Are gender role beliefs entrenched?
Empirical evidence from the Netherlands
during COVID-19.
Andrea D’Orta
Supervisor: Eva Ranehill
Master’s thesis in Economics, 30 hec
Spring 2024
Graduate School, School of Business, Economics and Law, University of Gothenburg, Sweden
Master Thesis in Economics*
Are gender role beliefs entrenched? Empirical evidence from
the Netherlands during COVID-19
Andrea D’Orta✝
May, 2024
Abstract
The development toward more egalitarian gender norms has in many countries
represented a fundamental shift in how societies are organized. However, while this
development is well documented, few academic studies address why such a change in
norms has occurred. In this paper, I rely on theories of motivated beliefs to study
whether gender norms respond to economic incentives. With women often seen as
the primary caregivers, adherence to gender-neutral roles becomes more convenient for
mothers, while more onerous for fathers. Such changes in the cost of upholding equality
would trigger individuals to adjust gender role attitudes in lowering their household
contribution. I use COVID-19 and the resulting closures of childcare centers and schools
as an exogenous shock to the cost of child-rearing under egalitarian gender norms. By
employing an Event Study approach, I assess the effects of interest in the short and
long term across various levels of household constraints. I draw on a representative
sample of Dutch individuals from the LISS panel to explore whether fathers (mothers)
of young children express less (more) egalitarian attitudes when impacted by school
closures, and how beliefs are shaped across various household working arrangements.
I find that men favour more unequal gender roles when having young children unable
to attend school. Women do not exhibit egalitarian reactions to the same extent.
Moreover, continuing to work leads to higher conservatism among parents, especially
fathers, when they are the sole earners in the household.
Keywords: Gender role attitudes, Gender equality, COVID-19, School closure, Childcare.
*Supervisor: Eva Ranehill ✝E-mail address: andreadorta001@gmail.com
Acknowledgements
I am deeply indebted to my supervisor Eva Ranehill for her insightful support and guidance throughout
this research. I extend my gratitude to the participants of the seminar held at the University of
Gothenburg for their valuable comments and suggestions.
Contents
Abstract 1
1 Introduction 4
2 Institutional Background 6
3 Literature 7
3.1 Socio-Demographic Influences on Gender Role Attitudes . . . . . . . . 8
3.2 Gender Role Beliefs and Household Constraints . . . . . . . . . . . . . 9
4 Theoretical Framework 10
5 Data and Empirical Approach 12
5.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
5.2 Classification of Gender Role Attitudes . . . . . . . . . . . . . . . . . . 14
5.3 Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.4 Empirical Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6 Results 19
6.1 Descriptive Evidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6.2 Event Study Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . 21
6.3 Young Children . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
6.4 Working Parents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7 Robustness Analysis 27
7.1 Only Constrained . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
7.2 Children of All Ages . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
7.3 Non-gender Role Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . 29
8 Conclusions 29
Bibliography 31
Appendix 36
1
List of Tables
1 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2 Gender Role Items . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
A1 Trends in Overall Gender Role Attitudes when having Young Children . . . . . . 36
A2 Trends in Statement-specific Gender Role Attitudes when having Young Children . 37
A3 Trends in Gender Role Attitudes among Parents with Young Children when Balanc-
ing Work and Childcare: a Comparison by Gender and Partner’s Working Status . 38
A4 Pre-trends analysis: Joint Statistical Significance of the Lead Estimates . . . . . . 39
A5 ”Only Constrained” Robustness analysis. Trends in Gender Role Attitudes when
having Young Children. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
A6 ”Only Constrained” Robustness analysis. Trends in Gender Role Attitudes among
Parents with Young Children when Balancing Work and Childcare: a Comparison
by Gender and Partner’s Working Status. . . . . . . . . . . . . . . . . . . . . 41
A7 ”Children of All Ages” Robustness analysis. Trends in Gender Role Attitudes when
having Children of All Ages. . . . . . . . . . . . . . . . . . . . . . . . . . . 42
A8 ”Children of All Ages” Robustness analysis. Trends in Gender Role Attitudes among
Parents when Balancing Work and Childcare: a Comparison by Gender and Partner’s
Working Status. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
A9 ”Non-Gender Roles” Robustness analysis. Trends in Attitudes both Related and
Unrelated to Gender Roles when having Young Children. . . . . . . . . . . . . . 44
2
List of Figures
1 Beliefs about Mothers’ Employment and Household Responsibilities in the Nether-
lands (2008-2023). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Distribution of Respondents’ Answers to the seven Gender Role Statements . . . . 20
3 The Development of Men’s Gender Role Attitudes when having Young Children at
home . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4 The Development of Women’s Gender Role Attitudes when having Young Children
at home . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5 Trends in Gender Role Attitudes among Fathers with Young Children when Balanc-
ing Work and Childcare: a Comparison by Partner Working Status . . . . . . . . 25
6 Trends in Gender Role Attitudes among Mothers with Young Children when Bal-
ancing Work and Childcare: a Comparison by Partner Working Status . . . . . . 26
A1 Share of Women working Part-Time Over Female Total Employment . . . . . . . 45
A2 Differences in Overall Gender Role Beliefs during and after School Closure (Wave
13). By Gender, Young Children, and Working Status . . . . . . . . . . . . . . 46
3
1 Introduction
Gender roles are a set of qualities and responsibilities that are commonly associated
with men and women based on their gender. Traditional beliefs generally predict that
men are more suited for employment while women have a natural propensity for child-
rearing. Opposed to this view, equal gender roles prescribe a shared responsibility of
contributing to the household sustenance in terms of paid and unpaid work. Attitudes
on these topics have progressively shifted toward greater egalitarianism over the last
decades of the previous century, coinciding with a steady rise in female participation
in the labour market and high education outcomes (Bolzendahl and Myers, 2004).
Such a pattern has nevertheless levelled off after the mid-90s. While many scholars
have pointed out a stall in the gender revolution (England, 2010; Cotter et al., 2011),
large survey instruments such as the General Social Survey (GSS) indicate more recent
liberal patterns in egalitarian views. Between 1997 and 2021, there has been a 40%
decrease in the share of respondents who agreed with conservative statements such as
”It is better for the man to be the primary breadwinner while the woman takes care
of the home and family” and ”A preschool child is likely to suffer if his or her mother
works”. A similar pattern has been observed for the same gender roles in the Nether-
lands during the last 15 years, as depicted in Figure 1.
Figure 1
Beliefs about Mothers’ Employment and Household Responsibilities in the Netherlands (2008-2023).
Source: Longitudinal Internet studies for the Social Sciences (LISS).
4
I use the COVID-19 outbreak in the Netherlands as an exogenous shock to exam-
ine whether trends in gender norm attitudes reverse due to economic incentives. The
restrictions on school facilities constrained many households’ ability to outsource child-
care activities. Grandparents were not able to provide informal care as an alternative
solution since they were at increased risk of illness. Consequently, parents were com-
pelled to stay home with their kids, leading many to adopt new household arrangements
to cope with the increased care burden. This often led to a more unequal distribution of
unpaid work. Despite women being already the primary caregivers, studies show that
mothers took on the majority of the additional housework and care duties during the
pandemic (Sevilla and Smith, 2020; Farre et al., 2020; Alon et al., 2020; Berghammer,
2022), with only few cases of a more even distribution of household chores (Hupkau
and Petrongolo, 2020). Not surprisingly, working mothers of young children reported
higher difficulties balancing work and family than fathers (Del Boca et al., 2020).
Fathers tend to be more involved in childrearing duties only when they are the sole
household members working from home. Mothers still shoulder most of the additional
work in symmetric working situations (Del Boca et al., 2020; Jessen et al., 2021) and
regardless of who was the top earner before the pandemic (Andrew et al., 2022). It
appears therefore that households did not allocate unpaid work hours based on time
availability and income factors, suggesting that gender norms may play an important
role in this matter.
In light of these circumstances, I use the perspective of motivated beliefs to ex-
plore whether such turmoil alters the cost of upholding equality and, in turn, beliefs
about the rightfulness of gender egalitarianism. Far from treating beliefs as exogenous,
a recent strand of work starts from the perspective that people may shape their beliefs
to align with their best interests and fulfill their ultimate needs (Bénabou and Tirole,
2016; Epley and Gilovich, 2016). One cornerstone of this line of work is that beliefs
adapt to the costs and benefits of acting accordingly. Applied to gender role attitudes,
these theories would predict a reversal in traditional opinions for fathers of young chil-
dren when they are constrained in delegating childcare tasks. Indeed, embracing equal
roles would prescribe higher costs in terms of increased caregiving contribution. More
conservatism can also serve as a justification for those who already do not contribute
enough. Conversely, mothers would advocate for greater equality to avoid carrying the
entire burden themselves. Consistent with this reasoning, I expect the effects would
exacerbate when individuals are the most time-constrained, particularly when both
parents continue working while having young children at home. The same logic sug-
gests that the curb of the pandemic restrictions decreases the cost to men of women’s
employment. With schools reopening and childcare facilities becoming more accessi-
5
ble, families find it easier to outsource their childcare responsibilities. This could lead
to fathers reverting to more gender-neutral views, as it would no longer be seen as
a threat to their (future) careers. This aligns with prior literature (Huebener et al.,
2022) suggesting that the pandemic altered gender ideology only in the short term.
Understanding how people set their beliefs on gender roles sheds light on the impor-
tance of public policies norming daycare provision in shaping such opinions. Moreover,
a reversal in beliefs during tough times would threaten the future of women’s labour
market outcomes. Equal gender roles are frequently associated with higher levels of
employment for women Corrigall and Konrad (2007); Cunningham (2008); Stickney
and Konrad (2007) and have been indicated as a probable explanation for the slow-
down in female labour force participation during the 2000s (Fortin, 2015). Even if
egalitarian attitudes towards female employment are immediately reinstated, a tempo-
rary shift may suffice to push women away from the labour market and reduce their
participation in the coming years.
The research is structured as follows. In Section 2, I illustrate the Dutch institu-
tional background and how the government tackled the pandemic emergency. Section
3 contains a review of the most relevant previous literature on gender role beliefs. Sec-
tion 4 delves into the motivated beliefs theory and its implications for gender roles
through a theoretical application. Section 5 describes the data and the empirical strat-
egy. Section 6 presents the main estimation results of the impact of multiple household
production constraints on attitudes. Section 7 assesses the robustness of the previous
results. Section 8 concludes.
2 Institutional Background
Nowhere do women work part-time as often as in the Netherlands: Figure A1 shows
that the share of working women opting for part-time employment is well above that
of any other OECD country. Informal care, housework and personal spare time are
among the most popular reasons reported by women, especially among mothers of
young children, for not working full-time (Portegijs and van den Brakel, 2018). It
suggests a structural preference of Dutch families to have children raised within the
household, mostly by mothers. The number of children in formal care has however risen
sharply in the last years following an expansion of the right to childcare allowances and
the construction of new childcare facilities in 2016 (ibid.). More and more parents are
also convinced that childcare is a good solution for their children (Brakel et al., 2020).
This rising interest in formal care makes the Netherlands case a salient framework where
to evaluate whether individuals’ gender role attitudes respond to increased household
6
constraints.
To curb the COVID-19 outbreak, the Netherlands’ initial strategy was to tackle the
epidemic by building population immunity1. More restrictive measures were gradually
enforced as soon as the number of reported infections rose. Childcare centres and
schools of any kind officially closed starting from March 16th2. A so-called ”intelligent
lockdown” was announced on March 23th, imposing stricter distancing guidelines and
banning groups of more than two in the public space. Individuals were only encouraged
to work from home as much as possible. Some relaxing measures were introduced
starting from May 11th: childcare facilities reopened and primary school attendance
was granted for half the time before transitioning to full on June 8th. Secondary and
special secondary schools officially reopened on June 2nd, followed by middle-level and
high education from June 15th onward3. As a result of the Second wave, new tightening
measures were instituted in October. A “partial” lockdown was implemented and
immediately followed by a “hard” lockdown that remained in place from December
15th, 2020, until March 15th, 2021. Additionally, childcare and in-presence educational
activities were suspended one day after the hard lockdown began. Initially set for
January 17th, the reopening date was deferred to February 8th for childcare, primary
and special primary schools4. Starting from March 1st, secondary education students
were allowed to attend school one day per week. On December 19th, 2021, a new
lockdown was imposed, and schools of any level closed again, even though to a smaller
window compared to the previous one. Students returned to school on January 10th,
2022 (with an exception for tertiary school) while the lockdown ended on January
14th. With the COVID-19 virus becoming less threatening, these have been the last
pandemic-related restrictions that affected the Dutch education system.
3 Literature
This section reviews the key relevant literature on gender role beliefs. Section 3.1
provides a general overview of how institutions such as marriage and employment
are associated with attitudes toward gender roles. Section 3.2 organizes the available
literature investigating how changes in household constraints lead individuals to shape
their support of egalitarianism.
1 Rutte (2020a)
2 Ministry of Health and Sport (2020)
3 Rutte and Hugo M. (2020)
4 Ministry of Justice and Security and Ministry of Health, Welfare and Sport (2021)
7
3.1 Socio-Demographic Influences on Gender Role Attitudes
My study builds upon a vast strain of literature that explores how attitudes toward
gender role division are formed. Marriage is generally associated with more traditional
attitudes for both genders, with divorced women showing more progressive traits to-
ward family responsibilities (Bolzendahl and Myers, 2004; Davis, 2007). Results are
nevertheless ambiguous as (Fan and Marini, 2000) reports a change only for women
of young age. In addition, they seem to uphold more egalitarian beliefs when having
children (Moors, 2003; Vespa, 2009), suggesting that marriage has a detrimental effect
only when combined with parenthood. Cunningham et al. (2005) does not detect any
correlation with cohabitation for young adults. Parenthood is another typical gendered
institution. Men with children tend to hold more traditional family expectations while
women register more equal gender roles in the public sphere and intrahousehold re-
sponsibilities (Bolzendahl and Myers, 2004). The relevance of timing in parenthood
is equivocal as contrasting effects are found among young cohorts (Davis, 2007; Fan
and Marini, 2000). Corrigall and Konrad (2007) reports instead a homogeneous effect
among genders, linking parenthood to reduced levels of gender egalitarianism. The
traditional impact is also found to decrease over multiple childbirths (Katz-Wise et al.,
2010). Recent findings in the UK limit the effects solely to fathers parenting daughters
of school age (Borrell-Porta et al., 2019). Even though to different extents, research
offers significant evidence of intergenerational transmission of gender ideology within
family borders (Farré and Vella, 2013; Myers and Booth, 2002). Employment status
is associated with more liberal opinions among women while slight to no influence is
found on men’s attitudes (Bolzendahl and Myers, 2004; Corrigall and Konrad, 2007;
Cunningham et al., 2005; Fan and Marini, 2000; Vespa, 2009). Men are typically more
liberal when having a working spouse but expect women to take on more responsi-
bilities when they are the primary breadwinners (Bolzendahl and Myers, 2004). The
relevance of the intrahousehold dimension is confirmed by Bertrand et al. (2015): as
soon as the wife earns more than the husband, her relative income drastically drops
by one-half, and she becomes more likely to quit the labour market in the following
year. The author links this pattern to the aversion to nontraditional household ar-
rangements. Lastly, young and more educated individuals tend to conform to more
egalitarian roles, as do those living in more urban areas and adults who attend reli-
gious gatherings (Bolzendahl and Myers, 2004; Cunningham et al., 2005; Davis, 2007;
Fan and Marini, 2000).
8
3.2 Gender Role Beliefs and Household Constraints
While there is considerable knowledge on how individual institutions shape attitudes, a
growing body of literature has been devoted to studying the ”supply-side” mechanisms
that contribute to shaping gender roles. Policy interventions such as childcare expan-
sion and parental leave reforms aim to promote women’s greater presence in the labour
market and ensure more balanced work-life arrangements between parents. The success
of these interventions varies widely, depending on their design and the national context
in which they are implemented (Bauernschuster and Schlotter, 2015; Dehos and Paul,
2017; Hook, 2006). On the other hand, great efficacy has been assessed in affecting
preferences on gender egalitarianism. Evidence in Norway suggests that a universal
childcare system made mothers consider childcare services the best form of care for
their children (Ellingsæter et al., 2017). Contrasting results are found instead among
German mothers following an expansion of childcare provision for children above three
years old (Zoch and Schober, 2018). Such policies also have indirect effects on those
who are close to the targeted groups: Unterhofer and Wrohlich (2017) observe that
German grandparents display egalitarian attitudes when their son had a child after in-
troducing a fathers’ quota in the parental leave scheme. Lastly, Pedulla and Thébaud
(2015) offers experimental evidence that individuals are likely to opt for an egalitarian
relationship structure when it is presented as an option. Women are found to be more
responsive than men to policies in support of dual-earner arrangements.
Taken together, this body of literature suggests that individuals tend to advocate
for more liberal gender roles when they are facilitated in outsourcing childcare chores
and balancing work and life aspects. Fewer insights are gained when restrictions are
instead enforced. The onset of COVID-19 has been providing a reliable framework
to disentangle such effects. Vandecasteele et al. (2022) closely relates to my research
purpose in that they evaluate the short-term impact of the first Dutch school closure
in the perceived trade-off between family values and full-time work. Although they
were limited in the data, it may be too narrow to neglect other dimensions such as the
division of childcare duties and gender roles in the public sphere (Scarborough, 2019).
No significant discrepancies are detected between men and women, where couples with
children experienced a halt in the progression toward egalitarian attitudes. Further
studies corroborate a similar detrimental effect on fathers of young children in France
(Boring and Moroni, 2023) and Germany (Huebener et al., 2022). The former explains
it by arguing that fathers avoid taking additional responsibilities for household produc-
tion when it is more costly to do so. Negligible effects are further detected in mothers.
To the best of my knowledge, Huebener et al. (2022) is the only study investigating how
disruptions in schools and daycare availability produce long-lasting effects on gender
9
role attitudes. Interestingly, those who experienced the largest drops in egalitarianism
reverted to the original opinions once the pandemic blew over.
In conclusion, my study aims to foster knowledge of the economic determinants
of gender role attitudes and their flexibility toward detrimental external shocks. The
effects of interest have traditionally been estimated using static approaches, with at-
titudes often measured at intervals of several years. This methodology may present
challenges in accurately identifying the true magnitude of a change in childcare avail-
ability. By relying on annual data and a more dynamic approach, I can better observe
the evolution of pre-pandemic trends and provide a deeper understanding of the un-
derlying processes driving belief adjustments. Furthermore, I extend the long-term
perspective by examining the impact of increased childcare constraints up to three
years after the pandemic outbreak. This dimension is often lacking in most previous
studies, which limited their focus on immediate changes.
4 Theoretical Framework
This study relies on the theoretical framework of motivated beliefs. These theories
suggest that individuals consider several factors when forming their own beliefs, ulti-
mately reflecting something different from their initial notions and thoughts. Epley
and Gilovich (2016) elaborates on the concept of “motivated beliefs” as a result of a
pragmatic exercise of human behaviour. One cornerstone of their work is that indi-
viduals do not act like rational agents, but reason about what is more convenient for
them to accept as true. Their ambitions and objectives play a similar role to their
inner inclinations in shaping the foundation of their beliefs.
