Difficult Times: The Division of Domestic Work under Lockdown in France

The lockdowns enforced in many countries to contain the spread of COVID-19 had important consequences for the domestic sphere. This paper analyzes the division of domestic work among heterosexual couples in France during the lockdown. In particular, we analyzed the role of time constraints and availability and expected to find a more egalitarian division of domestic work among couples in which the man had more time than his partner due to not working or working from home. We used data from the ELIPSS panel, a representative survey of the French population, and ran OLS regressions on the division of domestic work among 406 couples. The results show that men’s time availability was associated with a more egalitarian division of domestic work, even if gender inequalities persisted. However, we did not find any clear differences between men who did not work and men who worked from home, leading us to hypothesize that men’s presence at home is an important factor.


Introduction
The lockdowns enforced in many countries to contain the spread of COVID-19 had important consequences for the domestic sphere. As families were forced to spend more time at home and the outsourcing of domestic and care work became less available, confinement was likely to produce an increase in the domestic load of households. Given that the division of domestic work is gendered and that women continue performing the lion's share of domestic work (Altintas and Sullivan 2016), it is important to question the impact that lockdowns have had on gender inequalities within the family.
Lockdowns have also resulted in family members spending more time at home and possibly having more time available due to the elimination of commuting and other commitments. Time availability theories would lead us to expect that both men and women would engage in more domestic work during lockdowns. Existing research on some countries have pointed in this direction (Andrew et al. 2020;Biroli et al. 2020;Carlson et al. 2020;Craig and Churchill 2020;Farré et al. 2020;Fodor et al. 2020;Hank and Steinbach 2020;Hipp and Bünning 2020;Kreyenfeld et al. 2020;Safi et al. 2020;Seiz 2020;Sevilla and Smith 2020;Zhou et al. 2020). However, these increases do not necessarily translate into a more egalitarian division of work, and results from the above-mentioned studies show that gender inequalities persist.
To understand within-couple arrangements, we need to consider the partners' relative time constraints-particularly their employment situations, which may have significantly changed during the pandemic period. Lockdowns have entailed an increase in both unemployment and working from home, and these two situations are likely to have different impacts on couples' time use patterns and division of domestic work. Their effects are also likely to vary by gender. For instance, previous research has shown that unemployed women tend to use their free time to perform domestic work more often than men (Aguiar et al. 2013;Berik and Kongar 2013;Gough and Killewald 2011).
The aim of the current article is to investigate the division of domestic work in France during the lockdowns that occurred in response to the first wave of COVID-19. We hypoth-Soc. Sci. 2021, 10, 219 2 of 13 esized that relative time constraints are important obstacles for a more egalitarian division of work within couples. The study contributes to the emerging literature on the COVID-19 pandemic by examining the employment configuration of households as a key variable for understanding the division of work. We evaluated the impact of being unemployed, not working, and working from home during this particular period and expected employment situation to have gendered effects. We used data from a representative survey conducted during the lockdowns to investigate our hypothesis.

Theoretical Background
In many countries, the first wave of the COVID-19 pandemic led to lockdowns that aimed to stop the spread of the virus. Most European countries that implemented lockdowns did so in the spring of 2020, although their intensity varied according to country. In all cases, confinement entailed an increase in the amount of domestic work that households had to perform due to a lack of outsourcing options, and an increase in time spent at home. In places where schools closed, childcare needs also increased. Given that the division of domestic and care work remains profoundly gendered (Altintas and Sullivan 2016), it is important to investigate the extent to which this increase in domestic work had an impact on existing inequalities. Confinement has also been interpreted as a "natural experiment" (Carlson et al. 2020) for analyzing the division of work, because partners are forced to spend more time at home and may need to renegotiate the division of domestic labor.
Different theories have been used to explain the persistence of the gendered division of work, but here we focus on time constraints and availability because time at home is one of the most salient changes resulting from confinement measures. From this perspective, the effects of lockdowns on the division of domestic work are theoretically undetermined; we have reason to expect that lockdowns lead to a more equal division of domestic work, but the opposite may also be true.
