Chapter 2. France: Intergenerational mobility outcomes of natives with immigrant parents

The deep gender gap for the offspring of immigrants in France

For many years the educational outcomes of the children of immigrants remained unknown in France because data allowing them to be identified were not available (Simon, 2003). The first systematic and very influential study came out in 1996. Using a proxy measurement of origin in panel data from the French Ministry of Education, Vallet and Caille (1996) showed that among children who entered upper secondary school in 1989, those with an immigrant background clearly suffered from an early educational disadvantage that could be identified at the primary and lower secondary levels (higher dropout rates, lower academic performance). This group’s disadvantage was confirmed by all subsequent studies, whatever the source of data. Some of them showed that inequalities vary significantly by gender.

That girls perform better at school than boys is not an unusual result (Felouzis, 1993). Rather, what is striking is that the gender gap is much deeper with the children of immigrants than with the mainstream population. Using ministry panel data (cohort entering 6^th grade in 1995), Brinbaum and Kiefer (2009) showed that “most high achievers among immigrants’ children are girls, particularly of North African origin” (p. 541). In this group, boys are three times more likely than girls to drop out of secondary school without any diploma, with proportions reaching 28% for boys against 9% for girls (Figure 2.1), while the ratio is below two in the mainstream population (9% for boys and 5% for girls). Results of the TeO Survey confirm a gender gap that is deeper among immigrants’ children. While the proportion without the BEPC diploma (brevet des collèges, obtained in France at the end of lower secondary school) is 15% for men vs. 10% for women in the second generation, it is 8% and 7% respectively in the mainstream population (Table 2.2). TeO results also show striking differences by origin. Notably, the sons of immigrants from Africa and Turkey have the highest share of students that do not graduate at the end of lower secondary school1: between 19% and 27%, compared to “only” 8% in the male mainstream population (Table 2.2). In other terms, among visible minorities, one young man out of four or five lacks the lowest diploma that exists in France. On the other hand, sons and daughters of immigrant(s) of European origin have generally better outcomes than those of non-European origins, and – in some cases – even better than those of the mainstream population (Table 2.2).

Figure 2.1. Share of persons without qualification at the end of secondary school

Natives with two immigrant parents and mainstream population who entered lower secondary school in 1995, percentages

Note: No qualification means no diploma from secondary school – that is to say, neither the BEPC (diploma at the end of lower secondary school) nor any sort of baccalauréat (diploma at the end of upper secondary school), nor vocational qualification (including CAP [certificat d’aptitude professionnelle], BEP [brevet d’études professionnelles] and brevet de technicien).

Source: INSEE, DEPP (Direction de l’Evaluation, de la Prospective et de la Performance), 1995 panel of students entering lower secondary school in 1995 and its follow-ups the survey Famille 1998 and the survey Jeunes 2002. Adapted from Brinbaum and Kieffer, 2009, Table 7b.

In their pioneering analysis, Vallet and Caille (1996) showed that the difficulty for natives with immigrant parents mostly had to do with their social origin. This result is echoed throughout much of the literature, remaining valid in most subsequent quantitative studies in France. However, it is a debated issue whether some inequalities by origin persist with all other things being equal. Results vary depending on the indicator and, obviously, on the list of independent variables that are introduced in the models. According to Vallet and Caille (1996), children with a migration background had, other things being equal, greater results in primary and lower secondary schools than their native peers with no migration origins. Subsequent work with the same data following educational paths up to the end of upper secondary school confirmed, on average and again other things being equal, the net advantage of children with immigrant parents (Brinbaum and Kieffer, 2009).

However, the same data also showed that youths from that group more frequently goes onto vocational tracks, which are the least prestigious, than those of the mainstream population (Brinbaum and Lutz, 2017). PISA results (OECD, 2016) also show that students with a migration background (including immigrants and their offspring) are more likely to have a lower performance in science than children of natives after controlling for socio-economic status (SES), even though the socio-economic background matters relatively more in France than in other OECD countries (OECD, 2016, p. 252). Beyond the general effect of migration, these results obscure important inequalities by specific origin, gender and also education level.

