WORLD BANK GROUP Encouraging Women’s Economic Opportunities in Croatia Empirical Evidence of Determinants and Policy Advice Encouraging Women’s Economic Opportunities in Croatia: Empirical Evidence of Determinants and Policy Advice Monica Robayo-Abril, Anastasia Terskaya Kristina Noelle Vaughan Natalia Garcia Pena October 2020 The World Bank © 2020 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Acknowledgments This report was produced under the World Bank ECCEU FY20 Poverty Program as one of the knowledge products that aim to contribute to a better understanding of constraints to economic opportunities for women in Croatia. This work was prepared under the guidance of Arup Banerji (Country Director, ECCEU) and Salman Zaidi (ECA Practice Manager, Poverty and Equity GP). The team received invaluable guidance from Elisabetta Capannelli (Country Manager, Croatia), and is grateful for comments received from Lars Sondergaard, Mateo Uribe, Reena Badiani, Emilia Skrok, Lucas Cortazar, Valerie Morrica, Ana Simundza, and Josip Funda. Key Messages This report analyzes potential factors and determinants affecting female labor force participation in Croatia and identifies potential policy options to facilitate greater participation of women in the labor market. Our results show that the main reason for women’s inactivity in Croatia is child-rearing and other family responsibilities. While Croatia provides a generous maternity leave allowance in comparison with other EU countries, the absence of compulsory paternity leave does not encourage the distribution of child-rearing responsibilities between men and women. Childcare responsibilities also hinder mothers of school-age children from participating in economic activity, although this constraint is lower for mothers of children attending schools with longer school days. Our results also show that both informal and formal factors play a role. Patriarchal views—which were demonstrated to be negatively associated with women’s labor force participation—are more prevalent in Croatia than in many European countries; these views tend to be more prevalent among men, older people, and less-educated individuals. Labor market regulations also play an important role: despite recent reforms aimed at relaxing excessively strict employment protection legislation, introducing more flexibility in the labor market, and boosting active labor market policies, Croatia still lags behind its EU counterparts along these dimensions as is reflected in their low ranking on the ease of hiring and firing, a low proportion of flexible forms of employment, and low expenditure and coverage of active labor market policies. Finally, despite being a common reason in the literature for gender wage gaps, we did not find evidence that the systematic selection of women into low-wage occupations contributes significantly to the observed gender wage gaps. Contents Acknowledgments............................................................................................................................................................... 4 Key Messages....................................................................................................................................................................... 5 1 Introduction ................................................................................................................................................................. 9 2 Key Facts and Trends in Female Labor Force Participation ............................................................................. 16 3 Empirical Evidence: Factors and Determinants .................................................................................................. 19 3.1 Availability of Child and Elder Care and Family Leave Policies .............................................................. 20 3.1.1 How Female Labor Force Participation Varies with the Number of Children in Croatia ........... 25 3.1.2 How Maternal Labor Force Participation Varies with the Number of Hours Children Go to School ......................................................................................................................................................... 27 3.1.3 The Impact of Maternity and Paternity Leave Policies ...................................................................... 31 3.1.4 Elder Care and Female Labor Force Participation ............................................................................. 42 3.2 Labor Market Regulations .............................................................................................................................. 42 3.3 Gender Differences in Retirement Age........................................................................................................ 47 3.4 Social Norms Using an Epidemiological approach .................................................................................... 49 3.5 Occupational Choice: Are Croatian Women Sorting into Low-paying Sectors and Occupations? ... 55 3.6 Educational Choice: Are Croatian Women Sorting into Low-paying Educational Fields? ................. 58 3.7 Geographic Mobility ........................................................................................................................................ 61 4 Concluding Remarks and Policy Options ............................................................................................................. 62 References .......................................................................................................................................................................... 66 Annex 1. Tables................................................................................................................................................................. 71 Annex 2. Summary of Changes in Maternity and parental leave policy in Croatia between 2008 and 2020. .... 80 Figures Figure 1. Gender gaps and levels of female labor force participation, Croatia versus EU ................................... 10 Figure 2. As Croatia’s population ages, the working-age population is projected to decline over time ............. 11 Figure 3. Trends in female labor force participation rates (ages 20–64), selected EU countries, 2010–19 (percent) .............................................................................................................................................................................................. 17 Figure 4. Gender ratio in LFP in EU and Croatia by educational level. 2019 ........................................................ 18 Figure 5.Regional variation in Croatian female labor force participation by gender, 2019 (percent) ................. 19 Figure 6. Framework of barriers to female employability .......................................................................................... 20 Figure 7: The young-age dependency ratio is on the low side ................................................................................... 21 Figure 8: High old-age dependency ratio compared to other EU countries ........................................................... 21 Figure 9. Females in Croatia participate less in the labor force, especially at young ages, when family formation begins, 2019 (percent) ...................................................................................................................................................... 22 Figure 10. Use of institutional child and elder care among the general population is low in Croatia, even when compared with countries of similar development levels ............................................................................................ 23 Figure 11. Family responsibilities are cited as the most important reason middle-age females do not participate in the labor market in Croatia, 2019 (percent) ............................................................................................................. 25 Figure 12. Probability of being in labor force in Croatia as a function of the number of children, 2002 – 18(percent) ......................................................................................................................................................................... 26 Figure 13. Career break for childcare in Croatia, 2018 (percent) .............................................................................. 32 Figure 14. International comparison of family policies, 2018 ................................................................................... 35 Figure 15. International comparison of paid leave for fathers, 2018 ....................................................................... 36 Figure 16. Share of employed women on maternity leave. ........................................................................................ 39 Figure 17. Share of males among parents on parental leave ...................................................................................... 39 Figure 18. Full-rate equivalent (weeks) of paid parental leave reserved for fathers ............................................... 40 Figure 19. Share of males among parents on parental leave ...................................................................................... 41 Figure 20: Career break for incapacitated relatives ...................................................................................................... 42 Figure 21. Inactivity trap by select family types, Croatia and the EU-28, 2018 ...................................................... 45 Figure 22. Inactivity trap by tax and benefit components for select family types in Croatia, 2018..................... 46 Figure 23: Male-female ratio in LFP by age.................................................................................................................. 47 Figure 24: Decomposition of population by main activity status. The difference in retirement taking between men and women cannot entirely explain large gender differences in LFP among 55-64 years old. ................... 48 Figure 25: Share of inactive women fulfilling domestic tasks (Percent of women aged 20-64)........................... 50 Figure 26: Share of agreed with the statement regarding gender norms.................................................................. 51 Figure 27: Share agreed with the statement "man's job is to earn money; woman's job is to look after home and family" ................................................................................................................................................................................. 51 Figure 28: Percentage of population that agreed with "man's job is to earn money; woman's job is to look after home and family" by group (%) ..................................................................................................................................... 52 Figure 29: Gender norms beliefs of active and inactive women (%) ....................................................................... 52 Figure 30: Share of Females by Sector of Activity ...................................................................................................... 55 Figure 31: Selection into Industries by Gender. .......................................................................................................... 56 Figure 32: Gender gap in wage (in deciles) by industry, 2018 ................................................................................... 56 Figure 33: Share of females by field of education........................................................................................................ 59 Figure 34: Gender gap in wage (in deciles) by field of education ............................................................................. 60 Figure 35. Limited Geographical Mobility: Many workers fail to move to areas with higher job creation potential .............................................................................................................................................................................................. 61 Tables Table 1. Labor Market Indicators for Roma and non-Roma neighbors, by Sex 2011 .......................................... 71 Table 2. School shifts and labor market indicators by gender ................................................................................... 72 Table 3: Number of hours worked and labor income ................................................................................................ 73 Table 4. Labor income gender gap................................................................................................................................. 73 Table 5. Fathers’ labor supply ......................................................................................................................................... 74 Table 6. The Effect of Paternity Leave on Labor Market Outcomes ...................................................................... 75 Table 7. Determinants of Female Labor Force Participation 2006-2017 ................................................................ 77 Table 8. The Effect of Gender Norm Beliefs on Inactivity. Logit Marginal Effects. ........................................... 78 Table 9. The Effect of Female Labor Force Participation in the Country of Ancestry. Second-generation Immigrants ......................................................................................................................................................................... 78 Table 10: Oaxaca decomposition of within and across industry gender gap in wages. 2018 ............................... 79 Table 11. Oaxaca decomposition of within and across educational specialization gender gap in wages ........... 79 1 Introduction This report analyzes the potential determinants of gender inequality in economic opportunities and the main barriers for women to participate in the labor market in Croatia. While Croatia has experienced a narrowing of labor force participation rates and employment rates between men and women during the past decade, complete convergence has not been realized. In 2019, working-age (20-64 years old) women in Croatia were 10.4 percentage points less likely to be active in the labor market than working-age men, a figure below the EU average (11.7) but representing a gap larger than in 16 EU countries (Figure 1, panel a).1 Low labor force participation in Croatia is an issue not only for women but for men. Labor force participation rate in Croatia is the lowest in the EU and constituted only 71.3 percent in 2019 (for 20-64 years old population). Moreover, female labor force participation in Croatia is one of the lowest in the EU (after Italy, Romania and Greece), reaching only 66.1 percent in 2019 (EU average was 72.9 in 2019) and male force participation is the lowest in the EU reaching 76.5 percent in 2019 (Figure 1, panel b).2 This is particularly concerning given that Croatia the working-age population has already shrunk by 20 percent since its peak in 1990, and with a further 8.8 percent decline expected by 2040.3 Persistent low levels of female labor force participation in Croatia are a cause for concern, not only due to the low levels compared to other countries in the EU but also because of the expected decline in the number of workers in the labor force in the future due to the changes in the age composition of the population and high rates of outward migration of young people from Croatia shifting the population toward older ages. Women’s participation in the labor market varies significantly across countries in the EU, reflecting differences in economic development, labor market institutions, social norms, levels of education, fertility rates and the number of children, and access to childcare and other supportive services. This variation in factors across the EU is reflected in substantial variation in female labor force participation rates across countries in the EU, far exceeding the variation in the labor force participation of men. 1 For 15-64 years old, the gender gap in LFP constituted 9.9 percentage points in 2019 (61.6 percent of 15-64 years old women were active vs. 71.5 percent of men). Total LFP in 2019 for 15-64 years old was 66.5 percent). This gap is below the EU-28 average of 10.7 percentage points, but it is larger than the gap in 17 EU countries. 2 Source: Eurostat official estimates 3 Croatia’s working age population peaked in 1990 and has, since then, fallen by 20 percent. According to United Nations, Department of Economic and Social Affairs, Population Division (2019). According to World Population Prospects 2019, Online Edition. Rev. 1, Croatia’s working age population is expected to shrink by another 8.8 percent by 2040. 9 Figure 1. Gender gaps and levels of female labor force participation, Croatia versus EU Panel a. Gender Gap in Labor Force Participation (Age: 20-64), 2019 25 20.1 20.3 20.4 20 17.6 15.9 14.8 15.5 15 13.4 13.4 11.0 11.7 9.9 10.4 10.5 10 9.0 9.1 9.3 9.5 9.6 9.7 7.2 7.2 7.9 6.3 6.5 5.0 5.4 5 3.9 2.9 0 Panel b. Labor Force Participation Rates (Age: 20-64), Percentage, EU28 countries, 2019 100 90 80 70 60 50 40 30 20 10 0 Females Males Source: Eurostat official estimates based on 2019 EU-LFS. Age: 20-64 In addition to the impacts on growth and labor force size and composition, population aging puts pressure on the pension, health, and social care systems; if inactivity levels remain at current levels while the share of the non-working population increases meaningfully during the next decades due to aging populations and emigration, the sustainability of social security systems that depend on pay-as- you-go systems is likely to be greatly jeopardized. According to the medium variant population projections 10 of the UN, Croatia’s working-age population (15+)4 is expected to decrease by 8.85 percent by 2040 (Figure 2, panel b). Population aging has significant implications for the sustainability of the social security system such as the pension, health, and elderly care systems.6 The potential to increase labor supply through facilitating greater labor force participation of women, younger, and older workers, and ethnic minorities represents a source of untapped potential as well as a potential solution to the population aging problem, particularly since their labor market participation tends to be lower than that of working-age males. Increasing labor force participation is also particularly important in a country with emigration, considering the potential negative impact of emigration on labor supply. 7 Figure 2. As Croatia’s population ages, the working-age population is projected to decline over time Panel a. Croatia Population Pyramids, 2020 vs 2040 [70-74] [60-64] [50-54] [40-44] Age [30-34] [20-24] [10-14] [0-4] 10 5 0 5 10 Percent Males 2020 Females 2020 Males 2040 Females 2040 Panel b. Working-age Population Projections, Period 2020-2040 4 The official working-age population statistics in Croatia used 15+ as the reference age 5 This rate of decrease based on the UN population projections is slightly lower than predictions performed by the EC. (9.4 percent decrease by 2040). Source: European Commission. (2018c). 6 Recent pension reform in Croatia has counteract the positive impacts of rising dependency ratios on gross public pension expenditure. Source: European Commission. (2018c). 7 After the EU accession in 2013, Croatia experienced a rapid growth of emigration, with stronger emigration flows in the regions with high unemployment. According to Croatian Bureau of Statistics (CBS), prior to the crisis Croatia had a positive net migration balance (number of immigrants – number of emigrants). However, after 2008 net migration balance turned negative, remaining relatively low and stable until 2013 when Croatia joint the EU. After Croatia became a member of the EU in 2013, emigration outflows increased significantly, while the number of immigrants was relatively stable, which led to the negative net migration balance (Draženović et. al, 2018). 11 2 Change in Population 15+ 2020- 0 -2 -4 2040 (Percentage) -6 -8 -10 -12 -14 -16 Source: United Nations, Department of Economic and Social Affairs, Population Division (2015). World Population Prospects: The 2015 Revision. Medium variant projections. The potential benefits of gender inclusion are high. Investing in the skills and productive inclusion of the female population can herald important economic benefits in aging countries. First, international evidence shows that female labor force participation isan important driver and outcome of economic growth and development (Ostry et. al, 2018). Second, with an aging population pension and health care costs are expected to increase rapidly. Third, female inclusion means productivity gains.8 Finally, a skilled female workforce means positive fiscal effects. Equal labor market opportunities for women could enable faster productivity growth and could contribute to fiscal benefits through increased revenue from taxes and lower social assistance spending. The gender gap in labor market outcomes is a multifaceted phenomenon explained by a combination of factors. Gender differences in economic opportunities may be explained by gender differences in endowments (such as skills, access to productive inputs, information, and networks, time and access to services), differences in preferences (time use, family formation, migration or mobility), labor market discrimination, social norms, and institutions. Guided by this set of possible determinants of the gender gap in labor market outcomes, we analyze the degree to which low female employment in Croatia can be explained by the availability of child and elder care, family leave policies, labor market regulations and institutions, social norms, educational and occupational choices. We quantify the effect of these constraints on female labor market outcomes to identify potential policy options to promote women’s economic opportunities in Croatia. We draw conclusions and make policy recommendations by using a quantitative approach based on county-specific analysis, as well as cross-country comparisons. We provide a comprehensive description of key facts in female force participation and other labor market outcomes, as well as causal analysis of its determinants. 