UNICEF - Paid maternity Leave and Women’s Labour Force Participation

WORKING PAPER Paid maternity Leave and Women’s Labour Force Participation: Evidence from recent policy changes in 17 countries. Richard de Groot , Frank Otchere, Nyasha Tirivayi , and Joseph Darko March 2026Paid maternity leave and women’s labour force participation

2 Abstract This paper investigates the effect of paid maternity leave policies on women’s labour force participation using countrylevel interrupted time series analysis across 17 countries. These are the 17 countries with policy reforms extending paid maternity leave to at least 14 weeks between 2000 and 20 19 , as recommended by the International Labour Organization, and examine their effects on women aged 20 – 39, the prime childbearing age group accounting for over 80 per cent of all births . By contrasting this group with women aged 40 – 64 and employing the femaletomale labour force participati on ratio, we isolate policyspecific impacts from broader labour market trends. Our main findings indicate that extending maternity leave periods to the 14week threshold generates limited effects on aggregate female labour force participation rates which aligns well with established theoretical predictions regarding the duration of maternity leave. For selected countries, we also explore unintended shifts toward informal employment and variations in effective coverage and takeup of maternity leave. Our fi ndings contribute to the literature by providing crosscountry evidence on how maternity leave reforms can play a role in improving gender equality and labour market outcomes for women.Paid maternity leave and women’s labour force participation

3 Contents Abstract ................................ ................................ ................................ ................................ .................. 2 Introduction ................................ ................................ ................................ ................................ .......... 4 Theoretical framework and literature review ................................ ................................ .................. 8 Theoretical framework ................................ ................................ ................................ ................................ .......... 8 Literature review ................................

................................ ................................ .................. 2 Introduction ................................ ................................ ................................ ................................ .......... 4 Theoretical framework and literature review ................................ ................................ .................. 8 Theoretical framework ................................ ................................ ................................ ................................ .......... 8 Literature review ................................ ................................ ................................ ................................ ..................... 9 Data and methodology ................................ ................................ ................................ ...................... 11 Data and variables ................................ ................................ ................................ ................................ ............... 11

Empirical strategy: interrupted timeseries analysis (ITSA) ................................ ................................

........ 13 Sample of countries ................................ ................................ ................................ ................................ ............. 16 Results ................................ ................................ ................................ ................................ .................. 17 Descriptives ................................ ................................ ................................ ................................ ........................... 17 Estimation results ................................ ................................ ................................ ................................ ................. 19 Additional findings ................................ ................................ ................................ ................................ ............... 20 Discussion and conclusion ................................ ................................ ................................ ................ 29 Limitations ................................ ................................ ................................ ................................ .............................. 31 References ................................ ................................ ................................ ................................ ........... 33 Additional Figures and Tables ................................ ................................ ................................ ........... 374

Introduction Governments around the world are working to implement policies that improve the quality of life for their citizens. One policy that has gained attention for its potential to promote gender equality and increase female labour force participation and child w elfare is paid maternity leave, including benefits for unemployed households expecting a child. This policy allows mothers or expectant mothers to take paid time off work prior to or following the birth or adoption of a child in order to care for and bond with their children. Until recently, parental leave was essentially maternal leave, but there is growing expansion of paternal leave policies for expectant men as well, in furtherance of gender equality. Despite significant progress in extending parental l eave to fathers, however, the

care for and bond with their children. Until recently, parental leave was essentially maternal leave, but there is growing expansion of paternal leave policies for expectant men as well, in furtherance of gender equality. Despite significant progress in extending parental l eave to fathers, however, the majority of leave is still available and taken up by women (Earle et al., 2025) . Maternity leave policies around the world differ in terms of eligibility criteria, timing for the start and end of the leave, duration and entitlement, and financing modalities. The first formal proposal for paid maternity leave can be traced back to ILO C onvention No. 3 of 1919 on Maternity Protection, which established 12 weeks of maternity leave (six weeks before and six weeks after childbirth) among other provisions. Its application was limited however to women in industrial and commercial sectors (ILO, 1919) . In 1952, a second ILO maternity protection convention was adopted (No. 103) updating the one from 1919. While the minimum duration of maternity remained at 12 weeks, more flexibility was allowed in how the prenatal portion was used. In addition, this con vention expanded the scope to nearly all employed women in all industries (ILO, 1952) . The current minimum threshold of 14 weeks was set by the ILO Convention No. 183 in 2000, with an accompanying Recommendation (No.191) to provide up to 18 weeks of paid leave (ILO, 2000) . This threshold was based on a minimum period deemed necessary to allow a mother to rest and her physical health to recover and to reduce the risk of postpartum depression (Perera et al., 2022) . The ILO specifically mandates at least six weeks of compulsory postnatal leave to ensure recovery immediately following birth. In addition, the postpartum weeks are critical for parentchild bonding and the initiation of breastfeeding, which contribute to longterm child health

