The female-breadwinner well-being ‘penalty’: differences by men’s (un)employment and country

This article examines the relationship between female breadwinning and life satisfaction in heterosexual couples. We extend previous research by treating the man’s employment status as a variable that helps to explain rather than confounds this relationship, and by comparing multiple countries through regression analyses of European Social Survey data (Rounds 2–9). Results provide evidence of a female-breadwinner well-being ‘penalty’: men and women are less satisfied with their lives under the female-bread-winner arrangement versus the dual-earner and male-breadwinner alternatives. The penalty is marginal when the male partner is part-time employed but sizeable when he is jobless. However, there are gender differences: after controls for composition, gender-role attitudes, and partners’ relative incomes, the penalty becomes negligible for women while remaining large for men. Analyses suggest these gender differences are linked to high male unemployment among female-breadwinner couples: whereas women appear roughly equally adversely affected by a male partner’s unemployment as by their own, men report substantially higher well-being when she is unemployed instead of him. Country comparisons indicate that while this female-breadwinner well-being penalty is largest in more conservative contexts, especially Germany, it is fairly universal across Europe. So, even in countries where women’s employment is more widespread and cultural and institutional support for the male-breadwinner model is weaker, unemployed men with breadwinner wives are not immune from the social stigma and psychological difficulties associated with their gender non-conformity.


Introduction
Women are the main or sole breadwinner in an increasing minority of heterosexual couples across Europe (Klesment and Van Bavel, 2017;Kowalewska and Vitali, 2021).The unpredictability of health, relationships, and labour markets-laid bare by the COVID-19 pandemic-means that most women will be the breadwinner at some point in their relationship, even if only temporarily (e.g.Glass et al., 2021).However, research indicates that this emerging arrangement is associated with lower well-being.Although results for women are less conclusive, men's life satisfaction is significantly lower when she out-earns him (Rogers and DeBoer, 2001;Hajdu and Hajdu, 2018;Salland, 2018;Gash and Plagnol, 2021).
We extend previous research on women's breadwinning and couples' well-being in two ways.First, we examine the relationship between female breadwinning and life satisfaction through regression analyses of ESS data (2004ESS data ( -2018)).Previous literature has focused on one to three countries only with stronger male-breadwinning norms and higher female parttime employment, such as Germany and the United Kingdom (e.g.Luhmann et al., 2014;Gash and Plagnol, 2021).The same holds for studies on unemployment and/or employment hours and life satisfaction within couples (e.g.Booth and Van Ours, 2013;Flèche et al., 2020).Therefore, the generalizability of the association between female breadwinning and lower well-being to other kinds of societal contexts remains unclear.We contribute to addressing this gap by comparing nine European countries, including those in which the male-breadwinner model has less relevance culturally and/or in social practice.
Second, we take an approach that emphasizes employment as a psychological and symbolic resource.We compare life satisfaction for men and women across various breadwinning configurations-which is the man's employment status combined with the woman's-while controlling for total household income, partners' relative incomes, and other confounders.Prior research on female breadwinning and life satisfaction has overwhelmingly defined breadwinning Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcad034/7190495 by guest on 06 June 2023 based on between-partner differences in income or earnings (although see Blom and Hewitt, 2020).Yet, looking at employment status is important for unravelling how lower well-being in female-breadwinner couples may relate to the male partner's part-time or non-employed status, and not simply his lower relative income.The act of leaving the home and going out into the world to perform physical paid labour is key to the self-production of male identity for manual workers (Winlow, 2001).Likewise, men in professional occupations perform hegemonic masculinity through spending long hours at their desks and visibly displaying their physical exhaustion, commitment, and endurance (e.g.Williams et al., 2013).Conversely, labour within the home-and those who perform it-are deemed 'feminine', 'expressive', and 'emotional' (Demantas and Myers, 2015).
We distinguish between two subtypes of female-breadwinner couples: those in which the male partner is part-time employed ('one-and-a-half' female breadwinners) and those in which he is jobless ('pure' female breadwinners).This distinction is motivated by competing expectations.On the one hand, well-being is feasibly higher under the former.Any job is usually better than no job for well-being (e.g.Winkelmann and Winkelmann, 1998).Employment provides more than a wage: it is a mechanism of social inclusion and can provide social interaction, a structure to one's day, identity and social status, and a feeling of 'being useful' (Jahoda, 1982).On the other hand, well-being may be similarly low under both couple subtypes.The 'flexibility stigma' means part-time employed men are often labelled 'unmanly' and as lacking the stamina and discipline for full-time employment (Williams et al., 2013).Part-time employment could also reflect a lack of suitable full-time jobs rather than a genuine choice and is often lower quality than full-time employment (Kauhanen and Nätti, 2015).
For similar reasons, we further differentiate 'pure' breadwinners by whether the jobless partner is unemployed or inactive.We might expect lower well-being under the former, since male unemployment is generally associated with the biggest psychological costs, such as self-doubt, uncertainty, loneliness, and stigma (e.g.Winkelmann and Winkelmann, 1998;Knabe and Rätzel, 2011).Additionally, unemployment is (presumably) undesired, whereas inactivity may reflect a choice to, for instance, spend more time with family (our sample excludes individuals who are inactive for health or disability reasons, which is associated with lower well-being, e.g.Stam et al., 2016).Alternatively, pure female-breadwinner couples' well-being may be low regardless of his labour force status.Either way, the couple's arrangement is gender-atypical.Inactivity can also be unplanned and unchosen due to, for example, needing to suddenly care for an ailing relative or a lack of affordable childcare.Furthermore, many inactive men are the 'discouraged unemployed': individuals who exited from unemployment by redefining themselves as 'inactive'.In our sample, 78% of inactive men report having searched for employment in the last 5 years (versus 36% of inactive women).