Bénabou and Tirole (2016) provide further insights into the underlying mecha-
nisms. People like thinking about themselves as clever human beings who can control
every aspect of their lives. Egoistic feelings also induce people to judge others to pro-
tect and enhance their self-image (Beauregard and Dunning, 1998). Believing in one’s
capabilities is reassuring as it elicits self-efficacy when the risk of failure is high. Self-
serving beliefs may also be rewarding in other ways. One could directly benefit from
deceiving others into believing his or her views and, consequently, pursuing his or her
self-interest instead of their own. Convening to new beliefs is easier when being fre-
quently exposed to them and when they are perceived as genuine. When they are not,
the deceiver must alternatively give the impression of being truly convinced of them,
therefore learning to believe in them in the first place. Ultimately, the individual forms
a conviction based on a trade-off between truthfulness and desirability. Believing in
something may be highly beneficial but comes at the cost of it being false: the bal-
10
ance is determined by the extent of the associated gains from being convinced of it.
Hestermann et al. (2020) use these insights to investigate how meat consumers
use self-serving beliefs to alleviate the guilt from their dietary choices. The underlying
theoretical framework sets the following maximization problem:
maxU(c)− pc− ωx̃c
c∈R+
where the individual draws utility from meat consumption according to his or her pref-
erences and incurs two types of costs: the cost of the good (p) and moral concerns.
The latter stems from the fact that consumption produces external implications for the
animals, for which the individual feels responsible. The associated cost of such guilt
depends on the individual’s degree of morality (ω), i.e., to what extent the externality
is internalized in his or her behaviour and the expected magnitude of such externality
(x̃). This parameter is crucial as it reflects the agent’s attempt to reduce his or her
moral burden by aptly calibrating his or her views on the matter. To alleviate the
psychological distress of animal cruelty awareness, meat consumers deceive themselves
into believing that common farm practices are less harmful than one may think. This
setting is suitable to address my research question as it allows me to frame the in-
dividual response to an increase in the cost of “consuming” egalitarian gender roles.
As COVID-19 constrained households’ ability to outsource childcare duties, men find
it more costly to act according to equal gender role beliefs (p increases), as it would
require a larger household contribution. Assuming the consumption preferences and
degree of morality as fixed, they set their beliefs (x̃) to avoid such a detrimental effect
on his welfare. That is, they adopt more traditional opinions on gender responsibilities
to relieve them from the guilt of transferring the additional burden of parental care to
their partner or justify perhaps pre-existing unbalanced arrangements of unpaid work.
Oppositely, women find it more convenient to behave in more equal gender roles (p
decreases) because it exempts them from taking on the majority of the parental work.
Advocating for more egalitarian beliefs is a way to strengthen their aim and diminish
the distress of relegating their partner to such duties. While being rewarding on one
hand, the individual may suffer from cognitive distress when motivating their beliefs
persistently. It comes as a cost that the individual is likely to incur as long as the payoff
from his or her misleading beliefs is larger. They will no longer lie to themselves if the
cost of doing that is higher than the cost of believing the (true) opposite. Hestermann
et al. (2020) illustrates this result by linking the belief choice of the meat consumer to
his or her propensity to motivate his reasoning (σ ∈ [0, 1]). The maximization problem
is the following:
11
σ ∈ arg max tV (xh) + (1− t)V (x̃(σ))− k(1− t)
t∈[0,1]
where V (xh) is the indirect utility of setting true beliefs and k is the “psychic cost”
for eluding himself. For a certain value of σ, the individual finds it convenient to lie to
himself about the meat consumption externalities as far as the overall utility benefit is
larger than the associated cognitive distress, namely up to a point where:
V (x̃(σ))− V (xh) = k
This line of reasoning predicts that men (women) would cease altering their beliefs
once behaving under equal gender roles is no longer as costly (convenient), specifically
after COVID-19 restrictions are lifted.
5 Data and Empirical Approach
5.1 Data
To examine the implication of a change in household constraints on gender role beliefs,
I use panel data from the LISS panel (Longitudinal Internet Studies for the Social
Sciences) managed by the non-profit research institute Centerdata (Tilburg University,
the Netherlands). The LISS panel is a representative sample of Dutch individuals,
based on a true probability sample of households drawn from the population register
by Statistics Netherlands. Attitudes toward gender roles are tracked from the ”Politics
and Values” longitudinal study, surveying an average sample of 5,803 respondents over
fifteen annual waves from 2008 to 20235. Data from 2015 are missing in the analysis
since the survey was not implemented that year. Table 1 offers a detailed description
of the main sample demographic features.
General information on the respondents is retrieved from the ”Background Vari-
ables” section. As the latter is available on a monthly basis, I match this information
with the periods during which individuals’ opinions were recorded. This includes details
such as gender, age, civil status, level of education, occupation, net personal monthly
income and living residence. Religion and attendance to religious gatherings are simi-
5 In Wave 8, some of the respondents were randomly assigned to six groups that differed in both the
length and the delivery method of the questionnaire. Groups 1 and 2 received the full questionnaire at
once, groups 3 and 4 completed it in three parts with a week between each, and the rest in ten parts
one day apart. All groups received an email notification whenever a new part was about to start, while
even-numbered groups additionally received an SMS. From 2017, individuals have been receiving the
questionnaire in three fractions one month apart, with the third one targeting gender role attitudes.
12
13
Table 1
Descriptive Statistics
Time period: 2008 - 2023 [Count (%)] Wave 1 Wave 2 Wave 3 Wave 4 Wave 5 Wave 6 Wave 7 Wave 8 Wave 9 Wave 10 Wave 11 Wave 12 Wave 13 Wave 14 Wave 15
Gender
Male 3131 (46.14) 2745 (45.84) 2941 (46.43) 2468 (46.71) 2742 (46.87) 2648 (46.64) 2629 (46.51) 2768 (46.37) 2583 (46.69) 2688 (46.07) 2405 (45.96) 2737 (46.49) 2752 (45.89) 2418 (46.17) 2681 (46.59)
Female 3655 (53.86) 3243 (54.16) 3393 (53.57) 2816 (53.29) 3108 (53.13) 3030 (53.36) 3023 (53.49) 3201 (53.63) 2949 (53.31) 3146 (53.93) 2828 (54.04) 3150 (53.51) 3245 (54.11) 2819 (53.83) 3074 (53.41)
Age
15 - 34 years 1752 (25.82) 1512 (25.25) 1581 (24.96) 1177 (22.27) 1273 (21.76) 1218 (21.45) 1194 (21.13) 1317 (22.06) 1170 (21.15) 1284 (22.01) 1104 (21.10) 1322 (22.46) 1371 (22.86) 1054 (20.13) 1097 (19.06)
35 - 54 years 2945 (43.40) 2438 (40.71) 2272 (35.87) 1814 (34.33) 2040 (34.87) 1930 (33.99) 1872 (33.12) 1939 (32.48) 1745 (31.54) 1731 (29.67) 1478 (28.24) 1656 (28.13) 1721 (28.70) 1470 (28.07) 1583 (27.51)
> 55 years 2089 (30.78) 2038 (34.03) 2481 (39.17) 2293 (43.40) 2537 (43.37) 2530 (44.56) 2586 (45.75) 2713 (45.45) 2617 (47.31) 2819 (48.32) 2651 (50.66) 2909 (49.41) 2905 (48.44) 2713 (51.80) 3075 (53.43)
Domestic situation
Single, without child(ren) 953 (14.04) 895 (14.95) 1194 (18.85) 980 (18.55) 1073 (18.34) 1039 (18.30) 1069 (18.91) 1296 (21.71) 1174 (21.22) 1407 (24.12) 1212 (23.16) 1420 (24.12) 1443 (24.06) 1232 (23.52) 1304 (22.66)
Single, with child(ren) 316 (4.66) 320 (5.34) 305 (4.82) 254 (4.81) 279 (4.77) 296 (5.21) 278 (4.92) 333 (5.58) 313 (5.66) 289 (4.95) 272 (5.20) 319 (5.42) 357 (5.95) 277 (5.29) 292 (5.07)
Co-habitation, without child(ren) 2388 (35.19) 2164 (36.14) 2375 (37.50) 2066 (39.10) 2246 (38.39) 2190 (38.57) 2173 (38.45) 2190 (36.69) 2029 (36.68) 2104 (36.06) 1949 (37.24) 2121 (36.03) 2165 (36.10) 1972 (37.66) 2276 (39.55)
Co-habitation, with child(ren), living... 3085 (45.46) 2546 (42.52) 2382 (37.61) 1933 (36.58) 2180 (37.26) 2081 (36.65) 2054 (36.34) 1986 (33.27) 1836 (33.19) 1849 (31.69) 1628 (31.11) 1857 (31.54) 1866 (31.12) 1608 (30.70) 1713 (29.77)
Independently 915 (29.66) 884 (34.72) 920 (38.62) 687 (35.54) 762 (34.95) 719 (34.55) 724 (35.25) 715 (36.00) 685 (37.31) 591 (31.96) 558 (34.28) 811 (43.67) 691 (37.03) 642 (39.93) 640 (37.36)
At home, of age... 2170 (70.34) 1662 (65.28) 1462 (61.38) 1246 (64.46) 1418 (65.05) 1362 (65.45) 1330 (64.75) 1271 (64.00) 1151 (62.69) 1258 (68.04) 1070 (65.72) 1046 (56.33) 1175 (62.97) 966 (60.07) 1073 (62.64)
Below 12yo 843 (38.85) 697 (41.94) 638 (43.64) 571 (45.83) 676 (47.67) 672 (49.34) 678 (50.98) 597 (46.97) 558 (48.48) 545 (43.32) 504 (47.10) 487 (46.56) 531 (45.19) 478 (49.48) 539 (50.23)
Above 12yo 1327 (61.15) 965 (58.06) 824 (56.36) 675 (54.17) 742 (52.33) 690 (50.66) 652 (49.02) 674 (53.03) 593 (51.52) 713 (56.68) 566 (52.90) 559 (53.44) 644 (54.81) 488 (50.52) 534 (49.77)
Other 44 (0.65) 63 (1.05) 78 (1.23) 51 (0.97) 72 (1.23) 72 (1.27) 78 (1.38) 164 (2.75) 180 (3.25) 185 (3.17) 172 (3.29) 170 (2.89) 166 (2.77) 148 (2.83) 170 (2.95)
Occupation
Paid Employment 3555 (52.39) 3067 (51.22) 2977 (47.00) 2427 (45.93) 2672 (45.68) 2554 (44.98) 2436 (43.10) 2588 (43.36) 2387 (43.16) 2551 (43.73) 2243 (42.86) 2581 (43.85) 2671 (44.54) 2296 (43.84) 2573 (44.71)
Entrepreneur 383 (5.64) 311 (5.19) 326 (5.15) 268 (5.07) 300 (5.13) 263 (4.63) 282 (4.99) 288 (4.82) 257 (4.65) 270 (4.63) 230 (4.40) 306 (5.20) 292 (4.87) 243 (4.64) 292 (5.07)
Takes care of the housekeeping 722 (10.64) 562 (9.39) 565 (8.92) 478 (9.05) 501 (8.56) 483 (8.51) 486 (8.60) 431 (7.22) 422 (7.63) 443 (7.59) 416 (7.95) 447 (7.59) 450 (7.50) 408 (7.79) 419 (7.28)
Attends school or is studying 580 (8.55) 595 (9.94) 639 (10.09) 455 (8.61) 499 (8.53) 490 (8.63) 511 (9.04) 550 (9.21) 498 (9.01) 493 (8.45) 420 (8.03) 475 (8.07) 499 (8.32) 386 (7.37) 392 (6.81)
Unpaid work 112 (1.65) 112 (1.87) 138 (2.18) 120 (2.27) 133 (2.27) 137 (2.41) 145 (2.57) 149 (2.50) 125 (2.26) 140 (2.40) 117 (2.24) 138 (2.34) 111 (1.85) 98 (1.87) 108 (1.88)
Pensioner 909 (13.40) 895 (14.95) 1134 (17.90) 1045 (19.78) 1199 (20.50) 1215 (21.40) 1245 (22.03) 1378 (23.09) 1332 (24.09) 1420 (24.34) 1347 (25.74) 1434 (24.36) 1474 (24.58) 1415 (27.02) 1579 (27.44)
Work disability (partial) 232 (3.42) 222 (3.71) 254 (4.01) 231 (4.37) 242 (4.14) 217 (3.82) 216 (3.82) 213 (3.57) 213 (3.85) 251 (4.30) 230 (4.40) 252 (4.28) 259 (4.32) 221 (4.22) 229 (3.98)
Other 293 (4.32) 224 (3.74) 301 (4.75) 260 (4.92) 304 (5.20) 319 (5.62) 331 (5.86) 372 (6.23) 296 (5.35) 266 (4.56) 230 (4.40) 253 (4.30) 241 (4.02) 170 (3.25) 163 (2.83)
Income
0 EUR 703 (10.93) 644 (11.35) 699 (11.64) 560 (11.18) 595 (10.78) 605 (11.29) 627 (11.79) 665 (11.86) 629 (12.08) 547 (9.92) 509 (10.35) 564 (10.17) 556 (9.83) 476 (9.65) 423 (7.80)
1 EUR - 1500 EUR 3121 (48.53) 2673 (47.13) 2723 (45.36) 2264 (45.21) 2460 (44.57) 2320 (43.28) 2335 (43.92) 2287 (40.79) 2001 (38.44) 2044 (37.07) 1752 (35.62) 1866 (33.66) 1768 (31.25) 1469 (29.77) 1497 (27.61)
1501 EUR - 3000 EUR 2302 (35.80) 2071 (36.51) 2296 (38.25) 1933 (38.60) 2174 (39.39) 2152 (40.14) 2074 (39.01) 2319 (41.36) 2244 (43.11) 2510 (45.52) 2263 (46.01) 2595 (46.81) 2735 (48.34) 2377 (48.17) 2587 (47.72)
3001 EUR - 4500 EUR 216 (3.36) 222 (3.91) 224 (3.73) 194 (3.87) 231 (4.19) 233 (4.35) 235 (4.42) 284 (5.07) 284 (5.46) 359 (6.51) 343 (6.97) 441 (7.95) 516 (9.12) 533 (10.80) 783 (14.44)
> 4500 EUR 89 (1.38) 62 (1.09) 61 (1.02) 57 (1.14) 59 (1.07) 51 (0.95) 46 (0.87) 52 (0.93) 47 (0.90) 54 (0.98) 52 (1.06) 78 (1.41) 83 (1.47) 80 (1.62) 131 (2.42)
Education
Primary school 554 (10.12) 430 (9.14) 449 (9.46) 389 (9.74) 389 (8.81) 373 (8.76) 348 (8.26) 332 (7.97) 300 (7.78) 257 (7.01) 250 (6.52) 263 (6.64) 269 (6.70) 245 (6.09) 242 (6.88)
Intermediate secondary school 1472 (26.90) 1276 (27.12) 1268 (26.72) 1050 (26.30) 1165 (26.39) 1103 (25.89) 1048 (24.86) 915 (21.97) 818 (21.21) 807 (20.47) 723 (20.27) 748 (18.87) 704 (17.53) 624 (17.51) 668 (16.81)
Higher secondary school 544 (9.94) 511 (10.86) 529 (11.15) 420 (10.52) 466 (10.56) 478 (11.22) 463 (10.37) 447 (10.73) 408 (10.58) 408 (10.35) 371 (10.40) 411 (10.37) 425 (10.58) 358 (10.05) 401 (10.09)
Intermediate vocational school 1330 (24.30) 1126 (23.93) 1106 (23.30) 927 (23.22) 1054 (23.88) 994 (23.88) 1030 (24.44) 1043 (23.88) 963 (24.97) 966 (24.50) 857 (24.03) 1012 (25.54) 1005 (25.02) 900 (25.26) 1006 (25.31)
Higher vocational school 1184 (21.63) 1018 (21.64) 1036 (21.83) 909 (22.77) 985 (22.32) 949 (22.28) 957 (22.70) 1007 (24.18) 957 (24.82) 1037 (26.30) 932 (26.14) 1032 (26.04) 1083 (26.97) 964 (27.06) 1093 (27.50)
University 389 (7.11) 344 (7.31) 358 (7.54) 297 (7.44) 355 (8.04) 363 (8.52) 369 (8.75) 421 (10.11) 410 (10.63) 468 (11.87) 433 (12.14) 497 (12.54) 530 (13.20) 472 (13.25) 565 (14.21)
Living Residence
Extremely urban 24 (10.43) 485 (10.30) 509 (10.72) 408 (10.22) 430 (9.74) 396 (9.31) 411 (9.82) 453 (10.94) 417 (10.88) 471 (11.97) 452 (12.70) 509 (12.85) 517 (12.86) 433 (12.17) 639 (16.05)
Very urban 65 (28.26) 1220 (25.90) 1243 (26.18) 1031 (25.82) 1136 (25.74) 1104 (25.95) 1092 (26.09) 1089 (26.31) 980 (25.56) 991 (25.19) 735 (20.65) 800 (20.20) 820 (20.40) 734 (20.64) 1109 (27.85)
Moderately urban 60 (26.09) 1087 (23.08) 1085 (22.85) 936 (23.44) 1076 (24.38) 1073 (25.22) 1032 (24.65) 959 (23.17) 910 (23.74) 900 (22.88) 657 (18.46) 731 (18.46) 753 (18.74) 655 (18.41) 813 (20.42)
Slightly urban 50 (21.74) 1137 (24.14) 1144 (24.09) 954 (23.89) 1040 (23.56) 977 (22.96) 957 (22.86) 923 (22.30) 886 (23.11) 902 (22.93) 775 (21.77) 876 (22.12) 879 (21.87) 776 (21.82) 789 (19.81)
Not urban 31 (13.48) 781 (16.58) 767 (16.15) 664 (16.63) 732 (16.58) 705 (16.57) 694 (16.58) 715 (17.27) 641 (16.72) 670 (17.03) 941 (26.43) 1044 (26.36) 1050 (26.13) 959 (26.96) 632 (15.87)
Attendance to religious gatherings
Less than once a month 4333 (81.52) 3528 (75.24) 3648 (77.22) 3126 (78.25) 3516 (79.55) 3360 (78.85) 3355 (79.52) 3243 (78.07) 3030 (78.62) 3093 (78.76) 2826 (79.34) 2838 (72.66) 3199 (79.96) 2872 (80.56) 3204 (80.95)
More than once a month 982 (18.48) 1161 (24.76) 1076 (22.78) 869 (21.75) 904 (20.45) 901 (21.15) 864 (20.48) 911 (21.93) 824 (21.38) 834 (21.24) 736 (20.66) 1068 (27.34) 802 (20.04) 693 (19.44) 754 (19.05)
Religiousness
No 3743 (57.60) 3186 (54.46) 3532 (57.26) 3057 (58.91) 3447 (59.69) 3401 (60.86) 3387 (60.68) 3547 (60.23) 3321 (61.18) 3525 (62.17) 3181 (62.30) 3470 (79.13) 4098 (91.62) 3518 (91.12) 3985 (92.12)
Yes 2755 (42.40) 2664 (45.54) 2636 (42.74) 2132 (41.09) 2328 (40.31) 2187 (39.14) 2195 (39.32) 2342 (39.77) 2107 (38.82) 2145 (37.83) 1925 (37.70) 915 (20.87) 375 (8.38) 343 (8.88) 341 (7.88)
Source: Longitudinal Internet studies for the Social Sciences (LISS).
