Care More,Earn Less? The Association between Taking Paid Leave to Care for Sick Children and Wages among Swedish Parents

Abstract

Wages are related to parenthood and to child-related absences from work. The link between leave to care for sick children (CSC) and wages is understudied, however. CSC may negatively influence human capital and work capacity, and send the employer signals about work commitment. The short spells of CSC make this form of leave particularly suitable for testing the signalling theory. This study analysed data from Swedish population registers and showed that CSC use was associated with lower wages, particularly among men, up to 13 years after the birth of the first child. The association was strongest at high wage levels. Self-selection of parents with certain unmeasured characteristics into (high) CSC use was one, but not the only, explanation. The results support the idea that child-related time off negatively influences wages through a signalling effect. In addition, human capital or work capacity may suffer with frequent CSC use.

Keywords

gender equality labour market leave to care for sick children sick leave temporary parental leave wages

Introduction

Leave to care for sick children (CSC) in Sweden is part of an extensive system of welfare measures to facilitate work–family reconciliation. Despite this policy aim, CSC use and its consequences are seldom studied. The first aim of this study is empirical; it provides the first analysis of the association between CSC and wages with nationally representative Swedish data (cf. Evertsson, 2016; Stafford and Sundström, 1996). CSC allows parents to stay home from work to care for sick children without losing a significant amount of their income during the leave. However, absence from work has negative career and income consequences in the long run. CSC involves unpredictable spells of leave during a number of years. Although each spell usually lasts only one or a few days, recurrent leave-taking affects workplace organization over a long period, and influences employers as well as employees. Consequently, CSC use may affect an employee’s wage over time. The link between CSC and wages reflects the link between everyday childcare responsibilities and wage development (cf. Eriksson and Nermo, 2010). It is also an important piece of the puzzle when we want to understand gender differences in labour market outcomes.

The second aim of this study is theoretical. The characteristics of CSC use provide an opportunity to explore the mechanisms behind the link between care responsibilities and labour market outcomes. Parents use less than 10 days of CSC a year on average (Swedish Social Insurance Agency, 2018) and this form of leave enables a strong test of the signalling theory. If these short job absences are negatively related to wages, it will support the idea that employers interpret absence as a signal of a lack of work commitment, which in turn has negative wage consequences.

The high female labour force participation and low degree of selection of women into the labour market in Sweden (Korpi et al., 2013) make it a suitable country for studying the link between care and labour market outcomes. Another benefit is men’s relatively large share of the CSC, which is almost 40% (Swedish Social Insurance Agency, 2018). Many women as well as men in Sweden thus benefit from the CSC policy.

Leave to care for sick children and parental leave in Sweden

CSC is part of the Swedish temporary parental leave scheme and comprises 120 days per child annually that parents can use to stay home from paid work to care for a sick child or to take a child to a health care provider. To be entitled to CSC, a parent has to be employed or receiving unemployment benefits, and cannot be receiving other forms of benefits. The child’s other parent has to be in paid work. The leave can be used until the child turns 12 years old. The reimbursement level was 77.6% during most of the studied period (1998–2012). Reimbursement is capped at a certain income level that is tied to the consumer price index. In 2017, men’s share of CSC was 38%, with the share increasing by a mere two percentage points since 2003 (Swedish Social Insurance Agency, 2018).

The temporary parental leave scheme also includes paternity leave (i.e. temporary leave in connection to a child’s birth or adoption), which comprises 10 days that the other parent, usually the father, can use to stay home at the same reimbursement level as CSC following the birth of a child.

When parents in this study had their first child in 1999, the main parental leave commonly used to care for a small baby full-time consisted of 450 days per child. Each parent was entitled to half of these days, but could transfer all but 30 days to the other parent. A second reserved month was introduced in 2002 and the leave was extended to 480 days. For most of the studied period, the reimbursement level has been the same as for CSC, except for 90 days that are reimbursed at a flat rate. In 2017, men used 28% of the parental leave (Swedish Social Insurance Agency, 2018).

Labour market consequences of child-related time off

Although there is ample evidence that parenthood hampers women’s labour market outcomes as compared to men’s (e.g. Budig and England, 2001; Cooke, 2014; Magnusson, 2010), the negative association between parental leave and wages tends to be stronger for men than for women (e.g. Albrecht et al., 2015, 1999; Evertsson, 2016; Stafford and Sundström, 1996). In addition, Evertsson (2016) finds a negative association between temporary parental leave (i.e. the sum of CSC and paternity leave) and wages. Studying data from a large Swedish company in the 1980s, Stafford and Sundström (1996) perform the only prior study that singles out CSC from other forms of child-related time off. They find that CSC use is particularly negatively associated with men’s wages, and that the link is stronger than with other forms of child-related time off. The current study builds and improves on the two previous Swedish studies in two ways. First, this study analyses CSC, while Evertsson (2016) analyses the sum of CSC and paternity leave. Second, the present study uses nationally representative data, in contrast to Stafford and Sundström (1996) who analyse data from a single company in the 1980s. The next section discusses how this study contributes to the theoretical understanding of the relationship between family responsibilities and wages. Specifically, the study tests the importance of signalling theory.

Signalling

Absence from work may be interpreted by employers as a signal of how dedicated employees are to their job and about commitments outside work. Employers may therefore invest less in employees with a high rate of absence (cf. Albrecht et al., 1999; Stafford and Sundström, 1996). Previous research suggests that signalling may help to explain a negative association between parental leave and wages (Albrecht et al., 1999; Evertsson, 2016; Stafford and Sundström, 1996). This research finds a stronger negative association between parental leave and wages for men than for women (Albrecht et al., 1999; Stafford and Sundström, 1996), and a negative association even for very short periods of leave among men (Evertsson, 2016). Most mothers in Sweden take a considerable amount of parental leave, and such lengthy leave is rare among men; thus, employers may expect women, not men, to be absent for a long period. Employer expectations may affect all women’s or mothers’ wages through statistical discrimination, whereas only the wages of men who take parental leave are affected through signalling.