Our results indeed underscore the importance of differentiating female-breadwinner couples by the man's labour force status to account for the psychological, social, and symbolic dimensions of male employmentespecially for men themselves-beyond absolute and relative incomes.We find evidence of a female-breadwinner well-being 'penalty': life satisfaction is lower when she is breadwinning versus if he is the sole, main, or joint breadwinner. 1While this penalty is narrow when he is part-time employed, it is large when he is jobless, especially if he is unemployed.
Crucially, we observe variation by respondents' gender.After controlling for compositional characteristics and partners' relative incomes, the penalty becomes negligible for women while remaining sizeable for men.In fact, jobless men's well-being is higher when their female partner is also jobless rather than employed.We conclude that men attach greater importance to their own over their partner's employment status; moreover, her breadwinner status seemingly threatens jobless men's perceptions of their masculinity and amplifies the negative well-being consequences of their own joblessness.Cross-national comparisons indicate that while this well-being penalty is near universal, it is largest in more conservative societies.

Theoretical expectations
Six hypotheses underpin our research (Figure 1).Hypothesis 1 posits that life satisfaction is lowest for couples in which partners perform similar roles to one another, i.e. dual-earner couples (both in market work) and jobless couples (both in the home).Life satisfaction is highest for single or 'pure' breadwinner couples, when each partner 'specializes' in either market work or domestic work (Becker, 1985).According to the 'time-availability' perspective (e.g.Presser, 1994), complementarity between partners mitigates the stress of each trying to balance a job and domestic responsibilities, which leaves both partners with more time for leisure (e.g.Jacobs and Gerson, 2004).Specialization may also enhance interdependence and mutual obligation between partners (Blom and Hewitt, 2020) for pure-breadwinner couples, with equally high satisfaction for pure male-breadwinner and pure female-breadwinner couples.
Hypothesis 2 is based on 'role collaboration' (Rogers, 2004): life satisfaction is highest when partners share similar labour market experiences, lead similar daily lives, and hold a similar labour force status.Such similarities may promote empathy, mutual understanding, emotional intimacy, and cooperation between partners-with benefits for their well-being (Blom and Hewitt, 2020).Therefore, reflecting their dissimilar daily lives and experiences, pure-breadwinner couples should report lower life satisfaction than two-breadwinner couples, regardless of the sole breadwinner's gender.One-and-a-half breadwinner couples should report moderate well-being-again, regardless of the main breadwinner's gender-given their partial role collaboration.
According to this theory, pure-breadwinner couples should have lower life satisfaction than even jobless couples (although jobless couples have lower well-being than dual earners given the disadvantages of worklessness, discussed below).Compared Hypotheses 6a (inactive/unemployed) and 6b (inactive/unemployed + gender) Men and women under H6a Men and women under H6b with pure-breadwinner couples, jobless couples share more similar labour market experiences, social status, and daily routines-which should foster empathy (Luhmann et al., 2014).What is more, someone experiencing joblessness may perceive themselves more positively if their partner is also jobless, since joblessness becomes the household norm and less 'deviant' (Clark, 2003).By contrast, under the pure-breadwinner arrangement, the jobless partner must watch their partner 'go out to work' every day, which may heighten feelings of guilt and inadequacy.This can in turn impact negatively on the breadwinning partner's well-being, whether directly ('your pain is my pain') and/or indirectly (e.g. through tension in the household) (e.g.Demerouti et al., 2005;Song et al., 2011).
H2 (role collaboration): Life satisfaction is highest for dual-earner couples, moderate for one-and-ahalf breadwinner and jobless couples, and lowest for pure-breadwinner couples, with equally low satisfaction for pure male-breadwinner and pure female-breadwinner couples.
Hypothesis 3 instead posits that jobless couples will experience lower well-being than pure-breadwinner couples (and two-breadwinner couples).Couples face greater uncertainty and financial insecurity when two people are out of work rather than just one.A partner's joblessness may further deplete a jobless individual's 'coping resources', who must support the partner's psychological well-being on top of dealing with their own joblessness (Inanc, 2018).Conversely, having an employed partner can alleviate some of the negative well-being consequences of one's own joblessness through access to pooled household resources (e.g. higher household income, the employed partner's networks).Moreover, individuals often incorporate their romantic partner into their own identity and selfworth ('shared fate').Under the pure-breadwinner arrangement, then, the jobless partner may take pride in and feel partly responsible for their partner's successes, with their own joblessness paling in importance (Pinkus et al., 2012).
H3 (shared fate): Life satisfaction is lowest for jobless couples.
An additional relevant theoretical approach is the 'autonomy perspective' (e.g.Gupta, 2007;Tisch, 2021): life satisfaction depends only on one's own employment status and is unrelated to the partner's employment status.Having a 'good' job is associated with the highest life satisfaction (and job satisfaction, e.g.Drobnič et al., 2010); yet any job typically offers pecuniary and non-pecuniary benefits over joblessness, especially unemployment (e.g.Winkelmann and Winkelmann, 1998).Employment also provides social connections, which predict higher life satisfaction (Stiglitz et al., 2009), while earnings allow for social participation (e.g.going out with friends, attending gym classes).We further expect that full-time employed individuals should have higher life satisfaction than part-time employed individuals, again, independent of gender.Full-timers generally have higher earnings, better job security, greater access to statutory and occupational benefits, and improved training opportunities (Kauhanen and Nätti, 2015).
H4 (autonomy): Life satisfaction increases the 'more' employed one is-regardless of the partner's employment status: • Men report equally low satisfaction in pure female-breadwinner and jobless couples, moderate satisfaction in one-and-a-half female-breadwinner couples, and equally high satisfaction in male-breadwinner (both pure and one-and-a-half) and dual-earner couples.