Notes: Table 1 reports the main demographic features of the sample. Data is pooled from 15 annual waves of the LISS panel. Information on gender, age, domestic status, occupation, income, education and
living residence is gathered from the ’Background Variables’ section. Living residence is coded as Extremely urban if the surrounding address density per km2 (d) is > 2500; Moderately urban if 1000 < d < 1500;
Slightly urban if 500 < d < 1000; Not urban if d < 500. Furthermore, data on religiousness and attendance to religious gatherings is drawn from the ’Religion and Ethnicity’ section. Information on the former
category is missing in Wave 1 because the LISS panel has been collecting it only since March 2008. More information about the LISS panel can be found at: https://www.lissdata.nl/
larly collected from the ”Religion and Ethnicity” study. I further enrich the dataset by
collecting information on individuals’ children, contained in the ”Family and House-
hold” study. After removing some inconsistencies in the responses6, I narrow down
the sample to individuals reportedly cohabiting with a partner7 as I plan to study how
individuals shape their attitudes when allocating childcare and home duties within the
household. I end up with an average sample size of 4,201 individuals interviewed each
year, totalling 63,012 overall observations.
The core of the analysis rests upon the fact that Wave 13 respondents stated their
opinions while being constrained by the closure of schools, which occurred between
December 16th, 2020, and February 7th, 2021. It is worth noting that the survey period
spans until the end of March: 56,70% of the individuals (n = 2,277) reported an end-
ing date of the questionnaire when schools had already reopened. Nevertheless, I pool
together the two subgroups for two main reasons. Firstly, daycare and out-of-school
care centres remained closed up to March 2nd. Secondly, 91,83% of the ”school-open”
respondents (n = 1,949) had already been surveyed within two weeks by the end of
the restriction. Lastly, individuals exhibit slightly more progressive beliefs after the
school closure8 (see Figure A2 in Appendix). By including all Wave 13 respondents, I
would expect to capture an upper bound of the true constraining effects. Nevertheless,
I ensure the robustness of the main findings by replicating the entire analysis in Section
7.1 while retaining only those who were directly constrained by the school closure.
5.2 Classification of Gender Role Attitudes
Attitudes towards gender roles are measured by seven different items: (1) Child : ”A
child that is not yet attending school is likely to suffer the consequences if his or her
mother has a job” (Cronbach’s α = 0.75); (2) Childrearing : ”A woman is more suited to
rearing young children than a man” (Cronbach’s α = 0.79); (3) Responsibilities : ”The
father should earn money, while the mother takes care of the household and the family”
(Cronbach’s α = 0.77); (4) Full-Time: ”Overall, family life suffers the consequences if
the mother has a full-time job” (Cronbach’s α = 0.75); (5) Leadership: ”It is unnatural
for women in firms to have control over men” (Cronbach’s α = 0.80); (6) Relationship:
”A working mother’s relationship with her children can be just as close and warm as
that of a non-working mother” (Cronbach’s α = 0.77); (7) Contribution: ”Both father
6 11 individuals (111 obs.) reported an opposite gender to their previous statement at some point.
7 Respondents are selected if they either expressed one of the following answers: “(Un)married co-
habitation, without child(ren)”, “(Un)married co-habitation, with child(ren)”.
8 Before February 7th, men with young children show a 0.03 lower Belief Score than those without,
while the gap turns positive if focusing on those interviewed after. Women’s beliefs do not show any
change. Also, the attitude gap between working and non-working parents is positive and generally
more pronounced when the schools reopened, whether their partner works or not.
14
and mother should contribute to the family income” (Cronbach’s α = 0.82).
These items primarily encompass gender roles related to female employment and
childcare, which I believe may have been particularly impacted during the pandemic.
Individuals indicate their stances by choosing one of the following responses: ”Fully
Disagree”, ”Disagree”, ”Neither agree nor disagree”, ”Agree”, or ”Fully Agree”, each
corresponding to a score (”Statement Score”) ranging from 1 to 5. Statements such as
(1) Child, (2) Childrearing, (3) Responsibilities, (4) Full-Time and (5) Leadership are
re-coded on an inverted scale since they imply conservative views. Ultimately, a higher
value is associated with more liberal beliefs about gender roles. Table 2 provides a
detailed summary of the seven items and how gender role attitudes are classified.
Table 2
Gender Role Items
Item Statement Cronbach’s α Scale
Traditional
(1) Child A child that is not yet attending school is likely to suffer the 0.75
consequences if his or her mother has a job
Fully disagree = 5
(2) Childrearing A woman is more suited to rearing young children than a man 0.79
Disagree = 4
(3) Responsibilities The father should earn money, while the mother takes care of 0.77 Neither agree nor disagree = 3
the household and the family Agree = 2
Fully agree = 1
(4) Full-Time Overall, family life suffers the consequences if the mother has 0.75
a full-time job
(5) Leadership It is unnatural for women in firms to have control over men 0.80
Egalitarian
Fully agree = 5
(6) Relationship A working mother’s relationship with her children can be just 0.77
Agree = 4
as close and warm as that of a non-working mother
Neither agree nor disagree = 3
(7) Contribution Both father and mother should contribute to the family income 0.82 Disagree = 2
Fully disagree = 1
Source: Longitudinal Internet Studies for the Social Sciences (LISS). More information about the LISS panel can be found at:
https://www.lissdata.nl/
Notes: Table 2 describes the seven items capturing gender role attitudes are gathered. They are organized depending on the traditional or
egalitarian nature of the underlying gender role. Statements and Chronback’s alphas are reported in Columns 2 and 3. Column 4 displays
how responses are coded. Items (1) Child, (2) Childrearing, (3) Responsibilities, (4) Full-Time and (5) Leadership present an inverted
scale compared to the original since they imply conservative views. Overall, higher values are associated with more liberal beliefs.
I consolidate the statement-specific information into a comprehensive index (”Be-
lief Score”) to assess individuals’ overall attitudes. To construct the index, I perform
a Principal Component Analysis (PCA), a well-established statistical technique in the
literature (Reichelt et al., 2021). Compared to simply averaging out the single scores,
it is able to retain maximum information from multiple interrelated variables (Abdi
and Williams, 2010). Statements are standardized before computing the Kaiser-Meyer-
Olkin (KMO) measure, whose magnitude (0.85) indicates a high correlation among the
variables and, consequently, their adequacy for the analysis. The index is predicted
15
using the first principal component, which explains 47% of the total inertia9. The
resulting Belief Score ranges from -6.96 and 3.26, with higher values suggesting more
egalitarian overall beliefs about gender roles.
5.3 Identification
Identification relies on the heterogeneous effects across different groups generated by
exogenous combinations of COVID-19 restrictions that made people’s everyday lives
more challenging. On one hand, the “hard lockdown” measure in the Netherlands
precluded many from working at the office, regardless of their location preferences.
Furthermore, parental mobility and free time took a hit due to the subsequent closure
of schools. According to a communication from Dutch Prime Minister Mark Rutte on
December 14th, 202010, these measures were prompted by concerns over the rapid surge
in hospitalizations and the resilience of the overall healthcare system. Similarly, one
can read in a government note on February 2nd, 202111 that restrictions on primary
schools, special primary schools and childcare centres would be lifted in 6 days because
of a progressive decline in the daily number of newly reported infections and a gradual
easing of the pressure on hospitals. This makes the COVID-19 setting a valid source
of exogenous variation in the outsourcing ability of household production.
In addition, valid identification of the effects of interest rests upon the assumption
that COVID-19 restrictions drive no further differential impact on the groups of inter-
est. One could argue that the schooling interruption may have deeper repercussions
on young children in the early stages of their education. This concern might lead such
parents to favour a higher presence of the natural caregiver, traditionally perceived as
the mother, at home. While this could typically pose a threat to identification under
normal circumstances, I reason that this is not the case here since most parents were
already confined to their homes by the ”hard lockdown”. Parents might still be wor-
ried about the lack of schooling for their children, but I do not expect it to cause any
differential effect in beliefs when differentiating against young children.
Furthermore, parents may have possibly allocated working hours and employment
within the household to align with their inner gender role conceptions. It would imply
some reverse causality concerns that hinder the understanding of how working with
young kids at home affected parents’ beliefs. It is worth noting that gender gaps in
9 The squared loadings of the seven items are equal to 0.44 (”Child”), 0.35 (”Childrearing”), 0.42
(”Responsibilities”), 0.44 (”Full-Time”), 0.31 (”Leadership”), 0.40 (”Relationship”), 0.24 (”Contribu-
tion”). They represent the proportion of the single-statement variance explained by the first principal
component (Abdi and Williams, 2010)
10 Rutte (2020b)
11 Ministry of Justice and Security and Ministry of Health, Welfare and Sport (2021)
16
labour market outcomes did not widen to a large extent during 2020 (Meekes et al.,
2023). The authors attribute it to the relatively large proportion of women in essential
work, more resilient to the COVID-19 shock. Alon et al. (2022) found similar results
regarding employment but registered a large shrinkage in working hours for Dutch
mothers of young children. Differences are however limited to non-essential workers,
with negative results mostly concentrated on single mothers (Meekes et al., 2023). It
is reassuring since I restrict my analysis to individuals in a couple. Among the latter,
parents with children below 12 do not report significant changes in employment and
working hours due to the partner’s working status and, among those working, to the
presence and the age of children (ibid.)12.
To isolate the causal effects of increased household constraints on gender role
attitudes, I use a fully dynamic Panel Event Study design (Borusyak et al., 2024) and
estimate multiple OLS regressions with two-way fixed effects to allow for unobserved
heterogeneity. The decision to adopt this approach rests upon its appealing quality of
observing how the treatment effect varies across multiple periods. Compared to a static
DiD method, it allows me to assess the plausibility of the Parallel Trend assumption, i.e.
whether eventual pre-event patterns may drive the observed heterogeneity. Similarly, a
closer examination of the relationship of interest becomes possible after the event has
occurred. The longitudinal extent of the LISS data facilitates these attempts: the main
event study specification contains 12 leads (b=12) and 2 lags (a=2), corresponding to
the number of survey waves that occurred before and after the introduction of the school
closure restriction. The first lead coefficient (γ−1) is excluded as a normalization
13.
5.4 Empirical Approach
I evaluate to what extent an increase in household constraints affects gender role at-
titudes using a step-by-step approach. Firstly, I reason that individuals are the most
challenged when compelled to stay at home and take up a large portion of childcare
duties, usually outsourced to schools under normal circumstances. In particular, I set
my treatment group to be parents with at least a child under the age of 12 living in
the household. Children of young age typically demand greater support from parents
in terms of child nurturing. It is also reasonable to argue that higher educational as-
sistance is required during the early stages of their schooling, that is primary school,
which Dutch kids usually attend until the age of 12. I assess the sensitivity of this
12 An exception is given by partnered mothers of pre-school children with non-essential jobs, who
reported larger employment losses compared to the male counterparts
13 This is a popular choice in the event study framework (Miller, 2023). Aside from offering a
more intuitive comparison in the closest estimates to the event, a one-unit normalization follows my
intention to widen the pre-event window as much as possible.
17
assumption in Section 7.2, where I investigate how individuals’ beliefs respond to the
presence at home of at least one child, irrespective of the child’s age.
Equation 1 is specified as follows:
∑−2 ∑a
Beliefs =α + β + δChildBelow12 + γ Dl + γ Dlit i t it l i l i
∑ l=−b∑ l=0− (1)2 a
+ γlD
l
i × ChildBelow12 lit + γlDi × ChildBelow12it + θXit + ϵit
l=−b l=0
where the predictors are regressed on both the indexes previously computed for overall
and specific-statement beliefs. Dli represents both leads (l < 0) and lags (l > 0) dummy
variables, taking value 1 if the i-th individual answered to the survey l years away from
the school closure event14; ChildBelow12it is a dummy variable taking value 1 if the
i-th individual had at least a 12-years-old child younger living in the household at time
t; Xit is a vector of time-varying individual-specific confounding factors
15; αi and βt
are respectively individual and time fixed effects; ϵit is the idiosyncratic error term.
The model is estimated separately for men and women. This is because I assume
that having a young child(ren) affects gender role attitudes differently within gender
categories. The estimate of main interest is γ0, which identifies the heterogeneity in
the treatment effect linked to the school closure event among groups: in the absence
of significant pre-trends, I would interpret a negative (positive) value as evidence of
fathers (mothers) conforming their beliefs to lower their extra intrahousehold burden
given by staying-at-home children. The interpretation of γ1 and γ2 offers a broader
perspective in assessing the persistence of the change in attitudes and determining
whether it is immediate or takes time to manifest.
In the next stage, I deepen my analysis by focusing on a higher constraint degree:
working with young children at home. Some individuals kept working during the hard
lockdown while raising young children who could not attend school because of the
school closure. Innovative solutions such as working from distance may have helped
in combining these duties. Nevertheless, work duties likely consumed a considerable
amount of time, leaving less available for parenting young children. In addition, I posit
parents to be even more time-constrained when both continued to work simultaneously.
I restrict the sample to those parents with at least a child below 12 living at home. To
examine whether combining paid and unpaid work exacerbates their shifts in gender
role attitudes, Equation 2 is regressed separately on four specific subgroups varying
14 Dli = 1i{t− 2021 = l}
15 I control for age and age squared, civil status, level of education, occupation, net personal monthly
income, living residence, religiousness, and attendance to religious gatherings.
18
across gender and partner’s working status. The model is defined as follows:
∑−2 ∑a
Beliefsit =αi + βt + δEmployedit + γlD
l
i + γlD
l
i
∑ l=−b−2 ∑ l=0 (2)a
+ γ Dl × Employed + γ Dll i it l i × Employedit + θXit + ϵit
l=−b l=0
where Employedit is a dummy variable taking value 1 if the i-th individual is reported
to either have a paid employment, being an autonomous worker, a freelancer, or being
self-employed at time t.
6 Results
The main analysis results are illustrated in this section. In Section 6.1 I inspect the
evolution of gender role beliefs over time. Section 6.2 briefly explains the main assump-
tions of the Event Study framework. Beliefs are then assessed assuming two different
types of constraints: Section 6.3 examines the changes in respondents’ attitudes due to
having young children unable to attend school. Lastly, Section 6.4 further investigates
whether working during the school closure exacerbated parents’ shifts in gender role
views under different household working arrangements.
6.1 Descriptive Evidence
A graphical representation of trends in attitudes is provided in Figure 2. It depicts the
gendered distribution of respondents’ stated gender role beliefs from 2008 to 2023. It
emerges that attitudes tend to differ across the aspect of the gender role: no more than
half the respondents answer positively to a fair contribution to family income within
the household, while the great majority disagree with women exerting control over
men in firms being unnatural. Heterogeneity in the answers is found across genders,
with females unequivocally exhibiting more egalitarian beliefs than men, regardless of
the role type. Interestingly, the largest gender gap arises concerning statements about
mothers’ innate aptness to rear children and the implications of a working mother on
children of preschool age.
Relevant differences are also observed in the strength of the opinions: women
typically condemn traditional views more fiercely than men, especially when it comes
to female leadership, intrahousehold responsibilities and children-working mothers re-
lationships. Both men and women showed a clear-cut egalitarian trend before the
pandemic outbreak: traditional roles captured by ”Child”, ”Fulltime”, and ”Relation-
19
Figure 2
Distribution of Respondents’ Answers to the seven Gender Role Statements
a) Men b) Women
1) Child 1) Child
2) Childrearing 2) Childrearing
3) Responsibilities 3) Responsibilities
4) Fulltime 4) Fulltime
5) Leadership 5) Leadership
6) Relationship 6) Relationship
7) Contribution 7) Contribution
Notes: Figure 2 illustrates the evolution in time of the share of respondents who either ”Fully agree”, ”Agree”, ”Neither
agree nor disagree”, ”Disagree” or ”Fully disagree” to the seven gender role items. Disagreeing is associated with believ-
ing in more conservative gender norms, except for ”Relationship” and ”Contribution” as they convey more progressive
gender roles. The statements are the following: (1) Child: ”A child that is not yet attending school is likely to suffer
the consequences if his or her mother has a job”; (2) Childrearing: ”A woman is more suited to rearing young children
than a man”; (3) Responsibilities: ”The father should earn money, while the mother takes care of the household and the
family”; (4) Full-Time: ”Overall, family life suffers the consequences if the mother has a full-time job”; (5) Leadership:
”It is unnatural for women in firms to have control over men”; (6) Relationship: ”A working mother’s relationship
with her children can be just as close and warm as that of a non-working mother”; (7) Contribution: ”Both father and
mother should contribute to the family income”.
20
ship” report the largest decreases in the share of respondents agreeing with them. Only
statements like ”Childrearing” and ”Responsibilities” experienced a slight deflection in
the early years from 2009 to 2011. It is worth noting that strong answers drive most
of the progress. ”Fully” egalitarian responses often increased more than moderate
(”Child”, ”Responsibilities”), while the opposite holds for traditional ones. The main
takeaway is that opinions are getting less conservative over the last decade rather than
just polarizing. Such trends seem to become more stationary in recent years if not
descending, as it is for attitudes toward family income contribution (”Contribution”).
The next sections discuss a possible explanation for this curb: household constraints.
6.2 Event Study Assumptions
A valid interpretation of the average treatment and post-treatment effects requires that
two assumptions are met: individuals should not anticipate the effect of interest and
trends in the pre-treatment window should not be detected (Miller, 2023). Regarding
the former, I rely on the exogenous nature of the school closure event that happened in
2021. It is improbable that individuals knew in advance what would be the implications
of COVID-19 in the early stages of the outbreak. If they did, the impact associated
with household constraints might result in bias due to individuals strategically timing
the decision to have a child. Moreover, ruling out pre-trends existence is crucial to
ensure that any detected change in attitudes is likely induced by the treatment rather
than reflecting pre-existing dynamics. Pre-trends are investigated by discerning any
visible pattern of the lead estimates and assessing their individual and joint statistical
significance.
6.3 Young Children
First, I examine the average effect of having at least a child below 12 living in the
household on the development of overall gender role beliefs compared to how beliefs
evolve among individuals with older or no children at all. Figures 3 and 4 (see also
Table A1 in the Appendix) present the Event Study estimates from the first two-way
fixed-effects regression (Equation 1). The pre-event window shows an overall decreasing
pattern in the early leads coefficients that flattens in recent years. While one cannot
fully rule out the threat of pre-trends, a Wald test fails to reject the null hypothesis of
the last 7 leads being jointly equal to 0 (see Table A4 in the Appendix). This indicates
low chances of any pre-pandemic disruptions distinctly affecting the beliefs of either
group. Once restrictions were in place, men who had young children at home upheld
more conservative overall beliefs compared to those who didn’t. The difference in their
21
Belief Score is equal to 0.170 points (p = 0.032). The outcome is consistent with what
has been found in Germany (Huebener et al., 2022) and France (Boring and Moroni,
2023). Interestingly, beliefs do not worsen further (except for ”Leadership”) but remain
at this new traditional level attained: no positive rebound is detected in the coming
years, challenging Huebener et al. (2022) results that suggest an immediate recovery.
