 /index_en.html 6 Foreign-born individuals of Finnish ancestry born in the former USSR belong primarily to the Ingrian-Finnish community. Although officially considered returnees (Liebkind and Jasinskaja-Lahti, 2000), the parents of most of these children never lived in the area defined by the present-day borders of Finland. This is an important distinction from the children of Finnish-born migrants returning from Sweden or other Western countries. This is why, in this paper, the children with Finnish backgrounds born in the former USSR are, depending on their reported native language, assigned to the groups "former USSR, Finnish language" or "former USSR, Russian language". 7 In order to avoid the collinearity between arrival cohorts and immigrant groups, children born in the countries that gained independence in the 1990s are also assigned to the group defined by the borders of former countries, i.e., the USSR or Yugoslavia. For instance, an Estonian-speaking child born in Estonia in 1995 is classified as "former USSR, Estonian language". Yugoslavia, Serbo-Croatian language; former Yugoslavia, Albanian and other languages; other European countries; Iran, Farsi language; Iran, Kurdish and other languages; Iraq, Arabic language; Iraq, Kurdish and other languages; Somalia; Thailand; China, and other non-European countries. Because the study period saw substantial variations over time in the general economic trendsincluding income inequality and poverty levelsour regressions also control for the period of arrival. This variable is categorized as follows: 2000 or before (reference group); 2001-2005; and 2006-2009. Since immigrant children enter our study in their first full year in Finland, it may be that those who arrive in Finland in earlier months have better outcomes, as their parents have more time to adapt to the new country of residence. This especially concerns income poverty, which takes annual income into account. The models thus also control for the month of immigration to Finland. Finally, we control for differences in children's migration experiences by including a dummy variable for children with at least one parent already living in Finland at the time of immigration.

Immigrant Children's Disadvantage
The general trends in poverty and overcrowding among native and immigrant children are shown in Fig. 1. Over the entire observed period, immigrant children faced substantially higher poverty risks than native children, although the trends within both groups were fairly similar. For both groups, the poverty rate was lowest in 1995, when it stood at 5.5% for native children and 20.8% for immigrant children. The trend was largely negative in subsequent years, with poverty rates peaking in 2011 at 12.8 and 39.8 for, respectively, native children and immigrant children. In the following 3 years, poverty rates decreased somewhat in both groups. The overcrowding rates surpass poverty rates in each year among both native and immigrant children. For native children, there has been an almost steady decline, with the overcrowding rates ranging between 33.2% in 1995 and around 25% in the last 3 years of the observed period. The immigrant children's disadvantage is also evident when looking at housing conditions: the overcrowding rates among foreign-born children ranged between 40.9% and 48.3%. The trends in this group followed a somewhat more complex pattern than among native children, as the initial modest decrease in overcrowding rates was followed by an increase in overcrowding after 2007.
To what degree can the less favorable outcomes for immigrant children be explained by family and contextual characteristics? Starting with income poverty, two main conclusions emerge from Fig. 2. First, the relative disadvantage of immigrant children with respect to poverty remains substantial even after controlling for observable characteristics. More precisely, the odds of living in poverty were between three (in 2003) and three and a half (in 1997) times higher for immigrant children than for native children. Second, there is, however, no clear trend over the observed period. In contrast, when it comes to overcrowding, Fig. 2 indicates some negative trends for foreign-born children. More precisely, controlling for observable characteristics, the odds of living in overcrowded housing at the beginning of the observation period were 29% higher for immigrant children, which is arguably a moderate level of disadvantage. The gap between native and immigrant children has increased substantially since then, such that in 2014 the odds of living in overcrowded housing were 83% higher for the foreignborn children.

