Gender, immigration, and ethnicity

Abstract

We complement analyses of both gender and ethnic inequality by examining cleavages among women of different ethnic and migrant groups, with a particular focus on labour force participation. Such differences can be shown to be shaped by gender attitudes and norms, which are relatively slow changing, transmitted intergenerationally, and socially embedded. We also examine the benefits of taking a household and individual income perspective on economic inequalities, extending existing analysis of earnings and employment. This facilitates the investigation of the economic position of women both in and out of the labour market, and how their income relates to the resources of other family members. By these means we can shed light on the ways in which gendered processes that are common across ethnic groups can nevertheless result in differentiated outcomes that may persist across generations. This can enhance our understanding of persistence and change in gendered inequalities more widely.

Introduction

In their article in this collection on gender inequality, Andrew et al. (2024) highlight heterogeneity in women’s economic outcomes, focusing primarily on level of education. Participation, hours and earnings are all greater for more highly educated women, meaning that as education levels have increased among women, the total earnings gap between men and women has reduced. But nevertheless, earnings gaps persist. Moreover, as a result of their higher earnings potential, the impact of starting a family is relatively greater for more highly educated women, and wage gaps between more highly educated men and women have grown while those between less educated men and women have converged, due in part to the declining earnings of lower educated men. Their analysis points to the failure of arguments around specialization to explain the differential division of labour; and instead, they highlight the importance of norms in maintaining gendered inequalities in the division of paid and unpaid work, which drives much of the lifetime inequality gaps.

At the same time, Dustmann, Kastis and Preston (2024), in their article on immigration and inequality in this collection, note that there is greater earnings inequality among immigrants than among the native-born population, highlighting some of the diversity among immigrants, stemming in part from differences in cohorts and types of migration. This diversity in economic outcomes is also picked up in the article by Mirza and Warwick in this collection: they chart differences in education, employment, income and wealth across different ethnic groups, often distinguishing the first (immigrant) from the second (UK-born) generation, and revealing both convergence and persistent differences.

In this article, we aim, first, to add to the understanding of heterogeneity in women’s outcomes by highlighting some of the cleavages that operate across and among women of different ethnic groups. In doing this, we draw attention not only to some of the inequalities that exist between women, but also to the differences in the inequalities between men and womenfor different ethnic groups. Perhaps, more importantly, we also highlight the commonality of many of the processes and mechanisms that drive gendered inequalities across groups. The issues highlighted by Andrew et al.—educational attainment, family formation, differences in the domestic division of labour, gender norms and their national prevalence and intergenerational transmission, and the structure of policies—are relevant across ethnic groups, majority and minority, and across immigrants and native-born. We argue that the different outcomes we observe are consequent predominantly on differences in initial starting positions. For example, having a mother who was in paid work increases the chances of a woman maintaining paid work once she has a partner and family, and of a man facilitating his partner’s labour market participation. But women—and men—from different ethnic groups have different chances of having had a mother in paid work (Arcarons 2020). By focusing on the ways in which the consequences of differences in these common processes result in heterogeneous experiences across ethnic groups, we highlight the potential of such differentiated analysis for helping to clarify the relative role of the different factors, which we raise as a valuable area for further research.

Second, we aim to enhance understanding of immigrant and ethnic inequalities by focusing on the ways these are differentiated by sex, but also how—as also flagged in the commentary by Diehl (2024)—they are shaped by gendered processes, which play out differently in part according to origin country characteristics and patterns of migration. A focus on wages and wage inequalities across immigrants and between immigrants and native-born neglects, by definition, the experience of those not in paid work, even though the probability of being in paid work is, of course, highly gendered. This is especially so for immigrants, where visa statuses (such as family related or labour related) are differentially distributed across the sexes and are associated with particular relationships to paid work. The extent to which wages—the focus of Dustmann et al.’s analysis—translates into economic position is of course contingent not only on whether an individual is in paid work in the first place, but also on what others in the household or family unit do or do not earn. Household income inequalities may be greater or lesser than wage inequalities—or operate differently where safety nets are differentially accessible. At the same time, household incomes themselves disguise the extent to which income is or is not equitably distributed and who controls it. Thus, as urged by Bennett (2024) in her commentary in this collection, it remains important to consider not only men’s and women’s earnings and their household incomes, but also to evaluate individual incomes (including of those with zero earnings) alongside household income position if we are to understand the extent to which there is economic autonomy of individuals, and of partners in couples. She notes ‘the centrality of both individual financial autonomy and interdependence and caring to gender analysis.’ Paying attention to both autonomy and interdependence can be informed by considering the different ways the relationship between individual and household incomes play out across different groups as well as enhancing our understanding of inequalities faced by women and men. Autonomy and interdependence may be particularly relevant considerations for immigrants constrained by immigrant status that can limit their access to work or make residence conditional on a partner’s sponsorship.¹ Even for those with free movement, economic autonomy can be hampered by the tying of mobility to worker status. Shutes and Walker (2018), for example, discuss how EU citizens’ mobility rights in the UK were contingent on their status as workers, either as self-sufficient individuals or as family members of such workers, with gendered implications for their residence rights and access to social rights.

