Comparative analysis of occupational mobility between foreign men and women in Brazil’s formal labour market

In this article we analyse whether: i) foreign women and men exhibit different occupational mobility patterns in Brazil’s formal labour market; ii) foreign women are less likely to experience up ward mobility thantheir male peers; and iii) the chances of upward mobility among foreign women and men in Brazil’s formal labour market differ by nationality. To examine these questions, we use the data from the Annual Social Information Reports (RAIS) from 1995 to 2015, which refer exclusively to workers employed in the formal labourmarket. The results demonstrate that the chances of mobility of men and women do not differ among all groups of foreigners, since this difference is dependent on the nationality concerned. For those groups that do show differences, the chances of upward mobility are higher amongmen than women.


Introduction
Pioneering studies of the occupational mobility of foreigners in a host society, analysing just men, identified a U-curve: an initial decline from the socio-occupational status of the last job in the country of origin to the first job in the destination country, followed by a period of upward career mobility following assimilation and the accumulation of human capital after numerous years living in the new country (CHISWICK 1977, 1978, CHISWICK & PAUL 2005. However, Portes and Manning (2008) argue that the accumulation of human capital through the immigrant's 1 assimilation is insufficient to explain the mode of integration of foreigners in the destination country, but there are diverse factors that influence this process,classifiable into three types: 1) individual (sex, language proficiency. age, period since arrival, place of birth, and educational level, for example); 2) family (social and economic capital); 3) contextual (government policy, prejudices existing in the host society, among others).
1 In this article, we use the terms immigrant and foreigner to refer to all those workers whose nationality is other than Brazilian. Though aware that these concepts are not synonymous, to avoid repetitions and maintain the fluidity of the text, we consider the two terms interchangeable here given that we are analysing only international immigrants without Brazilian nationality, that is, foreigners. According to these studies, the country of origin, as well as the country of destination, are fundamental to comprehen-2 In terms of classifying foreign workers from the Global South, we assume that these are immigrants born in countries defined by the methodology of the World Economic Situation and Prospects as those with a transitional or developing economy. In other words, these are countries that are not part of Europe (whether or not members of the European Union), North America (United States and Canada) or other developed countries like Japan, New Zealand and Australia, southern hemisphere countries defined as belonging to the Global North (UN 2014).
ding the different forms of insertion, occupational mobility and salary progressionof immigrants. However, research shows that, compared to their male counterparts, women tend to have fewer chances of upward mobility (MAGUID & BRUNO 2010, BRUNO 2007. Specifically in the case of Brazil, Suziki (2019) argues that there are national groups of immigrants who accumulate more disadvantages than others in terms of the probability of acquiring better jobs over time. Foreigners originating from the Global South encounter larger barriers to occupational mobility in the Brazilian labour market than those coming from the Global North. However, the author does not compare the occupational trajectory of men and women. Another point to emphasize is that studies of the occupational mobility of foreigners in Brazil arequite scarce in general.
To add to this debate, in this article we intend to answer the following ques- 1. Comprehending the occupational trajectory of foreigners in the host society The first studiesof the occupational mobility of foreigners in their host society, which emerged from classic assimilation theory, showed a U-curve in the insertion of immigrants into the labour market. An initial fall in the immigrant's occupational status was seen to occur due to issuesfaced in the destination society, such as language difficulties, or problem in recognizing university qualifications and previous work experience. However, with more time spent living in the host country, combined with theconsolidation of social networks, learning the official language of the host society, and acquisition of social 4 and symbolic capitals, 5 as well as the possibleobtainmentof residence and work permits, along with recognition of academic qualifications, immigrants succeed in improving their occupational status and complete the U-curve (CHISWICK & PAUL 2005, RUEF 2017).
In debate with classic assimilation theory,segmented assimilation theory emerged, premised on the idea that modern societies are diverse and segmented, meaning that the form in which immigrants become inserted into the host society depends on three kinds of conditioning factors: 1) individual variables, including factors relating to the person's exposure tothe host society, such as the ability to speak the language, age on arrival, place of birth, educational level, and racial and gender characteristics; 2) family variables, which relate to social and economic capitals; 3) contextual variables, which concern the diversity found in the country of destination, including the different government policies, values and prejudices existing in the host society (PORTES & MANNING, 2008, EGREJA & PEIXOTO, 2011.
In general, the first studies from both these theoretical approaches analysed samples of foreign men, excluding women, due to evidence that the labour market shows distinct dynamics for men and women and due to the greater participation of men in the labour market compared to women (CHISWICK, 1977AND 1978, CHISWICK & PAUL, 2005. 4 Social capital can be defined as the investment and use of resources embedded in social relations for expected returns (Lin 2006). 5 The acquisition of symbolic capital by the foreigner is understood as the assimilation of tastes, values, 'ways of communicating,' 'good manners' and other notions prevalent in the destination country. However, the studies by Adserà, Ferrer and Herranz (2020), Adseràand Ferrer (2016) andAdseràand Chiswick (2007) demonstrate that this pattern is not the same in other countries such as Canada, the United Kingdom and Sweden,where these authors show that foreign women do not remain stuck in low-skilled jobs. They experienced upward mobility, more quickly among more highly educated women.Over time,foreign women end up occupying jobs that require analytic skills and less physical effort. Furthermore, such studies demonstrate that the countries of origin and destination are fundamental to comprehending the forms of insertion, occupational mobility and salary progression of immigrants.
However, specifically in relation to the group of highly qualified women, some authors indicate that the upward mobility of these foreign women in the host labour market is even more difficult than for men (PURKAYASTHA, 2005;KOFMAN & RAGHURAM, 2006). The main obstacles to the upward trajectory of women are the many fields in the labour market still dominated by men and, consequently, the devalorisation of roles occupied by women and the domestic responsibilities traditionally attributed to them.
It is worth emphasizing that this international literature refers to research on immigrants moving from developing or developed countries to developed countries (from the South or North to the North), unlike Brazil, which, over recent decades, had taken insome foreigners migrating North-South but mainly South-South.Moreover, many of these studies also analyse the informal labour market, which is not the focus of our analysis. Few studies exist on socio-occupational mobility that compare female and male foreigners originating from South-South migration.
The studies of mobility produced by Bruno (2007), Maguid and Bruno Seeking to complement this debate, we propose to test: i)whether immigrant women have different mobility patterns to men, caused by structural features of Brazil's formal labour market, characterized as sexually segregated; ii) whether foreign women generally have fewer chances of upward mobility than men with similar socioeconomic characteristics; iii) whether,however, the impact of the sex variable is heterogeneous among the differentgroups of nationalities migrating to the country.