Poverty and Overcrowding After Arrival in Finland
The results of the sequence analysis of poverty and overcrowding among newly arrived immigrant children are shown in Fig. 3. The trajectories are ordered by frequency, in descending order. The panel on the left shows that the most common outcome was the absence of poverty in each of the first 5 years after arrival, which was the case for just over 40% of immigrant children. In contrast, the second most common outcome was being exposed to poverty in all of the first 5 years in Finland. This outcome, however, was more than three times less frequent than the absence of poverty. The next most frequent outcomes were transitions from poverty after (in order of frequency) 3, 2, 1 and 4 years in poverty after arrival. Negative transitionsthat is, transitions into poverty after arrivalare more rare, which implies that the average socio-economic conditions of immigrant children do generally improve after arrival. Furthermore, as can be seen in the lower part of the panel, a considerable share of immigrant children experience unstable poverty trajectories. Almost 20% experience two or more poverty transitions in the first years of their stay in Finland. The right panel in Fig. 3 shows overcrowding patterns. Despite the substantial differences in the rates and trends that Fig. 1 shows in regard to poverty and overcrowding, the two panels in Fig. 3 are remarkably similar. The most common outcomefound among just under half of immigrant childrenwas living in adequate housing conditions in each of the first 5 years in Finland. Another similarity with income poverty is that the second most Source: Finnish register data, own calculations  Finland (in percent, 1995Finland (in percent, -2014 common outcome was living in overcrowded housing during each of the first 5 years, followed in descending order by positive transitions after 1, 2, 3 and 4 years in Finland. As a result, there is a positive association between average housing conditions and years of stay in Finland. Finally, a non-negligible share of immigrant children had unstable housing conditions in the years after arrival. Although the general trends in poverty and overcrowding appear similar, and despite those two outcomes being correlated, our additional analysis 8 shows that favorable poverty trajectories do not necessarily imply favorable housing trajectories, and vice versa. To illustrate, more than 10% of the children who experience no poverty after arrival spend all of their first 5 years in Finland in overcrowded housing, whereas 36% of these children spend at least 1 year in overcrowded housing. Among the children who live in chronic poverty after arrival, somewhat less than a third spend all 5 years in overcrowded housing, whereas 28% of them experience no overcrowding at all in the first years of settlement. Table 1 displays the results of the generalized ordered logistic model of poverty among immigrant children during their first 5 years in Finland. Since we report the coefficients in terms of odds ratios, this means that coefficients higher than one make the less favorable outcome more likely. A child's origin is an important predictor of the patterns of well-being in the first years of settlement. As expected, foreign-born children of Finnish background (who are also the reference group in this analysis) have the most favorable outcomes after arrival, followed by Chinese-born children. On average, immigrant children born in Europe have, all else equal, better outcomes than children born outside the continent. The advantage of foreign-born children of Finnish 8 Results not reported, but can be obtained upon request Source: Finnish register data, own calculations background relative to most other immigrant groups varies across outcomes. It is particularly pronounced when we look at the likelihood of experiencing no poverty during the first 5 years in Finland. To illustrate, the odds of experiencing at least some poverty are roughly 13 times higher for Arabic-speaking Iraqi children and Farsi-speaking Iranian children, as compared to the reference group. However, the relative advantage of foreign-born children of Finnish background is smaller when it comes to escaping chronic poverty: the odds of experiencing the least favorable outcome are around three times higher for Farsi-speaking Iranian children and only 35% higher for Arabic-speaking Iraqi children. Our results also suggest that considerable differences may exist between language groups originating from the same country. Arabic-speaking children from Iraq are more likely to experience some poverty, but less likely to experience chronic poverty than Iraqi-born children from the Kurdish and other minority language groups. The differences between the language groups from Iran are, however, less pronounced.