First, lockdowns spur changes in time-use patterns as a consequence of the cancellation of school, work, leisure, and social activities and a reduction in associated commuting times. Because mobility is restricted, most of this newly available time is spent at home. The time availability approach would lead us to expect that men and women would use part of this time to absorb the increase in domestic work in confinement and that the within-couple effects in terms of gender equality would differ according to which partner has more time. Spending more time at home also leads to more exposure to domestic work, which can lead to higher investments (Wray 2020). Therefore, we could hypothesize that, during the lockdown, we will find a more egalitarian division of work in households in which men have more time than women and a more traditional division of work in households in which women have more time than men (Hypothesis 1a).
However, this hypothesis relies on the assumption that the use of time is genderneutral; this is not necessarily the case. Research on unemployment during the 2007 financial crisis has shown that women's time seems more elastic than men's time with regard to domestic work: unemployed women performed more domestic work than unemployed men (Aguiar et al. 2013;Berik and Kongar 2013). The gender deviance neutralization hypothesis (Brines 1994) points at a different gendered effect: women compensating for men's unemployment by performing more domestic work, which both undermines women's status and asserts men's identities as breadwinners. However, gender deviance may not apply in the case of a pandemic, in which labor market outcomes are more clearly associated with a temporary and external cause than in other periods. In any case, the previous evidence on unemployment indicates the gendered effects of time availability and leads us to expect that the effect of greater time availability among men may be canceled out when women use more of their time to perform domestic work; thus, men's higher relative time availability may not be associated with a more egalitarian division of domestic work (Hypothesis 1b).
Using greater time availability to spend more time at home is undoubtedly a key explanatory factor, but we need to take into account the different time constraints related to paid work. In particular, unemployment and working from home are not equivalent situations. In principle, unemployment, as well as not working during this period, liberates more time than working from home. As mentioned before, existing research on unemployment has shown that unemployed women spend more time performing domestic work than unemployed men (Aguiar et al. 2013;Berik and Kongar 2013). The effect of unemployment during lockdown may differ from other periods, because unemployment is temporary in many cases (e.g., the company needs to close during the lockdown, but workers are expected to regain their jobs after it ends) and individuals do not spend time looking for another job. If unemployment during lockdown is permanent, then the time spent looking for a job or pursuing an education will most likely be lower than before the pandemic, given that most economic activities have ceased.
The amount of time liberated by working from home is much lower and mostly related to commuting time. Indeed, working from home may lead to longer working hours due to, for instance, interruptions, the necessity of addressing technical issues, and a lack of adequate work conditions. Previous literature has interrogated the effects of working from home and work flexibility on gender inequalities at home, although much of it focuses on the United States and uses qualitative methods that make the generalizability of the results problematic (Chung and van der Lippe 2020). Many of these studies focused on work-family conflict, and their results were inconclusive: some found that working from home decreased work-family conflict, and others found the opposite effect. Clark (2000) hypothesized that working from home may have different effects on men and women. She stated that, when working from home, the border between paid and unpaid work becomes flexible, which may lead to the expansion of one domain and the contraction of the other. For those who prioritize paid work, working from home may lead to longer working hours and less domestic work, whereas the opposite effect would be observed for those who prioritize the domestic sphere. Prioritizing one sphere or the other is not necessarily a personal choice but rather the result of other structural factors and gender norms. Some qualitative studies have shown that women are expected by others (e.g., family members or employers) to perform more domestic work when working from home (Sullivan and Lewis 2001;Hilbrecht et al. 2008). The effect of working from home on conflict may also depend on gender values and on the differences between expectations and outcomes (Bornatici and Heers 2020).
It is important to note that the majority of workers in the aforementioned studies chose to work from home; therefore, it is difficult to transpose their results to the lockdown period, when many workers have been "forced" to take up telework. In the case of the US, Carlson et al. (2021) have shown that fathers who work from home do more domestic and care work than fathers who do not, but the relationship between working from home and domestic work performance depended also on fathers' motivations: those who worked from home for personal reasons did more than those who stayed home for job mandated reasons. Their partners' employments were also crucial, because men who worked from home did more if their partners worked full-time. Lockdowns provide us with a type of natural experiment in which workers who would otherwise not work from home find themselves doing so.
