Migrants’ subjective well-being in Europe: does relative income matter? Open Access

Happiness may be the ultimate goal of life, but it is not an immutable concept and depends heavily on life circumstances. Questions about ‘happiness’ and ‘life satisfaction’ are typically used to measure subjective well-being (hereafter SWB), which is an individual’s evaluation of their own life as a whole (van Praag et al. 2003). In this article, we contribute to the study of SWB across Europe by testing the role of migrants’ relative economic position, compared with relevant reference groups. In their discussion of migrants’ happiness, relative to that of the general population, Hendriks and Bartram (2018) suggest that we need detailed information on the factors that contribute most to the happiness of this distinct group of people.

Prior studies have explored the association between income and SWB (Bartram 2011); variability of migrants’ SWB by their country of origin (Amit 2010; Amit and Risse 2014); decline of SWB with increased length of stay in the destination country (Obućina 2013); and change in SWB from the first to the second generation (Safi 2010). However, limited research has investigated migrants’ SWB, relative to reference groups in their destination country (Liu et al. 2019; Gelatt 2013; Bălţătescu 2014, 2007, and 2005).

This paper uses income, relative to that of natives and of migrants already present in the host country, as a means of understanding variation in migrants’ SWB. We operationalise migrants’ level of ‘relative income’ by defining two measures of economic distance that compare a migrants’ position with the average of two reference groups: natives and migrants, grouped by gender, age, educational level, actual country of residence and area of origin. Compared with previous research, our approach provides a more precise indication of the distance between each individual and the targeted reference group, because we rely on a large set of segmenting variables (e.g. Melzer and Muffels 2012; Firebaugh and Schroeder 2009; Brown et al. 2008; Clark et al. 2008a; Ferrer-i-Carbonell 2005; Luttmer 2005; Veenhoven 1991). We also extend the existing literature by scrutinising group-specific heterogeneity in the relation between relative income and SWB by migrants’ generation and educational level. Most studies (e.g. Clark et al. 2008a, 2008b; Hendriks and Burger 2019) have used objective measures to proxy subjective evaluation of an individual’s economic position, referring to this approach evaluation as ‘social comparison’. However, these objective data often lack subjective information that is necessary for social comparison. Here, we evaluate, in a final step, whether subjective and attitudinal variables mediate the relationship between relative income and migrants’ SWB.

Our main research questions were:

1. Is income relative to two reference groups (natives and other migrants) associated with migrants’ SWB and, if so, to what extent?

2. Are there group-specific variations in the relationship between relative income and SWB by educational level and generation of migration?

3. Do subjective and attitudinal variables mediate the relationship between migrants’ relative income and SWB?

To address these questions, we use data collected from 2002 to 2018 by the European Social Survey (ESS). These data include extensive coverage of European countries and were collected with a level of methodological rigour that ensures a high degree of comparability between countries (Jowell et al. 2007).

The seminal paper of Easterlin (1974) suggested that income plays a minor role in happiness, once an individual rises above subsistence level (the so-called ‘Easterlin paradox’). He also argued that happiness is strongly influenced by ‘relative status’. The theory of relative deprivation was originally described by Festinger (1954) and Runciman (1966), then applied to migration and further developed by Stark (1991) and Stark and Taylor (1989) to explain the decision to migrate more than the relationship between migration, integration and changes in SWB.

Clark et al. (2008a) state that individuals compare themselves to those they consider similar (i.e. external reference points) and, as a result, relative income is the ‘status return’ from income. Several scholars have asserted that SWB depends more strongly on people’s perception of income in relation to past income and on the perceived incomes of their peers than on their absolute income level (Hendriks and Burger 2019; Clark et al. 2008a, 2008b; Easterlin 2004; Clark and Etilé 2008; Ball and Chernova 2008; McBride 2001), while others (Hagerty 2000; Melzer and Muffels 2012; Caporale et al. 2009; Scoppa and Ponzo 2008) find that the effect of absolute income is larger than that of relative income. Caporale et al. (2009) found that the income of a reference group negatively affects SWB, even after controlling for absolute income and other personal and demographic characteristics. Similarly, Senik (2009) found that income comparisons impact SWB per se and that people suffer from relative deprivation, rather than from absolute income inequality.

