The Economic Case for More Gender Equality in Estonia

3. The role of firms in the gender pay gap in Estonia

Abstract

This chapter analyses the drivers of the gender pay gap, with a particular focus on the role of firms. Section 3.1 shows how the gender pay gap and its main determinants in Estonia compare with those in other European countries. Using comprehensive administrative data, Section 3.2 analyses to what extent the gender pay gap among similarly skilled female and male employees who perform similar tasks and hold similar responsibilities reflects differences in pay practices between firms (the fact that women are employed in lower-productivity and hence lower-paying firms or industries) and differences in pay practices within firms (the fact that female employees are paid less than their male colleagues). On basis of the literature and new evidence, Section 3.3 discusses how the gender pay gap evolves over the life course and looks at the role of career breaks in explaining wage developments following childbirth for mothers rather than for fathers.

Main findings

Cross-country evidence suggests that a better understanding of the gender pay gap in Estonia requires focusing on differences in the characteristics of the firms and jobs in which men and women are employed rather than differences in their skills.

• The gender pay gap between similarly skilled men and women reflects for about one‑quarter the sorting of women in low-wage firms (and industries), and one‑quarter the sorting of women in low-wage occupations within firms. The remaining half is unexplained and may reflect differences in tasks and responsibilities within occupations or differences in pay for work of equal value (discrimination). Differences in skills tend to mitigate the gender wage gap.

More detailed analysis for Estonia based on administrative tax records of the role of differences in pay practices within and between firms for similarly skilled men and women with similar tasks and responsibilities provides the following insights:

• Differences in firm pay practices account for about half of the unadjusted gender pay gap in Estonia. Three‑fifths of pay differentials are explained by the sorting of women into lower-paying firms, whereas the remaining two‑fifths reflect pay differences between men and women with similar skills, tasks and responsibilities (discrimination and bargaining).

• The gender pay gap is higher among high-wage workers. This is mainly driven by larger differences in pay practices between firms and reflects the difficulty of women in accessing top jobs in high-wage firms. Fewer options for high-wage women may further weaken their bargaining position, potentially exacerbating differences in pay practices for work of equal value within firms. Access for women to top jobs in top firms can be promoted through voluntary target setting, good management practices that make managers accountable and gender quotas for company boards.

• Differences in gender pay gaps across industries tend to reflect discriminatory pay practices within firms, especially in industries where labour market competition is weak. Differences in firm pay practices within firms (discrimination and bargaining) tend to be larger in industries with high rents (measured by the average firm wage premium in the industry), whereas differences in pay practices between firms tend to be smaller in such industries. This highlights the potential value of sectoral initiatives to raise awareness about the extent of gender discrimination and its consequences for firms and workers.

• The sorting of women into low-wage industries contributes to the gender wage gap. Men tend to be overrepresented in industries paying above average wages, conditional on worker characteristics, whereas women tend to be overrepresented in industries paying below-average wages. The sorting of women into low-wage industries is likely to reflect to some extent the role of past educational choices and traditional gender norms.

In Estonia, the gender pay gap temporarily increases following childbirth: it increases with age until it peaks among workers aged 30‑34, and gradually decreases afterwards resulting in the gender pay gap among workers aged 45‑54 being similar as those aged 25‑29. A number of different explanations are considered:

• There is no indication that the temporary increase in the gender pay gap can be explained by young mothers who sort into lower-wage firms that are more likely to offer family-friendly working conditions or employers who discriminate more strongly against young mothers.

• By contrast, career breaks are likely contribute to the temporary increase in the gender pay gap following childbirth due to the cost of lost experience and human capital depreciation. Women are much more likely than men to take a career break around the age of child birth and career breaks involve considerable costs in the form of foregone wage increases within firms. The incidence gap in career breaks and their wage effects tend to be more important among high-wage workers.

• The role of career breaks for the gender pay gap in Estonia may to some extent reflect strong employment protection. Workers are guaranteed to move back in the same or a similar position after career breaks of up to 3 years.

3.1. A cross-country perspective on the gender pay gap and its main components

This section provides a cross-country analysis of the gender pay gap and its main parts. More specifically, it decomposes the gender pay gap into four components which capture respectively: i) differences in skills, i.e. differences in education (low, middle, high) and potential work experience (as proxied by age); ii) differences in pay practices between firms and industries across similarly skilled workers; iii) differences in pay practices within firms across similarly skilled workers due to differences in occupations and working-time status; and iv) a residual that reflects – within the same firm, occupation and working-time status – differences in pay across similarly skilled workers due to differences in tasks and responsibilities, but also differences in pay for work of equal value. The analysis covers 23 European countries, including Estonia, using data from the Eurostat’s Structure of Earnings Survey for the period 2002‑18.