In bringing an ethnically differentiated perspective to gender inequalities and a gender perspective to immigrant inequalities, we aim to enrich understanding of both gender and immigrant/ethnic inequalities in the UK. In both cases, we draw attention not only to the ways in which economic processes operate at the household level, but also to the limits to a household perspective.

We use as our starting point one of the most-discussed aspects of inequality among women—differences in labour force participation, both among immigrants and the UK-born of different national origins/ethnic groups. We consider the various individual or family characteristics that contribute to these differences and why they might, nevertheless, not fully ‘account’ for raw gaps, as evidenced by the article by Mirza and Warwick in this collection. This leads us into a discussion of the role of partners and their circumstances, children, and parents (via intergenerational transmission) in shaping differentiated labour market outcomes across ethnic groups. We also consider the economic context in which decision-making arises, and why it is important to consider both individual and household income. We conclude that there remains substantial potential in future research to further illuminate gender inequalities by paying attention to those of migrant origin. Similarly, analysis of immigrant inequalities can benefit from further scrutiny of some of the paradoxes introduced by a gendered analysis, such as the fact that women have both better and worse labour market outcomes than men of the same ethnic group, depending on their specific origins, and that women’s labour market—but not educational—outcomes appear to reflect gendered norms.

Investigating differences in labour force participation

A great deal of academic and policy attention has been paid to the differences in labour force participation rates across ethnic groups, in particular the markedly lower participation rates of Pakistani and Bangladeshi women. For example, ‘Ethnicity facts and figures’ for 2021 showed that White British women aged 16–64 had participation (economic activity) rates of 76% compared to 49% among Pakistani and Bangladeshi women (Office for National Statistics 2022). At the same time, Black Caribbean women typically have high participation rates, higher than their White British counterparts (Khoudja and Platt 2018; Mirza and Warwick 2024). For example, the article by Mirza and Warwick in this collection shows that Black Caribbean women’s participation rates were around 2 percentage points higher than their White British counterparts, while Pakistani and Bangladeshi women’s participation rates were over 30 percentage points lower. Meanwhile, in their article, Andrew et al. highlight the relevance of women’s educational level in shaping overall earnings via participation, as well as hours and rates of pay. Part of what accounts for ethnic differences in participation is indeed educational levels, with average levels of education among immigrant Pakistani and Bangladeshi women typically low: as Mirza and Warwick show, more than a quarter have no qualifications and they have lower rates of degree holding than any other ethnic or immigrant groups. However, this is not the full story. Lower participation continues into the second generation, despite the increasing levels of education among second-generation Muslim women. Indeed, Diehl (2024) notes that this marks a difference from the picture in Germany, where lower participation in the second generation among those from Muslim majority countries largely follows on from lower educational attainment. It is, indeed, the case that participation rates are substantially higher among more highly educated compared to less highly educated second-generation Pakistani and Bangladeshi women, resulting in a bigger educational gap than for women from other ethnic groups (Zuccotti and Platt 2023). But participation is still lower among those both with and without degree-level qualifications than for other majority and minority groups. Controlling for a range of factors, including those that tend to be endogenous, such as family size and broad occupation, does not entirely ‘explain away’ these differences, as Mirza and Warwick’s analysis reveals.