• Women report equally high satisfaction in female-breadwinner (both pure and one-and-ahalf) and dual-earner couples, moderate satisfaction in one-and-a-half male-breadwinner couples, and equally low satisfaction in pure male-breadwinner and jobless couples.
Hypothesis 5 draws on 'gender-role ideology'.It predicts lower life satisfaction for men under the female-breadwinner arrangement versus all other couple-types due to the gendered meanings attached to breadwinning.Providing financially for their families is part of how men 'do gender' (West and Zimmerman, 1987).While cultural norms increasingly dictate that men will be involved in family life, it is expected that they are breadwinners first (Ranson, 2010): this is how men exhibit 'good' fathering (Townsend, 2002).Thus, staying 'in the home'-the domain historically associated with femininity-threatens men's perceptions of their masculinity (Meisenbach, 2010); and yet, masculinity is a strong predictor of men's sense of self and well-being (e.g.Burkley et al., 2015).Jobless men may also be vulnerable to isolation and loneliness, since they are less likely than women to have community or care-based social networks to draw upon (e.g.Russell, 1999).H5 likewise predicts lower life satisfaction for women under the female-breadwinner versus alternative arrangements.This partly reflects the 'crossover' of the jobless man's lower well-being to his partner (e.g.Baranowska-Rataj and Strandh, 2021), with women more adversely affected by a partner's unemployment than men (e.g.Inanc, 2018).It also reflects the mismatch between cultural ideals of 'good' mothering and the breadwinner role.The 'intensive mothering' ideology prescribes that dependent children's needs are prioritized, that the 'best' childcare is emotionally absorbing and labour intensive, and that the mothernot the father-performs these tasks, regardless of her labour force participation (Hays, 1996).However, in seeking to protect the family's main income source, breadwinner women may signal their 'ideal' worker status by, for example, working longer hours or taking on additional responsibilities that clash with homelifedespite breadwinner women's higher average domestic loads than breadwinner men (e.g.Latshaw and Hale, 2016).Relatedly, research has found that breadwinner women with jobless and part-time employed male partners enjoy less leisure and sleep/personal care time on average than their male partners and women in male-breadwinner couples (Latshaw and Hale, 2016).They also report higher stress than full-time employed fathers with stay-at-home wives (Zimmerman, 2000).These stresses and strains may, in turn, crossover to their male partners, thereby reinforcing jobless men's lower well-being under the female-breadwinner arrangement.
In highlighting social and cultural factors, H5 alludes to cross-national variation.The lower life satisfaction associated with the female-breadwinner couple-type should be amplified in stronger male-breadwinner contexts, where societal-level beliefs, norms, and expectations and/or policies continue to cast men as breadwinners (Germany, Ireland, Poland, Spain, United Kingdom; see 'Data and Method' for more detail).Here, the stresses associated with female-breadwinner couples' gender-atypicality should be greater.Their deviation from prevailing gender norms and expectations may lead to social sanctions and stigma-for example, ridicule, gossiping, criticism-which consolidate feelings of shame and impact negatively these couples' sense of social identity (Stam et al., 2016;Gonalons-Pons and Gangl, 2021).We anticipate a weaker association between female breadwinning and life satisfaction in more progressive countries with greater policy support for the dual-earner model (Finland, France) or where women's labour force participation is part of the social fabric (Portugal, Slovenia).H5 (gender-role ideology): Life satisfaction is lowest for women and men in female-breadwinner couples, especially in more conservative country contexts.
Hypotheses 6a and 6b focus strictly on pure-breadwinner couples.H6a posits that life satisfaction depends on whether the jobless partner in these couples is 'unemployed' or 'inactive'.Studies generally find that unemployment is associated with bigger declines in well-being than inactivity (e.g.Stam et al., 2016).Furthermore, when the jobless partner is unemployed, they are (presumably) seeking to 'specialize' in market work again, which may lead to role conflict from both partners trying to balance paid work/job-seeking and domestic work (Rao, 2020).Conversely, when the jobless partner is inactive, couples may experience reduced role conflict, especially if the inactive partner 'specializes' in domestic work and frees up the employed partner's time and energy for their paid job.
H6b instead posits that female-breadwinner couples always report lower well-being than male-breadwinner couples regardless of his labour force status.Like H5, H6b underlines the importance of gender.Arguably, male inactivity is just as 'deviant' as male unemployment from the male-provider role, if not more so when the inactive man is a homemaker or stay-at-home parent and explicitly assigned a feminized role.By contrast, when the male partner is unemployed, he is presumably seeking to (re-)enter employment and 'do masculinity' again; hence, female-breadwinner couples do not necessarily 'undo' the gendered division of domestic responsibilities on the assumption he will soon find employment again and things will be 'back to normal' (Legerski and Cornwall, 2010).
H6b anticipates higher well-being under the male-breadwinner versus female-breadwinner arrangement even when the former is due to her unemployment.Women typically have stronger neighbourhood and kinship links than men to draw upon during unemployment, particularly those (re)produced through performing family responsibilities, such as the 'school run' or 'playdates' (Russell, 1999).Additionally, social norms and patterns make it more acceptable for unemployed women than unemployed men to engage in alternate roles that provide purpose and self-esteem, like 'parent', 'carer', or 'volunteer' (Waters and Moore, 2002).H6a (inactive/unemployed): Among pure-breadwinner couples, life satisfaction is higher when the jobless partner is inactive rather than unemployed, regardless of the breadwinner's gender.
H6b (inactive/unemployed + gender): Life satisfaction is always lower when she rather than he is the only breadwinner, regardless of the jobless partner's labour force status.

