UvA-DARE (Digital Academic Repository) A Good Match? Education, Labour Market Position, and British South Asian Transnational Marriage

This study applies exchange theory to transnational marriages between descendants of migrants to Europe, and partners from their (grand)parents’ country of origin. Such marriages could offer socio-economic beneﬁts for the European partner/family, if the opportunity of migration attracts a more highly educated spouse. The translation of educational capital into socio-economic beneﬁts, however, is mediated by the labour market position of migrant spouses. In this study we explore the relationships between transnational marriage, education, and employment, by comparing the characteristics of spouses in transnational couples with those in intranational couples. Analyses are based on UK Labour Force Survey data (2004–2014) for two groups in which transnational marriage is common— Pakistani Muslims and Indian Sikhs. We ﬁnd that educational homogamy is the dominant pattern in both intranational and transnational couples, and that migrant spouses have a disadvantaged labour market position compared to non-migrant spouses with the same level of education—with variation across gender and ethnic groups. Our ﬁndings do not support a view of transnational marriage as socio-economic exchange but do suggest education plays a role in spousal choice in these marriages.


Introduction
Exchange theories of marriage suggest that spouses may exchange benefits in contracting a marriage.This approach to understanding spouse selection developed in the context of Black-White intermarriage in the United States to suggest that ethnic/racial 'prestige' might be exchanged for other desirable traits, such as education (Kalmijn, 1998).Studies on marriage between other 'racial' groups and in other countries have found little evidence of exchange (Kalmijn, 2010, but see Gullickson and Torche, 2014), but in this study we apply exchange theory to a different type of union: transnational marriage between descendants of migrants to Europe and partners from their parents' or grandparents' country of origin (hereafter referred to as the 'ancestral country').With family involvement common in such marriage choices, a more explicit evaluation of the relative attractions of potential partners might be envisioned than in contexts where marriages are expected to be based on love and physical attraction between individuals.Whilst Charsley (2013) has argued against purely strategic views of such marriages, and for the importance of understanding the emotional factors involved, in this study we explore exchange theory as a lens for interrogating the potential benefits of the choice of a partner from the ancestral country.
For most ethnic minority and migrant groups in Europe, the ancestral country has a lower level of economic development than the country of residence.Legislative changes on spousal migration are therefore often justified on the assumption that marriage migrants from these countries have low levels of education, and hence a weak position on the labour market, with consequences for their own structural integration, the socioeconomic position of the couple, and cumulatively the ethnic group (Kofman, Saharso and Vacchelli, 2013;Bonjour and Kraler, 2014).From the perspective of exchange theory, however, transnational marriages may offer possibilities for Europeans of migrant descent to attract a higher educated spouse by offering the opportunity for migration.If this is the case, transnational marriage could strengthen, rather than weaken the socio-economic position of the European partners.
Studies using exchange theory generally take education as an operationalization of labour market success, but the returns to education may vary with other traits involved in the exchange.Whether achieving a higher educated spouse through transnational marriage leads to a better socio-economic position for the couple depends on whether the migrant spouse obtains employment commensurate with their level of education.Migrants tend to receive lower returns to education (Kanas and Van Tubergen, 2009), which may reduce the socio-economic benefit of a highly educated migrant partner.Differing gender role attitudes in the ethnic group or ancestral country may also restrict the labour market participation of women in such couples (Joppke, 2009;Kofman, Saharso and Vacchelli, 2013).
In this study, we explore whether education status exchange occurs in transnational marriages and how partner choice relates to employment rates and the prevalence of low-level occupations.We focus on two of the largest ethnic minority groups in the United Kingdom involved in this form of transnational marriage: Pakistani Muslims and Indian Sikhs.As education and opportunities for migration are both highly valued in South Asian marriage markets (Jeffrey, Jeffery and Jeffery, 2007;Charsley, 2013), these groups form a good test case for exchange theory.Employing data from the UK Labour Force Survey (LFS) household files (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014), we compare the education and labour market profiles of both partners in transnational couples (a person born and/or raised in the United Kingdom, with a partner who arrived in the United Kingdom as an adult) with those in intranational couples (both partners UK-born/raised).We restrict our analysis to ethno-religiously endogamous heterosexual couples, both because of the differing dynamics which may underlie interethnic marriages, and the relatively low numbers involved.Given the suggestion of different dynamics depending on the gender of the migrant spouse (Lievens, 1999), and the frequent focus in policy discourses on women in transnational marriages, analyses are conducted separately by gender.
Studies of transnational marriages among European ethnic minorities have tended to focus on Muslim populations (Huschek, de Valk and Liefbroer, 2012;Carol, Ersanilli, andWagner, 2014 -but see Gonza ´lez-Ferrer, 2006).In including a non-Muslim group, we move beyond the implied particularity of Muslim marriage practices and processes of integration, to better reflect the diversity of contemporary marriage-related migration (Charsley, 2012).By looking at the socio-economic characteristics of both partners, we expand on studies that have generally focused on the characteristics of the European rather than migrant spouse, or on cultural issues.In contrast to the few studies examining education exchange in transnational marriage, and the broader literature on interracial marriage, we not only consider education as an exchangeable 'capital' in itself but also investigate the labour market position of the spouses.