In Sweden today, CSC may provide a stronger test of the signalling theory than parental leave does. Fathers in Sweden take more than three months of parental leave on average before the child’s second birthday (Duvander and Viklund, 2014). CSC normally involves significantly fewer days of leave, and employers could expect fathers to use few or no CSC days. Although frequent absences that interrupt work may negatively affect a parent’s opportunity to perform the job well, this should not happen for parents who only use a few leave days a year. If the small number of CSC days that parents usually use is negatively associated with wages, this speaks to the importance of signalling. Signalling is further supported if the association is stronger among men than among women. Specifically, the analyses test that:

Hypothesis 1a: The negative association between parents’ wage and use of CSC is stronger when parents use a small number of CSC days than when they use a larger number of CSC days.

Hypothesis 1b: The negative association between parents’ wage and use of CSC is stronger for men than for women.

An alternative interpretation of a negative association between CSC and wages is that parents sort themselves into a certain use of CSC and into certain jobs based on unmeasured, stable, individual characteristics. Parents who use a very small number of CSC days could, for example, be highly committed to their job and earn higher wages than parents who use more CSC. This would produce a negative association at low CSC usage rates but no substantial additional wage loss at higher usage rates. If there is such self-selection of parents with certain stable characteristics into low CSC usage rates and high wages, any association between CSC and wages should be weakened with individual fixed effects regression (FE) that takes into account unmeasured, time-constant heterogeneity.

The alternative story: The costs of staying at home

Division of CSC has been shown to reflect division of housework in Swedish households (Eriksson and Nermo, 2010). CSC, or housework that CSC reflects, may negatively influence career and wages through mechanisms that, unlike signalling, should mostly affect wages at a high workload (i.e. at high rates of CSC or long housework hours).

One of these mechanisms is decreased work capacity. Becker (1985) suggests that people who have numerous family responsibilities spend much of their energy at home and have less energy for paid work. The literature on work–family conflict shows how family responsibilities can interfere with labour market work (see Byron, 2005, for an overview) and a high workload from housework has been shown to be negatively associated with wages (e.g. Bryan and Sevilla-Sanz, 2011; Carlson and Lynch, 2017). Hence, if CSC or housework that CSC reflects influences parents’ work capacity, this is likely to happen only with frequent absences or a high workload at home. The association should be similar for men and women.

Another possible mechanism behind an association between CSC and wages is negative effects of CSC on human capital. According to human capital theory, knowledge gained from education and work may deteriorate during periods of absence from the labour market, due to a lack of practice. The employee may also miss on-the-job training and the implementation of new work practices or technologies (Mincer and Polachek, 1974). For example, studies of parental leave suggest that longer periods of parental leave may hamper wage development by influencing human capital (e.g. Albrecht et al., 1999; Stafford and Sundström, 1996). CSC generally does not involve the long periods of absence that parental leave involves. Recurrent CSC absences over a number of years may instead lead to a gradual deterioration or slower growth of human capital. This negative effect may be primarily noticeable among parents who use numerous CSC days and should be the same for men and women. Consequently, if a negative association between CSC and wages is present mainly among parents who use numerous CSC days, and if the association is similar for men and women, the results support the idea that family responsibilities negatively influence wages through either decreased work capacity or effects on human capital. The analyses test that:

Hypothesis 2a: The negative association between parents’ wage and use of CSC is stronger when parents use a large number of CSC days than when they use a smaller number of CSC days.

Hypothesis 2b: The negative association between parents’ wage and use of CSC does not vary by the parent’s gender.

Alternatively, a negative association between CSC and wages that is particularly strong at high CSC usage rates could be caused by selection; parents with low work commitment or other unmeasured, time-invariant characteristics may be inclined to use more CSC and they may be likely to earn lower wages irrespective of CSC use. In this case, FE regression models would show a weak or no association between CSC and wages.

Socio-economic position

Signalling, work capacity and human capital effects of CSC on wages may increase with wage level as expectations of commitment and availability are likely to be higher in high socio-economic positions. Furthermore, gendered expectations and the inherent masculinity of the unencumbered worker ideal mean that expectations may be particularly high on men in these positions (e.g. Acker, 1990; Cooke and Hook, 2018; Williams, 2000). In addition, parents with low-paying jobs or short working hours may be more willing to forgo paid work than parents with well-paying, high-status jobs because the former will lose less money, work time and/or human capital by staying at home. Consequently, using CSC may cost employees, and may especially cost male employees in high-paying jobs. In line with this argument, the negative association between parental leave and wages is stronger among higher-educated parents than among lower-educated parents in Sweden (Evertsson, 2016). Analysing US data, Cooke and Hook (2018) find support for the work capacity argument (see above) primarily among men with the highest wages. In the present study, socio-economic differences are tested for CSC with quantile regressions where the association between CSC and wages is estimated separately for different parts of the wage distribution. The analyses test that:

Hypothesis 3a: The higher the wage, the more negative the association between parents’ wage and use of CSC.

Hypothesis 3b: The negative association between parents’ wage and use of CSC is the strongest among high-earning men.

A negative association between CSC and wages may exist due to occupational differences that influence parents’ willingness and need to use CSC. For example, well-paying jobs may be flexible, making it easier for well-paid parents to care for a sick child without using formal leave. A control for occupation is therefore added to all analyses.