Data and method
We include nine countries with sufficient female-breadwinner couples aggregated from Rounds 2-9 of the European Social Survey (ESS) and that are representative of different welfare and gender regimes (Supplementary Table S1).Despite male-breadwinner tendencies-such as a home-care allowance for Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcad034/7190495 by guest on 06 June 2023 stay-at-home parents (read: mothers)-Finland represents the dual-earner (e.g.Korpi et al., 2013) or 'weak' male-breadwinner society (Lewis, 1992).Feminist scholars have likewise identified greater support for women's employment in France (e.g.Saraceno and Keck, 2010); yet, certain similarities with Continental European countries-such as on gender attitudes, women's employment rates, and taxes (Supplementary Table S1)-make France a 'moderate' male-breadwinner model.Meanwhile, Germany, Great Britain, and Ireland share a strong (albeit fading) male-breadwinner legacy (Lewis, 1992), although cultural and policy support for women's caregiving is strongest in (Western) Germany (Leitner, 2003; Supplementary Table S1).Great Britain and Ireland instead rely heavily on market provision of childcare and other welfare, meaning (mostly lower-skilled) women are sometimes unable to afford to take a paid job, despite the strong 'welfare-to-work' rhetoric (e.g.Korpi et al., 2013).
We include representatives from Southern and Eastern Europe, too.In Portugal and Spain, support for women's caregiving is more 'implicit' (Leitner, 2003).Underdeveloped state and market-based care services and benefits to financially support the family's caring function, amid strong inter-generational and gendered caregiving norms, leave care work to the family (women) 'by default' (Saraceno and Keck, 2010; Supplementary Table S1).Still, cultural and policy support for women's employment since the exodus of men to fight in the colonial wars has kept women's employment participation comparatively high in Portugal, despite women's disproportionate share of family labour (Tavora, 2012; Supplementary Table S1).Slovenia also has comparatively high female labour force participation, although this partly reflects more developed family policies.Poland instead lags behind and is more traditional (Javornik, 2014; Supplementary Table S1).
We model life satisfaction separately for men and women via linear regression using individual-level, weighted data.Our sample (N = 20,850 men and N = 22,028 women) comprises working-age (18-65) households with a 'male' and 'female' partner, whether married/unmarried or with/without children.We exclude couples living with other adults who are not their children and couples in which one or both partners are in education, permanently sick/ disabled, retired, or in community or military service.Our dependent variable, life satisfaction, is based on responses to: 'All things considered, how satisfied are you with your life as a whole nowadays?'Answers range from 0 ('extremely dissatisfied') to 10 ('extremely satisfied'), with sample averages of 7.24 for men and 7.31 for women.
Table 1 shows the shares of couple-type by country.Due to small sample sizes, 'one-and-a-half' female-breadwinner couples are excluded from the country comparisons; hence, these analyses are based on couple-types 1-3, 5, and 6 only.Supplementary Tables S2a and S2b present descriptive statistics for the couple-types across the pooled sample of countries and survey waves for men and women, respectively.We acknowledge that our analyses do not claim causal links between transitions into and out of employment and life satisfaction; rather, they describe the cross-sectional association between both partners' employment statuses and life satisfaction at the time of the survey.
Our four models are as follows.Model 1 controls for country and survey wave only.Model 2 then adds several controls: partners' ages, as life satisfaction is U-shaped over the life course; partners' education levels, as education and well-being are associated; a child under two and number of children, since having younger and more children is associated with lower life satisfaction (Layard et al., 2012); whether the respondent is married (positively related to life satisfaction), foreign-born, and in poor health (both associated with lower life satisfaction) (Layard et al., 2012); domicile, as living in urban contexts is associated with lower life satisfaction (Weckroth and Kemppainen, 2021); total household income, since higher incomes are associated with higher life satisfaction (Layard et al., 2012) and female-breadwinner couples have lower average incomes (e.g.Kowalewska and Vitali, 2021); and respondents' feelings about their household income nowadays to encompass a broader range of needs and resources (e.g.publicly funded childcare) than objective household income.
The sample sizes permitted by pooling data for all countries and waves allow for including 'oneand-a-half' female-breadwinner couples and several robustness checks.To account for possible attitudinal variation across and within countries and couple-types, Model 3 adds a control based on dis/agreement with: 'When jobs are scarce, men should have more right to jobs than women'.Model 4 adds a control for partners' relative incomes based on responses to: 'Around how large a proportion of the household income do you provide yourself?'We run Models 3 and 4 on a restricted sample from Rounds 2 and 5 (N = 5,192 men and 5,650 women), since these are the only ESS rounds in which data on relative incomes are collected.
The pooled sample also provides adequate sample size to assess how well-being varies for pure female-breadwinner versus pure male-breadwinner couples by whether the jobless partner is unemployed ('actively looking for job') or inactive ('not actively looking for job'/'doing housework, looking after children or other persons').Thus, we replicate Model 2 using four couple subtypes as our main independent variable: (i) MBW, W unempl: employed man, unemployed woman; (ii) MBW, W inact: employed man, inactive woman; (iii) FBW, M unempl: employed woman, unemployed man; (iv) FBW, M inact: employed woman, inactive man.

Pooled analyses
Figure 2 displays life satisfaction as a function of our six main couple-types based on the pooled data for all countries and waves.Supplementary Table S3 reports the coefficient estimates.Using these estimates, Table 2 shows the results of pairwise comparison tests to assess whether differences in life satisfaction between pairs of couple-types (based on Model 2) are statistically significant.A positive (negative) difference indicates higher (lower) average life satisfaction in the first couple-type.