Migration Context
Labour migration from India and Pakistan to Britain (the former colonial power) started to take off in the 1950s.Family reunification and the development of large British-born populations followed.Substantial numbers of these later generations have married partners from the Indian subcontinent-more than half of British Pakistanis, and over a third of British Indians, have spouses from overseas (Georgiadis and Manning, 2011).Sikhs are believed to be the most significant Indian group of second-/third-generation transnational marriages (Singh and Tatla, 2006).
British Pakistanis and British Indians differ in two aspects that may influence education exchange and labour market outcomes.The British Pakistani population is characterized by a weak labour market position and poor educational outcomes (Khattab et al., 2011;Platt, 2005).In contrast, British Indians are often represented as a 'model minority' in terms of socio-economic progress.While Sikhs have poorer socio-economic outcomes than Hindus, they have a higher socio-economic profile than Pakistani Muslims, even accounting for differences in education and age composition (Platt, 2005;Khattab et al., 2011;Heath and Martin, 2013).Labour European Sociological Review, 2019, Vol. 35, No. 1 Downloaded from https://academic.oup.com/esr/article-abstract/35/1/133/5116188 by University Library, University of Amsterdam user on 19 March 2019 market participation among women is higher among British Sikhs than among British Pakistani (Khattab et al., 2011;Heath and Martin, 2013), which may affect the importance attached to the wife's education.Furthermore, whereas among Pakistani Muslims kin marriage is common, this is not the case for Sikhs.This could affect the composition of the pool of potential partners.

Theorizing Relationships between Transnational Marriage, Education, and Employment
Education A commonplace observation is that spouses tend to have similar cultural and socio-economic characteristics (Blossfeld, 2009).Two mechanisms have been proposed to explain this outcome.The matching hypothesis states that people prefer partners who are similar, whilst the competition hypothesis postulates a preference for the highest possible amount of a certain trait.Thus, if both sides prefer to marry 'up' and are reluctant to marry 'down', the result is homogamy (Schwartz, 2013).Studies testing these hypotheses in a range of Western countries have found evidence for both, but for different traits: matching on cultural traits, but competition on economic traits (Kalmijn, 1994;Schwartz, 2013).
Ideas of homogamy and hypergamy are also culturally relevant in South Asia and its diaspora.Among the connotations of the Urdu/Hindi concept of a rishta is a marital 'match' in terms of a desirable (for the spouses and/or their families) similarity between spouses (Charsley, 2013).However, in the context of transnational marriage, certain cultural differences may be valued, such that the choice of a partner from overseas may reflect a search for a spouse with greater knowledge of religion or a 'mother tongue', for example, or more 'traditional' behaviour (Timmerman, 2006;Charsley, 2013).
As marrying 'down' in educational terms is generally more acceptable for men than women, lower educated men and higher educated women may face the greatest constraints in finding a partner within the local/national ethnic group (Blossfeld, 2009).Highly educated women in communities with low expectations of female labour force participation may be particularly affected, as co-ethnic men have less incentive to look for an educated partner (Strassburger, 2003;Blossfeld, 2009).Those who face such constraints may turn to an additional pool of potential partners in their ancestral countries of origin, where their ability to facilitate the migration of a spouse can enhance their position in the marriage market.Previous studies on migrant communities in Western Europe have consistently found that men with low education are more likely to be in a transnational marriage (Lievens, 1999;Gonza ´lez-Ferrer, 2006;Baykara-Krumme and Fuß, 2009;Muttarak, 2010;Huschek, de Valk and Liefbroer, 2012;Carol, Ersanilli, and Wagner, 2014).For women, results are mixed.Several studies have found higher educated women of Turkish and Moroccan origin are more likely to be in a transnational marriage (Lievens, 1999;Gonza ´lez-Ferrer, 2006;Carol, Ersanilli, and Wagner, 2014), but others have not found a clear relationship between education and propensity for transnational marriage (Baykara-Krumme and Fuß, 2009;Dale and Ahmed, 2011).
Transnational marriage not only offers an alternative to those unable to find a match in the country of residence but can also be a way to find a spouse with more advantageous characteristics.Exchange theory is a variant of the competition hypothesis; it assumes that there is a preference for a partner with the most desirable traits.According to the dyadic version of exchange theory, people are willing to trade down on one desirable trait in exchange for another (Kalmijn, 1998;Schwartz, 2013;Gullickson and Torche, 2014).The market version of exchange theory postulates that certain traits can act as penalties or advantages in the marriage market.In this case exchange does not occur at the couple level but is an outcome of sorting in the marriage market by preference and traits (Gullickson and Torche, 2014).
The opportunity of migration may attract potential spouses overseas with higher levels of education.This pattern has been found for Danish Pakistanis (though not for Danish Turks; C¸elikaksoy, Nielsen, and Verner, 2006) and for the second generation of Turkish, Moroccan, and of former Yugoslav origin across multiple Western European cities (Hamel et al., 2012, but see Baykara-Krumme andFuß, 2009).Among British Sikhs, Qureshi (2016) has identified what she calls 'shehri [city] brides'-educated women with career aspirations migrating as wives for men who are often less highly educated-whilst Charsley (2013) reports some similar cases amongst British Pakistanis.Our first hypothesis is therefore: UK-born/raised men and women who marry transnationally achieve a spouse with a higher level of education than themselves more frequently than those who marry intranationally (H1).
Labour market participation is considerably higher among UK Indian Sikh women than among UK Pakistani Muslim women, making women's educational capital for labour market engagement potentially more important in the former group.Furthermore, a preference for kin marriage among Pakistani Muslims may focus the search for a spouse on a more limited pool of potential spouses.We therefore expect that: For UK-born/raised Indian Sikh Downloaded from https://academic.oup.com/esr/article-abstract/35/1/133/5116188 by University Library, University of Amsterdam user on 19 March 2019 men, a transnational marriage (compared to an intranational marriage) is more often a marriage with a higher educated spouse than for UK-born/raised Pakistani Muslims who marry transnationally (H2).

