UNIVERSITY OF TARTU Faculty of Social Sciences School of Economics and Business Administration
Papuna Gogoladze
GENDER INCOME GAP OVER LIFE-CYCLE: CROSS-COUNTRY ANALYSIS
Master’s Thesis
Supervisors: Jaanika Meriküll (Senior research fellow, Ph.D.) Jaan Masso (Senior research fellow, Ph.D.)
Tartu, 2019
Name and signature of supervisor………………………………………………….
Allowed for defence on.........................................................
(date)
I have written this master's thesis independently. All viewpoints of other authors, literary
sources and data from elsewhere used for writing this paper have been referenced.
…………………………………..
(signature of author)
iii Acknowledgements First, I would like to express my utmost gratitude towards my research supervisors, Dr. Jaanika Meriküll and Dr. Jaan Masso. Their guidance and useful criticts have been monumental in the research work. The timely and detailed reviews provided during the study kept the progress on schedule. I am indebted to my supervisors as they set a perfect example of being a good researcher, or, simply, a person. I would like to extend my sincere appreciation to Dr. Ricardo Alfredo Vicente for his insightful comments and encouragement during interim reports of the study. I would also like to thank to the thesis committee and my opponent, Maryna Tver- dostup, who diligently reviewed and evaluated the paper. I thank Ministry of Education and Research of Estonia for funding my master’s studies and European Union’s Rights, Equality and Citizeship Programme (2014- 2020), No 820778, for providing research funds. Lastly, I would like to thank my family for their tremendous support and sym- pathetic ear. Despite the thousands of kilometers between us, their love and care are always with me. And my friends, who have been continuously cheering me up during this period and creating pleasant environment for me to pursue my goals.
v Abstract Gender Income Gap Over Life-Cycle: Cross-Country Analysis by Papuna Gogoladze Despite the vast literature on the gender disparities in the labor market participation and outcomes, there is a scarce literature on the gender gap in total income. This pa- per tries to fill the research gap and has threefold contribution to the existing litera- ture. First, the paper studies the gender gap in aggregate income instead of focusing only one component – wages. Second, the analysis of the gap in four age categories reveals how the income gap behaves throughout the “life-cycle”. And three, the gap is analysed in 25 countries that allows observing the institutional differences that are not apparent in case of single-country study. In most countries the unexplained me- dian total income gap is the lowest among the youngest and increases throughout the life-cycle. However, there are countries, where the gap peaks in the youngest age group, for example, in Bulgaria. There is a large heterogeneity among countries in the unexplained gap size. Among the working age people, the unexplained me- dian income gap is above 70% in Greece, while the lowest gap, approximately 4%, is reported in Slovenia. The paper suggests that trade union membership reduces and minimum wages increase the unexplained income gap of low-income individuals below age 45. More generous maternity leave increases it for low-income individu- als between age 25-44, while formal child-care has negative impact throughout the distribution for the oldest. The analysis shows that "one-size-fits-all" policies are unable to respond adequately the gender gap issue in different income sources.
- Introduction 1 1 Introduction A gender discrimination in numerous fields has laid a solid foundation for develop- ment of advanced econometric tools, which has enabled researchers to extensively study differences in the labor market outcomes and participation by gender. Most studies on gender inequality in labor market participation show strong convergence of economic status of males and females over time (e.g. see Balleer et al. (2009), Al- toji and Blank (1999)), However, there still exists substantial gap between earnings of men and women (e.g. see Boll and Lagemann (2018)). Since wages constitute the largest component of total earnings, the gender wage gap has become the most widely studied topic in terms of gender inequality. This process has been acceler- ated by the availability of the data on employment income. However, the absence of appropriate statistical data impeded the further examination of gender disparities in other fields, such as wealth and pensions. For example, until recently, little to no studies have been done on wealth inequality (see, inter alia, Siemerminska et al. (2010), Bonnet et al. (2013), D’Alessio (2018), Meriküll et al. (2018)). Most datasets collected information on wealth components at household level, which made is im- possible to differentiate the individual possessions. Due to very similar reasons as for the gender wealth inequality, there is a very scarce literature on the gender income gap. The study of wage differentials captures the gap only among wage-earners, which accounts for less than a half of the total population. Moreover, employment income is only one part of the total income. Em- ployment income itself may include wage and self-employment income, while the total income, in addition, includes property and capital income, and transfers. Also, the role of employment income is different for people from different age groups (less important for young and old individuals). The aim of the study is to investigate the gender income gap over age groups in Europe. The contribution of this paper to the existing literature is threefold. First, instead of focusing on discrepancies solely in employment income, the study aggregates 17 different income sources and studies the gender gap in the total income. Furthermore, the gender gaps in three differ- ent income categories are separately investigated. Second, the gap is analysed for different age groups that reveals how the income gap behaves throughout the “life- cycle”. And finally, the paper studies the gap in 25 countries and tries to observe institutional differences that are not apparent in case of single-country study. In the study the survey data of European Union Statistics on Income and Living Conditions (2016) is used. The dataset collects information about income at both personal and household levels along with other demographic and socio-economic data. The study spans the 23 EU and 2 non-EU countries, including individuals aged 16 and above. In order to study the impact of the observed characteristics on the total income throughout the whole distribution, the unconditional quantile re- gression is used, proposed by Firpo et al. (2009). The key advantage of this method is its ability to estimate the effects of marginal changes in the explanatory variables on the unconditional quantiles of the dependent variable. Next, the Oaxaca-Blinder (1973) decomposition method is applied to investigate the gender income gap and see what portion of the total gap can be explained by the differences in the observed characteristics. It is found that there is a high degree of heterogeneity among the countries. In 14 out of 25 countries, the unexplained median total income gap is the lowest for