In fact, the social background does not systematically or uniformly explain the overall disadvantage of natives with immigrant parents. In multivariate analyses of the TeO Survey, the remaining effect of origin varies by gender and also by level of education (Table 2.3). At this point it is instructive to look at some of the pivotal moments of educational trajectories in France. At the BEPC level (lower secondary diploma), results show that young males with Turkish and African origins have a significantly higher probability of not obtaining a degree than the French mainstream population whereas girls with parents from Africa show no such difference. Among girls, only those with parents from Turkey have a significantly higher probability of not obtaining a diploma compared to the mainstream population. At the baccalauréat level (upper secondary level), the male ethnic “penalty” disappears, except for those with Turkish parent(s).2 As for women, all things being equal and again with the exception of the Turkish group, they have in fact higher odds of obtaining their diploma than natives’ daughters. Finally, at the university level, almost all differences disappear between immigrants’ native offspring and the mainstream population, for both men and women.

Overall, these results show that gender and ethnic differences diminish at higher educational levels. Such a result raises an important interpretation issue, with policy implications. On one hand, Vallet (1996) interprets the convergence of the school results of the children of immigrants with those of the mainstream population as a sign of “schooling assimilation”. On the other hand, this convergence can also be interpreted in terms of institutional selection. TeO findings also suggest that there are important gender differences: Whilst boys achieve significantly lower results at the BEPC level, girls are indistinguishable at the secondary level and outperforming their peers with French native parents at the university level. What explains these differences in educational achievement among immigrants’ sons and daughters?

Based on the review of qualitative research conducted in the 2000s, Lorcerie (2011) shows that the French education system carries two distinct and opposite ethno-gender biases. One favours young women with non-European background and especially from Muslim families: they are seen as persons who deserve special attention in order to be protected from the supposed sexism of their family environment. On the other hand, boys of the same origins suffer from negative stereotypes associated with delinquency and deviant behaviours. These observations echo statistical results on perceived unfair treatment in the education system. While men and women in the mainstream population declare fairly similar treatment at school, immigrants’ sons report much more frequently than immigrants’ daughters the feeling that they were treated differently in terms of grading, discipline, the way they were addressed, and the school’s recommendation as to which stream they should follow at the end of lower secondary school (Figure 2.2).

Figure 2.2. Share of students declaring they felt unfairly treated while they were at school

Native-born with immigrant parents and mainstream population, aged between 18 and 35 in 2008

Notes: Natives with immigrant parents were born in France and have at least one parent who was born outside France without French citizenship. 8% of the women born to immigrants declare that they felt unfairly treated in the way they were addressed by their teachers.

Source: Trajectories and Origins Survey (TeO), INED-INSEE, 2008. Figures excerpted from Brinbaum and Primon, 2013. Weighted results.

Differences in the way young men and women are treated are observed not only in schools but also in families. Challenging the common wisdom that girls would be disadvantaged in families of extra-European background, Moguérou and Santelli (2015) have actually shown a reverse pattern of schooling support in immigrant families. While fathers in the mainstream population help their sons with their homework more frequently than their daughters, immigrant fathers, with the sole exception of those from sub-Saharan Africa, tend to help their daughters more than their sons (Table 2.4). Thus the question arises: why do immigrants’ parents invest more in girls’ education? Qualitative research among Maghrebi families show that girls are more controlled than boys, spend more time at home, and (thus) are less distracted from their schoolwork (Moguérou and Santelli, 2015). This is reflected in the fact that immigrants report more conflicts about school with their sons than with their daughters: 27% of both mothers and fathers report such conflicts with their boys, against 13% of the fathers and 15% of the mothers with their girls (Attias-Donfut and Wolff, 2009).