The literature on migrants’ economic and social integration and assimilation across Europe is extensive, most often investigating integration parameters such as education (e.g. Kunz 2014), occupation and wage (e.g. Semyonov et al. 2014; Creese and Wiebe 2012), or linguistic distance (e.g. Strøm et al. 2018; Chiswick and Miller 2002). Economics research on migrants’ integration uses Alba and Nee’s (1997) definition of assimilation; a migrant group assimilates if there is a ‘reduction of differences between similar native groups over time’. For example, migrant performance in the labour market is compared to that of a native worker with the same characteristics, such as gender, age, education, family structure and/or workload. Typically, there is an underutilisation of migrant workers skills which affects also their wages, and declines as their time in the destination country increases. However, the probability that a first-generation migrant’s wage equals that of a non-migrant counterpart is low (Algan et al. 2010). Frequently, segmentation of the labour market continues to negatively impact second-generation migrants’ wages more than real job discrimination does (Strøm et al. 2018). Despite the domination of integration and assimilation theories in the migration literature and the growing body of research on migrants’ SWB, few empirical studies have been conducted on the relationship between migrants’ income, relative to their specific reference groups, and SWB.

The relation between relative income and migrants’ SWB likely varies between upward comparisons – to others whose wealth, fame, or status is in some way superior – and downward comparisons to those whose status is inferior (Liu et al. 2019; Lahusen and Kiess 2018; Prilleltensky 2012; Oshio et al. 2011; Boyce et al. 2010). Downward comparison increases people’s positive self-perception, while upward comparison has the opposite effect (Fiske 2011; Delhey and Kohler 2006). Luttmer (2005) found that upward comparisons lead to more critical evaluations of one’s own life and reduce SWB, while downward comparisons lead to less critical evaluations and increase SWB. Gokdemir and Dumludag (2012) showed that, for Turkish immigrants residing in the Netherlands, absolute income does not affect the level of life satisfaction, while comparisons of income and social status to those other groups do, especially upward comparison with Dutch natives. Gelatt (2013) suggests that migrants simultaneously maintain reference groups in their destination country and their country of origin with a strength that varies by country of origin.

The relationship between relative income and SWB may be strongly moderated by generation and educational level. Several studies have found differences in the level of SWB by generation and length of stay in the receiving country. For instance, Arpino and de Valk (2018) observed that the gap in SWB is greater between first-generation migrants and natives than between second-generation migrants and natives. Safi (2010) confirmed this change in SWB from the first to the second generation, while Obućina (2013) found that migrants’ SWB declined as the length of stay in Germany, their destination country, increased. Hendriks and Burger (2019) suggest that, although migrants initially have as reference group comprised of people from their home countries, their frame of reference partially shifts over time toward natives and other migrants in their host country.

Education also influences SWB. Clark and Oswald (1996) suggest that this effect may be due to the positive relationship between education and aspirations. Moreover, migrants who have obtained higher educational levels may have more contacts with the native community (De Palo et al. 2007), with whom they may tend to build comparisons. Migrants’ level of education also influences their probability of being employed, the stability of their employment and the discrimination they experience in the labour market (van Tubergen et al. 2004; Heath et al. 2008). These labour market outcomes may shape the relationship between relative income and SWB.

Several studies have used relative income as a proxy for social comparison (e.g. Clark et al. 2008a, 2008b; Hendriks and Burger 2019). However, this practice goes beyond the informative capacity of relative income, referring to the evaluation of one’s societal position, relative to some reference groups. To gain a deeper understanding of social comparison mechanisms, subjective measures, such as feelings or attitudes, should be taken into account. For instance, Bălţătescu (2005) refers to ‘social comparison’, using subjective variables about an individual’s evaluation of the socioeconomic environment. He found that migrants have a higher degree of satisfaction with societal conditions than natives, supporting the model of social comparison. Hendriks and Burger (2019) used four subjective variables (economic satisfaction, government satisfaction, trust in public institutions, social trust) and found that wavering perceptions of the host country’s socio-economic and institutional conditions were associated with less positive SWB trajectories among diverse, first-generation immigrants in Europe. In summary, subjective measures may have an important role in mediating the relationship between income – both absolute and relative – and SWB.

The association between relative income and SWB may be confounded by several factors (Hooghe and Vanhoutte 2011). Demographic confounders include gender, age, marital status, and presence of children in the household. Women are generally less satisfied with their lives than men (e.g. Tesch-Römer et al. 2008). The effect of age is unclear, with research identifying a U-shape, others an inverted U-shape and others a linear relationship (for a review, see López Ulloa et al. 2013).