Analytical strategy

The association between CSC use and wages among parents who had their first child in 1999 was studied using population registers (see more on data below). Small changes in CSC between two years are unlikely to influence wages, so accumulated leave over a number of years was studied. A continuous measure of CSC was used with a quadratic term to test whether the association was particularly strong at low or high usage rates. Because parents use the highest number of leave days for 2–4-year-olds (Swedish Social Insurance Agency, 2007), CSC used during the five years after the birth of the first child was studied. In a long-term analysis, the CSC–wage association was analysed 13 years after the child’s birth (i.e. after the 12-year period during which parents could take leave to care for their first-born child). The models included controls for number and ages of children in the household, as these factors influence the number of CSC days used (Swedish Social Insurance Agency, 2007) and is associated with wages (e.g. Grönlund et al., 2017). The models also included standard controls in wage regressions that may change over time: education, sector, and (as measures of labour market absence other than CSC) parental leave, unemployment and sick leave.¹

First, Hypotheses 1a and 2a (i.e. whether there is a negative association between CSC use and wages and the functional form of the association) and Hypotheses 1b and 2b (whether the association differs between men and women) were tested with ordinary least squares (OLS) regression. Analyses were run separately for men and women and differences were tested with inclusion of an interaction between gender and CSC in analyses of the pooled sample (see the online Appendix, Part C). In a short-term analysis, the natural logarithm of wage in 2004 was regressed on number of CSC days accumulated in the period 1998–2003. Parental leave was controlled for and accumulated in the same way as CSC. The other control variables were measured in 2004. In a long-term analysis, log wage in 2012 and CSC days taken until 2011 were analysed.

Second, individual FE regression was applied to test whether some of the association found with OLS may be attributed to self-selection of individuals with certain, unmeasured time-invariant characteristics into different amounts of CSC. FE regression models such time-constant heterogeneity by exploiting the within-individual variation over time (Andreß et al., 2013). Each individual was represented by two person years, 1998 and 2004 or 2012. Hence, changes between two time points were studied. The same controls as in the OLS were included, adding a control for year. Notably, FE does not eliminate the influence of selection based on time-varying, unmeasured characteristics. Changes that occurred during parenthood may therefore influence wage and were not controlled for.

Each estimate in FE regression is based only on information from people who experienced a change in the variable in question. The transition matrix in the online Appendix (Table A1) shows that – because they became parents in 1999 – everyone in the sample experienced a change in number of children and age of the youngest child. A large majority experienced a change in CSC and parental leave. Smaller parts of the sample, 7.6–16.5%, experienced a change in number of days with sick leave benefits or unemployment benefits or in educational level or sector. The results for these variables in the FE should be interpreted with this in mind.

Third, quantile regression was applied to test Hypothesis 3a, whether the negative association between CSC use and wages is dependent on wage level, and 3b, whether the association is the strongest among high-earning men. Gender differences were tested with inclusion of an interaction between gender and CSC in analyses of the pooled sample (see the online Appendix, Part C). The conventional quantile regression method, conditional quantile regression, was used. This method models differences in the response to changes in the independent variables at different points in the conditional distribution of the dependent variable (Buchinsky, 1998). In contrast, the distribution of the outcome variable is not conditional on the control variables with unconditional quantile regression where the quantiles are defined pre-regression (Firpo et al., 2009; Killewald and Bearak, 2014). Unconditional quantile regression was applied as a robustness check (see the online Appendix, Part D). The 10th, 25th, 50th, 75th and 90th percentiles were estimated.

Standard errors were clustered on the family level in the OLS and on the individual level in the FE. The quantile regressions were estimated with robust standard errors.

Data

Annual data from Swedish population registers provided by Statistics Sweden were analysed. A group of parents was followed from 1998, the year before their first child was born, until 2012. All studied parents were employed during 1998 and in the years when wages were measured (2004 or 2012). The main data sources used were the Longitudinal integration database for health insurance and labour market studies (LISA) and the Structure of earnings survey (LSS). LISA covers the entire population under the age of 76. LSS covers the entire population under the age of 76 that works in the public sector or in private companies with at least 500 employees. Employees in private companies with fewer than 500 employees are randomly sampled each year. This generates a representative sample of employees in these smaller companies, but underrepresents them and makes them more difficult to follow over time. The data cover 5% of all private companies and 50% of all those employed in the private sector. The models required that information from LSS was available for each individual at two time points (1998 and 2004 or 2012). Of the initial sample with information available in 1998, 60% were followed up in 2004, and 64% in 2012. Of those working in the private sector in 1998, approximately 53% were followed up in 2004 or 2012. Thus, a smaller share of the privately employed than of the publicly employed persons was followed. Additional analyses of publicly and privately employed persons showed that, although the association between CSC and wages was somewhat smaller in the public sector, the results presented here were not driven by the publicly employed group (analysis available from the author).²

There may be a correlation between illnesses in children and in parents. The analyses therefore controlled for sick leave absence. Having a seriously ill child may hamper wage development by influencing work commitment or work capacity. Parents who received a care allowance were therefore excluded. This allowance reimburses anyone who needs to stay home for an extended period to care for a family member or relative and is an alternative to CSC for parents of children with serious illnesses or disabilities.

After the selection criteria were applied and individuals with missing information on any of the analysed variables were excluded, the sample consisted of 32,533 parents of which 15,789 were included in the analysis of wages in 2004 and 16,744 in the 2012 analysis. The sampling of employees in private companies with fewer than 500 employees in LSS means that the share of men was low: 43.8% in 2004 and 40.8% in 2012.

Variables

The dependent variable was full-time equivalent monthly wage, which included gross fixed wage, fixed wage supplements, variable wage, wage benefits and wage increments. The measure did not include compensation for working overtime and it was not affected by wage reductions due to CSC, parental leave, etc. The data provider has recalculated the income of employees who worked part-time or were paid by the hour to correspond to a full-time monthly wage using information on the employee’s standard full-time working hours and proportion of full-time worked.³ Wage was adjusted to the 2012 monetary value. The natural logarithm of full-time equivalent monthly wage was used in the regression analyses.