. The ILO specifically mandates at least six weeks of compulsory postnatal leave to ensure recovery immediately following birth. In addition, the postpartum weeks are critical for parentchild bonding and the initiation of breastfeeding, which contribute to longterm child health (Khan, 2020; UNICEF, 2019) . Moreover, 14 weeks is considered sufficient to retain attachment to the employer and increases the likelihood that women reenter the labour force after childbirth, hence maintaining job security (Addati et al., 2022) . As shown in Figure 1 (left panel) below, up until the 1970s, only 44 out of 189 countries (about 23 percent) had some form of maternal leave policy for at least 14 weeks. Since then, the number of countries with maternal leave policies for at least 14 week s has grown to 111 (nearly 60 percent) as of 2023, with an additional 67 countries (35 percent) having some form of maternal leave policies with a duration of less than 14 weeks. Eleven (11) countries, or about 6 percent of countries in the world, provide no national level policy on paid maternity leave. This category includes the United States, which has no paid maternal leave policy at the federal level, although several states have5

established statelevel leave policies. 1 The same information is graphed in a map in Appendix Figure A1, showing the geographic distribution and evolution over time of maternity leave policies globally. Considering the population size of women aged 1549 years old 2 in each country, these numbers look slightly different (Figure 1, right panel). By 1970, only 14 per cent of women 1549 years old were covered by a policy that granted them at least 14 weeks paid maternity leave. Big shifts occurred in 2010 and 2020, whe n China and India respectively crossed the threshold of 14 weeks of49 years old were covered by a policy that granted them at least 14 weeks paid maternity leave. Big shifts occurred in 2010 and 2020, whe n China and India respectively crossed the threshold of 14 weeks of paid maternity leave. By 2023, 73 per cent of women 15 – 49yearold were entitled to at least 14 weeks of paid maternity leave, with another 22 per cent of women having access to maternity leave less than 14 weeks. Globally, less than 5 per cent of women in the reproductive age range had no access to paid maternity leave. Figure 1 , Number of countries/territories with different degrees of paid maternity leave policies (left), and share of women aged 1549 years old covered by national maternity leave legislation (right)

Note: The figure includes information for 189 countries/territories covering >99% of the global female population aged 1549 years. The statistics in panel B are weighted by the population of women 1549 years old in each countryyear, taken from the World Population Prospects Database (2024). Source: Author’s calculations based on the Women, Business and the Law dataset (World Bank) and World Population Database 2024.

1 In addition, this category includes Australia, New Zealand, Norway, Portugal and Sweden . These countries provide generic ‘parental leave’ to which the mother has access, yet no clearly defined maternity leave policy. They are thus coded as having no maternity leave policy in the data, but in reality, these countries provide access to paid leave for women around childbirth (Dobrotić et al., 2025) . 2 The 1549 age group is considered as the typical reproductive ages for women and therefore, this age group is the most likely to benefit from paid maternal leave policies.6

Our main outcome in this paper is female labour force participation (FLFP). Historically, FLFP has

. 2 The 1549 age group is considered as the typical reproductive ages for women and therefore, this age group is the most likely to benefit from paid maternal leave policies.6

Our main outcome in this paper is female labour force participation (FLFP). Historically, FLFP has been increasing since the 1970s among women in their prime childbearing age (2039 years old) (Figure 2). The unweighted figure shows that across all countri es, FLFP for this age group was about 69 per cent in 2023. Participation among the older age group, women aged 4064 years old, also increased over time and almost closed the gap with the younger age group by 2023 in the unweighted graph. The femaletomal e LFP ratio also shows in increasing trend over the last decades, suggesting that more women are entering the labour force globally. The right panel in Figure 2 weighs these numbers by population size and shows a similar, albeit more erratic, pattern. Inte restingly, the LFPR of women 4064 years old surpassed the rate of women 2039 years old in the most recent years. Figure 2 , Labour force participation among different female population groups, unweighted (left), and weighted by population size (right)