The empirical literature suggests that relationship status – being in a partnership – is positively linked with SWB (e.g. Vignoli et al. 2014). Although some studies have found a positive association between parenthood and happiness (Kotowska et al.2010), others have shown that having children imparts either non-significant or negative effects on levels of SWB, when bias resulting from selection into parenthood is removed (Clark and Oswald 2002; Clark et al. 2008b).

Other confounding factors are related to an individuals’ health and socioeconomic position. SWB is positively connected with good physical health, a relationship that is bidirectional and mediated by various factors (for a review, see Cross et al. 2018). Education and employment status are markers of individual autonomy, intellectual ability and independence (e.g. Argyle 1999) and may affect migrants’ SWB. Residential area – urban, suburban, etc. – is also important, since migrants are often concentrated in relatively deprived areas (Simpson et al. 2009), a factor that may affect their SWB (Knies et al. 2016).

The relationship between religion and SWB has been widely explored in the literature (for a review, see Tay et al. 2014). Most findings suggest a positive association between SWB and religious involvement (Lelkes 2006), religious beliefs and intrinsic religiousness (Diener et al. 2011), and church attendance (Lim and Putnam 2010).

In addition, migrants’ country of origin is crucial, potentially reflecting varying job opportunities, social exclusion or discrimination, and affecting both relative income and SWB. Empirical research in Sweden (Rydgren 2004, le Grand and Szulkin 2002) and France (Silberman et al. 2007; Simon 2003) has found that immigrants from Western countries do not face particular problems to integrate into the labour market, while immigrants from other areas – especially Africa, Asia and Latin America, in the case of Sweden, and French former colonies and/or predominantly Muslim countries, in the case of France – are more disadvantaged in the labour market.

We used the cumulative dataset of the ESS, a repeated, cross-sectional survey involving 33 countries¹ conducted semi-annually from 2002 to 2018. We evaluated SWB using the question on life satisfaction: ‘All things considered, how satisfied are you with your life as a whole nowadays?’ measured on an 11-point scale, ranging from 0 (extremely dissatisfied) to 10 (extremely satisfied).

We first used the complete ESS dataset of 318,044 individuals (272,230 natives and 45,814 migrants) to form in-sample reference groups and compute relative income measures for migrants. We noted some differences between natives and migrants in the distribution of the absolute income variable. For natives the modal class was the 8th decile of income, the median was 5 and the mean value was 5.45 (SD 2.65); for migrants the modal class was the 4th decile, the median was 5 and the mean value was 5.27 (SD 2.62). As expected, migrants were income-disadvantaged, compared with natives. According to prior research (Liu et al. 2019; Prilleltensky 2012; Luttmer 2005), migrants’ income, relative to natives, can be considered an upward comparison, while comparison with other migrants’ income is a downward comparison. This interpretation should be taken cautiously, because it implies that our measure includes a subjective evaluation of relative income, while it is only an objective measure.

Then we conducted our analysis on the migrants’ sample². The mean age of migrants was 45.98 years, 52.60% were women (N = 24,097), 59.57% were employed an average of 39.2 h per week. The modal value of migrants’ life satisfaction was 8, the median was 7 and the average was 6.71 (SD 2.37). Complete descriptive statistics are reported in the Appendix (Table A1).

Extending existing studies, which individualise reference groups using a few characteristics simultaneously (e.g. Melzer and Muffels 2012; Firebaugh and Schroeder 2009; Brown et al. 2008; Clark et al. 2008a; Ferrer-i-Carbonell 2005; Luttmer 2005; Veenhoven 1991), the two reference groups we used to compare migrants’ income were: (i) natives of the same gender, age³ and level of education⁴, residing in the same host country and (ii) migrants of the same gender, age, level of education and area of origin⁵, residing in the same host country.

The first set of models was estimated without absolute level of income. In the second set, we included absolute income to account for its potential to inflate coefficients of relative income¹³.

To address research question 2, we split the sample by generation and level of education prior to running a set of models.

