Parenthood and the gender division of labour across the income distribution:

Abstract

This study employs a gendered relative resource approach to examine whether the importance of relative resources in shaping changes in the gender division of labor after first birth varies by couples’ household income. A substantial body of scholarship has argued that the gender division of labor within different-sex couples is influenced by partners’ relative resources. However, couples face class-based constraints that may alter the relevance of relative resources in shaping changes in gender divisions of labor following the transition to parenthood. This study compares couples’ divisions of paid work and housework before and up to four years after first birth, using 28 waves of the British Household Panel Survey and the UK Household Longitudinal Study (N = 1,606 couples). I find that the effect of relative resources on changes in couple’s paid work and housework behavior after first birth varies substantially by household income. Among higher-income couples, women’s paid work and housework time changes less among those with high relative earnings and more among those with low relative earnings, while men’s time allocation varies little after first birth. By contrast, among low-income couples, after first birth women’s paid work time and share decreases most among female breadwinners while their male partners’ paid work time increases substantially. These findings reflect the greater constraints that low-income couples face in making decisions about the division of household labor after first birth and highlight the need for greater attention to class interactions in the process of gender specialization in both research and work-family policy.

Introduction

A substantial body of scholarship has argued that partners’ relative resources (in terms of earnings, education or occupational prestige) are key determinants of the gender division of labor within different-sex couples. Either due to differential bargaining power resulting from unequal command of resources (e.g., England & Kilbourne, 1990; Lundberg and Pollak, 1996) or a couple’s desire to maximize household utility (e.g., Becker, 1981), partners with greater relative resources are theorized to specialize in market work while those with fewer resources specialize in unpaid work. Implicitly, relative resources are assumed to affect the division of labor in gender-neutral and universal ways. However, empirical research shows that women continue to perform the majority of unpaid labor within households even when they outearn their male partners (Bittman et al., 2003; Lyonette & Crompton, 2015; Tichenor, 2005), indicating that although relative resources may moderate patterns of gender specialization, gender remains a powerful organizing force shaping men and women’s roles within the household. At the same time, prior research on work-family dynamics across socio-economic classes in the UK and the US suggests there may be reason to question whether relative resources matter equally across households of varying socio-economic circumstances (Singley & Hynes, 2005; Warren, 2003, 2015). However, this potential heterogeneity has not been systematically explored. In this study, I examine whether the moderating effect of relative earnings on the gender division of household labor varies across the income distribution in Great Britain.

In liberal welfare states characterized by high levels of economic inequality and limited social support, couples face class-specific conditions that can constrain or promote gender-egalitarian divisions of household labor (Crompton & Lyonette, 2010; Lyonette & Crompton, 2015; Usdansky, 2011). For instance, although the need for two incomes may promote more equal divisions of paid labor among low-income couples, the quality and relative inflexibility of many low-paid jobs, poorer access to work-family policies and fewer resources with which to outsource unpaid work place substantial constraints on low-income couples’ decisions about paid and unpaid work and how it is divided within the household (Chung & van der Horst, 2018; Crompton & Lyonette, 2010; Damaske, 2011; Presser, 2003; Shows & Gerstel, 2009; Warren, Pascall, & Fox, 2010). At the same time, although the ‘ideal worker’ norms and workplace cultures of overwork that are characteristic of professional and managerial jobs may hinder gender equal divisions of labor among higher-income couples (Cha, 2010; Maume, 2006; Williams, 2000), these households have greater access to institutional and private resources to reconcile paid and unpaid work responsibilities (Crompton & Lyonette, 2010). These class-based constraints may alter the role of relative resources in shaping gender divisions of labor within couples. Among low-income couples, for instance, constraints may effectively “discount” (Blumberg & Coleman, 1989; Fuwa, 2004) the bargaining power that individuals are theorized to derive from their relative resources.

In this study, I examine whether the importance of relative resources in shaping the gender division of labor varies across the household income distribution in Great Britain, focusing on the transition to parenthood – a life stage during which gender specialization intensifies and solidifies among different-sex couples (e.g., Craig, 2006; Craig & Mullan, 2011; Fox, Pascall, & Warren, 2009; Schober, 2013). Using 28 waves of longitudinal data from the British Household Panel Survey and the UK Household Longitudinal Study (N=1,606 couples), I examine how wives’ pre-birth earnings relative to their husbands’ affects household divisions of paid work and housework up to four years after first birth and how these effects vary by class, measured using household income, during this important transition.

I find that the moderating effect of relative resources on changes in the gender division of labor after first birth depends in part on couples’ overall household resources. In general, wives’ relative earnings moderate changes in wives’ paid work and housework behavior among couples at the middle and upper end of the household income distribution in ways consistent with a gendered relative resource perspective – although these moderating effects are relatively minor. However, the effects of relative earnings run counter to relative resources expectations among low-income couples, where gender normative specialization increases most after first birth among couples where wives were the primary earner prior to pregnancy. The results of this study indicate that the drivers of gender specialization are not as universal as they are often assumed to be and highlight the importance of understanding how long-hypothesized mechanisms of specialization interact with other couple-level characteristics in influencing the division of paid and unpaid work.

Background

Relative Resource and Gender Theories of Household Specialization

A variety of long-standing theories have argued that partners’ within-couple relative resources (in terms of earnings, education or occupational prestige) are key determinants of the gender division of labor within different-sex couples. One line of thinking suggests that relative resources are indicative of partners’ comparative advantages in paid and unpaid work, with partners specializing according to these comparative advantages to maximize the couple’s joint welfare (Becker, 1981). Alternatively, relative resources may connote bargaining power within a relationship (e.g., England & Kilbourne, 1990; Lundberg & Pollak, 1996). Because housework is considered undesirable but necessary (Blood & Wolf, 1960; Brines, 1994), partners with greater relative resources – typically men – are able to bargain out of these tasks. Applied to the transition to parenthood, which necessitates greater household unpaid work and accentuates tradeoffs between paid and unpaid work, relative resource theories suggest increased specialization within couples following the transition to parenthood. Because relative resource theories are inherently gender-neutral, this approach predicts that, regardless of gender, partners with greater relative resources will specialize in market work while partners with lower resources will specialize in unpaid work.

Feminist scholars have long critiqued relative resource explanations of the gender division of labor, arguing that gender roles within a household are an important way that individuals “do” or enact their gender to meet societal norms and expectations rather than the results of a power negotiation between partners or a simple tradeoff based on utility maximization (e.g., Greenstein, 2000; Marx Ferree, 1990; West & Zimmerman, 1987). From this perspective, parenthood is theorized to increase household specialization in a gendered rather than gender-neutral way due to the starkly gendered nature of prevailing norms and expectations surrounding motherhood and fatherhood. And, indeed, empirical evidence widely acknowledges that the transition to parenthood increases gender specialization within different-sex couples (e.g., Craig, 2006; Craig & Mullan, 2011; Fox, Pascall, & Warren, 2009; Schober, 2013).

