Maternity leave versus early childcare—What are the long-term consequences for children?

Background and differences between countries

In theory, investments in formal, high-quality early childhood care and education (ECCE) should produce substantial long-term benefits. First, because the receptivity of the brain is at its highest during infancy, and, second, because skill acquisition has been shown to be a dynamic, self-productive, and cumulative process. The evidence on the long-term benefits of ECCE mainly originates from programs targeted toward children from disadvantaged backgrounds, and the long-term returns include decreased crime and increased earnings as adults. There is skepticism, however, as to whether universal formal childcare programs available to all children will yield high returns because they involve providing costly (potentially unnecessary) subsidies to middle- and high-income families. On the positive side, there would likely be broad popular support for such a program. Also, peer effects arising within a universal program may be quite different from those in a targeted program (e.g. positive spillover effects of having peers from better backgrounds).

An alternative to providing subsidized childcare is to pay parents to stay at home as caregivers. Over time, countries have expanded both maternity and, increasingly, paternity leave schemes, and some have even adopted cash-for-care policies. For instance, in 2008 Sweden adopted a cash-for-care policy under which municipalities compensated parents of children between the ages of one and three for not using public childcare. This policy was discontinued in 2016 because of negative effects on female employment, particularly in rural areas. Cash-for-care policies have also been instituted in Finland (since 1985) and Norway (since 1998) and still exist today. Generally, the payment is a flat rate and rather low, and therefore the take-up is mainly by low-educated mothers.

In many cases, reliable evidence on both types of policies originates from Scandinavia because of their early adoption of these programs and due to the availability of population register data that follows entire cohorts of children and their families over decades. In fact, the Scandinavian countries have for years been expanding and financing universal childcare and supporting long and generous maternity leave schemes. The correlation between spending on leave schemes and spending on childcare (based on the data in Figure 1) is negative when excluding the Nordic countries (−0.06) and strongly positive (0.49) when including them. Thus, apart from the Nordics, most OECD countries make a trade-off of investing in parents (mainly mothers) as childminders versus investing in public childcare.

Long-term effects of maternity leave expansions

In 1984, the Danish government increased birth-related leave by almost 50%, from 14 to 20 weeks. A study from Denmark compares outcomes of children born two months before and two months after this reform [2]. Results show no significant differences in reading scores, the probability of high school enrollment, or high school grade point average between these groups. Additionally, maternal career outcomes 10–15 years later are either unaffected or slightly positively affected. The article discusses some reasons for the non-findings: (i) the counterfactual type of care to parental care in Denmark is high-quality childcare; (ii) less pressure on the daycare system may have meant better conditions for those attending it; and (iii) parents make compensating investments—for instance giving more of their time (by cutting down on leisure time, say) or effort (“quality time”) or even more purchased goods to a child who attends daycare, so that child outcomes are the same regardless of care type.

Similar, though short term, evidence can be found in a study that exploits a change in maternity leave benefit entitlements in Canada, which impacted children at the sensitive ages of six months to a year [3]. It finds (at best) weak effects of the reform on child health and development up to two years later. Another Danish study investigates the impact of an increase in maternity leave on child and family health one and three years after birth for children born within a 60-day window around a reform that extended maternity leave entitlement with full pay from 24 to 46 weeks [4]. Overall, it finds little evidence that the leave expansion reduced either child hospitalizations or diagnoses of maternal depression.

While the above findings indicate rather minimal effects of increased maternity leave, could there be more pronounced effects resulting from different leave lengths? Three major maternity leave policy changes were enacted in Germany in 1979, 1986, and 1992, which increased the length of leave from 2 to 6 months, from 6 to 10 months and, later, from 18 to 36 months (successive increases from 10 to 18 months took place in between). A study evaluating these three reforms estimates difference-in-differences models comparing children born shortly before and shortly after the reform, in years in which the reforms took place, and further with those born in years when no reforms took place [5]. In no case (not even with the first reform) does it find significant positive effects on children’s long-term outcomes in terms of choice of school track, the probability of completing high school, wages, or years of employment. In fact, the third reform actually showed detrimental effects on children in terms of educational attainment. The study further brings time-use evidence from the German Time Use Survey (GTUS), indicating that when mothers work during a child’s infancy they do not reduce the time spent caring for children; instead, they reduce their leisure and household activities.