Also, in offering their support, parents would accord priority to the children who have the greater success at school. Once they realise that daughters are more likely to succeed and to fulfil their dream of upward social mobility, they switch attention from boys to girls. This process, especially observed in Maghrebi families, is labelled “the diagonal of the generations” (Delacroix, 2004). It could also reflect the fact that migrant families – at least some of them – see in formal education a way for the girls to escape the role traditionally assigned to women, and a path to financial autonomy and social emancipation. Although this aspiration for greater autonomy for women is usually attributed to the mothers and daughters themselves, the fathers’ involvement suggest that they also contribute to this social change (Table 2.4). This result tends to contradict received wisdom about gender roles in extra-European families, especially in Muslim culture. It calls for further research to better disentangle the role of personal, maternal and paternal aspirations to explain the relative success of immigrants’ daughters.

Whatever the gender of their child, immigrant parents are less able than other parents to help their children with their homework because of language barriers and other socio-economic handicaps. In the early 1990s, according to data from the French Ministry of Education, 81% of immigrant mothers declared they felt overwhelmed with their children’s homework vs. 51% among native mothers from the working class (Brinbaum, 2013). However, immigrant families use alternative resources to compensate for their own difficulties to monitor and support their children’s formal education. On the one hand, siblings play a great role. In immigrant families, the youngest are more often supported by their elder siblings than is the case among natives with a similar social background (Brinbaum, 2013; Moguérou and Santelli, 2015). On the other hand, immigrant families also use school support services more often, usually provided at no cost by associations. While only 8% of the mainstream population aged 18-35 in 2008 frequented free tutoring classes when they were at school, the average proportion is 19% among natives with immigrant parents, and the proportion rises to 35% among those of Sahelian origin, i.e. mainly from Mali and Senegal (Brinbaum, Moguérou, and Primon, 2015). This high frequentation of tutoring classes certainly reflects the fact that immigrants’ children have more difficulties at school than other children. It also suggests that parents with a migration background are concerned with their children’s school success.

In fact, the question of parental aspirations for their children has sparked a great deal of interest in migration studies in France. In their pioneer work, Vallet and Caille (1996) showed that immigrant families have, other conditions remaining the same, higher aspirations for their offspring in terms of educational achievement than native-born parents. This was observed by later panels following younger generations (Brinbaum and Kieffer, 2009) and confirmed by several studies combining qualitative and quantitative methods that focused particularly on extra-European migrants (Moguérou and Santelli, 2015; Brinbaum and Delcroix, 2016). The rationale for this native-immigrant difference is that the migration project is fuelled by an aspiration for upward social mobility. The transmission from the parents to their children of their family history, especially in terms of intergenerational social ascension, is thought by several authors to be a success factor for the children (Santelli, 2000; Moguérou and Santelli, 2015; Brinbaum and Delcroix, 2016). According to Vallet (1996), the French school system is especially adapted to take on board families’ expectations for two reasons. First, families’ wishes play a major role in the school’s recommendations for upper secondary school streaming. Second, the bifurcation between the academic and vocational tracks happens later than in other countries, such as Germany.3 That grants immigrant families and their children time to adapt to the French school system and adopt more informed strategies.

From one generation to the next: Educational mobility upward or downward?

To what extent do immigrant families’ expectations indeed translate into upward mobility? Are the native children of immigrants more educated than immigrants generally? Are they more educated than their parents? Answering these questions is not as straightforward as might appear; several approaches can be used.

A common but dubious approach consists in comparing education levels of immigrants and the children of immigrants in a cross-sectional perspective, as is done in Table 2.5. In this example, the offspring of immigrants does not actually correspond to the children of the immigrants taken into account in the table. For instance, in the TeO Survey, only one immigrant of ten coming from African countries surrounding the Gulf of Guinea (e.g. Benin, Congo, Ivory Coast) had a child or children who could be included in the sample, i.e. children born in France and over 18 years old at the time of the survey (Beauchemin, Lhommeau, and Simon, 2015). As a result, some groups of native offspring seem to be less educated than immigrants of the same origin (Table 2.5). This should not suggest that there is a downward assimilation trend (Moguérou, Brinbaum, and Primon, 2010). Actually, it instead reflects compositional changes in migration flows. For instance, the very high proportion of individuals with a higher degree among sub-Saharan migrants reflects the fact that African migration has recently become very selective. But these newly arrived and highly educated migrants are not the parents of the offspring of immigrants of the same origins included in the survey.