3.1.1. Differences in skills mitigate rather than contribute to the gender pay gap

In Estonia, women earn about 22% less than men on average, compared with 11% across European countries (Figure 3.1). Controlling for skills in terms of education and potential experience (age) increases the gender pay gap from 22% to 27% in Estonia and from 11% to 16% on average across European countries. This reflects the fact that working women tend to be better educated than working men on average (Chapter 2). Importantly, it implies that a better understanding of the gender pay gap requires focusing on differences in the characteristics of the firms and jobs in which men and women are employed rather than differences in their skills.

About one‑quarter of the wage gap between similarly skilled men and women results from the sorting of women into low-wage firms and industries

In Estonia as well as on average across European countries, approximately one‑quarter of the gender pay gap between similarly skilled men and women can be attributed to the sorting of women into firms and industries that pay low wages (“horizontal segregation”). This mainly reflects differences in pay practices between firms within industries rather than differences in pay practices between industries (see Annex 3.A). The sorting of women into low-wage firms reflects to some extent differences in non-wage working conditions, as women may be constrained to opt for firms with flexible working time arrangements due to childcare responsibilities and unpaid homework, but may also reflect the role of discriminatory hiring practices by employers in high-wage firms (OECD, 2021[1]). The sorting of women into low-wage industries may in addition reflect the tendency of women to sort into economic activities that are compatible with their past educational choices (e.g. privileging literacy over mathematical skills), driven by gendered socialisation processes earlier in life, stereotypes and social norms (OECD, 2017[2]).

Figure 3.1. Decomposing the gender pay gap across countries

Note: The figure reports estimates of the contribution of worker, firm and job characteristics to the gender wage gap, obtained estimating different specifications of the following equation: ${\log (w}_i)=\beta femal{e}_i+X_i\gamma +{year}_i+{\epsilon }_i\mathrm{ }\mathrm{ }$ where w_i is the hourly wage of individual i; $femal{e}_i$ is a dummy variable for women; $X_i$is a vector of additional individual-level characteristics, ${year}_i$ is a dummy capturing the survey wave’s year (2002, 2006, 2010, 2014 or 2018); and ${\epsilon }_i$ is the error term. The overall gender pay gap (black dots) is obtained estimating the $\beta$ coefficient from a specification in which the cohort dummy is the only variable in vector $X_i$. The skills component (blue bars) is obtained as the difference between the $\beta$ coefficients obtained from the first specification and an alternative one also including age bracket dummies (for age 14‑19, 20‑29, 30‑39, 40‑49 or 50‑59) and age dummies interacted with education dummies (lower secondary, upper secondary and post-secondary). The firm component (dark green bars) is obtained similarly, including industry (1‑digit NACE2) and firm dummies (obtained exploiting information on the firm identifier). The job component (orange bars) are obtained similarly, including a part-time dummy and occupation dummies (2‑digit ISCO), respectively. Finally, the unexplained component (red bars) is calculated as the $\beta$ coefficient obtained from the most comprehensive specification. See Annex 3.A for the more detailed decomposition.

Source: OECD calculations based on Eurostat Structure of Earnings Survey (SES).

Differences in occupations and working time arrangements within firms account for another quarter

Another quarter of the gender wage can be attributed to the sorting of women into low-wage occupations within firms (“vertical segregation”). In Estonia, this accounts for 6 percentages points of the gender pay gap or 27%, compared with 2 percentages points or 20% on average across European countries. The relative importance of vertical segregation in Estonia could reflect the role of educational choices for the sorting of women into certain occupations or differences in the probability of being promoted to higher wage occupations. The latter could be a reflection of the role of motherhood for the career progression of women (see Section 3.3). Working-time arrangements, as measured by part-time work status, do not play a significant role in the gender pay gap in Estonia, which may not be surprising given its limited incidence. Part-time work status plays a larger role in countries where part-time is more prevalent such as Germany, the Netherlands and the United Kingdom.

About half of the wage gap between similarly skilled men and women reflects unexplained factors, including differences in pay for work of equal value

The bulk of the gender pay gap between similarly skilled men and women cannot be explained by differences in the characteristics of the firms for which they work or differences in occupations and working time. This residual gender pay gap can reflect differences in pay due to differences in tasks and responsibilities, but also differences in pay for work of equal value. The latter are typically attributed to differences in bargaining or wage discrimination by employers.1 Employers’ greater tendency to exploit their wage‑setting power when negotiating with a female job candidate or employee is compounded by women’s internalisation of traditional gender norms according to which they are not expected to be as assertive as men when bargaining. Meriküll and Mõtsmees (2017[3]) confirm that women ask for lower wages than men in online job portals in Estonia. Limited competition in product and labour markets tends to increase the scope for discrimination either because it allows costly discrimination to persist or because it may translate in stronger wage‑setting power of firms and hence increase the scope for monopsonic discrimination (Box 3.1).