Before compositional controls (Model 1), men and women never report higher life satisfaction when one partner is employed (MBW/FBW) versus both (1.5MBW/DE/1.5FBW). 3While introducing individual and couple-level controls (Model 2) reduces this gap, it does not eliminate it (Figure 2).So, against H1 (role specialization), couples report higher life satisfaction when partners share breadwinning versus when only one partner 'specializes' in it.
Higher life satisfaction for two-breadwinner versus single-breadwinner couples seemingly supports H2 (role collaboration); yet, the breadwinner's gender matters more than this hypothesis predicts.Following H5 (gender-role ideology), life satisfaction increases as Notes: 'MBW' = man is employed, woman is not.'1.5MBW' = man works ≥30 h, woman works <30 h.'DE' = both members of the couple are employed a similar number of hours.'1.5FBW' = woman works ≥30 h, man works <30 hours.'FBW' = woman is employed, man is not.'JL' = jobless couple.Model 1 = the 'null' model, i.e., with no controls.Model 1 includes controls for country and survey wave.Model 2 adds basic socioeconomic and sociodemographic controls.Supplementary Table S3 reports the coefficient estimates.the couple's breadwinning configuration approaches the male-breadwinner norm and decreases as it strays further from it.More specifically, we identify a female-breadwinner well-being 'penalty': men and women report lower well-being when she is breadwinning versus if he is or breadwinning is shared.For all pairwise comparisons, this female-breadwinner well-being penalty is stronger for men than women (Table 2).The pairwise comparisons also indicate that the compositional factors controlled for in Model 2 explain some of this penalty, as the penalty is generally smaller than in Model 1. Indeed, the inclusion of compositional factors in Model 2 increases the adjusted R 2 by 0.112 for men and 0.115 for women (Supplementary Table S3).
The female-breadwinner well-being penalty appears to be associated mainly with the 'pure' female-breadwinner model, under which the male partner is jobless, rather than the 'one-and-a-half' female-breadwinner couple, under which he is part-time employed.Men in one-and-a-half female-breadwinner couples report only negligibly lower life satisfaction than men in dualearner and male-breadwinner couples (1.5FBW < DE: −0.088, 1.5FBW < MBW: −0.063; 1.5FBW < 1.5MBW: −0.130; all P < 0.10; Table 2).Women likewise report only slightly lower well-being, if not higher well-being, under the one-and-a-half female-breadwinner arrangement compared with the dual-earner and male-breadwinner alternatives (1.5FBW < DE: −0.016; 1.5FBW < 1.5MBW: −0.100; 1.5FBW > MBW: 0.061; all P < 0.10).Altogether, these findings offer limited support for H2 (role collaboration).
Further against H2, both partners' well-being is lower (P < 0.05)-rather than the same-when she is the sole breadwinner instead of him, especially for men (FBW < MBW: −0.567 for men and −0.240 for women; Table 2).Lower well-being under the 'pure' female-breadwinner versus 'pure' male-breadwinner arrangement also provides evidence against H3 (shared fate).In fact, jobless couples report higher well-being than pure female-breadwinner couples (JL > FBW: 0.200, P < 0.05 for men and 0.157, P < 0.10 for women) but lower well-being than pure male-breadwinner couples (JL < MBW: −0.367, P < 0.05 for men and −0.083, P < 0.10 for women)-again, with stronger effects for men.
Unlike for men, women's well-being is always higher when both partners are employed instead of just her.Her own employment status is relevant for her well-being (e.g.DE > MBW: 0.076, P < 0.05; 1.5MBW > MBW: 0.161, P < 0.05; Table 2), and some findings for women do align with the autonomy hypothesis (e.g.1.5FBW > MBW: 0.061, P < 0.10).Still, the evidence for women is, overall, less in favour of H4 and more against it.As aforementioned, women prefer that he is the sole breadwinner instead of her, in accordance with H5 (gender-role ideology).Breadwinning women also have lower average life satisfaction-and not higher satisfaction, as H4 would predict-than women in one-and-a-half male-breadwinner couples (FBW < 1.5MBW: −0.401, P < 0.05; 1.5FBW < 1.5MBW: −0.100, P < 0.10).Further against H4, breadwinner women report lower (not similar) life satisfaction than women in dual-earner couples (FBW < DE: −0.316, P < 0.05; 1.5FBW < DE: −0.016, P < 0.10), and higher (not similar) well-being when their partner has a part-time rather than no job (FBW < 1.5FBW: −0.301, P < 0.05).

Country comparisons
Figure 3 displays well-being for men and women separately as a function of couple-type interacted with country (excluding one-and-a-half female-breadwinner couples due to insufficient sample sizes by country).Again, we formally test whether differences between coefficients (Supplementary Table S4) are statistically significant through pairwise comparisons of life satisfaction (Supplementary Table S5).
Overall, the country-level analyses provide mixed support for H5 (gender-role ideology).Some results for men fail to support H5.While the well-being penalty associated with the pure female-breadwinner arrangement is smallest for men in Portugal and Slovenia, it is also small for men in more traditional Poland (e.g.FBW < MBW: −0.103 for Portugal, −0.212 for Slovenia, and −0.257 for Poland; all P < 0.10; Supplementary Table S5).Further against H5, the well-being penalty for men in pure female-breadwinner couples is present across all nine countries, including more gender-egalitarian France and Finland (France: FBW < MBW: −0.586, FBW < 1.5MBW: −0.882, and FBW < DE: −0.918; Finland: FBW < MBW: −0.566, FBW < 1.5MBW: −0.589, and FBW < DE: −0.669; all P < 0.05; Supplementary Table S5).Additionally, men in France and Finland report higher well-being under the jobless versus pure female-breadwinner arrangement (JL > FBW: 0.342, P < 0.10 for France and 0.657, P < 0.05 for Finland); yet, men in France report lower well-being under the jobless versus pure male-breadwinner arrangement (JL < MBW: −0.243, P < 0.05), while men in Finland report roughly equal well-being under these two couple-types (JL > MBW: 0.091, P < 0.10).