Note: The figure includes information for 189 countries/territories covering >99% of the global female population aged 2039 years. The statistics in panel B are weighted by population numbers of women 2039 years old, taken from the World Population Prosp ects Database. Source: Author’s calculations based on ILOSTAT and World

Population Database. We make four primary contributions to the literature: First, we analyse recent policy reforms (2000 – 20 19 ) where countries expanded paid maternity leave to meet the 14week threshold. This contemporary focus allows us to isolate the immediate and mediumterm effects of these policies, contrasting with studies of early adopters where such effects have long dis

– 20 19 ) where countries expanded paid maternity leave to meet the 14week threshold. This contemporary focus allows us to isolate the immediate and mediumterm effects of these policies, contrasting with studies of early adopters where such effects have long dis sipated. Second, we employ countrylevel Interrupted Time Series Analysis (ITSA) rather than traditional crosscountry pooled regressions. Crosscountry models are highly susceptible to confounding biases stemming from differing economic conditions, labour policies, and social norms. ITSA mitigat es these biases by examining nationwide policy changes within the same population over time, significantly7

reducing the influence of unobserved, slowmoving confounders. Third, we focus specifically on women in their peak childbearing years (ages 20 – 39), who account for 80% of global births and represent the primary demographic most likely to benefit from paid maternal leave policies . By using women aged 40 – 64 as a comparison in multigroup ITSA analysis , we successfully disentangle the specific impacts of maternity leave from broader labour market trends. To our knowledge, this is the first paper to utilize thi s specific age group comparison. Finally, we further utilize the femaletomale labour force participation ratio in multigroup ITSA analysis to account for overarching macroeconomic and general labour market trends that affect both men and women simultaneously. This latter analysis also provides a sense of whether FLFP substitutes or complements male LFP.8

Theoretical framework and literature review Theoretical framework There are several theoretical predictions regarding the impact of maternity leave policies on women's employment. A key prediction is that maternity leave has a nonlinear, specifically inverted Ushaped, impact on women’s labour force participation and wa ges. Ruhm (1998) posits that while

There are several theoretical predictions regarding the impact of maternity leave policies on women's employment. A key prediction is that maternity leave has a nonlinear, specifically inverted Ushaped, impact on women’s labour force participation and wa ges. Ruhm (1998) posits that while short to moderate leave increases female labour force participation by maintaining workforce attachment, extended leave decreases labour supply and depresses wages by eroding skills and signalling low career commitment. In addition, long er leave durations may disincentivize employers from hiring women of childbearing age, reducing labour demand for this demographic through discrimination (Olivetti & Petrongolo, 2017) . Furthermore, extended maternity leave has been linked to increased parttime work, which has hindered women’s advancement into highlevel positions in Europe compared to the United States (Blau & Kahn, 2013) . On the supply side, maternity leave provides minimum employment protection and income support around childbirth; this prevents the loss of firmspecific skills, reduces the friction of jobhunting, and simplifies the return to work, thereby encouraging wom en’s labour supply (Waldfogel, 1998) . Conversely, on the demand side, maternity leave can make the hiring of women of childbearing age riskier or costlier for firms due to potential long absences; which can reduce demand and foster statistical discrimination against women of childbearing age (Arrow, 1973; Phelps, 1972) , or lead to firms passing the costs to women via lower starting wages (Gruber, 1994) . Consequently, employers may only find it more attractive to hire young women, if they perceive a high degree of job attachment and a strong likelihood of employment continuity. At the household level, long, asymmetric maternity leave (where only the mother takes time off) can reinforce traditional gender roles; the mother specializes in childcare while the father gains an uninterrupted labour market advantage, leading to persiste

attachment and a strong likelihood of employment continuity. At the household level, long, asymmetric maternity leave (where only the mother takes time off) can reinforce traditional gender roles; the mother specializes in childcare while the father gains an uninterrupted labour market advantage, leading to persiste nt wage gaps long after the leave ends (Becker, 1985; Mincer & Polachek, 1974) . However, if leave is structured as a family benefit rather than a female only entitlement, it could prevent this specialization. Evidence suggests that shifting from maternity leave to a family/parental leave model increases women’s relative wages and en courage women to return to work (Patnaik, 2019) . Ultimately, because maternity leave duration, timing, and compensation interact with local labour market institutions in diverse ways, its net effect on women’s labour force participation is expected to vary significantly across countries and over time.9