Although relative resource and “doing gender” approaches are often treated as competing explanations of the gender division of labor, empirical evidence suggests that these forces are interactive rather than mutually exclusive (e.g., Lyonette and Crompton, 2015). A substantial body of empirical research spanning a variety of country contexts indicates that relative resources play a role in shaping the gender division of labor following first birth, but that they do so in a gendered way (Begall & Grunow, 2015; Herrarte, Moral-Carcedo, & Saez, 2012; Kanji, 2011; Kühhirt, 2012; Sanchez & Thomson, 1997; Wood, Kil, & Marynissen, 2018). Women’s greater relative resources reduce the magnitude of gender role-conforming specialization after first birth, but they do not eliminate gendered specialization altogether (e.g., Kanji, 2011; Kühhirt, 2012; L. Sanchez & Thomson, 1997; Wood et al., 2018). For example, analyzing dual earner couples’ employment patterns over the transition to parenthood in Belgium, Wood and colleagues (2018) found that women’s greater pre-pregnancy relative earnings were related to more egalitarian divisions of paid work following first birth but that substantial gender specialization remained. In this way, partners’ relative resources may moderate the specializing effects of parenthood on the household division of paid and unpaid work, but gender and gender normative expectations remain an important organizing force shaping divisions of household labor.

Still, some empirical evidence has questioned whether relative resources play any significant role at all in shaping changes in the gender division of labor following first birth (e.g. Argyrous et al., 2017; Schober, 2013). Instead, these studies highlight the importance of other mechanisms of specialization such as partners’ gender ideology and women’s absolute earnings, which is theorized to increase women’s ability to purchase substitutes for their domestic labor (Gupta, 2007; Gupta & Ash, 2008; Killewald & Gough, 2010).

Gendered Relative Resources Across the Income Distribution

Evidence from the UK and the US on class-specific work-family challenges suggest that the gendered effects of relative resources on changes in husbands’ and wives’ paid work and housework following first birth may also vary across the household income distribution. Although social class is a multifaceted construct relating to individuals’ economic, social and cultural resources, I focus primarily on the role of income, which is an important factor motivating decisions regarding labor supply and which can be directly leveraged to outsource unpaid work.

The United Kingdom is a market-oriented liberal welfare state that provides limited social supports to facilitate a dual-earner, dual-carer division of labor within couples (see, e.g., Korpi, 2000; Korpi, Ferrarini, & Englund, 2013). It is also characterized by high levels of economic inequality relative to other wealthy post-industrial societies (e.g., OECD, 2022). In this context, couples face classed barriers and facilitators to paid and unpaid work that affect how that work is shared within couples (e.g., Lyonette & Crompton, 2015; Usdansky, 2011; Warren, 2003). These factors, outlined in Table 1 and discussed in detail below, may work to constrain or promote gender egalitarian divisions of paid and unpaid work after first birth and may affect how gender and relative resources interact to shape the division of household labor following the transition to parenthood.

TABLE 1.

Class-Specific Factors Promoting and Constraining Gender Egalitarian Divisions of Labor

	Low-income households	High-income households
Equality-promoting	• Need for two incomes • Shift work	• High opportunity cost of staying home • Greater access to work-family policies • Greater resources to outsource childcare and domestic labor
Equality-constraining	• Quality and inflexibility of low-paid jobs • Limited access to work-family policies • Limited resources to outsource childcare or domestic labor	• Less need for two incomes • Ideal worker norms and workplace cultures of overwork

Economic theory suggests that different push and pull factors drive employment among women in low- and high-income households. Women in low-income households may be more likely to maintain employment because their families need the income (sometimes referred to as the “need for income” hypothesis (e.g., England, 2010)) whereas women with high earnings potential (who tend to be partnered with high-earning spouses (e.g., Gonalons-Pons & Schwartz, 2017; Schwartz, 2010)) may be more likely to continue working because the opportunity cost associated with staying at home is high (e.g., England, Gornick, & Shafer, 2012). These factors may alter the moderating effect of relative resources on the gender division of household labor after first birth. In low-income households, income need may mean that women (and men) are more likely to maintain employment following first birth regardless of each partners’ relative earnings. By contrast in higher-income households where couples can “afford” to specialize, relative resources may have a greater moderating effect on specialization following first birth, resulting in greater variability in wives’ paid work patterns following parenthood.

The nature, schedule and flexibility of one’s job also pose class-specific work-family challenges that may affect gender divisions of labor following first birth. Lower-paid workers are more likely to work shifts with a fixed start and end time and work non-standard or unpredictable hours and are less likely to have control over their schedules or have the ability to change their work hours to accommodate personal or family responsibilities (Harknett, Schneider, & Luhr, 2020; Presser, 2003; Shows & Gerstel, 2009; Usdansky & Wolf, 2008; Warren, 2015). Low-paid jobs are also less likely to offer conditions and opportunities that promote employment continuity, such as opportunities for advancement, employer recognition and social status (Damaske, 2011). Thus, although the “need for income” hypothesis suggests that lower-income couples are more likely to engage in paid work equally, the quality and relatively inflexible characteristics of many lower paid jobs may pose significant barriers to dividing paid work time equally within couples following first birth. These constraints may diminish the role of relative resources in moderating the gendering effects of parenthood on the division of household labor, and low-income couples may be more likely to specialize along gender traditional lines following first birth regardless of pre-pregnancy relative earnings. Still, some evidence suggests shift work may actually facilitate egalitarian divisions of unpaid work as partners can alternate paid and unpaid work shifts (Deutsch, 1999; Presser, 2003; Warren, 2003).

High-income workers face different challenges in balancing paid and unpaid work responsibilities. Rather than time inflexibility and lack of schedule control, high-income workers (especially those in professional and managerial occupations) are particularly likely face persistent “ideal worker” norms in the workplace that reward those who prioritize work above all else and who demonstrate their devotion through long work hours and accessibility outside of standard working hours (Chung & van der Horst, 2020; Williams, 2000). This culture of overwork among high-earning workers has been shown to systematically disadvantage women employees, who are also expected to take primary responsibility for unpaid work within different-sex couples (Cha, 2010; Maume, 2006). Therefore, although the opportunity costs of staying home are high among high-income workers, workplace expectations of long hours combined with normative expectations of mothers as primary caregivers discourage women’s employment following first birth and may increase within-couple specialization.

Work-family policies may counteract some of the work-family conflicts high and low-income workers face, but the effects of particular policies – and workers’ access to them – are classed (Evertsson et al., 2009; Korpi et al., 2013; Mandel, 2011; Warren et al., 2010). Although high-income women are more likely to face workplace cultures of overwork, they are also more likely to have access to work-family policies that promote their employment continuity – particularly in liberal welfare state regimes like the UK that lack strong state policies promoting a dual-earner, dual-carer model (e.g., Korpi et al. 2013).