Other results do, however, support H5. Notably, the well-being penalty is severest for men in more conservative contexts.In Germany, men report higher well-being when both partners are employed instead of just him; yet, their life satisfaction is 1.11-1.42points (of 10) lower when she is the sole breadwinner versus if he is the main or sole breadwinner or breadwinning is equally shared (Supplementary Table S5).Furthermore, although men prefer that at least one partner is employed, the well-being disadvantage associated with the jobless couple-type is smaller when compared with the pure female-breadwinner arrangement-i.e. when he is jobless-versus the pure male-breadwinner one (JL < FBW: −0.119, P < 0.10; JL < MBW: −1.231, P < 0.05).
The female-breadwinner well-being penalty is relatively large in the other strong male-breadwinner societies of Britain and Ireland as well as 'implicitly' familialist Spain (e.g.FBW < MBW: −0.616, P < 0.05 for Spain, −0.609, P < 0.05 for Ireland, and −0.423, P < 0.10 for Britain).Men's well-being is generally highest when they are employed with limited differences by the female partner's employment status (e.g.DE < MBW: −0.087 for Spain, −0.087 for Ireland, and −0.035 for Britain, all P < 0.10).In addition, men in Ireland and Britain report lower well-being when both partners are jobless versus if only she is (JL < MBW: Ireland: −0.571, P < 0.05; GB: −0.200, P < 0.10; Supplementary Table S5), but similar or higher well-being when both partners are jobless versus if only he is (JL > FBW: 0.038, P < 0.10 for Ireland and 0.223, P < 0.10 for GB).Meanwhile, although men in Spain apparently prefer the jobless couple-type, this preference is greater when the alternative is the pure female-breadwinner arrangement rather than the pure male-breadwinner one (JL > FBW: 0.761, P < 0.05; JL > MBW: 0.145, P < 0.10).
Well-being by couple-type follows similar cross-national patterns for women as for men.However, as in the pooled results, the well-being penalty associated with the pure female-breadwinner arrangement is generally smaller for women; hence, cross-national differences are also smaller.Across most countries, women's life satisfaction is 0.05-0.24points lower when she is the breadwinner instead of him (P < 0.10; Supplementary Table S5) except in Germany and Poland, where the penalty is larger (−0.637 and −0.553, respectively; P < 0.05).

Robustness checks (pooled sample)
The sample sizes permitted by pooling all countries allow for running various robustness checks.For both men and women, results are robust to including controls for gender-role attitudes (Model 3) and partners' relative incomes (Model 4 in Supplementary Figure S1 and Supplementary Table S6a and b).Since the latter control is collected in ESS Rounds 2 and 5 only, we replicate Model 1 (country and survey wave controls) and Model 2 (Model 1 + compositional controls) on the same restricted sample for comparability.While controlling for compositional factors increases the adjusted R 2 for men and women (Model 2 versus Model 1), including controls for gender-role attitudes and relative incomes does not change the adjusted R 2 (Models 3 and 4 versus Model 2; Supplementary Table S6a and b).Coefficients for Models 3 and 4 are not statistically significant either with one exception: life satisfaction is higher for women, on average, when she makes no contribution to total household income versus providing it all (P < 0.001; Supplementary Table S6b).
The pairwise comparisons confirm these patterns.For men, controlling for gender-role ideology and relative incomes has a limited impact (Models 3 and 4 versus Model 2; Supplementary Table S7).In fact, even after all controls, men remain significantly and substantially less pleased with how their lives are going when she is the sole breadwinner (FBW < MBW: −0.585; FBW < 1.5MBW: −0.712; FBW < DE: −0.653; all P < 0.05 in Model 4).Therefore, we suggest that for men, the female-breadwinner well-being penalty is about more than compositional factors: it reflects the non-pecuniary costs of their joblessness, too.
Controlling for gender attitudes also has limited impact on the pairwise comparisons for women (e.g.FBW < MBW: −0.209, P < 0.10 in Model 2 versus −0.210, P < 0.10 in Model 3; Supplementary Table S7), although relative incomes seemingly play a (n albeit limited) role.While other pairwise comparisons show limited change from Models 2 to 4, including the control for relative incomes virtually eliminates the well-being penalty for women in pure female-breadwinner couples versus pure male-breadwinner couples (FBW > MBW: −0.209, P < 0.10 in Model 2 versus 0.048, P < 0.10 in Model 4) and jobless couples (JL > FBW: 0.197, P < 0.10 in Model 2 versus 0.067, P < 0.10 in Model 4).For women, then, the gender 'deviance' of her providing all the income instead of him partly contributes to explaining the female-breadwinner well-being penalty.
As a final robustness check, we interact couple-type by household income.Supplementary Figure S2 shows that the results are robust to the household's income position: while the female-breadwinner well-being penalty is largest when families are in the bottom 20% of household incomes, the penalty persists even for households in the top 20%.

Unemployment versus inactivity (pooled sample)
Figure 4 focuses on pure-breadwinner couples to illuminate how well-being varies by whether the jobless partner is unemployed or inactive.Supplementary Tables S8 and S9 report the coefficients and pairwise comparisons.Results indicate that the female-breadwinner well-being penalty is driven mainly by male unemployment.The well-being penalty associated with the female-breadwinner versus male-breadwinner model is largest-especially for men-when the former is due to male unemployment and the latter is due to female inactivity (FBW, M unemp < MBW, W inact: −0.738 for men, P < 0.05 and −0.303 for women, P < 0.05; Supplementary Table S9).As the male partner is usually unemployed in pure female-breadwinner couples, whereas the female partner is usually inactive in pure male-breadwinner couples, it is unsurprising we observe a large female-breadwinner well-being penalty in the pooled sample and countries with high shares of male-breadwinner/female-caregiver couples (Germany, the United Kingdom, and Ireland; Table 1).