Table 2.6 offers a more realistic view of intergenerational mobility by comparing the level of education of the native children of immigrants with their actual parents. Here, as expected, immigrants’ children of all groups are significantly more educated than immigrants, with a high correlation between their respective education levels (a coefficient of 0.83 – see Figure 2.3). The pace of this upward educational mobility varies by origin. In groups dotted below the correlation linear regression line, the offspring of immigrants have reached an education level that exceeds what could have been expected considering their parents’ level. Children of immigrants from Southeast Asia exemplify this quick upward mobility. But those from Portugal, Morocco and Tunisia also moved very rapidly along the educational ladder. On average, immigrants’ children of all these groups experienced a quicker educational mobility than the mainstream population. On the other hand, groups dotted above the regression line lag behind: considering the education level of their parents, the children should be more educated. This is the case for the children of only two groups: the least educated group of immigrants (Turks) and the most educated one (Central Africa and Gulf of Guinea). Both cases could raise some concern, although for different reasons. Children of Turkish immigrants socially regress while their parents are already at the bottom of the educational scale, putting them at risk of a longstanding socio-economic exclusion. The downward integration of the sub-Saharan migrants’ children could be interpreted as a strong factor of perceived exclusion in this population and a socio-economic waste for the country.

Figure 2.3. Correlation between the educational attainment of immigrants and their children

Ages 18-35, not in education

Notes: The dotted line indicates the regression coefficient (0.83) between the education levels of immigrants’ children and their two parents. In groups dotted below the regression line, the offspring of immigrants reached an education level that exceeds what could have been expected considering their parents’ level. Groups dotted above the regression lag behind their parents’ education: considering the education level of their parents, the children are expected to be more educated.

Source: Trajectories and Origins Survey (TeO), INED-INSEE, 2008. (a) Figures excerpted from Tables 1 and 4 of Moguérou et al., 2010. (b) Figures excerpted from Table A of Brinbaum and Primon, 2013. Weighted results.

However, upward mobility might not be as pervasive as it seems in Table 2.6. Males and females might in fact experiment with different patterns of educational evolution. Without giving any detail relating to origin, Mainguené (2014) shows for example that upward educational mobility is less frequent among daughters of immigrants than among those of natives (Table 2.7). In any case, for both genders, further research is still needed to analyse the factors explaining upward or downward trajectories at the group and individual level.

From school to work: The segmented access to employment of children of immigrants

Unemployment during the first year following completion of initial studies varies notably by gender and origin. The male disadvantage observed at school translates into much higher probabilities to be unemployed among men than among women during the first year after the end of studies (Figure 2.4). Sons of African immigrants experience dramatic levels of unemployment: 50% of those of sub-Saharan descent were unemployed at this point and the proportions were 43% and 41% respectively among those of Algerian and Moroccan-Tunisian descent. In contrast, despite a low level of education, Turkish men are very quick to find employment, as are men of Portuguese origin. The latter even enter the labour market more quickly than men in the mainstream population. This reflects educational strategies oriented towards professional tracks leading to jobs in ethnic niches and family businesses, e.g. construction and trade (Domingues Dos Santos, 2005).

Figure 2.4. Unemployment rates one and seven years after the end of initial studies

Children of immigrants aged 18-50 who declared they were active, i.e. in employment or searching for employment after they finished their studies, percentages

Source: Source: Trajectories and Origins Survey (TeO), INED-INSEE, 2008. Adapted from Table 8 of Brinbaum, Meurs and Primon, 2015. Weighted results.

Women with Turkish parents stand out as displaying a particularly large unemployment gap to their male counterparts. They are the only ones among all groups to experience higher unemployment than men of similar origin (52% against 26%). Their level of unemployment is more than twice the mainstream women’s one (52% against 23%). Immigrants’ daughters are more frequently unemployed than their mainstream counterparts in all groups except those with Portuguese parents. Again, those of African descent appear to be more vulnerable in the labour market.