Paid maternity and paternity leave in the UK reflect normative assumptions of mothers as primary caregivers with 39 weeks of leave available to mothers (plus an additional 13 weeks unpaid) and two weeks available to fathers. While paid maternity leave facilitates employment continuity after first birth (e.g., Gornick & Meyers, 2009), when coupled with minimal paternity leave it may also encourage gender traditional divisions of labor by creating gendered care-related skill imbalances within couples (see e.g., Tanaka & Waldfogel, 2007). Eligibility for paid maternity and paternity leave in the UK is contingent upon “employee” status (excluding contractors and the self-employed), history of service with the same employer and meeting an earnings threshold. As a result, 16% of new mothers and 27% of new fathers are not eligible for these policies, with those with lower education and those in routine or intermediate occupations less likely to be eligible than more privileged groups (O’Brien, Aldrich, Connolly, Cook, & Speight, 2017). With less access to paid maternity leave, lower-income mothers may be more likely to leave the labor force altogether upon the birth of a child. Furthermore, high-income mothers and fathers alike are more likely to have access to more generously compensated company-specific paid leave policies to support mothers’ employment and fathers’ involvement in unpaid work (Chanfreau et al., 2011). Therefore, while the imbalance of paid maternity and paternity leave available in the UK may encourage gender traditional divisions of labor among different-sex couples across the income distribution, effects on household specialization may be most stark among low-income households.

Access to other work-family policies like flexible work are also classed. Flexible working arrangements (in terms of time and location) have been shown to help mothers maintain their paid work hours by providing greater freedom in how mothers reconcile paid and unpaid work (Chung & van der Horst, 2018). Although all parents were granted the statutory right to request flexible working arrangements in 2003, these requests are most likely to be granted to salaried employees, those with university degrees and those in professional occupations (Chung & van der Horst, 2018; O’Brien et al., 2017). With less access to flexible work arrangements, lower-income mothers may face greater difficulty reconciling paid and unpaid work following first birth compared with high-income women and may be more likely to reduce their paid work hours as a result.

Finally, high-income couples also have greater capacity to purchase substitutes for their own domestic labor, which may reduce women’s unpaid work time and promote women’s employment continuity after first birth. Maternal employment is positively associated with the availability and affordability of childcare, especially for very young children (Boeckmann, Misra, & Budig, 2015; Ruppanner, Moller, & Sayer, 2019). This is particularly important in countries with high costs and few public subsidies for childcare like the UK where the cost of childcare may push lower-paid mothers out of the labor force. Beyond childcare, in a high inequality context like the UK, high-income couples may achieve more gender egalitarian divisions of unpaid work by outsourcing housework (Heisig, 2011). Prior research has found a negative association between household expenditure on domestic outsourcing and the amount of time women spend on housework (Killewald, 2011; Sullivan & Gershuny, 2013) – although Killewald (2011) notes that the association is weaker than one might expect. Therefore, in theory, with greater ability to outsource, it may be easier to achieve a gender-equitable division of unpaid work within high-income couples. However, research shows that outsourcing has little effect on the gender division of housework within couples because it reduces men’s housework time as well as women’s (Gonalons-Pons, 2015; Lyonette & Crompton, 2015).

This Study

Recognizing the classed barriers and facilitators to gender egalitarian divisions of labor among different-sex couples in high inequality liberal welfare state contexts, in this study I examine whether the association between relative resources and changes in the gender division of labor after first birth varies across the household income distribution in Great Britain. Specifically, I examine two central hypotheses:

H1: The substantial constraints that low-income couples face in decisions about the division of paid work and housework reduce the moderating effect of relative resources on changes in the gender division of labor following first birth. Gender specialization of paid and unpaid work within low-income couples increases after first birth regardless of partners’ relative resources, resulting in greater uniformity in divisions of labor among low-income couples.

H2: With greater access to public and private resources to overcome class-specific work-family constraints, changes in the gender division of labor after first birth among higher-income couples are moderated by partners’ relative resources, resulting in greater variability in how paid work and housework are divided following first birth.

In addition to examining potential heterogeneity in the importance of relative resources in shaping changes in the gender division of labor after first birth, I also address several limitations of prior empirical research on gender specialization during the transition to parenthood. First, prior research has tended to focus on average changes in paid and unpaid work behavior after first birth among individuals (exceptions include: Sanchez & Thomson, 1997; Wood et al., 2018). However, understanding changes in the gender division of labor also requires analysis of changes in individuals’ behavior relative to their partners’. In this study I examine both absolute changes and within-couple relative changes in paid and unpaid work time to capture how relative resources affect both individual and couple behavior following parenthood. Second, while most prior studies have tended to focus on either paid work or unpaid work outcomes, precluding consideration of whether the mechanisms of specialization differ across paid and unpaid work, I examine the division of paid and unpaid work in tandem. Finally, a number of studies have focused exclusively on divisions of labor among dual-earner couples where both partners are in full-time employment (Wood et al., 2018; Yavorsky, Kamp Dush, & Schoppe-Sullivan, 2015). However, for some families the transition to parenthood may necessitate the transition to employment for one or both partners. Therefore, analysis of dual-earner couples only misses potentially informative heterogeneity within the population. By analyzing single-earner couples as well as dual-earner couples, I am able to examine changes in the gender division of labor among partners who may, for example, have particularly low intra-couple economic bargaining power.

Data and Sample

Data were drawn from a harmonized version of the British Household Panel Survey (BHPS) and its successor study, the UK Household Longitudinal Study (UKHLS). BHPS and UKHLS are nationally representative longitudinal household surveys collecting information from all household members on topics including work, education, income, social attitudes and, importantly for this study, weekly time spent on paid work and housework. I include waves 1–18 of BHPS, spanning from 1991 to 2008, and waves 1–10 of UKHLS, spanning 2009–2020.

My sample consists of 1,937 different-sex couples residing in Great Britain (England, Scotland and Wales) who became parents between the ages of 20 and 45 and for whom data is available for both partners at least one wave prior to and one wave following the birth of their first child. The selection of couples who became parents was based on having no biological or adopted children living in the household before the birth. Couples consist of both married and cohabitating partners (although partners are referred to as “husbands” and “wives” for simplicity). As Northern Ireland was not included in BHPS until 2001, it is excluded from the analysis for consistency. I limited my sample to couples whose weekly time spent on paid work and housework were available for both partners in each wave, excluding 71 couples missing paid work information and 552 couples missing housework information (which was not collected in some waves of data). Finally, I further restricted my sample to couples where income, employment and education data were available for both partners in each wave, rendering an analytic sample of 1,606 couples (3,212 individuals) for paid work and 1,255 (2,510 individuals) for housework. On average, couples were observed in 4.1 waves during my period of analysis, with a minimum of 2 and a maximum of 6 waves.