8 Heathcote et. al (2017) show that half of the growth in US earnings per capita over 1967-2002 can be attributed to growth in female labor force participation. 12 The key results are as follows. Young (15-24), low-educated women, women with more than two children, and Roma women have the highest prevalence of inactivity in Croatia. Inactivity rates among females with lower educational attainment in Croatia are among the highest in the EU and have been increasing over time. In contrast, the female labor force participation among highly educated females is above the EU average. The gender gap for low educated working age (20-64) population constitutes 20.5 percentage points, while this gap is just 1.4 percentage points for high educated. Moreover, the labor force participation rate among Roma females in Croatia is extremely low and the gender gap in labor force participation among the Roma population is significantly larger than in neighboring countries such as Bulgaria and Romania. Special consideration should be given to these groups when designing policies aimed at increasing female labor force participation given the persistently low participation rate among these groups of women and the likelihood of them facing high barriers to participation. When surveyed as to the reasons for inactivity, women in Croatia reported child-rearing and other family responsibilities as the main impediments. Specifically, in 2019, 27.9 percent of inactive women between 20 and 64 years old (vs. 9.4 percent of men) reported being inactive because of personal or family responsibilities and 6 percent of inactive women (vs. less than 1 of percent of men) reported being inactive because of child-rearing and care responsibilities.9 Among prime-age women (25-49 years old), the majority (61.4 percent) of women are inactive because of family, personal or childcare responsibilities (vs. 19.0 percent of men). This can be attributed to three main factors: (1) high old-age dependency ratios in Croatia by EU standards; (2) significant differences in the distribution of tasks between men and women potentially driven social norms; (3) the lack of affordable child and eldercare. Career interruptions due to family and child-rearing responsibilities can lead to substantial penalties in the labor market. This is likely to disproportionately affect Croatian women rather than men due to unequal gender parental leave policies. While Croatia provides a generous maternity leave10 allowance in comparison with most EU countries, the absence of compulsory paternity leave does not encourage a redistribution of child-rearing responsibilities between men and women. As a result, the share of men who participate in parental leave in Croatia is extremely low in comparison to other EU countries. The gender differences in uptake of maternity and paternity leave might result in lowering women’s economic activity for several reasons. First, international 9 In the EU-LFS, we can identify the following two categories: “Other family or personal responsibilities " and "Looking after children or incapacitated adults ". For 15-64 years old, 27.9 percent of women reported being inactive because of family, personal or childcare responsibilities vs. 7.9 percent of men. 10 Here by maternal leave we mean maternity or parental leave taken by mother (including unpaid leave). Women in Croatia are entitled to 3 years of job protected leave. 6 months is a paid maternity leave. After that, they can take parental leave that is 4 months to each parent with 2 months transferable from one parent to another (so, mothers typically take 6 months and fathers 2). So, in total, women have 1 year of paid leave and up to 2 years of unpaid leave. 13 evidence suggests that extended periods of absence from the labor market can result in labor market penalties. This is likely to disproportionately affect women in Croatia as they may be induced to take longer periods of absence from the labor force due to a low uptake of paternal leave by men and unequal care responsibilities. Very long leave may hurt female labor supply in the long run if it induces women to stay out of work for long by reducing their experience and human capital. Second, relatively longer maternity leave driven by social norms on care responsibilities may lead to discriminatory practices by employers in the hiring of women due to the additional costs associated with replacing the loss of the worker (e.g., expenses associated with hiring and training temporary replacement workers). Long parental/maternity leave for women relative to that of men may lower the demand for female labor by inducing additional costs of female workers on employers (e.g., expenses associated with hiring and training temporary replacement workers). In line with this, our cross- country analysis shows that there is a robust negative association between men’s participation in parental leave and the gender gap in labor force participation and employment. Therefore, inducing fathers to participate more in parental leave through transformation of the parental leave system and addressing social norms on care responsibilities might be an effective tool for reducing the gender gap in economic activity. Childcare responsibilities may also prevent mothers of school-age children from economic activity, although this constraint is likely to be lower for mothers of children attending longer school days. Having more children of compulsory school age decreases participation in the labor force for women but not for men. In Croatia, there is a large fraction of households with children not currently enrolled in the whole day school program, with this share especially large among low-income families. Our analysis of the Whole- Day-School program in Croatia suggest that women with children attending longer school days have a lower incidence of inactivity and unemployment, tend to work more hours, and earn more on average than women of children attending shorter school days. In contrast, men’s labor market outcomes are not significantly affected by the length of the school day. The results suggest that an extension of the Whole-Day-School program may positively affect the labor market outcomes of women with school-age children and help to reduce the gender gap in these outcomes. These results also suggest that the current COVID-19 crisis may significantly affect Croatian mothers and lead to an increase in gender inequality. In the event of large-scale school closures, women are expected to carry a large care burden during the crisis, given prevalent social norms surrounding the gender distribution of child and eldercare responsibilities. In addition to child and elder care and family responsibilities, labor regulation and institutions also play an important role. Croatia still lags behind its EU counterparts along these dimensions as is reflected in their low ranking on the ease of hiring and firing, the low proportion of part-time work, and low expenditure and coverage of active labor market policies, despite recent reforms aimed at relaxing excessively strict 14 employment protection legislation, introducing more flexibility in the labor market, and boosting active labor market policies. Additionally, though Croatia has reduced the burden of income taxation for individuals with lower earnings, the inactivity trap often remains above the EU average, primarily due to the abrupt withdrawal of certain benefits when transitioning from inactivity to employment particularly for single individuals with children towards the lower end of the earnings distribution. These constraints are likely to be particularly binding for women who often require flexibility due to care responsibilities and are often are seeking to reenter the labor force after extended periods of absences and may require greater assistance in navigating the transition from inactivity to employment. Improvements in the labor market outcomes of women in Croatia could be seen through a number of labor market reforms such as: increasing the expenditure on and efficacy of active labor market policies, phasing out the receipt of certain benefits over a longer period of time upon entering employment for low earners, and increasing the quantity and quality of flexible jobs. Although the gender gap in wages in Croatia is among the lowest in the EU, women earn less than men in almost all sectors of economic activity, which can be discouraging women to participate in the labor market. We show that this gap cannot be explained by gender-specific sorting into a sector of activity or field of education and that there is a large wage gap within industries.11 There are some differences in occupational choices between men and women so that women are less likely to be employed in construction and manufacturing than men and more likely to be employed in education, health services, and other service sectors. However, the decomposition of the gender gap in wages that we conduct suggests that women do not sort into lower-wage industries than men and that within-industry gender gap in wages is present almost in all sectors. There are also differences in selection into the field of education between men and women. However, the fields preferred by women do not yield lower returns or higher probability of inactivity than the ones preferred by men. In contrast, the gender gap in wages exists almost within all the fields of education. Due to data limitations, we cannot discern the reasons for within-occupation gender gaps, however gender differences in bargaining, wage discrimination and sorting across firms have been cited in the literature as potential reasons. Development of high-quality longitudinal survey that match employers with employees could facilitate an analysis of the potential explanations of within occupation gender wage gap. We show that patriarchal views---which were demonstrated to be negatively associated with women’s labor force participation---are more prevalent in Croatia than in many European countries; these views are more prevalent among men than women and this prevalence increases with age and decreases with educational level. We show that patriarchal views are associated with lower women’s activity rates but have no significant effect on men’s activity. To analyze the causal effect of cultural norms regarding women 11 Large gender wage gaps within industries that can be explained by several other factors including wage discrimination, sorting into firms (within the same occupation/industry) and differences in bargaining behavior. 15 working we compare second-generation immigrants from more and less gender unequal countries of ancestries but who reside in the same country ---this approach is often used in the literature in order to understand the role of informal institutional constraints (culture or social norms) distinct from environmental factors. The results of this analysis show that there is a negative association between female labor force participation in the country of birth of parents and second-generation immigrant women’s inactivity. Therefore, policies that affect formal institutions might be more effective if complemented by the policies that transform cultural norms into more gender neutral. The report is organized as follows. Section 2 introduces some stylized facts of the Croatian labor markets, focusing on labor force participation trends and EU comparisons. Section 3 details the empirical evidence analyzing the main factors and determinants, including the impact of the availability of childcare, eldercare and family policies, labor market regulations, social norms, and occupational and educational choice, and geographic mobility. Section 4 concludes and summarizes policy advice. 2 Key Facts and Trends in Female Labor Force Participation The levels of female labor force participation in Croatia are low compared with other EU countries12 and have only been improving slower than in other EU countries. When looking at the period 2010-2019, female labor force participation in Croatia (for 20-64 years old) has only increased by 2 percentage points, compared to 4.5 percentage points among EU28 countries, on average. This growth is also slow when compared to other countries such as Bulgaria and Poland, that have seen increases of 6.5 and 4.1 percentage points, respectively. Male labor force participation in Croatia is also one of the lowest in the EU and constituted 76.5 percent in 2019 (vs. 84.6 percent in EU on average). Still, men are about 10.4 percentage points more likely to participate in the labor market than women; this gap is below the EU average, but it is larger than in 16 EU countries. Male LFP has increased only by 0.7 percentage point between 2010 and 2019, which is similar to the trends observed in other EU countries (male LFP has increased by 1.0 percentage point on average among EU countries during the same period).13 Croatia had one of the longest recessions in the EU from 2008 until 2014, which was accompanied by a reduction in the gender gap in LFP due to the larger decrease in men LFP than in women LFP. 12FLFP was 66.1 percent in 2019 vs. 72.9 percent in EU-28 on average (for 20-64 years old). 13For 15-64 years old, female LFP has increased by 2 percentage points (vs. 4.3 in EU on average) and male LFP by 0.9 percentage points (vs. 1.8 in EU on average) between 2010 and 2019. 16 Figure 3. Trends in female labor force participation rates (ages 20–64), selected EU countries, 2010– 19 (percent) Panel a. All 75 70 65 Percent 60 55 50 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 European Union - 28 Bulgaria Croatia Poland Romania Panel b. Low educated 55 50 45 Percent 40 35 30 25 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 European Union - 28 Bulgaria Croatia Poland Romania Source: Eurostat official estimates for 2010-2019. Age: 20-64. Activity rates among females with lower educational attainment in Croatia are among the lowest in the EU and have been decreasing over time. Labor force participation rates among females with lower educational attainment in Croatia was only 34.7 percent in 2019 (for 20-64 years old), compared to 52.7 percent among the EU countries. While on average, inactivity among this group has been relatively stable over time in the EU, this is not the case in Croatia, where participation rates have dropped by 9.2 percentage points over the period 2010-2019.14 The gender ratio in LFP (males/females) rates is higher than the EU average (Figure 4). In contrast, for high educated, the gender ratio is below the EU average. 14 Similar patterns are observed for 15-64 years old. 17 Figure 4. Gender ratio in LFP in EU and Croatia by educational level. 2019 Panel a. By educational level 1.8 1.59 1.6 1.44 1.4 1.16 1.16 1.17 1.19 1.2 1.07 1.02 1.0 0.8 0.6 0.4 0.2 0.0 All Less than primary, primary and Upper secondary and post- Tertiary education lower secondary education secondary non-tertiary education European Union - 28 countries Croatia Panel b. By age 1.8 1.6 1.6 1.4 1.4 1.4 1.4 1.2 1.2 1.2 1.2 1.2 1.2 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.0 1.0 0.8 0.6 20 to 64 15 to 24 25 to 29 30 to 34 35 to 39 40 to 44 45 to 49 50 to 54 55 to 59 60 to 64 years years years years years years years years years years European Union - 28 countries Croatia Source: Eurostat 2019 estimates based on EU-LFS. Age: 20-64 Gender gap in labor force participation is particularly large for young (15-24) individuals and for those older than 55 years old. For these age groups the gender ratio in labor force participation exceeds the EU-28 average (Figure 6). On the other hand, Croatian women tend to be more active between 25 and 55 years old. The gender ratio in labor force participation is above the EU ratio at age. There is some heterogeneity among regions, with larger gender gaps in labor force participation among the old (55-64) living in Adriatic Croatia; by contrast the lowest gap is observed among those aged 25-34 living in the same region. In Continental Croatia, only about one-third of females aged 55-64 participate in the labor market, compared to 52.7 percent of the male counterparts. The gender gap is slightly lower in the Continental Croatia region than in Adriatic Croatia for 55-64-year-old. The gender gaps among those aged 25-34, 35-44 and 45-54 are slightly larger in Continental Croatia than in Adriatic Croatia. 18 Heterogeneity at the NUTS-3 level is not readily available, as the underlying Labor Force Survey Data is not representative at this level of disaggregation. Figure 5.Regional variation in Croatian female labor force participation by gender, 2019 (percent) 100 90 80 70 60 LFP, % 50 40 30 20 10 0 15-24 25-34 35-44 45-54 55-64 HR03 Males HR03 Females HR04 Males HR04 Females Source: Eurostat 2019 estimates based on EU-LFS. HR03: Adriatic Croatia; HR04: Continental Croatia. The EU-LFS is representative at the level of NUTS 2. Therefore, more disaggregate statistics at the level of NUTS 3 are not reliable. The labor force participation rate among Roma females in Croatia is extremely low. Evidence from the 2011 UNDP-WB-EC Roma Regional Survey15 shows that Roma females are much less likely to participate in the labor market than Roma males and are significantly less likely to participate than the majority population as a whole. Indeed, in 2011 the labor force participation rates among Roma females were only 29 percent, compared to 57 percent among neighboring non-Roma females. The gender activity gap among marginalized Roma populations in Croatia was also much wider than in Bulgaria and Romania, reaching 24 percentage points. (see Annex 1, Table 1 Panel a and b for details). 3 Empirical Evidence: Factors and Determinants Given the complex nature of female labor force participation, it is important to highlight how socio- economic factors affect the decision and ability of Croatian women to participate in the labor market. In this section, we present a quantitative analysis to understand key factors explaining the low activity rates. We first present a framework16 to understand many of the factors that have contributed to the key labor market challenges that females in Croatia face. The constraints that females in Croatia face when entering the labor 15More recent estimates among Roma are not available, as ethnicity is not available in the EU-LFS. 16This framework visually represents each of the barriers that women face when entering the labor market, so it focuses on constraints to economic opportunities. Notice this is different from the broader framework used in the World Development Report 2012: Gender Equality and Development, which is broadly used to analyze gender differences in endowments, agency, and economic opportunities. There are many overlapping elements between the two frameworks, but also some differences. For instance, endowments here not only refer to human capital but also access to productive inputs, information, and networks, time, and access to services. 19 market can be summarized by differences in endowments between males and females (such as skills, access to productive inputs, information and networks, time and access to services), different preferences (time use, family formation, migration or mobility), and contextual factors including social norms and institutions (work arrangements, legal rights) that determine the role of men and women in society. These barriers may affect female employability and can be summarized in the framework presented below. Figure 6. Framework of barriers to female employability Time and Access to services: child and elder care Access to productive Skills inputs, information and networks Attitudes and Work social norms arrangements Geographic mobility Source: Arias et al. 2014. In this section, we present empirical evidence to understand how some of these constraints affect female activity levels in Croatia, to identify potential policy options to tackle these constraints. Insufficient human capital accumulation may represent a major barrier to female employment, if females on the labor market are less educated than males. Attitudes and social norms can be a critical constraint to employability, both from the side of employers as well as those participating in the labor market. Low flexibility in work arrangements and other labor market regulations may also limit employability among Croatian females. Finally, lack of affordable child and elder care may also affect women’s decision to participate in the labor market. 3.1 Availability of Child and Elder Care and Family Leave Policies Child and elder care responsibilities may affect women’s decision to participate in the labor market. Consistent with the demographic structure of the population, old-age dependency ratios in Croatia are high by EU standards, while young-age dependency rations are below the EU average (Figure 7 and 8). In 2019, the young-age dependency ratio, defined as the ratio of younger dependents -people younger than 15- to the working-age population (those ages 15-64) was 22.6 percent, while the old dependency ratio was 32.3. Old age dependency ratio is projected to increase to 61.3 by 2070 (European Commission, 2018a). 