On the other hand, contrasting evidence comes from a maternity leave expansion from the 1970s that introduced paid maternity leave for the first time in Norway (paid leave increased from 0 to 4 months and unpaid leave rose from 3 to 12 months) [6]. The study finds a reduction in the dropout rate from high school and a 5% increase in wages at age 30 among the affected children. These effects are stronger for children whose mothers had fewer than ten years of education. Importantly, maternal income remained the same while maternal time spent with the child increased. It should be noted, however, that these positive effects are found at a time when the counterfactual quality of care was largely informal arrangements such as friends, relatives, or unlicensed caregivers.

Thus, most of the evidence points to weak or zero effects of increasing maternity leave on children’s educational or health outcomes in the medium to long term, especially when the counterfactual care is of high quality. In part, zero effects may arise because parents tend to make compensating investments of time (i.e. they take time away from other activities to spend more time with their children).

Long-term effects of childcare expansions

Few studies on the longer-term effects of childcare expansions estimate direct treatment effects rather than intent-to-treat effects, i.e. most estimate the effects of having had access to childcare, but not whether or not the childcare was actually used. One such study from Denmark investigates the effects of enrollment in different types of non-parental childcare on academic achievement in the final school-leaving year [7]. It does so by linking daycare registers to educational registers. It uses entire birth cohorts of ethnic Danish children, enrolled in either formal center-based daycare or the more informal family daycare at age two, and exploits the right to guaranteed access to preschool in some municipalities. In Denmark, both types of care are regulated with respect to safety, nutritional guidelines, playing facilities, health, and hygiene. However, while staff at centers hold a pedagogical degree (15–16 years of education), family daycare providers typically have either a vocational degree, a high school education, or fewer years of education. The family carers are almost exclusively female, and typically provide the care in their own homes to groups of four or five children. The results show that center-based daycare improves grades in the Danish language in the final year of compulsory school by 0.2 standard deviations. They also show some positive results for mathematics but these are not statistically significant. In terms of parental socio-economic status (SES), the results show that effects are strongest among children of low-SES backgrounds, but even children of high-SES backgrounds benefit from attending formal center-based care rather than informal family daycare. Additionally, boys gain consistently more than girls.

Over an even longer horizon, a study from Norway focuses on a period of rapid childcare expansion in the mid-to-late 1970s [8]. Since most Norwegian mothers post-WWII were already employed, the effect of the considerably greater provision of childcare over this period resulted in a substitution of informal care arrangements by formal care. Using a difference-in-differences design for comparing the long-term outcomes (approximately 30 years later) of three- to six-year-olds before and after 1975, the study finds significant and substantial improvements in education (measured by years of education, high school completion, and college attendance) and earnings and reductions in welfare dependency after the expansion. The increases in education were achieved mainly by children of low-educated mothers, while the higher earnings were reaped mostly by girls.

Although preschool education is not universalized in the US, two states, Georgia and Oklahoma, rolled out high-quality publicly funded preschool programs starting in the mid-1990s. One study compares outcomes of four-year-olds and their families in Georgia and Oklahoma to four-year-olds in other states both before and after the roll-out, and also to an age-ineligible comparison group in a triple-difference framework [9]. It too finds that impacts on children and families differ according to where on the income distribution they happen to be. For children of low-SES households, access to high-quality preschool led to an improvement in mathematics scores which were sustained through eighth grade. For high-SES children, no such increase in achievement was seen. Part of the mechanism for low-SES kids is that they seemed to spend more quality time with their mothers after the reform, although less total time, and also that their mothers were more likely to become employed. For high-SES children, these programs merely tend to crowd-out more expensive private care arrangements.

Putting this evidence together, it appears that when preschool substitutes for informal arrangements, the long-term effects are positive. When it substitutes for more formal arrangements (as it may do for high-SES children), preschool does not seem to have lasting effects.

