The 1970s witnessed disruptions both in economic conditions and French migration policy. Do we find any trace of the economic crisis and border closures in immigrant employment trajectories? The authors address this question by reconstructing these trajectories using data from the Trajectories and Origins survey. This original approach reveals the diversity of male and female pathways and shows how they are related not only to individual characteristics but also to the sociohistorical context.
1 In France, the employment trajectories of immigrants and their descendants are characterized by certain distinctive features (Beauchemin et al., 2015a). Compared with natives, their working careers are more frequently insecure, and they are more likely to occupy the status of manual worker (ouvrier). From a diachronic, spatial, and social point of view, immigrants from certain countries (particularly European countries) frequently occupy skilled and stable jobs; women often spend varying lengths of time as homemakers, which influences their professional trajectory; and individuals who arrived before the mid-1970s are more likely to have a stable job than those who immigrated more recently. Employment trajectories are connected to the historical contexts of arrival and to changes in general working conditions. They also differ according to immigrants’ social characteristics. This article focuses on these characteristics, using the Trajectories and Origins (Trajectoires et origines [TeO]) survey as a basis from which to examine how immigrants’ employment trajectories change and why they vary.
2 These questions have been raised in various types of research. Historical studies have analysed the relationship between migration policies and the composition of immigration flows, showing that the latter contribute to the production and reproduction of the working class (Noiriel, 1988). The occupational and social diversity of immigrants reflects the different ways successive waves of arrivals have populated regions and industry sectors (Wihtol de Wenden, 2016). Other studies of the contemporary period have relied on quantitative analysis of official statistical surveys: the Geographical Mobility and Social Integration survey (Mobilité géographique et insertion sociale), conducted by INED in 1992 (Tribalat, 1995); the Family History study (Étude de l’histoire familiale), conducted by INSEE in 1999; the Life History survey (Histoire de vie), conducted by INSEE in 2003 (Économie et statistique, 2006); the Génération surveys conducted since 1998 by the French Centre for Research on Education, Training, and Employment (Centre d’études et de recherches sur les qualifications); and the TeO survey, conducted by INSEE and INED in 2008 (Beauchemin et al., 2015a). These works have focused mainly on comparing immigrants, natives, and the descendants of immigrants rather than on the differences within the immigrant population. Regarding employment trajectories, these studies are generally based on econometric methods and the analysis of dichotomous states: working or not working, having a stable job or not (Glaude et al., 1996), employment or unemployment (Meurs et al., 2006), having a job or no job (Lacroix et al., 2017), promotion or lack of promotion (Tavan, 2006). The shape of trajectories, i.e. the succession of various states over time, is only rarely taken into consideration (Tucci et al., 2013) or done so qualitatively (Lendaro, 2013).
3 We extend this line of research by constructing a typology of the employment trajectories of immigrants arriving in metropolitan France between 1968 and 1988. The structure of these trajectories will be used to establish types of relationships to salaried employment and unemployment and to relate them to social characteristics (migration trajectories, social and geographical origins). This typology enables us to look beyond the dichotomous criteria for describing trajectories and to understand a broader set of employment states. Explaining how migrant workers’ trajectories are constructed and modified calls for a sociohistorical approach that can relate them, first, to alterations in the procedures for controlling immigration flows and to labour market changes and, secondly, to changes in migrants’ social characteristics. Since early 1970s, the composition of migration flows and immigrant workers’ conditions of entry and residence have been profoundly altered because of several developments: a tightening of the rules governing the entry and residence of workers and their families; the introduction of policies to encourage return and limit family reunification; an increase in unemployment and insecure forms of employment; feminization; and rejuvenation. These processes have had an impact on the occupational and social integration of immigrants in France.
4 This article has an assumed descriptive dimension. It uses the TeO survey—the most recent survey available on aspects of immigrants’ social life (work, marital status, migration trajectory)—to establish the various types of occupational and social integration of people arriving in France in the 1970s and 1980s, a period of major economic and social transformation and of profound change in the migration system.
5 In carefully considering the information provided by the TeO survey, this study will first describe the most significant characteristics of the employment states and trajectories of the immigrants within the sample. Secondly, it will analyse the various types of career pathways, anchoring each within the history of immigration in France, relating them to the characteristics of the migrants who experienced them, and modelling the transitions they reflect.