One commonplace interpretation is to simply attribute these remaining gaps to ‘cultural’ differences, which are themselves often elided with Muslim religiosity. However, this typically adds little to our understanding of the processes of gendered inequalities and tends, instead, to reinforce the idea of immutable group differences. This is not to say that culture does not matter for women’s labour market outcomes across first and second generations. Indeed, there is a body of work that demonstrates the role of culture in women’s labour market and fertility outcomes using the separation of immigrants from their institutional context to identify the causal role of culture (e.g. Fernández and Fogli 2009; Polavieja 2015; Finseraas and Kotsadam 2017). As these authors point out, such epidemiological approaches to the analysis of culture have great potential not only in demonstrating diversity in the economic outcomes of women from immigrant origins (typically second generation), but also in using the movement of migrants from their original settings to separate out the roles of beliefs and traits of individuals and others from their social context. After all, despite the way it can sometimes be presented, culture is not the preserve of those of immigrant origins—we are all embedded in cultural repertoires which inform our actions as well as our understanding of the world (Geertz 1973). Migration research thus has applicability to processes that are relevant to the population at large and of interest for labour market outcomes of all women. At the same time, these studies do not speak directly to the UK context, which has some distinctive features in both its immigrant composition as well as in institutional features and immigration policies that are likely to have a bearing on who migrates and how far women engage in paid work. Moreover, there is still much to be understood about how culture is transmitted, how it may jointly influence family and labour market behaviours and the implications of that for how we model labour market outcomes. Cultural differences are also relevant to helping us understand some of the heterogeneous outcomes or paradoxes that we observe in the experience of the UK’s ethnic minority women—particularly in terms of their educational success—and how these interact with institutional environment, such as the educational system (Jackson et al. 2012; Zuccotti and Platt 2023).

Focusing specifically on the UK case, early work by Angela Dale and others began to shed light on the noted gaps in participation of Pakistani and Bangladeshi women as a phenomenon worthy of sustained attention. They explored the role of expectations of young women themselves and others close to them in terms of work and family life, as well as the obstacles presented by high unemployment rates, in shaping participation decisions. Dale (2002) used both qualitative and quantitative data to explore these issues for Pakistani women, recognizing the turn of the millennium as a time of change, when an increasing number of young Pakistani women were continuing in education but it was still unusual for them to have tertiary qualifications. Her respondents discussed the different constraints and expectations they faced, and she noted that family circumstances and educational level influenced employment in the same way for White British and Pakistani women, but with the effects being much more pronounced (i.e. a greater reduction in participation consequent on partner and children and a greater gap between those with and without qualifications) for Pakistani women. She expected alignment in these effects over time. In further analysis, Dale et al. (2006) explored the role of family structure and life-course stage on participation across a number of ethnic groups. The authors identified that partnership was associated with higher participation for both White and Black Caribbean women but with lower participation for South Asian women, while Black Caribbean lone parents were substantially more likely to be in work than White British lone parents. Educational gradients were present across the board, but were steepest for Pakistani and Bangladeshi women. For those young women who were single and without children, activity rates were high across the board implying commonality of experience given equivalent circumstances. But of course, circumstances did vary systematically; and the authors’ expectations of generational change across groups were not well supported, suggesting limits to assumptions of simple adaptation. Somewhat in contrast to the earlier work, Dale et al. thus concluded that differences in participation could not be attributed simply to differences in family circumstances—instead implying group-specific explanations. However, the expectations of generational change did not take account of the very different starting positions in terms of potential intergenerational transmission: for example, few of the Pakistani women’s mothers would have been working and most Black Caribbean women would have had working mothers.

Twenty years later, Mirza and Warwick also find that while all women with a child under 16 in the household had lower employment than their non-parental counterparts, partnership was again associated with higher employment rates among White British and more especially Black Caribbean women, while it was associated with substantially lower employment rates among Indian, Pakistani and Bangladeshi women, other things being equal. As the authors caution, however, rates of cohabiting partnership are themselves very different across groups—it is the norm for Pakistani and Bangladeshi women and the exception for Black Caribbean women, meaning that those in the different situations are likely to be different in other ways. It also means that decisions on economic activity contingent on partnership are unlikely to be directly separable from the partnership decision itself. Moreover, while Mirza and Warwick control for UK-born status, they do not examine the immigrant generation separately from the second generation, with the former more likely to face specific constraints on employment, ranging from English language fluency to visa conditions, or differential (and more limited) access to labour market opportunities, with greater concentration in ‘niche’ employment (Platt 2021), partly fostered by greater geographical concentration. These constraints are themselves differentiated across national origins, where those from different countries of origin and cohorts differ in degree of selectivity, colonial connections, familiarity with English as a first language, and operation of immigration and citizenship laws.