Measures

I analyzed six dependent variables reflecting women and men’s absolute time allocations to paid work and housework as well as their relative allocations within couples.

• Wives’ and husbands’ weekly hours of paid work was derived from survey questions regarding respondents’ typical weekly hours of paid work at their main job, any second jobs, overtime and self-employment. Weekly paid work time was top coded at 80 hours per week, affecting approximately 1% of observations;

• Wives’ share of couple paid work was created by dividing wives’ weekly hours of paid work time by the sum of both partners weekly paid work time;

• Wives’ and husbands’ weekly hours of housework was derived from a survey question asking respondents’ typical weekly hours allocated to housework (specified in the questionnaire as activities such as cooking, cleaning and laundry). As with paid work, weekly housework time was top coded at 80 hours per week, affecting less than 0.1% of observations;

• Wives’ share of couple housework was created by dividing wives’ weekly hours of housework by the sum of both partners’ weekly housework time.

The main independent variables in this analysis include post first birth, share of pre-parenthood couple earnings and household income tertile. Post first birth is a binary measure indicating whether the observation occurred before first birth (which includes observations more than 9 months and less than two years prior to first birth) or after (including observations up to four years after first birth). Observations within one year following first birth were excluded from this analysis to avoid capturing atypical paid work and housework hours during periods of maternity leave. However additional analysis including observations within the first year of parenthood show substantively similar results to those presented here and are available upon request.

Share of pre-pregnancy couple earnings measures partners’ within-couple relative earnings prior to pregnancy and was derived from respondents’ gross monthly labor earnings the month prior to interview (including earnings from usual pay, second jobs and self-employment), expressed in 2020 GBP. Pre-pregnancy relative earnings is defined as a categorical variable equal to 1 where the husband earned 60% or more of couple earnings pre-pregnancy (referred to throughout as “male breadwinner couples”), equal to 2 where both partners earned between 40 and 60% of couple income (“dual-earner couples”) and equal to 3 where the wife earned 60% or more of couple income or more (“female breadwinner couples”). Alternative breadwinner thresholds as well as a continuous measure of relative earnings were also tested and are discussed further in “Robustness Analyses.”

Finally, household income tertile measures couples’ financial position relative to other households prior to pregnancy and was derived from a measure of gross monthly income from all household members from all income sources (including earnings, social benefits, investments, and private benefits), expressed in 2020 GBP. Household income tertiles were calculated within the analytic sample of couples transitioning to parenthood and tertiles are referred to “low-,” “middle-” and “high-” income couples throughout the text for simplicity.

Models controlled for age (centered at the sample mean, 33), whether any subsequent births occurred since the transition to parenthood, changes in gross monthly household income since pre-pregnancy and couples’ Government Office Region of residence. I also controlled for a series of pre-pregnancy characteristics including respondents’ own absolute gross monthly labor earnings and their own earnings squared, educational attainment (measured as a binary indicator equal to one if an individual had attained a university degree or higher), whether the respondent was employed during the week prior to interview, and their occupational class (measured as a binary indicator equal to one if an individual was currently or most recently employed in a professional occupation in the year prior to pregnancy and zero otherwise). Models also controlled for partners age (centered at the mean), and partners’ pre-pregnancy employment status, educational attainment and occupational class¹. Finally, models included a series of binary variables for survey year to control for unobserved temporal trends and to allow for time trend analysis.

Analytic approach

I used random effect panel regression analysis² with year fixed effects to explore the extent to which gender divisions of household labor vary by pre-pregnancy relative earnings and household income, using data from one wave before the transition to parenthood and at least one wave (and up to four waves) after for each individual in my sample. Although time allocation data was available for paid work in every wave, data on housework was only available in waves 2–20, 22, 24, 26 and 28. Analysis of housework is limited to those waves, and sample sizes are therefore smaller for these analyses than for analyses of paid work. Regressions are unweighted and standard errors are clustered at the primary sampling unit level to account for the multistage sampling design.

My analysis consists of two main specifications for each dependent variable:

Model 1 is the baseline model, assessing the extent that relative earnings moderate changes in men’s and women’s paid and housework time as well as how that time is divided within couples over the transition to parenthood. It includes all variables described above as well as an interaction term between post first birth and pre-parenthood share of couple earnings.

Model 2 adds a three-way interaction term between post first birth, pre-parenthood share of couple earnings and pre-parenthood household income tertile to Model 1. The addition of this interaction allows me to test H1 and H2 and assess whether the moderating effect of relative earnings on specialization following the transition to parenthood differs across the household income distribution.

Results

Descriptive results

Table 2 shows men’s and women’s average weekly time allocations to paid work and housework and the division of paid work and housework before and up to four years following first birth, disaggregated by household income tertile. Results show increased specialization within couples across all household income groups following first birth. Wives’ share of couple paid work time decreases substantially following first birth and wives’ share of housework increases. For paid work, this pattern is driven largely by large decreases in women’s hours of paid work after first birth (between 11.4 and 16.0 hours per week across income groups). Men’s hours of paid work also change following parenthood, although to a lesser extent than for wives and the direction of this change varies by household income. Paid work time among husbands in low-income households increased by around 2 hours following first birth while it decreased by a similar amount among middle income and high-income husbands. Notably, although wives’ hours of paid work were positively associated with household income both before and after first birth, statistically significant differences in wives’ share of total couple paid work by household income only emerged after parenthood, driven by particularly large specializing effects of parenthood among low-income couples.

Table 2.

Mean Division and Hours of Paid Work and Housework Before and After First Birth by Household Income

	Low income		Middle income		High income
	Pre-pregnancy After first birth		Pre-pregnancy After first birth		Pre-pregnancy After first birth
Wives’ share of couple paid work	41.6	28.7+	45.5	33.2+ *	46.4	33.5+ *
Wives’ weekly hours of paid work	30.8 (16.6)	19.4 (16.2)+	38.1 (12.3)*	23.5 (15.9)+ *	41.4 (13.1)*	25.4 (17.0)+ *
Husbands’ weekly hours of paid work	41.5 (16.5)	43.6 (15.0)+	44.7 (11.5)*	43.3 (13.3)+	46.6 (12.3)*	44.4 (12.7)+
Wives’ share of couple housework	65.1	74.2+	61.8	71.5+	61.0	70.0+ *
Wives’ weekly hours of housework	9.9 (6.2)	17.3 (10.2)+	8.7 (5.0)*	15.2 (9.4)+ *	8.4 (5.2)*	14.3 (9.5)+ *
Husbands’ weekly hours of housework	5.1 (4.0)	5.4 (4.8)	5.4 (4.0)	5.5 (4.7)	5.2 (3.6)	5.5 (4.3)

Note: N = 1606 for paid work hours and share and 1255 for housework hours and share.