Other results reinforce the stronger association between women's breadwinning and low well-being under male unemployment.Breadwinner women and their jobless partners report higher life satisfaction when he is inactive rather than unemployed (FBW, M inact > FBW, M unemp: 0.491, P < 0.05 for men and 0.167, P < 0.10 for women; Supplementary Table S9).When the man is inactive, female breadwinning carries a much smaller well-being penalty, if any (FBW, M inact < MBW, W inact: −0.246, P < 0.05 for men and −0.136, P < 0.10 for women; FBW, M inact > MBW, W unemp: 0.066 for men, P < 0.10 and 0.176 for women, P < 0.10).
Nevertheless there are gender differences.Men's well-being is lower under the female-breadwinner versus male-breadwinner arrangement even when both arrangements are due to unemployment (FBW, M unemp < MBW, W unemp: −0.425, P < 0.05; Supplementary Table S9), whereas women's well-being is no different (FBW, M unemp = MBW, W unemp).So, unlike men, women appear as sharply affected by a partner's unemployment as by their own.Overall patterns for women appear supportive of H6a: well-being is lower under the pure-breadwinner arrangement when the jobless partner is unemployed rather than inactive, with limited differences by the breadwinning partner's gender (Figure 4).For men, though, a combination of H6a and H6b explains the results: while men in pure-breadwinner households prefer that the jobless partner is inactive rather than unemployed, they would rather be the breadwinning partner.

Discussion
In examining variation in subjective well-being by heterosexual couples' breadwinning configuration, our study adds to a growing literature showing an association between female breadwinning and lower life satisfaction across industrialized countries (e.g.Rogers and DeBoer, 2001;Hajdu and Hajdu, 2018;Salland, 2018;Gash and Plagnol, 2021).We find evidence of a female-breadwinner well-being 'penalty': men and women are less satisfied with their lives when she is the sole breadwinner versus if he is breadwinning or breadwinning is shared.In defining breadwinning by employment status rather than relative incomes and comparing multiple countries, we uncover variation in this penalty by country and the male partner's labour force status, with further differences by respondents' gender.While female breadwinning yields a large penalty when the male partner is jobless, the penalty is smaller for men and negligible for women when he is part-time employed after all controls.Although these men may be involuntarily part-time employed, prior research generally agrees that any job is better than no job for well-being (e.g.Winkelmann and Winkelmann, 1998).Potentially, part-time employment confers the man sufficient resources and benefits-e.g.social contacts, independence, identity-to lessen the potential 'threat' to masculinity posed by the female partner's breadwinner status.At the same time, having a partner in part-time employment may reduce breadwinner women's role strain and stress from economically sustaining the family while managing their (usually high share of) domestic responsibilities (e.g.Latshaw and Hale, 2016).However, we cannot rule out a potential selection effect: for instance, the identities of men in this couple-type may emphasize non-work activities, meaning their self-worth is less bound up in their careers or holding breadwinner status.We also cannot rule out preference adaptation among these part-time employed men and their breadwinner partners.
The larger well-being penalty observed for female-breadwinner couples in which the man is jobless partly reflects these couples' composition.Compared with two-earner and male-breadwinner couples, these couples are more likely to have low household incomes, be unmarried, be migrants, and find it 'difficult' or 'very difficult' to cope with their household income.Additionally, a higher proportion of male partners in this couple-type report 'fair', 'bad', or 'very bad' health and are low-educated (Supplementary Table S2a and b).All these characteristics are associated with lower life satisfaction (e.g.Layard et al., 2012); indeed, controlling for them reduces the size of the well-being penalty (Model 2 versus Model 1 in Figure 2 and Supplementary Table S3).
While results are robust to controlling for individuals' gender-role attitudes (Model 3), controlling for partners' relative incomes shrinks the female-breadwinner well-being penalty further for women (Model 4, Supplementary Figure S1 and Supplementary Tables  S6a, b, and S7), to the extent women's well-being is virtually identical when either partner is the sole breadwinner (FBW > MBW: 0.048, Model 4 in Supplementary Table S7).Conversely, the well-being penalty associated with the pure female-breadwinner arrangement remains sizeable for men even after all controls.We suggest this remaining penalty reflects the social and psychological difficulties experienced by jobless men with breadwinner wives.As existing literature has shown, joblessness carries a heavy psychological penalty for men.Employment remains part of hegemonic masculinity and male identity (Connell, 1995) while providing non-pecuniary benefits (Jahoda, 1982).These men may face social sanctions and stigma for their gender-role non-conformity, such as judgement or ridicule, so that even men who personally hold gender-egalitarian views can suffer stress from violating societal gender norms (Gonalons-Pons and Gangl, 2021).The country-level analyses indicate that the well-being penalty experienced by jobless men with breadwinner partners is fairly universal.Although the penalty is largest in countries with stronger male-breadwinner norms-especially Germany-it is present in the less traditional contexts of Finland and France, too.Even here, jobless men with breadwinner partners are not immune from the social stigma and psychological difficulties associated with their gender non-conformity.While the female-breadwinner well-being penalty is smaller still for men in Portugal, Poland, and Slovenia, we suggest this is less to do with equality.Arguably, it is reflective of these countries' low-wage economies, which make two full-time wages essential for families' survival (Sánchez-Mira and O'Reilly, 2019).Under these circumstances, it is plausible that being in a single-breadwinner couple carries such economic risk that men's concerns about gender (non)conformity take a backseat.The smaller female-breadwinner well-being penalty for men in these countries may also reflect high rates of female unemployment among male-breadwinner couples (Table 1), which, as we show, is associated with men's lower well-being.