Literature review As summarized in recent reviews the empirical evidence regarding family leave policies is highly contextdependent (Olivetti & Petrongolo, 2017; RossinSlater, 2018; RossinSlater & Stearns, 2020) . However, the reviews, and both crosscountry studies and detailed countrylevel research have demonstrated that an inverted Ushaped relationship exists between the length of maternity leave and mothers’ labour market outcomes. Broadly, there is a consensus that maternity leave durations of less than one year can enhance women's job continuity and employment rates for several years after childbirth, whereas longer maternity leave decreases women's labour force participation, earnings, and career advancement. Looking at specific durations, introducing and extending parental leave rights and wage compensation for up to six months has been linked to improvements in mothers’ labour market outcomes. For exampl e, in their crosscountry study, Del Rey et al (2021) identified a threshold at approximately 30 weeks. Extending maternity leave rights to a year appears to have little e

improvements in mothers’ labour market outcomes. For exampl e, in their crosscountry study, Del Rey et al (2021) identified a threshold at approximately 30 weeks. Extending maternity leave rights to a year appears to have little e ff ect, whereas longer leave rights are associated with negative impacts on women ’ s wages and employment (Canaan et al., 2022) . While most of the research on the effect of maternity leave on women’s economic participation has been in highincome countries, a few recent studies focused on lowerand middleincome countries (LMIC). A study by Fallon et al (2017) covers 121 developing countries from 1999 to 2012. Using fixedeffects panel regressions, they examine how maternity leave affects fertility, infant and child mortality, and women’s labour force participation, specifically analysing how these effects are moderated by GDP per capita and secondary school enrolment. They find that maternal leave policies (both the existence of a policy, and the wage replacement rate) are not associated with increased female labour force participation in their sample. The prim ary reason for this finding is that women in LMICs usually work out of necessity, and a large portion of the labour force is in the informal sector without access to benefits. In contrast, a study by Aslim et al. (2021) focused on 21 emerging countries (middleincome or lower highincome) from 1995 – 2014 and found a positive but limited effect of the duration of maternity leave on women’s labour force participation. In addition, the effects on male labour force partici pation were stronger, leading to an economywide increase in labour force participation. The authors offer two potential explanations for this stronger effect on men compared to. First, structural changes like improved maternity leave policies enhance worklife balance, which could

partici pation were stronger, leading to an economywide increase in labour force participation. The authors offer two potential explanations for this stronger effect on men compared to. First, structural changes like improved maternity leave policies enhance worklife balance, which could encourage both men and women to enter and remain attached to the labour force. Second, maternity leave policies could shift intrahousehold divisions of labour, inducing more men to enter the labour force while their partners are on leave. Particularly if the maternity leave period is unpaid, more men might seek employment to make up for their partner’s lost income. Ahmed and Fielding (2019) analysed macropanel data from 18 African and Asian countries between 1995 – 2016 to investigate the effects of maternal leave duration and payment levels on the share of women in formalsector employment. Using this alternative indicator for women’s labour force participation, their analysis shows a negative effect between maternal leave and formal10

sector employment. This finding may be explained by a reduced demand for female workers in the formal sector, as employers would bear some of the costs associated with maternity provisions. Finally, Amin and Islam (2022) analysed firmlevel survey data from 111 developing and emerging countries to estimate the impact of paid maternity leave on female employment within registered private firms. Their analysis demonstrated a positive relationship between maternity leave dur ation and the share of female workers within a firm. Furthermore, this positive relationship is stronger when maternity leave is predominantly publicly funded.11

Data and methodology Data and variables Our main independent variable is the existence of a maternity leave policy granting at least 14 weeks paid maternity leave. We base our independent variable on information collected by the Women, Business and the Law dataset from the World Bank. This data series reports the number of paid leave days available to the mother for the birth of a child to be taken just before, during and immediately after childbirth. Countryweeks paid maternity leave. We base our independent variable on information collected by the Women, Business and the Law dataset from the World Bank. This data series reports the number of paid leave days available to the mother for the birth of a child to be taken just before, during and immediately after childbirth. Countryyear observations with at least 98 days (14 weeks) of paid maternity leave are coded as meeting the ILO recommendation, whereas those with fewer days are coded as not meeting it. Our key outcome variable is the labour force participation rate among women aged 2039 years. We obtain the female labour force participation rate from the ILO, more specifically the labour force participation rate by age and sex (ILOSTAT, 2024b) . This rate represents the share of employed and unemployed persons as a percent of the total number of people in the agesex group. Hence, the labour force includes persons that are employed or selfemployed, even if they are temporarily absent due to sic kness or leave, and those that are not employed but are available and actively seeking work. 3 For the purpose of our analysis, we are particularly interested in women of reproductive age in their prime childbearing age, so we combine several age groups to obtain an estimate of the labour force participation rate of women aged 2039 years old. To do so, we weigh the LFP rate for each age group by the population share of the age group in the total population, using population size figures from the World Population Database. Our rationale for focusing on women aged 2039 is that over 80 per cent of global births occur w ithin this demographic. Among the 17 countries in our analytical sample, this share is 84 per cent in 1990, rising to almost 87 per cent in 2023 (Appendix Table A1). 4 To isolate the specific effect on target demographic and exclude generic labour market trends, we augment our analysis by including a comparison group likely unaffected by maternity leave policies.