⁺indicates that after-first-birth values differ from pre-pregnancy at p<0.05;

^*indicates values differ from those among low-income couples at p<0.05.

Increased specialization in the division of housework after first birth is primarily driven by large increases in wives’ housework time, with the largest increases observed among low-income wives (7.4 hours per week, compared with 6.5 and 5.9 hours among middle- and high-income women). Consistent with prior research (e.g., Heisig 2011), women’s hours of housework time are negatively associated with household income both before and after first birth. However, as with paid work, differences in wives’ share of couple housework by household income only emerge after first birth. Husbands’ housework did not change meaningfully following first birth and did not vary statistically significantly by household income before or after first birth.

Table 3 shows pre-pregnancy descriptive statistics for husbands and wives in my sample, disaggregated by household income tertiles. Across household income groups, husbands are older, more likely to be employed and earn more, on average, than wives. In middle- and high-income couples, wives are more likely than husbands to hold a college degree. Across income groups, a similar proportion of couples (40.2–42.2%) are male breadwinner couples in which husbands’ earnings constitute 60% or more of total couple earnings. However, female breadwinner couples, where wives earn 60% or more of total couple earnings, are most common among low-income couples, accounting for around 20.4% of couples in that income group compared with 12.8% and 10.5% among middle- and high-income couples, respectively.

Table 3.

Pre-Pregnancy Descriptive Statistics

	Low income		Middle income		High income
	Wives	Husbands	Wives	Husbands	Wives	Husbands
Age - years	27.3 (4.9)	29.9 (5.3)+	29.3 (4.3)*	31.3 (4.6)+ *	30.9 (3.9)+	33.0 (4.4)+ *
College degree - %	21.2	18.1	39.8*	33.9+ *	65.8+	58.6+ *
Employed - %	84.1	92.2+	96.4*	98.9+ *	95.8+	98.9+ *
Professional occupational class - %	28.7	25.9	49.9*	51.9 *	79.4+	73.7 *
Monthly labor income - 2020 GBP	1134.4 (739.4)	1499.7 (801.9)+	1980.0 (815.4)*	2514.6 (800.0)+ *	3097.9 (1584.8)+	4311.8 (2170.0)+ *
Monthly household income - 2020 GBP		2793.6 (799.2)		4736.4 (534.6)*		8257.3 (2826.3)*
Couple type (relative earnings) - %
Male breadwinner		42.2		40.2		40.3
Dual earner		37.4		46.9*		49.3*
Female breadwinner		20.4		12.8*		10.5*
N waves observed		4.1 (1.1)		4.2 (1.1)		4.0 (1.1)
N	540	540	522	522	544	544

Note: Male breadwinner couple: wife earns ≤40% couple earnings; Dual-earner couple: wife earns 40–60% couple earnings; Female breadwinner couple: wife earns ≥60%.

⁺indicates that values differ by gender at p<0.05;

^*indicates values differ from those among low-income couples at p<0.05.

Model 1 Results: Changes in the Division of Labor after First Birth by Relative Resources

To aid interpretation, regression results are presented as changes in predicted outcomes after first birth and graphically as predicted outcomes over time. Full regression results are available in Appendix 1 Table 1. As shown in Table 4, the results from Model 1 for paid work outcomes indicate that, contrary to gendered relative resource expectations, shares of paid work decrease most among female breadwinners and least among women in male breadwinner couples. Among female breadwinner couples, wives’ share of couple paid work time decreases 17 percentage points after first birth compared with 11 and 10 percentage points among women in dual-earner and male breadwinner couples, respectively. This pattern can be observed visually in Figure 1 (left panel) where declines in wives’ predicted share of couple paid work are steepest among female breadwinner couples and shallowest among male breadwinner couples.

Table 4.

Model 1 results: Difference in Predicted Paid Work outcomes from Pre-Pregnancy to After First Birth by Relative Earnings

	Wives’ share of paid work	Wives’ hours of paid work	Husband’s hours of paid work
Male breadwinner	−0.10 +	−11.16 +	−0.38
Dual earner	−0.11 +	−13.97 + *	−0.73
Female breadwinner	−0.17 + *	−12.90 +	4.00 + *

Note:

⁺indicates that predicted changes from pre-pregnancy values differ from zero at p<0.05;

^*indicates that predicted changes from pre-pregnancy values differ from those for male breadwinner couples at p<0.05. Full regression results are available in Appendix 1.

FIGURE 1.

Model 1 Results: Predicted Share and Weekly Hours of Paid Work Over the Transition to Parenthood by Relative Earnings

Greater specialization in paid work among female breadwinner couples after first birth is driven by a combination of changes in both men’s and women’s paid work time. After first birth, women in female breadwinner couples reduce their paid work time by 12.9 hours per week and their husbands increase their paid work time by 4.0 hours. Wives in male breadwinner couples reduce their paid work hours slightly less than breadwinning wives (11.2 hours) and wives in dual-earner couples reduce their paid work time most (14.0 hours) after first birth, suggesting a non-linear relationship between relative earnings and changes in wives’ paid work. Meanwhile, husbands’ paid work time among dual-earner and male breadwinner couples remain stable following the transition to parenthood.

Model 1 results for wives’ share of housework, shown in Table 5, show little variation by relative earnings, with wives’ share of housework increasing by 7 percentage points after first birth among male breadwinner couples and 8 percentage points among dual earner and female breadwinner couples. However, the results for women’s housework time are largely consistent with gendered relative resource expectations. Housework time among women in male breadwinner couples increases the most after first birth – 5.9 hours per week compared with 4.6 and 3.3 hours among women in dual-earner and female breadwinner couples, respectively. Shown visually in the center panel of Figure 2, the steepest increase in housework hours after first birth can be seen among wives in male breadwinner couples while the shallowest increase is observed among breadwinning wives. By contrast, men’s housework time does not change appreciably after first birth, regardless of relative earnings.

Table 5.

Model 1 results: Difference in Predicted Housework outcomes from Pre-Pregnancy to After First Birth by Relative Earnings

	Wives’ share of housework	Wives’ hours of housework	Husbands’ hours of housework
Male breadwinner	0.07 +	5.86 +	0.24
Dual earner	0.08 +	4.63 + *	0.13
Female breadwinner	0.08 +	3.29 + *	−0.57*

Note:

⁺indicates that predicted changes from pre-pregnancy values differ from zero at p<0.05;

^*indicates that predicted changes from pre-pregnancy values differ from those for male breadwinner couples at p<0.05. Full regression results are available in Appendix 1.

FIGURE 2.

