The gender division of housework after the first child: a comparison among Bulgaria, France and the Netherlands

ABSTRACT

In our study, we adopt a comparative-longitudinal perspective on the gender division of housework before and after the birth of the first child, exploiting the first two waves of the Gender and Generation Survey and comparing three countries belonging to different gender and welfare regimes (i.e. Bulgaria, France and the Netherlands). We find that childrearing everywhere triggers a re-traditionalisation, generating a more inegalitarian gender division of housework, yet with interesting differences across countries. Fixed-effect regression analyses of the pooled data show that changes when becoming parents are less pronounced in France with respect to Bulgaria and the Netherlands, more pronounced when she is low-educated. Moreover, when countries are analysed separately, it emerges that it is only in the Netherlands that traditionalisation around first childbirth is significantly lower in couples where the woman is middle- and high-educated compared to those where she is low-educated. Economic, cultural and institutional contexts do matter. Traditionalisation and within-couples polarisation is weaker in contexts where non-traditional attitudes are widespread, social policies are more defamilialising and more explicitly addressed also to men, and where part-time is not the main reconciliation strategy, as in France.

RÉSUMÉ

Notre étude s’inscrit dans une démarche comparative-longitudinale appliquée à la répartition des tâches domestiques entre hommes et femmes avant et après la naissance du premier enfant, en exploitant les deux premières vagues du Generations and Gender Survey et en comparant trois pays appartenant à des régimes de genre et d’état-providence différents (la Bulgarie, la France et les Pays-Bas). Il en ressort que l’éducation des enfants provoque partout une re-traditionalisation, générant une répartition des tâches domestiques plus inégalitaire, avec cependant des différences intéressantes entre les pays. L’application du modèle à effets fixes à l’analyse des données recueillies montre que les changements liés au nouveau statut de parents sont moins prononcés en France par rapport à la Bulgarie et aux Pays-Bas, et plus prononcés lorsque la mère a un niveau d’éducation plus faible. De plus, si on analyse les pays séparément, il apparait que c’est uniquement aux Pays-Bas que la traditionalisation à la naissance du premier enfant est significativement plus basse dans les couples où la femme a un niveau d’éducation moyen ou élevé que lorsque son niveau d’éducation est faible. Les contextes économique, culturel et institutionnel ont une influence. La traditionalisation et la polarisation à l’intérieur du couple sont en effet plus faibles dans les contextes où les comportements non traditionnels sont répandus, où les politiques sociales sont plus dé-familiarisantes et plus explicitement adressées également aux hommes, et où le temps partiel n’est pas la principale stratégie de réconciliation, comme c’est le cas en France.

KEYWORDS:

• Parents
• housework
• gender inequality
• Bulgaria
• France
• Netherlands

MOTS-CLÉS:

• Parents
• travail domestique
• inégalité de genre
• Bulgarie
• France
• Pays-Bas

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Letizia Mencarini is Associate Professor of Demography at Bocconi University; Fellow at the DONDENA Center for Research on Social Dynamics and Public Policy at Bocconi University and at the Center for Demography and Ecology at University of Wisconsin–Madison. She is the principal investigator of SWEL-FER, an ERC funded research grant on subjective well-being and fertility, and chair of the IUSSP (International Union for Scientific Studies of Population) world panel on subjective well-being and life-course. She is an expert of Population Europe, among the Leading Women Scientists of AcademiaNet, and member of the editorial board of Genus and Population Review. She is also on the editorial board of Neodemos, an online Italian blog on population, society and politics and of the international demographic magazine N-IUSSP. Her research interests focus on family demography (life–course analysis, fertility, transition to adulthood, family formation and disruption) and its links with well-being, time use, gender diversity and migration in a policy and welfare state perspective.

Cristina Solera is Assistant Professor of Sociology in the Department of Cultures, Politics and Society at the University of Turin and fellow at Collegio Carlo Alberto, Moncalieri. She is member of the Scientific Committee of ECSR (European Consortium Sociological Research) Spring School and of MAPS (Master in Public Policy and Social Change) at Collegio Carlo Alberto. She is also member of the management committee of ‘Centro Interdisciplinare di Ricerche e Studi delle Donne e di Genere – CIRSDe’. She has taken part to various international projects and networks, such as bEUcitizen – Barriers Towards Eu Citizenship (FP7-SSH-2012 http://beucitizen.eu). Since her PhD at the European University Institute, through a life course and comparative perspective she has been working on the effect of education and occupational class on women's labour market participation and fertility, on the link between family formation and employment interruptions, on care-work reconciliation strategies (also towards the elderly), on gendered divisions of unpaid work and the emergence of new models of fatherhood and fathering, on gender inequalities in academia.