I – Quantifying employment trajectories using the TeO
1 – The challenges of a retrospective survey
6 The political and scientific expectations of a survey affect the conditions of its use and the types of sociological arguments that can be drawn from it. Using the TeO survey to explore the diverse employment trajectories of the immigrant population is not initially straightforward.  Without reviewing the controversy that arose when the survey was designed and its first results published,  we should note that the TeO survey was commissioned by the public authorities in 2003, on the occasion of the 20th anniversary of the March for Equality and Against Racism. It continues a tradition of major surveys on immigration, carried out since the late 1980s (the Geographical Mobility and Social Integration survey in 1992; the Life History survey in 2003), driven by the political and scientific interest in measuring foreign and immigrant presence in France, and in measuring discrimination (Spire, 1999a). The TeO survey reflects its designers’ desire to measure the discrimination against immigrants and their descendants, as well as the ways such discrimination is subjectively experienced. Questions relating to employment and work, particularly those aimed at a retrospective reconstruction of the career trajectory, constitute one of the themes addressed in the survey. From a life course perspective, the TeO survey provides us with occupational status before migration, at the time of arrival in France, and in 2008, when the survey was conducted, as well as annual information on the employment status of individuals (see below).
7 Another feature of the survey illustrates the difficulties potentially encountered when studying past phenomena through a retrospective survey. The subsample of immigrants includes 8,259 individuals aged 60 or under in 2008, who arrived in France between 1948 and 2007. Due to the way the survey is constructed, labour immigration in the 1950s and 1960s is under-represented in relation to its historical weight. The respondents who arrived during this period were at that point very young children. Moreover, the TeO survey does not provide information about certain specific types of workers; those living in communal facilities, including in migrant workers’ homes, were not surveyed. Finally, the requirements of the statistical survey mean that because only immigrants present in metropolitan France in 2008 were surveyed, we know nothing about those who left France after having spent some time there. Moreover, propensity to leave varies greatly by nationality and historical period. As such, there is no reason to believe that this variation would not have impacted the demographic and occupational structure of the immigrant population present in France at the time of the survey (Thave, 1999; De Coulon and Wolff, 2005; Brutel, 2015; Solignac, 2016), even if, strictly speaking, ‘such a hypothesis cannot be verified using the statistical sources available in France’ (Beauchemin et al., 2015b, p. 78). The population studied by the TeO survey is therefore the product of the political and statistical construction of the sample, and of the historical sedimentation of entry, settlement, and exit processes involving immigrant populations in France.
8 To analyse the employment trajectories of immigrants present in mainland France in 2008, the study population was limited to the 3,604 individuals who arrived between 1968 and 1988 and were aged 20–60 at the time of the survey. Several constraints informed this decision. First, sample design constraints (see above) meant that immigrants entering in the 1950s and 1960s were at that point no more than 10 years old, making them unrepresentative of the reality of migration flows during this period (Beauchemin et al., 2015a). As such, within the total sample population, 50% of individuals entered France before the age of 22. However, this distribution is not consistent over time. The average age at entry into France and the dispersion of ages increases gradually with year of arrival. For this reason, we needed to exclude individuals who arrived early (and as such at a young age, due to the sample design) and decided to use only those entering France after 1968. Secondly, the sampling procedure results in the right censoring of the event history data: at the time of the survey, in 2008, some individuals had reached the end of their working lives or were approaching it, whereas for others who entered France more recently or at a very young age, the TeO survey provided only a few years of observation. We therefore considered the same number of years for all individuals, in this case the first 21 years of their presence in France. The study thus excludes individuals who arrived after 1988. 
2 – Dynamics of employment status
9 When the survey was conducted, retrospective information was collected concerning employment trajectories. Beginning from their date of arrival in France or the completion of their studies (if later), each respondent had to fill in an annual timeline with his or her situation, using seven pre-coded states: ‘homemaker’, ‘salaried employee’, ‘unemployed’, ‘self-employed’, ‘in education’, ‘variable’, and ‘other’. The questionnaire specified that ‘salaried employee’ covers apprenticeships and paid internships, as well as salaried employment. ‘Selfemployed’ includes those working for themselves or assisting another person with their work. ‘Unemployed’ individuals may or may not be registered with the French Employment Agency.  ‘In education’ includes school or university students and individuals on unpaid internships. ‘Other’ is vague, as it includes national service, retirement, disability, and other unspecified situations. This state will not be taken into account in the analysis. The seventh state, ‘variable’, coded after the survey, is attributed to years in which the individual experienced several different states, including alternating periods of employment and unemployment. This information allows us to outline the employment trajectories of individuals arriving between 1968 and 1988, and to assess the relationship between trajectories, changing entry and residence conditions, and immigrants’ social characteristics.
10 Looking at the succession of these states over the first 21 years of presence in France, one is struck by the great diversity of trajectories. The 10 most common trajectories in the sample represent only 34% of individuals. However, three states clearly dominate: ‘salaried employee’ (average of 10 years out of 21), ‘in education’ (6 years), and ‘homemaker’ (3 years). Salaried employment thus remains a dominant experience in the social life of immigrants, with three-quarters of trajectories including at least 1 year of salaried employment and nearly half over 10 years. The most common trajectory (10%) consists of 21 years of continuous salaried employment.