Differences by origin in terms of unemployment are not only explained by lower educational outcomes. Indeed, multivariate results show that significant differences persist once education, social origin and demographic characteristics are controlled for (Table 2.8)4. In other terms, this result suggests that the educational premium of the daughters of immigrants paradoxically becomes a disadvantage: while they outperform the mainstream population at school (Table 2.3), they are more likely to be unemployed at the end of their studies (Table 2.8). This result supports the hypothesis that daughters of extra-European immigrants experience, upon entry in the labour market, a double discrimination because of both their origin and their gender. In other words, when they transition from the educational system to the labour market, they seem to lose the positive stereotypes they benefited from in the education system (Frickey and Primon, 2010).

Over time, unemployment levels decrease, although following paths that differ by origin and gender (Figure 2.4). Across groups, the reduction of unemployment is much more significant for men than for women. As a result, in each group, the gender gap tends to fade away and, in some groups, even morph into disadvantage (as in European groups). Seven years after the end of their initial studies, sons of immigrants from Africa (including Maghreb) and Asia (including Turkey) keep levels of unemployment that exceed 20%. Multi-variate analyses show that the penalty experienced by sons of Maghrebi students, when compared to the mainstream population seven years after having completed their studies, is not simply due to their education level, their social origin and other individual characteristics (Table 2.8). Differences between women of the same origin and those of the mainstream population are no longer statistically different at this point in time. This finding could suggest a slow integration into the labour market and/or a tendency to withdraw from the labour market, meaning that seven years after the entry in the labour market only the more successful women remain (Meurs, Pailhé, and Simon, 2006; Meurs and Pailhé, 2010). Unfortunately, so far the literature is not clear enough to disentangle these two possible and somehow contradictory processes. There is however some evidence that daughters of extra-European immigrants are more frequently inactive than mainstream women, even though they are much less likely to be inactive than their parents, e.g. 13% vs. 28% among those of Algerian origin, against 8% in the mainstream population (Brinbaum et al., 2015, p. 209).

Unemployment, salaries and discrimination: A cross-sectional comparison of immigrants and the offspring of immigrants

Comparing immigrants’ children with immigrants strikingly confirms the disadvantage of male natives whose parents originate from outside Europe. Among all immigrant groups with African origins, immigrants’ sons are more frequently unemployed than the immigrants themselves, with a notable peak among those of sub-Saharan origin (Figure 2.5). As immigrants in these figures are not necessarily the parents of the children in question, this comparison cannot be strictly interpreted in terms intergenerational mobility. It does however provide striking insights into the socio-economic position of natives with immigrant parents in France.

Figure 2.5. Unemployment among immigrants and native offspring of immigrants

By Gender, active and inactive individuals aged 18-50 in 2008, percentages

Source: Source: Trajectories and Origins Survey (TeO), INED-INSEE, 2008. Figures excerpted from Tables 1 and 3 of Brinbaum, Meurs et Primon, 2015. Weighted results.

Multivariate analyses show that demographics, family situation and social handicaps (education level, social origin, place of residence, health, etc.) together explain most of the high levels of unemployment. But significant differences by origin persist. After controlling for socio-economic factors, sons of Maghrebi immigrants remain more disadvantaged than the immigrants themselves, even though they were born, educated and socialised in France. Before controlling for socioeconomic characteristics, the probability of being unemployed is eleven points higher among immigrants’ sons than in the male mainstream population (19% vs. 8% in Figure 2.5), while the gap is “only” four and eight points for Moroccan-Tunisian and Algerian immigrants respectively. After controlling for socio-economic factors, the disadvantage of Maghrebi immigrants’ sons diminishes but remains: Their probability of being unemployed is still five points higher than the probability of their counterparts of the mainstream population (vs. four points for immigrants, Table 2.9). In contrast, native-born daughters of Maghrebi parents do better than immigrants. All in all, men born in France from Maghrebi parents underperform both men who were born in the Maghreb and daughters of the same origin.5 Although these results do not properly reflect intergenerational mobility, they do suggest that there might be reverse trends for males and females, with the former experiencing a socio-economic downgrading and the latter, on the contrary, experiencing an upgrading.