Socio-economic Position
Transnational marriage might affect the socio-economic position of UK-born/raised South Asians by influencing the labour market participation of the UK-born/raised South Asian spouse or by influencing that of their partner.

Labour market participation of the migrant partner
If a transnational marriage is indeed a way of attracting a higher educated spouse, this may have a positive effect on the couple's resources.A more educated spouse is more likely to be employed and have a higher-level occupation.Migrant spouses, however, might not attain the same returns on their education in the labour market as UKborn/raised South Asian spouses.Migrants suffer higher unemployment and higher levels of low-skilled employment than non-migrants with a similar level of education (Kogan, 2004;Ferna ´ndez and Ortega, 2008).This lower return on education is a result of multiple factors.Experience and qualifications obtained abroad may not be transferred given differing contexts of education and employment, language issues, and non-recognition of qualifications (Chiswick, 1978;Ferna ´ndez and Ortega, 2008).Discrimination negatively affects employment chances of ethnic minorities (Blommaert, Coenders, and Van Tubergen, 2014) and may particularly impact migrants given additional markers of 'difference'.
Another factor affecting migrants' ability to find work is their often limited social capital (Kanas and Van Tubergen, 2009).Here, marriage migrants may have a comparative advantage over other migrants, as their connection to a local family may provide valuable contacts for employment (Jasso and Rosenzweig, 1995).Studies in the United States, Sweden, and Spain, however, found that family migrants are less likely to be employed than labour migrants, even after accounting for human capital characterictics (Jasso and Rosenzweig, 1995;Rodrı ´guez-Planas and Vegas, 2011;Luik, Emilsson, and Bevelander, 2016).
We therefore hypothesize that: migrant husbands and wives are less likely to be employed than UK-born/ raised South-Asian husbands and wives, including after accounting for their level of education (H3) and that, migrant husbands and wives are more likely to be employed in low-level jobs, including after accounting for their level of education (H4).
Several studies on British South Asians have shown that Indian Sikhs fare better in the labour market than Pakistani Muslims-even when taking education and other group composition differences into account (Khattab et al., 2011;Heath and Martin, 2013).It is not clear whether these differences in labour market position are due to stronger discrimination against Muslims, differences in values, or in bridging and bonding social capital (Heath and Martin, 2013).
In both groups, economic activity rates among women are lower than among men (Cheung, 2014), but women's activity rate is particularly low among British Pakistani Muslims.Women in the second generation are more often active in the labour market than the previous generation, but this increase is more marked among Indian Sikh than among Pakistani Muslim women (Cheung, 2014).The variation between the two groups is likely to be in part a result of different ideas about women's domestic versus paid labour.We therefore expect that the employment gap between migrant wives and UK-born/raised South-Asian wives is larger among Pakistani Muslims than among Indian Sikhs, including when accounting for education (H5).
Labour market participation of British South Asians Verbakel and De Graaf (2009) suggest that one of the mechanisms explaining the greater labour market participation of women with highly educated husbands found in many European studies is that such men have less traditional gender role attitudes and thus are more supportive of their partner working.Indeed, Khoudja and Fleischmann (2015) find that women whose partners have traditional gender role attitudes are less likely to participate in the labour market, irrespective of women's own gender role attitudes.If, as often assumed, men from Pakistan and India have more traditional gender role attitudes 1 than British South Asians (Dale and Ahmed, 2011), we might expect to find that: UK Indian Sikh and UK Pakistani Muslim women married to migrant husbands are less likely to participate in the labour market than women in intranational couples (H6a).
However, marrying a partner from abroad may also boost female labour market participation.A preference for women not to work may be overcome if this leads to economic hardship (Cf.Khattab, Johnston and Manley, 2018).Migrants need to acquire destination country specific skills to be successful in the labour market (Chiswick, 1978;Kogan, 2004), so that Long (1980)  to transnational couples: British South Asian wives may enter the labour market to earn until the husband can find employment, and continue working if their husband can only find low-skilled, low-waged work-as is common for migrant men (Ferna ´ndez and Ortega, 2008;Longhi, Nicoletti, and Platt, 2013).This leads to the competing hypothesis that UK Indian Sikh and UK Pakistani Muslim women in transnational couples are more likely to participate in the labour market than women in intranational couples (H6b).