11 However, these findings must be differentiated according to sex and period of arrival. The TeO survey population, representative of the total immigrant population, became increasingly female over the period. This feminization had an impact on the characteristics of workers, who were increasingly female workers. While the most marked difference between men and women relates to the ‘homemaker’ state (not recorded for any men), salaried employment is still, for both women and men, the majority state after about 10 years of presence in France (even though, for women, the start of this employed period is deferred for several years). For women experiencing varying lengths of time as homemakers, the relationship with employment varies and requires special attention. Different social, professional, and family logics no doubt apply depending on whether the individual is at the beginning or end of their employment trajectory.
12 In addition, processes for entering employment and pursuing a salaried career are deferred and change over time. The TeO survey, due to its sample design constraints in terms of time and age of arrival, allows only partial observation of employment trajectories before the 1970s. Nevertheless, we see distinctions based on period of entry into France, particularly around the time of the suspension of labour immigration in July 1974. The proportion of individuals holding a salaried job upon arrival falls after 1974 (from 35% for 1968–1974 to around 20% for subsequent periods). However, entry into the labour market is only deferred. Irrespective of the time period, more than half of immigrants were salaried employees or self-employed after living in France for about 10 years. While the former state remained predominant, the 1980s–1990s saw a rise in self-employment, the determinants of which should also be studied.
13 The ‘end’ of labour immigration therefore did not signify the end of immigrants in work. The transformation of migration policies in the 1970s, while contributing to the rejuvenation and feminization of migrant flows—which doubtless explain the subsequent delays in labour market integration—is also associated with a change in the requirements for legal residence in France. Thus, in the survey sample, the fall in the number of individuals in paid employment upon arrival must be viewed in relation to changes in the statistics on the allocation of residence permits (Figure 1), which show a clear reduction in legal access to France (and the French labour market) granted for employment purposes. 
14 Finally, although the change is not on a large scale, we should note the increase over time in the proportion of unemployed individuals, particularly at the beginning of their careers.  Unemployment remains uncommon within the study population, but it seems to disrupt salaried employment trajectories (see below). This finding is corroborated by the fact that the proportion of individuals experiencing at least 1 year of unemployment (out of the first 21 years of presence in France) increases with more recent entry into the country. 
15 This initial description of the employment trajectories of immigrants who arrived in France between 1968 and 1988, and under age 60 in 2008, allows us to put the 1974 turning point into perspective (Laurens, 2008). It also allows us to highlight significant shifts over the period, despite the structural nature of work being the organizing principle of the employment trajectories: a rise in self-employment, the feminization of the population resulting in homemaker and/or in salaried employment trajectories, the increased difficulty in obtaining stable salaried employment after 1974, the structural dimension (related to the rejuvenation of migrant flows) of the transition from education to salaried employment for immigrants, the issue of unemployment, and the penetration of insecure forms of employment among immigrants.
Type of residence permit by year of arrival
Type of residence permit by year of arrivalCoverage: Immigrants arriving between 1968 and 1988.
16 Understanding employment trajectories involves taking both historical factors and individual characteristics into account. How do these types of trajectories relate to distinct sociodemographic and migration profiles? How can we explain the most frequent transitions between education, salaried or self-employment, unemployment, and the status of homemaker (not working)? What information do these types of trajectories and transitions provide about changes in migrant flows and the conditions of entry and residence of immigrants in France?
II – Social determinants of employment trajectories
17 To synthesize the complexity and variability of these employment trajectories, the optimal matching (OM) method was chosen, followed, as is customary in this type of analysis, by hierarchical ascendant classification,  enabling us to distinguish, initially, six main clusters or classes of employment trajectories among immigrants arriving in France between 1968 and 1988 (Figure 2).  We then related these trajectories to the sociodemographic properties of the individuals and to certain characteristics of their professional and migration trajectories (see Appendix Tables). Next, the probabilities of transition from one state to another were calculated for certain classes within which these transitions were the most frequent (see Box).