Prior studies based on specific country cases have reached different conclusions regarding whether women's life satisfaction is lower (e.g.Hajdu and Hajdu, 2018;Salland, 2018) or the same under the female-breadwinner arrangement versus the male-breadwinner model (e.g.Rogers and DeBoer, 2001;Gash and Plagnol, 2021).Our cross-national comparisons indicate that while women generally report lower well-being when they are the only breadwinner, this penalty is small across most countries except for Germany and Poland.In fact, Poland is the only country in which the female-breadwinner well-being penalty for women exceeds that for men.Descriptive statistics (not shown) reveal that in Poland, breadwinner women with an inactive partner report among the lowest average life satisfaction scores across the pooled sample of countries and couple-types (5.94), whereas their inactive male partners report among the highest scores (7.12).Such gender disparities in life satisfaction within this same couple-type may help to explain the larger female-breadwinner well-being penalty for Polish women than for Polish men.A country-specific investigation could illuminate whether these disparities are an artefact of the small ESS sample.
Poland aside, the female-breadwinner penalty is smaller for women than men.What is more, jobless men report higher well-being when their female partner also has no job, whereas jobless women report lower well-being when the male partner is out of paid work rather than employed (Model 4, Supplementary Figure S1 and Supplementary Tables S6a, b, and S7).Therefore, net of average household income, women apparently benefit from their partner's labour market successes even when they themselves are jobless.THE FEMALE-BREADWINNER WELL-BEING 'PENALTY' than the one who stays at home.Further progress toward changing gendered norms around breadwinning and the domestic sphere is critical, especially since more and more families are likely to experience female breadwinning amid increased labour market insecurity.

Notes
1. Similar to Gash and Plagnol (2021), who identify a 'psychological penalty' among secondary-earner men, 'penalty' denotes how individuals may be 'penalized' in terms of experiencing lower well-being when she is the breadwinner instead of him.It does not imply that female breadwinning is the cause of this penalty; instead, the various institutional, economic, and social disadvantages faced by female-breadwinner couples are at the root.2. Small sample sizes made subnational analyses impracticable.3. Differences between Model 1 coefficient estimates are statistically significant.

Figure 2
Figure 2Predicted life satisfaction scores by couple-type, separate models for men and women, pooled sample for nine European countries (N = 20,850 for men, N = 22,028 for women).

Figure 4
Figure 4Predicted life satisfaction scores for 'pure' breadwinner couples by whether the non-breadwinning partner is inactive or unemployed, separate models for men and women, pooled sample for nine countries (N = 5,711 men, N = 5,995 women) Notes: 'MBW, W unemp' = man is employed, woman is unemployed.'MBW, W inact' = man is employed, woman is inactive.'FBW, M unemp' = woman is employed, man is unemployed.'FBW, M inact' = woman is employed, man is inactive.Basic socioeconomic and sociodemographic controls are included (Model 2).Supplementary TableS8reports the coefficient estimates; Supplementary TableS9shows the statistical significance of differences across coefficient estimates.
. Technically, this theory is gender-neutral: it should not matter who specializes in what.Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcad034/7190495 by guest on 06 June 2023 THE FEMALE-BREADWINNER WELL-BEING 'PENALTY' H1 (role specialization): Life satisfaction is lowest for dual-earner and jobless couples, moderate for one-and-a-half breadwinner couples, and highest Hypothesized patterns of life satisfaction by couples' breadwinning arrangement.

Table 1
Distribution of household-level employment arrangements among heterosexual couples by country (male and female respondents combined), % Notes: 'MBW' = man is employed, woman is not.'1.5MBW' = man works ≥30 h per week, woman works <30 h per week.'DE' = both members of the couple are employed for a similar number of hours.'1.5FBW' = woman works ≥30 h per week, man works <30 h per week.'FBW' = woman is employed, man is not.'JL' = neither partner is in employment.Country legend: PT = Portugal, ES = Spain, SI = Slovenia, IE = Ireland, FI = Finland, PL = Poland, FR = France, DE = Germany, GB = Great Britain.Source: European Social Survey, Rounds 2 (2004) to 9 (2018).
Table 2 reports the statistical significance of differences across coefficient estimates.

Table 2 .
Pairwise comparisons of predictive margins from Model 2 in Figure2 Notes: *<0.05; †<0.10.'MBW' = man is employed, woman is not.'1.5MBW' = man works ≥30 hours per week, woman works <30 hours per week.'DE' = both members of the couple are employed for a similar number of hours.'1.5FBW' = woman works ≥30 hours, man works <30 hours.'FBW' = woman is employed, man is not.'JL' = neither partner is in employment.
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcad034/7190495 by guest on 06 June 2023 THE FEMALE-BREADWINNER WELL-BEING 'PENALTY' Predicted life satisfaction scores by couple-type and country, separate models for men and women (N = 20,429 for men, N = 21,716 for women) Notes: 'MBW' = man is employed, woman is not.'1.5MBW'=manworks≥30 h per week, woman works <30 h per week.'DE'=bothmembers of the couple are employed a similar number of hours.'FBW'=woman is employed, man is not.'JL'=joblesscouple.Basic socioeconomic and sociodemographic controls are included (Model 2).Supplementary TableS4reports the coefficient estimates.Supplementary TableS5reports the statistical significance of differences across coefficient estimates.