Figure 3
The Development of Men’s Gender Role Attitudes when having Young Children at home
Notes: Figure 4 shows the impact of a reduction in childcare and school availability on men’s gender role attitudes when
having living-at-home children below 12 unable to attend school (t = 0). Event Study coefficients (ChildBelow12 × Dit)
are obtained from Equation 1 and represented on the y-axis, time to event (2009-2023) on the x-axis. Using a Principal
Component Analysis (PCA), the Belief Score encompasses gender role attitudes measured by the seven items described
in Table 2. Equal gender roles correspond to high values such that a positive estimate implies more progressive stances
associated with working. Lead 12 estimates (γ−12) are unavailable due to excessively high confidence intervals resulting
from data limitations. Confidence intervals are computed at the 95% level.
Women experienced almost no significant downturn during the pandemic (-0.091,
p = 0.169), while significant effects were found before and immediately after the event.
The likelihood of differential trends before treatment makes it difficult to assess a
meaningful impact on mothers’ beliefs that is due to not being able to delegate childcare
tasks. At best, it could be interpreted as a slowdown in a conservative tendency
originating two years before16, which seems to curb slightly when schools were closed.
These findings contribute to an existing body of conflicted results. Huebener et al.
(2022) reports that childless mothers advocate for more conservative beliefs during the
16 The second lead estimate (γ−2) is -0.184 (p= 0.017), which would generally suggest a pitfall in
the No Anticipatory Effect assumption. Anyway, there was no clue yet about the upcoming pandemic
when mothers were interviewed between February and March 2019. In addition, there is no valid
reason why fathers did not anticipate the event in advance.
22
pandemic, while Boring and Moroni (2023) does not detect any relevant change for
mothers at all.
Figure 4
The Development of Women’s Gender Role Attitudes when having Young Children at home
Notes: Figure 4 shows the impact of a reduction in childcare and school availability on women’s gender role attitudes
when having living-at-home children below 12 unable to attend school (t = 0). Event Study coefficients (ChildBelow12
× Dit) are obtained from Equation 1 and represented on the y-axis, time to event (2009-2023) on the x-axis. Using a
Principal Component Analysis (PCA), the Belief Score encompasses gender role attitudes measured by the seven items
described in Table 2. Equal gender roles correspond to high values such that a positive estimate implies more progressive
stances associated with working. Lead 12 estimates (γ−12) are unavailable due to excessively high confidence intervals
resulting from data limitations. Confidence intervals are computed at the 95% level.
The understanding of such results is complemented by investigating the evolu-
tion of beliefs toward specific gender roles over the same samples (see Table A2 in the
Appendix). Fathers’ drop seems to be mostly driven by more conservative views on
women as natural caregivers. The ”Childrearing” score went down by 0.122 points (p
= 0.066) compared to individuals with older or no kids, and did not recover afterwards
as the lags estimates are almost null. They experienced a similar decrease in beliefs to-
ward women’s leadership (”Leadership”). Notice that beliefs kept dropping even when
kids were able to attend school again. The fact that no previous significant trends
are detected further implies that household constraints likely played a role in shaping
their perspectives on gender role issues. Beliefs regarding other gender roles remained
unaffected. Evidence also suggests that men with young children at home conformed
to a more unbalanced partition of intrahousehold tasks in the household (”Responsibil-
ities”) when schools were closed (-0.104, p = 0.052). Women show an opposite pattern
on the same item: while having young children has been typically associated with lower
stances, negative disparities level off when women are constrained by the school clo-
23
sure, only to reappear immediately afterwards. This effect is consistent with evidence
from Boring and Moroni (2023) on a similar belief item. More conservative opinions
are instead reported in attitudes toward women’s leadership, even though mothers’
decline started before the event of interest. No significant heterogeneity in the treat-
ment effect is detected in the remaining statements. The presence of children at home
does not affect either women’s or men’s beliefs toward ”Full-Time”. This contrasts the
main findings of Vandecasteele et al. (2022), which suggest that Dutch parents have
weaker progressive shifts in opinions on this statement compared to childless couples,
especially when older17.
6.4 Working Parents
Now I focus on parents with at least one child under 12 living at home. I split them into
two subsamples according to their partner’s working status and, for each subgroup, I
evaluate whether working during the school closure puts further pressure on parents’
beliefs in addition to taking care of young children. Results are presented in Figures
5 and 6 (see also Table A3 in the Appendix). Given the smaller sample size, it is
advisable to be cautious in their interpretation. I find however significant and varying
outcomes for fathers who are the sole household earners. When schools were closed,
combining work and childcare tasks made fathers more conservative on the relationship
between working mothers and children (”Relationship”) compared to those not work-
ing. The effect is only temporary, as I observe a reversal into more positive beliefs in
the following years. At the same time, they show more progressive and persistent views
against the idea that mothers are naturally suited for childrearing tasks (“Childrea-
ring”). Pre-event estimates are jointly statistically insignificant, suggesting that the
effect (+0.515, p = 0.083) is the result of the combination of household and working
constraints. Working fathers also tend to favour an equal contribution to family income
(“Contribution”) once children can attend school again. These findings only partially
confirm the hypothesis that fathers of young children turn to even more traditional
opinions when compelled to work. Evidence suggests that the effects could be limited
to beliefs concerning mothers’ employment. Contrary to my intuition, working fathers
do not shape their gender roles attitudes when being the most time-constrained, i.e.
when their partners also work. No significant changes are detected during the school
closure period but this group is found to advocate for more equal roles regarding state-
ments like “Responsibilities”, “Child” and “Childrearing” after the pandemic blows
17 Differences between my work and Vandecasteele et al. (2022) concern the methodology, sample
size and time period. Also, the authors do not put any restriction on the children’s age, except for
excluding adult respondents living with their parents.
24
Figure 5
Trends in Gender Role Attitudes among Fathers with Young Children when Balancing Work and
Childcare: a Comparison by Partner Working Status
Notes: Figure 5 shows the additional constraining effect of combining work and childcare tasks on gender role attitudes of
fathers with young children unable to attend school (t = 0). Event Study coefficients (Employed × Dit) are represented
on the y-axis, time to event (2009-2023) on the x-axis. Coefficients are obtained from Equation 2 and estimated
separately for two different subgroups: sole household earners and those with working partners. Using a Principal
Component Analysis (PCA), the Belief Score encompasses gender role attitudes measured by the seven items described
in Table 2. Equal gender roles correspond to high values such that a positive estimate implies more progressive stances
associated with working. Lead 12 estimates (γ−12) are unavailable due to excessively high confidence intervals resulting
from data limitations. Confidence intervals are computed at the 95% level.
over.
Figure 6 shows that mothers’ beliefs have been getting more egalitarian in the pre-
event window when they are the sole workers in the household. This progression came
to a halt as soon as school restrictions were enforced: no differences in the Belief Score
were detected between employed and unemployed mothers in this subgroup. Opinions
on statements such as “Childrearing”, ”Leadership”, and “Relationship” are the main
drivers of this pattern. Regarding the latter, a similarly detrimental effect is registered
for mothers as it was for fathers, though to a lesser extent. After school restrictions were
lifted, working mothers returned to advocate for more equal roles compared to those
unemployed. Notably, the most significant comebacks are observed in the same areas
where progressive beliefs dropped more dramatically, with positive effects also reported
25
for attitudes toward the division of childrearing tasks (”Childrearing”). Contrasting
results are found for attitudes of mothers living in dual-earner households. They tend
to decrease egalitarian stances on this matter after one year from the pandemic but
agree more that the family does not suffer if the mother has a full-time job.
Figure 6
Trends in Gender Role Attitudes among Mothers with Young Children when Balancing Work and
Childcare: a Comparison by Partner Working Status
Notes: Figure 6 shows the additional constraining effect of combining work and childcare tasks on gender role attitudes
of mothers with young children unable to attend school (t = 0). Event Study coefficients (Employed × Dit) are
represented on the y-axis, time to event (2009-2023) on the x-axis. Coefficients are obtained from Equation 2 and
estimated separately for two different subgroups: sole household earners and those with working partners. Using a
Principal Component Analysis (PCA), the Belief Score encompasses gender role attitudes measured by the seven items
described in Table 2. Equal gender roles correspond to high values such that a positive estimate implies more progressive
stances associated with working. Lead 12 estimates (γ−12) are unavailable due to excessively high confidence intervals
resulting from data limitations. Confidence intervals are computed at the 95% level.
Aside from that, the combination of working and raising young children did not,
on average, lead women to adopt more egalitarian views, aligning with what has been
observed for fathers. Beliefs did not become more progressive even in situations where
women might benefit the most from embracing more equal gender roles, such as when
both partners were working. This points out that mothers did not use a motivated-
reasoning logic to shape their beliefs and alleviate the additional burden of working.
26
7 Robustness Analysis
I perform three additional analyses using different empirical approaches to assess the
robustness of the main findings outlined in Section 6. This section is structured as
follows: first, I replicate the main analysis focusing only on those individuals directly
affected by the school closure (Section 7.1); Section 7.2 examines whether the young
age of the children is a relevant determinant of the changes in beliefs experienced by
parents; lastly, Section 7.3 investigates whether the observed shifts in beliefs are just
the outcome of a broader conservative tendency unrelated to gender roles.
7.1 Only Constrained
In the main analysis, I included those individuals who were not directly affected by
the school closure. Most were interviewed shortly after schools reopened and further
restrictions remained in place. One might however argue that the true size effect is
limited to those parents whose child(ren) were unable to attend school and, therefore,
compelled to stay at home. I address this concern by excluding Wave 13 respondents
who completed the gender roles questionnaire after February 7th, 2021. Equations 1
and 2 are estimated again using this revised subsample. Table A5 provides additional
evidence confirming the gendered impact of having young children on parents’ opinions.
Household constraints seem to be more effective in decreasing fathers’ support in equal
gender roles compared to the previous setting. Negative effects are now reported on
views toward a close relationship between working mothers and children (“Relation-
ship”), where fathers were originally found to be neutral. A novelty is also the more
discontinuous increase in mothers’ liberal attitudes toward “Responsibilities”. During
the school closure, they advocated for a more balanced distribution of household re-
sponsibilities, likely explained by the pressure experienced under those circumstances.
Differences with respect to the original analysis emerge also in the attitudes of fathers
in dual-earner households (see Table A6 in the Appendix). When children could not
attend school, employed fathers believed in more traditional gender roles compared
to those not working, hindering their progression toward egalitarianism started in the
pre-pandemic period. In most areas, the effects dissipate within two years, except
progressive attitudes toward family income contribution (“Contribution”), which have
permanently ceased to grow. Nothing changes for mothers compared to what has been
observed thus far, except that they tend to exhibit even lower scores when they are the
sole workers. Overall, findings confirm the robustness of the main analysis outcomes,
as restricting the attention to constrained individuals yields similar and more clear-cut
effects. I take these findings as reinforcing my hypotheses of men, and only partially
27
women, rationally adjusting their gender role beliefs to cope with increased household
constraints.
7.2 Children of All Ages
So far, I focused my attention on how a shortage in the availability of childcare and
school services affects the beliefs of parents with children below the age of 12. I reason
that children of young age require a greater effort by parents in terms of complementary
rearing tasks and educational support. I test the sensitivity of this assumption by
investigating whether individuals’ beliefs are prone to shift simply because they have
one or more living-at-home children, regardless of their age. Equation 1 is modified as
follows:
∑−2 ∑a
Beliefsit =αi + βt + δChild + γ D
l
it l i + γ
l
lDi
∑ l=−b− ∑ l=0 (3)2 a
+ γ l llDi × Childit + γlDi × Childit + θXit + ϵit
l=−b l=0
where Childit is a dummy variable taking value 1 if the i-th individual had at least a
child living in the household at time t. Equation 2 is then estimated on this revised
subgroup.
Results, as presented in Tables A7 and A8 (see Appendix), indicate that individ-
uals do not adapt their views on gender roles solely based on their parenting status.
When not considering children’s age, fathers with children at home lean toward more
equally distributed household responsibilities and do not report conservative ”Leader-
ship” views, as highlighted for fathers of young children. Likewise, mothers do not turn
more conservative as before in statements like ”Relationship” and ”Responsibilities”
after schools reopened. Less discontinuous effects are found when looking at mothers
who are the sole workers. When older children are also considered, they do not tend to
adopt less progressive attitudes as observed in Section 6. This is not however the case
for dual-earner households. The school closure led mothers with children of all ages to
believe that working worsens their relationship with the children (”Relationship”). I
further document that working fathers with unemployed partners are less likely to be-
come conservative toward the latter statement, compared to those having only younger
children. Instead, they tend to adopt more negative stances on women’s family income
contribution. Overall evidence suggests that the young age of children constitutes an
additional burden to parenthood when households cannot delegate childcare tasks to
schools. Mixed results are found regarding the impact of employment on gender role
28
beliefs. When at-home children are young, working predicts higher conservative set-
backs on parents’ opinions, mostly when they are the sole workers, and even more so
for mothers.
7.3 Non-gender Role Beliefs
This section investigates whether being constrained to supervise young children leads
to changes in beliefs unrelated to a gendered division of roles. Individuals adjusting
their stances on these issues may simply reflect a broader tendency in the Netherlands
toward different values from mothers’ employment and family attachment. To check
whether this is likely to happen, I estimate Equation 1 on three further indices that
encode opinions on tolerance toward foreigners ([I] Foreign: ”It does not help a neigh-
bourhood if many people of foreign origin or descent move in.”), parents’ marital status
([II] Marriage: ”People that want to have children should get married.”) and support
for euthanasia ([III] Euthanasia: ”Where would you place yourself on a scale from 1
to 5, where 1 means that euthanasia should be forbidden and 5 means that euthanasia
should be permitted?”). Responses are assigned to numbers from 1 to 5 so that a
higher value corresponds to more progressive opinions18.
Results in Table A9 (see Appendix) highlight that gender roles are the only domain
where men significantly lowered their progressive views when having young children at
home. This suggests that men likely choose to reduce their support toward egalitar-
ianism when they find themselves in a similar circumstance. Instead, women present
similarities in their response to overall gender roles and to marriage as a requirement
for parents (”Marriage”).
8 Conclusions
In this research, I study whether beliefs toward gender roles are entrenched or likely to
shift according to external shocks affecting individuals’ capability to outsource house-
hold production. I find that men are more responsive to childcare and school con-
straints than women. The presence of young children at home made fathers revert to
more conservative stances on gender roles when school attendance was limited due to
COVID-19. Notably, the highest diminished effect is reported on an equal division of
childrearing duties among parents. Consistent with prior studies (Boring and Moroni,
2023; Huebener et al., 2022), these findings suggest that men likely reduce their sup-
port for gender egalitarianism in circumstances where it requires a greater effort to do
18 [I] Foreign and [II] Marriage are re-coded similarly to (1) Child (see Section 5.4)
29
so, aligning with motivated reasoning perspectives (Bénabou and Tirole, 2016; Epley
and Gilovich, 2016). The children’s young age is a determining factor in this change,
as it likely increases the required parental workload. Women do not seem to behave
likewise to the same extent, as they are more hesitant to change beliefs in overwhelm-
ing circumstances. When I expected mothers of young children to advocate for higher
equality, they only temporarily adopted less conservative views regarding responsibili-
ties within the household and the relationship between children and mothers employed
full-time. Why women exhibited a lower response is debatable. It could be that gender
roles are deeply entrenched in Dutch individuals, leading to lower female bargaining
power. Taking on most of the unpaid work during the pandemic may be the result
of mothers simply indulging pre-existing gender role conceptions, without eliciting a
fierce response in their attitudes.
One more takeaway is that parents of young children react to the additional bur-
den of working depending on their household working dimension. Combining paid and
unpaid work is more likely to hinder parents’ progression toward equal gender roles
when they are the sole earners in the household. Working during the school closure
makes both of them, especially fathers, turn more negatively on mothers having a full-
time job. Contrasting effects are reported when asking about a balanced contribution
to family income, with working mothers getting more negative in the short term and
fathers agreeing more in the post-pandemic years. Surprisingly, beliefs hardly adjust
in situations where parents would gain the largest benefit in doing so, that is when
they are compelled to watch their young children at home and both working. There is
evidence that fathers in dual-earner households uphold even more progressive beliefs
after the school closure. This counteracts results in France (Boring and Moroni, 2023)
proving that men were more likely to become conservative when both partners con-
tinued working during the lockdown. One plausible explanation could be that being
exposed to women participating in the labour market encourages men to see it as a
positive model, thus leading them to adopt more progressive views. This aligns with
exposure-based theories proposed and tested by a segment of the literature (Bolzendahl
and Myers, 2004; Reichelt et al., 2021; Vandecasteele et al., 2022). Nonetheless, I do
not preclude reverse causality in the relationship, meaning that both parents might be
working because they are more egalitarian in the first place.
My study is not without limitations. The main analysis rests on a wide repre-
sentative sample of Dutch individuals. Yet, the sample size narrows considerably for
specific subgroups, especially parents of young children who are the sole earners in the
household. I am limited in looking at additional mechanisms within these categories.
In addition, I observe that large conservative shifts during the school closure period
30
do not typically anticipate immediate positive rebounds in the coming years. Despite
I already rely on a broader time frame compared to prior works, a higher number
of post-treatment periods would allow me to investigate further whether beliefs are
permanently altered or require a longer time to recover.
In conclusion, my research emphasises the importance of public programs that
expand childcare and school facilities, and promote easier access to these resources.
Likewise, enhancing dual-earner household arrangements helps prevent parents from
hindering their progression toward gender egalitarianism during challenging times.
References
Abdi, H. and Williams, L. J. (2010). Principal component analysis. WIREs Computa-
tional Statistics, 2(4):433–459.
Alon, T., Coskun, S., Doepke, M., Koll, D., and Tertilt, M. (2022). From Mancession
to Shecession: Women’s Employment in Regular and Pandemic Recessions. NBER
Macroeconomics Annual, 36:83–151.
Alon, T., Doepke, M., Olmstead-Rumsey, J., and Tertilt, M. (2020). The Impact
of COVID-19 on Gender Equality. Technical Report w26947, National Bureau of
Economic Research, Cambridge, MA.
Andrew, A., Cattan, S., Costa Dias, M., Farquharson, C., Kraftman, L., Krutikova, S.,
Phimister, A., and Sevilla, A. (2022). The gendered division of paid and domestic
work under lockdown. Fiscal Studies, 43(4):325–340.
Bauernschuster, S. and Schlotter, M. (2015). Public child care and mothers’ labor
supply—Evidence from two quasi-experiments. Journal of Public Economics, 123:1–
16.
Beauregard, K. S. and Dunning, D. (1998). Turning up the contrast: Self-enhancement
motives prompt egocentric contrast effects in social judgments. Journal of Person-
ality and Social Psychology, 74(3):606–621.
Berghammer, C. (2022). Childcare and housework during the first lockdown in Austria:
Traditional division or new roles? Journal of Family Research, 34(1):99–133.
Bertrand, M., Kamenica, E., and Pan, J. (2015). Gender Identity and Relative Income
within Households*. The Quarterly Journal of Economics, 130(2):571–614.
Bolzendahl, C. I. and Myers, D. J. (2004). Feminist Attitudes and Support for Gender
31
Equality: Opinion Change in Women and Men, 1974-1998. Social Forces, 83(2):759–
789.
Boring, A. and Moroni, G. (2023). Turning back the clock: Beliefs about gender roles
during lockdown. Labour Economics, 84:102363.
Borrell-Porta, M., Costa-Font, J., and Philipp, J. (2019). The ‘mighty girl’ effect:
does parenting daughters alter attitudes towards gender norms? Oxford Economic
Papers, 71(1):25–46.