1 – Students and young workers
18 Classes 1 and 4 (1,575 individuals, i.e. more than 40% of the sample) comprise immigrants who have had several years of education before making the transition to short periods of employment. Individuals in these two classes are more highly qualified (in each of these classes, at least 32% have a post– upper secondary qualification, compared with 22% of the full sample) and occupy relatively more highly qualified first jobs: higher level occupation for 11% of Class 1 and 13% of Class 4, compared with 7% of the full sample (they are also over-represented among technicians and supervisors). Unsurprisingly, they arrived in France at a younger age, especially those in Class 1, who immigrated on average at age 6, and 80% of whom immigrated before the age of 10. They are much less likely to be children of farmers, although they do not have higher-than-average social backgrounds, the majority being the children of manual workers. Although they mostly arrived in the 1970s, they differ even more markedly in terms of their residence permits. A larger proportion immigrated for family reunification (Class 1) or for study purposes (Class 4). Lastly, by 2008, more than two-thirds of them had French nationality, whereas this was the case for only 54% of the full sample.
19 These two classes undoubtedly combine heterogeneous migration motives and trajectories, i.e. the migration of individuals from middle or upper social classes for the completion of lengthy studies (in medicine, at the doctoral level, or at elite institutions called grandes écoles) and the school careers, potentially marked by repetitions of years,  of working-class immigrants who arrived very young. As such, matching solely based on the trajectory criterion, involving the transition from education to employment, masks a heterogeneity already well documented in other works. Classes 1 and 4 correspond to the second and third types of immigrants distinguished by Moguérou et al. (2015), i.e. those ‘arriving as children or teenagers’ and those arriving ‘for study’. The same significant differences can be observed for first residence permit obtained, country of origin, and date of arrival.
Chronograms of the six classes produced using OM
Chronograms of the six classes produced using OMNotes: The chronograms do not represent individual trajectories and therefore cannot be read horizontally. They show, for each year, the proportion of individuals in each activity state. Individuals in Class 1 are all in education at the beginning of their careers and then gradually enter salaried employment. At the end of the period, 60% of them are in the salaried employee state. At every point in the trajectory, at least 70% of Class 2 individuals are homemakers.
Coverage: Immigrants who arrived between 1968 and 1988.
Box. Modelling the transitions
- Education ➡ Unemployment versus Education ➡ Employment (salaried employee or self-employed). Field: Classes 1 and 4.*
- Employment ➡ Unemployment versus Employment ➡ Employment. Field: Classes 1, 3A, 3B, 4, and 5.
- Homemaker ➡ Employment versus Homemaker ➡ Homemaker. Field: Classes 1, 2A, 2B, 2C, 2D, 3A, 3C, and 4.
- Employment ➡ Homemaker versus Employment ➡ Employment. Field: Classes 1, 2A, 2B, 2C, 2D, 3A, 3C, and 4.
* Interruption followed by resumption of education remains uncommon. Around 85% of transitions out of education involve individuals from Classes 1 and 4 leaving formal education.
20 As highlighted by the large numbers of these classes, the rejuvenation of migration flows over the period has resulted in many young immigrants entering the French education system and thus contributing to the diversification of the social statuses occupied by immigrants, who are no longer immediately or solely workers. However, many students maintain a link with the world of work. The transition from the in-education state either to salaried employee or self-employed, or to unemployment or a variable status, is the second distinctive feature of Classes 1 and 4. The modelling of this transition (see Box and Table 1, Model 1) shows that, from the 1980s onwards, these students were affected by the employment crisis (Merckling, 1998), as were natives (Peugny, 2007) and descendants of immigrants (Brinbaum et al., 2015a). Indeed, the model indicates that the probability of transitioning from education to unemployment is lower, all else being equal, before the late 1970s. Those who completed their studies before the 1980s had a lower probability of experiencing unemployment (rather than employment) upon leaving education, compared with those who completed their studies in subsequent decades. In other words, the use of logistic modelling and consideration of year of leaving education reveal that difficulties in finding work were due partly to the composition of this immigrant student population (the significance, for example, of the geographical origin variable) but also to the specific year they left education and, therefore, to the economic context. 
2 – Figures on immigrant workers
Predominance of labour immigration
21 Classes 3 and 5 are particularly characterized by long periods of salaried employment (18.3 years on average vs. 10 years for the full sample) or self-employment (12 years on average vs. less than 1 year for the full sample) and a clear over-representation of men (64% of men in both classes combined vs. 48% for the full sample). Individuals in both classes share the feature of having migrated with a strong focus on work. They are more likely than others to have arrived in France with an employment contract (13% vs. 5%) or a work permit (41% vs. 20%).  They also worked more before migration (56% vs. 28%), having arrived in France at an older age. Individuals in these two classes are thus essentially characterized by their labour, recalling the archetypal immigrant worker extensively analysed by sociology (Morice and Potot, 2010) and whose sociohistorical reality is reflected here, in these two classes alone, by the sheer numbers: 1,358 individuals, i.e. over one-third of the sample (38%).