20 While the gender ratio in labor force participation for prime-age (25-54) individuals in Croatia is below the EU average (Figure 4a), women tend to participate in the labor force significantly less than men during the childbearing years, often not reaching peak participation (with respect to men) until they are in their 40s (See Figures 9 and 23). Moreover, women at this age face different constraints than men that prevent them from being active (Figure 11). The lack of affordable child and elder care, together with social norms that dictate that childcare and eldercare are women’s responsibilities, may represent binding constraints on women during the childbearing years.17 Figure 7: The young-age dependency ratio is on Figure 8: High old-age dependency ratio the low side compared to other EU countries Age dependency ratio, young (% of working- Age dependency ratio, old (% of working- age population), EU 28, 2019 age population), EU 28, 2019 IRL ITA FRA FIN SWE PRT GBR GRC BEL DEU EST BGR FIN FRA DNK SWE LVA HRV NLD LVA CZE MLT CYP EUU ROU EST EUU DNK LTU SVN SVN LTU BGR CZE NLD POL HUN SVK ESP HRV BEL LUX GBR ESP AUT HUN ROU GRC POL AUT SVK DEU IRL ITA LUX PRT CYP 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 40 Sources: World Bank World Development Indicators. The young-age dependency ratio = ratio of children ages 0–14 in the household to the working-age population (ages 15–64). The old-age dependency ratio = ratio of adults ages 65+ in the household to the working-age population (ages 15–64). 17In comparison to the EU, Croatia has a low share of children under 3 covered by formal childcare arrangements, ranking as the seventh lowest among EU member states. For children from age 3 to school age, the coverage of kindergartens and daycares is strikingly low, at 51 percent, placing Croatia at the very bottom of EU countries and far below the EU average of 86 percent. Source: World Bank (2019a). 21 Figure 9. Females in Croatia participate less in the labor force, especially at young ages, when family formation begins, 2019 (percent) Panel a. All Panel b. Low-educated 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 15 to 20 to 25 to 30 to 35 to 40 to 45 to 50 to 55 to 60 to 0 19 64 29 34 39 44 49 54 59 64 15 to 20 to 25 to 30 to 35 to 40 to 45 to 50 to 55 to 60 to 19 24 29 34 39 44 49 54 59 64 Males Females Males Females Panel c. Medium-educated Panel d. High-educated 100 100.0 80 80.0 60 60.0 40 40.0 20 20.0 0 0.0 15 to 20 to 25 to 30 to 35 to 40 to 45 to 50 to 55 to 60 to 15 to 20 to 25 to 30 to 35 to 40 to 45 to 50 to 55 to 60 to 19 24 29 34 39 44 49 54 59 64 19 24 29 34 39 44 49 54 59 64 Males Females Males Females Source: Eurostat 2019 estimates based on EU-LFS. High educated defined as those ISCED 5+, medium educated as those ISCED 3-4, and low educated as those ISCED 0-2. Increasing the coverage of good-quality, affordable childcare and eldercare may increase female labor force participation. There is robust evidence in the literature that women’s labor supply is positively correlated with the availability of affordable childcare and eldercare (Asai, Kambayashi, and Yamaguchi 2015; Borra 2010; Chevalier and Viitanen 2002; Lokshin 2004; Manley and Vásquez 2013; Maurer-Fazio et al. 2011). The provision of child and elder care facilities, but also parental leave and subsidies for childcare may increase work attachment by reducing the share of women leaving their jobs after giving birth or by helping women reenter the labor market after periods of absences due to child and care responsibilities. However, in Croatia, estimates from the 2016 Life in Transition Survey show that the availability of institutional childcare and elder care centers is limited, even relative to countries with similar levels of development (Figure 10). 22 Croatia has one of the lowest shares of children receiving formal care in the EU, based on Eurostat estimates (2016), constituting only 3.8 percent in 2016 (vs. 34.5 percent in the EU). This share is larger in cities (11.1 percent) but extremely low in small towns (1 percent) and in rural areas (1.6 percent).18 On the other hand, the main reason for not making use of formal childcare services in Croatia is that there is no need (92.3 percent report this reason). This may suggest that informal childcare---mainly provided by mothers of grandmothers---is preferred by parents. Figure 10. Use of institutional child and elder care among the general population is low in Croatia, even when compared with countries of similar development levels Panel a. Percentage of population in households using institutional care when there is a child needing care 70 institutional care when there is a child % of population in households using RUS 60 LAI BLR 50 SVN SVK needing care MNG POL CYP KAZ 40 UKR SRB LTU KGZ MNE DEU GEO ROU ITA 30 ALB BGR UZB 20 BIH HRV EST MKD GRC MDA ARM HUN TJK 10 KSV AZE TUR CZE 0 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000 GDP per capita, PPP (constant 2011 international USD), 2018 Panel b. Percentage of population in households using institutional care when there is an elderly needing care 30 SVK institutional care when there is an elderly % of population in households using 25 20 needing care 15 KYR GEO BGR HUN 10 CZE UKR ALB HRV LIT 5 TAJ MDA UZB ARM MKD AZE KSV BIH MNE BEL KAZ POL EST GRC SVN CYP ITA RUS ROM LAI 0 0 10000 SRB 20000 30000 40000 50000 MNG TUR -5 -10 GDP per capita, PPP (constant 2011 international USD) 2018 18 This estimate includes cost-free childcare as well as childcare paid in full or at a reduced price. 23 Panel c. Children receiving formal childcare (percentage) by degree of urbanization and cost 30 24.3 25.2 25.1 25 22.1 20 15 12.0 11.1 11.0 8.9 9.8 10 5 2.8 2.2 1.0 0.8 0.2 0.8 0.8 0 Full or reduced Cost free Full or reduced Cost free Full or reduced Cost free Full or reduced Cost free price price price price Total Total Cities Cities Towns and Towns and Rural areas Rural areas suburbs suburbs European Union - 28 Croatia Source: Panel a and b: World Bank estimates based on 2016 data of LiTS (Life in Transition Survey) (database), European Bank for Reconstruction and Development, London, http://www.ebrd.com/what-we-do/economic-research-and-data/data/lits.html; World Development Indicators and World Development Indicators. Panel c: Eurostat estimates [ilc_ats01] for 2016. The share is for children aged 12 years or less. Consistent with this evidence, one of the main reasons of middle-age (25-49) female inactivity in Croatia is child-rearing and other family responsibilities (see Figure 11). Moreover, there are structural differences in reasons of inactivity between men and women: 61.4 percent of inactive middle-age women vs. 18.2 percent of inactive middle-age men explain their inactivity by family responsibilities. The significant differences in reasons of inactivity between men and women are observed for all age groups except for young individuals (15-24), for whom the main reason of inactivity is education. These results suggest that there are significant differences in the distribution of tasks between men and women, which can partially explain gender differences in employment outcomes. 24 Figure 11. Family responsibilities are cited as the most important reason middle-age females do not participate in the labor market in Croatia, 2019 (percent) Panel a. Age 20-64 Panel b. Age 15-24 Females Males Other Think no work is available Other Think no work is available Retired Retired In education or training In education or training Looking after children or… Looking after children or… Other family or personal… Other family or personal… Own illness or disability Own illness or disability Lay-off Lay-off 0 20 40 60 80 100 0 20 40 60 80 100 Panel c. Age 25-49 Panel d. Age 50-64 Other Other Think no work is… Think no work is available Retired Retired In education or… In education or training Looking after… Looking after children or… Other family or personal… Other family or… Own illness or disability Own illness or… Lay-off Lay-off 0 20 40 60 80 100 0 20 40 60 80 100 Source: Eurostat estimates based on Croatia EU-LFS 2019. 3.1.1 How Female Labor Force Participation Varies with the Number of Children in Croatia To analyze whether female labor force participation is associated with presence of children we conduct a logistic regression analysis using Croatia LFS 2002-2018; the results indicate that the gender gap in LFP in Croatia is wider when children are present in the household and this gap is widening on the number of children (Figure 12, Panel a and b). Specifically, we regress the probability of being active on the number of children, controlling for year dummies, age group indicators, educational level indicators, and marital status. 25 Results show that that the probability of female labor force participation reduces sharply with the number of children while male labor force participation is stable. The gender gap in LFP for working-age population (20-64) is 7.9 percentage points larger for individuals with 2 children than for those with no children, conditional of educational level and age. For prime age individuals (25-54) the gender gap is 10.3 percentage points larger for those with two children than for those with no children. This may suggest that gender difference in child rearing responsibilities might be an important determinant of gender gap in labor force participation. Figure 12. Probability of being in labor force in Croatia as a function of the number of children, 2002–18(percent) Panel a. Age 20-64 Panel b. Age 25-54 Source: EU-LFS 2002-2018. Note: The probabilities are estimated with logit model controlling for year dummies, age group dummies, educational level dummies, and marital status. Standard errors are robust. About 50 percent of women report that they have had to interrupt their career (including maternity/parental leave or other forms of carrier interruptions) for at least 1 month in their employment history to take care of own or partner's children. For men this share constitutes just 1.5 percent. Among women who have interrupted their career, 71.2 percent of women stayed out of work for more than a year. For men this share is just 12.1 percent. Interestingly, among women 98.1 percent used parental leave and/or maternity as part of work interruption for childcare while among men 47.2 percent did not use any family leave. This may indicate that men in Croatia are not willing to use parental leave or that it is not easily available for them (Figure 13). 26 3.1.2 How Maternal Labor Force Participation Varies with the Number of Hours Children Go to School In this section, we aim to understand how the labor force participation of females with school-age children varies with the number of hours children go to school in Croatia. Primary schools in Croatia usually operate in a single shift (the half-day shift, the extended shift or the whole-day shift), but they are also allowed to operate in more than one shift (double shifts, and, exceptionally, in three shifts). Availability of a longer school schedule can affect females’ decisions to participate in the labor market by effectively substituting for childcare. This note analyzes the differences in gender labor market dynamics in households with compulsory-age children as well as the differential effects of the two different schooling programs on the female labor market. This analysis could be useful in the context of a potential school reform that would introduce the extended school schedule (rolling out Whole-Day-School program). We used pooled data from the European Union Statistics on Income and Living conditions (EU-SILC) between 2014 to 2017. Datasets provided information on children’s age as well as the hours they spend at school. The analysis focused on households with compulsory school-aged children and were classified into two categories by the average number of hours these children attend school: those where children attend 5 hours or less and those that attend 5 hours or more. We use logit regression models to estimate the gender differences in the probabilities of being employed, unemployed, having a temporary job and participating in the labor force for the two schooling programs. To understand the differences in the two programs on the gender income gap, hours worked and fathers’ responses to mothers’ shifts in labor supply we use OLS regressions with year fixed effects, controlling for adult demographic characteristics as well as the number of compulsory-aged children in the household. Our results show that females with children attending longer school days face a lower likelihood of unemployment and higher likelihood of employment, while males do not show significant differences in these indicators. In 2017, females with children who attended school for longer hours had an employment to population ratio of 60 percent compared to 55 percent for females with children with shorter school schedules. The unemployment rate was also significantly lower for the first group with 18 percent in comparison to 27 percent for the latter. Also, 22 percent of women with children in shorter school schedules had temporary jobs when compared to 17 percent of those with children in longer school schedules. When looking at women with similar demographic characteristics, results from pooled logit models show that females with compulsory- aged children attending school for less than 5 hours are on average 8.6 percent more likely to be unemployed, and 6.0 percent less likely to be employed. Conversely, there are no differences on labor market indicators for males with compulsory-aged children. 27 Box 1. Data Limitations: EU-SILC and EU-LFS • The only household-level data available to measure labor force participation among mothers with children ages 6-12, together with the hours spent at school is the EU-SILC data. In this data, comparisons can be made between mothers of children of compulsory school-age that spent 7 or 8 hours of school, with children that spend 4/5 hours of school. There is no information of hours for children older than 12, which likely excludes the students in lower secondary • The EU-SILC is representative at the NUTS1-1 level as well as at the NUTS-2 (Adriatic and Continental). Analysis of indicators at the level of NUTS3 (21 counties) or LAU (556 municipalities) is not reliable. To get reliable estimates of LFP at NUTS3 level, small area estimates of labor market indicators in a similar fashion to the poverty maps are needed (and not currently available. • The EU-LFS survey does not contain data on hours children spent at school, so it cannot be used for such an analysis. Besides, this data is representative also at the level on NUTS2. • With this data, we can use a quasi-experimental design to understand impacts of time children spent at school on LFP, while controlling for other observables. However, since the schooling decisions are endogenous and depend on many unobserved factors (such as preferences of women for certain child arrangements), to properly measure impact probably there is a need for instrumental variables approach. • An alternative method to measure ex-ante impacts is by a full impact evaluation taking advantage the of phased rollout of the school reform for an impact evaluation (experimental rather than quasi-experimental). The estimation strategy is essentially a comparison of relevant outcomes (or changes in those outcomes) between two groups, the treatment group and the control group. The treatment group will be defined as those households with children affected by the school reform (the treatment) as a result of the intervention. The control group is chosen to be as similar as possible to the treatment group, in terms of both observable and unobservable characteristics. In this case, a baseline survey should be available so that double difference estimates can be made. This is ideal since the evaluation is conducted ex-ante and not ex-post (before the reform is implemented). Alternatively, administrative data can tentatively be used. For example, a project data set could be collected and used for the treatment group and the national data set to establish the control using a matching method. The number of compulsory aged children also has a similar differential effect on women’s labor market dynamics, while these are not present in males. While having more compulsory-aged children in the household increases the likelihood of being unemployed, having a temporary job, while decreasing the likelihood of being employed and participating in the labor force for women, there is no effect of having more children on men’s labor market indicators. These results are consistent with the prior that women disproportionately bear the burden of childcare, while men’s labor supply is not significantly affected by these factors (see Annex Table 2 for details). Females with children in longer school days tend to work more hours and earn higher income than those with children in shorter school days. Results show that working-aged women with children attending 28 longer school days work on average 3.6 more hours per week and earn 7.5 percent more labor income than those with shorter school schedules for women of similar educational background, age and number of children. These differences do not hold for working-aged males with compulsory-aged children (see Table 3 in Annex 1 for details). While the labor income gender gap is significant, it is wider when children attend shorter school days. The conjunction between the labor income gap between women from the two schooling programs and the unaffected male income levels translates into a subsequent gender income gap. Results show that in the context of children attending school for more than 5 hours, women’s labor income is 3.2 percent lower than men in the same schooling category, while the gap rises to 3.5 percent in the case where children attend shorter school days, when controlling for similar levels of education, age and compulsory-aged children (See Table 4 in Annex 1 for details). Fathers’ labor supply slightly declines with an increase in mothers’ working hours and more so in the context of longer school days. In terms of intrahousehold labor supply allocation, the work analyzed whether an increase in mothers’ labor supply would translate into a subsequent decrease in fathers’ working hours. Results indicate increases in mothers’ working hours are accompanied by a decrease in the number of hours fathers work. This effect is larger in the context of longer school days, where fathers worked 0.19 hours less on average, for each additional hour worked by the mother, while this effect shifted to 0.29 hours less for households with children attending longer school days (See Table 5 in Annex 1 for details). From the large portion of households that are not currently enrolled in the whole day school program, many are from lower income households and on average have a greater number of children. Households that currently have children attending shorter school days would be the target group for an expansion of a Whole-Day-School program. These compose a large share of households with children. In 2017, 72 percent of households and 69.0 percent of mothers of compulsory-aged children had children attending shorter school days. Potential beneficiaries are also disproportionately from low-income households. For 2017, 27.3 percent of households with children attending school for 5 hours or less belong to the poorest 20 percent of the population, while only 12.3 percent belong to the richest 20 percent of the population. Also, the average number of children of a household with compulsory-aged children is higher for those attending school for 5 hours or less for all survey years. In 2017 households with 5 or less school hours had 1.45 children, those attending more than 5 hours had 1.22 children on average. Females with children attending longer school days face a lower likelihood of unemployment and higher likelihood of employment, while males do not show significant differences in these indicators. In 2017, females with children who attended school for longer hours had an employment to population ratio of 60 percent compared to 55 percent for females with children with shorter school schedules. The 29 unemployment rate was also significantly lower for the first group with 18 percent in comparison to 27 percent for the latter. Also, 22 percent of women with children in shorter school schedules had temporary jobs when compared to 17 percent of those with children in longer school schedules. When looking at women with similar demographic characteristics, results from analytical models demonstrate that females with compulsory-aged children attending school for less than 5 hours are on average 8.6 percent more likely to be unemployed, and 6.0 percent less likely to be employed. Conversely, there are no differences on labor market indicators for males with compulsory-aged children. The number of compulsory aged children also has a similar differential effect on women’s labor market dynamics, while these are not present in males. While having more compulsory-aged children in the household increases the likelihood of being unemployed, having a temporary job, while decreasing the likelihood of being employed and participating in the labor force for women, there is no effect of having more children on men’s labor market indicators. These results are consistent with the prior that women disproportionately bear the burden of childcare, while men’s labor supply is not significantly affected by these factors. Females with children in longer school days work more hours and earn higher income than those with children in shorter school