To address research question 3, we examined whether subjective perceptions mediate the relationship between relative income and SWB. We included additional variables, measuring ‘feeling about household income nowadays’ (coded as four dummies: ‘living comfortably’, ‘coping’, ‘difficult’, ‘very difficult’). These variables indicate an individual’s perception of their condition. Furthermore, to account for individuals’ perceptions of living conditions in their destination countries, we added five variables that describe how an individual feels about some aspects of well-being in the country. These variables, which account for the ‘immigrant’s societal perception’ (Hendriks and Burger 2019) include¹⁴ satisfaction with the present state of the national economy, the government, the way democracy works, the current state of education and the current state of health services, measured on an 11-point scale from ‘extremely dissatisfied’ to ‘extremely satisfied’. We ran a principal component analysis (PCA) on the five original variables, then using the score of the first component as a variable¹⁵. We refer to this new synthetic variable as ‘feelings about socio-economic-institutional conditions in the country’.

Table 1 reports the results of models run on the whole sample to assess the role of relative income and address research question 1. Our analysis shows that, as the distance between the respondent’s income and the median value of natives’ income decreases (column 1), eventually becoming positive, their SWB increases (β = 0.110, p-value <0.001). We observe a similar pattern when the main explanatory variable is income relative to other migrants (column 2), although the coefficient is smaller (β = 0.100, p-value <0.001). This result may be explained by upward versus downward comparison (Liu et al. 2019; Gelatt 2013; Prilleltensky 2012; Fiske 2011; Luttmer 2005), and is consistent with literature according to which upward comparison predominate and consequently is more important in explaining SWB (Lahusen and Kiess 2018; Delhey and Kohler 2006; Clark et al. 2008a).

When the same models were fit with absolute income added (columns 3 and 4), we observe a reduction in the magnitude of the coefficients of income, relative both to natives (β = 0.064, p-value <0.001) and to migrants (β = 0.038, p-value <0.001). This result is likely due to the fact that the absolute value of income is embedded in our relative measures.

We find that absolute income has a stronger association with migrant’s SWB than relative income. Although relative and absolute incomes are correlated (Scoppa and Ponzo 2008), relative income remains statistically significant even after the inclusion of absolute income, suggesting that our variable of interest is associated with SWB. The reduction of coefficients following inclusion of absolute income is more marked in models that include income relative to other migrants.

The other covariates included in the models are mostly in line with the classical findings in the happiness literature, which provides an indirect validation of our model¹⁶. Gender is not significant¹⁷, while age displays a non-linear pattern, indicating that young migrants are more satisfied. Living with a partner is also associated with an increase in the SWB, while the presence of children living in the household is negatively related to migrants’ SWB. SWB increases with education and employment, while being unemployed decreases SWB, as does the increase in the number of working hours. SWB is positively associated with better [perceived] health. Being religious is associated with higher levels of SWB, although religious affiliation itself is not significant. Living in suburban or smaller areas, rather than is large urban environments, is associated with higher SWB.

Compared with the second generation born in the host country, first-generation migrants from North America and Oceania report higher levels of SWB, as well as migrants from Central and South America (but the coefficients are not statistically significant when controlling also for absolute income). Migrants from Other African countries and North African Countries report lower levels of SWB. This finding suggests that forced migration, which may be more common from Africa, and chosen migration, which is more typically observed from wealthier regions, such as North America and Oceania, have an opposite relationship with SWB in the host country. These differences may also reflect initial economic conditions of migrants and different levels of socio-economic integration within the host society and its labour market. A sensitivity check (briefly described in the Appendix) assures the robustness of our findings to a different specification.

We present all following models including the absolute value of income, which partially absorbs the magnitude of relative income. The coefficients of both relative income variables in the estimates presented hereafter increase if we exclude absolute income from model equations (results available upon request).

Following suggestions in the literature, we have run the main models separately on the first and second-generation subgroups, additionally splitting the sample by gender (Table 2). These estimates show three main results. First, the importance of relative income for the second generation is higher than that for the first generation¹⁸. Second, the magnitude of the coefficient of relative income is larger for women than for men. Third, for both migrant generations, income relative to natives is more relevant than income relative to migrants.

Results (Table 3) show that the magnitude of relative income comparisons with natives and migrants is generally higher for those with a higher level of education, except for men in the model that includes income relative to other migrants. In particular, income relative to migrants is statistically significant only for medium-high educated migrants²⁰. In addition, for more highly educated migrants the role of relative income may be more important for women. Again, these estimates confirm that income relative to natives is more important than that relative to migrants, except for those who are less educated in the full sample and the men-only sample (where the estimates are not statistically precise).

In a final step, we explored whether individual’s perception of their situation and of the socio-economic and institutional conditions in their receiving country mediates the relationship between relative income and migrants’ SWB. Results (Table 4) show that SWB increases as ‘feelings about household income nowadays’ improve (from ‘difficult’ to ‘living comfortably’). Satisfaction with the current socio-economic and institutional conditions of the receiving country is also positively related to migrants’ SWB.