Apparent homogeneity covering distinctive profiles
22 However, this predominance should not mask certain distinctions within the immigrant worker population. Class 3 trajectories are of two types. Most individuals experience a continuous trajectory of employment, but some experience other states (homemaker, in education, unemployed). To develop this distinction and specify the types within this labour-driven class of immigrants, further OM of the employment trajectories of individuals in Class 3 only was performed, distinguishing four subprofiles (Classes 3A, 3B, 3C, and 3D; see Figure 3).
23 Most individuals in Class 3A have continuous employment trajectories (60% of them experience no other employment status and spend an average of 20 years in it) that begin as soon as they arrive in France. Much smaller than Class 3A (their combined numbers equate to only a quarter of the Class 3A population), Classes 3B and 3C also have long periods of employment (12 and 15 years, on average) but have different trajectory profiles. Class 3B includes the most convoluted trajectories, comprising more transitions and longer average times in states other than salaried employment, with a mean unemployment duration of 2 years.  Class 3C is almost exclusively composed of women, who begin by being non-working homemakers for a few years (6 years on average) before moving on to employment, which is consistent with the trajectories of the female homemakers analysed below. Finally, Class 3D, which includes trajectories with long-term unemployment, is too small (n = 27) to be interpreted, showing that long-lasting unemployment trajectories are infrequent among the TeO survey’s immigrant population.
24 These types of salaried-worker trajectories relate to differences in individual characteristics, migration trajectories, and relationships with employment. In particular, the distinction between continuous and ‘disrupted’ employment correlates with period of entry, as well as with social and migration origins. Individuals in Class 3A are more frequently children of farmers and worked before emigrating at a relatively older age, and before 1974. They are also more likely to have arrived in France with a work permit and an employment contract, and to have held a first job as a manual worker. Nearly a quarter come from Southern European countries (Portugal, Spain, and Italy) and 50% were, in 2008, in a relationship with a spouse they had met abroad. These are the characteristics of labour immigration, as described in particular by Sayad (1999) in relation to Algerian migrants.  Those in Class 3B, although overwhelmingly male manual workers, as in Class 3A, were more likely to have arrived after 1974 and are less likely to have a worker’s residence permit than a refugee card or French spouse visa. 
Chronograms of the four classes produced by the subdivision of Class 3 of employed immigrants
Chronograms of the four classes produced by the subdivision of Class 3 of employed immigrantsInterpretation: Individuals in Class 3A are all or almost all in employment from the beginning to the end of the trajectory. On the other hand, 30%–40% of Class 3D experienced, at different points in the trajectory, unemployment or a variable state.
Coverage: Immigrants who arrived between 1968 and 1988.
25 This overlapped with the policy changes of the early 1970s, which established different rules for legal residence and impacted the course of trajectories. Looking at the individuals in Class 3A, more than a third of whom arrived before 1974, for over 45% of them their first residence permit was a worker’s one (which was the case for only 35% of the individuals in Class 3B). The main distinction compared with Class 3B is that Class 3B trajectories are over- represented after the temporary suspension of labour immigration in 1974 and the general deterioration of economic conditions during the same period. Class 3B individuals in ‘disrupted’ employment thus experience, on average, 1.8 years of unemployment (compared with 0.4 years in the sample as a whole and 0.2 years in Class 1). This is also a characteristic feature of Class 3D, which includes the (admittedly rare) trajectories of long-term unemployment. Individuals belonging to this class are markedly over-represented among immigrants arriving in the 1980s. The second model estimated supports this idea by showing a statistically significant period effect from the late 1990s: all else being equal, transitions from employment to unemployment are 1.5 times more likely in the years 1999–2008 compared with 1982–1992 and earlier (see Table 1, Model 2).
Modelling of various employment state transitions
Modelling of various employment state transitions† The data to which we had access did not include country of birth. Only an individual’s first country of residence is known.
Interpretation: The coefficients correspond to odds ratios all else being equal. As such, according to Model 3, individuals experiencing the ‘end of a partnership’ were 4.5 times more likely to experience the homemaker– employment transition than to remain a homemaker. Union formation and the end of a partnership before transition refer to transition year n (year in which the change of state began) or year n − 1 (other codings were tested and yielded the same results).
Notes: In each model, age at transition and social origin (parents’ qualifications, father’s occupational code) were introduced as controls. As their coefficients were not significant, they are not reproduced here. (*): No coefficient estimated due to small numbers for the missing value for level of education (n = 3) and the period 1968–1974 (n = 0) in our sample for the study–unemployment transition.
* p < 10%, ** p < 5%, *** p < 1%.
