Descriptive statistics separated by migrants and non-migrants are presented in Table 1. Overall, about 8% of all observations (N = 104,589) in the sample are from respondents born in another country than the one they are living in at the time of interview. Regarding our dependent variable, migrants show on average only a slightly lower CASP value than natives. Comparing the sociodemographic characteristics, we see no striking differences between migrants and natives, with two exceptions: Migrants make up a higher share of people with financial difficulties and, unexpectedly, the educational level measured according to the International Standard Classification of Education (ISCED-97) is slightly higher among migrants. Latter holds for all migrant groups except Southern European migrants (tabulation not shown). Two-thirds of the migrants have the citizenship of the country of residence. They mostly migrated a long time ago. The mean length of residence in the host country is 40.3 years. While the majority of them migrated after the age of 18 or far beyond, one-third moved abroad in their early childhood or adolescence, most likely along with their families. This shows that the migrant population in SHARE is special not only in respect to age but also in the sense that most of the migrants have already been living in the host country for a very long period.
Table 2 shows the distribution of migrants and their origin regions (i.e., Northern/Central Europe, Eastern Europe, Southern Europe, and non-European areas) across all destination countries. For 85 migrants the information on region of origin is missing. The table shows that the distribution of all migrant groups is very heterogeneous across countries, which makes it necessary to control for country fixed effects in our regression models. Overall, migrants from Northern/Central Europe immigrating to other countries in Northern and Central Europe (DK, CH, SE, and BE) are the largest group with 36%, followed by non-Europeans with 31%. Especially non-European migrants might exhibit lower levels of SWB because high institutional barriers can hamper their social integration (e.g., legal access to labor market depending on citizenship). A closer look at the countries with the highest share of non-European migrants shows that in NL they are mainly from Indonesia and the former Dutch territories in the Caribbean, in FR and IT mainly from Northern Africa, and in ES mainly from Latin America and Morocco (not shown here). Both migrants from Southern and Eastern Europe make up about 16% in total, with the former representing the highest share in Luxembourg (mainly from Portugal) and the latter being the largest group in Austria and Germany (mainly from former Yugoslavia, former Czechoslovakia, and Poland).
As the first step of our analysis, we explore the differences in SWB between migrants and natives by running random effects regression models to estimate group-specific growth curves controlling for age, time of interview (wave), and country. In Fig. 1, it can be seen that within the older population and compared to natives, migrants show significantly lower levels of subjective well-being. The differences decrease with increasing age and become statistically insignificant beyond the age of 78.
Figure 2 displays the immigrant-native gap by origin regions. The horizontal line represents the CASP level of non-migrants. For Northern/Central European migrants no significant differences can be observed. Their SWB level is almost equal to the one of natives. Eastern European, Southern European, and non-European migrants show CASP levels that are significantly lower than the levels of the native population. Surprisingly, the gap is largest for Southern European and not—as expected—for non-European migrants.
Next, we examine individual factors that may have an impact on reducing the immigrant-native gap by estimating multivariate random effects regression models. As illustrated in Fig. 3, we start with a basic model (M1) controlling for age, time of interview (wave), country and then stepwise add additional control variables: sociodemographic characteristics and health (M2) and having financial difficulties (M3). Then we add our independent variables: having the citizenship of the country of residence (M4), having migrated before/after the age of 18 (M5), and finally length of residence (M6; for natives the latter equals age). It can be observed that each model contributes to explaining the variation in SWB between migrants and natives. While sociodemographic characteristics and health (M2) do not show large effects, the gap becomes considerably smaller after accounting for the financial situation (M3), having the citizenship of the country of residence (M4), and having migrated before the age of 18 (M5). The years migrants have resided in the destination country (M6) slightly contribute to reducing the gap. After all, even after controlling for all individual characteristics in the full model, the immigrant-native gap remains significant.
By moving our analysis to the country level, we first analyze the group differences between countries by controlling only for age, time of interview (wave), and country. The predictive margins in Fig. 4 illustrate that there are large variations concerning the size of the immigrant-native gap across countries. Migrants have a lower level of SWB than the respective native population in all countries with the exceptions of ES and IT. The differences are largest in NL and DK.
Since we observe great variation in terms of integration policies in Europe, we complete our analysis by exploring to what extent the country disparities are associated with their institutional framework. Controlling for all individual factors (M6), Fig. 5 plots the differences in SWB of migrants relative to natives (y-axis) against the country-specific average score in the MIPEX policy area family reunion (x-axis). The horizontal zero line represents the SWB level of natives. The slope of the graph clearly shows a positive association with family reunion policy context. The immigrant-native gap is comparably large in countries with low MIPEX scores (i.e., rather restrictive family reunion policies) and becomes smaller among countries with higher scores (i.e., more open family reunion policies). For instance, controlling for all individual factors, the CASP score of migrants in DK is on average one CASP point lower than the one of natives, whereas in ES it is one CASP point higher than in the native reference group.
The results turned out to be robust after running our analyses separated by gender and by replacing CASP with life satisfaction as a quality of life measure (not shown here).