Model 1 Results: Predicted Share and Weekly Hours of Housework Over the Transition to Parenthood by Relative Earnings

Taken together, Model 1 results show mixed findings regarding how relative earnings affects the gender division of labor following first birth and highlight the importance of examining both paid and unpaid work outcomes as well as the importance of examining absolute and relative time allocations. Although partners’ pre-pregnancy relative earnings moderate changes in wives’ housework time after first birth, this is not the case for wives’ paid work time, men’s paid work or housework time nor how paid work and housework time was divided within couples. In fact, paid work results often contradict gendered relative resource expectations, with larger changes observed among women with greater relative earnings pre-pregnancy. This results in greater similarity in paid work behavior within genders following first birth, despite differences in relative earnings, and greater time on and responsibility for housework among wives across the relative earnings distribution.

Model 2 Results: Heterogeneity by Household Income

The results from Model 2, which includes a triple interaction term between post first birth, pre-pregnancy share of earnings and pre-pregnancy household income, demonstrate that the relationship between relative income and changes in paid work and housework outcomes varies substantially by overall household income. In fact, they suggest that the counter-relative resource patterns observed in Model 1 are driven primarily by patterns among low-income couples.

Model 2 results for paid work outcomes are presented as changes in predicted outcomes after first birth in Table 6 and graphically as predicted outcomes in Figure 3. Full regression results are available in Appendix 1 Table 2. Shown in the left column of Table 6 and the left panel of Figure 4, changes in shares of couple paid work time among middle- and high-income couples largely follow gendered relative resource expectations in terms of the direction of association. Consistent with H2, wives’ share of paid work decreases more after first birth the lower their pre-pregnancy relative earnings. However, as indicated in Table 6 these differences by relative earnings are not consistently statistically significant. By contrast, in line with H1, the opposite pattern is observed among low-income couples. In the bottom income tertile, share of couple paid work time among female breadwinners decreases by 28 percentage points after first birth compared with 14 percentage points among women in dual-earner couples and just 6 percentage points among women in male breadwinner couples.

Table 6.

Model 2 results: Difference in Predicted Paid Work outcomes from Pre-Pregnancy to After First Birth by Relative Earnings and Household Income

		Wives’ share of paid work	Wives’ hours of paid work	Husbands’ hours of paid work
Low income	Male breadwinner	−0.06 +	−6.63 +	−0.88
	Dual earner	−0.14 + *	−15.37 + *	−0.99
	Female breadwinner	−0.28 + *	−15.62 + *	11.37 + *
Middle income	Male breadwinner	−0.13 +	−13.42 +	−0.58
	Dual earner	−0.11 +	−13.99 +	−0.91
	Female breadwinner	−0.08 +	−10.74 +	−1.69
High income	Male breadwinner	−0.13 +	−13.74 +	0.23
	Dual earner	−0.10 +	−12.89 +	−0.44
	Female breadwinner	−0.06 + *	−10.11 +	−3.47

Note:

⁺indicates that predicted changes from pre-pregnancy values differ from zero at p<0.05;

^*indicates that predicted changes from pre-pregnancy values differ from those for male breadwinner couples at p<0.05. Full regression results are available in Appendix 1.

FIGURE 3.

Model 2 Results: Predicted Share and Weekly Hours of Paid Work over the Transition to Parenthood by Relative Earnings and Household Income

FIGURE 4.

Model 2 Results: Predicted Share and Weekly Hours of Housework over the Transition to Parenthood by Relative Earnings and Household Income

Contrary to previous research that has suggested that couples’ paid and unpaid work responses to parenthood are driven primarily by changes in women’s behavior, with husbands’ behavior relatively fixed (e.g. Killewald & García-Manglano, 2016), results indicate that the dramatic shift in the division of paid work time observed among low-income couples is driven by changes in both men’s and women’s paid work hours. Notably, female breadwinners in low-income couples reduce their paid work hours by 15.6 hours per week after first birth (compared with 15.4 hours and 6.6 hours among women in dual-earner and male breadwinner couples) and men in low-income female breadwinner couples increase their paid work time by 11.4 hours per week (compared with decreases of 1.0 and 0.9 hours among men in dual-earner and male breadwinner couples).

These results constitute a stark reversal of pre-pregnancy divisions of paid work among female breadwinner couples at the bottom of the household income distribution. They also indicate that the convergence in paid work time by relative earnings (and the division thereof within couples) after first birth observed in Model 1 is driven by low-income couples. As shown graphically in Figure 3 (top panel), although women and men’s paid work time and share varied substantially by their relative earnings prior to pregnancy among low-income couples, this variation is almost entirely eliminated following first birth. By contrast, higher up in the income distribution, variation in women’s paid work time and share by relative earnings increased or remained stable following first birth.

Model 2 results for housework are shown in Table 7 and Figure 4. Similar to the results for paid work and consistent with expectations in H1, I find that among low-income couples, changes in women’s share of couple housework time contradict gendered relative resource expectations. Among female breadwinner couples, women’s share of couple housework time increases 12 percentage points following first birth compared with 8 percentage points in dual-earner couples and 5 percentage points in male breadwinner couples. However, in contrast with H2, changes in wives’ share of housework do not vary statistically significantly by relative earnings among middle- or high-income couples.

TABLE 7.

Model 2 results: Difference in Predicted Housework outcomes from Pre-Pregnancy to After First Birth by Relative Earnings and Household Income

		Wives’ share of housework	Wives’ hours of housework	Husbands’ hours of housework
Low income	Male breadwinner	0.05 +	6.86 +	0.56
	Dual earner	0.08 +	5.27 +	0.46
	Female breadwinner	0.12 + *	4.88 +	−1.15 + *
Middle income	Male breadwinner	0.09 +	5.97 +	−0.20
	Dual earner	0.09 +	4.64 +	−0.04
	Female breadwinner	0.06 +	3.43 + *	0.02
High income	Male breadwinner	0.05 +	4.59 +	0.31
	Dual earner	0.08 +	4.08 +	0.07
	Female breadwinner	0.05	−0.17*	−0.09

Note:

⁺indicates that predicted changes from pre-pregnancy values differ from zero at p<0.05;

^*indicates that predicted changes from pre-pregnancy values differ from those for male breadwinner couples at p<0.05. Full regression results are available in Appendix 1.

In general, increases in women’s hours of housework after first birth among low-, middle- and high-income couples reflect gendered relative resource expectations in terms of the relative magnitude of changes by relative earnings. Women in dual earner and female breadwinner couples increase their housework time less than women in male breadwinner couples, however only differences between female breadwinner and male breadwinner wives in middle- and high-income couples are statistically significant. Among middle-income couples, female breadwinning wives increase their housework time 3.4 hours after first birth compared with 6.0 hours among male breadwinner wives. Notably, housework time among high-income female breadwinner wives does not increase statistically significantly at all following first birth while high-income male breadwinner wives increase their housework time by 5.6 hours per week. Men’s housework time remains largely unchanged following first birth, regardless of relative earnings or household income. However, husbands in low-income, female breadwinner couples reduced their housework time by 1.2 hours after first birth.