Borusyak, K., Jaravel, X., and Spiess, J. (2024). Revisiting Event Study Designs:
Robust and Efficient Estimation. arXiv:2108.12419 [econ].
Brakel, M. v. d., Portgegijs, W., and Hermans, B., editors (2020). Emancipatiemonitor
2020. Centraal Bureau voor de Statistiek (CBS) ; Sociaal en Cultureel Planbureau
(SCP), Den Haag. OCLC: 1226692165.
Bénabou, R. and Tirole, J. (2016). Mindful Economics: The Production, Consumption,
and Value of Beliefs. Journal of Economic Perspectives, 30(3):141–164.
Corrigall, E. A. and Konrad, A. M. (2007). Gender Role Attitudes and Careers: A
Longitudinal Study. Sex Roles, 56(11-12):847–855.
Cotter, D., Hermsen, J. M., and Vanneman, R. (2011). The End of the Gender Rev-
olution? Gender Role Attitudes from 1977 to 2008. American Journal of Sociology,
117(1):259–289.
Cunningham, M. (2008). Influences of gender ideology and housework allocation on
women’s employment over the life course. Social Science Research, 37(1):254–267.
Cunningham, M., Beutel, A. M., Barber, J. S., and Thornton, A. (2005). Recipro-
cal relationships between attitudes about gender and social contexts during young
adulthood. Social Science Research, 34(4):862–892.
Davis, S. N. (2007). Gender ideology construction from adolescence to young adult-
hood. Social Science Research, 36(3):1021–1041.
Dehos, F. and Paul, M. (2017). The Effects of After-School Programs on Maternal
Employment. SOEPapers, 905.
Del Boca, D., Oggero, N., Profeta, P., and Rossi, M. (2020). Women’s and men’s work,
housework and childcare, before and during COVID-19. Review of Economics of the
Household, 18(4):1001–1017.
32
Ellingsæter, A. L., Kitterød, R. H., and Lyngstad, J. (2017). Universalising Childcare,
Changing Mothers’ Attitudes: Policy Feedback in Norway. Journal of Social Policy,
46(1):149–173.
England, P. (2010). The Gender Revolution: Uneven and Stalled. Gender & Society,
24(2):149–166.
Epley, N. and Gilovich, T. (2016). The Mechanics of Motivated Reasoning. Journal of
Economic Perspectives, 30(3):133–140.
Fan, P.-L. and Marini, M. M. (2000). Influences on Gender-Role Attitudes during the
Transition to Adulthood. Social Science Research, 29(2):258–283.
Farre, L., Fawaz, Y., Gonzalez, L., and Graves, J. (2020). How the Covid-19 Lockdown
Affected Gender Inequality in Paid and Unpaid Work in Spain. SSRN Electronic
Journal.
Farré, L. and Vella, F. (2013). The Intergenerational Transmission of Gender Role
Attitudes and its Implications for Female Labour Force Participation. Economica,
80(318):219–247.
Fortin, N. M. (2015). Gender Role Attitudes and Women’s Labor Market Participation:
Opting-Out, AIDS, and the Persistent Appeal of Housewifery. Annals of Economics
and Statistics, (117/118):379–401.
Hestermann, N., Le Yaouanq, Y., and Treich, N. (2020). An economic model of the
meat paradox. European Economic Review, 129:103569.
Hook, J. L. (2006). Care in Context: Men’s Unpaid Work in 20 Countries, 1965–2003.
American Sociological Review, 71(4):639–660.
Huebener, M., Spiess, C. K., Pape, A., Siegel, N. A., Danzer, N., and Wagner, G. G.
(2022). Cracking under Pressure? Gender Role Attitudes toward Maternal Employ-
ment during COVID-19.
Hupkau, C. and Petrongolo, B. (2020). Work, Care and Gender during the COVID-19.
Fiscal Studies, 41(3):623–651.
Jessen, J., Spiess, C. K., Waights, S., and Wrohlich, K. (2021). Sharing the Caring?
The Gender Division of Care Work During the Covid-19 Pandemic in Germany.
SSRN Electronic Journal.
Katz-Wise, S. L., Priess, H. A., and Hyde, J. S. (2010). Gender-role attitudes and
33
behavior across the transition to parenthood. Developmental Psychology, 46(1):18–
28.
Meekes, J., Hassink, W. H. J., and Kalb, G. (2023). Essential work and emergency
childcare: identifying gender differences in COVID-19 effects on labour demand and
supply. Oxford Economic Papers, 75(2):393–417.
Miller, D. L. (2023). An Introductory Guide to Event Study Models. Journal of
Economic Perspectives, 37(2):203–230.
Ministry of Health, W. and Sport (2020). Temporary closure of schools and childcare
centres in strategy to fight coronavirus.
Ministry of Justice and Security and Ministry of Health, Welfare and Sport (2021).
Coronavirus situation remains serious, lockdown extended.
Moors, G. (2003). Estimating the Reciprocal Effect of Gender Role Attitudes and Fam-
ily Formation: A Log-linear Path Model with Latent Variables. European Journal
of Population, 19:199–221.
Myers, S. M. and Booth, A. (2002). Forerunners of Change in Nontraditional Gender
Ideology. Social Psychology Quarterly, 65(1):18.
Pedulla, D. S. and Thébaud, S. (2015). Can We Finish the Revolution? Gender,
Work-Family Ideals, and Institutional Constraint. American Sociological Review,
80(1):116–139.
Portegijs, W. and van den Brakel, M. (2018). Emancipatiemonitor 2018. Sociaal en
Cultureel Planbureau, Den Haag. OCLC: 1081302463.
Reichelt, M., Makovi, K., and Sargsyan, A. (2021). The impact of COVID-19 on
gender inequality in the labor market and gender-role attitudes. European Societies,
23(sup1):S228–S245.
Rutte, M. (2020a). Television address by Prime Minister Mark Rutte of the Nether-
lands.
Rutte, M. (2020b). Television address by Prime Minister Mark Rutte of the Nether-
lands.
Rutte, M. and Hugo M., D. J. (2020). Literal text of the press conference of Prime
Minister Rutte and Minister De Jonge after the cabinet crisis meeting.
34
Scarborough, W. J. (2019). ATTITUDES AND THE STALLED GENDER REVOLU-
TION. Gender and Society, 33(2):173–200.
Scherpenzeel, A. and Das, M. (2010). “True” Longitudinal and Probability-Based Inter-
net Panels: Evidence From the Netherlands. In Das, M., P. Ester, and L. Kaczmirek
(Eds.).
Sevilla, A. and Smith, S. (2020). Baby steps: the gender division of childcare during
the COVID-19 pandemic. Oxford Review of Economic Policy, 36(S1):S169–S186.
Stickney, L. T. and Konrad, A. M. (2007). Gender-Role Attitudes and Earnings: A
Multinational Study of Married Women and Men. Sex Roles, 57(11-12):801–811.
Unterhofer, U. and Wrohlich, K. (2017). Fathers, Parental Leave and Gender Norms.
SSRN Electronic Journal.
Vandecasteele, L., Ivanova, K., Sieben, I., and Reeskens, T. (2022). Changing attitudes
about the impact of women’s employment on families: The COVID-19 pandemic
effect. Gender, Work & Organization, 29(6):2012–2033.
Vespa, J. (2009). Gender Ideology Construction: A Life Course and Intersectional
Approach. Gender & Society, 23(3):363–387.
Zoch, G. and Schober, P. S. (2018). Public Child-Care Expansion and Changing Gender
Ideologies of Parents in Germany. Journal of Marriage and Family, 80(4):1020–1039.
35
Appendix
Table A1
Trends in Overall Gender Role Attitudes when having Young Children
Dependent variable: Belief Score Men Women
(1) (2) (3) (4)
Child Below 12 × Dit:
-12 -0.123*** 0.468+ 0.144*** -0.555**
(0.019) (0.308) (0.018) (0.258)
-11 0.192*** 0.266** 0.199*** 0.123
(0.015) (0.104) (0.017) (0.086)
-10 0.105*** 0.183* 0.124*** 0.105
(0.016) (0.100) (0.017) (0.083)
-9 -0.156*** 0.031 0.010 0.043
(0.016) (0.097) (0.018) (0.081)
-8 0.147*** 0.209** 0.107*** 0.059
(0.017) (0.093) (0.018) (0.079)
-7 0.016 0.054 0.030+ -0.090
(0.017) (0.091) (0.019) (0.076)
-6 0.150*** 0.123 -0.008 -0.070
(0.018) (0.086) (0.018) (0.075)
-5 0.091*** 0.023 -0.103*** -0.125*
(0.011) (0.085) (0.010) (0.074)
-4 -0.122*** -0.057 -0.090*** -0.003
(0.011) (0.085) (0.009) (0.077)
-3 -0.189*** -0.010 -0.230*** -0.068
(0.010) (0.080) (0.012) (0.070)
-2 -0.158*** -0.057 -0.293*** -0.160**
(0.010) (0.078) (0.011) (0.067)
0 -0.199*** -0.170** -0.227*** -0.091
(0.006) (0.079) (0.008) (0.066)
1 -0.134*** -0.036 -0.198*** -0.177**
(0.006) (0.082) (0.005) (0.077)
2 -0.183*** -0.115 -0.102*** -0.107
(0.006) (0.099) (0.005) (0.080)
Observations 25,321 25,321 27,053 27,053
N° of groups (Years - Ind.) 15 4,981 15 5,634
Year FE YES YES YES YES
Individual FE NO YES NO YES
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1, + p<0.15
Notes: Table A1 reports the estimates from Equation 1. The model is estimated separately for male
and female partners. The outcome variable is the Beliefs Score, an index computed following a PCA
procedure (see Section 5.2). All columns control for age, age squared, level of education, 20 income
quantiles, employment, residential setting, religion, church attendance, marital status, and year-fixed
effects. Individual fixed effects are only included in even-numbered columns.
36
37
Table A2
Trends in Statement-specific Gender Role Attitudes when having Young Children
(1) Child (2) Childrearing (3) Responsibilities (4) Full-Time (5) Leadership (6) Relationship (7) Contribution
Men Women Men Women Men Women Men Women Men Women Men Women Men Women
Child Below 12 × Dit:
-12 0.573** 0.016 0.474** -0.652*** 0.160 -0.382** 0.289 -0.163 -0.161 -0.150 -0.285 -0.003 0.061 -0.244
(0.266) (0.228) (0.196) (0.228) (0.193) (0.189) (0.263) (0.277) (0.202) (0.143) (0.235) (0.208) (0.208) (0.261)
-11 0.297*** 0.292*** 0.029 -0.041 0.012 -0.080+ 0.093 0.038 0.059 0.034 0.232*** 0.077 -0.052 -0.008
(0.079) (0.066) (0.076) (0.066) (0.066) (0.053) (0.078) (0.073) (0.065) (0.046) (0.081) (0.065) (0.077) (0.068)
-10 0.222*** 0.281*** 0.024 -0.047 0.010 -0.049 0.076 0.073 0.012 -0.043 0.109 0.051 0.039 -0.001
(0.076) (0.064) (0.076) (0.063) (0.062) (0.051) (0.076) (0.072) (0.064) (0.045) (0.079) (0.064) (0.075) (0.066)
-9 0.131* 0.253*** -0.069 -0.010 -0.018 -0.107** -0.024 0.022 0.016 -0.012 0.066 -0.019 -0.062 -0.018
(0.074) (0.061) (0.074) (0.064) (0.062) (0.051) (0.075) (0.070) (0.062) (0.045) (0.076) (0.065) (0.075) (0.065)
-8 0.175** 0.202*** 0.010 -0.015 0.097+ -0.046 0.051 0.009 0.086+ 0.009 0.135* 0.010 -0.034 -0.040
(0.072) (0.060) (0.070) (0.061) (0.060) (0.049) (0.071) (0.068) (0.060) (0.044) (0.075) (0.063) (0.073) (0.063)
-7 0.147** 0.112* -0.028 -0.066 -0.031 -0.148*** -0.007 0.025 0.017 -0.040 0.029 -0.050 0.013 -0.097+
(0.071) (0.058) (0.069) (0.059) (0.059) (0.049) (0.071) (0.068) (0.060) (0.043) (0.073) (0.060) (0.070) (0.062)
-6 0.134* 0.060 -0.076 -0.059 0.074 -0.121** 0.092 0.005 0.048 -0.018 0.027 -0.040 -0.001 -0.013
(0.069) (0.057) (0.068) (0.059) (0.056) (0.048) (0.068) (0.065) (0.056) (0.043) (0.071) (0.059) (0.069) (0.061)
-5 0.021 0.089+ -0.006 -0.058 0.021 -0.105** -0.034 -0.035 0.047 -0.050 -0.000 -0.130** -0.013 -0.050
(0.064) (0.056) (0.069) (0.059) (0.057) (0.048) (0.066) (0.065) (0.058) (0.042) (0.067) (0.063) (0.070) (0.061)
-4 0.045 0.133** -0.061 -0.061 -0.095* -0.099** -0.010 0.080 -0.036 -0.049 0.033 0.021 -0.043 -0.033
(0.069) (0.058) (0.069) (0.060) (0.057) (0.047) (0.066) (0.066) (0.057) (0.045) (0.069) (0.061) (0.069) (0.062)
-3 0.032 0.037 -0.002 0.020 -0.042 -0.054 -0.027 0.020 0.026 -0.051 0.004 -0.106* -0.030 -0.041
(0.064) (0.055) (0.069) (0.059) (0.059) (0.044) (0.070) (0.063) (0.051) (0.040) (0.064) (0.063) (0.066) (0.060)
-2 -0.044 -0.023 -0.037 -0.066 0.005 -0.090** -0.045 -0.047 -0.035 -0.105** 0.007 -0.083+ -0.000 0.002
(0.066) (0.052) (0.065) (0.060) (0.055) (0.045) (0.065) (0.060) (0.056) (0.043) (0.070) (0.057) (0.065) (0.060)
0 -0.058 -0.055 -0.122* -0.022 -0.104* -0.020 -0.012 0.003 -0.107* -0.100** 0.001 -0.064 -0.074 0.063
(0.066) (0.054) (0.066) (0.055) (0.053) (0.044) (0.069) (0.062) (0.055) (0.040) (0.069) (0.058) (0.067) (0.059)
1 -0.047 -0.050 0.048 -0.051 -0.052 -0.156*** 0.046 0.034 -0.110* -0.051 0.077 -0.153** -0.051 -0.003
(0.066) (0.056) (0.075) (0.062) (0.059) (0.050) (0.069) (0.066) (0.061) (0.048) (0.070) (0.064) (0.074) (0.063)
2 -0.072 -0.132** 0.017 -0.018 -0.001 -0.032 -0.061 0.003 -0.103* -0.018 -0.060 -0.080 -0.035 0.011
(0.073) (0.061) (0.070) (0.064) (0.066) (0.055) (0.079) (0.077) (0.059) (0.049) (0.078) (0.061) (0.072) (0.067)
Observations 25,366 27,109 25,322 27,053 25,356 27,105 25,366 27,109 25,322 27,053 25,366 27,109 25,356 27,105
Number of individuals 4,988 5,639 4,981 5,634 4,985 5,638 4,988 5,639 4,981 5,634 4,988 5,639 4,985 5,638
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1, + p<0.15
Notes: Table A2 reports the γl estimates from Equation 1. The model is estimated separately for men and women in a couple. The outcome variables are the Statement Scores, whose values (1-5) correspond in
turn to the following responses: ”Fully Agree” ”Agree” ”Neither Agree nor Disagree” ”Disagree” ”Fully Disagree”. Statements ”Relationship (6)” and ”Contribution (7)” present an inverted scale, as agreeing
with them translates into more egalitarian gender role beliefs (higher scores). All columns control for age, age squared, level of education, 20 income quantiles, employment, residential setting, religion, church
attendance, marital status, individual and year-fixed effects. The statements are the following: (1) Child: ”A child that is not yet attending school is likely to suffer the consequences if his or her mother has a
job”; (2) Childrearing: ”A woman is more suited to rearing young children than a man”; (3) Responsibilities: ”The father should earn money, while the mother takes care of the household and the family”; (4)
Full-Time: ”Overall, family life suffers the consequences if the mother has a full-time job”; (5) Leadership: ”It is unnatural for women in firms to have control over men”; (6) Relationship: ”A working mother’s
relationship with her children can be just as close and warm as that of a non-working mother”; (7) Contribution: ”Both father and mother should contribute to the family income”
Table A3
Trends in Gender Role Attitudes among Parents with Young Children when Balancing Work and
Childcare: a Comparison by Gender and Partner’s Working Status
Belief Score (1) Child (2) Childrearing (3) Responsibilities (4) Full-Time (5) Leadership (6) Relationship (7) Contribution
Partner’s working status: Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working
Panel A: Men
Employed × Dit:
-12 1.824** -0.957+ 0.875 -0.151 0.431 0.117 0.811 -0.850 0.092 -0.158 0.953+ 0.112 0.892* -0.029 0.895 0.103
(0.772) (0.646) (1.036) (0.685) (0.413) (0.465) (0.723) (0.635) (0.620) (0.705) (0.634) (0.797) (0.501) (0.706) (1.081) (0.551)
-11 0.235 -0.799** -0.171 -0.362 0.049 0.064 -0.275 -0.333 -0.203 -1.030*** 0.303 -0.180 0.830** -0.493* 0.245 0.473+
(0.508) (0.345) (0.419) (0.315) (0.387) (0.368) (0.340) (0.242) (0.381) (0.376) (0.479) (0.212) (0.401) (0.263) (0.417) (0.295)
-10 0.405 -1.049*** -0.178 -0.547* -0.061 -0.202 -0.086 -0.537** -0.052 -1.088*** 0.270 -0.209 1.128*** -0.416+ 0.051 0.471*
(0.477) (0.313) (0.401) (0.297) (0.372) (0.371) (0.332) (0.226) (0.364) (0.385) (0.469) (0.202) (0.372) (0.255) (0.411) (0.281)
-9 0.549 -0.606* -0.222 -0.494* 0.072 0.060 -0.052 -0.424* 0.028 -0.894** 0.410 0.049 0.949*** -0.178 0.437 0.558**