Taking previous results into account, we would expect that working from home has a smaller impact on the division of work than not working (Hypothesis 2). Research on the division of domestic work during lockdown has provided mixed evidence on the role of time constraints and availability on men and women's performance of domestic work. Some surveys found that men have increased their time spent in domestic workespecially childcare-but that this increase has not compensated for women's higher share of domestic work (Andrew et al. 2020;Biroli et al. 2020;Carlson et al. 2020;Craig and Churchill 2020;Farré et al. 2020;Fodor et al. 2020;Hank and Steinbach 2020;Hipp and Bünning 2020;Kreyenfeld et al. 2020;Safi et al. 2020;Seiz 2020;Sevilla and Smith 2020;Zhou et al. 2020). Because of the exceptional circumstances, some of this research was conducted using convenience samples from internet surveys; therefore, the results need to be interpreted with care. In addition, most studies focused on childcare, which may follow a different logic than domestic work (Sullivan 2013). However, the few studies that used representative sampling and focused on domestic work came to a similar conclusion: men and women increased their time spent in domestic work during the pandemic, but this did not automatically increase the equality of the division of labor (Hank and Steinbach 2020;Zhou et al. 2020;Farré et al. 2020;Recchi et al. 2020a).
Most of these studies analyzed men's and women's time separately and could not provide direct conclusions about the division of work within couples. Only a few studies directly examined the division of domestic work within couples during the lockdown, which required having data on both partners. Research on the United Kingdom (Carlson et al. 2020) and Germany (Hank and Steinbach 2020) compared the division of domestic work before and during the lockdown and yielded mixed results about partners' time availability. Carlson et al. (2020) reported that higher time availability due to unemployment, reduced working hours, or working from home was associated with a more egalitarian division of domestic work. Hank and Steinbach (2020) used a representative sample and found mixed results regarding time constraints: women's share of housework decreased if their partner reduced their working hours, but women who worked from home increased their share of domestic work. Hipp and Bünning (2020) analyzed the lockdown period in Germany and found that available time was associated with more time invested in domestic work by both men and women, without any significant gender differences; thus, a more egalitarian division of work was found in couples in which men had more time. By contrast, Safi et al. (2020) found that the division of domestic work in France was more egalitarian in households in which women worked from home, although their analysis was restricted to couples in which both members were employed before the lockdown. Seiz (2020) analyzed a convenience sample of highly qualified Spanish women and found that non-traditional divisions of work were more frequent among couples in which both partners worked from home.
Overall, evidence from previous studies is therefore mixed. However, by following the time availability approach, we will investigate the following hypotheses: Hypothesis 1a. We expect to find a more egalitarian division of domestic work in households in which men have more available time than women, and vice versa.
Hypothesis 1b. This association between time and domestic work may not hold if men and women use their time very differently or if women engage in compensatory behaviors.

Hypothesis 2.
Regarding specific work arrangements, we expect to find a more egalitarian division of domestic work when men are unemployed than when they are working from home.
Our analysis contributes to the literature by focusing on the division of domestic work within couples and taking into account different employment configurations; in addition, we used a dataset that is representative of the French population.

The Lockdown in France
France is a continental welfare state, with relatively well-developed childcare policies, and a country where gender equality at work has recently become a policy goal (Anxo et al. 2017). As a reaction to the spread of COVID-19, schools were closed on 15 March 2020, and the general lockdown was established on 17 March. During the lockdown, only services that were considered essential were allowed to open, and telework was presented by the government as compulsory unless strictly necessary. Schools were closed, and French residents were only authorized to leave their homes to exercise (for an hour at a time and while staying close to home), shop for essentials, commute to work, or for care-related reasons. As in other countries, the lockdown entailed an increase in domestic and care work in all households.
The lockdown also had important effects on the labor market. Many companies had to cease operations because their economic activity was not deemed to be "essential". The French government implemented a plan that provided access to unemployment benefits for employees who were unable to work during the lockdown (chômage partiel). According to the Labor Force Survey (Institut National de la Statistique et des Etudes Economiques 2020), at the end of March 2020, 17 percent of employees were in this situation; a similar percentage were not working for other reasons (e.g., vacations and other types of leave). Approximately 37 percent of employees reported going to work as usual, and another third reported working from home at least part of the time. The survey also revealed that job searches were less active during the lockdown, signaling the particularities of the unemployment experience over this period. The ability to work from home was not equally distributed within the population: managers, professionals, and mid-level employees had more opportunities to work from home during the lockdown than others (Safi et al. 2020).