The role of gender attitudes and norms

Gender norms clearly shape the patterns of participation of women in general, and one potential influence linked to differences in origin cultures of immigrants that has received attention has been the role of individuals’ gender attitudes in contributing to differential outcomes. There are differences in more or less traditional views across ethnic groups that have been tied to differences in attitudes in countries of origin. As noted, work on the role of culture has seen cultural influence as working through gender norms, whether at the group, individual or partner level, though without directly measuring gender role attitudes themselves (e.g. Fernández and Fogli 2009; Polavieja 2015). However, UK studies do provide direct evidence of differences at the individual level in gender role attitudes. Specifically, measures show that Indian, Pakistani, Bangladeshi and Black African women tend to have more traditional views than White British women, while Black Caribbean women tend to have less traditional views on gender roles (Wang and Coulter 2019; Zuccotti 2021). Somewhat consistent with these attitudes, Mirza and Warwick show that Black Caribbean women’s labour market participation rates are higher, whereas those of Indian, Pakistani, Bangladeshi and Black African women are lower—though, as noted, the gap is substantially greater for Pakistani and Bangladeshi women. Trying to determine the relevance of gender norms for labour market outcomes and focusing on those held by women themselves, Khoudja and Platt (2018) looked at the role of religiosity, gender role attitudes and life events in shaping (differences in) women’s entries into and exits from economic activity. They showed distinct patterns of entry into and exit from the labour market across ethnic groups, with Pakistani and Bangladeshi women not only having lower entry rates but also higher exit rates than their White British counterparts. Black African women, by contrast, had higher entry rates even after a range of controls were taken into account. Religiosity was not however associated with any of these differences. Instead, the more traditional gender role attitudes that are typically associated with religiosity across affiliations contributed to explaining some of the differences between Pakistani and Bangladeshi women and White British women. Interestingly, across groups, these attitudes were associated with exits as much as entries, suggesting that more traditional attitudes may make women more responsive to circumstances under which opting out of the labour market becomes an option as well as inhibiting participation in the first place. Overall, while gender attitudes cannot explain all the differences in participation across groups, these analyses suggest that they do matter for women’s participation.

Individual attitudes of course may not capture all the ways in which gender norms facilitate or constrain economic activity—and the type of economic activity participated in. As noted by Andrew et al. (2024) in this collection, gender norms are held by other members of society—partners, parents, reference communities—such that these may influence behaviours as well as or independently of individual attitudes. As a result, individual gender role attitudes are not necessarily consistent with behaviour—and behaviour may be as, or more, important in perpetuating or undermining gender norms. This has been demonstrated by Platt and Polavieja (2016) in relation to intergenerational transmission of more or less traditional attitudes for the population as a whole. Two implications of their study, which are relevant for participation differences across ethnic groups, are the role of maternal employment and of housework shares in shaping attitudes and consequent behaviours.

If women from a certain ethnic group are less likely to have had mothers who were in paid employment, then we would expect them to be more likely to be economically inactive themselves, and vice versa. Analyses of labour market differences for immigrant and ethnic groups are rarely able to draw on information about maternal employment in considering women’s economic outcomes. However, social class background, which is also likely to shape opportunities and preferences, is increasingly being taken into account (e.g. Li and Heath 2016; Zuccotti and Platt 2023); see also the commentaries by van der Erve et al. (2024) and Lambert (2024) in this collection. Indeed, women with lower educational attainment (who are more likely to have had mothers not in paid work) have, as Andrew et al. in their article show, a greater probability of not being in paid work themselves, compared with higher educated women. Thus, disentangling the economic constraints they face from intergenerational transmission of expectations and behaviours is not necessarily straightforward. Nevertheless, innovative work by Arcarons (2020) has illustrated the importance of socialization of both women and their partners. He shows that maternal employment can play a role in explaining ethnic differences in women’s participation, but one that seems to operate through the transmission of educational expectations and gender role attitudes. However, in addition, he shows that having a working mother-in-law has a positive effect on women’s likelihood of participation, such that for Bangladeshi and Pakistani women who had a mother-in-law who participated in paid work, there is no remaining participation gap between them and other partnered women, other things being equal. While he cannot fully disentangle whether this is a selection effect (women who want to work select husbands who had working mothers) or a direct influence from the mother-in-law to the husband (men who grew up with a working mother expect or facilitate their wife’s labour market participation), it demonstrates the enhanced understanding that can be gained from considering general processes across the population—such as explored by Fernández et al. (2004) and Farré and Vella (2013)—and applying them to help understanding of ‘unexplained’ ethnic group inequalities. Families and partners matter. Without implying any determinism or neglecting the fact that migrants are selected in ways that may cause them to differ from the average in their countries of origin, those families and partners still cannot be divorced from the circumstances under which they were raised. That is, there are likely to be remaining influences from the norms and expected roles—and corresponding behaviours—in their countries of origin, even as those norms themselves change at origin, and even if migrants’ attitudes adapt and/or differ from those who do not migrate (Guveli et al. 2015). They are also likely to be shaped by the routes by which they migrated and the constraints that such migration placed on their possibilities for labour market participation.