Data
We test the hypotheses using UK Quarterly LFS household files, aggregating data from 2004 to 2014. 2 The sample is limited to heterosexual couples in which partners either both list Indian ethnic identity 3 and Sikh religion or both Pakistani ethnicity and Islam as religion.We only include couples in which at least one partner is either born in or migrated to the United Kingdom before the age of 18 4 .The sample includes a small number of non-married cohabiting couples. 5As labour market participation falls sharply after the age of 50, the sample is restricted to couples in which both partners were below 50 at the time of the survey.We also excluded couples for whom education data was missing and could not be imputed (see below).
The UK LFS does not contain information on the year of marriage, the start of cohabitation or reason for migration.It is thus not possible to determine whether migration preceded or followed the contracting of the marriage. 6Qualitative research, however, suggests that the ordering of marriage and migration may be an inaccurate guide to categorization, as an intended spouse may enter on a fiance ´(e), work or study visa.Given this fluidity of migration categories and trajectories, the term 'transnational marriage' is treated as encompassing all marriages in which one partner is an adult migrant (cf.Charsley, 2012).For the purposes of this exploration of exchange theory, there is also continuity, as marrying a British permanent resident or citizen has similar benefits in terms of permanent residence for migrants who enter the United Kingdom on a temporary visa route, as for aspiring migrants in the subcontinent. 7 Table 1 shows the proportions of endogamous transnational and intranational partnerships and partnerships with White British among the UK-born/raised sample by ethnic group.As expected, transnational couples are more common among Pakistani Muslims than among Indian Sikhs.The share of interethnic partnerships is too low for meaningful analysis and is dropped from the sample.

Measurement
Couple type: Persons born and/or raised in the United Kingdom, with partners who arrived in the United Kingdom as an adult, are classified as being in a transnational couple, with UK-born/raised persons with partners who are also UK-born/raised (intranational couple) as reference group.
Education match: Education is measured as highest qualification.In the UK LFS data, the highest qualification of about 40 per cent of those who migrated as adults and 10 per cent of those born in the United Kingdom or arrived before the age of 18 was coded as 'other'.For these cases, we used age at which the respondents left full-time education to estimate their highest qualification.The resulting categories are: This three-level education variable was used to generate an education match variable consisting of three categories: partner is higher educated than respondent, both partners have the same level of education, and partner is lower educated than the respondent.There are substantial differences in education levels between the United Kingdom and Pakistan and India, but we compare absolute rather than relative (to locality) levels of education, as we are interested in education as an influence on labour market position in the United Kingdom.
Employment: Those working at least 1 hour a week are considered employed (¼1).Unemployed and inactive are combined (¼0). 8 Low occupation level: Dummy variable of the lowest category of the LFS occupation groups: elementary occupations (unskilled work).This group includes occupations such as cleaners, domestics, maintenance workers, porters, shelf-fillers, and car park attendants.
As people who migrated as children tend to obtain lower qualifications and have a higher propensity to marry transnationally than members of the second generation, we created a dummy variable for whether a respondent was foreign-born (UK-born ¼ 0).In our Downloaded from https://academic.oup.com/esr/article-abstract/35/1/133/5116188 by University Library, University of Amsterdam user on 19 March 2019 analysis we furthermore control for year of birth (education matching), age at survey (employment), and survey year.In the analyses on employment we additionally control for the presence of dependent children under the age of 12 (1 ¼ present, 0 ¼ not present).
Descriptives by couple type, gender, and ethnicity are presented in the Online Appendix.