A number of strong associations between other family characteristics in the year of arrival and subsequent poverty trajectories indicate that the initial conditions just after arrival are very important for the well-being of immigrant children in their new country of residence. Older children are somewhat less likely to experience unfavorable poverty outcomes. Children arriving with parents aged 30 or younger are the most vulnerable in terms of income poverty. Children who arrive with married couples have the most favorable poverty outcomes after arrival, whereas children who arrive with single parents are the most vulnerable. It is noteworthy that children with a single father are more likely to experience chronic poverty than children arriving with a single mother. As expected, having more siblings increases family needs and thus increases the likelihood of negative outcomes, albeit modestly. The differences by the degree of urbanization are moderate, with immigrant children who live in rural areas upon arrival having somewhat worse poverty outcomes than children from peri-urban and urban areas. This is especially the case when looking at chronic poverty. Parental labor market participation in the year of arrival and parental education both considerably reduce the risk of negative poverty outcomes. While this finding is expected, it is somewhat more surprising to see that chronic poverty is positively associated with having a parent in Finland who migrated before the child (even if the magnitude of the association is fairly small). The period effects are considerable, with the most recent Source: Finnish register data, own calculations   Table 2. In accordance with our expectations, differences across migrant groups are considerable and also largely in line with those found for poverty. The finding that foreign-born children of Finnish background have the most favorable outcomes also corresponds to our expectations. They have a moderate advantage relative to children born in other Nordic countries and China, as well as to Finnish-speaking children born in the former USSR. However, relative to children from other immigrant groups, their advantage is considerable. Regarding the associations between, on the one hand, the degree of urbanization, education and time of parents' arrival and, on the other, overcrowding, these also resemble the associations for income poverty. Yet, some associations between independent variables and overcrowding differ considerably from the results of the poverty analysis. For instance, older children are more likely to live in overcrowded housing (this finding comes as no surprise, given how overcrowding is defined in our paper). The association between parents' age and overcrowding is not straightforward. Children who arrive in Finland with parents younger than 30 are more likely to experience some overcrowding as compared to children with older parents. At the same time, children who arrive with young parents are less likely to experience chronic overcrowding than children whose parents are between 31 and 50 in the year of arrival. An interesting result emerges when we look at the role of family type: whereas children arriving with single mothers are the least likely to experience overcrowding, those who are accompanied by a single father are the most disadvantaged group. Parental labor market participation helps escape overcrowding, but its role is much less prominent than when looking at income poverty. Also less pronounced are the period effects. Although the earliest arrivals are the most protected against income poverty, they are slightly disadvantaged in terms of overcrowding. Source: Finnish register data; * p < 0.10; ** p < 0.05; *** p < 0.01

The Levels of Deprivation by Family Characteristics
The previous results of this study clearly show that family and contextual characteristics can be a significant source of disadvantage for immigrant children in Finland. In this section, we conduct additional analyses to explore how the actual frequency of four outcomes after arrival varies by some socio-demographic characteristics and the timing of arrival. For the sake of clarity, we use less sophisticated classifications for some variables. When looking at the geographical origin of the family, we group immigrant children into only three groups: foreign-born with Finnish background, European, and non-European. With respect to family characteristics, we aggregate all the groups into two groups: children living in a single-parent household and children living with a couple. When it comes to employment, we distinguish between families with at least one employed adult and those with no employed adults. The period of arrival corresponds to the classification used in the multivariate analysis. These additional analyses for income poverty can be found in Table A.4 in the Appendix. In line with the results of our generalized ordered logistic regression, the analysis shows, first, that immigrant children arriving with a single parent (most commonly with a single mother) are more exposed to income poverty and, second, that later cohorts of children have less favorable outcomes. Differences across immigrant groups are sizeable, with foreign-born children of Finnish background being the most privileged group, followed by children of other European immigrants. However, employment seems to matter more than any other characteristic and can more than offset the disadvantages associated with geographical origin or family type. To illustrate, children of non-European origin arriving with a single parent who enters the labor market upon arrival have somewhat better poverty outcomes than children of Finnish origin who arrive with two non-employed parents. The cumulative impact of family characteristics on poverty trajectories can lead to enormous differences between immigrant children with advantaged backgrounds and those with disadvantaged backgrounds. When looking at children with Finnish origins who live with both parents (or in stepfamilies) and at least one of whom is employed, we can see in Table A.4 that more than four-fifths of them do not experience poverty during their first 5 years in Finland. On the other side of the spectrum are the non-European children who arrive with a single, non-employed parent. Within this group, the share of those who do not experience poverty is less than 10% in the earliest arrival cohort, and less than 5% in more recent cohorts. Table A.5 shows corresponding results for overcrowding. Relative to the poverty trends in Table A.4, the most pronounced difference is that overcrowding is somewhat more common among children who arrive with two parents. In line with the multivariate results, geographical origin and employment also matter. But, the positive impact of parental employment on housing conditions is less significant than it is for poverty, especially among children with Finnish and other European backgrounds. Recent cohorts of immigrant children of Finnish background have more favorable housing trajectories, whereas no clear trends are found in the other two groups. Similar to what we find for poverty, the cumulative impact of disadvantaged family backgrounds can be very large. The most privileged group in terms of housing conditions are immigrant children of Finnish background who arrive with an employed single parent. Three out of four children from this group do not experience overcrowding in any of the first Table 2 Overcrowding among immigrant children after arrival in Finland, 1995Finland, -2014  More than four out of five of children in this group lived in overcrowded conditions for at least a year during the first 5 years in Finland, and between 35% and 41% of children from this group lived in overcrowded housing in each of the first 5 years after arrival.