Data and Variables
We used data from ELIPSS, 1 a longitudinal panel that is representative of the French population. ELIPSS panelists are randomly selected and answer a monthly online survey on different topics. There is only one respondent per household. During the lockdown period, a research project (CoCo 2 ) was funded by the French research agency (ANR, Appel Flash Covid 19) to use the ELIPSS panel to research different sociological dimensions of the COVID-19 pandemic. During the third wave of this project, which was conducted in May 2020, the survey included questions on domestic work (Recchi et al. 2020b). Out of the 1404 panelists invited to respond to this particular survey, 1076 completed it, which represents a response rate of 76.4 percent.
The survey included information on the sociodemographic characteristics of respondents (e.g., sex, age, level of education, and marital status) and their household (e.g., income and number of children and adults). Because we are interested in the gendered division of domestic work, we only analyzed the responses of participants who were in a cohabiting or married relationship. The survey did not collect data on the sex of the partner; therefore, we imputed it by assuming that all couples were heterosexual. This is, of course, a biased assumption, but the most recent data from the Institut National de la Statistique et des Etudes Économiques indicates that same-sex couples amount to 0.9 percent of all French couples, so we do not expect our assumption to significantly bias the results. We also only selected couples in which at least one of partners was economically active. After cleaning the data, we had a sample of 406 households.
The CoCo survey gathered information on the domestic division of work between partners during the third wave. Respondents were asked who had been responsible for specific domestic tasks in their household since the beginning of the lockdown (i.e., the respondent, their partner, or someone else) and how often they did these tasks (i.e., always, often, occasionally, or never). The same questions were asked about cooking, shopping, doing laundry, ironing, cleaning, and performing repairs or gardening. Because these questions provided information on both the respondent and their partner, we used them to construct our dependent variable: the distribution of domestic work. We began by attributing a score to all possible responses for each task (3 if the person was always responsible for the task, 2 if they were often responsible for the task, 1 if they were sometimes responsible for the task, and 0 if they were never responsible for the task) for both the respondent and their partner. Then, we added up the values for the six tasks and created one variable for the respondent and one for their partner. To determine who performed more domestic work, we subtracted the total value for the woman from the total value for the man. The value of the final variable ranged from −18 (the most traditional outcome, in which the woman is responsible for all domestic tasks) and 18 (the reverse outcome, in which the man is responsible for all domestic tasks). Negative values indicate that the woman performs more domestic work than the man, while positive values indicate that the man performs more domestic work than the woman; a value of 0 would mean that both partners perform the same amount of work. Research on the division of work has tended to use percentages rather than scores for this type of question (Mencarini and Sironi 2012), but we prefer to use scores because the questions were not framed in terms of proportion but frequency. We tried an alternative specification that used proportion instead and also calculated the totals without taking into account ironing (since fewer households reported engaging in this task) or repairs (in order to obtain a measure of the more routine tasks), but the results did not change (results available from the authors).
Our main independent variable of interest was the paid work situation of each partner, particularly if one of them was unemployed or working from home. To construct this variable, we drew on data from different questions in the survey. First, respondents were asked about their employment situation (i.e., whether they were working, on leave, unemployed, unemployed as a consequence of COVID-19, or in a different situation). Second, employed respondents were asked about working from home over the two preceding weeks. Respondents could indicate that they were working as they did before the lockdown, that they alternated between working as usual and working from home, or that they always worked from home. The framing of this question was a bit problematic, because respondents who had worked from home before the lockdown were prompted to choose the first response (working as usual), which had to be taken into account when interpreting the results. The same two questions about employment situation and working from home were asked about the respondent's partner. We were interested in time availability and time spent at home, so we grouped respondents who were not working (i.e., unemployed, inactive, or on leave) into one category and respondents who worked from home (i.e., all the time or only on some days) into another, because the number of cases in these categories was small. It would have been extremely interesting to examine respondents' employment situation in greater detail, but the size of the sample did not allow for such an analysis. We combined the three possible situations (i.e., not working, working from home, and working as usual) for both partners into one variable with nine possible outcomes. In the first configuration, both partners had a similar amount of available time (i.e., both did not work, worked from home, or worked as usual). In the second configuration, the man had more available time than the woman (i.e., he did not work and she worked as usual, he did not work and she worked from home, or he worked from home and she worked as usual). In the last configuration, the woman had more available time than the man (i.e., he worked as usual and she did not work, he worked as usual and she worked from home, or he worked from home and she did not work). Table 1 