Models
We run multinomial logistic regression for the nominal dependent variable (education matching) and logistic regressions for the dichotomous dependent variables (employment, low-level occupation), with separate models for each gender and group.As logistic regression coefficients are difficult to interpret, we present figures with predicted probabilities for the outcome split by couple type (intranational vs. transnational).Tables with the full results are included in the Online Appendix.Figures and logistic regression tables are based on average marginal effects (AMEs), as these are more suitable for comparison across groups and models (Mood, 2010).AMEs are computed by calculating the predicted value for each case in the data set for the possible values of the variable of interest and then averaging the difference.

Education
Before examining the relationship between transnational marriage and education matching, we looked at the relationship between level of education of UK-born/raised Pakistani Muslims and Indian Sikhs and the likelihood of being in a transnational partnership.Controlling for year of birth, being foreign-born and year of survey, we find that men from both groups and UK Indian Sikh women 9 who have completed post-secondary education are significantly less likely to be in a transnational marriage than those with secondary education or no qualifications (see Online Table A2).For UK Pakistani Muslim women, however, we find a different pattern; women with secondary school qualifications are significantly more likely to be in a transnational marriage than either those without qualifications or those with post-secondary education.These findings are broadly in line with those by Dale and Ahmed (2011) using earlier waves of the UK LFS but different ethnic grouping (Pakistani/ Bangladeshi and Indians undifferentiated by religion).The findings also mostly fit the suggestion that people with limited education face most difficulties in attracting a spouse in the British South Asian marriage market.
Next we analysed educational matching.Figure 1 displays the predicted probabilities for UK Pakistani Muslims and UK Indian Sikhs of having a partner who has higher, similar, or lower levels of education to themselves (see Online Table A3).The figure shows that across all groups and couple types, educational homogamy is the dominant pattern, occurring in about 50 per cent of couples of both genders and groups.It also shows significant differences between transnational and intranational couples.For UK Pakistani Muslims, both men and women in transnational couples are significantly more likely to be higher educated than their partners, compared to those in intranational couples. 10UK Indian Sikhs demonstrate a more diverse pattern; for both genders, both having a higher educated and a lower educated partner are significantly more common in transnational partnerships. 11 We previously established that people with higher education are less likely to be in a transnational couple.As there is no higher category into which people educated to post-secondary level could marry, the apparent educational hypergamy (marrying up) among transnational Indian Sikh couples could be a result of the lower share of UK partners with a post-secondary education among these couples.Analyses by level of education show that the higher proportion of educational hypergamy in transnational marriages of UK Indian Sikh men is indeed a composition effect.For UK Indian Sikh women the picture is more complex.Women with less than secondary education (n ¼ 29) have a lower probability of marrying up in transnational marriage, women with secondary education (n ¼ 182) have a significantly higher probability of marrying up and down in a transnational marriage, and women with post-secondary education (n ¼ 169) have a significantly higher probability of marrying a lower educated partner when in a transnational marriage.Therefore, only for UK Indian Sikh women do we find some support for H1 that transnational marriages are more often with higher educated partners than is the case in intranational marriages.For the other groups, such marriages are more often with lower educated partners than when marriages are intranational.Hence, although the overseas partner is more educated in a substantial minority of transnational marriages, particularly among Sikhs, a status exchange of migration opportunity for education does not seem to be the dominant pattern in transnational unions.Rather, in all groups, educational homogamy is the most common outcome of both transnational and intranational marriage.
H2 predicted that for UK Indian Sikh men, compared to an intranational marriage, a transnational marriage would more often be with a higher educated partner than among UK Pakistani Muslim men.We tested this hypothesis by adding an interaction term to a pooled model for both groups.The results show that for UK Indian Sikh men transnational marriage is indeed significantly more often with a higher educated partner than for UK Pakistani Muslim men (see Online Table A3).However here too analyses by level of education show this is a composition effect driven by the stronger relation between education and transnational marriage among Indian Sikhs.H2 is therefore rejected.
Unexpectedly, we find that for UK Indian Sikh women, a transnational marriage (compared to an intranational marriage) is significantly more often a marriage with a higher educated partner than for UK Pakistani Muslim women.This difference remains when tested by level of education.