The main independent variable was the accumulated number of CSC days. Information was available on annual leave use. Accumulated CSC was calculated for the period 1998–2003 for the analysis of wages in 2004 and 1998–2011 for the 2012 analysis. CSC taken in 1999 and any subsequent years when children were born was excluded from the measure, because the data did not distinguish between CSC and paternity leave. The paternity leave seriously inflated fathers’ number of leave days during years when children were born. The measure underestimated the accumulated CSC for both parents, but this underestimation is likely to be smaller than the overestimation of fathers’ CSC that would have been introduced.⁴ A quadratic term for accumulated leave was included in the regression analyses. It was divided by 100 to facilitate interpretation (by reducing the number of decimal zeros).

Control variables were as follows:

Education: compulsory education ⩽ 9 years, upper secondary education, short tertiary education < 2 years and long tertiary education ⩾ 2 years.

Public sector versus private sector work.

Age of the youngest child and number of children in the household.

Number of accumulated parental leave days with economic compensation during 1998–2003 for the 2004 analysis and 1998–2011 for the 2012 analysis.

Number of days with unemployment benefits during a year.

Number of sick leave days covered by the sick leave insurance during a year. The employer pays for the first two weeks of absence, while the sick leave insurance pays for the third and subsequent weeks.

Occupation was coded according to SSYK 96 (the Swedish version of ISCO-88) on a one-digit level (1–9) and was included as a categorical variable.

The FE added a control for year (1998, 2004 or 2012).

Results

Descriptive statistics

The descriptive Table 1 shows that the gender difference in wages was less pronounced in 1998, when the women and men were childless, and particularly pronounced in 2012, after they had been parents for 13 years. By 2004, the women had used more CSC than the men, and this gender difference had increased considerably by 2012. A large gender difference was also observed for parental leave.⁵ The changes in average number and age of children by 2012 suggest that the parents had their children early in the period. The women were more highly educated than the men. In 2012, 35% of the men and 72% of the women worked in the public sector. The corresponding numbers in the population in 2013 were 18% of men and 46% of women (Statistics Sweden, 2014). The high proportions of publicly employed persons in the sample was due to the random sampling of employees in private companies with fewer than 500 employees in the LSS (see the description of the data above).

Table 1.

Descriptive statistics.

	2004				2012
	Men(n = 6913)		Women(n = 8876)		Men(n = 6830)		Women(n = 9914)
	Mean / %	SD	Mean / %	SD	Mean / %	SD	Mean / %	SD
Wage (median)^a	27.93		23.46		33.60		27.25
Wage 1993 (median)^a	22.56		19.31		22.40		19.19
CSC^b	11.68	16.04	16.87	17.62	31.41	33.87	54.37	44.08
Parental leave^b	86.12	87.31	549.35	173.46	170.38	136.10	726.77	246.42
Unemployment^c	2.21	18.74	3.55	21.21	1.15	12.25	1.52	15.69
Sick leave^c	3.29	22.20	6.42	30.42	2.68	18.77	5.69	27.88
Age of youngest child	3.16	1.45	3.09	1.39	10.01	2.61	9.88	2.55
No. of children	1.83	0.59	1.79	0.51	2.02	0.74	2.07	0.69
Education:
Compulsory	5.63		2.43		4.71		1.96
Upper secondary	45.44		40.29		42.84		34.34
Short tertiary	10.55		4.82		10.67		4.98
Long tertiary	38.39		52.46		41.77		58.73
Works in public sector	33.36		69.18		34.69		71.87

Notes: ^aIn 1000 SEK, adjusted to the 2012 monetary value; ^bin days accumulated during 1998–2003 and 1998–2011; ^cin days. CSC: care for sick children.

When their first child turned one year old in 2000, the women had used half a CSC day more leave than the men (Figure 1). CSC use then increased considerably more for women. When the child was 2–7 years old, the women averaged 6.5 days/year whereas the men averaged about 3.5 days/year. Annual CSC days then dropped, and the gender difference decreased. By the time the first child was 13 years old (in 2012), the men had used a total of 31 days on average and the women 54 days (Table 1).

Figure 1.

Annual number of CSC days for the 13 years following the first child’s birth in 1999.^a

Figure 1 shows a drop in the number of CSC days in 2005. This is due to a change in the method for data acquisition used by the Swedish Social Insurance Agency.

The association between CSC and wages

Table 2 displays the results of OLS and FE regressions of women’s and men’s log wages. Estimates for factors affecting labour market absence are shown. For full models, see the online Appendix (Part B).

Table 2.

Log wage in 2004 and 2012 regressed on accumulated CSC.