To what extent are the inequalities experienced in the job quest by the children of immigrants from outside Europe due to discrimination? At first sight, the groups who experience the higher penalties in matters of access to employment are also those who are more likely to declare they suffered from unfair treatment in the labour market. The Maghrebi case is especially telling: Male children of immigrants lead in discrimination declarations, with a level that exceeds declarations of their female counterparts and those of both male and female immigrants of the same origin (Figure 2.6). Such a coincidence at the group level is not sufficient to clearly attribute difficulties in the labour market to employers’ discriminatory practices. To explore the relationship between the two phenomena in more detail, Meurs discussed whether the actual probability of being unemployed is correlated – at the individual level – with the experience of discrimination, as reported by the interviewees (Meurs, 2017; Brinbaum, Meurs and Primon, 2015). Her results show that the people who are usually expected to be in employment (i.e. the people who have all characteristics that usually guarantee access to employment) but who are actually unemployed are also those who declare they suffer from unfair treatment in their search of a job. In other words, there is a strong correlation at the individual level between the objective measure of employment inequalities and subjective measures of discrimination.

Figure 2.6. Self-reported discrimination in the labour market

By gender and origin, immigrants, native-born children of immigrants and mainstream population, active and inactive individuals aged 18-50, percentages

Notes: Percentage of people who answered “yes” to at least one of the following questions, for a motive corresponding to one of those set down by the French law prohibiting discrimination: “During the past five years, were you ever unjustly refused employment? Have you ever been unfairly refused promotion? Been laid off unfairly?”

Source: Trajectories and Origins Survey (TeO), INED-INSEE, 2008. Adapted from Table 5 of (Brinbaum, Meurs and Primon, 2015). Weighted results.

Once in employment, do the offspring of immigrants experience a salary penalty when compared to the mainstream population? Is this potential penalty higher or lower than in their parents’ generation? A first look at Figure 2.7 reveals distinct negative gaps in terms of salaries for most groups of immigrants and immigrants’ native offspring compared to the mainstream population. Yet again, results differ markedly by gender and origin. The salary gap for the female offspring of immigrants is generally narrower than for immigrants themselves, except for those of Turkish origin where immigrants’ daughters experience the largest negative gap with close to 20% lower salaries than the mainstream population. On the other hand, among men, no clear pattern emerges from the comparison between immigrants and their offspring. Sons of immigrants from sub-Saharan Africa, Southeast Asia, Turkey, Spain and Italy have smaller salary gaps than immigrants, while those of other origins have higher gaps (Figure 2.7).

Figure 2.7. Differences in hourly salaries by comparison with the mainstream population

Immigrants and the native-born children of immigrants, percentages

Notes: The hourly salary of sons of Algerian immigrant(s) is inferior by 13% to the hourly salary of men in the mainstream population.

Source: Trajectories and Origins Survey (TeO), INED-INSEE, 2008. Reproduced from Figure 3 in (Brinbaum, Meurs, and Primon, 2015). Weighted results.

To what extent are the differences observed between immigrants and the children of immigrants attributable to compositional effects, i.e. to the fact that these populations have different characteristics? Based on the Oaxaca-Blinder decomposition technique, Table 2.10 shows what part of the salary difference between each group and the mainstream population can be attributed to individual characteristics, employment characteristics, or unexplained factors (Meurs, Lhommeau, and Okba, 2015). The effects of these three sets of characteristics are reviewed below, with the focus on the most vulnerable groups, i.e. men and women of Turkish, sub-Saharan and Maghrebi origin. First, in all groups, occupational segregation plays a major role in explaining wage gaps. That is to say, men and women from both groups – immigrants and the children of immigrants – are concentrated in less favourable jobs, a finding in line with results from previous studies (Aeberhardt and Pouget, 2010; Aeberhardt et al., 2010). Second, the effect of personal characteristics (age, educational attainment, place of residence, etc.) is less homogenous across groups. They have virtually no effect on male immigrants, but explain a significant part of the salary gap for other groups (male children of immigrants, female immigrants, and women with immigrant parents). Finally, once compositional effects are taken into account, differences with the mainstream population shrink among women,6 but unexplained wage discrepancies remain significant among men. Other things being equal, immigrants from sub-Saharan Africa, the Maghreb and Turkey still earn around 6% less than the mainstream population. The discrepancy amounts to 3% for the children of immigrants.7