3 – Entrepreneurship
26 Immigrant workers vary not only by type of employment trajectory. There is a striking difference between individuals in Classes 3 and 5, the latter spending periods of various lengths in self-employment. However, this profile remains uncommon. Whereas Class 3 contains 1,207 individuals, Class 5, the self-employed class, contains only 151. The state of self-employment often occurs at a particular point in a professional career: only eight individuals in Class 5 were self-employed throughout the 21 years of their trajectory, the average time being 12 years.
27 Social backgrounds in Class 5 are higher than those of individuals in the other classes, which supports the idea that becoming and remaining self-employed for several years requires ‘social support’ (Zalio, 2009): 44% had a father who was a craftsman (artisan), business owner, or in a higher level occupation (cadre), compared to 23% of the full sample; 29% had a father with at least an upper secondary qualification, compared to 15% of the full sample (similar findings apply to their mothers’ qualifications and socio-occupational categories). In addition, those who had a job before emigrating (about half of them) were more likely to have worked in categories other than farmer, manual worker, or non-management employee (22% in Class 5 vs. 11% in Class 3).
28 These distinctions probably reflect effects related to date of arrival in France, country of origin, and social background. These self-employed individuals tended to arrive after 1974 and particularly in the 1980s, reflecting both changes in the rules for obtaining the foreign trader’s card (abolished from 1984, following a period during which it was issued more readily) and the labour market crisis that characterized that period. Twenty-seven percent are from Asia (compared with 19% for the whole sample) or from Turkey and other European countries. In addition, their trajectories are more erratic than those of Class 3 and even the sample as a whole, as indicated by their higher number of transitions from one state to another. No doubt the transition to self-employment automatically adds at least one transition to the trajectory. But we might question the extent to which this greater variability of states reflects greater instability of trajectories, such as forced transitions to self-employment due to the lack of salaried employment.
29 Transitions to long-term self-employment usually occurred after a period of education or salaried employment. Because only small numbers of individuals in the TeO study experienced this state, it is difficult to subdivide this class and characterize the distinctions obtained. Several competing hypotheses remain open. Some salaried employees become self-employed because they have a starting capital from inheritance or savings; because they leave salaried employment (e.g. following redundancy); or because they have acquired enough experience in a certain field to work for themselves. However, these hypotheses cannot be tested due to the lack of sufficiently detailed information in the TeO survey on the employment trajectories of the self-employed.
4 – Women between home and employment
30 The TeO survey corroborates the ‘trend towards the feminization of the immigrant population’ (Beauchemin et al., 2015b, p. 68). As such, labour immigration is declining in favour of family reunification, usually but not exclusively by women. 
31 Trajectory analysis reveals one exclusively female class, reflecting the state of homemaker or housewife (Class 2), although women are also present in the other classes, particularly those reflecting trajectories in education (Classes 1 and 4) and salaried employment (Class 3 and more specifically subclass 3C).  Most of the women in Class 2 arrived in France after 1974 (fewer than 20% of them arrived before 1974 vs. 28% for the entire sample) with a residence permit for family reunification purposes or as the spouse of a French national. If we proceed as we did for the salaried employment class, we can distinguish four trajectory subtypes for immigrant women (Figure 4).
High diversity in homemaker trajectories
32 It is principally the relationship with salaried employment that distinguishes these subclasses. The most common profile (n = 271) is that of women in Class 2B who remain homemakers throughout their first 21 years of presence in France (20 years as a homemaker, on average), whereas all others have longer periods of salaried employment. However, women in Classes 2A and 3C  (n = 205) gradually enter the labour market and spend an average of 6 and 15 years, respectively, in salaried employment, while those in Classes 2C and 2D (n = 118) follow the opposite path.
33 As Tavan (2006) notes, these differences in trajectories echo the characteristics of the women in question and their migration trajectories. For example, women entering the labour market (Classes 2A and 3C) arrive younger; are less likely to have met and married their spouses before migrating; are less likely to have come for family reunification purposes (35% vs. 56% for women in Class 2B); and are slightly more likely to have worked in their country of origin (30% vs. 24% in Class 2B, but this comparison is partly truncated by non-responses due to younger arrivals). On the other hand, those who remain homemakers (Class 2B) are frequently of Algerian, Moroccan, Tunisian, or Turkish origin; this is family immigration strongly characterized by the labour immigration of men.
Chronograms of the four classes produced by the subdivision of Class 2
Chronograms of the four classes produced by the subdivision of Class 2Interpretation: Almost all individuals in Class 2A are homemakers at the beginning of the trajectory, while more than 80% of them are in salaried employment at the end of the trajectory.
Coverage: Immigrants who arrived between 1968 and 1988.