	Men				Women
	OLS		FE		OLS		FE
	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
2004
CSC^a	–0.0057***	–0.0039***	–0.0029***	–0.0026***	–0.0029***	–0.0022***	–0.0007***	–0.0007***
	(0.0004)	(0.0003)	(0.0003)	(0.0003)	(0.0003)	(0.0002)	(0.0002)	(0.0002)
CSC^2/100	0.0029***	0.0021***	0.0013***	0.0012***	0.0017***	0.0013***	0.0002	0.0002
	(0.0005)	(0.0004)	(0.0003)	(0.0002)	(0.0003)	(0.0002)	(0.0002)	(0.0002)
Parental leave^a	–0.0002***	–0.0002***	–0.0000	–0.0000	–0.0003***	–0.0002***	–0.0001***	–0.0001***
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Unemployment^b	–0.0013***	–0.0009***	–0.0005***	–0.0005***	–0.0008***	–0.0007***	–0.0002***	–0.0002***
	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)
Sick leave^b	–0.0005***	–0.0004***	–0.0002*	–0.0002*	–0.0003***	–0.0002***	–0.0001*	–0.0001*
	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0000)	(0.0000)
Constant	3.3262***	3.5498***	3.0699***	3.0835***	3.2846***	3.4410***	2.8926***	2.8989***
	(0.0206)	(0.0200)	(0.0112)	(0.0135)	(0.0185)	(0.0177)	(0.0108)	(0.0119)
Controlling for occupation	no	yes	no	yes	no	yes	no	yes
N persons	6913	6913	6913	6913	8876	8876	8876	8876
N person years			13826	13826			17752	17752
R²/R² within	0.3074	0.4541	0.5860	0.6010	0.3393	0.4655	0.6576	0.6658
2012
CSC^a	–0.0048***	–0.0034***	–0.0031***	–0.0028***	–0.0018***	–0.0013***	–0.0008***	–0.0006***
	(0.0002)	(0.0002)	(0.0002)	(0.0002)	(0.0001)	(0.0001)	(0.0001)	(0.0001)
CSC^2/100	0.0017***	0.0013***	0.0010***	0.0009***	0.0005***	0.0003***	0.0002***	0.0001***
	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0000)	(0.0000)	(0.0000)
Parental leave^a	–0.0001*	–0.0001*	0.0000	0.0000	–0.0004***	–0.0003***	–0.0002***	–0.0002***
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Unemployment^b	–0.0015***	–0.0013***	–0.0003*	–0.0003*	–0.0009***	–0.0007***	–0.0002**	–0.0002***
	(0.0002)	(0.0002)	(0.0001)	(0.0001)	(0.0002)	(0.0001)	(0.0001)	(0.0001)
Sick leave^b	–0.0007*	–0.0005	–0.0005*	–0.0005*	–0.0006**	–0.0005**	–0.0003*	–0.0003*
	(0.0003)	(0.0003)	(0.0003)	(0.0003)	(0.0002)	(0.0002)	(0.0002)	(0.0002)
Constant	3.4615***	3.6852***	3.1176***	3.1329***	3.5568***	3.7020***	2.9385***	2.9732***
	(0.0343)	(0.0315)	(0.0104)	(0.0127)	(0.0239)	(0.0224)	(0.0071)	(0.0097)
Controlling for occupation	no	yes	no	yes	no	yes	no	yes
N persons	6830	6830	6830	6830	9914	9914	9914	9914
N person years			13660	13660			19828	19828
R²/R² within	0.3265	0.4717	0.7365	0.7546	0.2888	0.4412	0.7861	0.8061

Notes: OLS and FE regression. Clustered standard errors in brackets. Variables not shown: age of youngest child, number of children, education, sector and, in the FE, year. ***p < 0.001, **p < 0.01, *p < 0.05. ^aIn days accumulated during 1998–2003 for the 2004 analysis and 1998–2011 for the 2012 analysis; ^bin days. CSC: care for sick children; FE: fixed effects; OLS: ordinary least squares.

Men’s wages in 2004 were negatively related to CSC, according to both the OLS and the FE (Table 2, upper panel). The positive quadratic term showed that the association was steepest at lower usage rates. The addition of a control for occupation in Model 2 reduced the association. Hence, occupational differences between men who used more or less CSC affected, but did not explain, the negative association between leave use and wages. Unmeasured individual characteristics were not the main cause of the association; the coefficient was smaller in the FE than in the OLS (compare Model 3 to Model 1, and Model 4 to Model 2), but was still large enough to be of some importance.

Figure 2 shows differences in wages, predicted at gender-specific averages of the control variables, for men and women with different amounts of CSC use, based on the OLS while controlling for occupation. The estimated wage of parents who used the gender-specific median number of accumulated CSC days was compared to the estimated wage of those whose CSC use corresponded to the gender-specific first quartile of leave, and of those whose CSC use corresponded to the third quartile of leave, respectively. A father who had used 1.5 days of leave during 1998–2003, the first quartile of fathers’ leave, had a 2.1% higher wage in 2004 than a father who had used seven days in total, the median among fathers. A father with leave use equivalent to the third quartile (15.5 days) had a 2.9% lower wage in 2004 than fathers who used the median amount of leave.

Figure 2.

Wage differences between parents by accumulated number of CSC days.^a

The negative association between women’s CSC use and wages in 2004 was significant but weaker and less curvilinear than the association for men (Table 2). Pooled OLS and FE regressions with an interaction between gender and CSC showed a significant gender difference (see the online Appendix, Table C1). Additional analyses where CSC was categorized showed a very steep negative association at the lowest levels of CSC use among men (< 6 days), and a much less steep association among women (analysis available from the author). The negative association between women’s CSC use and wages was somewhat reduced when controlling for occupation (Table 2, Model 2). The differences between the OLS and FE (compare Model 3 to Model 1, and Model 4 to Model 2), together with the small coefficient in the FE, suggest that self-selection of certain women into taking (numerous days of) leave was an important cause of the association. Figure 2 shows that women who used leave corresponding to the first quartile (five days) had a 1.4% higher wage than women who used the female median amount of leave (12 days). The corresponding difference between the third quartile (23 days) and the median is ‒1.9%.

The analyses for 2004 lent support to Hypothesis 1 and the signalling theory: the negative association between CSC use and wages was stronger at lower than at higher usage rates (Hypothesis 1a) and stronger for men than for women (Hypothesis 1b). The competing Hypotheses 2a and 2b, which suggest that CSC would be linked to wages mainly through effects on work capacity or human capital for parents who use numerous CSC days, was not supported. Work capacity or human capital can still be affected among parents who use numerous leave days, but signalling is likely to take place at lower usage rates.