presents the distribution for the sample that we analyzed, which only included respondents who cohabitated with a partner, married or not. The sample was balanced in terms of the sex of the respondent, as half were women and half were men. We observe a difference between men and women in terms of educational attainment, with more female respondents having some college education than male respondents. A bit more than half of the sample (56 percent) did not have any children living with them. Among those who did, a small number of households included children under 11 (117 households), which made the sample difficult to use for an analysis of childcare-related domestic work. The division of domestic work was relatively traditional, the average value is negative. In terms of the occupational structure of the households, the table shows that it was very diverse, although the most frequent situation reported during the lockdown was that both partners were not working. This category encompasses different situations, including people who were not economically active or unemployed before the lockdown but also those who were forced to stop working during the lockdown due to the characteristics of their jobs. By contrast, the number of households in which both partners continued to work as usual represented almost 12 percent of the sample. In 46 percent of households, at least one partner kept working as usual; this situation was more frequent among men (31.8 percent) than women (26.4 percent). We found a similar share of households in which at least one partner worked from home at least part of the time (48 percent). Within these couples, telework seemed to be slightly more common among women (33.2 percent) than men (27.8 percent). Finally, in 63.3 percent of households, at least one partner was not working; an equal number of men and women were in this situation (40 percent).

Method
We used OLS regression to analyze the association between different paid work arrangements and the division of domestic work. The dependent variable was the division of domestic work within the couple, and our main independent variable of interest was paid work arrangement in the household during the lockdown. Positive coefficients indicated a more egalitarian division of domestic work, while negative coefficients indicated a less egalitarian division. We also tried alternative specifications by running logistic regression analysis on the probability of having a traditional or non-traditional division of work and using slightly different cutoffs for the definition of traditional and egalitarian arrangements. The results were similar (not reported in this article, available from the authors). Thus, we chose to follow the OLS approach for simplicity.
We ran a first model while controlling only for the employment configuration of the household and a second model with other controls. Given the relatively small size of the sample, we include few controls. First, we controlled for the number of children in the household (i.e., no children, one child, two children, and three or more children). To control for the presence of small children, we also included a binary variable that indicated whether there was at least one child younger than 11 living in the household. Second, we controlled for the educational attainment for both partners in the form of a binary variable that indicated whether or not the respondent and their partner had a college education. Third, the age of the respondent was a categorical variable that was classified according to 10-year intervals. However, there were only three cases in the youngest category (18-24), which we grouped into the 25-34 category. We also ran models that included household income, but the results were not significant.

Results
We calculated the average distribution of domestic work in the surveyed households (the mean value of the dependent variable) for each employment configuration. The results are presented in Figure 1. We ordered the employment configurations according to relative time availability. It is important to note that, in all cases, the mean values were negative; therefore, women performed more domestic work than men in all cases. In other words, none of the configurations were egalitarian, but some were less egalitarian than others. The figure shows that, on average, less traditional divisions of domestic work were found when men have more available time than women. The most egalitarian (or least traditional) division of labor was found when the man did not work and the woman worked from home. There was, however, another configuration that was associated with a more egalitarian division of labor: when both partners worked from home. cated a more egalitarian division of domestic work, while negative coefficients indicated a less egalitarian division. We also tried alternative specifications by running logistic regression analysis on the probability of having a traditional or non-traditional division of work and using slightly different cutoffs for the definition of traditional and egalitarian arrangements. The results were similar (not reported in this article, available from the authors). Thus, we chose to follow the OLS approach for simplicity.
We ran a first model while controlling only for the employment configuration of the household and a second model with other controls. Given the relatively small size of the sample, we include few controls. First, we controlled for the number of children in the household (i.e., no children, one child, two children, and three or more children). To control for the presence of small children, we also included a binary variable that indicated whether there was at least one child younger than 11 living in the household. Second, we controlled for the educational attainment for both partners in the form of a binary variable that indicated whether or not the respondent and their partner had a college education. Third, the age of the respondent was a categorical variable that was classified according to 10-year intervals. However, there were only three cases in the youngest category (18-24), which we grouped into the 25-34 category. We also ran models that included household income, but the results were not significant.