Hypotheses on transitions from homemaker to employment
34 How can we explain that some women get a job, especially after several years as a homemaker, while others never do? We might assume that this is due to life-cycle effects (children leaving home) or disruptions in the life course (death of partner, job loss, separation). This is at least one of the explanations given in Lendaro’s (2013) description of the typical trajectories of immigrant women. Although we cannot test the first hypothesis using the TeO survey due to the lack of appropriate variables, the second hypothesis can at least be tested in part. The individuals within the TeO survey were asked to provide, where applicable, the year they stopped living together with their first partner.  By restricting this question to first partnership, ends of other unions are potentially missed, undoubtedly resulting in the underestimation of this phenomenon. Nevertheless, experiencing the end of a union more than quadruples the likelihood of experiencing the homemaker–employment transition. It is therefore a determining factor in explaining these women’s entry into the labour market after several years without salaried employment. Previous employment in the country of origin and this return to employment are positively correlated.
35 Similarly, it is hypothesized that employment–homemaker transitions are partly related to union formation or to the birth of a new child. For each woman, the number of children for each year of her trajectory can be reconstructed from their dates of birth; the underlying assumption is that the birth of a child has a negative effect over time on the homemaker–employment transition and a positive effect on the employment–homemaker transition. However, this is only an estimate because the number of children can only be inferred from the table of dwelling occupants and therefore only takes into account those still living in the parental home. This restriction, which disregards emancipated children and children living in the country of origin, results in an underestimation of parenthood’s effect on entry into employment. This likely explains why neither model can be used as evidence of the effect of the number or age of children at the time of the transition. Several different coding schemes were tried, and none was significant (hence the absence of these coefficients in the results table).
36 On the other hand, first union formation is highly correlated with an employment–homemaker transition, more than doubling its odds ratio (Table 1, Model 4). Withdrawal from working life after entering a union has a significant impact on the occupational trajectory of immigrant women. Working before or after arrival in France increases an individual’s likelihood of re-entering the labour market in France and decreases their likelihood of leaving it to become a homemaker.
37 Lastly, the level of education attained does not produce significant effects in the regression models, apart from a positive effect of lack of qualification on the employment–homemaker transition. Just as for women as a whole, women with no qualification are more likely to experience this transition (rather than remaining in salaried employment) compared to women with a primary school certificate (certificat d’études primaires) or vocational qualification (brevet d’études professionnelles), all else being equal in the model.
38 Analysing employment trajectories is a way to understand changes in the conditions of immigrants’ entry, residence, and employment in France. Historical approaches have shown the value of seeking an understanding of both the evolution of trajectories and changes in the worlds of production (Bruno, 2010). More modestly, and without being able to link trajectory characteristics with those of employment spaces in a dynamic way, this study has shown that certain transitions (employment–unemployment or deferred entry into salaried employment) are more common after the crisis of the 1970s and the closure of French borders to labour immigration in 1974. Employment trajectories are shaped by the system of legal and administrative restrictions imposed by the State apparatus to control the flow of arrivals (requirements for obtaining the various residence permits), but also the state of labour market opportunities (greater or lesser job stability, levels of unemployment) and the characteristics of the migrants (age, sex, level of qualification, emigration contexts). For women especially, employment trajectories are highly heterogeneous due partly to the combination of late entry into the labour market, marriages ending, and the effect of restricted employment opportunities. For young people in education, we also show that experiencing a transition to stable salaried employment depends not only on the characteristics of their pre-immigration social environment but also on this transition’s timing.
39 Methodologically, event history information can be looked at through two prisms of temporal analysis: a structural prism focused on both the trajectory as a whole and on similarities between trajectories (OM) and an event prism focused on the determinants of the probabilities of transitioning from one state to another (logistic regressions). The TeO survey provides a rich source of information for the sociohistorical analysis of immigration in France and for trying to explain the social determinants of a particular type of employment trajectory. However, as we sought to understand the employment trajectories we identified, we were confronted with certain limitations due to the challenges of determining past occupational trajectories from within a large statistical survey of the present. First, the annual scale of the information captured in the TeO survey’s retrospective timelines, while responding to the difficulties of reliably collecting complex information about past events, does not allow a fine distinction to be made between disruptions in the life course and rejects the most unstable periods as a ‘variable’ state. Secondly, it is impossible, from the TeO survey, to distinguish salaried employment situations over time according to the nature and qualification of the job (sector, type of employment contract, level of qualification). Thirdly, the retrospective collection of life course history data overlooks certain characteristics necessary to understand the transitions in detail: occupation, number of children, marital status, type of residence and work permit, and specific time of potential naturalization. The historical analysis of the dynamics of individual trajectories and production areas (sectors, size of companies, and composition of the workforce) must therefore be continued, no doubt by cross-referencing and diversifying the sources and locations of observation. 