In 2012, a negative association between wages and CSC existed among men and the association was considerably steeper at low than at high usage rates (Table 2, lower panel). Controlling for occupation reduced the association. The association was weaker in the FE than in the OLS, so selection was one, but not the only, explanation. The CSC coefficient in the OLS was slightly smaller than in the 2004 analysis, and the coefficient in the FE was equal to or a bit larger than that in 2004. As most CSC was taken early in the period (Figure 2), this indicated that the negative association between leave and wages did not greatly diminish during the studied time span. Among women, the negative association between CSC use and wages was weak in the OLS and even weaker in the FE. The gender difference was pronounced and significant (see the online Appendix, Table C2). Figure 2 shows that a man who had taken the equivalent of the first quartile of men’s CSC use (eight days) had a 4.4% higher wage in 2012 (estimated with OLS and controlling for occupation) than a man who had taken the median amount of leave among men (22 days). A man who had taken the third quartile of men’s leave (44 days), had a 5.5% lower wage. The corresponding wage differences among women were 2.2% (for 23 days compared to 44 days of leave) and ‒2.5% (for 73 days), respectively. Like the 2004 analysis, the 2012 analysis lent support to Hypotheses 1a and 1b and hence to signalling theory.

For the control variables, Table 2 shows the same results in 2004 and 2012: negative associations between wages and parental leave, unemployment and sick leave. The association between men’s parental leave and wages was, however, not significant in the FE (Models 3 and 4).

Although the negative association between CSC use and women’s wages was very weak in these analyses, the large gender difference in number of leave days may still contribute to the overall gender wage gap. In a pooled OLS regression, the gender wage gap, adjusted for all control variables, was reduced from 6.4% to 5.5% in 2004, and from 3.2% to 2.8% in 2012 when CSC was included (analysis available from the author). CSC is hence likely to contribute only slightly to the gender wage gap.

CSC and wages along the wage distribution

The negative association between men’s CSC use and wages was stronger in high-paying jobs than in low-paying jobs in 2004 (Table 3, upper panel). A similar pattern was found for women, though the negative association was weaker among women than among men at every point in the wage distribution. The gender difference was particularly large at high wage levels (significance test of the gender difference is shown in the online Appendix, Table C3). Table 3 (lower panel) shows similar results for 2012, though the negative association increased less with wage level for women. Thus, the gender difference was even larger at high wage levels in 2012 than in 2004. The results supported Hypothesis 3a, that the negative association between CSC use and wages increases with wage level, and Hypothesis 3b, that the association is particularly strong among high-earning men. This suggests that the signalling value and/or the costs of absence were particularly large in high socio-economic positions and especially among men in these positions.

Table 3.

Log wage in 2004 and 2012 regressed on accumulated CSC.

	Men					Women
	10th	25th	50th	75th	90th	10th	25th	50th	75th	90th
2004
CSC^a	–0.0013***	–0.0022***	–0.0031***	–0.0041***	–0.0057***	–0.0005**	–0.0007***	–0.0017***	–0.0028***	–0.0046***
	(0.0003)	(0.0002)	(0.0004)	(0.0003)	(0.0007)	(0.0002)	(0.0002)	(0.0002)	(0.0003)	(0.0006)
CSC^2/100	0.0005	0.0010***	0.0016**	0.0023***	0.0041***	0.0002	0.0002	0.0009***	0.0016***	0.0034***
	(0.0003)	(0.0001)	(0.0005)	(0.0002)	(0.0008)	(0.0002)	(0.0003)	(0.0002)	(0.0004)	(0.0009)
Parental leave^a	–0.0000	–0.0001*	–0.0001***	–0.0002***	–0.0002**	–0.0001***	–0.0001***	–0.0002***	–0.0002***	–0.0003***
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0001)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Unemployment^b	–0.0013***	–0.0009***	–0.0009***	–0.0006***	–0.0008***	–0.0010***	–0.0009***	–0.0008***	–0.0005***	–0.0004
	(0.0003)	(0.0003)	(0.0002)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0001)	(0.0004)
Sick leave^b	–0.0002	–0.0003***	–0.0004***	–0.0004**	–0.0004	–0.0001	–0.0002***	–0.0002**	–0.0002*	–0.0002*
	(0.0002)	(0.0000)	(0.0001)	(0.0001)	(0.0003)	(0.0001)	(0.0000)	(0.0001)	(0.0001)	(0.0001)
Constant	3.2456***	3.3458***	3.5193***	3.7093***	3.9064***	3.1181***	3.2404***	3.3710***	3.5810***	3.7756***
	(0.0190)	(0.0185)	(0.0232)	(0.0250)	(0.0365)	(0.0137)	(0.0136)	(0.0169)	(0.0220)	(0.0331)
N	6913	6913	6913	6913	6913	8876	8876	8876	8876	8876
2012
CSC^a	–0.0010***	–0.0016***	–0.0025***	–0.0040***	–0.0059***	0.0001	–0.0003***	–0.0008***	–0.0015***	–0.0025***
	(0.0001)	(0.0002)	(0.0002)	(0.0003)	(0.0002)	(0.0001)	(0.0001)	(0.0001)	(0.0002)	(0.0004)
CSC^2/100	0.0003***	0.0005***	0.0009***	0.0016***	0.0028***	–0.0001	0.0000	0.0002***	0.0004***	0.0008***
	(0.0001)	(0.0001)	(0.0001)	(0.0002)	(0.0001)	(0.0000)	(0.0000)	(0.0000)	(0.0001)	(0.0002)
Parental leave^a	–0.0000	–0.0000	–0.0001*	–0.0001*	–0.0001**	–0.0001***	–0.0001***	–0.0001***	–0.0003***	–0.0004***
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Unemployment^b	–0.0012***	–0.0011***	–0.0013***	–0.0011*	–0.0012**	–0.0010***	–0.0008***	–0.0006*	–0.0005**	–0.0003
	(0.0000)	(0.0002)	(0.0001)	(0.0006)	(0.0004)	(0.0001)	(0.0001)	(0.0002)	(0.0002)	(0.0003)
Sick leave^b	–0.0001	–0.0003***	–0.0002*	–0.0002	–0.0002	–0.0002	–0.0002***	–0.0002***	–0.0002***	–0.0003***
	(0.0001)	(0.0000)	(0.0001)	(0.0001)	(0.0009)	(0.0001)	(0.0000)	(0.0000)	(0.0001)	(0.0001)
Constant	3.3310***	3.4948***	3.6888***	3.8961***	4.0878***	3.3045***	3.4176***	3.5938***	3.8567***	4.1607***
	(0.0243)	(0.0265)	(0.0304)	(0.0282)	(0.0500)	(0.0149)	(0.0145)	(0.0145)	(0.0258)	(0.0426)
N	6830	6830	6830	6830	6830	9914	9914	9914	9914	9914