Conclusion
The goal of this study was to analyse the well-being of child migrants in Finland from a multi-dimensional perspective. We focused on two important measures of living conditions: income poverty and overcrowding. In this section, we start with a discussion on the findings related to poverty, after which we discuss our results on overcrowding. Lastly, we address the limitations of this study and avenues for future research.
Our analysis clearly shows that, as expected, foreign-born children are more susceptible to poverty than native children. After controlling for observables, the odds of living in a poor family were between three and over three and a half times higher for immigrant children. Although the magnitude of the poverty gap is sizeable, we find no clear trend. This is most likely due to substantial variation over the observed period in economic trends and income inequality, which are the major determinants of poverty trends in both native and immigrant populations. Moreover, it is possible that the lack of a clear trend in the poverty gap is also a result of migration dynamics and the changing composition of the immigrant population during the period of study. Our supplementary analysis of Finnish register data shows that the share of foreign-born children of Finnish background among all foreign-born children dropped from 51% in 1995 to 28% in 2014. On the other hand, the same period saw the share of European Source: Finnish register data; * p < 0.10; ** p < 0.05; *** p < 0.01 children increase from 31% to 41% while the share of non-European children more than doubled, increasing from 15% to 31%. When looking at the poverty trajectories after arrival in Finland, the single most common outcome is its absence in each of the first 5 years. However, the second most common outcome is being poor in all 5 years. In addition, most immigrant children do experience poverty at some point after arrival and a non-negligible share of them experience two or more poverty transitions. The generalized ordered logistic model shows that differences among children from different immigrant groups are substantial. As we expected, the foreign-born children of Finnish background have the most favorable poverty outcomes. However, it should be noted that their advantage relative to most other groups is more strongly pronounced in terms of the likelihood of experiencing the most favorable outcome (no poverty), than in terms of avoiding the least favorable outcome (chronic poverty). The results also show some important differences in poverty trajectories between the language groups from Iraq. Future integration efforts should thus consider multiple dimensions of social affiliation. Furthermore, our results suggest that the socio-economic characteristics of the country of origin are not perfect predictors of the social integration and well-being of immigrants in general, and of immigrant children in particular. This is especially true in light of the relatively favourable outcomes of Chinese-born children. The processes of selection into migration in the countries of origin should thus be taken into account when designing future integration policies.
While our hypotheses concern differences among children of different origins, the results we obtain for other independent variables are also important for our understanding of well-being among immigrant children. In line with the general literature on poverty, parental education and employment are associated with lower poverty risks. Moreover, our analysis shows that parental employment can more than offset the disadvantages associated with geographical origin or family type. Increases in income inequality and poverty in the general population during the period under study also imply that later cohorts of immigrant children were more susceptible to poverty. We also find that children arriving with a single parent have less favorable poverty trajectories as compared to children arriving with two parents. Although they constitute a small share of the population under study, it is interesting to note that immigrant children who arrive with a single father are even more exposed to poverty than those who arrive with a single mother. A supplementary analysis that we carried out (results not reported) suggests this is not due to single mothers marrying more often than single fathers in the first years of settlement in Finland. In actuality, those who arrived as single fathers are more likely than single mothers to be partnered after 5 years in Finland. However, single fathers partner with (or bring from the home country) economically inactive women more often than single mothers partner with (or bring from the home country) inactive men. It may thus seem even paradoxical that the vulnerable position of children who arrive with single fathers is associated with a low level of employment among immigrant women. According to some views, the widespread use of the child home care allowance (CHCA) in immigrant families decreases the incentives for immigrant mothers to enter the labor market while also having negative effects on the social integration of immigrant children (OECD, 2018b). However, the CHCA is not targeted to immigrant mothers, and it has been one of the central elements of Finnish family policy since the 1980s (Nyby et al., 2017). It is therefore questionable whether the extensive use of the CHCA among immigrant mothers alone will be a strong enough reason for policy makers to reconsider the availability of these benefits. However, Finnish integration policies should include more incentives for immigrant mothers to enter the labor market. These policies should be primarily aimed at recently arrived women as well as at women with long spells of inactivity in Finland. The incentives to enter the labor market should also include training programs that match immigrants' skills and experience. Furthermore, the existing language-learning programs have not shown satisfactory results (OECD, 2018b). Finnish does not belong to the Indo-European family of languages and is considered very difficult to learn. Thus, the acquisition of language skills for immigrant mothers and fathers in Finland is more challenging than for newcomers to most other European destinations. It can therefore be argued that developing innovative languagelearning programs should be one of the highest priorities on the integration agenda.