In sum, the gender pay gap reflects to an important extent differences in pay practices within firms

In sum, the cross-country analysis has shown that about half of the gender pay gap in Estonia reflects differences in pay between men and women with similar skills, occupations and working time status, who are employed in the same firm. An important question is to what extent the residual gender pay gap is due to any remaining differences in tasks and responsibilities and differences in pay for work of equal value (bargaining, discrimination). The analysis in this section accounts for differences in tasks and responsibilities between occupations, it cannot account for any such differences within occupations. Using linked employer-employee data for selected OECD countries, OECD (2021[1]) suggests that differences in tasks and responsibilities account for most of the gap in pay between men and women with similar skills, but that differences in pay for equal are also important role in some countries. The analysis in Section 3.2 uses state‑of-the‑art methods to provide a more precise decomposition of the role of pay practices within and between firms in the gender pay gap in Estonia.

Box 3.1. Gender wage gaps in state‑owned enterprises and private firms

The scope for gender discrimination is likely to be higher in product markets with limited competition. Employers who discriminate against women drive down the demand for female workers and their average wages relative to their male peers, while non-discriminating employers can increase profits by employing larger shares of lower-paid women (Becker, 1971[4]). Previous empirical work finds strong support for the hypothesis that increased product market competition drives out costly discrimination in the long run (Black and Brainerd, 2004[5]; Hirsch, Oberfichtner and Schnabel, 2014[6]).

New OECD research by Ciminelli et al. (2021[7]) suggests that the scope for discrimination may also be larger in state‑owned than privately owned firms in the Central and Eastern European countries, and notably in Estonia, Latvia and Lithuania (Figure 3.2). In principle, this could reflect the larger scope for gender discrimination in state‑owned enterprises due to weak budget constraints (the public sector is excluded from the analysis). However, in other countries, the gender pay gap tends to be similar in state‑owned and privately owned firms. Weak budget constraints in state‑owned firms may not translate into larger gender wage gaps in those countries because of the stricter enforcement of equal pay laws, the presence of pay transparency laws and the greater importance of collective bargaining and social dialogue.

Figure 3.2. Gender pay gaps in state‑owned enterprises are large in Central and Eastern Europe

Note: The figure compares gender pay gaps in state‑owned enterprises relative to privately-owned firms. It shows log-percent differences in women hourly wages relative to men with the same level of educational attainment, age and tenure, in these two broad sectors. Gaps are obtained estimating, separately for each country, the following regression specification: ${\log (w}_i)={\beta }_{F}femal{e}_i+X_i{\beta }_1+I_i+{\epsilon }_i$, where log denotes the natural log operator, $w_i$ is hourly earnings of individual i, $femal{e}_i$ is a 0/1 gender dummy, $X_i$ is a vector of control variables, including age‑by-education dummies and two dummies for apprentices and casual workers, $I_i$ are 2‑digit level industry dummies and ${\epsilon }_i$ is the error term. The specification is estimated separately over the two restricted samples of state‑owned enterprises and privately owned firms. State‑owned enterprises are defined as firms under public control. The Health, Education and Public Administration sectors are excluded from the sample. The equation is estimated on the 2018 sample.

Source: Ciminelli, Schwellnus and Stadler (2021[7]), “Sticky floors or glass ceilings? The role of human capital, working time flexibility and discrimination in the gender wage gap”, https://doi.org/10.1787/02ef3235-en; based on the Eurostat Structure of Earnings Survey.

The findings in this box complement those by REGE (https://rege.tlu.ee/) who look at differences between foreign and domestic ownership and differences between firms in the public and private sector. REGE (2020[8]) found that gender wage gaps tend to be larger in foreign-owned firms and smaller in public-sector firms than in other firms. The higher gender pay gap among foreign-owned firms may reflect the possibility that gender wage gaps tend to be higher in high performance firms (see the discussion in Section 3.2), despite being exposed to international competition.

3.2. An in-depth analysis of the gender pay gap and the role of firms in Estonia

The role of firms in the gender pay gap can be analysed in more detail by decomposing the gender pay gap between male and female employees with similar skills, tasks and responsibilities into a between-firm and a within-firm component. The between-firm component captures the role of differences in pay practices between firms due to the sorting of women into low-wage firms (labelled “sorting”). The within-firm component captures the role of differences in pay practices for work of equal value (labelled “discrimination and bargaining”). The decomposition is implemented empirically using comprehensive linked employer-employee data on the universe of workers and firms in Estonia from the Tax and Customs Board Register for the period 2002‑18. For more detail on the methodology, see Box 3.2).