Employment of Migrant Partners
We hypothesized that migrant wives and husbands have a lower likelihood of employment than UK spouses, even when accounting for their level of education.Figure 2 shows the predicted probabilities of employment of partners split by level of education of the partner and whether the marriage is transnational.Note: Estimates based on models presented in Online Table A3.
Source: UK LFS.Migrant wives are significantly less often employed than UK Pakistani Muslim and UK Indian Sikh wives. 12Lower levels of education among migrant wives only partly explain this difference (see Online Table A5).For Pakistani Muslims, migrant wives with post-secondary education have a similar likelihood of being employed as UK Pakistani Muslim wives with secondary education: around 30 per cent.Similarly, employment rates of migrant wives with secondary education are roughly equal to UK wives with less than secondary education.For Indian Sikh wives the difference between migrant and UK wives is present but less pronounced.
Unexpectedly, migrant husbands are as likely to be employed as UK Pakistani Muslim and UK Indian Sikh husbands, both before and after controlling for education.Lower education levels of migrant partners compared to intranational partners and problems with converting human capital do not therefore seem to lead to lower rates of employment for these men.
The response to H3-that migrant husbands and wives are less likely to be employed-is thus strongly gendered.H3 is confirmed for migrant wives but not for migrant husbands.The employment gap between migrant wives and UK-born/raised South-Asian wives is significantly larger among Pakistani Muslims than among Indian Sikhs (see Online Table A6).This is in line with H5 and supports the interpretation of the employment gap as being at least partly explained by gender norms, but may also be influenced by gendered forms of discrimination against Muslim women (Ghumman and Ryan, 2013).

Occupational Level of Migrant Partners
While migrant husbands have similar employment rates to their non-migrant counterparts, problems of converting human capital might restrict them to lower level occupations.Figure 3 shows the predicted probability of having a partner in low level occupation, again split by level of education of the partner and whether the marriage is transnational.It is clear that for both ethnic groups, migrant wives and migrant husbands are significantly more likely to work in elementary level occupations than UKborn/raised partners (see Online Table A8).Education only partly explains this result.The findings in with H4; migrant wives and husbands will more often be employed in low-skilled jobs, even after accounting for level of education.
For each level of education, migrant wives and husbands are more likely to be in low-level employment than their UK counterparts.For UK Pakistani Muslim men, the higher propensity of a migrant wife being in low-level employment is partially countered where the wife has a higher level of education.A migrant wife with less than secondary education has a probability of around 70 per cent of being in low-level employment compared to about 35 per cent for a UK-born/ raised wife with the same level of education.A migrant wife with secondary education has a probability of roughly 25 per cent of being in low-level employment. 13 For UK Indian Sikh men and women and for UK Pakistani Muslim women, a migrant spouse with  In sum, transnational marriage is associated with a higher likelihood of having a spouse in low-level employment.This relationship is more pronounced if the transnational marriage is with a spouse with low levels of education but is negated if the transnational union results in a higher educated spouse.

Employment of UK Spouse
Having considered the labour market position of migrant spouses, we turn to the relationship between transnational marriage and employment levels of UK-born/ raised individuals.While our hypotheses here relate only to women, we also ran the analyses for men for comparison.As H6b predicts a higher labour market participation to compensate for initial adjustment of migrant husbands, we ran the analyses splitting the migrant group into length of stay groups: 0-5, 6-10, 11-15, and 16þ years.Figure 4 shows the predicted probabilities of employment by couple type.What stands out are the low rates of employment among Pakistani Muslim women, compared to men in both ethnic groups, and to Indian Sikh women.
Contrary to H6a and H6b, being part of a transnational couple is unrelated to the likelihood of employment of UK Pakistani Muslim or UK Indian Sikh women.However, in line with the mechanism suggested in H6b, UK Pakistani Muslim women are more likely to work if their partner has been in the United Kingdom for 5 years or less, and less likely to work if their migrant partner has been in the United Kingdom for at least 16 years, compared to women with a UK partner (see Online Table A10).The differences remain when controlling for the husband's labour market position, which casts doubt on the idea that the higher employment rate of UK Pakistani Muslim women is a way to compensate for lower earnings of migrant husbands.There is thus only limited evidence that the labour market participation of British South Asian women varies with the migration status of their spouse.
UK Pakistani Muslim men and Indian Sikh men in transnational couples are equally likely to be employed as their counterparts in intranational couples.