Results
We calculated the average distribution of domestic work in the surveyed households (the mean value of the dependent variable) for each employment configuration. The results are presented in Figure 1. We ordered the employment configurations according to relative time availability. It is important to note that, in all cases, the mean values were negative; therefore, women performed more domestic work than men in all cases. In other words, none of the configurations were egalitarian, but some were less egalitarian than others. The figure shows that, on average, less traditional divisions of domestic work were found when men have more available time than women. The most egalitarian (or least traditional) division of labor was found when the man did not work and the woman worked from home. There was, however, another configuration that was associated with a more egalitarian division of labor: when both partners worked from home. Conversely, we observed a less egalitarian division of domestic work when the woman had more time than her partner. When partners had similar amounts of available time, the division of labor was heterogeneous. In this study, the most traditional division Conversely, we observed a less egalitarian division of domestic work when the woman had more time than her partner. When partners had similar amounts of available time, the division of labor was heterogeneous. In this study, the most traditional division of work was found when both partners worked as usual; at the same time, a relatively egalitarian division of work was observed when both partners did not work but especially when both teleworked. The four relatively egalitarian categories had two traits in common: the man had more time or a similar amount of time as his partner-either because he did not work or he worked from home-and the woman was active. According to this descriptive evidence, it is not enough for the man to stay at home; it would seem that the woman must also be present-while performing paid work-in order for the most egalitarian configuration to manifest itself. Figure 1 provides interesting descriptive evidence of the gendered association between time and presence at home and the performance of domestic work. However, to control for other intervening variables, we ran an OLS regression on the division of domestic work. The results are presented in Table 2. Model 1 only includes employment configurations, while Model 2 also includes control variables. With regard to the employment configurations, the reference category was the case in which both partners worked as usual, because it was the configuration with the most traditional division of work, according to Figure 1. In line with the descriptive data presented in Figure 1, the coefficients from the regression in Model 1 showed a positive association, which represents a more egalitarian division of domestic work in the three employment configurations in which the man has more time than his partner (Hypothesis 1a). The coefficients were positive and significant for all three configurations, but they were higher and more significant when the man did not work. This provides evidence in favor of relative time availability, but conclusions should be nuanced when observing the differences between the three broad types of relative time availability and Hypothesis 2. For instance, we found a significantly higher coefficient when the man did not work and the woman worked from home rather than onsite. From a time availability perspective, we would expect a stronger association between women who work onsite and men's involvement in domestic work. A tentative explanation could be that, when women work from home, their paid work may become more visible for their partners, who are also at home, and men respond to this visibility by performing more housework. When women work from home and their partners are also present, this also increases exposure to domestic work and opportunities to discuss what needs to be done and to eventually manage it.
When both partners have a similar amount of available time, either because both work from home or because they are not working, the division of domestic work is more egalitarian than in the reference category. Thus, it would seem that the employment configuration in which both partners worked onsite was the exception in terms of time availability. However, this finding should be interpreted with caution due to the nature of the data collection period. Couples in which both partners worked onsite during the lockdown likely consisted of essential workers who worked in healthcare or other sectors. Employment patterns may be very different depending on an individual's job during this period; therefore, it is difficult to draw conclusions about the couples in the survey, who held very specific jobs during an extraordinary period. Again, when we added the controls in Model 2, the results for this employment configuration did not change.
Finally, when women had more time than men, the division of work was not significantly different than in the reference category, which was indeed the least egalitarian. This is coherent with a time availability perspective and Hypothesis 1a. There was one exception: the configuration in which the man worked from home and the woman did not work. In this case, the division of work was more egalitarian than in the reference category. The coefficient stayed positive in Model 2, although it lost some significance, which indicates that the effect was partly explained by household composition and sociodemographic characteristics. We can only offer a tentative explanation for this effect; it may be that women take up more care work when they are not employed, although the effect persists when we control for the presence of young children in the household. There were only 39 cases of this configuration, which also needs to be taken into account for the interpretation.