Data and methods
We used data from the Annual Social Information Report (RAIS) from 1995 to 2015, which characterize the individuals and companies in whichBrazil's formal workforce are inserted. However, this database possesses some limitations that deserve highlighting: 1)it presents only information on workers employedin the formal market and by companies 6 , meaning that the article's conclusions apply exclusively to such workers; 7 2) there is no information onemployers and self-employed workers; 3) the year of entry into Brazil is shown only from 2011 onwards. As a solution we selected for analysis only individuals who appear in the databases from 2011 to 2015; 4) the existence of errors inthe data arising from the fact that the reportsare administrative records that are not checked for consistency (STERNBERG, 2001); 5) there is no information on the foreigner's earlier 6 In other words, it does not include domestic workers employed by individuals, thus excluding domestic workers in family homes. However, it does include cleaners and general service workers. 7 The data from the 2010 Demographic Census indicates that 55.76% of foreigners are employed in the formal labour market. occupation in the country of origin. In this article, therefore, we do not analyse the potential U-curve trajectory experienced by immigrants. Instead the immigrant's work trajectory is examined solely on the basis of the second phase of the curve, comparing the first occupation in the host country (more precisely, the first that appears in theRAIS records) with the current occupation (again more precisely, the occupation in the last year that the person appears in the database).
The In addition, the analysis considers just the fourteen groups of foreigners with the highest number of observations (all those over 15,000 observations, representing a minimum of3% of the total). We adopted this strategy for two reasons:1) to maintain a high enough number of countries for a diversified analysis, including the flows of immigrants from both the Global North and South, but not so broad that the fluidity of the text would be compromised; 2) to achieve a more parsimonious model insofar asthe origin variable will be included in the equation as a control variable.
To test our hypotheses, we used the multi-levelmixed-effects linear regressionmodel. We constructed the database in longitudinal form, classified as unbalanced: in other words, information does not exist for each individual in every year, since there are numerous situations that may remove a worker from the RAIS records, such as entering the informal market, being laid off, leaving to become an entrepreneur, retirement and death.
We based our construction of the dependent variable referring to the measure of socio-occupational mobility on theInternational Socio-Economic