Notes: quantile regressions. Robust standard errors in brackets. Variables not shown: age of youngest child, number of children, education, sector, occupation. ***p < 0.001, **p < 0.01, *p < 0.05. ^aIn days accumulated during 1998–2003 for the 2004 analysis and 1998–2011 for the 2012 analysis; ^bin days. CSC: care for sick children.

Occupation was controlled for in these analyses because occupational differences between parents at the top and the bottom of the wage distribution could be an important explanation for a difference in the negative CSC–wage association between these groups. Additional analyses without control for occupation showed stronger associations than the results presented in Table 3, particularly at high wage levels (analyses available from the author). Occupational differences between parents at different wage levels may hence be important, but as shown here, the CSC–wage association differed between wage levels even when controlling for occupation.

As a robustness check, unconditional quantile regressions were run. The online Appendix (Part D) shows that the results were similar to those in Table 3, although the negative association between CSC and wages was stronger at the highest wage levels for men.

Summary and conclusions

Studies of CSC are rare, and this is especially the case for studies of the association between CSC use and wages. The present study contributes to this research by analysing the link between CSC and wages in Sweden. The results showed a negative association between CSC and wages that was stronger among men than among women. It was stronger at a smaller than a larger number of leave days, and was largest at high-wage levels. The association was found five and 13 years after the first child’s birth and, among men, it did not appear to attenuate with time. The negative association between CSC and women’s wages was weak and can be regarded as insignificant at lower wage levels. Because the negative CSC–wage association was weaker among women than among men, CSC contributed only slightly to the gender wage gap, even though women used considerably more CSC than men did. The study adds to the growing literature showing that there are negative associations between family responsibilities and labour market outcomes, but that these associations are not simple; they are not the same for mothers and fathers or for parents in different socio-economic positions or with different working conditions (e.g. Boye et al., 2017; Cooke, 2014; Evertsson, 2016; Magnusson and Nermo, 2017).

Negative effects of labour market absence on human capital and work capacity have been put forward as explanations for links between parental leave and wages (e.g. Albrecht et al., 1999; Stafford and Sundström, 1996). On the one hand, the stronger negative association found here between CSC use and wages at higher wage levels than at lower ones could be due to a larger negative effect of CSC on human capital or work capacity in high-wage positions. On the other hand, because CSC involves short spells of leave taken over a number of years, the leave should mainly influence human capital or work capacity at high levels of absence, which were observed mostly among women. This would also be expected if CSC served as a proxy for housework (cf. Eriksson and Nermo, 2010). Housework that demands a great deal of time and energy is what mainly ought to influence work capacity. The finding of a weak negative association between CSC and wages among women, and of a negative association already at low usage rates among men, does not support the idea that human capital depreciation or reduced work capacity are the main mechanisms driving negative associations between child-related time off from work and wages. They may play a role among parents who use numerous CSC days and whose work is therefore frequently interrupted by leave, but other explanations are needed for those who use only a few days of leave.

These results are in line with signalling theory. Employers may expect women, but not men, to take a great deal of leave and hence perceive unexpected male behaviour to be a signal of a lack of work commitment (e.g. Stafford and Sundström, 1996). The findings of a negative association between CSC and wages, even among parents who used a small number of CSC days, of a stronger association at lower than at higher usage rates, and of a stronger association among men than among women, support signalling as a mechanism. The stronger association at higher wage levels could reflect a stronger signalling value and higher expectations of work commitment and availability in high-paying jobs.

Selection of parents with certain characteristics into (more) CSC use is an alternative explanation. Parents who use a small number of CSC days may be highly committed to their job and earn high wages for that reason. Parents who use numerous CSC days may be less committed to their job than other parents, and earn lower wages. The results suggest that such stable individual characteristics contributed to, but were not the only cause of, the association between CSC and wages. Notably, the analyses did not control for individual characteristics that changed over time. A change in parents’ work commitment during the studied period is either an alternative or a complementary explanation for the results. Occupational differences between parents who took different amounts of leave are another alternative explanation. Because of the nature of well-paying jobs, parents in these jobs may be less inclined than low-paid parents to use formal CSC. The results suggest that the negative association between CSC use and wages was linked to, but not explained by, occupational differences. The analysis of the importance of occupation presented here was limited though, so this question warrants further study.

One of the more important limitations of this study was the underrepresentation of employees in private companies with fewer than 500 employees caused by sampling of these employees in a dataset that otherwise included the entire Swedish population up to age 75. Robustness checks suggested that the results were not driven by the pattern displayed by publicly employed persons. The results may nevertheless be more representative of employees in the public sector and in large, private companies.

In conclusion, childcare that impacts workplace performance or is noticeable in the workplace, such as CSC, is likely to have a signalling effect that negatively influences the wages of parents, and especially fathers, even at low levels of care. Employers’ expectations of men’s constant availability and presence seem to imply that the labour market penalty of childcare is largest for fathers in well-paying jobs. It is reasonable to assume that an equal sharing of care for children could contribute to decreased statistical discrimination of women. This study suggests that equal sharing could also be beneficial to fathers. If more fathers, and fewer mothers, were to take a large number of leave days, the signalling value of fathers’ leave might decrease.