3.2.1. In Estonia, differences in pay practices within and between firms may explain up to half of the raw gender pay gap

Masso et al. (2022[9]) showed that the firm component of the gender pay gap associated with differences in firm pay practices within and between firms is substantial in Estonia and larger than in other countries for which similar estimations have been conducted. The firm component of the gender pay gap explains 35% in Estonia as opposed to 11% in France (Coudin, Maillard and Tô, 2018[10]), 16% in the UK (Jewell, Razzu and Singleton, 2019[11]), 21% in Portugal (Card, Cardoso and Kline, 2016[12]) for 26% in Germany (Bruns, 2019[13]) and 30% in Italy (Casarico et al., 2019[14]) Moreover, in Estonia, the firm component of the gender pay gap reflects in equal measure the within-firm (discrimination and bargaining) and the between-firm (sorting) dimensions, while it mainly reflects sorting in other countries (the contribution of discrimination and bargaining to the firm component of the gender pay gap varies from zero in France and Germany to roughly one‑third in Italy, Portugal and the United Kingdom).

New OECD results confirm that firm pay practices play a large role in the gender pay gap in Estonia (Figure 3.3), consistent with earlier results by Masso et al. (2022[9]) for a somewhat different time period.2 According to OECD estimates, firm pay practices account for about half of the gender pay gap. Three‑fifths of this is explained by the sorting of women into lower-paying firms (30% of the overall gender pay gap), whereas the remaining two‑fifths (20% of the overall gender pay gap) reflect pay differences between men and women with similar skills, tasks and responsibilities (discrimination and bargaining). The sorting component is comparable in magnitude with that documented in the cross-country analysis in Section 3.1 (about 25%). While the component associated with discrimination and bargaining is sizeable it appears to be small compared with the unexplained component of the gender wage gap in Section 3.1. This suggests a significant part of the unexplained gender pay gap is due to differences in tasks and responsibilities within occupations.

Box 3.2. Analysing the role of pay practices within and between firms

The role of firms in the gender pay gap can be analysed by decomposing this gap into a between-firm and a within-firm component, when male and female employees with similar skills, tasks and responsibilities are compared. The between-firm component (sorting) captures the role of differences in pay practices between firms in the gender pay gap between similarly-qualified men and women due to the sorting of women into low-wage firms. The within-firm component (bargaining and discrimination) captures the role of differences in pay practices between similarly-qualified men and women within firms for work of equal value.

The decomposition is implemented empirically using the approach put forward by Card, Cardoso and Kline (2016[12]). This first involves estimating wage equations with worker and firm fixed effects separately for men and women:

$\left(1\right) wit= {\alpha }_i+ {\mathrm{\phi }}_{J\left(i,t\right)}^{M,F}+{ \gamma }_t+ X_{it}^{\text{'}}{\beta }^{M,F}+ r_{it}$

where ${\alpha }_i$ denote worker fixed effects, which capture skill differences between male and female employees, ${\mathrm{\phi }}_{J\left(i,t\right)}^{M,F}$ denote gender-specific firm fixed effects, which capture differences in pay practices between firms as well as between men and women within firms, ${ \gamma }_t$ denote year fixed effects, $X_{it}^{\text{'}}$ denotes a third-order polynomial in potential work experience (age), and ${\beta }^{M,F}$ denotes gender-specific returns to potential experience.

To allow comparing the firm fixed effects between men and women, they are normalised with reference to the accommodation and catering sector. This is based on the assumption that there are no rents in this sector and hence firm wage premia are zero for both men and women. It has been verified that the firm fixed effects for both men and women are the lowest in the accommodation and catering sector. The normalisation also removes differences in tasks and responsibilities between men and women in the accommodation and catering sector. To the extent that differences in tasks and responsibilities are similar across sectors the analysis effectively abstracts from their role in the gender wage gap.

The role of firm pay practices – as measured by the gender-specific firm fixed effects – in the gender pay gap is quantified using the following decomposition:

(2) $E\left[{\mathrm{\phi }}_{J\left(i,t\right)}^{F}F\right]-E\left[{\mathrm{\phi }}_{J\left(i,t\right)}^{M}M\right]=\underset{discrimination and bargaining}{\underbrace{E\left[{\mathrm{\phi }}_{J\left(i,t\right)}^F-{\mathrm{\phi }}_{J\left(i,t\right)}^{M}F\right]}}+\underset{sorting }{\underbrace{E\left[{\mathrm{\phi }}_{J\left(i,t\right)}^{F}F\right]-E\left[{\mathrm{\phi }}_{J\left(i,t\right)}^{F}M\right]}}$

The first part on the right-hand side captures the role of differences in pay practices between men women within firms (discrimination or bargaining effect) or put differently how much the gender pay gap would change if women earned the same firm wage premium as men in the same firm. The second part on the right-hand side captures the role of differences in pay practices between firms (sorting effect), or put differently how much the gender pay gap would change if women worked in the same firms as men, weighted by male firm effects.