As noted by Dustmann et al. (2024), in their article in this collection, family reunification is associated with lower earnings potential, and indeed lower participation overall. Family reunification has typically been a highly gendered migration route—with immigration rules initially favouring overseas wives and only subsequently amended to treat overseas husbands similarly. Up to its abolition in 1997, the ‘primary purpose’ rule created a form of ‘Catch-22’ for those wishing to join a partner in the UK, which was particularly likely to penalize male applicants, given immigration officers’ assumptions about cultural practices (i.e. that wives should join their husband’s household rather than vice versa). Indeed, it was originally targeted at husbands specifically. Women have continued to dominate admissions via the family route or as dependants; and among the current population, those who migrate via this route, particularly women, continue to have poorer labour market outcomes than those migrating for work (Home Office 2011; Cooper et al. 2014; Walsh 2023). Of course, motives for migration are not straightforwardly separable, and work such as that by González-Ferrer (2011) has highlighted the ways in which women’s family and economic motivations can go hand in hand. Nevertheless, it remains the case that the mothers of the UK’s second generation who entered the UK via family reunification are less likely to have entered the labour market. Focusing on the UK, Ersanilli and Charsley (2019) have explored the extent to which male partners also join spouses in the UK, particularly for those groups where ethnic homogamy is the norm, such as Pakistani Muslims and Indian Sikhs. They show that the immigrant wives they covered in their study were less likely to be employed than immigrant husbands, or than wives sponsoring immigrant husbands. Such gendered migration patterns, then, can have consequences down the generations and at the level of communities. Conversely, where women migrated primarily independently and as workers—for example, those from the Caribbean who contributed to filling post-war labour demand in the health service and elsewhere—their employment status is also likely to have influenced their children through processes of intergenerational transmission. The historically markedly higher employment rates among Black Caribbean lone mothers compared with White British mothers also arguably reflects similar processes, and these processes do not need to be subsumed into larger historical–cultural narratives to account for the patterns we continue to see today.

As well as maternal work status, Platt and Polavieja (2016) also identified the domestic division of housework as a further behavioural influence on gender role attitudes and gendered behaviours down the generation. Andrew et al. (2024) in this collection describe the unequal distribution of unpaid work between men and women, as well as its variation cross-nationally. It is thus perhaps unsurprising to find that there is variation across those of different ethnic groups in the distribution of housework, variation that operates alongside—and may also contribute to—the charted differences in patterns of paid work. Kan and Laurie (2018) have described both differences in hours of housework undertaken across ethnic groups and differences in the shares undertaken by women with partners. All groups of women do a larger share of the housework than their partners on average, though higher levels of education and being employed reduce that share, while having an employed partner, children and more traditional attitudes towards gender roles increase it. But even net of these effects, Indian and Pakistani women do relatively larger shares of housework than their White British counterparts, suggesting that there are additional factors—and factors beyond individual attitudes to gender roles, which do not account for differences in housework hours—that shape the intra-household division of labour. While such models control for employment status, clearly there is likely to be endogeneity between expectations around housework and labour market participation and hours worked, both in terms of norms and practical constraints, which cannot be fully captured in this way. Analysis that can better identify the constraints that housework places on labour force participation, or that can jointly model decision-making around both, would make a valuable contribution to our understanding (see Bryan and Sevilla-Sanz 2011). But given that we know that there is intergenerational transmission of gendered behaviours as well as attitudes (Platt and Polavieja 2016), this evidence is consistent with such long-run transmitted effects. It also demonstrates the challenges in avoiding ongoing perpetuation of gendered behaviours and the consequences for the division of paid and unpaid work more generally for men and women in the UK.

One reason why the relationship between behaviours and individual attitudes is not necessarily consistent is because reference community norms may differ from personal beliefs, through practical constraints or necessity. Within a single country, such ‘horizontal’ transmission is hard to capture effectively, though Andrew et al. show that across countries there is a clear correlation between gendered attitudes and long-run child penalties in earnings. Within countries, it is possible to use the local concentration of those of different national or ethnic origins to proxy for the role of wider community influences on behaviours through social sanctions or support. Fernández and Fogli (2009), for example, showed the additional influence of larger neighbourhood shares of those from their country of ancestry on second-generation women’s participation in the US. Similarly, within the UK the impact of extra-familial influences can, to some extent, be inferred through co-ethnic concentration. At the same time, we can still recognize the wider contextual and institutional influences that help to explain the unsurprising fact that the gender role attitudes of those groups with more traditional views tend to become less traditional across the generations (Wang and Coulter 2019), reflecting both cohort change and the attitudes of the wider population surrounding the second generation.