Robustness Analysis

I conducted five supplementary analyses to test the robustness of my results. First, I tested whether results vary by time since the transition to parenthood, replacing the post first birth indicator in my main models with a categorical variable measuring years since first birth (ranging from one year pre-pregnancy up to four years after first birth). Results are presented in Appendix 2 and show remarkable consistency across post-first birth years. These results indicate that new gender divisions of labor forged following the birth of a child – and heterogeneity therein – have lasting consequences within couples.

Second, I tested models replacing pre-pregnancy relative earnings with pre-pregnancy relative hourly wages. My main analysis uses relative monthly earnings as a measure of relative resources because it most closely reflects the balance of economic power within a relationship and is commonly used in previous research on the gender division of labor (e.g. Bittman et al., 2003; Musick, Bea, & Gonalons-Pons, 2020). However, earnings are also dependent on hours of paid work, raising concern that paid work results may be driven by potential reverse causality in pre-pregnancy years. In this additional analysis, respondent hourly wages were constructed by dividing reported monthly labor earnings by reported typical weekly paid work hours multiplied by 4.33 and relative wages were measured by dividing respondents’ hourly wage by the sum of both partners’ wages. Models also controlled for pre-pregnancy hourly wages and hourly wages squared rather than pre-pregnancy earnings. Hourly wages could not be calculated for a small number of respondents who reported labor earnings in the month prior to interview but reported no hours of paid work due to employment changes. These respondents and their partners (11 couples for paid work outcomes; 20 couples for housework outcomes) were excluded from this analysis. Presented in Appendix 3, results from this analysis are consistent with my main results. Wives’ shares and hours of paid work tend to follow a gendered relative resource gradient among middle- and high- income couples while the opposite pattern is observed among low-income couples.

Third, I tested the sensitivity of my results to several alternative definitions of pre-pregnancy relative earnings. In my main analysis the threshold for male (and female) breadwinner couples was 60% of couple earnings, with dual-earner couples defined as those in which each partner earned 40–60% of couple earnings. In this analysis I tested two alternative breadwinner thresholds. First, I tested a lower breadwinner threshold of 55%, with dual-earner couples defined as those in which each partner earned between 45–55% of couple earnings. Next, I tested a higher breadwinner threshold of 65%, with couples earning between 35–65% of couple earnings defined as dual-earner couples. Finally, I tested models using a continuous rather than categorical measure of relative earnings.

The results from these analyses are presented in Appendix 4. For alternative threshold models, results are presented as predicted outcomes and changes in predicted outcomes after first birth, similar to my main models (Appendix 4 Figures 1-4 and Appendix 4 Tables 1-8). For models using a continuous measure of relative earnings, results are presented as regression coefficients (Appendix 4 Tables 9-10) and graphically as predicted outcomes at wife earns 25%, 50% and 75% of couple earnings (Appendix 4 Figures 5-6). The results from all three alternative analyses show similar patterns to my main specifications in both magnitude and statistical significance.

Fourth, I repeated my analysis, substituting measures of pre-pregnancy relative earnings, pre-pregnancy household income and pre-pregnancy individual earnings with measures averaged over two pre-pregnancy waves. Because partners’ incomes may fluctuate over time (e.g., Winslow-Bowe, 2006), averaging measures across two waves may more accurately capture longer-term patterns. However, requiring two pre-pregnancy waves reduced sample size and therefore precision: my sample size was reduced by 473 couples for paid work analysis and by 262 couples for housework analysis resulting in a total of 1,133 and 993 couples for each analysis, respectively. A comparison of relative earnings groups measured in one wave vs. averaged over two pre-pregnancy waves shows a high degree of consistency, particularly among male breadwinner and dual earner couples (shown in Appendix 5 Table 1). Female breadwinner couple was least consistent across the two measures of relative earnings group, however, more than two-thirds (69%) of couples categorized as female breadwinner using a single pre-pregnancy observation remained in this group after averaging earnings over two years.

The results for this robustness analysis are presented in Appendix 5 Tables 2-5 and shown graphically in Appendix 5 Figures 1-2 (the results from my main specifications on this limited analytic sample are available for comparison in Appendix 5 Tables 6-9 and Figures 3-4). These results show similar patterns to those found in my main paid work specifications.

Finally, I tested whether my main results were driven by single earner couples (either by choice or by temporary unemployment) by re-running my main specifications limiting my sample to couples where both partners were working for pay prior to pregnancy. Dropping 180 single-earner couples from my paid work sample and 118 couples from my housework sample, this resulted in analytic samples of 1,426 and 1,237, respectively. Results are shown in Appendix 6 Tables 1-4 and shown graphically in Appendix 6 Figures 1-2. Results for both paid work and housework outcomes are substantively consistent with those from in my main specifications. For paid work outcomes, however, excluding single-earner couples results in stronger relative resource gradients in changes in women’s share and hours of paid work after first birth among middle- and high-income couples and almost no relative earnings gradient among low-income couples. This suggests that my main results are partly driven by transitions to employment following first birth, particularly among low-income couples. Still, the results of this analysis support my hypotheses that the moderating effect of relative resources on the gender division of labor after first birth differs by household income.

Discussion

Most prior research on the gender division of household labor among heterosexual couples has sought to identify general, robust mechanisms explaining specialization. In this study, I examined whether the moderating effect of relative resources on changes in the gender division of labor after first birth varies across the household income distribution using 28 waves of longitudinal data from the British Household Panel Survey and the UK Household Longitudinal Study. The results show substantial heterogeneity in the relevance of relative resources in shaping specialization of paid work and housework after first birth.

Reflecting prior research findings (e.g., Kanji, 2011; Kühhirt, 2012; L. Sanchez & Thomson, 1997; Wood et al., 2018), my results indicate that relative earnings moderate (at least to some extent) the gendering effects of parenthood on wives’ paid and unpaid work behavior among middle- and high-income couples. However, this moderating role is relatively minor. Wives with high relative earnings tended to reduce their share and hours of paid work less than wives with low relative earnings after first birth. Similarly, the largest increases in housework time occurred among wives with low relative earnings and the smallest occurred among female breadwinners. However, these differences by relative resources tended to be small in magnitude and were not consistently statistically significant across outcomes, possibly due to the relatively small sample size available. Additionally, changes in paid work and housework among dual-earner wives (whose pre-pregnancy earnings were similar to their husbands) at times did not fit neatly between that of male breadwinner and female breadwinner couples, further complicating relative resource expectations. Changes in men’s paid work and housework time over the transition to parenthood varied little by relative earnings, indicating that any moderating effect of relative resources – however minor – is likely to operate through wives’ paid and unpaid work behavior rather than husbands’.