Index of Occupational Status (ISEI), an index developed by Ganzeboom, De
Graaf and Treiman (1992) In terms of independent variables, the models include information relating to: a) the worker's human capital and demographic characteristics; b)the process of assimilation in Brazil and the social and cultural differences; c) the operational sector of the companies employing the foreign workers; d) the status of the first occupation, based on the EGP classification (Carvalhaes 2015).Detailed information on these variables appears in the appendices.
Here it is worth explaining our inclusion of the predict_Formal variable as a control in the model, created in order to correct selection bias in the sample. This bias derives from the fact that the RAIS includes only workers from the formal job market. However, a large number of immigrants are known to be employedin informal jobs in Brazil. To correct this bias, the predict_Formalvariable was estimated, which refers to the probability of a worker being employedin the formal market rather than the informal. To create this variable, datafrom the 2010 Demographic Census relating to individuals in the informal sector was added to each year of the RAIS, and a logistic regression was applied to the data. This logistic regression included the following independent variables: age, sex, schooling, foreign or Brazilian, disabled, and region of the country in which the person is resident. This regression generated a score for the likelihood of the worker being employed in the formal sector, based on the observed variables.
In relation to the statistical models used in the analyses, we initially estimate a model including all the foreign workers. 8 Next, we estimate specific models for each group of foreigners.This process generatedfifteen equations.

Descriptive analysis
The We also analysed the existence of occupational niches on entering and leaving the labour market. In terms of entry occupation, neither sex presents high percentages of concentration in lower-skilled occupations -that is, in unskilled manual labour like general services, cleaning or construction.Furthermore, for men and women, there is an increase of foreigners employed in occupational groups in their final job that demand more analytic and management skills compared to their entry job. In the final employment, there is a higher percentage of workers occupying posts such as "directors and managers,""science professionals and intellectuals,""technicians and secondary education professionals." On the other hand, there is a notable decline in all those occupational groups demanding fewer qualifications when we compare the start and end of the labour trajectory.
This data suggests that foreign men and women demonstrate a tendency to upward mobility in the Brazilian labour market. These findings closely match the studies of Adserà, Ferrer and Herranz (2020), Adseràand Ferrer (2016),andAdseràand Chiswick (2007),which show that, with the passage of time, foreign women end up occupying jobs that demand more analytic skills and less physical effort.
In addition, we observe that foreign women in the formal market are generally not employed in jobs at the base of the occupational structure, but either in intermediary service activities or as science professionals and intellectuals. This data contradicts some of the literature's predictions, which classify foreign women as secondary workers, employed mainly in care professions.Onepossible explanation for this discrepancycould be that the literature refers to foreigners migrating from poor countries to rich countries, unlike foreigners migrating to Brazil who tend to come from developed or developing countries to a developing country. A second possible explanation is that we are dealing solely with the formal market, while previous studies do not specify whether the type of market is formal or informal, recognizing thatundocumented and heavily exploited foreign women potentially work in the informal sector. Another possible explanation is the fact that Brazil has a labour pool in precarious services (principally the general service, cleaning, and care sectors) generated by internal migration, making this space uncompetitive for foreign workers.
In terms of the average evolution of the ISEI with time recorded in the RAIS (Graph 1), we identified that the ISEI of women remains higher than those of men over time. The data suggest that the greater the experience on the formal work market, the higher the occupational status, principally for women.This result corroborates previous studies in both developing and developed countries, demonstrating that more experience in work, and more assimilation, increased the occupational status of immigrants (CHISWI-CKET al., 2005, POWERS et al., 1998. However, the result presents a higher average value of occupational status for women than for men. In terms of the mobility rates obtained through the ISEI,the results indicate thatmen and women are fairly closeacross all dimensions (immobility or upward and downward mobilities: see Table 1),with foreigners in Brazil generally displaying higherrates of immobility.Even so, it is worth highlighting a small percentage (0.06%) higher upward mobility among men compared to women. Based on these descriptive analyses, we can identify that women have average ISEIs higher than those of men and that women's trajectories of mobility are not very different to men's, but they do show a small disadvantage in terms of percentage of upward mobility. We can call attention to the fact that women have a higher percentage of immobility.
Perhaps this is not a significant disadvantage for foreign women, given that they present a higher average ISEI than their male peers over time. Nonetheless, these results do not allow us to infer whether foreign wo-men have more or fewer chances of upward mobility compared to foreign men in Brazil's formal labour market due to the fact of being women. To verify this possibility, a more advanced analysis is needed, controlling other factors.