Notes

1 Focusing only on childless couples obviously implies a selection: those not in a couple or already parents are excluded. The share of respondents aged less than 45 years old not cohabiting or being married amounted to 30% in Bulgaria and France and 24% in the Netherlands, while, among those in a partnership, those with one or more children amounted to 93% in Bulgaria, 75% in the Netherlands and 81% in France. Yet, because we were interested in gender patterns connected to the transition to parenthood, we had to focus only on couples and on couples without children, and thus incurred a strong reduction of sample sizes. Fortunately, no further reduction was caused by missing values which, in our selected sample, accounted for only around 3% for the housework tasks composing our dependent variable, and between 1% and 4% for the crucial covariates (having a child, his and her educational attainment and employment status).

2 By construction, the woman's share of domestic work can assume only the following values (percentages of share): 0, 8.3, 16.7, 25, 33.3, 41.7, 50, 58.3, 66.7, 75, 83.3, 91.7, 100. Consequently, differences across waves in the woman's share can be: (±) 0, 8.3, 16.7, 25, 33.3, 41.7, 58.3, 100. Because of these distributions, the following thresholds have been chosen when distribution of levels of share and change are shown in categorical rather than linear terms: for level of the woman's share in wave 1 or 2 we distinguish among egalitarian couple (her share < 50%), traditional couple (her share 50-74%) and very traditional couple (her share >74%); for level of change in her share (w2-w1) we distinguish among becoming much more equal (–100 to –10), becoming slightly more equal (–9 to –1), no change (0), becoming slightly more unequal (0–9), becoming much more unequal (10–100).

3 Unlike child status (having had or not the first child) and labour market status, which are time-varying covariates, educational level is in our models a time-constant covariate. Indeed, although it could theoretically change across waves (3 years time), since our focus is on women and men in stable partnership at risk of making the transition to parenthood (living together as cohabitants or spouses both at first and second wave, childless at first wave with the woman less than 45), level of education becomes de-facto time-constant. Indeed, average age in our sample is 29.9 for women, and 33.1 for men, when nearly all have completed education; and for the few that are still in education (1.3% in our selected sample) the highest level attained so far is considered.

4 Distinctions between types of reduction or types of increase, that is, distinguishing movements between fulltime to part-time work position from those between fulltime or part-time and not employment, would be interesting. Yet, small sample sizes and cells sizes, especially in Bulgaria where part-time employment is quite rare, do not allow for such distinctions.

5 Since GGS data are not are not dyadic data, that is, information are only asked to one adult member of the household (in our 3 countries couples sample respondents are 44% male and 56% female), the reported gender division of housework could be biased towards the respondent's own gender. Moreover, according to the relative resources perspective, also the relative income of one partner compared to the other, only partially captured by relative educational levels and working hours, can affect gender allocations to unpaid work.

6 In addition to “objective” behaviours, GGS also collects info on attitudes, which is, compared to other cross-country datasets, one of its strength. Yet, we have not included measures of attitudes in our models because in GGS attitudinal items are only asked to the respondent and they are not consistent across countries. In particular in the Netherlands, unlike in France and Bulgaria, only items measuring attitudes toward family formation are included (such as “It is all right for unmarried couples to live together”) rather than those toward gender roles (such as “A pre-school child is likely to suffer if his/her mother works” or “Children often suffer because fathers concentrate too much on work”).

7 Own extra calculations.

8 Since in Fixed-Effects models all time-invariant observed and unobserved heterogeneity is controlled for, the risk of the omitted variable bias is greatly reduced. Yet, it is not completely “eliminated” since FE models cannot control for time-varying unobserved factors, and reverse causality remains possible. Also in our case one could argue that changes in the division of housework can “push” couples to make the transition to first child. Although this is possible in theory, we think that it is much more plausible that the causality goes in the opposite direction, i.e. that the childbearing event drives changes in household sharing. The problem of unobserved heterogeneity and reverse causality remains for the covariates referring to changes in his and her labour market participation. Decisions on whether and to what extent women should work are closely intertwined with (put differently, endogenous to) decisions on whether and to what extent women and men should devote themselves to domestic and care work. Yet, as in other studies (e-g: Bittman et al., 2003; Fuwa, 2004; Geist, 2005), labour market participation is for us only a control variable conceived as a proxy for “time availability” whose introduction or not does not substantially change the effect of our crucial covariates (childbirth and education).

9 There is much evidence that nowadays it is the profile of the woman, rather than the profile of the male partner, that drives her labour supply and share of unpaid work (e.g.: Solera, 2018). Yet we also ran a model with the interaction of childbirth with the educational level of the man, or with the educational profile of the couple (both tertiary educated, only he, only she, neither) and the “story” did not change.