Discussion
Marriages between Europeans of ethnic minority group backgrounds and partners from ancestral countries of origin have been attracting increasing academic and political interest in recent years.This study, focussing on British Pakistani Muslims and Indian Sikhs marrying transnationally and within the ethno-religious group, examines relationships between transnational marriage, education, and labour market participation.We explored the use of exchange theory, conceptualizing ancestral countries of origin as additional marriage  markets, where education may be exchanged against the opportunity to migrate.As the benefits of education may be realized in employment and resultant socio-economic gains, we also examined the rates and levels of employment of migrant spouses.Our findings contribute both to developing understandings of transnational marriages among ethnic minorities of migrant background in Europe and to assessments of the utility of exchange theories of marriage.
In this study, we find that the dominant pattern is educational homogamy between spouses.The relatively high proportions of migrant spouses with lower levels of education are thus largely a reflection of the educational profile of British Indian Sikhs and British Pakistani Muslims entering into transnational marriages.Spouses from India and Pakistan have higher levels of education than their British partners in a substantial minority of transnational marriages.Educational hypergamy is however not more common among British Pakistani Muslims and British Indian Sikh men marrying transnationally.Instead, we found that the British-born/ raised spouses in a transnational marriage are more likely (than in an intranational marriage) to have a spouse with a lower level of education than themselves.
Our findings do not support the predictions based on exchange theory.This is all the more notable given that education is known to be valued in the marriage markets of the groups under study.It may be that those marrying transnationally have characteristics not included in our models that led them to find a partner abroad.It may also imply the absence of a hierarchical relationship between potential spouses from the United Kingdom and the ancestral countries.Kalmijn (2010Kalmijn ( : p. 1252) argues that the occurrence of status exchange in black-white marriages in the United States 'can be regarded as evidence that in the marriage market-and presumably also in society at large-race is treated as a hierarchical variable' (italics in original).In our study, however, a spouse from an ancestral country does not seem to hold a lower status in the marriage market requiring compensation through higher education.If the perception of marriage migrants becomes more negative, however, this may change (Charsley and Bolognani, 2017).
The mixed results on exchange in transnational marriage in other studies may be a consequence of different approaches to modelling education exchange.Given our interest in labour market position we examined absolute rather than relative levels of education of (migrant) spouses.Whilst in the UK context the educational profile of migrant spouses from the subcontinent does not appear high, it compares very favourably with origincountry levels of education.For those born in 1971-1975, only 3.3 per cent of women in Pakistan and 7.3 per cent of women in India attended post-secondary education (Barro and Lee, 2010).In comparison, about a third of Indian Sikh migrant wives and a sixth of  Pakistani Muslim migrant wives in the UK LFS sample have post-secondary education.The education of migrant husbands also compares favourably to that of the origin country population (Barro and Lee, 2010).This goes against common assumptions in public debates that migrant spouses have little education-instead they have a significantly higher educational profile in comparison to the general population in their countries of origin.Tests of exchange theory frequently take education as an indicator of socio-economic status.However, our findings show a lower return on education for migrant spouses.Migrant husbands have employment rates equal to their non-migrant counterparts but are significantly more often employed in low-level occupations.Migrant wives are both less often employed and more often work in low-level occupations.These findings hold when accounting for education.The common focus on education as proxy for socio-economic status in studies of exchange theory may thus misconstrue the exchange taking place.Future studies should take into account that the returns on education may depend on other traits such as parental social class, or migrant/ethnic minority status.
The dominance of educational homogamy even in Pakistani migrant-wife couples, where women are unlikely to take up employment, and despite migrant spouses' lower returns on education in the British labour market, suggests the need for a broader perspective on the role of education in these marriage choices.A certain level of education may be considered desirable in a spouse for reasons unrelated to immediate socio-economic gains-for example in enhancing or consolidating the educational environment for raising children.Education could also be viewed as part of the cultural traits involved in ideals of similarity between partners to lay the foundations for marital success (cf.Kalmijn, 1994;Charsley, 2013), rather than economic maximization through marriage as a form of exchange of capitals.
When it comes to differences between the two ethnoreligious groups, our analysis does not offer general support for the suggestion that kin marriage among British Pakistani Muslims limits the educational capital gained through transnational marriage compared to the situation for British Indian Sikhs.British Indian Sikh women are the only group for whom transnational marriage more frequently leads to a higher educated spouse.This is likely to be at least in part due to higher population levels of education in India compared to Pakistan, and of men compared to women within both countries.
We do find differences between the two groups in terms of gendered patterns of employment postmigration.Whilst employment is low among migrant wives in both groups, absolute levels are lower, and the difference between migrant and non-migrant wives employment is greater, in the Pakistani group.Although there is evidence of labour market discrimination against Muslim women (Ghumman and Ryan, 2013), differing expectations of women's employment between the two groups seem likely to be a key explanation here, with Indian Sikh families more often viewing paid work as compatible with domestic gender roles (Charsley et al., 2016).These findings show the importance of expanding studies of transnational marriage to non-Muslim groups.
Gender differences are also strongly in evidence in employment between migrant husbands and migrant wives in both groups-where lower employment among wives contrasts with high levels of employment among husbands from the subcontinent.Given the expectations of a migrant penalty in the labour market (Kogan, 2004;Ferna ´ndez and Ortega, 2008;Kanas and Van Tubergen, 2009), this result suggests both a gendered emphasis on employment for migrant men (who often carry a double responsibility to contribute to household incomes both in the United Kingdom and 'back home' in Pakistan- Charsley, 2013), but also the potential advantage for spousal migrants of their marital connections to a local family whose networks may provide employment opportunities, overcoming some conventional barriers to migrant employment.The fact that both male and female migrant spouses are more likely than UK-born/raised coethnics to find low-level employment, however, suggests that these opportunities are more likely to be for lower skilled (and likely lower paid) employment (cf.Charsley, 2013).
Like Dale and Ahmed (2011), we find no evidence that being in a transnational couple has a negative effect on the labour market participation of British South Asian women.This too counters a common assumption in policy debates in both the United Kingdom and mainland Europe.
While the UK LFS data used for the analyses have advantages for this kind of analysis, such as a large sample size, there are a number of limitations. 6The lack of information on foreign qualifications in the LFS led us to collapse the education variable into three categories and impute values, which may have resulted in overestimation of the share of homogamous marriages (Blossfeld, 2009).Some of our results also proved sensitive to imputation.Since 2011, data on foreign qualifications has improved.It would therefore be worthwhile to repeat our analyses, as more post-2011 waves become available.Analyses of later waves would furthermore allow further investigation into differences between