The analysis makes use of comprehensive linked employer-employee data on the universe of workers and firms in Estonia from the Tax and Customs Board Register. It covers the period 2002‑18 and all sectors of the economy. The sample is restricted to workers between 25 and 60 years old, earning at least the full-time minimum wage.

Figure 3.3. The role of firm pay practices in the gender pay gap in Estonia

Note: The figure presents the results from the decomposition of equation (2) in Box 3.2 for the average effect.

Source: OECD calculations based on Estonian Tax and Customs Board Register.

The relative importance of bargaining/discrimination in Estonia resonates well with previous research. For example, previous research attributes women’s lower ability to negotiate salaries as assertively as men to traditional gender norms (Meriküll and Mõtsmees, 2017[3]). Traditional gender norms are also closely intertwined with the prejudiced and stereotypical view that women cannot perform as well as men at managerial- and executive‑level positions. As such, they constitute an important driver of gender-based discrimination in the workplace. These harmful effects are compounded by a weak bargaining position of employees vis-a-vis employers in a context of flexible wage‑setting institutions (Masso, Meriküll and Vahter, 2022[9]). Estonia is characterised by a low minimum wage, a low unionisation rate and a weak coverage of collective agreements.

Larger gender wage gaps among high-wage workers reflect larger differences in pay practices between firms

The role of firms in the gender pay gap increases along the wage distribution, and particularly so over the bottom half (Figure 3.4). Among men and men and women in the top decile, firms account for as much as 70% of the gender pay gap. Differences in pay between firms explain about two‑thirds of the firm component, whereas the remaining one‑third is explained by differences in pay within firms. Among men and women in the bottom decile, the firm component of the gender pay gaps among low-wage workers is positive, as low-wage women tend to work in higher-wage firms than low-wage men, reducing the gender pay gap (this may also reflect the relative concentration of men in certain low-wage industries such as construction and agriculture, see Figure 3.6). There are no systematic differences in pay between low-wage men and women within firms.

Figure 3.4. The role of firms in the gender pay gap increases along the wage distribution

Note: The figure presents the results from the decomposition of equation (2) for the average effect and by decile of individual wages.

Source: OECD calculations based on the Estonian Tax and Customs Board Register.

Rising gender pay differentials along the wage distribution are not limited to Estonia, but tend to be present in the large majority of OECD countries (OECD, 2017[2]). However, so far the reasons for this pattern are not well understood. The present analysis suggests that gender wage gaps are higher for high-wage women because of their difficulty to access top jobs in high-wage firms. Indeed, top jobs in top firms are to a greater extent occupied by men than other jobs in top firms or top jobs in other firms (Chapter 2). This may also affect the bargaining position of high-wage women, and thus exacerbate differences in pay practices for work of equal value within firms.

Voluntary target setting, good management practices that make managers accountable and gender quotas for company boards are among the measures that could help to promote access for women to top jobs in top firms, while at the same time foster social norms that support gender equality. By contrast, the absence of a significant firm component among low-wage women or its tendency to reduce the gender pay gap suggests that strengthening collective bargaining or increasing the minimum wage is unlikely to be very effective in reducing the gender pay gap and potentially even counter-productive. Pay transparency measures hold more promise since these reduce the scope for discrimination against high-wage women and strengthen their relative bargaining position (OECD, 2021[15]).

Differences in the gender pay gap across industries tend to reflect discriminatory pay practices within firms, which tend to be more important in the presence of product and labour market frictions

The role of firm pay practices in the gender pay gap varies significantly across industries (Figure 3.5). The largest differences in firm pay practices are found in electrical, machinery and equipment and mining and quarrying, while the smallest differences are found in accommodation and catering, agriculture, and transportation. Differences in firm pay practices play an important role in explaining differences in the gender pay gap across industries. The correlation coefficient between the gap in firm pay practices and the overall gender pay gap across industries is 0.7. Differences in pay practices across industries mainly reflect differences in pay practices between men and women within firms (discrimination and bargaining) rather than differences in pay practices between firms (sorting). The corresponding correlation coefficients are 0.6 and 0.3 respectively.

Figure 3.5. The role of firms in explaining gender pay gaps by industry

Source: OECD calculations based on the Estonian Tax and Customs Board Register.