Table A1). 4 To isolate the specific effect on target demographic and exclude generic labour market trends, we augment our analysis by including a comparison group likely unaffected by maternity leave policies. This comparison group comprises women aged 4064 years old , who have passed the prime childbearing age. 5 By comparing the labour force participation trends of women aged 2039

3 Women that are on maternity leave but are still under contract with their employer are thus considered ‘employed’ in the ILO definition. 4 The only other age group that might be justified to include is the 15 – 19yearolds , which account for between 913% of births globally. However, since this group mostly represents minors and adolescents, this age group will not be a prime target for maternity leave policies, as they may not yet be in the labour force for a long time (if at all). 5 Only 34% of births globally are with mothers in this age range (see Table A1 for data source).12

against those of women aged 4064, we aim to separate the impact of maternity leave policies from broader labour market dynamics affecting women in general . In addition, we use the femaletomale labour force participation ratio, which is calculated as the female labour force participation rate divided by the male labour force participation rate for the same age group, 20 – 39yearolds. This additional outcome serves to control for general labour market developments over time that affect both men and women. An increase in the ratio would suggest that women’s labour force participation is increasing more rapidly than men’s (or decreasing less rapidly than men’s i f the overall labour force participation is trending downward). While we are mostly interested to examine the effect of maternity leave policies on women’s labour force participation, we recognize that such effects might be challenging to accurately capture, as

decreasing less rapidly than men’s i f the overall labour force participation is trending downward). While we are mostly interested to examine the effect of maternity leave policies on women’s labour force participation, we recognize that such effects might be challenging to accurately capture, as shown by previous literature on the topic. To assess some shortterm or immediate effects, we therefore also consider the coverage or takeup of maternity leave in the countries of our study. Data on coverage or takeup is hard to come by, but the OECD provides a data series on the recipients/users of publiclyad ministered maternity leave benefits or publiclyadministered paid maternity leave per 100 live births for the period including 20052023 (OECD, 2024) . We match these series to our dataset of countryyear observations, and two countries in our dataset (Israel and Switzerland) have corresponding data. For these two countries, we therefore also assess if the maternity leave policy change has induced a hig her takeup rate of maternity leave benefits. Finally, we also examine changes in the informal employment rate among women of the same age group (2039 years old). Previous studies have demonstrated that an unintended effect of paid maternity leave is a shift from formal to informal employment for wom en (Ahmed & Fielding, 2019) . Especially if employers bear some of the costs of the paid maternity leave, this could disincentivize the hiring of women of childbearing age, which could lead them to seek employment in the informal sector. In fact, some estimates suggest that globally, four out of five jobs for women are created in the informal sector (World Economic Forum, 2023) . As the total labour force participation rate would be the same, such a change would not be picked up in average labour force statistics, yet it would signal a decline in the quality of employment for women. We compile data from ILOSTAT on the informal em ployment rate by sex and age group, similarly to what we have for the

rate would be the same, such a change would not be picked up in average labour force statistics, yet it would signal a decline in the quality of employment for women. We compile data from ILOSTAT on the informal em ployment rate by sex and age group, similarly to what we have for the LFPR. 6 We focus again on two age groups, women 2039 years old as our prime age group of

6 According to the ILO, the informal employment rate represents “ the share of persons who are in informal employment as a percent of total employment. Informal employment comprises persons who in their main or secondary jobs were (a) ownaccount workers, employers and members of producers’ cooperatives employed in their own informal sector enterprises; (b) ownaccount workers engaged in the production of goods exclusively for own final use by their household (e.g. subsistence