days. Results show that working-aged women with children attending longer school days work on average 3.6 more hours per week and earn 7.5 percent more labor income than those with shorter school schedules for women of similar educational background, age and number of children. These differences do not hold for working-aged males with compulsory-aged children. While the labor income gender gap is significant, it is wider when children attend shorter school days. The conjunction between the labor income gap between women from the two schooling programs and the unaffected male income levels translates into a subsequent gender income gap. Results show that in the context of children attending school for more than 5 hours, women’s labor income is 3.2 percent lower than men in the same schooling category, while the gap rises to 3.5 percent in the case where children attend shorter school days, when controlling for similar levels of education, age and compulsory-aged children. Fathers’ labor supply slightly declines with an increase in mothers’ working hours and more so in the context of longer school days. In terms of intrahousehold labor supply allocation, the work analyzed whether an increase in mothers’ labor supply would translate into a subsequent decrease in fathers’ working hours. Results indicate increases in mothers’ working hours are accompanied by a decrease in the number of hours fathers work. This effect is larger in the context of longer school days, where fathers worked 0.19 hours less on average, for each additional hour worked by the mother, while this effect shifted to 0.29 hours less for households with children attending longer school days. 30 From the large portion of households that are not currently enrolled in the whole day school program, many are from lower income households and on average have a greater number of children. Households that currently have children attending shorter school days would be the target group for an expansion of a Whole-Day-School program. These compose a large share of households with children. In 2017, 72 percent of households and 69.0 percent of mothers of compulsory-aged children had children attending shorter school days. Potential beneficiaries are also disproportionately from low-income households, for which whole school programs may be expensive if not subsidized by local governments. For 2017, 27.3 percent of households with children attending school for 5 hours or less belong to the poorest 20 percent of the population, while only 12.3 percent belong to the richest 20 percent of the population. Also, the average number of children of a household with compulsory-aged children is higher for those attending school for 5 hours or less for all survey years. In 2017 households with 5 or less school hours had 1.45 children, those attending more than 5 hours had 1.22 children on average. This gender analysis sets an encouraging benchmark for the rolling-out of a Whole-Day-School program initiative. The program has the potential of targeting a large share of lower income households and those with a greater number of children. Also, households with children attending longer school hours present lower unemployment rates, higher employment, higher working hours and labor income which give a generally positive outlook for women’s labor market status when children attend longer school days and is also beneficial towards a lower labor income gender gap. This sets an encouraging benchmark to further analyze the effects of the program in future stages of implementation. This evidence also shed light into the significant challenges that the Croatian females may be facing in the labor market as a result of the COVID-19 crisis. Large-scale school closures, the already high burden of child and elder care responsibilities, and the prevalent social norms are all factors that may contribute to a higher gender inequality, including a wider gap in labor force participation and wages. 3.1.3 The Impact of Maternity and Paternity Leave Policies Maternity and parental leave policies are designed to protect parental employment for a limited time around childbirth and ensure that parents may return to their previous employer afterwards. On the one hand, by encouraging employment continuity, parental leave policies may promote gender equality and increase women’s earnings. Parental leave policy may also provide incentives to accumulate work experience to qualify for parental leave benefits. Moreover, leave policies may improve the health of the mother and the welfare of the child (Rossin, 2011). On the other hand, very long leave may have negative effect on female labor supply in the long run if it induces women to stay out of work for a long by reducing their experience and human capital (Lalive and Zweimüller, 2009). Moreover, long parental leave induces additional costs on employers (nonwage costs 31 such as, expenses associated with hiring and training temporary replacements), which may lower the demand for female labor and provoke labor market gender discrimination (Ruhm, 1998). The effect of maternity and parental leave policy is likely to interact with other institutional settings. For instance, the effect of an extension in maternal leave may reduce the female employment rate if there is equal pay legislation, since in this case employers may respond to the increase in costs of hiring women by substituting them by male workers. On the other hand, if firms have monopsony power, the costs of family policies may be paid by female workers through reduction in wages or worst work conditions without reducing female employment. Equal paternity and maternity leave policy may reduce incentives to discriminate female workers. In summary, family policies might have ambiguous effect on female employment and wages depending on wage elasticities of labor demand and supply and other labor market conditions. The main goal of reserving a share of the parental leave for fathers is to balance the distribution of childrearing costs within the family, achieve a less gender specialized home production model and foster the labor market prospects of women. Employers may anticipate that fathers will take time off to care for their children an modify their hiring and promotion decisions in favor of female workers (Phelps 1972, Lazear and Rosen 1990). Therefore, a family policy that shifts part of the childcare burden to fathers may positively affect women's longer-term labor market outcomes in terms of employment participation, career achievements and life-time earnings. Figure 13. Career break for childcare in Croatia, 2018 (percent) Panel a. Career Break for Children 70 60.14 60 49.87 50 37.44 40 30 26.25 16.94 20 10 1.53 3.23 0.85 3.72 0.05 0 Yes Never worked; for No (but was/is Never worked for other Never had children childcare reasons employed and has reasons children) Male Female Panel b. Complete length of career break for childcare 32 90 81.23 80 70 60 50 40 34.96 30 23.28 22.30 20 5.51 6.63 7.52 8.62 5.32 10 4.62 0.00 0.00 0 Up to 6 months 6 months-1 year 1-2 years 2-3 years 3-5 years More than 5 years Male Female Panel c. Use of parental leave and/or maternity/paternity as part of work interruption for childcare 100 86.69 90 80 70 60 47.19 48.36 50 40 30 20 11.42 10 0.00 0.00 4.45 1.89 0 Only used parental leave Combination of family Only maternity/paternity No family leave used leaves used Male Female Source: LFS 2018 Additional Module. Age 20–64. 33 Box 2. Current family policies in Croatia Maternity leave is 28 days before the expected day of birth and until the child turns six months of age. The period is mandatory for mothers for 98 days: 28 days before the expected date of delivery and 70 days after the birth without interruption. Maternity leave is compensated at 100 percent of average 6 months earnings with no ceiling on payments (Dobrotić, I., 2019). After that parental leave can be taken and it is four months (120 calendar days) per parent per child for the first- and second-born child. Two months of the leave can be transferred from one parent to the other (if father does not take the leave, mother has 6 months maximum; otherwise it is a total of 8 months with 2 months reserved for fathers). Parental leave is 15 months (450 days) per parent for multiple births, and the third and every subsequent child. Parental leave is compensated at 100 percent of average earnings for the first six (or eight) months, with an upper limit of 120 percent of the budgetary rate (532 euros per month) and at 70 percent of the budgetary base rate (310 euros)19 for the subsequent months (Dobrotić, I., 2019). Maternity and parental leave is also provided to unemployed parents – at a minimum flat rate of 310 euro per month. Paternity leave - exclusively reserved to fathers - does not exist in Croatia. However, fathers can take 7 days leave right after the child’s birth compensated at 100 percent of wage rate (Official Gazette, No. 30/04 -Labor Act, Article 57). Women are entitled to three years job protected leave. International comparison of family policies Croatia has a generous maternity leave policy in comparison with other OECD countries: 30 weeks maternity leave with 100 percent payment rate. However, the absence of compulsory parental leave does not encourage a redistribution of child rearing responsibilities between men and women. The replacement rate for paternity leave (which can be taken by fathers) is poorly compensated in comparison to other European countries, as shown in Figure 14. According to Dobrotić (2019), fathers account for 7.5 percent of all parental leave taken. One of the potential explanations for the low uptake of parental leave by fathers can be the relatively low compensation rate for parental leave, which can be used by fathers. Also, in the case of multiple births or a third child or more, there is no incentive for fathers to take the leave since the overall amount of leave does not change if it is not shared between parents (for first- and second-born children 2 months of parental leave are lost if not taken by fathers). Another explanation can be the prevalence of gender unequal social norms, since women are generally considered in charge of child related activities in the Croatia (Maskalan, 2016; Jugović, I. 2016). 19 This is a lower limit. 34 0 20 40 60 80 20 60 80 40 0 Bulgaria Croatia 100 100.0 Greece Estonia United Kingdom Luxembourg Slovak Republic Poland Croatia 30.0 Lithuania Czech Republic Austria Ireland Hungary Netherlands EU average Spain Italy Israel Estonia Slovenia Luxembourg Germany Poland Portugal Panel b. Length (weeks) of paid maternity leave OECD average Norway Denmark France Panel a. Average payment rate for maternity leave (%) Lithuania Bulgaria 35 Romania Romania Finland Italy Source: OECD 2018 (http://www.oecd.org/els/family/database.htm) Austria Latvia Figure 14. International comparison of family policies, 2018 France Sweden Latvia Slovak Republic Netherlands Finland Spain Hungary Turkey Iceland Belgium Turkey Israel Belgium Slovenia Czech Republic Germany Switzerland Switzerland Iceland Denmark Norway Greece Sweden United Kingdom Portugal Ireland 10 15 20 25 30 5 0 100 20 40 60 80 0 Israel Israel Slovak Republic Slovak Republic Switzerland Switzerland Greece France for fathers are included. Netherlands United Kingdom Italy Belgium Czech Republic Ireland Hungary Croatia 42.1 Turkey Denmark Latvia Portugal United Kingdom Czech Republic Ireland Finland Denmark Germany 36 Cyprus Iceland Estonia Luxembourg Poland Cyprus Bulgaria Sweden Lithuania Austria Slovenia Latvia Spain Slovenia Romania Panel a. Average payment rate for paid leave reserved for fathers (%) Bulgaria Panel b. Length (weeks) of paid leave reserved for fathers (%) Figure 15. International comparison of paid leave for fathers, 2018 EU average Norway OECD average Estonia Croatia 8.7 Greece Austria Hungary Germany Italy Finland Lithuania Norway Netherlands Iceland Poland Sweden Spain Luxembourg Source: OECD 2018 (http://www.oecd.org/els/family/database.htm). Only countries with at least some payment Belgium Box 3. International Evidence of potential impacts of maternity and paternity leave on female LFP There is evidence that the introduction of parental leave increases women’s employment but has the potentially negative effect of females’ wages, relative to their male counterparts (Ruhm, C. J., 1998; Low and Sánchez-Marcos, 2015). However, the evidence on the effects of the duration, income support, and availability to either parent of parental leave is mixed. To study the effect of parental leave policies on economic outcomes of females, some studies used a country- level approach, which considers interdependency between family policies and captures the general equilibrium effect. It is however problematic to interpret the estimates based on country-level data as causal since some unobservable country characteristics may simultaneously affect the legislation of family policies and female labor market outcomes (e.g., gender roles stereotypes). Most studies use a micro-level approach instead, focusing on the effect of a specific policy within a country relying on variation from a natural experiment, which allows identifying the causal effect. This approach in general does not consider policy interdependency and captures only partial equilibrium effect. Cross-country evidence suggests that there is a positive association between women’s employment and the duration of the protected parental leave, however, these relationships are nonmonotonic: female employment rises with job-protected parental leave duration up to some point and declines thereafter (Ruhm, 1998; Thévenon and Solaz, 2013; Olivetti and Petrongolo, 2017). Quantitatively, however, the estimated gain in employment associated with parental leave duration is small. Olivetti and Petrongolo (2017) also find a negative association between female employment and the percentage of total leave that is paid. In addition, Ruhm (1998) and Thévenon and Solaz (2013) find that an increase in parental leave duration widens the male-female earnings gap; this finding is not confirmed in Olivetti and Petrongolo (2017) who find the opposite for earnings gender gap. Many micro-level studies have found that an extension in job-protected leave causes women to delay their return to work but has no long-run effect on female labor market outcomes (Lalive and Zweimüller (2009) used the Austrian reform; Dustmann and Schönberg (2012) and Schonberg and Ludsteck (2014) provides the results for Germany). Dahl et.al (2016) show that the increase in the duration of paid maternity leave in Norway did not have any significant short and long-run effect on females’ labor market outcomes. Kluve and Tamm (2012) show that a change in parental leave regulations from a longer and less generous parental leave (24 months at a flat rate of 300 Euros) to a shorter and more generous leave (14 months with 2 months reserved for fathers and 67 percent replacement rate) increased the probability that mothers re- enter the labor force 1.5 years after the birth of the child and increased the take-up rate of fathers in Germany. The effect of maternity leave is likely to vary by women’s educational level. The effects of leave entitlement on female employment are beneficial for less-skilled women, while the earning of high-skill women fall because of longer leave. While there is evidence that parental leave reserved for fathers is associated with parental participation in paternity leave, most studies do not find a significant long-run effect of paternity leave on maternal labor market outcomes. Farré and González (2017) study the effect of the introduction of 2 weeks paid (at 100 percent wage replacement rate) "use-it or lose-it" parental leave reserved only for fathers in Spain in 2007 on fertility and labor market outcomes. They suggest that the reform was associated with increased participation of fathers in paternity leave and an increase in maternal employment probability a few months after giving birth caused by the parental leave reform, but the long-run effect is not significant (24 months after birth). The affected 37 mothers were less likely to take extended unpaid parental leave from the job they held at the time of birth. However, there is no evidence that fathers involve more in childrearing activities as a result of the reform Ekberg et al. (2013) analyze the effect of the Swedish “Daddy-Month” reform, which attempted to increase fathers' share of parental leave. Their results suggest that fathers in Sweden increased leave-taking as a result of the reform, but it did not transform into a long-run change in parental behavior and labor market outcomes. Cools et. al. (2015) confirm these results for Norway. Quantitative analysis for Croatia and other EU countries We conduct a cross-country analysis to study the association between several measures of parental and maternity leave take-off based on EU-LFS. The main advantage of cross-country analysis with respect to the micro-level analysis is that it considers the global equilibrium effect of family policy and that it allows capturing long-run effect. Moreover, in Croatia, there was no significant increase in maternity or paternity leave duration or benefit during the past decade, which would provide a natural experiment framework and allow us to conduct a micro-level analysis for Croatia (In Annex 2 we describe several changes in maternity and parental leave policy). For this analysis, we use data from EU Labor Force Survey (LFS) for years 2006-2018. The respondents of LFS were asked “What is the reason for not having worked at all during the reference week though having a job?” and they could answer “Maternity leave” and “Parental leave”. “Parental leave” option was included since 2006 and it was not included in some countries, therefore we analyze the period from 2006 to 2017 and we exclude several countries that did not have a “Parental leave” option in the survey. Using this question, we compute the share of female and male workers who report that they were on paternity or maternity leave during the reference week. We restrict our analysis for individuals between 25 and 44 years old not in retirement. Figure 16 shows that about 6 percent of female workers were on maternity and parental leave in 2018. This share is comparable to the EU average. 38 Figure 16. Share of employed women on maternity leave. 12 10 8 6.3 6 4 2 0 Source: EU-LFS 2018. Age:25-44. Retired individuals are excluded Among all workers who report being on maternity or parental leave was only 1.6 percent were males, which is one of the lowest shares among EU countries, see Figure 17. For comparison, in Sweden, this share constituted 28.0 percent and in Portugal 11.7 percent. In fact, Sweden and Portugal are EU countries with the highest full-rate equivalent of paid leave reserved for fathers (number of weeks reserved times payment rate) as it is shown in Figure 18. This may suggest that both the duration and the payment rate are important determinants of paternal use of paternity leave. Figure 17. Share of males among parents on parental leave 30 25 20 Percentage 15 10 5 1.60 0 Source: EU-LFS 2018. Age:25-44. Retired individuals are excluded 39 Figure 18. Full-rate equivalent (weeks) of paid parental leave reserved for fathers 14 12 10 8 weeks 6 3.6 4 2 0 Source: OECD 2018 (http://www.oecd.org/els/family/database.htm) To analyze the association between paternal and maternal participation in the gender gap in labor force participation and employment, we follow Olivetti and Petrongolo (2017). We regress different country-level labor market indicators of EU countries on (i) Share of males among workers on paternity leave; (ii) Share of workers on parental leave; (iii) Share of female workers on parental leave; (iv) Share of male workers on parental leave. First, we run simple regressions with no controls (see Panel A of Table 6 in Annex 1 for details), then we add country fixed effect to control for static country-specific characteristics, like geographical difference (see Panel B of Table 6 in Annex 1) and finally we add country and year fixed effect to control for country-specific characteristics and common time trend (see Panel C of Table 6 in Annex 1). The last specification is the most robust since it captures time variation in cultural norms, birth rates, etc. Column 1 of Table 6 shows that there is a strong negative association between the share of males among workers on paternity leave and the male-female gender gap in LFP. Particularly, a 1 percentage point increase in the share of males among workers on paternity leave decreases the male-female gender gap in LFP by 0.37 percentage points. This effect reduces to 0.19 percentage points when we add country and year fixed effect but remains statistically significant. Similarly, 1 percentage point higher share of males among workers on paternity leave is associated with a 0.20 percentage point decrease in the male-female gender gap in the employment rate. Column 2 of Table 6 suggests that a higher share of workers on parental leave is also associated with the lower male-female ratio in labor force participation, but this effect becomes not significant when country and year fixed effects are included. Similarly, the share of females on parental leave (Column 3) is not significantly associated with the gender gap in LFP and employment rate once county and year fixed effects are included. In contrast, Column 4 shows a significant and strong association between the share of male workers on parental leave and the male-female gender gap in LFP, female LFP, male-female gender gap in 40 employment, and female employment even when country and year fixed effects are controlled for. Particularly, in 1 percentage point increase in the share of male workers on parental leave is associated with 3.0 percentage points reduction in male-female gender gap and 4.4 percentage points reduction in the male-female gender gap. It is also associated with 2.4 percentage points increase in female labor force participation and 2.5 percentage points increase in the female employment rate. Therefore, our quantitative results demonstrate potential effectiveness in the increase of male’s enrollment in paternity leave as a tool to increase female LFP and employment. There is a clear tendency that in countries where parental leave systems were transforming into more gender-neutral (Sweden, Portugal, or Finland) the share of fathers who participate in parental leave has grown substantially during the past decade, see Figure 19. For instance, in Portugal, there were only 3.1 percent of males among parents on parental leave in 2006 while it was 17.8 percent in 2017 (now Portugal has one of the most generous conditions for fathers on parental leave in EU). In contrast, in Croatia, this share was stable since 2006 at about 1-2 percent. Transformation of the parental leave system into more gender-neutral might be an effective tool for reducing the gender gap in LFP. Our results suggest that by increasing the share of males in parental leave from the current 1.6 percent to EU average of 6.7 percent might decrease the gender gap in LFP in Croatia by 1 percental points which are 11.5 percent of the overall gender gap in LFP (the gender gap in LFP for 25-44 years old was 8.7 percent in 2018). An increase of this share to 15 percent could reduce the gender gap in LFP by 2.5 percentage points (29.2 percent of the overall gap). This transformation requires not only an increase in the duration of parental leave reserved to fathers but also an increase in payment rate, which is quite low in Croatia in comparison to other EU countries. Figure 19. Share of males among parents on parental leave 30 25 20 15 10 5 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 PRT SWE FIN HRV Source: EU-LFS 2006-2018. Age:25-44. Retired individuals are excluded. 