Figure 1 .
Figure 1.Predicted probability (AMEs) with 95 per cent confidence interval of type of educational match for UK-born/raised by couple type.Control variables: foreign-born, year of birth, and survey year.

Figure 2 .
Figure 2. Predicted probabilities (AMEs) with 95 per cent confidence interval of employment of partner by couple type and partner education.Control variables: age of partner, presence of dependent children <12 years, and survey year (dummies).Note: Estimates based on models presented in Online TableA5.Source: UK LFS.

Figure 3 .
Figure 3. Predicted probability (AMEs) with 95 per cent confidence interval of partner being in a low-level occupation (only of those employed) by couple type and partner education.Control variables: age of partner, presence of dependent children <12 years, and survey year (dummies).Note: Estimates based on models presented in Online TableA8.Source: UK LFS.
suggests that women in newly arrived immigrant couples participate temporarily in the labour market while their husbands invest in acquiring new skills.A similar scenario could apply European Sociological Review, 2019, Vol.35, No. 1 Downloaded from https://academic.oup.com/esr/article-abstract/35/1/133/5116188 by University Library, University of Amsterdam user on 19 March 2019

Table A5 .
Source: UK LFS.European Sociological Review, 2019, Vol.35, No. 1 Downloaded from https://academic.oup.com/esr/article-abstract/35/1/133/5116188 by University Library, University of Amsterdam user on 19 March 2019 post-secondary education has a similar chance of being in low-level employment to non-migrant spouses without post-secondary education.
-level occupation (only of those employed) by couple type and partner education.Control variables: age of partner, presence of dependent children <12 years, and survey year (dummies).Note: Estimates based on models presented in Online TableA8.Source: UK LFS.
Predicted probabilities (AMEs) with 95 per cent CI of employment for UK-born/raised by couple type.Control variables: foreign-born, age, education, presence of dependent children <12 years, and survey year (dummies).Note: Estimates based on models presented in Online Table A10.Source: UK LFS.European Sociological Review, 2019, Vol.35, No. 1 Downloaded from https://academic.oup.com/esr/article-abstract/35/1/133/5116188 by University Library, University of Amsterdam user on 19 March 2019