(0.475) (0.312) (0.395) (0.287) (0.364) (0.374) (0.332) (0.218) (0.368) (0.378) (0.477) (0.197) (0.363) (0.253) (0.414) (0.275)
-8 0.230 -0.665** -0.069 -0.358 -0.119 0.047 -0.307 -0.335+ -0.009 -0.975** 0.218 0.060 0.783** -0.420* 0.229 0.361
(0.469) (0.299) (0.395) (0.281) (0.369) (0.366) (0.327) (0.211) (0.357) (0.378) (0.476) (0.186) (0.358) (0.231) (0.398) (0.273)
-7 0.389 -1.169*** -0.144 -0.652** 0.016 0.012 -0.357 -0.585*** -0.178 -1.068*** 0.356 -0.330* 1.262*** -0.574** 0.202 0.353
(0.462) (0.287) (0.406) (0.281) (0.359) (0.356) (0.342) (0.191) (0.356) (0.370) (0.504) (0.181) (0.373) (0.234) (0.388) (0.258)
-6 0.789* -0.736*** 0.142 -0.411+ -0.188 0.056 0.055 -0.358* 0.313 -0.992*** 0.334 -0.203 1.199*** -0.351+ 0.291 0.599**
(0.468) (0.272) (0.401) (0.277) (0.368) (0.375) (0.313) (0.192) (0.350) (0.368) (0.473) (0.183) (0.347) (0.222) (0.385) (0.245)
-5 0.593 -1.598*** -0.530 -0.518* 0.485 -0.192 0.137 -0.794** 0.532 -0.858** -0.409 -0.850*** 1.439*** -0.649** -0.032 -0.300
(0.645) (0.383) (0.465) (0.311) (0.425) (0.489) (0.424) (0.343) (0.438) (0.394) (0.530) (0.299) (0.416) (0.306) (0.426) (0.419)
-4 0.352 -0.652* -0.487 -0.306 0.285 0.348 0.253 -0.683*** -0.324 -0.927** 0.261 -0.197 0.744+ -0.071 0.303 0.475
(0.501) (0.345) (0.413) (0.313) (0.287) (0.382) (0.418) (0.242) (0.397) (0.377) (0.522) (0.235) (0.509) (0.286) (0.420) (0.352)
-3 -0.045 -0.240 -0.062 -0.080 0.319 0.439 -0.562+ -0.664* -0.283 -0.507 0.073 -0.007 0.621* 0.003 -0.138 0.594
(0.412) (0.552) (0.430) (0.345) (0.249) (0.703) (0.371) (0.401) (0.337) (0.378) (0.323) (0.319) (0.346) (0.373) (0.322) (0.430)
-2 0.240 0.289 -0.316 -0.173 0.045 -0.045 0.199 0.361 -0.453 0.059 0.371 -0.080 0.701 -0.109 0.114 1.070***
(0.444) (0.353) (0.514) (0.243) (0.351) (0.382) (0.213) (0.318) (0.336) (0.336) (0.416) (0.288) (0.561) (0.295) (0.452) (0.390)
0 0.085 -0.044 0.037 -0.130 0.515* 0.298 0.020 -0.132 0.294 -0.276 0.086 -0.131 -0.617* -0.049 -0.064 0.600+
(0.329) (0.479) (0.353) (0.330) (0.297) (0.482) (0.255) (0.239) (0.430) (0.409) (0.238) (0.277) (0.367) (0.271) (0.344) (0.377)
1 0.697 0.239 -0.100 0.557* -0.173 0.370 0.155 -0.008 0.221 -0.102 0.251 0.113 1.086*** -0.301 0.538 0.053
(0.616) (0.292) (0.654) (0.306) (0.539) (0.295) (0.306) (0.131) (0.340) (0.192) (0.365) (0.128) (0.401) (0.309) (0.393) (0.201)
2 -0.820* 0.628** -1.326*** 0.279 -0.293 0.843** -0.768* 0.376* -0.721 -0.266 -0.487 0.223 0.793 0.072 1.235*** 0.255
(0.490) (0.295) (0.323) (0.336) (0.559) (0.409) (0.461) (0.217) (0.617) (0.314) (0.470) (0.180) (0.554) (0.337) (0.406) (0.245)
Observations 2,278 3,191 2,285 3,196 2,278 3,191 2,283 3,194 2,285 3,196 2,278 3,191 2,285 3,196 2,283 3,194
Number of id 859 1,171 861 1,173 859 1,171 861 1,173 861 1,173 859 1,171 861 1,173 861 1,173
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Panel B: Women
Employed × Dit:
-12 / -0.539 / -0.407 / -0.315 / 0.296 / 0.054 / -0.062 / -0.578 / -0.697
(0.522) (0.352) (0.327) (0.353) (0.552) (0.249) (0.473) (0.562)
-11 1.117** 0.064 0.340 -0.047 0.548+ -0.081 0.585** 0.105 0.407 0.179 0.062 -0.040 0.488 -0.020 0.715** 0.160
(0.470) (0.236) (0.367) (0.174) (0.340) (0.152) (0.288) (0.146) (0.327) (0.180) (0.341) (0.144) (0.395) (0.218) (0.304) (0.177)
-10 1.074** 0.131 0.493 0.087 0.546* -0.181 0.554** 0.099 0.133 0.107 0.056 -0.046 0.505 0.098 0.769** 0.294*
(0.470) (0.232) (0.368) (0.171) (0.316) (0.152) (0.282) (0.141) (0.323) (0.181) (0.340) (0.149) (0.392) (0.215) (0.301) (0.173)
-9 0.958** 0.121 0.388 0.057 0.473+ -0.112 0.391 -0.009 0.383 0.226 -0.008 -0.097 0.478 0.176 0.655** 0.187
(0.454) (0.233) (0.343) (0.176) (0.303) (0.153) (0.278) (0.144) (0.300) (0.172) (0.339) (0.147) (0.388) (0.224) (0.297) (0.169)
-8 1.200** 0.279 0.566+ 0.181 0.443+ 0.000 0.572** 0.070 0.513+ 0.264+ 0.039 -0.022 0.319 0.202 0.974*** 0.099
(0.469) (0.238) (0.353) (0.171) (0.300) (0.145) (0.265) (0.145) (0.316) (0.176) (0.345) (0.140) (0.384) (0.222) (0.300) (0.172)
-7 0.982** -0.060 0.431 -0.054 0.235 -0.237+ 0.522* 0.017 0.420 0.216 0.110 -0.193 0.313 0.076 0.682** 0.066
(0.481) (0.231) (0.362) (0.170) (0.304) (0.146) (0.272) (0.141) (0.305) (0.179) (0.348) (0.150) (0.386) (0.219) (0.266) (0.170)
-6 0.658 -0.036 0.248 0.093 0.051 -0.239+ 0.250 0.114 0.345 0.118 0.043 -0.154 0.438 -0.049 0.507* 0.008
(0.458) (0.233) (0.341) (0.164) (0.296) (0.152) (0.272) (0.140) (0.291) (0.181) (0.343) (0.148) (0.352) (0.214) (0.283) (0.176)
-5 1.116*** -0.215 0.398 -0.095 0.205 -0.106 0.437+ -0.129 0.343 0.062 0.084 -0.146 0.946** -0.180 0.724** 0.120
(0.414) (0.243) (0.339) (0.173) (0.380) (0.145) (0.282) (0.149) (0.330) (0.186) (0.359) (0.136) (0.391) (0.234) (0.308) (0.189)
-4 0.887* -0.029 0.537+ -0.004 -0.150 0.011 0.289 0.149 0.462 0.037 0.099 -0.062 0.394 -0.146 0.966*** -0.092
(0.481) (0.243) (0.340) (0.176) (0.397) (0.152) (0.284) (0.155) (0.398) (0.173) (0.357) (0.145) (0.355) (0.211) (0.310) (0.190)
-3 0.535 0.155 -0.115 0.172 -0.024 -0.168 0.108 0.139 0.267 0.097 -0.042 -0.099 0.885* 0.105 0.501+ 0.190
(0.390) (0.214) (0.394) (0.158) (0.342) (0.157) (0.276) (0.140) (0.315) (0.154) (0.321) (0.141) (0.482) (0.217) (0.312) (0.182)
-2 0.839** 0.156 0.419 0.128 0.115 -0.176 0.462* 0.277** 0.210 0.128 0.242 0.013 0.399+ -0.136 0.433 0.165
(0.402) (0.196) (0.331) (0.154) (0.325) (0.141) (0.279) (0.135) (0.264) (0.141) (0.350) (0.151) (0.248) (0.191) (0.316) (0.166)
0 0.144 0.054 0.195 0.167 -0.048 -0.086 0.204 -0.062 0.337 0.103 -0.231 0.041 -0.115 -0.209 0.002 0.206
(0.307) (0.171) (0.317) (0.153) (0.424) (0.144) (0.301) (0.128) (0.242) (0.151) (0.324) (0.119) (0.236) (0.188) (0.253) (0.172)
1 1.009* 0.048 0.332 0.110 0.344 -0.293** 0.520* -0.048 0.129 0.288* 0.385 0.083 0.453+ -0.174 0.685** 0.137
(0.525) (0.200) (0.353) (0.143) (0.340) (0.147) (0.306) (0.140) (0.319) (0.159) (0.378) (0.134) (0.311) (0.218) (0.297) (0.191)
2 0.554 0.186 -0.030 0.103 0.523 0.054 0.107 0.193 -0.153 0.021 0.203 0.044 0.514+ -0.089 0.557* 0.222
(0.662) (0.226) (0.504) (0.169) (0.378) (0.153) (0.433) (0.175) (0.496) (0.206) (0.397) (0.170) (0.329) (0.204) (0.291) (0.193)
Observations 1,646 4,078 1,651 4,086 1,646 4,078 1,651 4,086 1,651 4,086 1,646 4,078 1,651 4,086 1,651 4,086
Number of id 754 1,398 754 1,401 754 1,398 754 1,401 754 1,401 754 1,398 754 1,401 754 1,401
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1, + p<0.15
Notes: Table A3 reports the γl estimates from Equation 1. The model is estimated separately for men and women in a couple. The outcome
variables are the Beliefs Score, an index computed following a PCA procedure (see Section 5.2), and the Statement Scores, whose values
(1-5) correspond in turn to the following responses: ”Fully Agree” ”Agree” ”Neither Agree nor Disagree” ”Disagree” ”Fully Disagree”.
Statements ”Relationship (6)” and ”Contribution (7)” present a reserved scale, as agreeing with them translates into more egalitarian
gender role beliefs (higher scores). All columns control for age, age squared, level of education, 20 income quantiles, employment,
residential setting, religion, church attendance, marital status, individual and year-fixed effects. The statements are the following: (1)
Child: ”A child that is not yet attending school is likely to suffer the consequences if his or her mother has a job”; (2) Childrearing: ”A
woman is more suited to rearing young children than a man”; (3) Responsibilities: ”The father should earn money, while the mother takes
care of the household and the family”; (4) Full-Time: ”Overall, family life suffers the consequences if the mother has a full-time job”; (5)
Leadership: ”It is unnatural for women in firms to have control over men”; (6) Relationship: ”A working mother’s relationship with her
children can be just as close and warm as that of a non-working mother”; (7) Contribution: ”Both father and mother should contribute
to the family income”. γl estimates for sole-earners mothers could not be computed because of a limited sample size.
38
39
Table A4
Pre-trends analysis: Joint Statistical Significance of the Lead Estimates
Belief Score (1) Child (2) Childrearing (3) Responsibilities (4) Full-Time (5) Leadership (6) Relationship (7) Contribution
Panel A: Young Children
Men
All leads 0.00 0.00 0.15 0.01 0.27 0.28 0.02 0.80
Last 7 leads 0.25 0.09 0.78 0.03 0.32 0.43 1.00 0.98
Women
All leads 0.00 0.00 0.19 0.05 0.67 0.14 0.02 0.70
Last 7 leads 0.14 0.08 0.52 0.10 0.42 0.33 0.09 0.52
Panel B: Working Parents
Fathers
All leads 0.02 0.42 0.40 0.13 0.13 0.00 0.19 0.50
Last 7 leads 0.10 0.18 0.81 0.05 0.07 0.03 0.71 0.44
Mothers
All leads 0.02 0.05 0.56 0.55 0.13 0.67 0.53 0.18
Last 7 leads 0.10 0.19 0.98 0.20 0.16 0.81 0.57 0.37
Notes: Table A4 illustrates the P-values of multiple Wald tests with null hypothesis H0 : γl = 0. Joint statistical significance at the 95% level is assessed for all
model specifications. Each test is conducted separately for all leads and only the last 7.
40
Table A5
”Only Constrained” Robustness analysis. Trends in Gender Role Attitudes when having Young Children.
Belief Score (1) Child (2) Childrearing (3) Responsibilities (4) Full-Time (5) Leadership (6) Relationship (7) Contribution
Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women
Child Below 12 × Dit
-12 0.481+ -0.585** 0.602** 0.013 0.473** -0.646*** 0.164 -0.407** 0.297 -0.167 -0.165 -0.152 -0.292 -0.039 0.068 -0.258
(0.312) (0.261) (0.269) (0.229) (0.198) (0.228) (0.195) (0.190) (0.266) (0.279) (0.204) (0.143) (0.236) (0.208) (0.208) (0.262)
-11 0.274*** 0.104 0.305*** 0.294*** 0.023 -0.033 0.023 -0.090* 0.096 0.032 0.058 0.031 0.232*** 0.052 -0.047 -0.026
(0.105) (0.088) (0.080) (0.067) (0.077) (0.067) (0.067) (0.053) (0.079) (0.074) (0.066) (0.046) (0.082) (0.066) (0.078) (0.069)
-10 0.192* 0.088 0.230*** 0.282*** 0.019 -0.039 0.022 -0.058 0.080 0.066 0.012 -0.044 0.110 0.027 0.045 -0.017
(0.101) (0.085) (0.077) (0.064) (0.077) (0.064) (0.063) (0.051) (0.077) (0.074) (0.065) (0.045) (0.080) (0.065) (0.076) (0.066)
-9 0.042 0.023 0.141* 0.251*** -0.074 -0.003 -0.007 -0.115** -0.019 0.013 0.017 -0.013 0.068 -0.044 -0.055 -0.034
(0.098) (0.082) (0.075) (0.062) (0.075) (0.065) (0.063) (0.052) (0.076) (0.071) (0.063) (0.045) (0.077) (0.066) (0.076) (0.065)
-8 0.220** 0.037 0.184** 0.197*** 0.008 -0.009 0.108* -0.054 0.057 -0.002 0.088+ 0.008 0.134* -0.015 -0.028 -0.057
(0.094) (0.080) (0.073) (0.061) (0.072) (0.062) (0.061) (0.050) (0.072) (0.069) (0.061) (0.044) (0.076) (0.063) (0.074) (0.064)
-7 0.068 -0.112+ 0.157** 0.105* -0.029 -0.058 -0.020 -0.156*** -0.000 0.014 0.020 -0.040 0.028 -0.076 0.022 -0.114*
(0.092) (0.077) (0.072) (0.059) (0.070) (0.060) (0.060) (0.050) (0.072) (0.069) (0.061) (0.044) (0.074) (0.061) (0.072) (0.063)
-6 0.137+ -0.089 0.144** 0.054 -0.076 -0.051 0.085+ -0.129*** 0.098 -0.005 0.052 -0.018 0.027 -0.061 0.005 -0.030
(0.088) (0.076) (0.071) (0.057) (0.069) (0.060) (0.057) (0.048) (0.069) (0.066) (0.058) (0.043) (0.072) (0.059) (0.070) (0.062)
-5 0.036 -0.146* 0.029 0.078 -0.004 -0.049 0.033 -0.111** -0.028 -0.046 0.051 -0.052 -0.003 -0.151** -0.008 -0.065
(0.086) (0.076) (0.065) (0.057) (0.070) (0.060) (0.058) (0.049) (0.067) (0.066) (0.059) (0.043) (0.068) (0.063) (0.071) (0.062)
-4 -0.043 -0.025 0.056 0.120** -0.059 -0.053 -0.084+ -0.106** -0.005 0.068 -0.031 -0.051 0.030 0.001 -0.036 -0.046
(0.086) (0.078) (0.070) (0.058) (0.069) (0.060) (0.058) (0.048) (0.067) (0.067) (0.058) (0.045) (0.070) (0.062) (0.069) (0.062)
-3 0.007 -0.080 0.045 0.025 0.007 0.031 -0.034 -0.056 -0.019 0.017 0.032 -0.050 0.001 -0.123* -0.021 -0.051
(0.081) (0.071) (0.065) (0.055) (0.070) (0.060) (0.060) (0.045) (0.071) (0.064) (0.052) (0.041) (0.065) (0.063) (0.067) (0.061)
-2 -0.044 -0.171** -0.035 -0.035 -0.029 -0.054 0.010 -0.093** -0.037 -0.049 -0.033 -0.109** 0.002 -0.092+ 0.011 -0.008
(0.079) (0.068) (0.067) (0.052) (0.066) (0.060) (0.055) (0.046) (0.065) (0.060) (0.057) (0.043) (0.070) (0.058) (0.066) (0.061)
0 -0.250** -0.044 -0.104 -0.040 -0.265*** -0.019 -0.103 0.065 0.056 -0.003 -0.066 -0.103* -0.161+ -0.006 -0.048 0.025
(0.118) (0.089) (0.098) (0.075) (0.099) (0.089) (0.089) (0.070) (0.092) (0.083) (0.074) (0.060) (0.101) (0.086) (0.101) (0.085)
1 -0.008 -0.193** -0.032 -0.077 0.065 -0.037 -0.039 -0.149*** 0.059 0.012 -0.111* -0.053 0.089 -0.168*** -0.041 0.002
(0.083) (0.079) (0.066) (0.056) (0.076) (0.063) (0.059) (0.051) (0.070) (0.067) (0.062) (0.049) (0.071) (0.065) (0.076) (0.064)
2 -0.087 -0.137* -0.053 -0.158** 0.038 -0.009 0.014 -0.034 -0.046 -0.026 -0.099+ -0.017 -0.063 -0.089+ -0.027 -0.011
(0.100) (0.082) (0.075) (0.062) (0.071) (0.065) (0.066) (0.057) (0.080) (0.078) (0.060) (0.049) (0.078) (0.062) (0.074) (0.068)
Observations 24,571 26,116 24,616 26,172 24,572 26,116 24,606 26,168 24,616 26,172 24,572 26,116 24,616 26,172 24,606 26,168
Number of id 4,958 5,600 4,965 5,605 4,958 5,600 4,962 5,604 4,965 5,605 4,958 5,600 4,965 5,605 4,962 5,604
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1, + p<0.15
Notes: Table A5 reports the γl estimates from Equation 1. The model is estimated separately for men and women in a couple. Wave 13 respondents who were interviewed after schools reopened are discarded
from the sample. The outcome variables are the Beliefs Score, an index computed following a PCA procedure (see Section 5.2), and the Statement Scores, whose values (1-5) correspond in turn to the following
responses: ”Fully Agree” ”Agree” ”Neither Agree nor Disagree” ”Disagree” ”Fully Disagree”. Statements ”Relationship (6)” and ”Contribution (7)” present a reserved scale, as agreeing with them translates
into more egalitarian gender role beliefs (higher scores). All columns control for age, age squared, level of education, 20 income quantiles, employment, residential setting, religion, church attendance, marital
status, individual and year-fixed effects. The statements are the following: (1) Child: ”A child that is not yet attending school is likely to suffer the consequences if his or her mother has a job”; (2) Childrearing:
”A woman is more suited to rearing young children than a man”; (3) Responsibilities: ”The father should earn money, while the mother takes care of the household and the family”; (4) Full-Time: ”Overall,
family life suffers the consequences if the mother has a full-time job”; (5) Leadership: ”It is unnatural for women in firms to have control over men”; (6) Relationship: ”A working mother’s relationship with her
children can be just as close and warm as that of a non-working mother”; (7) Contribution: ”Both father and mother should contribute to the family income”.
Table A6
”Only Constrained” Robustness analysis. Trends in Gender Role Attitudes among Parents with
Young Children when Balancing Work and Childcare: a Comparison by Gender and Partner’s
Working Status.