41 3.1.4 Elder Care and Female Labor Force Participation In this section, we analyze whether eldercare responsibility is an important barrier to women’s economic activity. Using the data from 2018 LFS we estimate the logit model where we regress the probability to be active on the labor market on the indicator that there are elder relatives in the household, as well as on indicators for educational level and age group. The results suggest that prime-age females (25-54 years old) who live in households with elderly relatives are 2.7 percentage points less likely to participate in the labor market than those who do not have elderly relatives in the household, conditioning on the level of education and age.20 I n contrast, for men, there is no significant association between the presence of elderly relatives and inactivity. This may suggest that prime-age women have more eldercare responsibilities and that this may prevent them from being economically active and increase gender differences in labor force participation. Figure 20: Career break for incapacitated relatives Female Male 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Work interruption Only reduced working time No interruption or reduction Source: LFS 2018 Additional Module. Age: 20-64 Next, we analyze whether, among people who have an elder or incapacitated relatives' care responsibility, the share of those who had to interrupt the career is larger for women than for men. Figure 20 demonstrates that 5.2 percent of women vs. 2.2 percent of men with incapacitated relatives' care responsibilities had to interrupt work. On the other hand, the fraction of women who had career interruption is not large, it is significantly larger for women than for men. 3.2 Labor Market Regulations The institutional framework that governs the functioning of the labor market can provide incentives or disincentives for women to participate in the labor force. Within the EU, there is substantial variation 20 The effect is not significantly different from zero for working-age females (20-64 years old). 42 in these labor market regulations, features, and policies both across countries and across time, providing sources of variation that can be exploited to determine the relevance and magnitude of these institutional factors of interest on the female labor force participation rate. To analyze the effect of these institutional labor market factors, we exploit this source of variation and regress the country-level female labor force participation rate on country-level measures of institutional labor market factors and other controls for Croatia and other EU countries for the period 2006-2017. In the analysis that follows (see Table 7 in the Annex), the female labor force participation rate— the proportion of active women 15-64—is modeled as a function of (i.) financial disincentives to move from inactivity to employment—the inactivity trap for women earning 67 percent of the average wage, which measures the part of additional gross wage that is taxed away in the case where an inactive person (not entitled to receive unemployment benefits but eligible for income-tested social assistance) takes up a job (Eurostat Tax and Benefits Database); (ii.) labor market features—the lagged proportion of part-time employment and temporary employment, the lagged percent employment in services and the lagged unemployment rate; (iii.) general economic conditions in the country—GDP per capita growth; (iv.) the potential care burden—the age dependency ratio; (v.) the composition of the female workforce—the proportion of females with tertiary education; (vi.) active labor market policies—labor market measures expenditure per thousand unemployed; (vii.) labor market rigidities—ease of hiring and firing and (viii.) country and time fixed effects to control for unobserved factors across countries and time. Since the financial disincentives to participate in the workforce can differ across earnings and family types, we consider the inactivity trap associated with 4 main family structures: Column (1) considers single parents with two children, Column (2) considers a dual-earner couple with two children where the primary earner earns 100 percent of the average wage, Column (3) considers single parents without children and Column (4) considers a dual-earner couple without children where the primary earner earns 100 percent of the average wage. For single individuals with children and a couple with children, the inactivity trap is associated with a decline in the female labor force participation rate, most notably for a couple with two children. A 1 percentage point increase in the inactivity trap is associated with a 0.03 percentage point reduction in the female labor force participation rate for single individuals with children and a 0.10 percentage point reduction in the female labor force participation rate for couples with children signaling the importance of financial disincentives to move from inactivity to employment in determining female labor force participation. The effect may be stronger for a couple with children as women are disproportionately more likely to be second earners and absent for other family structures due to a lower proportion of women in these family structures. Across all specifications, the lagged proportion of part-time employment and the lagged proportion of employment in services is associated with an increase in the female labor force participation rate. 43 A 1 percentage point increase in the lagged proportion of part-time employment is associated with an increase of between 0.34 and 0.45 percentage points in the female labor force participation rate depending on family structure. The comparable range for the lagged proportion of employment in services is 0.44-0.45. This result could be due to the flexibility associated with part-time contracts facilitating greater female labor force participation and the percent employment in services capturing the availability of the types of jobs that women disproportionately participate in. Across all specifications, higher spending on active labor market policies and the ease of hiring and firing individuals is associated with higher levels of activity. A one-unit increase in labor market measures expenditure per thousand of unemployed is associated with a 0.30-0.35 percentage point increase in the FLPR. A one-unit increase in the ease of hiring and firing is associated with a 0.63-0.85 percentage point increase in the female labor force participation rate. This is likely due to reduced labor market frictions and rigidities that may prevent women from finding jobs as well as improvements in available opportunities and the probability and quality of job matches. Guided by the results of the cross-country regression analysis and the knowledge of the institutional framework governing the labor market in Croatia, the most salient constraints to female labor force participation can be identified. In the discussions that follow, we examine the institutional structure of the labor market in Croatia and its potential effect on the female labor force participation. Fiscal constraints associated with the loss of benefits and the incidence of taxation as the woman becomes employed can disincentivize a women’s labor force participation in the so-called inactivity trap. Figure 21 compares the inactivity trap in Croatia and the EU-28 for single individuals with two children and a dual-earner couple where the primary earner earns 100 percent of the average wage. As can be seen from the figures below, the inactivity trap in Croatia is generally higher than the EU-average, most notably for single individuals with two children with earnings below 67 percent of the average wage where the inactivity trap can be as high as 127 percent, more than twice the EU-average, indicating strong financial disincentives to work. The high burden of the inactivity trap is reflected in Croatia’s low ranking on the effect of taxation on incentives to work in the 2017 World Economic Forum’s Global Competitiveness Report where Croatia scored 2.6, where 1 is significantly reduced incentives to work and 7 does not reduce the incentive to work at all, ranking 133 out of 137 countries and last in the EU. To pinpoint the source of the inactivity trap, it is useful to disaggregate the inactivity trap into its benefit and tax components for the respective family types. Figure 22 compares the components of the inactivity trap in Croatia for the 2 family structures. For second earners earning 67 percent of the average wage, the inactivity trap is due to a reduction in child allowance and the mandatory 20 percent pension social security contributions. For single individuals with two children, the inactivity trap is largely due to the reduction of the 44 guaranteed minimum benefit,21 the housing benefit22, the 20 percent deduction for pension social security contributions, and to a lesser extent the child allowance. Whereas the guaranteed minimum benefit, and by extension the housing benefit is phased out over time, the child allowance benefit is means-tested with no special provisions for individuals starting a new job as employment income is included as part of the means test (The OECD Tax-Benefit Model for Croatia: Description of policy rules for 2018). This abrupt withdrawal of the child allowance benefits combined with the incidence of social security contributions could be disincentivizing women’s participation in the labor force. Figure 21. Inactivity trap by select family types, Croatia and the EU-28, 2018 Source: Eurostat tax and benefits database. Second Earner METR is computed based on a two-earner couple with a spouse earner at 100 percent of the average wage. The incidence of part-time employment is low in Croatia, for both male and female; in 2018, for women aged 20-64, Croatia had the joint third lowest proportion of female part-time employment at 6.6 percent of employment, compared with the EU-28 average of 30.8 percent. Part-time employment contracts offer flexibility to women, particularly women with care responsibilities and other constraints that may prevent them from engaging in full-time employment. In Croatia in 2018, among women 25-49, 45.7 percent of them reported having to care for their own or their partner’s children whereas, among women 50-64, 15.8 percent of them reported having to care for an incapacitated relative. The low incidence of part-time employment combined 21 The benefit is phased out over time upon entering employment, 0 percent in the first month, 25 percent in the second month, 50 percent in the third month and completely withdrawn in the fourth month if the average income exceeds the guaranteed minimum benefit. (The OECD Tax-Benefit Model for Croatia: Description of policy rules for 2018). 22 Individuals are entitled to this benefit if they are beneficiaries of the guaranteed minimum benefit. Individuals can continue to claim this benefit whilst employed as long as they remain beneficiaries of the guaranteed minimum benefit. (The OECD Tax-Benefit Model for Croatia: Description of policy rules for 2018). 45 with the relatively high burden of care likely constrains women’s ability to participate in the workforce in Croatia. Figure 22. Inactivity trap by tax and benefit components for select family types in Croatia, 2018 80 Percentage of gross wage taxed upon 70 60 entering employment 50 40 30 20 10 0 Single with two children Couple with two children Guaranteed Minimum Income Housing Benefit Child Allowance Social Security Contributions Income Tax Source: Eurostat tax and benefits database. Second Earner METR is computed based on a two-earner couple with a spouse earner at 100 percent of the average wage. In 2017, Croatia ranked 21 out of the EU-28 countries in expenditure on labor market measures per thousand unemployed indicating a low level of spending on active labor market policies. This was reflected in their ranking of 75 out of 140 countries, and third to last in the EU-28 in the 2018 World Economic Forum’s Global Competitiveness Report’s measure of the effectiveness of active labor market policies. Croatia scored 3.1, where 1 signifies that the policies are unhelpful in helping unemployed people find employment, and 7 signifies that the policies are helpful to a great extent. Spending on active labor market policies can help reduce search costs, facilitate job creation, and improve skills and training, resulting in increases in the probability and the quality of matches between firms and employees, reducing friction in the labor market and producing gains in the allocation of labor. The low level of spending on these measures to facilitate employment could be hampering the participation of women in the labor force, particularly those who have been out of the workforce for an extended period of time and may require additional help navigating reentry. In the 2018 World Economic Forum’s Global Competitiveness Report, Croatia ranked 135 out of 140 countries on the ease of hiring and firing and is ranked last in the EU. On a scale of 1 to 7, where 1 is heavily impeded by regulations and 7 is extremely flexible, Croatia scored a 2.6. In 2015, Croatia also recorded stricter levels of employment protection than the OECD average for both individual dismissals (regular contracts) (2.48 vs 2.12) and temporary contracts (2.00 vs 1.68). Excessively high levels of employment protection in Croatia has been shown to be associated with low levels of hiring and lower levels of job creation than job destruction making it hard for new entrants to successfully enter the labor force (Rutkowski, 2003). . 46 In addition to labor market rigidities, this rigidity contributes to inefficiencies in the labor market due to difficulties in allocating and reallocating labor to its most productive uses. Such labor market rigidities are likely adversely affecting the female labor force participation in Croatia, particularly for those who are seeking to reenter the labor force after an extended absence. Given the institutional constraints identified above, policies aimed at addressing these constraints can lead to improvements in the female labor force participation in Croatia. An increase in the availability and quality of flexible work such as telework, flexi-time, and part-time contracts could increase the female labor force participation, allowing for greatly needed flexibility, particularly for women with care responsibilities. Reducing frictions in the labor market both through making it easier to hire and fire individuals by reforming employment regulations as well as improving the probability and quality of employment matches through increased spending on active labor market policies can lead to improvements in the female labor force participation, particularly for women who have been out of the labor force for extended periods of time who may face high barriers to reentry. Phasing out of the child-care benefit in a manner similar to the guaranteed minimum benefit can reduce financial disincentives for women to work by smoothing the transition from inactivity to employment. 3.3 Gender Differences in Retirement Age The gender ratio in labor force participation increases substantially after 55 years old. Only 48.5 percent of 55-64 years old women were active in 2019, while among men from the same age group labor force participation was about 12 percentage points higher. One of the potential explanations for this pattern could be gender differences in retirement age. Figure 23: Male-female ratio in LFP by age 4.0 3.6 3.5 3.0 2.5 2.0 1.8 1.6 1.5 1.3 1.4 1.3 1.5 1.1 1.1 1.1 1.1 1.1 1.0 1.0 0.5 0.0 15 to 19 20 to 24 25 to 29 30 to 34 35 to 39 40 to 44 45 to 49 50 to 54 55 to 59 60 to 64 65 to 69 65 to 74 70 to 74 Source: Eurostat estimates based on Croatia EU-LFS 2019. 47 Currently, the retirement age in Croatia is lower for women than for men (62 vs. 65) and the share of retired women among 55-64 years old is higher than the share of retired men by about 6 percentage points. This difference has decreased substantially since 2010 when 55-64 years old men were 13.1 percentage points less likely to be retired when women from the same age category and when the women retirement age was 60 (vs. 65 for men). The retirement age for women is being raised by 4 months per year since 2010 in order to equalize it with the retirement age for men (65) by 2027. The retirement age for both men and women will be then raised to 67 years by 2033. Figure 24: Decomposition of population by main activity status. The difference in retirement taking between men and women cannot entirely explain large gender differences in LFP among 55-64 years old. Panel a. Men 85-89 99.3% 80-84 99.8% 75-79 98.5% 70-74 98.7% 65-69 94.2% 60-64 51.2% 55-59 22.3% 50-54 18.6% 45-49 14.1% 40-44 3.3% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Panel b. Women 85-89 4.7% 92.3% 80-84 9.8% 88.3% 75-79 8.5% 90.6% 70-74 9.0% 90.4% 65-69 7.5% 89.5% 60-64 12.9% 60.5% 55-59 16.0% 24.7% 50-54 13.5% 7.1% 45-49 11.5% 2.2% 40-44 8.1%1.8% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Active Domestic Tasks Retired Other Inactive Source: Croatia LFS 2018 48 Differences in retirement taking can explain about half of the gender gap in LFP between 60-64 years old individuals. Another half can be explained by the differences in domestic responsibilities. About 13 percent of women between 60 and 64 reports to be inactive because of domestic responsibilities. Therefore, pension reform might decrease the gender gap in labor force participation for the affected age group, but it is unlikely that it would close the gap entirely. More specifically, if the share of retired women aged 60-64 was equal to the share of retired men from the same age group, aggregate female labor force participation would increase by about 1 percentage point. In order to produce a noticeable increase, pension reform should be complemented by the policies aimed to improve the affordability of child and eldercare. 3.4 Social Norms Using an Epidemiological approach The need to better conciliate work and family lives can only be expected to grow in the future; the opportunity costs of childbearing and childrearing for women have been traditionally high in Croatia given their high educational attainment and labor market inclusion. These may have increased in recent years. Despite the evidence of slowly modernizing views over gender roles in childrearing and housework, Croatian women still spend much more time on these activities than men. Figure 23 shows that the share of women fulfilling domestic tasks and inactive decreased by about 4.5 percentage points in the past 15 years. However, while for men this share was negligible, for females this share was still substantial (8.8 percent) in 2018. In this section we analyze the importance of gender norms---that according to the is standards and expectations to which women and men generally conform, within a range that defines a particular society, culture, and community at that point in time (according to United Nations Statistics Division) --- in explaining female labor force participation. To analyze gender norms and its association with labor force participation in Croatia we use data from European Value Study (EVS) 2017 (latest year available) Integrated Dataset, which collects data for 30 participating countries including Croatia. The survey includes a rich set of questions related to beliefs regarding gender norms and personal characteristics.23 Moreover, it includes rich information on both parents, 23 To analyze gender norms, we use the following questions: For each of the following statements I read out, can you tell me how strongly you agree or disagree with each. Do you strongly agree, agree, disagree, or strongly disagree? 