Interestingly, the inclusion of subjective variables in our models reduced the magnitude and significance of both relative and absolute income. In particular, in the model including income relative to natives, the medium absolute income is significant at 10%, while the relative income variable remains statistically significant at 1%. In addition, the coefficient of income relative to natives is greater than income relative to other migrants. Overall, these results show that the self-perception of both individual condition and societal conditions matter, mediating the role of income, especially absolute income. The implication of this finding is that accounting for these subjective variables may reduce the income effect, which may be overestimated in studies that do not include subjective measures.

Migrants’ SWB is an expanding and important area of research, because a convergence of migrants’ SWB to the level of natives may generate other benefits (Hendriks and Burger 2019), such as improved integration (Richardson 1967; De Neve et al. 2013). In this paper, we concentrated on the relationship between SWB and migrants’ income relative to two reference groups – natives and other migrants – with the same characteristics. At least four findings have emerged from this paper in response to our three research questions.

First, we note that as the disadvantage between the migrant and the reference groups becomes smaller, eventually becoming an advantage, the migrant’s life satisfaction increases. This result holds even after controlling for absolute income. In contrast with some authors (Hendriks and Burger 2019; Clark et al. 2008a, 2008b; Easterlin 2004; Clark and Etilé 2008; McBride 2001), but in line with others (Melzer and Muffels 2012; Caporale et al. 2009), we find that absolute income has a stronger association with migrant’s SWB than the relative income level.

Second, we find that income relative to natives shows a stronger relationship with SWB than income relative to other migrants, a result that holds when looking at various subgroups. Consistent with the existing literature (Liu et al. 2019; Prilleltensky 2012; and Gelatt 2013; Fiske 2011; Delhey and Kohler 2006; Luttmer 2005), we assert that upward comparisons lead to more critical evaluations, while downward comparisons lead to less critical evaluations. Lahusen and Kiess (2018), referring to an extended concept of social comparison rather than simply relative income, explain that the way people form upward and downward comparisons is related to ‘salience’, which is what people feel to be more important. Thus, the relationship between SWB and relative income may be stronger when looking up [to natives] and weaker when looking down [on other migrants’] relative income.

Third, we detect relevant group-specific differences. Results show that, for second-generation migrants, who have lived in the receiving country and society throughout their lives, with a consequent rise in expectations and aspirations, the relationship between relative income and SWB is higher than for first-generation migrants, regardless of whether the income comparison is with natives or migrants. Migrants may initially compare their situation primarily to that of people in their home countries, then gradually shift their frame of reference toward natives and other migrants living in their host country (Hendriks and Burger 2019). This result may also be related to the fact that first and second-generation migrants report different levels of SWB (see, e.g. Arpino and de Valk 2018). Differences also emerge by level of education; the relationship between relative income and SWB is higher for those who have the highest level of education. Better-educated migrants may have higher expectations and aspirations and, therefore, may be more sensitive to their relative position in comparison with their reference groups. In addition, migrants with more education have more contacts with natives, while less-educated migrants tend to socialise inside their ethnic communities, a pattern that influences their perceptions (De Palo et al. 2007).

Finally, we explore the role of subjective measures, at both individual (feelings about household income) and societal (feelings about the socio-economic-institutional conditions in the host country) levels. We have found that these variables can be considered important mediators of the relationship between relative income and SWB. The introduction of these variables into our models reduced both the absolute and relative income coefficients. In particular, absolute income was no longer significant in some models. This result is consistent with previous findings on social comparison (Hendriks and Burger 2019; Bălţătescu 2005), which state that the subjective evaluation of one’s individual and societal conditions may have a relevant effect on SWB.

Some caution is necessary when interpreting our results. Although this study provides important insights into migrants’ SWB within the context of their relative income, it has limited power for inferring causal relationships. Importantly, none of our analyses accounted for the fact that individuals with a predisposition toward reporting a higher level of SWB may also systematically vary in their sense of belonging to certain comparison groups. In addition, the relationship between relative income and SWB may be the same for natives who experience the same labour market conditions as migrants. Finally, the raw data may be biased by selectivity. For example, only migrants who speak the receiving country’s language and, presumably, are relatively integrated are interviewed (Chiswick and Miller 2002; Strøm et al. 2018). Additionally, migrants at different stages of their immigration process or who differ by nationality are often segregated into areas and jobs with other migrants and, as a result, are at least as likely to compare themselves to other migrants. Although we controlled for employment variables, we cannot exclude that the association between relative income and SWB may be due to the specific situation of migrants in the labour market. A lower SWB may reflect [partial] labour market exclusion or precarity. Thus, our result may conceal differences across subgroups by country or by other characteristics.