Differences in firm pay practices tend to be larger in industries with high “rents”, i.e. industries with high labour market frictions and low job mobility. Industry “rents” are measured as the average firm wage premium for men in an industry, i.e. the part of the average firm wage that cannot be explained by the composition of its workforce (Box 3.2).3 A low responsiveness of job mobility to wages confers wage‑setting power to firms and increases the scope for monopsonic gender discrimination (Hirsch, 2016[16]). The scope for gender discrimination may be amplified by weak competition in product markets (see Box 3.1). Frictions in product and labour markets thus have the potential to increase differences in pay practices between men and women within firms (discrimination and bargaining). This pattern is supported by the data. The correlation coefficient between industry rents (as measured by average firm wage premia) and within-firm differences in pay practices is 0.8 (discrimination and bargaining).

The large heterogeneity in the importance of differences in firm pay practices across industries suggests that sectoral initiatives may be a useful complement to nation-wide initiatives to tackle the gender wage gap. Targeted initiatives directed at specific sectors where discriminatory pay practices are more pronounced can be used to raise awareness about the extent of gender discrimination and provide information about its consequences for firms and workers. To the extent that discriminatory practices derive from weak budget constraints (product market competition) and are costly for firms, this can be used as an argument to make the business case for gender-friendly pay practices. By contrast, if discrimination derives from monopsony, this can be profit-maximising and there may be a greater role for public policies that enhance competition in the labour market.

The sorting of women into low-wage industries also contributes to the gender pay gap

The Estonian labour market is quite segregated across industries (Figure 3.6). Men are over-represented in industries such as construction, basic metals and mining, while women are over-represented in accommodation and food services, textiles, and education and health. Moreover, men tend to be overrepresented in industries paying above average wages, conditional on worker characteristics, whereas women tend to be overrepresented in industries paying below-average wages (see Section 3.1). In other words, there is a strong negative relationship between the share of women in an industry and the “average wage premium” (the correlation coefficient is equal to ‑0.5).

The sorting of women into low-wage industries is likely to reflect to some extent the role of past educational choices. While women outperform men in terms of the level of education (women are more likely to hold a tertiary degree), there are important differences with respect to the field of study. More specifically, fewer women than men complete Science, Technology, Engineering and Mathematics (STEM) degrees (OECD, 2019[17]). To some extent educational choices may reflect the possibility that teenage boys on average still perform marginally better in mathematics than girls, but gender stereotypes also play an important role. This points to the importance of steering more girls in STEM and thus addressing stereotypes that drive the educational choices of girls.

Figure 3.6. Gender segregation across industries and its contribution to the gender pay gap

Note: Firm wage premia refer to the part of average firm wages that is unrelated to the characteristics of the firm’s workforce. Firm wage premia are estimated through the inclusion of firm fixed effects in a wage regression described by equation (1). As explained in Box 3.2, firm fixed effects are normalised with respect to the accommodation and catering sector. The blue line gives the fitted line, while the shaded area provides the 95% confidence interval.

Source: OECD calculations based on Estonian Tax and Customs Board Register.

3.3. The gender pay gap over the life‑course and the motherhood penalty

3.3.1. Motherhood is associated with significantly higher gender inequalities, including higher gender pay gaps

Gender pay gaps evolve over the life‑course, both between and within firms, and career breaks play an important role in explaining why childbirth entails slower wage growth for mothers than for fathers. Childbirth entails significant long-term motherhood penalties in labour income in all countries, but with significant differences in their magnitude (Kleven et al., 2019[18]; OECD, 2017[2]). The long-run motherhood penalty, defined as the shortfall in labour income growth of mothers since childbirth relative to fathers, ranges from 21‑26% in Denmark and Sweden to 31‑44% in the United Kingdom and the United States and 51‑61% in Austria and Germany (Kleven et al., 2019[18]). There are also important differences across countries in the evolution of the motherhood penalty over time with regard to the components of labour income, i.e. employment, working time and hourly wages.

In a number of countries, including France, Italy, Japan, and Portugal, motherhood penalties in hourly wages, as opposed to labour income more generally, tend to increase with the time elapsed since childbirth, as fathers continue to progress to higher-wage firms and occupations, while women increasingly move to lower-wage ones (OECD, 2021[1]). Evidence for Germany suggests that about a quarter of the long-term motherhood penalty results from differences in the sorting of men and women across firms (Bruns, 2019[13]). Findings for France also point to an important role of sorting in the motherhood penalty and suggest that this is closely related to the tendency of young mothers to move to firms close to home and firms with flexible working-time policies (Le Barbanchon, Rathelot and Roulet, 2020[19]). However, in other countries, including Austria and the Slovak Republic, the rise in the gender wage gap around the age of motherhood is temporary (OECD, 2021[1]).