Statistical description of the trajectory classes produced using OM
The technical characteristics of the survey and the associated documentation (sample design, questionnaire, dictionary of variables), are available on the dedicated website: http://teo.site.ined.fr.
See the debate in Revue française de sociologie, coordinated by Felouzis (2008).
Since this study concerns a non-representative subsample of individuals arriving in France between 1968 and 1988, the weighting designed to align the TeO sample with the 2008 EAR (annual census survey) was not used in the statistical processing. Our aim is not to infer reference population characteristics from the TeO survey, but to describe patterns and correlations within a population formed from within the TeO sample. Weighted statistical analyses were used to test the robustness of the results. This was done for optimal matching, the descriptive statistics used to characterize the classes produced using hierarchical ascendant classification, and the transition models. This did not affect the overall significance of the correlations observed.
Agence nationale pour l’emploi (ANPE), which became Pôle emploi when it merged with the Association pour l’emploi dans l’industrie et le commerce (Assédic), the Association for Employment in Industry and Commerce, in December 2008.
The 1970s and 1980s saw changes to the requirements for legal entry into the country: a decrease in the proportion of residence permits granted for work and an increase in those for asylum (until 1980, see Spire, 1999b), family reunification, reunification with French spouses, and education (Labat, 1993).
The probability of experiencing at least 1 year of unemployment is 15.2% for the period as a whole. It is only 11.1% for people who arrived between 1968 and 1974 and increases over time, peaking at 18.6% for people who arrived between 1984 and 1988. The probability of experiencing at least 1 year of unemployment during the first 5 years of presence in France also increases over time. It is 14.2% for individuals who arrived before 1974, 22.3% for those who arrived between 1974 and 1981, 20.4% for those who arrived between 1981 and 1984, and 29.7% for those who arrived between 1984 and 1988.
This observation is also consistent with the trend in unemployment among foreigners in France, which has been rising almost twice as fast as that among workers of French nationality since the mid-1970s and remains higher.
See Abbott and Tsay (2000) for a summary of the application of this method in sociology. The metric used to calculate the distances between the 3,604 individuals in the sample consists of setting the cost of insertion/deletion operations at a standard level of 1 and calculating substitution costs from the matrix of transition rates between states, actually observed in the sample. By doing so, the distance calculation algorithm favours substitution operations, best enabling the historical succession of states to be respected. For a discussion of methods of setting costs and their effects, see in particular Robette (2012).
Two other optimal matchings were performed: one on Class 2 alone, which resulted in subdivision into four subclasses (2A, 2B, 2C, and 2D); the other on Class 3 alone, which also resulted in subdivision into four subclasses (3A, 3B, 3C, and 3D) (see Appendix Tables A.2 and A.3).
Of individuals in Class 1, 57% repeated at least 1 academic year, compared with 20.4% of individuals in Class 4.
Future research should look at whether and to what extent this response to economic conditions is stronger or weaker among immigrant students than among natives or descendants of immigrants.
There is some overlap between these two characteristics: 82% of individuals with an employment contract before arrival in France arrived with a worker’s residence permit, but 63% of those with a worker’s residence permit did not have a contract.
This class would therefore be the closest to Class 6 (see below), which also has more trajectories including ‘variable’ and ‘other’ states but without long periods of employment. Class 6 is difficult to describe more precisely due to small numbers.
What is shown here is broader than the Algerian situation; it is labour immigration through the years of economic growth. During this period, this labour immigration was organized and encouraged by the State and large companies, particularly in the form of labour agreements with foreign countries.
Contrary to popular belief, the family reunification policy is not an invention of the 1970s. The right to family reunification was guaranteed by the 1945 Ordinance and then confirmed by a decree dated 29 April 1976, subject to guarantees of housing quality, before being suspended for 3 years in September 1977.
In the TeO survey, the ‘family reunification’ category for first residence permit obtained is recorded for both men and women. However, the arrival age of individuals migrating for family reunification differs greatly by sex: 90% of adults arriving for family reunification are women (who account for 66% of family-reunification residence permits), while 90% of men admitted to France for family reunification arrived under the age of 18 (the case for only 34% of individuals arriving for this reason). Therefore, male family-reunification migration involves mainly immigrant children.
In the initial statistical processing of the survey, the trajectories were analysed using OM to distinguish two subpopulations, one male and one female. The two resulting typologies are not fundamentally different from the one presented in this article (only the numbers of individuals are different, and the men do not have homemaker trajectories).
The subdivision of Class 3 (long-term salaried employees) revealed a profile of trajectories from homemaker to salaried employment (Class 3C) that was very similar to that of the women in Class 2A.
The reason for the end of cohabitation is provided in the survey, using three categories: separation, divorce, and death.
New data matching techniques provide a set of avenues for further investigation.