Belief Score (1) Child (2) Childrearing (3) Responsibilities (4) Full-Time (5) Leadership (6) Relationship (7) Contribution
Partner’s working status: Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working
Panel A: Fathers
Employed × Dit
-12 1.859** -0.797 0.902 -0.153 0.420 0.305 0.846 -0.783 0.116 -0.061 0.955+ 0.215 0.895* -0.106 0.896 0.194
(0.772) (0.627) (1.043) (0.682) (0.418) (0.456) (0.722) (0.631) (0.624) (0.690) (0.637) (0.798) (0.507) (0.704) (1.080) (0.541)
-11 0.271 -0.639** -0.148 -0.359 0.039 0.238 -0.244 -0.274 -0.179 -0.924*** 0.310 -0.089 0.839** -0.547** 0.245 0.564**
(0.511) (0.312) (0.425) (0.306) (0.392) (0.362) (0.341) (0.240) (0.389) (0.348) (0.483) (0.210) (0.402) (0.260) (0.418) (0.280)
-10 0.427 -0.892*** -0.168 -0.543* -0.080 -0.033 -0.060 -0.481** -0.039 -0.984*** 0.275 -0.119 1.137*** -0.468* 0.058 0.564**
(0.480) (0.278) (0.408) (0.288) (0.377) (0.369) (0.333) (0.226) (0.374) (0.357) (0.473) (0.198) (0.373) (0.253) (0.413) (0.265)
-9 0.569 -0.443+ -0.217 -0.491* 0.056 0.233 -0.030 -0.369* 0.032 -0.789** 0.419 0.141 0.959*** -0.226 0.450 0.657**
(0.479) (0.275) (0.403) (0.276) (0.368) (0.367) (0.332) (0.216) (0.379) (0.349) (0.483) (0.196) (0.364) (0.251) (0.416) (0.256)
-8 0.234 -0.516* -0.072 -0.358 -0.137 0.201 -0.292 -0.287 -0.015 -0.875** 0.223 0.143 0.783** -0.457* 0.245 0.454*
(0.473) (0.266) (0.403) (0.271) (0.373) (0.370) (0.326) (0.213) (0.368) (0.351) (0.481) (0.181) (0.358) (0.234) (0.400) (0.257)
-7 0.372 -1.021*** -0.156 -0.653** -0.003 0.160 -0.356 -0.538*** -0.200 -0.968*** 0.358 -0.246 1.252*** -0.607** 0.222 0.447*
(0.466) (0.254) (0.413) (0.271) (0.363) (0.361) (0.341) (0.194) (0.367) (0.343) (0.508) (0.175) (0.373) (0.239) (0.390) (0.241)
-6 0.776+ -0.586** 0.129 -0.410+ -0.214 0.196 0.063 -0.315+ 0.285 -0.889*** 0.337 -0.120 1.193*** -0.373* 0.325 0.699***
(0.473) (0.241) (0.409) (0.267) (0.372) (0.385) (0.313) (0.197) (0.362) (0.341) (0.478) (0.174) (0.347) (0.226) (0.387) (0.224)
-5 0.602 -1.450*** -0.544 -0.514* 0.454 -0.067 0.125 -0.766** 0.523 -0.757** -0.369 -0.770*** 1.457*** -0.650** 0.008 -0.195
(0.653) (0.359) (0.484) (0.308) (0.429) (0.484) (0.423) (0.345) (0.448) (0.373) (0.547) (0.290) (0.422) (0.311) (0.435) (0.426)
-4 0.343 -0.424 -0.492 -0.266 0.273 0.510 0.261 -0.626** -0.355 -0.819** 0.255 -0.126 0.748+ -0.035 0.324 0.698**
(0.518) (0.299) (0.424) (0.312) (0.284) (0.402) (0.421) (0.254) (0.401) (0.357) (0.542) (0.233) (0.509) (0.297) (0.423) (0.321)
-3 -0.038 -0.081 -0.038 -0.048 0.316 0.569 -0.549+ -0.637+ -0.299 -0.394 0.063 0.071 0.619* -0.014 -0.129 0.707*
(0.412) (0.518) (0.434) (0.349) (0.249) (0.693) (0.370) (0.403) (0.339) (0.364) (0.325) (0.306) (0.345) (0.383) (0.321) (0.405)
-2 0.294 0.449 -0.267 -0.125 0.051 0.191 0.230 0.436 -0.456 0.139 0.393 -0.027 0.721 -0.219 0.121 1.160***
(0.445) (0.358) (0.516) (0.260) (0.356) (0.410) (0.215) (0.332) (0.334) (0.357) (0.424) (0.253) (0.565) (0.332) (0.454) (0.399)
0 0.186 -0.753** 0.363 -0.607+ 0.885*** -0.380 0.109 -0.508** -0.014 -0.355 -0.090 0.072 -0.563 -0.165 -0.143 0.020
(0.500) (0.336) (0.512) (0.402) (0.323) (0.360) (0.295) (0.247) (0.726) (0.433) (0.423) (0.212) (0.502) (0.309) (0.471) (0.221)
1 0.606 0.187 -0.196 0.551* -0.221 0.301 0.130 0.002 0.144 -0.131 0.235 0.048 1.075*** -0.255 0.597+ -0.001
(0.630) (0.269) (0.673) (0.314) (0.541) (0.285) (0.304) (0.130) (0.353) (0.197) (0.397) (0.124) (0.400) (0.317) (0.396) (0.196)
2 -0.840* 0.751** -1.342*** 0.287 -0.314 0.914** -0.759* 0.442* -0.697 -0.191 -0.559 0.232 0.810+ 0.122 1.242*** 0.333
(0.462) (0.293) (0.342) (0.327) (0.577) (0.373) (0.459) (0.236) (0.628) (0.295) (0.503) (0.181) (0.558) (0.348) (0.419) (0.255)
Observations 2,206 3,090 2,213 3,095 2,206 3,090 2,211 3,093 2,213 3,095 2,206 3,090 2,213 3,095 2,211 3,093
Number of id 842 1,156 844 1,158 842 1,156 844 1,158 844 1,158 842 1,156 844 1,158 844 1,158
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Panel B: Mothers
Employed 12 × Dit
-12 -0.214 -0.593 0.228 -0.773* -0.634 0.022 -0.689*** 0.020 1.196*** 0.003 0.276 0.108 0.515* -0.713* -2.201*** -0.288
(0.302) (0.418) (0.258) (0.408) (0.596) (0.330) (0.231) (0.354) (0.296) (0.442) (0.345) (0.224) (0.276) (0.411) (0.440) (0.492)
-11 0.303 0.144 0.245 0.116 0.063 0.030 0.035 0.048 0.214 0.141 -0.163 -0.089 0.301 0.028 0.176 0.183
(0.252) (0.178) (0.194) (0.134) (0.192) (0.125) (0.152) (0.114) (0.187) (0.140) (0.174) (0.117) (0.217) (0.157) (0.188) (0.135)
-10 0.345 0.051 0.259 0.206+ 0.051 -0.102 0.028 -0.056 0.040 0.103 -0.067 -0.128 0.400* 0.007 0.319* 0.212+
(0.245) (0.181) (0.187) (0.139) (0.173) (0.123) (0.147) (0.114) (0.176) (0.145) (0.168) (0.122) (0.205) (0.157) (0.179) (0.132)
-9 0.234 0.128 0.151 0.178 -0.030 -0.022 -0.035 -0.054 0.104 0.215+ -0.099 -0.205* 0.335+ 0.152 0.335* 0.173
(0.240) (0.180) (0.181) (0.141) (0.172) (0.124) (0.145) (0.112) (0.172) (0.135) (0.166) (0.121) (0.206) (0.163) (0.178) (0.130)
-8 0.469* 0.242 0.270+ 0.275** 0.049 0.042 0.114 -0.001 0.230 0.132 -0.066 -0.046 0.324+ 0.157 0.476*** 0.150
(0.239) (0.181) (0.183) (0.137) (0.167) (0.118) (0.143) (0.112) (0.168) (0.140) (0.170) (0.116) (0.199) (0.161) (0.176) (0.131)
-7 0.215 -0.021 0.065 0.113 -0.064 -0.080 0.077 -0.029 0.083 0.143 -0.053 -0.172+ 0.190 0.008 0.346** -0.023
(0.248) (0.174) (0.188) (0.130) (0.171) (0.117) (0.148) (0.109) (0.174) (0.138) (0.181) (0.119) (0.205) (0.161) (0.168) (0.134)
-6 0.218 -0.029 0.211 0.177 -0.177 -0.129 -0.019 0.093 0.129 0.067 -0.057 -0.133 0.318* -0.120 0.265+ -0.071
(0.240) (0.175) (0.178) (0.130) (0.158) (0.119) (0.147) (0.109) (0.165) (0.138) (0.172) (0.116) (0.189) (0.158) (0.174) (0.134)
-5 0.481** -0.114 0.200 0.056 -0.006 -0.058 0.008 -0.098 0.374** 0.015 0.002 -0.170+ 0.554*** -0.103 0.210 0.118
(0.223) (0.178) (0.179) (0.134) (0.177) (0.116) (0.139) (0.114) (0.183) (0.141) (0.166) (0.105) (0.193) (0.167) (0.172) (0.141)
-4 0.414* 0.117 0.132 0.185 -0.068 0.076 0.026 0.068 0.320* 0.125 0.131 -0.061 0.298 -0.063 0.393** -0.025
(0.225) (0.177) (0.181) (0.133) (0.185) (0.115) (0.138) (0.123) (0.187) (0.140) (0.152) (0.117) (0.208) (0.160) (0.165) (0.137)
-3 0.140 0.005 -0.120 0.123 -0.020 -0.086 -0.122 0.094 0.254+ -0.058 -0.177 -0.104 0.459* -0.031 0.222 0.075
(0.218) (0.168) (0.189) (0.122) (0.159) (0.119) (0.142) (0.112) (0.172) (0.126) (0.148) (0.114) (0.239) (0.156) (0.178) (0.130)
-2 0.133 0.125 0.110 0.121 -0.003 -0.091 -0.039 0.204* 0.164 0.058 0.007 -0.007 0.009 -0.079 0.172 0.114
(0.226) (0.151) (0.186) (0.124) (0.170) (0.103) (0.143) (0.108) (0.162) (0.128) (0.159) (0.111) (0.159) (0.153) (0.184) (0.128)
0 -0.032 -0.009 0.037 0.042 -0.073 -0.007 -0.043 -0.046 0.096 0.107 -0.043 0.031 -0.034 -0.231* -0.010 0.089
(0.190) (0.137) (0.169) (0.112) (0.182) (0.112) (0.149) (0.107) (0.155) (0.115) (0.146) (0.095) (0.160) (0.138) (0.174) (0.135)
1 0.195 0.075 -0.065 0.074 0.107 -0.173+ 0.006 -0.069 0.249+ 0.166 0.192 0.098 0.012 0.020 0.078 0.070
(0.262) (0.171) (0.193) (0.113) (0.176) (0.120) (0.159) (0.120) (0.171) (0.128) (0.162) (0.100) (0.187) (0.174) (0.186) (0.138)
2 0.024 0.213 -0.093 0.152 0.157 -0.023 -0.092 0.160 0.054 0.128 0.194 0.039 0.069 -0.008 -0.188 0.134
(0.294) (0.178) (0.222) (0.138) (0.182) (0.118) (0.176) (0.126) (0.206) (0.156) (0.159) (0.116) (0.205) (0.170) (0.199) (0.145)
Observations 3,524 5,759 3,532 5,769 3,524 5,759 3,531 5,769 3,532 5,769 3,524 5,759 3,532 5,769 3,531 5,769
Number of id 1,239 1,757 1,239 1,760 1,239 1,757 1,239 1,760 1,239 1,760 1,239 1,757 1,239 1,760 1,239 1,760
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1, + p<0.15
Notes: Table A6 reports the γl estimates from Equation 2. The model is estimated separately for fathers (Panel A) and mothers (Panel B)
of young children across their partner’s working status. Wave 13 respondents who were interviewed after schools reopened are discarded
from the sample. The outcome variables are the Beliefs Score, an index computed following a PCA procedure (see Section 5.2), and the
Statement Scores, whose values (1-5) correspond in turn to the following responses: ”Fully Agree” ”Agree” ”Neither Agree nor Disagree”
”Disagree” ”Fully Disagree”. Statements ”Relationship (6)” and ”Contribution (7)” present a reserved scale, as agreeing with them
translates into more egalitarian gender role beliefs (higher scores). All columns control for age, age squared, level of education, 20 income
quantiles, employment, residential setting, religion, church attendance, marital status, individual and year-fixed effects. The statements
are the following: (1) Child: ”A child that is not yet attending school is likely to suffer the consequences if his or her mother has a
job”; (2) Childrearing: ”A woman is more suited to rearing young children than a man”; (3) Responsibilities: ”The father should earn
money, while the mother takes care of the household and the family”; (4) Full-Time: ”Overall, family life suffers the consequences if the
mother has a full-time job”; (5) Leadership: ”It is unnatural for women in firms to have control over men”; (6) Relationship: ”A working
mother’s relationship with her children can be just as close and warm as that of a non-working mother”; (7) Contribution: ”Both father
and mother should contribute to the family income”.
41
42
Table A7
”Children of All Ages” Robustness analysis. Trends in Gender Role Attitudes when having Children of All Ages.
Belief Score (1) Child (2) Childrearing (3) Responsibilities (4) Full-Time (5) Leadership (6) Relationship (7) Contribution
Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women
Child × Dit
-12 -0.015 -0.080 0.018 -0.330* 0.158 -0.171 0.038 0.036 -0.023 0.383 0.055 0.028 -0.088 0.020 -0.234+ -0.368
(0.235) (0.181) (0.191) (0.192) (0.140) (0.294) (0.184) (0.197) (0.227) (0.282) (0.188) (0.116) (0.212) (0.260) (0.148) (0.296)
-11 -0.080 -0.015 0.016 0.050 -0.036 -0.026 -0.035 0.015 -0.060 -0.015 -0.032 -0.018 -0.012 0.018 -0.088+ -0.093*
(0.073) (0.070) (0.058) (0.052) (0.057) (0.056) (0.054) (0.044) (0.058) (0.058) (0.055) (0.045) (0.061) (0.054) (0.058) (0.052)
-10 -0.060 -0.024 0.020 0.016 0.065 -0.032 0.021 0.028 -0.070 -0.024 -0.058 -0.009 -0.121* 0.004 -0.017 -0.077+
(0.074) (0.071) (0.057) (0.053) (0.058) (0.054) (0.056) (0.043) (0.059) (0.055) (0.056) (0.044) (0.062) (0.055) (0.059) (0.051)
-9 -0.037 -0.028 0.013 0.044 0.044 0.030 0.008 -0.008 -0.035 -0.026 -0.039 0.002 -0.071 -0.031 -0.019 -0.110**
(0.074) (0.067) (0.056) (0.050) (0.058) (0.053) (0.055) (0.042) (0.060) (0.052) (0.056) (0.043) (0.059) (0.055) (0.061) (0.051)
-8 -0.051 -0.061 -0.006 0.018 0.011 -0.008 0.018 0.016 -0.059 -0.043 -0.056 -0.029 -0.035 -0.057 -0.013 -0.081+
(0.070) (0.066) (0.055) (0.049) (0.055) (0.051) (0.051) (0.042) (0.058) (0.052) (0.052) (0.045) (0.056) (0.051) (0.056) (0.049)
-7 -0.029 -0.086 0.016 0.001 0.053 -0.013 0.018 -0.023 -0.092* -0.061 -0.035 -0.038 -0.037 -0.032 0.005 -0.093*
(0.071) (0.062) (0.054) (0.049) (0.054) (0.051) (0.054) (0.041) (0.055) (0.049) (0.054) (0.046) (0.055) (0.051) (0.055) (0.049)
-6 -0.039 -0.090 -0.008 -0.046 0.025 -0.002 0.064 0.003 -0.064 -0.032 -0.014 -0.031 -0.105* -0.061 -0.021 -0.088*
(0.066) (0.065) (0.055) (0.047) (0.055) (0.050) (0.051) (0.041) (0.053) (0.050) (0.052) (0.047) (0.055) (0.050) (0.058) (0.049)
-5 -0.077 -0.085 -0.024 0.003 0.035 -0.022 0.022 0.013 -0.075 -0.031 -0.047 -0.075* -0.102* -0.042 -0.017 -0.109**
(0.069) (0.059) (0.055) (0.046) (0.052) (0.047) (0.051) (0.041) (0.054) (0.047) (0.049) (0.041) (0.055) (0.054) (0.058) (0.048)
-4 -0.070 -0.026 -0.004 0.025 0.039 -0.023 -0.012 -0.032 -0.036 -0.003 -0.077 -0.048 -0.082 0.040 0.002 -0.042
(0.074) (0.060) (0.054) (0.047) (0.056) (0.048) (0.050) (0.041) (0.057) (0.052) (0.055) (0.041) (0.058) (0.054) (0.060) (0.048)
-3 -0.059 -0.035 0.008 0.008 0.052 0.069 -0.009 -0.024 -0.032 0.015 -0.020 0.010 -0.086* -0.065 -0.088+ -0.139***
(0.063) (0.058) (0.052) (0.047) (0.054) (0.050) (0.054) (0.040) (0.058) (0.046) (0.047) (0.040) (0.051) (0.061) (0.054) (0.046)
-2 0.007 -0.131** 0.057 -0.054 0.109** 0.014 -0.011 -0.090** -0.004 -0.046 -0.041 -0.010 -0.042 -0.087* -0.060 -0.074+
(0.063) (0.058) (0.053) (0.048) (0.052) (0.048) (0.046) (0.041) (0.050) (0.045) (0.049) (0.044) (0.053) (0.051) (0.052) (0.047)
0 0.062 -0.065 0.050 -0.029 0.085* -0.001 0.071* -0.024 0.049 -0.040 -0.073 -0.014 0.014 -0.032 -0.046 -0.033
(0.061) (0.050) (0.054) (0.046) (0.047) (0.045) (0.043) (0.038) (0.054) (0.040) (0.051) (0.039) (0.057) (0.050) (0.053) (0.046)
1 -0.013 -0.055 0.019 -0.090* 0.101* 0.013 0.010 0.020 -0.038 -0.069 -0.032 -0.022 -0.062 -0.008 -0.026 0.034
(0.068) (0.069) (0.053) (0.049) (0.055) (0.050) (0.045) (0.045) (0.051) (0.052) (0.049) (0.043) (0.059) (0.058) (0.054) (0.048)
2 0.008 0.009 0.001 -0.008 0.108* 0.001 0.079+ 0.007 -0.011 0.039 -0.030 0.007 -0.052 -0.037 -0.111* -0.002
(0.072) (0.062) (0.057) (0.050) (0.056) (0.048) (0.051) (0.043) (0.055) (0.055) (0.053) (0.040) (0.061) (0.053) (0.061) (0.053)
Observations 11,199 13,023 11,229 13,062 11,199 13,023 11,221 13,059 11,229 13,062 11,199 13,023 11,229 13,062 11,221 13,059
Number of id 2,740 3,274 2,747 3,279 2,740 3,274 2,745 3,279 2,747 3,279 2,740 3,274 2,747 3,279 2,745 3,279
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1, + p<0.15
Notes: Table A7 reports the γl estimates from Equation 3. The model is estimated separately for men and women partners. The outcome variables are the Beliefs Score, an index computed following a PCA
procedure (see Section 5.2), and the Statement Scores, whose values (1-5) correspond in turn to the following responses: ”Fully Agree” ”Agree” ”Neither Agree nor Disagree” ”Disagree” ”Fully Disagree”.