Zuccotti and Platt (2017) found that greater levels of co-ethnic concentration were associated with poorer employment outcomes for Pakistani and Bangladeshi women, in terms of both lower participation and worse occupational outcomes for those in work. They suggested that greater opportunities for interactions between members of an ethnic group fostering greater maintenance of more traditional gender norms could be the mechanism behind these findings. In subsequent work, Wang and Coulter (2019) and Zuccotti (2021) tested whether co-ethnic concentration was associated with more traditional attitudes to gender roles, which could imply greater constraints on behaviour. While Wang and Coulter suggested that greater co-ethnic concentration was associated with more traditional gender role attitudes for Pakistanis, Bangladeshis and Indians, and to a lesser extent Black Africans, Zuccotti only found a concentration effect for Bangladeshis and Indians. Moreover, her analysis suggested that the effect for Bangladeshis could be attributable, at least in part, to selection, that is, less traditional Bangladeshis being more likely to live outside areas of co-ethnic concentration. Nevertheless, to the extent that people with more traditional attitudes may tend to co-locate, this may not only help them to maintain their views and behaviours but also have consequences for those around them.

Of course, both location and occupational decisions are not independent of other actors. In a context where, as noted by Mirza and Warwick, there is substantial evidence for prejudice and discrimination, the literature has also explored how ethnic concentration can act as a resource for minorities, even as it can also limit opportunities (Knies et al. 2016; Zuccotti 2019). Conversely, moving out of such areas can enhance economic options, which can bring enhanced exposure to discrimination. For women, these trade-offs may be further compounded if accompanied by greater expectations to comply with more traditional gender norms.

This mixed evidence is clearly inferential and suffers from the challenges of assuming interactions from a measure of shares of different groups in a particular area; but greater insight into the role of peer and reference group attitudes and behaviours in the maintenance of gender norms could be informative not just in terms of helping to account for ethnic differences in women’s economic outcomes (including, potentially, the more positive effects on educational attainment) but also more widely. It could help us to understand further the stubborn persistence of gendered expectations relating to the division of labour at work and home, as well as, potentially, the possibilities for change. This is likely to be a fruitful area for further research.

In sum, research has demonstrated that attending to the role of norms and attitudes and their intergenerational transmission helps to shed light on the inequalities faced by ethnic minority women in particular. The evidence indicates both that norms and attitudes are malleable and changing over time, and that there is intergenerational transmission that affects both attitudes and behaviours. There are also important feedback loops between attitudes and behaviours of individuals and communities, with longer-term consequences for labour market outcomes than we might expect if we focused only on individual and family characteristics as drivers of inequalities.

At the same time, a focus on processes of intergenerational transmission brings its own challenges to interpretation of outcomes across multiple areas of the lives of the second generation. If low or high maternal (and mother-in-law) labour market participation has consequences for employment outcomes, it might also be expected to affect girls’ educational attainment. And for minority groups, this does not appear to be the case. Girls’ educational attainment does not vary systematically with the employment levels of women of the immigrant generation of their ethnic group. The ways in which immigrants are selected, occupational downgrading among immigrants, at least early in their working careers in the UK (Dustmann et al., 2024), including among marriage migrants (Ersanilli and Charsley 2019), and the motivation for the attainment of the second generation as a key reason for migration are all helpful in explaining the educational ‘overperformance’ that is extensively documented by Mirza and Warwick (2024) in this collection. Nevertheless, it still leaves the puzzle of the mismatch between differing gendered norms across ethnic groups, which appear to influence labour force participation—and potentially occupational choices as well—but which do not correlate with educational attainment. If we could better understand the factors moderating or reinforcing traditional gender norms across ethnic groups and how they play out at different stages, encompassing educational and occupational decision-making as well as employment, we might have better insight into how they operate for the population as whole. The topic is certainly worth further sustained attention.