1. When a mother works for pay, the children suffer 2. A job is alright but what most women really want is a home and children 3. All in all, family life suffers when the woman has a full-time job 4. A man's job is to earn money; a woman's job is to look after the home and family 5. On the whole, men make better political leaders than women do 6. University education is more important for a boy than for a girl 7. On the whole, men make better business executives than women do When jobs are scarce, men have more right to a job than women 49 including parental education and parental country of birth. The sample for Croatia includes 623 males and 865 females older than 18. Figure 25: Share of inactive women fulfilling domestic tasks (Percent of women aged 20-64). 14 12 10 Percentage 8 6 4 2 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Source: Croatia LFS 2002-2018. Women between 20 and 64. Box 4. International evidence on the role play by cultural norms There is a consensus that the relaxation of traditional gender norms has contributed to the reduction in the gender gap in labor market outcomes. Antecol (2000, 2001) and Fernandez and Fogli (2009) analyze labor market behavior of first and second-generation immigrants and find that a significant the overall variation in the gender gap in LFP is explained by country of ancestry LFP gap, providing evidence that culture has a strong effect on labor market outcomes. In addition, Fortin (2005) provides cross-country evidence for OECD countries on the role played by culture showing that anti-egalitarian views have a strong negative association with female employment rates. Rodríguez-Planas, Sanz-de-Galdeano, and Terskaya (2019) show that gender norms of high-school peer mothers affect the gender gap in labor market outcomes providing additional evidence that gender norms are an important determinant on individual behaviors. There is evidence that culture has a strong effect on other outcomes, including test scores (Pope and Sydnor, 2010; Nollenberger et.al, 2016), risky behaviors (Rodríguez-Planas et al., 2019; Rodríguez-Planas and Sanz-de- Galdeano, 2019), fertility (Fernandez and Fogli, 2009), etc. Figure 26 provides descriptive evidence regarding gender norms in Croatia. A substantial fraction of the population ---20.1 percent--- was agreed with the statement that “man's job is to earn money; a woman's job is to look after the home and family”. The majority of the population (51.0 percent) believes that “A job is alright but what most women really want is a home and children” and 33.9 percent believe that “All in all, family life suffers when the woman has a full-time job”. 50 There is evidence that culture has a strong effect on other outcomes, including test scores (Pope and Sydnor, 2010; Nollenberger et.al, 2016), risky behaviors (Rodríguez-Planas et al., 2019; Rodríguez-Planas and Sanz-de- Galdeano, 2019), fertility (Fernandez and Fogli, 2009), etc. Figure 26: Share of agreed with the statement regarding gender norms jobs are scarce: giving men priority 16.91 men make better business executives than women 9.74 university education more important for a boy than for a girl 2.87 men make better political leaders than women 14.36 man's job is to earn money; woman's job is to look after home 20.13 and family family life suffers when woman has full-time job 33.85 women really want home and children 50.98 child suffers with working mother 31.57 Source: EVS 2017, Croatia. Age: 20-64 Figure 27: Share agreed with the statement "man's job is to earn money; woman's job is to look after home and family" 60 51.35 50 40 30 20.13 20 10 3.68 0 Source: EVS 2017-18. Age: 20-64 Figure 27 compares views regarding gender norms (the share of those agreed with the statement “man's job is to earn money; a woman's job is to look after the home and family” across European countries. It shows that the patriarchal view is more prevalent in Croatia than in 12 other European countries included in the analysis and it is close to the median share. The lowest prevalence of patriarchal views is Sweden and the highest is in Bulgaria. 51 Figure 28 shows that patriarchal views are more prevalent among men than women and that this prevalence increases on age24 and decreases on education level. It also indicates that among active women the share of agreed with the statement “A man's job is to earn money; a woman's job is to look after the home and family” is about 10 percentage points lower than among non-active women but for men, this difference is reversed. Figure 29 shows that the share of women with patriarchal views measured with other questions available in EVS is consistently higher for inactive women than for active women. Figure 28: Percentage of population that agreed with "man's job is to earn money; woman's job is to look after home and family" by group (%) 70 60 57.1 50 45.3 40 31.3 30 24.2 24.3 25.3 25.0 24.9 24.5 21.5 21.6 20 17.3 15.1 14.9 16.6 16.6 13.0 12.5 14.7 11.3 7.9 7.8 10 0 All 20-24 25-29 30-39 40-49 50-64 Primary Secondary Tertiary active inactive female male Source: EVS 2017, Croatia. Age: 20-64 Figure 29: Gender norms beliefs of active and inactive women (%) jobs are scarce: giving men priority 23.3 18.9 men make better business executives than women 9.0 5.6 university education more important for a boy than for… 1.2 1.8 men make better political leaders than women 12.7 10.1 man's job is to earn money; woman's job is to look… 24.5 14.7 family life suffers when woman has full-time job 48.7 29.9 women really want home and children 60.0 49.3 child suffers with working mother 42.5 27.4 inactive active Source: EVS 2017, Croatia. Age: 20-64 24An exception is the group of females aged 25-29, which tend to have more patriarchal views on average than those aged 30-39 and 40.49. 52 To provide a more detailed analysis of determinants of gender norms views and its association with labor market outcome we conduct logistic regression analysis controlling for individuals age and parental education. We do not control for individuals’ educational attainment, marital status and the number of children since all these outcomes are likely to be affected by gender norm beliefs and can be seen are intermediate outcomes. To measure gender norms, we construct a summary index based on eight questions from EVS regarding gender norms described above. We use the first principal component obtained with polychoric principal component analysis, which explains 51 percent of the common variation (See Table 8 in Annex 1). The result of this analysis reported in Table 8 of Annex 1 suggests that 1 standard deviation increase in the patriarchal gender norms index increases the probability of inactivity by 3.6 percentage points for females, holding constant the level of education of parents (to control for family socio-economic status) and age. For males, the effect of the patriarchal gender norms index on inactivity is not significantly different from zero. This indicates that a high prevalence of patriarchal views to some extent might explain the gender gap in labor force participation and that the policies aiming at the relaxation of traditional gender social norms may induce women to participate more in the labor market. Interestingly, when we include controls for own level of education, the estimated effect of gender norms reduces to 2.7 percentage points, suggesting that the effect of gender norms on inactivity might be mediated via educational choices. We also found that one standard deviation increase in the patriarchal gender norms index is associated with an 8-percentage points reduction in the probability that women have tertiary education, holding constant parental education and age. Note that this relationship cannot be interpreted as causal since education may affect gender norm beliefs as well as gender norm beliefs that may affect educational choices. Next, we use an epidemiological approach to analyze the effect of gender norms on female inactivity following Fernandez and Fogli (2009) and applying an epidemiological approach. This approach helps to understand the role of informal institutional constraints (culture or social norms) apart from environmental factors (or formal economic and institutional constraints) in explaining female labor force participation. More specifically, we analyze how female labor force participation in the country of the ancestry of parents affects women’s LFP. To separate the effect of culture from the effect of formal institutions, we focus on second- generation immigrants --- individuals born in the country they live but those parents (at least none of them) born in a different country--- living in the same country. Thus, if only current formal institutional constraints in the host country determine employment decisions, country-of-ancestry female labor force participation should not matter, after controlling for individuals’ sociodemographic and family characteristics. The information on maternal and paternal country of birth asked in EVS allows us to conduct this analysis. However, we do not restrict this analysis only for Croatia given a very small sample of second- 53 generation immigrants. Therefore, for this analysis, we include second-generation immigrants from the following 30 countries: Albania; Armenia; Austria; Azerbaijan; Belarus; Bulgaria; Croatia; Czech Republic; Denmark; Estonia; Finland; France; Georgia; Germany; Great-Britain; Hungary; Iceland; Italy; Lithuania; Netherlands; Norway; Poland; Romania; Russian Federation; Serbia; Slovakia; Slovenia; Sweden; Spain; Switzerland. To proxy gender social norms, we focus on female labor force participation in country-of-ancestry using data from the World Bank Open Data for 2010. Therefore, we match the parental country of ancestry from EVS with these macroeconomic indicators. We estimate the following model for 20-64 years, old men and women: Inactiveiac= β0 + β1 FLFPa+ X’ijcα + λc + εicj (1) Where Inactiveiac is a variable which takes a value 1 if individual i with country of ancestry a living in county c is inactive. FLFPa is female labor force participation rate in country of birth of parents. X’ijc is a set of individual and parental characteristics as well as country of ancestry characteristics. λc is a full set of dummies that control for the individual’s country of residence c. Table 9 of Annex 1 reports the results of the OLS estimation of equation (1) showing that there is a negative association between female labor force participation in the country of birth of parents and second-generation immigrant women’s inactivity (Columns 1-4). For second-generation men, this association is also negative but not significant at 5 percent level (Columns 5-8) of Table 9. We report the results adding sequentially controls for (i) marital status, educational level, and the number of children; (ii) maternal and paternal educational level; (iii) GDP of the country of birth of parental. The results are robust across these specifications and show a similar pattern as our descriptive analysis: there is an association between culture and social norms and women’s employment choices. Our results indicate that although there is some evidence on the relaxation of traditional gender norms, patriarchal views are still prevalent in Croatia and this prevalence is high in comparison with many European countries. We also have shown that patriarchal gender norms are likely contributing to the gender gap in labor force participation. Therefore, policies aiming at shifting social norms (such as mentoring, role models and networks, curriculum revisions, dialogue on gender roles, etc.) might be considered. Findings from economic literature suggest that interpersonal influences ---including family (Alesina and Giuliano, 2010; Brenøe, 2018) and peers (Patacchini, Zenou, 2018, Rodríguez-Planas et al., 2019)- -- are key determinants of adolescents’ gender attitudes and have long term consequences on female labor market decisions. Moreover, these studies indicate that the development of gender conformity happens early in life when family and peers have a stronger influence. Therefore, policies aimed at promoting equitable 54 gender attitudes need to focus on interpersonal relationships and social environments of children or adolescents. 3.5 Occupational Choice: Are Croatian Women Sorting into Low-paying Sectors and Occupations? In this section we aim to answer the following questions: Are women concentrated in low wage/low productivity industries? Is there evidence of sectoral and occupational segregation? Is this the reason behind the wage gap? Is this discouraging females to participate? Figure 30 shows the share of female workers by industry and it also displays the level of wages within each industry (measured as average wage decile of workers employed in this industry in 2018).25 The results show that there is no clear association between the level of wages in the industry and the share of females employed. However, the share of females is low in the top 3 industries: mining and quarrying; information and communication and electricity, gas, steam supply. Figure 30: Share of Females by Sector of Activity 90% 12 80% 10 70% Average wage decil Share of femles 60% 8 50% 6 40% 30% 4 20% 2 10% 0% 0 Share of females Average wage decil in sector Source: Croatia LFS 2018. Women between 20 and 64. The gender gap in wage is unlikely to be explained by females sorting into low-paid industries and there is a gender gap within industries. In fact, the share of females employed in high paid industries is higher than this share for men and constitutes 27.5 percent (vs. 18.2 percent for males). This share is 50.7 25 EU-LFS does not provide exact information on individuals' wages, only the decile of the wage distribution. 55 percent for females with tertiary education vs. 38.8 percent for males. On the other hand, women were more likely to be employed in low-paid industries than men (35.9 percent females vs. 27.3 percent of males) and less likely to be employed in medium-paid industries than men (36.7 percent of females vs. 54.5 percent of males) , see Figure 31. Figure 31: Selection into Industries by Gender. Female Tertiary 15.1% 34.2% 50.7% Male Tertiary 17.1% 44.1% 38.8% Female All 35.9% 36.7% 27.5% Male All 27.3% 54.5% 18.2% 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0% Share in occuparion with av. wages lower than 30th pct. Share in occuparion with av. wages between 30th and 70th pct. Share in occuparion with av. wages higher than 70th pct. Source: Croatia LFS 2018. Women between 20 and 64. Figure 32: Gender gap in wage (in deciles) by industry, 2018 Other service activities Health and social work activities Public administration and defence Professional activities Financial and insurance avtivities Accommodation and food service Wholesale and retail trade; repair Water supply, sewerage, wast Manufacturing Agriculture, forestry and fishing -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 Gender gap in wages Tertiary All 56 Source: Croatia LFS 2018. Women between 20 and 64. In almost all industries the gender gap in wages is positive (males gain on average more than females) and this pattern holds when we compare only individuals with tertiary education (Figure 32). To explain differences in wages between males and females we use Oaxaca Decomposition (Oaxaca 1973), which allows us to explain the gap in wages between males and females. Particularly, we decompose the gap into the part that is due to male-female differences in industry and male-female differences in the effects of these industries on wages. For instance, females may have lower wages because they select low- wage occupations but also because they gain lower wages than males within the same industry. The results in Table 10 in Annex 1 suggest the systematic selection into occupation does not contribute to the gender gap in wages significantly (row “Endowments”) suggesting that females are not disproportionally distributed in low-wage occupations. However, the returns of being in well-paid occupations are lower for women than for men (row “Coefficients”), which may explain the high within- industry wage gender gap.26 Box 5: Gender-Based Occupational segregation: Previous evidence The Croatia gender assessment presents evidence on gender-based occupational segregation, measured as the share of females allocated in a particular sector or occupation, relative to the same share among males (using data from the European Institute for Gender Equality, Gender Statistics Database). The report also presents a raw positive gender wage gaps (using data from Eurostat). The report does not explain whether women or men are indeed clustering into high-wage occupations or sectors, and it does not analyze whether occupation or sectoral segregation is one of the main contributors to the observed gender wage gaps. This is because the statistics presented come from two separate data sources. Our decomposition analysis using the EU-LFS data (which contains data on the sector, occupations, and wage deciles) shows that: 1. Consistent with the previous findings, females self-select into different occupations than men (so there is gender-based occupation segregation); however, these occupations are not lower paid than occupations selected by men; in other words, females are not disproportionately sorting into low-paid 27 industries; indeed, the share of females employed in high paid industries is larger for females than for males 26 Table 10 presents the results for 20-64 years old. The similar patterns is observed for prime-age individuals (25-54). 27 Measured with average deciles as continuous data on wages is unavailable. 57 2. Systematic selection into low wage occupations and sectors does not contribute significantly to the observed wage gap. Oaxaca decompositions of the raw wage gap show that differences in observable characteristics (i.e. occupation and field of education) between males and females explain a small share of the observed gender gap. This means that if Croatian women had the same characteristics as Croatian men, the observed wage gap would only be reduced by a small amount. A large percentage of the gap remains unexplained and can be attributed to differences in “returns” to those characteristics, usually interpreted as discrimination. Similar results are obtained with fields of education (see section 3.6); we found that educated women tend to sort in different fields of education (i.e. disproportionately underrepresented in IT, engineering, manufacturing, and construction), but these are not necessarily the fields with higher wages and employment rates; decompositions show that differences in “returns” explain most of the gender wage differentials 3.6 Educational Choice: Are Croatian Women Sorting into Low-paying Educational Fields? In this section, we analyze whether women are more likely to study fields with lower returns than men. Particularly, we aim to answer the following question: can selection into fields of education explains the gender gap in earnings in Croatia? Is this discouraging females to participate in the labor market? To study this question, we focus on Croatian men and women with tertiary education using a similar approach as in the previous section. 58 Figure 33: Share of females by field of education Panel a. By earnings in field 90% 8.5 80% 8.0 Average wage decil share of females 70% 60% 7.5 50% 7.0 40% 6.5 30% 6.0 20% 10% 5.5 0% 5.0 Share of females Average wage decil in field Panel b. By employment rate in field 90% 35% Share of non-employed 80% 30% Share of females 70% 60% 25% 50% 20% 40% 15% 30% 10% 20% 10% 5% 0% 0% Share of females Share of non-employed Source: Croatia LFS 2018. Women with tertiary education between 20 and 64 There is no clear association between the level of wages and employment rate for the specific field of education and the share of females studying this field, which suggests that females do not select fields of education that yield low returns and with low demand in the labor market. Figure 33 Panel A shows the share of females by field of education and it also displays the level of wages within each educational filed (measured as average wage decile of tertiary-educated workers with this field of education in 2018). It shows 59 that the association between field-specific wages and the share of females in this field is not clear. Figure 33 Panel B depicts the association between the share of individuals not in employment and the share of females within each field. Again, we find that there is no clear association between the employment rates and share of females within each field, which provides suggestive evidence that women do not select fields that have low demand in the labor market. Next, we use Oaxaca Decomposition (Oaxaca 1973) to decompose the gender gap in wages into the part that is due gander difference in selection into the fields of education and the part that is due to male-female differences in the returns to these fields. For instance, females may have lower wages because they select low-wage specializations but also because they gain lower wages than males with the same specializations (See Table 10 in Annex 1 for details). The results suggest the systematic selection into low-paid specializations does not contribute to the gender gap in wages significantly (row “Endowments”) suggesting that females are not disproportionally distributed in low wage fields of education. However, the returns to educational specializations are lower for women than for men (row “Coefficients”), which may explain the high within- specialization wage gender gap. Figure 34: Gender gap in wage (in deciles) by field of education 3.0 9 2.5 8 2.0 7 1.5 Gender gap in wage 6 Average wage decil 1.0 5 0.5 4 0.0 3 -0.5 -1.0 2 -1.5 1 -2.0 0 Gender gap Average wage decil in field Source: Croatia LFS 2018. Women with tertiary education between 20 and 64 Men with tertiary education gain on average higher wages than women with a similar level of education almost in all specializations (except engineering, manufacturing, and construction where only a low fraction of women are employed). This gender gap in wages may discourage women to participate 60 in the market. Therefore, policies that encourage women to select STEM occupations might be not effective in Croatia, since we do find evidence of systematic selection of women into low paid specializations. 