The effects of the control variables show that the number of children in the household was associated with a less egalitarian division of domestic work, although the effect was only significant for households with two or more children. The age of the youngest child did not seem to have a significant impact, nor did the age of respondents. Finally, educational attainment had a negative coefficient, but only where men were concerned. This suggests that men with high educational attainment have greater bargaining power to negotiate domestic work; women do not appear to have the same power, as educational attainment was not significant in their case. The coefficient could also be related to the specific jobs that highly educated men perform, and this merits further research.

Discussion
The current paper investigates the division of domestic work among French couples during the first lockdown in 2020. We focused on the effect that partners' relative time availability from work and presence at home due to unemployment or telework had on the division of routine domestic tasks. We expected to find a positive association between men's relative time availability from paid work and a more egalitarian division of work (Hypothesis 1a), although the gender structure may cancel the effect of time availability (Hypothesis 1b). We also explored the effects of working from home or not working on the division of domestic labor during this period and expected telework to have smaller effects than not working (Hypothesis 2).
It is important to note that women performed more domestic work than men in all employment configurations; gender norms significantly mattered, and inequality was the rule. However, with that in mind, our results showed that men's time availability from paid work and their presence at home during the lockdown was positively associated with a more egalitarian division of domestic tasks (Hypothesis 1a). We found a more egalitarian division of domestic work in households in which men either did not work or worked from home. This was also the case when men and women had similar time constraints, except when both worked onsite. The latter finding should be interpreted with caution, given the exceptionality and employment characteristics of workers who continued to work onsite during the lockdown period. Our results add to those reported by Safi et al. (2020), and are in line with those of Carlson et al. (2020Carlson et al. ( , 2021 and Hipp and Bünning (2020): time availability is a key factor for understanding domestic divisions of work during the lockdown, although gender inequalities remained the rule.
We did not find evidence to suggest gender deviance neutralization or compensatory behaviors in terms of gender identities (Hypothesis 1b), but the lockdown period may not be the best context for testing this, at least in France. Different unemployment situations (e.g., long-term unemployment, COVID-related unemployment, and state-supported unemployment) coexist, which makes comparisons problematic.
However, results were not always as expected with regard to the differences between not working and working from home (Hypothesis 2). In some configurations, households in which men worked from home had a more egalitarian arrangement than households in which men did not work. It is possible that being at home-not merely having more time-is a factor for men. Spending significant amounts of time at home, even if part of that time is invested in performing paid work, may increase the visibility of domestic chores and exposure (Wray 2020). Telework may create more opportunities to discuss what needs to be done, how to manage the partner's domestic work (or to be managed), and perform tasks together, which lead to more egalitarian outcomes. Men may feel also more compelled to perform more tasks if their participation becomes more visible because they are at home, especially if their partners are also present. When the woman also works, her lack of time may be more apparent for the man if he is staying at home too.
According to our results, time constraints related to paid work but also the presence of men at home seem to be important factors in understanding the division of domestic work. An analysis of the lockdown period also allowed us to examine the effect of telework among men, and we found that this employment configuration was associated with a more egalitarian division of work. The line between paid and unpaid did not expand, as Clark (2000) suggested, for men who worked from home during the lockdown in France. Workers did not choose to work from home; thus, we think that this effect was not attributable to selection issues but points to an interesting variable that could contribute to increasing men's contribution to housework. These results point at the interest of considering both partners' employment status in analyses of the division of domestic work, because the effect of men's presence at home may depend on that of the partner, as shown by Carlson et al. (2021).
However, our study has some important limitations. The operationalization of the main independent variable was only a proxy for actual time constraints or availability, because we used the type of employment to infer time constraints. Working from home or working onsite may imply different situations that have an impact on time use; for instance, some individuals who work from home may choose to work partly during the night and be less available during the day, and we could not control for the heterogeneity of these situations. Differences may also be gendered, as women may work from home while simultaneously taking care of children or be interrupted by other family members more frequently than men (Andrew et al. 2020), or might experience more conflict because of the situation (Bornatici and Heers 2020). Taking all of these factors into account would undoubtedly enrich our knowledge of the division of work during this very particular period.
Funding: This research received no external funding.

Data Availability Statement:
The dataset used in this article is available at the Quetelet Progedo Platform, the French Data Archives for social sciences (http://quetelet.progedo.fr/). The syntax used to analyze the data is available from the author upon request.

Conflicts of Interest:
The author declares no conflict of interest.