How can this significant yet unexplained salary “penalty” observed for men be interpreted? To what extent can it be attributed to employers’ discriminatory attitudes or unobserved heterogeneity? Following the same method as the one used to study the role of discrimination in access to employment, Meurs (2015) analysed, at the individual level, the relationship between observed salary gaps and self-reported declarations of unfair treatment.8 She first estimated the wage that each individual would receive if they were paid as much as an individual of the mainstream population with the same characteristics. The difference between this theoretical wage and the actual wage of each individual was then used as an independent variable in a logistic regression explaining the declaration of unfair treatment, controlling for all variables used to estimate the theoretical salary. In other words, the analysis allows measuring, other things being equal, the extent to which the difference between the theoretical and actual salaries is explained by the self-reported experience of discrimination. Results show that the odds of declaring unfair treatment increase significantly with the gap between the theoretical and actual wages. This shows that self-reported discrimination is anchored in observed wage inequalities. It also suggests that the unexplained wage penalty can – at least partially – be interpreted in terms of discrimination.

Intergenerational socio-economic mobility

Although previous results compare immigrants and the children of immigrants, they do not properly address the question whether sons and daughters of immigrants experienced a downward or upward socio-economic mobility compared with their own parents. As already discussed in the section on education, that type of comparison is not easy to carry out. At least two methodological limitations need to be mentioned concerning social mobility. The first is that the socio-economic statuses of two successive generations (the children and their parents) are not really comparable because of the changes in the labour market structures. Typically, being a manual worker around 2010 obviously does not have the same meaning, in terms of social position, as it did in the early 1960s. The second limitation relates to the participation of women in the labour market. Among immigrant mothers, especially those whose origins are outside Europe, inactivity was and is still frequent,9 which hinders the possibility of comparing the socio-economic status of daughters with their immigrant mothers. Three approaches have been used to address these limitations in measuring social mobility.

Mainguené’s approach (2014) consisted in limiting the analyses to immigrants’ daughters whose mothers were active. As this approach significantly reduces the population size, results cannot be disaggregated by origin. They do however offer a first insight into the social mobility of the daughters of immigrants. The diagonal boxes in Table 2.11 show the proportion of daughters who maintained their mothers’ social position. Above the diagonals are those who experienced a downward mobility; below those who experienced an upward mobility. The matrix reveals mixed results. On one hand, remaining in the low qualification level category is more common among the daughters of immigrants than among the daughters of natives (17% vs. 12%). On the other hand, upward mobility happens more frequently among immigrants’ daughters (a total of 46% vs. 37%), although the type of mobility varies by origin. While moving to the highly qualified category from the medium category is less frequent among the daughters of immigrants (7% vs. 10%), moving from low to medium happens more often among immigrants’ daughters than among daughters of native French (35% vs. 23%). This probably reflects the fact that immigrant mothers are more concentrated in the lower qualified group, making it more likely for their daughters to move out from this category. Moving upwardly is less likely when the starting point is already a higher position, which is more common among natives born to natives. In any case, the major drawback of this approach comparing daughters and their mothers is that it says nothing about the social mobility of women who had an inactive mother. And, as inactivity is quite common in groups originating from outside Europe, these aggregated results actually reveal very little about the social mobility of daughters of extra-European immigrants.