Universal childcare does not always produce positive outcomes

Most studies focus on how preschool might enhance children’s cognitive skills. However, “soft” skills have also been shown to be important predictors of later success, and may even causally produce such success. A study from Denmark finds that child behavior at age seven (as measured along the Strengths and Difficulties scale) resulting from experiencing high-quality center-based daycare at age three is neutral compared to maternal care, while children experiencing informal daycare display worse behavior than those in maternal care [10]. This result is driven by boys and children of low-educated mothers. The Canadian study described earlier finds that a large-scale formal childcare expansion in Quebec caused mothers of preschool children to enter the labor market in large numbers, which in turn led to increases in hyperactivity, in particular among boys, as well as greater marital instability and lower family well-being [3]. In a follow-up paper, the authors revisited the Quebec universal daycare experiment and found negative and even larger long-term effects on non-cognitive skills, i.e. increases in anxiety and aggression and reductions in prosocial behavior during teenage years. This was driven mainly by boys, for whom they saw increased criminal behavior and worsened self-reported health, which increased with the “dosage” received—i.e. the number of years of exposure to the childcare program [11]. The study, however, does not report different long-term effects for parents with different SES.

The experience in Quebec differs from other examples in a number of ways. First, the Quebec reform had a substantial effect on maternal labor supply, particularly among post-secondary educated mothers. This contrasts with the Scandinavian and US expansions, where most mothers were already in the labor market. The hours spent in childcare also increased substantially for children in Quebec compared to the rest of Canada. Furthermore, the quality of care offered at the beginning of the program, in particular to low-income families, was not optimal, especially in terms of educational activities.

Subsidized childcare need not necessarily have a detrimental impact on children in all instances where mothers enter the labor force in large numbers. One study looks back at the Lanham Act of 1940 in the US, under which a subsidized, near-universal childcare program was rolled out during WWII [12]. The study finds that employment of mothers increased substantially and that employment and long-term earnings of treated children were higher and their rates of cash assistance (a form of social support) lower. Unlike the Quebec case, the childcare quality under the Lanham Act, though variable, was on average found to be fairly satisfactory by parents. Effects were largest for economically disadvantaged families and none or even negative effects on earnings were found for the most advantaged.

Thus, many studies find that the long-term positive effects of attending early formal childcare are mainly seen for children from low- or medium-SES backgrounds, and that although children from high-SES backgrounds do not gain much, they are not hurt either. Furthermore, results are not always negative when mothers enter the labor force in large numbers. While the Quebec experiment led to long-term behavioral deficits for affected cohorts of children compared to the cohorts not affected, the Lanham Act led to positive effects on employment and earnings, and lowered welfare dependence for the children exposed to the program compared to those not exposed to it. The evidence therefore seems to suggest that the quality of the childcare program being offered matters substantially.

When should children enter childcare?

Few studies have investigated the issue of the optimal starting age of ECCE. One convincing study exploits childcare assignment lotteries in Norway to determine the effect of an earlier starting age (15 vs 19 months) on children’s cognitive achievement at age seven [13]. Its estimates show significant gains associated with starting childcare at the younger age, but evidence on longer-term outcomes of the childcare starting age is lacking. It is of interest, therefore, to provide some original results on this question.

Cross-country data on (self-reported) ECCE starting ages and child cognitive (mathematics and reading test scores) and non-cognitive (achievement motivation—“I want to be the best, whatever I do”) outcomes at age 15 are available from PISA (the Programme for International Student Assessment, coordinated by the OECD). Figure 2 and Figure 3 show the PISA mathematics and reading scales by ECCE starting age, respectively. Compared to not attending ECCE or attending it at six years or older, attending at ages one to five produces higher PISA scores, with a maximum impact seen for those starting at ages two to three. However, these are just the raw correlations.

To establish more robust results, original cross-country regressions have been estimated using data from 30 countries. The results indicate that starting ECCE at any age from zero to five is associated with higher test scores compared to not starting at all or starting at the age of six or later, even within similar country groups. For both reading and mathematics, starting at ages one to two or two to three is associated with the largest gain in scores compared to not starting or starting at age six or later. The gains are on the order of 1.8 standard deviations in mathematics and 1.9 standard deviations in reading. Whereas for achievement motivation, starting at age zero to one is associated with the largest gain of 0.27 standard deviations, compared to not starting or starting at age six or later (results available from author).