Statements ”Relationship (6)” and ”Contribution (7)” present a reserved scale, as agreeing with them translates into more egalitarian gender role beliefs (higher scores). All columns control for age, age squared,
level of education, 20 income quantiles, employment, residential setting, religion, church attendance, marital status, individual and year-fixed effects. The statements are the following: (1) Child: ”A child that
is not yet attending school is likely to suffer the consequences if his or her mother has a job”; (2) Childrearing: ”A woman is more suited to rearing young children than a man”; (3) Responsibilities: ”The father
should earn money, while the mother takes care of the household and the family”; (4) Full-Time: ”Overall, family life suffers the consequences if the mother has a full-time job”; (5) Leadership: ”It is unnatural
for women in firms to have control over men”; (6) Relationship: ”A working mother’s relationship with her children can be just as close and warm as that of a non-working mother”; (7) Contribution: ”Both
father and mother should contribute to the family income”.
Table A8
”Children of All Ages” Robustness analysis. Trends in Gender Role Attitudes among Parents when
Balancing Work and Childcare: a Comparison by Gender and Partner’s Working Status.
Belief Score (1) Child (2) Childrearing (3) Responsibilities (4) Full-Time (5) Leadership (6) Relationship (7) Contribution
Partner’s working status: Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working Not Working Working
Panel A: Fathers
Employed × Dit
-12 0.486 -0.479 0.391 -0.368 0.156 0.171 0.189 -0.387 -0.264 0.634 0.547** 0.028 0.028 0.044 0.332 0.038
(0.590) (0.654) (0.889) (0.653) (0.282) (0.340) (0.731) (0.501) (0.499) (0.583) (0.267) (0.702) (0.399) (0.661) (0.596) (0.440)
-11 -0.165 -0.379 -0.164 -0.378** -0.316+ -0.014 -0.036 0.009 -0.350* -0.285+ 0.223 -0.217 0.070 -0.263 0.153 0.283
(0.297) (0.285) (0.246) (0.190) (0.216) (0.223) (0.199) (0.182) (0.203) (0.197) (0.235) (0.248) (0.265) (0.205) (0.228) (0.210)
-10 -0.057 -0.484* -0.099 -0.427** -0.377* -0.154 0.088 -0.112 -0.281+ -0.307+ 0.147 -0.194 0.287 -0.200 -0.026 0.242
(0.274) (0.267) (0.221) (0.182) (0.210) (0.223) (0.191) (0.173) (0.192) (0.199) (0.217) (0.246) (0.251) (0.206) (0.233) (0.208)
-9 0.092 -0.205 -0.154 -0.449** -0.225 -0.010 0.182 -0.084 -0.166 -0.144 0.230 -0.060 0.132 0.091 0.250 0.250
(0.276) (0.278) (0.215) (0.180) (0.197) (0.226) (0.188) (0.174) (0.198) (0.206) (0.230) (0.248) (0.247) (0.209) (0.236) (0.214)
-8 -0.213 -0.443+ -0.112 -0.462** -0.388* -0.038 -0.143 -0.070 -0.198 -0.272 0.095 -0.100 0.090 -0.262 0.084 0.042
(0.285) (0.284) (0.219) (0.182) (0.205) (0.225) (0.194) (0.169) (0.204) (0.201) (0.227) (0.243) (0.247) (0.202) (0.224) (0.206)
-7 -0.126 -0.802*** -0.156 -0.603*** -0.239 -0.114 -0.078 -0.268* -0.235 -0.329+ 0.060 -0.393* 0.209 -0.368* 0.123 0.085
(0.283) (0.276) (0.225) (0.187) (0.195) (0.216) (0.195) (0.163) (0.214) (0.201) (0.241) (0.236) (0.263) (0.211) (0.218) (0.208)
-6 0.104 -0.471* 0.088 -0.489** -0.392* -0.076 0.159 -0.115 0.018 -0.198 0.063 -0.205 0.189 -0.229 0.072 0.194
(0.260) (0.277) (0.212) (0.194) (0.202) (0.232) (0.176) (0.173) (0.193) (0.216) (0.229) (0.232) (0.253) (0.201) (0.212) (0.205)
-5 -0.136 -0.655** -0.204 -0.424** -0.132 -0.095 -0.015 -0.292+ -0.119 -0.254 -0.221 -0.383+ 0.471* -0.129 -0.251 -0.069
(0.286) (0.299) (0.251) (0.209) (0.234) (0.239) (0.207) (0.193) (0.199) (0.219) (0.241) (0.242) (0.282) (0.245) (0.269) (0.275)
-4 -0.482+ -0.525** -0.560*** -0.457** -0.398** -0.051 0.069 -0.212 -0.471** -0.250 -0.066 -0.234 -0.015 -0.083 0.176 -0.026
(0.301) (0.245) (0.214) (0.198) (0.196) (0.216) (0.193) (0.186) (0.237) (0.188) (0.280) (0.210) (0.254) (0.201) (0.210) (0.290)
-3 -0.276 -0.467+ -0.207 -0.302 -0.089 -0.179 -0.147 -0.468*** -0.280 -0.074 -0.101 -0.087 0.052 -0.102 0.090 0.126
(0.266) (0.320) (0.231) (0.233) (0.154) (0.328) (0.248) (0.160) (0.210) (0.230) (0.200) (0.212) (0.238) (0.245) (0.232) (0.203)
-2 0.127 -0.176 0.135 -0.144 -0.058 -0.246 0.098 0.041 -0.107 0.185 0.063 -0.275 0.194 -0.209 -0.030 0.287
(0.231) (0.277) (0.221) (0.209) (0.175) (0.228) (0.182) (0.164) (0.166) (0.193) (0.226) (0.233) (0.267) (0.220) (0.220) (0.221)
0 -0.226 0.148 -0.029 -0.127 -0.036 0.304 -0.085 0.151 -0.153 0.076 -0.097 -0.217 0.029 0.170 -0.343** 0.131
(0.214) (0.307) (0.192) (0.159) (0.164) (0.315) (0.153) (0.230) (0.195) (0.214) (0.184) (0.225) (0.237) (0.197) (0.163) (0.218)
1 0.337 0.075 0.049 -0.030 -0.144 0.649** 0.148 -0.170 0.019 0.072 -0.089 -0.268 0.686*** -0.006 0.263 0.175
(0.259) (0.262) (0.220) (0.248) (0.238) (0.301) (0.172) (0.131) (0.193) (0.141) (0.187) (0.212) (0.235) (0.183) (0.194) (0.228)
2 -0.366 -0.663 -0.394* -0.308 -0.203 0.002 -0.094 -0.168 -0.401* -0.070 -0.306+ -0.423+ 0.341 -0.285 0.145 -0.652*
(0.297) (0.811) (0.238) (0.504) (0.199) (0.341) (0.230) (0.292) (0.223) (0.556) (0.198) (0.290) (0.282) (0.498) (0.251) (0.355)
Observations 4,075 4,624 4,084 4,633 4,075 4,624 4,082 4,630 4,084 4,633 4,075 4,624 4,084 4,633 4,082 4,630
Number of id 1,239 1,497 1,241 1,501 1,239 1,497 1,241 1,500 1,241 1,501 1,239 1,497 1,241 1,501 1,241 1,500
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Panel A: Mothers
Employed × Dit
-12 -0.214 -0.593 0.228 -0.773* -0.634 0.022 -0.689*** 0.020 1.196*** 0.003 0.276 0.108 0.515* -0.713* -2.201*** -0.288
(0.302) (0.418) (0.258) (0.408) (0.596) (0.330) (0.231) (0.354) (0.296) (0.442) (0.345) (0.224) (0.276) (0.411) (0.440) (0.492)
-11 0.303 0.144 0.245 0.116 0.063 0.030 0.035 0.048 0.214 0.141 -0.163 -0.089 0.301 0.028 0.176 0.183
(0.252) (0.178) (0.194) (0.134) (0.192) (0.125) (0.152) (0.114) (0.187) (0.140) (0.174) (0.117) (0.217) (0.157) (0.188) (0.135)
-10 0.345 0.051 0.259 0.206+ 0.051 -0.102 0.028 -0.056 0.040 0.103 -0.067 -0.128 0.400* 0.007 0.319* 0.212+
(0.245) (0.181) (0.187) (0.139) (0.173) (0.123) (0.147) (0.114) (0.176) (0.145) (0.168) (0.122) (0.205) (0.157) (0.179) (0.132)
-9 0.234 0.128 0.151 0.178 -0.030 -0.022 -0.035 -0.054 0.104 0.215+ -0.099 -0.205* 0.335+ 0.152 0.335* 0.173
(0.240) (0.180) (0.181) (0.141) (0.172) (0.124) (0.145) (0.112) (0.172) (0.135) (0.166) (0.121) (0.206) (0.163) (0.178) (0.130)
-8 0.469* 0.242 0.270+ 0.275** 0.049 0.042 0.114 -0.001 0.230 0.132 -0.066 -0.046 0.324+ 0.157 0.476*** 0.150
(0.239) (0.181) (0.183) (0.137) (0.167) (0.118) (0.143) (0.112) (0.168) (0.140) (0.170) (0.116) (0.199) (0.161) (0.176) (0.131)
-7 0.215 -0.021 0.065 0.113 -0.064 -0.080 0.077 -0.029 0.083 0.143 -0.053 -0.172+ 0.190 0.008 0.346** -0.023
(0.248) (0.174) (0.188) (0.130) (0.171) (0.117) (0.148) (0.109) (0.174) (0.138) (0.181) (0.119) (0.205) (0.161) (0.168) (0.134)
-6 0.218 -0.029 0.211 0.177 -0.177 -0.129 -0.019 0.093 0.129 0.067 -0.057 -0.133 0.318* -0.120 0.265+ -0.071
(0.240) (0.175) (0.178) (0.130) (0.158) (0.119) (0.147) (0.109) (0.165) (0.138) (0.172) (0.116) (0.189) (0.158) (0.174) (0.134)
-5 0.481** -0.114 0.200 0.056 -0.006 -0.058 0.008 -0.098 0.374** 0.015 0.002 -0.170+ 0.554*** -0.103 0.210 0.118
(0.223) (0.178) (0.179) (0.134) (0.177) (0.116) (0.139) (0.114) (0.183) (0.141) (0.166) (0.105) (0.193) (0.167) (0.172) (0.141)
-4 0.414* 0.117 0.132 0.185 -0.068 0.076 0.026 0.068 0.320* 0.125 0.131 -0.061 0.298 -0.063 0.393** -0.025
(0.225) (0.177) (0.181) (0.133) (0.185) (0.115) (0.138) (0.123) (0.187) (0.140) (0.152) (0.117) (0.208) (0.160) (0.165) (0.137)
-3 0.140 0.005 -0.120 0.123 -0.020 -0.086 -0.122 0.094 0.254+ -0.058 -0.177 -0.104 0.459* -0.031 0.222 0.075
(0.218) (0.168) (0.189) (0.122) (0.159) (0.119) (0.142) (0.112) (0.172) (0.126) (0.148) (0.114) (0.239) (0.156) (0.178) (0.130)
-2 0.133 0.125 0.110 0.121 -0.003 -0.091 -0.039 0.204* 0.164 0.058 0.007 -0.007 0.009 -0.079 0.172 0.114
(0.226) (0.151) (0.186) (0.124) (0.170) (0.103) (0.143) (0.108) (0.162) (0.128) (0.159) (0.111) (0.159) (0.153) (0.184) (0.128)
0 -0.032 -0.009 0.037 0.042 -0.073 -0.007 -0.043 -0.046 0.096 0.107 -0.043 0.031 -0.034 -0.231* -0.010 0.089
(0.190) (0.137) (0.169) (0.112) (0.182) (0.112) (0.149) (0.107) (0.155) (0.115) (0.146) (0.095) (0.160) (0.138) (0.174) (0.135)
1 0.195 0.075 -0.065 0.074 0.107 -0.173+ 0.006 -0.069 0.249+ 0.166 0.192 0.098 0.012 0.020 0.078 0.070
(0.262) (0.171) (0.193) (0.113) (0.176) (0.120) (0.159) (0.120) (0.171) (0.128) (0.162) (0.100) (0.187) (0.174) (0.186) (0.138)
2 0.024 0.213 -0.093 0.152 0.157 -0.023 -0.092 0.160 0.054 0.128 0.194 0.039 0.069 -0.008 -0.188 0.134
(0.294) (0.178) (0.222) (0.138) (0.182) (0.118) (0.176) (0.126) (0.206) (0.156) (0.159) (0.116) (0.205) (0.170) (0.199) (0.145)
Observations 3,524 5,759 3,532 5,769 3,524 5,759 3,531 5,769 3,532 5,769 3,524 5,759 3,532 5,769 3,531 5,769
Number of id 1,239 1,757 1,239 1,760 1,239 1,757 1,239 1,760 1,239 1,760 1,239 1,757 1,239 1,760 1,239 1,760
Year FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1, + p<0.15
Notes: Table A8 reports the γl estimates from Equation 2. The model is estimated separately for fathers (Panel A) and mothers (Panel
B) of children of all ages, across their partner’s working status. Wave 13 respondents who were interviewed after schools reopened are
discarded from the sample. The outcome variables are the Beliefs Score, an index computed following a PCA procedure (see Section 5.2),
and the Statement Scores, whose values (1-5) correspond in turn to the following responses: ”Fully Agree” ”Agree” ”Neither Agree nor
Disagree” ”Disagree” ”Fully Disagree”. Statements ”Relationship (6)” and ”Contribution (7)” present a reserved scale, as agreeing with
them translates into more egalitarian gender role beliefs (higher scores). All columns control for age, age squared, level of education,
20 income quantiles, employment, residential setting, religion, church attendance, marital status, individual and year-fixed effects. The
statements are the following: (1) Child: ”A child that is not yet attending school is likely to suffer the consequences if his or her mother
has a job”; (2) Childrearing: ”A woman is more suited to rearing young children than a man”; (3) Responsibilities: ”The father should
earn money, while the mother takes care of the household and the family”; (4) Full-Time: ”Overall, family life suffers the consequences
if the mother has a full-time job”; (5) Leadership: ”It is unnatural for women in firms to have control over men”; (6) Relationship: ”A
working mother’s relationship with her children can be just as close and warm as that of a non-working mother”; (7) Contribution: ”Both
father and mother should contribute to the family income”.
43
Table A9
”Non-Gender Roles” Robustness analysis. Trends in Attitudes both Related and Unrelated to
Gender Roles when having Young Children.
Men Women
Belief Score [I]Foreign [II] Marriage [III] Euthanasia Belief Score [I] Foreign [II] Marriage [III] Euthanasia
Child Below 12 × Dit
-12 0.468+ -0.027 0.214 -0.055 -0.555** -0.193 0.002 -0.222
(0.308) (0.220) (0.194) (0.189) (0.258) (0.193) (0.216) (0.161)
-11 0.266** 0.074 0.056 0.015 0.123 0.017 0.104* 0.018
(0.104) (0.069) (0.066) (0.057) (0.086) (0.062) (0.054) (0.044)
-10 0.183* 0.068 0.133** 0.054 0.105 0.008 0.110** 0.012
(0.100) (0.066) (0.063) (0.054) (0.083) (0.061) (0.052) (0.043)
-9 0.031 0.022 0.050 0.074 0.043 0.014 0.032 -0.005
(0.097) (0.065) (0.064) (0.052) (0.081) (0.060) (0.050) (0.042)
-8 0.209** 0.053 0.116** 0.037 0.059 0.016 0.039 0.016
(0.093) (0.063) (0.057) (0.051) (0.079) (0.059) (0.049) (0.040)
-7 0.054 0.017 0.050 0.018 -0.090 -0.036 0.009 0.044
(0.091) (0.064) (0.058) (0.050) (0.076) (0.058) (0.048) (0.039)
-6 0.123 -0.031 0.063 0.025 -0.070 -0.036 -0.008 -0.013
(0.086) (0.062) (0.056) (0.048) (0.075) (0.057) (0.048) (0.040)
-5 0.023 0.019 0.037 0.025 -0.125* -0.039 -0.042 0.017
(0.085) (0.062) (0.056) (0.047) (0.074) (0.056) (0.048) (0.040)
-4 -0.057 -0.030 0.029 0.008 -0.003 0.005 -0.023 0.011
(0.085) (0.065) (0.054) (0.047) (0.077) (0.058) (0.048) (0.037)
-3 -0.010 -0.026 0.054 0.021 -0.068 -0.031 -0.015 0.052
(0.080) (0.062) (0.050) (0.044) (0.070) (0.056) (0.043) (0.038)
-2 -0.057 0.091+ -0.084+ -0.044 -0.160** 0.025 -0.092** -0.066+
(0.078) (0.061) (0.055) (0.047) (0.067) (0.055) (0.046) (0.040)
0 -0.170** -0.060 -0.046 -0.038 -0.091 -0.010 -0.043 0.001
(0.079) (0.063) (0.054) (0.045) (0.066) (0.057) (0.042) (0.039)
1 -0.036 0.017 -0.043 -0.011 -0.177** 0.043 -0.107** 0.015
(0.082) (0.066) (0.053) (0.048) (0.077) (0.063) (0.048) (0.038)
2 -0.115 -0.012 -0.008 -0.009 -0.107 -0.072 -0.061 0.012
(0.099) (0.067) (0.059) (0.048) (0.080) (0.062) (0.051) (0.040)
Observations 25,321 25,348 25,336 24,584 27,053 27,089 27,084 26,036
Number of id 4,981 4,984 4,983 4,902 5,634 5,637 5,637 5,533
Year FE YES YES YES YES YES YES YES YES
Ind. FE YES YES YES YES YES YES YES YES
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1, + p<0.15
Notes: Table A9 reports the γl estimates from Equation 1. The model is estimated separately for men and women in a couple. The
outcome variables are the Beliefs Score, an index computed following a PCA procedure (see Section 5.2), and three additional indexes
encoding opinions on tolerance toward foreigners ([I] Foreign: ”It does not help a neighbourhood if many people of foreign origin or descent
move in.”), parents’ marital status ([II] Marriage: ”People that want to have children should get married.”) and support for euthanasia
([III] Euthanasia: ”Where would you place yourself on a scale from 1 to 5, where 1 means that euthanasia should be forbidden and 5
means that euthanasia should be permitted?”). Responses are assigned to numbers from 1 to 5 so that a higher value corresponds to more
progressive opinions.
44
Figure A1
Share of Women working Part-Time Over Female Total Employment
Source: OECD
Notes: Figure A1 represents the evolution of the share of women engaged in part-time jobs among those
employed in the OECD countries from 2007 to 2022. The dashed line represents the OECD average, while
the blue line represents Dutch women.
45
Figure A2
Differences in Overall Gender Role Beliefs during and after School Closure (Wave 13). By Gender,
Young Children, and Working Status
a) Men
b) Women
Source: Longitudinal Internet studies for the Social Sciences (LISS).
Notes: Figure A2 illustrates the differences in overall gender role beliefs of Wave 13 respondents between those inter-
viewed during and after the school closure period in 2021. On the y-axis lies the average Belief Score, constructed as
a mean average of the seven gender role items described in Table 2. The variation in beliefs is reported separately for
men (a) and women (b) across young children and both partners’ working status.
46