In Estonia, motherhood is associated with a temporarily higher gender wage gap

In Estonia, gender pay differences increase with age until the late thirties, but decrease afterwards (Figure 3.7). The gender pay gap peaks among workers aged 30‑34, but is similar for workers aged 45‑54 as those aged 25‑29. Complementary evidence by Masso et al. (2022[9]) further shows that women with young children (between 0 and 6 years-old) earn less than fathers, but that the gap falls as children grow older. The gender gap in earnings among women with adult children (aged 18‑24) is half that of women with a young children (aged 0‑2). They also show that that the gender pay gap increases with the number of children.

Figure 3.7. The role of firm pay practices in the gender pay gap by age group in Estonia

Source: OECD calculations based on the Estonian Tax and Customs Board Register.

The temporary increase in the gender pay gap around the age of motherhood does not appear to reflect the role of firms (Figure 3.7). Indeed, the components of the gender wage gap due to sorting across firms or bargaining and discrimination are surprisingly stable over the life‑course. The temporary increase in the gender wage gap around the age of motherhood is more likely to reflect the possibility that some young mothers temporarily shift to part-time. This explanation cannot be considered here because of the absence of information on working time in the data. A complementary explanation could be that the temporary increase in the gender wage gap reflects the role of career breaks regarding the wage progression of women.

Career breaks contribute to the motherhood penalty by slowing within-firm wage growth

The role of career breaks for the motherhood penalty is analysed by focusing on the incidence of non-employment spells for men and women around the age of parenthood (25‑34) and their consequences for wages. The analysis starts by documenting the gap in the incidence of non-employment spells between men and women as proxied by the number of years spend out of work (Figure 3.8). It shows that women aged 25‑34 are much more likely than men to experience a non-employment spell of one or more years (by 2 percentage points), while the opposite is observed for older women aged 35‑54 (the average age at which women in Estonia have their first child is 28) (Panel A). This suggests that non-employment spells provide meaningful information about the incidence of career breaks around the age of motherhood rather than differences in the risk of unemployment. Moreover, among workers around the age of parenthood (age 25‑34), the gap in career breaks is larger between men and men and women with a high earnings potential (Panel B).

Figure 3.8. The gender gap in career breaks by age and skills

Source: OECD calculations based on the Estonian Tax and Customs Board Register.

Career breaks tend to be associated with significant wage losses (Figure 3.9). Wage losses may reflect the slower upward mobility within firms due to a lack of experience and the possible depreciation of relevant skills or the sorting of women into lower wage firms following a career break. To examine this, the analysis separately considers changes in within-firm wages and changes in firm-wage premia. The evidence suggests that wage losses due to a lack of experience or human capital depreciation can be sizeable, particularly for women, and especially those with a higher earnings potential (Panel A). Sorting to lower wage firms does not appear to contribute to the motherhood penalty in Estonia (Panel B). This presumably reflects the possibility that most women return to the same firm following a career break. Career breaks after childbirth therefore tend to contribute the motherhood wage penalty by slowing their wage progression within firms.

In Estonia, women taking a career break following childbirth benefit from strong employment protection, with their position being guaranteed for up to 3 years. While this looks like a generous policy for women, the present analysis suggests that taking long career breaks involves considerable costs in terms of lower wages upon re‑employment, with possibly long-lasting consequences for the quality of worker careers.

Figure 3.9. The wage effects of career breaks

Source: OECD calculations based on Estonian Tax and Customs Board Register.

References

[4] Becker, G. (1971), The Economics of Discrimination, University of Chicago Press Economics Books.

[5] Black, S. and E. Brainerd (2004), “Importing Equality? The Impact of Globalization on Gender Discrimination”, ILR Review, Vol. 57/4, pp. 540-559, https://doi.org/10.1177/001979390405700404.

[13] Bruns, B. (2019), “Changes in workplace heterogeneity and how they widen the gender wage gap”, American Economic Journal: Applied Economics, Vol. 11/2, pp. 74-113, https://doi.org/10.1257/app.20160664.

[12] Card, D., A. Cardoso and P. Kline (2016), “Bargaining, Sorting, and the Gender Wage Gap: Quantifying the Impact of Firms on the Relative Pay of Women”, The Quarterly Journal of Economics, Vol. 131/2, pp. 633-686, https://doi.org/10.1093/qje/qjv038.

[14] Casarico, S. et al. (2019), What Firms Do: Gender Inequality in Linked Employer-Employee Data.

[7] Ciminelli, G., C. Schwellnus and B. Stadler (2021), Sticky floors or glass ceilings? The role of human capital, working time flexibility and discrimination in the gender wage gap.

[10] Coudin, E., S. Maillard and M. Tô (2018), Family, firms and the gender wage gap in France IFS Working Paper W18/01.

[16] Hirsch, B. (2016), “Gender wage discrimination”, IZA World of Labor, https://doi.org/10.15185/izawol.310.