Economic context and influences

As Andrew et al. note, economic position is insufficient to account for the gendered inequalities they document, though this not to say—and nor do they suggest—that economic context does not matter for earnings gaps. More-educated women are more likely to be in paid work, for more hours and at higher rates of pay. While the specialization thesis clearly lacks support (see also Evertsson et al. 2023), higher earnings potential does change intra-household bargaining position and also informs options and decisions about the benefits of and disincentives to full- and part-time paid work. The amount of domestic labour to be distributed can also differ depending on circumstances. Not only do children clearly increase it, but, as Kan and Laurie (2018) showed, lower household income was associated with relatively more housework conducted by men and women as well as with a higher share conducted by women. The policy context that shapes options and their economic returns is also relevant—for example, through childcare provision and tax and benefit policies which implicitly discourage a second earner. It is therefore important not to lose sight in this discussion of the very different economic context faced by different women and how that is relevant to household participation decisions of different ethnic groups. For example, where earnings potential is low (and in the face of high unemployment rates, such as those very high rates faced by economically active Pakistani women), it may not make much economic sense for a woman, particularly one with family responsibilities, to take up—or attempt to take up—paid work. In the face of high marginal taxation rates, low partner earnings can disincentivize as much as incentivize the take-up of paid work. Conversely, the opportunity to work part-time (or not at all) may be facilitated (or incentivized) by a partner’s higher earnings. On the one hand, an individual’s low or no earnings might be comparable in these two situations, suggesting low economic autonomy; on the other hand, the level of household income is nevertheless highly germane for overall well-being. As Bennett (2024) in this collection argues, when considering incomes, providing insight into individual as well as household incomes may be especially important for understanding autonomy.

In earlier work, Nandi and Platt (2010) analysed concurrently the household and individual incomes of women from different ethnic groups, shedding light also on the extent to which there was within-group inequality and the implications also for how we might think about ‘group’ experience. Mirza and Warwick provide more up-to-date information on household income—though by ethnicity of household head rather than for the men and women living in those households—revealing similar patterns of highly compressed incomes among Pakistani and Bangladeshi households, mostly concentrated at very low levels; somewhat higher and somewhat more dispersed incomes among Black Caribbean and Black African households; and higher, more unequal incomes among White British, Indian and Chinese households. The particularly large dispersion among Chinese households could stem from very different origins and trajectories among this overarching group label (Mok and Platt 2020). Overall, as Dustmann et al. find for immigrants’ wages, inequality among the incomes of ethnic minorities is greater than inequality among the White British population—though this does not have a marked effect on overall inequality. Nandi and Platt (2010) used individual income data to show that the ranking of average incomes across ethnic groups differed according to whether an individual or a household measure was used—giving a stylized representation of the combinations of financial autonomy and financial well-being that different women might be considered to experience. The focus on income, rather than only earnings, is important because of the non-negligible role of taxes and transfers in shaping overall economic welfare. At the individual level, income also includes those without labour market earnings, as well as those with self-employment incomes, which form a particularly important component of income for some groups (Platt and Warwick 2020). Figure 1 compares the rankings of individual income for men and women by ethnic group. As it makes clear, there is not a consistent pattern across groups between men and women—relative economic autonomy for one sex does not necessarily correspond to relative economic autonomy for the other of the same group, cautioning about how we understand the position of ethnic ‘groups’ and the consistency of such group experience. At the same time, those women with relatively strong individual income positions may nevertheless not be in a position to translate this into high household incomes, depending on their household circumstances.

One way of stylizing this relationship between autonomy and economic welfare, and how it differs between men and women and across ethnic groups, is to consider household and individual income concurrently. That is, to distinguish between those who have below average household and individual income, so have neither economic autonomy nor economic well-being; those who have below average household but above average individual income, where they have economic autonomy but support others; those who have above average household income even though they have low individual income, giving economic security but on the basis of household rather than individual income; or those with higher individual income, where economic autonomy and economic well-being might be considered to go hand in hand.

Figure 2 shows such a representation by ethnic group for men and women, split additionally by whether or not they are UK-born. Overall, around a third of individuals have lower household and individual incomes and around a third have higher individual and household incomes, illustrating the ways individual and household incomes—or those of individuals and other household members—do tend to co-vary. The remaining third is split between the other two combinations. But there is clearly considerable variation across groups. In general, women have a somewhat greater chance of living in a lower-income household (blue and red bars) than men of the same ethnic group. But men tend to be more likely than women to live in a poorer household despite a higher individual income (red bars), that is, as a result of also supporting family or household. Women conversely are more likely to have a higher household income despite lower individual income (green bars). The higher proportions of women in this latter situation could be seen as reflecting a more ‘traditional’ household form, where a woman’s limited hours or returns to her work are facilitated by her partner’s income. This pattern is, though, reversed for Black Caribbean and Black African men and women, suggesting the greater importance of women’s incomes to the overall household in these cases, which aligns with the greater numbers of single women and lone-mother households among these groups. It is worth noting that the share of White British women with lower individual income but higher household income (green bars) is greater than the share for most other UK-born groups, while the share where both incomes are higher (yellow bars) is smaller than for other UK-born, apart from Pakistani and Bangladeshi women. This reinforces the point made elsewhere (see, e.g. Khoudja and Platt 2018) that consideration of gendered inequalities across minority ethnic groups also needs to recognize the extent of gendered beliefs and behaviours across the wider population: immigrants are not entering a ‘neutral’ context in terms of gendered expectations and their economic implications.