3.7 Geographic Mobility Even if substantial international migration takes place in Croatia, improving mobility within regions is also important considering some vulnerable population, including females, that may live in already deprived areas with limited job creation. Limited geographical mobility limits access to high productivity/high wage jobs, which can be a disincentive for participation in certain geographical areas. Some evidence for ECA countries has shown that incentives, demographics, and institutional factors may have played a role. An older and aging population, the social benefits that disincentive the lack of mobility since in many cases they lack portability28, lack of skills and insufficient labor market information, and rigid labor market institutions, and finally, underdeveloped housing and liquidity constraints associated with frictional credit markets may have played a role too. Figure 35. Limited Geographical Mobility: Many workers fail to move to areas with higher job creation potential Share of those unemployed willing to move within country for a job 70 60 50 Percentage 40 30 20 10 0 Kosovo Germany Serbia Greece Moldova Romania FYR Macedonia Kazakhstan Russia Latvia Kyrgyz Rep. Belarus Uzbekistan Czech Rep. Lithuania Georgia Albania Cyprus Italy Bosnia and Herz. Azerbaijan Estonia Bulgaria Armenia Tajikistan Turkey Hungary Slovenia Mongolia Ukraine Slovak Rep. Montenegro Poland Croatia Source: World Bank estimates based on 2016 Life in Transition Survey Source: Staff calculations based on Life In Transition Surveys (EBRD, 2016). The y axis measures the share of those who are unemployed and willing to move within the country for employment reasons More research is necessary to exploit how the extend of internal mobility can distinctive women living in municipalities or regions with less labor market opportunities. 28 Social benefits are sometimes linked to the place of residence and registration. 61 4 Concluding Remarks and Policy Options This report quantifies the most important barriers female face in the Croatian labor market and suggests policy options to address these constraints. Greater female participation in the labor market can potentially yield high long-term benefits for the Croatian economy, which is at risk of demographic crisis because of the aging of the population. The expected decrease of the labor force due to the changes in the age composition of the population puts pressure on the social care systems. Exclusion of the large share of women from the labor force together with the increase in the share of the non-working population due to aging a collapse of social security systems that depend on pay-as-you-go systems. These results also suggest that the current COVID-19 crisis may significantly affect Croatian mothers and lead to an increase in gender inequality. In the event of large-scale school closures, women are expected to carry a large care burden during the crisis, given prevalent social norms and child and eldercare responsibilities. The unequal distribution of unpaid care provision becomes a more binding constraint for female activity rates, as the responsibility for educating children at home may be disproportionately borne by women. In addition, considering the unprecedented challenges that the Croatian females are facing as a result of the current COVID-19 crisis, it is critical to formulating proper policy responses that acknowledge the specific challenges they are facing. In addition to the potential job losses they may face as a result of the economic crisis hitting sectors that disproportionately employ women, Croatian women are bearing the brunt of extra childcare and housework as a result of school closures, the already high burden of child and elder care responsibilities, and the prevalent social norms. Encouraging women’s economic opportunities may yield potentially high benefits for the Croatian economy but it requires policy support. We draw the following conclusion supplemented by the policy recommendations. The gender gap in labor force participation is particularly large for several groups of population and these groups should be primarily targeted by the policies. These are young, low-educated women, women with more than two children and Roma women. Promoting new jobs for low-educated women or providing incentives to women to continue education can reduce the gender gap in employment outcomes given that the gender gap in labor force participation is modest among highly educated. Provision of affordable childcare (including schooling reforms) and eldercare may effectively reduce female inactivity. The lack of affordable child and elder care, together with social norms that dictate that 62 childcare and eldercare are women’s responsibilities, thus represent binding constraints on women during the childbearing years. In fact, the main reason for women’s inactivity in Croatia is child-rearing and other family responsibilities, which is in line with limited availability of institutional childcare and alder care centers in Croatia. Moreover, the gender gap in labor force participation widens significantly on the number of children and when elderly relatives inhabit in the household. International evidence demonstrates that the provision of child and elder care facilities, as well as subsidies for childcare, may increase work attachment by dropping the share of women leaving their jobs after giving birth or by helping women reenter the labor market as their children grow. Extension of enrollment in the whole day school programs may increase the labor force participation of mothers of school-age children. In 2017 most households in Croatia had children attending short school days and this share is especially large among low-income families. We show that women with children attending longer school days face a lower likelihood of unemployment and a higher likelihood of employment, tend to earn a higher income, and to work more hours than those with children attending short school days. This gender analysis sets an encouraging benchmark for the rolling-out of a Whole-Day-School program initiative. Croatia needs to equalize incentives to participate in parental leave between mothers and fathers. The share of fathers participating in parental leave is low in Croatia relative to other EU countries. Moreover, this share was stable during the past decade, while in the countries that switched to a more gender-neutral parental leave system (e.g., Portugal, Sweden), this share has increased. This is consistent with the fact that Croatia provides a general and long maternity leave (on of most generous and long among EU countries) but the incentives for fathers to participate are low since the replacement rates for parental leave are low relative to other EU countries. Gender differences in child rearing responsibilities and parental/maternity leave participation may lower women's labor market activity by inducing women to stay out of work for long and reducing their experience and human capital. Moreover, it may reduce the demand for female workers by imposing higher costs on employers to hire women than men. In line with this, we show that a higher share of men participating in parental leave is associated with the reduction in the gender gap in employment and labor force participation. International evidence suggests that compulsory paternity leave with a generous replacement rate as maternity leave can increase father share of overall parental leave and decrease the gender gap in labor market outcomes in the long run. The reforms aimed at improving financial incentives to women to participate in the labor market by lowering the burden of income taxation and the inactivity trap are needed. Despite recent reforms aimed at relaxing excessively strict employment protection legislation, introducing more flexibility in the labor market, and boosting active labor market policies, Croatia still lags behind its EU counterparts along these dimensions as is reflected in their low ranking on the ease of hiring and firing, a low proportion of flexible forms of 63 employment and low expenditure and coverage of active labor market policies. Additionally, though Croatia has reduced the burden of income taxation for individuals with lower earnings, the inactivity trap often remains above the EU average, primarily due to the abrupt withdrawal of certain benefits when transitioning from inactivity to employment particularly for single individuals with children towards the lower end of the earnings distribution. These constraints are likely to be particularly binding for women who often require flexibility due to care responsibilities and are often are seeking to reenter the labor force after extended periods of absences and may require greater assistance in navigating the transition from inactivity to employment. Improvements in the labor market outcomes of women in Croatia could be seen through a number of labor market reforms such as:, increasing the expenditure on and efficacy of active labor market policies, phasing out the receipt of certain benefits over a longer period of time upon entering employment for low earners, and increasing the quantity and quality of flexible jobs can lead to gains in the participation of women and other previously excluded groups from the labor force as well as gains in the allocation of labor across jobs. The policies incentivizing women to select STEM fields of education to combat gender wage gap or gender gap in labor force participation might be less effective in Croatia than in some other counties since women in Croatia do not sort into low-wage occupations or choose the fields of education that yield lower returns and a higher probability of inactivity than those that men choose. Women earn on average less than men almost in all sectors of activities and almost in all fields of education and this gender gap in wages might discourage women to participate in the market. We conduct the decomposition of the gender gap in wages and the results suggest that women do not sort into lower-wage industries or lower-wage fields of education than men and that the gender gap is likely to be explained by other factors rather than sorting (e.g., wage discrimination). Therefore, lowering within occupation gender gap in wages would make economic activity more attractive for women. These policies should be complemented by the policies aimed at transforming cultural norms29 into more gender-neutral since patriarchal views are still prevalent in Croatia. Our results and evidence from economic literature shows that culture is an important determinant of the gender gap in labor market outcomes. In fact, one of the important determinants of gender convergence is the relaxation of traditional gender norms that happened almost in all countries during the past century. However, patriarchal views are still prevalent even in developed countries. For example, about 34 percent of Croatians believe that family life suffers when women have a full-time job. These views may force women into being home-staying wives and discourage them from labor market participation even when the formal institutions provide them the economic opportunities. 29Changing gender-based cultural norms is hard, but there is a broad research strand on behavioral economics that studies how influencing cultural norms can change behaviors in certain contexts and situations. A systematic review of over 90 empirical studies that have applied behavioral change interventions based on social norms in field settings is presented in Yamin et al (2019). 64 Findings from economic literature show that the development of gender conformity happens early in life when family and peers have a stronger influence and that interpersonal influences are key determinants of gender attitudes. Therefore, policies aimed at promoting equitable gender attitudes need to focus on interpersonal relationships and social environments of children or adolescents. 65 References Alesina, A. and P. Giuliano (2010), The power of the family," Journal of Economic Growth 15, 93-125 Antecol, H. (2000). An examination of cross-country differences in the gender gap in labor force participation rates. Labour Economics, 7(4), 409-426. Antecol, H. (2001). Why is there interethnic variation in the gender wage gap? The role of cultural factors. Journal of Human Resources, 119-143. Arias Diaz, Omar S., Cesar A. Cancho, Maria Eugenia Davalos, Carola A. Gruen, Gady Saiovici, Carolina Sanchez, Indhira Vanessa Santos, Erwin H. R. Tiongson, and Natasha De Andrade Falcao. 2014. Back to Work: Growing with Jobs in Europe and Central Asia (English). Europe and Central Asia Reports. Washington, DC: World Bank Group. http://documents.worldbank.org/curated/en/186521468029365216/Back-to-work- growing-with-jobs-in-Europe-and-Central-Asia Asai, Yukiko, Ryo Kambayashi, and Shintaro Yamaguchi. 2015. "Childcare Availability, Household Structure, and Maternal Employment." Journal of the Japanese and International Economies 38 (December): 172–92. Blau, F. D., & Kahn, L. M. (2013). Female labor supply: Why is the United States falling behind? American Economic Review, 103(3), 251-56. Borra, Cristina. 2010. "Childcare Cost and Spanish Mother’s Labour Force Part icipation." Hacienda Pública Española 194 (3): 9–40. Brenøe, Anne, Origins of Gender Norms: Sibling Gender Composition and Women's Choice of Occupation and Partner (July 2018). University of Zurich, Department of Economics, Working Paper No. 294. Available at SSRN: https://ssrn.com/abstract=3226843 or http://dx.doi.org/10.2139/ssrn.3226843 Borra, Cristina. 2010. "Childcare Cost and Spanish Mother’s Labour Force Participation." Hacienda Pública Española 194 (3): 9–40. Christiansen, L, Ferre, C., Matkovic, T. and Sharafudheen, T. (2019). “Jobs Challenges in Slavonia, Croatia – A Subnational Labor Market Assessment”. World Bank Jobs Working Paper Series No 35. Cools, S., Fiva, J. H., & Kirkebøen, L. J. (2015). Causal effects of paternity leave on children and parents. The Scandinavian Journal of Economics, 117(3), 801-828. Croatia Ministry of Social Policy and Youth. (2014). “Strategy for Combating Poverty and Social Exclusion in the Republic of Croatia, 2014-2020”. Dahl, G. B., Løken, K. V., Mogstad, M., & Salvanes, K. V. (2016). What is the case for paid maternity leave?. Review of Economics and Statistics, 98(4), 655-670. Dobrotić, I. (2019). Croatia country note. in Koslowski, A., Blum, S., Dobrotić, I., Macht, A. and Moss, P. (eds.) International Review of Leave Policies and Research 2019. Available at: http://www.leavenetwork.org/lp_and_r_reports/ Draženović, I., Kunovac, M., and Pripužić, D. (2018). “Dynamics and determinants of emigration: the case of Croatia and the experience of new EU member states”. Public Sector Economics, 42(4), 415-447. 66 Dustmann, C., & Schönberg, U. (2012). Expansions in maternity leave coverage and children's long-term outcomes. American Economic Journal: Applied Economics, 4(3), 190-224. European Commission. (2018a). “The 2018 Ageing Report: Economic and Budgetary Projections for the EU Member States (2016-2070). Institutional Paper No. 79. Luxembourg: Publications Office of the European Union. European Commission. (2018b). “Europe 2020 indicators - Poverty and social exclusion”. European Commission. (2018c). “The 2018 Ageing Report: Economic & Budgetary Projections for the 28 EU Member States (2016-2070)”, Institutional Paper 079, May 2018.European Commission. (2019). “2018 European Semester: Assessment of progress on structural reforms, prevention and correction of macroeconomic imbalances, and results of in-depth reviews under Regulation (EU) No 1176/2011”. EVS (2019): European Values Study 2017: Integrated Dataset (EVS 2017). GESIS Data Archive, Cologne. ZA7500 Data file Version 2.0.0, https://doi.org/10.4232/1.13314 Farré, L., & González, L. (2017). The effects of paternity leave on fertility and labor market outcomes. Fernández, R. (2013). Cultural change as learning: The evolution of female labor force participation over a century. American Economic Review, 103(1), 472-500. Fernandez, R., & Fogli, A. (2009). Culture: An empirical investigation of beliefs, work, and fertility. American economic journal: Macroeconomics, 1(1), 146-77. Fortin, N. M. (2005). Gender role attitudes and the labour-market outcomes of women across OECD countries. oxford review of Economic Policy, 21(3), 416-438. Heathcote, J., Storesletten, K., & Violante, G. L. (2017). The macroeconomics of the quiet revolution: Understanding the implications of the rise in women’s participation for economic growth and inequality. Research in Economics, 71(3), 521-539. Imbs, J. and Wacziarg, R. (2003). “Stages of Diversification”, American Economic Review, 93(1), pp.63-86. Inchauste, M. G, & Rubil, I. (2017). “The distributional impact of taxes and social spending in Croatia (English)”. Policy Research working paper; no. WPS 8203. Washington, D.C.: World Bank Group. http://documents.worldbank.org/curated/en/605341506024423991/The-distributional-impact-of-taxes-and- social-spending-in-Croatia Jugović, I. (2016). Beliefs about the Gender Division of Parental Leave and Characteristics Associated with Them. Institute for Social Research http://www.rsp.hr/ojs2/index.php/rsp/article/viewFile/1292/1342 Kluve, J., and Tamm, M. (2013). Parental Leave Regulations, Mothers’ Labor Force Attachment and Fathers’ Childcare Involvement: Evidence from a Natural Experiment. Journal of Population Economics 26(3): 983– 1005. Lalive, R., Schlosser, A., Steinhauer, A., & Zweimüller, J. (2014). Parental leave and mothers' careers: The relative importance of job protection and cash benefits. Review of Economic Studies, 81(1), 219-265. 67 Lokshin, Michael. 2004. "Household Childcare Choices and Women's Work Behavior in Russia." Journal of Human Resources 39 (4): 1094–1115. Low, H., & Sánchez-Marcos, V. (2015). Female labour market outcomes and the impact of maternity leave policies. IZA Journal of Labor Economics, 4(1), 14. Manley, James, and Felipe Vásquez. 2013. "Childcare Availability and Female Labor Force Participation: A Empirical Examination of the Chile Crece Contigo Program." Working Paper 2013–03 (December), Department of Economics, Towson University, Towson, MD. Mara, I., & Vidovic, H. (2015). Free movement of workers, transitional arrangements and potential mobility from Croatia (No. 402). wiiw Research Report. Maskalan, A. (2016). In the name of the father. Rev. soc. polit., god. 23, br. 3, str. 383-398, Zagreb 2016 Maurer-Fazio, Margaret, Rachel Connelly, Lan Chen, and Lixin Tang. 2011. "Childcare, Eldercare, and Labor Force Participation of Married Women in Urban China, 1982–2000." Journal of Human Resources 46 (2): 261– 94. Nepomnyaschy, L., & Waldfogel, J. (2007). Paternity leave and fathers' involvement with their young children: Evidence from the American ECLS-B. Community, Work & Family. Nollenberger, N., Rodríguez-Planas, N., & Sevilla, A. (2016). The math gender gap: The role of culture. American Economic Review, 106(5), 257-61. OECD. (2018). The OECD Tax-Benefit Model for Croatia. Description of policy rules for 2018. Olivetti, C., & Petrongolo, B. (2017). The economic consequences of family policies: lessons from a century of legislation in high-income countries. Journal of Economic Perspectives, 31(1), 205-30. Olivetti, C., Patacchini, E., & Zenou, Y. (2018). Mothers, peers, and gender-role identity. Journal of the European Economic Association. Ostry, M. J. D., Alvarez, J., Espinoza, M. R. A., & Papageorgiou, M. C. (2018). Economic gains from gender inclusion: New mechanisms, new evidence. International Monetary Fund. Pope, D. G., & Sydnor, J. R. (2010). Geographic variation in the gender differences in test scores. Journal of Economic Perspectives, 24(2), 95-108. Rodríguez-Planas, N., Sanz-de-Galdeano, A., & Terskaya, A. (2019). Gender Norms in High School: Impacts from Adolescence to Adulthood. Rodríguez-Planas, N., Sanz-de-Galdeano, A. (2019). Intergenerational Transmission of Gender Social Norms and Teenage Smoking, Social Science and Medicine, 222: 122-132. Rossin, M. (2011). The effects of maternity leave on children's birth and infant health outcomes in the United States. Journal of health Economics, 30(2), 221-239. Ruhm, C. J. (1998). The economic consequences of parental leave mandates: Lessons from Europe. The Quarterly Journal of Economics, 113(1), 285-317. 