Advanced analyses: model estimators
In argue that a period of 14 to 20 years is needed residing in the country of destination for more skilled occupations to be attained.
The analysis of the mobility rates through the calculation of the ISEIsfor each country of birth of the immigrants indicates that all these groups show higher levels of immobility, except Chilean men and Haitian women who have higher percentages of upward mobility, 38% and 37% respectively. The groups with the highest percentages of immobility are Chinese women and men, with 69% and 64% respectively, followed by Bolivian men and women with 62% and 59%. Meanwhile, downward mobility attainshigher percentages among Haitian men (29%), Chilean men (28%) and Haitian women (27%).
The central finding is that there are different occupational trajectories for men and women among distinct national groups, whether from the Global North or South. We did not identifya downward mobility pattern for foreign women. In general, men have a higher or similar rate of downward occupational mobility to women. But the question remains:is sex a factor that influences these differences?
In the econometric models separated by each nationality, the results indi-

Final considerations
The results found in this study arefairly significant. They dialogue with the international literature, which affirms that: i) immigrant women, in general, are employed in worse jobs than men; ii)they either tend to present fewer chances of upward social mobility in the labour market of the host society compared to men; iii) or their chances of upward social mobility are null -that is, they do not present chances of mobility in a U-curve pattern but in an L-shape.
We stress that, in the case of our study, it was not possible to measureU- Another point to highlight is that our data has limitations, the most significant being the fact that we analysed only those immigrants employed in Brazil's formal labour market by companies, thus excluding those working for individuals. Our results, therefore, may be quite different from those encountered in previous studies that coverthe labour market as a whole (formal and informal, thus excluding female domestic workers for individuals, an occupation held by many foreign women). Furthermore, the database does not contain any information on employers and self-employed workers, positions occupied by many international immigrants. This fact can generate an underestimation of the upward mobility (in the case of employers) and the upward and downward mobility (in the case of the self-employed) of those individuals being studied.
In the analysed data, in general, we did not observe, as indicated by some earlier studies, a concentration of foreign women in low status occu-pations in the care sector, except among Haitians, or a concentration of men in unskilled industrial jobs. We did find that the majority of foreign women are employed in Brazil's formal market in non-manual routine occupations, considering the period of assimilation into the host society. One probable explanation for this finding resides in the differences between migrations from poor to rich countries and from rich to poor countries. It was not expected, of course, that foreign women arriving from rich countries, like Germans, French and Americans, would become employedin more low-qualified occupations in Brazil (like the care sector). However, the same was not expected in the case of female immigrants from poor countries, like Bolivian, Paraguayan and Peruvian women, among others, which is what we found.
These results differ, therefore, from those encountered by Bruno (2007)  In terms of the insertion of foreign men in the formal market, we can notea concentration of workers in high-level professional posts and a tendency for an increase in male immigrants occupying skilled manual jobs.The explanations are the same as the case for women.
Men and women display a similar pattern in terms of the average evolution of the ISEI according to time recordedin the RAIS, a variable utilized as a proxy for experience in the labour market. It was also foundthat the ISEI of women remains higher than men's over time.For both sexes, there is a tendency for the ISEI to increase on average.
For foreigners as a whole, the sex variable has an effect on comprehending the occupational trajectories of foreigners in Brazil, with men generally showing higher probabilities of upward mobility than women. In other wor-ds, the model utilized for all foreigners confirms our overall supposition that female immigrants have different mobility patterns to men with lower probability of upward mobility, as shownby the international literature. However, when we analysed the fourteen national groups separately, the difference is positive for men among a higher portion (seven groups). For the six remaining groups no statistical difference is observed.Meanwhile, for Haitians the difference is negative for men. This case should be analysed in more detail in future studies.
This evidence demonstrates that detailed analyses are needed of the situation experienced by foreigners in the Brazilian labour market. In other words, if we only undertake a descriptive analysis of the occupational situation and mobility among men and women, we do not encounter a very different situation between them.In some cases, we might even conclude that women find themselves in a better situation, since, for instance, they presenthigher values of socio-occupational status than men over time.
However, as we proceed deeperinto the analyses, controlling other important variables in order to examine the real effect of sex on occupational mobility, we identify something very interesting: the pattern found among male and female foreigners is that female immigrants do not have more chances of upward mobility than their male counterparts (except for Haitian women).

Egp1
Binary 2 1= High level professionals 0= Agricultural workers and others in primary prod.

Egp2
Binary 2 1= Low level professionals 0= Agricultural workers and others in primary prod.

Egp3
Binary 2 1= Non-manual routine, higher education 0= Agricultural workers and others in primary prod.

Egp4
Binary 2 1= Services and sales 0= Agricultural workers and others in primary prod.

Egp5
Binary 2 1= Manual work supervisors 0= Agricultural workers and others in primary prod.

Egp6
Binary 2 1= Skilled manual work 0= Agricultural workers and others in primary prod.

Egp7
Binary 2 1= Semi-skilled and unskilled manual work 0= Agricultural workers and others in primary prod.