[6] Hirsch, B., M. Oberfichtner and C. Schnabel (2014), “The levelling effect of product market competition on gender wage discrimination”, IZA Journal of Labor Economics, Vol. 3/1, p. 19, https://doi.org/10.1186/s40172-014-0013-1.

[11] Jewell, S., G. Razzu and C. Singleton (2019), “Who Works for Whom and the UK Gender Pay Gap”, British Journal of Industrial Relations, Vol. 58/1, pp. 50-81, https://doi.org/10.1111/bjir.12497.

[18] Kleven, H. et al. (2019), Child Penalties Across Countries: Evidence and Explanations *.

[19] Le Barbanchon, T., R. Rathelot and A. Roulet (2020), “Gender Differences in Job Search: Trading off commute against wage”, The Quarterly Journal of Economics, Vol. 136/1, pp. 381-426, https://doi.org/10.1093/qje/qjaa033.

[9] Masso, J., J. Meriküll and P. Vahter (2022), “The role of firms in the gender wage gap”, Journal of Comparative Economics, Vol. 50/2, pp. 454-473, https://doi.org/10.1016/j.jce.2021.10.001.

[3] Meriküll, J. and P. Mõtsmees (2017), “Do you get what you ask? The gender gap in desired and realised wages”, International Journal of Manpower, https://doi.org/10.1108/IJM-11-2015-0197.

[15] OECD (2021), Pay Transparency Tools to Close the Gender Wage Gap, OECD Publishing, Paris, https://doi.org/10.1787/eba5b91d-en.

[1] OECD (2021), The Role of Firms in Wage Inequality: Policy Lessons from a Large Scale Cross-Country Study, OECD Publishing, Paris, https://doi.org/10.1787/7d9b2208-en.

[17] OECD (2019), “Why don’t more girls choose to pursue a science career? PISA in Focus #93”, https://doi.org/10.1177/0956797617741719.

[2] OECD (2017), The Pursuit of Gender Equality: An Uphill Battle, OECD Publishing, Paris, https://doi.org/10.1787/9789264281318-en.

[8] REGE (2020), The Analysis of the Gender Pay Gap in Estonia - A Brief Summary, Tallinn University, Tallinn, https://rege.tlu.ee/wp-content/uploads/Reducing-the-gender-pay-gap-kvant-1-page-summary-quantitative-study-ENG-TP3-1-1.pdf (accessed on 29 March 2022).

Annex 3.A. Additional data on the gender pay gap across countries

Annex Figure 3.A.1.  Decomposing the gender pay gap across countries

Note: The chart reports estimates of the contribution of worker, firm and job characteristics to the gender wage gap, obtained estimating different specifications of the following equation: ${\log (w}_i)=\beta femal{e}_i+X_i\gamma +{year}_i+{\epsilon }_i\mathrm{ }\mathrm{ }$ where w_i is the hourly wage of individual i; $femal{e}_i$ is a dummy variable for women; $X_i$is a vector of additional individual-level characteristics, ${year}_i$ is a dummy capturing the survey wave’s year (2002, 2006, 2010, 2014 or 2018); and ${\epsilon }_i$ is the error term. The overall gender pay gap(black dots) is obtained estimating the β coefficient from a specification in which the cohort dummy is the only variable in vector X_i. The age component (light blue bars) is obtained as the difference between the β coefficients obtained from the first specification and an alternative one also including age bracket dummies (for age 14‑19, 20‑29, 30‑39, 40‑49 or 50‑59). The education component (blue bars) is obtained as the β coefficients obtained from the earlier specification and an alternative one in which age dummies are interacted with education dummies (one each for lower secondary, upper secondary and post-secondary). Industry and firm components (dark and light green bars respectively) are obtained similarly, including industry (at the 1‑digit NACE2 level) and firm dummies (obtained exploiting information on the firm identifier). Part-time and occupation components (orange and gold components) are obtained similarly, including a part-time dummy and occupation dummies (for each 2‑digit ISCO occupation), respectively. Finally, the unexplained component (red bars) is calculated as the β coefficient obtained from the most comprehensive specification.

Source: OECD calculations based on Estonian Tax and Customs Board Register.

Notes

← 1.  Employers’ discriminatory behaviour may be rooted in conscious or unconscious biases against women (“taste‑based discrimination”), may reflect the perceptions of employers that women are on average less productive than men (“statistical discrimination”) or be based on profit considerations by paying lower wages to women given that women enjoy more limited outside options than men due to their greater unpaid work burden (“monopsonistic discrimination”).

← 2.  Masso et al. (2022[9]) focus on the period 2006‑17 whereas the present paper focuses on the period 2002‑18.

← 3.  Using the average firm effects of women does not change the results. The correlation coefficient between the average firm fixed effects of men and those of women across industries is 0.94.