68 Rutkowski, J. (2003). Does Strict Employment Protection Discourage Job Creation? Evidence from Croatia. World Bank Policy Research Working Paper 3104, World Bank, Washington, D.C. Schönberg, U., & Ludsteck, J. (2014). Expansions in maternity leave coverage and mothers’ labor market outcomes after childbirth. Journal of Labor Economics, 32(3), 469-505. Thévenon, O., & Solaz, A. (2013). Labour market effects of parental leave policies in OECD countries. Vidovic, H. and Mara, I. (2015). “Free Movement of Workers, Transitional Arrangements and Potential Mobility from Croatia”. The Vienna Institute for International Economic Studies Research Report, No. 402. World Bank. (2001). “Croatia - Economic vulnerability and welfare study (English)”. Washington, DC: World Bank. World Bank. (2005). World Development Report 2006: Equity and Development. Washington DC: World Bank and Oxford University Press. World Bank. (2007). “Croatia - Living standards assessment: promoting social inclusion and regional equity : Main Report (English)”. Washington, DC: World Bank. World Bank. (2010). “Croatia - Social impact of the crisis and building resilience (English)”. Washington, DC: World Bank World Bank. (2018). Demography and Family Policies in Croatia Experience in Selected EU Countries World Bank. (2018c). “Croatia - Systematic Country Diagnostic (English)”. Available online at: http://documents.worldbank.org/curated/en/452231526559636808/Croatia-Systematic-Country-Diagnostic World Bank. (2018d). “Moving for Prosperity: Global Migration and Labor Markets”. Policy Research Report. Washington, DC: World Bank World Bank (2019a). “Investing in Opportunities for All : Croatia Country Gender Assessment”. World Bank, Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/31937 License: CC BY 3.0 IGO World Bank. (2019b). “Demography and Family Policies in Croatia Experience in Selected EU Countries”. Policy Note prepared for the Croatian National Development Strategy World Bank Reimbursable Advisory Services. World Bank. (2019e). “Migration and Brain Drain” Europe and Central Asia Economic Update (Fall), Washington, DC: World Bank World Bank. (2019f). “EU Regular economic report (English). Including Institutions: Boosting Resilience in Europe”, Washington DC; World Bank Group” World Economic Forum (2017). “The Global Competitiveness Report”, Cologny/Geneva Switzerland. World Economic Forum (2018). “The Global Competitiveness Report”, Cologny/Geneva Switzerland. 69 Yamin, P., Fei, M.; Lahlou, S., and Levy, S. (2019). “Using Social Norms to Change Behavior and Increase Sustainability in the Real World: A Systematic Review of the Literature”, Sustainability 2019, 11, 5847; doi:10.3390/su11205847 Zaiceva, A. (2014). “Post-enlargement emigration and new EU members’ labor markets”. IZA World of Labor. 70 Annex 1. Tables Table 1. Labor Market Indicators for Roma and non-Roma neighbors, by Sex 2011 Panel a. Roma Levels by Sex Roma Females Roma Males Roma Gender Gap BGR HRV ROU BGR HRV ROU BGR HRV ROU Labor force participation 50 29 34 64 53 57 13 24 24 rate (ages 15-64) Employment to population ratio (ages 15- 26 7 19 42 22 42 15 15 22 64) Unemployment rate (% Labor of total labor force, ages 48 76 43 35 59 28 -13 -17 -15 Markets 15-64) Informal employment (% 45 49 53 54 48 68 9 -1 15 of total employment) Not in employment, education, or training 70 71 78 53 63 51 -17 -8 -27 (NEET) (ages 15-24) Notes: (1) All values shown are in percent (2) Cells in blue refer to gaps that are statistically significant at the 10 percent level. Source: WB estimates based on weighted 2011 UNDP-WB-EC Regional Roma Surveys. Panel b. Non-Roma Levels by Sex Non-Roma Neighbor Non-Roma Neighbor Non-Roma Neighbor Females Males Gender Gap BGR HRV ROU BGR HRV ROU BGR HRV ROU Labor force participation rate 64 57 42 71 71 67 7 15 24 (ages 15-64) Employment to population ratio 52 41 34 57 57 56 4 16 22 (ages 15-64) Unemployment rate (% of total 19 27 21 20 20 16 2 -7 -4 labor force, ages Labor 15-64) Markets Informal employment (% 8 13 15 13 13 27 5 0 11 of total employment) Not in employment, education, or 27 29 30 30 20 20 3 -9 -10 training (NEET) (ages 15-24) Notes: (1) All values shown are in percent (2) Cells in blue refer to gaps that are statistically significant at the 10 percent level. Source: WB estimates based on weighted 2011 UNDP-WB-EC Regional Roma Surveys. 71 Table 2. School shifts and labor market indicators by gender Working-aged females Working-aged males Temporary Temporary Unemployed Job LFP Working Unemployed Job LFP Working Less than 5 hr school shifts * - Coefficient 0.497*** -0.016 0.081 0.356*** -0.062 0.006 0.12 0.11 (-0.1292) (-0.1725) (-0.1417) (-0.1124) (-0.1598) -0.1846 -0.162 (-0.1309) - Marginal effect at means 0.086*** -0.002 0.008 0.060*** -0.008 0.001 0.011 0.018 Secondary complete -1.144*** -1.029*** 1.727*** ` -1.054*** -0.518* 0.653*** 0.974*** (- (-0.173) (-0.2621) -0.1403 (-0.1446) (-0.1853) (-0.2693) 0.1594) (-0.1482) Tertiary complete -2.320*** -1.579*** 3.235*** 2.939*** -1.892*** -0.463 1.508*** 1.889*** (-0.2409) (-0.2957) (-0.3009) (-0.2073) (-0.3525) (-0.3415) (-0.348) (-0.2746) Age -0.190*** -0.396*** 0.615*** 0.458*** -0.276*** -0.261*** 0.591*** 0.470*** (- (-0.0433) (-0.0644) (-0.0321) (-0.0319) (-0.0417) (-0.0518) 0.0318) (-0.0307) - - - Age^2 0.002*** 0.004*** -0.008*** 0.006*** 0.003*** 0.003*** 0.007*** 0.006*** (- (-0.0006) (-0.0008) (-0.0004) (-0.0004) (-0.0005) (-0.0006) 0.0004) (-0.0004) - Number of children 0.458*** 0.297** -0.587*** 0.548*** 0.127 0.138 -0.048 -0.133 (- (-0.1072) (-0.1452) (-0.1001) (-0.088) (-0.1114) (-0.1644) 0.1162) (-0.0938) - - - - Constant 2.884*** 7.542*** 10.214*** 8.550*** 4.922*** 4.626*** 9.336*** 8.758*** (- (-0.8466) (-1.2574) (-0.5804) (-0.6141) (-0.8497) (-1.086) 0.5558) (-0.5691) Number of observations 2,655 1783 3569 3,569 2,676 1914 3273 3,273 Year dummy Yes Yes Yes Yes Yes Yes Yes Yes Notes: Logit regression model using pooled 2014-2017 EU-SILC. Robust standard errors in parentheses; (***), (**), and (*) indicate levels of significance at 1%, 5%, and 10%respectively. * These are the marginal effects with respect to the baseline of individuals with children attending school for more than 5 hours. 72 Table 3: Number of hours worked and labor income Hours worked Log labor income Female Male Female Male More than 5 hr school shifts * 3.678*** -0.308 0.753*** -0.318* (-0.8957) (-0.942) (-0.1884) (-0.1876) Secondary complete 11.179*** 8.072*** 2.039*** 1.678*** (-1.5978) (-1.6392) (-0.3333) (-0.3118) Tertiary complete 18.039*** 10.712*** 3.787*** 2.498*** (-1.6846) (-1.78) (-0.3531) (-0.3502) Age 1.379*** 2.068*** 0.425*** 0.541*** (-0.3879) (-0.3708) (-0.0801) (-0.0697) Age^2 -0.017*** -0.025*** -0.005*** -0.006*** (-0.0049) (-0.0047 (-0.001) (-0.00090) Number of children -3.253*** -0.281 -0.737*** -0.226 (-0.9876) (-0.8018) (-0.1988) (-0.1412) Constant -5.506 -15.341** -2.929* -3.676*** (-7.7087) (-7.3856) (-1.5599) (-1.404) Number of observations 2,655 2,676 2,655 2,676 Year dummy Yes Yes Yes Yes Notes: OLS regression model using pooled 2014-2017 EU-SILC. Robust standard errors in parentheses; (***), (**), and (*) indicate levels of significance at 1%, 5%, and 10%respectively. * These are the marginal effects with respect to the baseline of individuals with children attending school less than 5 hours. Table 4. Labor income gender gap Log labor income >5 hrs <=5 hrs Female -0.315*** -0.347*** (-0.0455) (-0.0301) Secondary complete 0.221*** 0.232*** (-0.0726) (-0.0547) Tertiary complete 0.726*** 0.673*** (-0.0819) (-0.06020) Age 0.116*** 0.083*** (-0.0215) (-0.0136) Age^2 -0.001*** -0.001*** (-0.0003) (-0.0002) Number of children 0.044 -0.068*** (-0.0389) (-0.0246) Constant 6.789*** 7.648*** (-0.4528) (-0.2817) Number of observations 1,532 3,799 Year dummy Yes Yes 73 Notes: OLS regression model using pooled 2014-2017 EU-SILC. Robust standard errors in parentheses; (***), (**), and (*) indicate levels of significance at 1%, 5%, and 10%respectively. * These are the marginal effects with respect to the baseline of males with compulsory aged children. Table 5. Fathers’ labor supply Father's hours worked >5 hrs <=5 hrs Mother's hrs worked -0.286** -0.193*** (-0.1124) (-0.0619) Secondary complete -6.737 1.433 (-6.5711) (-5.6977) Tertiary complete 5.315 5.398 (-7.0771) (-3.8816) Age 0.189 1.029 (-1.0812) (-0.804) Age^2 -0.004 -0.012 (-0.0143) (-0.0096) Number of children 4.18 2.217 (-2.7001) (-2.3279) Constant 42.943** 13.912 (-17.0001) (-14.5602) Number of observations 69 144 Year dummy Yes Yes Notes: OLS regression model using pooled 2014-2017 EU-SILC. Robust standard errors in parentheses; (***), (**), and (*) indicate levels of significance at 1%, 5%, and 10%respectively. 74 Table 6. The Effect of Paternity Leave on Labor Market Outcomes Panel A: No controls Share of males Share of Share of female Share of male among workers workers on workers on workers on on paternaty parental leave parental leave parental leave leave Male-female ratio in LFP -0.366*** -1.981*** -0.890*** -4.993*** Labor Force Participation female 0.286*** 1.503*** 0.679*** 3.938*** Labor Force Participation male 0.037* 0.079 0.031 0.439* Male-female ratio in employment -0.343*** -2.510*** -1.160*** -5.264*** Employed male 0.077* 0.233 0.123 1.518*** Employed female 0.281*** 1.874*** 0.888*** 4.781*** Inactive fulfilling domestic tasks female -0.387*** -1.659*** -0.745*** -5.269*** Panel B: Country fixed effects Male-female ratio in LFP -0.336*** -1.107*** -0.353** -6.546*** Labor Force Participation female 0.137*** 0.976*** 0.389*** 3.757*** Labor Force Participation male -0.119*** 0.150 0.128* -1.183** Male-female ratio in employment -0.387*** -1.780*** -0.589*** -8.918*** Employed male -0.208** -0.778* -0.175 -4.229** Employed female 0.073 0.506** 0.241* 2.321** Inactive fulfilling domestic tasks female -0.265*** -0.640*** -0.179 -4.689*** Panel C: Country and year fixed effects Male-female ratio in LFP -0.192*** -0.221 0.065 -3.018** Labor Force Participation female 0.068* 0.677*** 0.239** 2.356*** Labor Force Participation male -0.081*** 0.515*** 0.290*** 0.012 Male-female ratio in employment -0.201*** -0.538 -0.012 -4.308*** Employed male -0.097 0.336 0.317* -0.652 Employed female 0.049 0.683*** 0.298** 2.522** Inactive fulfilling domestic tasks female -0.111** 0.190 0.190* -1.030 Note: number of observation: 282. Results are based on LFS 2006-2018. Panel A: No controls Share of males Share of Share of female Share of male among workers workers on workers on workers on on paternaty parental leave parental leave parental leave leave Male-female ratio in LFP -0.366*** -1.981*** -0.890*** -4.993*** Labor Force Participation female 0.286*** 1.503*** 0.679*** 3.938*** Labor Force Participation male 0.037* 0.079 0.031 0.439* Male-female ratio in employment -0.343*** -2.510*** -1.160*** -5.264*** Employed male 0.077* 0.233 0.123 1.518*** Employed female 0.281*** 1.874*** 0.888*** 4.781*** Inactive fulfilling domestic tasks female -0.387*** -1.659*** -0.745*** -5.269*** Panel B: Country fixed effects Male-female ratio in LFP -0.336*** -1.107*** -0.353** -6.546*** 75 Labor Force Participation female 0.137*** 0.976*** 0.389*** 3.757*** Labor Force Participation male -0.119*** 0.150 0.128* -1.183** Male-female ratio in employment -0.387*** -1.780*** -0.589*** -8.918*** Employed male -0.208** -0.778* -0.175 -4.229** Employed female 0.073 0.506** 0.241* 2.321** Inactive fulfilling domestic tasks female -0.265*** -0.640*** -0.179 -4.689*** Panel C: Country and year fixed effects Male-female ratio in LFP -0.192*** -0.221 0.065 -3.018** Labor Force Participation female 0.068* 0.677*** 0.239** 2.356*** Labor Force Participation male -0.081*** 0.515*** 0.290*** 0.012 Male-female ratio in employment -0.201*** -0.538 -0.012 -4.308*** Employed male -0.097 0.336 0.317* -0.652 Employed female 0.049 0.683*** 0.298** 2.522** Inactive fulfilling domestic tasks female -0.111** 0.190 0.190* -1.030 Note: number of observation: 282. Results are based on LFS 2006-2018. 76 Table 7. Determinants of Female Labor Force Participation 2006-2017 Single with Couple with Single Couple two children two children Inactivity trap -0.0333*** -0.0960** -0.0081 -0.0036 (0.0099) (0.0453) (0.0288) (0.0782) Lagged part-time employment 0.4539*** 0.3434** 0.4173** 0.4182** (0.1743) (0.1695) (0.1778) (0.1768) Lagged temporary employment 0.0350 0.0047 0.0091 0.0072 (0.1026) (0.1026) (0.1069) (0.1085) Lagged percent employed in services 0.4375** 0.4506*** 0.4356** 0.4377** (0.1717) (0.1677) (0.1770) (0.1750) Age dependency ratio 0.0463 0.0254 0.0309 0.0312 (0.0750) (0.0762) (0.0791) (0.0768) Lagged unemployment rate -0.1793** -0.1475* -0.1965** -0.1996** (0.0773) (0.0774) (0.0824) (0.0797) Proportion of women with tertiary education 0.1665*** 0.1628*** 0.1549*** 0.1556*** (0.0564) (0.0559) (0.0582) (0.0575) GDP per capita growth -0.0898** -0.0892** -0.0893** -0.0887** (0.0388) (0.0418) (0.0429) (0.0438) Labor market measures per thousand unemployed 0.3023*** 0.3494*** 0.3016*** 0.3013*** (0.0930) (0.0985) (0.0940) (0.1015) Ease of hiring and firing 0.8532*** 0.6259** 0.7708*** 0.7607*** (0.2727) (0.3070) (0.2806) (0.2536) Constant 21.1353 24.7896 21.5532 20.9933 (16.2447) (16.2767) (17.3041) (16.8603) R-squared 0.9560 0.9560 0.9540 0.9540 Observations 307 307 307 307 * p<0.10, ** p<0.05, *** p<0.0. Robust standard errors are in parentheses. All specifications include country and time fixed effects. Ease of hiring and firing comes from the 2004-2017 World Economic Report. All other variables come from Eurostat. 77 Table 8. The Effect of Gender Norm Beliefs on Inactivity. Logit Marginal Effects. (1) (2) (3) (4) Females Females Males Males Gender norms index (Normalized) 0.0362** 0.0274* 0.0142 0.00310 (0.0149) (0.0147) (0.0176) (0.0188) Personal controls yes Yes Observations 577 569 404 394 Source: EVS 2018-2019. Note: Logit marginal effects are reported. Gender norms index is normalized to have mean 0 and standard deviation 1 and higher values are associated with more patriarchal views. All regressions control for age, age squared, and the indicators for paternal and maternal educational level. Columns (2) and (4) include in addition indicators for educational level, marital status and the number of children. Source: EVS, Croatia 2017. Age: 20-64. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Table 9. The Effect of Female Labor Force Participation in the Country of Ancestry. Second- generation Immigrants Women Men (1) (2) (3) (4) (5) (6) (7) (8) female labor force participation in country of ancestry -0.368*** -0.338*** -0.356*** -0.480*** -0.208 -0.193 -0.236 -0.334* (0.076) (0.090) (0.106) (0.150) (0.132) (0.152) (0.160) (0.165) Individual controls Yes yes yes yes yes yes Parental education yes yes yes yes GDP of country of ancestry yes yes Observations 1,412 1,412 1,412 1,412 1,126 1,126 1,126 1,126 R-squared 0.144 0.184 0.230 0.233 0.196 0.236 0.279 0.281 Source: EVS 2018-2019. Note: Table reports the effect of female labor force participation rate in the country of birth of parents measured in 2010 (Source: World Bank Open Data). All specifications control for the host country fixed effects, age, and age squared. Columns (2) and (6) include indicators for marital status, educational level, and the number of children. Columns (3) and (7) include in addition indicators for maternal and paternal educational level. Columns (4) and (8) include in addition GDP of the country of ancestry measured in 2009. Only second-generation immigrants between 20 and 64 years old are included. Robust standard errors in parentheses account for clustering at country level. *p<0.1, ** p<0.05, *** p<0.01. 78 Table 10: Oaxaca decomposition of within and across industry gender gap in wages. 2018 Tertiary educated only. 2018 All. 2018 Males 8.245*** 6.137*** (0.219) (0.118) Females 7.282*** 5.146*** (0.196) (0.132) Difference 0.963*** 0.991*** (0.294) (0.177) Endowments 0.0327 -0.0592 (0.160) (0.159) Coefficients 0.959*** 1.308*** (0.328) (0.202) Interaction -0.0290 -0.258 (0.224) (0.187) Observations 2,202 8,691 Source: LFS 2018. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Age: 20-64 Table 11. Oaxaca decomposition of within and across educational specialization gender gap in wages Tertiary educated only. 2018 Males 7.994*** (0.305) Females 6.704*** (0.280) Difference 1.290*** (0.414) Endowments 0.0992 (0.331) Coefficients 1.137** (0.478) Interaction 0.0539 (0.420) Observations 1,080 Note: Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Age: 20-64 79 Annex 2. Summary of Changes in Maternity and parental leave policy in Croatia between 2008 and 2020. In July 2008, Act on maternity and parental leave was introduced. The most significant changes are related to equalization of employed and self-employed parents` rights, flexibilization of leave and the extension of beneficiaries entitled to adoption leave. More specific changes include: • Equalization of the rights of employed and self-employed parents. Under the former legislation, self-employed parents were not entitled to the same rights as employed parents. Compulsory Maternity leave for the mother has been shortened, with leave from the forty-third day after birth made available to the child’s mother or father and on a full-time or part-time basis. Under former legislation, the leave could only be taken full time and the father could use it before the child was six months of age only in exceptional circumstances. Before that the mandatory maternal leave period was 6 months and it was regulated by Labor Law Act 66. Parental leave can be exercised until the child turns eight years of age; it is a personal right of both parents and can be used in equal shares; and it can be used in a more flexible way. Under former legislation, the leave was the mother’s right, which the father could only use with the mother’s agreement; and there was no flexibility in its use. • Employers were given opportunity to postpone Parental leave for 30 days; • Leave for employed pregnant women or a mother breastfeeding the child is available to all companies. Under the former legislation, larger companies were excluded. Responsibility for paying mothers taking this leave has been transferred to employers and the payment rate increased. • Adoptive parents have the same rights as biological parents, regardless of employment status, and can take leave until the child turns 18 years. Under the previous legislation, they were not able to use adoptive leave if they were self-employed and could only take leave until a child was 12 years. In 2011 there were several changes in legislation. Obligatory Maternity leave was extended from 70 days to 98 days in order to harmonize with EU legislation. Conditions for transmission of all leave rights (i.e. except for the obligatory period of Maternity leave) from one parent to another, if they are in different activity/working status, have been relaxed (e.g. parents do not have to share the same labor law status in order to transmit the rights related to childcare as before; the condition to work at least three months continuously was abolished as a precondition to transmit the unused share of parental exemption from work/parental care for the child from one parent to another). Parental allowance for the additional two months of Parental leave - in those cases when the period of leave has been extended due to the father`s use of at least three months of leave - was equalized with the amount of parental allowance paid during the first six months of Parental leave (i.e. 100 per cent of average earnings, with a ceiling of 80 per cent of the budgetary base rate; before the ceiling was set at 50 per cent) 80 In 2013 there were few changes in leave-related legislation (all effective since July 2013). Parental leave for first and second born children has increased to eight months – four months per parent, with two months transferable to the other parent and two months that are non-transferable. Previous leave legislation provided three months leave for each parent plus two additional ‘bonus’ months if the father used his three-month entitlement. To harmonize with the EU Council Directive 2010/18/EU on Parental leave, from December 2014 the Parental leave rights of (self)-employed adoptive parents were equalized with those of (self)-employed biological parents. Now they also become entitled to eight months of Parental leave (two of them non-transferable) until their child turns eight years of age. Previously employed and self-employed adoptive parents were entitled to Parental leave of six months until their child turns eight years of age. Since July 2017, new amendments to the Maternity and Parental Benefits Act entered into force, according to which the parental benefits and Maternity/Parental allowances have increased. The ceiling on parental benefit has increased from 80 to 120 per cent of the budgetary base rate (i.e. from HRK2,660 [€360.25] HRK3,991 [€540.51] per month), and the flat-rate amount of maternity/parental allowance paid to inactive and unemployed parents from 50 per cent to 70 per cent of the budgetary base rate (i.e. from HRK1,663 [€225.22] to HRK2,328 [€315.29] per month). These changes became effective in July 2017, and they apply to all parents, regardless if they already exercise the rights or not. It was also stated that the further plan is to completely abolish the ceiling on earnings-related Parental benefit, but that they are not going to abolish it now due to budgetary constraints. There have been no changes in maternal or parental leave policies between 2018 and 2019. In April 2020, the Croatian Parliament adopted amendments to the maternity and parental allowance law which will increase the ceiling on parental benefit for the second six months of the child's life from 3,991 HRK to 5,654 HRK (a 41.5% increase). The amendments also reduce the duration of years of service required to claim the benefit from 12 to 9 months and from 18 to 12 months with interruptions in the last two years. Source: https://www.leavenetwork.org/annual-review-reports/ https://www.sabor.hr/en/press/news/amendments-increase-maternity-benefits-second-six-months 81