This simple illustration is intended to draw attention to the ways that individual earnings potential and income constraints interact not only with family circumstances, other sources of income, and policy incentives and disincentives, but also with the earnings and incomes of other household members to shape both decisions and overall economic well-being. At the same time, household incomes are not necessarily enjoyed equally. The chances of having a low income differ for men and women across ethnic groups, as do the degrees of control over income. Both low income and autonomy may have implications for subsequent generations, not only in terms of the known risks of poverty for later-life outcomes and the reduced opportunities associated with more limited resources, but also in terms of exposure to more or less traditional forms of the division of labour.

In conclusion, the experience of immigrant and ethnic minority women cannot be straightforwardly mapped on to that of all immigrants. Nor is it fully consistent with the gendered economic landscape experienced by women in general in the UK. While both Dustmann et al. and Andrew et al. in their articles in this collection tell us a lot that is relevant to immigrant women’s experiences, and while the article by Mirza and Warwick illustrates many of the distinct experiences of women from different ethnic groups, an analysis of immigrant inequalities that fully encompasses the experience of women and an analysis of gendered inequalities that engages with ethnic differences have the potential to provide a more complete account.

Conclusion

In this article, we have shown the value of attending to the role of norms and attitudes in a way that emphasizes them as common processes when analysing ethnic inequalities in women’s outcomes, while also acknowledging the importance of economic circumstances for the perpetuation of inequality—and the relationship between the two. The evidence indicates that norms and attitudes are malleable, and that for those groups with more traditional attitudes, there is intergenerational change towards the level of the overall population. Nevertheless, differences remain. It is clear that intergenerational transmission of labour market behaviour and attitudes (which are necessarily linked) contributes to the continuing divergence among women of different ethnic groups in both attitudes and behaviours across the generations—both for those with more traditional attitudes and for those with less traditional attitudes. It is also evident that a focus on women’s own attitudes alone can provide some explanation—but only an incomplete one—of labour market outcomes; the attitudes of partners and community also act as enabling or constraining. While we have not focused on these in this article, the expectations and attitudes of others, such as employers, teachers and wider society, also play a role in sustaining women’s unequal economic situation. We have shown how more or less traditional gender norms can foster particular divisions of paid and unpaid work, and are associated with patterns of residential and occupational concentration, which have both immediate economic consequences and themselves feed into the expectations and decisions of subsequent generations. The result is that these norms and feedback loops can help account for unequal labour market outcomes to a greater degree than when the focus is only on individual and family characteristics as drivers of inequality. It might be argued that the degree of change that we do observe across the generations, particularly in terms of women’s educational attainment, which is of course then important for other elements of their lives, is perhaps as much to be remarked as the persistent differences in economic inequalities, important though they are to women’s and their families’ well-being. Given the relatively high and unchanging prevalence of more traditional attitudes in the UK as a whole, as noted in the article by Andrew et al. (2024) in this collection, the conditions under which such change arises can be considered more widely informative.

Attention to inequalities at the intersection of gender and ethnicity, by capitalizing on the differences in women’s experiences, can thus help to further unravel some of the ways in which gendered norms are perpetuated or change, and the different influences on these. It could additionally shed further light on transmission processes and the relative influence of different forms of modelled behaviour and expressed beliefs on children’s educational, occupational and family aspirations and attainment. Polavieja and Platt (2014) illustrated how the extent of parents’ gendered occupations can influence the gendered occupational aspirations of their children. Given the continuing distinctiveness of the occupational distributions of men and women, alongside the existence of over-representation of certain ethnic groups in particular occupations, even across the generations, there is clearly potential to disentangle how these distinctive patterns arise and the extent to which influences on the next generation of men and women of different ethnic groups are consistent or distinctive.

ACKNOWLEDGEMENTS

We are very grateful to Rob Joyce for his feedback on an earlier version of this article.

FUNDING

Alita Nandi is funded by the ESRC, Understanding Society Waves 13-15 grant, ES/T002611/1. This article was written for the IFS Deaton Review of Inequalities, funded by the Nuffield Foundation (grant reference WEL/43603). The views expressed are those of the authors and not necessarily the Foundation.

Footnotes

REFERENCES