At the bottom of the household income distribution relative resources do not appear to play the moderating role suggested by gendered relative resource perspectives and, in fact, parenthood appears to negate the effect of pre-pregnancy relative earnings on the division of paid work and housework behavior after first birth. I find that among low-income couples, wives’ share of couple paid work decreased more after first birth among women with greater relative earnings than among those with lower relative earnings, resulting in little variation in couple divisions of paid work by relative earnings in the years following first birth. These steep changes in the division of paid work among low-income female breadwinner couples were driven by both sharp decreases in women’s hours of paid work and substantial increases in men’s paid work, adding nuance to prior studies that have found little variation men’s paid work behavior over the transition to parenthood (Killewald & García-Manglano, 2016; Kühhirt, 2012; Sanchez & Thomson, 1997; Schober, 2013). While this was true among low-income dual-earner and male breadwinner couples (as well as for men in middle- and high-income couples) the transition to parenthood appears to motivate greater time allocations to paid work among men in low-income female breadwinner couples. For housework, too, I found that increases in women’s share of couple housework at the bottom of the household income distribution were largest among female breadwinner couples, driven partly by decreases men’s housework time after first birth among husbands in female breadwinner couples.

These findings are robust to a series of alternative specifications including 1) modelling results by years since first birth; 2) measuring relative resources using relative hourly wages rather than relative earnings; 3) using alternative measures of relative earnings; and 4) averaging income variables over two time periods. Additional analysis limiting the sample to couples where both partners were employed pre-pregnancy showed a stronger moderating effect of relative resources on changes in wives’ paid work outcomes among middle- and high-income couples but no moderating effect among low-income couples. This finding suggests that my main results are driven in part by transitions to employment following first birth, particularly among low-income couples. However, it also supports the hypothesis that the effect of relative earnings in shaping changes in the gender division of labor after first birth differs across the income distribution.

The findings from this study reflect prior research on the classed barriers and facilitators to gender equality within the home. While couples at either end of the household income distribution face a variety of conditions that could promote or constrain gender equality in the division of paid and unpaid work, low-income couples face barriers that are likely to substantially limit their options regarding how to divide paid and unpaid work after the birth of a child. Barriers such as schedule inflexibility and unpredictability of many low-paid jobs (e.g., Harknett, Schneider, & Luhr, 2020; Presser, 2003; Shows & Gerstel, 2009; Usdansky & Wolf, 2008), limited access to work-family policies (e.g. Evertsson et al., 2009; Korpi, Ferrarini, & Englund, 2013; Mandel, 2011) and limited capacity to outsource childcare and domestic responsibilities (e.g. Killewald, 2011; Sullivan & Gershuny, 2013) may explain why gender specialization in paid work and housework increases after first birth among low-income couples regardless of partners’ relative resources. These barriers may effectively “discount” the bargaining power that women derive from their relative earnings, resulting in greater uniformity in divisions of paid and unpaid work regardless of relative earnings (Blumberg & Coleman, 1989; Fuwa, 2004). Alternatively, the relative prevalence of low-quality employment among lower-paid individuals (e.g., Damaske, 2011; Warren & Lyonette, 2018) combined with prevailing gender norms may mean that low-income wives are less likely to draw on their relative resources to bargain out of unpaid work.

Couples with greater household incomes also face barriers to gender egalitarian divisions of labor including workplace cultures of overwork and ideal worker norms (e.g., Cha, 2010; Maume, 2006; Williams, 2000). However, these couples benefit from substantially greater access to institutional and private resources to overcome these barriers, resulting in the greater (albeit still modest) variation in paid work and housework responses to parenthood by relative earnings observed among middle and high-income couples in this study.

Alternatively, these results could reflect class differences in gender role attitudes. Research indicates that greater education and higher occupational status are associated with more gender egalitarian views on women’s paid work and men’s unpaid work (see e.g., Usdansky, 2011). It is possible that the transition to parenthood may activate or make more relevant existing gender-traditional values among low-income couples, even among those with gender non-traditional arrangements prior to first pregnancy. However, previous research shows that there are substantial class differences in the extent to which couples’ divisions of labor align with their gender role attitudes, with “lived” gender egalitarianism exceeding “spoken” egalitarianism among couples with lower incomes or lower-skilled occupations (Usdansky, 2011; see also Lyonette & Crompton, 2015). Furthermore, prior research indicates that the relationship between gender role attitudes and paid work and housework behavior is reciprocal rather than unidirectional: individuals do alter their behavior to fit their gender role attitudes but they also alter their gender role attitudes to match their behavior (Carlson & Lynch, 2013; Schober & Scott, 2012). Data limitations and small sample sizes precluded further analysis of why the effects of relative resources on changes in the division of household labor after first birth varies across the household income distribution, however testing the mechanisms driving this heterogeneity is an important area for future research.

There are a number of limitations to this study. First, this analysis relies on stylized self-reports of time allocation, which are known to suffer from measurement error (Kan, 2008). However, in the absence of large-scale, nationally representative panel time diary studies, this study provides insight into changes in gender divisions of labor over time that cannot be captured with existing time diary data. Second, the data collected on unpaid work time in BHPS and UKHLS is limited, covering female-typed routine housework tasks at the exclusion of other types of housework and, critically, childcare. Future research is needed to understand how relative resources and household income relate to divisions of non-routine housework and childcare in the transition to parenthood. Third, data on respondent use of domestic outsourcing and of work-family balance policies are not consistently available in BHPS and UKHLS, however this information could help shed light on why patterns of relative income and specialization vary so starkly across the income distribution. Finally, the period of analysis in this study (1991 to 2020) spans a period of substantial change in British work-family policy. This includes expansions of paid statutory maternity leave, the introduction of two weeks of statutory paternity leave, transferrable maternity leave, the right to request flexible working and the introduction and expansion of partially subsidized childcare. Samples sizes precluded analysis of whether and how the mechanisms of specialization change in response to these policy interventions, however this is an important question for future study.

Despite these limitations, this study provides valuable new insight into the factors driving increased gender specialization in the transition to parenthood, demonstrating that the importance of relative resources in explaining changes in the gender division of labor varies substantially by household income in Great Britain. The results of this study highlight that, in the context of a liberal welfare state with limited social supports and high levels of economic inequality, the class-specific constraints that low-income parents face have important consequences for gender equality in the household division of labor. Future cross-national research examining whether and how the role of household income varies across inequality, welfare state and work-family contexts could provide important insight into the mechanisms driving these results. Additionally, research on whether the importance of relative resources in moderating the gendering effects of parenthood varies by husbands’ and wives’ own incomes (rather than total household income) could provide further insight into gendered decision-making processes within different-sex couples.

Overall, this study underscores the need for greater attention to heterogeneity in both theoretical and empirical consideration of the factors driving specialization. While this study focuses on the role of household income, further research on other potential sources of heterogeneity, such as race and ethnicity, would help provide a more nuanced perspective regarding when and for whom classically hypothesized drivers of specialization, such as relative resources, apply.