An alternative approach was adopted by Okba (2012): as a reference point to measure social mobility, he used the fathers’ socio-economic status for both immigrants’ sons and daughters. This choice allowed him, to a certain extent, to disaggregate results by origin (Figure 2.8). The results both confirm and expand what had been observed for mothers and daughters by Maiguené (2014). First, upward mobility is generally higher among children of immigrants than in the mainstream population; that trend is observed for both women and men, whatever the origin. Again, this might be explained by the lower position of immigrant fathers on the social ladder: starting from the bottom, it is easier to move upward. Second, the immigrants’ children’s social mobility needs to be tempered, as it is largely due to changes within the employee/worker category. Social moves to higher socio-professional categories are proportionally rarer than in the mainstream population. In other words, natives with immigrant parents do progress on the social ladder, but climb fewer steps than natives born to natives. Third, looking at gender differences, the relative success of Maghrebi immigrants’ daughters is especially striking: the proportion experiencing upward mobility is higher than those of all other women groups and that of men of the same origin. Conversely, results point to the social disadvantage of sons of Maghrebi immigrants, who have lower outcomes than their sisters and European counterparts.

Okba analysed the factors explaining upward mobility. He showed that the fathers’ country of origin has virtually no effect on the probability of upward movement, as such a social move is essentially determined by socio-demographic variables. The probability of moving upward increases with the father’s socio-economic status, the mother’s activity status, and the person’s own education level and age. They are reduced for individuals who experienced an interruption in their career and for women. As interactions were not tested between origin and other variables, it is not known to what extent the factors facilitating or hindering upward mobility vary by origin among the children of immigrants.

Figure 2.8. Socio-professional upward mobility of native-born children of immigrants

By gender and father’s origin, ages 35-50, percentages

Notes: - The figure includes only individuals in salaried employment or with previous work experience at the time of the survey. - Socio-professional mobility is measured by comparing the fathers' socio-professional category when their children were 15 years old with the socio-professional category of their children in their most recent job. Changes measured used the following categories: unqualified employee/worker, qualified employee/worker, intermediate profession, managerial and professional occupation. - Southern Europe includes Portugal, Spain and Italy.

Interpretation: In their last employment, 36% of the children of immigrants had experienced upward social mobility, i.e. they were employed in a more qualified job than their fathers. 12% experienced upward mobility within the employee/worker category (from an unqualified to a qualified job), while 24% had moved upwardly from and to any of the other categories (e.g. from the unqualified employee/worker category to intermediate category, or from this latter category to a managerial occupation, etc.).

Source: Trajectories and Origins Survey (TeO), INED-INSEE, 2008. Adapted from Table 8 of Meurs, Primon and Okba, 2015. Weighted results.

The two approaches (Mainguené’s and Okba’s) described above have two limitations in common. The first is that they measure social mobility through changes in overly large categories. More precise measures of the socio-economic status, such as the International Socio-Economic Index (ISEI), could temper the conclusions. The second is that the intergenerational mobility is measured without taking into account the fact that parents and children do not share the same labour market or socio-economic contexts. Taking into account their relative position could also modify the results, as was observed above for education. (Attias-Donfut and Wolff, 2009) explored another approach, consisting in analysing subjective answers given by parents about their children’s social mobility in a survey exclusively dedicated to immigrants aged 45-70 (Table 2.12). The pervasive view of immigrants in 2003 was that their children experienced social ascension by comparison with themselves (59% for sons and 61% for daughters), a view that was shared by only a third of the native population ten years earlier, according to a previous survey that included this population (Attias-Donfut and Wolff, 2009, 264). This view depends on parents’ geographic and social origin. By definition, the children who have parents with the highest positions cannot experience upward mobility. This is probably why, for instance, only a third of the sons of Northern European origin were considered as having socially progressed, with another third being viewed as occupying a similar social position. Conversely, immigrants with lower social positions have a more optimistic view of their children’s social mobility. Social mobility again appears as a gendered phenomenon. Except among those of Turkish origin, immigrants’ daughters are less frequently involved in a downward social trajectory than sons. This disadvantage men have in relation to women is especially salient among those originating from the Maghreb and sub-Saharan Africa. Importantly, the same handicap persists in multivariate analyses that control for personal and parents characteristics (Attias-Donfut and Wolff, 2009, 268).