When comparing the two indicators of well-being, the trajectories in terms of income poverty and overcrowding look similar in the first years of immigration. However, the prevalence of and trends in income poverty and overcrowding among immigrant children clearly differ over the observed period. In addition, favorable trajectories in terms of one indicator of well-being do not necessarily imply favorable trajectories in terms of the other. This suggests that the mechanisms determining poverty and overcrowding do not entirely overlap, which accentuates the importance of including multiple indicators when analysing children's well-being. As compared to the native-immigrant poverty gap, the relative disadvantage of immigrant children with respect to overcrowding is smaller in magnitude, but the results we obtain suggest a negative trend. The odds of living in overcrowded housing are around 30% higher at the start and more than 80% higher at the end of the observation period. Contrary to what we saw in the analysis of income poverty, the fact that a trend exists isat least to some extentmost likely due to the more stable nature of determinants of overcrowding. Whereas the poverty threshold changes every year, the definition of overcrowding remains the same over the observed period. On a similar note, while disposable income changes every year in almost all households, only a small share of households moves to another dwelling or experiences a demographic change that may trigger a transition from overcrowded to adequate housing, or vice versa. The finding that the trend is negative is partly a result of price increases in the Finnish housing market. Moreover, the price increases were clearly more pronounced in the rental market (where an overproportional share of immigrants are concentrated) than in the sales market (Eskelä, 2018;Eurostat, 2019). The negative trend was in all likelihood also shaped by changes in the composition of the immigrant population. As discussed above, the groups that are economically more vulnerable and more exposed to discrimination were more represented in later arrival cohorts. In light of the evidence that discrimination in Finland exists in both the labor and housing markets (Larja et al., 2012;Öblom and Antfolk, 2017), it may be necessary to expand on existing legal and institutional antidiscrimination mechanisms.
As predicted, and similar to the findings on poverty, sizeable cross-group differences are found in the experience of overcrowding among immigrant children. Foreign-born children of Finnish background have the most favorable housing trajectories in the years after arrival. However, the role of parental employment and the period of arrival are considerably less important than when looking at poverty. Another noteworthy result in the overcrowding analysis is that children arriving with single mothers have the most favorable outcomes in the years after arrival. To some degree, this result is due to the fact that single-parent families need less living space than couples with children. However, this finding also suggests that access to social housing as well as meanstested housing allowances have had some effect on securing adequate housing conditions for children with single mothers.
Although the previously mentioned Nordic studies on immigrant child poverty have different research designs, all of them have found that immigrant children are heavily disadvantaged in terms of poverty risks. Our study shows that Finland is no exception to this trend. This unfortunately suggests that ethnic stratification may be an important part of social inequalities in Finland in the decades to come. Moreover, the high prevalence of child poverty among immigrants in a country with one of the lowest child poverty rates in the world may contribute to the stigmatization of the foreign-born population. The overall picture becomes even more negative in light of immigrant children's increasing disadvantage in terms of housing.
Although we show that the initial conditions just after arrival can be good predictors of well-being trajectories in the years following arrival in Finland, we do not claim causality in this study. Except when discussing in this section the differences between children with a single mother and those with a single father, we do not explicitly address labor market transitions, demographic changes within the family or other events that can trigger changes in the two indicators of well-being during the first years of immigration. What is more, we do not have information on the countryspecific skills and experiences of discrimination among immigrants to Finland, both of which are important factors responsible for native-immigrant differentials in wellbeing. We also lack information on the reason for migration, aspirations, and dominant values in the families of immigrant children. Yet, these unobserved factors can be important sources of variation in the integration trajectories not only across individuals, but also across groups. Future research should thus focus on events after arrival in order to obtain a fine-grained picture of their relative importance to the well-being of immigrant children. In order to address the factors that are usually unobserved in registers, the future research should, whenever possible, combine registers with survey data and qualitative research.