2 the youngest age-group and gradually increases over the life-cycle. However, there are countries in which the gap peaks among the youngest individuals, for exam- ple, in Bulgaria. The study also revealed significant variation within the age groups. Among the individuals below age 25, in Bulgaria the unexplained median income gap is the largest and in favour of men (0.726 log points), while in Greece there is the evidence of the largest unexplained median income favouring women. In the following age group of 25-44, the largest unexplained median gap is observed in Greece (0.557 log points), while the lowest gap is reported for Denmark (0.115 log points). Similar results are reported for the individuals between 45-65: there is the evidence of the largest and the lowest unexplained median income gaps favouring men in Greece and Slovenia, respectively. The fact that in Greece there are the largest gap in these two age groups could be attributed to the shift from public to private sector. Earlier literature found that in Greece there is a negative relationship between public sector employment and wage gap (e.g. see Christofides et al. (2013)). Since the crisis in 2008, the public sector employment has been decreasing at high rates. Therefore, moving towards private sector could have contributed to the enlargement of the wage gap and, subsequently, total income gap. In the oldest age group, the largest unconditional median income gap in favour of men is reported for Austria, while it is the lowest in Estonia, though statistically insignificant. Since the public transfers have the lion’s share in the total income for the oldest individuals, these findings are further strengthened by the gaps in public transfers: in Estonia the un- explained median gap in public transfers is the lowest, while in Austria it is one of the largest. In addition, the paper analyzed the gender gaps in employment income, private transfers and capital income, and public transfers. The raw and unexplained median gaps in employment income are always in favour of men. Among the individuals between age 25-44, the largest unexplained median employment income gaps are reported in Latvia and Estonia (0.463 and 0.456 log points, respectively), while it is the lowest in Romania (0.078 log points). In the following age group of 45-65, there is the evidence of the largest and the lowest unexplained median employment income gaps in the Netherlands and Slovenia (0.425 and 0.091 log points, respec- tively). Also, glass ceiling and sticky floor effects are found in Belgium, Czechia, France, Greece, and Norway for both age groups, indicating a presence of positive selection (Olivetti and Petrongolo (2008)). The analysis for private transfers and cap- ital income provides evidence that the largest total and unexplained median gaps are present in France, while they are the lowest in Hungary, among the youngest. In the age groups of 25-44 and 45-65, the explained part of the total gap systemati- cally favours women over men. Among the oldest, the largest unexplained median gap is reported for Serbia, which is also in favour of women. The analysis of the public transfers showed that the differences in the observed characteristics predom- inantly favour women for the individuals below age 65. This could be attributed to the women’s increased participation in the education and higher education-related allowances compared to men. On the contrary, both explained and unexplained me- dian gaps in public transfers favour men over age 65. Finally, the paper studied impact of eight institutional factors on the unexplained total income gap: union density, employment protection legislation, Kaitz index (ra- tio of minimum wage to average wage), maternity pay entitlement, formal child care for children under 3, pensions system design, minimum wage setting, and na- tional minimum wage. Union density and minimum wage setting are found to have
- Related Literature 3 significant negative and positive impact, respectively, on the unexplained gap for low-income individuals. The negative impact of union density on the gap is also highlighted by other studies (e.g. see Blau and Kahn (1992, 1996)). Moreover, the analysis showed that minimum wages have negative impact on the unexplained gap in employment income (e.g. see Bargain et al. (2018)). However, as mentioned above, it enlarges the unexplained total income gap. Since many low-income indi- viduals receive public transfers, once minimum wages are set, they may not qualify for those social benefits anymore, which might be larger than the marginal increase in employment income. It is found that the maternity pay entitlement (product of maternity leave length and payment rate) has positive impact on the unexplained gap for individuals between 25-44: increased burden encourages employers to offer lower wages to women compared to their male counterparts. Among the oldest indi- viduals, there is the evidence of the negative relationship between formal child-care and unexplained total income gap. This could easily be explained by widespread culture of informal care (for example, grandmothers), especially in Eastern and Cen- tral European cultures. The structure of the paper is as follows: Section 2 overviews the literature re- lated, Section 3 explaines the application of the unconditional quantile regression and Oaxaca-Blinder decomposition, Section 4 describes the data and its transforma- tions, Section 5 presents the findings of the paper, and Section 6 concludes. 2 Related Literature A study of the economics of discrimination is thought to be pioneered by Becker (1957) in his seminal study. As Weichselbaumer and Winter-Ebmer (2005) describe, availability of microdata allowed labour economists to conduct numerous studies on gender inequality in the last decades. Most studies on gender inequality depict strong convergence of economic status of males and females over time. Lion’s share of this reduction can be attributed to increasing trends in women’s participation in the labour market and their educational levels. Altoji and Blank (1999) summa- rized the literature of gender and race inequality and showed dramatic changes in the labour force participation in the United States – there had been a steady decline in men’s involvement in the labour force, especially for black men, while, women showed increased labour force participation. Despite this convergence, there exists a gap between men’s and women’s earnings. They distinguished two main factors contributing to the gender earnings gap: human capital accumulation and discrim- ination. The largest part of the differential was due to the discrimination even af- ter controlling for individual and job characteristics. Goldin (2006) introduced the term "quiet revolution" to describe how women changed their views about career perspectives and their role in the family. On the other side of the labour market, computerization has had a great impact on relative labour demand for females as compared to males. Weinberg (2000) estimates that more than half of the increase in demand for female workers in the United States can be accounted for computeriza- tion. Until the second half of the 20th century, it was legal to differentiate employees by gender and offer them different wages for certain jobs. It was a common practice to publish job advertisements for each gender separately. However, some countries started promoting equal pay for both genders. For example, in the United States, the