Despite these limitations, our analysis reveals some of the axes on which relative income is related to migrants’ SWB. Overall, the size of the coefficients for both type of relative income is relatively small, a result that may depend on data structure (e.g. deciles of income) or analytical strategy (e.g. distance from the median value of the reference group). Nonetheless, our analysis adds to the existing literature, in particular by exploring the differences that emerge across different groups. Moreover, our results suggest that subjective variables should be taken into account in analyses devoted to exploring the relationship between income (both absolute and relative) and SWB.

Future efforts should be directed at verifying the associations identified in this cross-sectional study through the use of panel data and adoption of causal approaches. Further developments of this work might also consider using a measure of income level in their country of origin to assess whether this factor could affect migrants’ SWB especially in the initial phase of their immigration. A clearer understanding of distance from reference groups is necessary for designing policy favourable to migrants’ integration into their host country.

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

Manuela Stranges is Assistant Professor of Demography at the University of Calabria (Italy).

Daniele Vignoli is Full Professor of Demography at the University of Florence (Italy).

Alessandra Venturini is Full Professor of Political Economy at the University of Turin (Italy).

A set of sensitivity checks supported the robustness of our results. Complete results are not shown for space constraints but are available from the authors upon request.

First, we fit the complete model including both measures of relative income (with natives and with migrants). We found that, while the coefficient of ‘relative income with natives’ maintains magnitude and significance, the coefficient of the ‘relative income with migrants’ loses both magnitude and significance.

Second, we considered that those who are in the very low or very high deciles of the distribution may perceive less their income relative to peers. Thus, we ran two additional sets of models, excluding from the analysis, first individuals in the lowest and the highest decile of income, then individuals in the two lowest and two highest deciles of income. In both tests the coefficients of the relative income variables remained significant, except for ‘relative income with migrants’, when the two lowest and two highest deciles of income were excluded.

We conducted additional checks to test whether our decisions about variable construction influenced the results. To verify our strategy of harmonisation of income variables between ESS rounds 1–3 and ESS rounds 4–9 (see Note 6), we ran all models presented in the paper, while excluding the first three rounds of the survey. No major changes occurred in the results; signs, magnitude and significance were stable. This outcome reassured us that our choice to include rounds 1–3 was valid.

In addition, we changed the composition of the reference groups, using two different thresholds for education: EISCED 3 in a first test and EISCED 4 in a second one. The variables describing relative income remained significant in all models. We also ran models splitted by educational level using the new thresholds. Results confirm that, for more highly educated migrants, relative income is more relevant in determining SWB.

Although our grouping choices for migrants’ area of origin is based on both geographical and political criteria, these groups may not be realistic. For example, among EU27 immigrants, that an immigrant from Norway would compare themselves to an immigrant from Bulgaria seems unlikely, and vice versa. The same would apply to Asia; immigrants from Japan have different experiences and concerns than immigrants from Syria. For this reason, we conducted a robustness check by creating alternative groupings of countries within areas. We grouped European countries by gross domestic product into in ‘medium’ and ‘rich’ countries (per capita annual GDP >=€30.000) and ‘poor’ countries (per capita annual GDP <€30.000). We used the same classification, expressed in US dollars ($33.000), for Asian countries. We maintained African, Central-South America (all countries in these groups were below the individualised threshold) and North America plus Oceania (all the countries were above the individualised threshold) in their original groups. Results with this new classification did not show differences from the estimates reported in the main text; North America plus Oceania was positively and significantly related to SWB, while North African and other African countries were negatively related. More importantly, the coefficient of our main independent variable did not change in magnitude and significance.

Finally, to test our use of first and second-generation groups, we ran alternative regressions using a different specification (as proposed in Arpino and de Valk 2018): G1, first-generation migrants (born outside the country); G2, ‘strict’ second-generation migrants (born in country to two migrant parents); G2.5, mixed second-generation (born in country to one migrant parent). The results still show generational differences, with both measures of relative income being more related to SWB for the second generation than the first, with no substantial differences between G2 and G2.5.