Supplemental Material

WES868138_Appendix – Supplemental material for Care More, Earn Less? The Association between Taking Paid Leave to Care for Sick Children and Wages among Swedish Parents

Supplemental material, WES868138_Appendix for Care More, Earn Less? The Association between Taking Paid Leave to Care for Sick Children and Wages among Swedish Parents by Katarina Boye in Work, Employment and Society

Footnotes

Acknowledgements

Helena Holmlund, Michael Tåhlin and the WES editor and anonymous reviewers have generously contributed with valuable comments and suggestions.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was funded by the Institute for Evaluation of Labour Market and Education Policy (IFAU no. 123/2010) and the Swedish Research Council for Health, Working Life and Welfare (FAS no. 2010-1076).

ORCID iD

Katarina Boye

Supplemental material

The supplementary material is available online with the article.

Notes

Katarina Boye is Associate Professor of Sociology and lecturer at the Swedish Level of Living Project, Swedish Institute for Social Research. She studies the intersection between family and labour market and focuses particularly on consequences of the division of work between men and women.

References

Acker

(1990) Hierarchies, jobs, bodies: a theory of gendered organizations. Gender & Society 4(2): 139–158.

Albrecht

Skogman Thoursie

Vroman

(2015) Parental leave and the glass ceiling in Sweden. IFAU Working Paper 2015:4, March. Uppsala: IFAU.

Albrecht

Edin

Sundström

, et al. (1999) Career interruptions and subsequent earnings: a reexamination using Swedish data. Journal of Human Resources 34(2): 294–311.

Andreß

Golsch

Schmidt

(2013) Applied Panel Data Analysis for Economic and Social Surveys. Berlin; Heidelberg: Springer.

Becker

(1985) Human capital, effort, and the sexual division of labor. Journal of Labor Economics 3(1): 33–58.

Boye

Halldén

Magnusson

(2017) Stagnation only on the surface? The implications of skill and family responsibilities for the gender wage gap in Sweden, 1974–2010. The British Journal of Sociology 68(4): 595–619.

Bryan

Sevilla-Sanz

(2011) Does housework lower wages? Evidence for Britain. Oxford Economic Papers 63: 187–210.

Buchinsky

(1998) Recent advances in quantile regression models: a practical guideline for empirical research. Journal of Human Resources 33(1): 88–126.

Budig

England

(2001) The wage penalty for motherhood. American Sociological Review 66(2): 204–225.

10.

Byron

(2005) A meta-analytic review of work-family conflict and its antecedents. Journal of Vocational Behavior 67(2): 69–198.

11.

Carlson

Lynch

(2017) Purchases, penalties, and power: the relationship between earnings and housework. Journal of Marriage and Family 79(1): 199–224.

12.

Cooke

(2014) Gendered parenthood penalties and premiums across the earnings distribution in Australia, the United Kingdom, and the United States. European Sociological Review 30(3): 360–372.

13.

Cooke

Hook

(2018) Productivity or gender? The impact of domestic tasks across the wage distribution. Journal of Marriage and Family 80(3): 721–736.

14.

Duvander

Viklund

(2014) Kvinnors och mäns föräldraledighet. In: Boye

Nermo

(eds) Lönsamt Arbete – Familjeansvarets Fördelning Och Konsekvenser (SOU 2014:28). Stockholm: Fritzes, 23–61.

15.

Eriksson

Nermo

(2010) Care for sick children as a proxy for gender equality in the family. Social Indicators Research 97(3): 341–356.

16.

Evertsson

(2016) Parental leave and careers – women and men’s wages after parental leave in Sweden. Advances in Life Course Research 29: 26–40.

17.

Firpo

Fortin

Lemieux

(2009) Unconditional quantile regressions. Econometrica 77(3): 953–973.

18.

Grönlund

Halldén

Magnusson

(2017) A Scandinavian success story? Women’s labour market outcomes in Denmark, Finland, Norway and Sweden. Acta Sociologica 60(2): 97–119.

19.

Killewald

Bearak

(2014) Is the motherhood penalty larger for low-wage women? A comment on quantile regression. American Sociological Review 79(2): 350–357.

20.

Korpi

Ferrarini

Englund

(2013) Women’s opportunities under different family policy constellations: gender, class, and inequality tradeoffs in western countries re-examined. Social Politics 20(1): 1–40.

21.

Magnusson

(2010) Why is there a gender wage gap according to occupational prestige? An analysis of the gender wage gap by occupational prestige and family obligations in Sweden. Acta Sociologica 53(2): 99–117.

22.

Magnusson

Nermo

(2017) Gender, parenthood and wage differences: the importance of time-consuming job characteristics. Social Indicators Research 131(2): 797–816.

23.

Mincer

Polachek

(1974) Family investments in human capital: earnings of women. The Journal of Political Economy 82(2): 76–108.

24.

Stafford

Sundström

(1996) Time out for childcare: signalling and earnings rebound effects for men and women. Labour 10(3): 609–629.

25.

Statistics Sweden (2014) På tal om kvinnor och män. Lathund om jämställdhet 2014. Stockholm: Statistics Sweden.

26.

Swedish Social Insurance Agency (2007) VAB – Vård Av Barn. Tillfällig föräldrapenning 1974–2006 (Försäkringskassan Redovisar 2007:10). Stockholm: Swedish Social Insurance Agency.

27.

Swedish Social Insurance Agency (2018) Socialförsäkringen i siffror. Stockholm: Swedish Social Insurance Agency.

28.

Williams

(2000) Unbending Gender: Why Family and Work Conflict and What to Do about It. Oxford: Oxford University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.76 MB